You are on page 1of 223

RRI (ISR)

The Rapid Response Instrument


toBankruptcy Risk
in theNon-Financial Sector:
Design and Implementation
Edited by
Piotr Adam Boguszewski

Warsaw 2014

This publication was prepared within the framework of Rapid Response Instrument project implementing within the framework
of Human Capital Operational Programme, Priority II. Development of human resources and adaptation potential of enterprises
and improvement in the health condition of working people, Measure 2.1 Development of human resources for modern economy,
Sub-measure 2.1.3 System support for increasing the adaptation potential of employees and entrepreneurs systemic projects
cofinanced by the European Union from European Social Fund.

This publication reflect only the opinions of the authors.

This publication is free of charge.

Reviewers:
Mirosawa Lasek, PhD
Jarosaw Grniak, PhD

Publisher:
Polish Agency for Enterprise Development
Ul. Paska 81/83, 00-834 Warszawa
Phone: +48 22432 80 80
Fax: +48 22432 86 20
biuro@parp.gov.pl,
www.parp.gov.pl

Copyright by Polish Agency for Enterprise Development, Warsaw 2014


ISBN: 978-83-7633-265-9
1st Edition
Circulation: 200 copies

Printed by:
Institute for Sustainable Technologies National Research Institute
K. Puaskiego 6/10, 26-600 Radom, Phone: + 48 48 36 442 41, Fax +48 48 36 447 65
e-mail: instytut@itee.radom.pl http://www.itee.radom.pl

Table ofcontents
Foreword ............................................................................................................................................................................................. 7
Jerzy Hausner
Introduction........................................................................................................................................................................................ 9
Piotr Adam Boguszewski, Jerzy Hausner

Part I. Bankruptcy theconcept, its measurement, causes and effects........................................................... 15


Chapter 1. THE CONCEPT OF BANKRUPTCY SELECTED ECONOMIC AND LEGAL ASPECTS ................................ 17
Piotr Adam Boguszewski, Marta Kdziorska
Chapter 2. MEASURING THE SCALE OF BANKRUPTCY IN THE ECONOMY ................................................................ 33
Piotr Adam Boguszewski, Jarosaw Kaczmarek, Konrad Kolegowicz, Pawe Krzemiski
Chapter 3. DETERMINANTS OF BANKRUPTCY AN OVERVIEW OF APPROACHES ................................................. 47
Piotr Adam Boguszewski
Chapter 4. ECONOMIC AND SOCIAL CONSEQUENCES OF BANKRUPTCY ................................................................. 57
Piotr Adam Boguszewski

Part II. Bankruptcy vs. social policy........................................................................................................................... 69


Chapter 5. AN OVERVIEW OF THE SECOND CHANCE POLICY ..................................................................................... 71
Anna ebrowska
Chapter 6. THE LEGAL FRAMEWORK FOR STATE AID ................................................................................................... 81
Mikoaj Stasiak
Chapter 7. THE FORMS AND INSTRUMENTS OF STATE AID ......................................................................................... 89
Mikoaj Stasiak

Part III. Models oftheRapid Response Instrument................................................................................................. 97


Chapter 8. THE DESIGN PRINCIPLES OF EARLY WARNING AND RAPID RESPONSE SYSTEMS
IN THE NON-FINANCIAL SECTOR .................................................................................................................. 99
Piotr Adam Boguszewski, Stanisaw Mazur, Marcin Zawicki
Chapter 9. RAPID RESPONSE SYSTEMS INTERNATIONAL EXPERIENCE .................................................................. 109
Stanisaw Mazur, Marcin Zawicki
Chapter 10. BANKRUPTCY RISK ASSESSMENT A MICROECONOMIC PERSPECTIVE ............................................. 117
Kamil Fijorek, Jarosaw Kaczmarek, Konrad Kolegowicz, Pawe Krzemiski
Chapter 11. MACROECONOMIC MODELLING OF BANKRUPTCY RISK ....................................................................... 133
ukasz Lenart, Baej Mazur, Krystian Mucha, Mateusz Pipie, Justyna Wrblewska
Chapter 12. THE STARTING MECHANISM ........................................................................................................................ 159
ukasz Postek
Chapter 13. COMMUNICATION IN THE RAPID RESPONSE SYSTEM ............................................................................ 167
Piotr Adam Boguszewski, Monika Dawid-Sawicka, Magorzata Kazubska

Part IV. Medium-term bankruptcy risk inthePolish economy forecasts and recommendations............... 175
Chapter 14. FORECASTS OF THE MACROECONOMIC SITUATION AND THEIR IMPACT ON BANKRUPTCY RISK . 177
ukasz Lenart, Baej Mazur, Krystian Mucha, Mateusz Pipie, Justyna Wrblewska
Chapter 15. BANKRUPTCY RISK AT ENTERPRISE LEVEL A MICROECONOMIC ASSESSMENT .............................. 191
Kamil Fijorek, Jarosaw Kaczmarek, Konrad Kolegowicz, Pawe Krzemiski

Bibliography .................................................................................................................................................................. 217


3

Ladies and Gentlemen,


At the times of progressive economic globalization, Polish companies are not only exposed to
various uncertainties in their local microeconomic and macroeconomic environments, but to
the fluctuations in the global and European economic situation, as well.
Conclusions drawn from the economic downturn in Europe are highly valuable to governments,
as they serve as directions in the shaping of economic policies and policies in support of
enterprises. The economic slowdown has shown that, today, we must be able to react quickly
to early symptoms of recession; the crisis has also proved that a flexible and effective economic
policy should not only be based on good planning, but primarily on being able to carefully
observe trends and uncertainties before companies or industries are caught in a spiral of
difficulties.
To minimize the effects of recession, states engage in different intervention measures. In 2007,
the European Commission launched the so called second chance policy. The conviction behind
the policy is to support those industries which are potentially endangered by business distress,
but are not yet "at risk" in the meaning of the guidelines on the provision of aid to secure and restructure enterprises. One of the
instruments to implement the policy is a rapid response tool.
It is better to monitor, issue early warnings and react fast, rather than cure the consequences, i.e. help the companies and industries
jeopardized by bankruptcy. The Rapid Response Instrument stems from such an approach. It is a pilot project developed by
the Polish Agency for Enterprise Development (PARP) and the Malopolska School of Public Administration, Cracow University of
Economics (MSAP UEK). The project shows that analytical tools, which facilitate decision making under conditions of increased risk,
are extremely important and from the practical point of view may lead to operational stability of Polish companies, especially
the small and medium-sized enterprises.
Experiences gained under the project are the source of numerous inspirations as to how a state may effectively intervene in the
period of economic slowdown. They are also a starting point for the development in Poland of a modern early warning and rapid
reaction system in the circumstances of economic instability.
PARP's systemic change is based on the belief that intervention should be replaced with prevention. I hope that this publication
will not only explain it further, along with the rapid response instruments mechanism, but will also become an inspiration to
develop future solutions.

Boena Lubliska-Kasprzak
President of the Polish Agency for Enterprise Development

Foreword
The Rapid Response Instrument (RRI, Polish acronym ISR) isa modern tool ofeconomic policy or, ina broader sense,
a tool inthedomain ofpublic policies concerned with short-term operational activities. It isan instrument expressly designed
toundertake focused public interventions, which, under certain conditions, seek toprevent thenegative effects ofeconomic
downturns related, inparticular, totherisk ofbankruptcy faced by a significant number ofmajor enterprises.
Contemporary economic policy (indeed any public policy) must be carried out simultaneously along several spatial and temporal
dimensions. Authorities that implement such policies must be able torecognize thefact that thedevelopments inquestion occur
simultaneously along those dimensions and torespond appropriately. Certain important phenomena intheeconomy occur
instantaneously, such ashuge cash flows, while others take anepoch, such asclimate change. Some appear locally, while others
have a global reach.
It appears, therefore, that economic policy should have atleast five time horizons: thecurrent, short-term one (28 quarters),
medium-term (210 years), long term (1020 years orone generation) and extra long-term orlong lasting one (over 20 years
oratleast two generations). And atthesame time, itmust be conducted along a number ofterritorial dimensions: local, regional,
national, cross-border, international, global and extra-global ones (e.g. problems related tospace exploration). If we were tobuild
a matrix (model) ofeconomic policy based on this assumption and use ittorecognize economic phenomena, itisclear that no
one will ever be able tograsp it(regardless ofthetechnical equipment attheir disposal). Thepoint isnot that such a pattern
should be used toanalyse everything that ishappening ata given time, but that itshould be applied totheproblems that appear
on theagenda ofeconomic policy and assuch constitute anarea ofits operation. It must be theresult ofa conscious choice,
a recognition what isreally important fortheoperation and development ofa given economy.
A specific public policy responds tothree kinds ofquestions: (1) What isthepurpose oftheaction tobe taken?, (2) What are
therules under which we operate?, and finally (3) How toact, by what means orusing which instruments? The third question
isrelated totheoperational dimension ofpublic policy. Without descending tothis level nothing can be achieved, yet bringing
policies down totheinstrumental (technical) level condemns ittofailure.
In order toappropriately answer these questions we need knowledge. It isa varied knowledge, one which isgenerated and
obtained ina different way. When we go down totheoperational level, we need expert, hands-on knowledge generated through
practical experimentation and research. In this case, invoking theconcept ofevidence-based policy isjustified. Armed with
such knowledge, we can select appropriate instruments (measures), calibrate them properly and verify theeffectiveness oftheir
application. Such knowledge can systematically accumulate over time thereby increasing theeffectiveness ofpublic interventions.
Instrumental knowledge essentially consists inprofessional expertise itisgenerated by people who specialize ina given field
ofstudy, moreover, its practical application also requires such specialization. It circulates primarily within a given occupational
group and among representatives ofdifferent professional groups. Policymakers who want touse itmust be experts themselves
and must be able towork with experts.
The problem isessentially posed by correctly targeting economic policy, which occurs attheoperational level and consists
ininstrumental and parametric action. In this case, thedecision-maker shaping economic policy primarily seeks toaffect thecurrent
behaviour ofmarket participants. Thedesired response horizon isrelatively short, although interventions inthis area generally tend
tohave longer-lasting effects. Thedecision-maker operates by producing specific and specifically targeted impulses; therefore he
must adopt a certain response model and toa greater orlesser extent adjust toit theplanned actions.
This particular dimension ofeconomic policy islegitimized by thefact that themarket economy operates through business
cycle fluctuations. It ischaracterized by inherent processes ofan oscillatory nature, but there are also automatic business cycle
regulators. Accordingly, all interventions along this dimension ofthemarket economy may be fairly effective, but must be cautious
and calculated so asnot todisrupt thenatural economic cycle without a good reason. A serious disruption oftheexisting balance
primarily on themacroeconomic but also on themicroeconomic scale, especially one that may lead toa deep economic downturn,
isperceived assuch a good reason. In this respect, theobjective ofeconomic policy isto mitigate thecyclic fluctuations. Thepolicy
isnot so much meant tostimulate theeconomy astolimit its fluctuations which may throw itoff its natural cyclical balance
and result ina loss ofits self-steering capacity. Hence theobjective ofeconomic policy inthis area tends tobe a negative one
toprevent theoccurrence ofconditions which can be defined using certain parameters. A positive aspect ofsuch a policy consists
inensuring thenecessary level ofstability interms ofinternational openness and globalization ofeconomic flows.
7

A good example ofthis kind ofpolicy ismonetary policy. In this case aslong asthepolicy iscarried out ina conventional way
we parametrically designate a direct target and determine thetolerance band, which should be interpreted asa justification
forabandoning corrective interventions, since inflation remaining within thedesired range indicates that themarket mechanism
isefficient. Here we are dealing with a clear example ofevidence-based policy. It isimportant, however, tonote that insuch
a case itismore important not tointervene provided that inflation remains within theprescribed range intheperspective under
consideration (in this case itisa short-term perspective). It istherefore anexample ofevidence-based non-policy. In thecase
ofa policy that affects theeconomic cycle, theprimary aim isnot totake unnecessary action, especially, todo no harm. Action
isto be taken only when necessary, ideally asa warning signal (an Early Warning System, EWS) rather than a follow-up intervention.
The idea ofa Rapid Response Instrument (RRI) consists intransplanting this way ofthinking into thearea ofthereal economy
and thesphere ofentrepreneurial activity. In general, itisbased on thebelief that certain public policy response algorithms can
be developed, asisthecase with fiscal policy and other aspects ofcurrent (cyclical) policy. TheRRI isprimarily about thehigh
bankruptcy risk and theresulting potential disruptions on thelabour market. TheRRI ismeant toeliminate thediscretionary aspect
from theresponse ofpublic authorities inthis area, todetermine its admissibility and offer theright tools. Thebelief that during
aneconomic slowdown orrecession intervention inthesphere ofbusiness activity can be avoided isobviously false. Experience
consistently shows that such interventions tend tocome too late, be disorganized, expensive and mostly ineffective.
Undeniably, a policy geared towards preserving economic stability cannot be separated from its other dimensions and objectives,
particularly from thestructural policy, but thelatter has different aims and methods ofoperation. Only by accomplishing its
primary mission ensuring stability oftheeconomy can business cycle policy promote theachievement oftheother objectives
ofoverall economic policy, such aseconomic growth. If itisdirectly subordinated tosuch objectives, itoften results ina disruption
ofeconomic stability and fails toachieve those objectives inthemedium term.
The difficulties inexecuting economic policy along theoperational dimension involve thefact that we cannot easily distinguish
between cyclical phenomena cyclical oscillations around equilibrium points deemed tobe natural fora given economy from
structural change processes that result inshifting theequilibrium points. Moreover, conducting aneffective economic cycle policy
requires comprehensive expert knowledge and not inconsiderable skills toapply itinpractice. It also requires something else,
something that isdifficult toachieve inparliamentary-cabinet political systems theindependence ofexpert and operational
activities from thepolitical rivalry forpower. What works fairly well inthecase ofmonetary authorities and independent central
banks, isnotoriously difficult toapply toe.g. fiscal and government structures. Hence theidea ofcreating national fiscal councils
based on themodel ofmonetary policy councils.
Today a good economic policy stems from thebelief that its structural dimension ismore important than its cyclical dimension.
Ifeconomic policy isto be effective, itmust be carried out within theinstitutional order rather than within theoperational order
since theobjectives ofstructural policy cannot be achieved within thestimulus-response pattern. They may not be induced
directly, because they result from much more complex relationships, including interactivity, i.e. theprocess inwhich crucial
roles are played by communication, interpretation and reflective learning. This process leads toan identity transformation on
thepart ofparticipants intheeconomy and only thus may bring about thedesired and expected economic changes. These
are developmental, not oscillating changes, and, assuch, disrupt thebalance oftheeconomic cycle. They are not automatically
reversible and shift thenatural economic balance points. We do not deal with these issues inthis publication, even though inour
quarterly research reports we invoke them ina systematic and practical manner.
The RRI also constitutes a tool that facilitates appropriate matching ofthemacro- and microeconomic public policy planes.
Theinitial analyses are carried out atthemicro level and concern thedegree ofbankruptcy risk faced by enterprises representing
individual NACE sections and regions ofPoland. When combined with supplementary macroeconomic analyses, they permit
theformulation ofinferences regarding thegeneral economic climate inPoland. Analyses prepared within theframework ofour
project are based on original econometric models developed by our team and are explained inmore detail below.
The results ofeconometric modelling are regularly commented upon and discussed by a panel ofeminent experts, who
themselves are excellent economic analysts, represent varied backgrounds and expertise inthepractical and theoretical aspects
ofeconomic policy. As a result, we provide decision-makers both inpublic administration and inenterprises with reliable analyses
and recommendations.
The practical application oftheRRI requires various kinds ofexperts toliaise with decision-makers. Theinstrument itself proves
tobe useful only when professional expertise iscombined with good communication and a high level oftrust. Thus, theRRI
constitutes a good model tool ofmodern public governance asit facilitates combining thepowers ofa public authority (the state)
and themarket.
Jerzy Hausner

Introduction
Bankruptcies ina market economy constitute aninevitable and insome cases even a beneficial phenomenon since they
lead oratleast should lead tothepermanent elimination ofinefficient enterprises, which may contribute toa better use
oftheir resources inother ways. Therecent global financial crisis, however, painfully reminded us that one ofthemost serious
effects ofstrong negative shocks may be a significant increase inthescale ofbankruptcies intheeconomy and their economic
consequences. Such events demonstrate that insuch circumstances thesaid phenomena and processes may affect not only
single entities, but even entire sectors and give rise tounacceptable social consequences. Additionally, individual victims
ofsuch disturbances are not only firms ofminor importance; crises may also reveal structural flaws orother kinds ofproblems
inthelargest enterprises, which would find iteasier todevise survival strategies orotherwise adapt inmore normal times.
Economic problems affecting large companies usually cause further consequences via theso-called domino effect, resulting
insecondary waves ofbankruptcies (e.g. among cooperating entities) and a sharp rise inunemployment thus triggering a vicious
circle ofproblems. A spectacular example ofsuch a situation was thebankruptcy ofGeneral Motors announced on June 1, 2009.
TheU.S. administration considered thepotential consequences ofthecompanys liquidation tobe so serious that on July 11 ofthat
year thecompany was bought by theTreasury Department. Theaid granted by theU.S. administration reached theastronomical
amount ofapproximately 49.5 billion dollars (of which ca. 38 billion was subsequently recouped), but themove isestimated tohave
saved a total ofapproximately 1.2 million jobs.
Bankruptcies may cause serious economic and social problems not only intheeconomy ofa given country asa whole, but itmay
also exert extremely detrimental local effects on individual cities, municipalities orcounties. Such situations occur intheevent
ofbankruptcy ofthelargest local employer orif insolvency problems are experienced by several smaller businesses atthesame
time. Poland has seen its share ofsimilar developments. One ofthebest known examples are theconsequences oftheliquidation
in200809 ofthetwo main employers inapy (a city intheRegion ofPodlasie) oftheRolling Stock Repair Works a company
with almost 150 years oftradition and theclosure ofthelocal sugar factory.
Both thepossible undesirable macroeconomic effects aswell aslocal bankruptcies raise thequestion ofwhether and how
economic policy should respond. It cannot be easily answered, not least due tothefact that economists are sharply divided
on thesubject. However, despite such differences ofopinions one thing appears tobe beyond dispute itisalways worth
knowing inadvance where and how large therisk ofbankruptcy is, and what its consequences may be, including its economically
quantifiable costs. Thus, early warning systems against bankruptcies appear toembody universal values. Owing totheeffects
oftherecent financial crisis mentioned above and thecurrent level ofdevelopment ofthePolish bankruptcy system, which still
has a number ofelements that reduce its effectiveness, itisreasonable todevelop not only nationwide warning solutions, but
also those comprising procedures aimed atan active economic policy response tosuch events known asrapid response systems.
The construction ofsuch a system isa complex, multi-layered task, which requires a proper synergy ofanalytic and research
procedures, business practices and institutional legitimacy. Thepilot systemic project called theRapid Response Instrument (RRI,
Polish acronym ISR) constitutes anundertaking ofthis kind, which takes into account theabove prerequisites. This project was
executed from April 1, 2009 toJune 30, 2014 by thePolish Agency forEnterprise Development inpartnership with theMalopolska
School ofPublic Administration attheUniversity ofEconomics inKrakow under theHuman Capital Operational Programme
Priority II: Development ofhuman resources and adaptation potential ofenterprises and improvement inthehealth condition
ofworking people financed by theEuropean Social Fund. Its primary purpose isto create a comprehensive system ofsupport
forenterprises and their employees aimed atmitigating thenegative impact ofeconomic downturns on theeconomic situation
ofthese entities, including threats tothecontinuity oftheir operation. Theaim ofthescheme isto ensure positive outcomes
ofsuch interventions both ex-post and, tothegreatest possible extent, toprevent untoward developments.
As was already mentioned, itisa challenging task both interms ofdesign and implementation. Therefore theproject combined
inputs from several closely related areas and consisted ofseveral stages. Thepurpose ofthis monograph isto describe its main
pillars both inthearea ofits starting premises and thesolutions developed. Since even such a narrowing down ofthescope
ofwork due tothecomprehensive character oftheissue would have been insufficient foraneffective treatment ofthetopic,
certain additional restrictions have been adopted. First, we focused on thepresentation oftheabsolutely crucial components from
thepoint ofview ofthestructure ofthesystem. In this context, thefollowing appear tobe particularly significant:

The premises justifying theconstruction ofsuch a system,

Feasibility ofits objectives, including appropriate sets ofinstruments,

Admissibility ofactive measures implied by theRRI under theexisting law, especially due topossible conflicts between
theinterventions themselves and theprinciple offree competition,

The crucial methodological solutions and tools adopted inthesystem,

The key empirical results arising from theapplication ofmethods and tools.
Second, when choosing theareas forfurther study, we assigned high priority totopics asuniversal aspossible toensure that
thesolutions developed could be used by different institutional structures. In thecase oftheRRI, itisparticularly important that
such solutions can be included inbusiness practice ina manner largely independent ofthevarying political and institutional
circumstances.
Third, particular attention was given tosolutions developed not only inconceptual terms, but forwhich we managed tocollect
a sufficiently large body ofexperience during theproject and which yielded results that may already provide thebasis forspecific
business recommendations.
Considering that one ofthefirst and essential steps oftheRRI project istheresearch component, itwas given themost extensive
treatment inthis monograph. Due tothegeneral objective ofRRI systems, itisparticularly important topredict and determine
thelevel ofrisk ofsuch distortions that justify theimplementation by thegovernment and its agencies ofspecial preventive and
assistance measures on thescale ofthenational economy and its various sectors. Thenatural centre ofgravity oftheresearch
component presented here therefore involves problems associated with measuring and forecasting such risk.
The extent and complexity oftheRRI, on theone hand, and organizational factors, timing, and volume associated with
thepreparation ofthis monograph on theother hand, meant that despite thestrong substantive links between its parts thebook
does not constitute a complete presentation ofRRI system concepts, while individual chapters are mainly original essays devoted
tokey issues. It should also be noted that despite thebreakdown ofissues related totheRRI, their complexity ina number
ofchapters isstill considerable enough that their subject matters themselves could constitute thebasis forfurther extensive
investigations. As a result, our limited resources meant that a number ofthematic areas may have only been acknowledged
inpassing inindividual chapters, although inthis case they are usually supplemented by suggestions forfurther reading.
The monograph consists offour sections containing a total offifteen chapters. Thelayout and design are subject totheprinciples
stated above.
Part I: Bankruptcy theconcept, its measurement, causes and effects begins thediscussion oftheconcept ofbankruptcy
itself. We demonstrate that from aneconomic point ofview itisambiguous, while its criteria and their verification procedures
are, toa certain extent, arbitrary. Such conventionality results from several factors, but themain one istheimportance ofcertain
axiological choices underlying thedesign ofa bankruptcy system. Thechoices affect especially thefundamental goals ofinsolvency
proceedings which, given theessence oftheproblem, may grant more prominence either totheprotection ofcreditors ordebtors.
In view oftheabove-mentioned dilemmas, we attempt totake a more general look attheissue ofthebankruptcy system.
From thepoint ofview oftheeconomy, bankruptcy systems can be built on different principles, hence itisdifficult toobjectively
indicate thebest, model solutions. In business practice, however, itisnecessary toendow such systems with a dimension
ofuniversal applicability. Such a function, among others, isfulfilled by bankruptcy law, whose Polish model isalso outlined
inthis chapter. It should be emphasized that thesolutions adopted inthelegal system have a number ofconsequences relevant
totheobjectives and thepossibilities ofbuilding early warning and early response systems. Here we mainly emphasize thefact
that thechoice ofspecific legal procedures within a given socio-economic system isrelated totwo basic outcome parameters
ofbankruptcy procedures, which remain inharmony with each other, namely their cost effectiveness and duration.
Even unsatisfactory efficiency ofbankruptcy procedures does not constitute sufficient reason fortheconstruction ofearly
warning/early response systems (EWS/ERS) aslong asthey affect only individual, isolated cases, since their total social costs
are not big enough tojustify interventions. Thediscussion presented inChapter 2 ofPart I isthus anattempt todetermine
thescale ofthephenomenon, including theconstruction ofvarious ways ofmeasuring it. This would not be possible without
a deeper analysis oftheinformation systems on enterprise demographics, including bankruptcy. Theresults obtained demonstrate
that thePolish system still has a long way togo before itachieves theresults considered asstandard inother countries. Due
tothesolutions adopted inPoland, including those arising directly from theprovisions ofthebankruptcy law, itisquite possible
that theavailable data on theincidence ofbankruptcy significantly underestimate its magnitude. Early warning systems can
therefore fulfil anadditional function by expanding thescope ofinformation regarding theactual scale oftheproblem and its
intensity indifferent socio-economic cross-sections.
The necessary condition fortheconstruction ofany warning system istheawareness ofthecircumstances and mechanisms
forpredicting therisk ofinterest tothesystem. Such a system requires theprovision ofrelevant advance information assoon
aspossible, which allows fora timely implementation ofEWS activities. Systems that contain anadditional response component
also require appropriate instruments with which toinfluence thephenomenon inquestion. For this reason, Chapter 3 contains

10

anoverview ofresearch results inboth areas. Thechapter ismostly synthetic because ofextensive worldwide research inthis
area and noteworthy achievements inPoland. Such a synthesis ispossible since, despite theabundance ofresearch publications,
interms ofmethodology they belong toonly a few principal categories. Due totheanticipated use oftheplanned EWS, a strong
emphasis was placed on macroeconomic circumstances ofbankruptcy, including thephenomenon ofsynergy ofindividual
bankruptcies. Given that many models ofbankruptcy are constructed primarily formicroeconomic use, including theprediction
ofbankruptcies ofindividual companies, such anenhancement appeared tobe necessary. In particular, itisshown that
thecorrelation oftheeconomic condition ofseveral entities may considerably increase thelikelihood oftheir collapse.
In thefinal chapter ofPart I we strive todeepen theanalysis ofbankruptcy costs and against this background toexamine
theefficiency oftheprocess. It appears that one ofthemost common reasons forunderestimating theimportance ofEWS/
RRI systems inthearea ofbankruptcy isrooted inoverestimating theefficiency ofthenatural selection mechanisms while
underestimating themagnitude ofcosts that arise when such mechanisms work less than perfectly. It isimportant that thereason
forsuch anunderestimation isquite often thefact that bankruptcy costs are often limited only totheso-called direct costs. Likewise,
not all indirect costs are recognized, and even if they are enumerated appropriately, itisoften extremely difficult tomeasure some
ofthem inpractice. As a result, they tend tobe overlooked inthecost-effectiveness account. Moreover, theanalyses presented
inChapter 1 demonstrate that bankruptcy law invariably comprises a number ofelements that may limit its effectiveness and
generate excessive costs. If this conclusion issupplemented with thedata on theactual scale ofbankruptcy inthePolish economy,
theissue ofadvisability ofdeveloping RRI-type solutions appears tobe settled. This statement was verified empirically using data
on theeffectiveness ofthebankruptcy process inPoland, which show that although our country ranks average inthis respect,
there isstill a considerable distance separating us from thevery best, including several European countries.
Part II ofthemonograph Bankruptcy vs. public policy isdevoted totheissues ofpotential opportunities available within
public policy inreducing thescale ofbankruptcy risk and its consequences. We note that theexisting barriers arise not only
from thefinancial constraints which we always deal with inthepublic sphere, but also from another fundamental problem. One
ofthemost obvious effects ofstate aid directed toindividual companies granted with a view toproviding temporary support and
enabling them toregain full economic efficiency after surviving a difficult period, may be thedistortion oftheprinciple offree
competition incertain areas oftheeconomy. State aid may therefore be granted only totheextent and insuch a way that such
a distortion does not occur, and if itdoes occur, itshould be within anacceptable scale and ina controllable form.
One ofthesolutions that meet theabove-listed criteria iscalled thesecond chance policy. It aims, among others, toincrease
thesurvival rate ofbusinesses and help those who have suffered a business failure tostart a new business. These issues are dealt
with inChapter 5 ofthemonograph, which characterizes, among others, theassumptions behind this policy ensuing from relevant
EU regulations, aswell astherecommendations oftheEuropean Commission. It offers a comparative analysis oftheconsequences
ofjudicial and extrajudicial procedures concerning such entities. This constitutes thebackground fora brief outline ofthespecial
role ofEWS/RRI-type systems inproviding effective support forsuch a policy.
In view ofthefact that state aid constitutes aninfluential intervention instrument, and, aswas already mentioned, itcannot
interfere with theprinciples offree competition, appropriate regulations constitute anextremely important component
ofthesecond chance policy. Chapter 6 ofthebook isdevoted tothese issues. First ofall, itdefines state aid inits legal sense,
aswell astheconditions forits admissibility. Theregulatory aspect plays a crucial role intheactual practice ofRRI-type systems,
since under EU regulations state aid isinadmissible inprinciple. Nevertheless, itmay be granted inspecific situations detailed inthis
chapter. Theaccompanying legal analysis, however, concerns not only theconditions oftheTreaty (Treaty on theFunctioning
oftheEuropean Union), but also thenational legislative frameworks governing theprovision ofsuch aid.
Part II ofthemonograph ends with a chapter devoted tostate aid instruments and thesynthetic evaluation ofits scale and scope
inPoland with particular attention paid tode minimis aid. It also presents therules forregional, horizontal and sectoral aid. It
isworth noting that ina RRI system interventions are frequently triggered by thesocial consequences ofbankruptcy attheregional
orsectoral level. Theanalysis ofstate aid inPoland shows a clear need fora programme forrescuing and restructuring firms
intheSME sector. In particular, itisnecessary todevelop systemic solutions inthearea ofstate interventions with respect toSMEs
indifficulty sector. TheRRI project isan example ofan attempt toaddress this need.
Part III ofthebook Models oftheRapid Response Instrument istechnical innature and includes both a discussion ofvarious
models ofsuch a system and a description oftheir fundamental structural components. It begins with Chapter 8 outlining thebasic
aspects ofearly warning systems. Theproposed approach isbased on theso-called cost avoidance rule, which means that
thesystem isactivated only when thecosts ofinaction are greater than theexpenses incurred inthereverse situation. Animportant
element ofthediscussion inthis chapter isthecomplex matter ofselecting theoptimal sensitivity level ofthewarning system.
Ifthesystem isdesigned so that itonly responds tovery expressive stimuli, thenegative effect may involve omitting theearly and
often faint indications oflater disturbances. If, inturn, such a system istoo sensitive, itwill raise false alarms. This may undermine
theconfidence ofEWS stakeholders intheoperation ofthesystem and asa result, lead toignoring information about theactual
11

risks and threats. Thechapter also discusses examples and classifications ofresponse systems, aswell astherelationships between
early warning systems and early response systems. Thechapter ends with theanalysis ofeffective implementation ofsuch systems.
Quite often, this extremely important element tends tobe overlooked insimilar projects, which results ina limited practical
effectiveness ofsuch aninstrument, oreven its complete absence. On theother hand, we must emphasize very strongly that such
circumstances tosome extent determine therange ofpossible models and tools tobe used inan EWS. What itmeans inpractice
isthat there do exist more accurate kinds ofanalysis orforecasting ofcertain phenomena, but they are oflimited practical utility
from thevantage point ofEWS construction. This may occur, forexample, if such models are too complex interms ofsocial
communication, their preparation istakes too much time orthey are expensive tooperate.
The next chapter (Chapter 9) provides anoverview ofthemost important practical experiences oftheUnited Kingdom and
theUSA intheconstruction and application ofEWS/RRI-type systems. This review isvaluable foratleast a handful ofreasons. First,
such systems are not very common, which means that intheprocess ofdevising a new one itisdifficult torefer toa repository
ofready-made solutions and practices. Likewise, and quite understandably, literature on thesubject isstill fairly limited. Hence
theconstruction ofsuch a system inPoland isinmany ways a pioneering challenge. Second, thereview reveals a certain important
feature ofthese systems although they should ideally be created inorder toachieve certain specific objectives within a pragmatic
structure ofaction, they often reflected attempts tofind immediate solutions toserious problems. TheRRI project isalso unique
inthat itprovides a consistent structure operating from top tobottom, yet atthesame time capable ofdynamic adjustments.
Third, international experience, inspite ofits limitations and shortcomings, may offer certain insights relevant totheconstruction
ofsuch a system inPoland.
The next two chapters (10 and 11) contain detailed accounts ofdata sources, models and procedures fortheassessment and
prediction ofbankruptcy risk inthemicro- and macroeconomic perspectives, respectively. It isworth noting that although fairly
classical solutions outlined inChapter 3 have been applied inmicroeconomic analyses, theapproach adopted intheproject has
two distinguishing characteristics, especially incomparison with other microeconomic bankruptcy models. First, themodels used
here were estimated using much larger learning data sets than those presented inother publications. Second, itmakes effective
use ofthedouble forecasting method. Traditional approaches utilize current economic data on businesses inorder toanticipate
their problems on thebasis ofmodels which allow them todo so effectively within thetime horizon ofapproximately one
year. Numerous international experiences show that this class oftools rarely permits thehorizon tobe extended since such
attempts usually result ina considerably decreased accuracy offorecasts. In early warning systems this isunacceptable, since
itcompromises their credibility. On theother hand, forecasting risks well inadvance isextremely important forwarning systems.
In theapproach discussed here itwas thanks tothedevelopment ofeffective procedures forforecasting theexplanatory variables
inbankruptcy prediction models. An additional advantage ofthesolutions adopted inthemicroeconomic block isthecapacity
toobtain bankruptcy risk assessments across different sections oftheeconomy (including theterritorial ones) and thedirect
costs ofsuch risk (excluding theelements resulting from economic linkages) forvarious cost categories (loss ofjobs, decreased
industrial output, etc.).
The chapter on macroeconomic analyses ofbankruptcy risk focuses on three groups oftopics. First, itcontains and overview
ofmethods foranalysing theeconomic cycle with a particular emphasis on thespecific features ofthePolish economy. Second,
itpresents thedesign principles underlying theforecasts ofthebasic macroeconomic indicators inPoland. International experience
shows that both factors are extremely important inpredicting bankruptcy risk intheentire economy. Third, itcontains a description
ofthescenario analyses methodology oftheimpact ofpotential impulses affecting Polands economy on its performance,
especially intheenterprises sector. Analyses ofthis kind are especially important fortheRRI foratleast two reasons. Firstly, they
allow researchers toassess thefuture effects ofthebasic kinds ofmacroeconomic shocks. Secondly, they offer a convenient tool
foraninitial assessment ofcertain macroeconomic policy options from theperspective oftheir impact on economic growth and
thedegree ofbankruptcy risk.
It isworth noting that inthemacroeconomic analyses and forecasts block theauthors adopted measures toensure thescalability
ofcertain tools inorder toensure their applicability not only atthelevel oftheentire economy (macroeconomic aggregates), but
also mesoeconomic ones (individual industries orsectors). Theselection oftools was also guided by other principles contributing
totheeffectiveness oftheEWS. Substantial emphasis was placed, among others, on thetransparency ofsolutions, which included
on theone hand good communicability ofprocedures, and on theother hand, thehighest possible intersubjectivity ofthose tools,
including their significantly greater resistance toexogenous assumptions than isthecase with theclassical structural approaches.
It should be noted, however, that these principles were used intheforecasts ofmicro- and macroeconomic bankruptcy risk
ina dynamic way, i.e. inorder toensure that thetools based on them could be subjected toongoing verification from thepoint
ofview oftheir relevance and where necessary tothenecessary modifications. Both chapters include a documentation
oftheprocedures used.

12

The subject matter ofChapter 12 istheso-called starting mechanism, i.e. a set ofprocedures on thebasis ofwhich decisions are
made regarding when tostart implementation activities inspecific enterprises qualified forsupport under theRRI. Themechanism
employs, among others, information provided by themicro- and macroeconomic components oftheProject. Thesolutions
used are based on theprinciple ofa three-level scoring system which provides a synthesis ofinformation from themacro-,
meso- and microeconomic components. We have also adopted theprinciple that thesystem should make use ofall available
information, including feedback from outside theRRI. As a result, theprimary scoring based on data generated by theRRI modules
issupplemented by anauxiliary information system.
Part III ofthemonograph ends with Chapter 13 devoted totheissues ofcommunication within theRRI system. Let us start by saying
that, paradoxically, numerous warning systems neglect oreven ignore them, although forobvious reasons good communication
isa prerequisite fortheeffectiveness ofsuch systems. At least three interlinked characteristics ofthecommunication processes must
be analysed: thecredibility ofthemessage, its timeliness and coverage. All these elements are subject tocontinuous monitoring
within theRRI. Chapter 13 explains both theconstruction principles ofthis systemic functionality and theaccompanying ratings
ofthecommunication process. It also provides a description ofa unique solution devised fortheRRI, namely theso-called
interpretation panel. Thepanel consists ofa group ofeminent experts intheareas covered by RRI analyses, whose responsibilities
include: a systematic validation oftools and solutions employed inthesystem, qualitative and quantitative evaluations ofanalyses
and forecasts generated inmacro- and microeconomic blocks aswell asanticipatory inspirations.
The final, fourth part ofthemonograph Medium-term bankruptcy risk inthePolish economy forecasts and recommendations
presents theevaluations and predictions ofbankruptcy risk obtained using themacro- and microeconomic modules covering
both theentire economy and its selected sub-aggregates. In particular, thepart offers a thorough analysis ofthecurrent situation
and levels ofbankruptcy risk broken down by individual divisions oftheeconomy. A particularly important result ofthesynergies
ofmacro- and microeconomic procedures developed within theRRI istheso-called Top 10 a list ofsectors most vulnerable
tobankruptcy, yet atthesame playing a crucial role inthePolish economy. Within themicroeconomic component theauthors have
included heat maps forindividual NACE divisions reflecting thedegree ofretrospective bankruptcy risk, i.e. from 2007 to2013. Such
maps permit, among other things, aneasy identification ofareas ofbusiness activity consistently exposed toa high bankruptcy risk.
In conclusion ofthis brief introduction itshould be added that, although most authors ofthis monograph are professionally
affiliated with institutions that directly orindirectly deal with theproblems ofbankruptcy intheeconomy, theviews expressed
here are individual and represent only personal beliefs, including those developed during theProject.
Apart from the authors of the present monograph, a number of people, who unquestionably formed an effective interdisciplinary
research and project team, were involved in the RRI (ISR) project. Here we wish to thank all the project participants for their
contributions:





Anna wiebocka-Nerkowska, Head of the Project (PARP); Jerzy Hausner, Project Coordinator (MSAP).
Stanislaw Mazur, Project Manager (MSAP); Micha Bonin, Project Coordinator (PARP).
Piotr Boguszewski, Lead Expert on Research; Marcin Zawicki, Expert on the Analysis of the Legal and Institutional Environment.
Members of the Microeconomic Analysis Team: Kamil Fijorek, Jarosaw Kaczmarek, Konrad Kolegowicz, Pawe Krzemiski.
Members of the Macroeconomic Analysis Team: ukasz Lenart, Baej Mazur, Krystian Mucha, Mateusz Pipie, Justyna
Wrblewska.
Members of the Team for the Preparation and Modification of Methods and Tools for the Integration and Interpretation of
Results of the Monitoring of the Economy in the Micro- and Macro-level Components: Kamil Fijorek, Jerzy Marzec, Jacek
Osiewalski, Andrzej Sokoowski.
Members of the RRI (ISR) Development and Implementation Team: Irena Herbst, Tomasz Jagusztyn-Krynicki, Mikoaj Stasiak,
Marek Szczepaski, Marcin Zawicki.
Experts in the RRI (ISR) action research component: Mariusz Andrzejewski, Pawe Biaynicki-Birula, Edyta Bieliska-Dusza,
Jan Brzozowski, Marek wiklicki, Marek Dziura, Konrad Grabiski, Magdalena Jelonek, Jacek Klich, Konrad Kolegowicz, Pawe
Krzemiski, Micha Kudacz, Norbert Laurisz, ukasz Mamica, Piotr Markiewicz, Wojciech Pajk, Tomasz Rojek, Tomasz Sapeta,
Wojciech Szymla, Krzysztof Wach, Krzysztof Woniak, Marcin Zawicki, Andrzej Zygua.
Coordination team for the RRI (ISR) action research component, including experts in administration and logistics: Tomasz
Geodecki, Jarosaw Plichta, Sylwia Kodras, Rafa Sukowski.
Research Assistants: Piotr Kopyciski, Marek Oramus; Specialist for Organizational Affairs: Magdalena Nalepa; IT Tools Specialists:
Grzegorz Kolegowicz, Aleksandra Zadora, Maciej Zadora.
Coordination team (PARP): Anna Nikowska, Joanna Bazylak, Monika Karkowska.
Piotr Adam Boguszewski,
Jerzy Hausner

13

Part I
Bankruptcy theconcept,
its measurement, causes and effects

Piotr Adam Boguszewski, Marta Kdziorska

Chapter 1
THE CONCEPT OF BANKRUPTCY SELECTED ECONOMIC
AND LEGAL ASPECTS
1.1. Introduction
It may appear that formal and legal issues related tobankruptcy law do not play a leading role inthedesign ofearly warning
systems (EWS) against bankruptcies intheeconomy. This isnot thecase, however, because thesolutions adopted inthecontext
ofinsolvency procedures affect several extremely important elements related tothephenomenon inquestion.
First, itisdifficult toimagine such a system without anin-depth analysis ofthephenomenon ofbankruptcy. This raises thenumber
ofissues, but thekey one isthat we witness mostly bankruptcies inthelegal sense. Causes ofbankruptcy, however, are economic
innature. Therefore, without reflection on themutual relationships between thestrictly economic and legal components we
will be unable toachieve thenecessary knowledge fortheconstruction ofan EWS. In particular, itwill be difficult todetermine
thecausal relationships necessary formodelling, including forecasting bankruptcies. We must also remember that bankruptcy
isalways tosome extent a condition determined by convention. Thus, without a proper understanding oftherelevant existing
conventions, itisnot possible tomodel thephenomenon.
Second, warning systems should be developed forphenomena that may entail significant costs orother negative impacts on
society. Thelegal system, aswe will show, has a significant effect on thelevel ofsuch costs and theefficiency ofthebankruptcy
process. We therefore need tounderstand thesources ofthose costs. It isalso important, because itisoften argued that a wellfunctioning legal system may render any EWS unnecessary. By discussing thebankruptcy system in-depth we will show, however,
that although such a course ofaction isunquestionably required, thesystem always contains a number ofsubjective and
discretionary elements. This situation isto a large extent independent ofthesolutions adopted ina particular bankruptcy law.
For that reason, inthis section we present not only anoutline oftheprovisions ofthePolish law, but we shall do so against
thebackground ofa more general reflection on theconstruction ofbankruptcy systems themselves.
Third, all bankruptcy systems always generate a certain temporal structure oftheprocess both interms ofits duration and
individual stages. This feature ofthesystem isalso crucial from thepoint ofview ofthedesign ofan EWS. Such systems are
developed, among others, inorder toenable us topromptly anticipate risks on thebasis oftheavailable information. For a number
ofeconomic phenomena, including bankruptcy, italso makes sense tohave sufficient long-term knowledge, since activities
undertaken aspart ofeconomic policy are usually characterized by delayed effects oftheinstruments applied.
Fourth, both theanalysis ofthebankruptcy system and thePolish law are necessary inorder todevelop anappropriate
terminological apparatus.

1.2. Thebankruptcy system an outline


The problem ofunpaid financial obligations appeared concomitantly with thefirst loans and necessitated thedevelopment
ofprocedures todeal with such cases. Solutions inthis area have evolved over thecenturies. In ancient times, failure topay was
often treated asa crime, inkeeping with theRoman tradition. Such a view led tothedevelopment ofappropriate remedies
creditors were typically treated ina very restrictive manner, with dominant tools being different types ofretribution, including
capital punishment. Failure topay ones obligations was also often associated with loss offreedom, and almost always with
considerable damage tothesocial reputation ofthedebtor (cf. e.g. Levinthal 1918, Efrat 2006). Still, we must remember that certain
ancient legal and ethical systems provided fora periodic forgiveness ofdebts, which insome ways, constituted a further-reaching
solution than themodern rule oflimited liability fordebts.
A number ofsuch components ofthesystem survived inEurope until theend ofthe19th century including thepractice
ofimprisonment fordebt, and thestrong social stigmatization ofinsolvent debtors (cf. e.g. Gratzer, Stiefel et al. 2008). In modern
times, however, bankruptcy istreated toa considerable extent asan economic phenomenon inherent intheoperation ofthefree
market, competition and various other mechanisms resulting from these risks, not asan ethical flaw on thepart ofthecreditor
(although intheopinion ofvarious social groups theeffect ofthestigma associated with bankruptcy remains quite strong, cf. e.g.
Oldfield 2005).
The objective ofthis section isto briefly characterize thefundamental elements ofthebankruptcy system, by which we mean a set
ofrules and procedures essential and useful indealing with bankruptcy. This obviously isa very complex system; consequently,
17

we have tosettle forsketching out only its most important elements, focusing on those whose significance from theperspective
ofconstruction ofearly warning systems against bankruptcy appears tobe crucial. We shall discuss especially those solutions
that have a significant impact on thecharacteristics ofbankruptcy procedures, such astheir time, cost and other elements
contributing toits efficiency. This review, although itnaturally touches upon legal issues, isconducted mainly from thevantage
point ofeconomic and process analysis ofthematter athand. Their judicial implications will be covered inthenext section. Here
we shall discuss only theissues ofbankruptcy ofnon-financial companies.
The basic components ofthesystem are shown inFigure 1. Thesystem rests on three main pillars: thenormative base, thedesign
principles ofbankruptcy procedures and theprocedures ofbankruptcy proceedings.
As regards thenormative base, itspecifies thebasic constitutive elements ofany bankruptcy system. Thestarting point forits
construction isa conceptualization ofthebasic purpose ofthebankruptcy system (cf. e.g. Gutirrez, Olalla et al. 2005), since
theoccurrence ofunpaid obligations initself does not have toprovide a stimulus forthedevelopment ofdedicated tools tosolve
theproblem.

Pillars of the
bankruptcy system
The normative base

Bankruptcy procedures

Construction of the
bankruptcy procedure

Model of conflicts
of interests

Principles of dealing
with debtor assets

Entitlement to
participate in
proceedings

Causes
Principles governing
priority of claims
Protection of value
of debtors assets

Rules governing
priority of claims
(PoC)
Of debtors
Rules governing
priority of claims
Of owners and other
stakeholders

Function(s) of the system


objectives
Enterprise protection
(SURVIVAL)

Authorized entities and


relevant procedures

As a plaintiff

Asset protection procedures

Enforceability
of principles

Current management
Asset liquidation
procedures

Bankruptcy petitions
processing principles
Entity conducting
the proceedings
Criteria for petition
acceptance/dismissal
Appeal procedures
Forms of proceedings
Types of bankruptcy
Liquidation

Debtor protection

Negotiated

Other

Procedure selection
principles

The concept
of bankruptcy

Authorized entities

Formal criteria

Selection criteria

Assets and obligations


Valuation principles

Asset valuation
and disposal
Principles of dealing with
debtor obligations
Entities entitled to submit
negotiation proposals
Negotiation principles
Liability settlement funds
Priority of claims
settlements
Process supervision and
control principles
Procedures of changing
the form of bankruptcy
proceedings
Entities entitled
to file motions

Prognostic model

Procedure completion
criteria (stop rules)

Fig. 1.1. Themain components ofthebankruptcy system anoverview


Source: own study.

It isassumed that thefundamental premise behind thecreation ofsuch a system isthedesirability todevelop a policy forsolving
theproblem ofcreditor competition (race forcompanys assets), especially inthecase oflimited opportunities topursue those
claims on thedebtors assets. It isbelieved that thesocial cost ofsolving theproblem within theframework ofsuch a system
islower than itwould be if each creditor individually tried tosatisfy their claims.
There are a number ofreasons why themechanism ofexternal coordination imposed by bankruptcy ismore effective than
bilateral agreements between creditors and thedebtor concluded inthespirit oftheCoase Theorem: thenumber ofcreditors,
asymmetry ofinformation, incomplete contracts etc. (cf. e.g. Blazy, Martel et al. 2014). Such circumstances would inevitably give
rise toconflicts ofinterests amongst individual creditors. Themain task ofthebankruptcy system istherefore tosolve it(cf. e.g.
Longhofer and Peters 2000). It isobvious that fulfilling this task requires anaccurate diagnosis ofthecauses ofconflict and
a submission ofproposals oftools toavoid orto mitigate it. Therefore, both elements intrinsically linked together are called
theconflict ofinterests model.

18

As was mentioned before, thesecond component oftheconflict ofinterests model includes thetools aimed toresolve this conflict,
which we shall call theclaim settlement principles. While this isa significant simplification ofthecomplex and multidirectional
operation ofreal bankruptcy systems, here we assume that therules are made up oftwo main groups ofprocedures protection
ofthedebtors assets and priority ofcreditor claims (Priority ofClaims PoC). Both are intrinsically interdependent, which
sometimes causes errors ofomission thecreditors problem consists not only inwhether theprinciples governing theallocation
ofthedebtors assets will leave something forhim, but that theorder ofsuch allocation, and hence thepossible date ofactual
satisfaction ofhis claims aswell asother system-based procedures may have a negative impact on thereal value ofthedivided
assets. A system that does not resolve this problem will not constitute a socially acceptable way toresolve conflicts ofinterest,
since creditors occupying further places intheline ofpriority may claim tobe discriminated against not solely because oftheeffect
ofthechanging value ofmoney intime but also due tothefact that thevalue ofthedebtors assets are likely toerode, which could
conceivably be avoided oreven reduced if better systemic solutions were tobe chosen.
Despite theclose connections between both elements, intheliterature more attention appears tobe devoted totheproblem
ofPoC. It has been widely discussed, but more extensive comments would go beyond thescope ofthis paper. Let us examine
only three elements deemed fundamental inthecontext ofan EWS. First, there are several possible rules governing thePoC.
Itisworth mentioning three classic solutions: thecanonical rule (absolute priority rule APR), theequal priority rule (equal priority
rule EPR), theFamaMiller rule (cf. e.g. White 1989). Second, such rules may lead tovarious costs and influence theefficiency
ofbankruptcy systems (cf. e.g. White 1980). Specifically, ithas long been known that exceptions tothecanonical rule (APR), which,
inthepractice oftheoperation ofreal bankruptcy systems are not uncommon, are accompanied by fluctuations inthestock
market valuations ofrisks associated with assets (cf. e.g. Warner 1997). Third, intheliterature there are attempts todesign optimal
systems (cf. e.g. Schwartz 2005), but they still tend tobe theoretical constructs difficult totranslate into real operating regulations.
Thus, since theoperation ofbankruptcy systems may be subject tonumerous disturbances and deviations from theoptimum
resulting, forexample, from disruptions intheproper application ofPoC rules, anappropriately designed EWS may decrease some
ofthesocial costs associated with these imperfections.
The last remark isparticularly important against thebackground ofanother element ofthesystem ofnormative base ofbankruptcy
thefunction ofthesystems objectives. It isworth noting that thevery model ofconflict does not have toconstitute thebasis
forthefinal solutions adopted inthepractice ofoperation ofa given bankruptcy system there can be a number ofthem. Usually,
however, thelegislature imposes on theentire system some major conditions (cf. e.g. M. Hummelen 2013, White 2011). Without
going into too much detail ofthis complex subject matter, itcan be assumed that there are three such main objectives. They are
orshould be present invirtually all systems operating inpractice, although their priorities may vary considerably (c.f. Korobkin
1992). These objectives are: toprotect thecreditors, toprevent unjustified liquidation ofenterprises and other preventive measures
aimed ate.g. precluding owners from undertaking activities that may lead tofinancial difficulties and impending bankruptcy, but
such that can be avoided by implementing remedial measures (cf. e.g. Haviv-Segal 2005). When discussing theissue we shall limit
ourselves toobserving that anEWS can fulfil animportant role ineach ofthese three areas. For obvious reasons itisintheinterests
ofcreditors tohave advance knowledge about thepotential risk ofbankruptcy, especially if theadvance warning issufficiently
long toafford anopportunity toprotect thevalue ofthedebtors assets prior totheir unnecessary anunwanted depreciation.
Such a system not only improves theefficiency ofliquidation procedures, but may also warrant a timely restructuring intervention
by theenterprise itself and, asa result, may help ittoavoid bankruptcy. An EWS supports thepreventative role ofthebankruptcy
system by including pre-emptive identification ofareas ofhigher investment risk. An EWS may also accelerate information flow and
improve data transparency within theenterprise itself, which undoubtedly aids thebankruptcy system inits precautionary capacity.
The last element ofthenormative base ofthesystem istheconcept ofbankruptcy itself. Since this isalso one ofthemost important
categories from thevantage point ofan EWS system design, we shall devote a little more attention toit (a more in-depth treatment
ofthesubject isoffered by e.g. Armour 2001).
Let us start by saying that, contrary tothecommon misconception, bankruptcy is, toa considerable extent, a conventional and
ambiguous notion. These features come tolight upon a more detailed scrutiny offormal bankruptcy criteria, which combine
toform anessential component ofany bankruptcy system.
Bankruptcy isa systemic (legal) consequence ofinsolvency. Thus, itisbased on theconcept ofbankruptcy and ways toderive legal
action from it. There are two main types ofinsolvency: balance-based (i.e. balance sheet insolvency) and loss ofliquidity (i.e. cash
flow insolvency). Thefirst one can be formally represented as:

) (

E Dt +H | It1 > E Vt +H | It2

(1.1)

where: Dt+H represents liabilities due attime t+H; Vt+h thevalue ofassets attime t+H; It1 and It2 sets ofinformation available
respectively tothecreditors and owners ofthecompany; and E () istheexpectation operator. Thecondition means that
theexpected value ofliabilities ofan enterprise ishigher than thevalue ofits assets.
19

With respect toformula (1.1) itisworth focusing on several issues. First, itissomething ofa prediction ora forecast. Open issues
include horizon H and theway theforecast itself isconstructed. Bankruptcy law must define both components. For reasons
discussed among others insubsection 1.4, thedebtor may have anincentive tomove forward themoment ofverification ofthis
condition. This isoften due tothestrong conviction on thepart oftheowners ofa given firm that inthelonger run ittheir
enterprise will financially recover and its current problems will disappear. On theother hand, they are aware that intheevent
ofbankruptcy, thepresent owners ofthefirm may be deprived ofall existing benefits. A very common argument raised by
thedebtors incourt oron other forums isthat bankruptcy was declared too early, that this was thefinal blow that finished
off a good company etc. Thus, thedebtor has a rather natural tendency touse H values which are significantly greater than
zero. Conversely, looking atH through theeyes ofthecreditors, who, even inthecase ofa good bankruptcy law, may be even
subconsciously motivated by thesyndrome ofracing forassets, theparameter isbound tobe fairly small. When talking about
thepotential conflict ofinterests between creditors and debtors associated with thechoice ofthelength ofhorizon H, itisalso
worth noting that thehigher thevalue ofthis parameter, thestronger thedebtors position may be. A longer horizon H may,
infact, increase thevalue ofa firm inquestion due toe.g. thefuture effect ofcompleted investment1 (cf. e.g. Tong and Reuer 2007),
whereby thedebtor may appeal tonew advantageous business circumstances etc.
The parameter H also affects theleft side ofinequality (1.1). Here arises, among others, theproblem with thefuture valuation
ofliabilities. Quite often, a firms liabilities may, infact, be conditional. A number ofvery well-run Polish enterprises experienced
theso-called futures crisis (cf. e.g. Konopczak, Mielus et al. 2011). Thefinancial situation ofa large group ofsuch entities
unexpectedly deteriorated toa significant extent when theappreciation trend ofthezloty was suddenly interrupted due
tothepresence intheir portfolios offutures contracts optimized along thepath ofthestrengthening domestic currency.
A practical solution totheabove problems associated with thechoice ofH often leads totheadoption ofH = 0. In such
a situation, itissufficient that thecondition issatisfied now (which, given theoften time-consuming valuation procedures,
means, strictly speaking, very recent past). Whether and which party benefits from thesolution depends, among other things, on
therespective rights and obligations ofcreditors and debtors toverify this condition and towhat extent they can ormust take
action if itissatisfied. Too stringent solutions might be disadvantageous tothecreditors, forexample, when there are reasonable
grounds forthem toexpect that inthenear future inequality (1.1) changes direction. Unfounded restrictions might also be used
by thedebtors, forwhom too automatic fulfilment ofsuch a condition would facilitate their escape from liabilities through thesocalled planned bankruptcy. All inall, a way toverify this condition, including theselection ofhorizon H isa sensitive issue, which
should be carefully regulated, among others; within theframework ofrelevant bankruptcy procedures (see Fig. 1). We will shortly
return tothis issue.
The other fundamental problem associated with condition (1.1) isthat inthemost general case theinequality isbased on
conditional expectations and draws on two sets ofinformation: available tothecreditors (left side) and totheowners/debtors
(right side). Due tothepossibility ofinformation asymmetry (which ismore often than not thecase under such circumstances)
they do not have tobe identical sets. Bankruptcy law must decide on which set(s) ofinformation thecondition isto be verified and
how. Depending on thepurpose ofthebankruptcy system, itcan be based toa greater extent on theset ofinformation available
tothecreditors (if this isthemain entity protected by thebankruptcy system) orto theowners (which may be thecase otherwise).
In practice, itisimpossible toverify condition (1.1) without theadoption ofcertain assumptions about thevaluation ofliabilities
and assets on both sides oftheinequality (for H = 0) orforecasting their values (for H > 0). In thefirst case, their book value
isoften adopted. Such a solution issimple, but itmay lead toundue liquidation ofenterprises generating a positive added value,
especially those with a large investment portfolio (see footnote 1). This basis forvaluation may also be detrimental tothecreditors,
forexample, inthecase oftheso-called rare assets (with very specialized applications), theliquidation value may even be negative
(disposal costs). At theother end ofthespectrum, there are purely market-based concepts, inwhich, inmany cases, anauctionbased valuation ofassets isrecommended (cf. e.g. Thornburn 2000). A distinct group ofproblems isencountered inthecase
ofnegotiated bankruptcy caused by theestimation ofthevalue oftheso-called reorganized firm (restructuring value) (cf. e.g.
Bebchuk 1988). Due toinsufficient space we shall not deal with thebroader issues ofvaluation. Suffice ittonote, however, that
any bankruptcy system, including thelegal one, must adopt specific rules ofconduct whose task isnot so much totheoretically
estimate those values astopass judgement based on available information concerning thefirm inquestion.
An analysis ofthenature ofinformation sets I1,2 has another important consequence. It helps todraw a distinction between
financial and economic bankruptcy. Although intheliterature there isconsiderable conceptual confusion, we say that a company
isbankrupt inthefinancial sense if condition (1.1) issatisfied given thepresent set ofI, but itwould not be thecase if thecompany
pursued a different financing strategy. By financial bankruptcy, therefore, we understand a situation that occurs only because
ofa specific financing structure. Although within theM-M paradigm theenterprise value does not depend on such a structure,
1

The strength ofthis effect may significantly depend on themethod ofmeasuring thevalue ofV, e.g. using theaccounting, market orDCF
(discounted cash flow) methods.

20

so if (1.1) holds fora certain structure, itholds forany other, but thelimitations ofthis theory have been known fora long time.
By economic bankruptcy we understand economic circumstances under which inequality (1.1) holds regardless ofthefinancing
structure. It can therefore be perceived asa situation inwhich anenterprise does not have a permanent capacity togenerate
a surplus sufficient tocover thedebt that was incurred tofund its activity (absence ofeconomic profitability).
The other type ofbankruptcy isinsolvency due totheloss ofliquidity. We can formally define it(adopting a simplified situation
ofa single creditor) asthecircumstances under which:

CFt < PDt PDt > 0

t <T H ,T>

(1.2)

This relationship describes a situation inwhich financial flows generated by firm CF do not cover its unpaid financial obligations due
by time t PDt. It isgenerally acknowledged that such a situation must continue fora period oftime H, since points ofinsolvency
may often occur inalmost any enterprise. Models ofbankruptcy systems usually outline such a horizon orother conditions that
confirm thepermanence and importance ofthesituation.
Definition (1.2) issometimes modified. Let us note fact that itcontains animplicit assumption oftheobjective economic nature
ofliquidity problems, thecause being insufficient cash flow. In practice, however, debtors sometimes deliberately avoid repaying
their obligations oreven play games with their creditors, forexample, by paying back some ofthem while disregarding others, orby
attempting tobankrupt their own company. For these reasons, this condition isoften formulated insuch a way that thesufficient
condition fortheoccurrence ofbankruptcy istheexistence fora certain time ofoutstanding obligations due by time t:

PDt > K KRY 0

t <T H ,T>

(1.3)

The basis forinsolvency adopted here isconstituted only by theexistence ofdue and unpaid obligations. In general, condition 1.3)
issupplemented by anadditional requirement concerning thescale ofthese obligations they should not be less than a certain
critical value KKRY. Theinequality itself may also refer toabsolute and normalized values (e.g. thesize ofcompany assets). When using
thecritical value KKRY itishard toimagine a threshold not defined asprecisely this kind ofrelationship.
A number ofobservations formulated inthecase ofbalance-sheet insolvency remain valid forrelationships (1.2) and (1.3).
Nonetheless, there also exist important differences. Thebasic one isthat this criterion isclearly anex-post one innature, while
theprevious one may have anex-ante version. It should be remarked that this criterion, especially insystems focused mainly on
theprotection ofcreditors, may play anadditional role. Now, formula (1.1) seems tobe quite difficult toverify quickly. Theprocess
ofsuch verification, if only due tothepresence ofexpected value operators, may be complicated and isalways arbitrary, atleast
tosome extent. Since theverification ofthis criterion requires theestablishment ofa number ofitems contained inthebalance
sheets, itmay be quite time-consuming toapply inlarger companies. In contrast, conditions (1.2) or(1.3) can be tested and applied
fairly easily.2 The loss ofliquidity criterion can thus be considered asa sort ofshortcut that a creditor may use toquickly initiate
theprocedure and avoid thedifficulties associated with theverification ofcondition (1.1).
An interesting and pragmatically important problem isthequestion ofthemutual relationships that hold among conditions
(1.1), (1.2) and (1.3). Let us start by saying that inbankruptcy law they often occur together, i.e. thefulfilment ofany one ofthem
istreated asa situation that justifies thedeclaration ofan entity asbankrupt. One ofthereasons behind using such a solution
was discussed above. Theother reason may be that condition (1.2), even fora fairly long H, may not be met by a firm ina poor
economic condition, thevalue ofwhose assets isless than that ofits debts. Maintaining liquidity isactually possible even under
such circumstances aslong asthefirm has enough liquid assets and/or maturity periods ofsome ofits obligations are more remote.
So if bankruptcy law was limited toconditions (1.2) or(1.3), itwould not be sufficiently selective.3
A question arises whether a firm may lose liquidity and yet not be considered bankrupt asprovided by condition (1.1). Such
a situation isquite common on a temporary basis. Naturally, inthelong run theaccumulation ofunpaid debts will eventually meet
condition (1.2), but even inthis case itisnot inevitable. Thecompany may, infact, have assets with a prospective value (ongoing
investments, natural resources etc.).
As we can see, conditions (1.2) and (1.1) are not automatically interdependent, hence thebankruptcy procedure must provide
fora way ofdealing with situations when only one ofthem ismet. This issue isanalysed inthenext section.

It should not be construed tomean that itoccurs without difficulties. Thecontentious issue may involve determining outstanding liabilities.
For example, thecompany inquestion may argue that itwill not pay forcertain services because their quality does not meet thepreviously
agreed standards.
More information on thesubject can be found inChapter 4.

21

Virtually all oftheabove-mentioned parts ofthebankruptcy model are basic and necessary structural components ofthedecision
ofcreditor I asregards theinitiation ofthebankruptcy procedure and thechoice ofits forms and strategies s. The rationality ofsuch
a decision must meet thefollowing condition:

uS ( i )\ {s}

E (w i , s, Vt +H ( s ) | It ) {E (Dt +H (i ,u) | It ) + C (i,s,H)}

(1.4)

where function () characterizes rule PoC within strategy s; V (s) represents thecompany value generated inthecase ofselecting
thestrategy, whereas function C (i,s,H) characterizes thecosts ofparticipation inbankruptcy proceedings forentity i conducted
intheform ofs. D (i,u) represent thevalue ofliabilities oftheentity, which can be recovered by selecting strategy u.
Condition (1.3) means that thecreditor chooses bankruptcy proceedings intheform ofs if he expects that theresulting priority
ofclaims and company value V and theanticipated net share (after deduction ofcosts C) inthis value exceeds theexpected
value ofclaims that could be obtained by pursuing any other strategy s regarding u available tohim4.
Note that fora number ofcreditor firms theexpected recovery value ofclaims atrisk may be zero if they do not undertake
bankruptcy proceedings. In this case, thesufficient condition forparticipation intheprocedure isthat theleft side oftheinequality
isgreater than its costs. On theother hand, itisoften thecase that these costs, if understood sufficiently broadly, greatly exceed
theleft side ofinequality. A number ofhypothetical situations can be quoted, but one ofthemost common isthat thefirm
that chooses toinitiate bankruptcy proceedings may experience negative reactions on thepart ofthemarket, forexample, by
losing certain customers, become ostracized, ortheentity inquestion may be themain recipient ofthecompany products. Such
complications affect especially smaller enterprises.5 Consequently, a very common way ofsolving theproblem ofnon-timely
payments involves more orless informal activities conducted outside bankruptcy law. Naturally, there may exist many more reasons
for(not) meeting condition (1.4), but their discussion isbeyond thescope ofthis introduction.6
As regards condition (1.4) and its cost function C, we would like tomention yet another aspect, namely thequestion how
itthemodel should operate if condition (1.4) ismet only fora certain group ofcreditors. This isa very complex problem, which
cannot be further addressed here. Let us, however, point totheconsequence offulfilment ofthis inequality foreach ofthecreditors.
Adding up thesides we obtain:

E (Vt +H ( s ) | It ) C (s,H)

(1.5)

where C (s, H) istheaggregated function ofbankruptcy costs. It isa prerequisite forsocial rationality (given such assumptions)
ininitiating thebankruptcy procedure. If we remember that bankruptcy costs include7 both direct and indirect costs, and if
condition (1.5) ismet, theexpected value ofa firms assets which can be obtained under thebankruptcy procedure s must also be
greater than thedirect bankruptcy costs. If not, initiating thebankruptcy procedure isnot socially justified. Since thedirect costs
ofbankruptcy are the most tangible and easily measurable, theprinciple that theleft side ofinequality (1.5) should be greater
than those costs, itconstitutes a natural candidate fora rule ofthumb governing social rationality ofbankruptcy proceedings.
Condition (1.4) also has another extremely serious consequence. Its verification clearly requires which can already be seen
tosome extent even inthelight oftheprevious considerations theforecasting ofcategories occurring intheformula aslong
aswe want toconsistently follow theprinciple that thebankruptcy system should have a favourable effect on thelevel ofsocial
prosperity, i.e. eliminate only those entities which are economically inefficient. In particular, these procedures must establish
whether a sufficiently long horizon H and/or a different strategy may offer greater chances ofdebt recovery than bankruptcy.
A similar analysis should be carried out once bankruptcy has been considered tobe theoptimal way ofdealing with thedebtor;
then theproblem becomes one ofchoosing its form. If negotiations are thepreferred option (i.e. theenterprise may live), such
a choice must be based on its future business opportunities.
A wide range ofpotential solutions can be discussed. They may include individual methods and procedural solutions, including
issues related totheregulation ofthepowers ofindividual entities participating intheprocess toformulate such forecasts and
take action on their basis. Consequently, inthecontext ofthenormative base ofa bankruptcy system, we are talking about
a forecasting model. Even though analysing bankruptcy usually does not include expressions related directly tosuch a term,
themodel ispresent there atleast implicitly. If, forexample, thevalue ofa firms liabilities exceeds thepresent value ofits assets

4
5

6
7

Derived from set S(i).


This case often represents a forced business credit. Thesupplier becomes reconciled with thefact that therecipient does not pay on time.
Although thesupplier could formally apply fora declaration ofbankruptcy, equation (1.3) stipulates that such anaction may not be beneficial
tothecreditor.
More information on thesubject can be found, among others, in: (Gilson 1990).
Cf. Chapter 4.

22

and a decision toliquidate itismade, itisdifficult toimagine such anaction asrational, if, atthesame time, we believe that such
a relationship ofdebts and assets islikely toreverse intheforeseeable future.
Now we shall discuss theother two pillars ofFig. 1 thedesign ofthebankruptcy procedure and insolvency procedures. We shall
confine ourselves toseveral crucial observations fortwo reasons. First, they constitute themore practical elements ofthewhole
system and, assuch, will be characterized and discussed in-depth (based on thePolish bankruptcy law) inthenext section. Second,
several elements ofthese procedures are closely related tothepreviously discussed model ofa bankruptcy system.
The design ofany bankruptcy procedure includes mainly theverification ofinsolvency conditions (1.1) (1.3). Thekey issues
include thedetermination ofwho has theright and obligation toverify these conditions and what may constitute thepossible
sanctions intheevent that theobligation has not been fulfilled. Thecreditors rights are obvious inthis respect. We should,
however, remember theasymmetry ofinformation that has anadverse effect on theinterests ofthefirm, and quite often also on
public interest. It isnatural, therefore, toimpose anobligation on debtors totake appropriate action. Unfortunately, aswas already
mentioned, theverification procedure ofthese conditions isgenerally not assimple asit may appear atfirst glance. Additionally,
debtors often have a vested interest inmanipulating these assessments. A very important component ofa bankruptcy system
isthecollection ofinstruments and procedures todiscipline debtors inthearea oftimely provision ofobjective information on
thecompanys problems. Theless restrictive and accurate theadopted solutions, thelower theefficiency ofthesystem asa whole.
The second group ofcomponents ofthebankruptcy procedure includes systemic solutions that define theentities entitled
toparticipate intheproceedings, thecriteria foracceptance/dismissal ofthebankruptcy petition and appeal procedures.
Theacceptance/rejection criteria involve both specific rules fortheverification ofconditions (1.1) (1.5) and thealready mentioned
problem ofmutual relationships amongst criteria (1.1) (1.3), including their equal status: do they have tobe met concurrently
orisit sufficient if only one ofthem ismet? Of particular importance may be theanalysis ofcondition (1.5) since its result provides
anassessment ofsocial efficiency ofthebankruptcy process. Theprocedure should therefore expressly include solutions tobe
implemented intheevent that such anefficiency evaluation isnegative. Thenatural direction oftheproceedings isto dismiss
thebankruptcy petition, but other options may be considered aswell.
Since theexamination ofconditions (1.1) (1.5) isa complex process that may include a number ofsubjective, prognostic and
other factors, thebankruptcy system must also comprise appeal procedures on thepart oftheentities involved.
The third family ofelements ofthebankruptcy procedure comprises solutions inthearea offorms ofbankruptcy and principles
governing their choice. In todays bankruptcy systems, inprinciple, there are two basic forms ofbankruptcy liquidation and
negotiated settlement. A much more complex problem isposed by theoperationalization ofchoice between these two
possibilities. Three issues are especially open todebate: theentities authorized tomake such a choice (the court oflaw, thecreditors,
thedebtor), rules ofconduct when thepreferences oftheauthorized entities are atvariance and thereasons formaking such
a choice. Thelast one may pose a substantial challenge since, asour previous remarks have shown, theprocedure forverifying
conditions (1.1) (1.3) can be very complicated, require forecasting, expert knowledge etc.
Finally, thelast element oftheprocedure involves therules ofprocessing thebankruptcy petition, i.e. actions taken once bankruptcy
has been declared. This group ofsolutions isofparticular importance from thepoint ofview ofefficiency ofthebankruptcy
process, including its cost and duration. Individual elements contributing tothese rules are synthetically presented inFig. 1. Here
we shall confine ourselves totherole ofactivities related totheprotection ofthedebtors assets (in thecase ofliquidation) and
theorganization and supervision oftherestructuring process (if negotiated bankruptcy isthepreferred option). Specific solutions
inthis area are extremely important tothequality oftheentire bankruptcy system since some ofthekey issues which have tobe
dealt with by thesystem atthis stage include theasymmetry ofinformation and theproblem ofagency. We should be aware that
itisvery difficult todevelop optimal solutions inthis area. It suffices tonote that individual participants intheprocedure have
different knowledge and do not always represent shared interests. If theentire process issupervised by thecourt oflaw, one may
doubt whether itpossesses sufficient knowledge ofthespecific nature oftheenterprise tobe restructured. In turn, thecreditors
frequently have no similar general knowledge, and almost asa rule have no access toinside information related tothespecific
circumstances, history, contracts, customer base etc. ofa particular firm. This group may also exhibit a tendency torecover their
debts asquickly aspossible, which isoften thereason fornon-optimal asset management. By and large, this isnot a group with
homogeneous interests, but this feature may paradoxically constitute thepredisposing factor. It ishard toimagine aneffective
and reasonable bankruptcy procedure inisolation from theinterests ofthis group, internal negotiations amongst its members
and ways toreconcile such interests. Finally, theprevious owners usually have themost complete relevant knowledge about
their firm and its economic environment. Unfortunately, quite often this isa group whose mismanagement was themain cause
ofbankruptcy. Under such conditions itisdifficult toexpect ittoimplement thedesired changes. As we can see, there isusually
no ideal candidate tobe entrusted with theresponsibility oforganizing and supervising thebankruptcy proceedings.
We have clearly explained thecomplexity ofthebankruptcy process and its dependence on theresults ofits adopted form,
theforecasts, managing theprocess itself, etc., consequently, itisnecessary tosupervise theprocess not only informal and
23

legal terms, but also ineconomic ones. Procedures atwork must provide forthepossibility ofcorrecting errors, among others,
committed due tothecomplexity ofthesubject matter, including changing theform ofbankruptcy.
Our reflections on thepillars ofthebankruptcy system shall terminate with several conclusions concerning terminology. Thenatural
consequences oftheadopted model ofa bankruptcy system rest intheir implied conceptual and terminological network. It
must be stressed that this isan area where neither sufficient accuracy nor unification ofterminology has been achieved. For
example, one ofthefundamental concepts isthenotion ofbankruptcy. It isquite commonly accepted that theterm derives
from thecombination oftwo Latin words bancus and ruptus and originally referred toa situation whereby a person conducting
banking activity was unable tomeet its financial obligations orto continue lending activity. This was symbolized by thebreaking
ofa bench on which such activity was conducted. Currently, however, theterm isused ina wide range ofcontexts and inpractice
means not only bankruptcy inits legal sense, but also refers toa number ofproblems associated with thediscontinuation ofdebt
service, poor financial standing and others. Likewise, we use theterm quite freely throughout this study whenever itdoes not
cause interpretation problems.
Simplifying theissue a little we can say that theRapid Response Instrument isa system that warns against impending bankruptcies,
hence itisquite important how thelatter term isunderstood. We tend toassume that bankruptcy inits broad sense refers
toa situation that meets any oftheconditions (1.1) (1.3). In contrast, bankruptcy inits strict sense describes thestate resulting
from a decision made by theCommercial Court. Our previous considerations concerning themodel ofa bankruptcy system show
clearly that these situations do not have tooverlap. In this study, however, we often use thegeneral term bankruptcy inboth
contexts with clarifications only where necessary. In such cases we also use, with anidentical reservation, theterm insolvency.
The above convention also implies that thecategory ofbankruptcy inits broad sense may overlap with theterm insolvency, which
means that any oftheconditions (1.1) (1.3) ismet. It isworth noting that theformer term isprimarily used insituations where
conditions (1.1) (1.3) have a deeper justification forthebusiness circumstances ofan economic entity. In theEWS discussed here,
ananalysis ofthis situation usually completes thefinal diagnosis.
The Polish terminology lacks a really good translation ofthemuch-needed term financial distress. We interpret this term quite
broadly (cf. e.g. Opler and Titman 1994) asan event orsituation which significantly increases theprobability ofa firms insolvency
(i.e. one that meets atleast one oftheconditions (1.1) (1.3)). It may also be interpreted asa period offinancial distress, a negative
financial shock, economic problems orimpending bankruptcy. Moreover, intheliterature on thesubject theterm financial distress
issometimes identified with theconcept ofbankruptcy (cf. e.g. Casey and Bartczak 1985).
Situations (1.2) or(1.3) are thus called insufficient liquidity ordefault and thesubject thus affected isdeemed tobe illiquid.

1.3. Insolvency under thePolish legal system8 basic components


Insolvency under thePolish legal system isdefined asa property standing ofan enterprisewhich, foreconomic reasons, isunable
tomeet its obligations tocreditors asruled by a competent court following insolvency proceedings. During theproceedings
thecourt analyses theenterprises financial standing inorder todetermine whether itisappropriate todeclare insolvency ina given
case. Thedeclaration ofinsolvency by a court oflaw starts often lengthy insolvency proceedings. In practice, theterm insolvency
isoften used interchangeably with theterm bankruptcy. Such anapproach isnot always correct from a formal point ofview since
ina strict sense bankruptcy isprimarily a concept ineconomics, while insolvency isa legal term (Britannica 2014).
As regards thefunctions and objectives ofinsolvency law, they are inline with thegeneral rules described inthesubsection 1.2.
In principle, theinsolvency law regulates theproceedings against firms which have become insolvent orwhose financial situation
may bring about thedanger ofinsolvency inthenear future.9 The primary function ofinsolvency law isto satisfy thecreditors
ofan insolvent debtor. This should occur tothelargest possible extent bearing inmind, if possible, theprospects ofrestoring
theoperation ofthebusiness ofan insolvent debtor orone already indanger ofinsolvency. For this purpose thelegal regulations
provide fortwo solutions: liquidation orarrangement insolvency. Thechoice between them depends on theindividual enterprises
capacity tosatisfy thecreditors using a given procedure. In other words, theprimary goal isto protect theinterests ofcreditors
while thesecondary one isto sustain thecompanys activities.
The law also prevents a enterprises insolvency from getting worse if ithas already found itself infinancial distress. An enterprise
operates inan economic environment i.e. themarket and its insolvency more often than not has a significant impact on other
entities with which ithas formed economic and legal ties. For that reason, itisimportant forthedebtor threatened with insolvency
tofile aninsolvency request ina timely and effective manner and tocooperate with theauthorities involved intheproceedings.
8

Regulations relating toinsolvency and restructuring law are described intheAct ofFebruary 28, 2003 Insolvency and Restructuring Law
(Journal ofLaws ofApril 9, 2003, asamended; hereinafter: Insolvency and Restructuring Law), which starting from Octover 1, 2003 replaced
thePolish Presidents decree from 1934 Insolvency Law and Law ofarrangement proceedings
Article 1 oftheInsolvency and Restructuring Law.

24

Additionally, theInsolvency and Restructuring Law aims topromote fair trade attitudes among thecompany executives. It
isa way offorcing them toact ina manner that does not expose thecompany totherisk ofinsolvency and if ithas occurred,
tofile aninsolvency requestas soon aspossible and tocooperate with theauthorities conducting theproceedings. It istherefore
necessary toinstitute sanctions against any instances ofdishonest behaviour on thepart ofthefirms executives (Zedler 2004, p. 31).
In summary, theprimary goal ofinsolvency law isto protect theinterests ofcreditors and theeconomy atlarge by eliminating
dishonest orincompetent entrepreneurs.
The Insolvencyand Restructuring Law isbased on two main principles: theprinciple ofoptimization and theprinciple ofdominance
ofthecollective interest ofcreditors. In accordance with theprinciple ofoptimization, theabove-mentioned Law provides that
inorder tosatisfy creditors tothegreatest extent possible, thebest conditions fortheuse oftheassets ofinsolvent companies
orthose facing insolvency risk must be ensured. As long asobjective conditions permit, any activities carried out intheprocess
should be conducted ina way that permits thecompany tocontinue doing business.
Insolvency proceedings require a compromise amongst theindividual, often conflicting interests ofcreditors oftheenterprise
inquestion. Accordingly, theprinciple ofdominance ofcollective interest ofcreditors focuses on optimal and inprinciple
proportional settlement ofcreditors claims (at least within a single category ofcreditors).
An important element ofinsolvency law isthecategory ofcapacity forinsolvency. Subjective capacity forinsolvency ofan entity
isdefined asa priviledge toact ininsolvency orrestructuring proceedings and tobe declared insolvent (Cyman 2008, p. 15). Under
Polish law, such a capacity isgranted toentrepreneurs, physical persons not involved inbusiness activities whose insolvency was
due toexceptional circumstances independent oftheir actions (i.e. consumer insolvency) aswell aslimited liability companies,
joint stock companies not involved inbusiness activity, shareholders ofcommercial partnerships accountable fortheliabilities
ofthepartnership without limitation on their assets and members ofpartnerships10.
An insolvent debtor issaid tohave anobjective capacity forbankruptcy. Thedeciding factors here are theboth economic facts
(failure tomeet ones financial obligations and/or a situation inwhich thevalue ofliabilities exceeds thevalue ofproperty) and
legal ones (the existence ofliabilities ofa certain value that have been unmet indue time thecourt may dismiss theinsolvency
requestif thedelays inpayment do not exceed three months aslong asthetotal outstanding liabilities do not exceed 10%
ofthebalance-sheet value ofthedebtors business). However, even if such legal facts do not occur, thecourt should declare
insolvency if non-payment ispermanent innature orif thedismissal ofan insolvency requestmay lead tothecreditors detriment11.
The Insolvency and Restructuring Law also lists entities that do not have thecapacity forinsolvency. These include theState
Treasury, local government units, independent public health care institutions and legal entities established by law, unless thelaw
provides otherwise, and those established toperform obligations imposed by legal provisions, physical persons running farms and
universities12. Theentities mentioned can neither be declared insolvent nor participate ininsolvency orrestructuring proceedings.
By excluding theabove-mentioned entities from thecapacity forinsolvency thelegislators grant legal protection from creditors
assuming therisk ofdefault incase ofbusiness failure. This isdue tothedistinctive nature oftheir activities related totheexercise
ofstate responsibilities orthepursuit ofpublic benefit activities. However, thePolish insolvency law does not define clearly
thegroup ofentities that do not have thecapacity forinsolvency. This applies especially totheprovisions dealing with legal entities
established by law ortheperformance ofobligations imposed by law. Thedifficulty inestablishing which entities are accorded such
a privilege isprimarily due tothefact that there isno registry orauthority competent toconfirm it. Moreover, this kind ofprotection
isoften enjoyed by entities pursuing profit-oriented activities, which may raise doubts astothecompatibility ofthis provision with
theprinciples offair competition (Sciska 2006, p. 27-33).
One ofthekey groups ofprovisions contained intheInsolvency and Restructuring Law are theprerequisites forinsolvency.
Insolvency isdeclared by thecourt oflaw when all theformal and substantive law conditions occur together. Under Polish
insolvency and restructuring regulations, thesubstantive law prerequisites include theentitlement toinitiate insolvency
proceedings13 and basis forinsolvency declaration pursuant toArticle 10 oftheabove-mentioned Law. Thus, only debtors who
10
11
12
13

Articles 1 and 5 oftheInsolvency and Restructuring Law, op. cit.


Article12, items 1 and 2, ibid.
Article 6, ibid.
In accordance with Article 7 oftheInsolvency and Restructuring Law, intheevent ofthedeath ofan entrepreneur his company may be
declared insolvent if theinsolvency request was filed within one year ofthedate oftheevent. Theinsolvency request may be filed by a creditor,
theheir, thespouse and each ofthechildren orparents ofthedeceased even if they do not inherit anything. However, according toArticle 20
items 1 and 2 oftheInsolvency and Restructuring Law, theinsolvency request may be filed by a debtor orany ofhis creditors. Therequest may
also be filed by:
1) with respect toa general partnership, limited partnership, limited partnership and limited joint stock partnership each partner having
unlimited liability fortheobligations ofthecompany;
2) with respect tolegal persons and organizational units without legal personality, recognized ashaving legal capacity by other legal
provisions anyone who has theright torepresent them individually orjointly with other people;
3) with respect toa state-owned enterprise thegroup also includes its founding body;

25

have become insolvent can be declared bankrupt. Any debtor who does not meet his outstanding liabilities isdeemed tobe
insolvent. At thesame time, itisirrelevant whether ornot he fails tomeet all oronly some ofhis financial obligations. If thedebtor
isa legal person oran organizational unit without legal status, whose legal capacity can be established by separate legal provisions,
thedecision astoits insolvency ismade on thebasis whether its liabilities exceed its assets, even if itpays its current financial
liabilities14.
In consequence, thelegal perspective distinguishes between two approaches tothedeclaration ofinsolvency: theperception
ofinsolvency interms ofproperty and balance-sheet insolvency. Insolvency interms ofproperty isdue totheinability ofthedebtor
topay his debts, which have become due intheordinary course ofbusiness. Balance-sheet insolvency refers toa situation inwhich
thedebtors total liabilities exceeds thevalue ofhis assets (Fletcher 2002, p. 429).
Next, inorder tomeet theprocedural conditions, thedebtor must have thecapacity forinsolvency,15 be subject toPolish
jurisdiction,16 thecourt must be material competent (District Court Commercial Court ortheso-called Insolvency Court) and
territorial competent (i.e. within whose jurisdiction themain businesses ofthedebtor isconducted or, if thedebtor has no company
inPoland, a court corresponding totheresidence ofthedebtor or, intheabsence thereof, thecourt within whose jurisdiction
thedebtors assets are located)17 toinstitute appropriate proceedings and certain formal requirements oftheinsolvency request18
must be fulfilled.
The fundamental basis forthedeclaration ofinsolvency isconstituted by permanent failure torepay debts. Insolvency cannot be
declared solely on thebasis ofa debtors inability tosatisfy creditors aslong asdebts are paid even if was proved beyond reasonable
doubt that inthenear future thesaid debtor will be unable tosatisfy all his creditors. Thecessation ofrepayments occurs when
a debtor has no funds athis disposal and fails torepay most ofthedebts due. Such a situation may also occur if thedebtor does
not repay a single creditor who possesses a significant liability (Ruling 2004).
As a rule, once a debtor has ceased all business activity, itisimpossible toinitiate theproceedings leading tothedeclaration
ofinsolvency with theexception ofa natural person anentrepreneur intheevent ofhis death orcessation ofbusiness activity.
This possibility isopen fortheperiod ofone year from thedate ofdeath orremoval ofa business entity from therelevant register.
This provision also applies topersons who have ceased tobe shareholders ofcommercial partnerships.19
In terms ofboth legal and economic consequences, a very important role isplayed by theso-called entitlement toparticipate
inbankruptcy proceedings, i.e. tofile aninsolvency request. Theinsolvency request can be filed by thedebtor and each ofhis
creditors. Nevertheless itshould be borne inmind that, while thefiling oftheabove-mentioned request constitutes theright
ofthecreditor, thedebtor isobliged todo so within two weeks from thedate on which thecause forinsolvency accrued.
If thedebtor isnot a natural person, theinsolvency request with respect tocommercial partnerships can be filed by each partner
accountable without limitation forthecompanys obligations; with respect tolegal persons and organizational units without
legal status with a legal capacity granted by separate provisions anyone who has theright torepresent them; with respect
toa state-owned enterprise thegroup also includes its founding body; with respect tothesole shareholder company oftheState
Treasury thegroup also includes theminister responsible fortheState Treasury; with respect tolegal persons and commercial
partnerships undergoing liquidation each oftheliquidators; with respect toa legal person registered intheNational Court
Register a representative appointed pursuant totheNational Court Register Act; with respect tothedebtor who was granted
aid inexcess ofeuro 100,000 euros theauthority which provided thegrant (Jakubecki, Zedler 2006, p. 65-79).
A significant component ofPolish legislation involves enterprises too poor tobe declared insolvent. The court may dismiss
theinsolvency request despite thedebtors insolvency if thedelay inthesettlement ofliabilities does not exceed three months,
thetotal amount ofdefault does not exceed 10% ofthebalance sheet value oftheenterprise and if thedebtors property that
ismortgaged, put inpledge orregister pledge, placed under a tax lien ormaritime mortgage isnot sufficient tocover thecosts
ofinsolvency proceedings. Moreover, thecourt will dismiss therequest if thedebtors assets are insufficient tocover thecosts
ofsuch proceedings.20 These regulations have important implications21 both from theperspective ofmeasurement ofthescale

14
15
16
17
18
19
20
21

4) with respect tothesole shareholder company oftheState Treasury thegroup also includes theminister responsible fortheState Treasury;
5) with respect toa corporation, partnership, limited partnership and partnership limited by shares undergoing liquidation each
oftheliquidators;
6) with respect toa legal person entered intheNational Court Register a representative appointed pursuant toArticle 26 item 1
oftheNational Court Register Act of20 August, 1997 (Journal ofLaws 2007, no. 168, item 1186, asamended3));
7) with respect tothedebtor who was granted public aid inexcess ofeuro 100,000 theauthority granting theaid.
Article 11 oftheInsolvecy and Restructuring Law, op. cit.
Articles 59 ibidem.
Article 382 ibidem.
Articles 1819 ibidem.
Articles 2225 ibidem.
Articles 7 and 8 oftheInsolvency and Restructuring Law, op.cit.
Articles 12 and 13 oftheInsolvency and Restructuring Law, op.cit.
More information on thesubject can be found inChapter 2 ofthis study.

26

ofinsolvency (the so-called hidden insolvencies occurs) aswell asthefactual assessment oftheeffects ofinsolvency, which may
be underestimated if based solely on insolvency statistics.
The actual practice and numerous studies show that intheoperation oftheinsolvency system, including its efficiency,
animportant role isplayed by theregulations governing theselection ofthetype ofinsolvency proceedings. Following thefiling
oftheinsolvency request, thecourt extends protection over thedebtors assets (if thedebtor has submitted a request tothat
effect) orattherequest ofa creditor. For this purpose, thecourt may appoint a temporary court supervisor, manager, receiver/
trustee (in theevent that thedebtor hides his assets orotherwise acts tothedetriment ofhis creditors), suspend all claims against
thedebtor subject tonegotiated agreements under relevant legal provisions orinstitute other measures toprotect theassets.22
Subsequently, thecourt may convene theinitial meeting ofcreditors unless itentails excessive costs, thetotal disputed claims
do not exceed 15% ofthetotal debt and there isa prospect that creditors will be satisfied toa greater extent by arrangement
than by liquidation ofthedebtors assets. At themeeting, thecreditors may adopt a resolution on how toconduct further
insolvency proceedings (with thepossibility ofreaching anarrangement ordeciding on theliquidation ofthedebtors assets),
on theappointment ofa creditor committee and may express their opinion on theofficial receiver/trustee, court supervisor
ormanager. Thecourt conducts insolvency proceedings inaccordance with theresolutions made attheinitial meeting ofcreditors
provided they comply with thelaw.23
The court oflaw declares thedebtors insolvency with anoption ofan arrangement if such a solution offers anopportunity tosatisfy
thecreditors claims toa greater extent than they would be after insolvency proceedings involving theliquidation ofthedebtors
assets. Otherwise, and if there isa reasonable risk that thearrangement isnot likely tobe executed orit does not comply with
thelaw, thecourt declares insolvency involving theliquidation ofthedebtors assets. However, even inthecourse ofinsolvency
proceedings thecourt isentitled tochange theprocedure if thegrounds forsuch analteration emerge only after insolvency has
been declared. As ofthedate ofdeclaration ofinsolvency, thedebtors assets become aninsolvencyestate.24
After declaring insolvency with theoption ofarrangement, theinsolvent debtor should submit theproposals forsuch
anarrangement within a month. Such proposals can also be submitted by thecourt supervisor, manager orthecreditor if thecourt
has declared bankruptcy with option tonegotiate a settlement attheir request. Theproposals should include a way ofrestructuring
theliabilities oftheinsolvent debtor, forexample postponement ofrepayments, debt rescheduling, repayment ininstalments,
a reduction inthetotal amount ofdebt, conversion ofdebt into shares, a modification orcancellation ofa decision safeguarding
a specific claim. Theterms ofdebt restructuring should be thesame forall creditors, and if a group ofcreditors votes on its
specific provisions, they should be thesame forcreditors belonging tothesame group. These conditions may be less favourable
toa particular creditor only on condition that he expresses his explicit consent. It ispossible togrant more favourable restructuring
conditions tocreditors who have smaller debts and tothose who after insolvency will provide theloans needed toexecute
thearrangement. Theproposed arrangement approved by a majority ofcreditors holding a total ofat least two-thirds ofthetotal
amount ofdebt and entitled tovote isthen subject toapproval by thecourt.25
Arrangement include claims which arose before thedate ofdeclaration ofinsolvency. After thefinal ruling approving
thearrangement comes into force, any security and enforcement proceedings are discontinued by law (Adamus 2009, p. 419-420
and 453-454). There isalso anoption toenter into anarrangement providing fortheliquidation ofthedebtors assets inorder
tosatisfy thecreditors (the so-called liquidation arrangement). 26 Once thearrangement has been executed, thecourt issues
a decision tothat effect, which istantamount tothecompletion ofinsolvency proceedings.27
The adopted form ofinsolvency proceedings need not tobe final. Thecourt changes theway inwhich theproceedings have been
conducted from thearrangement option toliquidation ofthedebtors assets if no arrangement has been reached.28
In thecase ofinsolvency which consists intheliquidation ofthedebtors assets, theofficially appointed receiver/trustee
takes over its management. Thereceiver/trustee isresponsible forthepreparation oftheinventory and financial statements
asattheday preceding thedeclaration ofinsolvency orsubmits a written report tothejudge responsible and conducts
theliquidation procedure oftheinsolvency estate. Liquidation involves, inparticular, thesale oftheinsolvent company asa whole
orinorganized parts, thesale its property, collection ofdebts owed totheinsolvent company and execution orsale ofits property
rights. Theinsolvent company issold free from all encumbrances. Theproperty issold by tender orauction inaccordance with
theprovisions oftheCivil Code of23 April 1964.29
22
23
24
25
26
27
28
29

Articles 3643 ibidem.


Articles 4448 ibidem.
Articles 1417, 61 and 288 ibidem.
Articles 267, 269, 270, 279, 285 and 287 ibidem.
Article 271 oftheInsolvency and Restructuring Law, op.cit.
Article 297 ibidem.
Article 286 ibidem.
Articles 57, 306308, 311, 317 and 320 ibidem.

27

The proceeds from theliquidation oftheinsolvent estate are used tosatisfy theclaims ofdebtors asprovided by thelaw. Funds
from thesale ofmortgaged property, property held inpledge, inregistered pledge, tax lien and maritime mortgage are used
topay off thedebts secured by theabove-mentioned methods (after deduction ofrelevant costs oftheproceedings). Funds
from theinsolvency estate are distributed according toa plan drawn up by theliquidator. Once thedistribution offunds has been
completed, thecourt issues anofficial decision tothat effect.30
Another important aspect ofinsolvency law isthesubjective structure oftheproceedings. As far astheentities involved
intheinsolvency proceedings are concerned,31 cases involving insolvency request are heard by theinsolvency court (District Court
Commercial Court) competen according tothemain seat ofthedebtors company. Thecourt declares thedebtors insolvency,
convenes theinitial meeting ofcreditors, approves thearrangements made by them, designates thesupervising judge, appoints
and approves theremuneration ofthereceiver/trustee, thecourt supervisor and/or themanager, hears appeals against rulings
and decisions made by thesupervising judge. After thedeclaration ofinsolvency, all thestages oftheprocedure that remain
outside thejurisdiction ofthecourt are carried out by thesupervising judge. Thejudge oversees thecourse oftheproceedings,
activities ofthereceiver/trustee, thecourt supervisor and themanager, indicating actions that they are allowed perform only with
his consent, points totheerrors made inthecourse oftheproceedings and investigates complaints against theinsolvency-related
actions ofthebailiff.
If thecourt declares insolvency involving liquidation ofthedebtors assets, all further actions are taken by theappointed receiver/
trustee. Thereceiver/trustee and thecourt supervisor ormanager carry out operations on behalf oftheinsolvency estate intheir
own name. Thereceiver/trustee manages theassets oftheinsolvency estate, protects itfrom destruction, damage orremoval by
unauthorized persons and carries out theliquidation oftheassets. Additionally, thereceiver/trustee isobliged tonotify thecreditors
whose contact data are known tohim.
If thecourt declares insolvency with theoption ofan arrangement with continued management on thepart ofrepresentatives
oftheinsolvency estate, itappoints a supervisor. Thecourt supervisor oversees all theactivities oftheinsolvent debtor and
his company until thefinal approval ofthenegotiated settlement oruntil theproceedings have been terminated otherwise.
If thecourt prohibits theinsolvent debtor from managing his company, itmust appoint a manager. Themanager takes over
theadministration oftheinsolvency estate, protects itfrom destruction, damage orremoval by a third party, draws up aninventory
and financial statements asattheday preceding thedeclaration ofinsolvency and makes all management decision associated
with therunning ofan insolvent enterprise.
The creditors meeting may be invited toparticipate intheinsolvency proceedings and decide whether toagree on a negotiated
settlement orliquidate theassets oftheinsolvent company. Additionally, thecreditors meeting passes resolutions, appoints a board
ofcreditors and issues opinions.
If themeeting ofcreditors fails toappoint a board ofcreditors, thesupervising judge isentitled todo so. Theboard ofcreditors
isa subsidiary body that oversees theactions ofthereceiver/trustee, thecourt supervisor ormanager, examines theinsolvency
estates funds, and authorizes activities that may be conducted only with its consent and issues opinions atthebehest
ofthesupervising judge, thereceiver/trustee, thecourt supervisor ormanager.
The main objective oftheInsolvency and Restructuring Law isto protect thecreditors oftheinsolvent debtor. Thus, during
theinsolvency proceedings they are protected by law. In accordance with Article 189 oftheabove-mentioned Law, all creditors
are entitled toobtain satisfaction from theinsolvency estate even if their claims are not notifiable. If a creditor wishes toparticipate
intheinsolvency proceedings, he should submit his claim within thespecified period (no less than one month and no more than
three months), unless theclaim does not require notification inorder tobe included on thelist ofreceivables (e.g. employment
receivables, claims oftheGuaranteed Employee Benefits Fund forreimbursement ofthebenefits paid totheemployees
ofthebankrupt company).32 Notification ofclaims entitles creditors toparticipate inthemeeting ofcreditors and intheboard
ofcreditors.
Debts subject tocollection during theinsolvency proceedings are be conventionally divided into two groups: insolvency claims
(arising before theinsolvency), and claims on theinsolvent estate (basically arising after insolvency has been declared). Theclaims
ofthelatter group do not require notification and tend tohave a higher priority. In turn, theinsolvency claims are satisfied inorder
ofpriority (i.e. by category) (Jakubecki, Zedler 2006, p. 687).
If aninsolvent enterprise isliquidated, thecreditors receivables are paid off from theproceedings obtained from theliquidation
ofthedebtors assets. If thefunds are insufficient tosatisfy all thecreditors infull, thecreditors belonging toa lower-priority
category are repaid only after a full repayment ofhigher-priority creditors. If thefunds are insufficient tosatisfy infull all thecreditors
30
31
32

Article 368 ibidem.


Articles 76, 150-153, 156, 160, 173, 176, 180-184, 191-213 ibidem.
Articles 236238 ibidem.

28

belonging tothesame group, their claims are repaid proportionally.33 If thecourt has approved anarrangement, thecreditors are
repaid asprovided by thearrangement (Jakubecki, Zedler 2006, p. 686). Furthermore, inaccordance with Article 93 oftheInsolvency
and Restructuring Law itispermissible tosatisfy some claims by way ofa deduction from theclaims oftheinsolvent company
aslong asboth debts existed on theday insolvency was declared. It isa privileged way ofpaying off creditors outside thedivision
ofproceeds.
The institution ofinsolvency proceedings isnot prejudicial totherights ofthecreditors against theguarantors and co-debtors
oftheinsolvent enterprise. Both theguarantors and theco-debtors have anobligation torepay thedebt oftheinsolvent
entity atthecreditors behest. However, they are entitled tointhis respect toclaim back theamount that they have repaid.
Theclaim arising thereby isrecorded on thelist ofdebts and specifies theamount that theguarantor ortheco-debtor has paid
tothecreditor.34 As regards thelegal position ofcompany executives who have failed toexercise due diligence and may be subject
toliability fortheir actions, thedebtor responsible forfiling theinsolvency request who does not submit itwithin two weeks from
thedate on which theevent constituting thecause forthedeclaration ofinsolvency occurred, isliable forthedamage inflicted
asa result offailure tomeet thelegal deadline.35 Additionally, thecourt may ban them fora period ofthree toten years from doing
business on their own account and toact asa board members, representatives orproxies incommercial companies, state-owned
enterprises, cooperatives, foundations orassociations. Such a ban isimposed on a person who through his/her own fault has not
filed theinsolvency request ina timely manner despite theobligation todo so; who after thedeclaration ofinsolvency has not
released thedata concerning thecompany assets ordid not identify them, including account books, correspondence orother
documents oftheinsolvent company despite thelegal obligation todo so; who after thedeclaration ofinsolvency conceals,
destroys orencumbers theproperty that constitutes part oftheinsolvency estate, oras aninsolvent person fails tofulfil other
obligations provided forby thelaw, a ruling by thecourt oflaw ora decision made by thesupervising judge. Thecourt may also
order theabove-mentioned punishment with respect totheperson who has atleast once been declared insolvent previously with
debt forgiveness attheend oftheproceedings ordeclared insolvent within five years ofthesubsequent declaration ofinsolvency.36
Providing false information intheinsolvency request, ina statement toinitiate recovery proceedings orduring therestructuring
proceedings ispunishable by imprisonment from three months up tofive years. Failure todisclose tothereceiver/trustee all
theassets oftheinsolvency estate, accounting books orother documents relating tothose assets and failure toprovide
thesupervising judge with information concerning theinsolvent debtors assets ispunishable by similar prison terms.37

1.4. Thetimeframe ofthebankruptcy process


The legal and institutional system has a significant impact on theefficiency/efficiency ofbankruptcy mechanisms, including their
costs and duration. It should be emphasized that very fast-acting, low-cost bankruptcy proceedings that meet stakeholder needs
render therole ofan EWS much smaller. A very important element inthis context includes thetime and structure ofbankruptcy
proceedings since EWS systems are always characterized by a certain time horizon intheforecasting ofthreats, orto be more
precise, aneffective warning timeframe. On theone hand such systems are characterized by a certain delay between theinput
signal and thewarning message, while on theother hand they have a limited perspective ofrational prediction. Here, by identifying
theindividual stages ofthebankruptcy process, we shall also specify thetime horizons within which EWS systems can be useful.
The basic steps inthebankruptcy process are presented inFigure 1.2. It isbased on two perspectives: thetimelines applying
tointernal and external stakeholders, respectively. It should be emphasized that inanalysing bankruptcy thekey external
stakeholders are thecompany creditors. Our analysis shows, however, that thegroup may also include potential investors, public
authorities, etc.
The starting point oftheprocess isa set ofevents that trigger certain processes within theenterprise inquestion which may result
inbankruptcy. Theperiod inwhich these events occur and begin tosignificantly affect thesituation ofthecompany shall be called
thecontamination (contagion) period.
It isworth noting that theduration oftheprocess may vary asdo thetype ofstimuli that have caused theenterprises problems.
To illustrate thesequence ofevents, we have included three options: a shock event (negative stimulus on a large scale occurring
within a short time), a systematic problem (a distinct trend with a varied slope) and shockwaves (fluctuations indifferent directions,
but within anoverall negative trend). It isobvious that thefirst effect islikely tobe felt faster and ina more pronounced way than
thelast one, where periods ofdeterioration are interconnected with periods ofrecovery. In fact, more such sequences ofevents
may lead toan enterprises problems (cf. e.g. Crutzen 2009, Lukason 2012).
33
34
35
36
37

Article 344 oftheAct ofFebruary 28, 2003, op.cit.


Articles 248 and 291 oftheAct ofFebruary 28, 2003, op.cit.
Article 21 ibidem.
Article 373 ibidem.
Articles 522523 ibidem.

29

Fig. 1.2. Thetimeframe ofthebankruptcy process vs theefficiency ofan EWS


Source: Own study.

The next stage comprises theincubation period. At that time, theeffects oftheevent that has triggered thefirms financial
problems are revealed. It can be quite a complex process asthere are a number ofpotential factors affecting thespeed and
clarity ofidentification ofarising threats by thefirms management. Certain reasons may be purely technical and be related e.g.
totheoperation offinancial and accounting systems reporting some data with delay. Other kinds ofthreats may become apparent
only later against thebackground ofexternal information such asmacroeconomic data, theperformance ofcompetition etc.
In thecase ofunfavourable information, animportant role may be played by psychological and sociological processes within
theframework ofethical and behavioural models prevalent intheenterprise, including a whole range ofbehaviours such
asattempts tocover up certain uncomfortable facts, fear oftransmission ofnegative e information tosuperiors etc. Moreover,
thecontamination and incubation periods may overlap with certain actions taken during theincubation period, which causes
additional problems. Theincubation period ends with thefirms decision-makers having sufficient knowledge oftheactual
problems and their extent.
Next comes thestage ofadaptation and corrective action. It covers theactivities implemented by theowners and managers
ofa firm atrisk genuinely aimed atsolving its problems. They may involve internal reactions (e.g. cost reduction, modifications
intheproduction structure, new investments) orexternal ones (usually raising additional capital, consultations with creditors etc.).
Activities whose goal isto buy time have specific implications fortheefficiency and thetimeframe ofbankruptcy procedures.
There are numerous analyses ofsuch behaviours, yet most ofthem reflect a sequence ofactivities illustrated inFigure 1.3 below.
Let us assume that anentrepreneur attime T = 0 made aninvestment funded with debt D tobe repaid intwo instalments: D1 and
D2 (disregarding theinterest rate). At that time he finds out whether theincome from business activity after period X1 enables him
torepay instalment D1 due atthetime (X1 > D1, probability p) orit does not (X1 = 0, probability 1p). If debt repayment ispossible,
thecompany continues tooperate until time T = 2, where thefinal result inperiod X2 depends on factors known only tothefirm
and may reach level X2H intheevent ofsuccess and favourable conditions fordoing business, orlevel H2L if theinvestment itself
issuccessful, but economic conditions are unfavourable. Theproject may also fail (without losing generality, we can then assume
theprofit toequal 0). Let us also assume that only inthecase ofprofits atthelevel ofX2H thecompany iscapable ofrepaying its
liabilities (X2H > D). At T = 0 thecreditor and theentrepreneur are ina situation ofinformation symmetry. In contrast, attime T =1,
thecreditor does not know thefinal outcome ofinvestment attime T2.

30

Fig. 1.3. Debtors actions intheperiod ofbankruptcy risk


Source: own study based on: (Sautner, Valdimirov 2013).

Let us consider thepossible actions taken by theentrepreneur inquestion if attime T = 1 he cannot repay theamount ofD1.
Clearly, he has two options. First, he may choose togive up fighting and file a bankruptcy petition. Second, he may try tonegotiate
anagreement with his creditors. These negotiations may take place between thepoints represented by T = 1 and T = 1.5. Note that
from thepoint ofview oftheentrepreneur such negotiations may have two consequences. One, they may result inan agreement.
Two, even though anagreement may not be reached, time T = 1.5 may bring about extremely beneficial circumstances forthefirm,
asa result ofwhich itbecomes solvent again (probability w). On theother hand, thefirms situation may become even more
complicated and lead toits permanent insolvency (Y < C + D, where K isthebankruptcy cost). It should be emphasized that if
bankruptcy law provides that intheevent ofbankruptcy ofthefirms owner isleft with nothing, he will be always willing toenter
negotiate since inthis way he either avoids bankruptcy ordelays it.
The creditor also benefits from negotiating with thedebtor between T = 1 and T = 1.5. Thenegotiations help him toatleast partly
redress theinformation asymmetry, inother words, thecreditor obtains additional information about theactual performance
ofthefirm inT = 2. If theknowledge thus gained allows thecreditor tobelieve that thecompany iscapable ofperforming
atthelevel ofX2H, he knows that theagreement will permit him torecover thedebt infull. It isa tempting alternative tofiling
a bankruptcy petition because more often than not w islikely tobe small. It isworth stressing that thecapacity and theduration
ofsuch negotiations depend, among others, on theposition ofthedebtor.38 Naturally, theparticular strategy tobe adopted also
depends on specific probabilities ofevents described inFigure 3.
To recapitulate, theadaptation period may be characterized by a variety ofinteractions between thedebtors and their creditors,
which are not always conducive totheefficiency ofthesubsequent possible bankruptcy proceedings. In theperiod spanning
T= 1 T = 1.5, thedebtor may, infact, demonstrate a range ofbehaviours with clearly negative consequences interms ofsocial
impact, including a tendency toover- orunder-invest. Theconsequences ofsuch actions are examined inmore detail inChapter 4.
If theabove-mentioned negotiations do not result inan agreement, thebankruptcy petition isfiled. In Poland, itmay be done by
thedebtor (within two weeks ofT = 1.539) orby thecreditor (essentially starting atT = 1). Afterwards, thecourt within approximately
two months (under thePolish law thedeadline isapproximate) issues anorder declaring thebankruptcy proceedings open and
their form. Appeals and other procedures may take another month. Consequently, theoretically within three months offiling
thebankruptcy petition a legally binding final ruling inthematter should be issued.

38
39

It may take longer if thedebtor isthemain recipient/buyer ofproducts from thecreditor.


It should be emphasized that theidentification ofcircumstances inwhich such anaction must be regarded asabsolutely mandatory is,
inpractice, difficult.

31

Figure 1.2 shows that there are atleast four useful warning horizons inEWS systems. Thefirst one (H1) may be advantageous
due totheopportunity toshorten theincubation period. Quite often, thecomplicated and unclear internal procedures applied
by a firm indistress may actually be replaced by anassessment oftheeffects ofshocks occurring during thecontamination
period using freely available tools. This option isalso attractive toexternal stakeholders. In their case, thelength oftheincubation
period depends, among others, on market efficiency, which may be significantly improved by anEWS. It should be noted that
thebasis forsolutions used inthis process istheir public nature. These tools are not only universally available, but also theindividual
predictive functions are published and can be applied by any stakeholder.
EWSs do not lose their advantages even when thewarning horizon extends toH2. In such a case, thesolutions offered by anEWS
can be very helpful intheprocess ofdetermining probabilities inFigure 1.2, which decidedly rationalises theprocedures involved.
Horizons H3 and H4 are also very useful, especially topublic sector stakeholders. Their advantages include:

In Poland, a number ofbankruptcies occur intheform ofout-of-court settlements. Accordingly, a warning system whose
horizon iseven longer than H4 constitutes a valuable source ofknowledge about thescale ofthephenomenon;

Information from theNational Court Register isscattered, and although individually itappears only with a slight delayed
relative tothetime ofissuance ofrelevant decisions, obtaining aggregated data isnot easy;

Polish law recognizes situations inwhich a firm istoo poor togo bankrupt. Although a portion ofinsolvency proceedings
enter thejudicial system, they also fall out ofit without anadequate adjudication that corresponds with theeconomic
situation oftheentities involved;

Each and every bankruptcy system needs tools toverify its quality inorder toassess whether itcorrectly identifies companies
which are healthy, bankrupt intheeconomic sense and bankrupt only inthefinancial sense.

An EWS may help thecourts inadjudication, especially intheperiod extending between H3 and H4.
In consequence, thearea ofEWS application can be quite broad and includes systems with markedly different effective warning
horizons.

Piotr Adam Boguszewski, Jarosaw Kaczmarek, Konrad Kolegowicz, Pawe Krzemiski

Chapter 2
MEASURING THE SCALE OF BANKRUPTCY IN THE ECONOMY
2.1. Introduction
One ofthe basic conditions necessary forthedevelopment ofaneffective early warning system (EWS) involves thequantification
ofbankruptcy risk. In thecase ofsmall-scale and/or limited-impact phenomena, thecost ofanEWS may actually be out
ofproportion toits potential benefits. Therefore, themeasurement ofthe scale ofbankruptcy inthe economy constitutes one
ofthe key objectives ofthe project and thepurpose ofthis chapter istoprovide both theconceptual background forsuch
measurement, including thedefinition, tools and knowledge concerning metadata related tothe measurement ofbankruptcy
aswell asbasic information concerning theintensity ofthe phenomenon.
Let us begin our discussion with ananalysis ofdata sources related tobankruptcy and their quality. Thefact that inPoland there are
no official and, atthe same time, comprehensive databases on thesubject proves that this isnot aneasy task. Thesituation ispartly
due toobjective reasons that exist inother countries aswell, since thecategory ofbusiness demographics includes more potential
states than just their birth, period ofactivity and death. To some extent, however, itis also theresult ofthe lack ofcentralization,
orat least inadequate harmonization ofsuch data sources inPoland. As a result, themost complete demographic picture ofPolish
companies involves their entries and removals from theofficial register ofcompanies. For reasons discussed below, removals must
not be identified with bankruptcies.
It isalso hard todevelop relative measures ofbankruptcy, i.e. thescale ofthe phenomenon relative tothe size ofthe population.
There are several reasons forthis problem, but two ofthem appear tobe offoremost importance. First, fora long time thePolish
legal system did not sufficiently well regulate theissue ofupdating company demographics databases inthe National Official
Register ofBusiness Entities (REGON). This especially applies tosmaller, single-person firms. Second, fora large group offirms
litigation isnot thedominant method ofsolving their financial problems. As a result, therelative measures ofbankruptcy inPoland
may be heavily biased.
We will also demonstrate that itis difficult toconstruct international statistics ofbankruptcy which would meet thecomparability
requirement. Poland compares favourably with most ofsuch statistics interms ofthe scale ofbankruptcy. A closer scrutiny,
however, reveals that such a conclusion may be misleading.
Significant problems associated with obtaining reliable and comparable data on bankruptcy are not only a challenge that applies
exclusively tothe planned EWS. Paradoxically, they also explain why theresults ofanalyses ofbankruptcy inthe economy may
tosome extent compensate forthese problems. This chapter discusses some estimations ofthe scale ofthis phenomenon based
on themethodology provided by theRapid Response Instrument (RRI, Polish acronym ISR).

2.2. Business demographics sources and quality ofdata


One ofthe main features expected ofearly warning systems against bankruptcy istheir ability toprovide a complete, rapid
and accurate identification ofthe phenomenon ofdeaths ofcompanies. Theanticipatory nature ofthese systems means
that their proper operation requires specific additional knowledge that permits us toanalyse thecauses ofsuch events and,
asa result, topredict them. Very important elements ofthe EWS are thus various demographic databases that reflect thelifecycle ofbusinesses: their births, growth and deaths, and theoption ofcombining1 these databases with strictly economic data
sets that reflect theeconomic causes ofcertain events inthe system. Unfortunately, this brief overview ofthe issues related
tobusiness data sources necessary toconstruct anefficient EWS forthis purpose, must start with anobservation that there are
a number ofsignificant problems and limitations which inthemselves have become one ofthe challenges facing thesystem.
Thedemographic databases constitute a complex system whose quality depends (at least currently inPoland) on thecooperation
amongst several registers and other subsystems. Thefunctioning ofthis mechanism isformally governed by a fairly large and
complex set ofdifferent legal provisions which are not entirely consistent with one another. In practical terms, even given thebest
legal regulations, theefficiency ofsuch a system affected by a number ofsociological factors; among others, by those related
tobusiness peoples attitudes towards disclosure requirements, including theprovision oftimely and reliable data on economic

An even more convenient solution would be tointegrate thedatabase ofdemographic metrics with theappropriate set ofeconomic data,
including financial data, forthefirms atrisk atthe level ofthe system. However, such a solution may be difficult and costly toimplement.

33

activity ortheir attitudes towards theproblem ofbankruptcy. Moreover, theinventory ofissues difficult tosolve has a permanent
and transnational nature since similar problems were discussed nearly 30 years ago even inthe USA (Sullivan, Warren et al. 1987).
If, inturn, we were toaccept that themeasure ofcurrent quality and international comparability ofdemographic data can be
gleaned from theinformation published by Eurostat, theimage isnot likely tobe optimistic either. Data on business deaths are
currently (i.e. 2014) available fortheperiod covering 20082011 only for24 ofthe 29 countries included inthe reporting system.
Moreover, all thedata for2011 are flagged aspreliminary and/or estimates. It isalso clear that itis easier tocollect data on thebirths
ofcompanies, because thesame database includes theinformation dating back to2004. Such long delays and short time series
lead tothe conclusion that theusefulness ofdata from this source from theperspective ofEWS design isvery limited. Indeed,
there are a number ofreasons why theconstruction ofinternational databases ofbusiness demographics, especially with respect
tocompany deaths can be very difficult (cf. e.g. de Koning 1998).
An overall assessment ofthe extent and quality ofdemographic data fornon-financial enterprises inPoland shall start with a brief
discussion ofthe key sources ofdata. Polands fundamental and official business demographics database isthe REGON system (the
National Official Register ofBusiness Entities). It was founded pursuant toArticle 41 ofthe Official Statistics Act (Journal ofLaws
1999, no. 69, item. 763) with a view toproviding a general and consistent characteristic ofbusiness entities operating inPoland
inbroad classification groups. Among others, itcontains information concerning:

The name and location ofthe entity,

Its legal status and form ofownership,

The activities pursued,

The founding date, thedates ofstarting, suspending and resumption ofoperation, entry inthe relevant registers orrecords,
termination ofbusiness orremoval from therelevant records orregistry and removal from theregister ofbusiness entities,

The anticipated number ofemployees.
The system isprimarily fed by forms RG-1 and RG-2 which report theregistration and removal ofanentity from theregister (Journal
ofLaws 99.69.763, 1999). Theprocess includes thelinking ofthe REGON registry with other registers, mainly with CEIDG (Central
Register and Information on Business Activity) and KRS (National Court Register). TheCEIDG register includes, among others, data
on civil law associations, and theNCR on general partnerships.
Only form RG-1 contains a box corresponding tothe state oflegal and business activity ofa given business enterprise, including
activities related toits liquidation orbankruptcy. Information on this subject isabsent from form RG-2, likewise, itis impossible
todetermine on its basis other above-mentioned demographic categories ofevents which have caused thesigning-off ofa given
entity. In thelight ofthe applicable regulations, theinformation fed into theREGON system iscontinuous and therefore should
be reasonably up todate. It should also be emphasized that registers ofofficial statistics are governed by certain principal rules
(European Communities 2003), inwhich thenotion ofchange practically comes down tothe issue ofthe so-called administrative
change which isnot always consistent with actual change (e.g. with respect tothe economic condition ofanentity). Itdemonstrates
that REGON, even though itmay be deemed valuable from thevantage point ofbusiness demographics, isaffected by significant
limitations interms ofits adaptability tothe needs ofanEWS.
It isalso worth noting that theCentral Statistical Office (GUS) collects information on thestate ofeconomic and legal activity using
other forms, such asF-02, SP and SP-3 (to a lesser extent). These are comprehensive (in thesense explained infootnote 1) data
sources containing demographic data, but one oftheir drawbacks isthat they are updated only on anannual basis. It constitutes
a significant limitation on theanalysis ofthe phenomenon ofbankruptcy, and therefore options touse data available from sources
updated more frequently are being considered (Chava and Jarrow 2004). Another drawback isthe rather long delay associated
with making such information publicly available (even up tothree quarters from theend ofthe calendar year covered by a given
report). We should also bear inmind theconsequences oflegislation concerning theprotection ofindividualised information inthe
official statistics system inprinciple, individualised data cannot be disclosed.
A fairly rich and comprehensive collection ofdata, including demographics, which partly meets theneeds arising from their
potential use inthe planned EWS, isthe database ofthe National Court Register (KRS) (Journal ofLaws 2011, no. 273, item. 1616
2011) whose functionality isavailable asthe Central National Court Register Information. This register could even be considered
asthe fundamental one if thephenomenon bankruptcy isinterpreted inits strictly legal sense. It isanintegrated electronic system,
which includes theinformation on: theformation ofpartnerships/companies, their capital, declaration ofbankruptcy, termination
ofinsolvency proceedings (forms, dates), suspension and resumption ofbusiness activity, liquidation ofentities, tax arrears and
overdue debts.2 The register also contains strictly financial information provided by theAccounting Act, including Articles 64
and 69 (Journal ofLaws 1994, no. 121, item 591 1994). Given therestrictions on access toindividualised GUS data, theKRS may
be theprimary source ofpublicly available information inwhich demographic data are linked with economic ones atthe level
2

The range ofdata submitted depends on thetype ofcompany.

34

ofindividual records. However, anexternal user ofthe system may face a practical problem that consists inhaving tomanually
re-create thedatabase using publicly available information. It must be remembered that information on bankruptcy based on
court rulings may be available with a delay resulting, among other things, from theprovisions ofbankruptcy law. Theproblem
isdiscussed ingreater detail inSection 1.4.
From our point ofview, a much more modest scope ofinformation iscollected aspart ofthe CEIDG database (Central Registration
and Information on Business). Theform CEIDG-1 contains only data related tothe commencement, suspension, resumption
ofbusiness and removal ofanentry from thedatabase. As a result, itis possible that inthe case ofcommercial-law companies we
have more demographic information than with respect tocivil-law partnerships. Since thelatter are usually smaller, italso means
that, atleast inPoland, thecontent ofdemographic data relating tothe SME sector ispoorer than that that oflarger enterprises.
Despite theexistence ofseveral registers containing information on demographic events inenterprises, itcomes asa surprise that
none ofthe organisations that maintain those registers publicizes focussed data on issues related tobankrupt firms. In particular,
when itcomes todata collated by GUS, its publication includes only demographic information related tothe current state
ofregistrations and removals ofentities by different sections (GUS Zmiany strukturalne... 2013). Apart from theabove-mentioned
official demographic business databases, other more comprehensive registers are being developed, among others, by consultancy
firms, banks and business organizations. These databases are fed by data from both thedifferent portions ofofficial databases and
research and analyses that meet thespecific needs ofthese institutions. These are often valuable collections, thesource oftheir
value added being generally theinterlinking and structuring ofinformation from different official sources, moreover, they convert
thebasic data, which isless convenient toperuse, into anelectronic format.
It isimportant tonote that a number offeatures ofthese demographic databases are not specific tothe adopted solutions. Here,
we are also dealing with some universal phenomena. First ofall itshould be noted that maintaining businesses demographics
databases ismuch more complex than ordinary demographics. In thelatter case we deal essentially with thethree states
ofentities (i.e. birth, life and death), whereas existing enterprises go through many more states. According tothe methodological
recommendations by Eurostat and theOECD (OECDEurostat 2007) asmany as13 types ofdemographic events should be
distinguished,3 10 ofwhich may affect theentry and/or removal ofanentity inregisters ofthis kind. Thecase isfurther complicated
if one remembers thedifferences between theremoval ofa specific entity from theregister (of active companies) and its actual
death. For example, acquisition ofone firm by another results inits disappearance from thedatabase, but isnot tantamount tothe
liquidation ofits resources. Another element that makes itmore difficult toevaluate ofthe demographic status ofanenterprise
isthe fact that businesses often suspend their activities. Such suspension may not, however, result from their economic problems.
Themain reasons may be due tocircumstances ofa seasonal nature (e.g. seaside tourist firms inwinter months), interruption
ofactivity due tothe necessity toobtain orupdate a license and a number ofother reasons. This raises thepractical question
ofhow todistinguish a temporary suspension from permanent cessation ofbusiness activities if a number ofentities fail tofulfil
theobligation toupdate such data inthese systems. What comes naturally tomind are thevarious types ofcomplex supplementary
procedures that would force businesses toreport more detailed data more frequently. Such a solution, however, would be costly
and burdensome not only toentrepreneurs themselves, besides, itwould not necessarily be more effective than theexisting legal
solutions. As a result, inthis type ofregisters itis often easier tofollow theprinciple that, forexample, a period ofinactivity lasting
fortwo-years (or longer) can be interpreted asthe death ofa firm inthe system4 (Journal ofLaws 1997, no. 121, item. 769 and 2004,
no. 173, item 1807). Nonetheless, from theperspective ofanEWS design this approach has noticeable drawbacks. Certain events
relevant tothe operation ofsuch a may be reported with long delays oreven ina discretionary manner.
Another, equally important group ofproblems, partly related tothe above-mentioned issues, isposed by theambiguity ofthe
concept ofbankruptcy. In particular, thephenomenon inquestion may have a purely legal sense. In such a case, its best
descriptions may be based on thecourt records (KRS-type). We must remember, however, that Polish bankruptcy law precludes
thepossibility ofdeclaring bankruptcy ofanentity which istoo poor togo bankrupt. On theother hand, there are a number
ofsituations, especially insmall firms, when thelegal procedures ofbankruptcy are dispensed with, and they terminate their
business activity themselves. Quite often, they are simply instances ofa silent death, which shall be discussed inmore detail below.
What itmeans isthat theregisters contain no information regarding thecauses ofsuch a decision, hence a more broadly conceived
bankruptcy would be better served by REGON-type databases. In this case, however, theregister includes more general information
concerning theremoval ofa given entity. Obviously, this need not be tantamount toits bankruptcy and many systems consistently
attempt totake into account this difference inthe design ofbankruptcy indicators (The Dun & Bradstreet Corporation 1997). Thus
we can see that thefull details ofbankruptcies actually lie somewhere between theKRS and REGON databases.

3
4

Such asthe founding ofa business, its liquidation, division, merger oracquisition by another entity.
In Poland, appropriate verification procedures inthe KRS are initiated after 24 months pursuant toArticle 20d ofthe National Court Register
Act. In specific cases regulations also limit thelength ofsuspension periods (cf. e.g. Article 14 ofthe Freedom ofEconomic Activity Act ).

35

Our previous considerations demonstrate additionally that thequality ofdemographic data about businesses largely depends
on theattitudes ofbusinesspeople toa reliable and timely supply ofinformation tosuch databases. It appears that inthis respect
larger companies tend tobe more compliant, toanextent because itis easier toget tothem toobserve theregulations by using
appropriate legal tools. Incidentally, this raises yet another problem related tothe EWS, namely that theSME sector plays aneverincreasing role inmodern economies. Poorer quality data concerning this sector would thus negatively affect theefficiency ofthe
system asa whole. Failure toupdate therecords by self-employed individuals affects numerous countries. Although theregistration
ofanentity pays off, itis more difficult tomotivate thefirms toremove data from thedatabase inthe event oftermination
ofbusiness. Single-person firms die not only because ofthe entrepreneurs decision toterminate activity ortheir failure toupdate
therecords fora long time, but also theactual death ofits owner. In such a situation, thefirm itself will not update theregister;
consequently, theefficiency ofthe entire procedure will depend mainly on thequality ofinterlinkage between such a database
and other registers. In thecase ofREGON, these mechanisms did not work sufficiently well, which resulted inanostensible increase
inthe number ofentries rooted inits sub-optimal dealing with thephenomenon ofdead souls. One ofthe long-term effects ofthe
problem was theensuing need toremove theaccumulated errors, which made itmore difficult tocompare demographic data
from this source over time.5
Up tothis point we have focused on issues related tothe registration ofcertain demographic events inthe business sector. It
should be noted, however, that bankruptcy has its roots ineconomic matters. Hence, if we plan toconstruct anEWS, we need
databases that should also contain financial data ofadequate quality inorder toenable us toglean thedemographic determinants
ofspecific events. Theproblem ofquality financial data offirms infinancial distress orbankrupt isvery complex and exceeds
thescope ofthis publication; therefore we shall limit ourselves toselected fundamental issues. Such data on this group offirms
may be subject toa number ofdistortions. First, inaccordance with Article 5, item 2 ofthe Accounting Act (Journal ofLaws 1994,
no. 121, item. 591 1994) theprinciple ofcontinuity ofbusiness activity applies. This principle may, ofcourse, stand instark contrast
tothe actual situation ofthe enterprise facing bankruptcy risk, not tomention thealready bankrupt one. Although this provision
also states that therule does not apply if such anassumption isinconsistent with thefacts asat thedate ofdrawing up thebalance
sheet, often itdoes not solve theproblem. After all, atthe moment inquestion thebankruptcy ofa given firm can be either
a foregone conclusion ora state only tobe expected with a considerable error margin. Second, if bankruptcy risk affects entities
with a positive option to wait, they may have a significant tendency toadjust their balance sheets tomake them look better than
inreality. Third, even marginal symptoms ofbankruptcy risk may cause quite significant adjustment reactions not only on thepart
ofa given firms creditors but also its other stakeholders. Further, given theexistence ofdifferent types ofrisk scenarios, thebalance
sheet and other elements offinancial statements ofthe companies affected may be subject tovarious distortions and if they are
tobe used forthepurposes ofanalysis and research, including forecasting toall intents and purposes, they should be subject
toadditional verification.
All theabove-mentioned problems build up when attempts are made todevise databases on aninternational level due tothe
complex issue ofcomparability ofinformation obtained from various countries. As a result, all international statistics on bankruptcy
should be approached with caution, likewise, theactual reliability ofinformation, particularly from thepoint ofview ofEWS design
may be, aswas already mentioned, problematic (cf. e.g. de Koning 1998).
In summary itcan be stated that:

The business demographic information system inPoland consists ofand isfed by a number ofdatabases whose operation
isgoverned by various and not always consistent legislation from theperspective ofthe bankruptcy phenomenon;

In some instances, theregistration ofdemographic events itis not anefficient process; more considerable errors affect
thestatistics that apply tothe death ofenterprises rather than their birth; thesources oferrors include, among others,
thesuspension ofbusiness activities;

More serious problems may affect thedeath statistics ofsmaller entities, whose frequency may be underestimated. This has
led (and may still lead) tooverestimating thenumber ofentities inthe REGON database;

Obtaining a comprehensive database ofdemographic data (i.e. one that combines strictly business demographic information
with relevant economic data) on thebasis ofregisters is, forusers from outside thesystem ofpublic statistics, limited tothe
data contained inthe KRS;

The overall quality ofdemographic information on SMEs ispoorer than that on larger companies;

Observations incomprehensive demographic databases usually occur atmost on anannual basis (although strictly
demographic data can be virtually continuous);

The economic data included inthese comprehensive databases are by nature prone tovarious kinds ofdistortion;

The process ofremoving thedead souls from registers has (apparently) resulted ina decrease inthe number ofactive entities. Theprevious
data underestimated both theabsolute number ofactual deaths ofbusinesses and thepercentage ofthose events.

36

It isdifficult totalk about international comparability ofbusiness demographic data between Poland and other EU countries,
atleast before 2008. Thedelays inthe publication ofthe official (Eurostat) statistics inthis area are quite considerable (over
two years)

Therefore thequestion arises whether, taking into account thecurrent conditions inPoland, thesystem ofinformation on
bankruptcies inthe sense defined by Williamson and Ginsberg (Williamson, Ginsberg et al. 1997) could be easily adapted tothe
EWS needs. Theanswer must be negative with certain reservations. Positive examples ofsuch systems are not far tobe found.
Thesolutions adopted inPoland pale especially incomparison with thecomprehensive information systems inEstonia6 that offer
detailed and cross-linked databases on thesubject. There theproblem was solved pragmatically across theboard, i.e. thesystem
was implemented together with theadoption ofthe new bankruptcy law. It should also be noted that atpresent neither GUS
(with its public programme ofstatistical surveys) nor theMinistry ofEconomy conduct dedicated research on thephenomena
ofbankruptcy.
In designing theRapid Response Instrument itwas necessary totake into account theabove-listed considerations, which was
achieved intwo ways. On theone hand, a choice ofexisting databases was made with a view tooptimally exploiting their
potential.7 On theother hand, theproject enabled us todevelop system functionalities which tosome extent fill gaps and remedy
certain shortcomings ofthe currently operating systems. Thetarget solution was touse individualised data from databases
F-01/F-02 GUS inclose cooperation with theinstitution. These data sets include, among others, individualised information on:

A broad range ofdemographic events;

The economic condition ofenterprise which can be related tothe above-mentioned events;

The classification keys that appear tobe very important fortheEWS (including information on thesize ofa given entity,
including employment levels, its geographical location, branch ofindustry represented etc.).
It isalso important that thedata sets include a representative (since itis complete) population ofentities that employ atleast 10
staff. Thecooperation with GUS guaranteed on theone hand thehighest professional competence inthe area ofdata processing,
while on theother hand ensuring statistical confidentiality. As a result, itwas possible toobtain thelargest and most comprehensive
data set inthe area ofresearch on thescale, causes and consequences ofbankruptcy inPoland. In theearly stages ofthe project,
a simpler approach was utilized based on theso-called micro-aggregates, i.e. aggregated individualised data foranalytical purposes
that met theprinciple ofstatistical confidentiality. These data were combined with theproprietary database ofinformation on
bankruptcy based, among other things, on theavailable KRS reporting system. Expertise thus gained can be valuable 8 from
theviewpoint ofimplementation ofthe future EWS incases where solutions based on individualised statistical data are impossible
toimplement.

2.3. Aggregated measures ofbankruptcy


The problem ofex-post measurement ofthe scale ofdeath/bankruptcy offirms inthe economy asa whole (or its sectors)9 appears
tobe deceptively simple: why not use aggregated measures constructed asthe total ofindividual cases identified asbankrupt
companies (or subject toa serious risk ofinsolvency)? We must, however, mention atleast two groups ofproblems arising atthe
very first stages ofreflection. One ofthem concerns theextent ofthe analysis. Some ofthe difficulties occurring inthis area stem
from theproblems which were already described and are related tothe structure ofthe relevant demographic databases. It
isobvious that themeasurement method depends, among other things, on thedefinition/identification ofbankrupt firms and
on theway ofdetermining thepopulation ofcompanies. Thelatter group contains animportant new element which results
from theneed toprovide not just point-in-time scores ofthe phenomenon, but attempts tocompare them along a timeline. It
isanextremely important element inthe EWS design, because thewarning signs are always identified on thebasis ofthe observed
deviations from their (historical10) mean values. Accordingly if we plan toconstruct a measure ofintensity ofthe bankruptcy
phenomenon, we must have access not only toabsolute variables, but also torelative indicators which take into account
thedynamics ofchange inthe population ofbusinesses.

6
7

10

Cf. E.g. [https://www.lursoft.lv/exec?act=MNR_LSTAT&lang=EN&aid=#aid#].


It should be emphasized that theproject provided fortheconstruction ofproprietary databases only toa very limited extent (for example, itwas
hardly expected toinclude thedevelopment ofstate registers). Therefore, theselection process has tobe based on theexisting opportunities.
Naturally, such a simplified system has a number offlaws and weaknesses with respect tothe target solution, but itgenerates series
ofsufficiently well-approximated results incomparison with thecomplete ones.
Note that theultimate objective ofthe EWS isnot todetermine individual bankruptcy risk but toevaluate theextent ofthe phenomenon and
its effects from a broader perspective.
It isdifficult totalk about another source ofsuch a mean value.

37

380000
360000

1000000

340000

950000

320000
300000

900000

280000

850000

260000

800000

240000

750000

220000

Half-year period

200000

I.04
I-II.04
I.05
I-II.05
I.06
I-II.06
I.07
I-II.07
I.08
I-II.08
I.09
I-II.09
I.10
I-II.10
I.11
I-II.11
I.12
I-II.12
I.13
I-II.13

700000

Number of enterprises

Number of enterprises

1050000

Legend:
Total excluding natural persons
Exponential trend

Structure by employment class (% share)

400000

1100000

100
90
80
70
60
50
40
30
20
10
0
2003 2004 2005 2006 2007 2008 20092010 20112012 2013 2014
Joint stock companies:
1049 employees
Up to 9 employees

Total general partnerships (right axis)


Exponential trend

Above 50 employees

Fig. 2.1. Thenumber ofenterprises excluding natural persons and general partnerships (left panel) and thebreakdown ofgeneral
partnerships by number ofemployees (right panel) in20042013
Source: own calculations based on: GUS Zmiany strukturalne... (2013).

The simplest relative measure ofbankruptcy (WMU) can be defined asfollows:


WMUt =

DEt
AEt K

(2.1)

where DEt represents thenumber ofbankrupt enterprises attime t (an absolute measure), whereas AEt-K thenumber ofentities
active inthe period t-K (i.e. those that reported a non-zero turnover oremployment). As was already mentioned, thedefinition
ofDE depends on thechoice ofthe demographic database. If itwas a REGON-type register, we would use thefollowing absolute
measure:

DEt = Zt \ (Rt K t )

(2.2)

where theindividual elements are defined asfollows:

Zt = {e AEt e AEt +1}

Rt = {e AEt e AEt +1 e AEt +2 }

(2.3)

Set Z can be interpreted asa family ofcompanies suspended (removed) up totime t+1; R companies active inperiod t+2 and
inactive one year before. Theset K reflects theother adjustments. Note that thestructure ofmeasure DE isbased on thenumber
ofoperators inactive inthe time horizon t, t +1 and its adjustment forthenumber ofentities suspending their activity attime t+1,
recommencing itin period t+2 aswell asother reasons forremoval from theregister unrelated tobankruptcy ofa given entity
(e.g. takeover, etc.). As a result:
DEt = {e AEt e AEt +1 e AEt +2 }

(2.4)


The application ofdefinition (2.1) actually implies theneed touse thedata coming from two reporting periods before (two years
inour case), which constitutes a serious drawback ofthis approach from theperspective ofEWS design. Themeasure, however,
has atleast one advantage theconsistency ofthe numerator and thedenominator ofthe fraction inthe sense that both values
are calculated on thebasis ofthe same population (sample). For practical reasons, a simplified measure can also be used (WMUS):
WMUSt =

Zt
AEt 1

(2.5)

but incomparison with (2.1) this measure isobviously distorted by thefact that itdoes not take into account theprocess
ofresumption ofactivity ofsome such firms orother causes ofremoval from theregister (e.g. change ofcompany name).
Quite serious measurement problems also arise when anattempt ismade toreplace thestatistics ofthe firms deaths
indemographic registers with metrics based on court data related tobankruptcies which have already been declared. Even though
thesituation issomewhat easier interms ofabsolute measurements, theissue becomes even more complicated when relative

38

measures are sought. As regards theformer situation, themeasure based on thestatistics ofcourt decisions strongly depends
on theefficiency ofthe judicial apparatus and theinflux ofcases into thesystem. This may cause theeffect ofsmoothing out
ofdata ofthis kind which consists inthe fact that during theperiod characterised by a marked increase inthe number ofrelevant
cases, we do not notice thefull effect ofthe phenomenon inthe form ofgrowing numbers ofcourt decisions, since a proportion
ofthose cases will be addressed with a delay due tothe systems limited processing capacity. On theother hand, when theintensity
ofsuch disruptions inthe economy abates, judicial statistics may still demonstrate large numbers ofrulings due tothe backlog
ofaccumulated cases being processed. Therefore anabsolute measure based on thenumber ofdeclarations ofbankruptcy may
be quite problematic from thepoint ofview ofits applicability inthe EWS. Given these weaknesses itappears that a better solution
might be based on thenumber ofbankruptcy petitions filed.
Let us, however, remember that regardless ofwhether we utilise thenumber ofbankruptcy petitions orthe number ofactual
declarations ofbankruptcy, there remain atleast two principal weaknesses inherent insuch measures, especially if one planned
touse them forinternational comparisons orin order toassess thelong-term national trends. First ofall, thenumber ofbankruptcies
depends not only on theoverall situation ofthe economy and thefirms operating init, but also on theprevailing social tradition/
culture inthis respect. It should be stated clearly that inPoland thebankruptcy procedure does not constitute thedominant
mode offirms exit from themarket and, consequently, theincidence ofbankruptcies inPoland reported annually issimilar tothat
reported monthly inBelgium (Kartus 2012). Another factor confounding themeasure isbankruptcy law itself. Solutions adopted
inPoland are deemed tobe inclusive, which, according tocertain judges dealing with this issue, means that about ofdomestic
firms consistently meet thestatutory conditions forbankruptcy (Pilitowski 2013).
The structure ofrelative measures (MWU_KRS) isasfollows:
MWU _ KRSt =

MBU _ KRSt
POPt K

(2.6)

where: MBU_KRS isthe absolute measure ofbankruptcy based on theKRS, whereas POP isthe regulating factor. An additional
problem associated with relative measures consists inthe proper selection ofthe regulating factor, inother words, thepopulation
from which theentities involved insuch litigation are derived. This isnot a trivial issue if we remember that thetendency (and
economic capacity) offirms tocomply with insolvency proceedings are not uniformly distributed throughout their population.
Such actions tend tobe pursued by larger enterprises. Accordingly, itwould be a questionable solution debatable tobase
thedenominator informula (2.6) on theAE variable (cf. formula 2.1 thesize ofthe full population ofactive enterprises). Using
such a solution would be fraught with anadditional disadvantage. Figure 2.1 shows that not only thesize but also thestructure
ofthe population ofactive enterprises changes intime, with a clear increase inthe share ofsmall firms, and therefore less willing11
toparticipate inthe insolvency proceedings. As a result, itwould constitute a source ofadditional burden on theindicator.
So far, we have only discussed theex-post indicators. As expected, they are useful ina number ofsituations, but their weaknesses
appear tobe very difficult, if not impossible toeliminate, since their primary causes stem from theabove-mentioned problems
associated with theconstruction and updating ofrelevant databases. One possible way out ofthis predicament istodevise
indicators based on individual assessments ofbankruptcy risk. Such anapproach not only has a prognostic dimension but may
also be used toassess thecurrent situation and conduct retrospective analyses. In this case, theoften questionable and incomplete
information on bankruptcy, which additionally concerns firms deriving from anunidentifiable population,12 isreplaced with
assessments/forecasts ofbankruptcy risk drawn up inaccordance with theestablished methodology forfirms that belong tothe
population/sample defined inadvance. Different approaches are possible inthis respect, but they are often based on thefollowing
formula (Das, Freed et al. 2006):
GLOB _ PDt =

1
N (t )

N (t )

PD (t )

(2.7)

i =1


where: GLOB_PD isthe aggregated indicator ofbankruptcy risk; N(t) thenumber offirms inthe sample attime t, forwhich
theindividual indicators ofprobability ofbankruptcy PDi are estimated. Naturally, these idiosyncratic measures can be
constructed drawing on different models, including those that make use ofboth individual financial data and measures ofa more
aggregated nature, e.g. forindividual branches ofindustry orfor thewhole economy etc. (cf. e.g. Sobehart, Stein et al. 2000). One
ofthe justifications forsuch a design solution involves its certain statistical properties (Lando and Nielsen 2009). If theintensity
ofbankruptcy risk fora given firm i isdefined as:

11
12

By thesame token, such companies tend tobe less attractive tothe creditors.
In thegeneral population there isa group offirms that never resort tothe formal bankruptcy procedures.

39

i (t) = lim

t 0

P (t < i t + t | i t )
t

(2.8)

where isthe moment ofbankruptcy ofthe firm and N(t) thenumber ofsuch events up totime t, where N isthe number
ofanalysed firms, then:
N

N(t) =

1| (t t)

(2.9)

i =1

and assuming theexistence oforthogonality ofthese events forcouples offirms understood as:
P (t i = t j ) = 0 i j

(2.10)



then thecumulative function ofevent intensity (t) forprocess (2.9) can be represented as:
N

(t) =

(t )* 1| ( t)
i

(2.11)

i =1


After integrating, thefunction:

(t ) =

l(s)ds

t 0

(2.12)

Po (t ) = N (t ) t 0
(2.13)


we may obtain valid results. Given theassumptions adopted above, thecumulative measure ofbankruptcy risk described by
formula (2.9) isa single-parameter Poisson process (2.13). Further, itshould be noted that inorder toobtain this result itsuffices
that theorthogonality condition be met (2.10), whereas conditional independence ofthese events isnot necessary. This result,
therefore, also remains true inthe case ofcontagious bankruptcy inthe corporate sector, which may be a common situation tobe
diagnosed by theEWS. Consequently, formula (2.9) issuitable foruse insuch a system.

The primary direction ofresearch highlighted inthe study involves theanalysis ofthe phenomenon ofbankruptcy both
inretrospective and inprospective terms illustrated by thereview ofbankruptcy proceedings conducted inPoland inaccordance
with appropriate formal and legal provisions. It also analyses thecourse and structure ofbankruptcy proceedings from
theeconomic point ofview, drawing attention toasset liquidation (winding up ofbusiness activity) and negotiated settlements
(a business continues its independent economic existence) broken down by region and type ofbusiness. Subsequently, regional
(voivodship) analyses ofbankruptcies have been conducted using indicators such asthe percentage ofbankruptcies (OU) and
theregional bankruptcy barometer (RBU). One ofthe analytical strands included thevariations inthe course ofthe bankruptcy
process depending on thelegal and organizational form ofthe participating entities. Thefinal layer isconstituted by theanalysis
ofbankruptcies by branch ofindustry (in accordance with theNACE classification) using measures such asthe percentage
ofbankruptcies and theindustry bankruptcy barometer (BBU). Theanalysis included theabsolute number ofbankruptcy
proceedings announced, thedynamics oftheir change interms oftotal figures and cross-sections by region, by industry and by
organizational and legal form. As was thecase with thereview ofthe structural changes, our research consisted inthe calculation
ofrelevant indicators reflecting thepercentage shares ofthe analysed components and evaluated structures.
In order todetermine thescale ofbankruptcies we used thepercentage ofbankruptcies (OU). It isthe ratio ofthe number
ofcompanies forwhich bankruptcy proceedings were initiated tothe total number ofenterprises per 10,000 calculated on
anannual basis. It may take value from theinterval <0; 10,000> and indicate a deteriorating economic situation orits improvement
infirms. TheOU value iscalculated using thefollowing formula:

OU =

Lu x10000
Lp

where:
OU percentage ofbankruptcies,
Lu number ofenterprises subject tobankruptcy proceedings,
Lp total number ofenterprises.

40

(2.14)

Changes inthe calculated values ofthose indicators over time assists inthe identification ofcomponents ofthe analysed structure
whose shares increase, decrease, do not change oronly slightly change. Theanalyses ofstructural changes inthe compared
structures incorresponding periods oftime are based on thedifferences between thevalues ofindicators characterizing thesame
indicators that apply tothe compared structures.
Additionally, theanalysis comprises theextent ofbankruptcies expressed intotal figures and then divided by sector and by region.
Its aim was tocreate a ranking list ofactivities and regions (voivodships) characterized by thehighest level ofconcentration
ofbankruptcy proceedings. In order toanalyse theconcentration ofbankruptcies we used a measure applied ineconomic
geography taking into account thedensity ofdistribution and known asthe location quotient (LQ).13 By surveying thephenomenon
ofbankruptcy not only on a national scale but also inthe regions, theintensity ofthe phenomenon can be related tothe number
ofentities registered ina given area. In this way, themeasure ofbankruptcy proceedings can be related tothe actual intensity ofthe
demographic changes ina given region. For thepurposes ofEWS research, theRBU indicator reflects theconcentration ofbankrupt
firms ina particular region [Luw/Lpw] relative tothe total number offirms that belong tothe analysed population inthe economy
inrelation toall entities that operate inthe national economy [Luo/Lpo]. This measure can be expressed inthe following way:

Lu
Lp
RBU =
Lu
Lp

(2.15)

where:
RBU rate ofregional concentration ofbankruptcies,
Luw number ofenterprises placed inreceivership ina given region (voivodship),
Lpw number ofenterprises inthe region,
Luo total number ofenterprises placed inreceivership,
Lpo total number ofenterprises.
Since theRBU serves todetermine theintensity ofthe studied phenomenon inthe region inrelation tothe overall economy,
itis assumed that its values above 1 indicate anabove-average concentration ofthe analysed phenomena inthe region. By
using theindicator itis possible torank theregions, thus indicating areas with anabove and below average concentration
ofbankruptcies.
In order toassess changes inthe relative extent ofthe phenomenon inthe different types ofbusiness we have used thegeographic
location quotient (LQ) and called itthe Industry Bankruptcy Barometer (BBU) by way ofananalogy tothe Regional Bankruptcy
Barometer (RBU). Thecomputed BBU values enabled us tocompile a ranking list ofeconomic activity indicating areas ofbusiness
with anabove and below average concentration ofbankruptcies by industry (manufacturing, trade and services).
This measure iscalculated on thebasis ofthe following formula:

Lu
Lp
BBU =
Lu
Lp

rd

rd

(2.16)

where:
BBU industry bankruptcy barometer,
Lurd number ofenterprises placed inreceivership by type ofbusiness (manufacturing, trade, services),
Lprd number ofenterprises operating ina particular type ofbusiness (manufacturing, trade, services),
Luo total number ofenterprises placed inreceivership,
Lpo thetotal number ofenterprises.
Accordingly, both theBBU and theRBU contribute toa relative assessment ofthe concentration ofinitiated bankruptcy
proceedings with respect tothe number offirms operating inthe analysed population. It isassumed that their values above 1
indicate anabove-average concentration ofthe analysed phenomena inthe region.

13

Cf. (Antonowicz 2010).

41

2.4. Bankruptcy inPoland selected characteristics


Bankruptcies inthe entire Polish economy shall be analysed based on two sources ofinformation: bankruptcy court statistics and
data concerning removals from theREGON register (see Figure 2.2). When interpreting theresults presented below, thelimitations
applicable tothese data and mentioned inthe previous section should be borne inmind.

Fig. 2.2. Selected information regarding bankruptcies inPoland in20032013. Thenumber and dynamics ofbankruptcies according
toCOFACE (top left). Thenumber and dynamics ofremovals from theREGON register excluding natural persons (top right).
Therates ofremovals from theGUS register forthetotal population and excluding natural persons (bottom left) and
thebankruptcy rate based on total COFACE data and excluding natural persons (bottom right) data inpercentage points
standardized tothe number ofentities inthe REGON database
Source: own calculations based on data from COFACE and GUS.

When analysing both absolute and relative data on bankruptcies (COFACE) itshould be noted that after a long period ofdecline
(since 2003) theend of2008 marked a growing, although non-monotonic trend affecting both measures with a very strong increase
noted in2009, and a certain slowing down ofthe process in2010 related, among other reasons, tothe fairly good macroeconomic
situation in20102011. Although theincrease inthe number ofbankruptcies after 2008 can be seen quite distinctly, thescale
ofthe phenomenon issignificantly smaller than in2003, when bankruptcies resulted from themarked economic slowdown and
major structural changes that occurred place inPoland atthe beginning ofthis century.
A similar picture ispainted by theanalysis ofremovals from theREGON register aslong asnatural persons engaged ineconomic
activity intheir own name are excluded. Likewise, after a period ofanalmost uniform decline until 2008, asof 2009 we can see
a sharp rise inthe value ofthe variable, with thecurrent removal rate much lower than in2003. A certain difference isnoticeable
only towards theend ofthe time horizon under review, where theCOFACE-based index shows a decreasing trend while theREGONbased one indicates a gradual increase. It isdifficult toascertain thecause ofthis discrepancy, but itmay be due tothe higher feed
rate ofthe bankruptcy database than that ofthe demographic register with new events.
Statistical data on removals from theregister involving natural persons conducting economic activity (self-employed) show
different trends than thedata on bankruptcies. In 20032009 quite a clear upward trend can be observed interms ofthe relative
measure ofthe phenomenon and anextra strong peak in2011. This difference may be due toseveral coinciding reasons. First,
itmay have been affected by theevolution ofthe entrepreneurship model inPoland that prompted anincreasing tendency

42

towards self-employment, which also resulted ina greater number ofbusiness failures. Second, therestructuring processes
undertaken by larger entities improved their financial standing oreven allowed them toavoid bankruptcy by reducing costs
through outsourcing. Third, thereadings may have been affected by anundue accumulation ofdead souls inthe system and their
subsequent elimination through intensive verification and updating efforts on thepart ofGUS.
In summary, both data sources show that theeconomic crisis caused a perceptible increase inthe bankruptcy indicators. In 2009,
itwas quite marked, but because ofthe favourable macroeconomic situation inPoland inthe previous period thescale ofthe
phenomenon was moderate. In 2003, theimpact ofthe economic downturn on bankruptcies was much stronger, due, among
others, tothe much worse initial situation ofindividual firms atthe time. It should also be remembered that long-term analyses
(e.g. from theUnited States) demonstrate that even though theaverage bankruptcy rate totals several percent even during
a slump inthe economy there have been crises when theindex reached double digits (Giesecke, Longstaff et al. 2011). Thedata
thus show that anEWS due tothe potential scale ofthe problems promises tobe aneffective solution toimprove thesecurity
ofthe economy.
The total number ofinitiated bankruptcy proceedings (since early 2007) ischaracterized by a strong irregular upward trend with
a peak inthe fourth quarter of2012. Afterwards, thenumber ofinsolvency proceedings has slightly decreased. Nonetheless,
neither thedecrease inthe number ofcases reported nor thelow values ofthe rate allow us foroptimistic forecasts that might
indicate a significant improvement inthe economic situation offirms, which isdue tothe still very high numbers ofbankruptcy
proceedings announced. It isworth emphasizing that theincrease inthe number ofinsolvency proceedings was highly dynamic,
which has not been observed inthe Polish economy since 2003. It isnecessary, however, tonote that due totheir complexity
theannounced bankruptcy proceedings constitute only a small portion ofall insolvency proceedings inrelation tothe scale
ofliquidation processes inthe entire economy, moreover, thephenomenon isdifficult tomeasure using currently available
reporting. Thebankruptcy rate (OU) and thenumber ofbankruptcy proceedings has increased since 2009. This confirms both
theincreasing intensity ofthe analysed phenomenon and thecontinued negative trend during thelast two years under review.
Therate ofOU growth has slightly slowed down, which undoubtedly indicates theweakening ofthe trend.

Fig. 2.3. Thebankruptcy rate (OU) and its y/y change inPoland from thefirst quarter of2007 tothe third quarter of2013
Source: own calculation based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek, Raport z oceny stopnia zagroenia przedsibiorstw upadoci komponent
mikroekonomiczny, Raporty 112, MSAP UEK, Krakw czerwiec 2011 kwiecie 2014; Oglnopolskie Informatory Upadociowe , Raporty COFACE.

Regardless ofthe fact that thelaw provides fortwo kinds ofbankruptcy proceedings liquidation ofthe debtors assets
ornegotiated settlements with creditors thelatter isalmost five times less popular. Thestructure ofbankruptcy proceedings
shows that asset liquidations accounted forasmany as83% ofthe total bankruptcy proceedings. Therates ofboth types
ofbankruptcy proceedings are characterized by similar patterns ofchange, which indicate their marked growth in2009 and 2012.
Therate started todecrease gradually in2013.

43

Fig. 2.4. Thenumber ofbankruptcies by type ofproceedings inPoland from thefirst quarter of2007 tothe third quarter of2013
Source: own calculation based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek, Raport z oceny Raporty 112, op.cit.; Oglnopolskie Informatory Upadociowe, Raporty COFACE.

Despite thefact that natural persons conducting economic activity constitute thelargest group offirms, they account foronly
20% ofall theinitiated bankruptcy proceedings. Undoubtedly, most bankruptcy proceedings involved limited liability companies,
which account for63% ofall theproceedings followed by joint-stock companies and registered partnerships. Thetotal number
ofinitiated bankruptcy proceedings against limited liability partnerships has increased dramatically since 2009. Thetrend peaked
in2009 and slowed down in2013, corresponding tothe total number ofproceedings.

Registered
partnerships
4%

Cooperatives
3%

Other forms
2%
State-owmed firms
0%

Joint-stock
companies
8%

160
140
120
100
80
60
40

Natural persons
20%

20
0

I2
0
II 2 07
III 007
2
IV 007
2
I 2007
II 2008
III 008
2
IV 008
20
I 2 08
II 2009
III 009
2
IV 009
20
I 2 09
0
II 2 10
III 010
20
IV 10
2
I 2010
II 2011
III 011
2
IV 011
20
I 2 11
II 2012
III 012
2
IV 012
20
I 2 12
II 2013
III 013
20
13

Limited liability
partnerships
63%

Limited liability partnerships


Natural persons

Joint-stock companies
Registered partnerships

Fig. 2.5. The percentages (left panel) and numbers (right panel) ofbankruptcy proceedings by organizational and legal forms
inPoland from thefirst quarter of2007 tothe third quarter of2013
Source: own calculation based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek, Raport z oceny Raporty 112, op.cit.; Oglnopolskie Informatory Upadociowe, Raporty COFACE.

A strong upward trend inthe number ofannounced bankruptcy proceedings was observed inall areas ofbusiness activity
(manufacturing, trade and services). Theshapes ofrespective curves are different inservices than inmanufacturing and production
with changes intheir values indicating a deteriorating situation inservices and a slight improvement intrade and manufacturing.
Thenumber ofbankruptcies inservices was undoubtedly influenced by thedifficult situation inthe construction sector, where
thenumber ofbankruptcy proceedings constituted 53% oftotal proceedings involving service firms. A retrospective analysis
ofmanufacturing and trade indicates two periods ofgrowing numbers ofbankruptcy proceedings: 2009 and 2013. In contrast
toservices, therate ofbankruptcies decreases intrade and manufacturing. OU curves forindividual types ofbusiness are similar
tothe shape ofthe curve representing theabsolute number ofinsolvency proceedings, which indicates a strong upward trend
and high intensity, especially inthe services sector. Unquestionably, theincrease inOU inservices was influenced by thedifficult
situation inthe construction industry. Theanalysis ofthe data leads tothe conclusion that services was thesector characterized
by thehighest average OU levels while thehigh value ofthe standard deviation indicated a high level ofintensity ofchange. Trade
has thelowest average OU levels and simultaneously thelowest intensity ofbankruptcy processes.

44

Fig. 2.6. The percentage ofbankruptcies (left panel), its mean value and standard deviation (right panel) by type ofbusiness inPoland
from thefirst quarter of2007 tothe third quarter of2013
Source: own calculation based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek, Raport z oceny Raporty 112, op.cit.; Oglnopolskie Informatory Upadociowe, Raporty COFACE.

0
tu
rin
uf
ac
an

Services

Trade

Manufacturing

es

0.00

vic

Se
r

1.00

Tra
de

I2
0
II 2 07
0
III 07
2
IV 007
20
I 2 07
0
II 2 08
III 008
2
IV 008
20
I 2 08
0
II 2 09
III 009
2
IV 009
20
I 2 09
0
II 2 10
0
III 10
20
IV 10
20
I 2 10
0
II 2 11
0
III 11
20
IV 11
20
I 2 11
0
II 2 12
0
III 12
20
IV 12
20
I 2 12
0
II 2 13
III 013
20
13

2.00

The intensity ofbankruptcies inselected areas ofbusiness activity (BBU) combined with thechanges intheir environment leads
tothe conclusion that anabove-average (i.e. greater than 1) concentration ofthe number ofbankruptcy proceedings occurred
inservices (which was also thecase with theOU). In manufacturing, theBBU was lower with a clear downward trend. Theshape
ofcurves representing theBBU formanufacturing and services differed with respect toboth theintensity ofthe process and its
direction. Theaverage BBU values indicate thehighest levels and maximum variability inservices. A high level ofconcentration
ofbankruptcies (greater than 1) was also observed inmanufacturing.

Fig. 2.7. The Industry bankruptcy barometer (left panel), its mean value and standard deviation (right panel) by type ofbusiness
inPoland from thefirst quarter of2007 tothe third quarter of2013
Source: own calculation based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek, Raport z oceny Raporty 112, op.cit.; Oglnopolskie Informatory Upadociowe, Raporty COFACE.

The highest concentration ofbankruptcy proceedings was noted inMazowieckie, lskie, Dolnolskie and Zachodniopomorskie
Regions (voivodships) where 56% ofall proceedings were conducted, which appears tobe related tothe numbers ofbusiness
entities operating inthose regions. Theregion least affected by bankruptcies was Opolskie Voivodship with 13 times
fewer cases than Mazowieckie Voivodship. OU values inindividual regions indicate that theprocess was themost intense
inZachodniopomorskie and Dolnolskie Voivodships, which isconfirmed by a threefold increase inthe measure since 2007. Only
Lubelskie Voivodship was characterized by a decreased intensity ofbankruptcies. Theaverage OU and its standard deviation show
thehighest values and their variability inZachodniopomorskie and Dolnolskie Voivodships.
The highest concentration ofbankruptcies was not recorded inMazowieckie Voivodship (as might have been expected given its
highest number ofactive bankruptcy proceedings), but inZachodniopomorskie Voivodship. In seven regions theconcentration
ofbankruptcies was above average. Thegeographical distribution ofregions with above-average RBU values centres

45

around thenorth-eastern areas ofPoland. Above-average RBU values since 2007, aswas thecase with theOU, were noted
inZachodniopomorskie and Dolnolskie Voivodships.
400

1000
900
800

300

700
600

200

500
400
300

100

200
100
ec
k ie
l
ski
ch do
e
od lno
nio l
po ski
w i mo e
elk rsk
op ie
ol
ku
m
jaw
ao skie
sko po
-po lski
mo e
rs
lub kie
els
ki
d e
wa
rm
zk i
p
is om e
ko
o
-m rski
p o azur e
s
dk
arp kie
ac
k ie
l
w ubus
it
k ie
ok
rz y
s
p o k ie
dla
s
op kie
ols
k ie

rm
i

wa

za

za

ch
od
nio
p

zo
wi
ma

om
o
d
s olno rskie
ko
-m lsk
az ie
u
ku
jaw mazo rski
sko wi e
-po eck
mo ie
rs
lub kie
els
ki
l e
po skie
ma dlas
w op kie
it ols
ki
o
wi k r z y e
elk sk
op ie
ol
lub skie
us
ki
po d e
d k zk
arp ie
ac
op kie
ols
k
po
mo ie
rsk
ie

Fig. 2.8. The number ofdeclared bankruptcies (left panel), theaverage OU value and its standard deviation (right panel) by
voivodship inPoland from thefirst quarter of2007 tothe third quarter of2013
Source: own calculation based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek, Raport z oceny Raporty 112, op.cit.; Oglnopolskie Informatory Upadociowe, Raporty COFACE.

ie
ski
e
ok
rz y
ski
e
lub
us
po
kie
dk
arp
ma ackie
op
ols
kie
po
dla
wi
elk skie
op
ols
kie
op
ols
kie
d
zki
po
e
mo
rsk
ie
l

it

rsk

mo

kie
sko

-po

kie

iec

els

zo
w

lub

ma

jaw

below 1
above 1

ku

kie
ski
e
l

lno

urs
az

-m

do

ko
is

wa
rm

za
ch

od

nio

po

mo

rsk

ie

Fig. 2.9. Themean RBU value with its standard deviation (left panel) and its geographical distribution (right panel) by voivodship
inPoland from thefirst quarter of2007 tothe third quarter of2013
Source: own calculation based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek, Raport z oceny Raporty 112, op.cit.; Oglnopolskie Informatory Upadociowe, Raporty COFACE.

Information regarding thescale ofbankruptcies inPoland incomparison with other European countries isdifficult toobtain,
isfragmentary and characterized by long delays. Thestatistics are also virtually incomparable. In thegroup ofcountries analysed
by Mazurek (2013), thebankruptcy rate inPoland in2010 (per 10,000 ofactive firms) was thelowest and totalled 2. For thesake
ofcomparison, atthe same time itexceeded 200 inAustria and inHungary. Considering thedata for20042010, thestudy only
warrants theconclusion that thePolish statistics are comparable tothose ofthe Czech Republic and Spain. Theproblem does not
arise from themethods ofprocessing statistical data but from considerable differences inbankruptcy laws invarious countries
and insocial attitudes towards bankruptcy on a transnational scale. A genuine solution inthis respect may consist inreplacing
bankruptcy statistics with measures ofbankruptcy risk. In this context, itshould be emphasized that thedesigned EWS can be fairly
easily adapted toobtaining data about such a risk tofacilitate comparisons on a transnational scale. Theeasiest solution, although
not devoid ofcertain disadvantages,14 may rely on theapplication ofproduction functions (discussed inthis study and developed
aspart ofthe project) tointernational microeconomic data available insuch databases asAMADEUS.

14

Studies ofAltmans classic models show their acceptable invariance intime and space, yet thequestion arises towhat extent nowadays given
such phenomena ase.g. globalization orwidespread financial crises such solutions can still remain timeless and universal.

46

Piotr Adam Boguszewski

Chapter 3
DETERMINANTS OF BANKRUPTCY AN OVERVIEW OF APPROACHES
3.1. Introduction
The serious and diverse effects ofbankruptcy (see Chapter 4) prompted thenatural need toseek instruments that might enable
people topredict such events. Tools tothat end have been constructed forat least several decades. Thepurpose ofthis chapter
is toprovide a brief overview ofthe solutions commonly applied inthis area. Due tothelong history ofthe search and huge
numbers ofpublications, here we can only highlight some problems inthis respect and selected possible ways ofsolving them.
For a more in-depth treatment ofthe issues, theReaders are kindly directed totheReferences section at theend ofthis chapter.
Theselection ofapproaches discussed here was guided by theneed toapproximate thelogic and necessary intuitions contained
inmodel solutions described inmore detail inChapters 10 and 11, and adopted asthebasis fortheRapid Response Instrument
(RRI). Since theset ofpotential tools forforecasting thethreat ofbankruptcy is quite extensive, somewhat more attention was
devoted tothemain premises behind theselection ofthe most appropriate solutions from theperspective ofthe objectives and
characteristics ofthe RRI inquestion.
This chapter begins with a discussion ofthe most common classes ofmodels forpredicting bankruptcies presented intheliterature.
Two ofthem are characterised here inmore detail: thediscriminatory models and binary regressions. This is done due tothefact
that, first, these two approaches dominate research on predicting bankruptcy, and second, binary regressions constitute themain
tool used inChapter 10. Next, theissue ofthe selection ofexplanatory variables used inthese models is discussed insome detail,
indicating, among other things, theimportance ofdata used tofeed theindicators, ideally validated by large-scale international
research.
Then we move on totheissue ofthe synergic effects ofbankruptcy and its impact on aggregated measures ofthis risk fora given
industrial sector ortheeconomy asa whole. Based on a simple model we show that thecorrelation ofbankruptcy risk may
increase severalfold thelikelihood ofa negative scenario ofwidespread bankruptcies. Theexistence ofthis effect is an additional
motive fortheconstruction ofRRI-type systems since warnings generated by them may serve tointerrupt such a chain ofevents.
Further we show that there are a number ofnon-diversifiable reasons forthecommon threat ofbankruptcy, including therole
ofcertain macroeconomic variables. These observations, among other things, entail theneed tointegrate models combining
bankruptcy with themacro state ofthe economy insuch an RRI system. We discuss several types ofsuch models and indicate
thecriteria fortheselection ofthose deemed most compatible with RRI needs. We also include a brief review ofboth thepotential
explanatory and explained variables insuch models. Thelast issue is particularly complex, and we feel that ithas gone completely
underappreciated intheliterature, which is partly due totheproblems associated with measuring thescale ofbankruptcies
intheeconomy described inmore detail inChapter 2.

3.2. Fundamental aspects ofmicroeconomic modelling ofbankruptcy


Modern microeconomic research on predicting bankruptcies started with theworks by Beaver and Altman (cf. e.g. Altman 1968),
although attempts toapply ratio analysis tothestudy ofbankruptcies had been made before (cf. e.g. Horrigan, 1968). These
approaches were based mainly on discriminant methods (Multiple Discriminant Analysis, MDA). Thenext strand ofresearch was
initiated by Ohlson and Zmijewski (cf. Ohlson 1980, Zmijewski 1984), inwhich Tobit and logit models were used. It soon turned
out that itwas a very useful method foranalysing thegroup ofissues inquestion. Numerous studies involving theuse ofthis
tool also revealed its weaknesses. It can be primarily criticized forbeing based on observations that do not reflect theevolution
path ofthe object analysed. At thesame time, these models were dominated by indicator-type variables based on thefinancial
statements ofcompanies. Potential predictors ofbankruptcy, such asmarket characteristics, were rarely taken into account (cf.
e.g. Shumway 2001). An answer tothis criticism was provided by theso-called survival models, which examined theprobability
ofsurvival (or death) ofan enterprise until a certain point intime (cf. e.g. Manjn-Antoln and Arauzo-Carod 2008). They employed
both approaches based on modelling a single cause ofdeath (single-hazard models) and a number ofsuch causes (competing risks
models, starting with theclassic survival models proposed by Cox). In the1990s neural networks began tobe more extensively used
formodelling bankruptcy. Their attractiveness consisted, among other things, inthefact that they made itpossible ina relatively
simple way totake into account non-linearity inpredicting bankruptcy inmultilayer structures, (cf. e.g. Zhang, Hu et al. 1999).
In summary, currently intheliterature there are at least six distinct kinds ofmodels used forthemicroeconomic forecasting
ofbankruptcy:

47

1.
2.
3.
4.
5.
6.

Linear and nonlinear discriminant models,


Binary regression models,
(Dynamic) survival models,
Option models ofbankruptcy risk (distance todefault),
Neural networks,
Hybrid approaches.

An in-depth discussion ofthese approaches exceeds thescope ofthe present publication. More detailed treatment ofthe issues
can be found e.g. inAltman and Narayanan (1997), orina very good and comprehensive overview by Gruszczyski (2012),
especially Chapter 4 which offers valuable insights based on thePolish context. We should, however, take note ofseveral issues
important especially inthecontext ofthe applicability ofthese toEWS design. First, review papers tend tosuggest a certain
equivalence ofthese solutions inthat they usually yield fairly similar results; moreover, there is insufficient evidence toaccord
unquestionable preference toany ofthose considered interms ofquality ofthe forecasts generated by them (cf. e.g. Thevnin 2003,
Hall 2002). Undeniably, there are situations inwhich some models turn out tobe more useful than others. Each ofthese tools also
has its natural limitations, forexample, traditional discriminant approaches may not be optimal when explanatory variables are
not normally distributed (or even log-normal). In thecase offinancial indicators this is thedominant situation (cf. e.g. Karels and
Prakash 1987). Second, from theperspective ofEWS design two features ofsuch models offer a significant advantage, namely
their relative simplicity and canonicity. What we mean by thelatter is theexistence ofa sufficiently large body ofwork, research
and analyses conducted over at least a few dozen years based on such an approach. In this way we obtain an extremely valuable
additional source ofverification ofthe quality oftools at our disposal. Additionally, theuniversal application ofa certain class
ofmodels makes their results more intuitive and more easily acceptable by a wide range ofusers. Finally, thesuitability ofa specific
model foran EWS is influenced by thedata economics: theapproaches should use fairly easily accessible data both with respect
totime and thepossibilities ofobtaining them, aswell astheir cost. A violation ofthis principle has several effects, all ofthem
extremely unfavourable forEWS operation. First ofall, data access problems usually significantly shorten thesystems warning
horizon, because they often require intricate procedures foraccessing/obtaining information on which theentire system is based.
On theother hand, high costs ofobtaining data may deny some EWS beneficiaries access totheir results (if they are supplied
commercially) ortotheoriginal sources should they wish toverify them.
According tothis rationale, we have decided that two approaches are likely tobe themost attractive from theperspective
ofpracticability ofour EWS, namely thediscriminant and regression models (especially logit and probit). A review ofthe literature
proves that approximately 70% ofall studies related tobankruptcies were conducted using these two types ofmodels (cf. e.g.
Cheng Lim and Gan 2012). Below, we shall characterize both classes oftools ina very synthetic and practical way.
Both approaches operate ina fairly simple way,1 which is shown inFigure 3.1.

Fig. 3.1. A graphical interpretation ofthe probit model (left panel) and theprinciple oflinear discriminant (right panel)
Source: own study.

Nonetheless, their application associated with quite a few technical and methodological problems ofvarying degrees ofdifficulty. Their
discussion, however, exceeds thescope ofthis study.

48

As far asmodels ofprobit/logit binary variables are concerned, let us assume that thestate ofsubject i is characterized by
thefollowing unobserved variable:
y i = x i + + i

(3.1)

where x is a vector ofobservable variables affecting y i , and ei represents a random variable characterizing uncertainty inmodel
(3.1). Let us further assume that yi thealready observed behaviour ofsubject i covering two states, conventionally defined
as0and 1 results from thefollowing rule:
1 y i > l
y i =
0 y i l

(3.2)

It means that theentity finds itself instate 1 if thehidden variable takes a value greater than a certain critical level characterized
by parameter l (it can be assumed tobe equal to0), otherwise itfinds itself instate 0. Note that model (3.2) is very attractive from
thepoint ofview ofour objective, which is tomodel bankruptcy risk. If we interpret variable y i asthelevel ofeconomic difficulties
faced by a certain firm, thecondition means that if these difficulties exceed a certain critical value, thefirm will go bankrupt.
Obviously, thefact that thefirm inquestion has collapsed (or continues tooperate) can be observed. Conversely, thevariable
defined asthe level ofeconomic difficulties faced by a firm (or financial distress) cannot be observed directly, but each company
analyst will indicate observable economic variables x, which can quite well approximate theeconomic condition ofthe company.
Equation (3.1) states that theimpact ofthese variables onto themeasure ofthe firms condition is linear.
It follows from equations (3.1) and (3.2) that theconditional probability ofthe firm inquestion finding itself instate 1, aslong asits
financial parameters are described by vector x, is:
Pr ( y = 1| x ) = Pr ( y i > l | x )

(3.3)

hence:
Pr ( y = 1| x ) = Pr ( >[ x + ] | x )

(3.4)

Further application ofequation (3.4) obviously depends on our assumptions about theerror distribution ei . It should be
emphasized that since theleft side ofequation (3.1) is unobserved, theerror remains unknown and forthis reason itis necessary
tomake additional assumptions about its nature. In themodelling ofbankruptcy (and a number ofother binary variables), itis
generally assumed that this error is distributed normally with theexpected value 0 and unit variance, orithas a logistic distribution.
In thefirst case, from (3.4) we obtain theso-called probit model:
x +

Pr ( y = 1| x ) =

z 2
1
exp dz
2
2

(3.5)


In turn, forthelogistic distribution oferror with variance p 2 / 3 we obtain thefollowing relationship:
Pr ( y = 1| x ) =

exp ( x + )
1+ exp ( x + )

(3.6)

or its equivalent:
Pr ( y = 1| x ) =

1
1+ exp( x + )

(3.7a)

When comparing (3.5) and (3.6), itis clear, however, that thelatter formula is somewhat simpler. It is one ofthe reasons why logit
models tend tobe slightly more common inthis line ofresearch. In our EWS bankruptcy model, thefamily ofthe type (3.7) is used
(see Chapter 102). Both formulae also have one thing incommon, namely:

Pr ( y = 1| x ) = F ( x b )

(3.7b)

where F () is thecumulative distribution function ofvariable y i due to(3.1) being strongly correlated with thedistribution
ofvariable ei . Let us remember that inour examples these are normal and logistic distributions.
Equation (3.5) has a simple graphical interpretation (see Figure 3.1, left panel). According toformula (3.1), if thevalue offactor x
increases from x1 tox2, thedistribution ofthe unobservable variable y i also shifts totheright (if theerror has a normal distribution,
2

Complete with more references devoted totheconstruction and estimation ofthis class ofmodels.

49

thedistribution ofthe unobservable variable is also normal). Probabilities (3.3) fordifferent levels offactor x are represented
by areas P1 and P2 equal tothose lying above thecritical value of l under thecurves showing thedistribution density ofthe
unobservable variable y i . It is clear that asthevalues factor x (which characterize thesituation ofentity i) increase, there is an
increasing probability ofoccurrence ofevent 1 (P2 > P1).
Formula (3.6) may also be approached inother ways, which may be attractive forat least two reasons. First, itis not always easy
touse a research paradigm based on unobservable variables. Some researchers refuse todo iton philosophical grounds. Second,
thanks toan alternative method ofderivation, equation (3.6) turns out tobe easier tointerpret. Let us suppose that we intend
tomodel thefollowing relationship:
(x)=

Pr ( y = 1| x )
Pr ( y = 1| x )
=
Pr ( y = 0 | x ) 1 Pr ( y = 1| x )

(3.8a)

Function ( x ) (odds ratio) expresses therelationship ofconditional probabilities ofevents 1 and 2, respectively, aslong
astheentity is characterized by a state vector x. It is easy tosee that thevalues offunction ( x ) belong toset < 0, ). Finding
thelogarithm ofthis function (to thebase e) we therefore obtain an expression can assume real values. Hence itmay be justified
touse thefollowing model:

ln ( x ) = x b

(3.8b)

Note that theformulae ofthe type (3.8b) cannot be used formodelling probability Pr(y = 1|x) itself. Theright side ofexpression (3.8b)
can infact take any real value, whereas probability is always limited totheinterval <0, 1>. In this case, only using therelationship
ofthe type (3.7) is justified.
Mutual correspondence ofmodels (3.6) and (3.8) can be proved, which entails thefollowing natural interpretation ofcoefficients
. Equation (3.8) can be rewritten as:
n

( x ) = exp ( x b ) =

e x b

i i

i =1

(3.9)

hence:
( x1, x 2 , x 3 ....., x i + 1,....... x n )
( x1, x 2 , x 3 ....., x i ,....... x n )

=e

bi

(3.10)


Coefficient
i indicates therefore how theodds ratio changes if thevalue ofvariable x has increased by one unit, other things being
equal (the values ofother variables remain constant).
The basic ideas ofthe discriminant approach which underlie thecanonical bankruptcy prediction model proposed by Altman can
also be presented inquite a simple way. Let us imagine that we have two sets S1 and S2 on a plane and are looking fora realvalued linear function that will effectively separate them (see Figure 31). Let us further assume that itis a projection of:

y = vT x

v, x R n

(3.11)

In thefigure, we can see that thetask can be solved with varying degrees ofsuccess. In thecase ofprojection onto axis X1, both
images ofsets S1 and S2 have one part incommon (the red square). We can obtain a better result by projecting them onto axis
X2 theimages are completely disjoint. We must note, however, that even better results can be obtained by projecting onto axis
Y: not only are theimages disjoint, but their centres are more distant from each other than those projected onto axis X1 (D2 >D1).
Obviously, itis not always possible toobtain separate images. A good separation mentioned above means only that thecentres
ofthe projections are asdistant from each other aspossible, whereas theprojections themselves are asfocused aspossible around
those centres. Mathematically, itcan be expressed interms ofsearching fora vector v such that will maximize thevalue ofthe
expression:
G ( ) =

2
1 2

r12 + r22

(3.12)

1
1
1
where thecentres ofsets S1 and S2 are represented by , =theprojections
,2
= themeasures
v T x = v T i
sets by i and
x i = 1ofthese
y = N ofdispersion
i
Ni xSi
Ni xS

x
S
i
i
i
fortheprojections are defined asfollows:

i =

1
Ni

xSi

i = 1, 2

(3.13)

50

i =

1
Ni

y = N vT x = vT

xSi

i xSi

ri =

(3.14)

( y )

(3.15)

ySi

In order todetermine themaximum ofthe function G (v ) it is, therefore, sufficient tofind thefunctional relationship ofprojection
centres and their measures ofdispersion from vector v. This can be done fairly easily using thefollowing relationships:
S

( 1 2 )

= vT 1 vT 2

BC


T
T
= v ( 1 2 ) ( 1 2 ) v = vT SBC v

(3.16)

and:
ri2 =

( vT x vT )

xSi

vT ( x i ) ( x i )

i
v = vT SWC
v

xSi

i = 1, 2

(3.17)

where:
i
SWC
=

( x i ) ( x i )

xSi

(3.18)

Thus, by introducing thedesignation:



we obtain thefollowing:

1
2
SWC = SWC
+ SWC

(3.19)

r12 + r22 = vT SWC v

(3.20)

By combining equations (3.12), (3.16) and (3.20), we can determine theoptimal vector v asa solution totheproblem:

vT SBC v
Max
G (v ) = T
v SWC v
v

(3.21)

where thematrices SBC and SWC are represented by formulae (3.16) and (3.20), respectively. Linear discriminant analysis (LDA)
is based on theabove concept. As was already mentioned, however, itrequires thefulfilment ofquite restrictive assumptions
regarding thedistribution ofobservations and characteristics ofmatrices SBC and SWC. Good communicability ofresults ofthis
technique3 makes ita popular tool forpredicting bankruptcy risk.
Two very important considerations apply toboth approaches and both have a significant impact on theselection ofpractical
solutions used inRRI system design (see Chapter 10). These models turn out tobe sensitive both totheperiod inwhich their
estimation is made and tosome characteristics ofthe sample itself, such asthebusiness sector (cf. e.g. Grice and Dugan 2001).
Apart from selecting thetype ofmodel tostudy therisk ofbankruptcy, another extremely important part ofthe procedure involves
thechoice ofexplanatory variables. It is obvious that this choice is tosome extent dictated by thetype ofthe model itself. There
are, however, more general reasons. First ofall, inthelight ofthe concept ofbankruptcy presented inChapter 1, thesearch forits
crucial advance symptoms should naturally focus on such financial characteristics ofa given enterprise that enable us toanalyse
theprocess ofdepletion ofits resources under different assumptions regarding their possible replenishment from external sources.4
The justification forsuch an approach is straightforward and is based on theclassic model proposed by Wilcox (cf. Wilcox 1970).
If we assume that a firm may find itself instates S0, S1, Sn and thetransitions between these states comply with thefollowing
Markov process:

It is enough toprovide theanalytical form ofthe linear discriminant function (i.e. thedefinitions ofvariables and coefficients) aswell asits
critical values. Likewise, themodel is not only easy tounderstand but also touse inself-assessing therisk ofbankruptcy ofa given firm.
In thelight ofremarks presented inChapter 1, bankruptcy is not an objective state inthesense that, intheevent ofprovision ofextra capital
toa firm at risk, changes intheway itis managed etc., virtually any firm can be kept alive foralmost any length oftime. In centrally planned
economies numerous permanently unprofitable enterprises operated inthis way forover half a century.

51

P (S0 | S0 ) = 1

P (Si | Si1) = p

P (Si1 | Si ) = q = 1 p

P (Si | S j ) = 0 when i j > 1

(3.22)

where P (Si|Sj) denotes theprobability oftransition from state j tostate i (as long astheentity was instate j at time t1). Subsequent
states ofthe firm inquestion can be interpreted assuccessive stages ofits development, whereas state S0 asits ultimate collapse.
Note that itis an absorbing state, inother words, once thefirm has found itself init, there is no exit. The probability p is a measure
ofthe chance thefirms development, whereas q that ofits involution (i.e. return toan earlier stage ofdevelopment). Using such
a model itcan be shown that theprobability ofbankruptcy ofa firm inthestate Sk is:
1 when p q = 1 p
P (S0 (t) | Sk (t w )) =
(3.23)
(q / p)k


Thus, if thecurrent value offirm equals W and assuming that each year itcan increase thevalue by with probability p orlose

thesame value with probability q, and if p>q, then theprobability ofbankruptcy is:
W /s
P (S0 (t) | Sk (t w )) = (q / p)


The expression W/ can be interpreted asthenumber ofsuccessive failures that exhaust all thefirms assets.

(3.24)

In thecited work by Wilcox itis assumed that:


W = l1A l2L

(3.25)

i.e. thecompany has a buffer ofliquidity dependent on thecapacity l1 toliquidate assets A and postpone l2 its liabilities L. Let us
note that theparameters l1 and l2 may depend on thecharacteristics ofthe company and possibly on theoverall macroeconomic
situation. In theevent ofa financial crisis we may assume that l2 = 0 (the repayment ofliabilities cannot be postponed), while l1
may correspond totheproportion ofcash inthefirms assets. On theother hand, inthecase ofaverage business economic climate
orinmedium-size firms, W may be equal thenet value ofits working assets.
If thechanges inthefirms value approximate random walk, theaverage increase inthevariable per unit oftime is (p q)s. When
observing thefinancial situation ofthe firm, change equals A ROA z, where ROA is thereturn on thefirms assets, whereas
zrepresents thepart ofthe (net) profit which is neither paid out intheform ofdividends nor reinvested inilliquid assets. On
thebasis ofthis information, we can calculate theq/p ratio:

A ROA
1
( p q ) = A ROA

q/ p=

A ROA
p + q = 1
1+

(3.26)

It can be easily shown that if theformulae (3.24) (3.26) hold, then ina certain area ofthe analysed variables we can use
thefollowing approximation:

P(S0 ) 1 2

A2 ROA
2

(1 2 )

(3.27)

where x is a measure offinancial leverage, and s2 an estimator ofthe variance ofliquid funds flowing into thecompany.
In view offormula (3.27), theprobability ofbankruptcy is explained by several kinds offinancial indicators: thesize ofthe firm
inquestion, its profitability, liquidity, leverage and ability toretain profits. It should be emphasized that thenature ofrelationship
(3.27) constitutes thestarting point forthedesign ofexplanatory variables fora large number ofif not formost models
ofbankruptcy. In specific applications thechoice depends on theneed totake into account a number ofother factors, including,
among others: theforecast horizon, industry-specific features and availability ofdata. A detailed discussion ofthese issues exceeds
thescope ofthis study. More information on indicators used insuch analyses and their popularity can be found e.g. inHossari and
Rahman (2005). It should be emphasized, though, that theprocedures ofselecting theexplanatory variables used bankruptcy risk
prediction models inRRI take these experiences into account, and themodels are based on financial indices.
Our discussion ofissues related totheselection ofvariables concludes with a remark concerning theselection ofthe dependent
variables. It is worth remembering that theconcept ofbankruptcy may be defined invarious ways. A review of84 empirical
studies conducted in19682004 and devoted tobankruptcy modelling shows that they were dominated by thelegal definition
ofbankruptcy (cf. (Hossari 2007). Such an approach was also adopted inChapter 10.

52

3.3. Synergy inbankruptcy proceedings and macroeconomic approaches


Formula (3.27) indicates that thelikelihood ofbankruptcy may be significantly influenced by macro- ormesoeconomic factors.
After all, coefficients l1 and l2 reflect, among other things, a phenomenon which we might call pressure on a firms liquidity. If
thesituation ofthe industry inwhich thefirm operates is difficult, thecreditors ofthat particular enterprise may insist on quick
repayment ofliabilities. In this case, l1 is small (e.g. similar totheproportion ofcash and demand deposits intotal assets), while l2
relatively large (and may include all current liabilities ofthe firm inquestion). In times offinancial crisis ora deep recession l2 can
be even greater, because itmay comprise all thefirms liabilities.
Financial leverage may work ina similar way (cf. e.g. Hol, Westgaard et al. 2002). In periods ofincreased financial instability
ofthe economy, increased inflation etc. raising external capital may be more difficult foran extended period oftime. In addition,
higher interest rates on loans being repaid may significantly reduce theROA, increase thevolatility of s2 , which, according
toequation (3.27) will clearly lead toan increased bankruptcy risk. Thus we see that macroeconomic variables may significantly
affect bankruptcy risk at thelevel ofa single company.
It is easy toshow (cf. Zhou 2001) that thelinks between firms may significantly increase therisk ofbankruptcy intheeconomy.
Let us assume that therandom variables with theBernoulli distribution ofB1 and B2 describe theprobability ofcollapse offirms
1 and 2 by time t:

1 if bankrupt by t
Bi (t ) =
i = 1, 2
0 if no

(3.28)

If thebankruptcies ofthe two companies are independent, then:


P (B1 = 1 B2 = 1) = P (B1 = 1) P (B2 = 1)

(3.29)

Let us, however, consider how such a probability might change if both events were not independent (which is thepredominant
situation, if only forreasons described intheintroduction tothis section). Let us define thecorrelation coefficient inthefollowing
way:
CORR (B1, B2 )

E (B1B2 ) E (B1)E (B2 )


VAR (B1)VAR (B2 )

(3.30)

We must remember that since thevariables B1 and B2 follow theBernoulli distribution, then B1 B2 does well, and:

E (Bi ) = P (Bi = 1)

VAR (Bi ) = P (Bi = 1) 1 P (Bi = 1)

i = 1, 2

(3.31)

If we assume that thebankruptcies ofboth firms are interdependent, then from (3.30) and (3.31) itfollows that.
P(B1 = 1 B2 = 1) = E (B1B1) = E (B1)E (B2 ) + CORR (B1, B2 ) VAR (B1)VAR (B2 ) =

= P (B1 = 1)P (B2 = 1) + CORR (B1, B2 ) P (B1 = 1) * 1 P (B1 = 1) P (B2 = 1) 1 P (B2 = 1)

(3.32)

Hence, if we compare theprobabilities resulting from theformulae (3.29) independent events, and (3.32) dependent events,
we obtain:

1 P (B1 = 1) 1 P (B2 = 1)
P(B1 = 1 B2 = 1)
1
1

= 1+ CORR (B , B )
= 1+ CORR (B1, B2 )

1 2
P (B1 = 1)P (B2 = 1)
ODDS (B1) ODDS (B2 )
P (B1 = 1 B2 = 1)

(3.33)

where ODDS(Bi) represents therelative odds ratio ofbankruptcy offirm i.


Formula (3.33) leads toseveral important conclusions. First, due tothenon-zero correlation therisk ofsimultaneous bankruptcies
may increase severalfold incomparison with a situation without such a correlation. If, forexample, therisk ofbankruptcy ofthe
first firm is 0.05 and that ofthe other 0.01, and thecorrelation equals 0.3, then formula (3.33) demonstrates that thesituation
leads toan almost 15-fold increase intherisk ofsimultaneous bankruptcies ofboth firms. Second, insuch a context therole
ofEWS-type systems becomes particularly clear, especially if they permit taking pre-emptive action. It may involve both
solutions toreduce such a correlation coefficient and theprobabilities ofbankruptcy at thelevel ofindividual firms. In thelatter
case, theeffect ofthe correlation will still be strong, but theprobability itself (3.32) will be small. It should be emphasized that
thereduction inthecorrelation coefficient often requires systemic action at macro ormeso levels. Third, due tothepresence ofthe
phenomenon ofsynergy, theanalysis ofthe risk ofbankruptcy, including one inthis RRI project, must take into account not only

53

themicroeconomic variables discussed intheprevious chapter, but also themacroeconomic environment and thesituation at
thelevel ofindividual branches ofindustry.
The impact ofGDP changes on thedynamics ofbankruptcies inPoland is clearly visible (see Figure 3.2), but italso points
tothecomplexity ofthe relationship. Thechart shows a decline inthenumber ofbankruptcies when GDP growth approached
(or exceeded) 4% y/y. At thesame time, slower economic growth (without a recession inPoland) below 3% led tomarkedly
accelerated bankruptcy processes. It is thus clear that examining therelationships between themacroeconomic situation and
thedegree ofbankruptcy risk is a difficult task and requires theapplication offormal tools.
Meeting theabove requirement consists primarily intheselection ofappropriate macroeconomic variables strongly associated
with aggregated measures ofbankruptcy risk. Since thedesign ofthe RRI was informed by theresults ofresearch on bankruptcy
inother economies, a valuable clue was provided by thefact that macroeconomic topics have appeared intheliterature fora fairly
long time. Furthermore, itturned out that themost important research directions forthesolutions adopted intheRRI include two
main planes ofsuch interactions. Historically speaking, theearliest studies concerned therelationships between thedifferently
defined bankruptcy rates aggregated forthewhole economy and selected macroeconomic variables. In this context itis worth
mentioning theclassic work by Altma, which had numerous followers5 (cf. Altman 1983a), devoted tothestudy ofthe relationship
between thebusiness cycle and thebankruptcy rate. We wish toemphasize that inour approach toRRI design therelationships
between thebankruptcies and theeconomic cycle are examined with particular attention.
Analyses conducted by representatives ofresearch trend comprised not only thecyclical position ofthe economy, but also
themacroeconomic variables such asinterest rates, unemployment, wages and exchange rates (cf. e.g. Dewaelheyns and Van
Hulle 2007). They also demonstrated theimportance ofunanticipated changes ina number ofmacroeconomic variables (cf. e.g.
Hunter and Isachenkova 2006).

Fig. 3.2. GDP growth (y/y) and thedynamics ofbankruptcy (y/y) inPoland in19982013
Source: own study based on data from GUS and COFACE.

The other current ofresearch dealing with therelationships between theintensity ofbankruptcies and macroeconomic factors
incorporates analyses conducted at theinstitutional level, including regulations governing thebankruptcy procedures themselves.
When focussing on theimpact ofthe cyclical changes, itis worth remembering that thediscussion revolves around theissue
oftheir endo- orexogenous relationships with bankruptcies. Authors ofthe work cited above (Suarez and Sussman 2007) mention
that certain solutions inthearea ofbankruptcy proceedings may themselves affect thefluctuations ineconomic activity. Theissue
is analysed even more broadly by Peri and Rachedi (2013).
Most ofthe above approaches tomodelling theimpact ofmacroeconomic variables on bankruptcy rates are based on one ofthe
following solutions. Thesimplest approach is represented by thefollowing formula:

A more extensive review ofthe literature may be found ine.g. (Hol 2006).

54

BRt =

i1X t1i + +

i =1

k k
i t i

+ t

i =1

(3.34)

where BR is the(variously defined, cf. section 2.3) aggregated measure ofbankruptcy intheeconomy; whereas X 1.... XN
themacroeconomic variables under consideration. In practice, equation (3.34) is often estimated astheAlmon-type model, which
yields good results insome economies (cf. Dewaelheyns and Van Hulle 2007).
More general approaches are based on VAR models:
t = B0 +

B
i

t i

i =1

+ et

(3.35)

where Bi are coefficient matrices; et theerror vector; while vector t is defined as:

BRt

1
Xt
t =


X tN

(3.36)

Non-linear generalizations ofequation (3.35) are also used (cf. e.g. (Drehmann, Patton et al. 2005):

t = (t , t 1,t p ) + et

(3.37)

Bearing inmind theperspective ofRRI design, thechoice ofthe most useful class ofmodels from those listed here requires taking
several criteria into account. First, theselection must reflect theavailability and other characteristics ofPolish data on bankruptcy.
Thequality ofsuch information is limited asis thelength ofthe time series (see Chapter 2). Second, an extremely important issue
related toRRI design is thegood communicability ofresults ofmodel analyses. In general, itmay be assumed that itdecreases
with thedegree ofmodel complexity, theuse ofnon-standard variables, theuse ofniche approaches etc. Third, themodel should
also meet other conditions supporting its credibility: be capable ofreplication (at least forprofessional economists dealing with
similar issues), contain not too many assumptions about variables etc. Fourth, such a model should allow a comprehensive
evaluation ofpossible impacts ofthe selected macroeconomic policy scenarios if theRRI signalled theneed totake such action.
Fifth, themodel should be relatively simple when itcomes toits re-estimation, modification ofthe list ofvariables etc. Finally,
theprocess offeeding data into themodel, estimations, calculations and analyses ofresults cannot be overly time-consuming.
It can be said that theabove conditions are best met by a model ofthe type (3.35). TheVAR model used inthemacroeconomic
module meets theabove conditions (see Chapter 11).
The above-listed criteria also generate some concomitant costs. Among others, they fail totake into account thewell-documented
work on bankruptcy based on long time series ortheexistence ofseveral macroeconomic bankruptcy regimes which significantly
differ with respect tothemean bankruptcy rates intheeconomy (cf. Giesecke, Longstaff et al. 2011), thetransitions between
which are not continuous. Nevertheless, such a problem can be solved using alternative methodologies which include switchover
mechanisms (cf. e.g. Bhattacharjee, Higson et al. 2009, Das, Freed et al. 2006). Neither does theapproach inquestion take into
account thechanging systemic circumstances such asthose involving changes inbankruptcy legislation. In more general terms,
these formulae do not usually contain explicit qualitative variables which often play an important role inbankruptcy processes
(cf.e.g. Becchetti and Sierra 2003). It should be emphasized that theRRI project comprises procedures aimed toovercome
theeffects ofthe adopted methodology and its potential limitations. One ofthem is theso-called Interpretation Panel whose
responsibilities include theidentification, assessment and expert correction ofsuch situations.

55

Piotr Adam Boguszewski

Chapter 4
ECONOMIC AND SOCIAL CONSEQUENCES OF BANKRUPTCY
4.1. Introduction
Bankruptcy procedures constitute one ofthe most important selection mechanisms intheeconomy. As a result, they eliminate
permanently inefficient enterprise whose assets could be better used innew ways, whereas certain financially inefficient but
economically sound firms implement restructuring and recovery programmes allowing them topermanently regain acceptable
performance. In a well-designed bankruptcy system, such changes also benefit thefirms stakeholders. As a result, insuch an ideal
world bankruptcy proceedings do not entail particularly serious negative consequences, themost important task ofeconomic
policy being not todisturb them when they are doing their job; inconsequence, theimportance ofearly warning systems (EWS)
is quite limited. However, an in-depth analysis ofthe operation ofbankruptcy systems shows that theconsequences ofbankruptcy
contrary toprevalent opinions about their low costs shaped inthe1970s and 1980s can be considerable, and spread over
a wide area ofeconomic and social life. Moreover, a detailed generic analysis ofthese effects shows that EWSs are capable not only
ofanticipating such consequences but also ofrationalizing thepossible costs ofimperfections inherent inbankruptcy systems.
In particular, they may reduce thecosts oftype I and type II errors described here aswell asshorten thetime ofsuboptimally
adjourned decisions regarding bankruptcy and its form.
In previous chapters we synthetically presented thefunctioning ofthe bankruptcy system asa selection mechanism and
characterized some ofits limitations, which was followed by an overview ofthe direct and indirect costs ofbankruptcy.
Theliterature shows that thetotal bankruptcy costs may range from 10 to25% ofa given firms value. It was also noted that simple
aggregation ofmicroeconomic costs does not reflect thefull macroeconomic impact ofbankruptcy, including, among others,
its domino effect. An additional cause ofunderestimation ofthe phenomenon is thefact that inthelight ofresearch thelargest
companies (including non-financial ones) are usually protected from bankruptcy. As a result, therisk monitored by theEWS may
be even greater than that suggested by these estimates.
The existence ofseveral kinds ofbankruptcy costs and their substantial magnitude raises theproblem ofanalysis ofthe efficiency
ofsuch procedures. This chapter reviews some performance measures and, on their basis, provides an attempt toevaluate
theefficiency ofthe Polish system with that ofother European countries. Theresults show that Poland ranks average inthis respect.
At thesame time, itlags quite far behind thevery best. In general, itcan be said that theefficiency ofa countrys bankruptcy system
depends on several factors, including institutional and other structural characteristics. All inall, theoverall efficiency ofthe process
changes slowly, although improvements inthesystem should constitute a permanent objective, such measures are not likely
tolead toa rapid reduction ofthe EWS role asone ofthe tools contributing tothereduction ofcosts ofbankruptcy procedures.

4.2. Bankruptcy asa selection mechanism


In modern economies, thetermination offirms business activity is quite a common occurrence. It is estimated that ina large group
ofcountries 20 to40 % ofbusinesses fail tosurvive thefirst two years since their founding (Bartelsman, Haltiwanger et al. 2004).
According totheclassical views ofthe followers ofSchumpeterian tradition, bankruptcy provides theprimary selection mechanism
which results intheelimination ofeconomically weaker entities and simultaneously facilitates therelease oftheir productive
resources infavour ofmore efficient business use conducive toa higher efficiency ofthe system asa whole. Naturally, this opinion
raises some important questions. First, does such a mechanism actually work inthesense that a world without bankruptcies is less
effective than one that allows theselection mechanism tooperate? Second, is theselection mechanism itself effective inthesense
that itactually eliminates weaker entities? And third, is there really no alternative solution? The most interesting issue, however,
may be expressed by yet another question, which we shall address inthefollowing sections: is itthemost efficient mechanism
available, and if not what would a better one be like? All these problems are undeniably very important inthecontext ofthe
EWS design, because if thenatural mechanism performs well and there are no obviously better solutions, theobjective ofsuch
a system would be less than obvious. A full discussion ofthese issues significantly exceeds thescope ofthis study; hence we shall
synthetically address selected aspects ofthese problems.
As far astheissue ofhow a world without bankruptcies might operate is concerned, now itis quite difficult tofind valid current
experiences that would be helpful inthis respect. Since most countries have a bankruptcy law ofone sort ofanother, inprinciple,
we now have only two options at our disposal. First, itmay be helpful toobtain information about theoperation ofthe areas
excluded from bankruptcy law. Such cases exist ina number ofcountries and often include even entire sectors ofthe economy

57

(e.g. strategic ones, such asenergy production, mining and defence industries). Unfortunately, they constitute a thankless subject
ofresearch, since any knowledge ofthe actual economic situation ofthose entities is generally limited and not always objective.
Second, one may focus on analysing historical experiences related totheoperation ofeconomies intransition from thecentrally
planned model tothemarket one. This is only seemingly an easy task. Although thecentrally planned economy model has clearly
historically lost ground tomarket-based solutions, thepotential causes ofthe situation were so many that theselective bankruptcy
mechanism under market conditions was theonly one ofthem. In thecontext ofthis remark, very valuable knowledge can be
obtained e.g. from theexperience ofthe Chinese economy,1 inwhich thefirst bankruptcy ofa state-owned enterprise took place
in1986 with more comprehensive systemic solutions inthefield ofbankruptcy law were introduced in1988. There are a handful
ofworks examining theSchumpeterian effect ofbankruptcy using data from theChinese economy inboth periods. Their results
can be summarised asfollows (Parker 1995): inan economy without bankruptcy enterprises grow significantly faster intheinitial
phase oftheir development. Then theeffects ofdecreasing efficiency start toappear due tothelack ofselection mechanisms. In
sectors ofthe economy where selection mechanisms are allowed tooperate thereverse phenomenon can be observed, namely
slower growth intheearly stages ofdevelopment, but greater dynamics and efficiency when firms have matured. One ofthe
practical conclusions fortheEWS from this kind ofexperience may be that although such solutions should not inany case involve
attempts toabolish mechanisms eliminating theweaker actors ortocreate a model ofa world without bankruptcies, such systems
and a more active economic policy towards theareas with such risks may be desirable toinclude at certain stages ofdevelopment
ofthese segments.
Likewise, thequestion ofthe efficiency ofbankruptcy understood asthecapacity toeffectively eliminate theweakest actors is
quite complex. A detailed study involving non-financial companies conducted intheUnited States from 1979 to1983 suggests
a good efficiency ofthe bankruptcy process inAmerica at that time (Kahl 2001). Other studies also show that inthecase ofless
radical actions, without eliminating bankrupt entities but based on remedial and restructuring processes, bankruptcy law can
be regarded asa fairly effective way ofrestoring performance (Djankov, Hart et al. 2008). On theother hand, one may not fail
torecall that theterm too big tofail (TBTF syndrome) was first used intheUSA asearly asin thefirst half ofthe 1980s with respect
tofirms inthefinancial sector. Therecent global financial crisis revealed even more egregious cases ofTBTF indifferent countries
worldwide. This phenomenon also included firms from thenon-financial sector with a spectacular example being thebankruptcy
ofGeneral Motors on June 1, 2009. At first, itcould have even been considered tobe event contradicting theTBTF phenomenon
had itnot been forthefact that merely 40 days later thecompany was bought out by theU.S. Treasury Department. Naturally,
thephenomenon ofTBTF produces not only positive consequences such asavoiding thedomino effect, saving jobs, but also
various types ofcosts and other side effects (cf. e.g. Roe 2014). This does not alter thefact that thephenomenon under discussion
is always present intheeconomy, and inmany cases thenet balance ofTBTF policy, despite thecosts, is likely tobe positive. From
theEWS vantage point itis therefore important toensure that thesystem is not used asa tool forreinforcing thenegative effects
ofTBTF. On theother hand, theEWS can significantly reduce such effects. Theanticipatory and warning functionality ofthe system
allows e.g. fortheimplementation ofa series ofpre-emptive actions and defining scenarios and policies inrelation totherisk
ofbankruptcy oflarge companies. In theabsence ofsuch tools all decisions inthis area will have tobe made on an ad-hoc basis,
which certainly limits their rationality.
The third group ofquestions concerns potential mechanisms fortheelimination ofentities that fail tooptimally use themeans
ofproduction other than theclassical mechanism ofbankruptcy. Let us first note that all themodern bankruptcy systems,
inprinciple, include such solutions associated with differently designed non-liquidation procedures. In many cases, however,
creditors prefer liquidation. In this way, a natural additional research direction appears. It should be remembered that one ofthe
main reasons fortheinclusion oftraditional bankruptcy procedures inlegal systems is thethreat ofan ineffectual creditors race
fortheassets offirms infinancial distress, which would occur intheabsence ofthose solutions. Researchers may try todevelop
formulae forcontracts between debtors and creditors which would almost automatically facilitate thesolution ofthe problem
ofclaims without theabove-mentioned side effects. This often happens inpractice when parties reach an agreement outside
bankruptcy law2 asto how tosettle theoutstanding financial obligations. In theliterature, thepresent trend ofsystemic
considerations employs, among others, theprinciples and achievements ofcontract theory, examining solutions which are deemed
optimal from this point ofview, forexample works by Gennaioli and Rossi (2012). Moreover, itmust be admitted that formal
insolvency procedures are not matched equally well with theentire population ofenterprises, which applies especially totheSME
sector. In thecase offirms belonging tothelatter group, classic bankruptcy law is often too expensive and too complicated a way
1

Theoretically, similar insights could be gained from theexperience ofpost-socialist European economies, including Poland. It must be
remembered, though, that inthese cases changes occurred much faster, included a radical reconstruction ofthe political model and an
important role was played by theeffects ofintegration with thedeveloped Western countries. In such a situation, itis difficult toseparately
assess theimpact ofvarious factors.
This is another reason forthefact that bankruptcy statistics are notoriously inaccurate, beyond those described inChapter 2. Using medical
terminology itmay be compared toa silent heart attack. Although itis not recorded itany official statistics, itmeets theformal bankruptcy
criteria.

58

ofsolving theproblem ofclaims. Accordingly, inthis group we can see theemergence ofalternative methods ofprotecting
theinterests ofcreditors (cf. e.g. Garca-Posada and Mora-Sanguinetti 2013).
This very sketchy overview ofsome issues related totheoperation ofselection mechanisms based on bankruptcy procedures
proves their indispensability. It is clear, however, that alternatives can and insome areas even must exist. Those solutions generally
demonstrate a satisfactory performance. Still, theproper positioning ofthe role ofEWSs ina world where bankruptcy law is
universal and operates quite effectively, we must also consider their efficiency. This issue will be dealt with inthefollowing sections.

4.3. Direct and indirect costs ofbankruptcy a review


Until theearly 1980s, theliterature on thecosts ofbankruptcy was dominated by an approach based only on theconcept ofdirect
costs (Bhabra and Yao 2011). Such costs usually represent thetotal expenditures associated with thepreparation and conduct
ofthe insolvency proceedings. They normally include such items asrequisite court fees, costs ofexpert opinions and valuations
ofassets, costs oflegal services and accounting orcosts ofprotecting theproperty from devastation and misuse. Although various
studies yield heterogeneous results regarding themagnitude ofthese costs, most ofthem show3 that they add up toonly a few
percent ofthe companys assets (cf. e.g. Altman 1984, Thornburn 2000, Weiss, 1990). This may be considered asone ofthe reasons
forwhich theimportance ofEWS is often underestimated, inother words, if thepotential losses resulting from thematerialization
ofa risk are small, itis easy toquestion thedesirability ofdeveloping warning systems against it. Thesituation changes significantly
if we take into account thefull costs ofbankruptcy by analysing not only themicroeconomic aspects ofthe phenomenon but also
its macroeconomic effects. Thefollowing brief discussion follows up on this train ofthought.
The indirect costs ofbankruptcy are usually defined asfollows:

CIND = E (V | NB )V

(4.1)

where CIND represents theindirect costs ofbankruptcy; E (V|NB) theassessment ofthe conditional expected value ofthe firm if
unaffected by bankruptcy and (or) optimally managed, and V theactual value ofthe company. It is clear that thekey requirement
forthepractical application offormula (4.1) involves thedevelopment ofa method ofdetermining theconditional expected
value ofthe firm inquestion. Approaches will differ depending, among other things, on whether we wish toevaluate thecosts
from theex-ante orfrom theex-post perspective. In theformer case we need toanalyse thedisciplining impact ofthe model
ofa bankruptcy system i on thewillingness and motivation ofthe firms stakeholders, especially managers and employees, towork
more effectively forthegood ofthe firm and control theagency problem. It is worth noting that inthecase ofthese costs we focus
on relative costs compared with thebest bankruptcy law.4 As a result, inthis context formula (4.1) assumes thefollowing form:

CIND = V = E (V | j)V (i) (4.2)

where j represents theoptimal bankruptcy law and V(i) is thevalue ofthe firm inquestion inthecase ofapplication oflaw i. The
aggregated ex-ante cost (CEX-ANTE) relative totheideal solution fortheentire economy inwhich N firms operate is thus:
N

CEX ANTE ( j ,i) =

(E (V | j)V (i))
k

(4.3)

k =1

An estimation ofex-post costs can also be performed using different approaches, but themethod proposed by Altman (cf. Altman
1984) and a number ofits adaptations are considered standard. A disadvantage ofthis approach is that itcan only be applied
tofirms with a sufficiently long history; including a period offinancial stability.
Besides theproblem ofmeasuring indirect costs ofbankruptcy, an equally important issue involves theanalysis ofsources ofthese
costs including their macroeconomic effects, which is extremely important forthedesign ofEWS. It is a comprehensive issue,
but forthesake ofclarity ofpresentation we shall confine ourselves toits synthetic discussion (see Figure 4.1) noting that its
excellent and detailed description can be found e.g. inWhite (1996). However, at theoutset, itis worth emphasizing that inthis
case a significant portion ofthese costs results from theagency problem and constitutes theconsequence ofspecific stakeholder
behaviours, which are not always optimal forthefirm.
To illustrate themechanisms ofhow these costs accrue, let us assume that inits current form thefirm generates a revenue stream5
P1F, while line K represents thecost level. When we examine theshape ofthe two lines itbecomes clear that we are dealing
with an entity that is not only bound toenter a period offinancial distress (within a certain time horizon after time T1), but itis
also economically inefficient and should be liquidated. Let us further assume that itis possible tomake better use ofthe assets
3
4
5

A broader overview and more detailed information on thebreakdown ofthese costs are offered by e.g. (Branch 2002).
The question how todefine them still remains. Thenormative model ofbankruptcy is discussed somewhat more broadly inChapter 1.
Or, tobe more accurate, itrepresents tothefirms revenue forecast.

59

ofthe liquidated firm, which would afterwards produce thenet income AF (i.e. after thededuction ofnecessary expenses I) and,
forthesake ofsimplicity, let us further assume that after such a restructuring thecosts line would remain unchanged. Note that
based on such assumptions theoptimization ofthe firms value should prompt a decision about its restructuring even at time
T0. Theabsence ofappropriate action may be due toa number ofcauses such asthelack ofresources I, orthefact that such
a decision depends on thecapacity toforecast revenues and theaccuracy ofsuch forecasts. However, if theopportunity tomake
such a decision at time T0 is missed, thecosts due tothedelay equal thearea under theappropriate curves designated by points
ABC assuming that at time TBANKR thefirms bankruptcy was actually declared.

Fig. 4.1. Sources ofex-ante indirect costs ofbankruptcy a qualitative analysis


Source: own study based on: (White 1996).

When analysing thesituation shown inFigure 4.1 we must also note that theperspective ofsuch developments may generate
two other kinds ofcosts overinvestment orunderinvestment. At a certain point intime (which is difficult toidentify clearly within
thefigure, but itcan be assumed tocoincide with theformulation ofa reliable forecast P1F), when themanagement ofthe company
become aware ofsuch a profile ofcosts and revenues, they may decide tomake risky but profitable investments inorder todelay T1.
Thedelay may also be achieved by way ofreduction ofexisting and future investment programmes ordisinvestment inthefactors
ofproduction (sale ofassets, layoffs ofthe best qualified but also expensive staff ). Such activities often markedly improve thefirms
liquidity, but inthelong run they may cause e.g. a further decline ofline AF. It is clear that theaccrual ofthese two types ofcosts
may significantly depend on thespecific provisions ofbankruptcy law. Such costs are themore likely, themore likely itis that at
thetime ofdeclaration ofbankruptcy themanagement gets nothing. It also depends on theactual accountability ofmanagers
formaking such sub-optimal decisions (although such intentions are very difficult toprove).
Actually, itis often difficult todetermine a priori whether thecompany should be liquidated. For example, at time T2 itmay turn
out that curve P2F is going tobe theactual revenue path. Under such circumstances, theoptimal solution would be torestructure
thefirm. Nevertheless, since itis not known whether thefirm can actually regain its long-term efficiency, type I and II errors may be
committed intheprocess. In theformer case, theeconomically bankrupt firm will be treated asa viable business tobe restructured,
whereas inthelatter case a company that may recover sustainable performance will be liquidated. In thecase oftype I errors, if
a company is kept alive, thecost ofthe incorrect decision correspond tothearea designated by BCDE inFigure 4.1. On theother
hand, thecosts oftype II errors undoubtedly amount tothediscounted value ofthe difference between thefirms revenues and
expenses after T4. It is difficult topinpoint thearea ofthese costs beforehand asitdepends on, amongst other things, whether path
P2F has been observed before TBANKR. Thelocation ofpoint H is also important thefirm may regain economic efficiency within
a longer timeframe (after TLIKW). One more remark is inorder: if themarket runs smoothly, then type II error costs are not significant.
A liquidated firm may quickly find a new owner who within a relatively short time will restore its profitability.

60

The total costs ofbankruptcy obviously include direct and indirect costs. Their full breakdown can be found e.g. inthework cited
above (White 1996). It may be well worth pointing out that on theaggregated scale a crucial role is played by CEXANTE costs since
they relate toall firms operating ina given economy, while others affect only thegroup offirms6 intheprocess ofbankruptcy.
Theliterature offers different assessments oftotal costs ofbankruptcy, but they are usually estimated at 1025% ofthe firms value
(cf. e.g. Altman 1984, Andrade and Kaplan 1997). Quite often a universal measure ofthese costs is adopted intheform ofthe
duration ofbankruptcy proceedings (see e.g. Franks and Torous 1988).
The kinds ofbankruptcy costs described above, although they may be universal and widespread, should rather be construed
asmicroeconomic consequences. But itmakes sense, especially inthecontext ofEWS, topay attention totheserious
macroeconomic consequences that may result from theapplication ofcertain legal and institutional solutions inthearea of
bankruptcy. Theexistence ofsuch consequences appears tobe well documented, tomention only theclassical work by Djankov
and Hart (Djankov, Hart et al. 2008), which shows, among other things, that one ofthe primary catalysts ofpositive impact7 ofthe
degree ofdevelopment ofthe financial system on thereal economy are thesolutions provided by bankruptcy law. This point
ofview seems indeed natural if we note that credit market imperfections, certainly important fortheoperation ofindividual
enterprises, may toa considerable extent arise from theprovisions ofbankruptcy law, thedegree ofinvestor protection etc.
Consequently, if we consider theimportance ofbankruptcy law, itmay be surprising that its macroeconomic dimension is
relatively rarely mentioned intheliterature and research. An additional problem intheanalysis ofthis issue is that themeasures
ofmacroeconomic effects ofspecific bankruptcy policies may be skewed by thefact that, generally speaking, countries tend
toprotect larger companies from bankruptcies (cf. e.g. Bris, Welch et al. 2006). This may lead totheerroneous impression that
thegeneral consequences ofbankruptcies are not too significant intheeconomy asa whole. Themicroeconomic picture even
after certain aggregations may also be insufficient, because bankruptcies often cause various multiplier effects. A good example
is theconstruction sector whose problems affect virtually theentire economy themanufacturing offurniture, household
appliances, steel, financial services etc.
From theperspective ofEWS design, when surveying recent research results we should pay attention totwo fundamental
conclusions concerning therelationships between bankruptcy systems and themacro-economic sphere. First, effective operation
ofsuch systems largely depends on other elements oftheir regulatory environment, including thedegree ofinvestor protection
(Peri and Rachedi 2013). This observation leads toan important conclusion. As was mentioned inChapter 1, itis often argued
that improving bankruptcy systems may constitute a competitive solution tothedevelopment ofEWSs. Without questioning
theneed forsuch strategic actions, we must bear inmind thelimitations ofthis approach that apply inthelight ofthe above
observations and considerations presented inChapter 1. Improving thelegal systems, although necessary, is likely tobring gradual
effects inthelong run. Therefore, during such a period oftransition EWSs appear toconstitute a solution that has no alternatives.
Second, research results also suggest that EWSs may not have a viable alternative forat least one more reason. Some analysts think
that if themacroeconomic effects ofbankruptcies are serious, one ofthe ways tocushion them may be tosoften theprovisions
ofbankruptcy law. Theabove-mentioned studies indicate, however, that theoverall costs ofsuch a softening can be high and,
therefore, a better solution may include aid offered toindividual businesses (cf. e.g. Suarez and Sussman 2007).
In this context, itis worth paying attention toa specific kind ofmacroeconomic costs ofbankruptcy related tothesocial effects
ofthe process, which are notoriously difficult tomeasure. In modern economies, a very important role is played by small businesses.
These entities are often separated from households only by a very thin border since quite often they are simply family businesses.
Thebankruptcy ofsuch an entity may therefore have similar effects toan instance ofconsumer bankruptcy. Furthermore, thesame
effect may occur even inthecase ofinsolvent larger firms. If thelocal unemployment rate is high, both therank-and-file staff and
themanagement ofbankrupt firms often become insolvent orat least illiquid consumers. Thenegative impacts ofsuch a situation
may be very serious and far-reaching (cf. e.g. McIntyre 1989). Likewise, we should not forget that bankruptcy is still associated with
a sort ofsocial stigma and multiple attendant negative consequences (see COM 2007).
For these reasons, theEWS designed here makes extensive use ofthe concept ofthe bankruptcy costs function. Since its practical
implementation depends on obtaining relevant data, at thepresent stage we only apply theindicators based on themeasure
ofpotential job losses, i.e. thenumber ofemployees ina firm facing bankruptcy risk. This measure can be fairly easily adapted
toother categories that assess thepossible effects ofbankruptcy ofa given group offirms, such asadded value, exports etc.

6
7

Quite often, they do not comprise all ofthem since thefrequency oftype I ortype II errors does not add up to100%.
Although itmust be noted that thedirection ofcausality ofthis relationship is thesubject ofcontroversy; see e.g. (Levine 2005).

61

4.4. Theefficiency ofbankruptcy procedures


The efficiency ofbankruptcy procedures constitutes an important category not only from thepoint ofview ofthe objectives
ofbankruptcy law, but is also vital inthecontext ofEWSs tasks. If such procedures are ineffective, thegeneral objective ofthe
system is not likely tobe achieved (Gine and Love 2006), inwhich case theanalyses ofbankruptcy risk provided by theEWS play
a dual role: they are themain tool forthedetection ofsuch conditions and allow fortheimplementation ofpreventive measures.
Prevention orsufficiently rapid implementation ofinformal procedures tosolve theproblem ofoverdue claims is a better way
tosolve this issue than inefficient provisions ofbankruptcy law.
Despite thecrucial role played by theefficiency aspect ofbankruptcy law, there is no uniform concept ofsuch efficiency orits
universal measures (with thelatter toa certain extent deriving from theformer).
There are several approaches totheissue ofdefining theefficiency ofbankruptcy law. First ofall, its provisions may be effective
forpreventive (ex-ante) purposes. Provisions ofbankruptcy law are considered tobe effective if they affect theowners and
managers8 offirms (as well astheir other stakeholders) ina way that prevents these stakeholder groups from making excessively
risky decisions (Cornelli and Felli 1997, Claessens and Klapper, 2002). It is one ofthe fundamental dimensions ofbankruptcy law,
because thelack ofefficiency inthis area leads toat least two kinds ofcosts. In economies (or their specific segments) that lack this
mechanism thecosts ofcapital are certainly higher because thelenders bear a greater risk. A very serious problem is also posed
by reckless executives who suffer no consequences oftheir errors. Themacroeconomic costs ofthis phenomenon can be quite
high (Brogi and Santella 2004).
The adoption ofcertain general assumptions makes itpossible todetermine ex-ante efficiency conditions based on theclassic
model proposed by Bulow and Shoven (Bulow and Shoven 1978). Let us assume that thefinancial standing ofa given business
entity is influenced by a coalition ofshareholders (E) and banks which have agreed tofinance a certain project (B). This assumption
can be challenged under different conditions because, among other things, itdisregards therole played by bondholders (O), but
itis attractive inthat intheevent ofinsolvency proceedings thedecisions made by this group may be critical9 interms offiling
a bankruptcy petition and choosing theform ofthe proceedings. In order forinvestment I toachieve social efficiency itmust fulfil
thefollowing inequality:
PN IN

(4.4)

where P is thevalue ofthe firm inquestion (subscript N indicates that itis thevalue ofthe firm after theinvestment has been
made). On theother hand, from thepoint ofview ofthe financial coalition (the shareholders and banks) theefficiency ofthe project
depends on thefollowing condition:


since thefollowing relationship holds:

FIN KOALICJI

(P OK ) = (E K + BK ) > IN DN

P = E K + OK + BK

(4.5)

(4.6)

Subscript K intheabove formulae means that thefirm continues tooperate. Therelationship represented by (4.5) can be
interpreted asa condition stating that theincrease invalue ofthe firm (as a result ofinvestment I) less thenecessary increased
transfers tobondholders has tobe greater than theportion ofnecessary capital expenditure which must be financed by this group
(assuming risk neutrality). Accordingly, thecondition necessary toensure that thecoalition funds only such projects that are not
only beneficial from its point ofview but also socially effective, can be expressed inthefollowing way:

(P OK ) + ON = PN

(4.7)

The other dimension ofbankruptcy law is known asits ex-post efficiency, which should be understood asallowing only
fortheliquidation ofunprofitable (nonviable) firms while simultaneously preventing theelimination ofeconomically healthy
enterprises. This kind ofefficiency thus presupposes theexistence ofa sufficient separation ofthe adopted solutions interms
oftheir discriminative resolution. Theseparation aspect ofbankruptcy law was discussed inmore detail subsection 4.2. It should
also be noted that intheliterature and inpractice analysts point totheneed toadequately balance thetwo aspects ofregulations.
On theone hand, unduly restrictive legal provisions may lead totheelimination ofotherwise economically healthy business
entities which are temporarily experiencing financial distress. An opposite situation may also occur, whereby thelaw will be too
8

In thelight ofmore recent work this effect may be weaker than previously thought, since reductions inCEO incomes proved tobe overestimated
(see e.g. (ECBO, EB, et al. 2012).
Bondholders generally constitute a more diverse and spatially dispersed group entitled tofiling a bankruptcy petition. In consequence,
reaching an agreement with this group is likely tobe much more difficult and expensive forthefirms owners.

62

lenient towards permanently ineffective firms. Since type I and II errors tend tooccur inevery classification/selection system,
theprocedural solutions adopted should ensure an appropriate balance between them (White 1989).
Conditions similar to(4.7) can also be formulated forex-post efficiency. In this case, thecoalition ofbanks and owners are willing
tolet a given firm continue its operations (and banks are willing toprovide therequired funds) if:
(4.8)
E K + BK > BB


where subscript B represents thefirms bankruptcy. Note that condition (4.8) means that intheevent ofbankruptcy shareholders
receive nothing. Owing totherelationship expressed in(4.6) thecondition can also be expressed as:

P > BB + OK

(4.9)

In turn, thecondition forthesocial efficiency ofthe continued operation ofthe firm inquestion is asfollows:

(4.10)

P>L

where L is theliquidation value ofthe firms assets. Thus, thesocial efficiency ofcontinued operation ofthe firm co-occurs with
thecost-efficiency ofsuch an option from thepoint ofview ofshareholders and creditors when:

L = BB + OK

or OK = OB

(4.11)

It is easy toshow (cf. e.g. White 1980) that whether ornot conditions (4.7) or(4.11) are met depends toa considerable extent
on thesolutions adopted inthebankruptcy system, especially those applying totherules governing thepriority ofclaims (see
Chapter 1).
One may also consider theso-called process efficiency (interim efficiency) ofbankruptcy law (Mirowska 2013). It is defined
asa construction that ensures effective and asharmonious aspossible cooperation ofstakeholders inthecourse ofbankruptcy
proceedings, especially by reducing opportunistic behaviours ofmanagers offirms involved insuch proceedings.
Finally, a very broad spectrum ofconcepts ofefficiency ofbankruptcy procedures can be described asrecovery efficiency. In this
approach, theanalysis focuses on theimpact ofbankruptcy law on therelationships between theclaims ofspecific stakeholder
groups and theactual benefits that accrue tothese groups asa result ofthe bankruptcy procedure.
It should be emphasized that foreach group ofstakeholders theterm recovery efficiency may mean something different. Thus,
creditors may focus on variously defined capital recovery rates10 adjusted forthenecessary expenses incurred inthecourse ofthe
procedure, its duration, theeffects on reputation etc. Employees may be more interested intheproportion ofoverdue salaries
tobe paid, and intheevent ofrestructuring, inwhether ornot they will keep their jobs, thescale ofpossible layoffs etc. (Thornburn
2000). In turn, forthecurrent owners theterm may mean a sufficiently high degree ofpreservation oftheir existing shares,
thecapacity ofthe restructured firm toregain sustainable economic performance etc. Individual efficiency measures thus defined
may sometimes be inconflict, especially if we consider theinterests ofowners vs. those ofthe employees11 (cf. e.g. Morgado and
Garoupa 2006). Especially important are theconflicts ofinterests between thecreditors and theowners, which, incidentally, have
a considerable impact on theoperation ofthe entire bankruptcy system (Valdimirov 2007). In its empirical layer, thephenomenon
manifests itself, among other things, by various reactions ofstock prices offirms involved inbankruptcy proceedings, depending
on theactual design ofbankruptcy law (Gutirrez, Olalla et al. 2005).
In conclusion ofthis brief review ofthe concept ofefficiency ofbankruptcy law itis worth mentioning holistic approaches,
inwhich attempts are made todevelop normative solutions conducive tothemaximization ofsocial welfare (e.g. by combining
theinterests ofdifferent groups), cf. theso-called Objective 1 ofbankruptcy procedures (Hart 2000, Hummelen 2013). Their starting
point is often an analysis ofthe main objectives ofbankruptcy law, including, inparticular, appropriate solutions totheproblem
ofcoordinating redress if thefirm inquestion has numerous creditors. A thorough analysis ofthe behaviour ofthese entities may
constitute a source ofuseful tips forstreamlining theprocess.

10

11

Sometimes such criteria are formulated not interms ofrelationships but interms ofmaximization ofthe absolute amounts obtained by
creditors asa result ofthe proceedings. There are several reasons foradopting such an approach, but thepractical ones focus on thefact that
efficiency measures based on relationships are difficult toconstruct owing totheproblems inselecting thedenominator forthefraction.
Usually, thedenominator includes items such asreceivables and thefirms assets. It is very difficult toaccurately assess their value if thefirm
inquestion is infinancial distress, which is thecase especially during bankruptcy proceedings, since itis subject tosubstantial changes
throughout.
Some studies suggest theexistence ofa negative correlation between theefficiency ofbankruptcy procedures and theextent oflabour
market protection.

63

Fig. 4.2. Distributions ofchanges intheduration ofbankruptcy procedures (in years), their costs (% offirms assets) and recovery
rate (% ofliabilities) in20042013 in138 countries worldwide
Source: own study based on: (Doing Business 2013).

Summing up this topic, we should also mention that theissue ofthe efficiency ofbankruptcy law has not only a theoretical
dimension but primarily constitutes a matter ofenormous economic importance. This is supported by theintensity and depth
ofthe changes inbankruptcy law after World War 2, especially inthe1990s introduced by over 90 countries (Gine and Love 2006).
Theeffects ofthese modifications, however, were markedly different (cf. e.g. Franks and Loranth 2005). One ofthe underlying
causes ofthe problems, which were already mentioned above, is thedilemma related tothechoice between restrictive regulations
and thedesire toavoid their accidental victims resulting inpressure on theimplementation ofsofter solutions. One area
ofparticularly strong contradictions may involve thebalance between liquidation and restructuring provisions. This particularly
emphasizes theimportance ofEWS-type solutions: since thedevelopment ofan optimal bankruptcy system entails overcoming
certain difficulties, moreover, some ofthem appear tobe immanent innature, such early warning procedures offer more time
forrational decision-making regarding how toproceed with firms facing bankruptcy risk. If we look at efficiency parameters such
astheduration ofbankruptcy procedures, their costs and recovery rates (see Figure 4.2) surveyed by theWorld Bank in20042013
innearly 140 countries (Doing Business 2013), what is striking is that thefirst two ofthese variables have practically remained
unchanged (in both cases themedian values are zero, and theaverage is close tozero). Only therecovery rate has improved
insome countries (mean value = 5.0, median = 2.0).
Despite such an important role played by theissue ofthe efficiency ofbankruptcy procedures and theinclusive character
ofthis category, thedesign ofefficiency measures is difficult inpractice and, toa considerable extent, arbitrary. This is partly due
toa certain vagueness ofthe concept ofbankruptcy law and partly tothedifficulty inassigning toitspecific economic impacts that
should affect thebusiness entities involved. As a result, a significant portion ofthose measures is based on theanalysis ofqualitative
features ofbankruptcy systems and on attempts toassign todiscrete variables thus obtained specific economic significance on
thebasis ofvarious kinds ofquantitative modelling. Theterm economic significance also requires clarification, although itusually
refers tocertain cost categories ofbankruptcy, which were discussed intheprevious section.
We shall start our brief overview ofmeasures by recalling theidea proposed by Guiterezz (Guiterezz, Olalla et al. 2002) whose
elements were already present inearlier works (cf. e.g. (Rafael La Porta, Lopez-de- Silanes et al. 1998). Theapproach employed
theso-called ex-ante efficiency index constructed inthefollowing way: all thecharacteristics ofa bankruptcy system, which,
intheauthors opinion, contributed totheimprovement ofex-ante efficiency were assigned 1 point, while others were scored at
0. Five groups ofsystem features were evaluated, including: thegroup ofentities entitled tofiling a bankruptcy petition, whether
ornot thetaking ofsuch action was mandatory, thescope ofthe absolute priority rule (APR), thegroup ofentities entitled

64

tomaking decisions asto thefirms liquidation orrestructuring. Thesystem was also characterized by another measure, thesocalled reorganization index, theaim ofwhich was toprovide a synthetic reflection offeatures which could help toimprove
thesurvival offirms infinancial distress. Its design principle was similar and theindex itself was obviously an attempt toapproximate
ex-post efficiency. As a result, both indicators adopt values from 0 to5 and are stimulants.
Doing Business reports prepared by theWorld Bank offer practically useful efficiency measures which are based on theclassic
methodology developed by Djankov and Hart (Djankov, Hart et al. 2008). In particular, they employ such variables asbankruptcy
duration and costs aswell asa measure ofefficiency based on theconcept ofterminal value:
E=

VGC CONT + VLIK (1 CONT ) CBANKR


(1+ r )T
P

(4.12)

where: VGC represents theterminal value ofa firm if, asa result ofthe bankruptcy process, its financial health is restored and
itcontinues todo business (variable CONT assumes thevalue of1); VLIK theliquidation value ofthe company (CONT=0), CBANKR
thecosts ofthe bankruptcy procedure; T theduration ofthe procedure, P thevalue ofthe company before itis declared
bankrupt. It should be noted that inthis approach costs represented by CBANKR include only thedirect costs.
Other efficiency measures are based on theanalysis ofthe results ofindividual stages ofbankruptcy proceedings. For example,
Mirowska (Mirowska 2013) examines therelationships among theassets recovered (the ratio offunds recovered during
theproceedings tototal assets intheopening balance) and thedegree towhich creditors claims were satisfied (the ratio offunds
received by thecreditors totheamounts recovered). They allow fora decomposition ofthe total process efficiency ratio (the ratio
ofthe recovered funds tototal assets), which is their product.

4.5. Theefficiency ofthe bankruptcy process inthePolish economy


the considerations presented intheprevious section demonstrate that theefficiency ofthe bankruptcy process can be assessed
using different measures. Since such an assessment may depend on a number offactors, a good practice would involve taking
into account both partial measures and an attempt todesign an aggregated measure based on them. Here we seek toanalyse
theefficiency ofthe bankruptcy process inPoland using data from Doing Business 2014 (World Bank 2014). Three indicators are
used inthereport: theduration ofthe bankruptcy procedures, their cost and therecovery rate. Unfortunately, owing tochanges
inthemethodology ofthe study, results are not fully comparable throughout theentire study period (20042013), i.e. since its
beginning.
As was mentioned before, theresults will be presented intwo ways: asan aggregated measure and asa set ofpartial measures.
Theaggregated measure makes use ofthe principal components technique. Theglobal bankruptcy index thus created (the
first component) explains about 75% ofthe variance intheentire sample. Because ofthe way theindex was constructed, itis
a destimulant, i.e. thehigher its value, thelower theefficiency ofthe bankruptcy process. Thedistribution ofthis measure
throughout Europe is shown inFig. 4.3. Clearly, Poland occupies a position in themiddle behind most developed Western
countries. On theother hand, Poland ranks clearly higher than most post-socialist countries led by theCzech Republic, Latvia and
Lithuania.
It is worth comparing theresults obtained here with theoriginal classification12 included inthereport. In thecase ofthe leaders
20 countries with thehighest efficiency level ofthe process inquestion theclassifications are quite similar. More noticeable
differences appear aswe go down theranking list and are probably due tothefact that inour model, theshorter total duration
ofthe procedures is scored higher, whereas themethodology applied by theWorld Bank appears toattach more weight
totherecovery rate. In theoriginal World Bank publication of2013, Poland ranked 37th, second only totheCzech Republic among
theformer socialist countries.
This ranking should, however, be supplemented by information on thepartial indicators since itdoes not show theactual distance
between analysed countries, only their succession intheclassification. Hence, if we take into account thepartial indicators, we can
see that theefficiency ofthe Polish bankruptcy system is much lower than that ofthe leading countries. In thecase ofthe leaders
(Japan, Norway), theaverage procedure takes six tonine months. In Poland itlasts three years, and its cost (in percentage terms)
is several times higher. As far astherecovery rate is concerned, inthetop ten countries itis no less than 87%, whereas inPoland
itis only 55%.

12

Its main drawback is thelack ofdata on thecomposite index on which theranking is based.

65

EFFICIENCY INDEX:
-3.06 -2.71
-2.71 -1.36
-1.36 -0.96
-0.96 -0.39
-0.39 -0.04
-0.04 1.85

Fig. 4.3. Thedistribution ofthe bankruptcy proceedings aggregate efficiency index inselected European countries in2013
Source: own study based on: (Doing Business 2013).

Fig. 4.4. Time density distribution functions13 in2012 (in months): bankruptcies declared by thelower court (left graph) and
theduration ofbankruptcy proceedings (right graph)
Source: Authors own analysis based on Information (2013).

13

The negative results arise only from theapplied estimator and their weight inthedistribution is marginal. However, there were districts
inwhich theduration ofthe procedure amounted tozero due tothefact that none were actually conducted there. One extreme district was
excluded from theanalysis ofduration since theprocedure inquestion lasted 220 months. This was due tothefact that itwas theonly one
conducted there and therefore was considered tobe an outlier.

66

Based on theDoing Business survey, theefficiency ofthe Polish bankruptcy system can be classified asaverage, yet we lag far
behind thecountries with thebest solutions. It should also be noted that theindex inquestion is an aggregated one and hence
refers tothecountry asa whole. Disaggregated data about thecharacteristics ofbankruptcy proceedings conducted by district
courts (GU and GUp registers) indicate significant differences intheduration ofproceedings inindividual districts (see Figure 4.4).
In 2012, thetime required todeclare bankruptcy was approximately 2.6 months with thestandard deviation ofabout 1.1 months,
whereas bankruptcy proceedings took on average about 29 months with thestandard deviation of11 months. It should also be
noted that in2012 theduration ofbankruptcy proceedings was longer by nearly two months when compared to2011.

Part II
Bankruptcy vs. social policy

Anna ebrowska

Chapter 5
AN OVERVIEW OF THE SECOND CHANCE POLICY
5.1. Introduction
The European Union (EU) has firmly placed small and medium-sized enterprises (SMEs) at thecentre ofthe Lisbon Strategy
forgrowth and jobs implemented since 2000, and especially since 2005 through theapplication ofthe partnership approach.1 To
confirm theimportance ofSMEs inEU policies and fulfil thevision ofthe heads ofstate and government ofEU Member States
in2000, and thus make theEU a world-class environment forsmall and medium-sized enterprises, theEuropean Commission (EC)
on June 25, 2008 issued a Communication Think Small First. A Small Business Act forEurope (COM 2008 (394) (henceforth
referred toasSBA), which is crucial foractions under thenew opportunity policy.
SBA is themain strategic document at theEU level, which sets out theframework forSME policy. It comprises a coherent strategy
forsmall and medium-sized enterprises, included intheproposals forspecific actions implemented at both Community and
national level.
The second priority area ofthe Small Business Act refers explicitly tothesecond chance policy and aims toensure that honest
entrepreneurs who have faced bankruptcy quickly get a second chance.
The issues related tothesecond chance were also raised intheCommunication from theCommission totheCouncil ofthe
European Parliament, theEuropean Economic and Social Committee and theCommittee ofthe Regions ofOctober 18, 2007:
Overcoming thestigma offailure inbusiness fora second chance policy. Implementing theLisbon Partnership forgrowth and
employment (COM (2007) 584), and theCommunication of23 February 2011, Review ofthe Small Business Act forEurope (COM
(2011) 78).
EU policy aswell astherecommendations ofthe European Commission totheMember States inrelation totheissue
ofentrepreneurship development are based on theassumption that itis necessary tocontinuously improve thebusiness
environment, inorder tosimplify thesetting up ofsuch activities and toreduce barriers tobusiness development. Simultaneously,
thepriority is thecreation and implementation ofa second chance policy.
One ofthe objectives ofthis policy is toencourage those who have suffered a business failure. First, theexperience gained by
such people often allows them toresume business activities on a larger scale than previously. Second (but related tothefirst),
theactivation ofentities after a business failure, prepares theground formaintaining orincreasing employment, which is
particularly important fortheeconomies ofthose Member States that have been affected most by theconsequences ofthe crisis.2
Even though there are quite a few positive aspects ofresuming business activities by entities which have previously failed, this is not
a common occurrence. This is due, among others, totherisk inherent indoing business at any stage aswell asthegeneral public
stigmatization ofpeople who have experienced a business failure, which discourages them from taking up further challenges.
The Second chance policy cannot be limited toactions tofacilitate theresumption ofeconomic activity, but should provide
a comprehensive support system forentrepreneurs indifficulties. It should be supplemented by tools that help toprevent business
failures. This is especially important forsmall and medium-sized enterprises, which are affected by theconsequences ofeconomic
crises toa much larger extent than thelarge enterprises sector.
Problems related tothebankruptcy ofenterprises and resumption ofeconomic activity after sustaining a business failure have
been given special attention by theEuropean Commission. As part ofthe second chance policy, theEuropean Commission invites
theMember States to:
Promote a positive attitude inthesociety totheresumption ofbusiness activities by e.g. public information campaigns,
Strive toensure that all legal procedures related tobusiness closure intheevent ofa bankruptcy unrelated tonon-compliance
with thelaw last no longer than a year,
Ensure that business people re-starting economic activity are treated on an equal footing with thenew firms, which also includes
access toassistance programmes (ECC 2008).
1

In June 2010, theEuropean Council adopted a new strategy forsmart, sustainable and inclusive growth Europe 2020, which replaced
theLisbon Strategy
An expert review commissioned by theMinistry ofEconomy concerning theprevention ofbankruptcy ofenterprises and a second chance
policy, Warszawa 2012

71

As is clear from themonitoring ofthe implementation ofSBA principles developed fortheannual SBA Fact Sheet3 forindividual
Member States, inall countries, thelowest scores among theten SBA principles is given tothearea ofa second chance policy.
Emphasizing theessence and importance ofensuring that viable firms infinancial difficulty have an opportunity torestructure
at an early stage, theEC has issued recommendations on a novel approach toa business failure and insolvency (ECR 2014).
Therecommendations are aimed at preventing insolvency, thus ensuring maximization ofthe total value forcreditors, employees,
owners and theentire economy. Theaim ofthe recommendations is also toenable honest entrepreneurs who have been declared
bankrupt totake advantage oftheir second chance intheEU(ECR 2014, art. 1).
The international financial crisis occasioned a downturn in the Polish economy, and, as a result, destabilised a number of domestic
companies. Considering the phenomenon in question and its attendant problems, on July 22, 2014, the Council of Ministers
adopted a resolution concerning the New Opportunity Policy submitted by the Minister of the Economy. The document provides
a development programme4 and aims to create a comprehensive system in support of SMEs in a difficult financial situation and
entrepreneurs who intend to resume business activity.
The policy is intended toincrease thesurvival rate ofenterprises and tofacilitate theresumption ofeconomic activity by people
who have faced a business failure. Ensuring optimal functioning and development capacity ofenterprises constitutes an essential
element ofeconomic development strategies and programmes constituting an effective way toovercome orat least mitigate
theeffects ofeconomic crises.

5.2. Passivity orcommitment?


In recent years, increasing risk and uncertainty intheeconomy asa whole comprised themain negative factors affecting
theoperation ofbusinesses. Enterprises are increasingly subject tounexpected events series ofevents aswell asnegative economic
phenomena, which are difficult toanticipate.
The frequency ofoccurrence ofuncertainty and instability inthemodern economy is increasing due tothetechnological advances
and changes taking place all over theworld. In Poland, civilizational and technological transformation overlaps with theincreasing
internationalization and globalization ofthe economy. Under such circumstances, firms face a variety ofbarriers which are hard
toovercome, such asdifficulties with theshaping oflong-term development strategies, theabsence ofwhich exposes thefirm
tomanagement errors, which, inturn, may lead tobankruptcies (Mczyska 2009).
The ongoing globalization processes, thedevelopment ofcapital markets, multinational corporations and a complex network
ofrelationships make not only theeconomy, but also therisks subject toglobalization (Mczyska 2005). In a knowledge-based
economy, factors related toknowledge and information management play an increasingly important role. At thesame time,
knowledge and information management aswell astheir flow faces a number ofobstacles, which may result inmeasurable
business losses.
Given such uncertainty, economic policy requires long-term planning, flexibility and adaptation totherapidly changing economic
conditions. Thus, actions taken by central governments should be aimed at overcoming theabove-mentioned factors. Thepoor
culture ofstrategic approach also adversely affects economic policies and conditions inwhich enterprises operate. For those
reasons, comprehensive research on theissues ofuncertainty and risk may explain themechanisms ofsurvival and bankruptcy
ofenterprises aswell ascontribute toa rationalization ofeconomic policy.
Numerous entrepreneurs inthecourse ofday-to-day business activity experience financial difficulties and liquidity problems.
Itshould be noted, however, that financial distress does not necessarily mean unprofitability orlack ofprospects forbusiness
growth. New and small businesses are often financially unstable. Their problems may result from a number ofreasons, such
asrecession, loss ofa key supplier orcustomer, unpaid obligations, poor management, lack ofcontrol over liquidity, lack ofexpertise
insuch critical areas asaccounting and marketing, high growth rate aswell asfrom personal reasons, such asillness.
It makes much more sense torestructure a firm and maintain theexisting jobs intheevent ofits temporary financial distress
oreven insolvency, forthesake ofthe firm itself, its employees and theeconomy rather than toinstantly liquidate it. Therefore,
critical importance is attached toeffective protection procedures which help enterprises facing insolvency tomaintain current
liquidity, continue their operations and, subsequently, torestore profitability and torepay debts totheir creditors.

The document prepared annually by a group ofexperts forResearch on Entrepreneurship and SME Sector (Policy-relevant Research on
Entrepreneurship and SMEs), acting under theguidance ofthe European Commission. A separate publication ofthe SBA Fact Sheet For is
prepared foreach ofthe Member States. Monitoring provides thebasis forpeer review and exchange ofgood practices between Member
States inthearea ofcompetition policy.
Pursuant toArticle 15 ofthe Principles ofDevelopment Policy Act ofDecember 6, 2006 (Journal ofLaws 2009, No. 84, item. 712, asamended).

72

Successful management ofan enterprise consists not only intheperception ofand adequate response toexternal factors that
influence its development, such asits specific environment, demographic, social and cultural factors, but also depends on internal
factors, such asgovernance, staffing policy and therelationships with thefirms business partners.
Practice shows that a confronting difficulty ina timely manner increases thelikelihood ofsaving thefirm theearlier a problem
is diagnosed and acted upon, thebetter restructuring results can be achieved.
A key issue is toaccurately identify themain crises-generating factors and an early warning against threats tobusiness activity and
impending insolvency, which may cause an entrepreneur togo bankrupt. These are thenecessary conditions formaking sound
business decisions (Mczyska 2009).
Entrepreneurs should be able torespond totheidentified threats inan inconspicuous way, since thedisclosure ofproblems may
not only reduce theprestige oradversely affect theimage ofthe firm, but above all, itmay trigger a chain ofevents accelerating
thenegative trends within it, such astheloss ofsome business partners asa result ofthe disclosure ofinformation related toits
poor financial situation. Theabove constitute some themost important reasons why entrepreneurs conceal their problems and
take ineffective attempts torepair their firm by themselves.
State action targeting viable enterprises indifficulty should focus on minimizing thenegative effects ofa crisis situation within
thefirm and encourage a more proactive approach inresponding tochange and emerging challenges. This does not mean, however,
that thestate should prevent bankruptcy at all costs, since itconstitutes a natural component ofthe economy based on market
principles. Therole ofthe state is tocreate legal and institutional framework allowing thefirms toimplement remedial measures. Only
intheabsence ofevidence that might warrant restructuring should they be able togo bankrupt asefficiently aspossible.
Ensuring optimal capacity forenterprises tooperate and develop aswell aspromoting a more proactive approach topreventive
measures constitute indispensable elements ofstrategies and economic development programmes aswell asan effective way
toovercome orat least mitigate a business failure. Greater commitment torapid intervention torectify previously identified
problems allows forthestabilization ofan enterprise indifficulty and minimises thenegative effects ofchange.

5.3. Therole ofpreventive action


In 2003, theEuropean Commission emphasized theimportance ofearly warning systems,5 whose implementation is meant todetect
emergencies and financial difficulties before firms at risk actually become insolvent. With theright combination ofeconomic and
financial indicators characterizing an economic entity its failure can be predicted with a good degree ofprobability. Likewise, there
are tools with which thewillingness ofentrepreneurs toexpose thefirm tocrisis-generating factors increasing therisk ofbankruptcy
can be determined. Unfortunately, this kind ofrisk management is not very popular, especially among micro-entrepreneurs, hence
rarely used inpractice. Supporting thedevelopment and use ofearly warning systems offers one ofthe ways ofimproving thesurvival
rates ofenterprises and maintain economic activity without theneed toresort toreducing creditor protection.
In order toimplement an effective system fordiagnosing thefinancial health ofthe SME sector, itis necessary totake into account
thedifficulties interms oftheir access toaccounting and financial information.6
Due tothefact that entrepreneurs who experience difficulties inmanaging their firm often cannot afford expensive legal
consultancy, they should be given access tothefullest possible range ofsuch services. To that end, a good solution is touse
thesupport opportunities currently offered on themarket by business environment networks, particularly inthecontext
ofinformation and consultancy services.
The category ofpreventive instruments, which constitute part ofthe preventive measures inthearea ofthe second chance
policy, includes theoption togrant de minimis aid7 and aid covered by block exemptions.8 The legal system makes itpossible
toprovide such assistance regardless ofwhether theaid comes solely from thenational budget orwhether itwas provided forin
aid programmes tothepublic under individual operational programmes and regardless ofwhich agency provides such aid.
It is reasonable, however, tonote that inorder toensure effectiveness itis necessary toassume that any de minimis aid
forentrepreneurs intheearly stages ofdifficulties aswell asthose meeting theconditions stipulated forenterprises indifficulty

5
6

The best project on Restructuring, Bankruptcy and a Fresh Start, 2003.


Expert report commissioned by theMinistry ofEconomy Instrument badania kondycji przedsibiorstwa (An Instrument todiagnose firms
economic condition), Allerhand Institute, Krakw 2010.
Commission Regulation (EU) No 1407/2013 ofDecember 18, 2013 on theapplication ofArt. 107 and 108 ofthe Treaty on theFunctioning
ofthe European Union tode minimis aid (OJ. EU L 352 of24.12.2013).
Commission Regulation (EC) No 800/2008 of6 August 2008 declaring certain categories ofaid compatible with thecommon market
inapplication ofArt. 87 and 88 ofthe Treaty (General Block Exemption).

73

asprovided by EU regulations on rescuing and restructuring9, aid should be used tosolve their problems and can be granted only
tosome ofthose entities.
De minimis aid is limited totheamount of200,000 euro within 3 calendar years, which restricts thenumber ofpotential
entrepreneurs who could restructure their firms on thebasis ofthis principle. It should be added, however, that theapplication
ofthese solutions will primarily minimize theobligation todraw up relevant documentation with respect toobtaining theconsent
ofthe European Commission togrant state aid.
Another advantage ofthis solution is thefact that no special support measures dedicated toentrepreneurs who encounter
difficulties, are needed. Thesolution makes itpossible todraw on theexisting instruments, so that they can be used fora new
purpose, namely solving theproblems faced by entrepreneurs.10 This means that such measures could be implemented without
theneed forlegislative action todevelop new regulations (support programmes) governing aid tofirms indifficulty.
Nonetheless itmust be emphasized that therestructuring ofa firm indifficulty, which is not eligible foraid under thede minimis
rule, will still require an individual notification procedure,11 which significantly slows down therestructuring process and increases
therisk ofits failure. It should be noted, however, that inprinciple thelegislation governing state aid admits ofthe preparation and
notification ofan assistance programme,12 which may constitute a systemic solution forbusinesses that either cannot be granted
assistance under thegeneral block exemption ortheaid exceeds theamount set forth inthede minimis rule.
Providing small and medium-sized enterprises with an opportunity toreceive assistance without a long waiting period
fortheCommissions decision may be crucial inensuring that they regain profitability more quickly aswell asminimizes therisk
ofbankruptcy.
It should be emphasized that thedevelopment ofa restructuring assistance programme whose effect is purely procedural and
not substantive, does not establish an additional legal basis forgranting state aid and is not related toits amount. Moreover, such
an assistance programme has thefollowing benefits:
1. It shortens theaid-granting procedure, which increases thelikelihood ofsuccess ofthe restructuring process and,
consequently, reduces thenumber ofbankrupt entrepreneurs,
2. It reduces thecosts related toobtaining thePresident ofthe UOKiKs (Office ofCompetition and Consumer Protection, OCCP)
opinion and costs ofproceedings involving theEuropean Commission, which will allow fora significant shortening ofthe
procedures inthearea ofgranting restructuring aid, asthey will only involve thenational aid-granting agency.
Entrepreneurs inadequate knowledge and awareness ofpossible threats and ways inwhich they can respond tosuch threats 13
constitutes an argument infavour ofthe institution ofexpert consultancy under thede minimis rule, which would ensure
theidentification ofpossible solutions and implementation ofappropriate action inenterprises inorder toavoid further problems.
It can therefore be stated that at thestage ofcrisis prevention inan enterprise itis extremely important tocreate conditions that
will allow theentrepreneurs toimprove their knowledge and skills inbusiness management, including crisis management. It
should be remembered that consultancy provided tofirms which still are ina good financial condition may be more effective
than such activities designed forfirms experiencing difficulties (II szansa 2011). Consequently, training and consultancy activities
within theframework ofbankruptcy prevention should focus primarily on raising theentrepreneurs awareness ofthe risks and
failures inherent inbusiness activity, theability torecognize thesigns ofa forthcoming crisis aswell astheexisting forms ofaid and
restructuring when their firms face financial distress. Such services should also be largely based on thetransmission ofpractical
skills with theparticipation ofpeople who have had similar experiences.

5.4. Judicial and non-judicial proceedings


Entrepreneurs infinancial difficulty who have limited resources are often unable toafford a long restructuring process involving
external consultants, because itgenerates significant additional costs. Affordable and simple restructuring procedures are therefore
necessary. Non-judicial procedures are often cheaper and faster and thus particularly attractive forstart-ups and small businesses.
Amicable settlements ofdisputes and reorganization implemented without involving thejudiciary occur outside statutory
regulations. Their objective is torestructure thefirm which currently finds itself infinancial distress. They often involve unofficial
settlements, because they are not directly regulated by thelaw. According toDutch researchers, such procedures often comprise
9

10

11
12
13

Communication from the Commission Guidelines on State aid for rescuing and restructuring non-financial undertakings in difficulty
(OJ C249, 31.07.2014).
Expert report commissioned by theMinistry ofEconomy concerning theprevention ofbankruptcy ofenterprises, and a second chance policy,
Warszawa 2012, p. 62.
Council Regulation (EC) No 659/1999 of22 March 1999 laying down detailed rules fortheapplication. 93 ofthe EC Treaty.
Ibidem.
Expert report commissioned by theMinistry ofEconomy concerning theprevention ofbankruptcy offirms, and a second chance policy,
Warszawa 2012, p. 5658.

74

two processes: 1. therestructuring ofeconomic activity, when an attempt is made torestore a firms viability, and 2. financial
restructuring, where repayments are deferred, thenominal debt is reduced and new capital is brought in(Adriaanse 2005).14
Moreover, according totheEC report Business Dynamics: Start-ups, Business Transfers and Bankruptcy, with respect tonon-judicial
proceedings, particular attention was paid tothefact that thepurpose ofsuch proceedings forthecreditors should be toachieve
a joint agreement on theliabilities. Theagreements are concluded out ofcourt, making itfaster and cheaper, but itdoes not mean
that they entail no costs. Thereport indicated that inonly five Member States (Poland is not one ofthem) thetax authorities participate
insuch proceedings. Furthermore, itwas noted that theaverage length ofsuch proceedings inPoland is 7.7 months, which is
considered average. Thesystem ofnon-judicial proceedings inPoland was similarly rated based ofthe outcome coefficient.15
In order toachieve themost satisfactory agreement, thestate should allow entrepreneurs facing insolvency tonegotiate
agreements with their creditors without involving thejudiciary, e.g. through arbitration. Entrepreneurs should enjoy an easy
access toinformation on alternative modes ofdispute resolution, inparticular, especially on negotiation and mediation. Based on
international experience, non-judicial settlements often follow certain established standards orcodes ofconduct. An important
role inpromoting codes ofconduct may be played by public authorities such asthecentral bank ortheMinistry ofFinance,
asexemplified by theso-called London approach promoted by theBank ofEngland since 1980.16 The state may also support such
solutions by establishing a general legal framework and favourable tax systems toencourage such settlements.17
State action toreduce therisk ofpremature bankruptcy targets entrepreneurs whose economic situation is so serious that
itrequires them totake active remedial measures. At this stage, thekey issue forthesurvival ofthe firm is toprovide conditions
conducive toits effective restructuring, which will restore its ability tocompete on themarket and thereby contribute toincreasing
thenationwide business survival rate. At thesame time, we must not forget that intheearly stages ofproblems affecting a given
firm, all kinds ofnon-judicial settlement ofdisputes can be used, which may lead torepairing thefirms condition inparallel with
therestructuring ofits debt.
Currently, inPoland thepopularity ofnon-judicial ways ofreaching agreements between entrepreneurs and creditors, such
asmediation18 orarbitration,19 is negligible. Although judicial mediation was included intheCode ofCivil Procedure in2005 and
about 4 thousand mediators are active on themarket, only a fraction ofeconomic matters is settled inthis way. Nonetheless, there
is no verifiable data on thescale ofnon-judicial mediation. Meanwhile, a proper application ofthose instruments would help
toreduce thecosts and duration ofjudicial procedures involving economic issues.
At present, under Polish law, therestructuring offirms is governed by Part Four ofthe Bankruptcy and Restructuring Act ofFebruary
28, 2003.20 Still, both thestatistics and thedaily practice indicate a negligible use ofthe institution ofits provisions by Polish
entrepreneurs.
The main objective ofthe restructuring process is toimprove thecondition ofthe firm, which should restore its capacity tocompete
on themarket. This means that entrepreneurs should treat theprocedure asan instrument tobe used when thefirm is not yet
eligible tostart insolvency proceedings. It is therefore necessary topromote a way ofthinking about therestructuring procedure
asa tool torespond toemergent issues, even non-financial ones, but those that may inthefuture have a negative financial impact.
Difficulties indetermining when a firms situation actually requires theimplementation ofremedial measures cause that
entrepreneurs tend totake such action too late, moreover, they file bankruptcy petitions, which is even more unfavourable.
Currently, due tothedynamic changes inthemarket and errors made directly by entrepreneurs, numerous firms find themselves
infinancial distress which manifests itself asinsolvency.
In Poland, theconcept ofinsolvency based on which an entrepreneur may be eligible forrestructuring proceedings orbankruptcy
proceedings is governed by Article 11 ofthe Bankruptcy and Restructuring Act ofFebruary 28, 2003.21 According toits provisions,
an entrepreneur is deemed insolvent if his firm does not cover thefinancial obligations which have become due. This provision
applies to:

The firms monetary obligations,

All theliabilities regardless oftheir extent, causes and aswell aswhether ornot they are related totheeconomic activity. In
regard ofthe shareholders ofunlimited partnerships, inwhich partners are responsible fortheir firms liabilities with all its
14
15
16

17
18
19
20
21

The legal options forthecapital solution remain a moot point.


Final report ofthe (Business Dynamics 2011).
Entrepreneurship policy indicators forbankruptcy legislation inOECD member and non-member economies. Final Report, Working Party forSmall
and Medium-sized Enterprises and Entrepreneurship, OECD CFE / SME (2006)3.
Commission Recommendation ofMarch 12, 2014 on a new approach tofailures inbusiness and insolvency (C (2014) 1500).
Code ofCivil Procedure ofNovember 17, 1964 (Journal ofLaws 2014.101 uniform text), Articles 18318315.
Code ofCivil Procedure ofNovember 17, 1964 (Journal ofLaws 2014.101 uniform text), Part Five The Court ofArbitration.
Bankruptcy and Restructuring Act ofFebruary 28, 2003 (Journal ofLaws No. 60, item. 535, asamended).
Bankruptcy and Restructuring Act ofFebruary 28, 2003 (Journal ofLaws No. 60, item. 535, asamended).

75

assets, itis immaterial whether these entities fail toperform theobligations on behalf oftheir firm orfail toperform their
other obligations,
The liabilities due asset by thepublic law and thecivil law.22

Entrepreneurs facing bankruptcy risk may declare incourt that they have planned and initiated remedial proceedings.
Theentrepreneur is considered tobe at risk ofinsolvency if, despite thepayment ofdue liabilities, itis obvious that his firm will
soon become insolvent according toa reasonable assessment ofits economic situation.23
Both thestatistics and thedaily practice indicate that Polish entrepreneurs make a negligible use ofthe provision ofthe Act. This
fact contrasts sharply with thedifficulties that they have had toface inrecent years, related totheeconomic situation (which
resulted, among other things, inan increase inthenumber ofbankruptcies) and with thepopularity ofvaried remedial solutions
available inother countries. Leaving aside thereasons forthis state ofaffairs on thepart ofthe entrepreneurs themselves, itmeans
that theregulations governing Polish remedial procedures and therestructuring infrastructure are not functioning properly and
require a critical analysis and subsequent modification (Analiza moliwoci 2010).
The remedial orcorrective procedure is implemented by theentrepreneur assisted by thedesignated court supervisor.
Theprocedure should involve simple rules ofconduct, which entails short proceedings, since a recognised risk orthreat may
rapidly develop into palpable difficulties. Moreover, thedebtor is entitled toprotection from creditors, who are not interested
inprolonging theproceedings and thereby delaying thesatisfaction oftheir claims. Quick proceedings are possible if court actions
are limited toa few necessary steps crucial forthelawfulness ofthe final ruling. Currently, they include:

The capacity toprohibit theentrepreneur from implementing remedial measures asa result ofoccurrence ofthe conditions
specified intheregulations,

The capacity toappoint a court supervisor, whose presence guarantees that all themeasures undertaken will lead
toa negotiated agreement ina professional manner and incompliance with legal and economic knowledge,

Approval orrefusal toapprove theagreement,

Revocation ofthe agreement.
The remedial procedure is actually implemented inthedebtors interest, or, tobe more specific, intheinterest ofthe potential
debtor, since theentrepreneur who plans toembark on such an undertaking is still paying his obligations (which may soon
cease tobe possible). Theprocedure is also intheinterests ofcreditors orrather suppliers ofan entrepreneur threatened with
insolvency. It is important, therefore, that therules clearly present theposition ofthe legislator, stating that remedial action
toprotect theentrepreneur fora specified period oftime against too hasty debt collection efforts on thepart ofpotential creditors
is meant toenable thefirm toimprove its condition, perhaps saving itfrom bankruptcy. Preventing such firms from bankruptcy
subsequently rescues and protects current turnover ofother suppliers.
As a result ofthe initiation ofremedial measures, theenforcement ofobligations owed by theentrepreneur and theaccrual
ofinterest on thedebt is suspended (deduction ofdebts is permissible). No security and enforcement proceedings against
theentrepreneur can therefore be initiated, and those already underway are subject tosuspension by law except those involving
claims not covered by theagreement.24 In theevent that a creditors files thebankruptcy petition, thecourt will postpone hearing
thecase until thecompletion ofthe remedial procedure orwill combine itwith theagreement approval procedure.25
It is worth noting that theinclusion ofremedial legislation inthesame legal act astherules governing bankruptcy procedures
has a very negative impact on thepopularity ofrestructuring measures asa means ofrescuing enterprises from difficulties,
since remedial procedure tend tobe strongly associated with bankruptcy (insolvency). Legally sanctioned separation ofremedial
legislation from bankruptcy provisions will serve toimprove theway itis perceived and applied inpractice (II szansa 2011).

5.5. Systemic significance ofearly warning mechanisms


Preventive and pre-emptive measures on thepart ofthe state inthearea ofearly warning systems should be readily available
forentrepreneurs, complete with transparent and clear rules fortheir employment. At thesame time, entrepreneurs must be
aware ofwhat phenomena constitute warning signals that should prompt them totake action under such a system, and possess
theskills toemploy those solutions.
In many cases where preventive measures have not been instituted orhave failed toproduce theexpected results, itis necessary
toimplement a procedure under which theentrepreneurs debt is restructured, including further remedial action taken toeliminate
22

23
24
25

Expert report commissioned by theMinistry ofEconomy concerning theprevention ofbankruptcy ofenterprises, and a second chance policy,
Warszawa 2012.
Bankruptcy and Restructuring Act ofFebruary 28, 2003 (Journal ofLaws No. 60, item 535, asamended), Articles 492 &ff.
Bankruptcy and Restructuring Act ofFebruary 28, 2003 (Journal ofLaws No. 60, item 535, asamended), Article 498.
Ibidem, Article 499.

76

thecauses ofsimilar difficulties inthefuture. A significant role inthis regard could potentially be played by non-judicial proceedings
due totheir simplicity, pace and informal character. Notwithstanding theforegoing, inpractice, they should be organized so
that entrepreneurs indifficulties could actually see them asan opportunity toeffectively restructure their firms and eliminate
theproblems that affect them.26
The role ofthe state should be limited tohelping entrepreneurs toidentify threats and prevent theoccurrence ofcrisis situations
intheir firms, but should not involve counteracting bankruptcy at all costs. It is intheinterest ofthe state that firms should be
strong and flexibly react toany changes on themarket. Ensuring optimal operation and development capacity ofenterprises
constitutes an essential element ofeconomic development strategies and programmes and an effective way toovercome orat
least mitigate theeffects ofeconomic crises.
An important role inovercoming thedifficulties inherent ineconomic activity is played by preventive measures, both inthearea
ofidentification/anticipation ofcrisis situations inenterprises and averting premature bankruptcies. Therefore, state actions should
focus on providing entrepreneurs with thepossibility ofimplementing preventive measures with thefollowing objectives:

To promote ongoing monitoring ofthe economic and financial situation offirms,

To promote early warning systems detecting thesusceptibility offirms tocrises and financial difficulties before they become
insolvent,

To provide a fully featured system ofconsultancy and training forentrepreneurs, enabling them tocope effectively with
thethreat offinancial difficulty, which could result inbankruptcy.
An example ofsuch a solution is thepilot project carried out by thePolish Agency forEnterprise Development (PARP)
directed toentrepreneurs who experience economic problems, but their firms do not yet meet thecriteria ofan enterprise
indifficulty. Theproject called theRapid Response Instrument (RRI, Polish acronym ISR) involves a system ofpreventive measures
intended tostabilize thecondition offirms indistress and mitigate thenegative effects ofeconomic change by undertaking
rapid intervention thanks toan early identification ofproblems. Thus, entrepreneurs may obtain comprehensive assistance
inthedevelopment and implementation ofdevelopment plans, which will provide thebasis forcorrective ordevelopment action.
A very important component ofa comprehensive early warning system consists ofpreventive instruments, such ason-line tools
toenable entrepreneurs tomonitor their financial situation. Theidentification ofeconomic entities intheearly stages ofdifficulty
will allow entrepreneurs tobecome eligible forany state aid within thelimits set by therules ofadmissibility laid down at EU
level fordifferent kinds ofpublic assistance. These solutions primarily reduce theneed toprepare relevant documentation and
do not require prior consent ofthe Commission togive aid. As far asprevention instruments are concerned, itis possible togrant
de minimis aid and public assistance covered by theblock exemption regardless ofwhether such support derives solely from
thenational budget orwhether itis provided forby public assistance programmes under operational programmes and regardless
ofthe body that gives such aid.
It should be noted that de minimis aid is limited totheamount of200,000 euro within 3 calendar years, which limits thenumber
ofpotential entrepreneurs who could restructure their firms on thebasis ofthis principle. Consequently, restructuring firms
indifficulty not eligible foraid under thede minimis rule will require thecompletion ofan individual notification procedure, which
significantly slows down therestructuring process and increases therisk ofits failure.
Identification ofthe problems ofentrepreneurs who find themselves ina difficult economic situation and ofmeasures toovercome
them should constitute one ofthe priorities ofthe second chance policy pursued by thestate. For example, itwould be possible
tocreate a refundable aid programme, which would allow SMEs toimplement thenecessary restructuring measures (e.g. change
their production profile orrestructure employment) and ensure appropriate conditions fortheprocess.
Consequently, thedesign ofnational legislation should make iteasier toconduct a comprehensive restructuring ofenterprises
(including structural and financial restructuring). New normative acts should institute an aid programme that includes rescuing
firms from difficulties, restructuring, temporary support forrestructuring and therestructuring ofcertain public and legal dues.
Thedevelopment ofan assistance programme will offer entrepreneurs clear and transparent eligibility rules forpublic assistance, and
hence itwill also speed up theprocedure, reduce administrative burdens aswell asimprove theefficiency ofrestructuring measures.
The idea ofa second chance involves not only theprovision ofeffective support toentrepreneurs indifficulty, but also making use
oftheir potential and experience incase their firm ceases its activity. Thanks tothedissemination and promotion ofthe possibility
toresume business activity itis possible toachieve a positive impact on societal attitudes towards entrepreneurship. Hence,
awareness-raising activities should be undertaken aimed at overcoming thebad reputation ofthe phenomenon ofbankruptcy
and at theprevention ofstigmatisation ofentrepreneurs who have experienced a business failure. This issue is important since
itinvolves a sensitive matter such astheconsciousness not only ofentrepreneurs, but also society asa whole.
26

Expert report commissioned by theMinistry ofEconomy concerning theprevention ofbankruptcy ofenterprises, and a second chance policy,
Warszawa 2012.

77

A very important feature ofactivities inthearea ofcountering thestigmatisation ofentrepreneurs who have experienced a business
failure is first ofall thecreation ofopportunities fortheresumption ofbusiness activity using their previous experience.
Needs analyses show that people who have been forced out ofbusiness require different services than active entrepreneurs
orpeople starting new businesses. Therefore, inorder toallow people who have experienced a business failure toresume economic
activity, a dedicated support system forformer entrepreneurs (a new start) should be established.27
Interventions undertaken aspart ofthe new start programme should target exclusively honest entrepreneurs whose firms have
been liquidated, who would like toremain active and apply foraid inorder toaccomplish their business plans taking into account
their past experience. Moreover, targeting support tosuch businesses will have a positive impact on thesocietal perception
ofthe concept ofa new start. Such actions are likely toeliminate thenegative connotations ofbankruptcy and will constitute an
important step incountering thestigmatisation ofentrepreneurs who have experienced a business failure.
In view ofthe above considerations, thecurrent economic downturn and theresulting problems affecting many Polish firms,
theMinistry ofEconomy prepared the document New Chance Policy. Thedocument aims tobuild a comprehensive support
system forentrepreneurs (SMEs) who find themselves infinancial difficulty and those who intend toresume economic activity.
Thepolicy is designed toincrease thesurvival rate ofenterprises and tofacilitate theresumption ofeconomic activity by people
who have suffered a business failure.
Thus, theNew Chance Policy provides foractivities with thefollowing objectives:
1. To reduce therisk ofbusiness setbacks,
2. To minimize theimpact ofnegative connotations related tobankruptcy via liquidation through thedissemination and
development ofrestructuring procedures,
3. To streamline bankruptcy procedures, especially interms ofshortening theprocedure and reducing its costs,
4. To provide conditions toensure that every honest entrepreneur who has experienced a business failure is offered a chance
tostart a new business.
The New Chance Policy involves not only theprovision ofeffective support toentrepreneurs indifficulty, but also aims tomake use
oftheir potential and experience incase their firms have toterminate their activity. Firms founded by people who resume business
activity record higher turnover and employment growth, aswell asexhibit greater prudence indecision-making compared with
firms run by start-up entrepreneurs orby those who have never suffered a defeat. At thesame time, thepromotion and support
ofthe resumption ofbusiness activity makes itpossible toachieve a positive impact on societal attitudes toentrepreneurs.
The New Chance Policy advocates thedissemination ofremedial measures aimed at enterprises, which would involve, first ofall,
restructuring followed by negotiated settlements. Liquidation would be undertaken only intheabsence ofeconomic reasons
forrestructuring. One ofthe primary aims ofthe proceedings should be tocontinue sustainable operation ofeconomic entities,
which offers and opportunity, among others, tomaintain existing jobs. Such a model requires appropriate changes inbankruptcy
legislation.
Changes inthelegislation should also include relevant insolvency proceedings with a view tomaking itless formal and simpler,
which may translate into its lower costs aswell asshorten its duration.
Conclusions from theanalysis ofthe practical application ofthe existing regulations inthefield ofbankruptcy and restructuring
law and therecommendations ofthe European Commission have been taken into account intheproposals ofnew regulations
prepared by theMinistry ofJustice, which is currently working on drafting new legislation (the Restructuring Act) and on
amendments toin theBankruptcy and Restructuring Act ofFebruary 28, 2003.
The adoption ofa completely new law theRestructuring Act will ensure institutional autonomy ofthe restructuring proceedings
from thebankruptcy proceedings, asthelatter may unduly stigmatize entrepreneurs.
According tothedraft document,28 theprimary change will involve theway ofdealing with insolvent businesses, including theaim
ofthe proceedings. Bankruptcy and liquidation ofthe debtors assets will be treated asa last resort measure. Themain aim ofthe
changes is toensure toentrepreneurs ina difficult financial situation conditions conducive tosuccessful restructuring, allowing
them toresume normal activity within several toa dozen months.
Consequently, theinventory ofrestructuring proceedings will be expanded, which will be more flexible and better-adapted
totheactual needs ofdebtors. Thenew Restructuring Act will introduce four brand new restructuring institutions, based inpart
on thebest international practices adapted tothePolish economy. Entrepreneurs experiencing financial problems should be given

27

28

Expert report commissioned by theMinistry ofEconomy concerning theprevention ofbankruptcy ofenterprises, and a second chance policy,
Warszawa 2012.
The Restructuring Bill adopted by theCouncil ofMinisters on February 11, 2014.

78

unobstructed access toflexible procedures, which, on theone hand will offer them an opportunity foreffective restructuring, while
on theother ensuring creditor protection.29
The development ofnew types ofrestructuring proceedings has thefollowing primary aims:
a) To simplify theprocedures asmuch aspossible,
b) To enable entrepreneurs toindividually and independently choose themost convenient restructuring path,
c) To reduce thetime required toconclude a successful agreement with creditors,
d) To enable actual restructuring ofthe overall activities ofthe firm under court protection.
All restructuring procedures should be directed both toinsolvent entrepreneurs and those threatened with insolvency inline with
theview that theliquidation ofa firm is only permissible intheabsence ofeconomic reasons forrestructuring.
Moreover, inaccordance with theprovisions ofthe draft bill, a debtors insolvency will be defined asinability tomeet his financial
obligations forover 3 months, which means that such inability should meet thecondition ofpermanence (the liquidity premise).
Theopinion asto theloss ofability tomeet financial obligations should result from theevaluation ofthe financial condition ofthe
debtors business. An additional criterion ofinsolvency is met if thetotal amount ofliabilities exceeds thetotal value ofthe debtors
saleable assets (the liability premise).
The Bankruptcy and Restructuring Act emphasizes theprimacy ofinterests ofcreditors over theinterests ofthe insolvent
debtor. In thenew Act, this provision should apply only at theliquidation stage. To enable theentrepreneur tocarry out an
effective restructuring programme, during negotiated settlements and reorganization theinterests ofcreditors and thedebtor
should be protected toa similar extent. Thelack ofbalance between those two kinds ofprotection makes itvirtually impossible
forentrepreneurs topursue business activity that would result inovercoming thethreat ofinsolvency and ensure their firms
survival.
It should be emphasized that financial difficulties affecting a firm, which may lead toits liquidation, do not necessarily mean defeat.
New attempts toconduct economic activity should be based on past experience, which will have a substantial influence on
thechange ofsocietal attitudes towards entrepreneurs. Accordingly, theNew Chance Policy provides foran extensive information
campaign aimed at changing thepublic perception ofthe phenomenon ofbankruptcy.

29

Ibidem.

79

Mikoaj Stasiak

Chapter 6
THE LEGAL FRAMEWORK FOR STATE AID
6.1. Introduction
State aid constitutes a powerful intervention instrument thanks towhich desired economic phenomena can be strengthened
orthephenomena perceived asnegative can be counteracted. In particular, such aid may assist inthedesign ofeffective policies
against bankruptcies.
In its Communication totheEuropean Parliament, theCouncil, theEuropean Economic and Social Committee and theCommittee
ofthe Regions The modernization ofEU state aid policy1 theEuropean Commission notes that theEuropean development strategy
laid down inthedocument Europe 2020, sets three objectives fortheEuropean economy, which should achieve smart, sustainable
and inclusive growth. TheCommission points out that thebest instrument forachieving these goals is thesingle market. Its
efficient operation requires a more modern state aid control, so that thespending ofthe Member States inthearea is aimed at
theimplementation ofpolicies foreconomic growth while limiting thedistortion ofcompetition.
The Commission stresses that themodernization ofthe system ofstate aid control is aimed to: i) stimulate sustainable, smart and
inclusive growth inthecompetitive internal market, ii) focus theCommissions attention on ex-ante control activities involving cases
with thegreatest impact on theinternal market while strengthening cooperation between Member States intheenforcement
ofstate aid rules, iii) optimize therules and shorten thedecision-making process.
The first ofthe stated objectives requires that state aid control shall facilitate theprocessing ofaid that is well designed, targeted
at identified market failures and objectives ofcommon interest and cause theleast disturbance (appropriate assistance). This shall
ensure that public support stimulates innovation, thedevelopment ofgreen technology and human capital, without causing damage
totheenvironment, and ultimately shall stimulate economic growth, employment and competitiveness intheEU. Such assistance shall
best contribute toeconomic growth if itis targeted at a specific market failure, and thus shall complement, not replace, spending from
private sources. State aid shall contribute effectively totheachievement ofthe desired objectives ofpublic policy only if itprovides an
incentive effect, i.e. when thebeneficiary ofthe aid is induced totake actions which they would not have taken without such aid. Thebiggest
impact ofstate aid on economic growth shall occur only when itis constructed ina way that constrains distortions ofcompetition, and
theinternal market remains open and competitive.
According totheCommission, state aid control, if thequality and efficiency ofpublic support is duly emphasized, can also help Member
States improve budgetary discipline and improve thequality ofpublic finances and ultimately ensure a better use ofthe taxpayers
money. It is particularly important tobe able toperform intelligent fiscal consolidation; reconciling therole that targeted public spending
is expected toplay ingenerating economic growth with theneed tocontrol thebudget situation.
Modernization ofthe system ofstate aid control shall be achieved by:

Establishing uniform rules applicable totheassessment ofcompatibility ofall theaid with thecommon market (including,
inparticular, thedefinition and evaluation ofactual market failure, theincentive effect and thenegative effects ofpublic
intervention),

The revision and harmonization ofguidelines on state aid, so that they are consistent with these common principles.
TheCommission draws attention tothefact that therevision ofthe rules is tobe used primarily forstrengthening theinternal
market and increasing theefficiency ofexpenditure, aswell asa more accurate identification ofmarket failures, inrelation
towhich theaid is tobe addressed and a more detailed examination ofthe incentive effect. TheCommission announced
inthis Communication that theassessment ofstate aid should be carried out taking into account thefact that all its negative
and positive effects,

The prioritization ofcontrol ofaid and further focus on theassessment by theCommission ofsuch aid only that has
a significant impact on trade on theinternal market. Hence, themodernization ofthe control system is associated with
a change inregulations on block exemptions and de minimis aid,

Simplification and modernization ofprocedures under which theCommission shall study aid granted by theMember States.
Driven by thedemands delineated intheabove Communication, theCommission proposed amendments toCouncil Regulation
(EC) No 659/1999 ofMarch 22, 1999 laying down detailed rules fortheapplication ofArticle 108 ofthe Treaty on theFunctioning
ofthe European Union (Official Journal ofthe European Union (OJ) L 83, 27.3.1999, p. 1, asamended). These changes were
1

Brussels, 8.5.2012 COM (2012) 209 final.

81

adopted Council Regulation (EU) No 734/2013 ofJuly 22, 2013 amending Regulation (EC) No 659/1999 laying down detailed
rules fortheapplication ofArticle 93 ofthe EC Treaty (OJ L 204, 31.7.2013, p. 15). In addition, theCouncil issued Regulation (EU) No
733/2013 ofJuly 22, 2013 amending Regulation (EC) No 994/98 on theapplication ofArticles 92 and 93 ofthe Treaty establishing
theEuropean Community tocertain categories ofhorizontal state aid (OJ L 204, 31.7.2013, p. 11).
The Commission also adopted:

Regulation (EU) No. 1407/2013 ofDecember 18, 2013 on theapplication ofArticles 107 and 108 ofthe Treaty on
theFunctioning ofthe European Union tode minimis aid (OJ L 352, 24.12.2013, p. 1),

Guidelines on national regional aid for20142020 (OJ C 209, 23.7.2013, p. 1),

Union Guidelines on state aid topromote risk finance investment (OJ C 19, 22.1.2014, p. 4),

EU Guidelines fortheapplication ofstate aid rules inrelation torapid deployment ofbroadband networks (OJ C 25, 26.1.2013,
p. 1),

Guidelines on state aid toairports and airlines (OJ C 99, 4.4.2014, p. 3),

Guidelines on state aid forenvironmental protection and energy-related targets in20142020.2
In addition, theCommission is currently consulting with Member States drafts ofthe following legislation:

Regulation declaring certain categories ofaid compatible with thecommon market inapplication ofArticles 107 and 108
ofthe Treaty,

Communication from theCommission Criteria forthecompatibility analysis ofthe internal market state aid topromote
theexecution ofimportant projects ofcommon European interest,

The Framework forstate aid forresearch, development and innovation,

Guidelines on state aid forrescuing and restructuring ofnon-financial enterprises ina difficult situation,

Commission Notice on theconcept ofstate aid asprovided by Article 107, paragraph 1 ofTFEU.

6.2. Theconcept ofstate aid


Pursuant toArticle 107 paragraph 1 ofTFEU: Save asotherwise provided intheTreaties, any aid granted by a Member State
orthrough State resources inany form whatsoever which distorts orthreatens todistort competition by favouring certain undertakings
ortheproduction ofcertain goods shall, inso far asitaffects trade between Member States, be incompatible with theinternal market.
That provision establishes a general prohibition togrant aid by Member States. At thesame time, theprovisions ofthe Treaty admit
ofexceptions from theprohibition and thus constitute thebasis fortheadmissibility ofstate aid. Over theyears, theCommission
(and tosome extent theCouncil) adopted a number ofinstruments that specifically indicate what types (and categories) ofaid
may be granted by Member States within theexceptions totheprohibition.
Pursuant toArticle 107 paragraph 1 ofTFEU, theEU courts have developed a definition ofstate aid. It is assumed that aid constitutes
an advantage conferred: i) either by thestate orits resources that ii) gives therecipient a selective economic advantage, iii) distorts
orthreatens todistort competition and iv) affects theterms oftrade between Member States. For any state action tobe considered
state aid, all theabove conditions must occur jointly. Although theabove definition ofstate aid was developed decades ago, its
individual elements (conditions) are still subject todetailed analyses by theCommission and theEU courts. Thus, theinterpretation
ofArticle 107 paragraph ofTFEU is subject toconstant evolution.
First, granting aid entails state activity. Assistance may therefore be provided by central regional and local governments.3
Additionally, theprovisions ofArticle 107 paragraph 1 ofTFEU indicates that aid may be provided from state resources. This
distinction serves merely tohighlight that intervention through State resources was intended tocover, inaddition toadvantages
granted directly by theState, those granted through a public orprivate body4 appointed orestablished by that State toadminister
theaid.5 The concept ofstate resources refers toall funds by which public authorities might actually support undertakings, irrespective
ofwhether ornot those means were permanent assets ofthe public sector. Therefore, even if thesums corresponding tothemeasure
inquestion were not permanently held by theTreasury, thefact that they constantly remained under public control, and therefore available

2
3

Adopted by theCommission on April 9, 2014 (not yet published).


ECJ judgment of14October 1987 on C-248/84 Germany v Commission and CFI judgment of6 March 2002 inJoined Cases T-127/99, T-129/99
and T-148/99 Territorio Histrico de lava and Others. v Commission (Demes).
EGC Judgment of21 May 2010 inJoined Cases T-425/04, T-444/04, T-450/04, T-456/04, theFrench Republic, France Telecom SA, Bouygues SA,
Bouygues Telecom SA, AFORS Telecom v Commission.
ECJ Judgment of30 May 2013 on C-677/11 Doux Elevage SNC Cooperative agricole UKL ARREE against theMinistere de l Agriculture, de l
Alimentation, de la Peche, de la Ruralite et de lAmenagement du territoire, Comite interprofessionnel de la francaise dinde (CIDEF).

82

tothecompetent national authorities.6 At thesame time, incases where assistance is not provided by theadministrative authorities
ofthe state, itis necessary that assistance could be imputed tothestate.7
Second, state aid must entail a selective economic advantage. This means that itmust be primarily provided inconnection with
theeconomic activity ofthe beneficiary, which involves offering goods orservices on themarket.8 It is irrelevant whether such
activity is conducted forprofit.9 With reference toentities established inorder toconduct economic activity, EU competition law
uses theterm undertaking. Theterm includes every entity that offers goods orservices on themarket regardless ofits legal status
orhow itfinances its activities. Theconcept ofenterprise has a functional character. It means that even if themajority ofthe activities
ofa given entity involve projects ofother than economic nature and, therefore, itis not, asa whole, an undertaking, such entity may
nevertheless qualify asa one due its certain functions if they are economic innature.10 The status ofan undertaking does not depend
on how theentity inquestion allocates its profits (consequently, a non-profit organization may also qualify asan undertaking).11
The Commission notes that economic activity does not occur inconnection with theexercise ofpublic authority understood
asa task which is part ofthe essential functions ofthe state, orif itis associated with these functions by its nature, its aims and principles
towhich itis subject.12
In order forstate aid tooccur, an undertaking must receive an advantage from thestate. Theadvantage is defined asa benefit
which is not available under market conditions. Conferring advantages is, inpractice, thesubject ofdetailed analyses by theEU
Commission and theEU courts owing tothefact that Member States undertake a number ofactivities intheeconomic sphere
and often argue that their conduct should be considered identical toactions that can be expected on themarket on thepart
ofinvestors, creditors orpurchasers ofgoods orservices (Kaznowski 2013, p. 816). It should be emphasized that thekey factor is
theactual impact ofthe advantage on theundertaking, not theobjectives ofstate intervention.13
The absence ofan advantage, ortheability todemonstrate that thestate acts inthesame way asa market entity, is synonymous
with theabsence ofstate aid despite its action.
The said advantage must be selective. Theprovision ofArticle 107 paragraph 1 ofTFEU provides foraid favouring certain
undertakings ortheproduction ofcertain goods. Selective instruments are contrasted with general instruments. Thedistinction
inthis regard, however, is very complicated and inpractice theCommission rarely admits that a measure introduced by a Member
State has a general character (Kurcz 2009, p. 9&ff ).
Selectivity can be either territorial (geographical selectivity), sectoral, be related tothesize ofan undertaking (e.g. aid targeted
at small and medium-sized enterprises) (object-based selectivity) orderive from thediscretionary power ofan authority, which
does not mean, however, that theinstruments implemented on thebasis ofobjective criteria (based on theexistence ofobjective
circumstances) are general by definition (Van den Casteele, Hocine 2008, p. 247&ff ).
Conversely, general measures are those which are available toall undertakings operating ina given Member State on thesame
basis, provided, however, that they are not actually subject tolimitations (including e.g. sectoral ones, which, aswas indicated
above, may lead toselectivity).14 Determining whether a given instrument is selective requires theestablishment ofa reference
system, ascertaining whether thestudied instrument constitutes a deviation from thereference system, and whether thedeviation
is justified by thenature ofthe system. If theinstrument inquestion is found toconstitute such a deviation but its application is
found tobe justified by its nature, itis not be regarded asselective, and therefore does not constitute an instrument ofstate aid.15
6

10
11
12
13
14

15

ECJ judgment of16 May 2002 inCase C-482/99 French Republic v Commission. Cf. also theexplanation given by theCommission intheDraft
Commission Notice on thenotion ofstate aid pursuant toofArticle 107 paragraph 1 ofTFEU: Regulation which leads toa redistribution offunds
from one private entity toanother without any further involvement ofthe state, does not involve thetransfer ofstate resources, where there is a
direct transfer offunds from one private entity toanother, without theparticipation ofpublic orprivate entity designated by thestate forthepurpose
ofmanaging such transfer. For example, an obligation imposed by a Member State inprivate electricity suppliers topurchase energy produced from
renewable energy sources at fixed minimum prices does not involve direct orindirect transfer ofState resources toundertakings which produce that
kind ofelectricity. In this case, thecompanies (i.e. private electricity suppliers) are not appointed by thestate tomanage theaid program, but they are
only bound by theobligation topurchase a certain type ofenergy using its own financial resources.
Judgment ofthe ECJ of15 July 2004 inCase C-345/02 Pearle BV, Hans Prijs Optiek Franchise BV, Rinck Opticiens BV v Hoofdbedrijfschap
Ambachten and theECJ judgment of16 May 2002 inCase C-482/99 French Republic v Commission.
ECJ judgment of23 April 1991 on C-41/90 Klaus Hfner and Fritz Elser v Macrotron GmbH, theECJ judgment of22 January 2002 on C-218/00
Cisal di Battistello Venanzio & C. Sas v Istitutonazionale per lassicurazione contro gliinfortunisullavoro (INAIL), thejudgment ofthe ECJ of12
September 2000 inJoined Cases C-180/98 toC-184/98 Pavel Pavlov et al. v Stichting Pensioenfonds Medische Specialisten, and thejudgment
ofthe ECJ of26 March 2009 on C-113/07 P Selex Sistemi Integrati v Commission.
Judgment ofthe ECJ of21 September 1999 on C-67/96 Albany International BV v Stichting Fonds Bedrijfspensioen Textielindustrie, ECJ,
judgment of1 July 2008 on C-49/07 MOTOE.
European The Commissions Decision inCase N 470/ 08 Poland Aid forrevitalization ofdegraded areas inPoland.
Judgment ofthe ECJ of11 July 2006 inCase C-205/03 P Federacin Espaola de Empresas de Tecnologa Sanitaria (FENIN) v Commission.
Draft Commission Notice on thenotion ofstate aid pursuant toArticle 107 paragraph 1 ofTFEU.
ECJ Judgment of17 July 2008 on C-206/06 Essent Netwerk Noord.
ECJ judgment of8 November 2001 inCase C-143/99 Adria-Wien Pipeline GmbH and Wietersdorfer & Peggauer Zementwerke GmbH v
Finanzlandesdirektion fr Krnten.
Draft Commission Notice on thenotion ofstate aid pursuant toArticle 107 paragraph 1 ofTFEU.

83

Selective advantage conferred by theState must distort orthreaten todistort competition and affect trade. In theCommissions
decision-making practice, thedistortion ofcompetition and theimpact on trade are very often considered together. This is due
tothefact that thevast majority ofcases they are actually very closely interlinked, moreover, if an advantage distorts competition,
italso implies that ithas an impact on trade. This does not mean, however, that there are no cases inwhich theCommission finds
that a distorting action taken by a given state is limited totheMember State granting theaid and therefore has no impact on trade
asprovided by theTFEU. In such cases, where despite thefinding that aid distorts ormay distort competition, itis decided that ithas
no impact on theterms oftrade between Member States, state aid asprovided by Article 107 paragraph 1 ofTFEU does not occur.16
State aid distorts competition when itleads (or may lead) tofavouring an undertaking (beneficiary) at theexpense ofits
competitors. What itmeans inpractice is that if a state grants a selective economic advantage toan undertaking operating
ina liberalized sector, thecondition ofdistortion orthreat ofdistortion ofcompetition is met.
Actual orpotential advantages accorded toa beneficiary undertaking over its competitors may be related todifferent spheres ofits
activities depending on thepurpose ofthe aid. Therefore, itmay result from thefact that thebeneficiary (unlike its competitors)
does not have toincur all expenses related toinvestment, modernization and restructuring processes, development orexpansion
oreven (although itis possible only toa limited extent) its current operation.17 In principle, thesize ofthe advantage is immaterial.18
With regard tothedistortion oftrade itshould be noted that: When aid granted by a Member State strengthens theposition ofan
undertaking compared with other undertakings competing inintra-Community trade, itis considered that theaid has an impact on this
trade.19 It is necessary toemphasize that thedistortion oftrade may also apply toundertakings which: i) are not involved incrossborder exchange, ii) conduct only local activity, iii) only export tocountries outside theEU.20
As was already mentioned, theCommission accepts, however, that incertain circumstances no effect ofaid on trading conditions
is present. This occurs inparticular if itcan be demonstrated that theeconomic activity inquestion is only relevant totheterritory
(region) ofa single state (the services orgoods offered attract no interest inother countries), orif itis shown that theaid does not
lead tomarket closure. For example, theCommission found no impact ofstate aid on trading conditions inrelation totheatrical
productions,21 theactivities ofhospitals,22 museums,23 cultural infrastructure24 orconference centres.25

6.3. Principles ofstate aid intheEU


As was mentioned above, state aid is inprinciple inadmissible. However, thetreaties provide forexceptions tosuch prohibition,
which along with associated conditions oftheir use, constitute therules ofadmissibility ofstate aid. Such rules, developed
inthedecision-making practice ofthe Commission, are contained primarily intheguidelines, communications and framework
principles. They refer tocategories ofstate aid, theadmissibility ofwhich ensues from theprovisions ofthe Treaty. It should be
mentioned that theCommission has not issued relevant documentation governing theadmissibility ofaid with respect toall its
kinds ofaid authorized under thetreaty provisions (i.e. excluded from theban). In such cases, Member States shall apply therules
arising from theCommissions decision-making practice.
The table below contains a list ofTreaty provisions on theadmissibility ofstate aid and therelevant Commission documents further
specifying its conditions.
Treaty provision
Article 107 paragraph 2 a) TFEU:
aid having a social character, granted
toindividual consumers

Guidelines/communications/frameworks
Guidelines on state aid toairports and airlines (OJ C 99, 4.4.2014, p. 3).

Article 107 paragraph 2 b) TFEU:


aid tomake good thedamage caused by
natural disasters orexceptional occurrences

16

17

18
19
20

21
22
23
24
25

Cf. theCommissions decisions inthefollowing cases: SA.34466 (2012/N) Cyprus State support totheCentre forVisual Arts and Research,
SA.34576 (2012/N) Portugal North East oflong-term care unit ofJean Piaget and SA.34891 (2012/N) Poland state aid totheAssociation
ofMunicipalities ofthe Przemyl Fortified Area.
The reason is that aid tocover operating costs is generally inadmissible. To a very limited extent itis permitted only where justified by thenature
ofa business and its attendant factors. See e.g. records intheGuidelines on state aid toairports and airlines (not yet published).
Judgment of24 July 2003 inCase C-280/00 Altmark Trans GmbH, Regierungsprsidium Magdeburg v Nahverkehrsgesellschaft Altmark GmbH.
Judgment of17 September 1980 inCase 730 /79 Philip Morris Holland BV v Commission ofthe European Communities.
The Commission discusses theissue indetail intheDraft Commission Notice on thenotion ofstate aid pursuant toofArticle 107 paragraph 1
ofTFEU, p. 52&ff.
The Commissions Decision inCase N 257/2007 Spain Grants fortheatrical productions intheBasque country.
The Commissions decision SA.34576 Portugal North-eastern long-term care unit ofJean Piaget.
The Commissions Decision inCase N 630/2003 Italy Sardinia Local Museums.
The Commissions decision SA.34891 (2012/N) Poland state aid totheAssociation ofMunicipalities ofthe Przemyl fortified area.
The Commissions Decision inCase N 486/2002 Sweden congress hall inVisby.

84

Treaty provision

Guidelines/communications/frameworks

Article 107 paragraph 2 c) TFEU:


aid granted totheeconomy ofcertain areas
ofthe Federal Republic ofGermany affected
by thedivision ofGermany

Article 107 paragraph 3 a) TFEU:


aid topromote theeconomic development
ofareas with economic orsocial problems

Guidelines on national regional aid for2014-2020


EU Guidelines fortheapplication ofstate aid rules inrelation torapid deployment
ofbroadband networks (OJ C 25, 26.1.2013, p. 1)
The Framework on state aid forshipbuilding (OJ C 364, 14.12.2011, p. 9)

Article 107 paragraph 3 b) TFEU:


aid topromote theexecution ofan
important project ofcommon European
interest ortoremedy a serious disturbance
intheeconomy ofa Member State

Communication from theCommission Criteria forthecompatibility analysis ofthe internal


market state aid topromote theexecution ofan important project ofcommon European
interest (draft)
Community framework forstate aid forresearch, development and innovation (OJ C 323,
30.12.2006, p. 1)1
Communication from theCommission on theapplication, from 1 August 2013, ofstate aid
rules tosupport measures infavour ofbanks inthecontext ofthe financial crisis (Banking
Communication), (OJ C 216, 30.7.2013, p. 1)

Article 107 paragraph. 3 c) TFEU:


aid tofacilitate thedevelopment ofcertain
economic activities orofcertain economic
areas

Guidelines on national regional aid for2014-2020


Community framework forstate aid forresearch, development and innovation
Communication from theCommission Criteria forthecompatibility analysis ofstate
aid toworkers indisadvantaged and disabled workers subject toindividual notification
(OJ C 188, 11.8.2009, p. 6)
Communication from theCommission Criteria forthecompatibility analysis ofstate aid
fortraining subject toindividual notification (OJ C 188, 11.8.2009, p. 1)
Guidelines on environmental and energy State aid for201420202
Guidelines on certain state aid measures inthecontext ofthe greenhouse gas emission
allowance trading scheme post-2012 (OJ C 158, 5.6.2012, p. 4)
EU Guidelines on state aid topromote risk finance investment (OJ C 19, 22.1.2014, p. 4)
Community guidelines on state aid forrescuing and restructuring firms indifficulty
(OJ C 244, 1.10.2004, p. 2, asamended)3
EU Guidelines fortheapplication ofstate aid rules inrelation torapid deployment
ofbroadband networks (OJ C 25, 26.1.2013, p. 1)
The Framework on state aid forshipbuilding (OJ C 364, 14.12.2011, p. 9
Guidelines on state aid toairports and airlines (OJ C 99, 4.4.2014, p. 3)
Community guidelines on state aid tomaritime transport (OJ C 13, 17.1.2004, p. 3)
Communication from theCommission providing guidance on state aid complementary
toCommunity funding forthelaunching ofthe motorways ofthe sea (OJ C 317, 12.12.2008, p. 10)
Communication from theCommission providing guidance on state aid toship management
companies (OJ C 132, 11.6.2009, p. 6)

Article 107 paragraph 3 d) TFEU:


aid topromote culture and heritage
conservation

Commission Communication on state aid tofilms and other audio-visual works (OJ C 332,
15.11.2013, p. 1)
Commission Communication on theapplication ofstate aid rules topublic service
broadcasting (OJ C 257, 27.10.2009, p. 1)

Article 107 paragraph 3 e) TFEU:


other categories ofaid

Council Decision of10 December 2010 on state aid tofacilitate theclosure ofuncompetitive
coal mines (OJ L 336, 21.12.2010, p. 24)

Article 93 TFEU

Council Regulation (EC) No 1370/2007 ofthe European Parliament and ofthe Council of23
October 2007 on public passenger transport services by rail and by road and repealing
Council Regulations (EEC) Nos 1191/69 and 1107/70 (OJ L 315, 31.12.2007, p. 1)
Community guidelines on state aid forrailway undertakings (OJ C 184, 22.7.2008, p. 13)

Article 106 paragraph 2 TFEU

Commission Communication on theapplication ofEU rules on state aid inrelation


tocompensation forservices ofgeneral economic interest (OJ C 8, 11.1.2012, p. 4)
Commission Decision of20 December 2011 on theapplication ofArticle 106 paragraph 2
ofthe Treaty on theFunctioning ofthe European Union tostate aid intheform ofpublic
service compensation granted tocertain undertakings entrusted with theoperation
ofservices ofgeneral economic interest (OJ L 7, 11.1.2012, p. 3)
The framework ofthe European Union state aid intheform ofpublic service compensation
(2011) (OJ C 8, 11.1.2012, p. 15)
EU Guidelines fortheapplication ofstate aid rules inrelation torapid deployment
ofbroadband networks (OJ C 25, 26.1.2013, p. 1)
Commission Communication on theapplication ofstate aid rules topublic service
broadcasting (OJ C 257, 27.10.2009, p. 1)
Commission Notice on theapplication ofcompetition rules tothepostal sector and on
theassessment ofcertain State measures relating topostal services (OJ C 39, 6.2.1998, p. 2)
Guidelines on state aid toairports and airlines (OJ C 99, 4.4.2014, p. 3)
Community guidelines on state aid tomaritime transport (OJ C 13, 17.1.2004, p. 3)

Thedocument is valid until 30 June 2014 and will be replaced by a new framework.
Adopted by theCommission on 9 April 2014 (not yet published).
3
Thedocument will be replaced by new guidelines.
1
2

85

The rules governing theadmissibility ofstate aid also include therules laid down forspecific aid instruments contained particularly
inthefollowing documents:

Commission Notice on theapplication ofArticles 87 and 88 ofthe EC Treaty tostate aid intheform ofguarantees (OJ C 155,
20.6.2008, p. 10),

Commission Communication on state aid elements insales ofland and buildings by public authorities (OJ ofthe EC C 209,
10.1997, p. 3),

Communication from theCommission totheMember States on theapplication ofArticles 107 and 108 ofthe Treaty on
theFunctioning ofthe European Union toshort-term export-credit insurance (OJ C 392, 19.12.2012, p. 1),

Commission Notice on theapplication ofstate aid rules tomeasures relating todirect business taxation (OJ ofthe EC C 384,
10.12.1998, p. 3),

Commission Communication on therevision ofthe method forsetting thereference and discount rates (OJ C 14, 19.1.2008,
p. 6).
Apart from those acts and documents devoted totheadmissibility ofstate aid rules, EU law also regulates theprocedural issues
related tothegranting ofstate aid by Member States. As a matter ofprinciple, state aid (within theframework defined by exceptions
totheban) requires theconsent ofthe Commission.
Article 108 paragraph 3 TFEU provides that theMember States shall not implement proposed aid until theCommission gives its
approval. In consequence, theEU aid law contains thefollowing legal acts governing theprocedure through which Member States
shall notify theCommission oftheir aid plans:26

Council Regulation (EC) No 659/1999 of22 March 1999 laying down detailed rules fortheapplication ofArticle 108 ofthe
Treaty on theFunctioning ofthe European Union (OJ ofthe EC L 83, 27.3.1999, p.1, asamended),

Commission Regulation (EC) No 794/2004 of21 April 2004 implementing Council Regulation (EC) No 659/1999 laying down
detailed rules fortheapplication. 93 ofthe EC Treaty (OJ L 140, 30.4.2004, p.1, asamended).
Not every kind ofaid, however, requires notification asprovided by article 108 paragraph 3 TFEU. Acting on thebasis ofArticle
94 ofthe Treaty establishing theEuropean Community (now Article 109 TFEU), theCouncil adopted Regulation No 994/98 of7
May 1998 on theapplication ofArticles 92 and 93 ofthe Treaty establishing theEuropean Community tocertain categories
ofhorizontal state aid (OJ ofthe EC L 142, 14.5.1998, p.1, asamended). It authorizes theCommission toadopt regulations specifying
thecategories ofaid that are compatible with thecommon market and not subject tothenotification requirements ofArticle 108
paragraph 3TFEU (block exemption), and also by means ofappropriate regulations, tostipulate that certain categories ofaid do
not meet all thecriteria under Article 107 paragraph 1 TFEU (evidence ofstate aid) and, therefore, are exempt from thenotification
requirement ofArticle 108 paragraph 3 TFEU, provided that theaid granted tothesame undertaking over a given period does
not exceed a certain fixed amount (de minimis). On thebasis ofthis mandate, theCommission adopted theexisting Regulation
No 800/2008 of6 August 2008 declaring certain categories ofaid compatible with thecommon market pursuant toArticles 87
and 88 ofthe Treaty (General Block Exemption Regulation), (OJ L 214, 9.8.2008, p. 3, asamended)27 Commission Regulation No.
1407/2013 of18 December 2013 on theapplication ofArticles 107 and 108 ofthe Treaty on theFunctioning ofthe European Union
tode minimis aid (OJ L 352, 24.12.2013, p. 1), aswell asRegulation No 360/2012 of25 April 2012 on theapplication ofArticles 107
and 108 ofthe Treaty on theFunctioning ofthe European Union tode minimis aid granted toundertakings providing services
ofgeneral economic interest (OJ L 114, 26.4.2012, p. 8).

6.4. Domestic regulations


In Poland, theissues at hand are regulated by theProcedural Issues Pertaining toState Aid Act of30 April 2004, including therules
ofconduct inmatters concerning state aid, thenotification procedure, therules ofcooperation between thePresident ofthe UOKiK28
(Office forCompetition and Consumer Protection, OCCP) and theinstitutions developing aid programmes, aid donors, entities
applying forid and beneficiaries ofsuch aid aswell astheprinciple ofrepresentation ofRepublic ofPoland before theEuropean
Court ofJustice (ECJ) and theEuropean Court ofFirst Instance (ECFI), aid refund and monitoring therules and procedures.
It should be emphasized that theAct inquestion does not lay down therules ofaid admissibility since they are defined inEU legislation.
TheAct sets out theprocedures necessary fortheproper implementation ofthe provisions ofEU law, especially inthecontext

26

27
28

In addition, thefollowing documents have been adopted: theCommission Notice on a simplified procedure fortreatment ofcertain types
ofstate aid (OJ. EU C 136, 16.6.2009, p. 3) and theCode ofBest Practice fortheconduct ofstate aid control procedures (OJ C 136, 16.6.2009,
p.13).
Theregulation will expire on 30 June 30 2014 and will be replaced by a new one concerning state aid covered by theblock exemption.
The Act also defines thepowers ofthe minister responsible foragriculture inmatters concerning state aid. Theissue ofstate aid inagriculture
remains beyond thescope ofthis chapter.

86

ofthe requirement toinform theCommission ofplans togrant aid inaccordance with Article 108 paragraph 3 TFEU (notification
requirement). Thenotification requirement applies toaid schemes29 orindividual aid (i.e. remaining outside aid programmes).30
In thenotification procedure, thePresident ofthe UOKiK issues opinions regarding theplanned state aid, i.e. with regard toplanned
aid schemes aswell asindividual aid projects. In theopinions, thePresident takes a stand regarding thequestion ofwhether
theproject actually constitutes state aid, and whether theaid is compatible with theinternal market. In thecase ofan aid scheme,
such notification requires theconsent ofthe Council ofMinisters. Thepre-notification procedure thus designed is aimed at
thedevelopment ofan appropriate notification form toensure that itcomplies asfully aspossible with therules applied by
theCommission inits assessment ofadmissibility ofstate aid.
The draft aid scheme orindividual aid project is then be forwarded by thePresident ofthe UOKiK tothePermanent Representation
ofthe Republic ofPoland totheEU inBrussels, which officially submits itto theCommission. Notification occurs inelectronic form
(by means ofthe State Aid Notification Interactive, SANI).
The President ofthe UOKiK is thecompetent authority inthearea ofproceedings before theCommission. Accordingly, therelevant
stakeholders, including public authorities, aswell asapplicants foraid, submit tothePresident ofthe UOKiK information necessary
todraw up explanations, comments ordefine thejoint position on issues related toquestions asked by theCommission
inthecourse ofproceedings. Using theICT network, thePresident publicizes information concerning thedecisions made by
theCommission Decisions related totheaid instruments subject tonotification.
The President ofthe UOKiK is also responsible forrepresenting theRepublic ofPoland inall aid-related proceedings before
theCourt ofJustice ofthe European Union and theGeneral Court ofthe European Union.
The President ofthe UOKiK monitors all state aid granted inPoland. For that reason, donors are required tosubmit tothePresident
all information on assistance provided. ThePresident draws up an annual report concerning theaid granted inPoland and presents
itto theCouncil ofMinisters.
The Procedural Issues Pertaining toState aid Act also sets theframework rules related tothescope ofinformation submitted by
applicants forstate aid orforde minimis aid.
Second, Polands domestic legal provisions related tostate aid include those that provide direct legal basis forgranting aid. Such
regulations may be contained inthealready mentioned aid programmes. In accordance with Article 2 item 7 ofthe Procedural
Issues Pertaining toState Aid Act, an aid scheme is a normative act that meets theconditions set out inArticle 1 d) ofthe Council
Regulation (EC) No 659/1999 of22 March 1999 laying down detailed rules fortheapplication ofArticle 108 ofthe Treaty on
theFunctioning ofthe European Union (OJ L 83, 27.3.1999, p.1, asamended), and therefore, inparticular, a normative act on
thebasis ofwhich funds may be disbursed toindividual undertakings defined intheAct ina general and abstract manner without
further implementation measures being required. This means that an aid programme is a normative act on thebasis ofwhich state
aid may be granted aid, and which specifies:
i. The beneficiaries ofsuch aid e.g. small and medium-sized enterprises,
ii. A body (entity) providing theaid, e.g. thePolish Agency forEnterprise Development,
iii. The purpose ofthe aid (what theaid is granted for), e.g. investment, training, consultancy services etc.,
iv. The form inwhich theaid is granted, e.g. a grant, a preferential loan, a loan guarantee,
v. The amount ofaid, most often defined with respect totheso-called eligible costs, e.g. 50% ofthe eligible costs, such
asthecost ofpurchasing consultancy services,
vi. The principles governing theaccumulation ofaid received from various sources,
vii. The aid application procedure.
An aid programme may be based on theprovisions ofseveral regulations (e.g. an act orseveral acts with their attendant enactment
and implementation provisions). In thePolish system ofstate aid, such schemes involve primarily theaid granted under operational
programmes, which specify support areas under theStructural Funds. Currently, Article 21 item 1 ofthe Development Policy
Principles Act of6 December 2006 (Journal ofLaws No. 227, item 1658, asamended) provides that totheextent towhich thefunding
ofprojects under programmes constitutes aid that meets theconditions set out inArticle 87 paragraph 1 ofthe Treaty establishing
theEuropean Community [now Article 107 paragraph 1 ofTFEU] orde minimis aid, such funding shall be governed by specific conditions
and procedures forgranting aid. In addition, paragraph 3 ofthe Article provides that thecompetent minister ortheminister responsible
29

30

Pursuant toArticle 2 item 7 ofthe Procedural Issues Pertaining toPublic Aid Act, an aid programme is normative act on thebasis ofwhich funds
may be disbursed toindividual undertakings defined intheAct ina general and abstract manner without further implementation measures
being required aswell asany act on thebasis ofwhich an indefinite amount ofaid that is not linked toany specific project may be granted
toone ormore undertakings forand/or foran indefinite period oftime.
Individual aid asprovided by Article 2 ofRegulation No 800/2008 of6 August 2008 declaring certain categories ofaid compatible with
thecommon market inapplication ofArticles 87 and 88 ofthe Treaty (General Block Exemption) (OJ L 214, 9.8.2008, p. 3, asamended)
constitutes ad hoc aid (aid granted outside aid schemes) and notifiable aid.

87

forregional development interms ofoperational programmes under which thevoivodship [regional] government serves asthemanaging
authority orintermediate body, shall determine, by way ofstatutory instruments, thespecific purpose, conditions and procedures
forgranting aid referred toin paragraph 1, especially bearing inmind theneed toensure compatibility ofthe aid with theconditions
ofadmissibility, where separate regulations do not specify thedetailed conditions and procedures forgranting theaid. Accordingly, such
secondary legislation constitutes aid programmes asprovided by theProcedural Issues Pertaining toState Aid Act pursuant toEU
law.
Other than this, state aid derived exclusively from thedomestic budget is also granted on thebasis ofaid programmes (see e.g.
theSpecial Economic Zones Act of20 October 1994) (Journal ofLaws 2007, No. 42, item. 274, 2008, No. 118, item 746, 2009, No.18,
item 97).
As was indicated above, theprovisions governing state aid inPoland will soon be subject tomodification due totheextensive
changes inEU law setting out theconditions fortheadmissibility ofsuch aid.
EU regulations and domestic legal provisions under which aid is granted may serve asa state policy tool aimed at preventing
bankruptcies. Thesystem contains no documents orlegislation that would expressly indicate theadmissibility ofaid forpreventing
bankruptcy (see thetabulated overview insection 6.3). EU law provides, however, inparticular fortheadmissibility ofthe following
kinds ofaid: de minimis aid forrescuing and restructuring, training and consultancy services. Such aid may be inline with
theadopted policy focused exclusively (or among other things) on respond toproblems offirms which find themselves inbroadly
defined difficulties orare undertakings indifficulty asprovided by thecurrent Community guidelines on state aid forrescuing and
restructuring offirms indifficulty (OJ C 244, 1.10.2004, p. 2, asamended).31

Abbreviations
TFEU Treaty on theFunctioning ofthe European Union
The Commission theEuropean Commission
ECJ European Court ofJustice
SPI theCourt ofFirst Instance
EGC The Court ofthe European Union
ECJ The Court ofJustice ofthe European Union

31

The guidelines provide that a firm is considered tobe indifficulty where itis unable, whether through its own resources orwith thefunds itis
able toobtain from its owner/shareholders orcreditors, tostem losses which, without outside intervention by thepublic authorities, will almost
certainly condemn itto going out ofbusiness intheshort ormedium term (point 9 ofthe Guidelines).
In particular, a firm is considered tobe indifficulty if at least one ofthe following circumstances occurs (point 10 ofthe Guidelines):
a) in thecase ofa limited liability company where more than half ofits registered capital has disappeared and more than one quarter ofthat
capital has been lost over thepreceding 12 months, or
b) in thecase ofa company where at least some members have unlimited liability forthedebts ofthe company, where more than half ofits
capital asshown inthecompany accounts has disappeared and more than one quarter ofthat capital has been lost over thepreceding
twelve months, or
c) whatever thetype ofcompany concerned where itfulfils thecriteria under its domestic law forbeing thesubject ofcollective insolvency
proceeding.
Even when none ofthe circumstances set out inpoint 10 are present, a firm may still be considered tobe indifficulties, inparticular where
theusual signs ofa firm being indifficulty are present, such asincreasing losses, diminishing turnover, growing stock inventories, excess
capacity, declining cash flow, mounting debt, rising interest charges and falling ornil net asset value (point 11 ofthe Guidelines).

88

Mikoaj Stasiak

Chapter 7
THE FORMS AND INSTRUMENTS OF STATE AID
7.1. Introduction
According tothepreferences expressed by theCommission inits Communication totheEuropean Parliament, theCouncil
ofEurope, theEuropean Economic and Social Committee and theCommittee ofthe Regions The modernization ofEU state aid
policy,1 one ofthe objectives ofthe reform ofthe rules governing public assistance is tofocus theCommissions ex-ante scrutiny
on cases with thebiggest impact on theinternal market aswell asto streamline therules and provide forfaster decisions.
The Commission thereby intends that thescope ofits control be even more limited than before and that an increasing scope
ofstate aid should not require such ex-ante scrutiny and, therefore, notification by theMember States. Accordingly, theCommission
extends theinventory ofstate aid targets that will be exempted from thenotification requirement referred toin Article 108
paragraph 3 TFEU.
Due totheneed forcontinued focus by theCommission on themost important issues forthesingle market, theimportance ofde
minimis aid and aid covered by theblock exemption increases. It is assumed that inthenear future this kind ofsupport shall be
themost frequently provided by theMember States.

7.2. De minimis aid and block exemption regulations


As was mentioned above (cf. Section 6.3 above), itshould be noted that, acting under theauthority ofthe Council, theCommission
has adopted:
i) De minimis regulations,
ii) Regulation No 360/2012 of25 April 2012 on theapplication ofArticles 107 and 108 ofthe Treaty on theFunctioning ofthe
European Union tode minimis aid granted toundertakings providing services ofgeneral economic interest (OJ L 114, 26.4.2012,
p. 8.), aswell as
iii) Regulation No 800/2008 of6 August 2008 declaring certain categories ofaid compatible with thecommon market
inapplication ofArticles 87 and 88 ofthe Treaty (General Block Exemption), (OJ L 214, 9.8.2008, p. 3, asamended).2
De minimis aid is defined asaid which does not fulfil all thecriteria under Article 107 paragraph 1 TFEU (aid awarding conditions)
and therefore is exempt from thenotification requirement ofArticle 108 Paragraph 3 TFEU, due tothefact that aid granted
tothesame undertaking over a given period does not exceed a certain fixed amount.3
Pursuant toArticle 3 item 1 ofthe De minimis Regulation, itwas decided that aid funds do not meet all thecriteria set out inArticle
107 paragraph 1 TFEU and are therefore exempted from thenotification requirement provided that they are granted inaccordance
with theRegulation.
First, thetotal amount ofde minimis aid granted toany one undertaking shall not exceed EUR 200,000 (EUR 100 000 foran
undertaking inthefield ofroad freight ofgoods and may not be used fortheacquisition ofvehicles fortheroad freight ofgoods)
over any period ofthree fiscal years.4 The ceilings ofthe de minimis rule shall apply regardless ofthe form (e.g. grant, loan guarantee)
and theobjective pursued (such asinvestment activities, operations, etc.) and regardless ofwhether theaid granted by theMember
state is financed entirely orpartly by resources ofUnion origin. Theceilings applicable tothede minimis rule shall be expressed
interms ofgross cash grant. Where aid is awarded ina form other than a grant, theaid amount shall be its gross grant equivalent.5

1
2
3

Brussels, 8.5.2012 COM (2012) 209 final.


Theregulation will expire on 30 June 30 2014 and will be replaced by a new one on state aid covered by block exemption.
Cf. Regulation No. 994/98 of7 May 1998 on theapplication ofArticles 107 and 108 ofthe Treaty on theFunctioning ofthe European Union
tocertain categories ofhorizontal state aid.
If an undertaking performs road freight transport forhire orreward and other activities towhich theceiling ofEUR 200 000 applies, theceiling
ofEUR 200 000 shall apply totheundertaking, provided that theMember States ensure by appropriate means such asseparation ofactivities
ordistinction ofcosts that thebenefit totheroad freight transport activity does not exceed EUR 100 000 and that no de minimis aid is used
fortheacquisition ofroad freight transport vehicles.
Pursuant to2 item 1) ofthe Regulation ofthe Council ofMinisters of11 August 2004 on thedetailed rules forcalculating theaid granted
invarious forms (Journal ofLaws No. 194, item. 1983, asamended), thegross grant equivalent shall be thegross amount ofaid received by
theaid recipient ortheapplicant entity if received aid intheform ofa grant without regard toincome tax expressed accurate totwo decimal
points.

89

Pursuant toArticle 6 ofthe De minimis Regulation, theMember States are obliged toinstitute a mechanism toensure that specified
ceilings forde minimis aid shall be observed ineach case. Thus, theappropriate authority intheMember state shall inform
thebeneficiary inwriting ofthe estimated amount ofaid (expressed asgross grant equivalent) and ofits de minimis character,
making express reference tothede minimis Regulation. Prior togranting theaid, thecompetent authority shall also obtain
a statement from theundertaking concerned, inwritten orelectronic form, about any other de minimis aid received towhich this
Regulation orother de minimis regulations apply during theprevious two fiscal years and thecurrent fiscal year. In implementing
this obligation, pursuant toArticle 5 paragraph 3 ofthe Procedural Issues Pertaining tostate Aid Act, bodies awarding de minimis
aid issue tothebeneficiary an official statement declaring that theaid obtained meets thede minimis conditions. According
toArticle 37 ofthe Act, an entity applying forde minimis aid is obliged tosubmit totheaid granting body certificates confirming
theaward ofde minimis aid, a statement ofthe amount ofde minimis aid, ora statement ofnon-receipt ofsuch aid.
Second, de minimis aid must be transparent innature. Article 4 item 1 ofthe De minimis Regulation provides that itshall apply
only toaid inrespect ofwhich itis possible tocalculate precisely thegross grant equivalent ofthe aid ex ante without need toundertake
a risk assessment (transparent aid).
For obvious reasons, transparent aid is awarded intheform ofgrants.
De minimis aid tobe provided intheform ofloans is considered transparent aid if itmeets thefollowing criteria:
i. Thebeneficiary is not subject tocollective insolvency proceedings ordoes not meet theappropriate domestic law criteria
regarding collective insolvency proceedings at thebehest ofits creditors. Beneficiaries who are large companies must be
ina situation comparable toat least B credit rating, and
ii. Theloan does not exceed EUR 1 000 000 (or EUR 500 000 forundertakings performing road freight transport), has a duration
not exceeding five years and is secured by collateral covering at least 50% ofthe loan. If a loan only accounts forless than
this amount and/or is granted fora period ofless than five years, thegross grant equivalent ofthat loan shall be calculated
asa proportion ofthe applicable ceiling (200000 EUR of100000 EUR), or
iii. Gross grant equivalent has been calculated on thebasis ofthe reference rate applicable at thetime ofthe grant.6
Aid intheform ofcapital injection is considered transparent de minimis aid, if thetotal amount ofthe public injection does not
exceed thede minimis ceiling.
Aid comprised inrisk finance measures shall not be considered astransparent de minimis aid, unless themeasure concerned
provides capital not exceeding thede minimis ceiling toeach target undertaking.
Aid comprised inguarantees shall be treated astransparent de minimis aid if:
i. The beneficiary is not subject tocollective insolvency proceedings ordoes not meet theappropriate domestic law criteria
regarding collective insolvency proceedings at thebehest ofits creditors. Beneficiaries who are large companies must be
ina situation comparable toat least B credit rating, and
ii. The guaranteed part ofthe underlying loan does not exceed EUR 1 500 000 (or EUR 750 000 forundertakings performing road
freight transport) and theduration ofthe guarantee does not exceed five years. If theguaranteed part ofthe underlying loan
only accounts forless than this amount and/or theguarantee is fora period ofless than five years, thegross grant equivalent
ofthat guarantee shall be calculated asa proportion ofthe applicable ceiling (EUR 200 000 orEUR 100 000), or
iii. The gross grant equivalent has been calculated on thebasis ofsafe-harbour premiums laid down intheCommission Notice
on theapplication ofArticles 87 and 88 ofthe EC Treaty tostate aid intheform ofguarantees (OJ C 155, 20.6.2008, p. 10), or
iv. Before being implemented, themethodology tocalculate thegross grant equivalent ofthe guarantee has been accepted
following notification ofthis methodology totheCommission under a regulation adopted by theCommission inthestate aid
area, and theapproved methodology explicitly addresses thetype ofguarantee and thetype ofunderlying transaction at stake
inthecontext ofthe application ofthis Regulation.
Aid comprised inother instruments shall be considered astransparent de minimis aid on condition that theinstrument provides
fora cap ensuring that theapplicable ceiling is not exceeded.
Third, we need toanalyse thecumulation rules contained inArticle 5 ofthe De minimis Regulation. Their purpose is primarily
toensure that thetotal de minimis aid granted toone company does not exceed therelevant ceilings (EUR 200 000 and EUR
100000, respectively). De minimis aid pursuant totheRegulation may thus be cumulated with theCommission Regulation (EU)
No 360/2012 of25 April 2012 on theapplication ofArt. 107 and 108 ofthe Treaty on theFunctioning ofthe European Union tode
minimis aid granted toundertakings providing services ofgeneral economic interest (OJ L 114, 26.4.2012, p. 8). This means that
ina period ofthree years (calculated astheprevious two fiscal years and thecurrent fiscal year) an undertaking may receive de
6

In such a case, therules arising from theCommissions Communication on therevision ofthe method forsetting thereference and discount
rates (OJ C 14, 19.1.2008, p. 6) shall apply, which states that thereference rate corresponding tothebase rate and theapplicable margin can be
established foreach individual case.

90

minimis aid from a number ofsources, provided that thecumulated de minimis aid within that time the(expressed interms ofthe
gross grant equivalent) does not exceed EUR 200 000 or100000, respectively.
Cumulative rules also apply tode minimis aid combined with state aid. Therefore, de minimis aid cannot be cumulated with state
aid inrespect ofthe same eligible costs, orwith thestate aid forthesame measure ofrisk financing, if such cumulation would result
inan aid intensity exceeding that fixed inthespecific circumstances ofeach case by a block exemption regulation ordecision
adopted by theCommission. De minimis aid, which has not been granted inrelation tospecific eligible costs, orwhich cannot be
attributed tosuch costs, can be combined with other state aid granted inaccordance with theRegulation on Block Exemption
orinaccordance with thedecision made by theCommission.
Accordingly, any applicant forde minimis aid must provide information not only on de minimis aid obtained so far, but also any
public aid if de minimis aid is tobe granted tocover thesame eligible costs which have already covered by theprevious aid, orif
is tocomplement theaid fixed intheRegulation on block exemption ortheCommissions decision.
Assistance under thecurrent de minimis Regulation may be granted inprinciple, regardless ofthe financial situation ofthe
beneficiary (c.f. however, theprovisions relating toaid intheform ofloans and guarantees). Thus, therecipient ofde minimis aid
can also be a company indifficulty asprovided by theCommunity guidelines on state aid forrescuing and restructuring firms
indifficulty (OJ C 244, 1.10.2004, p. 2, asamended), forwhich until now only rescue and restructuring aid has generally been
available. Such a change is not without significance fortheEU bankruptcy prevention policy.
As already mentioned, theCommission was authorized by theCouncil toadopt regulations specifying thecategories ofaid which
are compatible with thecommon market and not subject tothenotification requirements ofArticle 108 paragraph 3 TFEU (block
exemption). Therefore, theCommission adopted theexisting Regulation No 800/2008 of6 August 2008 declaring certain categories
ofaid compatible with thecommon market inapplication ofArticles 87 and 88 ofthe Treaty (General Block Exemption), (OJ L 214,
9.8.2008, p. 3, asamended).7 The regulation will be replaced by a new regulation related toblock exemptions.
The draft ofthe new Block Exemption Regulation, 8 theCommission provides fortheadmissibility ofand exemption from
thenotification requirement forthefollowing uses:
1) Regional aid (investment and operating activity9 forthedevelopment ofurban areas),
2) Small and medium-sized enterprises (SMEs) (investment aid, participation intrade fairs, costs ofproject-related cooperation
under European territorial cooperation),
3) Access ofSMEs tofinance (risk financing, start-ups, alternative trading platforms specializing inSMEs, costs ofdue diligence),
4) Research, development, innovation (research and development projects, investment intheresearch infrastructure
forinnovation clusters, tosupport SME innovation, process and organizational innovation, research and development
inthefishery and aquaculture sector),
5) Training,
6) Disadvantaged workers and workers with disabilities (wage subsidies, recruitment ofworkers ina particularly difficult situation,
wage subsidies forworkers with disabilities, tooffset theadditional costs ofemploying disabled workers),
7) Environmental protection (investment aid enabling undertakings togo beyond Union standards forenvironmental protection
orincrease thelevel ofenvironmental protection intheabsence ofUnion standards, early adaptation tofuture standards
ofEU investment inmeasures topromote energy savings forefficiency projects energy investments incogeneration systems,
promotion ofenergy from renewable sources, promotion ofelectricity from renewable energy sources ininstallations
operating on a small scale intheform ofreductions inenvironmental taxes on Under Directive 2003/96/EC, investment
intheremediation ofcontaminated sites, investment inenergy-efficient heating and cooling systems, environmental studies),
8) To make good thedamage caused by certain natural disasters (aid schemes toremedy some ofthe damage caused by
natural disasters),
9) Social aid fortransport forresidents ofremote regions,
10) Broadband infrastructures,
11) Culture and heritage conservation (aid schemes foraudio-visual works),
12) Sport and multifunctional recreational infrastructure.
Clearly, therange ofaid targets exempted from thenotification requirement asprovided by Article 108 paragraph 3 TFEU is
therefore much broader than even theapplicable provisions ofthe Block Exemption Regulation (Regulation 800/2008). First,
itresults from thegreater experience ofthe Commission, which has enabled theestablishment ofcriteria foradmissibility and
exclusion ofthe notification requirement inrelation tothenew allocation ofaid. Second, itis dictated by theprovisions (described
7
8

Theregulation will expire on 30 June 2014, and will be replaced by a new regulation on state aid covered by theblock exemption.
Draft Commission Regulation (EU) of18 December 2013 declaring certain categories ofaid compatible with thecommon market inapplication
ofArticles 107 and 108 ofthe Treaty.
Regional operating aid is, however, permissible toa limited extent, i.e. only inthemost remote orsparsely populated areas (cf. art. 15 ofthe
Draft).

91

intheprevious part) ofthe reform inthearea ofregulations governing state aid,10 which aim topromote: (i) sustainable, smart and
inclusive growth inthecompetitive domestic market, while contributing totheprogress ofthe activities ofthe Member States
tomake more efficient use public finance, (ii) thefocus intheCommissions ex-ante control measures on theissues that have
thegreatest impact on theinternal market, while strengthening cooperation between theMember States intheenforcement
ofstate aid rules, and (iii) optimizing therules and allow faster and more robust decisions based on clear economic reasons,
a common approach and clear responsibilities. New rules forgranting aid covered by theblock exemption meet thespecific
priorities ofthe Commission.
At thesame time, especially inthepart related toSMEs, thenew rules allow fora broader scope ofstate intervention, which may
affect theprevention ofbusiness failure and does not require prior approval ofthe Commission.
It should be noted, however, that therules ofblock exemption will not apply (as is thecase today) tofirms indifficulty asdefined
inrelevant Community guidelines on state aid forrescuing and restructuring firms indifficulty (OJ C 244, 1.10.2004, p. 2,
asamended).

7.3. Regional, horizontal and sectoral aid


The admissibility ofregional aid, beyond thealready mentioned Block Exemption Regulation, is determined by regional guidelines.11
In theIntroduction totheGuidelines, theCommission points out that, pursuant toArticle 107 paragraph 3 points a) and c) ofthe Treaty
on theFunctioning ofthe European Union (TFEU), theCommission may declare aid topromote theeconomic development ofcertain
disadvantaged EU regions compatible with thecommon market. This kind ofaid is known asregional aid. Theguidelines therefore
indicate theconditions ofadmissibility ofpublic aid aimed at regional development. 12 Such aid is compatible with theinternal
market pursuant toArticle 107 paragraph 3 points a) and c) ofthe TFEU. Prior totheawarding ofsuch aid, theMember States
should obtain theCommissions decision confirming thecompatibility ofthe aid with theinternal market.
According totheguidelines, regional aid may be granted asaid fornew investment and, toa limited extent, asoperating aid, if
theinvestment aid alone is not enough tostimulate economic development ina given region.
The guidelines are generally applicable toall areas ofeconomic activity with theexception ofiron and steel, synthetic fibres,
fisheries and aquaculture, agriculture and transport aswell astheairports and theenergy sector.13 No regional aid is admissible
forinvestments oflarge undertakings inareas classified asregions specified inArticle 107 paragraph 3 point c) TFEU, unless such
investment ushers innew economic activities inthearea ordiversifies theexisting activity toinclude new products orinnovative
processes. Regional aid tofirms indifficulty is also inadmissible.
The Regional Guidelines (as all thenew documents laying down therules ofadmissibility ofpublic aid), theCommission sets out
thegeneral criteria foraid eligibility, which result from theadopted general rules forthemodernization ofthe policy on state
aid. Thus, i) theaid must contribute totheachievement ofa well-defined objective ofcommon interest, ii) there must be a need
forstate intervention, iii) themeasure applied must be appropriate iv) theaid must include an incentive effect, v) theaid must
be limited totheminimum (needed toinduce additional investment oractivity inthearea), vi) undue negative effects ofthe aid
should be limited, vii) transparency ofaid must be ensured. In this context, ofparticular importance are theprovisions related
totheproportionality ofthe aid:

In thecase where aid provides an incentive tomake theinvestment, theamount should not exceed theminimum necessary
tomake theproject profitable enough,

If theaid is an incentive tocarry out theproject intheregion (since itcompensates forthelimitations and net costs arising
from such a location), theamount should ofaid not exceed thedifference between thenet present value ofthe investment
inthetarget region and thenet present value ofsuch investment inanother location.
Moreover, theCommission has set out themaximum aid intensities inrelation toeligible costs, which are primarily used
asthemaximum allowable ceilings ofaid calculated using theabove principles14 aswell asin calculating aid forSMEs.
The formula ofthe new guidelines, which contributes totheobjectives ofthe reform ofstate aid (see Communication from
theCommission totheEuropean Parliament, theCouncil, theEuropean Economic and Social Committee and theCommittee ofthe
Regions The modernization ofEU policy inthefield ofstate aid) clearly indicates that therules related tothegranting ofstate aid
10

11

12

13
14

Cf. Communication from theCommission totheEuropean Parliament, theCouncil, theEuropean Economic and Social Committee and
theCommittee ofthe Regions, The modernization ofEU policy inthefield ofstate aid Brussels, 8.5.2012 COM (2012) 209 final.
In order toavoid all doubt, itshould be emphasized that theconditions forgranting regional aid are governed by regional guidelines. TheBlock
Exemption Regulation lists theconditions forgranting regional aid whose application eliminates thenotification requirement.
Thus, thepurpose ofsuch aid is different from thespecial categories ofaid tocertain sectors, oraid which refers tospecific purposes, such
astraining, research and development, environmental protection, etc.
In this respect, theaid admissibility rules contained inother documents, apply.
With theprovision that forlarge investment projects theamount ofaid shall be calculated based on theso-called theadjusted aid amount.

92

are increasingly used toinfluence theeconomic policies ofthe Member States, toset their priorities and toverify thelegitimacy
ofpublic spending.
There is no single document ofthe Commission which regulates awarding horizontal aid. For each ofthe objectives ofhorizontal
aid, theCommission issued separate documents specifying theconditions fortheadmissibility ofsuch aid. Currently thefollowing
are inforce:

Union Guidelines on state aid topromote risk finance investment (OJ C 19, 22.1.2014, p. 4),

Community framework forstate aid forresearch, development and innovation (OJ C 323, 30.12.2006, p. 1)15 (subject torevision,
tobe replaced by a new document),

Communication from theCommission Criteria forthecompatibility analysis ofstate aid todisadvantaged and disabled
workers subject toindividual notification (OJ C 188, 11.8.2009, p. 6),

Communication from theCommission Criteria forthecompatibility analysis ofstate aid fortraining subject toindividual
notification (OJ C 188, 11.8.2009, p. 1),

Guidelines on state aid forenvironmental protection and energy 20142020,16

Community guidelines on state aid forrescuing and restructuring firms indifficulty (OJ C 244, 1.10.2004, p. 2, asamended.)17
(the document is currently being revised).
Awarding state aid referred toin theabove documents requires theconsent ofthe Commission; therefore, theMember States are
obliged tonotify itoftheir intentions togrant such assistance.
The general principles governing theadmissibility ofeach category ofhorizontal aid are (or will be, inthecase ofdocuments
subject tochange) based on uniform assumptions underlying thereform ofEU policy inthefield ofstate aid.
In thecontext ofthe present study, attention needs tobe drawn totherules related torescuing and restructuring aid. As was
mentioned previously, thecurrent guidelines inthis regard are subject tochange. TheCommission is consulting a draft version
ofthe new guidelines with theMember States. In its proposal, theCommission emphasizes that Rescue and restructuring aid are
among themost distortive types ofState aid. It is well established that successful sectors ofthe economy witness productivity growth not
because all theundertakings present inthemarket gain inproductivity, but rather because themore efficient and technologically advanced
undertakings grow at theexpense ofthose that are less efficient orthat have obsolete products. Exit ofless efficient undertakings allows
their more efficient competitors togrow and returns assets tothemarket, where they can be applied tomore productive uses. By interfering
with this process, rescue and restructuring aid may significantly slow economic growth inthesectors concerned. (Point 6 ofthe Draft
Guidelines).
Where parts ofa failing undertaking remain essentially viable, theundertaking may be able tocarry out a restructuring that leads toits
exit from certain structurally loss-making activities and allows theremaining activities tobe reorganised on a basis that gives a reasonable
prospect oflong-term viability. Such restructuring should usually be possible without State aid, through agreements with creditors orby
means ofinsolvency orreorganisation proceedings. Modern insolvency law should help sound companies tosurvive, help safeguard jobs
and enable suppliers tokeep their customers, and allow owners toretain value inviable companies. Insolvency proceedings may also return
a viable undertaking tothemarket by way ofacquisition by third parties, whether ofthe undertaking asa going concern orits various
production assets. It follows that undertakings should only be eligible forState aid when they have exhausted all market options and
where such aid is necessary inorder toachieve a well-defined objective ofcommon interest that could not be achieved without theaid.
Undertakings should be allowed toreceive aid under these guidelines only once within ten years (the one time, last time principle). (Points
7 and 8 ofthe Draft Guidelines).
Under theDraft Guidelines, an undertaking is considered tobe indifficulty when, without intervention by thestate, itwill almost
certainly be condemned togoing out ofbusiness intheshort ormedium term. In particular, itmeans that at least one ofthe
following circumstances occurs: i) inthecase ofa limited liability company, where more than half ofits subscribed share capital
has been lost asa result oflosses incurred, ii) inthecase ofa company where at least some shareholders have unlimited liability
forthedebts ofthe company, where more than half ofits capital asshown inthefinancial statements has been lost asa result
oflosses incurred, iii) where theundertaking is subject tocollective insolvency proceedings orfulfils thecriteria under its domestic
law forbeing placed incollective insolvency proceedings at therequest ofits creditors, iv) where theundertaking is rated
theequivalent ofCCC+ (payment capacity is dependent upon sustained favourable conditions) orbelow by at least one registered
credit rating agency, v) where: (1) theundertakings book debt toequity ratio is greater than [7.5] [and]/[or] (2) theundertakings
[EBIT]/[EBITDA] interest coverage ratio has been below [1.0] forthepast [two] years.18

15
16
17
18

Thedocument is valid until 30 June 2014, and will be replaced by a new framework.
Adopted by theCommission on 9 April 2014 (not yet published).
Thedocument will be replaced by thenew guidelines.
These provisions shall be specified intheadopted document.

93

Where none ofthe circumstances applies, theCommission (and only theCommission) may exceptionally consider that an
undertaking is indifficulty if there is evidence that without an intervention by thestate, theundertaking will almost certainly be
condemned togoing out ofbusiness intheshort ormedium term.
Assistance referred toin theDraft Guidelines may not be granted tonew businesses, i.e. during thefirst three years after thestart
oftheir activity.
The Draft Guidelines deal with three types ofaid: rescue aid, restructuring aid and temporary restructuring support. Rescue aid is
temporary innature and is intended toenable an undertaking tosurvive forthetime necessary todevelop a restructuring plan
ora liquidation plan. Theaim ofthe restructuring aid is torestore theundertakings ability tocompete inthemarket on thebasis
ofa restructuring plan, which should determine thecauses ofthe difficulties, appropriate methods oftheir solutions, time, and
budget restructuring, with theprovision that state aid inthebudget should be limited totheminimum necessary. Temporary
restructuring support is liquidity assistance designed tosupport therestructuring ofan undertaking by providing theconditions needed
forthebeneficiary todesign and implement appropriate action torestore its long-term viability. Temporary restructuring support may
only be granted toSMEs (point 30 ofthe Draft Guidelines).
As a rule, therescue and restructuring aid cannot be given ineach individual case without theconsent ofthe Commission.
Therefore, theMember States prior toawarding such aid are required tomake thenotification, but this rule does not apply toSMEs,
forwhich theMember States may develop aid programmes.19 Approval ofan aid programme on thepart ofthe Commission
inprinciple excludes theobligation tosubmit individual notifications foreach case.
The Commission shall assess theadmissibility ofaid (its compatibility with theinternal market) taking into account thefollowing
criteria:
1) Contribution toa well-defined objective ofcommon interest. In relation toSMEs, theCommission expects tobe shown that
thedisappearance ofan undertaking would result insocial hardship ormarket failure. In addition, theCommission expects
that therestructuring plans will comprehensively address all theissues which are thesource ofproblems20 (in particular,
theaid will not be solely focused on financial aid tooffset losses), and ways toeliminate them through thenecessary
restructuring ofevery sphere ofbusiness. Restructuring plans must specify reliably expected effects ofcorrective actions,
2) The need toprovide assistance. It is necessary todemonstrate that without such aid theobjectives ofcommon interest will
not be achieved,
3) Relevance. Rescue aid must fulfil thefollowing conditions: aid may be provided instrictly defined forms (loan guarantees
orloans) asset out inthedraft remuneration, and fora specified period oftime (cf. point 57 ofthe Draft Guidelines). In
thecase ofrestructuring aid, theMember States are free tochoose its form aslong astheinstrument chosen is appropriate
totheissue that itis intended toaddress. In regard tothetemporary restructuring support forSMEs, itmay be limited toloans
orloan guarantees forwhich level ofremuneration should reflect theunderlying creditworthiness ofthe beneficiary, and be
provided fora limited time,
4) The incentive effect. It is necessary todemonstrate that intheabsence ofthe aid, thebeneficiary would have been
restructured, sold orwound up ina way that would not have achieved theobjective ofcommon interest,
5) Proportionality. Rescue aid must be limited totheminimum necessary tokeep thebeneficiary inbusiness forsix months.
Theamount ofrescue aid shall be determined by theformula set out inAnnex 1 totheDraft Guidelines. Theamount
ofrestructuring aid should also be kept toa minimum, which shall be determined by theburden sharing formula21 ormethod
ofdetermining theextent towhich thebeneficiary should bear therestructuring costs. Temporary restructuring support must
be restricted totheamount needed tokeep thebeneficiary inbusiness for[12] [18] months.22 In determining that amount
regard will be had totheoutcome ofthe formula set out inAnnex 1.
6) Avoidance ofundue negative effects on competition and trade between theMember States. In this respect theDraft
Guidelines provide that rescue, restructuring and temporary restructuring support should be granted inprinciple only
inrespect ofa single restructuring action (one time, last time principle). Secondly, theDraft provides fortheapplication
ofmeasures tolimit distortions ofcompetition: structural measures intheform ofa sale ofassets orreduction ofproduction

19
20

21

22

A definition ofwhat constitutes an aid programme can be found inPart VI.


In thecase ofSMEs, theassessment ofthe circumstances remains at thediscretion ofthe Member States. In other cases such assessment shall
always be conducted by theCommission.
Currently, theGuidelines provide fortwo options forburden sharing: i) theamount ofcontribution totherestructuring shall be at least equal
totheamount ofaid, and thecontribution made by theexisting shareholders and creditors shall be justified inthelight ofany losses which
they may have suffered intheevent ofinsolvency. Under exceptional circumstances and incases ofparticular hardship, theCommission may
authorize a contribution which is lower than theamount ofthe aid, ii) theamount ofcontribution totherestructuring shall be at least 50%
ofthe restructuring cost. In exceptional circumstances and incases ofparticular hardship, theCommission may authorize a contribution that is
less than 50% ofthe restructuring cost. Detailed terms and conditions associated with each ofthese options are set out insection 3.5.2 ofthe
Guidelines.
As previously indicated, this condition shall be ultimately determined intheadopted guidelines.

94

7)

capacity ormarket presence and behavioural measures, which aim toensure that theaid is directed solely tofinance longterm viability, aswell asmarket opening measures.
Avoiding excessive negative effects ofstate aid also occurs through theapplication ofthe principle that where unlawful
aid has previously been granted toan undertaking indifficulty inrespect ofwhich theCommission has adopted a negative
decision with a recovery order, and where no such recovery has taken place, theassessment ofany rescue aid, restructuring
aid ortemporary restructuring support tobe granted tothesame undertaking shall take into account, first, thecumulative
effect ofthe old aid and ofthe new aid and, second, thefact that theold aid has not been repaid.
Transparency. TheMember States must publish on a designated website information on notified aid programmes, including
thegranting authority, theform and amount per beneficiary and theprincipal economic sector.

In addition, with regard toSME programmes submitted totheCommission, theMember States should specify themaximum
amount ofaid that can be awarded toany undertaking aspart ofthe activities ofrescue aid, restructuring aid ortemporary
restructuring support. Themaximum total amount ofaid granted toany one undertaking may not be more than EUR 5 million,
including any aid obtained from other sources orunder other schemes.
Conditions forawarding rescue aid, restructuring and temporary restructuring support toSMEs aswell astheconditions
forgranting de minimis aid (including firms indifficulty) should allow theMember States tocreate more flexible than previously
instruments toprevent bankruptcy inorder toenable them torespond effectively toproblems ofundertakings whose continued
operation is particularly important forsocio-economic development.
The policies against bankrupt companies should therefore take into account therules related torescue aid, restructuring and
temporary restructuring support forSMEs, inparticular theprinciple that theCommission may approve aid programmes that target
SMEs, but inrelation toother undertakings itmust make individual assessments ofaid foreach beneficiary.
It should also be noted that EU law lays down specific rules forawarding aid tosectors inwhich theprevailing business conditions
justify a particular approach totheprovision ofaid. In this regard, thefollowing documents have been adopted:











Guidelines on state aid toairports and airlines (OJ C 99, 4.4.2014, p. 3),
EU Guidelines on theapplication ofstate aid rules inrelation torapid deployment ofbroadband networks (OJ C 25, 26.1.2013,
p. 1),
Framework on state aid toshipbuilding (OJ C 364, 14.12.2011, p. 9),
Communication from theCommission on theapplication, from 1 August 2013, ofstate aid rules inrelation tosupport
measures infavour ofbanks inthecontext offinancial crisis (Banking Communication), (OJ C 216, 30.7.2013, p. 1),
Guidelines on certain state aid measures inthecontext ofthe greenhouse gas emission allowance trading scheme post-2012
(OJ C 158, 5.6.2012, p. 4),
Community guidelines on state aid tomaritime transport (OJ C 13, 17.1.2004, p. 3),
Communication from theCommission providing guidance on state aid complementary toCommunity funding
forthelaunching ofthe motorways ofthe sea (OJ C 317, 12.12.2008, p. 10),
Communication from theCommission providing guidance on state aid toship management companies (OJ C 132, 11.6.2009,
p. 6),
Council Decision of10 December 2010 on state aid tofacilitate theclosure ofuncompetitive coal mines (OJ L 336, 21.12.2010,
p. 24),
Commission Notice on theapplication ofcompetition rules tothepostal sector and on theassessment ofcertain state
measures relating topostal services (OJ C 39, 6.2.1998, p. 2),
Community guidelines on state aid forrailway undertakings (OJ C 184, 22.7.2008, p. 13),
Council Regulation (EC) No 1370/2007 ofthe European Parliament and ofthe Council of23 October 2007 on public passenger
transport services by rail and road passenger transport services and repealing Council Regulation (EEC) No 1191/69 and (EEC)
No 1107/70 (OJ L 315, 31.12.2007, p. 1).

In thecontext ofpolicies aimed at preventing business failure ofparticular importance is theCommunication on theapplication,
from 1 August 2013, ofstate aid rules inrelation tosupport measures infavour ofbanks inthecontext offinancial crisis.

7.4. Structure and scale ofstate aid inPoland


Pursuant toArticle 31 ofthe Procedural Issues Pertaining toPublic Aid Act, thePresident ofthe UOKiK (Office forCompetition and
Consumer Protection, OCCP) monitors state aid granted inPoland23 and prepares an annual report on theresults ofthe process.
Thelatest report pertains to2012, while a separate report contains information on aid granted in2012 under thede minimis rule.
23

Monitoring state aid inagriculture falls under theresponsibility ofthe Minister ofAgriculture and Rural Development.

95

The total value ofaid granted inPoland in2012 amounted to20379.124 million PLN, which represented 1.28% ofGDP.
In 20082012, thelargest aid was granted in2010 (24 087.3 million PLN, which represented 1.70% ofGDP).
In 2012, rescue aid amounted to401.2 million PLN, and restructuring aid totalled 17.7 million PLN, while de minimis aid was 4 329
PLN 5 million (0.27 % ofGDP).
As far aspolicies against bankruptcy are concerned, itshould be noted that currently inPoland there are no rescue orrestructuring
aid programmes targeted at SMEs. It means that there is no systemic solution that would permit thestate tointervene and help
theabove group ofundertakings indifficulty.25 It should be remembered, however, that theavailability ofrescue and restructuring
aid constitutes an essential instrument fortheprevention ofbankruptcies.

Abbreviations
TFEU Treaty on theFunctioning ofthe European Union
The Commission theEuropean Commission
ECJ The European Court ofJustice
SPI The Court ofFirst Instance
EGC The Court ofthe European Union
ECJ The Court ofJustice ofthe European Union
De minimis Regulation No. 1407/2013 of18 December 2013 on theapplication ofArticles 107 and 108 ofthe Treaty on
theFunctioning ofthe European Union tode minimis aid (OJ. EU L 352, 24.12.2013, p.1)
Regional Guidelines Guidelines on national regional aid for20142020 (OJ. EU C 209, 23.7.2013, p.1)
Design Guidelines Guidelines on state aid forrescuing and restructuring non-financial undertakings indifficulty26
SME small and medium-sized enterprises

24

25

26

This amount does not include aid intheform ofremuneration forGdask Transport Company SA and forAutostrada Wielkopolska II SA. In its
decisions on thecompatibility ofaid with theinternal market, theCommission did not indicate theextent ofremuneration that constituted
state aid.
It is possible that thelack ofa rescue orrestructuring aid programme forSMEs reflects thepolicy offocusing exclusively on large firms that can
actually affect thesocio-economic development.
http://ec.europa.eu/competition/consultations/2013_state_aid_rescue_restructuring/draft_guidelines_pl.pdf.

96

Part III
Models oftheRapid Response Instrument

Piotr Adam Boguszewski, Stanisaw Mazur, Marcin Zawicki,

Chapter 8
THE DESIGN PRINCIPLES OF EARLY WARNING AND RAPID RESPONSE
SYSTEMS IN THE NON-FINANCIAL SECTOR
8.1. Introduction
Early warning, rapid alert, early response orrapid response systems are just some ofthe names given tomechanisms that serve
toanticipate and prevent risks developed at thelevel ofstrategic management ina given country ortheir components within interand transnational systems. These names refer todiverse analytical and decision-making structures aswell asformal organization
which share common goals. Warning systems focus on prediction, whereas response systems concentrate on avoiding dangers
and risks. An early identification and elimination ofthe causes ofthreats before they develop into full-scale crises ordisasters is
a doctrine on which various institutions ofpublic governance are built. This way ofthinking also incorporates thedevelopments
inscience and technology enabling us toperform observations more effectively and limit therisks ofuncertainty, instability and
unpredictability occurring intheareas ofthe economy, society, policy and theenvironment.
Early warning and response systems usually include:

Mechanisms toanalyse therisk ofoccurrence ofdisruptions innational economies orintheir sectors which justify
theemployment ofspecific preventive measures and assistance activities by thegovernment and its agencies (early/rapid
alert systems),

Anticipatory research on theeconomic and financial condition offirms inorder todiagnose thesymptoms oftheir worsening
standing, including thethreat ofbankruptcy, using discriminatory prediction models ofsuch bankruptcy (early/rapid alert
systems),

Comprehensive systems providing proactive support tofirms infinancial distress, employees ofsuch enterprises, usually
taking into account assistance intheform ofpreparation and implementation ofrestructuring plans (early/rapid response
systems),

Warning systems against currency crises implemented by banks and financial institutions (early/ rapid alert systems),

Hazard warning systems ininternational goods and services markets concerning e.g. dangerous food products, animal feed
etc. (early/rapid alert systems),

Early warning systems against natural disasters and catastrophic events including: air quality, fires innatural areas, nuclear and
chemical accidents, geological hazards (earthquakes, tsunamis, volcanic eruptions, sinkholes, landslides), hydro-meteorological
hazards (desertification, droughts, floods, impact ofclimate change, extreme weather conditions, storms, tropical cyclones),
epidemics (early/rapid alert systems),

Early warning/response systems torisks related tothepolitical situation, security, theoperation ofeducational systems, health
care, communications, transportation and a number ofothers.
In this chapter we shall focus on early warning and rapid response systems used toanticipate and prevent threats totheeconomy.

8.2. An outline ofthe morphology ofearly warning systems


Contemporary economies are served by a number ofearly warning systems (EWSs) with differing characteristics and objectives
(cf. e.g. Walas-Trbacz 2012). Nevertheless, they have some features incommon, which shall be discussed below.
An outline ofthe structure ofsuch a system is shown inFigure 8.1. It should be emphasized that this is a far-reaching simplification
since they tend tobe, infact, quite complex structures. In particular, inour diagram most connections between theelements have
been assumed torun one-way, whereas inreality they often are two-way ones. Moreover, inreal EWSs certain direct relationships
exist even among themajority oftheir component parts. It must also be remembered that theoperation ofsuch systems
especially inthesocial space often requires multiple iterations ofthe procedures described here. In our discussion we do not
differentiate1 between EWS and RRS systems since, although we have simplified thesubject matter somewhat, they can be
assumed todiffer mainly with respect tothescale ofthe reaction. After all, theoperation ofan EWS results incertain reactions,
occasionally involving only thetransfer ofinformation from thesystem toits stakeholders. Obviously inthecase ofRRSs thescope
ofsuch activities, their impacts and further procedures may be much more complex compared with theEWSs.

Such differences, however, have been discussed inmore detail inother sections.

99

Models nad
procedures

Patterns and
standards

Instruments

Warning
signals

Acton scenarios
planning
module

Prevention
Advance
indicators
detection and
measurement
system

Input
signals

World situation
forecasting
system

Threat costs
Choice
of channels, units
and methods

Outcomes
evaluation
module

Effectiveness
evaluation
module

Outcomes

Communication
and
implemenation
process

Selection of
response path

Efficiency
evaluation
module

Diffuson
And synergy
Models

Ews
decision
making
rules

Procedures
and criteria
Correction and
self-learning
module

Final
recommendation

Cost-Benefit
Balance

Decsionmakers

Resources
and other
limitations

Model of the porcess


Planning phase
Implementation

Fig. 8.1. A structural outline ofan early warning/rapid response system (EWS/RRS)
Source: own study.

The first element ofan EWS is a component that serves todetect and measure advance signals. It should be noted that
theexistence ofsuch signals constitutes thenecessary condition forthedesign ofsuch a system. International experience shows
that those signals include both macro- and microeconomic variables. Their list can be found e.g. inReport 2011 (Report 2011) and
most ofthem are also analysed intheRRS. They must be strong enough tobe measured and occur early enough toensure that by
activating theprognosticdecision-making procedures shown inthediagram above we can be reasonably certain that our actions
stop orblunt theunfavourable course ofevents that would take place if thesystem was not used. In thecontext ofthis remark,
particular attention should be paid tosubsection 1.4 and Chapters 3, 10, 11, which characterize theforecast horizons and actions
ensuring that thecondition is met. Likewise, we can better realize thesignificance ofproblems inherent inthemeasurement ofthe
phenomenon ofbankruptcy and its attendant circumstances, including theprevalent economic conditions, tothedesign ofEWS
described inChapter 3.
The next stage ofEWS operation involves prognostic procedures, which, on thebasis ofthe identified advance signals permit
theidentification ofthreats ahead oftime. At this stage, a variety oftools may also be used such asformal statistical and
econometric models, expert reviews, opinion surveys etc.
It should be emphasized, however, that theforecast ofa future situation inno way constitutes a sufficient condition fortriggering
further actions on thepart ofthe EWS orRRS inquestion. Such a decision can only be made if thepresence ofat least one ofthe
circumstances is detected, i.e. theforecast diverges from a certain standard ormodel and/or a high social cost associated with
theoccurrence ofsuch a situation.
Both above-mentioned criteria may be either complementary orsubstitutable. For example, we are often unable tocalculate
thecosts (the magnitude ofa threat) associated with certain future situations and remedial action is instituted only because we
believe that they are unusual. Let us also note that thedecision may be based both on models (i.e. normative elements) and on
departures from conditions considered astypical. Thelatter element ofan EWS is based on statistical regularities, and thus has
strong empirical foundations.
As was shown inChapter 4 inthecase ofour EWS, bankruptcy costs inPoland can be substantial. Thesystem assuch, incomparison
with standards, is not very efficient. At thesame time, thecosts may arise invarious areas ofthe economy and affect numerous
elements ofsocial structures.

100

In thediscussion ofthe process ofgenerating warning signals we must also mention another issue. In theprocess ofcost
examination, a very important role may be played by analyses ofdiffusion and synergies ofthreats. Naturally, one may imagine that
such procedures belong tothemodule that forecasts a certain state ofthe world. In practice, this tends tobe quite difficult. Usually,
we can make best predictions about individual components ofthe system at an earlier stage. To that end, aggregation procedures
are needed, which do not just add up theindividual threats. EWS have todeal with theproblems ofinterdependence ofthreats.
How seriously they may affect thestate ofthe entire system intheevent ofbankruptcy was demonstrated insubsection 3.4.
The rules outlined above have also been subject toattempts at formalization, but they originally concerned natural disasters rather
than disruptions intheoperation ofeconomic systems. A good illustration ofsuch rules was offered e.g. by Grasso (2005), whose
idea is reproduced below.

Fig. 8.2. Types oferrors inearly warning systems


Source: compilation based on: (Grasso 2005).

First ofall, we must note that inthecase ofEWSs two types oferroneous decisions may be taken. Obviously, different rules may
apply toalarm triggers and their deactivation, but itcan be assumed that they are similar tothemechanism shown inFigure8.2.
Such steps are, infact, taken if theprojected state ofthe object exceeds a certain level ofthreat. First, a decision toinitiate
theprogramme tocounter thethreat may be taken ina situation inwhich itdoes not materialize. If themeasure ofthreat is
thecondition ofan object which is worse than a certain critical level K, thesituation is illustrated intheleft panel inFigure 8.2.
Theprobability ofoccurrence ofsuch an event known asa false alarm is represented by PFALSE and is determined by formula (8.1).

PFALSE = Pr F < K |F a K

= Pr F K |F < a K
MISSED

(8.1)

The opposite situation, presented intheright panel, may occur if thesystem does not signal theneed foraction when thestate
ofthe object turns out tobe worse than thecritical level K. This is thecase ofa hazard overlooked by thesystem whose probability
is PMISSED. One ofthe factors affecting thelikelihood ofoccurrence such errors is thechoice ofthe hazard level that triggers alarm
activation.2 The choice reflects an obvious dilemma if thetrigger level is low, theprobability ofPMISSED is small. However, at
thesame time itincreases thelikelihood ofgenerating false alarms. Theissue ofappropriate selection procedures fortheactivation
ofsuch a system is discussed inmore detail inChapter 12 devoted tothetriggering mechanisms. Here we should only mention that
itis particularly difficult inthecase ofEWS since such systems operate ina social environment inwhich a high incidence offalse
alarms would completely undermine confidence insuch systems and thus therationale forits existence. Moreover, ineconomics
itis not always clear how todefine themeasured variable K and how todetermine its critical level, especially ina situation oflimited
access tosufficiently long observation sequences toeasily establish their distribution and toidentify theareas ofrisk on their basis
(cf. e.g. Bussiere and Fratzscher 2002).
2

Here we have assumed that itis proportional tothemagnitude ofthe critical state K, but other options are possible.

101

Each ofthe above-mentioned errors entails certain costs, but EWSs generate them even when they properly respond tothreats.
In thelatter case, they are associated with themaintenance ofthe system inits standby mode. Thematrix ofthese costs is shown
below:

AKTYW
PASYW

F <K
CFA
CGM

F >K
CGA
C MA

(8.2)

The rows inthematrix correspond tosituation inwhich a decision was made within thesystem totake action (AKTYW) ornot
torespond tothethreat (PASYW). In turn, thefirst column (F<K) reflects a situation inwhich themagnitude ofthe threat does not
justify therisks oftaking action within a given EWS, whereas thesecond one deals with a situation that requires a response (F>K).
Theexpected total costs ofaction inboth cases are described by formulae (8.3) and (8.4):

E (C | AKTYW ) = CFAPFALSE + CGA (1 PFALSE ) = CFAPFALSE + (CBANK COPOR )(1 PFALSE )

E (C | PASYW ) = CGM (1 PMA ) + C MA PMA C MA PMA

(8.3)
(8.4)

where CBANK denotes costs associated with thebankruptcy ofthe entity inquestion (or a sector ofindustry orspecific segments
ofthe economy), COPOR thebenefits ofavoiding risks, and PMA theprobability ofmissing an actual threat by thesystem.
Theapproximation which entails therelationship presented in(8.4) is theresult ofour assumption that thecosts associated with
system inactivity when itshould not trigger any actions constitute standby costs and are insignificant.
On thebasis offormulae (8.3) and (8.4) we can define themain condition foraction within an EWS/ RRS based on thecost
avoidance approach (cf. e.g. Klafft and Meissen 2011):

E (C | PASYW ) > E (C | AKTYW )

(8.5)

Owing torelationships (8.3) and (8.4), theinequality can be rewritten as:


CBANK PMA > (COPOR + CFA CBANK )PFALSE + (CBANK COPOR )

(8.6)

From formula (8.6) itfollows that one ofits important elements is thequality ofa given EWS expressed by both PMA and PFALSE,
thecosts oftaking action intheevent ofa false alarm CFA, aswell asthebenefits that thesystem affords interms ofavoiding threats
COPOR. Assuming that theabove EWS/RRS parameters are appropriate, itdirectly follows from (8.6) that if bankruptcy is socially
expensive, itis worth avoiding ina broad range ofcontexts (determined by theother components ofthe inequality). Incidentally,
itshows theimportance ofthe reflections on theproper design ofdecision-making rules forEWS/RRS systems contained inPart
I ofthis book, especially Chapters 1, 2 and 4.
The stages inEWS operation described above result intheprovision ofinformation on theareas and magnitudes ofthreats. Let
us then assume that after relevant analyses and tests have been completed, itwas found that condition (8.5) holds, but at least3:
COPOR > CBANK

(8.7)

Now, theEWS/RRSs initiate procedures known astheplanning stage. Thefirst element ofthe process constitutes an attempt
tomap out theinstruments at thesystems disposal over thearea under threat. Let us make here a trivial, but important note that
EWS/RRSs without instruments ofaction virtually turn into mock systems since thecan do nothing toavoid orreduce theeffects
ofa threat. Theother important issue is that EWS instruments can either correct and compensate (i.e. by minimizing theeffects
ofalready present negative consequences) orprevent threats before they materialize (an action represented by thedashed line
inFigure 8.2).
A technical solution ofthis problem is not easy, but inEWS/RRSs one ofthe canonical approaches may include theso-called
MIMC class models (see e.g. Rose and Spiegl 2009) inwhich theobserved symptoms/costs ofthreat yi,j are modelled asa hidden
function ofthe unobservable variable ofthe harmfulness ofcrisis i ina given area (sector, country etc.) i, where harmfulness itself
is measured asa function ofthe observable causes and instruments w affecting crisis xi,w:

y i , j = j i + i

i = w x i ,w + i

(8.8)

In fact, verifying condition (8.5) requires analyses ofthe successive stages presented inFigure 19: thecosts inmatrix (8.2) depend, among other
things, on thechoice ofinstruments. Therefore Figure 8.2 is, inpractice, aswas already mentioned, a procedure that requires multiple iterations.

102

Hence, if we can assign specific instruments ofaction toidentified threats, we can move on tothenext stage: calculating thecostsbenefits balance associated with their use. Thecosts ofconducting specific activities within any EWS are infact always present,
even if itis only meant toconvey warnings.
Such a cost-benefit balance ofusing specific tools should be compared with thecosts tobe incurred intheevent offailure
topursue an active policy toprevent threats (see matrix 8.2). Obviously this does not involve a simple comparison oftwo numbers,
but a complex procedure, which we have called theEWS decision-making rules. Whereas thepreviously described stages ofEWS
operation can be implemented by experts oreven automatically, at this stage theneed toinvolve stakeholders inthedecisionmaking process is becoming clear. In calculating thefinal balance ofthe advisability ofresponding tothreats they must be guided
not only by a map oftheir goals and interests orconsensus procedures inthis respect, but also by thematerial resources and
other constraints available ina given EWS. Let us emphasize that at this stage our system will often require theactivation oflinks
operating intheopposite direction. This may occur e.g. when theinstruments selected at an earlier stage prove tobe too costly
orunacceptable by all stakeholders.
If theabove process ofanalysing thecosts and opportunities associated with thespecific mode ofconduct leads topositive results
tofinding a path forEWS instruments we may move on tothestrategy implementation phase. It is a complex task, which,
inthecase ofEWS, requires further specification ofa number ofcomponents. It is vital toselect communication strategies, target
groups ofactivities, detailed courses ofaction etc.
This is an extremely subtle and important component ofEWS from thevantage point oftheir effectiveness. Poorly targeted
assistance only generates costs both direct losses related toineffective outlays and indirect ones. A system which does not warn
indue time those who are actually indanger loses its credibility. In any EWS itis theconfidence inthequality ofits solutions that
constitutes a prerequisite forits effective implementation. Hence theextensive discussion ofthe issues ofcommunication and
perception ofthis instrument asapplicable toRRSs inChapter 13.
The final components ofan EWS are theperformance evaluation modules complete with procedures fortheir correction and
self-learning. Such an evaluation occurs along two dimensions efficiency and effectiveness. In designing an EWS one needs
todetermine thereciprocal relationship between these two elements. In thecase ofbankruptcy itis particularly important,
because, aswe have shown inChapter 1, a given bankruptcy system itself may be designed toachieve different objectives and
indifferent ways. Very often theeffectiveness ofa given EWS requires efficiency asits prerequisite. An inefficient intervention system
is characterized by low social credibility, which, inturn, may reduce its efficiency. Second, a very important element ofall EWSs is
theinclusion ofself-correcting and self-learning procedures. These characteristics are highly desirable, especially intheprognostic
section. Despite thedynamic development ofmethods offorecasting social and economic phenomena, thequality ofpredictions
is still an area inwhich thelearning processes, including those based on previous forecasting experience, significantly contributes
toimproving their quality.

8.3. Classification ofearly response systems


An early response system is defined asan institutionalized, public administration-based preventive mechanism aimed at providing
support toenterprises indistress oremployees ofsuch enterprises triggered inresponse toperturbations intheeconomic cycle.
Such a system comprises all activities undertaken from themoment an emergency situation is recognised (alert triggered by
an early warning system), thespecification ofthe scope ofthe intervention needed, thedecision tostart such an intervention,
theintervention itself until theevaluation ofits outcomes. Theessence ofan early intervention consists not only ina rapid
implementation ofpre-emptive activities but also no less important inachieving their objectives with maximum efficiency
within a short span oftime.
Early response systems originated inthe1970s and 1980s intheUnited States, where instruments offering comprehensive support
totheemployees ofrestructured enterprises especially intheheavy and automotive industries, came tobe used more and
more widely. Thepurpose ofthose solutions was toassist workers facing redundancy, companies and their closest business
environment. Thesystem comprised rapid response units equipped with pre-emptive action capacity. At thefederal level, these
mechanisms were sanctioned by theWorker Adjustment and Retraining Notification Act of1989. Since then, numerous countries
have built similar systems. Theworldwide financial crisis, which started in2007, made such systems even more popular and led
totheintensification offurther development ofrapid response systems. With time, these systems have become thestandard
ofgood governance ineconomic policy.
Identification, analysis, but especially categorization and classification ofearly warning systems intheeconomy is beset by
a number oflimitations. Two ofthem appear tobe themost important. First, such systems noticeably differ among one another

103

not only due tothefact that sometimes they perform other related functions beyond rapid response. Second, some ofthe early
warning systems have names that do not directly correspond with their functions, which makes them difficult torecognize.
It should be emphasized that no universally accepted convention categorizing early response systems has been developed so
far. Theaccomplishment ofsuch a plan exceeds thescope ofthis book, we wish, however, tosuggest a handful ofcriteria useful
indeveloping a typology ofearly warning/rapid response systems (Table 8.1).
Table 8.1. Thecriteria fordifferentiating early warning systems intheeconomy
Criterion

RRS characteristics

1. Place occupied intheadministrative system

government, centralized
government, deconcentrated
decentralized
mixed

2. Compliance with therestructuring trends

include basic restructuring trends


disregard restructuring trends

3. Notification about planned redundancies

legal acts (usually labour market regulations)


judicial decision
collective bargaining
governed by customs
mixed

4. Jobs

preservation ofexisting jobs


finding new jobs
creation ofnew jobs
mixed

5. Role oftrade unions

major
insignificant

6. Scale and scope ofinstruments used

numerous and varied


few and uniform

7. Flexibility and innovation

high
low

Source: own research.

Now let us analyse theposition ofthese solutions within theadministrative system. Early response systems are managed by
central government administration orina decentralized manner. Where theresponsibility forrapid response rests with
thegovernment, such systems are directly subordinated tocentral authorities ortheir local representatives (a de-concentrated
system). Analternative solution involves thedevelopment ofearly response systems at thelevel ofdecentralized administration.
There are also mixed solutions inwhich theresponsibilities rest both with thecentral government and decentralized administration.
An example ofa mixed system attached tocentral government structures and tode-concentrated administration is theBritish
Rapid Response Service. In contrast, early response systems operating intheUnited States are essentially decentralized, managed
at thelevel ofstate administration.
Todays rapid response systems are expected toform part ofthe mainstream restructuring processes inenterprises. Currently used
early response systems should therefore be consistent with theongoing nature ofsuch processes and operate on theassumption
that they are carried out inmulti-level arrangements and with theparticipation ofnumerous actors (European Restructuring 2010,
p. 11).
Corporate restructuring is increasingly becoming a perpetual process. This involves theneed toimplement mechanisms
forprogramming and overseeing changes inenterprises. These changes should, however, be programmed ina rational manner.
Especially inlarge enterprises there exist certain individual cells that do not tolerate well excessive change since thequality oftheir
performance requires stability over time. In some enterprises such cells include central management orproduction facilities, while
inothers they may comprise research centres orcustomer service. Early response systems should therefore take into account
thelimited utility ofimmediate remedial measures aslong asthey are targeted at segments ofenterprises which prefer a shortormedium-term stability and inertia.
The effects ofeconomic perturbations are observed primarily inindividual enterprises orparts oflarge companies affected by
theconsequences ofcrises affecting local orsupralocal markets. In contrast, due totheprogressive intensification ofcooperation
among enterprises and theties between thecomponents oflarge companies and corporations, itis necessary forearly response
systems tobe aware ofthe multi-level (local, regional, national and transnational) context ofremedial actions which start at
thelocal level. For example, possible interventions involving a manufacturing firm which belongs toa transnational corporation
necessitate negotiations with its headquarters on thepart ofthe institution that manages a given early intervention system.

104

Finally, restructuring processes have become an arena where, inaddition toenterprises requiring support, there are a number
ofterritorially-rooted actors. One may even go asfar asto say that itis not an enterprise indistress but theterritory within therange
ofimpact ofsuch an enterprise which has become themost important target ofearly interventions. Gone is thevision ofcorporate
restructuring conducted at theaffected enterprises own expense asis theperception ofcorrective actions inenterprises only
ineconomic terms. Thereasons forthis are fairly obvious. Problems ofenterprises cause a range ofsocial, economic, financial
and environmental perturbations fortheresidents ofa given territory, forthechain oflocal sub-contractors and affect local
development at large. That is why rapid alert systems should allow local social and economic actors tobe involved inthedesign
and implementation ofinterventions.
Notification about theplanned redundancies constitutes an important part ofour considerations. Employers obligations tonotify
public authorities oftheir intention toalter theemployment structure differ by country.
Such obligations are laid down not only by relevant legislation that regulates thelabour market, but also result from judicial
decisions, collective bargaining and local customs. In practice, each early response system employs all theabove mentioned ways
tovarying degrees and with varying intensity. Essentially, legal provisions dominate incontinental European countries. Theformula
ofcollective bargaining has been extensively used inSweden, inGermany and intheUnited States, while inJapan such issues are
forthemost part resolved by custom (Hurley, Mandl 2011, p. 16).
As a rule, notifications about planned layoffs are submitted topublic authorities, trade unions and employees inadvance ofthe
deadlines stipulated by law ordictated by custom. In most cases thenotification period is at least 30 days.
Most early response systems operating incontinental European countries feature a strong orientation towards thepreservation
ofexisting jobs (e.g. Germany, France, and Austria). In these countries redundancies are considered justified only if all other methods
ofprotecting jobs have proved tobe ineffective.
Concerted efforts topreserve jobs are also evident inJapan, where extensive instrumentation tothat end is used intherestructuring
process. This logic derives from thebelief that employment stability (or even lifelong employment) is thesupreme value, which is
rooted inJapanese culture and constitutes a natural duty ofentrepreneurs.
The focus ofthe early response system intheUnited States is quite different. Despite thepresence oflegislation that is quite
favourable toemployees interms ofrestructuring, corporate reorganisation is a much more frequent phenomenon than inEurope.
Employers more readily use employment restructuring asa means ofoptimizing employment. In turn, public authorities more
often implement measures aimed at creating new jobs rather than at theprotection ofexisting ones (Evans-Klock, Kelly P, Richards,
Vargha 1998).
The position oftrade unions inearly response systems varies considerably. In some ofthem (e.g. inGermany orFrance), thetrade
unions play an important role indetermining theconditions ofemployment restructuring and theredundancy criteria (socially fair
asprovided by theGerman Protection against Redundancies Act). They are also extensively involved intheprogramming ofpreemptive actions involving training schemes aimed at theacquisition ofnew skills by employees who may need them incase they
are made redundant.
In theUnited States, theposition oftrade unions inrestructuring processes is less prominent. This does not mean, however, that
employees are deprived oflegal protection inthearea ofemployment. On theone hand, such protection is provided by, among
others, thecollective bargaining formula. On theother hand, analysts note theincreasing popularity ofagreements made between
employers and trade unions concerning pre-emptive training aimed at developing thecompetencies and flexibility ofemployees
inorder toprotect them from losing their jobs inthecourse ofrestructuring (a practice common intheautomotive industry)
(Evans-Klock Kelly P, Richards, Vargha 1998).
We shall now turn tothequestion ofthe scale and scope ofapplication ofthe instruments used. From theperspective ofearly
intervention, instruments tosupport restructuring can be divided by stage oftheir application:

Pre-emption and prevention (before layoffs),

Implementation ofremedial activities inenterprises (after layoffs).
A survey conducted in2011 by theEuropean Foundation fortheImprovement ofLiving and Working Conditions shows4 that
themost frequently used pre-emptive and preventive instruments tosupport restructuring include: counselling (30% ofthe
total), training (27%), other support fortheSME sector (26%), information concerning thesituation on thelabour market (20%),
access tofunding (17%), supporting thegrowth ofcompanies (17%), support forinnovation (14%), internationalization activities
(9%), territorial coordination (7%), and support forstart-ups (5%). Less frequently used instruments include: attracting investment,
pro-employment incentives, greater remuneration flexibility, needs analyses inthefield oflifelong learning, financial support
ofemployment inenterprises and increasing wage and working time flexibility (Hurley, Mandl 2011, p. 12).
4

The survey was conducted in27 EU countries and inNorway.

105

Conversely, themost popular instruments insupport ofrestructuring used at thestage ofimplementation ofremedial measures
include respectively: training (38% ofthe total), counselling (36%), benefits foremployees (27%), job matching (23%), increasing
working time flexibility (20%), support forstart-ups (10 %), access tofunding (11%) and pro-employment incentives (8 %). Other
instruments play a smaller role inthis respect.
In 2011, thefollowing European countries had themost numerous and diverse support instruments: Belgium (35 instruments),
France (25), Finland (24) Denmark (21) and Austria (19). Few instruments were available at thetime inRomania (5 instruments),
Slovakia (5) and Lithuania (6). Poland, with a 11 instruments identified inthesurvey, belonged tothegroup ofcountries using fewer
support instruments (Hurley, Mandl 2011, p. 12).
The best contemporary early warning systems are characterized by flexibility and innovation. Flexibility means theability tomake
changes inhow theinterventions already underway are being implemented, whereas innovation reflects their openness tononroutine, non-standard modes ofaction. Flexibility and innovation relate both totheinstruments used intherestructuring process
and totheprocedures fortheir implementation.
Examples ofinnovation include employee pooling, communication plans and pro-employment employee innovation. Employee
pooling comprises agreements among enterprises under which they jointly employ their staff and jointly cover theemployment
costs. Such pools are created mainly when companies need small quantities ofspecific, specialist labour and thedemand
forspecific jobs inindividual enterprises is cyclical. Employee pooling is popular, among others, inFrance, Germany, theNetherlands
and Belgium.
Communication plans include detailed communication programmes tonotify about planned redundancies, whose implementation
reduces thesocial impact ofsuch actions. Communication plans are already used inmany countries, but Spanish and British
solutions are considered tobe particularly interesting.
Pro-employment employee innovation, inturn, comprises incentives meant toencourage employees toseek and report ways
toprevent loss ofjobs ortocreate new ones intheir enterprises. Such mechanisms are used, among others, inDenmark and
Germany (Hurley, Mandl 2011, pp. 202, 232, 278).
Early warning systems are subject toongoing development, adaption tochanging expectations and economic change.
Characteristically, such systems evolve towards increasing therole ofenterprises intheearly response process. These trends
partly explain theinfluence ofethical action models derived from themainstream ofcorporate social responsibility. Enterprises
not only assume a passive role by submitting tostreamlining employment asrecommended by public services, but also become
theoriginators and leaders ofearly intervention, e.g. intheprocess ofbuilding local partnerships foremployment or, inthecase
ofcorporations, by initiating transnational agreements.5

8.4. Rapid response systems vs. early warning systems


the operation ofearly warning systems entails thecapacity tointervene once certain conditions have been met. In thevast majority
ofearly warning systems, signals that trigger interventions come directly from companies concerning impending difficulties. This
is usually done through themandatory filling and submission offorms specifying planned redundancies. Thus, early warning
interventions insuch cases are implemented incompanies which have both indicated their difficulties, and at thesame time
meet thecriteria forsupport (e.g. minimum number ofplanned layoffs). This kind ofrelationship between early warning and early
response systems is illustrated inthefirst quadrant ofTable 8.2 (Decentralized EWS and RRS).
Table 8.2. Types ofmodel relationships between early warning and rapid response systems
Early Response System

decentralized

centralized

Source ofsignals about threats enterprises


Intervention targets enterprise and employees

Source ofsignals about threats enterprises


Intervention targets economy, sectors ofeconomy

centralized

Early Warning System

decentralized

Source ofsignals about threats economy, sectors


ofeconomy (macro- ormicroeconomic data)
Intervention targets enterprise and employees

Source ofsignals about threats economy, sectors


ofeconomy (macro- ormicroeconomic data)
Intervention targets economy, sectors ofeconomy

Such initiatives were implemented e.g. by Vauxhall inEngland, European plants owned by Ford and Schneider Electric headquartered inFrance.

106

The Rapid Response Instrument developed inPoland involves theintegration ofa centralized early warning system (macro- and
microeconomic data) with theintervention ofan early response system undertaken ina decentralized manner, i.e. one that targets
specific companies. Thestrength ofsuch an approach lies intheavailability ofaggregated knowledge about a given situation
intheeconomy and its sectors, on thebasis ofwhich a decision is made tostart an intervention and determine its objectives
(the so-called starting mechanism). A weaker spot insuch an approach is, however, theinertia ofthe early response system. Firms
indistress do not receive support until at least several dozen days have passed.
According tothematrix, two other kinds ofmodel relationships between early warning and early response systems can be
distinguished. In thefirst one, information about thethreats come directly from enterprises (decentralized EWSs), while
interventions are directed towards theeconomy asa whole orits sectors (centralized RRSs). In thesecond one, information
about thethreats come from data on theeconomy and its sectors (centralized EWSs), and interventions are also directed towards
theeconomy orits sectors (centralized RRSs). It means, however, that theresponse occurs at thelevel ofeconomic policy whose
effects are naturally delayed intime and less appropriate than individual aid. For that reason, theobjectives ofthe previous two
kinds ofresponse should involve making rapid changes ineconomic conditions rather than rapid interventions on thelabour
market and inenterprises indistress.

8.5. Conditions foran effective implementation ofearly warning systems


The results ofanalysis ofinternational experience enable us toidentify thegeneral conditions affecting theeffectiveness ofRRS.
These conditions, tosome extent, are universal and, assuch, appear invarious early warning systems.
First, an RRS must be holistic, which means that itis capable ofperforming three mutually complementary functions:
a) Programming thedesign ofpublic aid packages forrestructuring actions tomitigate/eliminate therisk ofbankruptcy and
loss ofjobs,
b) Implementation implementation ofthe aid packages aimed at supporting therestructuring measures tomitigate/eliminate
therisk ofbankruptcy and loss ofjobs,
c) Evaluation ofthe effects ofpublic aid insupport ofthe restructuring measures designed toreduce/eliminate therisk
ofbankruptcy and loss ofjobs.6
Second, an RRS must be dynamic innature. In this context, theterm means both thecapacity toquickly take pre-emptive action
based on theinformation coming from an early warning system and flexibility ofsuch action (it should be founded on a wide range
ofdifferent instruments tofacilitate rapid response appropriate tothesituation). Thedynamic nature ofan RRS is its ability tolearn
and adapt inresponse tobroader economic and social changes intheenvironment inwhich thesystem operates.
Third, an RRS must be transparent. In particular, itis important toprovide a specific and clear division ofpowers and responsibilities
ofits participants and assign appropriate roles tothem (strategic, concerted action, coordination, implementation, monitoring).
It is also important toprovide clear rules and mechanisms ofcommunication within theRRS and between itand its environment.
Fourth, an RRS should be hybrid innature. On theone hand, itshould be characterized by a fairly high degree ofcentralization (in
thesphere ofstrategic decisions on thescope and form ofsupport provided tocertain sectors ofthe economy and monitoring
theoverall health ofthe economy). On theother hand, itshould be decentralized and based on local knowledge (in thefield
ofrecognition ofspecific needs and targeting appropriate implementation activities).
Fifth, an RRS must be based on thelogic ofnetwork interoperability. In such a system, itis essential toensure interaction among
central government and its agencies, regional and local structures ofterritorial government, businesses, employers organizations,
trade unions and organizations offering appropriate training and consultancy services. Each ofthose entities has theresources
(power, knowledge, financial resources, information, processing potential etc.) without theintegration ofwhich itis impossible
toachieve theobjectives forwhich an RRS was established.
Sixth, an RRS must be capable ofsystemic learning and evolution. This implies theneed forregular evaluations ofthe outcomes
ofthe measures implemented under thesystem and forthesubsequent modification ofits mechanisms based on theresults
ofsuch evaluations. In this case, itis advisable pursue theapproach known asevidence-based policy.
Seventh, an RRS must be credible. Its credibility is verified by thepromptness ofits interventions, their relevance totheneeds
ofentrepreneurs and their employees inaddition tolow transaction costs ofusing its services.

A function complementary tothethree listed here is themonitoring function. It is attributed toan early warning system and its purpose is
tostudy theeconomic and financial situation ofenterprises inorder toassess their condition and identify phenomena that may point toits
significant deterioration (early warning) aswell asmonitoring thestate ofthe economy asa whole.

107

Stanisaw Mazur, Marcin Zawicki

Chapter 9
RAPID RESPONSE SYSTEMS INTERNATIONAL EXPERIENCE
9.1 Introduction
This chapter outlines theobjectives and structures ofrapid response systems (RRS) operating intheUnited Kingdom and
intheUnited States. These systems constitute crucial elements intheportfolio ofstate interventions aimed at reducing thenegative
consequences ofthe economic cycle. Their anticipatory nature helps toavoid thehigh economic and social costs, which inevitably
arise if interventions are initiated after thefact. An RRS is an exponent ofmature, anticipatory impacts on thechanging economic
conditions and their consequences, inparticular forthelabour market. They provide a form ofaid toenterprises and their
employees affected by economic downturns and global financial crises. Theaid is meant mainly tosupport therestructuring
process.
This chapter begins with a discussion ofthe British experience ofRRS and then presents an overview ofsimilar solutions used
intheUnited States. Theanalysis covers theobjectives ofsuch systems, their components and mechanisms. In conclusion,
a synthetic assessment ofthe utility ofthese RRS is provided.
RRS descriptions inthis chapter are, toa certain degree, fragmentary. This is due totwo main reasons. First, theRRSs only tosome
extent assume theform ofa system intheclassical sense ofthe term. Often they comprise sets ofsolutions which facilitate
immediate reaction toeconomic processes and, assuch, they are implemented without an overall plan forstate intervention
intheeconomic sphere. Second, theliterature lacks studies which would address thematter at hand ina holistic manner.
Another issue dealt with inthis chapter concerns thequestion ofevaluating theeffectiveness ofRRS with a special focus on
theidentification ofthe key factors that influence theutility ofthese systems. These include reliability, speed, sensitivity, specificity,
transparency and communication.
Conclusions and recommendations forPoland based on theforegoing review make up thefinal part ofthis chapter.

9.2. TheRapid response service intheUnited Kingdom


The early warning system operating intheUnited Kingdom, called Rapid Response Service (RRS), deserves special attention among
theEuropean solutions ofits kind. It is offered and managed by Jobcentre Plus, an executive agency ofthe Department forWork
and Pensions specializing inlabour market services.
Jobcentre Plus is a network ofagencies operating throughout thecountry that offers comprehensive support services ranging from
job search tothedisbursement ofbenefits. Its business profile reflects its mission work forthose who can, security forthose who
cannot. The key objective ofJobcentre Plus is toprepare therecipients ofits services tocompete forexisting jobs and toprevent
long-term unemployment. Theagencys activities focus on improving job search strategies and on increasing employability
especially ofthose whose labour market situation is difficult forvarious reasons. Theagency is also committed topromoting
co-operation among public sector, market-based and non-government entities inorder toimprove its performance interms
ofachieving its objectives.
The Rapid Response Service was launched by Jobcentre Plus inApril 2002. 1 It was one ofthe activities undertaken aspart
ofWelfare-to-Work labour market reforms implemented by Tony Blairs Cabinet. Apart from theestablishment ofthe RRS, thereform
included decentralization ofemployment services, introduction ofthe minimum wage and linking welfare benefits with an active
job search. As part ofthe policy, active job search instruments were integrated with welfare support tools (Evans, Millar 2006, p.
6263).
The RRS is designed tohelp entrepreneurs planning collective redundancies, and through them, their employees affected, who
are themain recipients ofsupport. Themost important goal and priority ofthe service is toquickly step into action inorder tohelp
thelargest number ofemployees tofind a new job before they lose their current one. Thescope ofassistance offered toredundant
workers by theRRS is quite broad. In addition tobasic forms ofsupport, theRRS offers services which are rarely available topeople
who look forjobs inthetraditional way. This testifies totheactual primacy ofpolicies toprevent theloss ofjobs over theother
pro-employment solutions.
1

Certain elements ofthe service were offered before 2002 under thename ofthe Rapid Response Fund (RRF).

109

In order tofully understand theoperation ofthe RRS mechanism, we shall now review thestages that precede theactual initiation
ofthe service. Thepath involves thefollowing steps:

Notification ofplanned redundancies by a firm,

Consultations on thescale and principles ofplanned redundancies,

Contact ofJobcentre Plus consultants with theemployer todetermine thescope ofthe RRS tobe implemented,

The RRS steps into action.
The obligation toreport anticipated redundancies ina firm rests with theemployer. Such notification is made using theadvance
notification ofredundancies (HR1 form). Therequirement applies toplanned redundancies involving 20 ormore employees at one
establishment within a period of3 months. Thecompleted forms are submitted totheDepartment forBusiness, Information and
Skills. Then thedata concerning planned redundancies are sent toan institution called Insolvency Service 2 and then toRegional
Development Agencies3 and local Jobcentre Plus branches. (Jefferys, Clark 2009, p. 19).
According totheprovisions ofthe Trade Union and Labour Relations Act, thenotification ofplanned redundancies is followed by
consultations with theaffected employees.4 They are held inorder toexplore thepossibilities ofpreventing layoffs, and if this is
not possible, tominimize their effects.
The employer who intends toproceed with collective redundancies is obliged toconsult theplan with representatives
ofemployees affected by his action. If theemployees belong toa trade union, theemployer is also obliged toconsult theplanned
redundancies with therelevant unions. If no trade unions5 are active inthefirm inquestion, such consultations are carried out
directly with therepresentatives ofthe affected workers.
Consultations must take place before theformal notification ofemployees ofdismissals. When theplanned staff reduction affects
20 to99 people, theconsultation should begin at least 30 days before issuing thefirst dismissal notice. In thecase of100 ormore
redundancies, consultations shall begin at least 45 days before issuing thefirst dismissal notice.
For consultation purposes, employers are obliged toprovide information on thereasons ofthe planned reduction inemployment,
thenumber and characteristics ofemployees expected tobe dismissed, themethod forselecting employees fordismissal,
thedismissal procedure and theamount ofseverance pay if itis governed by other than statutory regulations. Themain purpose
ofsuch consultations is todetermine ways toavoid redundancies, reduce their scale ormitigate their effects. Theregulations
stipulate that theconsultations should be genuine and conducted with a view toreaching an agreement between theemployer
and therepresentatives ofemployees.
During theconsultations, a number ofsolutions are being considered that can prevent theforced redundancies. They include
finding employees who are willing tovoluntarily become redundant oraccept a shorter working time, early retirement, dismissal
ofself-employed staff ortemporary employees, restriction orabandonment ofovertime, withholding hiring new staff, filling
vacancies with staff already employed by thefirm, reduction intheworking time, temporary reduction ofsalaries, temporary leaves
and suggestion ofemployment ina different position.
In theevent ofbankruptcy ofan entrepreneur, a specialized government agency enters thescene. It takes over part ofthe
companys financial obligations toits employees. For example, theService disburses allowances totemporarily unemployed people
(e.g. on maternity leaves), money equivalents forunused vacation, compensation forunfair dismissal and statutory severance
pursuant tocourt decisions (Jefferys, Clark 2009, p. 1214).
If theresults ofconsultations with employees tobe made redundant ortrade unions have verified that dismissals are inevitable,
regional centres ofJobcentre Plus intervene. Thecentres employ RRS managers whose duties involve accepting, giving opinions
and processing aid applications and thecoordination ofactions undertaken by theadministration and its partners. Thebasic
condition forreceiving aid by a firm under theRRS involves its planned downsizing that may cause orworsen thesituation
on thelocal labour market. Managers oflocal Jobcentre Plus centres are authorized toassess thescale ofsuch risk and enjoy
a considerable autonomy inthis respect.
The provision ofRRS services begins even before actual dismissals take place. RRS managers liaise with theemployer and agree
on thetype and scope ofassistance. If theoffer is accepted, aid specialists visit thefirm inorder todetail thescope ofthe service
offered (Jefferys, Clark 2009, p. 23). If, however, dismissals have already started, eligible persons receive assistance under theRRS
fora period up to13 weeks from thetime ofdismissal (Robinson 2009, p. 5).
2

4
5

A government executive agency which specializes ininvestigating theissues ofbankruptcy and financial distress intheeconomy. It deals
with, among others, improving bankruptcy laws, limiting barriers toits use and thepromotion ofeffective bankruptcy procedures. TheBritish
Insolvency Service has 35 branches throughout thecountry.
Government agencies responsible forsupporting economic development and investment, enterprise development and economic
competitiveness, promote employment and improve professional competence at regional level.
The consultations are conducted under Part IV ofthe Trade Union and Labour Relations Act of1992.
In 2009, itapplied toca. 73 % offirms.

110

State support under theRRS scheme is extended toall employees at risk oflosing their jobs and includes, inparticular:

Those under threat ofredundancy, but still employed,

Those under notice ofredundancy, but still employed,

Those who lose their jobs incompanies which are inthesupply chain ofa larger company making redundancies,

Those who lose their job ina locality designated ashaving RRS status by theJobcentre Plus District Manager orCustomer
Services Director. (Provider Guidance, 2010).
The Rapid Response Service has been designed asa flexible tool. In practice, itmeans that theinventory ofservices provided
by itis unlimited. As a rule, RRS support theprocess ofmanaging redundancies infirms and help employees about tolose their
job primarily through a wide range ofprofessional counselling. Theprocess includes providing information on available forms
ofsupport, welfare benefits and pensions, self-employment counselling, assistance inthepreparation ofCVs, cover letters, job
interviews and job search.
The fundamental element ofthe package offered by theRRS toemployees is a high-quality personalized analysis ofskills and
competencies (Skills Transfer Analysis). It helps todetermine towhat extent theexisting employee qualifications meet theneeds
oflocal labour markets and toidentify theskills tobe acquired, which are likely toimprove thecompetitiveness ofpeople losing
their jobs. If theresults ofthe analysis indicate theexistence ofskills gaps, theRRS framework offers access totraining and education
programmes (Job Focused Training). Measures enacted under such programmes may involve retraining, but only if there is excess
demand over supply ofskilled employees on a given market. Thevocational training offered allows those interested, among others,
toobtain skills certificates which are not subject toconfirmation by other official documents.
Eligible people may also be granted financial support intheform oftravel expenses (to work) and toovercome other barriers
toobtaining gainful employment, such asthepurchase ofspecialist tools orclothing. Funds forthese purposes come from
a separate operational fund (the Action Fund). If, however, theexisting forms ofsupport prove tobe insufficient, inaccordance
with theprinciple ofopenness and flexibility ofRRS, special solutions may be applied which are offered ordesigned with
theparticipation ofthe Specific Programme Support Centre.
On thepremises offirms planning a significant reduction inemployment an ad hoc consultancy and job centre can be established,
which inthecase ofshift work, may be active up to24 hours a day (Provider Guidance 2010, p. 913, Robinson 2009, p. 3).
Since its launch, thebudget ofthe Rapid Response Service has steadily grown. Thehighest growth rate was noted in20072010.
In the20072008 fiscal year, its budget totalled 3 million pounds todouble only a year later and inthe20092010 fiscal year
itamounted to12 million pounds (Jefferys, Clark 2009, p. 24).
The overall conclusion is that after more than 10 years ofits operation UKs Rapid Response Service constitutes a useful tool that
has lived up tothehopes ofits creators. Themost immediate and tangible results ofthe existence ofthe RRS is its real impact
on reducing thenumber ofjobs lost and prevention ofthe dangerous and costly labour market phenomenon oflong-term
unemployment.
The Rapid Response Service has also become an important source ofstrategic change fortheeconomy asa whole. More and
more stakeholders ineconomic policy believe inthedesirability ofanticipatory approaches and tools, aimed at preventing and
reducing thenegative consequences ofeconomic turmoil. TheRapid Response Service is also subject toconstant evolution and
development. For several years, a considerable emphasis has been placed on building theright atmosphere and theformation
ofa mature relationship between employers and employees ofenterprises undergoing restructuring. New elements ofRRSs
procedures already contain a number ofmechanisms reflecting innovative developments ingovernance and management
psychology. They consist, forexample, inencouraging employers topay attention not only toeconomic but also toemotional
aspects ofjob cuts, taking notice ofthe moods and feelings ofthe affected workers, giving them support infinding new
employment, demonstrating concern fortheir present and future, ensuring proper communication with thepublic, trade unions
and other external stakeholders aswell aspreventing emotional isolation ofthe affected workers. RRSs procedures attempt
tosupport theemployees ofrestructured enterprises. This involves thedissemination ofinformation about new opportunities
involving employment change, consideration fortheinterests and general well-being ofemployees and thelocal community,
stimulating participation inmanaging therestructuring process, which basically aims toallay thefears and reduce uncertainty
ofstaff (How tomanage 2013, p. 2931). Thesystems flexible procedures, but above all, its evolution dynamics consistent with
therhythm ofsocial and economic change, makes theBritish rapid response system all themore remarkable forthose who wish
todevelop similar systems at home.

111

9.3. Theearly response system intheUnited States


The history ofefforts tobuild an early response system (ERS) intheUS economy dates back tothe1950s. Theefforts intensified
inthe1970s and the1980s due totherestructuring ofheavy and automotive industries, which led toa marked increase
inthenumber ofredundancies. In order toprevent increased unemployment, instruments aimed at both protecting jobs and
reallocation ofresources and labour were used.
An important step inthedevelopment ofan RRS intheUnited States was theTrade Act of1974. It imposed, among others, on
companies planning torelocate their activities outside theUnited States an obligation todeliver prior notification ofsuch an
intention. This information was tobe communicated totheemployees concerned and theSecretaries ofLabour and Commerce.
Implementation ofthe Acts provisions was accompanied by much controversy and concern, ofwhich they key ones involved
theintroduction ofinefficient and bureaucratic mechanisms intheUS labour market (Levine 2007).
Another important stage inthedevelopment ofRRSs solutions came with thepassage ofthe Worker Adjustment and Retraining
Notification Act of1989.6 It imposed on employers with over a hundred people, among other things, theobligation toinform about
planned redundancies (involving more than fifty people) orabout a planned closure ofa firm thefollowing entities: theemployees,
thestate rapid response unit and thelocal government. It was decided that such information should be provided at least sixty days
before theplanned action. This Act constitutes thecore ofthe rapid response system by mandating early notification ofproblems
and theplanned solutions, which offered specialized public institutions an opportunity totake pre-emptive action.
The principles ofinteraction between businesses and public authorities, asprovided by theWorker Adjustment and Retraining
Notification Act, have been further elaborated inthedocuments adopted by thestate authorities. As a rule, they clarify and often
introduce stricter notification criteria related totheprovision ofinformation concerning planned redundancies.
In theEmployers Guide prepared by theDepartment ofLabour, thesolutions contained intheWorker Adjustment and Retraining
Notification Act was summarized asfollows: In our dynamic economy, many companies are streamlining their operations
tomaintain a competitive position inthemarketplace. Although such actions can help your company become more efficient,
this may result intheelimination ofexisting jobs and facilities. In 1988, Congress passed theWorker Adjustment and Retraining
Notification (WARN) Act toprovide workers with sufficient time toprepare forthetransition between thejobs they currently hold
and new jobs. This transition may involve theprovision ofinformation about where new jobs may be found, oritmay involve
providing workers with other employment orretraining opportunities before they lose their jobs.7
The next step intheprocess ofbuilding theRRS was theEconomic Dislocation and Worker Adjustment Assistance Act adopted
in1988. It imposed on state authorities theobligation toestablish specially designated units toaid employees changing their place
ofwork, within which rapid response units were tobe organized (Salzman, 2009).
Experience with theoperation ofthe RRS inthe1980s and the1990s showed its fragmentation, low coordination potential and
redundancy ofadopted measures. Those critical insights led totheadoption ofthe Workforce Investment Act in1998. Its key feature
constituted theestablishment ofrapid response units (RRU) by state administrations. Their range ofactivities includes:

Conducting on-site visits (immediately after receiving information about planned redundancies) inorder toassess thescale
ofredundancies, thepossibility ofavoiding them, alternatives, potential needs interms ofretraining, job opportunities on
thelocal labour market and available resources,

Informing redundant employees about therange ofservices that they can use, including unemployment insurance,
employment centres and integrated assistance offered by Trade Adjustment Assistance/NAFTA Transitional Assistance
programme, providing financial support toemployers affected by importer competition,

Setting up local committees toimplement strategies aimed at assisting theemployees offirms indifficulty,

Supporting local administration insolving current problems and helping create comprehensive strategies forsolving
problems related tothelayoffs (including theapplication fora special federal grant called theNational Emergency Grant).
In addition, theRRU aspart ofstate administration can perform additional tasks, including:
1. Together with other public entities, employers organizations and trade associations they may develop labour market
intervention strategies inthecontext ofrising unemployment risks, create programmes todiscourage theuse ofredundancies,
develop instruments toexchange information concerning theplans ofliquidation ofenterprises ortools available tohelp
workers at risk ofredundancy.
2. Together with thecompetent state authorities, they may collect and analyse information related tothedisturbing phenomena
that occur on thelabour market, with a particular emphasis on layoffs and plant closure, aswell asaggregate information
6

Its adoption, aswas thecase with theTrade Act, was accompanied by concerns about theUS labour market losing its flexibility, poorer
economic growth and a reduced job creation capacity (Addison, Blackburn in1994).
The Worker Adjustment and Retraining Notification (WARN) Act, Employers Guide toAdvance Notice ofClosings and Layoffs, United States
ofDepartment ofLabor, July 2003.

112

3.
4.

inorder tocreate thebasis foreffective management ofthe rapid response programme, aswell asdevelop control and
evaluation tools forsuch programmes.
Engage ineducational activities toimprove thequality ofassistance toworkers affected by redundancies.
Participate inthepreparation and implementation ofstrategies for:

The prevention ofcollective redundancies (including thefunding ofstudies ofthe possibilities ofrestructuring affected
firms ortheir acquisition by other operators, including theits employees),

Retraining,

The provision ofloans toworkers with a view toimproving their qualifications.

The Workforce Investment Act specifies thesequences ofactions within theearly response system. They consist of:
I. The diagnosis: an analysis based on thedata on planned redundancies provided by firms, thesecond information channel
makes use ofthe results oflabour market research (and theeconomy asa whole) inthefollowing respects:

Analysis and evaluation ofthe impact ofdisturbances caused by thefinancial and economic turbulence on thelabour
market. Theaim ofthe analysis is toidentify thevulnerable sectors ofthe economy and toensure theaccess ofstate
and local administration tothenecessary information inorder totake action toreduce thethreat ofcompany failure,

Evaluation ofregional and national economy inorder toidentify opportunities tooffset job losses inother industries
and/or inother areas,

Devising employee competence profiles inorder toassess their chances offinding another job intheregion,

Opportunities toprovide education and training services intheregions forpeople at risk oflosing their jobs.
II. Impact assessment ofthe effects ofa given companys failure on thelabour market using an econometric model. Themodel
is based on input-output analysis and, once certain assumptions have been adopted, allows foran assessment ofthe impact
on a given regions economy ofthe elimination ofa certain number ofjobs inspecific firms. Theimpact is measured interms
oflost jobs and declining revenues. This enables theadministration toperform a cost-benefit analysis ofindividual policy
options. Themodel is supplemented with information from theemployers affected by thefinancial and economic disruption
(e.g. wage levels, lists ofsuppliers).
III. Analysis and evaluation ofthe regional economy inorder toidentify employment opportunities inother sectors ofthe
economy, itbeing understood that such action will strengthen theregional economy and protect itagainst new
disruptions. Thelikelihood offinding an alternative that meets all ofthese criteria is low, but theextent towhich they are
fulfilled determines thepossible options. Thefirst stage involves thesearch forcompanies inthesame ora related industry
totheone inwhich thefailing firm operates, which narrows down thescope ofthe necessary adaptation action. If, however,
thedifficulties affect theentire sector ofindustry, efforts should be redirected toother, more promising sectors. In order
tominimize retraining costs, industries which employ workers with similar skills are sought (an analysis ofkey professional
groups intheliquidated company is followed by application ofthe industryoccupation matrix). An inputsresults model
may help inthesearch forindustries complementary totheregional structure ofthe economy.
IV. Profiling employees with respect totheir profession. In order toprovide effective assistance toworkers at risk ofredundancy,
theanalysis includes thefollowing data: employment history, education, skills and qualifications, individual preferences,
readiness tochange their place ofresidence etc. Understanding thestructure ofemployment ofthose at risk ofredundancy
according totheSOC (Standard Occupational Classification) system allows fortheidentification ofindustries capable ofhiring
laid-off workers with minimal retraining. Statistical data collected allows foran assessment where inthestate orneighbouring
states selected industries are represented, and subsequently, based on theNAICS (North American Industry Classification
System) statistical register specific employers, potentially interested inemploying workers can be chosen using theavailable
databases such asinfoUSA.
V. Profiling employee qualifications interms ofemployers expectations. Thesimplest method is toidentify theskills required
toperform a specific job based on theSOC-O*NET model that contains 6 major domains ofinformation (employee profile,
expectation ofemployees including skills and education, professional experience, specific job position and its attributes,
characteristics ofthe position from thepoint ofview ofthe parameters ofthe labour market and thesocial and cultural
context ofthe job). Similarities interms ofthe above-mentioned attributes help intheidentification ofprofessions that require
theleast outlay on training. Likewise, employees themselves may individually use thejob evaluation tests inorder todevelop
a list ofrelated occupations forwhich they could be retrained.
VI. Diagnosis oftraining opportunities at post-secondary level. Preliminary testing can be performed using theIPEDS
(Integrated Post-Secondary Education Data System) database maintained by theDepartment ofEducations National Centre
forEducation Statistics, towhich all post-secondary schools submit mandatory reports. Theidentification ofopportunities
ina given area may include thecurrently available programmes and their training opportunities aswell astheprospects
fortheimplementation ofnew programmes tailored totheneeds ofworkers and employers. Where employees may have
trouble finding a job corresponding totheir experience and skills, thekey measures include vocational training and theuse
ofrapid response instruments.

113

The advantage ofthe Workforce Investment Act is that itlays down therules fortheimplementation ofhuman resource
development programmes focused on workers. All investment inhuman capital should meet thefollowing requirements:
1. Training, educational and employment services should lead tothecreation ofquality jobs that guarantee themaintenance
ofstable families and local community development.
2. The services should be administered by public institutions that provide fair and equal access toall employees.
3. They should be used toreduce theexisting income gap intheeconomy, while creating conditions conducive
totheimplementation ofcareer paths through access tobetter paid and more skilled professional jobs.
4. Educational services inthefield ofvocational retraining and reintegration offered toadult workers should guarantee longterm improvement incareer opportunities and earnings.
5. Support provided toyoung people should be based on guidance from adults. In addition, itis necessary tocreate conditions
conducive totheacquisition ofacademic knowledge and professional experience by young people that foster their
development, and not constitute a tool todelay their entry inthejob market.
6. Services should be delivered inan environment where theworkers inneed occupy a central place, and their rights, dignity
and privacy are fully respected.
7. Users ofthe service should be informed ofthe employee rights intheworkplace and other instruments oflegal protection
such astheFair Labour Standards Act, theOccupational Health & Safety Act, theNational Labour Relations Act and theCivil
Rights Act.
8. Providers ofthe services should respect and uphold theprinciple ofparticipation ofall stakeholders employees, employers,
educational institutions and training local organizations both inthecourse ofpreparation ofprogrammes and inthedelivery
ofservices.
In order toensure effective implementation ofthe solutions contained intheWorkforce Investment Act, theDepartment ofLabour
established working groups, including theNational Rapid Response Group. Its role is toprovide support aswell asto coordinate
thecomponents ofthe rapid response system intheUS economy (Salzman 2009).
In summary, thelegislation governing theoperation ofUS companies provides a set oftools which aim toprovide systemic
support toemployees facing therisk ofredundancy. Thelegislation has allowed fortheestablishment ofa comprehensive RRS. Its
effectiveness is ensured by advanced research tools and procedures fordata transmission, legitimacy provided by thelegal system,
efficient coordination mechanisms and thecapacity toutilize experience accumulated thanks toits previous actions. This system
has a strong institutional legitimacy intheform ofRapid Response Units, whose activity allows formulti-faceted impacts aswell
aspre-emptive and effective response toemerging issues.

9.4. Conclusions and recommendations forPoland


The results ofresearch and analyses can be combined toprovide six key recommendations, which, inour opinion, should be taken
into consideration when developing a RRI inPoland. They include thefollowing:
1. The RRI should be based on theassumption that itis only authorized tocounter negative changes intheeconomy caused
by periodic fluctuations and economic downturns. Such a system should not be used torespond tostructural changes
intheeconomy.
2. Actions and measures contained ina RRI are subject toa number ofwell-defined conditions. While avoiding excessive
intervention ofstate aid inmarket economy mechanisms, they should also contain safeguards against abuse by individual
firms.
3. The planned RRI include three components:
(I) Programming,
(II) Implementation,
(III) Evaluation.
(I) The aims ofthe programming component should be toformulate recommendations forpublic policies, toprepare policy
plans forthepurposes ofthe decision-making process and todesign support instruments.
(II) The implementation component should serve toexecute public interventions relevant tothediagnosis ofrisks and
tocarry out actions insupport ofrestructuring firms indifficulty. Thepotential range oftools available topublic
authorities inthis regard is quite broad. Theimplementation tools most relevant tothecurrent economic situation
include:

Community framework forstate aid measures tosupport access tofinance,

The Second chance policy tools,

Instruments toassist therestructuring offirms indifficulty.

114

(III) The evaluation component should involve a scrutiny ofimplemented policies. It is assumed that all theindividual tools
ofthe Rapid Response Instrument shall be evaluated. Therationale forconducting such evaluations is dictated by
theneed forongoing improvement ofinterventions and their adaptation totheactual circumstances. Another task
subsumed under this component involves theperiodic evaluation ofthe functioning ofthe RRI inits two essential
dimensions: substantive and procedural-organizational.
4. The RRI architecture should be multi-layered and should consist ofdifferent entities with various functions and resources that
are complementary toone another. The five levels ofinstitutional architecture should include:
(I) The Prime Minister and theCouncil ofMinisters,
(II) Appropriate ministers, particularly ofeconomy, finance and regional development,
(III) Institutions which report totheMinister ofEconomy (e.g. theAgency fortheRestructuring ofIndustry), especially
thePolish Agency forEnterprise Development and thefuture Economic Change Monitoring Centre established within
its structures,
(IV) Regional (voivodship) governments with their associated regional development agencies and local labour market
institutions,
(V) Employers organizations, trade unions and trade representations.

The institutional actors listed above should have thefollowing roles inthesystem:
a) The Prime Minister and theCouncil ofMinisters: setting thestrategy,
b) Appropriate ministers, particularly ofeconomy, finance and regional development: coordination,
c) Institutions that report totheMinister ofEconomy (especially thePolish Agency forEnterprise Development): analyses,
projections and coordination,
d) Local and regional governments, development agencies, labour market institutions: consultancy and implementation,
e) Employers organizations, trade unions and trade representations, including thebanking sector: information and
consultancy.
5. The primary operator ofthe Rapid Response Instrument is a unit operating unit within thePolish Agency forEnterprise
Development (it has appropriate resources interms ofstaff and experience). We propose tocall ittheEconomic Change
Monitoring Centre.
6. In order toensure theeffectiveness ofthe system, itshould be holistic, dynamic, transparent, hybrid, network-based, capable
oflearning and evolution, and reliable (the cited characteristics have been described inmore detail above).

Kamil Fijorek, Jarosaw Kaczmarek, Konrad Kolegowicz, Pawe Krzemiski

Chapter 10
BANKRUPTCY RISK ASSESSMENT A MICROECONOMIC PERSPECTIVE
10.1. Introduction: purposes and features ofearly warning systems and rapid response
systems
From themicroeconomic perspective, forresearch purposes, a crisis is seen asa culmination ofadverse events, asa turning point
between development stages, which distorts orthreatens normal operation ofan enterprise. Businesses vary interms ofexposure
and vulnerability tosuch a threat and theparameters may be subject tomodelling. As numerous studies demonstrate, Polish firms
are quite vulnerable inthis respect, which may increase thecontagion effect intimes ofcrisis.
A crisis is not an aberration either forfirms ortheeconomy asa whole and can be subject torisk management. Thelatter can
be effective if itconstitutes part ofan integrated solution. A crisis may have negative impacts, possibly leading tobankruptcy,
aswell aspositive ones a threat toa firms existence may be turned into an opportunity forits development through restructuring.
Determinants ofcrisis include numerous factors that differ interms oftype and source oforigin. A characteristic feature ofa crisis
is its complexity and thefact that itis caused by a co-occurrence ofseveral factors ina sequence ofevents based on identifiable
cause-effect relationships aswell asa clear escalation path. External factors associated with thestages ofthe life-cycle ofsectors
ofindustry and their economic condition are particularly dangerous. Even intensive individual remedial actions undertaken by
a firm indifficulty do not guarantee its survival.
Symptoms ofa crisis ina firm appear asa result ofdiverse internal and external causes and usually do not occur inisolation. Such
symptoms tend toinclude failures inits general condition and management. Theproblem is that crisis has a rather nebulous
character itcan be felt rather than described, moreover, its symptoms are difficult toquantify. For thepurposes ofthe present
study, such symptoms have been defined forthearea ofthe financial condition ofthe firm with reference totheeconomic aspect
ofits behaviour.1
In order tocounter crises effectively, we must possess thecapacity toanticipate and prevent. Theformer condition was
a prerequisite forthedevelopment ofthe concept ofan Early Warning System (EWS). It is an integrated tool formeasuring risk
focused on theidentification ofchanges intheexposure tofinancial risk that may entail bankruptcy ofthrough theidentification
and quantification ofsymptoms asstructured financial signals.
As a tool, theEWS is integrated into thefirst ofthe four RRS functions themonitoring component, whose aim is toscrutinize
thefinancial situation ofundertakings inorder toevaluate their condition and identify phenomena that may indicate its significant
deterioration (early warning function).
The degree ofrisk subject tomeasurement expresses thecategory ofbankruptcy ineconomic terms, inother words, asthefinancial
risk offailure ofa firms operation and bankruptcy (bankruptcy inits broad sense, also called financial distress). Theevaluation
also addresses thephenomenon ofbankruptcy inlegal terms (in its strict sense, also called insolvency proceedings), i.e. courtsupervised bankruptcy proceedings.
The measures ofbankruptcy inits economic sense and inits legal sense, respectively, cannot be related without making
a distinction between their substantive content. This does not mean, however, that there is no correlation between them. This
relationship is thesubject ofresearch, moreover, itallows forcomparative analyses offirms using a measure offinancial distress
expressed inunits corresponding tothepercentage ofcourt insolvency proceedings.
The primary task ofthe EWS RRI is toevaluate therisk offinancial distress using original methodologies that involve empirical
induction and discriminant analysis (multivariate logit models). Conclusions are reinforced by themulti-criteria evaluation ofthe
potential and financial results ofa firm inquestion supported by data from court insolvency proceedings.
In order toevaluate thedegree offinancial distress theEWS RRI prediction models apply classic and Firth logistic regression. This
is due totheweakness ofthe previous prediction models (shortcomings intheir design and applicability limitations), especially

Further discussion ofthe impacts ofcrisis on firms, its characteristics, determining factors and symptoms can be found, among others, in:
(Mitroff 2001, p. 5), ( Hamel, Prahalad 1994, p. 134), (Smart 1978, no 9 p. 4648), (Mczyska 2011, p. 14), (Gierszewska, Romanowska 1995, p.44),
(Wieczerzyska 2009, p. 35), (Altman 1983b, p. 40, 180), (Argenti 1976, p. 123, 132135), (Wawrzyniak 1999, p. 112118), (Zavgren 1983, nr2,
p.133).

117

inthecontext oflearning from very small data sets. Conducting population surveys ofthe institutional sector ofnon-financial
undertakings numbering close to50 thousand entities thus necessitated thedevelopment ofnew research tools.2
The measure obtained using thetools listed above thedegree offinancial risk (PR and MW) reflects thelikelihood ofbankruptcy
ofanalysed firms within one year. It facilitates a quantitative description ofthe dynamic rate ofchange inthearea ofbankruptcy
and a comparison ofthe degree ofrisk among theundertakings ofdifferent sizes (total, small, medium-sized, large) and generic
groups offirms (total, manufacturing, trade and services). Themodels previously constructed and described intheliterature do
not possess such properties, moreover, thenewly-developed discriminant models may be considered unique inPoland thanks
totheapplication ofinnovative methods and certain detailed techniques.
Due tothespecific function ofthe RRI monitoring component inmicroeconomic terms (ongoing analysis and diagnosis ofthe
situation offirms), studies inthearea ofmonitoring condition offirms using theEWS RRI have several overlapping aims:

To define thesymptoms offinancial distress ina firm,

To develop methods and tools forthequantification ofemergencies described by specific symptoms ofthreats,

To evaluate thedegree offinancial distress using multivariate discriminant models,

To evaluate thecourse, intensity and structure ofbankruptcy offirms inlegal terms,

To conduct a multivariate analysis oftrends inthevalue and structure ofthe measures ofpotential and financial performance
offirms.
The type, method and approach employed intheEWS RRI are characterized by thefollowing features:

Its task is toreveal thedeteriorating situation offirms, inparticular thedetection and measurement ofwarning signals
indicating thethreat offinancial distress,

It serves toassess thedegree ofrisk represented by thelikelihood ofbankruptcy within one year and scaled inunits
corresponding tothepercentage ofcourt insolvency proceedings,

It is not a predictor ofbankruptcy, thewarning signals identified should be interpreted asa component ofthe warning
forecast.
The premises on which theEWS RRI was founded, its objectives and features aswell asthemethods and tools that make up
thepath ofstudy offinancial distress asapplied tonon-financial undertakings determine thescope and manner ofpossible
deduction procedures and assessments formulated on their basis, taking into account theprinciples ofperiodization.

10.2. Characteristics ofmicroeconomic data sources


The operation ofthe EWS RRI, theextent and frequency ofobservations, analytical itemisation and classification criteria are
determined by theplatform that feeds information and numerical data into thesystem. Its core resource is constituted by
thedatabase created and processed by theCentral Statistical Office (GUS) inWarsaw. This resource is unique interms ofdescribing
theresearch methodology, thescope and quality ofthe results obtained. It contains information and numerical data on individual
firms present inthenational statistical system. Theresources are processed toachieve theobjectives ofEWS RRI observing
theprinciples ofstatistical confidentiality.
The group ofeconomic entities included intheanalysis consists ofnon-financial firms with 10 ormore employees, which have
submitted themandatory statistical report (referred toasF-01/-01, F-02 and SP forms). As at theend ofthe second half of2013,
there were 51 269 such firms. Theshare ofmanufacturing, trade and services enterprises intheoverall population ofnon-financial
firms employing more than 10 people was at least 93.4% interms ofthe basic economic categories (number ofemployees 95.3%,
total assets 93,4%, sales revenue 96.5%, net profit 99.8%, equity 93.8%, added value 96.6%, capital outlays 94.5%). Additionally, their
share inadded value ofthe whole institutional sector ofenterprises inPoland (broadly speaking, interms oftheir contribution
toGDP) was 91.7% (total) and 88.5% (manufacturing, trade, services). Thebalance was theshare ofeconomic entities with
up to9 employees, whose 4% random sample was included inthenational statistical survey. These parameters indicate that
theconclusions based on thesurveyed population should be treated ashighly reliable and, assuch, they justify thegeneralizations
made with reference totheentire institutional sector offirms inPoland.
The facts and figures that characterize individual firms within thedata set processed by theGUS inWarsaw include three basic
groups:

General information and identification (statistical unit number, type ofeconomic activity according toNACE, ownership form,
organizational and legal form, location (region), activity status),

Numerical data reflecting thefinancial situation ofthe firm (set of29 relative measures),
2

The concept of evaluating the degree of financial distress, including the paths and research methods along with their updates, are discussed
in detail in: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (May 2011), Kaczmarek, Kolegowicz, Krzemiski, Fijorek (June 2011), Kaczmarek,
Kolegowicz, Krzemiski, Fijorek (December 2013).

118

Numerical data describing thedynamics ofchange inkey financial figures (set of10 measures assessing time series with
a variable base).

The analysis, evaluation and conclusions related tothefinancial standing aswell asfinancial distress offirms are carried out
inthemultidimensional space ofnon-financial enterprises described by thefollowing criteria:
Firm size, i.e.:

small (1049 employees),

medium-sized (50249 employees),

large (250 employees ormore),
Type ofactivity:

manufacturing (NACE divisions 539),

trade (NACE divisions 4547),

services (NACE divisions 4143, 4968, 7782, 9496),
NACE sections (i.e. theselected branches ofthe national economy):

industry (NACE sections: B, C, D, sections 535),

construction (NACE section range: F, sections 4143),

wholesale and retail trade (NACE sections range H, sections 4547),

transportation and storage (PKD scope ofSection H, sections 4953),

information and communication (NACE section range: J, sections 5863),
NACE divisions (32 manufacturing ones, 3 trade and 27 services, ranges asin thetypes ofactivity),
Regions (16 voivodships).
Another layer ofinformation and numerical data is provided by theeconomic and financial parameters describing thecondition
ofindividual NACE divisions (economic background intheform ofsectoral analyses), NACE sections and kinds ofactivity taking into
account thecriterion offirm size and its location (voivodship). These are processed, analysed and evaluated using databases provided
by theInfo Gospodarka SiDP Pont Info Ltd. system inWarsaw, whose primary data sources are theGUS databases inWarsaw.
The third layer ofinformation supply is based on theCourt and Economic Monitor ofthe Ministry ofJustice and Coface Bankruptcy
Reports. Thedata collected, processed and analysed characterize thebankruptcy processes interms ofinitiated insolvency
proceedings under theBankruptcy and Restructuring Act of28 February 2013 (Journal ofLaws 20013, No. 60, item 535).
The periodization ofresearch within theEWS RRI and presentation ofits results is determined by thecyclical nature ofnational
statistical surveys conducted by GUS. Theresults ofsurveys offinancial distress intheform ofreports are compiled on a quarterly
basis starting from thefourth quarter of2010 until thethird quarter of2013. Theresults formedium-sized and large enterprises
are updated quarterly, whereas small enterprises and enterprises total update thedata semi-annually.
For thepurpose ofcomparison and detection oftrends, theanalysis involves apart from thecurrent period a retrospective
approach starting inthefirst quarter of2007. As regards theforecasting oftime series ofvariables (factors) fortherisk prediction
models, a short-term annual forecast period was adopted with quarterly and semi-annual updates, respectively (for individual
classes offirms).
Table 10.1. NACE divisions that provide theinformation and basis forresearch conducted aspart ofthe EWS RRI
NACE division

Description

05

Mining ofcoal and lignite

08

Other mining and quarrying

09

Mining support service activities

10

Manufacture offood products

11

Manufacture ofbeverages

12

Manufacture oftobacco products

13

Manufacture oftextiles

14

Manufacture ofwearing apparel

15

Manufacture ofleather and related products

16

Manufacture ofwood and ofproducts ofwood and cork, except furniture, manufacture ofstraw and plaiting materials

17

Manufacture ofpaper and paper products

18

Printing and reproduction ofrecorded media

19

Manufacture ofcoke and refined petroleum products

119

NACE division

Description

20

Manufacture ofchemicals and chemical products

21

Manufacture ofpharmaceutical, medicinal chemical and botanical products

22

Manufacture ofrubber and plastics products

23

Manufacture ofother non-metallic mineral products

24

Manufacture ofbasic metals

25

Manufacture offabricated metal products, except machinery and equipment

26

Manufacture ofcomputer, electronic and optical products

27

Manufacture ofelectrical equipment

28

Manufacture ofmachinery and equipment n.e.c.

29

Manufacture ofmotor vehicles, trailers and semi-trailers except motorcycles

30

Manufacture ofother transport equipment

31

Manufacture offurniture

32

Other manufacturing

33

Repair, maintenance and installation ofmachinery and equipment

35

Electricity, gas, steam, hot water and conditioning supply

36

Water collection, purification and supply

37

Sewerage and wastewater treatment

38

Waste collection, treatment and disposal activities, materials recovery

39

Remediation activities and other waste management services

41

Construction ofbuildings

42

Civil engineering

43

Specialized construction activities

45

Wholesale and retail trade and repair ofmotor vehicles and motorcycles

46

Wholesale trade, except ofmotor vehicles and motorcycles

47

Retail trade, except ofmotor vehicles and motorcycles

49

Land transport and transport via pipelines

50

Water transport

51

Air transport

52

Warehousing and support activities fortransportation

55

Accommodation

56

Food and beverages service activities

58

Publishing activities

59

Motion picture, video and television programme production, sound recording and music publishing activities

60

Programming and broadcasting activities

61

Telecommunications

62

Computer programming, consultancy and related activities

63

Information service activities

64

Financial service activities, except insurance and pension funding

66

Activities auxiliary tofinancial service and insurance activities

68

Real estate activities

77

Rental and leasing activities

78

Employment activities

79

Travel agency, tour operator, reservation service and related activities

80

Security and investigation activities

81

Services tobuildings and landscape activities

82

Office administrative, office support and other business support activities

94

Activities ofmembership organizations

95

Repair ofcomputers and personal and household goods

96

Other personal service activities

Source: own study.

120

10.3. A system forevaluating thefinancial condition offirms


Using multifaceted, cross-sectional information and figures characterizing thepopulation ofnon-financial enterprises aswell
asthenew models and specific measures, themain structure ofthe EWS RRI was developed, complete with relevant research
paths. Theconceptually coherent structure with its appropriate instruments, structural elements, objectives ofscrutiny and an
organizing factor intheform ofa defined measure ofthe degree offinancial distress meets therequirements ofan integrated
(systemic) solution.
The EWS RRI structure comprises six planes ofscrutiny and evaluation ofnon-financial undertakings interms oftheir financial
condition and thedegree ofrisk:

An analysis ofchanges ingroups enterprises, their potential and financial performance,

An analysis ofinstituted insolvency proceedings supervised by courts,

A multifaceted analysis ofchanges inthefinancial condition offirms,

An analysis ofthe degree offinancial distress,

An analysis ofrisk factors faced by individual NACE divisions (the so-called TOP 10 ofhighest risk),

Developing forecasts offactors driving change and thedegree offinancial distress.
Each ofthe above-listed EWS RRI planes ofscrutiny constitutes a separate path ofinquiry which measures, assesses, classifies and
predicts thedegree offinancial distress. Each plane has well-defined, measured and analysed key exponents (comprising its path
ofinquiry) aswell asits characteristic evaluation process.
In terms ofchanges affecting thepopulation ofenterprises (demographics), their potential and financial performance, thefollowing
are subject toevaluation:

The number offirms,

The number ofemployees,

Sales revenue,

Investment outlays,

Total assets,

Short-term liabilities,

Receivables and claims,

Net financial result.

Measurement
and evaluation

Financial condition

Classification

Bankruptcy
proceedings

Dziay I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II
PKD 2007 2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013
HANDLOWE

45
46
47

8 4

6 2

8 5

7 9

8 3

8 7

1 0 5

7 3

1 0 0

1 0 7

8 5

1 2 7

1 3 2

9 7

1 2 0

1 0 9

1 4 2

1 0 5

1 1 4

8 7

1 2 8

1 1 4

1 7 6

1 5 9

1 8 4

1 1 8

9 4

1 2 4

1 0 5

1 3 4

1 1 8

1 2 9

1 6 3

1 0 7

1 3 0

1 1 9

1 5 2

1 2 2

1 5 9

1 9 1

1 4 2

1 5 9

1 5 3

1 7 5

1 6 5

2 4 8

2 0 7

2 6 5

2 3 2

2 9 7

2 8 6

3 3 3

2 6 7

1 3 4

2 8 3

1 8 8

3 1 9

1 5 2

3 3 6

1 8 3

3 7 8

2 6 3

5 5 9

3 5 6

5 4 3

1 9 3

USUGOWE

Degree of risk

Common plane of
measurement

41
42
43
49
50
51
52
53
55
56
58
59
60
61
62
63
66
68
78
79
80
81
82
94
95
96

8 7

6 2

8 9

5 4

1 0 1

7 3

1 8 9

1 1 9

1 5 5

1 7 0

2 4 0

2 3 8

1 7 5

2 1 9

1 0 2

1 1 4

4 9

2 5 2

3 6 3

2 2 3

8 4

1 1 7 3

1 0 9 0

1 0 0 7

5 3 9

1 3 4 4

1 2 2 3

1 2 3 2

1 0 5 0

8 2 0

5 5

2 1 1

9 0 6

1 5 1 0

7 0

1 5 7

2 4 4

3 3 0

8 4

4 3 8

7 1 5

5 9 0

1 0 2 2

2 9

4 4

8 9

4 6

5 4

5 4

1 7 7

1 8 0

1 7 8

3 5 8

2 4 9

3 3 0

3 1 7

7 6

9 1

1 6 1

3 1 3

2 2 3

2 8 9

1 4 7

6 9

3 6 4

3 7 3

3 7 9

3 5 9

3 9 8

8 8

1 0 7

1 0 7

1 8 1

1 5 7

1 1 5

6 5

2 5

1 4 4

1 5 5

1 0 3

1 3 3

2 0 4

5 4

2 1 2

1 0 3

2 8 7

1 0 8

1 2 6

6 2

2 8 4

7 8

1 0 1

1 7 5

4 5

1 2

2 3

8 6

1 1 0

1 8 7

2 6 4

1 0 0

7 7

2 9 9

1 4 8

1 7 8

2 6 9

1 7 7

5 0 8

4 8 8

5 8 5

8 1 1

6 6 5

3 0 0

2 7 2

9 7

9 8

1 3 6

1 0 7

7 3

1 2 5

1 1 1

1 4 7

1 7 0

2 0 8

1 5 0

1 6 2

6 4

9 3

6 8

1 0 6

1 2 1

1 3 7

1 1 1

1 1 7

9 9

7 5

4 5

5 6

5 5

7 2

6 1

1 0 1

8 9

1 1 8

8 0

1 6

4 6

3 3

1 0 3

5 7

1 0 2

1 0 6

3 1 9

2 6 9

2 3 1

2 6

3 8

1 9

5 1

6 8

6 3

4 7

6 3

5 4 7

1 7

1 4 4

1 1 8

8 9

2 4 0

1 4 7

1 6 9

1 1 1

1 4 7

1 2 3

2 6 6

1 3 1

1 9 3

9 7

1 8 4

2 0 2

1 6 5

2 8 8

2 8 0

2 9 8

4 1 6

3 0 8

2 9 4

2 8 1

2 9 9

3 2 1

4 6 4

2 5 1

6 0

3 7 4

9 0

7 3 1

3 8 1

5 2 5

3 1 7

3 9 2

2 8 3

5 5 6

2 1 9

3 4 5

1 7 1

1 7 0

7 8

1 7 8

1 5 0

1 7 0

1 1 1

1 3 0

4 1

2 1

1 6 3

4 2

1 0 3

2 3

1 8 0

4 5

1 3 6

1 6 3

9 1

1 6 5

2 8 3

6 1

7 5

2 7

7 3 3

2 1

4 5

2 1 0

1 0 2

1 0 9

1 6 8

1 6 6

2 0 1

1 7

2 8

1 8

3 6

1 8

1 6

2 5

1 0

7 2

3 0

1 0 1

6 4

1 2 8

6 3 8

2 8

3 1 1

1 5 3

2 3 7

2 9

7 4 7

1 4 8

3 9 8

5 5

2 1

2 9 6

6 4

2 4

4 7

2 7

5 2

1 8

4 7

5 2

1 6 1

1 7 3

1 4 8

2 2 3

1 2 0

2 3 1

7 1

7 7

2 4

3 0

2 4

2 3 3

4 3

Demographics,
potential and results

Fig. 10.1. Architecture, stages and processes comprising theEWS RRI


Source: own study.

121

101

1 1 2

5 7

2 4 7

1 4 8

3 1

2 1 6

4 1

1 0 9

4 2

1 8 4

1 5 2

2 9 6

3 0 4

1 4 8

6 2

6 4

Forecasting

In thearea ofevaluation ofthe bankruptcy phenomenon inlegal terms (i.e. bankruptcy inits strict sense insolvency proceedings),
theanalysis and evaluation ofongoing proceedings supervised by courts involved:

The number and percentage ofproceedings,

The types ofinsolvency proceedings,

Regional analyses by region (regional bankruptcy barometer RBU),

Analysis ofsectors ofindustry by type ofactivity (industry bankruptcy barometer BBU).
The multifaceted analysis ofchanges inthefinancial condition offirms is carried out using thefollowing key measures:
The overall condition ofthe firm,
Financial liquidity,
Total debt,
Ability toservice debt,
Performance measures:

The inventory cycle,

The accounts receivable collection cycle,

The short-term liabilities cycle,

The cash cycle,
Asset productivity,
Operating return on sales,
Return on assets.
The central plane and, at thesame time, thekey evaluation process involves establishing thedegree offinancial distress ofan
enterprise inquestion (bankruptcy inits broad sense financial distress) using theestimated predictive models (classes PR and
MW). It is carried out ina multidimensional space ofnon-financial firms designated by thefollowing criteria:

Firm size,

Type ofactivity,

NACE sections, i.e. theselected branches ofthe national economy,

NACE divisions, i.e. manufacturing (32), trade (3) and services (27),

Regions (16 voivodships).
Thanks tothedetermination ofrisk bankruptcy parameters forindividual firms itwas possible (starting with the11th edition ofthe
study) tobroaden thecapacity and scope ofthe analysis offinancial distress. Hence, theanalyses carried out at thegeneral level
ofaggregation, types ofactivity and NACE sections, include thefollowing:

The time series ofthe mean value and standard deviation ofthe degree ofbankruptcy risk,

The time series ofthe mean value, theninth decile and theproportion ofthe maximum tothemean value,

The value ofdeciles fortwo periods under comparison,

The change ininterdecile ranges fortwo periods under comparison,

The cumulative distribution function fortwo periods under comparison,

The time series ofpositive deviations ofdeciles from themean and interdecile range between theninth and thefirst decile,

The time series ofthe mean value, theupper and lower limit and theninth decile,

The time series ofthe decile distribution.
The analysis ofthe heat map (representing thedegree ofrisk) and thesynthetic evaluation measure (MSO) provide thecriteria
foroutlining areas ofincreased risk NACE divisions at particular risk (called theTOP 10). Theprocess involves a multi-criteria
analysis ofrisk factors with respect to:
The number offirms,
The number ofemployees,
The degree ofrisk and thefactors that influence it:

Asset productivity,

Operating return on assets,

Operating return on sales,

Return on equity,

Current liquidity,

Self-financing,

Short-term debt,
Financial performance (the cash cycle and its components: inventory conversion, receivables and short-term liabilities cycles).
The application offorecasting time series variables (factors) inrisk prediction models made itpossible tofurther extend their
timeframe beyond theimmediate predictive horizon. As a result, assumptions concerning thefuture levels offinancial distress can
122

be made and thetime horizon can be extended toinclude another year. Theprocess employs theclassification and evaluation
criteria applicable tosets ofnon-financial firms:
Their size,
Type ofactivity,
NACE sections represented (i.e. selected sectors ofthe national economy).
Thanks totheconceptual consistency and reliability ofinstruments, thestructural elements, theobjectives ofscrutiny and
theorganizing factor, theEWS RRI constitutes an integrated solution that performs a key function and provides a monitoring
component ofthe RRI supported by themacroeconomic approach, i.e. studying theeconomic cycle.
The core component ofthe EWS RRI involves theevaluation offinancial distress, which is additionally reinforced by five analytical
planes. As a result ofthe inquiry, they combine togenerate a stream ofinformation which describes thefinancial risk, scaled (MW
class models) by thepercentage ofcourt bankruptcy proceedings, inother words, bankruptcy inits broad sense (financial distress)
is adjusted forbankruptcy inits strict sense (insolvency proceedings).

10.4. Estimation ofbankruptcy risk models


The condition ofa firm is usually described synthetically by means ofvarious financial indicators. Themodels ofbankruptcy
risk employed inthesystem under discussion comprise a set of24 indicators from thefollowing areas: productivity, liquidity,
funds, returns, debt and performance.3 Apart from thebasic versions fortheexplanatory variables, their nonlinear functions and
interactions between pairs ofvariables are also included intheanalysis.4
Table 10.2. Financial indicators used fortheconstruction ofpredictive models
Area ofanalysis

Financial indices

Liquidity

Current liquidity index, quick liquidity ratio, cash ratio

Funds

Overall financial situation, self-financing, coverage offixed capital assets, ability toservice debt, coverage ofliabilities by
financial surplus, period ofdebt repayment, creditworthiness/ debt capacity

Profitability

Return on sales, operational profit margin, return on assets, return on operating assets, return on equity

Debt

Total debt, short-term debt, long-term debt

Productivity
/ efficiency /
performance

Asset productivity, inventory conversion, receivables, payables, cash cycles, net working capital cycle

Source: own study.

In order tocompile thetraining data set5 forthefirst stage ofthe study (editions 110) under theEWS RRI (2011 1st half of2013),
thefullest possible data on a group offirms at risk (cases) was prepared. Then, using thecase-control method, they were matched
with enterprises deemed not tobe at risk (controls).6 It was assumed that thecriterion ofrisk was tantamount tothedeclaration
ofbankruptcy, i.e. theinitiation ofbankruptcy proceedings. Thefigures needed forthetraining set were obtained from public
databases on firms. As a result, theoriginal data set included 15 thousand firms not at risk and about 2 thousand at-risk ones.
Following theelimination ofincomplete data and taking into account thematching criteria, thefinal training set consisted of426
firms at risk and 1936 firms not at risk.
The development ofmodels forprediction was preceded by an analysis ofone-dimensional distributions ofexplanatory variables
conducted separately forfirms indifficulty and ina good condition, respectively. Thedistributions were analysed with respect
totheir numerical descriptive characteristics (mean, deciles, measure ofdispersion) and thecorresponding diagrams (histograms,
3

The indicators constitute a set ofexplanatory variables which, inthecase ofa firm indifficulty, describe its financial condition one year before
thelegal declaration ofbankruptcy (the initiation ofcourt bankruptcy proceedings).
For a further discussion of the course, methods and techniques for estimating such models, see: Kaczmarek, Kolegowicz, Krzemiski, Fijorek
(May 2011), Kaczmarek, Kolegowicz, Krzemiski, Fijorek (June 2011).
In thepreparation oftraining data sets, standard practice involves collating numerical data on firms at risk, and then matching them with
firms which are not at risk. Small data sets typically involve matching based on expert knowledge and a careful analysis ofeach individual
observation. Such an approach, however, is shows limited effectiveness inanalyses involving larger data sets.
This method consists inthedetermination ofseveral key characteristics ofstatistical units and matching each one with a certain highlighted
feature with a unit without such a feature, but most similar interms ofthe variables used formatching (pairing). It was assumed that each firm
indifficulty should be matched with firms ina good condition similar interms ofthe value oftheir assets and net sales, and thepairing was
done at NACE level with respect tolegal and organizational forms. Additionally, financial data on these firms came from thesame financial
year. Frequently 1 to1 matchings were applied, but from a theoretical point ofview, itis reasonable touse even 1 to5 matchings. In thestudy
under discussion, itwas possible toassign toeach firm indifficulty 4 or5 similar firms deemed not at risk. See also: (Hosmer, Lemeshow 1989,
p. 145162).

123

box-and-whisker plots, density functions determined by nuclear-type estimators). Moreover, univariate analyses determined
thepredictive capacity ofsimple decision-making rules based on thevalues ofindividual explanatory variables. Thepredictive
ability ofsimple rules and themodels described below were measured using theparameters ofsensitivity (proportion offirms at risk
considered tobe at risk) and specificity (percentage offirms not at risk deemed not tobe at risk). An equally important component
ofthe initial data analysis involved theevaluation ofthe correlation ofall potential explanatory variables (financial indicators).
Theprocedure helped us todefine a group ofvariables with a similar information load about therisk ofbankruptcy.
Despite thefact that theset offirms was quite large (nearly 2500 entities), theconcept ofmodelling taking into account thetype
ofeconomic activity (manufacturing, trade, services) required a special effort toensure that thedata were used with maximum
efficiency, reasoning was affected by theleast systematic error and theuncertainty ofparameter estimates was reliably measured.
Numerous studies have shown that insmall samples, theestimations oflogistic regression model7 parameters obtained using
theclassical method ofmaximum likelihood are characterised by a significant bias. Bearing this inmind, forthepurposes ofthe
study we decided touse Firths logistic regression model, which could be regarded asa relatively minor modification ofthe classical
logistic regression model. Theestimations ofparameters inthis model are almost unbiased, while theconfidence intervals are
characterized by superior probabilistic properties. Theresearch required an in-depth study ofthe logistic regression model and
thedevelopment ofspecialized software.8 In order todetermine theoptimal set ofvariables forming a logistic regression model,
we used thebest subset method (models ofup toeight explanatory variables were considered). Theadequacy ofmodel-data fit
was assessed using theAkaike Information Criterion (AIC).
In keeping with themethodology described above, three models ofbankruptcy threat were estimated, one foreach ofthe types
offirms (i.e. manufacturing, trade and services). Theeffectiveness ofthose models was evaluated with respect totheir sensitivity,
specificity and AUC.9 The resulting models are characterized by a high predictive capacity.
Table 10.3. Measures ofefficiency ofPR class models representing thedegree ofbankruptcy risk
Model

Entities at risk

Manufacturing firms
Trade firms
Service firms

Entities not at risk

Sensitivity

Specificity

AUC

207

916

82.1 %

81.3 %

0.890

65

318

84.6 %

84.5 %

0.918

154

702

86.3 %

80.0 %

0,886

Source: own study.

The PR-class models, represented by formulae used fordetermining thedegree ofbankruptcy risk formanufacturing, trade and
service firms, respectively, are asfollows:

Logistic regression is a common tool foranalysing binary data and is usually used toassess theimpact ofindependent variables on
theprobability ofoccurrence ofa particular event. Theclassical logistic regression model assumes that thedependent variable yi {0,1}
(i= 1,..., n) is subject toBernoulli distribution with theprobability ofa success F ( x i q ) , where F is thecumulative distribution function ofthe

1
, where xi is a p-dimensional vector ofexplanatory variables, and Rp is a p-dimensional vector
1+ exp x i q

ofstructural parameters containing an absolute term. In order toestimate theparameters ofthe model thelikelihood function and its natural
logarithm are determined, then thepartial derivatives ofthe logarithm ofthe likelihood function ofthe model parameters U() are calculated.
Thesolution ofthe equation U() = 0 is equivalent tofinding thevector ofparameter estimates NW that maximize thelikelihood function.
TheNW vector is obtained using iteration. In thecase ofFirths logistic regression model, thefunction U() is replaced by its modified version:

( )

logistic distribution F x i q =

1
2

1
2

y F ( x q)+ h ( 2 F ( x q)) x , where h are thediagonal elements ofmatrix H = W X ( X WX ) X W , X is a data matrix and W

is a n n diagonal matrix, where thei-th diagonal element equals F ( x i q )(1 F ( x i q )) . Themodification ofthe system ofequations U * (q ) is
U * (q ) =

i =1

1/ 2

identical with themodification ofthe likelihood function L (q ) = L (q ) Iq


, where Iq , is theinformation matrix, while thefunction L* (q ) is
called thepenalized likelihood function. It can be shown that Firths logistic regression model, despite thefact that itwas derived on thebasis
ofclassical statistical inferencing, has its Bayesian counterpart. It is equivalent totheclassical model oflogistic regression with Jeffreys noninformative prior distribution imposed on theparameters. See also: (Long 1997, p. 5668), (Fijorek 2011, no 2).
A broader discussion ofthe problems ofthe parameter estimates, aswell asin-depth analysis ofthe logistic regression model and its
instrumentation development, can be found, inter alia, in: (Firth 1993, no 80, p. 2738), (Heinze 2006, no 25, p. 42164226), (Fijorek 2011, no 27,
p. 34393440), (Fijorek, Sokoowski 2012, no 47), (Fijorek 2012).
9 AUC (Area Under theCurve) is themost commonly used measure ofthe quality ofthe classifier associated with theROC curve, defining
1

thearea under thecurve, and taking values intherange (0,1): AUC =

y ( x )dx , where y(x) is a function describing theROC curve (Receiver


0

Operating Characteristic). ROC is a two-dimensional graph that shows theand 1specificity calculated fordifferent values ofthe cut-off point.

124

Manufacturing firms PRP =


=

0, 41)
W
W
1
64

,
(
(W 0, 45)
(W 2,12)
(
)
1+ exp 0, 51 0, 44 1
0, 80 3
+ 0, 65 6
0, 70 19

0, 85
0, 32
0, 29
13, 51

(10.1)

where:
W1 asset productivity index,
W3 self-financing index,
W6 short-term debt index,
W19 return on operating assets (%).
Trade firms PRH =
=

1

, )
(W 0, 33)
(W 1, 58)
(W 179
(W 0, 00)
1+ exp 0, 80 0, 20 1
1, 57 3
0, 69 8
0, 84 17
0, 85
0, 30
0, 98
12, 20

(10.2)

where:
W1 asset productivity index,
W3 self-financing index,
W8 current liquidity index,
W17 operating return on sales (%).
Service firms PRU =
=

0
39
W
,
W
2
26

,
(
)
(W 0, 48)
(W 7, 86)
(
)
(W 1, 93)
1+ exp 0, 82 0, 46 1
115
+ 0, 69 6
0, 38 17
0, 35 20
, 3

,
1, 39
0, 29
0, 27
9, 26
5110

(10.3)

where:
W1 asset productivity index,
W3 self-financing index,
W6 short-term debt index,
W17 operating return on sales (%),
W20 return on equity (%).
The measures calculated using themodels presented above indicate thedegree offinancial distress (understood inits broad
economic sense) established under thecurrent conditions. It is an estimate ofthe probability ofbankruptcy threat arising within
one year ofthe currently analysed period.
Based on theanalysis ofhistorical data regarding bankruptcy risk at thehighest level ofaggregation, three levels ofsuch risk have
been identified: 020 % low (acceptable) risk, above 2040 % warning level, above 40 % high risk level. Thecut-off points were

determined arbitrarily on thebasis ofthe volume and distribution ofthe degree ofbankruptcy risk based on turning points and
peak values (a retrospective analysis ofthe 1st half of2008 2nd half of2009). Thedegree ofrisk at thegeneral level (aggregation
ofresults from specific models corresponding toindividual types ofactivity) is determined using theweighting ofresults obtained
at lower levels.
The second stage ofthe study (from the2nd half of2013, edition 1112) resulted intheestimation ofnew models offinancial
distress (MW class). Their distinguishing feature is thefact that they were developed on thebasis ofthe complete database ofnonfinancial enterprises subject toa large-scale nationwide statistical survey conducted by GUS in20072012. Thetraining data set
comprised 42 479 active firms (non-bankrupt, mean value intheperiod studied) observed forsix years on a semi-annual basis (small
enterprises) and on a quarterly basis (medium-sized and large enterprises) and 4 975 firms identified asbankrupt (same frequency
ofobservations) classified into three activity groups (production, trade and services). Comparative analyses included thetotal
of641 370 entity-observations (firms observed at specified intervals), including 353 282 entity-observations which comprised
thetraining data set ofpredictive models.

125

Table 10.4. Measures ofefficiency ofMW class models representing thedegree ofbankruptcy risk
Model
Manufacturing firms

Active entities1

Entities at risk2

13 047

1 377

Entity-observations
forthetraining set3

AUC

130 204

0,914

Trade firms

13 197

906

120 673

0,796

Service firms

11 823

1 306

102 405

0.893

Notes: 1 themean annual number ofenterprises in20072012, 2 thenumber ofenterprises in20072012, 3 thenumber ofentity-observations
conducted with a given frequency (semi-annual and quarterly) in20072012.
Source: own study.

The previously developed system forbuilding predictive models remains unchanged. Thenew models are highly efficient
theAUC was 0.914 formanufacturing firms, 0.796 fortrade ones and 0.893 forservice firms10 which means that they are
good orvery good. Moreover, their predictive quality is higher than that ofother models thanks tothelarge scale and number
ofanalytical levels aswell asthesize ofthe training set (nearly 50000 enterprises ascompared with a few dozen usually used
tofeed comparable models).
The MW-class models, represented by formulae used fordetermining thedegree ofbankruptcy risk formanufacturing, trade and
service firms, respectively, are asfollows:
Manufacturing firms MWP =
1
10000

1+ exp (5, 85 + (1, 231 R1 0, 492 R3 1, 947 R6 + 0, 62 R9 + 0, 004 R13 ))

(10.4)

where:
R1 operating return on assets,
R3 debt service coverage ratio,
R6 operating return on sales,
R9 total debt ratio,
R13 short-term liabilities conversion cycle.
Trade firms MWH =
=

1
10000
1+ exp (6, 22 + (0, 053 R4 1, 804 R6 117
, R7 + 0, 594 R10 0, 427 R16 ))

(10.5)

where:
R4 current (liquidity) ratio,
R6 total debt ratio,
R7 return on sales,
R10 short-term debt,
R16 return on equity.
Service firms MWU =
=

1
1+ exp (6, 24 + (0, 342 R3 + 0, 544 R9 + 0, 004 R13 1, 44 R14 1, 088 R15 ))

10000

(10.6)

where:
R3 debt service coverage ratio,
R9 total debt ratio,
R13 short-term liabilities conversion cycle.
R14 return on sales,
R15 return on assets.
10

A further discussion of the course, methods and techniques for estimating these models can be found in source: Kaczmarek, Kolegowicz,
Krzemiski, Fijorek (December 2013).

126

The application ofthe estimated logistic regression models yields theprobability ofbankruptcy ofa given firm (degree ofrisk)
within one year assessed on thebasis ofthe conditions prevalent intheinquiry period and its current economic and financial
condition. Theaggregation ofindividual probabilities constitutes thebasis fordetermining theproportion ofcorporate
bankruptcies scaled by thenumber ofinitiated court bankruptcy proceedings. Thedegree ofrisk ofbankruptcy is expressed per
10 000 firms (a quotient scale).
Three brackets ofbankruptcy risk at thehighest level ofaggregation have been identified: 0160 low (acceptable) risk, above
160200 warning level, above 200 high risk level. Thecut-off points were

determined arbitrarily on thebasis ofthe volume and


distribution ofthe degree ofbankruptcy risk based on turning points and peak values (a retrospective analysis ofthe 1st half of2007
2nd half of2012). Thedegree ofrisk at thegeneral level (aggregation ofresults from specific models corresponding toindividual
types ofactivity) is determined using theweighting ofresults obtained at lower levels.
Features which make these models unique interms ofsignificantly higher-quality predictions ofchange infinancial distress levels
and bankruptcy risk than models applied so far include: estimation on a large sample offirms, multiple parameters used toassess
thefirms condition and toidentify phenomena indicating a possible financial distress (early warning) coupled with theapplication
ofinnovative methods and techniques.

10.5. The rationale and methods fordetermining thevalidity period ofthe measured risk
The application ofmodels predicting thedegree ofrisk offinancial distress constitutes an effective tool used intheassessment
ofthreats faced by firms which may occur intheevent ofdeteriorating business conditions inwhich they operate. Theproblem
ofpoor predictive capacity, inaccuracy and outdatedness ofmodels available intheliterature was resolved thanks toestimating
new, unique and highly efficient PR and MW models. Another limitation, which consisted insignificant delays inthesupply
ofrelevant survey data inPoland (up toa year), was overcome by theprovision ofinformation and numerical data from national
statistics (GUS) on a quarterly and/or semi-annual basis.

12,8

12,6

12,4

12,2

12,0

11,8

11,6

11,4
II 2009

III 2009

IV 2009

Period 1

I 2010

II 2010

III 2010

IV 2010

Period 2

I 2011

II 2011

III 2011

Period 3

IV 2011

I 2012

II 2012

Forecast

Fig. 10.2. Predicting thedegree ofrisk on an annual basis using themoving average model
Source: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (May 2011), Kaczmarek, Kolegowicz, Krzemiski, Fijorek (June 2011).

The solution developed Early Warning System Rapid Response Instrument allows fortherecognition ofadverse changes
inthedegree ofrisk within a much shorter timeframe. Thanks tothefact that surveys are conducted on a quarterly and/or semiannual basis, thedegree ofrisk estimated inadvance remains valid fora single survey period, since a subsequent survey provides
new information about themeasures inquestion and refers tothenext period.

127

The estimated models themselves are tools (methods) ofa prospective analysis offinancial distress (one year ahead). However,
theuse oftime series forecasting methods allowed us toextend thevalidity ofthe forecasts. In forecasting thechange inthedegree
ofrisk instead ofusing themeasure ofrisk itself (PR orMW values) we have decided toapply thevariables (factors defined
asfinancial indicators) reflecting thefinancial condition offirms. Thepredicted values ofthese variables are fed into prediction
models, which yield thelikely future levels offinancial distress and allow us toextend thetime horizon onto thenext period
(one year).
The variables determining thedegree ofrisk are predicted using one-dimensional time series models. For this purpose,
thefollowing models have been used: SARIMA (Seasonal Autoregressive Integrated Moving Average) and Hyndmans exponential
smoothing11 (ETS). Themain reason fortheuse ofone-dimensional models instead ofmultidimensional ones is therelative
shortness ofthe time series studied. Thedecision was also justified by theresults ofempirical studies showing that multivariate
models applied toshort time series typically generate forecasts subject togreater errors than those generated by onedimensional models.
At present, ETS-class models are not yet popular; hence itmakes sense toprovide a brief introduction. Classic exponential
smoothing belongs toa group offorecasting methods. A forecasting method should be understood asa set ofprocedures used
toobtain point forecasts ofa studied phenomenon ina future period oftime. In contrast, theSARIMA-class model is a full statistical
model with a defined stochastic process generating numerical data; consequently, theprobability distribution ofa forecast
forthesubsequent period oftime can be estimated. As a result, thestatistical model apart from point forecasts is capable
ofdelivering interval forecasts with a probabilistic interpretation. Theclassic exponential smoothing methods did not have this
dimension, which changed thanks tothework ofHyndman and his associates. In order tospecify theexponential smoothing
model, three elements need tobe determined. Thefirst one is thecomponent responsible forthetrend. Thecomponent consists
ofthe level factor and thegrowth factor. Both factors may be combined inseveral ways todefine a process that either has no
trend orhas an additive trend, a damped additive trend, a multiplicative trend ora damped multiplicative trend. Thesecond
component is thecomponent ofseasonal fluctuations which may occur inadditive and/or multiplicative versions, oritmay
not exist. Thethird element is therandom fluctuations component that may occur inits additive ormultiplicative version. Thus,
specific components ofthe ETS model define 30 different models, from which theoptimal model can be selected by using e.g.
information criteria.
In theanalyses oflevels ofrisk, each time series offinancial indicators studied was matched with thebest SARIMA and ETS models
interms ofthe AIC criterion. Thepredicted values were calculated asthearithmetic average ofthe forecasts yielded by SARIMA
and ECJ models. Thesolution adopted is justified by theresults ofempirical research, which show that averaging models usually
leads tosmaller forecast errors.
As part ofthe quantitative analysis ofuncertainty associated with theprojections offinancial distress, a tool called scenario
analysis was used: inaddition totheneutral scenario, pessimistic and optimistic scenarios were also constructed. For this purpose,
thelower and upper limits of80% ofthe confidence interval ofthe predicted degree ofrisk were used based on theanalysis
ofhistorical deviations ofthe degree ofrisk adjusted forthepercentage ofinitiated court bankruptcy proceedings.

10.6. A synthetic assessment measure in the identification of risk areas


An important aspect ofstudying thedegree offinancial distress under theEWS RRI is thesectoral analysis by NACE divisions.
Theanalysis includes 62 individual types ofproduction, trade and services enterprises. Structurally, itis carried out intwo stages
type ofeconomic activity and thesynthetic evaluation measure (MSO) using two classification criteria: enterprise size (total
enterprises, small, medium-sized, large) and thetype ofactivity (manufacturing, trade, services).
As a result, changes are described and risk levels are assessed using a heat map. Moreover, areas ofparticular risk are indicated
using thesynthetic evaluation measure (MSO). Theprocess leads tothedevelopment ofa ranking list ofNACE divisions
characterized by thehighest risk adjusted fortheshare ofindividual NACE divisions inthetotal number ofemployees, thesocalled TOP 10.

11

An in-depth discussion ofissues concerning prediction using ETS class models can be found e.g. in: (Hyndman 2008).

128

NACE I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II
divisions 2007 2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013

NACE I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II
divisions 2007 2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013

MANUFACTURING
5
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
35
36
37
38
39

338

18

371

37

185

43

182

12

270

211

321

114

245

273

210

121

TRADE
74

124

137

154

109

291

219

353

935

126

132

95

110

78

114

141

149

111

162

307

360

606

125

157

173

183

127

133

118

150

137

155

140

146

131

146

143

166

163

199

165

160

176

197

232

250

281

286

223

270

289

605

339

196

141

145

163

182

146

118

157

199

215

207

380

191

215

156

196

179

166

181

171

139

161

162

176

158

173

198

187

174

172

151

172

159

158

136

132

161

157

165

160

187

204

177

219

249

172

100

85

98

179

211

219

199

258

215

203

194

197

200

177

118

102

137

135

98

106

142

134

110

120

90

92

94

126

123

145

191

232

202

216

215

193

179

204

201

160

112

95

79

167

261

191

179

156

139

164

157

142

221

70

77

62

124

150

166

127

131

87

95

90

131

101

65

55

20

37

69

65

67

82

74

77

110

114

42

148

146

163

184

158

135

152

154

168

159

161

150

148

28

24

53

70

170

125

196

132

150

115

188

148

248

46

73

136

183

343

380

318

259

188

204

259

275

310

151

149

171

174

172

161

220

193

201

181

189

177

198

327

315

283

303

188

286

305

248

255

270

213

223

182

209

183

158

197

160

146

196

171

212

205

220

204

147

154

174

229

187

188

198

174

177

164

137

137

140

170

149

202

195

183

165

179

142

177

159

177

163

563

788

827

883

590

532

784

637

554

567

502

346

337

169

192

235

261

147

127

140

128

155

138

169

147

117

107

143

136

155

112

103

152

114

175

180

131

129

184

164

179

154

196

169

156

270

243

249

220

223

201

246

87

109

65

109

72

128

80

81

70

186

221

217

213

187

245

181

158

182

162

219

177

233

194

236

207

214

175

174

141

66

80

58

87

116

138

106

125

119

126

119

149

148

57

52

67

83

125

58

194

77

192

122

253

264

253

88

57

169

92

210

184

208

84

62

100

127

132

97

114

87

128

114

176

159

184

85

79

83

87

120

109

118

94

124

105

134

118

129

105

73

107

85

142

105

163

107

130

119

152

122

159

333

SERVICES

41
42
43
49
50
51
52
53
55
56
58
59
60
61
62
63
66
68
78
79
80
81
82
94
95
96

215

127

195

45
46
47

332

67

194

191

142

159

153

175

165

248

207

265

232

297

286

267

134

283

188

319

152

336

183

378

263

559

356

543

87

62

89

54

101

73

189

119

193

163

204

212

287

155

170

240

238

252

363

223

84

55

211

906

1510

70

157

244

330

84

438

715

590

1022

29

44

89

46

54

54

177

180

178

358

249

330

317

175

219

102

114

49

1173

1090

1007

539

1344

103

1223

54

1232

103

108

1050

820

76

91

161

313

223

289

147

69

364

373

379

359

398

65

88

107

107

181

144

157

115

155

133

126

62

284

45

78

101

175

136

163

91

165

25

45

12

23

171

86

110

187

264

170

299

148

178

269

78

178

97

98

42

23

100

77

180

177

508

488

585

811

665

300

272

136

107

73

125

111

147

170

208

283

103

150

170

150

162

64

93

68

106

121

137

111

117

111

130

99

75

45

56

55

72

61

101

89

118

80

41

16

46

33

103

57

102

106

319

269

231

21

26

38

19

51

68

63

47

63

547

17

144

118

89

240

147

169

111

147

123

266

131

193

97

184

202

165

288

280

298

416

308

294

281

299

321

464

251

60

374

90

731

381

525

317

392

283

556

219

345

61

75

27

733

21

45

210

102

109

168

166

201

17

28

18

36

18

16

25

10

72

30

101

64

128

29

747

233

638

148

398

28

311

112

148

31

153

237

30

24

43

55

21

296

64

24

57

216

41

109

42

47

27

52

18

47

52

161

173

247

184

152

296

304

148

223

120

231

71

77

24

148

62

64

101

Fig. 10.3. Heat map differences inbankruptcy risk levels by NACE division
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

The synthetic evaluation measure (MSO), which constitutes thebasis fortheranking list, employs two categories: therelationship
ofthe risk level and thenumber ofemployees ina specific NACE division toaverage values inindividual types ofeconomic activity.
Thenext steps intheanalysis and MSO calculation are asfollows:
1. Determining thedeviation ofthe degree ofrisk (MW) foreach NACE division relative tothat ofthe corresponding type
ofeconomic activity.

oMW Dit =

MW Dit
MW RD jt

(10.7)

where:
oMW deviation ofthe degree ofbankruptcy risk fora given NACE division,
MW D degree ofrisk fora given NACE division,
MW RD degree ofrisk by type ofeconomic activity,
i NACE division (i = 1,..., n),
j type ofeconomic activity (j = manufacturing, trade, services),
t period analysed.
2. Determining thedeviation ofthe number ofemployees foreach NACE division relative totheaverage value forthetype
ofeconomic activity.

oLP Dit =

LP Dit
sLP RD jt

where:
oLP D deviation ofthe number ofemployees fora given NACE division,
LP Dit number ofemployees ina given NACE division,
sLP RD average number ofemployees by type ofeconomic activity,
i NACE division (i = 1,..., n),
j type ofeconomic activity (j = manufacturing, trade, services),
t period analysed.
129

(10.8)

3. Assigning normative standards toindividual NACE divisions.


Based on theresults ofoMW D and oLP D forNACE divisions, normative standards (1, 2, 3 or4) are assigned toindividual
NACE divisions, including theassessment ofthe deviation ofthe degree ofbankruptcy risk fora given NACE division (oMW D)
and thedeviation ofthe number ofemployees ina given NACE division from theaverage fora given type ofactivity (oLP D).
Table 10.5. Normative standards designated by thedegree ofbankruptcy risk and thenumber ofemployees applicable toindividual
NACE divisions
Criterion 1

Standard

Criterion 2

Deviation ofthe number ofemployees fora given NACE


division
(OLP Di) > 1

Deviation ofthe degree ofrisk fora given NACE division


(oMW Di) > 1

II

Deviation ofthe number ofemployees fora given NACE


division
(OLP Di) 1

Deviation ofthe degree ofrisk fora given NACE division


(oMW Di) 1

III

Deviation ofthe number ofemployees fora given NACE


division
(OLP Di) > 1

Deviation ofthe degree ofrisk fora given NACE division


(oMW Di) 1

IV

Deviation ofthe number ofemployees fora given NACE


division
(OLP Di) 1

Deviation ofthe degree ofrisk fora given NACE division


(oMW Di) > 1

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

The features ofrespective standards (i.e. theNACE divisions assigned toeach group) are interpreted inthefollowing way:
Standard I a NACE division characterized by an above-average risk and an above-average number ofemployees
a high-risk area,
Standard II a NACE division characterized by an above-average risk and a below-average number ofemployees an
average-risk area,
Standard III a NACE division characterized by a below-average risk and an above-average number ofemployees
a low-average-risk area,
Standard IV a NACE division characterized by a below average risk and a below-average number ofemployees
a low-risk area.
4. Calculation ofthe synthetic evaluation measure MSO Di
The synthetic evaluation measure (MSO) is used toassign theabove-listed normative standards toindividual NACE divisions.
Themeasure is based on theevaluation oftwo categories: therelationships between thebankruptcy risk and thenumber
ofemployees ina given NACE division relative tothemean values forrespective types ofactivity. Thecalculation ofMSO
value involves measuring thedistance ofa given NACE division (whose coordinates are oMW Di and oLP Di, respectively) from
thepoint oforigin ofthe coordinate system.

MSO Di = 2 oMW Di2 + oLP Di2

(10.9)

where:
MSO Di synthetic evaluation measure ofa NACE division (object),
5.

Compilation ofa ranking list ofNACE divisions


The synthetic evaluation measure (MSO) enables us todraw up a list ofNACE divisions organized according totwo criteria:
thestandards discussed above (in ascending order) and MSO values within a given standard (in descending order).

130

3,00

2,50

oLP Di

1,50

oMW Di

Degree of bankruptcy risk

2,00

1,00

SO

0,50

Di

0,00
0,00

1,00

0,50

1,50

2,00

2,50

3,00

Number of employees

Fig. 10.4. Determining thevalue ofthe synthetic evaluation measure (MSO) forindividual NACE divisions inaccordance with thepredefined normative standards
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

Table 10.6. A sample ranking ofNACE divisions according tothedegree ofrisk, thesynthetic evaluation measure (MSO) and
normative standards
NACE division

MW
1st half of2013

STANDARD
1st half of2012

MSO

1st half of2013

1st half of. 2012

1st half of2013

Change

25

198

3.08

3.10

332

III

2.04

2.63

23

248

1.86

2.06

27

215

1.85

1.82

935

II

II

1.80

5.55

8
9

606

II

II

1.88

3.60

30

337

II

II

3.13

2.08

24

310

II

II

1.80

2.02

10

131

III

III

4.50

4.52

29

163

III

III

2.44

2.53

22

148

III

III

2.25

2.27

28

140

III

III

1.80

1.73

20

101

IV

IV

1.14

1.16

13

139

II

IV

1.21

0.95

12

157

IV

IV

0.90

0.93

15

100

II

IV

1.54

0.63

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

131

ukasz Lenart, Baej Mazur, Krystian Mucha, Mateusz Pipie, Justyna Wrblewska

Chapter 11
MACROECONOMIC MODELLING OF BANKRUPTCY RISK
11.1. Introduction
The main aim ofthis section is todiscuss indetail themethodological approach adopted by theMacroeconomic Team ofthe
project titled Rapid Response Instruments. Theoretical and empirical analyses focus on three issues: first, themethods ofdrawing
inferences about business cycles with a particular emphasis on thespecifics ofthe Polish economy; second, theprinciples
ofconstructing projections ofbasic macroeconomic indicators which constitute thebasis forthedescription ofprospects
foreconomic growth inPoland; and third, themethodology ofscenario analyses ofthe impact ofpotential impulses affecting
thePolish economy and its cyclical changes, with regard tothecondition ofthe corporate sector.
Part 1 ofthe study is devoted totheformal-statistical approach tomodelling business cycles. Theproposed approach is based
on econometric analysis ofselected macroeconomic indicators, which belong totheso-called cyclic indicators. These indicators
shall provide thebasis fortheformulation ofstylized facts about thevariations inbusiness activity inPoland. Theanalysis will aim,
among other things, toidentify thedominant business cycle length inPoland, toestimate theamplitudes ofidentified fluctuations
and toprovide thepossible interpretations ofthus estimated phases.
Within themacroeconomic block, theprimary objective shall be achieved through theuse ofproprietary solutions inthefield
ofmathematical statistics. Thebusiness cycle will be analysed based on theresults obtained inthedomain ofthe almost
periodically correlated time series. New tools forthestatistical analysis ofbusiness cycles will draw both on thelong-established
spectral theory fortheclass ofstationary time series and thetheory ofharmonized almost periodically correlated time series.
Thekey outcomes oftheoretical work inthefield ofmathematical statistics are related totheparameters and their estimators,
which characterize both thediscrete spectrum and thecontinuous almost periodically correlated time series.
In our analyses we shall assume that thefluctuations inbusiness activity can be modelled using a non-parametric approach with
concurrent discrete and continuous characteristics ofthe spectrum. In most studies on thebusiness cycle itis assumed that cyclical
fluctuations constitute, infact, a time series with its expected value equal tozero, which means that thediscrete spectrum forthis
time series is identical and equal tozero. Theapproach developed and applied tothematter at hand is original and innovative
incomparison with thestandard methods foranalysing thecyclical nature ofthe economy based on thespectral theory and
filtration methods applied toobservable discrete-time processes.
In Chapter 2 we propose touse theBayesian approach totheproblem ofshort-term forecasting ofmacroeconomic indicators
and thecyclical position ofthe economy. Due totheshort forecast horizon, we focus on themodelling ofone-dimensional time
series. An important distinguishing feature ofthis approach incomparison with themethods widely used inpractice is that we
formulate our forecasts inprobabilistic terms, which, inaddition tothefuture indicators designating point estimates, includes
theuncertainty offorecasts. Theresulting forecasts intheform ofpredictive probability intervals will be based on theBayesian
approach totheestimation ofmodel parameters and forecasting.
Chapter 2 introduces theidea ofBayesian inference and discusses thespecifications ofmodels used inforecasting macroeconomic
indicators. Theadded value consists inproposing a novel method offorecasting thecyclical position ofthe economy based on
theBayesian model with respect totheproblem ofparameter estimation ofalmost periodically correlated processes (which
are discussed inPart 1). Theapproach commonly used inpractice is based on filters that enable theestimation ofthe cycle, yet
itprecludes any cyclical forecasts beyond theperiod considered. Our team has developed an approach that allows forshort-term
probabilistic forecasts ofthe cyclical component, which will facilitate an analysis ofthe condition ofthe Polish economy, which
includes forecasts about thebusiness cycle.
Chapter 3 is devoted totheissue ofmodelling vector-based time series with a special emphasis on thecointegration approach
that makes use ofvector autoregression models (VAR) and vector error correction (VECM). These methods focus on a formal
statistical approach totesting long-term relationships among theobservable variables, forecasts and scenario analyses. One ofthe
subsections presents both theclassical approach and theBayesian one totheestimation ofmodel parameters and forecasts.
Multi-dimensional modelling, discussed inPart 3, is applied with a view toassessing theeffects ofabrupt changes
inmacroeconomic variables. Such abrupt changes are referred toasimpulses, and theassessment oftheir occurrence is
described interms ofhypothetical macroeconomic scenarios (as is done intheliterature on quantitative methods) reflecting
theestimation ofthe impulse response functions both inclassical and Bayesian terms. In empirical analyses, this function will

133

constitute theprimary tool toassess thesensitivity ofthe condition ofthe corporate sector tohypothetical impulses occurring
intheeconomy.
As inthecase offorecasts, quantitative assessments ofthe effects ofcertain impulses are presented inprobabilistic terms.
Accordingly, theanalyses include a formal assessment ofstatistical uncertainty asto thevalue ofthe impulse response function,
which allows us toevaluate theimpact ofstatistical uncertainty on our conclusions.
All parts ofthe study are illustrated by empirical examples ofthe proposed econometric and statistical methods.

11.2. Modelling fluctuations inbusiness activity using cyclostationary processes


In this section we shall introduce our original proposals fortheanalysis offluctuations inbusiness activity. Theapproach discussed
here does not provide tools forforecasting fluctuations which can be identified with fluctuations inbusiness activity, but focuses
only on thedetailed characteristics ofcyclical changes affecting thebasic macroeconomic indicators.
Intensive research on theempirical identification ofbusiness cycles, which over theyears has yielded a wide range ofmethods and
tools toanalyse theprocess inquestion, has not led tothedevelopment ofa dominant method that would be acceptable tomost
scholars. In theliterature, most popular approaches totheextraction offluctuations identified with thefluctuations ofbusiness
activity employ filtration methods, including those developed by Hodrick and Prescott (Hodrick and Prescott 1997), Baxter and King
(Baxter and King 1999), Christiano and Fitzgerald (Christiano and Fitzgerald 2003), and Butterworth (Gmez 2001). Theproposed
solutions, especially inregard oftheir automatic application, have attracted serious criticism, since their inappropriate use may
lead totheidentification ofspurious business cycles (cf. e.g. Nelson and Kang 1981, Kaiser and Moravall 1999, Gmez 2001, ShenkHopp 2001).
In particular, themethods developed and tested inhighly-developed economies may not supply data that would allow researchers
toformulate clear conclusions applicable e.g. toeconomies intransition. One ofthe reasons may involve thecharacteristics
ofcyclical fluctuations affecting theeconomies intransition incomparison with thecharacteristics offluctuations inhighlydeveloped economies. Thefeatures ofcyclical fluctuations intransition economies have been described by Barczyk et al. (2006),
emphasizing their separation and substantial differences from theprogression ofthe business cycle indeveloped economies.
Inthelatter, fluctuations inbusiness activity (around thedownturn/upturn path) have been, infact, flattened, hence thedominant
view about theneed tomodel them using stationary processes with expected value zero. This feature does not have toaccompany
thecyclical fluctuations intransition economies (due tothelikelihood ofgreater intensity and amplitude offluctuations). Hence
themotivation toadopt themore general assumptions typical ofthe APC model, which, ina particular case, can be reduced
totheassumption that theprocess is stationary. In theproposed approach, theproblem ofchoice between stationarity and APC
assumptions shall be resolved using appropriate statistical tests.
The analysis offluctuations inbusiness activity may also be made more difficult due tothelack ofsufficiently extensive statistical
material justifying strong conclusions about thenature ofthe business cycle. Theissue ofuncertainty regarding theinferences on
thecyclical nature ofthe Polish economy having its source inthevery short time series was highlighted ina study by Adamowicz
et al. (2008), where alternative methods offiltration were applied together with an analysis ofidentified fluctuations both
inthedomains oftime and frequency. Theproblem with identifying thelength ofbusiness cycles inPoland complete with an
analysis ofthe degree ofsynchronization with theEurozone countries was investigated, among others, by Skrzypczyski (2006),
Skrzypczyski (2008), Adamowicz et al. (2008), Skrzypczyski (2010), Konopczak (2009), Gradzewicz et al. (2010).
This chapter offers a method foranalysing fluctuations inthebusiness cycle which has a solid theoretical grounding. Thesmallsample properties ofstatistical methods, on which thealgorithm is based, were tested by Lenart (2010). Of special interest are
thetools fortheestimation ofparameters which can be identified with thelength ofbusiness cycles, specifically their amplitude
and phase. Themethodology is based on thefundamental properties ofa subclass ofnon-stationary stochastic processes, i.e. almost
periodically correlated processes. Thekey results ofthe proposed algorithm procedure are estimators ofthe parameters identified with
fluctuations inbusiness activity. This allows us todetermine thenecessary parameters using theHodrick-Prescott filter.
The proposed methodological approach is based on an econometric analysis ofmacroeconomic time series. Modelling comprises
historical data, which makes itvery difficult toformally take into account (possible) significant changes inthenature ofthe business
cycle ofthe Polish economy that result, among other things, from theconvergence process.
The methodological approach presented inthefollowing sections is based on more in-depth studies on thesubject by Lenart
(2010) and Lenart and Pipie (2011). Some ofthe definitions inthis section are quoted directly from Lenart (2010), where all
theattributions can be found. However, we devote more attention totheavailable interpretations inthecontext ofresearch on
fluctuations inbusiness activity informed by thecomments and conclusions tobe found inLenart (2010) and Lenart and Pipie
(2011).
134

11.3. Almost periodically correlated stochastic processes basic features


This section starts with thedefinition ofthe class ofperiodically and almost periodically correlated time series. Next, we discuss
thedefinitions and properties ofalmost periodic functions, which are closely related tothedefinition and properties ofalmost
periodically correlated time series.1 The final part ofthe section contains thedefinition ofthe amplitude and phase offluctuations
with respect totherepresentation ofthe expected value function ofthe process under consideration asa Fourier series.
The concepts ofperiodic and almost periodic functions are primary with respect tothedefinitions oftime series periodically
and almost periodically correlated ina broader sense. For thefirst time theclass ofperiodically correlated time series was
defined by Gladyshev (1961). Often, periodically correlated and almost periodically correlated time series, respectively, are called
cyclostationary series oralmost cyclostationary time series, respectively (cf. Gardner et al. 2006). Let us consider thetime series
{ X t : t Z} with theexpected value function:

m(t ) = E ( X t ) <

and thefollowing autocovariance function:


B(t , t ) = cov( X t , X t +t ) <

where t Z. Let us now derive thedefinition ofperiodically correlated time series.


Definition 1.1 (Gladyshev 1961). Thetime series { X t : t Z} is said tobe periodically correlated ina broader sense (abbreviated PC) with
a period equal toT, if forany t Z functions m(t) and B(t , t ) are periodic functions ofan integer variable with period T.
In order todefine theclass ofalmost periodically correlated time series ina broader sense, we should first quote a definition
ofalmost periodic functions taken from themonograph by Corduneanu (1989).
Definition 1.2 (Corduneanu 1989, p. 45) The function ofthe integer variable f : Z R is said tobe almost periodic if forany e > 0 there
exists an integer Ne such that ineach set containing Ne elements ofconsecutive integers there is an integer p forwhich thefollowing
holds true:
sup | f (t + p ) f (t ) |< e
t Z

To introduce theconcept ofalmost periodic functions, let us consider thefunction h : Z R such as


h(t ) = sin( 2t )

(11.1)

This function is not a periodic function ofthe integer variable t, which can be easily justified using indirect proof. Conversely,
function h is almost periodic.
Any periodic function with period T is also an almost periodic function. Thefamily ofalmost periodic functions is closed under
addition and multiplication (cf. Corduneanu 1989, p. 46). Thebasic property ofthe almost periodic function f : Z R is thefact
that thelimit
1
n n
lim

t0 +n

f (t )

t =t0 +1

(11.2)

exists and does not depend on thevalue ofthe integer parameter t0 (cf. Corduneanu 1989, Theorem 1.28, p. 48). Thelimit is
called theaverage value ofthe almost periodic function. As shown by Lenart (2010) and Lenart and Pipie (2011), this property is
fundamental totheestimation ofthe parameters identified with thefrequencies responsible forfluctuations inthebusiness cycle.
The definition ofthe class ofalmost periodically correlated time series is a natural generalization ofthe definition ofthe class
ofperiodically correlated time series.
Definition 1.3 The time series { X t : t Z} is said tobe almost periodically correlated ina broader sense (abbreviated APC), if forany t Z
functions m(t) and B(t, t) are almost periodic functions ofthe integer variable t.
Each stationary time series is also periodically correlated, and each periodically correlated time series is almost periodically
correlated.
In order tointuitively explain thedifference between theperiodic and almost periodic functions, let us consider a simple example
ofa periodic function f(t) with period T = 12. Using elementary properties ofperiodic functions, function f can be represented as
1

Cf. also (Besicovitch 1932) and (Corduneanu 1989).

135

T 1

f (t ) =

b e
k

it wk

(11.3)

k =0



where bk are complex numbers and frequencies wk equal wk =2kp/T, fork = 0,1, ..., T 1. As can be seen, thecharacteristic property
ofa periodic function is that inits representation (11.3), frequencies wk depend on period T (are strictly determined). Assuming,
however, that function f(t) is an almost periodic function represented as
K

f (t ) =

b e
k

it wk

(11.4)

k =0

we only assume that wk are numbers from theinterval [0,2p) and K is an arbitrary natural number.
Subsequently, theexpected value and autocovariance functions, respectively, can be presented asa Fourier series:
(t ) :

i t

(11.5)


B (t , ) :

m()e

a(, )e

it

(11.6)

where coefficients m(y) and a (l, t) qual


1
n n

m( ) = lim

1
n n

a( , ) = lim

(t )e

i t

(11.7)

t =1
n

B( j , )e


i j

j =1

Thus, sets:

/ 0}
= { [0,2 ) : m( ) =

and

/ 0}
= { [0,2 ) : a( , ) =

are countable (cf. Corduneanu 1989). If a given time series is periodically correlated, then its representations (11.5) and (11.6)
become equal, whereas sets z and L are contained intheset

{2k p / T : k = 0,1,, T -1}

where T represents thelength ofthe period.


The classical analysis ofthe fluctuations inbusiness activity begins with theassumption that theisolated fluctuations identified
with fluctuations inbusiness activity are stationary. This assumption, however, is not verified inmost cases, except when unit root
tests are used. Thecondition ofstationarity ofthese fluctuations is equivalent tothespecial case ofan almost periodically correlated
process inwhich thefollowing equalities apply: Y = {0} and L = {0}. These equalities need not be justified, given thenumerous
cases ofmodelling macroeconomic time series which contain a component identified with fluctuations inbusiness activity using
models with thenon-stationary, periodic structure ofthe autocovariance function (cf. e.g. Franses and Paap 2004). Therefore,
theproposed approach tothestudy offluctuations inbusiness activity constitutes a generalization, inwhich itis possible totest
thewidely adopted assumption ofstationarity.
In theapproach proposed by theMacroeconomic Team, themodelling ofcyclical fluctuations is accomplished on thebasis
ofmore general assumptions (note that theclass ofAPC time series contains theclass ofstationary time series). It is assumed that
thebasic properties ofcyclical fluctuations, such asfluctuations inlength, amplitude, asymmetry and phase, can be modelled
using theparameters ofa non-trivial spectrum ofa discrete APC time series. Using thebasic Fourier representation ofthe expected
value ofthe APC time series, certain frequencies inthis representation shall be identified asthose responsible forthefluctuations
caused by thebusiness cycle, while others shall be interpreted asseasonal variations. Thefact that thefunction ofthe expected

136

value ofthe APC is almost periodic may contribute toa better description ofthe characteristics offluctuations ofbusiness activity
(in terms offitting theempirical data tothemodel).
The assumption ofalmost periodicity ofthe expected value function is also very convenient from thepoint ofview ofthe economy.
Thediscussion on thecauses ofeconomic fluctuations may include theimpact ofendogenous and exogenous variables, theeffects
ofbusiness cycle transmission between different economies, theinvestment multiplier effect aswell astheeffects ofinteractions
and overlaps among these reasons. If each ofthose multiple reasons generates cyclical (wavelike) changes, theassumption
ofalmost periodicity ofthe expected value function intuitively seems not toimpose restrictions on theconcepts and theories
ofthe business cycle.

11.4. Estimation ofApc processes using subsampling methods


Parameters which are thought todirectly correspond with thefluctuations inbusiness activity belong toset Y. In this section,
we shall focus on theissues related totheestimation ofselected elements ofthis set. In mathematical statistics, Lenart (2010)
formulated and proved theorems related totheidentification and estimation ofthe parameters ofset Y. Themethod foridentifying
these parameters is based on thewell-known subsampling method (in theliterature on mathematical statistics, cf. Politis et al.
1999).
What follows below is a summary ofthe subsampling method based on Lenart (2010), which, inturn, is based on theresults
contained inthemonograph by Politis et al. (1999).
Let { X1, X 2 ,, X n } be an n-element sample ofan actual time series { X t : t Z} whose total distribution is P. 
q n ( X1, X 2 ,, X n ) = 
qn

q n,b,t and
denotes theestimator ofthe unknown real parameter q = q(P) . Thesubsampling version ofestimator q n is denoted by
calculated asthevalue ofthe estimator
q n based on subsample { X t , X t +1,, X t +b1} , where 1 t n b + 1 can be interpreted
asthestarting point forthesubsample, and b = b( n) asits length. One ofthe basic assumptions ofthe subsampling procedure is
that b / n 0 and b . Let { un }nN be a sequence ofreal numbers. If we are interested inobtaining an approximation using
thesubsampling distribution method
n (
n -)

(11.8)

then thecumulative distribution function ofthis distribution forany point x R can be approximated using thesubsampling
estimator below
Ln,b ( x ) =

1
n b +1

nb+1

1{ (
b

n, b , t n ) x }

(11.9)

t =1

Another basic assumption ofthe subsampling method is that theasymptotic distribution ofthe size of(11.8) exists. To explain
n - ), and J(x) thevalue ofthe
thecompatibility ofthe subsampling method, let J(P) denote thelimiting distribution of n (
c
(1
-a
)
, respectively, be the 1-a -th order quantiles from thedistribution
distribution function J(P) at point x. Let c(1-a ) and n,b
ofJ(P) and thesubsampling distribution, that is
c(1 a ) = inf { x : J( x ) 1 a }

cn,b (1 a ) = inf { x : Ln,b ( x ) 1 a }

With respect totheabove itcan be said that a subsampling procedure is consistent if:
p

(i) Ln,b ( x ) J( x ) for each continuity point ofthe distribution function J() ,
p

(ii) sup | Ln,b ( x ) J( x ) | 0,


x R

in case thedistribution function J() is continuous along thereal axis,


(iii) forany point ofcontinuity c(1-a ) ofthe distribution function J() thefollowing relationship holds

P( n (
n ) cn,b (1 )) 1

where a (0,1).
Sufficient conditions forthecompliance ofthe subsampling method fornon-stationary time series were formulated asAssumptions
4.2.1 and 4.2.2 ina monograph by Politis et al. (1999, pp. 102103). Let us note that thestructure ofestimator (11.9) does not require

137

theknowledge ofthe asymptotic distribution of(8), which is very convenient if such a distribution is not exactly known oris very
complicated.
The practical application ofthe subsampling method, however, is not entirely devoid ofproblems. One ofthe limitations is
undoubtedly introduced by condition b / n 0 without which, according tothemonograph by Politis et al. (1999), theproof
ofprocedural compliance appears tobe impossible, since thesubsampling method may be sensitive tochanges inparameter b.
Lenart et al. (2008), studying theproperties ofthe test based on a subsampled distribution forPC time series, reached thesame
conclusion. Theoptimum value ofparameter b is not known, except fora few special cases.
The study by Lenart (2010) proved thecompliance ofthe subsampling procedure intheclass ofnon-stationary time series
fortheFourier coefficient estimator m( y ) for any [0,2 ), which, inturn, leads totheconstruction ofa statistical test2 based on
which thefrequencies ofset Y are identified. Themethodology will be described inthenext section.

11.5. The four-step business cycle extraction procedure


In this section we shall discuss a four-step business cycle extraction procedure described indetail by Lenart (2010) and Lenart and
Pipie (2011).
Let { Pt : t Z} be theobserved time series. For thesake ofclarity, let us focus on time series observed at monthly intervals. 3
We assume that { Pt : t Z} is a time series with a finite unconditional expected value forany point intime t Z. Theequation
representing this model is based on theassumption that theexpected value functionmP (t ) = E ( Pt ) is thesum offunction f (t , b ) of
argument t Z, parameter b R s , and thealmost periodic function
g(t ) =

m (y)e
P

i yt

y P

(11.10)

of argument t Z, where P [0,2p ) . Theequation forthemodel, therefore, has thefollowing form:


P (t ) = f (t , ) + g(t ) = f (t , ) +

m ()e
P

i t

(11.11)

For set Y P we shall use thenatural representation intheform of


(11.12)
P = P ,1 P ,2 P ,3


where set Y P,1 is a set containing frequencies identified with those responsible forbusiness fluctuations, set Y P,2 is a set
offrequencies responsible forseasonal variations, whereas set Y P,3 represents all other frequencies. Set Y P,1 includes such
frequencies that correspond tocycles whose length ranges from half a year toinfinity,4 which means that

P ,1 (0; 0.35) = P ,1

Set Y P,1 includes, therefore, thefrequencies ofinterest. Set Y P,2 identified with theintersection ofset Y P with thefrequencies
set {2k p / T : k = 1,2,, T -1}, where T is thenumber ofperiods intheyear inthedata under consideration. Conversely, set Y P,2
contains other frequencies, such asthose responsible fortheinteraction between seasonal variations and cyclical fluctuations
identified with thefluctuations caused by changes inthebusiness cycle.5
In our considerations we shall assume that function f (t , b ) is a polynomial ofat most thesecond degree, that is
f (t , b ) = b0 + b1t + + b pt p , where p 2 . Thevalue ofparameter p will be assumed toequal 1 or2, depending on themacroeconomic indicator after theobservation ofthe graphics representation ofthe first orthesecond differences ofthe time series
under consideration. Let us note that function f (t , b ) may not contain an intercept, because such a term constitutes a natural
expression ofa trigonometric series g(t ) .6 Let us consider a sample time series that meets theassumptions formulated inmodel
equation (11.11).
Example 3.1 Let thetime series { Pt : t Z} be ofthe form

2
3
4
5
6

Pt = Pt 1 + et

The test is asymptotically consistent.


The time series can also be observed on a quarterly basis orat other intervals.
Not all thefrequencies that belong tothis set can be identifies with fluctuations inbusiness activity.
An example ofa time series satisfying model equation (11) can be found in: (Lenart 2010).
A sufficient condition foritto hold is that P {0 } . .

138

where E ( P0 ) = b and { et : t Z} are APC time series whose expected value function (unconditional) () that satisfies thefollowing
equation
(t ) = a + g(t ) g(t 1)

where g : Z R is an almost periodic function ofthe form:


g(t ) =

m(y)e

i yt

a R and card( ) < . . Note that forany t 1 we obtain


Pt = P0 + e1 + e2 + + et

Hence
t

E ( Pt ) = b +

E( ) = b + at g(0) + g(t ) = f (t , ) + g(t )


j

j =1

Where f (t , b ) = b0 + b1t , b0 = b - g(0), b1 = a, which means that thetime series under consideration { Pt : t Z} satisfies equation
(11.11). Note that if we assume g(t ) 0, then (t ) = a, which proves that time series { Pt : t Z} satisfies equation I(1) with
E ( P0 ) = b being an additional condition.
Below we present an overview ofa procedure fortheidentification and estimation ofmodel parameters (presented intheform
ofa four-stage algorithm) developed by Lenart (2010). In most cases, before theapplication ofthe algorithm, therealization
ofa given time series will be logarithmised tothebase e.7

Step 1. Reducing seasonal variations


Given thefact that most monthly macroeconomic indicators reflect seasonal variations, thefollowing observation appears tobe
obvious
/
P {2k p / 12 : k = 1,2,,11} =

which means that set Y P contains certain frequencies responsible forseasonal variations. Theidentification ofmodel parameters
corresponding with cyclical fluctuations is not, however, thekey objective ofthe study ofbusiness activity. In order toeliminate
seasonal variations and, at thesame time, strengthen thebusiness cycle fluctuations, we shall apply a filter from thefamily ofthe
centred moving averages totheindicator inquestion. One option is theabove-mentioned 2x12MA-type operator (cf. Makridakis
et al. 1998, Brockwell and Davis 2002). Using theoperator 2x12MA we obtain a time series {Yt : t Z} forwhich
Yt = L212 (B )Pt

where
L212 (B ) = (B6 + 2B5 + + 2B1 + 2 + 2B + + 2B 5 + B 6 ) / 24

and B Pt = Pt -k forany integer values oftime t and shift k. Let us note that forthetime series {Yt : t Z} there exists an unconditional
expected value. Applying Theorem B.1, we obtain an unconditional expected value function ofthe time series {Yt : t Z} intheform of
Y (t ) = E (Yt ) = 0 + 1t + + pt p +

f(t ,  )

m ()e
Y

i t

(11.13)

where
Y {2k p / 12 : k = 1,2,,11} =

and

YY = Y P \ {2k p / 12 : k = 1,2,,11}

After logarithmisation, theidentified cycle (using Step 4) will enable us tointerpret these fluctuations asa cycle ofdeviations. Another purpose
ofthe operation is toequalize thepossible amplitude ofseasonal variations, which will facilitate their weakening inStep 1 ofthe algorithm.

139

This means that thetime series {Yt : t Z} contains thesame frequencies inset YY asa time series { Pt : t Z} inset Y P , after
deducting thefrequencies responsible forseasonal variations. Additionally, forcoefficients mP ( y ) and mY ( y ) the following
relationship obtains:
mY ( y ) = L212 ( eiy )mP ( y )

(11.14)

For coefficients bk and bk we obtain bk = bk , fork = p and k = p1, which means that forp = 0 and p = 1 coefficients
ofpolynomials remain unchanged. If p = 2, what changes is only theintercept ofthe polynomial.
Step 1 shall be executed forall macroeconomic indicators under consideration intheanalysis ofthe business cycle regardless
ofthe justification forthepresence ofseasonal variations. This is due totheproperties offilter 2x12MA, after whose application
therandom fluctuations are reduced, which facilitates subsequent stages ofthe analysis.

Step 2. Trend elimination


The next step proposed inthealgorithm foridentifying thefrequency ofset Y P represented by (11.11) involves theuse ofan
operator that eliminates a possible trend from thetime series { Pt : t Z}. Let us recall that we are only considering a case where
function f (t , b ) is a polynomial ofthe p-th degree, where p 2. Accordingly, we shall consider two cases.
Case 1. Let p = 1. Using theoperator L1(B ) = (1- B ) forthetime series ofmoving averages we obtain a time series { X t : t Z} in
theform

X t = L1(B )Yt = Yt Yt 1 = ( Pt +6 Pt 6 + Pt +5 Pt 7 ) / 24

The expected value forthetime series { X t : t Z} exists and is an almost periodic function represented by theformula
X (t ) = 1 +

m ()e
X

i t

(11.15)

where

Y X = Y P \ {2k p / 12 : k = 0,1,2,,11}

which follows from Theorem B.1. Additionally,


mX ( y ) = L1( eiy )mY ( y ) = L1( eiy )L212 ( eiy )mP ( y )

(11.16)

and

X (0; 0.35) = P ,1

(11.17)

It should be noted that theexpected value ofthe time series is an almost periodic function, forwhich condition 1.5.1 is satisfied.
Theresultant series thus obtained { X t : t Z} can be identified with thedynamics ofchange inthetime series { Pt : t Z}.
Case 2. For p =2 we use theoperator L2 (B ) = (1- B )2 . In this way we obtain thefollowing series

X t = (1 B )( Pt +6 Pt 6 + Pt +5 Pt 7 ) / 24
X (t ) = 2 +

m ()e
X

i t

(11.18)

where

mX ( y ) = L2 ( eiy )mY ( y ) = L2 ( eiy )L212 ( eiy )mP ( y )

Y X = Y P \ {2k p / 12 : k = 0,1,2,,11}

(11.19)

Similarly, if p = 1, we obtain

X (0; 0.35) = P ,1

(11.20)

At this point itis worth explaining why, before thedetermination ofthe first and second differences, we do not perform theADF
(Augmented Dickey-Fuller) unit root tests known intheliterature. Thereason is theabsence ofjustification fortheuse ofADF tests,

140

where we assume that theexpected value ofthe time series under consideration depends on thevalues ofa nontrivial almost
periodic function. Theoperation ofdifferentiation, which is executed inStep 2, is perceived interms ofa filter that eliminates
thepossible component ofa linear trend ora stochastic trend. Unwarranted use ofthe differentiation filter does not cause theloss
ofany information (from a formal point ofview) whose frequencies are identified with changes inthebusiness cycle (cf. formulae
(11.17), (11.20)).

Step 3. Identification and estimation offrequencies


The identification and estimation offrequencies inset Y P forthetime series { Pt : t Z} thus comes down totheir identification
on thebasis ofthe realization ofthe time series { X t : t Z}. 8 For thepurpose offurther reasoning we shall formulate an additional
assumption regarding thetime series { X t : t Z}. This assumption allows fortheidentification offrequencies inset Y P ofthe time
series {Yt : t Z} based on therealization ofthe time series { X t : t Z}. We shall, therefore, assume that thetime series { X t : t Z}
is an APC time series, which means that itis also characterized by a nearly periodic autocovariance function. This assumption is
necessary toidentify thefrequencies ofset Y P using thesubsampling method. Theassumption should not be treated astoo
restrictive, because this series was produced using thedifferentiation operator and themoving averages operator fororiginal data.
In addition, ina number ofpreviously cited articles and monographs on theanalysis ofmacroeconomic indicators (cf. e.g. Parzen
and Pagano 1979, Osborn and Smith 1989, Franses and Boswijk 1996, Franses 1996, Franses and Ooms 1997, Franses and Dijk 2005)
adopt a slightly stronger assumption concerning theperiodic structure ofthe autocovariance function.
 ({ y }) = n |r n ( y ) | and
In order toidentify theparameters subject toinference inset Y P,1 we use thetest statistic P
n
({ y }) can be
thecorresponding critical values gn,b (0.99%) for theseries { X t : t Z}. We should remember that thetest statistic P
n
n ( y ) | based on a sample { X - X n , X - X n ,, X - X n }, where X n
identified with thevalue ofthe test statistic Pn ({ y }) = n | m
n
1
2
is themean ofthe sample { X1, X 2 ,, X n }. Thecritical value is calculated as
gn{ ,b} (1 ) = inf { x : Gn{ ,b} ( x ) 1 }

where
Gn{ y,b} ( x ) =

1
n b +1

nb+1

1{

t 1,b

b (|r n

( y ) | |r n ( y ) |) x }

t =1

and
r tn1,b ( y ) = 1
b

t +b1

(X X
j

n )e

j =t

iy j

In our analyses we assume that b = 2.5 n . Thevalues ofthe test statistic together with its critical values are determined
forthefrequency y from a discrete set9 contained intheinterval (0; 0.35). If thevalue ofthe test statistic exceeds thecritical
value on a certain interval I (0; 0.35), we assume that theinterval includes a frequency from set Y P,1. Thenext step involves
theestimation ofthe unknown frequency.

Step 4. Extracting thecycle using theHodrick-Prescott filter


This step involves theselection ofthe filtering parameter using theHodrick-Prescott method (HP filter) inorder tocorrectly
extract thefluctuations ofinterest. TheHP filter shall be understood asa filter (extracting cyclical fluctuations) intheform

a j B j = 11/ (1+ l(1 B )2 (1 B1)2 ) , where filter 1-L(B ) is a filter that extracts thelong-term growth ordecline
of L(B ) =

k =

path (trend).
In empirical applications ofthe filtration method proposed by Hodrick and Prescott, itis particularly important toproperly
select thesmoothing parameter l. Changing theparameter affects thesmoothness ofthe isolated trend line. Thelarger its
value, thesmoother thetrend line, and thereby theextracted fluctuations (i.e. thedifference between thedata and thevalues
along thetrend line) include cycles ofgreater length. An algorithm fortheselection ofthe smoothing parameter l depending
8

The literature on theanalysis ofbusiness activity fluctuations also offers examples ofthe use ofFourier series asa method fortheapproximation
ofthe trend function. Milas et al. (2006) (Chapter 9, pp. 221246) considered a model with a time-varying intercept, which was approximated
using Fourier series. This approach, however, is somewhat different from those where Fourier analysis is used toidentify and estimate
theparameters which may affect theplot ofeconomic fluctuations. Thetrend function is not subject toestimation, aswas thecase with
theresults discussed by Milas et al. (2006).
This set shall be selected toensure that its cardinality is several times larger than thesample size.

141

on thelength ofthe cycles ofinterest was presented by Maravall and del Rio (2001). In their work (p. 17, formula 5.3) thevalue
ofparameter l constitutes thefollowing function offrequency w0 .
= [4(1- cos( 0 ))2 ]-1

(11.21)

Frequency w0 can be interpreted asthelower limit ofthe frequencies ofinterest. Determining the w0 , which so far intheliterature
has been done arbitrarily, also determined themorphological features ofthe obtained cycle estimators. On thestrength
ofthe construction, thevalue ofparameter l can be interpreted asthevalue forwhich, after theapplication ofthe HP filter,
thecorresponding variations infrequencies above thevalue of w0 will be amplified, whereas thecorresponding frequencies
below thevalue of w0 will be diminished. Such an interpretation is possible by demonstrating that theHP filter can be obtained
asa special case ofthe Butterworth filter (cf. Gmez 1999, Gmez 2001).
The approach proposed inthis study consists inthedetermination ofthe smoothing parameter l using thevalues ofstatistically
significant frequencies identified inStep 3 ofthe algorithm. Therefore, we select thesmoothing parameter so asto reduce
thefluctuations which do not result from changes inthebusiness cycle while strengthening those fluctuations which have been
identified with such changes. Theselection ofparameter l thus becomes a statistical procedure with asymptotic properties, and
is no longer a purely arbitrary decision, which has been thecase so far.
In order toexplain theabove mechanism let us consider an example inwhich fluctuations have been isolated using theHP filter
fordifferent values ofparameter l.
Example 3.2. Let us consider thetime series { Pt : t Z} ofthe form
Pt = cos(0.262t ) / 2 + 4 cos(0.15t ) + 10 cos(0.06t ) + 5 sin(2t / 12) + t

(11.22)

where t = t 1 + t ,, h0 = 0 and { et : t Z} is Gaussian white noise with a mean which equals 0.6 and variance which equals one.
For monthly data, thefrequency of0.262 corresponds toa 2-year cycle, thefrequency of0.15 toa 3.5-year cycle, and thefrequency
of0.06 fora 8.7-year cycle. Based on (11.12), itcan be rewritten as {0.262, 0.15, 0.06} P ,1 and 2pt / 12 P ,2 . Thecardinality
ofthe realization n = 160 was determined using model (11.22) and thecentred moving average filter 2x12MA with a view
toweakening theseasonal variations. Fluctuations were isolated using theHP filter fordifferent smoothing parameters. Theresults
obtained are shown inFigure 11.1.
1

60
50
40
30
20
10

50

0,8

40

0,6

30

0,4

20

0,2
0,05

(b) 2x12MA

(a) Realization of the time series

0,15

0,2

0,25

0,3

0,35

(c) HP filter amplification for


= 5500, 12000, 32000, 55000, 133100
(from right to left)

(1.22), n = 160.
3
2
1

0,1

-1
-2
-3

-2

-2
-4

-4

(e) = 12000, (0 5.5 years)

(d) = 5500, (0 4.5 years)

4
2
-2
-4
-6

(g) = 5500, (0 8 years)

(f ) = 32000, (0 7 years)

6
4
2

6
4
2

-2
-4
-6

-2
-4
-6

(h) = 1331000, (0 10 years)

(i) A comparison of results

Fig. 11.1. Effects ofthe HP filter fordifferent values ofthe smoothing parameter l on therealization ofa time series ofthe form
(11.22) after theapplication ofthe centred moving average
Source: own study.

142

Note that forthevalues ofparameter l = 5500,12000,32000 (which, according toformula (11.21), correspond to4.5-, 5.5-, and
7-year cycles) theextracted cyclical fluctuations are characterized by an average length of34 years (cf. Figure 11.1 (d)(f )). When
analysing fluctuations extracted using l= 55000,13310 (which, according toformula (1.21), correspond to8- and 10-year cycles),
we can see a clear impact offrequency 0.06 (in equation (11.22) itcorresponds toa 8.7-year cycle) on theplot ofthe fluctuations
obtained. For thesmoothing parameters thus selected, fluctuations ofthe corresponding length of8.7 years are not effectively
separated due totheproximity ofthe frequency 0.06 toa frequency corresponding with thesmoothing parameter l and due
tothelarge amplitude ofthose fluctuations (cf. model equation (22)). Therefore, theHP filter has reduced these variations toa lesser
extent forl = 55000,13310 (which resulted inan artificial cycle whose length was approximately 68 years) than forparameters
l =5500,12000,32000. Figure 11.2 (c) shows linear filter amplification | L( e-iy ) | significant fortheinterpretation ofthe amplitude
ofthe extracted fluctuations.

2p/12

4p/12

6p/12

8p/12

10p/12

12p/12

 ({ y }) = n |r n ( y ) | fortherealization ofprocess { X : t Z }. The values


Fig. 11.2. Identification ofsignificant values ofstatistic P
t
n


ofthe test statistic Pn ({ y }) = n | r n ( y ) | for y from thediscrete set {(k -1)p / 720 : k = 1,2,,720} ; thevalues of
gn{ y,b} (99%) for y from thediscrete set {(k -1)p / 720 : k = 1,2,,720}
Source: (Lenart 2010).

Analysing theidentified frequencies from set Y P,1, parameter l is selected so that thefrequency determined by formula (21) finds
itself between thefrequencies identified with thefluctuations inthebusiness cycle and thefrequencies identified with thelongterm growth trend. Thereason is that absence ofinformation on significant frequencies that belong toset Y P,1 may infact lead
toerroneous conclusions. Setting an arbitrary smoothing parameter may significantly distort thepicture ofcyclical fluctuations
(in terms ofconclusions reached) identified with changes inbusiness activity.
 ({ y }) = n |r n ( y ) | and
Fig. 11.2 illustrates how Stage 3 ofthe proposed algorithm works. It contains thevalues ofthe test statistic P
n
{
}
y
thevalues of gn,b (99%) for y from thediscrete set {(k -1)p / 720 : k = 1,2,,720} for thefirst process differences ofthe centred
moving average asapplied totheproduction index inPoland.10
Areas inwhich theidentified values ofthe static test exceeded thesubsampling quantile are shown inFigure 11.2. asred rectangles.
These areas are also shown inthefigure below. Thevertical lines inFigure 11.3 (a) represent thesampled estimator values ofthe
unknown frequencies from set Y1,P . These values are summarized inthetable below.
Table 11.1. Estimated frequency values with their corresponding cycle lengths (in years) fortheindustrial production index inPoland
Estimated frequency value
Corresponding cycle length (in years)

= 0.062
y
n,1

= 0.153
y
n,2

= 0.258
y
n,3

8.5

3.4

Source: own study.

On thebasis ofthe obtained estimates ofthe unknown frequencies from set Y1,P theparameter ofthe Hodrick-Prescott filter is
selected. Theselection is thus based on results justified from a formal point ofview. This allows us toavoid choosing theparameter
inan arbitrary manner. Thefigure below shows fluctuations intheindustrial production index inPoland11 extracted inthis way.
Thefollowing Hodrick-Prescott filter parameters have been adopted: l = 5500, l = 12000, l = 32000, l = 55000, which corresponds
totheweakening offluctuations which exceed (respectively): 4.5 years, 5.5 years, 7 years and 8 years. Theselection ofsuch a set
ofparameters was due totheabsence ofidentified variations inthe48-year interval.
10
11

A single base index (2005 = 100%) from January 1995 toDecember 2009.
A single base index (2005 = 100%) from January 1995 toDecember 2009, after theapplication ofthe centred moving average 2x12MA.

143

a)

b)

0.04 0.09

0.13

0.17 0.22

0.26

0.31

0.35

0.39 0.44

2.01 2.05 2.09 2.14 2.18

2.23

2.27 2.31 2.36

2.4

Fig. 11.3. Identification ofessential values ofthe test statistic

n |r n ( y ) | fortherealization ofa discrete process { X t : t Z} : (a) thevalues ofthe


 ({ y }) = n |r n ( y ) | (black line) and thevalue of gn{ ,b} ( ) (blue line) for a {92%, 95%, 99% } and y from thediscrete
test statistic P
n
 ({ y }) = n |r n ( y ) | (black line) and thevalues of g{ } ( ) for
set {(k -1)p / 720 : k = 1,2,,100} ; (b) thevalues oftest statistic P

Identification ofsignificant values oftest statistic

n, b

a {92%, 95%, 99% } and y from thediscrete set {(k -1)p / 720 : k = 450, 451,,550} .
Source: (Lenart 2010).

A more detailed discussion ofthe methods ofestimation ofthe amplitude and phase offluctuations identified with fluctuations
inbusiness activity (and their graphical representation and interpretation) can be found inLenart (2010).
A detailed description offluctuations inbusiness activity Polish appears tobe only a preliminary step inmodelling business
cycles. Naturally, thenext step should involve forecasting thephases ofthe business cycle based on appropriately constructed
econometric models. This issue, however, is quite complex due tothecharacteristics ofcyclical fluctuations. Theinnovative design
methodology offorecasting cycles using theBayesian approach is thesubject ofthe next section.

0.06
0.04
0.02
`95

`96

`97

`98

`99

`00

`01

`02

`03

`04

`05

`06

`07

`08

`09

-0.02
-0.04
-0.06
Fig. 11.4. Theprocess isolated from themoving average ofthe logarithm ofindustrial production using theHodrick-Prescott filter
for: l = 5500 (black line), 12000 (red line), 32000 (blue line), 55000 (green line)
Source: own study.

11.6. The Bayesian forecasts ofmacroeconomic indicators using one-dimensional models


The proposed methodology forobtaining short-term projections ofkey microeconomic indicators involves theapplication
ofBayesian inference tothebasic parametric models ofa given time series. This is dictated by two fundamental reasons. First,
itpermits a fully formal comparison ofcompeting models and combining knowledge. Problems related tothechoice ofa model
do not require arbitrary decisions involving one specification and rejecting other, competing models. Thefinal projections result
from formal inferences using all themodels considered together, with uncertainty related tothechoice ofa specification being
automatically reflected intheforecasts. Second, Bayesian inference provides a useful, formally-grounded and non-arbitrary
description ofuncertainty regarding theprojected valued intheform ofa predictive distribution. In contrast tocertain other
procedures involving thestudy ofex-post error distribution, itis not necessary toinvoke arbitrary methods adopted on an ad-hoc
basis. In theBayesian model, predictive distribution is defined inan unambiguous manner. When examining error distributions
ex-post, even given a fixed form ofthe sampling model, one may consider various types oferrors, moreover, theresults obtained
can be highly sensitive tothelength ofthe series ofexpired forecasts under consideration.

144

The primary forecasting tool comprises one-dimensional models forindividual time series. This assumption stems from thefinding
that themain goal ofinference is theexamination ofthe future values ofindicators which characterize thestate ofthe economy.
An analysis ofthe future relationships among them poses a more advanced problem. In such an approach itmay be justified
tolimit ourselves toone-dimensional models. Due tothefact that theprimary objective ofinference is prediction, we have
excluded exogenous variables from theanalysis, since itwould require thespecification ofan additional model, which inevitably
leads toundue complexity.
For thepurposes ofthe present analysis, itis assumed that variable yt is subject tomodelling, which represents increments
(or logarithmic rates ofchange) ofmacroeconomic indicator xt under consideration. It is further assumed that T observations
are available (realizations ofrandom variables) representing thevalues of yt insuccessive equidistant time intervals (periods).
Thevalues of yt insubsequent H periods are treated asunavailable and are subject tostatistical inference (prediction).
For thesake ofclarity we have adopted a generic notation, inwhich p(.) represents thedensity functions ofan absolutely
continuous probability distribution, forwhich theidentification ofdistributions ofdifferent variables follows from thecontext and
thedesignation adopted forthearguments offunction p(.). Likewise, inthecase ofdifferent sampling models, thevector ofall
parameters shall be designated as without a subscript identifying themodel. To simplify thenotation, we have also decided that
thesequence ofrandom variables { yt } (a stochastic process with discrete time) can be represented by a single typical element yt ..
Each ofthe competing sampling models corresponds toa family ofT-dimensional distributions ofa random vector y which groups
theavailable observations parameterized using thefinite-dimensional parameter vector . It is assumed that theadopted form
ofa sampling model also determines theform ofthe conditional distribution ofvector y f that groups H unavailable observations,
which constitute thesubject ofstatistical inference (with a fixed past ofthe observable process and set parameter values).
The Bayesian model represents thecombined distribution ofparameters , available observations y, and unavailable observations
f
y f , which corresponds totheprobability density function p( y , y , q ). It should be added that theelements ofvector may involve
unavailable observations which do not constitute thesubject ofinference (e.g. those representing theunknown initial conditions
ofa dynamic process) aswell ashidden variables.
The following decomposition is being considered forthepurpose ofthe specification based on theBayesian model:
p( y f , y , q ) = p( y f | y , q ) p( y | q ) p( q )

where: p( y | y , q ) represents thesampling-theoretical predictive distribution, p( y | q ) corresponds with thesampling model,


and p( q ) represents theso-called prior distribution. Thefirst two elements are used inthenon-Bayesian inference and can be
specified by determining theform ofthe observation equation. p( q ) (the prior distribution) reflects theassumption ofrandomness
ofparameters adopted inBayesian inference and on thestrength ofthe construction itconstitutes is themarginal distribution
ofunknown parameters (relative toobservations). Uncertainty about thepossible parameter values (before

accessing data) is here


represented formally ina probabilistic manner. Thespecification ofa particular form of p( q ) inresearch practice usually refers
totheidea (formal oronly approximate) that reflects theabsence ofinitial knowledge about theparameters ofthe model. This
clause requires, ofcourse, a definition ofthe space inwhich theparameters and observations assume their values.
Statistical inference is possible according tothefollowing decomposition ofthe combined distribution ofobservations and
parameters:

p( y f , y , q ) = p( y f | y , q ) p( q | y ) p( y )

This reflects thebasic idea ofBayesian inference, which is conditioning with respect toobservable parameters (y). The p( q | y )
distribution is called theposterior distribution. It represents theknowledge concerning theparameters ofthe model after access
torelevant data. Within Bayesian inference, initial knowledge about parameters represented by theprior distribution is influenced
by observations and undergoes modification totheform ofthe posterior distribution. Differences between theprior and posterior
distributions reflect theimpact ofobservable data y on theknowledge ofunknown parameters .
If thelikelihood function, which is analytically identical with p( y | q ) is denoted by Ly ( q ) , thecore ofthe posterior distribution
can be rewritten as

p( q | y ) Ly ( q )p( q )

If thegoal ofinference is only toreach a prediction, thepredictive distribution ofthe following form is considered:
p( y f | y ) =

p(y | y, )p( | y )d
f

145

In theabove formula, thetotal distribution ofthe unknown parameters (i.e. unavailable observations and parameters) conditioned
by theknown values (i.e. theavailable observations) were integrated. This corresponds totheidea ofBayesian inference which
involves marginalization with respect tounknown parameters which remain outside theresearchers interest. In this case itmeans
that all theuncertainty associated with theinference concerning unknown parameters has been formally included intheforecast.
Thepredictive distribution may also be interpreted inthefollowing way: thedistribution ofunavailable observations is obtained
asa mix oftheory-sampled predictive distributions (conditional with respect to ) and mixed according totheposterior distribution
f
ofparameters . In this approach, thepredictive distribution p( y | y ) represents both uncertainty associated with thestochastic
mechanism generating theobservations and uncertainty related todrawing inferences about theunknown parameters ofthe
model.
In thebasic model employed inthis study, theobservation equation defining thesampling model has thefollowing structure:
yt = g (t ; g ) + et ,

t =

i t i +

j =1

t = 1,2,..., T

i t i

k =1

+ t

t | q : iiN 0, 1

It means that itwas assumed that therates ofchange (increments) under consideration are described by function g (t; g ) , which
reflects theaverage level ofchange xt , including its possible regular (e.g. seasonal) fluctuations, and thedeviation process et .
Thelatter is defined by theconditional distributions (in terms ofconditioning with respect tothepredetermined parameters and
thepast ofthe process). Their structure corresponds tothestationary and causal Gaussian autoregressive moving averages model
ARMA(J, K) (with reversible part MA) and strict white noise zt with precision t (cf. Brockwell and Davis 1986, Gelfand et al. 1995).
Thedistribution ofvectors y and is obviously determined in RT . Thedeviation process et , strictly speaking, does not have tobe
a stationary process. This is due totheassumption adopted above concerning thetime domain limited from below.
In such a case, itis necessary tospecify theassumptions concerning theinitial conditions ofthe observable process and white
noise (we shall denote them, respectively, by y(0) and e(0) ). Here we have assumed that theinitial conditions are unknown, and
therefore must be represented by hidden variables. We have also assumed that they are stochastically independent ofthe future
ofthe white noise zt , but thedistributions resulting from analysing thestationary process ARMA inthedomain ofintegers (for
theobservation index) are assumed not toapply. From theviewpoint ofBayesian inference, theinitial conditions may thus be
treated ina similar way totheother parameters inthemodel.
The adoption ofdistributions ofinitial conditions which do not reflect thefull homogeneity intime ofprocess et allows
fora certain weakening ofassumptions about thestructural stability ofthe data generation mechanism under consideration. For
example, theimpact ofincreased variability of xt intheperiod oftime preceding theone under review can be visualized inan
initial sequence ofmodelled observations asthe(decreasing) impact ofinitial conditions.
The model adopted forprocess et implies at least its asymptotic stationarity (in thesense that with increasing t, thedistribution
of et converges with thestationary distribution ofthe ARMA process with zero astheunconditional expected value). Due
totheassumption concerning theGaussian distribution ofrandom components zt , inthecase under consideration (for theARMA
process with time inthedomain ofintegers) strict stationarity is equivalent tocovariance stationarity.
Stationarity ofprocess yt itself depends on theform ofthe function g (t; g ) which defines theessential component ofthe
sampled expected value yt . Theexpected value yt also depends on theexpected value of et , which can be non-zero due
totheinfluence ofthe initial conditions. As mentioned above, ast increases, theexpected value of et tends tozero, so that
thesampled expected value ofthe observed process is dominated by theform of g (t; g ) . We are talking about theunconditional
sampled expected value of yt inthesense that given a fixed and known t, and fixed parameters ofthe model, is conditioning with
prior realizations ofthe process is not considered (i.e. realizations up tothemoment including theindex t 1). Thus understood
unconditional expected sampled value yt with increasing t converges with g (t; g ) .
If function g (t; g ) is not constant with respect to t, itis possible that thesampled model fortheobserved process will not exhibit
even asymptotic stationarity. However, this will only be associated with thepossible fluctuations intheabove-mentioned expected
value; unconditional variance insuch a model cannot grow indefinitely. In this sense, thenon-stationarity generated inthesampled
distribution of yt is different innature from thenon-stationarity caused by single elements inan ARMA-type process associated with
theconcept ofintegration I(1). If yt is defined asincrements of xt , themodel considered, infact, allows fora certain generalization
ofproperty I(1) with respect totheimplied process describing theevolution ofthe basic macroeconomic indicator xt .
146

In accordance with theassumptions set out above, thevector ofall theparameters ofthe model has theform

0
0
= , h , , , y( ) , ( ) . Theinitial conditions have been previously defined ashidden variables, but from a formal point
ofview inBayesian analysis they are treated inthesame way asparameters. We assume a fixed form of g (t; g ) aswell asvalues
J and K. Due tothedifferent possibilities ofspecification of g (t; g ) and theselection ofvalues J and K itis necessary toperform
a formal comparison ofmodels.

In thefirst stage ofthe analysis, we assume that function g (t; g ) forsuccessive t adopts a repetitive sequence S offixed values
(s=1, 2, ..., S). For S = 1 we are dealing with a constant function with respect tot. For S =12, function g (t; g ) if we take into account
monthly data assumes thesame value g s each time inthesame calendar month. Since theincrements ofmacroeconomic
indicators are subject tomodelling, there is no need todeal with g (t; g ) intheform ofpolynomial trends. In summary, vector g
S
comprises S unknown parameters. It is assumed, therefore, that g R .
Vectors h and y represent parameters ofpart of ARMA(J, K). Due totheconditions ofcausality and reversibility, thevectors belong
tosets ofa fairly complex shape inthegeneral case. These conditions imply h H J and y yK , aswell as Hi Ri and Hi = Yi .
Between vectors h and y therelation offree variation holds (i.e. , H J K ).

(0)

The initial conditions are met by y and e which assume values in R and R K . It is understood that both y( ) and e( ) are
random vectors formed by variables with identical independent Students t-distributions with thenumber ofdegrees offreedom
greater than 2. Obviously, thedistribution parameters oftypical elements y(0) and e(0) are different. Theassumption ofidentity
and independence ofthese distributions is dictated by thedesire tosimplify thestructure ofthe model.
(0)

The proposed structure ofthe prior distributions inthis model is asfollows (the initial conditions below will be treated inthesame
way asother parameters):

0
0
p( ) = p( ) p( )p( ) p( t )p( y( ) ) p( ( ) )

All these distributions are proper distributions with thegroups ofparameters exhibiting prior independence. For parameters
g , a prior multivariate normal distribution was assumed. Parameters h and y are assumed tohave uniform distributions on
H J and YK (it can be shown that these are proper distributions). Theprecision parameter t is assumed tohave prior gamma
(0)
0
distribution, while theinitial conditions y( ) and e can be treated asparameters forwhich prior identical, independent Students
t-distributions are assumed.
The adoption ofsuch a prior structure has significant advantages. First, ithelps toensure thecoherence ofthe prior structure,
which is important because ofthe need tocompare models. Uniform distributions for h and y , themultidimensional normal
(0)
0
distribution for g and distributions with independence for y( ) and e allow us, inprinciple, tomeet thecoherence postulate
incomparisons ofmodels with different values J, K and S. Second, since all theprior distributions are proper distributions, themodels
can be compared using theNewton-Raftery estimator (cf. Raftery et al. 2007 and thecited references). Unfortunately, itcannot
be demonstrated that such a prior distribution structure satisfies theformal conditions ofnon-informativeness (derived e.g. from
theJeffreys rule). In thecase of ARMA structural parameters, thesolution applied here (the uniform distribution) is probably less
arbitrary than thenormal distribution truncated by therestrictions ofcausality orreversibility often adopted inmultidimensional
applications. Theadoption ofthe Gaussian distribution where theshape ofsets H J and YK is quite complicated does not allow
foran easy control ofthe adopted assumptions. Certain areas ofthe parameter space can thus be a priori strongly preferred
orstrongly dispreferred ina purely random manner.
It can be shown that if theassumptions ofthe Bayesian model described here are adopted, theanalysis ofthe posterior distribution
can be carried out using MCMC methods (cf. Tierney 1994, Robert and Casella 2004). Unfortunately, due tothereversibility and
causality restrictions (and theadopted prior structure) itis not possible tofind a complete conditional posterior distribution
ofparameters h and y ofa standard form toenable easy randomization. Therefore we propose touse a hybrid algorithm

(0) (0)
. Thecomplicated appearance
intheform ofthe Gibbs sampler together with theMetropolis algorithm forvector , y , y ,
ofsets H J and YK suggests a re-parameterisation invoking themutually unambiguous mapping Hi Ri (cf. Gelfand et al. 1995
and thecited references). Such a re-parametrization forestimation purposes allows forthereplacement ofparameters h and


,
, y(0) , (0)
2 J +2K
and y
, which allows fortheassumption that thedomain ofvector
y by their counterparts h
is R

and theuse ofthe Gibbs sampler with theMetropolis algorithm inits random walk version. Such a re-parameterisation may be
necessary forthestudy ofmodels characterized by a large number ofdelays inthestructure. Sampling from thedistribution
intheMetropolis algorithm while rejecting thevalues that

do not meet therestrictions ofcausality orreversibility may be extremely


inefficient insuch a case.

The step ofthe Gibbs algorithm associated with parameter t can be implemented inmany ways. Thescalar dimension ofthe
parameter facilitates sampling. For parameters g thefull conditional distribution is a multivariate normal distribution, which makes
thelatest step intheGibbs cycle easy toimplement.
147

The use ofa hybrid Gibbs sampler allows us toobtain a (dependent) sample ofthe distribution p( q | y ) . Generally, MCMC
procedures generate a correlated sample from theposterior distribution. Thesimulation tool is a Markov chain type ofan
innumerable number ofstates whose stationary distribution is p( q | y ) . In connection with theasymptotic character (with respect
tothenumber ofiterations ofthe algorithm) ofconvergence and thepotentially strong correlation ofthe sampler, itis necessary
togenerate numerous pseudo-random samples and monitor (at least with a degree ofapproximation) convergence ofthe chain
used.
f
Due tothestandard form of p( y f | y , q ) obtaining a sample from thepredictive distribution p( y | y ) is relatively easy.
For parameters randomly drawn from theposterior distribution ofvector realisations and denoted by q( i ) , a sample from
p( y f | y , q = q( i ) ) is obtained by direct sampling using conventional procedures.
f

Having sampled thepredictive distribution p( y | y ) , we can estimate theposterior probabilities ofevents referring by definition
tothefuture values ofthe indicator inquestion. Thediagnosis ofthe state ofthe economy may therefore depend on thevalues
ofposterior probabilities obtained forthespecific configuration ofthe future value ofthe variable under consideration. It is
also possible torepresent uncertainty ofthe future values ofthe modelled process, i.e. yT +h or xT +h (for h = 1, 2,..., H) using,
forexample, a fan chart.

11.7. Bayesian forecasts ofthe business cycle using a model with cyclical fluctuations
The model proposed intheprevious section was built on theassumption that thestructure offunction g (t; g ) reflects seasonality
that occurs intheincrements ofa given macroeconomic indicator inquestion. It is possible toextrapolate this concept by invoking
an analogy with theidea ofAPC processes discussed previously. This allows us toredefine theobservation equation ofthe model
under consideration tothefollowing form:
yt = d (t ; d, ) + t ,

t = 1, 2,..., T

d (t ; , ) = g (t ; ) +

f sin(tf ) +

f =1

cos(t )
f

f =1

where itis assumed that j = (j1,..., jF ) and F 0, , while thevalue ofF is finite (fairly small inpractice). For function
g (t; g ) a specification discussed intheprevious section was adopted, namely that its value is represented by parameters g s

which belong tothea repetitive sequence containing S elements. It is also assumed that a = (a1,..., aF ) , b = (b1,..., bF ) . Vector d
F

groups parameters with respect towhich function d (t;.) is linear: = , , . It is further assumed that d R2F +S . Themodel
developed inthis way allows us torepresent more complex cyclic structures intheunconditional sampled expected value yt
. If we assume that S = 1 and yt = Dxt , thedifference d (t ; ,j ) g(t ; ) may represent cyclical deviations from thelinear trend
intheunconditional sampled expected value xt (for thesake ofsimplicity, we have disregarded theeffect ofinitial conditions
which diminishes intime, and also adopted some weak technical assumptions toensure theexistence ofthe expected values
under consideration).

Bayesian inference insuch a model needs tobe supplemented with thestructure ofthe prior distributions asdescribed
intheprevious section. In this paper, we propose toadopt forthevector parameters d ofa multi-dimensional normal distribution,
while forvector theuniform distribution on F (which, due totheassumptions adopted above, is its proper distribution).
A separate issue is related todetermining theacceptable range ofvalues forparameters jf . Admitting values too close tozero
may allow theexistence ofcomponents ofthe cyclic dynamics whose period is shorter than thelength ofthe vector ofavailable
observations. It appears reasonable tolimit thedomain ofparameters jf from below, e.g. by k 2p / T , for k {1,2,3} . Additionally,
itis possible toexpress theprior knowledge ofthe planned structure ofthe modelled dynamic process by further limiting
thedomain of jf to certain sub-intervals of 0,p . Preservation ofmethodological consistency ofBayesian inference, however,
requires that thedomain inwhich theprior distribution ofthe vector assumes a positive density should be determined before
accessing data.
Moreover, certain considerations ofa methodological nature may justify imposing a priori restrictions on thetotal space occupied
by parameters d and j . This may be dictated by thedesire toensure identifiability. Thedefinition offunction d (t; d,j ) adopted
above allows forF > 1 theexistence ofmultiple observationally equivalent (in thesense ofthe sampled model) combinations
ofparameter values. From theviewpoint ofaccuracy ofBayesian methodology used forconstructing predictions, therequirement
is not essential. Ensuring identifiability may, however, be dictated by thedesire fora clear representation ofa posteriori knowledge
about theparameters ofthe model. Thelack ofidentifiability may, infact, lead toa posterior distribution ofa very complicated
shape. In such a case thedescription ofa posteriori information applicable toelements ofvector using traditional characteristics
(such ascentral moments) may be far from satisfactory.
148

Sampling from theposterior distribution insuch a case may also invoke thehybrid Gibbs algorithm. Thesampling process
described intheprevious section requires some modifications. Thefirst one consists inthereplacement ofvector with theentire
vector d (in this case, thestructure ofthe full a posteriori conditional distribution is analogous). Thesecond difference is more
basic inthat theestimation ofvector j requires an additional step intheGibbs cycle. Thefull conditional posterior distribution
forvector j is non-standard, hence sampling from itrequires a step invoking theMetropolis algorithm inits independence chain
orrandom walk versions. In thelatter case itmay be convenient tore-parameterise vector j which would allow forsearching
thedomain intheform of RF .
In principle itis also possible touse a version ofthe Metropolis algorithm forall parameters ofthe model together. It seems,
however, that when considering larger values ofJ, K, S and F, such a solution may be oflimited utility. Thesize ofthe combined
space ofthe parameters inthecase under consideration is S + 3F +2J + 2K +1. In thecase ofmonthly data, is difficult toexclude
values close to12. A similar argument may refer toS, which may be replaced toa certain extent by allowing F > 1.
The use ofMCMC algorithms tosample from theposterior distribution ofthe model discussed inthis section, however, requires
special attention. Thecombined posterior distribution may have a complex multimodal structure. In turn, algorithms such
astheGibbs sampler may encounter difficulties intheexploration ofhighly multimodal distributions. Obtaining a practically
useful Markov chain may therefore require fairly detailed research ineach case.
It is assumed that thestructure analogous tothat described by theabove observation equation is defined by theH-dimensional,
f
sample-theoretical distribution using thepredictive p( y | y , q ) . Sampling from theBayesian predictive distribution is done by
direct sampling using a sample from theposterior distribution. This allows us toobtain a predictive distribution of yt or xt
t =T + 1, ..., T + H) and its characterization using thetechniques discussed intheprevious section.
Adopting thespecification of g (t; g ) described above, one may consider theposterior distribution ofa T + H-dimensional function
vector ofthe parameters ofthe model designated as c (d , j ) with a typical element ofthe form:
ct (d , ) = d (t ; d , )

1
S

s=1

The posterior distribution ofthis parameter may represent theknowledge ofthe cyclic component indicated by theanalysed
model of yt distribution (including its predictive distribution). It is a matter ofdebate towhat extent predictions forsuch a function
ofparameters can be identified with theprojection ofcomponents representing thebusiness cycle. Forms c (d , j ) and d (t; d,j )
may require themodification depending on theempirical and theoretical properties ofthe time series under consideration. It
appears, however, that theanalysis ofthe posterior distribution offunction c (d , j ) (or functions ofa similar structure) can provide
important information related tothepotential future value ofthe cyclical nature ofthe time series studied. Knowledge of c (d , j )
(especially intheprediction period) can be represented by point characteristics (expected value, quantiles) orusing a fan chart
made up by marginal posterior distributions. It is also possible toanalyse thedecomposition offuture values into components
d (t;d,j ) and et , and analyse therelationships that hold among them.
Bayesian inference allows fora formally grounded examination ofa posteriori uncertainty that affects inferences concerning
thecyclical component. Naturally, itis also possible tostudy posterior probabilities ofspecific events defined with reference
toeither thefuture value ofvariable xt under consideration ortoan otherwise defined cyclical component. Theidentification
ofthe cyclical component ineach case and its suggested decomposition into seasonal variation components and business cycle
fluctuations must be made individually, taking into account thenature ofthe series under consideration inthecontext ofeconomic
knowledge.
We shall illustrate themethodology presented here using an analysis ofthe annual dynamics ofindustrial production (using
monthly statistical data from GUS). Theanalysis ofthe macroeconomic time series updated on a monthly basis enables us
tominimize thedelays associated with data publication. This is especially important forthemonitoring ofthe current economic
situation interms ofthe possibilities and advisability ofintervening.
Figure 11.5 shows theyear-on-year percentage change inindustrial production from January 1997 toMarch 2011 (number
ofobservations T= 171).
Let us consider a model where F = 3, J = 12, K = 0, S = 1, which corresponds with theadoption ofthree harmonic components
infunction d (t; d,j ) , twelve delays intheautoregressive part ofthe model, theexclusion ofthe moving averages process and
theadoption of g (t; g ) representing a single intercept fixed intime. Due tothedesire toextract long-term fluctuations we assume
that theparameter domain jf is limited totheinterval 0;0,52 . This corresponds totheexclusion offluctuations with a period
ofless than one year.

149

Industrial production: change y/y (%)


30

30

25

25

20

20

15

15

10

10

-5

-5

-10

-10

-15

-15

-20

I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

-20

2010 2011

Fig. 11.5. Year-on-year change inindustrial production (%) from January 1997 toMarch 2011
Source: own study.
Industrial production y/y (%) and the scoring of regular cyclical fluctuations from January 1997 to March 2011

30

30

25

25

20

20

15

15

10

10

-5

-5

-10

-10

-15

-15

-20

I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I IV VII X I

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

-20

2010 2011

Fig. 11.6. Year-on-year change inindustrial production (%) and theregular cyclical component (point assessments) from January
1997 toMarch 2011
Source: own study.

In themodel above, thefunction ofparameters d (t;d,j ) represents theregular cyclical component. It differs from ct (d,j ) only
with respect toa shift by a constant value, which represents theaverage change inindustrial production. Thefigure shows theplot
ofthe posterior expected value offunction d (t; d,j ) for t = 1, ...171, which corresponds totheperiod studied. In Figure11.6
thedata values observed

have also been shown.

150

An informal analysis ofFigure 11.6 appears toindicate that theyear-on-year changes inoutput exhibit cyclical regular cyclical
fluctuations. Theplot ofratings ofthe cyclical component shows a substantial overlap with thedata plot.
The model shown here allows us toobtain both theposterior distribution forfuture values ofthe regular cyclical component
represented here by thevalues of d (t; d,j ) and thepredictive distribution ofthe explanatory variable, i.e. thepercentage
ofyear-on-year change inindustrial production. Based on themodel, thepredictive distribution ofthe rate ofchange inindustrial
production was obtained forApril 2011 (forecast horizon H = 1). Theindustrial production growth rate inApril 2011 published by
GUS (compared with April 2010) is 6.6%. It is located between thequantile ofthe 0.55th order ofthe predictive distribution (which
equals 6.38%) and thequantile ofthe 0.6th order (which equals 6.85%). Theexpected value ofthe predictive distribution is 5.93%
with a standard deviation of3.7 percentage points.
Industrial production: Actual values compared with the forecast

20

20

15

15

10

10

-5

-5

-10

-10

-15

II III IV V VI VII VIII IX X XI XII I

2008

II III IV V VI VII VIII IX X XI XII I

2009

II III IV V VI VII VIII IX X XI XII I

2010

II III IV

-15

2011

Fig. 11.7. Year-on-year changes inindustrial production (%): A fan chart ofpredictive distribution and actually observed values;
(%); sampling period toMarch 2010, prediction period toApril 2011
Source: own study.

Industrial production: actual values compared with theforecast


In order toillustrate thepredictive properties ofthe model ina longer timeframe, itwas assumed that thelast available
observation is thevalue corresponding toMarch 2010. Themodel was then re-estimated. Figure 11.7 shows a fan chart based on
thepredictive distribution ofindustrial production with themarked realized data values. Thefan chart was plotted using an informal
approximation: relevant values were obtained from themarginal predictive distributions foreach forecast horizon. Thebands on
thefan chart represent respectively 30%, 60% and 90% prediction intervals (produced by themarginal quantiles ofthe predictive
distributions). Forecasts presented inFigure 11.6 appear tohave a good ex-post accuracy. In thecase ofa sequence offorecasts
for13 consecutive periods, only three observations lie outside the60% prediction intervals. Theproposed methodology provides
fairly accurate forecasts ofturning points.
It appears that thepresented methodology allows us both toanalyse regular cyclical fluctuations inthesampling period and
toobtain forecasts ofthe observable variable, which makes extensive use ofthe information concerning thecyclical fluctuations
identified inthesample.

11.8. Var and vecm models ofmultivariate time series intheanalysis ofshock scenarios
Vector autoregression models (VAR) whose widespread use inempirical macroeconomic research was initiated by Sims (cf. e.g. Sims
1980) constitute an easy toestimate statistical tool. At thesame time, years ofexperience confirm their suitability forthedescription
ofthe dynamic properties ofmacroeconomic time series and scenario analyses. VAR models can be built without imposing
restrictions arising from economic theory, which allows forthetesting ofhypotheses arising from this theory.

151

Let us consider an n-dimensional autoregressive process ofthe k-th order (VAR (k)):
xt = 1x x 1 + 2 x x 2 + + k x x k + Dt + et , t = 1,2,, T

(11.23)

where xk +1, xk +2 ,, x 0 are theinitial conditions, et is thevector white noise, i.e. E ( et ) = 0 , E ( et et ) = S ( nn is a positively
defined matrix) and E ( et es ) = 0 , when t s. n n - dimensional matrices Pi , i = 1,2, ..., k group parameters at subsequent process
delays. Vector Dt contains non-random variables, such asa constant, a deterministic trend, seasonal variables, other variables of0-1
type, while matrix F contains theparameters occurring at these variables.
A VAR process ofany order can be represented asVAR (1):
X + FD + E
Xt =
t 1
t
t

(11.24)

where

1 2
xt
Dt

et

In

0
xt 1
0

0 I


Xt =
FD
=
,
,
,
E
=

=
t
t
n

x
0

t k +1

0
0

k 1 k

0
0

0
0




In
0

In further considerations, we shall confine ourselves toprocesses that meet thefollowing assumptions:
lie inside theunit circle orare exactly equal toone.
Assumption 1.1. Eigenvalues ofmatrix P
Assumption 1.2. Thecomponents ofvector Dt are polynomials ofvariable t (of a finite order).
A VAR process is covariance stationary12 if and only if all thecharacteristic roots ofthe matrix lie inside theunit circle.
there are unit roots, i.e. whose modulus equals 1, then process { x } is non-stationary. Assumption
If among theroots ofmatrix P
t
(1.1) reduces further considerations tostationary processes orintegrated inthefirst degree { xt } : I(1) , i.e. such that those their first
differences are covariance stationary. These processes play an important role inthemodelling ofmacroeconomic time series,
since a number ofthe observed series have features that support their treatment asrealizations ofsuch processes. Using ordinary
regression tomodel time series that constitute therealizations ofnon-stationary processes may lead tofalse associations whose
symptoms may include a good model-data fit with a simultaneous high autocorrelation ofresiduals.
The search foran appropriate method ofmodelling non-stationary time series has led totheidea ofco-integration. Processes
are said tobe cointegrated if there exists their non-zero linear combination leading toa reduction inthedegree ofintegration.
In practice, stationary combinations are particularly important. Further we shall restrict ourselves tocase CI(1, 1), i.e. we shall seek
non-zero linear combinations ofcomponents ofthe vector process I(1)which are covariance stationary processes. TheGranger
representation theorem (cf. e.g. Engle and Granger 1987) states that any multivariate cointegration process can be represented
intheform ofan error correction mechanism, i.e. from therepresentation ofVAR (see equation 23) we move totheVEC model:
k 1

xt = xt 1 +

x
i

t i

+ Dt + et , t = 1, 2,, T

i =1

(11.25)

j , whereas
i In , i =
where xk +1 , xk +2 , ..., x 0 are theinitial conditions, { et } is a vector white noise, =
j =i +1
i =1
FDt denotes thedeterministic part ofthe model. While formulating thedeterministic part ofthe VEC model, one should carefully
investigate theimpact ofthe adopted assumptions on both theincrements and thelevels ofthe process inquestion (cf. e.g.
Johansen 1996, Juselius 2007). If theanalysed process { xt } is an integrated process, then matrix P array is an incomplete matrix
( r ( P ) = r < n ). For r = 0 process { xt } is integrated, but itis not cointegrated; for0 < r < the process is cointegrated, hence matrix
P can be (equivocally) represented intheform ofa full-column product oftwo matrices anr and bnr :

P =

(11.26)

The columns ofmatrix b contain thecointegration vectors that define thelong-term dynamic equilibrium. In a system which
remains inan equilibrium no internal factors appear, which would change thesituation, however, when asa result ofexternal
factors thesystem is thrown out ofbalance, its components, inan attempt torestore balance, react totheemerging error through
theadjustments parameters stored inmatrix a .
12

A stochastic process { X t : t Z} is said tobe covariance stationary if forany given finite integer t thefollowing conditions are met:

E ( xt ) = , < < +, E (( xt )( xt ) ) = 0 < + and E (( xt )( xt ) ) = h < +, h 1, 2,... .


152

One ofthe main advantages ofmacroeconomic models with error correction mechanisms is thus a clear separation ofshort-term
correlations (described by theparameters contained inmatrices Gi , i = 1, 2, ..., k 1) from thelong-term ones, usually supported
by economic theory and constituting themain focus ofthe analysis.
Parameters ofVAR/VEC models do not have a direct interpretation, but intherepresentations ofmoving averages we can
investigate theproperties ofthe modelled processes and their relationships, forexample by analysing theresponse torandom
noise function, which allows us totrace theimpact ofthe innovation shock on its components ina selected variable ofa given
multi-dimensional process.
Let us introduce a covariant stationary process VAR(k) with no deterministic components (for thesake offacility
i J ,
ofnotation) asa vector moving average model ofan infinite order, MA() : xt = et + 1et 1 + , where i = J
Jnnk = [In : 0 : : 0] . Element y jk ,i , i.e. element jk ofmatrix Y i represents thereaction ofthe j-th variable from
thesystem on one unit shock13 ofvariable k after i periods, assuming theabsence ofother disturbances (cf. e.g. Luetkepohl 2007).
When conducting a similar analysis forthecointegrated process:
xt = xt 1 +

k 1

x
i

t i

+ t , t = 1, 2,, T

i =1

first we should provide its VAR representation:


xt = 1xt 1 + 2 xt 2 + + k xt k + et , t = 1, 2,, T

where: 1 = + In + 1 , i = i i 1, i = 2,3,, k 1and k = k 1 . Next we determine therandom disorder response


functions according tothepattern shown above.
Now we shall move on totheissue ofparameter estimation forVAR and VECM models using themaximum likelihood method
(MLM). If we know thedistribution type ofthe analysed process, we can use this very method toestimate its parameters. Let us
suppose that process VAR(k) given by equation (11.23) has an nT-dimensional normal distribution, i.e. et : iiN n (0, S) and, forthesake
ofsimplicity ofnotation, let us rewrite process (11.23) asfollows:
xt = B Zt + et , t = 1, 2,, T

(11.27)

where:
] , Zt = xt 1 xt 2 xt k Dt , and theinitial conditions X 0 = x 0

established. We obtain thelogarithmised likelihood function14 (cf. e.g. Juselius 2007)


B = [1 2 k

n
1
1
lnL(B, ; X ) = T ln(2p ) T ln | |
2
2
2

( x B Z )
t

x
1 xk +1 have been

( xt B Zt )

t =1

(11.28)

which reaches a maximum at point

B=

( xt Zt )(

t =1

= 1

Z Z )

1
1
= MxZ M
t t
ZZ

t =1

t =1

1
B Zt )( xt 
B Zt ) =
( xt 
T

(11.29)

e e

t t

t =1

(11.30)

As far astheestimation ofthe VEC model is concerned, we should note that themain problems arising intheanalysis
ofcointegration, both on classical and Bayesian grounds, involve thenon-identifiability ofthe vector model parameters with
theerror correction mechanism. Matrix P , which is an incomplete-rank matrix, is decomposed into thecointegrating vectors
matrix b and theadjustment coefficients matrix a . Such decomposition is not unambiguous, because forany full-rank matrix
thefollowing relationship obtains: = = CC1 ; inother words, thedata contain information only about thecointegrating
space, and not thevectors that span it.
The use ofthe maximum likelihood method toestimate theVEC model parameters is based on theprocedures proposed in1951
by T.W. Anderson toanalyse incomplete regressions. This approach intheanalyses ofVEC models was proposed and developed
13
14

Shocks that equal one standard deviation are also considered, especially when thesystem variables are presented on different scales.
Ln(.) is an increasing function and, assuch, itdoes not alter theposition ofthe extremum ofthe likelihood function.

153

by S. Johansen, hence itis called theJohansen procedure. Let us represent theVEC model asequation (11.25) using thefollowing
conventionally adopted designations (cf. e.g. Johansen 1996, Juselius 2007):

Z 0t = Dxt

Z1t = xt -1

Z 2t = xt 1 xt 2 xt k +1 Dt

that is:
Z 0t = Z1t + Z 2t + t

(11.31)

where
= [1 2 k 1 ]

n
Assuming that et : iiN (0, S) , we obtain thefollowing logarithmised likelihood function:

n
1
1
ln L( , , , ; X ) = T ln(2 ) T ln | |
2
2
2

( Z

0t Z1t Z 2t ) ( Z 0t Z1t Z 2t )

t =1

(11.32)

Due tothecomplex form ofthe likelihood function given by equation (11.32), searching forits maximum insubsequent steps
involves building a concentrated likelihood function.
Function (32), given a fixed r, reaches its maximum at thefollowing point (cf. e.g. Johansen 1996):

b = S112 [v1 v 2 v r ]

(11.33)

= S ( S )-1

01
11

(11.34)


  M )M1
= ( M02
12 22

(11.35)

= S S b ( b S b )1b S

00
01
11
10

(11.36)

T
1
-1
Z it Z jt , Sij = Mij - Mi 2M22
M2 j fori, j = 0,1, whereas v1, v 2 ,, v r are eigenvectors ofmatrix
t =1
T
corresponding r first eigenvalues ofthe matrix arranged ina non-increasing order.

where Mij =

1
1
-1
S112 S10 S00
S01S112

Before commencing thedescription ofthe estimation ofVAR and VEC models using theBayesian approach, let us briefly review
theidea. Themain distinctive feature ofBayesian inference methods, which makes itdifferent from theclassical ones, is its
interpretation ofprobability. Classical statistics employs its frequency interpretation, whereas theBayesian approach makes use
ofits subjective interpretation, because probability is intended toexpress thedegree ofconviction asto thetruth ofa given
proposition (cf. e.g. Osiewalski (2001)). In theBayesian model, parameters () constitute random variables, and theBayesian
statistical model is unambiguously characterized by thecumulative density function ofthe probability distribution ofthe observed
vector, theprojected vector and theparameter vector:

p( x , x f , q ) = p( x f | x , q ) p( x | q ) p(q )

where x is thevector ofobserved values, x f thevector offuture realizations, theparameter vector, p( x f | x , q ) represents
thesampled predictive density, p(. | q ) thedensity function defined on theset ofpossible realizations ofthe phenomenon
inquestion, and p(q ) thedensity ofthe prior distribution that reflects our initial uncertainty asto thevalue ofparameter .
Inferencing is carried out inaccordance with theprinciples ofprobability theory, using, among others, theBayes theorem.
Theparameter vector is estimated on thebasis ofthe conditional distribution of given thefixed vector ofobservations x, i.e.
based on theso-called posterior distribution, whose density is set by thefollowing formula:
p(q | x ) =

p( x | q ) p(q )

p( x | q )p(q )dq

154

Let us rewrite theVAR model (11.27) inits matrix notation:


(37)

X = ZB + E

where, X = x1 x 2 xT , Z = z1 z 2 zT , E = e1 e2 eT and let us assume thefollowing prior distributions:

1. for S theinverse Wishart distribution with thematrix parameter S and q degrees offreedom: S : iW ( S , q ),

2. forB thematrix normal distribution conditional with respect to S , centred at B*: B | S : mN(B * , S, G ) , where G( nk +l )( nk +l ) is
a known positively defined matrix and l is thenumber ofnon-random variables.
Those distributions together with theassumption ofnormality applied totheanalysed process ( et : iiN n (0, S) ) lead tothefollowing
system offull conditional posterior distributions (which makes iteasier toobtain a pseudo-random sampling from theposterior
distribution):
1. theinverse Wishart distribution for S :
p( | B, X ) = iW ( S + (B B * ) G1(B B * ) + E E , q + nk + l + T )

2. thematrix normal distribution forB:


mBB , ,(G1 + Z Z )1)
p(B | , X ) = mN(


where

mB = (G1 + Z Z )1(G1B * + Z X )

In thesection devoted totheBayesian estimation ofVEC models, we shall present themodel proposed by Strachan and Inder (cf.
Strachan and Inder 2004) with theposterior distribution sampling design derived from Koop, Len-Gonzlez, Strachan (cf. Koop
et al. 2010). Theauthors ofthe model fortheestimation ofthe cointegrating space employ theclass ofMACG distributions with
thematrix parameter mm (matrix angular central Gaussian distributions, cf. e.g. Chikuse 2002). In order todescribe thespace
inquestion, Strachan and Inder (cf. Strachan and Inder 2004) use vectors that satisfy thecondition b b = Ir , where Ir is theidentity
matrix ofthe r-th level, i.e. theestimated matrix ofcointegrating vectors is given with an accuracy toa certain orthogonal matrix
Q ofthe r-th level, i.e. Q O(r ) .
Theorem 3.1. If therandom matrix Z has dimensions mr and a matrix normal distribution mNmr (0, Ir , ) , then its orientation
-

HZ = Z ( Z Z )

1
2 has an MACG distribution with parameter

W with thefollowing density function:

p(H Z ) =| |

r
m

2 | H H | 2
Z
Z

Any matrix Z nr ofrank r (almost everywhere) can be uniquely (almost everywhere) represented intheform ofthe product
Z

1
2
= H Z TZ ,

where H Z = Z ( Z Z )

1
2

is a matrix with orthonormal columns and TZ = Z Z is a symmetrical square matrix ofrank r


1

(Cadet 1996, Muirhead 1982). Matrix H Z is called theorientation ofmatrix Z, TZ2 theelliptical part (concentration) ofmatrix Z.
Spaces spanned by thecolumns ofmatrices Z and HZ are thesame sp( Z ) = sp(H Z ) (). For any symmetric positive definite real matrix
1

M we adopt thefollowing definition ofthe square root: M 2 = C LC and M

1
2

= C 1C , where L is a diagonal matrix which

contains on themain diagonal thesquare roots ofthe eigenvalues ofmatrix M, while C contains theorthonormal eigenvectors
of M (cf. e.g. Johansen 1996).
Using the MACG(W) distribution parameter, itis possible tointroduce into themodel theinitial information concerning thecointegrating space orabsence thereof (then W is an identity matrix ofthe m-th rank). In thedistribution ofinformation we assume

, where Hms , s r (H H = Ir ) is a matrix containing initial information on thecointegrating space,


that = Pt = HH + tHH

H^ thematrix spanning theorthogonal complement ofspace sp(H ) , and t a parameter representing theweight ascribed
tospace sp(H ) .

155

In order tocombine theabove-described distribution with a convenient and efficient sampling design from theposterior
distribution, Koop, Len-Gonzlez and Strachan (see Koop et al. 2010) propose tostart from parameterization = AB ,
1

where B R mr , A R nr and B = b (B B ) 2 , A = a(B B ) 2 , assuming forB and A thefollowing normal matrix distributions:
B | , r : mN(0, m1Ir , P ), A | , r : mN(0, Ir , ) . Starting from thedistribution forB, we thus obtain thefollowing MACG( Pt )
distribution for b :
For theremaining parameters ofthe model thefollowing prior distributions can be adopted:
1. for S theinverse Wishart distribution with thematrix parameter S and q degrees offreedom: S : iW ( S , q ) ,
2. for G thematrix normal distribution centred at zero conditional with respect to S : | , h : mN(0, , hIn( k 1)+l ) , where h>0 is
theparameter governing thedegree ofdispersion ofthe distribution, and I denotes thenumber ofnon-random variables,
3. fort,n and h , if we decide toestimate them inverse gamma distributions: gamma: t : iG( st , nt ) , n : iG( sn , nn ) , h : iG( sh , nh ) .
Thea priori distribution fort such parameters should be adopted that theentire probability mass is within the0 to1 interval, so
that more probability could be assigned tospace sp(H ) .
The adoption ofthe above-mentioned prior distributions leads tothefollowing system offull conditional posterior distributions
(cf. e.g. Wrblewska 2010):
1. theinverse Wishart distribution for S :

1
1
p( | A, B , , , , h, r , X ) = iW ( S + + AA + E E , q + n(k 1) + l + r + T )
h

2. thematrix normal distribution forA:


1
p( A | , B , , , , h, r , X ) = mN( A ,(B Z1 Z1B + Ir )1, )

where
1
A = ( Z 0 Z 2) Z1B(B Z1 Z1B + Ir )1


3. thenormal distribution forvector vec(B):

p(vec(B ) | , A, , , , h, r , X ) = N( vB , vB )

Where
vB = ([( A 1A) ( Z1 Z1)] + [ mIr P1 ])1, vB = vBvec( Z1 ( Z 0 Z 2)1A)


4. thematrix normal distribution for G :

1
p( | , A, B , , , h, r , X ) = mN( , ,( In( k 1)+l + Z 2 Z 2 )1)
h

where

1
= ( In( k 1)+l + Z 2 Z 2 )1Z 2 ( Z 0 Z1 )
h

5. theinverse gamma distribution for n and h:



1
nr
p( | , A, B , , , h, r , X ) = iG( s + tr (1AA ), n + )
2
2
1
n( n(k 1) + l )
p( h | , A, B , , , , r , X ) = iG( sh + tr (1 ), nh +
)
2
2

6. fort:

p( t | B , r , X ) | Pt |

1
m

r t nt 1 exp( 1 ( s
tau + tr (BHHB )))

t
2


Pseudorandom sampling from theposterior distribution can be obtained using a parameter-augmented Gibbs sampling
alogorithm (cf. Koop et al. 2010, Wrblewska 2010):
156

(0)
(0)
(0)
(0) (0)
(0)
(0)
1. Setting an arbitrary vector ofinitial values: ( , , , , , h , , ) .
(s)
(s)
(s)
(s) (s)
(s)
(s)
2. Random selection ofa single realization ( , A , B , , , h , , ) consists ofthe following steps:
(s)
(a) Sampling S from thedistribution

p( | A = A( s1) , B = B( s1) , = ( s1) , = ( s1) , h = h( s1) , = ( s1) , r , X )

(s)

(b) Sampling A

(c) Sampling B

from thedistribution
p( A | = ( s ) , B = B( s1) , = ( s1) , = ( s1) , h = h( s1) , = ( s1) , r , X )


(s)

from thedistribution
p(B | = ( s ) , A = A( s ) , = ( s1) , = ( s1) , h = h( s1) , = ( s1) , r , X )

(d) Calculation of b and a assuming that = B(B B )


(e) Sampling G( s ) from thedistribution

1
1
2 , = (B B ) 2 ( = = AB ),

p( ( s ) | = ( s ) , = ( s ) , = ( s ) , = ( s1) , h = h( s1) , = ( s1) , r , X )


(s)

(s)

(s)

( s1)

( s1)

, =
, r, X )
(f ) Sampling n ( s ) from thedistribution p( | = , = , = , h = h
(s)
(s)
(s)
(s)
( s1)
(s)
p
(
h
|

,
r
,
X
)
(g) Sampling h from thedistribution
(h) For thesampling ofparameter t from theposterior distribution we can use e.g. theMetropolis-Hastings algorithm with
thegeneration ofinitial values from thefollowing distribution:

m
iG st + tr (BH H B ) , nt

11.9. The application ofBayesian var and vecm models inforecasting


VEC models used intheanalysis ofcointegration may differ with respect tothenumber ofdelays inshort-term dynamics (k),
thenumber ofcointegrating relationships (r), thekind oftrends (d), additional restrictions imposed on thecointegrating space
(o,o 0,1, ..., , where o = 0 represents no restriction), a form ofexogeneity ofselected variables (e). A detailed description ofthe
most commonly considered linear restrictions imposed on thecointegrating space, on thespace ofthe adjustment factors and
considerations concerning thetrends can be found e.g. inJohansen (1996). At this point itis worth noting that using theBayesian
VEC model these restrictions can be expressed by conditions applied directly totheco-integrating space.
Let x=(k, r, d, o, e) X be a vector describing thecharacteristics under consideration and Mx a model with these characteristics.
Before determining theposterior probabilities, themodels examined must be assigned prior probabilities, so that e.g. all
theconsidered models have equal prior probabilities orthat themore economically parameterized models are more likely.
Knowing prior probabilities, we can, using theformula below, determine theposterior probability:
p( Mx ) =

p( Mx )p( X | Mx )

p(M )p( X | M )
x

, x

(11.38)

where p( X | Mx ) is themarginal density ofthe vector ofobservations inmodel Mx . An analytical determination ofthe marginal
density ofthe vector ofobservations is usually very difficult oreven impossible, hence numerical methods orapproximations are
applied tothat end.
Using theBayesian method ofcombining knowledge, theposterior probabilities determined can be used toperform a further
analysis ofthe studied phenomenon, accordingly, theconclusions drawn will not be based on a single selected model, but will
reflect theuncertainty associated with its specification (cf. e.g. Osiewalski 2001).
In this section, theBayesian idea ofcombining knowledge shall be presented inthecontext offorecasting. To produce theforecast
ofa cointegrated process, we move from theVEC representation totheVAR representation according tothepattern presented
insubsection 2.3.
Bayesian prediction is based on predictive distribution, i.e. on theconditional distribution ofthe future values ofthe analysed
process with respect toobservations. Let us denote thevector ofexpected values by X = ( xT +1, xT +2 , ..., xT +h ) , where h is
theforecast horizon. Thepredictive distribution relative totheobserved values ofXis obtained by averaging (on theparameter

157

space) theso-called predictive sampling density p( X | q , X ) using thedensity ofthe posterior distribution density fortheparameters
ofthe model asa weight function (Osiewalski 2001, Zellner 1971):
p( X | X ) =

p( X , q | X )dq =

p( X | q , X )p(q | X )dq

(11.39)

In many cases itis impossible tocalculate analytically theintegral set by formula (11.39), therefore, inorder toobtain a sample
from thepredictive distribution numerical methods are used. Thepredictive sampling density can be represented asthefollowing
product ofdensities:

p( X | q , X ) = p( xT +1 | q , X ) p( xT +2 | q , X , xt +1)... p( xt +h | q , X , xt +1, xt +2 ,..., xT +h1)

In other words, inorder toobtain a sample from thepredictive distribution, after sampling theparameter vector from theposterior
distribution, we sample xT +1 according tothedistribution p( xT +1 | q , X ), and then xT +2 from p( xT +2 | q , X , xt +1). Such a sampling
design is continued up totheadopted forecast horizon. In theBayesian VAR/VEC models discussed, theconditional distribution
p( xT +1 | q , X , xt +1, xt +2 ,..., xT +i 1) is normal, so sampling is relatively easy.
The predictive distribution takes into account theuncertainty associated with both thepredicted values and
with thevalues ofthe
parameters. In addition, using theBayesian technique ofcombining knowledge, determining theforecast may be based not on
a single Bayesian model, but on a set ofsuch models:
p( X | X ) =

p( M | X )p( X | X , M ) = p( M | X )

p( X | , X )p( | X )d

(11.40)

As a result ofcombining knowledge, we obtain a final predictive distribution, which is a weighted average ofthe predictive
distributions inthemodels under consideration with respective weights corresponding totheposterior probabilities ofthese
models.

ukasz Postek

Chapter 12
THE STARTING MECHANISM
12.1. Introduction
The starting mechanism ofthe Rapid Reaction Instrument comprises information about thestate ofthe economy and its individual
sectors. Theinformation derives from thesystems research (monitoring) component based on which political decisions are made
tostart theimplementation activities inspecific enterprises eligible forsupport under thesystem.
The purpose ofthe starting mechanism is toprovide policy-makers with information about theassumptions trigger implementation
activities inindividual enterprises, industries and groups ofbusinesses that should be supported.
The source information on thesubject inquestion should come from reports produced within theRRIs micro- and macroeconomic research component, while thedecision-making procedure should be consistent with other activities undertaken within
theinstrument.

12.2. Thegeneral concept ofthe starting mechanism


It was assumed that thebasic decision-making procedure should comprise a three-level scoring system formicro-, meso- and
macro-level variables. It means that theRRI starting mechanism will be based on point evaluations ofinformation on thestate
ofthe economy and its individual sectors originating from theresearch component ofthe RRI converted into a probability score.
In order tomaximize theamount ofinformation analysed inthereports ofthe RRI research component two complementary
systems are proposed:

The basic scoring system,

The auxiliary information system.
The basic scoring system shall provide thedirect basis fora decision tostart implementation activities inenterprises, and thereby
meet thepostulated requirement ofmethodological consistency incombination with aggregation and disaggregation capacities
among all levels (micro-, meso- and macro). Thesystem would thus indicate thespecific groups ofenterprises, with regard towhich
RRI interventions should commence.
Due tothespecific features ofstatistics compiled by different sources some data are not calculated oravailable at all three levels,
orare expressed from a certain point ofview theutilization ofsuch input inthebasic scoring system is not possible orwould not
make economic sense. After all, thedata often contain valuable information about thestate ofthe economy orofcertain aspects

Fig. 12.1. A schematic comparison ofthe basic scoring system and theauxiliary information system
Source: own study.

159

ofits operation and, therefore, their complete exclusion would not be recommended. Hence theproposal toinclude an auxiliary
information system, which would provide additional input into thebasic scoring system and offer a broader background
inthedecision-making process. Thesystem could suggest areas on which aspects ofenterprise operation RRI support should be
focused (information concerning theobject ofintervention inquestion). Such an auxiliary information system should be open
and flexible, which would make iteasier, if necessary, toincorporate new information without having tomodify theentire scoring
system. Simultaneously, data within theauxiliary information system would be presented inthesame way asin thebasic scoring
scheme.

12.3. The basic scoring system


Below we shall describe thegeneral algorithm forthebasic scoring. With a view topreserving scoring consistency at all three
levels, we propose that thebasic algorithm forcalculating thepoint values within thebasic scoring system should be sequential
and run from themicro-, through themeso- tothemacroeconomic level. Accordingly, thesequential scoring algorithm would
run asfollows:
1. Calculation ofa point value at themicroeconomic level MicroScore j ,t (0,1) based on a set ofsource information,
2. Thecalculation ofa point value at themesoeconomic level MesoScorei ,t (0,1) based on theaggregation ofresults from
themicroeconomic level,
3. Thecalculation ofa point value at themacroeconomic level MacroScoret (0,1) based on theaggregation ofresults from
themesoeconomic level.
(i and j cross-sectional indices, t thetime index).

Basic scoring system

Aggregation formula

MacroScore

3. Macroeconomic level

2. Mesoeconomic level

MesoScore i,t

1. Microekonomic level

Macro
WeightiMeso
MesoScorei , t
,t

i =1

Weight i,tMeso Macro

i =1

Weightj,tMicro Mesoi MicroScorej,t

j =1

Weightj,tMicro Mesoi

j =1

MicroScorej,t

Full set of source information

Fig. 12.2. A general ofalgorithm ofthe basic scoring system


Source: own study.

Since individual data from studies F-01/I-01 are protected by statistical secrecy, research inthemicro component is based on
data aggregated into classes defined by theintersection ofthe cross-sections by NACE divisions and thenumber ofemployees
(small enterprises: 1049 employees, medium-sized enterprises: 50249 employees, large enterprises: 250 employees ormore).
Itshould be noted that statistical confidentiality prevents access toofinformation if thedata list consists offewer than three
entities ortheshare ofone ofthem intheaggregate is greater than three-quarters. Consequently, theavailable data do not include
certain classes ofenterprises, and are incomplete with respect tocertain others categories. An additional complication is posed
by thefact that thedata on small enterprises are updated on a semi-annual basis, whereas thedata on medium-sized and large
enterprises on a quarterly basis.
Taking into account theavailability ofdata, itis recommended that problematic classes ofenterprises be aggregated by NACE
section with respect totheir size interms ofemployee numbers (small, medium-sized, large) so asto avert theproblem ofthe lack
ofdata. Due tothedifferent rules governing theupdating ofinformation on small enterprises (semi-annual) vs. medium-sized and
large enterprises (quarterly) itis not advised toaggregate theclasses ofdifferent size within theNACE sections.

160

Table 12.1 shows thesuggested breakdown, inwhich theproblematic NACE sections have been combined with those ofthe most
similar profile ofactivity or if such are missing included intheresidual category. As a result, theproposed breakdown, comprise
99 individual classes ofenterprises which will be scored at themicroeconomic level:
MicroScore sc
j , t forthej-th class ofenterprises under NACE section (j {1,2,..., 33}), size class SC {S, M, L}, inperiod t
Table 12.1. Proposed classification ofenterprises at themicroeconomic scoring level (xxx absence ofdata fortheentire period
from 2007, x absence ofdata forselected periods)
NACE
class

Small
(semi-annually)

Medium-sized
(quarterly)

Large
(quarterly)

6. Extraction ofcrude petroleum and natural gas

xxx

xxx

xxx

7. Mining ofmetal ores

xxx

xxx

xxx

xxx

xxx

NACE division
Industry

Section B Mining and quarrying


1

5. Mining ofcoal and lignite

8. Other mining and quarrying


9. Mining support service activities

Section C Manufacturing
2

10. Manufacture offood products

11. Manufacture ofbeverages

13. Manufacture oftextiles


14. Manufacture ofwearing apparel

12. Manufacture oftobacco products

xxx

19. Manufacture ofcoke and refined...

21. Manufacture ofpharmaceutical, medicinal...

15. Manufacture ofleather and related products


5

16. Manufacture ofwood and ofproducts ofwood and cork,


except furniture...
31. Manufacture offurniture

17. Manufacture ofpaper and paper products

18. Printing and reproduction ofrecorded media

20. Manufacture ofchemicals and chemical products

22. Manufacture ofrubber and plastic products

10

23. Manufacture ofother non-metallic mineral products

11

24. Manufacture ofbasic metals

12

25. Manufacture offabricated metal products...

13

26. Manufacture ofcomputer, electronic and...

14

27. Manufacture ofelectrical equipment

15

28. Manufacture ofmachinery and equipment n.e.c.

16

29. Manufacture ofmotor vehicles, trailers and semi-trailers

17

30. Manufacture ofother transport equipment

18

32. Other manufacturing

19

33. Repair and installation ofmachinery and equipment

20

35. Production and supply ofelectricity, gas...

Section D Production and distribution ofelectricity, gas...

Section E Water supply, sewerage, waste management and remediation activities


21

36. Water collection, purification and supply

37. Sewerage
38. Waste collection, treatment and disposal...
39. Remediation activities and other...

161

xxx

NACE
class

Small
(semi-annually)

NACE division

Medium-sized
(quarterly)

Large
(quarterly)

Trade
Section G Wholesale and retail trade, repair ofmotor vehicles and motorcycles
22

45. Wholesale and retail trade ofmotor vehicles...

xxx

xxx

xxx

46. Wholesale trade, except ofmotor vehicles...


47. Retail trade, except ofmotor vehicles...
Services
Section F Construction
23

41. Construction ofbuildings

24

42. Civil engineering...

25

43. Specialized construction activities


Section H Transportation and storage

26

49. Land transport and transport via pipelines


50. Water transport

51. Air transport


52. Warehousing and support activities fortransportation

xxx

Section I accommodation and food service activities


27

55. Accommodation

xxx

56. Food and beverage service activities


Section J Information and communication
28

58. Publishing
59. Motion picture, video and TV programme production,...

29

60. Programming and broadcasting activities

30

61. Telecommunications

31

62. Computer programming, consultancy and related activities


63. Information service activities

Section K Financial and insurance activities


Section L Activities related toreal estate
32

64. Financial service activities, except...


65. Insurance, reinsurance and pension funding, except...

xxx

66. Activities auxiliary tofinancial services...

xxx

xxx

68. Real estate activities


Section N Activity intheservices administration and support service activities
Section S Other service activities
33

77. Rental and leasing activities


78. Employment activities

79. Travel agency, tour operator,...

80. Security and investigation activities

81. Services tobuildings and landscape...

82. Office administrative, office support and...


94. Activities ofmembership organizations

xxx

xxx

95. Repair ofcomputers and personal and

96. Other personal service activities

Source: own calculations based on information from theRRI microeconomic research component reports.

162

In order tocalculate thevalue ofthe MicroScore, we suggest that thepoint value at themicroeconomic level should be based on
thedegrees ofbankruptcy risk estimated and forecast within theRRI microeconomic research component. For thepurposes ofthe
basic scoring system, these values should

be converted toclasses defined inTable 10.


Since thebankruptcy probabilities ofdifferent classes ofenterprises estimated inthemicroeconomic component are not directly
comparable with each other innormative terms, a three-stage procedure should be applied whose purpose is torelativize thedata
both cross-sectionally and along thetemporal dimension:
1. Removing seasonal variations from thetime series ofestimated and predicted bankruptcy probabilities (suggested method:
TRAMO/SEATS, cf. Gomez and Maraval 1996).
2. Filtering theadjusted time series inorder toremove thelong-term trends and high-frequency fluctuations (suggested
method: theChristiano-Fitzgerald filter (cf. Christiano and Fitzgerald 2003), with thefluctuation period from 1.5 to8 years,
assuming that filtered series are trend stationary).
3. Estimation ofthe random distribution ofthe cyclic component ofthe time series and calculation thevalue ofthe cumulative
distribution function foreach observation inthesample (suggested method: assuming that thecyclical component ofthe
time series has a normal distribution with parameters estimated on thebasis ofthe sample).
As a result ofthe proposed procedure, 99 (33 x 3) time series ofpoint values MicroScore sc
j , t will be calculated, which will be
subsequently aggregated tothemesoeconomic level, and later tothemacroeconomic one.
In order tocalculate thevalue ofthe MesoScore, theinitial division into 3 classes ofenterprises by number ofemployees should
be preserved at themesoeconomic level and an aggregation into 3 classes designated under NACE sections should be applied
(industry I, trade T, services S). Commonly, data are presented insuch a way intheRRI microeconomic component.
According totheproposed sequential algorithm, scoring at themesoeconomic level is based on theinformation calculated at
themicroeconomic level and on a weighting system. Two scoring paths should be created comprising two weighting schemes
based on thefollowing:

Gross value added (GVA),

Number ofemployees (NE).
The use ofthe proposed weighting schemes based on thegross value added and on thenumber ofemployees is justified
since both parameters constitute fundamental business categories ofconcern toeconomic policymakers. On theone hand,
microeconomic information reflects theestimated bankruptcy probability ofenterprises, on theother hand, theweighting scheme
contains information about theeconomic importance ofthose enterprises, inother words, thepotential cost interms ofthe lost
gross value added and thenumber ofjobs generated by these enterprises, forwhich thebankruptcy probability is measured.
Thegross value added represents an insight through thelens ofthe final good market, while thenumber ofemployees represents
an insight through theprism ofthe labour market(s). Since these two approaches do not always coincide, both should be analysed
simultaneously, which results intwo parallel scoring paths:
SC
MesoScore _ GVATA
,t a scoring path based on thegross value-added forenterprises whose size class SC {S, M, L} and which are
engaged intype ofactivity TA {I, T, S }.
SC
MesoScore _ NETA,t
a scoring path based on thenumber ofemployees inenterprises whose size class SC {S, M, L} and which
are engaged intype ofactivity TA {I, T, S}.
Adopting theprevious representations, thepoint values at thefor mesoeconomic level fora weighting scheme based on gross
value added can be calculated asfollows (a similar aggregation can be performed fora weighting scheme based on thenumber
ofemployees by replacing GVA with NE intheformulae):

MesoScore_GVAISC
,t

21

21

33

(industry)

GVA j,t

j =1

SC
MesoScore _ GVATSC,t MicroScore22
,t

MesoScore _ GVASSC,t

GVA j ,t MicroScore SC
j ,t

j =1


(trade)

GVA j ,t MicroScore SC
j ,t

j =23

33

GVA j ,t

j =23

(services)

The formulae presented above demonstrate that aggregation applies only toindustrial and service enterprises, since theshare
ofcommercial enterprises is relatively small.

163

Another problem is thecalculation ofthe MacroScore. Since small businesses are scored semi-annually, whereas information about
medium-sized and large enterprises is updated on a quarterly basis, a distinction should be introduced between macroeconomic
scoring paths with a quarterly periodization (excluding small enterprises) and a semi-annual one (including small enterprises).
The same indications asbefore should be introduced:
MacroScore _{ path}1tQ a macroeconomic scoring path based on thegross value added ({path} = GVA) orthenumber ofemployees
({path} = NE) updated on a quarterly basis (excluding small enterprises)
MacroScore _{ path}t2Q a macroeconomic scoring path based on thegross value added ({path} = GVA) orthenumber
ofemployees ({path}= NE) updated on a semi-annual basis (including small enterprises).
Adopting theprevious representations, thepoint values at thefor macroeconomic level fora weighting scheme based on gross
value added can be calculated asfollows (a similar aggregation can be performed fora weighting scheme based on thenumber
ofemployees by replacing GVA with NE intheformulae):

MacroScore _ GVAt1Q

MacroScore _ GVAt2Q

SC{M, L}

SC { S , M ,L }

SC
SC
(GVATA
,t MesoScore _ GVATA,t )

TA

M
L
(GVATA
,,t + GVATA,t )

TA

SC
SC
(GVATA
,t MesoScore _ GVATA,t )

SC
S
(GVATA
,t
TA

M
L
+ GVATA
,t + GVATA,t )

Now we shall discuss thepremises forthedeployment ofimplementation activities. Theformulae presented intheprevious
subsections allow forscoring at themicro-, meso- and macroeconomic levels. Since thesystem is based on theenterprise
bankruptcy probability, thenumerical values ofscores belong tothe<0,1> interval. Thesmaller thevalue adopted by thescoring,
thebetter thecondition ofindividual enterprises inquestion and thesmaller theneed todeploy implementation activities.
The threshold value that suggests theneed todeploy implementation activities should be determined with respect totheaverage
percentage oftime inwhich thePolish economy was inrecession, according tohistorical data. As defined by theNational Bureau
ofEconomic Research (NBER)1 applied totheOECD based Recession Indicators forPoland,2 thePolish economy was ina recession
during 29.6% (approx. 30%) ofthe time between 2003m1 and 2013m6. In view ofthis fact, thedeployment ofimplementation
activities is advisable if thecurrent orforecast scoring forthenext half ofthe year exceeds 70%.
In terms ofheuristics, we suggest that theanalysis ofthe need todeploy implementation activities should be based on estimated
and projected scoring values forclasses defined at themicroeconomic level, then aggregates at themesoeconomic level should
be analysed, and finally aggregates ofall theenterprises.
Scoring paths with a quarterly and semi-annual periodization are available forall theanalysed levels. We suggest that any decision
toinitiate implementation activities ina given category ofenterprises should be taken when current pertinent data are available.
In thecase ofsmall enterprises, thedecision-making process would involve a semi-annual periodization, whereas inthecase
ofmedium-sized and large enterprises a quarterly one.
At themeso- and macroeconomic levels there are two scoring paths based on thegross value added and thenumber
ofemployees. We suggest that intheir case thescore (its current orsemi-annual projection) exceeding 70% inat least one ofthe
scoring paths should be interpreted asa signal justifying intervention.
Given theabove, thefollowing algorithm ofaction is proposed:
1. Implementation activities should start inenterprises which belong toindividual classes defined at themicroeconomic level,
whose scores have exceeded a certain threshold.

In theevent that itis necessary toset priorities, higher priority should be assigned toenterprises which belong toindividual
classes defined at themicroeconomic level with thehighest scores and those with scores over and above a certain threshold
also at themesoeconomic level.
2. If thepossibilities forlaunching implementation activities have not been exhausted, such activities should start inenterprises
that belong toindividual classes defined at themesoeconomic level, forwhich thescores have exceeded a certain threshold
(even if at themicroeconomic level thescores have not been exceeded).
3. If thepossibilities forlaunching implementation activities have not been exhausted and thescores exceed thethreshold
values at themacroeconomic level, implementation activities should start inall enterprises eligible fortheRRI.
1

Here recession is understood ineconomic terms asa phase ofthe business cycle, not inits technical sense (i.e. at least two consecutive quarters
ofnegative GDP growth).
[http://research.stlouisfed.org/fred2/series/POLREC?rid=242&soid=4].

164

12.4. Using auxiliary information from therri research component


As was indicated at thebeginning ofthe study, theauxiliary information system should fulfil a complementary role inthedecisionmaking process concerning theneed tostart implementation activities inenterprises. While thebasic scoring system indicates
theneed orits absence forimplementation activities inspecific groups ofenterprises, itis theauxiliary information system
which is supposed toprovide additional information, a broader decision-making background and ultimately signpost theareas
ofbusiness activity and theeconomy asa whole on which special attention should be focused. In particular, theauxiliary
information system could provide thebasis fortheintroduction ofpossible expert modifications and adjustments tothebasic
scoring system.
The authors propose that themethod ofpresentation ofthe auxiliary information should correspond tothebasic scoring system
(i.e. comprise thethree-stage procedure 3 that involves removing seasonal variables, filtering thetime series and calculating
thecumulative distribution function foreach observation inthesample) with theprovision that insome cases itis impossible
topresent a given variable at all three requisite levels: micro-, meso- and macroeconomic.
Since theauxiliary information system does not indicate whether implementation activities should start ornot, setting
theintervention threshold values is purposeless. Thethreshold of70% (or 30%) may, however, provide a certain nominal limit
indicative ofthe severity ofthe problems ina given area.
Table 12.2 lists thevariables analysed within theRRI microeconomic research component, which may be included intheauxiliary
information system at all three levels micro-, meso- and macroeconomic ones inthecorresponding cross-sections and
inthebasic scoring system. We have highlighted especially important variables due totheir statistically significant role inpredicting
enterprise bankruptcy along with theinformation on thephenomenon ofenterprise bankruptcy and theformation ofnew
enterprises, creation ofnew jobs and gross value added. Additionally, only at themacroeconomic level (i.e. from theperspective
ofthe whole corporate sector) theauxiliary information system may include negotiated bankruptcy and liquidation bankruptcy
rates, respectively, which are presented inthereports submitted by theRRI microeconomic research component.
Table 12.2 also shows thevariables analysed by theRRI macroeconomic research component, which may be included inancillary
information system, however, thenature ofthese data makes their aggregation purposeless.
Table 12.2. Theproposed array ofvariables tobe included intheauxiliary information system (the most important variables from
theRRI microeconomic research component which should be given special attention are marked with x)
Information from theRRI microeconomic research component
Enterprises
Industry

Trade

Services

Overall financial condition indicator (OFCI)

Debt sustainability indicator (DSI)

Current ratio (CR)

Quick ratio (QR)

Cash conversion cycle times (CCCT)

Cash conversion cycle days (CCCD)

Asset productivity (AP)

Operating return on sales (OROS)

Operating margin assets (ROOA)

10

Return on assets (ROA)

11

Return on equity (ROE)

12

Total debt ratio (TDR)

13

Self-financing rate

14

Short-term debt rate (STDR)

15

Number ofenterprises

16

Number ofemployees

17

Gross value added

18

Sales revenue (net)

19

Capital expenditure

x
x
x

x
x

Variables which are not expressed inpercentage terms (e.g. thenumber ofemployees, gross value added) should be logarithmised.

165

Information from theRRI microeconomic research component


Enterprises
Industry
20

Total assets

21

Short-term liabilities

22

Receivables and claims

23

Net profit

Industrial production index intheanalysed NACE divisions and sections

Construction production index intheanalysed NACE divisions and sections

Retail sales index intheanalysed NACE divisions and sections

General business climate index intheanalysed NACE divisions and sections

Unemployment rate

Gross domestic product divided into constituent sources:*

domestic demand

consumption

private consumption

10

public consumption

11

accumulation

12

gross fixed capital formation

13

changes ininventories

14

net exports

15

export

16

import

Trade

Services

Information from theRRI macroeconomic research component

Since thechanges ininventories and net exports may assume negative values, thecontributions ofindividual categories toGDP growth should
be analysed rather than their levels orgrowth rates.

Source: own study.

Piotr Adam Boguszewski, Monika Dawid-Sawicka, Magorzata Kazubska

Chapter 13
COMMUNICATION IN THE RAPID RESPONSE SYSTEM
13.1. Introduction
Communicating theresults ofmonitoring economic change will always pose a challenge. Thekey factor inthis respect
is theresponsibility not only forwhat is actually said, but also forwhat remains unsaid orinsufficiently clarified. Of strategic
importance is also thecapacity toanticipate thepossible cuts and simplifications made by themedia orthebusiness environment
organizations inan effort toextract theessence ofthe message, which may significantly affect its gist. Thestakes inmonitoring
economic change are high, because paraphrases oromissions intheprocess ofcommunication orpublication ofmaterials may
lead toan incorrect assessment ofthe potential ofindividual NACE sectors ortoa grave misinterpretation ofeconomic forecasts.
The subject matter intended forcommunication aspart ofthe Rapid Response Instrument is, by its very nature, complicated and
lengthy. TheRRI comprises a broad range ofmethods and procedures forassessing thecondition ofthe economy asa whole
and groups ofenterprises, programming and implementation ofwell-defined public interventions targeted toenterprises, their
groups and sectors ofthe economy affected by theeconomic downturn. Theinterpretation offorecasts presented inRRI Reports
is associated with understanding theidea behind itand an appreciation ofits fundamental methodological assumptions.
At theconceptual stage itwas assumed that theRapid Response Instrument should consist oftwo subsystems themonitoring
component (or theEarly Warning System) and theprogramming and implementation component (or Rapid Response System).
Thevast majority ofcommunication activities carried out inaccordance with theguidelines prepared by thePolish Agency
forEnterprise Development (PAED, Polish acronym PARP) were assigned totheEarly Warning System component, however,
informing about theresults ofthe Early Warning System monitoring component served toachieve theobjectives ofthe Rapid
Response System. Thecommunication objectives followed theobjectives that guided theexamination ofkey macroeconomic
trends and other external determinants ofcorporate bankruptcy processes, which accounted forthecontents ofreports devoted
tomacroeconomic issues aswell astheidentification ofeconomic phenomena at thesectoral level contained intheReport on
theassessment ofbankruptcy risk, which constitutes a microeconomic component.
First, thekey aspect ofcommunication activities involved theway ofpresenting information on thesymptoms and theworsening
situation ofindividual sections and NACE divisions, inparticular thedirect threat oflosing liquidity followed by thesymptoms
ofimpending bankruptcy risk aswell ascommunicating thebankruptcy risk toenterprises and NACE sections described
intheReport. Communication activities ofthe macroeconomic part ofRRI Reports focused on communicating thebusiness
cycles ofthe Polish economy, forecasts ofthe basic macroeconomic indicators and scenario analyses ofthe impacts ofpotential
impulses affecting thePolish economy.
The implementation ofthis challenging and wide-ranging objective was done according tothesaying monitoring and prevention
are better than cure, which became themain theme ofthe entire communication strategy pursued by thePolish Agency
forEnterprise Development from March 2013 toMarch 2014.

13.2. Target groups


Monitoring Economic Change aspart ofthe RRI project is a component ofthe Early Warning System which will allow decisionmakers totake pre-emptive action, remedial action, but above all, toreduce theeconomic and social costs ofperiodic disturbances
inthefunctioning ofthe economy. Therefore, one ofthe target groups includes government representatives, whose responsibilities
include theimplementation ofpublic interventions that address cyclical economic risks. Some ofthese tasks at theregional level
are faced by local governments; hence they have also become one ofthe target groups. Communication activities also included
representatives ofregional development agencies, labour market institutions, employers organizations, trade unions, industry
associations and analysts from financial institutions. Another important group are theentrepreneurs themselves, especially those
responsible formanaging their firms, their financial policies and decision-makers inthearea ofhuman resources. Communication
addressed tothem was, however, different innature and focused on theprovision ofinformation about thecondition ofthe
economy, thesituation oftheir industry orsector ofthe economy inorder toenable them totake appropriate action within their
own companies based on proven and reliable data.

167

13.3. The challenges what are we up against when communicating theresults ofmonitoring
economic change?
A number ofcommunication challenges, which arose during theimplementation ofthe RRI project, were similar tothose related
toother research projects ofkey importance forthePolish economy and worth considering when preparing communication
strategies forvarious projects pursued by government agencies. However, there were also unique challenges.
The first challenge involved compliance with thegovernment documents. Theaims ofthe project Rapid Response Instrument
include providing decision-makers with theinformation which they need inorder totake pre-emptive and remedial action inorder
toreduce theeconomic and social costs ofcyclical disturbances intheeconomy. Therefore, all communication efforts had totake
into consideration theactivities anticipated by theCouncil ofMinisters (mainly intheLong-term National Development Strategy and
theMedium-Term National Development Strategy, which provided forthedevelopment ofstate capacity topre-emptively respond
tophenomena and processes occurring inthesocio-economic system). Communication activities also needed tobe consistent
with theobjectives ofthe New Opportunity Policy a document prepared by theMinistry ofEconomy.
The threat ofbankruptcy is also a topic ofincreasing concern totheEuropean Commission. It seeks toensure efficient procedures
leading toconvergence among thenational bankruptcy regulations, which would protect theinterests ofboth debtors and
creditors. Consequently, intheprocess ofcommunication activities itwas important totake note ofand comply with themeasures
implemented inthis area by theEuropean Union.
Another challenge was theperception ofbankruptcy. In theprocess ofcommunication, thetopic ofbankruptcy itself posed
a challenge on theone hand, due toits ardent exploitation by themedia, on theother hand, by creating a barrier indealing
with entrepreneurs. Even firms which find themselves ina serious financial distress tend toavoid thesubject ofbankruptcy. Their
representatives were clearly reluctant tobe associated with support/restructuring tools orproblems that affect businesses during
theeconomic downturn. This was also noticeable when theTOP 10 ofthe most vulnerable NACE divisions (i.e. characterized by
a serious bankruptcy risk) was compiled and communicated. Thequarterly updates ofthe TOP 10 list made by theRRI team also
take into account thegrowth rate ofbankruptcy risk adjusted forthenumber ofemployees ina given NACE division. Thepurpose
ofcommunication inthis case was not only towarn ofa high orincreasing degree ofbankruptcy risk, but also todraw attention
tothepossible consequences that may result from therelatively high share ofthese sectors intheoverall employment structure.
It was important totalk about thethreat ofbankruptcy inthecontext ofthe warning system and tostrengthen thebelief that
what is happening intheeconomy aswhole is naturally reflected inthesituation on thelabour market. Thechallenge consisted
inpresenting thetopic insuch a way asto avoid associations merely with theeconomic issues while shifting thefocus onto
labour and employment issues. Accordingly, communication regarding theRRI project also took place during events dedicated
tothelabour market with theparticipation ofpolicy-makers involved inthearea ofhuman resources.
The next challenge was posed by thelanguage used by theauthors ofRRI reports, which did not facilitate their reception by
theaverage interested party. Hence thematerials produced aspart ofthe communication activities had topresent inan accessible
way thecontent that had originally been described using scientific language. It should be noted that one ofthe target groups
ofthe project comprises Polish entrepreneurs representing theSME sector and, therefore, theissues at hand had tobe presented
ina meaningful way inorder toexplain thephenomena that may have a direct impact on their daily activities. Moreover, the
language used inthepronouncements had tobe highly disciplined interms ofphrases and expressions used inorder toavoid
simplification orover-interpretation ofthe forecasts.
Another important challenge is also theissue ofcompetition. RRI reports face serious competition intheform ofwell-known
statistics compiled by public institutions, such asthePolish National Bank (NBP) ortheCentral Statistical Office (GUS) aswell
astheperiodically updated and widely commented inthemedia reports by such companies asCoface Poland, Euler Hermes,
IBnGR, which prepare forecasts ofmicro-and macro-economic indicators and bankruptcies inindividual sectors ofindustry.
Thedifficulty was compounded by thefact that RRI reports are based on existing data and are published with a delay with respect
tothedata on bankruptcy published by private companies involved inthemonitoring ofthe economy. Thesolution was toprovide
an accurate plan ofcommunication activities including thepublication dates ofcompeting research results and take advantage
ofthe synergy effect interms ofcommunication on individual topics.
The final challenge involved theambiguity ofthe acronym ofRapid Response Instrument inPolish (Instrument Szybkiego
Regowania, ISs). Thesearch engine most popular among Polish internet users returns a number ofother Polish and international
institutions unrelated toPARP activities, such astheInstytut Studiw Regionalnych (Institute forRegional Studies), Inteligentny
System Ratunkowy (Smart Rescue System), IRS Brakes Sweden, International Schools Reviews, Information Systems Research
orInternational Society ofRadiology. For this reason, especially during thefirst period ofcommunication activities, thefull name
ofthe project was often used. As therecognition ofRRI reports increased, itwas gradually replaced by theacronym.

168

13.4. What aided thecommunication activities strengths


Strength: recognised analysts. One ofthe main strengths ofthe communication activities consisted inthefact that comments
by leading economists members ofthe research team and theinterpretation panel lent credence tothepresented results.
Their authority, knowledge and often recognition by thegeneral public facilitated attracting media attention, increased its
reach and built thecredibility offorecasts presented inRRI reports. Theformula ofthe interpretation panel implemented aspart
ofthe project and described elsewhere inthetext, allowed us togather comments from experts and arguments tobe used
inthecommunication process.
Strength: confidence. Thefact that theRRI project was implemented by thePolish Agency forEnterprise Development attracted
interest and confidence on thepart ofthe recipient groups. PARP is thelargest and most important implementation institution
inthesystem forthesupport ofenterprises inPoland, and theresearch component is executed by a team ofrecognized experts
and analysts from theMalopolska School ofPublic Administration headed by Professor Jerzy Hausner.
Strength: innovation. TheRRI project represents an innovative process prompting change inthearea ofplanning and
implementation ofintervention and aid programmes tocounter thenegative effects ofeconomic change on thelabour market
and support enterprises threatened by temporary difficulties. Theinnovative aspect ofthe system facilitated communication
with themedia and reaching out toits target groups. Theelements that contribute totheuniqueness ofRRI Reports were
therefore crucial inthecommunication process. Theoriginal method, thelargest research sample inPoland, thecyclical nature
ofobservations and thepossibility ofanalysing theresearch results from a variety ofperspectives by region, by industry orby
structure are theelements that exerted a positive influence on theeffects ofcommunication activities.
Strength: thesubject matter. Communication activities were carried out during theeconomic downturn, theeffects ofwhich were
felt by theaverage citizen and were reported every day by themedia. Thedata on symptoms ofa deteriorating financial condition
ofindividual NACE sections and divisions, communication ofinformation concerning bankruptcy risk, theforecasts ofbasic
macroeconomic indicators, theperspectives foreconomic growth and scenario analyses ofthe impact ofpotential impulses on
thecondition ofthe Polish economy will always be sought out by themedia.
Strength: reliability offorecasts. Most research presented inthemedia focuses on theanalysis ofthe current economic situation.
Theforecasts contained inRRI Reports guaranteed uniqueness ofthe message, whereas thecyclic presentation ofresearch results
made itpossible tocompare theforecasts with thecurrent economic situation. Theverifiability ofnumerous forecast elements
was supported by media coverage even insituations where thepredicted values differed from other widely publicised analyses.
Accordingly, communication activities focused on theforecasting aspect complete with thedescription ofthree forecast options:
optimistic, neutral and pessimistic ones with an emphasis on their high accuracy.

13.5. Relations with themedia planning activities, database ofjournalists


and topical diversity
Planning communication with themedia started with thecompilation ofa media database, which enabled us toreach out
tointerested journalists with dedicated information. ThePR strategy inthedraft Rapid Response Instrument provided fora variety
ofpress releases delivered toeditors, which was facilitated by thepresence ofextensive documentation contained inthepublished
Reports. Analyses and forecasts contained inthem allowed us toprepare materials tailored todifferent recipients, hence relevant
texts were printed inthenational, regional, and trade media.
Accordingly, thedatabase ofjournalists was continuously updated, depending on thetopic ofthe press release so that, forexample,
analyses ofspecific NACE divisions reached professional and trade magazines who might potentially be interested inthem, while
more general macroeconomic forecasts were circulated toa broader group ofjournalists working fornewspapers and websites
devoted tobusiness issues. Beforehand, thedatabase was verified inorder tomake sure that thescope ofthe material matched
theprofessional interests ofindividual journalists. It was thetopical diversity ofpress releases which allowed us toinitiate and
maintain effective media relations with opinion-forming editorial boards, professional and trade media.
When selecting thetopics and dates ofcirculation ofeach press release, thefollowing factors were taken into consideration:
importance ofindividual events fortheeconomy (e.g. thedecisions ofthe Monetary Policy Council), theschedule and subject
matter ofGUS reports aswell asthepublications ofother research centres and business intelligence agencies. Thanks totheabove,
our research results were directly related tothecurrent economic situation and thecurrent problems ofindividual branches
ofindustry.
Below there is a list oftitles ofpress releases circulated tothemedia from April 2013 toMarch 2014.

169

04/2013 Economic growth oran increase inbankruptcy rates, Services face increased bankruptcy risk
05/2013 Sectors unaffected by thecrisis
06/2013 Is consumption going topull us out ofthe recession?, Social unrest likely inSilesia
07/2013 When will thecondition offirms improve?
08/2013 Manufacturing and exports will overcome theslowdown
09/2013 2014 will bring a revival oftrade
10/2013 Deferred consumption orsustainable growth?
11/2013 Favourable forecasts forindustrial production
12/2013 Hausner: construction and mining still closely watched, 2014: expected recovery intheautomotive industry
01/2014 PAED: economic recovery is a fact
02/2014 PAED: thecontribution ofexports toGDP will gradually decline
03/2014 PAED: industrial production growth will decline in2015, Likely increase inbankruptcy risk intheconstruction
industry

The principal objective ofcontinuing communication activities was thepresentation ofthe RRI asa warning system that consumes
fresh data and produces information about what may happen intheeconomy inthecourse of612 months. At thesame time,
thesubject matter ofRRI Reports, theissues offorecasting bankruptcies and recession imposed certain discipline interms
ofdrawing conclusions, oreven thetitles ofpress releases themselves so asto avoid theeffect ofscaring thetarget recipients
and prevent misinterpretation. Even so, thetitles ofpublished materials given by theeditorial staff, sometimes had more edge
than thetitles ofthe circulated press releases, ascan be seen intheheadlines, forexample, conclusions from therelease titled
Social unrest likely inSilesia were published by theportal wnp.pl asBankruptcies ofmining forms will cause social unrest, whereas
Trybuna Grnicza (The Mining Tribune) ran an article with thefollowing headline Is mining going under?
The specialized language ofRRI Reports and certain topics sometimes turned out tobe hard tounderstand fortheaverage
recipient and required additional statements by themembers ofthe research team and theinterpretation panel, which clarified and
interpreted theobserved phenomena, while at thesame time explaining thebasis ofthe methodology used. Similarly, dedicated
comments and specialized texts on particular industries, regions and issues were prepared inresponse tocurrent questions raised
by therepresentatives offirms affected by theeconomic slowdown and transmitted tous by therepresentatives ofthe media.
Owing tothein-depth analysis ofthe condition ofthe various NACE sections and divisions contained intheReport on theassessed
bankruptcy risk ofenterprises themicroeconomic component, a strong emphasis inthefield ofmedia relations was placed
on establishing relationships with theprofessional and trade media. In cooperation with their editorial boards, expert articles
were prepared containing comments by members ofthe research team, in-depth analyses ofselected branches ofindustry and
statements by members ofthe interpretation panel. Long-term cooperation was established with several professional and trade
magazines, which resulted incycles ofpublications devoted totheprospects forthePolish economy and its individual sectors
based on theRRI Reports.
Below there is a list oftitles ofexpert articles, which were published by selected professional and trade media from April 2013
toMarch 2014 asa response totheneed forin-depth editorial materials related tothecondition ofindividual industries.
The uniqueness ofthe RRI Reports and their forecasts about various economic indicators mean that their main recipients were
economic journalists who present theRRI projections incomparison with other available analyses and predictions. A key role
inthis respect was played by maintaining contact with theeditorial staff ofeconomic news agencies (Polish Press Agency (PAP),
Newseria Business, Reuters) due totheir influence and impartiality. A PAP representative was granted exclusive right toparticipate
asan observer inmeetings ofthe interpretation panel and was thefirst toreceive thelatest data and comments from theresearch
team. ThePAP representative had theopportunity tolisten tothediscussions on theworking versions ofRRI Reports, gain a better
understanding ofthe structure ofthe model and directly interview theexperts. Thanks tothis solution theagency despatches
were subsequently quoted by major dailies thanks totheir high factual quality and credibility.
05/2013 Can employers afford tobe optimistic inthecoming months? Portal nf.pl
06/2013 Not a paper crisis Przegld Papierniczy (Paper Review) monthly
07/2013 Increased bankruptcy risk despite optimistic forecasts Gazeta samorzdu i administracji (Biweekly paper oflocal
government and administration)
09/2013 Human capital incrisis an interview with Professor Jerzy Hausner Personel Plus (Staff Plus) monthly
10/2013 Food perspectives Rynek Spoywczy (The Food Market) magazine
11/2013 NACE Section 10 gaining orlosing ground? portalspozywczy.pl/dlahandlu.pl (internet portal)
11/2013 An optimistic outlook forthenear future Gazeta Finansowa (Financial Newspaper) weekly
11/2013 Sectors with a high degree ofbankruptcy risk rynekpracy.pl (internet portal)
12/2013 Why economic recovery has not translated into increased employment? Personel Plus (Staff Plus) monthly
12/2013 Will thePoles spend theautumn shopping? Rynek Spoywczy (The Food Market) magazine
12/2013 The economy accelerates asis theindustry Polski Przemys (Polish Industry) magazine
02/2014 The end ofthe slowdown insight Think Tank Magazine
02/2014 Will Section H benefit from therecovery? Industry Perspectives forbusiness in2014 Polski Przemys (Polish
Industry) magazine

170

03/2014 Deferred consumption orsustainable growth? Forecasts forthefood industry for2014 Magazyn Spoywczy
(Grocery Magazine)
03/2014 The automotive industry inwhich gear? Forecasts fortheindustry an interview with Dr Andrzej Kondratowicz,
reviewer ofRRI Reports wnp.pl (a portal specialising inbusiness issues)
Likewise, theissues taken up by RRI Reports constituted a platform forcontacts with analysts and economists from various financial
institutions. Communication activities were preceded by thepreparation ofdatabase ofcontacts, who were then provided with
selected content specialist releases along with theRRI Report Synthesis by Professor Jerzy Hausner. This resulted inbuilding
thebrand ofRRI Reports asan analytical tool tobe used by therepresentatives ofthe above group intheir daily work.

13.6. The Interpretation Panel objectives and principles of operation


One of the special and unique communication platforms within the RRI (ISR) system is the so-called Interpretation Panel (IP).
In Chapters 3, 8 and 9 we have identified a number of basic conditions for an effective implementation of such a system.
Unfortunately, some of these conditions, such as a considerable restraint in the modification of model specifications one of the
reasons why its users trust it1 also has apparent drawbacks in that it may sometimes cause excessive convergence of the results
obtained. Conventional forecasting often makes use of specific ad-hoc modifications. However, an effective warning system should
examine the long-term values of the tools used. The necessary adjustments and comments are thus introduced into the results
without necessarily changing the models themselves too often. This was one of the reasons why the Interpretation Panel was
established within the RRI (ISR) structure.
The Interpretation Panel constitutes a two-way communication platform since it ensures both expert verification of RRI (ISR)
anticipatory research results and the transfer of knowledge to professional circles in the relevant problem areas.
The operating practice of RRI (ISR) demonstrated that the experts cooperated, among others, in the following areas:

Evaluation of the results obtained as part of the RRI (ISR) research component from alternative research and analytical
perspectives,

Inclusion amongst the phenomena to be diagnosed by the RRI (ISR) of those elements considered to be important from the
vantage point of bankruptcy risk assessment in the economy and its potential consequences, which cannot be included
directly in the project due to the limitations (of each) methodology,

Updating parts of ratings: it should be emphasised that forecasting tools and their practical application requires specific cutoff points, beyond which no research or analyses are carried out; however, warning systems should always take into account
the most recent knowledge, hence, if there are no technical or organizational possibilities of including it in applied formal
and model tools, such knowledge should constitute expert input,

Provision of anticipatory, non-forecastable knowledge based on model approaches; it should be noted that irrespective of
their class, models are based on historical knowledge and its capacity to be transferred onto future events; in the event of
deep structural changes, certain shocks in the economy, etc. traditional modelling may prove to be insufficient.
The selection of experts guaranteed a successful completion of these IP tasks. The process was based on three fundamental
principles:
1. First-rate professionalism.
2. Extensive knowledge/practical experience in the field of public, academic or consultancy activities related to specifics of the
methods and tasks of the RRI (ISR) project.
3. Research independence.
These conditions not only guaranteed a successful completion of IP tasks, but also the achievement of the all-important expert
knowledge pooling effect2 (cf. e.g. Blinder 2007). Without going into formal considerations of this effect, it does not work in
the absence of independent views. Naturally, the operation of such a panel would be adversely affected by differences in the
competencies of its individual members. Likewise, the size of the team 8 members was set in accordance with the principles
of optimization of such an effect (cf. e.g. Erhart, Lehment et al. 2007).
As far as the organization of IP activities is concerned, the developed solutions and procedures were based on one of the
fundamental characteristics of warning systems, namely that the exchange of expert knowledge within an EWS is not supposed
to resemble a seminar. What is important in such systems is the response time (a short response time means low inertia) and the
accuracy of conclusions (clarity and concreteness of views). Therefore, the experts worked using RRI (ISR) reports supplied to them
1

Frequent changes in this area generally undermine the trust of users of such a system in its inner wisdom; among others due to the impression
that the tool tends to adapt rather than anticipate.
The activities of the central banks and their monetary policy are now, thanks to the so-called monetary policy councils, one of the betterresearched areas in terms of objectives, design principles and methods of operation of decision-making and advisory collegial bodies. This was
one of the reasons why this area was chosen as the main source of bibliographical references

171

with sufficient (but not excessive) notice before the IP plenary meeting. Plenary meetings were organized on a quarterly basis in
accordance with the schedule of individual RRI (ISR) project editions.
IP members were, among others, required to submit prior to the meeting (3 working days) written, concise comments on the
materials received. IP meetings were held in strictly pre-determined timeframes. Their leading topics were also determined based
on the syntheses of materials supplied by the IP members. Within 7 working days after each meeting, IP members expressed their
final opinion on the topics under discussion. Both the written opinions and the plenary discussion were used in drafting the final
recommendations of each RRI (ISR) edition.
The experience with the IP operation can only be judged as very positive. It constituted a valuable source of verification of the
anticipatory RRI (ISR) research results, strengthened the holistic effects of the tool as well as improved and expanded the research
and analytical perspectives of the system.

13.7. Other public relations tools


Online debates
The RRI Reports also analyse therisk ofbankruptcy by administrative area, thedata is thus relevant toenterprises located
throughout Poland. That is why itwas so important toundertake communication efforts targeting therepresentatives oflocal
administration, business support organizations aswell astheentrepreneurs themselves who operate insmaller agglomerations.
One ofthe contact channels was provided by online debates with well-known Polish journalists. Online broadcasts allowed us
toreach out tothetarget group with conclusions ofthe RRI Reports without having totravel long distances. Three debates were
held by means ofthis medium:

When is thebusiness sector likely tosee a revival? conducted incooperation with Dziennik Gazeta Prawna (a daily focussing
on business and legal affairs) inOctober 2013,

Bankruptcy and restructuring whether and how tosupport firms affected by theeconomic downturn? organized
incooperation with Rzeczpospolita (a nationwide daily) inNovember 2013,

Challenges tothePolish economy in2014 held during a symposium organized by theMSAP inJanuary 2014.
Conferences
The planned PR activities included talks by RRI experts during conferences, which offered opportunities fordirect interaction with
representatives ofthe target groups. Opinions and questions formulated by practitioners provided clues asto which elements
ofthe Early Warning System should be emphasized, which forecasts and research results attracted themost interest. A presentation
ofconclusions and recommendations worked out thanks totheproject totherepresentatives ofthe business environment took
place during a debate organized by theBCC inApril 2013. Other ones, addressed torepresentatives offirms (especially financial
and human resources policymakers inenterprises), took place during theConference on Restructuring inJune 2013 and theHR
Congress inOctober 2013. Regional meetings with themanagers ofenterprises, inparticular with their representatives responsible
formaking financial decisions, were planned forMay 2014 inKatowice, Krakw and Pozna (at thetime ofwriting).
An important target group ofadditional communication activities constituted decision-makers working instate administration.
Activities inthis area offered opportunities topresent theRRI project and thefounding principles ofthe Early Warning System
at themeeting ofthe Parliamentary Commission ofEconomy inJanuary 2014. Apart from attending a presentation by Boena
Lubliska-Kasprzak, President ofPAED, Members ofParliament participating inthework ofthe Commission were given materials
presenting theidea, conclusions and recommendations formulated aspart ofthe RRI project. Communication activities also
involved theparticipants ofthe Convention ofMarshals ofthe Republic ofPoland. Special publications presenting theproject
objectives and themost important results ofRRI Reports were prepared fortheConvention organized by theauthorities ofthe
witokrzyskie Voivodship inApril 2014.

13.8. Effects ofcommunication activities


The implemented ways ofreaching thetarget groups and topics raised inRRI Reports proved tobe effective and corresponded
totheneeds ofopinion-forming press. From April 2013 toMarch 2014 inpaper and electronic media at least 585 materials tailored
inthis way were published. Articles and news about RRI Reports appeared, among others, inthefollowing:

Dailies: Rzeczpospolita, Gazeta Wyborcza, Puls Biznesu, Dziennik Gazeta Prawna, Polska The Times, Dziennik Trybuna, Dziennik
Polski, Nasz Dziennik, Gazeta Giedy Parkiet,

Weeklies: Wprost, Forbes,

Horizontal portals: onet.pl, wp.pl, interia.pl,

172

Thematic websites: money.pl, bankier.pl, inwestycje.pl, obserwatorfinansowy.pl, biznes.pl, wyborcza.biz, nf.pl, stooq.pl, waluty.
com.pl, abc.com.pl, portalsamorzadowy.pl,
Professional and trade internet portals an printed magazines: Nowy Przemys, Gazeta Finansowa, wiat Druku, Rynek Spoywczy,
Magazyn Przemysowy, Spka z o.o., Wiadomoci Kosmetyczne, Personel Plus, Poligrafika, Polski Przemys, Przegld Papierniczy,
Wiadomoci Handlowe, samar.pl, portalspozywczy.pl, spedycje.pl, formutransportu.pl, rynekpracy.pl, gospodarkamorska.pl,
dlahandlu.pl, pulshr.pl, poradnikhandlowca.pl, rolnicy.com, forummleczarskie.pl, propertynews.pl,
Regional press and portals: Dziennik Zachodni, Gos Wielkopolski, Tygodnik Ostrocki, Gos Koszaliski, Trybuna Grnicza,
Kurier Lubelski, Dziennik Batycki, Gazeta Wrocawska, Nowiny, Dziennik dzki, Kurier Szczeciski, gospodarkapodkarpacka.pl,
wrotapodlasia.pl, gazetalubuska.pl, eostroleka.pl.

Activities inthearea ofmedia relations also included ongoing contacts with television and radio stations. Theissues raised by
theRRI were mentioned and commented on innewscasts, information broadcasts and feature programmes devoted tobusiness
issues, among others inPolsat Biznes, TVN24 Biznes i wiat (d. TVN CNBC), TVP Info, Polskie Radio, Radio PIN, Radio ZET and
regional radio stations. Thevariety ofcontent intheRRI Reports facilitated theproduction ofsuch programmes, including direct
presentations and thefollowing:

RRI experts opinions regarding thecorrelations between forecasts forspecific branches ofindustry and thequotations
ofshares oflisted companies (recorded forTVN CNBC),

RRI experts comments on regional differences inbankruptcy risk (recorded ina regional radio station Radio RDC),

RRI experts participation ina discussion on thecondition and forecasts forselected branches ofindustry (recording forPolish
Radio).
Data taken from theRRI Reports also appeared inthecontext ofa tool tosupport students inpreparing term papers, Bachelors
and Masters degree theses. An interview with a member ofthe research team was broadcast by Akademickie Radio Kampus.
The effects of operation of the Interpretation Panel are discussed in more detail in Section 13.7.
The success ofany research project consists inthedissemination ofits research results and analyses by people not directly involved
intheresearch process. In thecase ofcommunication activities intheRRI project, thegoal was not only toinform thepublic about
theconclusions contained inthereports, but also tocreate a positive climate forthediscussion ofbankruptcy-related matters,
forecasting theeconomic situation and theneed topursue pre-emptive activities aimed at mitigating thenegative effects ofthe
economic downturn.
Can any information campaign contribute toa systemic change inwhich theintervention-based approach should be replaced
by a preventive approach? Although communication activities themselves are not sufficient, they may pave theway fora change
ofthinking about thebankruptcy ofenterprises and theprovision ofsupport during theeconomic downturn. Virtually everyone
agrees with thesaying that prevention is better than cure. However, theanswer tothequestion ofhow toprevent and who exactly
tocure raises far more problems. Therefore itseems important topursue a well thought-out and systematic communication
campaign inthearea ofmonitoring theeconomic situation.
Thanks totheeffective communication strategy we have also succeeded inchanging theperception ofthe reports themselves.
Thepresentation ofresearch results carried out aspart ofthe RRI ceased tobe associated exclusively with academic discourse and
was given a more accessible form. It was achieved thanks toa series ofpublications which translated theinitially difficult contents
into relevant and comprehensible messages. Thetopical diversity ofmaterials allowed us toreach a wide audience with our
message. TheRRI Reports themselves have become thebasis fordiscussion inthebusiness community and among policy makers
at thecentral and regional level. Themost important determinants ofsuccess ofthe implemented measures include thecirculation
ofthe conclusions ofthe RRI Reports, their recognition among theleaders ofopinion-forming circles and numerous citations by
themedia.

173

Part IV
Medium-term bankruptcy risk inthePolish
economy forecasts and recommendations

ukasz Lenart, Baej Mazur, Krystian Mucha, Mateusz Pipie, Justyna Wrblewska

Chapter 14
FORECASTS OF THE MACROECONOMIC SITUATION AND THEIR IMPACT
ON BANKRUPTCY RISK
14.1. Introduction
This part ofthe monograph presents empirical examples illustrating theapplication ofthe proposed econometric and statistical
methods foranalyses conducted inthemacroeconomic component ofthe project. These methods are described indetail
inChapter 11.
On thebasis ofthe tools described intheprevious parts ofthe study, a detailed analysis ofthe current economic situation by
individual divisions ofmanufacturing and trade is performed. It is also possible toconstruct short-term forecasts concerning
thedevelopment prospects forindividual production divisions under consideration. In order toillustrate theresults ofthe
approach, this chapter focuses on thesituation intwo selected divisions: one taken from theproduction sector and theother
from theconstruction sector. Manufacturing asa subdivision ofproduction and construction ingeneral were subject toan indepth analysis.

14.2. Analysis ofbusiness activity inselected nace sections (manufacturing, trade and
construction)
The tools described intheprevious sections enable us toperform a detailed analysis ofthe current economic situation
inproduction and trade. It is also possible toproduce short-term forecasts about theprospects fordevelopment inindividual
divisions under consideration. In order toillustrate theresults ofthe approach discussed, this section focuses on theeconomic
situation intwo selected divisions one inthemanufacturing sector and theother intheconstruction sector. Thein-depth analysis
focuses on manufacturing and construction ingeneral.
Table 14.1 contains a sample report on theanalysis ofthe economic situation inthemanufacturing offabricated metal products
except machinery and equipment. Theanalyses are based on observations made until June 2012. They are presented asa business
cycle clock, a series ofidentified deviations, a fan chart ofthe predictive distribution ofthe production change rate production
intheanalysed division (based on observation until June 2012) and a fan chart based on observations made until March 2012,
together with observations involving April, May and June. Table 14.2 presents synthetic information about theeconomic situation
and development prospects inconstruction ingeneral. Each report on theeconomic situation comes complete with a brief
description ofthe business cycle clock, theforecasts and projections prepared fortheprevious cycle.
The deviation cycles identified have been analysed incorrelation with thetotal manufacturing cycle. Tables 14.3 and 14.4 show
thecorrelation coefficients between thedeviation cycles inmanufacturing and construction and delayed oranticipated deviation
cycles inmanufacturing. Thetables include maximum absolute values ofthe correlation coefficients (above 0.5).

177

Table 14.1. Theeconomic cycle report forthemanufacture offabricated metal products except machinery and equipment
Manufacture of fabricated metal products except
machinery and equipment

Deviation cycle
(lambda~4,5 year)

3,0%
2,0%
1,0%
0,0%
-1,0%
-2,0%
-3,0%
-4,0%
-5,0%
-6,0%
-7,0%

-1,5%

-1,0%

-0,5%

0,0%

0,5%

1,0%

1,5%

2,0%

2,5%

The individual predictive distributions demonstrate a slight


slowdown inbusiness activity inthis division inthesecond
half of2012 and inearly 2013. This marks a continuation ofthe
downward trend intherate ofchange ofproduction initiated
inJanuary 2011. A continued decline inproduction, however,
is unlikely and does not exceed 0.4 inall forecast horizons.
New observations led tothemodification ofthe predictive
distributions incomparison with theprevious forecasting
round. Predictive medians are variable and fluctuate around
5% y/y, which, given their highly dispersed distributions is
associated with considerable uncertainty asto theeconomic
situation over theforecast horizon.

Dynamics of change
2011M12

2011M12

Variant 2

Manufacture of fabricated metal products except


machinery and equipment

9%
6%
3%

Y/y change (%)

sty-09

sty-11

The observed production data inthedivision under


consideration were clearly anticipated intheprevious
scenario.

-9 %
-1 2%
Periods of recession (total manufacturing)
12000 (5,5 year)
55000 (8 year)

35%
30%
25%
20%
15%
10%
5%
0%
-5%
-10%
-15%

35%
30%
25%
20%
15%
10%
5%
0%
-5%
-10%
-15%
-20%

5500 (4 ,5 year)
32000 (7 year )

Manufacture of fabricated metal products except


machinery and equipment

VII

IV

2009

Y/y change (%)

sty-10

sty-07

sty-08

sty-05

sty-06

sty-03

sty-04

sty-01

sty-02

sty-99

sty-00

sty-97

sty-98

-6 %

sty-95

0%
-3 %

sty-96

Deviations from the trend line

Variant 1

VII

IV

2010

VII

IV

2011

VII

2012

IV VI

2013

Manufacture of fabricated metal products except


machinery and equipment

VII

2009

IV

VII

2010

IV

VII

2011

IV

VII

2012

The final points on thebusiness cycle clock inits classical


variant appear along theaxis between thesecond and
third quarters ofthe coordinate system, which indicates a
deterioration intheeconomic situation inthis manufacturing
division ascompared with theresults ofthe previous report
(which signalled entry into thevicinity ofthe upper turning
point ofthe deviation cycle). A correlation analysis indicates
a high degree ofsynchronization between thedeviation
cycle ofthis variable with thedeviation cycle oftotal
manufacturing. Thecorrelation coefficient fordelays of2 and
3 months, respectively, exceeds 0.89.

III

2013

Source: own study.

178

Table 14.2. Theeconomic cycle report inconstruction industry


Construction industry

Deviation cycle
(lambda~4,5 year)

4,0%
3,0%
2,0%
1,0%
0,0%
-1,0%
-2,0%
-3,0%
-4,0%
-5,0%
-6,0%
2,0%

-3,0%

-2,0%

-1,0%

0,0%

1,0%

Dynamics of change
Variant 1

2011M12

2011M12

Variant 2

6%
3%
sty-11

sty-10

sty-09

sty-08

sty-07

sty-06

sty-05

sty-04

sty-03

sty-02

sty-01

0%
sty-00

Deviations from the trend line

9%

-6%
-9%

Periods of recession (total manufacturing)


12000 (5,5 year)
55000 (8 year)

5500 (4 ,5 year)
32000 (7 year )

Construction industry
40%

Y/y change (%)

30%
20%
10%
0%
-10%
-20%
-30%
-40%

VII

IV

2009

VII

IV

2010

VII

IV

2011

VII

X XIII III

2012

VI

2013

Construction industry
40%
Y/y change (%)

30%
20%
10%
0%
-10%
-20%
-30%
-40%

VII

2009

IV

VII

2010

IV

VII

2011

Predictive distributions ofthe indicator under consideration


are characterized by changes inthelocation ofthe central
tendencies throughout theforecast horizon. Therate
ofchange ofthe indicator under consideration will oscillate
around zero, with small positive values until theend of2012.
Theprobability ofa decline inconstruction activity equals 0.5
almost across theentire forecast horizon. This is associated
with considerable uncertainty asto thedevelopment ofthe
industry over thenext 18 months.
The observation ofMay 2012 was accurately forecast at
thelevel ofthe predictive median intheprevious edition
ofthe report. Other observations from thesecond quarter
of2012 differ from thecentral tendencies. What was surprising
was theunexpected improvement intheindustrys condition
inApril 2012, aswell asits marked deterioration inJune 2012.

Construction industry

-3%

The analysis ofthe business cycle clock indicates theslowing


down ofeconomic recovery intheconstruction industry
ingeneral. Thelast point ofthe clock inits classical variant is
tobe found inthesecond quadrant ofthe coordinate system.

IV

VII IX X

2012

III

2013

Source: own study.

179

Table 14.3. Relationships between thedeviation cycles inindividual NACE manufacturing divisions and thedelayed oranticipated
deviation cycles intotal manufacturing
Section/Division Anticipated cycle (months)

-1

-2

-3

-4

-5

-6

-7

-8

Mining and quarrying, manufacturing,


0,15 0,30 0,45 0,58 0,71 0,82 0,90 0,96 0,99 0,98 0,93 0,86 0,76 0,65 0,52 0,38 0,24
electricity, gas, steam and hot water,
construction
Intermediate goods

0,27 0,41 0,54 0,66 0,77 0,86 0,92 0,96 0,97 0,93 0,87 0,78 0,66 0,53 0,39 0,24 0,09

SECTION C

SECTION B

Energy-related goods (excluding Section E) -0,04 0,08 0,20 0,32 0,43 0,53 0,61 0,68 0,72 0,74 0,73 0,70 0,64 0,56 0,47 0,36 0,25
Energy-related goods (excluding Sections
D and E)

-0,05 0,06 0,18 0,30 0,41 0,51 0,60 0,67 0,72 0,73 0,72 0,68 0,61 0,52 0,41 0,29 0,17

Investment goods

-0,07 0,07 0,22 0,36 0,49 0,62 0,73 0,81 0,87 0,90 0,90 0,88 0,83 0,76 0,67 0,57 0,45

Durable consumer goods

0,42 0,53 0,63 0,72 0,78 0,82 0,83 0,81 0,77 0,69 0,59 0,47 0,33 0,18 0,03 -0,12 -0,26

Non-durable consumer goods

0,46 0,54 0,61 0,67 0,70 0,72 0,72 0,69 0,64 0,56 0,47 0,36 0,25 0,13 0,01 -0,11 -0,22

Mining and quarrying

0,03 0,12 0,20 0,29 0,37 0,45 0,51 0,58 0,62 0,64 0,63 0,60 0,55 0,48 0,40 0,30 0,20

Mining ofcoal and lignite

-0,06 0,02 0,10 0,19 0,27 0,34 0,41 0,47 0,51 0,53 0,53 0,51 0,46 0,40 0,32 0,23 0,13

Other mining and quarrying

0,07 0,16 0,25 0,34 0,44 0,53 0,61 0,68 0,73 0,76 0,76 0,73 0,69 0,64 0,57 0,49 0,39

Mining and quarrying support service


activities
Manufacturing

0,34 0,42 0,49 0,54 0,57 0,59 0,59 0,58 0,55 0,51 0,47 0,41 0,35 0,29 0,22 0,15 0,07
0,19 0,34 0,49 0,63 0,75 0,85 0,93 0,98 1,00 0,97 0,92 0,83 0,72 0,60 0,46 0,31 0,16

Manufacture offood products

-0,04 0,06 0,16 0,26 0,35 0,42 0,48 0,52 0,54 0,53 0,50 0,46 0,40 0,33 0,26 0,19 0,11

Manufacture ofbeverages

0,33 0,27 0,20 0,12 0,04 -0,04 -0,12 -0,19 -0,26 -0,33 -0,39 -0,42 -0,45 -0,46 -0,45 -0,43 -0,39

Manufacture oftobacco products

0,33 0,30 0,25 0,20 0,14 0,08 0,01 -0,05 -0,12 -0,17 -0,21 -0,24 -0,26 -0,26 -0,26 -0,24 -0,21

Manufacture oftextiles

0,23 0,36 0,49 0,60 0,70 0,78 0,84 0,87 0,88 0,87 0,82 0,75 0,66 0,56 0,44 0,31 0,18

Manufacture ofwearing apparel

0,13 0,21 0,30 0,39 0,47 0,55 0,61 0,66 0,69 0,69 0,67 0,64 0,58 0,51 0,43 0,33 0,23

Manufacture ofleather and related


products
Manufacture ofwood and ofproducts
ofwood and cork, except furniture,
manufacture ofstraw and plaiting materials
Manufacture ofpaper and paper
products
Printing and reproduction ofrecorded
media
Manufacture ofcoke and refined
petroleum products
Manufacture ofchemicals and chemical
products
Manufacture ofpharmaceutical, medicinal
chemical and botanical products
Manufacture ofrubber and plastics
products
Manufacture ofother non-metallic
mineral products
Manufacture ofbasic metals

0,32 0,42 0,51 0,59 0,66 0,71 0,74 0,76 0,75 0,73 0,68 0,62 0,54 0,45 0,36 0,26 0,15
0,44 0,56 0,66 0,73 0,79 0,83 0,84 0,83 0,78 0,71 0,61 0,49 0,35 0,21 0,08 -0,06 -0,18
0,55 0,64 0,71 0,77 0,80 0,81 0,79 0,75 0,68 0,59 0,48 0,35 0,21 0,08 -0,06 -0,19 -0,32
0,43 0,52 0,58 0,64 0,68 0,70 0,70 0,70 0,67 0,64 0,58 0,52 0,43 0,35 0,25 0,15 0,05
0,04 0,15 0,26 0,37 0,46 0,55 0,62 0,68 0,71 0,72 0,71 0,67 0,61 0,53 0,44 0,33 0,22
0,44 0,55 0,64 0,72 0,78 0,82 0,84 0,84 0,81 0,75 0,66 0,56 0,44 0,31 0,17 0,03 -0,11
0,07 0,06 0,05 0,03 0,01 -0,01 -0,04 -0,08 -0,12 -0,18 -0,24 -0,29 -0,35 -0,39 -0,43 -0,46 -0,48
0,36 0,48 0,60 0,70 0,78 0,84 0,88 0,90 0,87 0,82 0,73 0,62 0,50 0,36 0,22 0,08 -0,06
0,28 0,39 0,50 0,59 0,68 0,75 0,80 0,83 0,83 0,80 0,75 0,67 0,58 0,48 0,37 0,25 0,13
0,06 0,20 0,35 0,49 0,63 0,74 0,84 0,91 0,95 0,95 0,92 0,86 0,77 0,65 0,52 0,38 0,23

Manufacture offabricated metal products,


-0,12 0,01 0,14 0,27 0,40 0,52 0,64 0,74 0,82 0,87 0,89 0,89 0,86 0,81 0,75 0,66 0,56
except machinery and equipment
Manufacture ofcomputer, electronic and
0,46 0,53 0,59 0,63 0,65 0,66 0,65 0,61 0,56 0,48 0,38 0,26 0,14 0,01 -0,12 -0,25 -0,38
optical products
Manufacture ofelectrical equipment
Manufacture ofmachinery and
equipment n.e.c.
Manufacture ofmotor vehicles, trailers
and semi-trailers except motorcycles

0,17 0,32 0,46 0,59 0,70 0,78 0,85 0,89 0,89 0,87 0,82 0,74 0,65 0,54 0,41 0,28 0,15
-0,11 0,02 0,15 0,28 0,40 0,51 0,61 0,68 0,73 0,76 0,77 0,75 0,70 0,64 0,55 0,45 0,35
0,14 0,27 0,41 0,53 0,64 0,74 0,82 0,87 0,89 0,87 0,83 0,77 0,68 0,58 0,46 0,33 0,20

SECTION D

Manufacture ofother transport equipment -0,49 -0,41 -0,32 -0,22 -0,11 0,00 0,12 0,23 0,33 0,43 0,51 0,57 0,62 0,66 0,69 0,70 0,70
Manufacture offurniture

0,27 0,36 0,43 0,50 0,55 0,60 0,63 0,65 0,65 0,62 0,58 0,53 0,46 0,39 0,32 0,24 0,16

Other manufacturing

-0,14 -0,08 -0,01 0,05 0,11 0,17 0,23 0,29 0,33 0,36 0,37 0,37 0,37 0,36 0,34 0,32 0,29

Repair, maintenance and installation


ofmachinery and equipment

-0,40 -0,27 -0,14 0,00 0,15 0,29 0,42 0,54 0,64 0,72 0,78 0,81 0,83 0,82 0,79 0,74 0,67

Electricity, gas, steam, hot water


and conditioning supply

-0,13 -0,04 0,05 0,13 0,20 0,26 0,30 0,34 0,36 0,37 0,37 0,36 0,33 0,31 0,27 0,24 0,20

Source: own study.

180

Table 14.4. Relationships between theNACE subdivisions ofconstruction and thedelayed oranticipated deviation cycles
inindustrial production asa whole
8

Variable Anticipated (months)

-1

-2

-3

-4

-5

-6

-7

-8

Construction (total)

-0,18 -0,16 -0,13 -0,10 -0,06 -0,01 0,05 0,11 0,18 0,25 0,31 0,37 0,43 0,49 0,54 0,59 0,62

Construction ofbuildings

-0,56 -0,49 -0,41 -0,30 -0,18 -0,06 0,06 0,19 0,30 0,43 0,53 0,63 0,70 0,76 0,79 0,80 0,79

Civil engineering

0,42 0,37 0,30 0,22 0,14 0,07 0,00 -0,06 -0,10 -0,16 -0,21 -0,23 -0,24 -0,22 -0,19 -0,14 -0,09

14.3. Forecasts ofmacroeconomic aggregates


The methodological approach outlined above enables its users togenerate short-term forecasts ofeconomic aggregates. This
section presents examples ofpredictions made forboth fordata updated on a monthly and on a quarterly basis. Theforecast
monthly time series include therate ofchange intotal production and therate ofchange inretail sales. In thecase ofquarterly
data, theforecast includes thepercentage change (year on year) inGDP, domestic demand, exports and gross value added
intheindustry.
Figure 14.1 shows fan charts representing thepredictive distributions constructed fortheseries ofrates ofchange ofproduction
and sales, respectively. Theanalyses are based on data collected until October 2013. Thered dots on thegraph indicate theactual
values ofthe projected time series inNovember, December 2013 and inJanuary 2014. Thegraph also shows theplot ofthe cyclic
factor estimated on thebasis ofthe expected posterior values ofthe model (black-and-white line).
Figure 14.2 shows fan charts ofpredictive distributions constructed fortheseries ofchanges intheGross Domestic Product,
domestic demand, exports and gross value added inindustry. Thefan charts are contrasted with theforecasts with thecyclical
factor excluded (blue line). As is thecase with themonthly time series, thecyclic component is shown asa black and white line.
Variant projections with thecyclical factor excluded assume that thecomponent is constant intime and its value approximates
thearithmetic mean ofthe sample taken from thetime series. Actual values ofthe forecast series are marked inred. Figure 14.2
shows forecasts related totheXI RRI Macroeconomic Report based on observations comprising thethird quarter of2013.
The analysis ofthe impact ofnew observations, dynamically incorporated into thetime series, on theplot ofthe forecasts is
shown inFigure 14.3. Thecyclical component resulting from theeight forecasting rounds is plotted on thebasis ofobservations
comprising thetime period from thefirst quarter of2012 (1st forecast) tothefourth quarter of2013 (8th forecast) and is marked
ingreen.
The shape (plot) ofthe regular cyclical component changes asa result ofnew observations included intheupdated time series
under consideration. Theimpact ofchanges affecting this component after factoring intheobservations from thefourth quarter
of2013 concerning thefour macroeconomic aggregates ofinterest is shown inFigure 14.4. TheRRI reports discussed such changes,
but adding a single observation did not usually significantly alter theshape ofthis component ofthe model. In consequence,
theeconomic development scenarios presented insubsequent macroeconomic reports were relatively similar. With theexception
ofdata which indicate thepresence ofa turning point and an imminent trend reversal, theoverall tone ofthe forecast generally

20

20

20

20

15

15

15

15

10

10

10

10

-5

-5

-5

-5

-10

-10

-10

XI

III

V VII IX XI

2012

2013

III

V VII IX XI
2014

III

XI

2012

2015

new observations vs the previous forecast

a) industrial production (%) y/y

III

V VII IX XI
2013

III

V VII IX XI
2014

new observations vs the previous forecast

b) retail sales (%) y/y

Fig. 14.1. Monthly data: forecasts with observations added ex-post


Source: own study.

181

III
2015

-10

-1

-1

-2

-2

-3

-3

-4

-4

-5

II

III

IV

II

2012

III

IV

2013

II

III

IV

2014

I
2015

25

20

20

15

15

10

10

III IV
2012

IV

II

III

IV

II

2013

III

IV

2014

70%

90%

new observations
vs the previous forecast

-5

50%

forecast from a model without


the cyclical component

2015

b) domestic demand (%) y/y

25

II

III
2012

a) GDP (%) y/y

-5

II

30%

II

III
2013

IV

II III
2014

IV

I
2015

15

15

13

13

11

11

-1

-1

-3

-3

-5

-5

II

III

IV

2012

c) exports (%) y/y

II

III

IV

II

2013

III

IV

2014

30%

50%

90%

70%

new observations
vs the previous forecast

-5

forecast from a model without


the cyclical component

2015

d) gross value added in industry (%) y/y

Fig. 14.2. Quarterly data: forecasts with observations added ex-post


Source: own study.

-1

II

III

2008

IV

II

III

2009

IV

II

III

2010

IV

II

III

IV

2011

II

III

IV

II

2012

Fig. 14.3. Gross GDP (%) y/y: a comparison ofthe current and previous point forecast paths
Source: own study.

182

III

2013

IV

II

III

2014

IV

II

-1

2015

-1

-1

II

III

IV

2012

II

III

IV

II

2013

III

IV

2014

-2

2015

II

III

IV

2012

The regular cyclical component in the current forecast


(including the new observation)

II

III

IV

2013

II

III

IV

2014

-2

2015

The regular cyclical component in the current forecast

a) GDP (%) y/y

b) domestic demand (%) y/y

25

25

25

25

20

20

20

20

15

15

15

15

10

10

10

10

-5

-5

-5
I

II

III

2012

IV

II

III

2013

IV

II

III

-5
II

IV

III
2012

2014

IV

II

III

IV

2013

II

III

2014

IV

I
2015

The regular cyclical component in the current forecast


(including the new observation)

The regular cyclical component in the current forecast


(including the new observation)
c) exports (%) y/y

d) gross value added in industry (%) y/y

Fig. 14.4. Quarterly data: the impact ofnew observations on thepoint forecasts ofthe regular cyclical component
Source: own study.

remains unchanged, nevertheless adjustments are made asto thepredictive probability ofa recession. Themethodology
developed quite accurately predicted theeconomic slowdown and thetrend reversal which occurred inthesecond quarter
of2011. Similarly, theend ofthe downturn and a small likelihood ofrecession inthefirst quarter of2013 were fairly accurately
predicted.

14.4. Sensitivity analyses


Using theresponse torandom shock function, we analysed theeffect ofexogenous shocks/innovations affecting thevariables
under consideration on thenumber ofbankruptcies, which, asin theprevious reports, allowed us totrace theinfluence ofselected
economic categories on enterprises.

183

Table 14.5. Quantiles ofthe posterior distribution ofthe response function ofthe number ofentities inbankruptcy torandom
shocks affecting selected variables (shock size equals one posterior standard deviation foreach variable, which
entails an increase inthelogarithm (ln) ofthe GDP by about 0.011 inthezero period, thereal interest rate by 0.393%,
thelogarithm ofthe exchange rate by 0.039 and thelogarithm ofthe number ofbankruptcies by 0.239).
IRF

WIBOR3M no of
bankruptcies

GDP no ofbankruptcies
Quantile

Horizon

EUR/PLN no of
bankruptcies

Quantile

no ofbankruptcies no of
bankruptcies

Quantile

Quantile

0.16

0.5

0.84

0.16

0.5

0.84

0.16

0.5

0.84

0.16

0.5

0.84

-0.053

-0.021

0.011

-0.046

-0.016

0.016

-0.039

-0.009

0.021

0.219

0.239

0.262

-0.031

-0.010

0.008

-0.061

-0.023

0.014

-0.023

-0.003

0.016

0.107

0.139

0.174

-0.036

-0.013

0.009

-0.052

-0.018

0.014

-0.027

-0.004

0.018

0.144

0.168

0.194

-0.030

-0.010

0.008

-0.055

-0.018

0.017

-0.024

-0.003

0.018

0.111

0.142

0.175

-0.031

-0.010

0.009

-0.054

-0.015

0.021

-0.025

-0.002

0.019

0.112

0.143

0.177

-0.029

-0.009

0.008

-0.055

-0.013

0.023

-0.025

-0.002

0.019

0.096

0.132

0.170

-0.028

-0.008

0.009

-0.056

-0.012

0.026

-0.025

-0.002

0.020

0.089

0.129

0.168

-0.026

-0.008

0.009

-0.057

-0.011

0.027

-0.025

-0.002

0.021

0.080

0.122

0.164

-0.026

-0.007

0.009

-0.057

-0.011

0.029

-0.026

-0.001

0.022

0.073

0.116

0.162

-0.025

-0.006

0.009

-0.057

-0.010

0.030

-0.026

-0.001

0.022

0.065

0.110

0.159

10

-0.024

-0.006

0.010

-0.057

-0.009

0.031

-0.027

-0.001

0.024

0.060

0.106

0.157

Tables 14.5 and 14.6 and Figures 14.4 and 14.5 show thequantiles ofthe posterior distribution function ofthe response ofthe
number ofbankruptcies toa single isolated shock (i.e. assuming that other variables remain constant) intheeconomic figures under
consideration. When identifying shocks (orthogonalization), thefollowing order ofvariables was adopted: variables representing
Polands economic environment (EUR/USD exchange rate, Germanys GDP, theprice ofcrude oil), theGDP, thebasic interest rate,
PLN/EUR exchange rate and thenumber ofcompanies inbankruptcy. It was assumed, therefore, that theGDP has an immediate
effect on theinterest rate, theexchange rate and on thenumber ofbankrupt entities. Theinterest rate, inturn, has an immediate
effect on theexchange rate and thenumber ofactors, while theexchange rate on thenumber ofentities inbankruptcy.1
GDP number of bankruptcies

real WIBOR3M rate number of bankruptcies


0,05

0,02

0
0

-0,02
-0,05

-0,04

-0,1

-0,06
real exchange rate PLN/EUR number of bankruptcies

number of bankruptcies number of bankruptcies

0,3

0,04
0,02

0,2
0
0

9
0,1

-0,02
-0,04

-0,06
0
*

The solid line represents themedian, whereas thedotted lines represent 0.16th and 0.84th quantiles.

Fig. 14.5. Quantiles ofthe posterior distribution ofthe response function ofthe number ofentities inbankruptcy torandom shocks
affecting domestic variables (size ofthe shock equals one posterior standard deviation)
Source: own study.
1

The order ofthe variables domestic GDP, basic interest rate, PLN/EUR exchange rate was adopted from an article by (Haug 2012).

184

For comparison, thecharts below show theresponse torandom shock functions, which have been normalized so that thesize
ofthe shock is equal to1.
real WIBOR3M rate number of bankruptcies

GDP number of bankruptcies


0,1

0
0

-2
-0,1
-4

-0,2

-6
real exchange rate PLN/EUR number of bankruptcies

number of bankruptcies number of bankruptcies

1,5

1
0,5

0
0

-0,5
0,5
-1
-1,6

0
0

The solid line represents themedian, whereas thedotted lines represent 0.16 and 0.84 quantiles.
th

th

Fig. 14.6. Quantiles ofthe posterior distribution ofthe response function ofthe number ofentities inbankruptcy torandom shocks
affecting domestic variables (unit shock)
Source: own study.

Only responses ofthe number ofbankruptcies todomestic shocks are important. Thescores ofresponses ofthe remaining variables
toshocks are associated with a very high degree ofuncertainty. However, most ofthe posterior probability mass fortheresponse
function ofbankruptcies toan increase inthereal interest rate remains on thenegative side, which makes itmore likely that
themodel shows a weak negative correlation between theinterest rate and thenumber ofinitiated bankruptcy proceedings.
Assuming that theresponse function remains at thelevel ofthe median posterior distribution, we can conclude that a single
isolated (i.e. assuming that other variables remain constant) WIBOR3M rate increase by one percentage point will result ina decline
inthenumber ofbankruptcies by approximately 3.8% inthesame period. Themost noticeable response occurs after one quarter
(5.7%) and then gradually disappears (to about 1.4% after 4 years). Thesupply ofnew information slightly strengthened
theassessment ofthe response intensity ofthe number ofbankruptcies tointerest rate changes.
It can also be argued that GDP growth is very likely toresult ina decrease inthenumber ofenterprises at risk ofbankruptcy.
According tothemedian ofthe posterior distribution, a one-percent increase inGDP results ina decline inthenumber
ofbankruptcies by approximately 1.9% inthesame period. In subsequent quarters, theobserved relationship gradually diminishes
(non-monotonically) toreach thelevel of0.2% after 5 years. No significant changes with respect totheresults presented
intheprevious report have been observed.
The effect ofa one-percent increase inthenumber ofbankruptcies inthenext quarter drops toabout 0.6 %. In thesubsequent
period itincreases slightly (to 0.7 %) and then diminishes toabout 0.3% after 5 years. Despite thesupply ofnew information,
thereaction has not changed ascompared with theresults obtained intheprevious edition ofthe report. Themedian ofthe
posterior distribution ofthe number ofbankruptcies toPLN/EUR exchange rate shocks response function is close tozero.
Theresponse ofthe number ofbankruptcies toa one-time increase ofvariables representing thesituation intheworld was also
analysed (Figure 14.3).

185

Table 14.6. Quantiles ofthe posterior distribution ofthe response function ofthe number ofentities inbankruptcy torandom
shocks affecting variables related totheinternational situation (shock size equals one posterior standard deviation
foreach variable, which entails an increase inthelogarithm (ln) ofthe EUR/USD exchange rate by about 0.044 inthezero
period, thelogarithm ofGermanys GDP by 0.010 and thelogarithm ofthe price ofcrude oil by 0.162).
IRF
Horizon

EUR/USD no ofbankruptcies

Germanys GDP no ofbankruptcies

Price ofcrude oil no ofbankruptcies

Quantile

Quantile

Quantile

0,16

0,5

0,84

0,16

0,5

0,84

0,16

0,5

0,84

0.010

0.043

0.078

-0.063

-0.031

0.002

-0.017

0.015

0.047

0.001

0.025

0.051

-0.043

-0.020

0.001

-0.022

0.011

0.044

0.006

0.032

0.059

-0.049

-0.024

0.001

-0.024

0.009

0.041

0.002

0.027

0.054

-0.044

-0.020

0.002

-0.028

0.008

0.044

0.002

0.028

0.056

-0.044

-0.019

0.003

-0.032

0.007

0.046

0.000

0.026

0.054

-0.043

-0.018

0.004

-0.035

0.007

0.049

-0.001

0.026

0.055

-0.042

-0.017

0.005

-0.039

0.007

0.051

-0.003

0.025

0.055

-0.041

-0.016

0.006

-0.041

0.006

0.054

-0.005

0.024

0.055

-0.041

-0.015

0.006

-0.043

0.006

0.056

-0.006

0.024

0.055

-0.040

-0.014

0.008

-0.044

0.006

0.057

10

-0.008

0.023

0.055

-0.040

-0.014

0.008

-0.047

0.005

0.058

Shock size one posterior standard deviation, which entails an increase in the logarithm of the EUR/USD exchange rate by about 0.044 in the zero
period, the logarithm of Germany's GDP by 0.010 and the logarithm of the price of crude oil by 0.162
ln (EUR/USD) ln (number of bankruptcies)
0,1
0,08

ln (Germany's GDP) ln (number of bankruptcies)


0,02

ln (price of oil) ln (number of bankruptcies)


0,1

0
0,06 0
-0,02

0,06

0,05

0,04
-0,04

0,02

-0,06

0
-0,02

-0,05

-0,08

Unit shock
ln (EUR/USD) ln (number of bankruptcies)
2
1,5

ln (Germany's GDP) ln (number of bankruptcies)


2

0,4

0,2

0
1

-2

0,5

-4

ln (price of oil) ln (number of bankruptcies)

-6
0

-0,5

-0,2

9
-0,4

-8

*
The solid line represents themedian, whereas thedotted lines represent 0.16th and 0.84th quantiles.
Fig. 14.7. Quantiles ofthe posterior distribution ofthe response function ofthe number ofentities inbankruptcy toa one-time
increase invariables describing Polands economic environment

Source: own study.

As is thecase with therelationships described above, theimpact assessments ofselected parameters describing theglobal
environment on thenumber ofbankruptcies are associated with a considerable uncertainty. Within this group ofmodels, itis
impossible tojudge thestrength and direction ofthe impact ofcrude oil prices on thenumber ofenterprises inbankruptcy,
although theshift ofthe posterior distribution slightly above zero suggests that a positive correlation is more likely, therefore an
increase inthecrude oil prices is slightly more likely toresult inan increase inthenumber ofbankruptcies.

186

A negative correlation can be observed between thenumber ofbankrupt enterprises inPoland and Germanys GDP. After one
quarter, a one-time, isolated, one-percent increase inGermanys GDP results ina decline inthenumber ofbankruptcies by about
2%. Thestrength ofthis response is maintained inthesubsequent three quarters, and then slowly diminishes to0.8% after 5 years.
Conversely, a positive correlation can be observed between thenumber bankruptcies inresponse toan increase intheEUR/USD
exchange rate: theweakening ofthe euro against thedollar increases thenumber ofbankruptcies, forexample, an increase by
1% results inan increase inthenumber ofbankruptcies by approximately 1% inthesame period. After 5 years, thestrength ofthis
correlation still exceeds 0.4%. Both above-mentioned correlations appear tobe consistent with intuition.
It should be emphasized that those relationships may be helpful inpredicting thedirection ofchange inthenumber ofbankrupt
entities inresponse tochanges observed intheanalysed economic categories, and, assuch, may provide useful insight todecisionmakers. Thepicture can be made more complete by invoking historical analyses, i.e. a forecast error variance decomposition, and
simulations ofthe hypothetical paths excluding shocks intheanalysed economic data.
Arranging thevariables according tothepattern adopted fortheidentification ofrandom shocks, below theabove-mentioned
analyses are presented with respect tothenumber ofbankruptcies and thegross domestic product.
Table 14.7. Forecast error variance decomposition forthenumber ofbankruptcies and GDP
Number ofbankruptcies (ln)
Share (%)

100%

Crude
German
PLN/
oil
GDP WIBOR3M
Bankruptcy
PKB
EUR
price

Horizon

EUR/
USD

4.37

2.95

1.85

2.13

1.76

1.47

85.46

4.46

3.09

2.61

2.08

3.34

1.50

82.92

4.57

3.15

2.80

2.06

3.54

1.51

82.39

4.68

3.20

3.19

2.03

3.93

1.55

81.43

90%
80%

No of bankruptcies

70%

PLN/EUR

60%

WIBOR3M

50%

4.81

3.23

3.58

2.01

4.20

1.59

80.59

40%

4.94

3.27

4.04

1.99

4.51

1.64

79.61

30%

5.08

3.30

4.52

1.97

4.79

1.71

78.64

20%

5.22

3.32

5.01

1.95

5.08

1.78

77.63

10%

5.37

3.35

5.51

1.93

5.35

1.87

76.63

0%

10

5.52

3.38

5.99

1.92

5.61

1.95

75.63

GDP
Crude oil price
Germany's GDP
EUR/USD

7 8

9 10

Over 80% ofthe forecast number ofbankruptcies error variance forthefirst period is constituted by domestic shocks. In subsequent periods,
their share gradually decreases, reaching thelevel ofabout 75% inthe10th quarter. Thehighest increase inexplaining thevariance intheforecast
number ofbankrupt enterprises is related totheshocks related totheinterest rate from approximately 1.8% inthefirst quarter to5.6%
inthetenth quarter, and theprice ofcrude oil (from 1.85% to6%). Theoverall share ofinternational variables inexplaining theforecast error
variance increases from 9% intheinitial period up toabout 20% at 5 years, whereas theshare ofdomestic variables (excluding domestic shocks)
increases from 5 to12.5%.
In theperiod under consideration, theweight ofGDP-related innovations and thePLN/EUR exchange rate remained close to2%.
The results discussed do not differ significantly from those obtained intheprevious edition ofthe report.
An additional analysis ofthe importance ofshocks originating from thedifferent economic parameters inexplaining change inthenumber
ofbankruptcies was based on a simulation ofhypothetical bankruptcy paths after excluding theshocks associated with themodelled variables.
Such a historical analysis combines theknowledge resulting from theshock response function, variance decomposition and thepathways taken
by individual categories included inthemodel.
Shocks are reset tozero starting with theninth quarter inthesample.

187

Excluding shocks related to the EUR/USD exchange rate

Excluding shocks related to Germany's GDP

1
0.8
0.6
0.4
0.2
0
-0.2
-0.4

Excluding shocks related to crude oil prices

1
0.8
0.6
0.4
0.2
0
-0.2
-0.4

1
0.8
0.6
0.4
0.2
0
-0.2
-0.4

2013

2012

2010

2011

2013

2012

2011

2010

2009

2008

5
2013

2012

2011

2010

2009

2008

2007

2006

4.5

2013

2012

2011

2010

2009

2008

2007

4.5

5.5

2005

2004

5.5

1
0.8
0.6
0.4
0.2
0
-0.2
-0.4

6.5

2003

2006

2007

Excluding shocks related to the PLN/EUR exchange rate

6.5

2005

2006

4.5

Excluding shocks related to the interest rate

2004

2009

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

4.5

5.5

2005

2004

5.5

1
0.8
0.6
0.4
0.2
0
-0.2
-0.4

6.5

2003

2003

2008

Excluding shocks related to GDP

6.5

2003

2007

4.5

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

4.5

5
2006

5.5

2005

5.5

2004

1
0.8
0.6
0.4
0.2
0
-0.2
-0.4

6.5

2003

6.5

Excluding shocks related to the number of bankruptcies

1
0.8
0.6
0.4
0.2
0
-0.2
-0.4

6.5
6
5.5
5
2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

4.5

The solid line represents theactual path ofthe number ofbankruptcy petitions, thedotted line a hypothetical evolution trajectory assuming
no individual shocks intheperiod 2004q42013q4 (left axis), while thebars represent thedifferences between theactual and simulated values
(right axis).
In accordance with thedecomposition ofthe variance, thegreatest differences totheactual plot ofthe path are observed after theexclusion
ofdomestic shocks. Thegreatest differences between thereal plot and all theseven simulated plots can be observed in20072009 and inmost
cases theobserved differences are negative. Eliminating theshocks associated with theEUR/USD exchange rate causes themost perceptible
negative differences at theturn of2007/2008 and positive ones in2013. Theformer period was dominated by negative shocks, which confirms
thepositive relationship between theaverage number ofbankruptcies and thepath observed. Theelimination ofshocks caused by changes
inGermanys GDP in20062009 and again in201120012 results inan increase inthehypothetical number ofbankruptcies inrelation tothose
actually observed. Thegreatest negative differences can be seen in2008, which is associated with thehigh rate ofGDP growth inGermany
inthequarters preceding theunder consideration. Thenegative growth rate, which was observed in2009, resulted ina decrease ofthis
difference leading toa reversal ofthe relationship in2010. Subsequent recovery of20112012 improved thesituation ofenterprises.

188

In both above-mentioned cases, what appears tobe mysterious is thelast year ofthe survey, inwhich theobserved differences have theopposite
sign totheexpected one.
The exclusion ofshocks due tothePLN/EUR exchange rate results inan increased number ofbankruptcies, which is particularly evident in2005
2011 and inthefourth quarter of2012. Although theresulting negative differences are conspicuous, when we take into account both thepath
ofthe exchange rate and theestimated shocks, itis difficult toindicate thedirection ofthe relationship, which is also reflected intheassessment
ofthe uncertainty response function toshocks due tothePLN/EUR exchange rate. Since thepopulation ofbankrupt enterprises includes both
net importers and net exporters, theobserved uncertainty asto thedirection ofthe relationship appears tobe natural.

Gross Domestic Product (ln)


Share (%)
Horizon EUR/
USD
1
2
3
4
5
6
7
8
9
10

German Oil
GDP
price

4.73
4.83
4.89
4.90
4.91
4.91
4.91
4.92
4.93
4.95

2.38
2.51
2.61
2.69
2.76
2.83
2.88
2.93
2.98
3.02

2.76
3.10
3.35
3.63
3.93
4.24
4.57
4.91
5.24
5.57

GDP

WIBOR3M

86.36
84.91
83.40
82.07
80.72
79.40
78.10
76.84
75.62
74.45

1.03
1.75
2.34
2.94
3.53
4.09
4.63
5.14
5.62
6.06

PLN/
Bankruptcies
EUR
0.10
0.15
0.20
0.25
0.31
0.38
0.44
0.51
0.58
0.66

2.64
2.75
3.19
3.50
3.84
4.15
4.46
4.75
5.03
5.29

100%
90%
80%

No of bankruptcies

70%

PLN/EUR

60%

WIBOR3M

50%
40%

GDP
Crude oil price

30%

Germany's GDP

20%

EUR/USD
gross domestic product (ln)

10%
0%

7 8

9 10

In theperiod under consideration, theshare ofdomestic shocks intheGDP forecast error variance exceeds 70% (ranging from 86% forthefirst
quarter forecast toabout 75% forthe2.5-year forecast horizon). Theshare ofshocks related tothePLN/EUR exchange rate is low 0.1%0.7 %
which indicates little relevance ofthis parameter forGDP changes. Thelonger theforecast horizon, thegreater theimportance ofinterest rate
shocks (increase from 1% to6 %) and those involving thenumber ofbankruptcies (increase from 2.6% to5.3%). Thetotal share ofthe remaining
variables increases from 8% toabout 13.5% forthe10-quarter forecast horizon and 17% forthe5-year forecast, with thelargest increase observed
inthecase ofshocks related totheprice ofcrude oil.
Compared with theresults ofthe previous report, we have noted a slight decrease intheimportance ofdomestic shocks inexplaining
thevariance inthefirst quarters ofthe forecast.
In order toobtain a more in-depth analysis ofthe impact ofshocks originating from different economic parameters on GDP, a simulation
ofa hypothetical GDP path was run totheexclusion ofshocks associated with theanalysed variables, just aswas done forthenumber
ofbankruptcies. Shocks are reset tozero at thebeginning ofthe ninth quarter inthesample.
Excluding shocks related to Germany's GDP

Excluding shocks related to the EUR/USD exchange rate

5.6
5.4
5.2

Excluding shocks related to crude oil prices

2013

2012

2011

2010

2009

2008

2007

0.15
0.1
0.05
0
-0.05
-0.1
-0.15

5.6
5.4
5.2

189

2013

2012

2011

2010

2009

2008

5
2007

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

5.2

6
5.8

2006

5.4

0.15
0.1
0.05
0
-0.05
-0.1
-0.15

2004

5.6

Excluding shocks related to GDP

2003

6
5.8

2006

5
2005

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

5.2

0.15
0.1
0.05
0
-0.05
-0.1
-0.15
2004

5.4

6
5.8

2005

5.6

0.15
0.1
0.05
0
-0.05
-0.1
-0.15

2003

6
5.8

Excluding shocks related to the interest rate

Excluding shocks related to the PLN/EUR exchange rate

5.2
2013

2012

2011

2010

2009

2008

5
2007

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

5.2

5.4

2006

5.4

5.6

2005

5.6

0.15
0.1
0.05
0
-0.05
-0.1
-0.15

6
5.8

2004

0.15
0.1
0.05
0
-0.05
-0.1
-0.15

2003

6
5.8

Excluding shocks related to the number of bankruptcies

6
5.8
5.6
5.4

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

5.2

0.15
0.1
0.05
0
-0.05
-0.1
-0.15

The solid line represents theactual path ofthe number ofbankruptcy petitions, thedotted line a hypothetical evolution trajectory assuming
theabsence ofshocks in2004q42013q4 (left axis), while thebars represent thedifferences between theactual and simulated values (right axis).
The greatest differences between hypothetical GDP paths and theactually observed ones emerge after theexclusion ofdomestic shocks,
thebasic interest rate and Germanys GDP. Resetting domestic shock tozero shifts theentire plot down with theprovision that in2012
and inthefirst two quarters of2013, theobserved difference is clearly decreasing. In this period, we can observe a decline inGDP growth;
consequently, theidentified shocks are negative.
The elimination ofshocks related toGermanys GDP also leads toa downward shift ofthe path ofPolands GDP, with thegreatest differences
observed in20062008, in20102011 and again inthelast two quarters of2013. In other periods, theactual values ofPolands GDP are located
above thehypothetical values, but thedifferences are considerably smaller. These periods correspond with theyears ofa marked slowdown
intheGerman economy.
Likewise, positive differences arise after theexclusion ofshocks related totheshort-term interest rate. Thedifferences disappear in20112012,
when therates were low and mainly negative shocks were identified intheequation.

Kamil Fijorek, Jarosaw Kaczmarek, Konrad Kolegowicz, Pawe Krzemiski

Chapter 15
BANKRUPTCY RISK AT ENTERPRISE LEVEL A MICROECONOMIC
ASSESSMENT
15.1. Introduction
The aim ofassessing theoverall financial standing ofenterprises inthemicroeconomic component consists ineffectively identifying
thesymptoms oftheir steadily deteriorating economic condition, with a view tomapping out theareas most vulnerable tobankruptcy,
while making use ofthe well-established analytical methods, instruments and procedures, inclusive ofbankruptcy prediction
modelling (i.e. theSZU models estimated on theaggregated values inthefirst stage ofthe study, ascompleted up tothe2nd half
of2012, and then inthesecond stage, pursued asofthe 1st half of2013, theMW models estimated against therespective values
pertaining totheset ofnearly fifty non-financial enterprises studied and included inpublic domain statistics). All theabove facilitated
thequantification, analyses and theactual estimation ofpotential bankruptcy risk affecting theenterprises under consideration.
The basic analytical cross-section applied throughout thepresent study consists intheappraisal ofenterprises interms ofspecific
type ofbusiness activity (manufacturing, trade, services) and their respective sizes (small, medium-sized, large). Some additional
profiles (structures) have also been addressed indue consideration ofNACE terminology (inclusive ofsections and divisions), their
respective legal status and regional location. Their presentation inthis study is somewhat limited, however, itremains inline with
theadopted editorial constraints.1

15.2. Financial standing ofenterprises


The assessment offinancial standing ofenterprises was carried out against thekey components ofeconomic potential and
theeconomic and financial indicators deemed viably predictive ofthe degree ofbankruptcy risk, while making due reference
torelevant values within thecorresponding period ofthe previous year. Theactual analytical scope ofthe economic potential was
augmented with thedescriptive statistics pertaining tothevariability ofthe analysed values, expressed

through standard deviation


and minimum and maximum values, asreported throughout theperiod under consideration. Theadopted set ofanalytical
cross-sections, apart from theanalytical total, consists intheir diversity interms ofthe actual type ofbusiness activities pursued
(manufacturing, trade, services), and therespective sizes (small, medium-sized, large).
In terms ofthe revenue generated from sales, capital expenditure, value ofassets, and net financial result actually achieved,
manufacturing firms had by far thelargest share, followed by theservices sector (with theexception ofrevenues and thenet
profit) and trading firms. Both intheaggregated group ofenterprises, aswell asin theones differentiated by therespective types
ofbusiness activity pursued, theactual change trends inpertinent variables throughout theperiod under study (1st half of2007
first half of2013) were similar, even though they still differed interms ofintensity and time ofoccurrence. Following theadverse
impact ofthe 2008 economic crisis, by mid-2010 thecapital expenditure, revenues and net financial results were reported todrop
(even though overall value ofassets was not subject toany significant changes throughout theperiod under consideration).
In thecase ofcapital expenditure, this trend occurred earliest and lasted longest intheservices sector, i.e. asearly asofmid-2008,
inthetrade sector asofmid-2009, whereas inthemanufacturing sector itcame latest and had theshortest duration, i.e. asofmid2009. Theabove notwithstanding, theservices sector was theonly one not tohave experienced any drop intherevenues, whereas
some reduction inthenet financial result was encountered inthelate 2008 and late 2010 only (relatively themost favourable
situation among theremaining types ofbusiness activity).
An appreciably worse situation inthis respect befell thetrading entities (decrease intherevenues throughout 2009, poorer financial
result inthelate 2008, early 2009, and early 2010), aswell asthemanufacturing ones (despite a drop inrevenues intheearly 2009
only, these entities experienced not only a reduction inearnings throughout 2008 and inthe1st half of2010, but also loss-making
business activities inthe2nd half of2008).
As ofearly 2010, intheservice companies (as thefirst ones), and asoflate 2010 inthetrading and manufacturing companies,
thesituation improved, although thesigns ofdeterioration, i.e. a perceptible drop inthevalue ofpertinent indicators, began
1

A broader presentation of the results of analysis and evaluation of economic and financial situation and the exposure to the degree of
bankruptcy risk of enterprises, completed with the aid of widely acknowledged economic indicators, methods and tools, models for predicting
bankruptcy assuming the analytical cross-sections, as defined for the period spanning early 2007 early 2013 and the third quarter of 2013,
can be found in: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (June 2011 April 2014).

191

toreappear asearly asin the2nd half of2011. Thefirst and most prominently visible adverse changes, still tobe encountered
up tomid-2013, were recorded intheservices sector, unlike inthetrading and manufacturing companies, where certain signs
ofimprovement were noted inthe1st half of2013, i.e. theyear rounding off thepresent analytical span.
In thecompanies highlighted interms oftheir respective sizes with regard totheindicators under study (revenues, capital
expenditure, assets, and net profit) thelarge companies had by far thelargest share, followed by medium-sized and small
firms. Amongst thelarge entities, thegreatest share is claimed by manufacturing firms (followed by services and trading ones),
inthemedium-sized ones manufacturing and trading companies (service companies claiming thesmallest proportion),
whereas inthesmall ones itwas theservices sector that claimed thelions share (followed by thetrading and manufacturing
ones). Thesymptoms ofdeteriorating economic climate occurred theearliest inthemedium-sized entities (reduced net profits
intheperiod spanning early 2008 early 2009, capital expenditure asofmid-2008 until early 2010, revenues asofmid-2009).
In small entities, adverse changes were noted inthe1st half of2009 (in thecase ofcapital expenditure up totheend of2010),
whereas inthelarge entities this happened latest and had theshortest duration, i.e. asofmid-2009 until early 2010 (although
accompanied by loss-making business operations inthelate 2009). In thelarge entities, a period ofrelative improvement, lasting
from mid-2010 until early 2012, was followed by a renewed deterioration intheir financial standing. In thesmall entities thesigns
ofimprovement occurred a little earlier, i.e. asofmid-2009, although they did not refer toall economic indicators, aswas thecase
with thelarge, although with perceptible signs ofimprovement intheearly 2013. By far theworst situation inthis respect
characterized themedium-sized entities (further reduction ofnet profits occurred asearly asin mid-2010 and lasted until theend
of2012, reduction inassets asofmid-2010 until early 2011, capital expenditures throughout 2012, revenues from mid-2012
until early 2013).
Total

20

Manufacturing

25
20
15
10
5

15
10
5
0 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013
-5
-10

0 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013
-5
-10
-15
-20

-15

-25

-20
Revenues

Capital expenditure

Assets

Revenues

Trade

50

Assets

Services

30

40

Capital expenditure

20

30

10

20
10
0 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013

-10

0 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013
-10
-20

-20

-30

-30

-40

-40
Revenues

Capital expenditure

Assets

Revenues

Capital expenditure

Assets

Fig. 15.1. Rate ofchange insales revenues, capital expenditure and assets owned by enterprises inPoland from the1st half of2011
tothe1st half of2013 intotal and by respective types ofbusiness activity (in %, y/y)
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

A closer scrutiny ofthe variability ofthe basic economics indicators implies that itwas inthemanufacturing activity that thehighest
average values ofall variables under study were encountered (revenues, capital expenditure, assets, net financial result), aswell
astheir highest variability. Thehighest average values are consistent with thefavourable situation inthemanufacturing sector,
although their high variability indicates at thesame time certain instability inthis particular sector ofnational economy, which is
only further corroborated by thenegative values inrecorded earnings (loss-making periods) inthemanufacturing activities only.
In thetrading business, on theother hand, thelowest average values were noted incapital expenditure and assets. Trading was
also characterized by thelowest variability ofthe indicators under study, thus effectively corroborating rather stable situation inthis
sector. Service activities were characterized by thehigher mean values than inthecase ofthe trading activities, with theexception
ofthe sales revenue.
192

Capital expenditure

Revenues

600

30

400

20

200

10

0
Manufacturing

Trade

Services

Manufacturing

Assets
30

1200

25

1000

20

800

15

600

10

400

200

Notes:

Services

Financial result

1400

Trade

-5
Manufacturing

Trade

Services

Manufacturing

Trade

Services

standard deviation; mean value represented by the centre of the figure


minimum and maximum values, respectively

Fig. 15.2. Descriptive statistics on thevariation insales revenues, capital expenditure, assets and financial results ofenterprises
inPoland by type ofbusiness activity from the1st half of2011 tothe2nd half of2013 (in billions ofPLN)
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

The analysis ofthe rate ofchange intheoverall financial standing index (WOSF value) forcompanies intotal, and asstratified by
respective types ofbusiness activity, brings similar indications aswhen making use ofbasic economic indicators oftheir economic
potential. It may infact be indicated that thefallout ofthe 2008 financial crisis initially affected most thetrading and manufacturing
sectors (WOSF value was significantly reduced intheperiod spanning mid-2008 and early 2009), whereas theservices sector was
least affected. As a result, theimprovement inoverall financial situation was first recorded by theservices sector, i.e. inthe1st half
of2009, while by thetrading and manufacturing sectors inthe2nd half of2009. This situation prevailed until 2011, whereupon
thesymptoms ofa deteriorating overall financial situation ofthe enterprises re-emerged, lasting until mid-2013. Thefirst one
toexperience theeconomic adversity was theservices sector (as ofmid-2011), then themanufacturing (as ofthe 1st half of2012)
and thetrading one (as ofthe 2nd half of2012) followed inquick succession. Among thelatter, certain signs ofimprovement were
noted intheearly 2013. In themanufacturing sector thedownward trend was effectively arrested, whereas intheservices one
theWOSF value was reported todrop (by far theleast favourable situation).
Having made due allowances totherespective sizes ofthe enterprises, since early 2012 there seems tobe an appreciable
improvement intheoverall financial situation ofsmall firms, unlike thelarge ones inwhich by theend of2011 a perceptible
deterioration was recorded. In this respect, themedium-sized firms seemed tohave had thebumpiest ride, i.e. assince thelate
2009 (with a notable exception ofearly 2011) a consistent reduction intheir WOSF value was reported.
Assessment ofthe variability rate intheoverall financial standing index (WOSF) ofcommercial enterprises indicates its highest
average and maximum values with regard tothetrading activities. Theminimum WOSF value actually reached inthetrading
activities was greater than themaximum value attained inall other types ofbusiness activity, which only goes tocorroborate
themost favourable overall situation inthis particular group ofenterprises. Thelowest mean WOSF value, while at thesame time
its smallest variability characterized thebusiness activities pursued by theservices sector, where its lowest value was also achieved.
Analysis ofthe variability ofthis index gives credence totheearlier analyses and appraisal ofthe companies overall financial
standing, aspursued with theaid ofbasic economic indicators oftheir economic potential.

193

1.4

15
10

1.2

5
1

0 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013
-5

0.8
0.6
0.4
Total
Trade

Total

Manufacturing
Services

Manufacturing

Trade

Services

Notes: asfor Figure 15.2.


Fig. 15.3. Therate ofchange intheoverall financial standing index (left panel, expressed in%, y/y) and its variability (right panel)
inenterprises inPoland from the1st half of2011 tothe2nd half of2013: total values and by type ofbusiness activity
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

The figure below shows therate ofchange inthevalues ofselected key economic and financial indicators intheperiod
spanning early 20122013, forthebusiness enterprises intotal, and then divided by therespective types ofactivities that may be
ofindubitable significance intheoverall assessment oftheir exposure tothedegree ofbankruptcy risk.
Total

Manufacturing
1/WZO
5%

1/WZO
5%
ROA

0%

ROA

WZOZ

-5%

-5%

-10%

-10%

-15%

PM

III 2012

WPB

-20%

ROOS

PM

WPS

CGOTr
IIIIV 2012

III 2013

III 2012

5%
0%

ROA

WZOZ

-5%
-10%
-20%

PM

III 2012

ROOS

WPB

CGOTr
IIIIV 2012

10%
0%
-10%
-20%

PM

WPS

CGOTr
IIIIV 2012

WPS

III 2013

III 2013

WZOZ

-30%
-40%

-15%
ROOS

WPB

Services
1/WZO
30%
20%

Trade
1/WZO
10%
ROA

WZOZ

-15%

-20%

ROOS

0%

III 2012

WPB

WPS

CGOTr
IIIIV 2012

III 2013

Note: on thecharts, inorder tostandardize thedirections ofchanges intheindicators under study, CCC value was itemised and theinverse value
ofthe TDR was adopted.
Fig. 15.4. Therate ofchange infundamental groups ofeconomic and financial indicators inenterprises inPoland from the1st half
of2011 tothe1st half of2013: total and by type ofbusiness activity
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

194

Within theanalytical period spanning the1st half of2007 the1st half of2013, thehighest level oftotal debt (TDR) was shown
by thetrading companies, whereas thelowest one by themanufacturing ones. A systematic and significant increase indebt
over thesame period (with theexception ofthe 4th quarter of2012) characterized firms intheservices sector, notably irrespective
oftheir sizes, asopposed toother types ofbusiness activity and company sizes. In thetrading and themanufacturing enterprises,
thechanges intotal debt alternated, although without any significant changes intheactual total debt level (excluding themediumsized manufacturing firms, inwhich theinitial rise intotal debt until theend of2011 was eventually arrested). This was not without
an impact upon theactual level ofdebt inthecompanies intotal, inwhich a slight increase was also duly noted.
Concomitant adverse changes consisted inreduced debt sustainability (DSI) intheenterprises pursuing all types ofbusiness
activity, whereas themost, i.e. nearly by one-half, intheservices sector (companies ofall sizes) and themanufacturing one with
regard tothemedium-sized companies. Thereduction ofdebt sustainability was noticeable inthecompanies ofall sizes, especially
inthemedium-sized entities.
These disturbing signals with regard tothechanges intotal debt (TDR) and theability toservice it(DSI) were somewhat alleviated
by theresults ofthe analysis intheareas ofcurrent and quick liquidity (CR and QR), which, despite its slight deterioration up
totheearly 2009 forthecompanies pursuing all types ofbusiness activity, underwent a systematic improvement intheperiods
spanning late 2009 early 2011, and then intheearly 2013. This was most noticeable intheservices sector, then inthetrading
firms, but least inthemanufacturing sector. In this particular respect, small companies seemed tolead theway, asopposed
tothemedium-sized and large ones, inwhich only a slight improvement was noted.
Along with theimprovement inliquidity parameters within theperiods indicated with regard toall types ofbusiness activity and
company sizes, thewaiting time forcash (CCC) was slightly reduced, which also was most noticeable inservice firms ofall sizes
(perceptibly less so inthetrading firms, with no appreciable changes inthemanufacturing ones). It is hard todetermine theactual
sustainability ofthe observed improvement, though, asparticularly intheservice companies itwas achieved through theextension
ofdeadlines forthepayment ofshort-term receivables (a negative trend forservice firms).
In themanufacturing firms, unlike intheservices and trading ones, asset productivity (AP) showed a downward trend within
theperiod under consideration, whereas intheservices and trading companies its value fluctuated. There was no correlation
whatsoever between those trends and theactual sizes ofthe companies, even though itwas inthelarge trading firms that
a perceptible increase intheproductivity ofassets was noted, asopposed tothesmall and medium-sized ones, where itwas
reduced.
Despite thefact that during theperiod under consideration inneither group ofcompanies broken down by respective types
ofbusiness activity was there any loss-making activity reported, theoperating return on sales (OROS) and thereturn on assets
(ROA) inall theentities (irrespective ofthe actual type ofstratification) exhibited fluctuations, with a discernible downward trend
(to thegreatest extent by almost half ofthe service companies, and least inthemanufacturing sector). Slight improvement
inthereturn on sales (very much like inthecase ofsome ofthe above referenced indicators) was observed inthe1st half of2013
with regard tothemanufacturing and theservices sectors.
The highest average total debt radio (TDR) and its low variability throughout theperiod under consideration pertained
tothetrading activities. Thehigh variability ofthe debt also applied totheservices sector, while its average value was at a much
lower level than inthetrading one. Themost favourable situation regarding theamount ofdebt throughout theperiod under
study pertained tothemanufacturing activities, asitwas characterized by thelowest average value ofthe TDR, and thehighest
value ofthe DSI. Theunfavourable situation inthetrading activities is corroborated by thelowest average liquidity ratios (CR and
QR) and thelow variability indicative ofstagnation. Thehighest average and maximum values ofthe CR and theQR were observed
intheservices sector.
In themanufacturing sector theaverage waiting time forcash (CCC) was at thehighest level, whereas its minimum values were
close totheaverage forthetrading activities. This type ofactivity was also characterized by thehigh average operating return
on sales (OROS) and assets (ROA). Thehigh average values ofboth asset productivity (AP) and operating return on sales (OROS)
inthetrading activities and their low variability corroborate thestable and favourable position within this sector. In theservices
sector theaverage cash turnover cycle was theshortest, and also accompanied by high variability. Furthermore, thelowest mean
value ofboth theasset productivity (AP) and thereturn on sales (OROS) was also observed there. High variability ofthe above
referenced indicators implies rather unstable situation intheservices sector throughout theperiod under consideration.
In summary, itmight well be pointed out that intheaftermath ofthe financial crisis of2008, and after a period ofdeteriorating
economic and financial situation offirms until 2009, a slow but steady improvement was encountered inthespecifically indicated
areas only (e.g. liquidity and efficiency ofoperation), aswell asonly within theperiods under consideration. It should be noted at
this point that within theactual period under consideration theunfavourable trends which occurred and persisted across different
types ofbusiness activity and respective company sizes were actually indicative ofthe increase indebt and a decrease intheability

195

TDR

0.7

DSI

0.35
0.3
0.25

0.6

0.2
0.15
0.5

0.1
0.05

0.4

0
Manufacturing

Trade

Manufacturing

Services

CR

1.8

Trade

Services

QR

1.2

1.6
1
1.4
0.8
1.2

Manufacturing

0.6
Trade

Manufacturing

Services

Trade

Services

Notes:
15.2. standardowe; rednia wyznaczona przez rodek figury
Uwagi:asfor Figure
odchylenie
Fig. 15.5. Descriptive statistics on thevariability ofTDR, DSI, CR and QR values inenterprises inPoland from the1 st half of2011
st
wartoci
i maksimum.
tothe1
halfminimum
of2013 by
type ofbusiness activity
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

CCCDays

30

AP

2.5

25
2
20
1.5

15
10

5
0

Manufacturing

Trade

0.5

Services

OROS

0.1

0.08

0.06

0.06

0.04

0.04

0.02

0.02

Manufacturing

Trade

Services

Trade

Services

ROA

0.1

0.08

Manufacturing

Manufacturing

Trade

Services

Notes: asfor Figure 15.2.


Fig. 15.6. Descriptive statistics on thevariability ofCCC (in days), AP, OROS and ROA intheenterprises inPoland intheperiod
spanning the1st half of2011 the1st half of2013 by type ofbusiness activity
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

196

toservice it(especially intheservices sector), aswell asofa reduction intheproductivity ofassets (with theexception ofthe large
trading firms), and thereturn on assets inall types ofbusiness activity and company sizes. Certain positive signs indicating slight
improvement intheselect areas interms ofrelevant economic and financial indicators regarding specific groups ofcompanies
may be observed only inthelast period under study, i.e. early 2013. It would be quite hard todetermine theactual sustainability
ofthose symptoms, though.

15.3. The key aspects of bankruptcy risk assessment


current models forestimating potential bankruptcy risk (MW), thesubject matter ofthe present study, are based on theactual sizes
ofindividual firms. In relation tothepreviously estimated and applied models based on theaggregated sizes ofthe enterprises
(SZU), no significant changes occurred intheindications forthedegree ofbankruptcy risk (as was also pointed out inChapter
10 ofthe present study), be that interms ofgeneral estimation, ortheone inconsideration ofthe respective types ofbusiness
activity. Theplots ofthe SZU and MW curves run quite close together (linear correlation coefficient r = 0.98), whereas thetrend
lines are almost parallel toeach other. Theresults obtained with theaid ofa new class ofmodels, and based on theindividual
values, corroborate thepreviously formulated conclusions which made use ofthe models based on theaggregate sizes, and thus
indicate their utility asviable predictors ineffectively assessing thedegree ofbankruptcy risk.
210

24%

23%
190
22%
170

21%

20%

150

19%
130
18%

110

III 2007 IIIIV 2007 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013 IIIIV 2013 III 2014
Degree of bankruptcy risk MW
Degree of bankruptcy risk SZU

17%

Trend line MW
Trend line SZU

Note: thetotal value was determined asa weighted mean (number ofemployees) out ofthe values calculated forrespective types ofbusiness
activity. Theblue dashed line represents thepredictions.
Fig. 15.7. Comparison ofthe degree ofbankruptcy risk faced by total companies from the1 st half of2011 tothe2nd half of2013
assessed with theaid ofthe following models: SZU (aggregated values) and MV (respective individual values)
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

The estimates ofthe degree ofbankruptcy risk were conducted using models specifically tailored totheactual type ofbusiness
activity pursued by firms (cf. Chapter 10). Theanalyses ofthe changes inthedegree ofbankruptcy risk broken down by respective
types ofbusiness activity, leading NACE divisions and therespective company sizes, were augmented with thepredictions
ofthe degree ofbankruptcy risk with respect toa one-year horizon.2 Additionally, with regard tothequantitative analysis ofthe
uncertainty associated with thepredicted degrees ofbankruptcy risk, theinstruments ofa scenario analysis were applied. Apart
from a neutral scenario, both a pessimistic scenario and an optimistic one were developed. For this purpose, theupper and
thelower confidence limits (80% value, adjusted foroutliers) fortheprojected magnitude ofthe degree ofbankruptcy risk were
used, developed inline with theanalysis ofhistorical deviations intheparameter inquestion relative tothepercentage ofcourt
bankruptcy proceedings.

These predictions are estimated on the basis of the variables determining the degree of bankruptcy risk using one-dimensional time series.
Abroader discussion of this issue is to be found in Chapter 10 of the present paper, as well as in: Kaczmarek, Kolegowicz, Krzemiski, Fijorek
(May 2011); Kaczmarek, Kolegowicz, Krzemiski, Fijorek (June 2011).

197

The analyses ofchanges inthedegree ofbankruptcy risk forenterprises intotal broken down by thetype ofbusiness activity
and respective sizes, retrospectively indicated an increase intheestimated measurement since the1st half of2007, although
with perceptible differences interms ofvalue and thenature ofMW changes pertaining toselected groups ofentities and their
respective sizes.
Since early 2007 until mid-2008 itwas themanufacturing companies that reacted fastest and strongest toan increase inbankruptcy
risk. Successive periods spanning the1st half of2009 1st half of2011 indicated a different, i.e. decreasing trend intheMW changes
inrelation tothetrading and service firms, reaching a low level intheearly 2011 level (MW value below 160). Despite a slight
increase inthedegree ofbankruptcy risk asofmid-2011, and theMW value reaching thewarning level (i.e. intherange of160
200), themanufacturing firms (very much like thetrading ones), both with regard tothepast and theprojected periods, were
characterized by overall stability. Thetrading firms also reacted toa rapid increase inthedegree ofbankruptcy risk spanning
mid-2007 and early 2009. At this point itshould also be noted that among theselected groups theentities pursuing trading
activities were actually characterized by changes ofa substantially different nature inthevariables under study resulting from
theseasonality factor, which, having been offset forthepurpose ofthe present assessment by thetrend line, indicated a slower
increase inthedegree ofbankruptcy risk since mid-2009, followed by stabilization, both fortheperiod spanning early 2012 and
early 2013, and theprojected periods studied (mid-2013 early 2014).
It should also be noted that despite theincreasing trend inthedegree ofbankruptcy risk forthetrading companies, ascompared
totheservices and manufacturing ones, that risk was thesmallest, while additionally assuming its low levels (below 160),
thesituation appeared favourable enough. Thegeneral conclusion that can be drawn fortheservice companies, despite their
rather sluggish first reaction, is a continuous and intense increase inthedegree ofbankruptcy risk within theperiod inquestion
and theachievement ofthe highest MW values (over 200 a high level) out ofthe three groups under study. This has been
particularly noticeable since mid-2010, and these results are owed primarily toprevalent situation inthegeneral construction
industry, transport, IT, and communications. Furthermore, forthecurrent period ofanalysis, i.e. early 2013, only theservices sector
was characterized by an increase inthedegree ofbankruptcy risk, and only forthat particular sector thechanges within theneutral
scenario projected by theend of2014 indicated a further increase inthebankruptcy risk. Accordingly, thesituation inthis particular
group offirms must be deemed asnegative.
Note: The value Total is determined asa weighted average (number ofemployees) out ofthe values calculated forrespective types
ofactivities. Thevalues fortheoptimistic scenario are marked ingreen, and forthepessimistic one inred.

210

Total

Manufacturing

200

190

180

170
160
150
140

130
110

290

120

III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III
2007 2007 2008 2008 2009 2009 2010 2010 2011 20112012 2012 2013 2013 2014
Services

III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III
2007 2007 2008 2008 2009 2009 2010 2010 2011 20112012 2012 2013 2013 2014
Trade

170
150

240

130
190
110
140

90
70

190
III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III
2007 2007 2008 2008 2009 2009 2010 2010 2011 20112012 2012 2013 2013 2014

III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III
2007 2007 2008 2008 2009 2009 2010 2010 2011 20112012 2012 2013 2013 2014

Fig. 15.8. Evolution ofthe degree ofbankruptcy risk from the1st half of2007 tothe1st half of2014 by type ofbusiness activity
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

198

The changes occurring within thebusiness environment prompted an altogether different reaction from thecompanies depending
on their respective sizes. Theresults yielded by theanalyses indicated that small entities were highly susceptible toadverse changes
intheir business environment, moreover, they appeared tobe unable toreact with sufficient flexibility when compared with larger
entities. They were characterized by multi-directional, alternating changes inresponse tothedegree ofbankruptcy risk, with its
strong increase intheearly 2010 followed by a decline intheamplitude offluctuations asofthe 1st half of2011, and a stabilisation
ofthe absolute values asprojected forthe2nd half of2013 and the1st half of2014, although without a prospective growth path
(with theexception ofthe trading sector and construction asa subdivision ofservices). Accordingly, their situation must be assessed
asneutral. Medium-sized enterprises, despite their alternating fluctuations intheperiod under study, demonstrated a relatively evenpaced increase intheMW value, with thearrested upward trend within thefinal period under study (i.e. theplot ofthe curve reflecting
thechanges intheexposure tothedegree ofbankruptcy risk is similar totheplot ofthe curve representing firms intotal).
The projected changes (from the3rd quarter of2013 tothe2nd quarter of2014) indicate stabilization and a departure from
thehistorical growth ofMW value, which prompts a careful transition from a negative assessment toa neutral one (it should
be pointed out that theresults achieved by themedium-sized entities are owed totheprevalent situation within theservices
sector, including construction industry ingeneral, forwhich theforecasts remain unfavourable). An altogether different response
toadverse changes inthebusiness environment broken down by company size was demonstrated by thelarge entities, which,
having noted a sudden rise intheMW value toa high level at thebeginning of2009, intheperiod from the2nd quarter of2009
totheend of2011 effectively mobilised their existing restructuring potential and managed tosignificantly reduce their exposure
tobankruptcy risk toa low level. From 2012 tothe2nd quarter of2013, thesituation oflarge entities, characterised by changes
intheMW value, indicated a steady increase intheir exposure tobankruptcy risk (approaching high levels), which was caused
by an overall deterioration ofthe situation intheservices sector (including IT services and communications). Theprojected
changes fortheperiod from the3rd quarter of2013 tothe2nd quarter of2014 indicate a likelihood ofstabilization inthedegree
ofbankruptcy risk, which is going tobe highly dependent, however, on thepositive changes within theservices sector. A truly telltale observation, though, consists intheconvergence ofthe degree ofbankruptcy risk among all thefirms regardless oftheir size.
Despite perceptible signs ofimprovement inthe3rd quarter of2013, theassessment ofthe degree ofbankruptcy risk both
retrospective and interms ofcurrent predictions serves asthebasis fora negative overall assessment, and well justifies sounding
thealarm bells with regard tothefinancial condition ofthe companies within thesector under review, astheexposure tothedegree
ofbankruptcy risk, asdetermined under theconditions ofthe early 2013, was higher both inrelation tothecorresponding period
ofthe previous year, and theprevious half-year.
Predictions regarding thelikely direction ofchanges inthedegree ofbankruptcy risk intheperspective ofup tomid-2014,
intheneutral scenario, suggest a slight change, aswell asenvisage thevalue in2014 tobe identical with theone ofthe early
2013, whereas under thepessimistic scenario intheearly 2014 thevalue ofthe parameter inquestion is likely tobe higher than that
ofthe early 2013 (maintaining a slight upward trend). Theoptimistic scenario, despite thesigns ofan overall upturn intheeconomy,
appears tobe even less likely. Theidentified increase inthedegree ofbankruptcy risk also affects a possible increase inthenumber
and percentage ofthe instigated court insolvency proceedings within thetwo successive half-years, relative tothecurrent period
(early 2013). It should be noted, nevertheless, that theactual extent ofthose changes will depend on multiple factors, i.e. thefiled
bankruptcy petitions ultimately running out, changes intherelationship between thenumber ofactually instigated insolvency
proceedings and thenumber offiled petitions, theduration offormal proceedings aswell asthenumber ofthe already processed
cases inthemost vulnerable sectors (mainly intheconstruction industry ingeneral, transport and tourism).
Apart from theestimated MW value calculated with theaid ofmodel variables, thepossibility tohave those values determined
forrespective firms (for theprevious 11 research editions) effectively extended thescope ofthe presented assessment ofthe
degree ofbankruptcy risk, both interms ofthe total value foreach specific type ofbusiness activity and indetail fortherespective
sections ofbusiness activities and NACE. Due tothelimited framework ofthe present study, thedescriptive statistics ofthe
bankruptcy risk forthemanufacturing firms from the1st half of2007 early 2013 are shown below.
The average value ofthe degree ofbankruptcy risk forthemanufacturing firms intheearly 2013 increased from 163 by 3.1% year on
year. Thevalue ofthe ninth decile ofthe degree ofbankruptcy risk increased within thesame period from 231 to261 (i.e. by 13.0%).
This means that theoverall situation ofthe manufacturing firms facing a relatively high bankruptcy risk has actually deteriorated
over thelast year by more than was thecase with themanufacturing companies facing medium bankruptcy risk.
The distribution ofthe degree ofbankruptcy risk is characterized by a significant right-sided asymmetry. In this type ofdistribution,
themean value is clearly greater than themedian, i.e. thecentral value. In right-sided asymmetric distributions, themean value
asa rule is strongly affected by a relatively small part ofthe population under study boasting by far thehighest values ofthe
analysed characteristics, i.e. theactual degree ofbankruptcy risk. It should be noted that this rather general observation applies
not only tothemanufacturing firms, but also toother business activities under study and therespective NACE divisions/sectors
ofthe domestic economy.
199

A
190

320
300
280
260
240
220
200
180
160
140
120

76
74

180

72

170

70

160

68
66

150

64

140

62
60

Mean value
Standard deviation

130
120

IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III
2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013

58

Mean value
pct 90
Max./Mean value

27
26
25
24
23
22

IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III
2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013

21

20%

III
2013

15%
10%

III
2012

5%
0%

300

E
III 2013
III 2012

150
100

pct 50

pct 60

pct 70

pct 80

pct 90

8
007 008
IV 2 III 2 IV 200 I 2009 2009 010
0
II IIIIV III 2 V 201 2011 011
III
12
I
III IIV 2 II 20
012
III
13
I
II
IV 2 II 20
III
I
pct 70 pct 80 pct 90 Range (of statistical dispersion) 90/10

III

50
0%

10%

20%

30%

40%

50%

60%

70%

80%

90% 100%

H
350

350

300
250
200
150
100
50
pct 90
0
pct 70
pct 50
007 08 8 9
IV 2 20 00
III IIIIIIV 2 II 200 2009 10
pct 30
I
I IV II 20 010
1
III
I IV 2
201 011
2
III
pct 10
III IIV 2
201
II
012
III
13
V2
I
0
2

I
II
III

300
250
200
150
100

pct 40

300
250
200
150
100
50
0

200

50

pct 30

250

pct 20

pct 90
Upper range

Mean value
Lower range

IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III IIIIV III
2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013

Notes: A thetime series ofthe mean value and standard deviation, B thetime series ofthe mean value, theninth decile and therelationships
ofthe maximum tothemean value, C values ofdeciles forthetwo periods under comparison, D change intheinterdecile ranges forthetwo
periods under comparison, E thecumulative distribution function forthetwo periods under comparison, F thetime series ofpositive deviations
from themean ofpositive deciles and theseries ofthe range between theninth and thefirst deciles, G thetime series ofthe mean value,
theupper range (mean value + standard deviation) and thelower range (mean value standard deviation) and theninth decile, H thetime
series ofthe decile distribution.
Fig. 15.9. Descriptive statistics on thedegree ofbankruptcy risk faced by manufacturing firms inPoland from the1st half of2007
tothe1st half of2014
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

200

The decile distribution indicates thehigh value ofthe eighth and ninth decile (Diagram H inFigure 45), which toa large extent
contributes tothehigh mean value located between theseventh and theeighth deciles. As a result, 20% ofthe firms inthegroup
under study toa large extent determine theactual direction and intensity ofchanges inthedegree ofbankruptcy risk, asmeasured
synthetically, with theaid ofthe mean value which is deemed representative ofthe overall situation within theset ofenterprises
studied.
The distribution function forthetwo comparative periods (1st half of2012 and 1st half of2013) changed slightly (cf. Diagram C
and E). Particularly small changes inwere noted inthefirst up totheeighth deciles, whereas intheninth decile themagnitude
ofchanges was relatively higher, asthehiatus between theninth and theeighth decile increased by approximately 19%,
ascompared with thecorresponding period ofthe previous year (Diagram D). Therange for80% ofthe observations actually
conducted, i.e. between theninth and thefirst decile increased from 194 intheearly 2012 to219 intheearly 2013 (by 12.9%), with
a 3.1% increase inthemean value, i.e. from 163 to168. Therange offluctuations inthevalue ofthe ninth decile (i.e. thedifference
between themaximum and theminimum value over theperiod under study) is 73, and is actually broader than thefluctuation
range ofthe mean value equalling 38, i.e. 92%. Thestandard deviation shows alternating fluctuations without a significant
correlation with thechanges inthemean value (linear correlation coefficient r = 0.31).

15.4. Diversity ofdegree ofbankruptcy risk across thenace divisions


An important research direction on how thedegree ofbankruptcy risk may affect thetrading sector and how itcan be managed
under theEarly Warning System Rapid Response Instrument Project, is theactual assessment ofthe degree ofbankruptcy
risk made indue consideration ofspecific types ofbusiness activity pursued inPoland (in line with NACE terminology), and
therespective sizes ofthe enterprises concerned (small, medium-sized and large ones). This assessment was carried out inthethree
stages, i.e. assessment based on thedegree ofbankruptcy risk (MW) value, allocation ofrelevant normative standards (IIV) toeach
type ofbusiness activity pursued based on theassessment ofthe deviation from thevalue expressing thedegree ofbankruptcy
risk fora specific NACE section (oMW D), thedeviation ofthe number ofemployees ina specific NACE section from theaverage
forthat particular type ofbusiness activity (oLP D), and thedevelopment ofa ranking list ofNACE divisions taking into account
thevalue ofthe synthetic measure called MSO.
A retrospective analysis ofthe changes inthedegree ofbankruptcy risk (MW) oftrade firms operating inPoland from the1 st half
of2007 tothe1st half of2013, asillustrated by theheat maps, facilitates theassessment ofthe level and evolution ofthe parameter
inquestion. In themanufacturing sector, theabove-average value reflecting potential bankruptcy risk forwas observed infirms
involved inthemanufacture ofother transport equipment (30) and themining ofcoal and lignite (05). Assessment ofchanges
intheactual degree ofbankruptcy risk adjusted fortherespective company sizes indicates above-average levels with regard
tosmall companies falling into thefour NACE divisions, an average one ineight NACE divisions, and a large one inseven NACE
divisions. An appreciably more favourable situation was reported inthetrading sector than inthemanufacturing and theservices
ones, which is further attributable tothefact that itwould be impossible toidentify a specific NACE section inwhich thedegree
ofbankruptcy risk would remain consistently above average throughout theperiod under study. Thehighest value ofbankruptcy
risk was noted intheservices sector, whereas thevalues above

theaverage fortheservice activities were observed inconstruction


industry (41, 42 , 43), transport (51, 52), intheactivities related toemployment (78), and intheactivities ofthe tour operators, travel
agencies, and travel agents (78). With regard tothesmall companies theexposure tothedegree ofbankruptcy risk throughout
theperiod under study remained above theaverage for3 NACE divisions, forthemedium-sized ones for4 NACE divisions, and
forthelarge ones for6 NACE divisions.
The heat maps indicate a distinctly greater exposure tobankruptcy risk intheservices sector than inthemanufacturing and
thetrading ones, especially after 2010, whereupon a significant increase occurred.
An analysis ofvariability ofthe degree ofbankruptcy risk indicates its highest average values and thegreatest variability
inthebusiness activities associated with: manufacturing ofother transport equipment (30), other mining and quarrying (08),
manufacture oftobacco products (12), and service activities supporting mining and quarrying (09). Thefollowing NACE divisions
applied tosmall enterprises: manufacture ofother transport equipment (30), manufacture ofbasic metals (24), manufacture
ofelectrical equipment (27), travel agency activities, tourism (79). In medium-sized enterprises, very much like inthesmall ones,
thehighest average value and thegreatest variability was observed inthemanufacture ofother transport equipment (30), travel
agency activities, travel agents (79). With regard tothelarge entities thesame observations applied tomanufacture ofother
transport equipment (30), manufacture ofcomputer, electronic and optical products (26), and themining ofcoal and lignite (05).
Themost stable situation inthesmall business entities was noted intheservice activities related tothemaintenance oforder
within thebuildings and greenery (81), among themedium-sized companies inthemanufacture ofbasic pharmaceutical products,
medicines and other pharmaceutical products (21), whereas inthelarge companies intheactivities involved inthemanufacture
and supply ofelectricity, gas, steam, hot water and air conditioning supply (35).
201

Table 15.1. Variations inthedegree ofbankruptcy risk faced by enterprises inPoland by NACE division from the1st half of2007
tothe1st half of2014
NACE
I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II
divisio
ns 2007 2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013

MANUFACTURING

NACE
I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II
divisio
ns 2007 2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013

TRADE

5
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
35
36
37
38
39

45
46
47

SERVICES
41
42
43
49
50
51
52
53
55
56
58
59
60
61
62
63
66
68
78
79
80
81
82
94
95
96

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).
1000
900
800
700
600
500
400
300
200
100
0

10 11

12 13 14 15 16 17

18 19 20

21 22 23 24 25 26
NACE division

27 28

29 30 31

63 66

68

32 33

35 36 37 38 39

1400
1200
1000
800
600
400
200
0

41

42

43 45 46

Notes: As forFigure 15.2.

47

49

50

51 52

53

55 56

58

59

60 61

62

78

79 80

81

82

94 95 96

NACE division

Fig. 15.10. Descriptive statistics on thedegree ofbankruptcy risk faced by enterprises inPoland by NACE division from the1st half
of2007 tothe1st half of2013
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

202

The next step ofthe analysis consisted inallocating a specific type ofnormative standard (I, II, III orIV) toeach NACE division within
a specific period under study based on theassessment ofthe deviation from thedegree ofbankruptcy risk fora particular NACE
division (oMW D), and thedeviation ofthe number ofemployees within such a NACE division from theaverage forthat type
ofbusiness activity (oLP D).

3.00

3.00

2.50

2.50

2.00

2.00

Degree of bankruptcy risk

Degree of bankruptcy risk

The dispersion ofNACE divisions pertaining tomanufacturing activities is shown graphically inthefigures below. Respective NACE
divisions are placed on thecross graph within thefour areas (normative standards IIV) where thehorizontal and vertical axes
represent themean values ofcorresponding measures (the degree ofbankruptcy risk and thenumber ofemployees). Theactual
location ofa particular NACE section on thegraph makes itpossible tocharacterize its position inrelation tothetwo measures
covered by theanalysis.

1.50

1.00

1.00

0.50

0.50

0.00
0.00

1.50

0.50

1.00

1.50
2.00
No of employees

2.50

0.00
0.00

3.00

0.50

1.00

1.50
2.00
No of employees

2.50

3.00

Fig. 15.11. NACE manufacturing divisions (left panel) and services divisions (right panel) by degree ofbankruptcy risk and
thenumber ofemployees inthe1st half of2013
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

A comprehensive picture ofthe normative standards forvarious NACE divisions intheperiod spanning early 2007 early 2013
is shown intheheat maps inwhich thefour colours correspond totherespective normative standard (red standard I, yellow
standard II, pale green standard III, and dark green standard IV).3 This particular way ofrepresentation clearly demonstrates how
theabove-mentioned standards were established over theentire period under study fortherespective NACE divisions. Thanks
totheheat maps ofthe normative standards itwas possible tosee theconcentration ofstandard I (high risk area) inconstruction
industry since 2010, and intheactivities related toretail trade throughout theperiod under consideration.
The final step ofthe analysis consisted inestablishing a ranking list ofthe respective NACE divisions on thebasis ofa synthetic
measure MSO. An analysis ofthe distribution ofNACE divisions intherespective cross graphs (attributable tostandards IIV), and
theresultant MSO value measures facilitate thedrawing up ofa ranking list ofrespective NACE divisions starting with theones
most exposed tobankruptcy risk totheleast exposed ones (the lower therank, thehigher therisk; thehigher thevalue, thelower
therisk). Additionally, forthepurpose ofa comparative analysis, therespective ranks were allocated tothecorresponding NACE
divisions forsmall, medium-sized and large business enterprises, thus indicating theactual position ofparticular NACE divisions
within a particular class ofcompany size. Theanalysis ofthe ranking indicates a rather difficult economic situation, with a significant
share inthenumber ofemployees inactivities related tocoal mining inall enterprises, inlarge enterprises aswell asin business
activities related tothemanufacture ofother transport equipment regardless ofcompany size. Despite thegenerally sound
situation inthetrading sector, particular attention should be paid towholesale trade. Within theservices sector, on theother
hand, thekey sections that merit due monitoring are those related totransport and construction. An overall difficult situation and
a high level ofworkforce concentration were reported intourism (79).

The rationale behind thenormative standards is discussed inChapter 10 section 6.

203

Table 15.2. Heat map ofnormative standards inNACE divisions by type ofbusiness activity inPoland from the1 st half of2007
tothe1st half of2013
NACE
I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II NACE I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II III-IV I-II
divisi
divisi
2007
2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013
ons
ons 2007 2007 2008 2008 2009 2009 2010 2010 2011 2011 2012 2012 2013
MANUFACTURING

TRADE

5
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
35
36
37
38
39

45
46
47
SERVICES
41
42
43
49
50
51
52
53
55
56
58
59
60
61
62
63
66
68
78
79
80
81
82
94
95
96

where:
standard I

standard II

standard III

standard IV

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

Table 15.3. Ranking list ofNACE manufacturing divisions by MSO measure and by company size from the1st half of2007 tothe1st
half of2013
Position in the ranking
NACE divisions
TOTAL

Small

Mediumsized

Large

05 Mining of coal and lignite

27

30 Manufacture of other transport equipment

25 Manufacture of fabricated metal products, except machinery and equipment

27 Manufacture of electrical equipment

17

25

16 Manufacture of wood and of products of wood and cork, except furniture;


manufacture of straw and plaiting materials

15

28 Manufacture of machinery and equipment n.e.c.

16

31 Manufacture of furniture

23

11

11

26 Manufacture of computer, electronic and optical products

13

10

10 Manufacture of food products

14

MANUFACTURING

204

Position in the ranking


NACE divisions
TOTAL

Small

Mediumsized

Large

29 Manufacture of motor vehicles, trailers and semi-trailers except motorcycles

10

15

19

10

24 Manufacture of basic metals

11

33 Repair, maintenance and installation of machinery and equipment

12

20

12 Manufacture of tobacco products

13

17

13 Manufacture of textiles

14

11

14

22

08 Other mining and quarrying

15

18

36 Water collection, purification and supply

16

26

13

28

18 Printing and reproduction of recorded media

17

28

12

24

37 Sewerage and wastewater treatment

18

30

30

13

11 Manufacture of beverages

19

10

17

16

23 Manufacture of other non-metallic mineral products

20

12

09 Mining support service activities

21

16

14 Manufacture of wearing apparel

22

21

23

35 Electricity, gas, steam, hot water and conditioning supply

23

20

29

19

22 Manufacture of rubber and plastics products

24

19

20

39 Remediation activities and other waste management services

25

22

18

15 Manufacture of leather and related products

26

14

24

18

19 Manufacture of coke and refined petroleum products

27

12

15

21

32 Other manufacturing

28

27

22

27

20 Manufacture of chemicals and chemical products

29

25

31

25

17 Manufacture of paper and paper products

30

29

23

29

38 Waste collection, treatment and disposal activities; materials recovery

31

24

26

26

21 Manufacture of pharmaceutical, medicinal chemical and botanical products

32

21

28

30

MANUFACTURING

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

Table 15.4. Ranking list ofNACE trade divisions by MSO measure and by company size from the1st half of2007 tothe1st half of2013
Position in the ranking
NACE divisions

Mediumsized

TOTAL

Small

47 Retail trade, except of motor vehicles and motorcycles

46 Wholesale trade, except of motor vehicles and motorcycles

45 Wholesale and retail trade and repair of motor vehicles and motorcycles

Large

TRADE

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

205

Table 15.5. Ranking list ofNACE services divisions by MSO measure and by company size from the1st half of2007 tothe1st half
of2013
Position in the ranking
NACE divisions
TOTAL

Small

Mediumsized

Large

SERVICES
49 Land transport and transport via pipelines

17

42 Civil engineering

50 Water transport

20

41 Construction of buildings

78 Employment activities

51 Air transport

25

79 Travel agency, tour operator, reservation service and related activities

59 Motion picture, video and television programme production, sound recording and music
publishing activities

18

52 Warehousing and support activities for transportation

14

12

80 Security and investigation activities

10

15

19

53 Postal and courier activities

11

68 Real estate activities

12

11

20

82 Office administrative, office support and other business support activities

13

11

17

43 Specialized construction activities

14

13

10

13

95 Repair of computers and personal and household goods

15

10

18

96 Other personal service activities

16

22

22

61 Telecommunications

17

14

14

58 Publishing activities

18

13

18

63 Information service activities

19

16

23

16

81 Services to buildings and landscape activities

20

24

21

10

66 Activities auxiliary to financial service and insurance activities

21

23

15

21

94 Activities of membership organizations

22

21

62 Computer programming, consultancy and related activities

23

12

20

19

60 Programming and broadcasting activities

24

19

17

15

55 Accommodation

25

11

56 Food and beverages service activities

26

22

16

12

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

15.5. Areas highly exposed tobankruptcy risk


In identifying themost vulnerable NACE divisions, theMSO measure was used, including an evaluation oftwo categories: 1.
therelationship ofthe degree ofbankruptcy risk and 2. thenumber ofemployees within a particular NACE section relative
totheaverage value forthat particular type ofbusiness activity inNACE. TheNACE divisions which require special monitoring
were identified astheones assigned tostandard I, i.e. those characterized by thedegree ofbankruptcy risk above theaverage value
assumed forthat particular type ofbusiness activity and those where thenumber ofemployees is above average. Theidentification

206

was done separately formanufacturing, trade and services sectors, aswas outlined intheprevious section with theaid ofthe
ranking lists fortherespective NACE divisions on thebasis ofthe synthetic MSO measure. Consequently, a set of10 NACE sections
(i.e. TOP 10) is presented, although due tothesignificance ofinformation on theareas exposed toparticular bankruptcy risk
(NACE divisions); itwas assumed that this set may be subject tosome variations. Thetable below shows theareas ofactivity (NACE
divisions) deemed particularly vulnerable (since the3rd research edition aspart ofthe Early Warning System Rapid Response
Instrument Project).

Table 15.6. Selected areas ofNACE divisions with thehighest exposure tobankruptcy risk by MSO measure from the1st half of2011
tothe1st half of2013
NACE divisions

I-II 2011 III-IV 2011 I-II 2012 III-IV 2012 I-II 2013
Manufacturing

05 Mining of coal and lignite

08 Other mining and quarrying

09 Mining support service activities

24 Manufacture of basic metals

25 Manufacture of fabricated metal pro ducts, except machinery and equipment

26 Manufacture of computer, electronic and optical products

27 Manufacture of electrical equipment

29 Manufacture of motor vehicles, trailers and semi-trailers except motorcycles

30 Manufacture of other transporte quipment

31 Ma nufa cture o f furni ture

33 Repair, maintenance and installation of machinery and equipment

45 Wholesale and retail trade and repair of motor vehicles and motorcycles

47 Retail trade, except of motor vehicles and motorcycles

Trade

Services
41 Construction of buildings
42 Civil engineering

43 Specialized construction activities

49 Land transport and transportvia pipelines

51 Airtransport

52 Warehousing and support activities for transportation

78 Employmentactivities

79 Travel agency, tour operator, reservation service and related activities

Notes:
did not occur during theperiod inquestion, occurred during theperiod inquestion, occurred during theperiod inquestion and was
additionally characterized by thespecification comprised within a NACE section.
Source: Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (June 2011 April 2014).

The selected NACE divisions are characterized by such parameters as: thenumber ofemployees, share intotal employment,
number ofcompanies, value ofsales, assets, and equity. Exposure tobankruptcy risk was presented and discussed, followed by
theestimations ofthe same interms offuture likelihood, whilst taking into account changes inthevariables describing bankruptcy
risk and their predicted values (asset productivity, short-term debt, self-financing, operating return on assets). Additionally,
theactual length ofthe cash turnover cycle and its components (cycle inventories, receivables and liabilities), aswell ascurrent
liquidity were also taken into consideration.
Due tothelimited scope ofthe present publication, below follows a detailed description ofNACE division 41 (Construction
ofbuildings) from the1st half of2007 tothe1st half of2013 along with thepredictions regarding bankruptcy risk forthe2nd half
of2013 and the1st half of2014, thecharacteristics and descriptions ofpertinent model variables, aswell asbasic information on
theNACE division under study.

207

NACE division

CONSTRUCTION OF BUILDINGS

No ofcompanies

1 909

No ofemployees

108 079

Share inthenumber
ofemployees

2.23%

Sales revenues

PLN 40922.50 m

Assets

PLN 56803.48 m

Net book value

PLN 22387.60 m

Degree ofbankruptcy risk


III 2012

IIIIV 2012

III 2013

Tempo r/r

297

286

333

12.06%

60%

400
350

40%

300
250

20%

200
150

0%

100
50

-20%

0
III 2007 IIIIV 2007 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013 IIIIV 2013 III 2014
Rate y/y (right axis)

Degree of bankruptcy risk (left axis)

No of enterprises and employees

AP and OROA

160 000

2600

1.20

12%

150 000

2400

1.10

10%

2200

1.00

2000

0.90

1800

0.80

1600

0.70

1400

0.60

1200

0.50

140 000
130 000
120 000
110 000
100 000
III 2007 IIIIV 2007 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013

No of employees (left axis)


No of entities (right axis)

SFR and STDR


0.48
0.44
0.39
0.35
0.30

III 2007 IIIIV 2007 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013

8%
6%
4%
2%
III 2007 IIIIV 2007 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013

0%

Asset Productivity (left axis)


Operating Return on Assets (right axis)

0.38

160

0.37

140

0.36

120

0.35

100

0.34

80

0.33

60

0.32

40

0.31

20

0.30

CCC Days

III 2007 IIIIV 2007 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013

Self-Financing Ratio (left axis)


Short-Term Debt Ratio (right axis)

CCC (days)
ICP (days)

RCP (days)

PCP (days)

Fig. 15.12. Detailed characteristics of a single NACE division using the basic information, MW values and model variables reflecting
the degree of bankruptcy risk from the 1st half of 2011 to the 1st half of 2014
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

208

This particular NACE division is characterized by an increased exposure tobankruptcy risk (+12.06% y/y) expressed by thevalue
ofup to333 intheearly 2013. Theprojected changes inthis parameter indicate its continued rise inthelate 2013 and early
2014 (312 and 356, respectively), which thus merits diligent monitoring until early 2015. Thenumber ofenterprises decreased
(5.26% y/y, i.e. by 106 entities), with an accompanying significant reduction inthenumber ofemployees (17.67% y/y, i.e. by
23,202 persons). An increase inbankruptcy risk was partly due tothediminished asset productivity (20.88 % y/y) and a similar
decline intheoperating return on assets (16.01 % y/y). There was a slight increase incurrent liquidity (7.02% y/y), and a reduction
inshort-term debt (11.11% y/y). Thecash conversion cycle increased (30.67% y/y) to102 days asa result ofthe extended inventory
conversion cycle (25.58 % y/y) and thereceivables conversion cycle (10.33 % y/y), which only toa certain extent offset theextended
time fortherepayment ofshort-term liabilities (9.89% y/y).
These NACE divisions addressed inmore detail inthemicroeconomic component Microeconomic Reports along with
theanalysis ofeconomic conditions (cyclical fluctuations) and thedevelopment prospects (forecasts with predictive distributions)
inthemanufacturing, trade and construction sectors ofthe Polish economy (addressed inmore detail inthemacroeconomic
component Macroeconomic Reports) offer a comprehensive evaluation ofeach NACE division complete with indications asto
its specific development prospects.

15.6. Bankruptcy risk a regional profile


Another important direction ofresearch aimed at assessing thedegree ofbankruptcy risk under theEarly Warning System Rapid
Response Instrument Project consists intheassessment ofthat risk on a regional basis. It involves 16 Polands regions and specifies
therisk level fortotal enterprises, by type ofbusiness activity (manufacturing, trade and services). Theassessment was carried
out inthethree stages: on thebasis ofthe MW value and its variability within therespective regions with respect totheadopted
analytical criteria, theallocation ofnormative standards (standards IIV) foreach region based on theassessment ofthe deviation
from thedegree ofbankruptcy risk fora particular region (oMW D), and thedeviation ofthe number ofemployees within a given
region from theaverage forthat type ofbusiness activity (oLP D), aswell asthedevelopment ofa NACE-based ranking list taking
into account thevalue ofthe synthetic MSO measure.
The degree ofbankruptcy risk forbusiness entities intotal is highest inZachodniopomorskie and Dolnolskie regions (voivodships).
This measure intheselected regions was at its highest level throughout theentire period ofanalysis. In six regions located
ineastern Poland, theaverage degree ofbankruptcy risk remained at thewarning level. A graphical illustration ofthe distribution
ofthe mean values across theregions clearly points tothepresence ofa north-south divide ofthe country. Regions located
inwestern Poland, with theexception ofZachodniopomorskie and Dolnolskie, are characterized by a low degree ofbankruptcy
risk, whereas theeastern part ofthe country remains at thewarning level. Variability ofthe exposure indicates thehighest levels
inDolnolskie and witokrzyskie regions. Thelowest levels ofbankruptcy risk aswell asthemost stable situation was reported
inMaopolskie, dzkie, Lubuskie, Wielkopolskie and Opolskie regions.

400
300
200
100

nio
od
za
ch

20

160

po
mo
r
wa
d
oln skie
rm
o
is
ko lsk
-m
ie
az
urs
ma
kie
zo
wi
ec
kie
l
po skie
mo
rsk
po
ie
ku
dk
jaw
arp
ac
sko
kie
-po
mo
rsk
ie
po
dl
w
it askie
ok
rz y
ski
e
lub
els
ma
kie
op
ols
ki
d e
zki
e
lub
wi
elk uskie
op
ols
kie
op
ols
kie

Notes: As forFigure 15.2.


Fig. 15.13. Regional variation inthedegree ofbankruptcy risk faced by all enterprises inPoland (right panel) and its descriptive
statistics (left panel) from the1st half of2007 tothe1st half of2013
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

209

When assessing theregional degree ofbankruptcy risk formanufacturing firms, itshould duly be noted that itis structured
differently than was thecase forbusiness entities intotal. In Zachodniopomorskie alone thehighest values ofbankruptcy risk
were reported. Unlike thebusiness entities intotal, high values ofbankruptcy risk among themanufacturing companies were
reported inWarmisko-Mazurskie. Analysis ofthe changes intheMW value indicates its largest variability (standard deviation)
inZachodniopomorskie, witokrzyskie, Lubuskie and Opolskie. Themost stable situation within theperiod under consideration
was reported inmanufacturing (low standard deviation) located across Maopolskie and Wielkopolskie.

500
400
300
200
100

za

20

160

ch
od
nio
wa
po
rm
mo
is
rsk
ko
ie
-m
az
urs
k ie
do
lno
l
ski
e
l
s
k
po
i
mo e
rsk
ie
po
dla
po
s
k
i
dk
arp e
ac
k ie
d
zk
lub ie
u
m
sk
ku
jaw azow ie
sko
iec
k ie
-po
mo
rsk
lub ie
e
ma lskie
op
ol
wi
elk skie
op
o
w
lsk
it
ie
ok
rz y
ski
op e
ols
k ie

Notes: As forFigure 15.2.


Fig. 15.14. Regional variation inthedegree ofbankruptcy risk (right panel) and its descriptive statistics (left panel) formanufacturing
companies inPoland from the1st half of2007 tothe1st half of2013
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

The degree ofbankruptcy risk fortrading companies, when assessed regionally, is characterized by significantly lower
values, ascompared totheother types ofbusiness activity. Thehigh value ofexposure tothedegree ofbankruptcy risk
inthewitokrzyskie region should duly be noted, though, asitis not reported forany other type ofbusiness activities (warning
level) within this region. In theremaining regions potential bankruptcy risk remained at a low level. Despite theso structured MW
values forthetrading entities intherespective regions (with theexception ofthe above referenced witokrzyskie region), their
variability was relatively high within theperiod under study, except fortheMaopolskie region. Themost stable situation inthis
respect was observed inthedzkie region.

300

200

100

ma

0
20

160

w
i

to

k rz

ysk
op ie
ols
k
ku
lub ie
jaw
els
sko
kie
-po
mo
rsk
ma
ie
zo
wi
e
ck
do
i
lno
e
l
sk i
e
l
ski
e
po
mo
po
rsk
dk
arp ie
ac
wa
kie
rm
lub
is
us
ko
kie
-m
az
urs
wi
kie
elk
op
ols
kie
za
ch
po
od
dla
nio
ski
po
e
mo
rsk
ie
d
zki
op e
ols
kie

Notes: As forFigure 15.2.


Fig. 15.15. Regional variations inthedegree ofbankruptcy risk (right panel) and its descriptive statistics (left panel) fortrading
companies inPoland from the1st half of2007 tothe1st half of2013
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

210

The regionally assessed degree ofbankruptcy risk intheservices sector indicates that most ofregions characterised by its elevated
levels are located incentral-eastern and central-western Poland. Despite thefact that thehighest average MW value fortheservices
sector was recorded inMazowieckie, its highest maximum levels were actually reported inDolnolskie. In thenorthern regions
ofthe country low MW values were reported, which means that this part ofthe country is least exposed tobankruptcy risk
asregards thesector under consideration. Warning recommendations with regard tothis sector were given towitokrzyskie
and Maopolskie, though. Thegreatest stability intheservices sector (low risk and low variability) was observed inPomorskie and
Zachodniopomorskie.

700
600
500
400
300
200
100

ma

zo
w
do ieck
ie
lno
l
ski
e
d
w
zki
it
e
ok
rz y
s
k
po
mo ie
r
wi
za
elk skie
ch
o
od
nio pols
ki
po
mo e
rsk
i
lub e
els
po
k
ie
dk
arp
a
c
ma
k
op ie
ols
k ie
po
dla
wa
ski
rm
e
is

ls
ko
k
-m
az ie
urs
k ie
ku
jaw
o
sko pols
k ie
-po
mo
rsk
lub ie
us
k ie

20

160

Notes: As forFigure 15.2.


Fig. 15.16. Regional variations inthedegree ofbankruptcy risk (right panel) and its descriptive statistics (left panel) fortheservices
sector inPoland from the1st half of2007 tothe1st half of2013
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

The next step intheanalysis and assessment ofthe degree ofbankruptcy risk faced by firms operating indifferent regions ofPoland
consisted intheallocation ofan appropriate standard (IIV) toeach region. This was done on thebasis ofthe MSO value obtained
throughout thedifferent periods ofanalysis inconjunction with theparameter ofeconomic significance ofa given region reflected
inthesize ofits resident workforce (i.e. concentrated employment structure). Theheat maps depicting theactual distribution
ofstandards across different regions indicate diverse characteristics fortherespective types ofbusiness activity and business
entities intotal. Particular attention should be focused on regions inwhich standard I (marked inred) prevailed throughout
theperiod studied. As far asmanufacturing and trading enterprises are concerned, a certain improvement was noted across
theregions, whereas theservices sector since 2010 has seen a steady growth inthenumber ofregions characterized by a high
and moderately high bankruptcy risk (standards I and II).
Table 15.7. Regional differences in MSO measures of by reference pattern and type of business activity in Poland from the 1st half
of 2007 to the 1st half of 2013
Regions (voivodships)

I-II
2007

III-IV
2007

I-II
2008

III-IV
2008

I-II
2007

III-IV
2007

I-II
2008

III-IV
2008

I-II
2009

III-IV
2009

I-II
2010

III-IV
2010

I-II
2011

III-IV
2011

I-II
2012

III-IV
2012

I-II
2013

I-II
2010

III-IV
2010

I-II
2011

III-IV
2011

I-II
2012

III-IV
2012

I-II
2013

dolnolskie
mazowieckie
lskie
zachodniopomorskie
warmisko-mazurskie
lubelskie
pomorskie
witokrzyskie
wielkopolskie
dzkie
maopolskie
podkarpackie
podlaskie
lubuskie
kujawsko-pomorskie
opolskie
Manufacturing
Regions (voivodships)
lskie
dolnolskie
zachodniopomorskie
pomorskie
warmisko-mazurskie
podlaskie
dzkie
kujawsko-pomorskie

I-II
2009

211

III-IV
2009

maopolskie
podkarpackie
podlaskie
lubuskie
kujawsko-pomorskie
opolskie
Manufacturing
Regions (voivodships)

I-II
2007

III-IV
2007

I-II
2008

III-IV
2008

I-II
2007

III-IV
2007

I-II
2008

III-IV
2008

I-II
2007

III-IV
2007

I-II
2008

III-IV
2008

I-II
2009

III-IV
2009

I-II
2010

III-IV
2010

I-II
2011

III-IV
2011

I-II
2012

III-IV
2012

I-II
2013

III-IV
2009

I-II
2010

III-IV
2010

I-II
2011

III-IV
2011

I-II
2012

III-IV
2012

I-II
2013

III-IV
2009

I-II
2010

III-IV
2010

I-II
2011

III-IV
2011

I-II
2012

III-IV
2012

I-II
2013

lskie
dolnolskie
zachodniopomorskie
pomorskie
warmisko-mazurskie
podlaskie
dzkie
kujawsko-pomorskie
lubuskie
mazowieckie
maopolskie
wielkopolskie
podkarpackie
witokrzyskie
opolskie
lubelskie
Trade
Regions (voivodships)

I-II
2009

dolnolskie
mazowieckie
maopolskie
lubelskie
witokrzyskie
lskie
pomorskie
kujawsko-pomorskie
wielkopolskie
dzkie
lubuskie
podkarpackie
podlaskie
warmisko-mazurskie
zachodniopomorskie
opolskie
Services
Regions (voivodships)

I-II
2009

dolnolskie
mazowieckie
wielkopolskie
maopolskie
pomorskie
lubelskie
witokrzyskie
podkarpackie
lskie
warmisko-mazurskie
dzkie
lubuskie
podlaskie
zachodniopomorskie
opolskie
kujawsko-pomorskie
where:
standard I

standard II

standard III

standard IV

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

The final step ofthe present analysis consists indrawing up a ranking list ofregions based on theMSO values throughout
theperiod under consideration. Theregions were thus allocated respective ranks of1 to16, whereby thelower ranks indicate
regions facing thehighest bankruptcy risk, whereas thehigher ranks represent regions least exposed toit. Theregions were then
arranged inline with theallocated ranks forbusiness entities intotal and by specific types ofbusiness activity. Theranking list
clearly indicates that thefollowing regions merit special attention: Dolnolskie with respect tobusiness entities intotal, lskie
with respect tomanufacturing activities aswell asDolnolskie and Maopolskie with respect tothetrading and services sectors.

212

Table 15.8. Ranking list of regions by MSO measure score and type of activity in Poland from the 1st half of 2007 to the 1st half
of 2013
Position in the ranking
Regions (voivodships)

Mediumsized

TOTAL

Small

Large

dolnolskie

mazowieckie

10

lskie

zachodniopomorskie

15

14

warmisko-mazurskie

14

10

lubelskie

16

pomorskie

witokrzyskie

14

wielkopolskie

12

dzkie

10

10

11

maopolskie

11

11

podkarpackie

12

13

12

podlaskie

13

13

13

lubuskie

14

11

12

kujawsko-pomorskie

15

16

opolskie

16

15

16

15

Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

15.7. Verification of bankruptcy risk assessments


An empirical verification ofthe assessment ofchanges intheoverall condition ofthe sector under consideration interms
ofrespective degrees ofbankruptcy risk (MW) consists ina comparison ofthe obtained parameters with thepercentage ofinitiated
court-supervised insolvency proceedings over time. Both time series are statistically significant and highly correlated and thetrend
line fortheinitiated insolvency proceedings determines thevalues falling

within a limited range ofvalues reflecting

thedegree
ofbankruptcy risk (i.e. an optimistic and a pessimistic scenario).
Values reflecting

thedegree ofbankruptcy risk and thepercentage ofcourt-supervised bankruptcy proceedings initiated inthe1st
half of2013 essentially reached identical levels (197 and 194, respectively) with thecorrelation coefficient r = 0.82. Another
observation, inlate 2013, corroborates thehigh empirical verifiability ofbankruptcy risk assessment results asyielded by thepresent
analysis ofthe percentage ofthe actually instigated court-supervised bankruptcy proceedings, because theMW value was 178,
compared toOU 182 (the difference between them is only ca. 2.2%). Thecoefficient oflinear correlation MW and OU also increased
up tor = 0.85.
The other plane ofthis empirical verification reflects theaccuracy ofpredictions concerning thedegree ofbankruptcy risk (cf.
Figure 15.18).
The plot ofthe curve describing thetime series ofthe actual MW values implies that itfalls well within thepredicted range defined
by theoptimistic and thepessimistic scenarios, respectively. Theobservation fortheperiod spanning late 2011 early 2012 was
deemed theonly aberration, astheactual MW values exceeded theprediction limits anchored inthepessimistic scenario ofthe
early 2011. Further observations substantiate theconclusion that theactual MW values are, infact, much closer totheupper limit

213

ofthe prediction range (the pessimistic scenario). Theavailable data forthe2nd half of2013 clearly indicate (an orange line) that
theactual MW value differed from thepredicted one only by 1.7% (178 vs 181), which only goes toprove high accuracy ofall
predictions fortheconcluding period ofthe present study.
210

190

Percentage of court-supervised bankruptcy proceedings


Degree of bankruptcy risk (MW)

170

150

130

110

90

III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013 IIIIV 2013 III 2014

Note: thetotal value is determined asa weighted mean (number ofemployees) out ofthe values calculated forrespective types ofbusiness activity.
Thevalues fortheoptimistic scenario are marked ingreen, whereas those forthepessimistic one inred. Thepredicted values are represented
by thedashed line.
Fig. 15.17. Thedegree ofbankruptcy risk and thepercentage ofinitiated court-supervised bankruptcy proceedings fortotal
companies from the1st half of2008 tothe1st half of2014
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).
210

190

170

150

130

110

III 2007 IIIIV 2007 III 2008 IIIIV 2008 III 2009 IIIIV 2009 III 2010 IIIIV 2010 III 2011 IIIIV 2011 III 2012 IIIIV 2012 III 2013 IIIIV 2013 III 2014

Note: thetotal value is determined asa weighted mean (number ofemployees) out ofthe values calculated forrespective types ofbusiness activity.
Thevalues fortheoptimistic scenario are marked ingreen, whereas those forthepessimistic one inred. Thepredicted values are represented by
thedashed line. Theactually observed value forthe2nd half of2013 is marked inorange.
Fig. 15.18. Comparative analysis ofthe degree ofbankruptcy risk faced by all types ofcompanies and theforecast intervals based
on respective scenarios from the2nd half of2010 tothe1st half of2014
Source: own study based on: Kaczmarek, Kolegowicz, Krzemiski, Fijorek (December 2013).

214

Prospectively identified warning signals (subjected toan in-depth analysis in12 successive Report on theassessment ofenterprise
bankruptcy risk themicroeconomic component) indicating a likely deterioration intheoverall financial standing offirms have been
positively verified forthefollowing sectors ofthe domestic economy: construction industry, air transport, mining and quarrying,
tourism activities, IT services and communications.
The positive verification ofidentified warning signals along with theproposals forrevising theassessment ofthe degree
ofbankruptcy risk by adjusting itfor thepercentage ofactually initiated court-supervised bankruptcy proceedings, aswell
astheaccuracy ofpredictions regarding thedegree ofbankruptcy risk interms ofa selection ofscenarios effectively confirm
thehigh efficacy ofthis particular assessment method which employs novel estimated prediction models, whilst making due
allowances forrespective types ofbusiness activity.

Bibliography
Adamowicz E., Dudek S., Pachucki D., Walczyk W. (2010), Synchronizacja cyklu koniunkturalnego polskiej gospodarki z krajami strefy euro
w kontekcie struktury tych gospodarek, Instytut Rozwoju Gospodarczego SGH [http://www.nbpnews.pl/r/nbpnews/Pliki_PDF/
NBP/Publikacje/analityczne/irg_sghP.pdf, 2008].
Addison J.T., Blackburn M. (1994), The Worker Adjustment and Retraining Notification Act, Journal of Economic Perspectives 8.
Adriaanse J. (2005), Restructuring in the Shadow of the Law, Informal reorganization in the Netherlands, Doct. Dissertation, University
Leiden.
Altman E.I. (1968). Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy, The Journal of Finance XXIII (4).
Altman E.I. (1983a), Why Business Fail, Why Businesses Fail 3.
Altman E.I. (1983b), Corporate Financial Distress. A Complete Guide to Predicting, Avoiding, and Dealing with Bankruptcy, John Wiley
& Sons, New York.
Altman E.I. (1984), A Further Empirical Investigation of the Bankruptcy Cost Question, Journal of Finance 39(4).
Altman E.I., Narayanan P. (1997), Business Failure Classification Models: An International Survey, Financial Markets, Institutions
& Instruments 6(2).
Analiza (2010), Analiza moliwoci wykorzystania instrumentu prawa naprawczego jako narzdzia ratowania dziaalnoci gospodarczej,
Wardyski i Wsplnicy sp. k., Warszawa.
Andrade G., Kaplan S.N. (1997), How Costly is Financial (not Economic) Distress? Evidence from Highly Leveraged Transactions that
Became Distressed, NBER.
Argenti J. (1976), Corporate Collapse. The Causes and Symptoms, McGrawHill Book, London.
Armour J. (2001), The Law and Economics of Corporate Insolvency: a Review, University of Cambridge Working Paper 197, ESRC
Centre for Business Research.
Bankruptcy (2003), Bankruptcy and Restructuring Act of February 28, 2003, Journal of Laws 60(535).
Barczyk R., Ksek L., Lubiski M., Marczewski K. (2006), Nowe oblicza cyklu koniunkturalnego, PWE, Warszawa.
Bartelsman E.J., Haltiwanger J., Scarpetta S. (2004), Microeconomic Evidence of Creative Destruction in Industrial and Developing
Countries, World Bank.
Baxter M., King R.G. (1999), Measuring Business Cycles: Approximate Band-Pass Filters for Economic Time Series, The Review
ofEconomics and Statistics 81(4).
Bebchuk L.A. (1988), A New Approach to Corporate Reorganizations, Harvard Law Review 101.
Becchetti L., Sierra J. (2003), Bankruptcy Risk and Productive Efficiency in Manufacturing Firms, CEIS Tor Vergata, Research paper series 30.
Besicovitch A.S. (1932), Almost Periodic Functions, Cambridge University Press, London.
Bhabra G.S., Yao Y.Y. (2011), Is Bankruptcy Costly? Recent Evidence on the Magnitude and Determinants of Indirect Bankruptcy Costs,
Journal of Applied Finance & Banking 1(2).
Bhattacharjee A., Higson C., Holly S., Kattuman P. (2009), Macroeconomic Instability and Business Exit: Determinants of Failures and
Acquisitions of UK Firms, Economica 76.
Blazy R., Martel J., Nigam N. (2014), The Choice Between Informal and Formal Restructuring: The Case of French Banks Facing Distressed
SMEs, Journal of Banking and Finance 44.
Blinder A.S. (2007), Monetary Policy by Committee: Why and How?, European Journal of Political Economy 23.
Branch B. (2002), The Costs of Bankruptcy. A Review, International Review of Financial Analysis 11.
Bris A., Welch I., Ning Zhu (2006), The Costs of Bankruptcy: Chapter 7 Liquidation versus Chapter 11 Reorganization, The Journal
ofFinance LXI(3).
Britannica (2014), Encyclopaedia Britannica, online [http://www.britannica.com/EBchecked/topic/289166/insolvency].
Brockwell P.J., Davis R.A. (1986), Time Series: Theory and Methods, Springer-Verlag New York Inc., New York.
Brockwell P.J., Davis R.A. (2002), Introduction to Time Series and Forecasting, Springer-Verlag New York Inc., New York.
Brogi R., Santella P. (2004), Two New Measures of Bankruptcy Efficiency, SUERF The European Money and Finance Forum.
Bulow J.I., Shoven J.B. (1978), The Bankruptcy Decision, The Bell Journal of Economics 9(2).
Business Dynamics (2011), Business Dynamics: Start-ups, Business Transfers and Bankruptcy, European Commission.
Bussiere M., Fratzscher M. (2002), Towards a New Early Warning System of Financial Crises, EBC Working Paper 145.
Cadet A. (1996), Polar Coordinates in the Rnp; Application to the Computation of the Wishart and Beta Laws, Sankhya: The Indian
Journal of Statistics 58.
Casey C., Bartczak N. (1985), Using Operating Cash Flow Data to Predict Financial Distress: Some Extensions, Journal of Accounting
Research 23(1).
Chava S., Jarrow R.A. (2004), Bankruptcy Prediction with Industry Effects, Review of Finance, Oxford 8(4).
Cheng Lim T.C., Gan S. (2012), Bankruptcy Prediction: Theoretical Framework Proposal, International Journal of Management Sciences
and Business Research 1(9).

217

Chikuse Y. (2002), Statistics on Special Manifolds, Springer, New York.


Christiano L.J., Fitzgerald T.J. (2003), The Band-Pass Filter, International Economic Review 44.
Claessens S., Klapper L.F. (2002), Bankruptcy around the World. Explanations of its Relative Use, Policy research working paper 2865,
The World Bank.
Code of Civil (1964), Code of Civil Procedure of November 17, 1964, Journal of Laws 2014.101 uniform text.
COM (2007), Overcoming the Stigma of Business Failure For a Second Chance Policy. Implementing the Lisbon Partnership for Growth
and Jobs, Commission of the European Communities.
Corduneanu C. (1989), Almost Periodic Functions, Chelsea, New York.
Cornelli F., Felli L. (1997), Ex-ante Efficiency of Bankruptcy Procedures, European Economic Review 41.
Crutzen N. (2009), The origins of small business failure: A Taxonomy of Five Explanatory Business Failure Patterns, HEC-Ecole de Gestion
de l'Universit de Lige.
Das S.R., Freed L., Geng G., Kapadia N. (2006), Correlated Default Risk, The Journal of Fixed Income 16(2).
de Koning A.C.P. (1998), Business Failure and Entrepreneurship in International Perspective, EIM.
Dewaelheyns N., Van Hulle C. (2007), Aggregate Bankruptcy Rates and the Macroeconomic Environment: Forecasting Systematic
Probabilities of Default, Katholieke Universiteit Leuven.
Djankov S. Hart O., McLiesh C., Shleifer A. (2008), Debt Enforcement around the World, Journal of Political Economy 116(6).
Doing Business (2013), Doing Business 2014. Understanding Regulations for Small and Medium-Size Enterprises, World Bank Publications.
Drehmann M., Patton A.J., Sorensen S. (2005), Corporate Defaults and Large Macroeconomic Shocks, ECB Draft.
Eckbo E.B., Wang W. (2012), How Costly is Corporate Bankruptcy for Top Executives?, Tuck School of Business Working Paper 109.
ECC (2008), Think Small First. A Small Business Act for Europe (COM 2008 (394).
ECR (2014), European Commission Recommendation on a New Approach to a Business Failure and Insolvency of March 12, 2014
(C (2014) 1500).
Economic Dislocation (1988), Economic Dislocation and Worker Adjustment Assistance Act.
Efrat R. (2006), The Evolution of Bankruptcy Stigma, Theoretical Inquiries in Law 7.
Ekspertyza (2012), Ekspertyza dotyczca przeciwdziaania upadoci przedsibiorstw oraz polityki drugiej szansy, Ministerstwo
Gospodarki, Warszawa.
Employers Guide (2003), Employers Guide to Advance Notice of Closings and Layoffs, United States of Department of Labor, July 2003.
Engle R.F., Granger C.W.J. (1987), Cointegration and Error Correction: Representation, Estimation and Testing, Econometrica 55.
Entrepreneurship (2006), Entrepreneurship Policy Indicators for Bankruptcy Legislation in OECD Member and Non-member Economies.
Final Report, Working Party for Small and Medium-sized Enterprises and Entrepreneurship, OECD CFE / SME.
Erhart S., Lehment H., Vasquez Paz J.L. (2007), Monetary Policy Committee Size and Inflation Volatility, Kiel Institute for the World
Economy Working Paper 1377.
European Communities (2003), Business Register Recommendations Manual, European Communities, Office for Official Publications
of the European Communities.
European Restructuring Toolbox (2010), The Restructuring Toolbox. Capitalising on Past and Present Innovations to Anticipate
Restructuring Processes, Bruggeman F., de Lavergne F. (coordinators), Progress Project VS/2009/0319, Amnyous.
Eurostat-OECD (2007), Eurostat OECD Manual on Business Demography Statistics, "Methodologies and Working Papers, Eurostat.
Evans M., Millar J., (2006), Welfare to Work in the United Kingdom, [in:] Hoefer R., Midgley J., International Perspectives on Welfare to
Work Policy, The Haworth Press, Inc. Binghamton.
Evans-Klock Ch., Kelly P., Richards P., Vargha C. (1998), Worker displacement: Public Policy and Labour-management Initiatives in Selected
OECD Countries, Employment and Training Papers 24, International Labour Organization, Geneva.
Fijorek K. (2012), Aproksymacja modelu regresji logistycznej Firtha za pomoc waenia obserwacji, Zeszyty Naukowe seria Metody
analizy danych, Uniwersytet Ekonomiczny w Krakowie, Krakw 2012 (w druku).
Fijorek K., Fijorek D. (2011), Dobr zmiennych objaniajcych metod najlepszego podzbioru do modelu regresji logistycznej Firtha,
Metody Informatyki Stosowanej 2.
Fijorek K., Fijorek D., Winiowska B., Polak S. (2011), BDTcomparator: A Program for Comparing Binary Classifiers, Bioinformatics 27.
Fijorek K., Sokoowski A. (2012), Separation-Resistant and Bias-Reduced Logistic Regression: STATISTICA macro, Journal of Statistical
Software 47, Code Snippet 2.
Firth D. (1993), Bias Reduction of Maximum Likelihood Estimates, Biometrika 80.
Franks J.R., Torous W.N. (1988), An Empirical Investigation of U.S. Firms in Reorganization, The Journal of Finance 44(3).
Franks J., Loranth G. (2005), A Study of Inefficient Going Concerns in Bankruptcy, CEPR Discussion Paper 5035.
Franses P.H. (1996), Stochastic Trends in Economic Time Series, Oxford University Press, New York.
Franses P.H., Boswijk H.P. (1996), Temporal Aggregation in a Periodically Integrated Autoregressive Process, Statistics and Probability
Letters 30.

218

Franses P.H., Dijk D. (2005), The Forecasting Performance of Various Models for Seasonality and Nonlinearity for Quarterly Industrial
Production, International Journal of Forecasting 21.
Franses P.H., Ooms M. (1997), A Periodic Long-Memory Model for Quarterly UK inflation, International Journal of Forecasting 13.
Franses P.H., Paap R. (2004), Periodic Time Series Models, Oxford University Press, Oxford.
Garca-Posada M., Mora-Sanguinetti J.S. (2013), Are There Alternatives to Bankruptcy? A Study of Small Business Distress in Spain,
Documentos de Trabajo 1315, Banco de Espagna.
Gardner W.A., Napolitano A., Paura L. (2006), Cyclostationarity: Half a Century of Research, Signal Processing 86.
Gelfand A.E., Marriott J., Ravishanker N., Pai J. (1995), Bayesian Analysis of ARMA Processes: Complete Sampling Based Inference under
Exact Likelihoods, [in:] Berry D., Chaloner K., Geweke J., Bayesian Statistics and Econometrics: Essays in Honor of Arnold Zellner,
JohnWiley and Sons, New York.
Gennaioli N., Rossi S. (2012), Contractual Resolutions of Financial Distress, CEPR.
Gierszewska G., Romanowska M. (1995), Analiza strategiczna przedsibiorstwa, PWE, Warszawa.
Giesecke K., Longstaff F.A., Schaefer S., Strebulaev I. (2011), Corporate Bond Default Risk: A 150-year Perspective, Journal of Financial
Economics 102.
Gilson G.C. (1990), Troubled Debt Restructurings. An Empirical Study of Private Reorganization of Firms in Default, Journal of Financial
Economics 27.
Gin X., Love I. (2006), Do Reorganization Costs Matter for Efficiency? Evidence from a Bankruptcy Reform in Colombia, World Bank
Policy Research Working Paper 3970.
Gladyshev E.G. (1961), Periodically Correlated Random Sequence, Sov. Math. 2.
Gomez V., Maravall A. (1996), Programs TRAMO and SEATS. Instructions for the Users, Banco de Espana Working Paper 9628.
Gmez V. (1999), Three Equivalent Methods for Filtering Finite Nonstationary Time Series, Journal of Business and Economic
Statistics 17(1).
Gmez V. (2001), The Use of Butterworth Filters for Trend and Cycle Estimation in Economic Time Series, Journal of Business and
Economic Statistics 19(3).
Gradzewicz M., Growiec J., Hagemejer J., Popowski P. (2010), Cykl koniunkturalny w Polsce wnioski z analizy spektralnej, Bank
i Kredyt 41(5).
Grasso V.F. (2005), Seismic EarlyWarning Systems: Procedure for Automated Decision Making, Doct. Dissertation, Universit degli Studi
di Napoli Federico II.
Gratzer K., Stiefel D. (red.) (2008), History of Insolvency and Bankruptcy from an International Perspective, Sdertrn Academic Studies 38.
Grice J.S., Dugan M.T. (2001), The Limitations of Bankruptcy Prediction Models: Some Cautions for the Researcher, Review of Quantitative
Finance and Accounting 17.
Gruszczyski M. (2012), Empiryczne finanse przedsibiorstw. Mikroekonometria finansowa, Difin, Warszawa.
GUS Zmiany strukturalne (2013), Zmiany strukturalne grup podmiotw gospodarki narodowej w rejestrze REGON, Informacje
i opracowania statystyczne GUS.
Gutirezz C.L., Olalla M.G., Olmo B.T. (2002), Shareholder's Value and Inefficiencies in European Bankruptcy Systems, 7th Global
Conference on Business & Economics.
Gutirrez C.L., Ollala M.G., Olmo B.T. (2005), Insolvency Problems in the European Union: Bankruptcy Law Orientation and Market
Valuation, SSRN.
Hall S.G. (2002), The Macroeconomic and Microeconomic Factors Influencing Financial Distress: A Comparison of Altman's 1983 and 1993
Bankruptcy Prediction Models, Doct. Dissertation, Nova Southeastern University.
Hamel G., Prahalad C.K. (1994), Competing for the Future Breakthrough, Harvard Business School Press, Boston.
Hart O. (2000), Different Approaches to Bankruptcy, NBER Working Paper Series 7921.
Haug A.A., Smith C. (2012),Local Linear Impulse Responses for a Small Open Economy, Oxford Bulletin of Economics and Statistics
74(3).
Haviv-Segal I. (2005), Bankruptcy Law and Inefficient Entitlements, Berkeley Business Law Journal 2(2).
Heinze G. (2006), A Comparative Investigation of Methods for Logistic Regression with Separated or Nearly Separated Data, Statistics
in Medicine 25.
Hodrick R.J., Prescott E.C. (1997), Postwar U.S. Business Cycles: An Empirical Investigation, Journal of Money, Credit and Banking 29.
Hol S. (2006), The Influence of the Business Cycle on Bankruptcy Probability, Statistics Norway Discussion Papers 466.
Hol S., Westgaard S., van der Wijst N. (2002), Capital Structure and the Prediction of Bankruptcy, Norwegian University of Science and
Technology.
Horrigan J.O. (1968), A Short History of Financial Ratio Analysis, The Accounting Review 43(2).
Hosmer D.W., Lemeshow S. (1989), Applied Logistic Regression, John Wiley & Sons Publishing, New York.
Hossari G. (2007), Modelling Corporate Collapse: Definitional Issues of the Collapse Event, Deakin University 5(2).

219

Hossari G., Rahman S. (2005), A Comprehensive Formal Ranking of the Popularity of Financial Ratios in Multivariate Modeling of Corporate
Collapse, Journal of American Academy of Business 6(1).
How to Manage (2013), How to Manage Collective Redundancies, Acas [www.acas.org.uk].
Hummelen J.M. (2013), Efficient Bankruptcy Law. Establishing an Assessment Framework, University of Groningen Faculty of Law
Research Paper Series 13.
Hunter J., Isachenkova N. (2006), Aggregate Economy Risk and Company Failure: An Examination of UK Quoted Firms, Kingston
University.
Hurley J., Mandl I. (2011), Public Instruments to Support Restructuring in Europe, European Foundation for the Improvement of Living
and Working Conditions, Dublin, Luxembourg.
Hyndman R., Koehler A.B., Ord J.K., Snyder R.D. (2008), Forecasting with Exponential Smoothing: The State Space Approach, Springer
Series in Statistics, Springer-Verlag, Berlin-Heidelberg.
II szansa (2011), II szansa dla przedsibiorcw. Raport z bada, PARP, Warszawa.
Informacja (2013), Informacja statystyczna o ewidencji spraw i orzecznictwie w sdach powszechnych oraz o wiziennictwie. Sprawy
gospodarcze w 2012 r., Ministerstwo Sprawiedliwoci.
Instrument badania (2010), Instrument badania kondycji przedsibiorstw, ekspertyza na zlecenie Ministerstwa Gospodarki, Instytut
Allerhanda, Krakw.
Jefferys S., Clark N. (2009), National Restructuring Seminars UK Anticipating and Managing Restructuring, National Background Paper
UK, International Training Centre International Labour Office.
Johansen S. (1996), Likelihood-based Inference in Cointegrated Vector Auto-regressive Models, Oxford University Press, New York.
Juselius K. (2007), The Cointegrated VAR Model. Methodology and Applications, Oxford University Press, New York.
Kaczmarek J., Kolegowicz K., Krzemiski P., Fijorek K. (May 2011), Koncepcja bada stanu i wynikw przedsibiorstw dla potrzeb
Instrumentu Szybkiego Reagowania. Zaoenia rozwizania System Wczesnego Ostrzegania Metody i narzdzia monitorowania
gospodarki w komponencie mikroekonomicznym, MSAP UEK, Krakw.
Kaczmarek J., Kolegowicz K., Krzemiski P., Fijorek K. (June 2011), Raport z oceny stopnia zagroenia przedsibiorstw upadoci
komponent mikroekonomiczny, Raport 1, MSAP UEK, Krakw.
Kaczmarek J., Kolegowicz K., Krzemiski P., Fijorek K. (December 2013), Raport z oceny stopnia zagroenia przedsibiorstw upadoci
komponent mikroekonomiczny, Raport 11, MSAP UEK, Krakw.
Kaczmarek J., Kolegowicz K., Krzemiski P., Fijorek K. (June 2011 April 2014), Raport z oceny stopnia zagroenia przedsibiorstw
upadoci komponent mikroekonomiczny, Raporty 112, MSAP UEK, Krakw.
Kahl M. (2001), Financial Distress as a Selection Mechanism: Evidence from the United States, Anderson School University of California.
Kaiser R., Moravall A. (1999), Estimation of the Business Cycle: A Modified Hodrick-Prescott Filter, Spanish Economic Review 1.
Karels G.V., Prakash A.J. (1987), Multivariate Normality and Forecasting of Business Bankruptcy, "Journal of Business Finance
& Accounting 14(4).
Kartus K. (2012), Upado w Polsce jest rzadko wykorzystywanym narzdziem, Forum Obywatelskiego Rozwoju 10.
Kaznowski A. (2013), Pomoc publiczna. Aspekty podatkowe, [in:] Litwiczuk H. (red.), Prawo podatkowe przedsibiorcw, Wolters Kluwer
Polska SA, Warszawa.
Klafft M., Meissen U. (2011), Assesing the Economic Value of Early Warning Systems, Proceedings of the 8th International ISCRAM
Conference, Lisbon.
Konopczak K. (2010), Analiza zbienoci cyklu koniunkturalnego gospodarki polskiej ze stref euro na tle krajw Europy rodkowo-Wschodniej oraz Pastw Czonkowskich strefy euro, Projekty Badawcze, Cz III", Narodowy Bank Polski [http://www.nbportal.
pl/r/euro/materialy_papierowe_pdf/raport_euro/re13n.pdf ].
Konopczak M., Miklus P., Wieprzowski P. (2011), Rynkowe aspekty problemw na pozagiedowym rynku walutowych instrumentw
pochodnych w Polsce w czasie globalnego kryzysu finansowego, Bank i Kredyt" 2.
Koop G., Len-Gonzlez R., Strachan R. (2010), Efficient Posterior Simulation for Cointegrated Models with Priors on the Cointegration
Space, Econometric Reviews 29.
Korobkin D.R. (1992), Value and Rationality in Bankruptcy Decisionmaking, William and Mary Law Review 33(2).
Kurcz B. (2009), Midzy ustami a brzegiem pucharu gdzie jest pomoc pastwa? Definicja pomocy pastwa w wietle orzecznictwa
ETS-u, [in:] Kurcz B. (red.), Pomoc pastwa. Wybrane zagadnienia, Wydawnictwo C.H. BECK, Warszawa.
Lando D., Nielsen M.S. (2009), Correlation in Corporate Defaults: Contagion or Conditional Independence? Copenhagen Business School,
Copenhagen.
Lenart . (2010), Procesy stochastyczne prawie okresowo skorelowane w badaniu cyklicznoci wskanikw makroekonomicznych,
Rozprawa doktorska, Uniwersytet Ekonomiczny w Krakowie, Krakw.
Lenart ., Pipie M. (2011), APC Time Series and Subsampling in Business Fluctuations Analysis: The Case of Industrial Production in
Poland, mimeo.

220

Levine L. (2007), The Worker Adjustment and Retraining Notification Act (WARN), Congressional Research Service, Domestic Social
Policy Division.
Levine R. (2005), Finance And Growth: Theory And Evidence. Handbook of Economic Growth, The Netherlands: Elsevier Science.
Levinthal L.E. (1918), The Early History of Bankruptcy Law, University of Pennsylvania Law Review and American Law Register 66(5/6).
Long S.J. (1997), Regression Models for Categorical and Limited Dependent Variables, Thousand Oaks, SAGE Publications, London.
Longhofer S.D., Peters S.R. (2000), Protection for Whom? Creditor Conflicts in Bankruptcy, Working Papers, Federal Reserve Bank of
Cleveland 9909.
Luetkepohl H. (2007), New Introduction to Multiple Time Series Analysis, Springer-Verlag, Berlin Heidelberg.
Lukason O. (2012), Pre-Bankruptcy Financial Performance of Firms: Do Size and Industry Matter?, 2nd International Conference on
Economics, Trade and Development 36.
Makridakis S., Wheelwright S.C., Hyndman R.J. (1998), Forecasting: Methods and Applications, Wiley, New York.
Manjon-Antoln M.C., Arauzo-Carod J.M. (2008), Firm Survival: Methods and Evidence, Empirica 35.
Maravall A., del Ro A. (2001), Time Aggregation and the Hodrick-Prescott Filter, Banco de Espaa Servicio de Estudios Documento
de Trabajo 0108.
Mazurek S. (2013), Statystyka upadoci (podmiotw biznesowych), Zeszyty Naukowe Uniwersytetu Szczeciskiego 760.
Mczyska E. (red.) (2005), Ekonomiczne aspekty upadoci przedsibiorstw w Polsce, SGH, Warszawa.
Mczyska E. (red.) (2009), Meandry upadoci przedsibiorstw. Klska czy druga szansa? SGH, Warszawa.
Mczyska E. (2011), Upado przedsibiorstw w kontekcie ekonomii kryzysu, [in:] Morawska S. (red.), Zarzdzanie przedsibiorstwem
w kryzysie, SGH, Warszawa.
McIntyre L.J. (1989), A Sociological Perspective on Bankruptcy, Indiana Law Journal 65(1).
Milas C., Rothman P.A., Dijk D. (2006), Nonlinear Time Series Analysis of Business Cycles, Elsevier, Amsterdam.
Mirowska D. (2013), On the Efficiency of Corporate Bankruptcy Proceedings in Poland: An Exploratory Analysis, WNE UW Conference,
Kazimierz n. Wis.
Mitroff I.I. (2001), Managing Crises Before They Happen, American Management Association, New York.
Morgado A.J., Garoupa N. (2006), Employment Policy and Bankruptcy Law Does Legal Tradition Matter?, UPF 6th Workshop on
Institutional Analysis.
Muirhead R.J. (1982), Aspects of Multivariate Statistical Theory, Wiley, New York.
Nelson Ch.R., Kang H. (1981), Spurious Periodicity in Appropriately Detrended Time Series, Econometrica 49(3).
Oglnopolskie Informatory Upadociowe.
Ohlson J.A. (1980), Financial Ratios and the Probabilistic Prediction of Bankruptcy, Journal of Accounting Research 18(1).
Oldfield L. (2005), Attitudes to Bankruptcy, The Insolvency Service.
Opler T.C., Titman S. (1994), Financial Distress and Corporate Performance, The Journal of Finance 49(3).
Osborn D.R., Smith J.P. (1989), The Performance of Periodic Autoregressive Models in Forecasting Seasonal U.K. Consumption, J Bus.
Econ. Stat. 9.
Osiewalski J. (2001), Ekonometria bayesowska w zastosowaniach, Wydawnictwo Akademii Ekonomicznej w Krakowie, Krakw.
Parker E. (1995), Schumpeterian Creative Destruction and the Growth of Chinese Enterprises, China Economic Review 6(2).
Parzen E., Pagano M. (1979), An Approach to Modeling Sezonally Stationary Time-Series, J.Econometrics 9.
Peri A., Rachedi O. (2013), The Macroeconomics of Corporate Bankruptcy Reforms, mimeo
Pilitowski B. (2013), Polski system upadociowy instytucja spoeczna i jej funkcjonalno, Studia Socjologiczne 1 (208).
Politis D., Romano J., Wolf M. (1999), Subsampling, Springer-Verlag, New York.
Provider Guidance (2010), Provider Guidance, Jobcentre Plus, Version 18 V1.0 March.
Rafael La Porta R., Lopez-de-Silanes F., Shleifer A., Vishny R.W. (1998), Law and Finance, Journal of Political Economy 106.
Raftery A.E., Newton M.A., Satagopan J.M., Krivitsky P.N. (2007), Estimating the Integrated Likelihood via Posterior Simulation Using the
Harmonic Mean Identity, Bayesian Statistics 8.
Raport z bada (2011), II szansa dla przedsibiorcw. Raport z bada, PARP, Warszawa.
Raporty COFACE.
Raport z oceny stopnia zagroenia przedsibiorstw upadoci komponent mikroekonomiczny (20112014), Kaczmarek J. (red.),
Fijorek K., Kolegowicz K., Krzemiski P., Raporty 112, Maopolska Szkoa Administracji Publicznej Uniwersytetu Ekonomicznego
w Krakowie, Krakw.
Report (2011), Early Warning Systems: Can More be Done to Avert Economic and Financial Crises?, The Institute of Chartered
Accountants in Australia.
Restructuring (2003), Restructuring, Bankruptcy and a Fresh Start, Komisja Europejska.
Robert C.P., Casella G. (2004), Monte Carlo Statistical Methods, Springer, New York.
Robinson H. (2009), Rapid Response Service, JobCentre Plus, Yorkshire & The Humber.
Roe M.J. (2014), Structural Corporate Degradation Due to Too-Big-to-Fail Finance, Harvard Law School.

221

Rose A.K., Spiegl M.M. (2009), Cross-Country Causes and Consequences of the 2008 Crisis: Early Warning, Federal Reserve Bank of San
Francisco 17.
Ruling (2004), Ruling of the Supreme Court of November 19, 2004, V CK 231/2004, [in:] Gazeta Prawna 2006/67.
Salzman J. (2009), Rapid Response and TAA. An Occasional Paper Prepared as Part of the Evaluation of the Trade Adjustment Assistance
Program, Social Policy Research Associates, Oakland, CA.
Sautner Z., Valdimirov V. (2013), Indirect Bankruptcy Costs and Bankruptcy Law, report, Universiteit van Amsterdam.
Schwartz A. (2005), A Normative Theory of Business Bankruptcy, Yale Law School Faculty Scholarship Series 303.
Shenk-Hopp K.R. (2001), Economic Growth and business Cycles: A Critical Comment on Detrending Time Series, Studies in Nonlinear
Dynamics and Econometrics 5(1).
Shumway T. (2001), Forecasting Bankruptcy More Accurately A Simple Hazard Model, Journal of Business 74.
Sims C.A. (1980), Macroeconomics and reality, Econometrica 48.
Skrzypczyski P. (2006), Analiza synchronizacji cykli koniunkturalnych w strefie euro, Materiay i Studia Narodowego Banku Polskiego 210.
Skrzypczyski P. (2008), Wahania aktywnoci gospodarczej w Polsce i strefie euro, Materiay i Studia Narodowego Banku Polskiego 227.
Skrzypczyski P. (2010), Metody spektralne w analizie szeregw czasowych narzdziem badania waha aktywnoci gospodarczej,
Materiay i Studia Narodowego Banku Polskiego 252.
Smart C.F., Thompson W.A., Vertinsky I. (1978), Diagnosing Corporate Effectiveness and Susceptibility to Crisis, "Journal of Business
Administration 9.
Sobehart J.R., Stein R.M., Mikityanskaya V., Li Li (2000), Moodys Public Firm Risk Model: A Hybrid Approach To Modeling Short Term
Default Risk, Moodys Rating Methodology (March), Moodys Investors Service, Inc., New York.
Strachan R.W., Inder B. (2004), Bayesian Analysis of the Error Correction Model, Journal of Econometrics 123.
Suarez J., Sussman O. (2007), Financial distress, bankruptcy law and the business cycle, Annals of Fonance 3(1).
Sulivan T.A., Warren E., Westbrook J.L. (1987), The Use of Empirical Data in Formulating Bankruptcy Policy, Law and Contemporary
Problems 50(2).
The Dun & Bradstreet Corporation (1997), Business Failure Record, The Dun & Bradstreet Corporation.
Thevnin C. (2003), A Comparative Examination of Bankruptcy Prediction: Altman MDA Study versus Luther ANN Study. A Test of Predictive
Strength between the Two Techniques, Doct. Dissertation, Nova Southeastern University.
Thornburn K.S. (2000), Bankruptcy Auctions: Costs, Debt Recovery, and Firm Survival, Journal of Financial Economics 58.
Tierney L. (1994), Markov Chains for Exploring Posterior Distributions, Annals of Statistics 22.
Tong T.W., Reuer J.J. (2007), Corporate Investment Decisions and the Value of Growth Options, Managerial and Decision Economics
28(8).
Valdimirov V. (2007), Soft vs. Tough Bankruptcy Law Evidence from the U.S. and Germany, Goethe University.
Van de Casteele K., Hocine M. (2008), Favouring Certain Undertakings of the Production of Certain Goods: Selectivity, [in:] Mederer W.,
Pesaresi N., Van Hoof M., EU Competition Law. Volume IV. State Aid. Book One, Claeys & Casteels Publishing BV, Deventer.
Walas-Trbacz J. (2012), Systemy wczesnego ostrzegania w przedsibiorstwie, Ekonomika i Organizacja Przedsibiorstwa 9.
Warner J.B. (1997), Bankruptcy, Absolute Priority, and the Pricing of Risky Debt Claims, Journal of Financial Economics 4.
Wawrzyniak B. (1999), Odnawianie przedsibiorstwa, Poltext, Warszawa.
Weiss L.A. (1990), Bankruptcy Resolution: Direct Costs and Violation of Priority of Claims, Journal of Financial Economics 27(2).
White M.J. (1980), Public Policy toward Bankruptcy: Me-First and Other Priority Rules, The Bell Journal of Economics 11(2).
White M.J. (1989), The Corporate Bankruptcy Decision, The Journal of Economic Perspectives 3(2).
White M.J. (1996), The Costs of Corporate Bankruptcy: A US-European Comparison. Corporate Bankruptcy, Cambridge University Press.
White M.J. (2011), Corporate and Personal Bankruptcy Law, Annual Review of Law and Social Science 7.
Wieczerzyska B. (2009), Kryzys w przedsibiorstwie, CeDeWu, Warszawa.
Wilcox J.W. (1970), A Simple Theory of Financial Ratios as Predictors of Failure, MIT Working Paper 495.
Williamson B.C. i in. (1997), Data Compilation and Dissemination, [in:] Bankruptcy: The Next Twenty Years, National Bankruptcy Review
Commission Final Report.
Wrblewska J. (2010), Modele i metody bayesowskiej analizy kointegracji, Wydawnictwo Uniwersytetu Ekonomicznego w Krakowie,
Krakw.
Zavgren C. (1983), The Prediction of Corporate Failure: The State of the Art, "Journal of Accounting Literature 2.
Zellner A. (1971), Introduction to Bayesian Inference in Econometrics, John Wiley, New York.
Zhang G., Hu M.Y., Patuwo B.E., Indro D.C. (1999), Artificial Neural Networks in Bankruptcy Prediction: General Framework and CrossValidation Analysis, European Journal of Operational Research 16.
Zhou C. (2001), An Analysis of Default Correlations and Multiple Defaults, The Review of Financial Studies 14(2).
Zmijewski M.E. (1984), Methodological Issues Related to the Estimation of Financial Distress Prediction Models, Journal of Accounting
Research 22.

222

Legal Regulations
Economic Dislocation and Worker Adjustment Assistance Act 1988.
Fair Labor Standards Act (2009).
Commision Regulation (EU) No 1407/2013 of 18 December 2013 on the application of Articles 107 and 108 of the Treaty on the
Functioning of the European Union to de minimis aid (OJ L 352/1, 24.12.2014).
Commision Regulation (EC) No 800/2008 of 6 August 2008 declaring certain categories of aid compatible with the common market
in the application of Articles 87 and 88 of the Treaty (General block exemption Regulation) (OJ L 214/3, 9.8.2014).
Council Regulation (EC) No 659/1999 of 22 march 1999 laying down detailed rules for the application of Article 93 of the EC Treaty
(OJ L 083, 27.3.1999).
Rozporzdzenie Rady Ministrw z 27 lipca 1999 r. w sprawie sposobu i metodologii prowadzenia i aktualizacji rejestru podmiotw
gospodarki narodowej, w tym wzorw wnioskw, ankiet i zawiadcze, oraz szczegowych warunkw i trybu wspdziaania
sub statystyki publicznej z innymi organami prowadzcymi urzdowe rejestry i systemy informacyjne administracji
publicznej (Dz.U. z 1999 r., Nr 69, poz. 763).
Rozporzdzenie Ministra Sprawiedliwoci z 30 listopada 2011 r. w sprawie szczegowego sposobu prowadzenia rejestrw
wchodzcych w skad Krajowego Rejestru Sdowego oraz szczegowej treci wpisw w tych rejestrach (Dz.U. z 2011 r.,
Nr273, poz. 1616).
The Civil Rights Act 1964.
The National Labor Relations Act 1935.
The Occupational Health&Safety Act 1970.
The Worker Adjustment and Retraining Notification (WARN) Act 1989.
WARN (2003) The Worker Adjustment and Retraining Notification (WARN) Act, Employers Guide to Advance Notice of Closings and
Layoffs, United States of Department of Labor, July 2003.
Ustawa z 17 listopada 1964 r. Kodeks postpowania cywilnego (tekst jednolity, Dz.U. z 2014 r., poz. 101).
Ustawa z 29 wrzenia 1994 r. o rachunkowoci (Dz.U. z 1994 r., Nr 121, poz. 591).
Ustawa z 20 sierpnia 1997 r. o Krajowym Rejestrze Sdowym (Dz.U. z 1997 r., Nr 121, poz. 769).
Ustawa z 28 lutego 2003 r. Prawo upadociowe i naprawcze (Dz.U. z 2003 r., Nr 60, poz. 535 z pn. zm.).
Ustawa z 2 lipca 2004 r. o swobodzie dziaalnoci gospodarczej (Dz.U. z 2004 r., Nr 173, poz. 1807).
Ustawa z 6 grudnia 2006 r. o zasadach prowadzenia polityki rozwoju (Dz.U. z 2009 r., Nr 84, poz. 712 z pn. zm.).
Workforce Investment Act 1998.
Commity Guildelines on State aid for rescuing and restructuring firms in difficulty (OJ C 244, 1.10.2014).

223

You might also like