Does Decentralization Improve Perceptions of

Accountability? Attitudinal Evidence from Colombia

Maria Escobar-Lemmon Texas A&M University
Ashley D. Ross Sam Houston State University

Decentralization is argued to create incentives for local and regional politicians to be more responsive and accountable to their
constituents, but few studies have directly tested this claim. We use survey data from Colombia to examine individual-level
evaluations of the degree to which decentralization prompts citizens to view department government as more accountable. We
estimate the effect of administrative, fiscal, and political decentralization, controlling for participation, political knowledge,
confidence in government, education, and income on perceptions of accountability. Our results indicate that administrative
and fiscal decentralization improve perceptions of accountability, while political decentralization does not.

F
ew recent policy reforms have generated as much
enthusiasm as the decentralization of power from
national to subnational governments. Decentral-
ization promised to increase governments responsive-
ness to citizen needs, improve the effectiveness of the
allocation of public goods, mobilize citizens through
new venues of local political participation, increase po-
litical accountability, and generally improve democracy
from below (Bardan and Mookherjee 2006; Blair 2000;
Daughters and Harper 2007). In both the global north
and south, subnational governments were given increased
fiscal, political, and administrative responsibilities. This
move came at the requests of activists seeking more lo-
cal control, national governments seeking to enhance
states rights (e.g., Reagans New Federalism) and/or re-
formers hoping to improve the provision of services.
Meanwhile, international organizations such as the World
Bank and the International Monetary Fund (IMF) pro-
moted decentralization as part of a strategy to realize the
twin goals of accountability and efficiency (Rondinelli
1989).
Decentralization could, consequently, be thought of
as the architecture making more accountable local gov-
ernment possible. If decentralization works as predicted,
in countries (or parts of a country) where it has become
most institutionalized or implemented most extensively,
government should be (and be seen as) more account-
able by citizens. Places where decentralization has not
been as extensive, in contrast, should be characterized
by a consensus view that there is less accountability of
government.
Despite the prediction that decentralization should
improve accountability, previous work has failed to fully
test if this is in fact the case. At its broadest, accountability
is associated with the process of being called to account
to some authority for ones actions (Jones 1992, 73; cited
in Mulgan 2000, 555). This has led to studying account-
ability as voters control of politicians through elections

Maria Escobar-Lemmon is Associate Professor of Political Science, Texas A&M University, College Station, TX 77843-4348
(m escobar@pols.tamu.edu). Ashley D. Ross is Assistant Professor of Political Science, Sam Houston State University, Huntsville, TX
77341 (ashleydross@gmail.com).
Kim Yi Dionne, Tulia Falleti, Kim Q. Hill, Marisa Kellam, Paul Kellstedt, and Michelle Taylor-Robinson as well as the journals anony-
mous reviewers provided helpful comments on this article. Previous versions were presented at the (2011) annual meeting of the
Midwest Political Science Association and the (2011) annual meeting of the Public Choice Society, where we also received helpful
comments and suggestions. The staff at LAPOP and especially Juan Carlos Rodrguez Raga provided useful information about the
LAPOP data. We are also grateful to Leopoldo Fergusson for help in obtaining the fiscal decentralization data. All errors and omis-
sions are nonetheless our own. All data and replication files can be found at http://dvn.iq.harvard.edu/dvn/faces/study/StudyPage.xhtml?
globalId=hdl:1902.1/21236&versionNumber=1.
American Journal of Political Science, Vol. 58, No. 1, January 2014, Pp. 175188

C 2013, Midwest Political Science Association DOI: 10.1111/ajps.12043

175
176
by sanctioning poor performance and rewarding good
behavior.1 For instance, Manin, Przeworski, and Stokes
define governments as accountable if citizens can dis-
cern representative from unrepresentative governments
and can sanction them appropriately, retaining in office
those incumbents who perform well and ousting from
office those who do not (1999, 10). Elections then are
a retrospective tool (Cheibub and Przeworksi 1999) for
shaping politicians behavior.2 Certainly, there have been
efforts to examine whether there is greater electoral ac-
countability in subnational elections (e.g., Gelineau and
Remmer 2006; Khemani 2001), but this focuses on only
one aspect of accountability.
ODonnell argues that in addition to vertical ac-
countability between voters and elected officials, there
exists what he calls horizontal accountability, which is
the existence of state agencies that are legally enabled and
empowered, and factually willing and able, to take actions
that span from routine oversight to criminal sanctions
or impeachment in relation to actions or omissions by
other agents or agencies of the state that may be qualified
as unlawful (1999, 30). ODonnells work makes clear
that voters are not the only ones who can sanction and
reminds us that unelected officials also play a role. For
example, the behavior of unelected bureaucrats may also
shape citizens views of whether government is account-
able. Other work has moved beyond government to em-
phasize civil society and societal accountability whereby
citizen associations, social movements, and media work to
bring public officials misdeeds to light, leading to public
condemnation and possible action by watchdog groups
within government (Rich and Gomez 2012; Smulovitz
and Peruzzotti 2000). Alternatively, Kitschelt et al. (2009)
propose measuring the existence of democratic account-
ability by studying the provision of clientelistic goods by
parties.
For us, accountability applies to government broadly,
not only elected officials, but also bureaucrats, and in-
cludes providing information and responsiveness as well
as sanctioning. Consequently, we see accountability much
as Schedler does: [T]he notion of political accountability
1
There is a voluminous literature examining whether voters punish
politicians for poor economic performance based on the U.S. case.
Remmer and Gelineau (2003) extend this work to Argentina, and
Gelineau and Remmer (2006) and Khemani (2001) offer evidence
that voters are more vigilant in monitoring local than national
officials in Argentina and India, respectively.
2
Elections are not universally seen as retrospective. Fearon (1999,
82) argues that voters see elections more as a chance to pick good
types, making them prospective, although he still sees this as im-
plying accountability.
MARIA ESCOBAR-LEMMON AND ASHLEY D. ROSS
carries two basic connotations: answerability, the obliga-
tion of public officials to inform about and to explain
what they are doing; and enforcement, the capacity of ac-
counting agencies to impose sanctions on powerholders
who have violated their public duties . . . It implies sub-
jecting power to the threat of sanctions; obliging it to be
exercised in transparent ways; and forcing it to justify its
acts (1999, 14).
In this article, we focus on the answerability dimen-
sion of accountability and directly test whether adminis-
trative, fiscal, and political decentralization determine if
citizens see government as more open, transparent, and
responsive. By emphasizing this aspect of accountability,
we fill a significant hole in the literature and advance our
understanding of the consequences of decentralization.
Using survey data from Colombia, we test whether more
extensive decentralization increases citizens perceptions
of the accountability of government after taking account
of individual-level attributes and attitudes. Our first step,
however, is to review the extant literature to ground our
expectations.
Context Matters: The Incentives to Be
Accountable under Decentralization
Falleti defines decentralization as a process, as the set
of policies, electoral reforms or constitutional reforms that
transfer responsibilities, resources or authority from higher
to lower levels of government (2010, 34; italic in origi-
nal). For her, administrative decentralization entails the
set of policies that transfer the administration and de-
livery of social services such as education, health, so-
cial welfare, or housing to subnational governments and
may involve devolution of decision-making authority
over these policies, but this is not a necessary condi-
tion (Falleti 2005, 328). Fiscal decentralization refers
to the set of policies designed to increase the revenues or
fiscal autonomy of subnational governments, whereas
political decentralization includes reforms to open sub-
national representation and policies to devolve political
authority or electoral capacities to subnational actors
(Falleti 2005, 32829). Because she is fundamentally con-
cerned with whether decentralization really increases sub-
national power, autonomy is central to her conceptual-
ization of decentralization. While attitudes about gov-
ernment accountability will be shaped by an individuals
experience, they will also be determined by the context in
which that individual lives. We expect the implementation
of administrative, fiscal, and political decentralization to
DOES DECENTRALIZATION IMPROVE ACCOUNTABILITY?
produce outcomes that affect individual attitudes about
subnational government.
Decentralization, in general, is argued to enhance ac-
countability by bringing government closer to citizens,
thereby making subnational government more transpar-
ent, involving citizens in decision making, and facilitating
the monitoring of subnational government (Blair 2000;
Burki, Perry, and Dillinger 1999; Manor 1999; Peterson
1997). Decentralization is argued to reduce the scale of the
principal-agent problem, as informational problems are
less severe at the local level and smaller constituencies
facilitate the monitoring of the performance of elected
representatives (Lederman, Loayza, and Soares 2005, 5).
The improved ability to monitor local officials and de-
mand change can help control corruption (Shah 2006),
and some scholars find that decentralization reduces cor-
ruption (Fisman and Gatti 2002; Shah, Thompson, and
Zou 2004), which could in turn improve accountability
or perceptions of accountability.3
Decentralization can engender accountability as sub-
national governments become more transparent and
share more information; it can also improve perceptions
of accountability by involving citizens in decision mak-
ing. The hallmark example of this is the highly successful
participatory budgeting process in the city of Porto Ale-
gre, Brazil, made possible by the decentralized nature
of the Brazilian state. This process involved many who
were previously marginalized, and it also improved per-
ceptions of government accountability as citizens became
aware of the citys budget constraints and were involved
in determining which projects should be funded (Smith
2009). The participatory budgeting experience in Porto
Alegre provides one clear example of the way in which
fiscal decentralization has the potential to improve ac-
countability. Another example comes from the work of
Faguet (2004), who shows that in Bolivia, giving munic-
ipalities control over public investment budgets allowed
them to target resources to a few high-priority projects
where local need was greatest, rather than spreading in-
vestment across multiple sectors. His analysis indicates
that local decisions were more suited to local needs than
central decisions, especially in the smallest and poorest
municipalities.
Political decentralization, specifically, is argued to in-
crease the accountability and responsiveness of locally
elected officials by establishing local elections and open-
3
However, Treisman (2000) finds that federal states score a half to
one point higher (more corruption) on the Transparency Interna-
tional Corruption scale, and others such as Prudhomme (1995)
argue that decentralization creates more opportunities for corrup-
tion to occur.
177
ing meaningful spaces for subnational political competi-
tion. Linz and Stephan (1996) and Rose-Ackerman (1999)
show that competitive elections are helpful in making sure
that elected officials are accountable for their actions.4
Faguet argues compellingly,
Substantively competitive politics is character-
ized by a greater diversity of ideas and policy
proposals competing for public favor, and hence
a broader representation of the publics needs. A
direct result of this is improved responsiveness
and public accountability from government of-
ficials, as opposition parties continuously search
for advantage over their rivals. A substantively
uncompetitive politics, in contrast, leads to lower
levels of policy innovation and entrepreneurial-
ism, which, in turn, reduce the level of oversight
that local government institutions are subject to.
This will tend to result in a less responsive, less
accountable local government. (2009, 43)
Local officials (or parties) concerned about reelec-
tion become more accountable. Where there is a lack of
electoral competition, Bardhan and Mookherjee (2000)
argue that local governments are more likely to be cap-
tured by elites, resulting in distortions in the provision
of public goods, which Faguet (2009) finds happened in
noncompetitive municipalities in Bolivia.
Fiszbeins (1997) study of Colombian municipalities
following the implementation of decentralization found
that political competition was critical in electing leaders
who engaged in innovation, successfully developed local
capacity, and thus improved the overall quality of gov-
ernment. More recent work by Magaloni, Daz-Cayeros,
and Estevez (2007) demonstrates that heightened politi-
cal competition reduced clientelism in Mexico, although
Weitz-Shapiro (2012) suggests that competition must be
combined with lower levels of poverty to effectively re-
duce clientelism. Thus, where political decentralization
has created spaces for electoral competition, we would
expect accountability to improve.
The establishment of local elections (or the conver-
sion of appointed positions into elected posts) is fre-
quently used as a measure of political decentralization,
but the existence of those elections may not be a suffi-
cient indicator of political decentralization. Nor may it
be sufficient to cause improved accountability. Eaton and
Schroeder (2010, 171) argue it is not, preferring instead
4
Canes-Wrone (2006) and Ferejohn (1986) also link competi-
tiveness to government responsiveness. Hobolt and Klemmensen
(2008) find that higher levels of political contestation lead to more
responsive executives in the United States (but not Britain).
178
to focus on features of electoral rules and internal char-
acteristics of political parties that empower local lead-
ers as measures of political decentralization. Like Faguet
(2009), they see parties and the dynamics of electoral
competition as important for connecting civil society and
government.
Of course, not everyone agrees that decentralization
will improve accountability (or bring other promised ben-
efits). Some are critical of decentralization in general (i.e.,
Prudhomme 1995); others caution that the details (or
specifics of its implementation) and not the quantity de-
termine if it is beneficial (Prudhomme and Shah 2002).
Keefer, Narayan, and Vishwanath (2006) show that ad-
ministrative decentralization can backfire; decentraliza-
tion may have matched policies to preferencesbut this
meant reducing the availability of girls schools in rural
Pakistanwhich was not the national governments goal.
Of the nine countries Crook and Sverrisson (2001) ex-
amine, they conclude that decentralization only reduced
poverty in West Bengal. If outcomes worsen as a result
of administrative (or fiscal) decentralization, it might not
improve perceptions of accountability.
Treisman (2007) concludes that political decentral-
ization is not clearly or consistently better than cen-
tralization at making it easier for voters to coordinate
and sanction government or for improving accountabil-
ity in general. If elite capture (Bardhan and Mookerjee
2000) occurs, political decentralization might not pro-
duce increased accountability. Moreover, Falleti (2005,
2010) warns that the sequence in which political, ad-
ministrative, and fiscal decentralization are adopted
determines whether decentralization empowers local gov-
ernments or leaves them beholden to the national govern-
ment, and a dependent government may be seen as less
accountable.
The implementation of administrative, fiscal, and po-
litical decentralization offers opportunities to enhance
subnational government accountability, but these might
or might not be realized. We seek to test empirically
whether this has actually occurred. Accordingly, we ex-
pect:
H1: Citizens living in a context of more extensive admin-
istrative decentralization will perceive their subna-
tional government as more accountable;
H2: Citizens living in a context of greater fiscal decentral-
ization will perceive their subnational government as
more accountable; and
H3: Citizens living in a context of greater political de-
centralization will perceive their subnational govern-
ment as more accountable.
MARIA ESCOBAR-LEMMON AND ASHLEY D. ROSS
Individual-Level Determinants: Who
Sees Government as Accountable?
The above discussion highlights the extent to which de-
centralization matters, positing that context has a direct
impact upon how accountable citizens perceive govern-
ment to be. However, we do not expect perceptions to
be constant across all respondents in a department. In-
stead, two individuals living in the same department may
think about accountability differently based on their at-
titudes about government, the extent to which they are
involved and participate in government, and their knowl-
edge about politics. We discuss each of these in turn.
The poor behavior of government officials (such as
not providing information) obviously damages govern-
ment evaluations, but, if it becomes extreme (corrupt,
morally suspect), it can actually damage support for
democracy in general (Weitz-Shapiro 2008). Thus, at one
extreme, complete mistrust of politicians might be linked
to an unwillingness on the part of citizens to monitor
politicians behavior and engage in democratic activities
(Cleary and Stokes 2006). At the same time, a degree of
skepticism is healthy for political life. It reflects a vigilant
and active society willing to exercise control over govern-
ment if politicians have violated the rules of the game
(Hardin 1999). Therefore, since some skepticism is good
for democracy and is indicative of mature democracies
(Cleary and Stokes 2006), but too much skepticism (or
cynicism) undermines democratic functioning, we hy-
pothesize:
H4: Respondents who are optimistic or skeptical, as op-
posed to cynical, about government will be more
likely to perceive government as accountable.
In addition to optimism about government, political
participation should affect perceptions of accountabil-
ity. In her examination of how marginalized groups used
democratic decentralization in Uganda and South Africa
to demand accountability from leaders, Dauda (2006)
highlights the role and importance of community par-
ticipation. Similarly, Fiszbein notes that in addition to
political competition, which brought to office innovative
leaders who worked to improve capacity, community par-
ticipation was also important. He finds participation in-
creased demands for effective local governments . . . and
forced government accountability (1997, 1034).
This suggests that individual political participation
may be an important predictor of perceptions of account-
ability. Those who are more actively involved in local gov-
ernment may perceive greater accountability than those
DOES DECENTRALIZATION IMPROVE ACCOUNTABILITY?
who are either not involved or simply vote. Moreover,
part of how accountability is envisioned to work involves
citizens taking steps to monitor government and demand
transparency. While independent media and opposition
political parties fulfill this role, the studies mentioned
above highlight how community meetings and town halls
also serve this function. Therefore, to the extent that indi-
viduals have realized that accountability is a result of their
actions, they may view government as more accountable.
Thus, we hypothesize:
H5: Respondents with higher levels of local political par-
ticipation will be more likely to perceive government
as accountable.
Finally, we anticipate that the amount of information
citizens have about government will influence their per-
ceptions of government accountability. Because we are
interested in examining accountability as answerability,
we expect an individuals level of political knowledge to
matter. For example, if local government posts the bud-
get online and citizens have a chance to review it and
leave feedbacksomething we would say represents a
high degree of accountabilitywe would expect those
individuals who knew this to see their government as
accountable, whereas their less informed counterparts
might perceive lower accountability. This leads to our
final hypothesis:
H6: Respondents with higher levels of political knowl-
edge will be more likely to perceive government as
accountable.
Modeling Accountability: Case
Selection, Data, and Methods
Our analysis focuses on Colombia, specifically on exam-
ining accountability in the context of decentralization to
the Colombian departments.5 Decentralization is rela-
tively advanced in Colombia, a fact highlighted by the
countrys self-description in its constitution as a decen-
tralized, unitary republic.6 Falleti (2005, 2010) argues de-
centralization has proceeded in Colombia in a manner
that offers the greatest chance for the successful trans-
fer of autonomy to subnational governments, as political
5
Government in Colombia consists of three levels: the national,
the department (an intermediate level), and the municipal (which
includes both urban and rural areas).
6 and health expenditures (2005, 183).
For an excellent overview of the evolution of decentralization in
Colombia, see Angell, Lowden, and Thorp (2001) or Falleti (2010).
For fiscal decentralization, see Alesina, Carrasquilla, and Echavarra
(2005).
179
decentralization preceded administrative and fiscal de-
volution. This sequence has given governors and may-
ors those legal, economic, and administrative tools they
need for good, creative government. The extent to which
this is realized varies widely with the heterogeneous con-
stellation of Colombian departments and municipalities
(2010, 149). This variation in decentralization across de-
partments, but within a country where decentralization
policies have progressed successfully and to a high de-
gree, provides an excellent set of cases to accurately test
if accountability perceptions are indeed shaped by fiscal,
administrative, and political decentralization.
Significant variation in decentralization exists at the
departmental level in Colombia, as illustrated in Figure 1.
While laws decentralizing responsibility for service pro-
vision or establishing subnational elections have applied
everywhere, the implementation of those reforms and
their success have varied considerably across the country,
providing important intracountry variation necessary for
hypothesis testing while holding many national factors
constant (see Fiszbein 1997). In one sense, political de-
centralization is uniformly well advanced in the country,
with mayoral elections having been adopted in 1986 and
gubernatorial in 1991. However, there has been signifi-
cant variation across departments (and municipalities
see Angell, Lowden, and Thorp 2001) in the extent to
which new parties and individuals have taken advantage
of the opportunities offered by these elections to govern
at the subnational level (meaning the competitiveness of
the electoral environment varies). Moreover, fiscal decen-
tralization has advanced unequally across departments.
Most departmental revenue comes from formulaic and
guaranteed revenue sharing under the Situado Fiscal7 and
royalties based on the production of minerals (almost ex-
clusively oil and coal) that especially benefit departments
where they are produced. Departments can minimize this
dependency by taxing automobile registration as well as
liquors and tobacco; they can also contract debt, enabling
them to run fiscal deficits.8 Different resource endow-
ments combined with more or less aggressive behavior in
tax collection and borrowing have, therefore, resulted in
variation in fiscal decentralization across the country.
7
The national government collects most taxes (around 80%) and is
required to transfer a significant share to departments (and munic-
ipalities) through what is known as the Situado Fiscal. It is highly
conditional, with Alesina, Carrasquilla, and Echavarra estimating
that about 80% of all transfers are allocated by rules to education
8
Taxes on liquor (33%), beer (32%), tobacco (7%), and motor vehi-
cles (19%) account for 91% of departmental tax revenue (Alesina,
Carrasquilla, and Echavarra 2005, 17879).
180 MARIA ESCOBAR-LEMMON AND ASHLEY D. ROSS

FIGURE 1 Variation in Decentralization Types Across Colombian

Departments

100

80

Fiscal Decentralization 60

40
Admin. Decen.

20

0
-10 0 10 20 30 40
-20
Political Decentralization

Note: All 32 Colombian departments are plotted. Bubble size marks the extent of adminis-
trative decentralization, which is measured as the DANE administrative autonomy ratings
(extent of sufficiency of resources and quality of public services, among other assessments)
for the year 2007 based on survey responses from departmental functionaries. The adminis-
trative autonomy ratings have been raised to the tenth power to visually maximize variation
among cases. Political decentralization on the x-axis is measured as the average of the margin
of victory of gubernatorial elections for the years 1997, 2000, and 2003. Fiscal decentraliza-
tion on the y-axis is measured as the percent of own-source revenue to total departmental
revenue, averaged over the years 20002006. See the online supporting information for a
more detailed description of variable coding and sources. See also Table SI.2 for a table
reporting the values of administrative, fiscal, and political decentralization plotted above
for each department.

We examine accountability in nonelectoral forms,
focusing specifically on answerability or the obligation
of public officials to inform about and to explain what
they are doing (Schedler 1999, 14). To explore account-
ability, we rely on data from the Latin American Public
Opinion Project (LAPOP) Democracy Audit and Cul-
ture of Democracy surveys of Colombia for the years
2004 and 2006.9 These LAPOP surveys include citizen
assessments of department-level government in consult-
ing them, making decisions and plans public, and shar-
ing information.10 Our measure of accountability is the
average of three citizen ratings: In your opinion, your
departmental government: (1) consults citizens before
making decisions, (2) makes plans and decisions public,
and (3) shares information openly and on time. Possible
9
These surveys are ideal for testing how perceptions of accountabil-
ity vary across departments because there are extensive questions
exploring accountability at multiple levels of government which
are not found in a comparable way in other LAPOP surveys.
10
Our empirical measure is perceived accountability, even though
most of the theoretical predictions generated from the literature
would seem to imply actual accountability. Because we assert that
citizens are relatively good judges of whether government is being
accountable in terms of answerability, we feel the two can be treated
interchangeably.
responses were never (1), almost never (2), once in
a while (3), almost always (4), and always (5).11 Be-
cause this measure includes multiple ratings, it is a more
robust indicator than one that relies on a single survey
response.
In order to test our hypotheses that citizens living
under regimes with more extensive administrative, fiscal,
and political decentralization will be more likely to per-
ceive government as being more accountable (H1, H2,
and H3), we include indicators of decentralization that
capture the results fromrather than the existence or
adoption ofdecentralization policies. We focus on rat-
ings of departmental autonomy in the administration of
responsibilities as an indicator of the results of adminis-
trative decentralization. We examine fiscal autonomy of
departmental governments as the degree to which they
have been successful at generating independent income
resulting from the opportunities afforded by fiscal de-
centralization, and we utilize electoral competition, mea-
sured as margin of victory, as an indicator of the out-
come of political decentralization within a department.
11
For more details on variable coding and full question wording,
see the online supporting information. For descriptive statistics,
see Table SI.1 of the online supporting information.
DOES DECENTRALIZATION IMPROVE ACCOUNTABILITY?
Measuring decentralization in this manner has the ad-
vantage of testing claims about this policy program from
multiple angles. It also pushes the scholarship on decen-
tralization to focus on more than parchment institutions
by examining the outcomes produced by decades of im-
plementation.
We employ an indicator of administrative autonomy
constructed from expert surveys conducted by the Ad-
ministrative Department of National Statistics (DANE)
to measure administrative decentralization. The survey
queried numerous public functionaries in each depart-
ment, including elected and appointed officials as well as
career bureaucrats, asking questions to gauge departmen-
tal performance in multiple areas related to decentraliza-
tion, such as resources, management, labor, and public
services. Responses to survey questions were averaged by
DANE to produce a score for each department ranging
from 1 to 5 (very low, low, medium, high, very high) in 10
areas, which was aggregated into two conceptual dimen-
sions: administrative autonomy and political autonomy.12
We use the average of the administrative autonomy indi-
cators only to measure administrative decentralization.
The DANE administrative autonomy measure is an
average of ratings across five areas of administrative de-
centralization.13 The first is credibility of rules, which cap-
tures perceptions about procedures for hiring and con-
tracting of goods and services. Respondents were asked
about transparency in hiring personnel and staff as well
as the frequency with which contracts were awarded on
the basis of personal/familial ties or political reward. The
second area is resource credibility, which assesses percep-
tions about the sufficiency and ownership of fiscal and
human resources. For example, respondents were asked
to gauge the sufficiency of the departmental budget to
meet responsibilities (handed down as a result of admin-
istrative decentralization) and how personnel turnover
affected the development of departmental programs. The
third theme is management for results, which focused
on the implementation of management tools to achieve
departmental government objectives. Questions in this
area spanned topics from evaluation systems to training
programs to strategic actions aimed at achieving depart-
mental government goals. The fourth area included in
administrative autonomy is public territory management,
which explores perceptions about the obstacles that exist
in the management of departmental tasks. Respondents
12
The full report is available at http://www.dane.gov.co/files/
EDI/INFORME RG EDI DPTAL 2007.pdf.
13
See the online supporting information, Table SI.3, for details on
the scores for each component of the administrative autonomy
rating by department.
181
were asked, among other items, about the degree to which
their department has fulfilled its administrative and fis-
cal commitments to the national government. Also asked
were questions about the quality of public services pro-
vided by the department. The final theme encompassed in
administrative autonomy is labor welfare, which taps into
perceptions about the degree of well-being or satisfaction
the respondent has with regard to the work incentives
the department government provides. Questions in this
area explored, for example, the respondents perceptions
of improvement in her job skills, contributions to depart-
mental citizen satisfaction through her work, and fairness
of her salary. Taken as a whole, the components of the ad-
ministrative autonomy rating (which vary from 2.97 to
3.86) capture functionaries perceptions of departmen-
tal government authority and control over employment,
service provision, and other management issuesall ar-
eas that have been recommended as sound measures of
administrative decentralization (see Eaton and Schroeder
2010).
We measure fiscal decentralization as fiscal auton-
omy or own-source revenues as a percentage of total
subnational revenues, which is a widely accepted indica-
tor of this type of decentralization (Eaton and Schroeder
2010, 181). We calculate fiscal autonomy as 100 minus
the percentage of departmental government transfers and
royalties (provided by the national government) of total
department income, averaged for the years 20002006.
This measure, which ranges from 18 to 84%, captures the
fiscal independence of the departmental government
the closer this measure is to 100%, the more fiscally au-
tonomous the departmental government. These data were
obtained from the National Planning Department (DNP)
of Colombia.
In addition to administrative and fiscal decentraliza-
tion, we test the effects of political decentralization, which
we operationalize using electoral competition. Political
decentralization is defined as the devolution of authority
and/or electoral capacities to subnational actors (Falleti
2005). The most common measure of political decen-
tralization, whether or not a country holds subnational
elections, does not vary within Colombia and is insuf-
ficient to capture the extent of political decentralization
(Eaton and Schroeder 2010, 172). As an alternative, Eaton
and Schroeder (2010) recommend focusing on electoral
dynamics. In a similar vein, Schneider (2003) frames de-
centralization in terms of representationmapping cit-
izen preferences on policy decisions at the local level.
For Schneider, the clearest measure of representation is
subnational elections because they require candidates to
compete and make appeals to citizens in local juris-
dictions (2003, 40). We agree with these scholars that
182
capturing the full extent of political decentralization in-
volves subnational elections but in a manner that takes
into account electoral competition. Therefore, we exam-
ine subnational electoral competition as an observable,
measurable, and varying (within a country) measure of
political decentralization.14
We employ the average margin of victory in the gover-
nors race in three recent elections (1997, 2000, and 2003)
as a measure of political decentralization.15 The margin
of victory is calculated as the difference between the win-
ning partys vote share and the second-place partys vote
share and varies from 3 to 35%. We treat smaller margin
of victory averages as representing more closely contested
elections. Thus, we expect the coefficient for this vari-
able to be negative. The data are taken from the National
Registrys voting records.
We include numerous individual-level indepen-
dent variables to test our hypotheses regarding op-
timism/skepticism versus cynicism about government
(H4), political participation (H5), and political knowl-
edge (H6) on perceptions of accountability. We hypothe-
size that those individuals who are generally optimistic (or
skeptical) about governmentas opposed to cynical
perceive government as more accountable. To measure
optimism versus cynicism toward government, we con-
struct a dichotomous variable where 1 represents a re-
spondent who is optimistic or skeptical about govern-
ment and 0 indicates she is cynical. We construct this
variable by combining individual responses to three cor-
ruption evaluationsof politicians in general, of depart-
mental officials, and of municipal officials. Responses to
each question ranged from not widespread (1) to very
widespread (4). If a respondent evaluated government
at all three levels as having widespread (3) or very
widespread (4) corruption, we coded them 0, meaning
he or she is cynical about government; 1 indicates that
a respondent is optimistic or healthily skeptical about
government.16
14
Faguet (2009) finds evidence that in the context of decentral-
ization, greater electoral competition in Bolivia induced higher
responsiveness to the electorate. Because we treat electoral compe-
tition as equivalent to political decentralization, we expect similar
relationships and expect more electoral competition to be linked
to higher levels of accountability.
15
Prior to 2003, governors served a three-year term. Beginning in
2003, they were elected for a four-year term (20032007) but during
both periods were prohibited from being immediately reelected,
although they could run for nonconsecutive terms.
16
This variable was chosen over corruption evaluations (at the mu-
nicipal, departmental, and national levels as well as an average of all
three) because it is empirically and theoretically distinct from our
dependent variable while measures of corruption were not. Given
that some may equate noncorrupt and accountable governance,
MARIA ESCOBAR-LEMMON AND ASHLEY D. ROSS
We expect political participation and political knowl-
edge to increase evaluations of departmental accountabil-
ity. To measure political participation, we create an ad-
ditive index of the following components: attendance at
a municipal meeting, participation in a meeting to plan
the budget or annual municipal plan, and solicitation
of help from some municipal office, official, or council-
man. The index ranges from 0no participationto 3
participation in all three activities. We use the municipal-
level participation measure because similar questions at
the department level were not asked in the survey, and we
have theoretical reasons to believe that participation on
the local level engenders perceptions that government on
the subnational level is more accountable.17 To measure
political knowledge, we create an additive index of two
political knowledge questions specific to Colombiathe
number of departments in Colombia and the term limit
of the Colombian president. Individuals were assigned a
value representing the number of correct answers they
gave (i.e., one of two correctly answered = 1).
We also include a series of individual-level control
variables which we anticipate may affect perceptions of
accountability. These control variables include age (mea-
sured in years), education (measured in years), and in-
come (measured in ranges on a 10-point scale, with higher
values indicating greater monthly family income). Finally,
because we pooled data from two different surveys, con-
ducted in 2004 and 2006, we include a control variable
for the year 2004.
We estimate a multilevel model because we expect
individual perceptions of accountability to be influenced
by departmental decentralization and our data are struc-
tured with individual-level observations nested in depart-
mental groups. Considering the nature of our depen-
dent variable, we estimate a random-intercept18 model
reverse causality analyses indicated that corruption predicted ac-
countability and vice versa. The same is not true for this measure
of optimism.
17
Our measure of participation assumes that positive affect toward
government generated by participating in local government will
spill up to the department level, which is consistent with Putnams
(1993) argument suggesting that social capital can spill up to
trust in institutions. While ideally we would be able to measure
participation at the departmental level, we believe this measure is
a reasonable proxy.
18
We estimate a random-intercept model because we hypothesize
that decentralization shifts the baseline assessment of government
accountability, with more administratively, fiscally, or politically
decentralized departments having higher baseline accountability.
We do not estimate a varying slopes model because we have no
strong theoretical expectations that any of the three forms of de-
centralization will change the way in which participation, optimism
versus cynicism, and information affect perceptions of accountabil-
ity.
DOES DECENTRALIZATION IMPROVE ACCOUNTABILITY?
(restricted maximum likelihood) with the normal link
function. Due to limitations in Level 1 data taken
from LAPOP, the number of Level 2 unitsColombian
departmentsis reduced from 32 to 24.19 Given the small
number of Level 2 units, we take a conservative approach
and estimate the multilevel model separately for each
Level 2 variableadministrative, fiscal, and political de-
centralization.20
Results
The model results are reported in Table 1.21 The three
models estimated are reported in separate columns, with
administrative decentralization as the Level 2 variable in
the first column, fiscal decentralization in the second,
and political decentralization in the third. Most of our
theoretical expectations are confirmed, although not all.
Most strikingly, we observe differences in the significance
of the types of decentralization in explaining variation
across departments in perceptions of accountability. Ad-
ministrative22 and fiscal autonomy are statistically signif-
icant and positively related to accountability perceptions
of departmental government, whereas political decentral-
ization is not. We first discuss the Level 2 results and then
consider the Level 1 results.
Level 2 Effects
We expected departmental decentralization to shape
individual-level accountability perceptions. As citizens
have more opportunities to be involved in their local
government and see greater transparency as a result of
19
The departments included in our regression analysis are Antio-
quia, Atlantico, Bolvar, Boyaca, Caldas, Caqueta, Casanare, Cauca,
Cesar, Cordoba, Cundinamarca, Huila, Magdalena, Meta, Narino,
Norte de Santander, Putumayo, Quindio, Risaralda, Santander,
Sucre, Tolima, Valle, and Vaupes.
20
The results of the same model, but with all three Level 2 variables
estimated together, are discussed in the online supporting infor-
mation and reported in Table SI.5. The results are consistent with
those reported in Table 1.
21
See the online supporting information, specifically Table SI.6,
for a discussion of regression results with the dependent variable
disaggregated into its three components. The Level 1 results do not
change considerably, but the Level 2 results fluctuate somewhat
from the results presented in Table 1.
22
Because this measure of administrative decentralization is an
average of five ratings, the model was reestimated to include the
measure each time excluding one of its five components. The results
of these robustness checks are consistent with those reported in
Table 1all of the alternative measures of administrative autonomy
were statistically significant and positively related to the dependent
variable. See the online supporting information for a discussion
and presentation of the results (Table SI.7).
183
decentralizationadministrative, fiscal, and political
they should perceive their departmental government as
more answerable or accountable. We find that adminis-
trative and fiscal autonomy affect accountability percep-
tions; however, the link between political decentralization
and accountability is not evident in our findings. 23 The
results of these department-level effects are shown in Ta-
ble 1 under Fixed Effects: Level 2.24
Administrative and fiscal decentralization are posi-
tive and statistically significant. Their coefficients indicate
that baseline accountability scores by department are im-
proved with greater levels of administrative and fiscal au-
tonomy. A one standard deviation increase in administra-
tive and fiscal autonomy is associated with a 0.07 and 0.08,
respectively, increase in departmental intercepts. Put an-
other way, if the least autonomous departmentVaupes
with an administrative autonomy rating of 2.07 and fiscal
autonomy rating of 17.63were to raise its autonomy to
the levels of the highest performersRisaralda with a 3.86
administrative autonomy rating and Casanare with fiscal
autonomy of 84.07then its intercept for the adminis-
trative decentralization model would increase by 0.49 and
its intercept for the fiscal decentralization model would
increase by 0.33. Because accountability is calculated as
the average of three ratings, these increases are equivalent
to a one-point improvement in the accountability rating
prior to averaging it. This corresponds, for example, to a
positive shift in thinking the departmental government is
accountable in one of the three components (consulting
citizens, making plans public, and sharing information)
included in the aggregate measure.
While we find that administrative and fiscal
autonomy increase baseline accountability for all depart-
ments, we also find that departmental intercepts vary
considerablyillustrated by Figure 2. The 24 depart-
ments analyzed are reported on the x-axis, sorted by their
mean accountability scores shown in parentheses. The
y-axis displays the departmental intercept (or baseline
level of accountability) for each of the three models
23
We tested alternative measures of political decentralization, in-
cluding minimum margin of victory in gubernatorial races for each
department (years 1997, 2000, and 2003) as well as the margin of
victoryaverage and minimumfor presidential and senate races
(years 1998, 2002, and 2006). We also examined political auton-
omy ratings from DANE (the same source for the administrative
autonomy measure) as an alternative measure. The regression re-
sults employing these alternative measures do not vary considerably
from those reported in Table 1.
24
The number of observations by department varied from a mini-
mum of 18 to a maximum of 208, with an average of 73. The 95%
confidence intervals for none of the coefficients overlap zero. For
the number of observations by department, see Table SI.4 of the
online supporting information.
184 MARIA ESCOBAR-LEMMON AND ASHLEY D. ROSS

TABLE 1 Predictors of Departmental Accountability

Dependent Variable: Accountability
Average rating of responses to the following. In your opinion, your departmental government:
Consults citizens before making decisions; Makes its plans and decisions public; and Shares information openly
and on time.
Ratings: 1 = never, 2 = almost never, 3 = sometimes, 4 = almost always, and 5 = always
Admin. Fiscal Political
Fixed Effects: Level 1 Optimistic/skeptical vs. cynical toward government 0.171 0.166 0.166
(0.066) (0.066) (0.066)

Subnational political participation 0.204 0.203 0.203
(0.031) (0.031) (0.032)
Political knowledge 0.030 0.029 0.029
(0.037) (0.037) (0.037)
Income 0.035 0.034 0.036
(0.019) (0.019) (0.019)
Age 0.005 0.005 0.005
(0.002) (0.002) (0.002)
Education 0.003 0.003 0.003
(0.007) (0.007) (0.007)
Year 2004 0.235 0.229 0.231
(0.076) (0.075) (0.075)

Intercept 1.075 1.806 1.969
(0.444) (0.143) (0.133)
Fixed Effects: Level 2 Administrative Decentralization: DANE admin. 0.276
autonomy rating (0.125)
Fiscal Decentralization: Percent of own-source revenue 0.005
(0.002)
Political Decentralization: Margin of victory in 0.002
governors races (0.005)
Random Effects Level 1 Variance (standard deviation) 1.009 1.009 1.008
(1.005) (1.005) (1.004)
Level 2 Variance (standard deviation) 0.022 0.022 0.028
(0.150) (0.148) (0.168)
Number of Level 1 1748 1748 1748
Observations Level 2 24 24 24
Note: Random intercept model (restricted maximum likelihood) estimated with normal link function. Model estimated for each Level
2 variable (administrative, fiscal, and political decentralization) separately. Coefficients are reported with robust standard errors in
parentheses. For number of observations by department, see the online supporting information, Table SI.4.
Level 1 variables with are centered around their grand means. Statistical significance denoted as p < 0.10, p < 0.05, p < 0.01.

reported in Table 1. The three types of decentralization
are marked with a distinct symbol. Figure 2 shows that
in general, the largest intercepts are those associated with
the fiscal and administrative decentralization models.
In departments with higher-than-average accountability
ratings (greater than 2.266), fiscal and administrative
decentralization model intercepts tend to be larger. This
highlights the positive influence of fiscal and adminis-
trative autonomy on perceptions of accountability. It
also underscores that how decentralization has been
translated into policy by individual departments within
their own unique contexts has a significant bearing on
how accountable citizens perceive government to be.
Individual-Level Results
As discussed in the previous section, administrative and
fiscal decentralization significantly influence individual-
level accountability perceptions as they shift baseline de-
partmental intercepts. On the individual level, we find
that optimism toward government, subnational political
DOES DECENTRALIZATION IMPROVE ACCOUNTABILITY? 185

FIGURE 2 Departmental Intercepts by Decentralization Type

2.25
2.2
2.15
Accountability
2.1
2.05
2
1.95
1.9
1.85
1.8
1.75 Fiscal decen.
1.7
Administrative decen.
Casanare (2.50)
Boyac (2.49)

Crdoba (1.86)
Cundinamarca (2.17)

Bolvar (2.01)
Vaups (2.35)

Quindo (1.68)
Cesar (2.39)

Cauca (2.21)
Nario (2.21)
Meta (2.40)
Caldas (2.44)

Tolima (2.25)

Magdalena (2.20)

Sucre (1.86)
Huila (2.17)
Antioquia (2.40)

Risaralda (2.34)

Santander (2.29)
Putumayo (2.30)

Atlntico (2.13)
Valle (2.44)

Caquet (1.83)
N. de Santander (2.20)
Political decen.

Level 2 Units: Colombian departments

with mean of accountability in parentheses

Note: Intercepts plotted based on the models reported in Table 1. Decentralization is measured as (1)
AdministrativeDANE administrative autonomy ratings for the year 2007 based on survey responses from depart-
mental functionaries; (2) Politicalthe average of the margin of victory of gubernatorial elections for the years 1997,
2000, and 2003; and (3) Fiscalthe percent of own-source revenue to total departmental revenue, averaged over
the years 20002006. The number of departments is reduced to 24 due to limitations in Level 1 data (LAPOP). For
number of observations by department included in the regression analyses, see the online supporting information,
Table SI.4.

participation, income, and age explain perceptions of de-
partmental accountability. These individual-level results
are reported in Table 1 under the heading Fixed Effects:
Level 1.
General optimism about government and subna-
tional political participation are positive and significant
in each model, indicating that optimistic (or skeptical)
and participatory individuals evaluate departmental gov-
ernment as more accountable than cynics or those who
do not participate. A shift from cynicism to optimism
is associated with a 0.17 increase in accountability rat-
ings across all three models. A one standard deviation
increase in participation is associated with a 0.14 increase
in accountability ratings. No subnational political partic-
ipation is correlated with a 0.08 decrease in accountability
ratings, whereas participation in all of the activities mea-
sured boosts accountability by 0.52 points. While these ef-
fects seem small, taken together they have the potential to
considerably shift accountability perceptions. For exam-
ple, an individual who is optimistic and participates fully
in subnational politics has a 0.70-point higher account-
ability rating than an individual who is cynical and non-
participatory. This is roughly equivalent to a two-point
increase in the accountability rating prior to averaging it,
which means, for example, this individuals responses in
two of the three components measured (consulting citi-
zens, making plans public, and sharing information) are
scored one point higher.
We also expected the amount of information citizens
have about government to influence their perceptions
of government accountability. But political knowledge
does not explain perceptions of accountability; however,
income and age do. Individuals with higher monthly in-
comes have more positive perceptions of departmental ac-
countability; a one standard deviation increase in income
is associated with a 0.06 increase in accountability rat-
ings across all three models. In addition to greater wealth,
younger individuals have higher accountability evalua-
tions; a one standard deviation increase in age decreases
accountability scores by 0.07 across all three models. We
also find that when the survey was conducted is statisti-
cally significant, with respondents in 2004 having higher
accountability ratings by 0.235 in the administrative de-
centralization model, 0.229 in the fiscal decentralization
model, and 0.231 in the political decentralization model.
In all, the findings of the regression analyses indicate
that individual-level perceptions of departmental govern-
ment accountability are shaped by contextthe extent
of administrative and fiscal autonomyand individual-
level beliefs and characteristicsoptimism about
186
government, political participation, income, and age. De-
partments that have successfully achieved administrative
and fiscal autonomy have higher baseline assessments of
departmental government accountability, and individuals
within these departments who are optimistic toward gov-
ernment, participate in subnational politics, have higher
incomes, and are younger are more likely to have higher
accountability ratings. While the impacts of these ex-
planatory factors, indicated by their respective coeffi-
cients, tend to be small, together they can substantively
shift individual perceptions of departmental government
accountability.
Conclusions
Decentralization was adopted widely across the world
amid much fanfare that it would increase accountability,
improve government, and bring representation closer to
citizens. Given the amount of promise the policy held, it is
important to evaluate the effectiveness of these reforms.
Yet, few attempts have been made to identify the way
in which decentralization fulfilled its grander promise in
terms of improved accountability. Our research repre-
sents an attempt to determine if and how decentraliza-
tion changed whether citizens saw their government as
accountable in terms other than their ability to sanction
through elections.
We find support for our expectation that the decen-
tralization context determines if individuals see their gov-
ernment as more answerable, but only in its fiscal and ad-
ministrative forms. Despite the theoretical assumptions
that politicians in politically decentralized (e.g., highly
competitive) environments will be more honest and re-
sponsive to citizens and this, in turn, will affect how indi-
viduals perceive government, we do not find evidence of
this. Our analysis also shows that perceptions result from
what an individual has experienced personally. Optimism
or healthy skepticism about government and participa-
tion both improve perceptions of accountability.
This article makes two important contributions.
First, we offer an explicit test of whether the three types of
decentralization alter citizens perceptions of subnational
government accountability. In doing so, we provide em-
pirical evidence that one of the most popular, and yet
frequently untested, assumptions about the benefits of
decentralization is partly, but not totally, correct. Com-
petence (and perhaps good governance), which is clearly
related to administrative decentralization and fiscal em-
powerment, matters more for improving perceptions of
accountability than does political decentralization. Of
MARIA ESCOBAR-LEMMON AND ASHLEY D. ROSS
course, the absence of a link between political decen-
tralization and perceived government accountability is
something worthy of future study. Political decentraliza-
tion may enhance accountability where citizens couple
opportunities for sanctioning with information to deter-
mine if sanctioning is required. Therefore, political de-
centralization might be found to be successful if account-
ability is examined as both sanctioning and answerability.
Second, by measuring accountability as answerabil-
ity, we move beyond electoral institutions to provide a
more nuanced view of accountability. Elections are im-
portant for calling governments to account for their ac-
tions and remain a critical requirement for accountabil-
ity; however, government that is accountable in the eyes of
citizens may require more than simply periodic opportu-
nities to toss the bums out. Decentralization enhances
subnational governments ability to engage citizens be-
tween elections. The existence of these opportunities and
citizens awareness that they existregardless of whether
they personally take advantage of themhelp explain
how decentralization improves outcomes.
While our results suggest that administrative and fis-
cal decentralization can have a significant positive effect
on accountability, they also suggest that this might not
be automatic. As the Colombian evidence indicates, even
when decentralization offers the same opportunities to
all units (or imposes the same restrictions upon them),
what individual departments do with those opportuni-
ties matters. Some departments clearly sought to develop
significant fiscal autonomy, some ambitiously sought to
reinvent the way their government functioned by im-
proving capacity and trying to respond to citizen needs,
and certainly others sought to do as little as possible, ex-
cept perhaps steal from public coffers. That said, those
who used the opportunity afforded by decentralization
to increase fiscal and administrative autonomy were re-
warded with more favorable citizen evaluations of their
performance. A national policy of decentralization is not
a panacea for improving accountability; however, in the
right hands it can be a useful tool.
References
Alesina, Alberto, Alberto Carrasquilla, and Juan Jose Echavarra.
2005. Decentralization in Colombia. In Institutional Re-
forms: The Case of Colombia, ed. Alberto Alesina. Cambridge,
MA: MIT Press, 175207.
Angell, Alan, Pamela Lowden, and Rosemary Thorp. 2001. De-
centralizing Development: The Political Economy of Institu-
tional Change in Colombia and Chile. Oxford: Oxford Uni-
versity Press.
DOES DECENTRALIZATION IMPROVE ACCOUNTABILITY?
Bardhan, Pradeep, and Dilip Mookherjee. 2000. Capture and
Governance at Local and National Levels. American Eco-
nomic Review 90(2): 13539.
Bardan, Pranab, and Dilip Mookherjee. 2006. The Rise of Local
Governments: An Overview. In Decentralization and Local
Governance in Developing Countries: A Comparative Perspec-
tive, ed. Pranab Bardhan and Dilip Mookherjee. Washington,
DC: Inter-American Development Bank, 152.
Blair, Harry. 2000. Participation and Accountability at the
Periphery: Democratic Local Governance in Six Countries.
World Development 28(1): 2139.
Burki, Shahid Javed, Guillermo E. Perry, and William R.
Dillinger. 1999. Beyond the Center: Decentralizing the State.
Washington, DC: World Bank.
Canes-Wrone, Brandice. 2006. Who Leads Whom? Presidents,
Policy and the Public. Chicago: Chicago University Press.
Cheibub, Jose Antonio, and Adam Przeworski. 1999. Democ-
racy, Elections, and Accountability for Economic Out-
comes. In Democracy, Accountability, and Representation,
ed. Adam Przeworski, Susan C. Stokes, and Bernard Manin.
Cambridge: Cambridge University Press, 22249.
Cleary, Matthew, and Susan Stokes. 2006. Democracy and the
Culture of Skepticism. New York: Russell Sage.
Crook, Richard C., and Alan Sturla Sverrisson. 2001.
Decentralisation and Poverty-Alleviation in Developing
Countries: A Comparative Analysis or, Is West Ben-
gal Unique? IDS Working Paper no. 130, http://www
.washdoc.info/docsearch/title/122620.
Dauda, Carol L. 2006. Democracy and Decentralisation: Lo-
cal Politics, Marginalisation, and Political Accountability in
Uganda and South Africa. Public Administration and De-
velopment 26(4): 291302.
Daughters, Robert, and Leslie Harper. 2007. Fiscal and Political
Decentralization Reforms. In The State of State Reform in
Latin America, ed. Eduardo Lora. Washington, DC: Inter-
American Development Bank, 21361.
Departamento Administrativo Nacional de Estadstica (DANE).
2007. Encuesta sobre ambiente y desempeno in-
stitucional departamental. http://www.dane.gov.co/files/
EDI/INFORME RG EDI DPTAL 2007.pdf.
Eaton, Kent, and Larry Schroeder. 2010. Measuring Decen-
tralization. In Making Decentralization Work: Democracy,
Development, and Security, ed. Ed Connerley, Kent Eaton,
and Paul Smoke. Boulder, CO: Lynne Rienner, 16790.
Faguet, Jean-Paul. 2004. Does Decentralization Increase Gov-
ernment Responsiveness to Local Needs? Evidence from Bo-
livia. Journal of Public Economics 88(3-4): 86793.
Faguet, Jean-Paul. 2009. Governance from Below in Bolivia:
A Theory of Local Government with Two Empirical Tests.
Latin American Politics and Society 51(4): 2968.
Falleti, Tulia. 2005. A Sequential Theory of Decentralization:
Latin American Cases in Comparative Perspective. Ameri-
can Political Science Review 99(3): 32746.
Falleti, Tulia. 2010. Decentralization and Subnational Politics in
Latin America. Cambridge: Cambridge University Press.
Fearon, James D. 1999. Electoral Accountability and the Con-
trol of Politicians: Selecting Good Types versus Sanction-
ing Poor Performance. In Democracy, Accountability, and
Representation, ed. Adam Przeworski, Susan C. Stokes, and
187
Bernard Manin. Cambridge: Cambridge University Press,
5597.
Ferejohn, John. 1986. Incumbent Performance and Electoral
Control. Public Choice 50(1-3): 525.
Fisman, Raymond, and Roberta Gatti. 2002. Decentralization
and Corruption: Evidence across Countries. Journal of Pub-
lic Economics 83(3): 32545.
Fiszbein, Ariel. 1997. The Emergence of Local Capacity:
Lessons from Colombia. World Development 25(7): 1029
43.
Gelineau, Francois, and Karen L. Remmer. 2006. Political De-
centralization and Electoral Accountability: The Argentine
Experience, 19832001. British Journal of Political Science
36(1): 13357.
Hardin, Russell. 1999. Do We Want Trust in Government?
In Democracy and Trust, ed. Mark E. Warren. Cambridge:
Cambridge University Press, 2242.
Hobolt, Sara Binzer, and Robert Klemmensen. 2008. Govern-
ment Responsiveness and Political Competition in Com-
parative Perspective. Comparative Political Studies 41(3):
30937.
Jones, G. W. 1992. The Search for Local Accountability. In
Strengthening Local Government in the 1990s, ed. Steve Leach.
London: Longman, 4978.
Keefer, Philip E., Ambar Narayan, and Tara Vishwanath. 2006.
Decentralization in Pakistan: Are Local Governments Likely
to Be More Accountable Than Central Governments?
In Decentralization and Local Governance in Developing
Countries: A Comparative Perspective, ed. Pranab Bardhan
and Dilip Mookherjee. Cambridge, MA: MIT Press, 285
303.
Khemani, Stuti. 2001. Decentralization and Accountability:
Are Voters More Vigilant in Local Than in National Elec-
tions? World Bank Policy Research Working Paper 2557: 1
23.
Kitschelt, Herbert, Kent Freeze, Kiril Kolev, and Yi-Ting Wang.
2009. Measuring Democratic Accountability: An Initial Re-
port on an Emerging Data Set. Revista de Ciencia Poltica
29(3): 74173.
Lederman, Daniel, Norman V. Loayza, and Rodrigo R. Soares.
2005. Accountability and Corruption: Political Institutions
Matter. Economics and Politics 17(1): 135.
Linz, Juan, and Alfred Stepan. 1996. Toward Consolidated
Democracies. Journal of Democracy 7(2): 1433.
Magaloni, Beatriz, Alberto Daz-Cayeros, and Federico Estevez.
2007. Clientelism and Portfolio Diversification: A Model
of Electoral Investment with Application to Mexico. In
Patrons, Clients, and Policies: Patterns of Democratic Ac-
countability and Political Competition, ed. Herbert Kitschelt
and Steven I. Wilkinson. Cambridge: Cambridge University
Press, 182205.
Manin, Bernard, Adam Przeworski, and Susan C. Stokes. 1999.
Introduction. In Democracy, Accountability, and Represen-
tation, ed. Adam Przeworski, Susan C. Stokes, and Bernard
Manin. Cambridge: Cambridge University Press, 126.
Manor, James. 1999. The Political Economy of Democratic De-
centralization. Washington, DC: World Bank.
Mulgan, Richard. 2000. Accountability: An Ever-Expanding
Concept? Public Administration 78(3): 55573.
188
ODonnell, Guillermo. 1999. Horizontal Accountability in
New Democracies. In The Self-Restraining State: Power and
Accountability in New Democracies, ed. Andreas Schedler,
Larry Diamond, and Marc F. Plattner. Boulder, CO: Lynne
Rienner, 2951.
Peterson, George E. 1997. Decentralization in Latin America:
Learning through Experience. Washington, DC: World Bank.
Prudhomme, Remy. 1995. The Dangers of Decentralization.
The World Bank Research Observer 10(August): 20120.
Prudhomme, Remy, and Anwar Shah. 2002. Centraliza-
tion v. Decentralization: The Devil Is in the Details.
Presented at Fiscal Federalism in the MERCOSUR: The
Challenges of Regional Integration Conference, Porto Ale-
gre, Brazil. Available at http://www.forumfed.org/libdocs/
FedMerco02/116-FEME0206-int-PrudhommeShah.pdf.
Putnam, Robert. 1993. Making Democracy Work: Civic Tradi-
tions in Modern Italy. Princeton, NJ: Princeton University
Press.
Remmer, Karen L., and Francois Gelineau. 2003. Subnational
Electoral Choice: Economic and Referendum Voting in
Argentina, 19831999. Comparative Political Studies 36(7):
80121.
Rich, Jessica A. J., and Eduardo J. Gomez. 2012. Centralizing
Decentralized Governance in Brazil. Publius: The Journal of
Federalism 42(4): 63661.
Rondinelli, Dennis A. 1989. Decentralizing Public Services in
Developing Countries: Issues and Opportunities. Journal of
Social, Political, and Economic Studies 14(1): 7798.
Rose-Ackerman, Susan. 1999. Corruption and Government:
Causes, Consequences, and Reform. Cambridge: Cambridge
University Press.
Schedler, Andreas. 1999. Conceptualizing Accountability. In
The Self-Restraining State: Power and Accountability in New
Democracies, ed. Andreas Schedler, Larry Diamond, and
Marc F. Plattner. Boulder, CO: Lynne Rienner, 1328.
Schneider, Aaron. 2003. Decentralization: Conceptualization
and Measurement. Studies in Comparative International De-
velopment 38(3): 3256.
Shah, Anwar. 2006. Corruption and Decentralized Public Gov-
ernance. World Bank Policy Research Working Paper 3824:
128.
Shah, Anwar, Theresa Thompson, and Heng-Fu Zou. 2004.
The Impact of Decentralization on Service Delivery, Fiscal
Management and Growth in Developing and Emerging Mar-
ket Economies: A Synthesis of Empirical Evidence. CESIFO
DICE Report 2(Spring): 1014.
Smith, Graham. 2009. Democratic Innovations: Designing In-
stitutions for Citizen Participation. Cambridge: Cambridge
University Press.
MARIA ESCOBAR-LEMMON AND ASHLEY D. ROSS
Smulovitz, Catalina, and Enrique Peruzzotti. 2000. Societal Ac-
countability in Latin America. Journal of Democracy 11(4):
14758.
Treisman, Daniel. 2000. The Causes of Corruption: A
Cross-National Study. Journal of Public Economics 76(3):
399457.
Treisman, Daniel. 2007. The Architecture of Government: Re-
thinking Political Decentralization. Cambridge: Cambridge
University Press.
Weitz-Shapiro, Rebecca. 2008. The Local Connection: Local
Government Performance and Satisfaction with Democ-
racy in Argentina. Comparative Political Studies 41(3):
285308.
Weitz-Shapiro, Rebecca. 2012. What Wins Votes: Why Some
Politicians Opt Out of Clientelism. American Journal of
Political Science 56(3): 56883.
Supporting Information
Additional Supporting Information may be found in the
online version of this article at the publishers website:
Measurement of accountability;
Measurement of decentralization;
Measurement of individual factors related to account-
ability; and
Robustness analyses.
The Supporting Information also includes the following
tables:
Table SI.1: Descriptive Statistics
Table SI.2: Decentralization Measures Across the 32
Colombian Departments
Table SI.3: Administrative Autonomy Disaggregated
into Five Components by Department
Table SI.4: Number of Observations by Department
Included in Regression Analysis
Table SI.5: Predictors of Departmental Accountability
Estimated with all Level 2 Variables in a Single Model
Table SI.6: Regression Results with Accountability Dis-
aggregated into Three Components
Table SI.7: The Effect of the Five Components of Ad-
ministrative Decentralization on Accountability