You are on page 1of 31

Money, Ination and Monetary Policy in the Euro Area

By S.:\i A\o.i-Do\i .: Ji.:-Giii.:i S.ic

Both an empirical breakdown of the quantity theory of money and


a shift in the volatility of nominal variables occurred in the euro
area over the period 19802007. A dynamic stochastic general
equilibrium model with money is proposed and estimated over dif-
ferent subsamples to assess the drivers of these empirical facts.
Our estimation results and counterfactual exercises provide evi-
dence that a change in the degree of responsiveness of monetary
authorities and the transition from a money growth to an inter-
est rate rule explain much of the observed properties of nominal
variables and their relationships.
Keywords: Quantity theory of money, monetary policy, DSGE
model, Bayesian methods.
JEL: E31, E51, E52.
The empirical breakdown of the quantity theory of money and the radical shift
in the volatility of most macroeconomic data are two prominent stylised facts char-
acterizing the euro area over the period 19802007. Recent literature has focused
on such developments for several OECD countries. For instance, Sargent and Surico
(2011) and Teles and Uhlig (2010) explain the large departures from a unitary money
growth/ination relation by the dependence of the coecient estimate associated
with the regression of the two variables to the policy rule parameters. In more recent
times, the lowination countries should then see the relationship between ination
and the growth rate of money becoming tenuous at best. Other studies, such as
Andres et al. (2006) or Justiniano and Primiceri (2008), suggest that the decline in
macroeconomic volatility over the past thirty years is largely the result of smaller
shocks impinging on the economy (usually called the good luck hypothesis), with
structural changes having played at most a secondary role. There is therefore dis-
agreement about the origin of changes both in the statistical properties of nominal
variables and in their relationships.
In this paper, we argue that a change in the degree of responsiveness of the mon-
etary authorities and the transition from a money growth to an interest rate rule
explain much of the stylised facts. To reach this conclusion, we rst show the signif-
icant dierence in the way monetary policy was conducted pre and postEuropean
Monetary Union (EMU). We then go on to demonstrate that this dierence in policy
behavior allows an understanding of the shift in macroeconomic behavior.

Avouyi-Dovi: Banque de France, 31 rue Croix des Petits Champs, 75049 Paris, France (e-mail:
sanvi.avouyi-dovi@banque-france.fr); Sahuc: Banque de France, 31 rue Croix des Petits Champs, 75049 Paris,
France (e-mail: jean-guillaume.sahuc@banque-france.fr). We are grateful to Christophe Cahn, Patrick Fve,
Benoit Mojon, Christian Pster, Thomas Sargent and Richard Summer for comments. Special thanks to
Julien Matheron, whose suggestions greatly improved the paper. The views expressed herein are those of the
authors and do not reect those of the Banque de France.
1
The intuitions behind this result are as follows. The quantity theory implies that
any change in the monetary aggregate induces a variation in the same direction and
same magnitude of prices and of nominal interest rates. If price stability is the
objective of the monetary authorities, any price increase should result in a sharp
contraction in the money supply. As a consequence, both variances and covariances
between nominal variables decrease. But, implementing price stability or ination
targeting while the underlying theory establishes similar variations between prices
and the monetary aggregate necessarily induces a stronger fall of the covariances.
This mechanical eect blurs the two quantity propositions. In addition, with an
interest elastic demand for real money balances, any shock that aects the path for
expected ination or the real interest rate causes money demand to shift. When
the central bank follows an interest rate rule rather than a money growth rule, the
money stock is endogenous and ination is xed by the policy rule. Money absorbs
the adjustment and the central bank can accommodate this jump in the money stock
almost instantaneously and with little cost. Consequently, the money stock responds
by shifting to clear the money market. It allows to explain why, in an economy with
stochastic ination and an interest rate rule for monetary policy, the money growth
rate is much more variable than the ination rate.
To demonstrate the role of these mechanisms, we consider a structural monetary
model of the business cycle in which money is allowed to play a relevant role. The
model combines a neoclassical growth core with several shocks and frictions. It
includes features such as habit formation, money in the utility, investment adjust-
ment costs, variable capital utilization, monopolistic competition in goods and labour
markets, and nominal price and wage rigidities. We follow the Bayesian approach to
estimate several versions of the model. Unlike all previous papers that use an ad-
hoc calvo-type employment adjustment equation to translate hours worked into the
observed employment series, we directly use a new series of quarterly hours worked
for the euro area. We consider two subsamples: 1980Q11998Q4 in imposing a
money growth rule and 1999Q12007Q4 in imposing an interest rate rule. Indeed,
during the 1980s and 1990s, many European countries have employed either ocial
money growth targets (through a broadly dened monetary aggregate) or exchange
rate policies in order to tie to the Deutschmark within the Exchange Rate Mecha-
nism (Bernanke and Mishkin, 1992). In Germany for instance, the setting of targets
explicitly takes into account the Bundesbanks long-term ination goal, estimated
potential output growth and expected velocity trends, which are combined using the
quantitytheory equation to determine the desired money growth rate. From 1999,
the European Central Bank (ECB) steered shortterm money market rates in order
to inuence the spending decisions of the private sector, monetary and nancial de-
velopments and, ultimately, prices. This change in the monetary policy instrument
follows the idea that money and hence credit does not have any crucial and construc-
tive roles to play in monetary policy design (Woodford, 2008).
1
The money market is
1
The designation monetary policy instrument is a shortcut in macroeconomic modeling to represent in
fact intermediate targets which are variables that are neither under the direct day-to-day control of the
central bank nor are the ultimate goals of policy, but that are used to guide policy. Values for instruments are
usually set so that, given estimates of behavioral parameters such as the interest elasticity of money demand,
intermediate targets for variables are reached in the longer term (quarter-to-quarter or year-to-year).
2
then only useful for determining the supply of money which responds endogenously
to the demand of money. This consensus substituted the one put forward by Milton
Friedman that ination is always and everywhere a monetary phenomenon.
The article is structured as follows. Section I presents the empirical evidence
that motivates the paper. Section II describes the structural model and Section III
describes the estimation procedure and reports the estimation results. Section IV
analyses the drivers of the two stylised facts. A last section concludes.
I. Stylised Facts
A. The empirical breakdown of the quantity theory of money
The quantity theory of money can be expressed in terms of lowfrequency comove-
ments between money growth and ination as well as between money growth and
nominal interest rate. The lowfrequency approach, that does not require a host of
encumbering theoretical or econometric assumptions, is achieved by using a lter
that extracts a longrun signal from time series data. Lucas (1980) has suggested
the following lter (revisited by Whiteman, 1984, and Sargent and Surico, 2011):
x
t
() =
n

k=n

|k|
x
t+k
,
where x
t
is the variable of interest, is a parameter comprised between 0 and 1, and
= (1 )
2
/(1 2
n+1
(1 )) is selected such that the sum of the weights is
equal to 1. As approaches zero, no ltering occurs, while as approaches unity,
the ltered series x
t
() approaches the sample mean of the original series. Following
Lucas, we set = 0.95.
2
Given the size of our sample and the fact that long averaging
does not appear in practice to deliver any greater improvement in t (McCallum and
Nelson, 2010), we set n = 4.
Unit slopes of graphs of long twosided moving averages are used to characterise
the implications of the quantity theory of money. According to this theory, a plot of
ination or nominal interest rates against money growth should produce data points
that lie along a 45degree line.
Figure 1 shows scatter plots of ltered M2 growth, ination and nominal short
term interest rate (see Section III.A for details on the data used in our analysis).
We split the sample in 1999 when the nal stage of the European Economic and
Monetary Union (EMU) was successfully launched. This date ocially translated
regime change in monetary policy in the euro area. During the 1980s and 1990s,
many European countries have either a monetary target (through a broadly dened
monetary aggregate) or exchange rate policies in order to tie to the deutschmark
through the Exchange Rate Mechanism. From 1999, the conversion rates of the
currencies of the 11 members states initially participating in monetary union were
irrevocably xed and the ECB took over responsability for conducting the single
monetary policy in the euro area. The ECB steers shortterm interest rates by
2
Sargent and Surico (2011) have tested other values of without aecting the variability of the elasticity
of money relative to price.
3
signalling its monetary policy stance and by managing the liquidity situation in the
money market.
y = 1.34x - 3.02
0
2
4
6
8
0 2 4 6 8
Money growth
1980-1998
y = 1.38x + 0.15
0
2
4
6
8
10
12
0 2 4 6 8 10 12
Money growth
1980-1998
y = 0.03x + 1.33
0
2
4
6
8
0 2 4 6 8
Money growth
1999-2007
y = -0.12x + 3.07
0
2
4
6
8
0 2 4 6 8
Money growth
1999-2007
Figure 1. Scatter plots of ltered money growth, ination and interest rate
The graphs reveal that the quantity theory of money can be veried or not ac-
cording to the reporting period. The scatters of points corresponding to the period
19801998 show a concentration of points around the diagonal (45degree line), val-
idating the quantity theory of money over that period. This result is conrmed by
calculating the slope of the regression associated with these points: 1.34 for the in-
ation on money growth regression and 1.38 for the interest rate on money growth
regression. This illustrates the fact that periods of sustained high ination are always
accompanied by high growth rates of money, reinforcing the dictum that ination is
always and everywhere a monetary phenomenon. It was in this context that the
architects of the ECB assigned a prominent role to money. In doing so, they wanted
to acquire the credibility of the Bundesbank prior to the adoption of the euro.
But since 1999, the two scatters of points are at, resulting in slopes of the regres-
sion lines close to zero. The quantity theory of money seems to have disappeared
over the recent period. There is a marked deterioration in the ltered series once
the postEMU period is considered. While the ination rate remained practically
constant at the 2% level, the rate of money growth took on an upward trend. This
4
pattern created a divergent gap between the long term component of money growth
and ination after the introduction of the euro.
B. Changes in volatilities
In the period between the mid1990s and 2007, European economies enjoyed one
of the greatest economic growth periods, known as the Great Moderation due to
the low macroeconomic volatility in those years. The high growth rates of economic
variables with low volatility came simultaneously with ination under control and
low interest rates across the board of nancial assets, with practically inexistent risk
premia in many cases as a result of the underassessment of risk.
Table 1 summarises the evidence on volatility changes in ination, interest rate
and money growth by showing their respective standard deviations for the preEMU
and postEMU periods as well as the ratio between the two. We observe two striking
features. First, the volatility of ination and the shortterm interest rate has declined
sharply, by a factor of 3. This resulted from the high credibility achieved by the
ECB in maintaining a low and stable ination rate, in line with its denition of price
stability. Second, the volatility of money growth increased from 0.52 to 0.60.
3
Table 1Standard deviation of quarterly ination, interest rate and money growth (19802007)
PreEMU PostEMU
PostEMU
PreEMU
Ination 0.718 0.189 0.263
Interest rate 0.769 0.236 0.307
Money growth 0.515 0.601 1.167
There are two main reasons underlying this development. From a theoretical point
of view, when the interest rate becomes the monetary instrument, money becomes
endogenous and clears the money market. This property makes it automatically more
volatile to meet money demand. From a cyclical point of view, the emergence of new
nancial players and an array of innovative nancial instruments make the traditional
money supply gures harder to interpret. That increase in the volatility of money
growth is our second piece of evidence pointing to the presence of changes beyond
those that would result from a scaling down of volatility in all nominal variables.
II. A mediumscale model for the euro area
The present section describes our microfounded model of the Euro area economy,
which is close to Christiano, Eichenbaum, and Evans (2005) and Smets and Wouters
(2007). The model combines a neoclassical growth core with several shocks and fric-
tions. It includes features such as habit formation, investment adjustment costs,
variable capital utilization, monopolistic competition in goods and labour markets,
3
This increase in volatility is valid whatever the choice of the monetary aggregate (see Appendix A1).
5
and nominal price and wage rigidities. The economy is populated by ve classes of
agents: producers of a nal good, intermediate goods producers, households, em-
ployment agencies and the government. We adopt the specication investigated by
Justiniano, Primiceri and Tambalotti (2010), except that (i) we allow for money in
the utility function and (ii) we introduce two types of monetary policy rules.
A. Household sector
L:iio.:i:1 .oi:cii
Each household indexed by j [0, 1] is a monopolistic supplier of specialised labor
N
j,t
. At every point in time t, a large number of competitive employment agencies
combine households labor into a homogenous labor input N
t
sold to intermediate
rms, according to
(1) N
t
=
__
1
0
N
j,t
1

w,t
dj
_
w,t
,
Prot maximization by the perfectly competitive employment agencies implies the
labor demand function
(2) N
j,t
=
_
W
j,t
W
t
_


w,t

w,t
1
N
t
,
where W
j,t
is the wage paid by the employment agencies to the household supplying
labor variety j, while
(3) W
t

__
1
0
W
j,t
1

w,t1
dj
_
w,t1
is the wage paid by intermediate rms for the homogenous labor input sold to them
by the agencies.

w,t
measures the substitutability across labor varieties and its steadystate is the
desired steadystate wage markup over the marginal rate of substitution between
consumption and leisure. It is assumed to follow an ARMA(1,1) process in order to
capture the moving average, high frequency component of wages,
log (
w,t
) = (1
w
) log (
w
)+
w
log (
w,t1
)+
w,t

w,t1
,
w,t
i.i.d.N
_
0,
2
w
_
Hoiioi iniiini:ci
The preferences of the jth household are given by
(4) E
t

s=0

b,t+s
_
log (C
t+s
hC
t+s1
) +

m,t+s

1
z,t+s
1
_
M
t+s
P
t+s
_
1

N
1+
j,t+s
1 +
_
,
6
where E
t
denotes the mathematical expectation operator conditional upon informa-
tion available at t.
4
C
t
denotes consumption, M
t
/P
t
represents real balances, N
j,t
is
labor of type j. The parameter is the subjective discount factor, h [0, 1] denotes
the degree of habit formation, is related to the interest rate elasticity of the money
demand, and > 0 is the inverse of the Frisch elasticity.
b,t
and
m,t
are a dis-
turbance of the discount factor and a velocity shock respectively, evolving according
to
log (
b,t
) =
b
log (
b,t1
) +
b,t
,
b,t
i.i.d.N
_
0,
2
b
_
,
and
log (
m,t
) =
m
log (
m,t1
) +
m,t
,
m,t
i.i.d.N
_
0,
2
m
_
.
As we explain below, households are subject to idiosyncratic shocks about whether
they are able to reoptimise their wage. Hence, the above described problem makes
the choices of wealth accumulation contingent upon a particular history of wage
rate decisions, thus leading to households heterogeneity. For the sake of tractability,
we assume that the momentary utility function is separable across consumption,
real balances and leisure. Combining this with the assumption of a complete set of
contingent claims market, all the households will make the same choices regarding
consumption and money holding, and will only dier by their wage rate and supply
of labor. This is directly reected in our notations.
Household js period budget constraint is given by
P
t
(C
t
+I
t
) +T
t
+B
t
+M
t
R
t1
B
t1
+M
t1
+Q
j,t
+D
t
+W
j,t
N
j,t
+R
k
t
u
t

K
t1
P
t
(u
t
)

K
t1
, (5)
where I
t
is investment, T
t
denotes nominal lumpsum taxes (transfers if negative),
B
t
is the one-period riskless bond, R
t
is the nominal interest rate on bonds, Q
j,t
is
the net cash ow from households j portfolio of state contingent securities, D
t
is
the equity payout received from the ownership of rms, and R
k
t
is the rental rate of
capital. The capital utilization rate u
t
transforms physical capital

K
t
into the service
ow of eective capital K
t
according to
(6) K
t
= u
t

K
t1
,
and the eective capital is rented to intermediate rms at the nominal rental rate
r
k
t
. The costs of capital utilization per unit of capital is given by the convex function
(u
t
). We assume that u = 1, (1) = 0, and we dene

(1) /

(1)
1 +

(1) /

(1)
.
4
As tested by Ireland (2004) and Andrs et al. (2006), the assumption of non separable preferences in the
utility function is rejected.
7
Later, we estimate
u
rather than the elasticity

(1) /

(1) to avoid convergence


issues.
The physical capital accumulates according to
(7)

K
t
= (1 )

K
t1
+
i,t
_
1 S
_
I
t
I
t1
__
I
t
,
where is the depreciation rate of capital, and S (.) is an adjustment cost function
which satises S (
z
) = S

(
z
) = 0 and S

(
z
) =
k
> 0,
z
is the steadystate
growth rate of technology, and
i,t
is an investment shock, evolving according to
log (
i,t
) =
i
log (
i,t1
) +
i,t
,
i,t
i.i.d.N
_
0,
2
i
_
.
Households set nominal wages according to a staggering mechanism. In each period,
a fraction
w
of households cannot choose its wage optimally, but adjusts it to keep
up with the increase in the general wage level in the previous period according to the
indexation rule
(8) W
j,t
=
z

1
w

w
t1
W
j,t1
,
where
t
P
t
/P
t1
represents the gross ination rate, is steadystate (or trend)
ination and the coecient
w
[0, 1] is the degree of indexation to past wages.
The remaining fraction of households chooses instead an optimal wage, subject to
the labor demand function (2).
B. Business sector
Fi:.i ooo inocin
At every point in time t, a perfectly competitive sector produces a nal good Y
t
by combining a continuum of intermediate goods Y
t
(i), i [0, 1], according to the
technology
(9) Y
t
=
__
1
0
Y
i,t
1

p,t
di
_
p,t
,
Final good producing rms take their output price, P
t
, and their input prices, P
i,t
,
as given and beyond their control. Prot maximization implies the following Euler
equation
(10) Y
i,t
=
_
P
i,t
P
t
_


p,t

p,t1
Y
t
.
Integrating (10) and imposing (9), we obtain the following relationship between
8
the nal good and the prices of the intermediate goods
(11) P
t

__
1
0
P
i,t
1

p,t
1
di
_
p,t1
.

p,t
measures the substitutability across dierentiated intermediate goods and its
steady state is then the desired steadystate price markup over the marginal cost of
intermediate rms. It is assumed to follow an ARMA(1,1) process in order to capture
the moving average, high frequency component of ination
log (
p,t
) =
_
1
p
_
log (
p
) +
p
log (
p,t1
) +
p,t

p,t1
,
p,t
i.i.d.N
_
0,
2
p
_
.
I:1in:ii.1i-ooo iin:
Intermediate good i is produced by a monopolist rm using the following produc-
tion function
(12) Y
i,t
= K
i,t

[Z
t
N
i,t
]
1
Z
t
F,
where denotes the capital share, K
i,t
and N
i,t
denote the amounts of capital and
eective labor used by rm i, F is a xed cost of production that ensures that prots
are zero in steady state, and Z
t
is an exogenous laboraugmenting productivity factor
whose growthrate, denoted by
z,t
Z
t
/Z
t1
, evolves according to
log (
z,t
) =
z,t
,
z,t
i.i.d.N
_
0,
2
z
_
.
In addition, we assume that intermediate rms rent capital and labor in perfectly
competitive factor markets.
Intermediate rms set prices according to a staggering mechanism. In each period,
a fraction
p
of rms cannot choose its price optimally, but adjusts it to keep up
with the increase in the general price level in the previous period according to the
indexation rule
(13) P
i,t
=
1
p

p
t1
P
i,t1
,
where the coecient
p
[0, 1] indicates the degree of indexation to past prices. The
remaining fraction of rms chooses its price P

i,t
optimally, by maximizing the present
discounted value of future prots
(14) E
t

s=0
(
p
)
s

t+s

t
_

p
t,t+s
P

i,t
Y
i,t+s

_
W
t+s
N
i,t+s
R
k
t+s
K
i,t+s
__
,
where
(15)
p
t,t+s
=
_

s
=1

1
p

p
t+v1
s > 0
1 s = 0
,
9
subject to the demand from nal goods rms given by equation (10) and the produc-
tion function (12).
t+s
is the marginal utility of consumption for the representative
household that owns the rm.
C. Public sector
We assume that public spending G
t
is set according to
(16) G
t
=
_
1
1

g,t
_
Y
t
,
In order allow extensive feedback from endogenous variables to money growth,
monetary policy is set according to the following rule
(17)
t
=

t1
_
_

_
Y
t

z
Y
t1
_

y
_
(1

r,t
,
where
r,t
is a monetary policy shock, evolving according to
log (
r,t
) =
r,t
,
r,t
i.i.d.N
_
0,
2
r
_
.
We consider two types of monetary policy rules depending on the instrument the
central bank uses. The rst is a money growth rule according to which the central
bank adjusts smoothly the growth rate of money,
t
= M
t
/ (
z
M
t1
), in response
to movements in ination and output growth. The second is a Taylor rule according
to which the shortterm nominal interest rate,
t
= R
t
/R, is adjusted smoothly in
response to movements in ination and output growth.
D. Market clearing
Market clearing conditions on nal goods market are given by
Y
t
= C
t
+I
t
+G
t
+(u
t
)

K
t1
, (18)

p,t
Y
t
=
_
u
t

K
t1
_

[Z
t
N
t
]
1
Z
t
F, (19)
where
p,t
=
_
1
0
_
P
i,t
Pt
_


p,t

p,t1
di is a measure of the price dispersion.
III. Quantitative analysis
In this section, our formal econometric procedure is expounded. We then discuss
our results and check the ability of our models to reproduce the two stylised facts.
A. Data and econometric approach
The quarterly euro area data used in our empirical analysis are extracted from
the AWM database compiled by Fagan et al. (2005), except the monetary aggregate
10
and hours worked. Regarding nominal variables, ination is measured by the rst
dierence of the logarithm of GDP deator (YED), the shortterm nominal interest
rate is a three month rate (STN), and money growth is the rst dierence of the
logarithm of M2.
-1
0
1
2
3
4
Inflation
0
1
2
3
4
5
Short-term interest rate
0
1
2
3
4
M2 growth
-2
-1
0
1
2
Output growth
-2
-1
0
1
2
Consumption growth
-4
-2
0
2
4
Investment growth
-2
-1
0
1
2
Wage growth
-2
-1
0
1
Growth of total hours worked
Figure 2. Quarterly data for the euro area (1980Q12007Q4)
11
Regarding real variables, output growth is the rst dierence of the logarithm
of real GDP (YER), consumption growth is the rst dierence of the logarithm of
real consumption expenditures (PCR), investment growth is the rst dierence of
the logarithm of real gross investment (ITR), wage growth is the rst dierence of
the logarithm of nominal wage (WRN) divided by GDP deator, and growth of
total hours worked are the rst dierence of the logarithm of total hours worked.
Real variables are divided by the working age population, extracted from the OECD
Economic Outlook. Ohanian and Rao (2012) have build a new dataset of quarterly
hours worked for 14 OECD countries. We have then made an average of their series
of hours worked for France, Germany and Italy to obtain a series of total hours for
the euro area. Interestingly, the series thus obtained is very close to that provided by
the ECB on the common sample, i.e. 1995Q12007Q4. The series for M2 is available
from the ECB statistical warehouse since 1980Q1, which is therefore the starting
date for our analysis. The data are reported in Figure 2.
We are interested in two versions of the loglinearised model:
5
A model with a money growth rule (MG), estimated from 1980Q1 to 1998Q4;
A model with an interest rate rule (IR), estimated from 1999Q1 to 2007Q4.
The choice of the nal date prevents our estimates from being distorted by the
nonlinearities induces by the dierent size of the shocks and the zero lower
bound on nominal interest rates.
We follow the Bayesian approach to estimate the models (see An and Schorfheide,
2007, for an overview). Letting denote the vector of structural parameters to be
estimated and S
T
{S
t
}
T
t=1
the data sample, we use the Kalman lter to calcu-
late the likelihood L(, S
T
), and then combine the likelihood function with a prior
distribution of the parameters to be estimated, (), to obtain the posterior distrib-
ution, L(, S
T
)(). Given the specication of the model, the posterior distribution
cannot be recovered analytically but may be computed numerically, using a Monte-
Carlo Markov Chain (MCMC) sampling approach. More specically, we rely on the
MetropolisHastings algorithm to obtain a random draw size of 1,000,000 from the
posterior distribution of the parameters.
We use growth rates for the non-stationary variables in our data set (output, con-
sumption, investment, money and the real wage) and express gross ination, gross
interest rates and the rst dierence of the logarithm of hours worked in percent-
age deviations from their sample mean. We write the measurement equation of the
Kalman lter to match the eight observable series with their model counterparts.
Thus, the statespace form of the model is characterised by the state equation
X
t
= A()X
t1
+B()
t
,
t
i.i.d.N (0,

) ,
where X
t
is a vector of endogenous variables, and
t
is a vector of innovations to the
eight structural shocks; and the measurement equation
S
t
= C() +DX
t
+
t
,
t
i.i.d.N (0,

) ,
5
See Appendix A2 for further details on the procedure used to induce stationarity.
12
where S
t
is a vector of observable variables, that is,
S
t
= 100[log Y
t
, log C
t
, log I
t
, log (W
t
/P
t
) , log N
t
, log (M
t
/P
t
) ,
t
, R
t
];
and
t
is a vector of measurement errors.
The model contains nineteen structural parameters, excluding the parameters rela-
tive to the exogenous shocks. We calibrate six of them : the discount factor is set to
0.99, the capital depreciation rate to 0.025, the capital share in the CobbDouglas
production function is set to 0.30 (McAdam and Willman, 2008), the steadystate
price and wage markups
p
and
w
are set to 1.20 and 1.35 respectively (Everaert and
Schule, 2008), and the steadystate share of government spending in output is set to
0.20 (the average value over the sample period). The remaining thirteen parameters
are estimated. The prior distribution is summarised in Table 2. Our choices are in
line with the literature, especially with Smets and Wouters (2007), Sahuc and Smets
(2008) and Justiniano et al. (2010). As regards the interest rate elasticity of the
money demand , we assign it a Gamma density prior with mean 10 and standard
deviation 5.
B. Estimation results
The estimation results for the two models are summarised in the righthand side
panels of Table 2, where the posterior mean and the 90% condence interval are
reported. Several results are worth commenting on. First, as regards the interest
rate elasticity of the money demand, , we nd a value of 22 for the model with a
money growth rule and of 17 for the model with an interest rate rule. Such values,
combined with the respective steadystate values of ination and of economic growth,
imply an interest semi-elasticity of money demand at 1.59 for the rst sub-sample
and at 2.92 for the second one
6
.
As regards the behavior of households, we rst nd that the habit persistence
parameter h diers between periods, indicating that the reference for current con-
sumption was about 42% (resp. 68%) of past consumption from 1980 to 1998 (resp.
from 1999 to 2007). Second, the inverse of the elasticity of labor disutility, , is
similar across the samples and is approximately equal to 2.2. The wage indexation
parameter is
w
0.40 in the two model versions, slightly higher than the price in-
dexation parameter
p
0.35. This reects a now standard result that the euro area
data do not require too high a degree of price indexation. The probability that rms
are not allowed to re-optimise their price is
p
= 0.89 (resp.
p
= 0.71) in the rst
sub-sample (resp. in the second sub-sample). It implies an average duration of price
contracts of about 36 months on the period 19801998 and 14 months on the period
19992007. The probability of no wage change is
w
0.70, implying an average
duration of wage contracts of about 13 months. All these numbers are consistent
with the results reported in the survey done by Druant et al. (2012).
6
These values are close to the point estimates found in recent papers, see for instance Reynard (2004) and
Dorich (2009).
13
Table 2Prior Densities and Posterior Estimates
Parameter Prior Posterior
Model with a money growth rule
Sample: 1980Q11998Q4
Model with an interest rate rule
Sample: 1999Q12007Q4
h B[0.60,0.10] 0.417 [0.309,0.525] 0.677 [0.568,0.789]
G[10.00,5.00] 22.213 [16.113,28.049] 17.555 [9.992,25.047]
G[2.00,0.75] 2.139 [0.937,3.277] 2.234 [1.028,3.410]

u
B[0.50,0.10] 0.650 [0.536,0.769] 0.719 [0.604,0.838]

k
G[4.00,1.00] 3.418 [2.188,4.601] 4.447 [2.837,6.059]
log (
z
) G[0.50,0.10] 0.367 [0.288,0.447] 0.447 [0.347,0.544]

p
B[0.66,0.10] 0.896 [0.870,0.923] 0.712 [0.607,0.827]

w
B[0.66,0.10] 0.702 [0.587,0.813] 0.693 [0.584,0.801]

p
B[0.50,0.15] 0.338 [0.139,0.536] 0.361 [0.131,0.579]

w
B[0.50,0.15] 0.404 [0.198,0.609] 0.371 [0.166,0.575]

B[0.60,0.20] 0.493 [0.377,0.612] 0.811 [0.747,0.875]

N[0.00,0.50]
G[2.00,0.50]
0.655

[0.501,0.811]


1.742

[1.344,2.128]

y
N[0.00,0.50]
G[0.125,0.10]
1.057

[1.456,0.649]


0.198

[0.075,0.315]

w
B[0.60,0.20] 0.959 [0.928,0.989] 0.698 [0.499,0.909]

b
B[0.60,0.20] 0.625 [0.482,0.765] 0.309 [0.119,0.493]

m
B[0.60,0.20] 0.318 [0.154,0.477] 0.959 [0.924,0.996]

x
B[0.60,0.20] 0.158 [0.048,0.260] 0.616 [0.432,0.809]

p
B[0.60,0.20] 0.643 [0.437,0.861] 0.689 [0.479,0.909]

p
B[0.60,0.20]
B[0.60,0.20]
B[0.60,0.20]
0.982
0.789
0.589
[0.967,0.996]
[0.789,0.671]
[0.327,0.963]
0.969
0.619
0.506
[0.936,0.999]
[0.405,0.842]
[0.165,0.785]

w
IG[0.25,2.00] 0.161 [0.120,0.199] 0.172 [0.131,0.211]

b
IG[0.25,2.00] 0.095 [0.066,0.121] 0.085 [0.057,0.113]

m
IG[0.25,2.00] 0.928 [0.804,1.051] 0.996 [0.835,1.115]

x
IG[0.25,2.00] 0.719 [0.606,0.834] 0.247 [0.168,0.322]

p
IG[0.25,2.00] 0.165 [0.121,0.209] 0.155 [0.113,0.198]

z
IG[0.25,2.00] 0.705 [0.608,0.795] 0.543 [0.441,0.644]

g
IG[0.25,2.00] 0.416 [0.360,0.472] 0.272 [0.219,0.324]

r
IG[0.25,2.00] 0.352 [0.251,0.449] 0.113 [0.088,0.138]
Note: This table reports the prior distribution, the mean and the 90 percent condence interval
(within square brackets) of the estimated posterior distribution of the structural parameters.
14
The policy parameters
_

,
y
_
(0.66, 1.06) and

= 0.49 indicate that, during


the period 19801998, money growth was moving smoothly with a little weight on
ination and a larger weight on output growth. As expected, the gure is dierent
for the period 19992007. Indeed, the policy parameters
_

,
y
_
(1.74, 0.20) and

= 0.81 indicate that the ECB acts very gradually with a large weight on ination,
consistent with its mandate.
The estimates of the serial correlation of shocks display a dierence between the two
samples. For instance, the serial correlation of wage markup and preference shocks is
stronger in the rst subsample while the serial correlation of velocity and investment
shocks is higher in the second subsample. Finally, notice that the standard error of
the velocity shock is slightly higher in the recent period.
C. Model evaluation
In this subsection, we analyze the performance of the models at replicating the two
stylised facts. To do so, we generate 1000 samples of size consistent with the empirical
counterpart (after a burn-in period of 1000 observations) from the two model versions
using the posterior estimates. For each simulation, we compute the lowfrequency
relationships between ltered money growth, ination and the nominal interest rate.
The results of this exercise are displayed in Figure 3.
0
10
20
30
40
50
60
70
80
0.78 0.83 0.87 0.92 0.97
Inflation on money growth
0
10
20
30
40
50
60
70
80
90
0.51 0.58 0.66 0.73 0.80
Interest rate on money growth
Panel a. Model with a money growth rule (1980Q11998Q4)
0
10
20
30
40
50
60
70
80
-0.23 -0.15 -0.06 0.03 0.11
Inflation on money growth
0
10
20
30
40
50
60
70
80
90
-0.37 -0.25 -0.12 0.00 0.12
Interest rate on money growth
Panel b. Model with an interest rate rule (1999Q12007Q4)
Figure 3. Simulated coecients of the regressions on ltered data
15
First, we observe that the model with a money growth rule (period 19801998)
replicates quite well the two quantity theory propositions: the mean of the coecient
of the regression of ination on money growth is 0.91 and the mean of the coecient
of the regression of the nominal interest rate on money growth is 0.70 (panel a).
Second, the model with an interest rate rule (period 19992007) also reproduces the
empirical fact of absence of quantity theory. Indeed, the mean of the coecient of
the regression of ination on money growth is 0.04 and the mean of the coecient
of the regression of the nominal interest rate on money growth is 0.03 (panel b).
This exercise conrms that our structural model is able to reproduce the rst stylised
fact.
Let us focus now on the second stylised fact, the shift in macroeconomic volatility.
To do so, we use the simulated data and compute their standard deviations. Table
3 reports the simulated standard deviation for the two models. Given that Bayesian
estimation operates by trying to match the entire autocovariance function of the
data, there is a tension between matching standard deviations and other second
moments of the data. Therefore, the researcher should not expect perfect accounting
of the observed volatilities. Despite this, the models are able to replicate to a large
extent the empirical evidence at hand. Indeed, the theoretical framework successfully
delivers the dierences in size of the slowdown in the volatility of ination and nominal
interest rate, as observed in the data.
Table 3Model Fit: Standard Deviations
1980Q11998Q4 1999Q12007Q4
Data Model (MG) Data Model (IR)
Mean 90% CI Mean 90% CI
Ination 0.718 0.736 [0.551,0.954] 0.189 0.248 [0.186,0.314]
Interest rate 0.769 0.784 [0.627,0.979] 0.236 0.227 [0.174,0.310]
Money growth 0.515 0.674 [0.555,0.830] 0.601 0.805 [0.727,0.912]
Note: For each 1000th parameter draw from the posterior distribution, 1000 samples with the same
length as the data are generated (after discarding 1000 initial observations). This table reports the
mean and the 90 percent condence interval (within square brackets).
However, the modelimplied standard deviations for money growth is larger than
that in the data. The reason is that the model imposes a common trend between
output and money. This constraint is quite strong since the two series show dier-
ent trends. To compensate, the volatility of money growth should increase. This
dierence in trends is identical in both models. Thus, it does not alter the relative
dierence between the two variances of money growth. We can conclude that the
model proposed in this paper is a good candidate for analyzing the two stylised facts.
16
IV. Assessing the drivers of both stylised facts
In this section, we analyze the model drivers of both the breakdown of the two
quantity propositions and the change in the macroeconomic volatility. Our aim is to
understand whether changes in monetary policy, shocks or private sector coecients
across subsamples are responsible for the two empirical facts. To do so, we start
from the model (MG) hereafter referred to as benchmark and perform two sets
of counterfactual exercises. The rst set is summarised below:
Counterfactual (1): we analyze the role played by monetary policy (shifts in
the coecients and the instrument).
Counterfactual (2): we study the relevance of the good luck hypothesis, i.e. the
role of the size and source of the shocks hitting the economy.
Counterfactual (3): we assess the relevance of a change in the structure of the
economy (i.e. a modication of preferences and technology).
For illustration purposes, let us consider Counterfactual (1). We proceed by per-
forming 1000 simulations for each 1000th draw in the posterior simulator using the
following procedure. We simulate the model economy for 80 periods (after a burn-in
of 1000 observations) using the parameter estimates vector characterizing the 1980
1998 period but with the estimated interest rate rule obtained on the 19992007
sample period. We then compute (i) the low frequency relationships between ltered
money growth, ination and the nominal interest rate, and (ii) the standard devia-
tion of the endogenous variables. The other counterfactual exercises are performed
in the same way.
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
-0.2 0 0.2 0.4 0.6 0.8 1 1.2
Inflation on money growth
(1)
(2)
(3)
(6)
(4)
(5) (IR)
(MG)
Figure 4. Coecients of the regressions on ltered data: Counterfactuals.
Note: The two crosses correspond to the models (MG) and (IR); The
bullets correspond to the counterfactuals 16.
17
Figure 4 reports the lowfrequency relationships between ltered money growth,
ination and the nominal interest rate. First, there is no signicant dierence be-
tween Counterfactual (3) and the benchmark, indicating that the potential changes
in preferences and technology by the private sector did not inuence the two quantity
propositions. Second, although the good luck hypothesis Counterfactual (2) does
not modify the relationship between money growth and nominal interest rate, that
between money growth and ination deteriorates. Indeed the slope of the regression
of money growth on ination changes from 0.909 to 0.562. Finally, Counterfactual
(1) clearly shows that monetary policy is the key factor explaining the breakdown
of the quantity theory of money. The slope of the regression of money growth on
ination (resp. money growth on nominal interest rate) changes from 0.909 (resp.
0.702) to 0.286 (resp. 0.172).
Table 4 reports the standard deviations generated in each counterfactual simu-
lation. The simulations lead to conclusions in line with previous remarks. In-
deed, imposing the economic structure from the model (IR) into the model (MG)
Counterfactual (3) increases strongly the standard deviations of the nominal vari-
ables, which is inconsistent with the empirical regularities. By contrast, Counterfac-
tual (1) and Counterfactual (2) allow a dramatic reduction of the standard deviations
of ination and nominal interest rate. However, only Counterfactual (1), i.e. the ex-
ercise in which the interest rate rule from model (IR) is imposed in the model (MG),
leads to an increase in the standard deviation of money growth.
Table 4Counterfactuals: Standard Deviations
Specication Variable
Ination Interest rate Money growth
Mean 90% CI Mean 90% CI Mean 90% CI
(MG) 0.736 [0.551,0.954] 0.784 [0.627,0.979] 0.674 [0.555,0.830]
(1) 0.337 [0.257,0.420] 0.452 [0.385,0.528] 0.990 [0.895,1.099]
(2) 0.600 [0.392,0.961] 0.572 [0.434,0.763] 0.583 [0.450,0.848]
(3) 1.121 [0.763,1.659] 0.991 [0.649,1.493] 0.852 [0.579,1.287]
(4) 0.736 [0.586,0.915] 0.820 [0.672,1.014] 1.130 [0.981,1.310]
(5) 0.255 [0.196,0.322] 0.641 [0.518,0.788] 0.494 [0.461,0.528]
(6) 1.011 [0.692,1.397] 1.004 [0.718,1.372] 0.786 [0.550,1.098]
Note: For each 1000th parameter draw from the posterior distribution, 1000 samples with the same
length as the data are generated (after discarding 1000 initial observations). This table reports the
mean and the 90 percent condence interval (within square brackets).
18
The signicant eect of monetary policy to reproduce the two stylised facts led to
scrutinise its characteristics in order to evaluate their respective contributions. To
do that, we carry out the following set of counterfactual exercises:
Counterfactual (4): We analyze the role played by the type of policy instru-
ment (money growth, M
t
, versus interest rate, R
t
). To realise this exercise,
we target the standard deviation of ination calculated in the model (MG) and
deduce the coecients of an interest rate rule consistent with this value. We
then simulate as explained above.
Counterfactual (5): We study the relevance of the degree of reactivity, i.e.
the greater or lesser interest rate response to ination and to the growth rate
of output
_

,
y
_
. To do that, we target the standard deviation of ination
calculated in the model (IR) and deduce the coecients of a money growth rule
consistent with this value. We then simulate as explained above.
Counterfactual (6): We assess the relevance of the degree of gradualism, i.e.
a change in the value of the parameter

.
Figure 4 shows that the regressions obtained in Counterfactuals (4) and (6) do not
dier signicantly from the one obtained with the benchmark model. While the policy
instrument and the degree of gradualism seem to play a minor role, a monetary policy
shift towards a more aggressive antiinationary stance, as implemented from 1999 in
the euro area, leads to a breakdown of the quantity theory of money (Counterfactual
(5)). Turning to the analysis of standard deviations, we note that the stronger
reactivity by the monetary authority allows to explain the sharp drop in the volatility
of all the nominal variables. However, only a shift from a money growth rule to
an interest rate rule provides a marked increase in the volatility of money growth
(Counterfactual (5)).
The underlying mechanisms are the following. On the one hand, the breakdown
of the quantity theory of money when reacting suciently aggressively to incipient
inationary pressures is linked to the change in the variancecovariance matrix of
the macroeconomic variables. The quantity theory implies that any change in the
monetary aggregate induces a variation in the same direction and same magnitude
of price and of nominal interest rate. If price stability is the goal of central bank,
any increase in prices should result in a sharp contraction in the money supply. As
a consequence, both variances and covariances between nominal variables decrease.
But, implementing price stability or ination targeting while the underlying theory
establishes similar variations between prices and the monetary aggregate necessar-
ily induces a stronger fall of the covariances. This mechanical eect blurs the two
quantity propositions.
On the other hand, the interest rate as an instrument of monetary policy allows
understanding of the increase in the volatility of money growth. With an interest
elastic demand for real money balances, any shock that aects the path for expected
ination or the real interest rate causes money demand to shift. When the central
bank follows an interest rate rule, the money stock is endogenous and ination is
xed by the policy rule. Money absorbs the adjustment and the central bank can
19
accommodate this jump in the money stock almost instantaneously and with little
cost. It is the money stock, rather than the price level, that responds by shifting
downward to clear the money market. Hence, in an economy with stochastic ination
and an interest rate rule for monetary policy, the money growth rate is much more
volatile than the ination rate (Gavin et al., 2005). This explanation comes on top
of the possible composition eect of the monetary aggregates. Indeed, M2 comprises
two distinct categories (associated with portfolio motives and transaction purposes,
respectively) that are expected to move in opposite directions following a movement
in the short-term interest rate. If the remunerated category of M2 exercises a stronger
inuence than the most liquid category, an increase in the interest rate will result in
even higher M2 growth. We observe, however, that M1 volatility is greater than M2
volatility over the postEMU period (see Appendix A.1).
V. Conclusion and nal remarks
We have estimated a structural monetary model of the euro area business cycle to
examine the sources of both the empirical breakdown of the quantity theory of money
and the shift in the volatility of nominal variables over the period 19802007. Our
results suggest that a more antiinationary monetary policy and the transition from
a money growth to an interest rate rule explain these macroeconomic developments.
More generally, we shed light on the strong link between the quantity theory of money
and the conduct of monetary policy. Periods of improvements (resp. impairments)
in the conduct of monetary policy or in the functionning of the money market are
characterised by an empirical breakdown (resp. return) of the quantity theory of
money.
A very recent illustration is the return of the quantity of money concomitantly with
the implementation of nonstandard measures (see Appendix A3). But, contrary to
historical changes in the degree of responsiveness or instrument of the central bank,
unit slopes in scatter plots emerge because the usual interest rate channel is broken.
Since 2008, the relationship between the expected path of policy rates and market
rates broke down because the liquidity premia widened and became volatile. The
Lehman Brothers bankruptcy caused a freeze in the interbank market, forcing the
ECB to inject its funding capacity into it. The predominance of this liquidity channel
mainly aected banks and central bank balance sheets. After lowering its interest
rates drastically to a level close to zero, the ECB has eased monetary conditions
by increasing sharply the size of its balance sheet. However, given the atony of
the economy and the heightened uncertainty, banks prefer hoarding cash instead
of spending it. Liquidity has then circulated among nancial institutions but was
not transmitted to the real economy. Consequently, the huge increase in liquidity
did not trigger ination pressures. In addition, such a liquidity can be sterilized
by symmetrical operations of withdrawal. The ination rate remained practically
constant at the 2% level over the past four years. Moreover, the presence of excess
liquidity in the overnight market and the resulting recourse to the deposit facility
implied a fall in the euro interbank overnight money market rate. The evolution of the
interest rate has then become independent of ination. The traditional relationships
between ination, money growth and interest rate seem no longer valid.
20
With a larger sample available, the model and our approach would be used and
probably extended to examine more carefully the crisis period. For now, the recent
behavior of the monetary aggregates suggests that the quantity theory of money is
useful for detecting nancial stress, supporting McCallum and Nelson (2010).
REFERENCES
An S. and Schorfheide F. 2007. Bayesian Analysis of DSGE Models, Econo-
metric Reviews, 26, 113172.
Andres A., Hansen G. and Ohanian L. 2007. Why Have Business Cycle Fluc-
tuations Become Less Volatile?, Economic Theory, 32, 4358.
Andrs J., Lopez-Salido J. and Valls J. 2006. Money in an Estimated Busi-
ness Cycle Model of the Euro Area, The Economic Journal, 116, 457477.
Bernanke B. and Mishkin F. 1992. Central Bank Behavior and the Strategy of
Monetary Policy: Observations from Six Industrialized Countries. in O. Blanchard
and S. Fischer (eds), NBER Macroeconomics Annual, 183238, Cambridge: MIT
Press.
Christiano L., Eichenbaum M., and Evans C. 2005. Nominal Rigidities and
the Dynamic Eects of a Shock to Monetary Policy, Journal of Political Economy,
113, 145.
Dorich J. 2009, Resurrecting the Role of Real Money Balance Eects, Working
Paper #2009-24, Bank of Canada.
Druant M., Fabiani S., Kezdi G., Lamo A., Martins F. and Sabbatini R.
2012. Firms Price and Wage Adjustment in Europe: Survey Evidence on Nominal
Stickiness, Labour Economics, 19, 772782.
Everaert L. and Schule W. 2008. Why It Pays to Synchronize Structural Re-
forms in the Euro Area Across Markets and Countries, IMF Sta Papers, 55,
356366.
Fagan G., Henry J., and Mestre R. 2005. An Area-Wide Model (AWM) for
the Euro-Area, Economic Modelling, 22, 3959.
Gavin W., Keen B. and Pakko M. 2005. The Monetary Instrument Matters,
Federal Reserve Bank of St. Louis Review, 87, 633658.
Ireland P. 2004. Moneys Role in the Monetary Business Cycle, Journal of
Money, Credit and Banking, 36, 969983.
Justiniano A. and Primiceri G. 2008. The Time Varying Volatility of Macro-
economic Fluctuations, American Economic Review, 98(3), 604641.
Justiniano A., Primiceri G., and Tambaloti A. 2010. Investment Shocks and
Business Cycles, Journal of Monetary Economics, 57, 132145.
21
Lucas R. 1980. Two Illustrations of the Quantity Theory of Money, American
Economic Review, 70, 1005-1014.
McAdam P. and Willman A. 2008. Medium Run Redux: Technical Change,
Factor Shares and Frictions in the Euro Area, Working Paper #915, European
Central Bank.
McCallum B. and Nelson E. 2010. Money and Ination: Some Critical Issues,
in: Benjamin M. Friedman and Michael Woodford (ed.), Handbook of Monetary
Economics, volume 3, chapter 3, pages 97153, Elsevier.
Ohanian L. and Rao A. 2012. Aggregate Hours Worked in OECD Countries:
New Measurement and Implications for Business Cycles, Journal of Monetary Eco-
nomics, 59, 4056.
Reynard S. 2004. Financial Market Participation and the Apparent Instability of
Money Demand, Journal of Monetary Economics, 51, 12971317.
Sahuc J.-G. and Smets F. 2008. Dierences in Interest Rate Policy at the ECB
and the Fed: An Investigation with a Medium-Scale DSGE Model, Journal of
Money, Credit and Banking, 40, 505521.
Sargent T. and Surico P. 2011. Two Illustrations of the Quantity of Money:
Breakdowns and Revivals, American Economic Review, 101, 109128.
Smets F. and Wouters R. 2007. Shocks and Frictions in US Business Cycles:
A Bayesian Approach, American Economic Review, 97, 586606.
Teles P. and Uhlig H. 2010. Is Quantity Theory Still Alive?, Working Paper
#16393, National Bureau of Economic Research.
Whiteman C. H. 1984. Lucas on the Quantity Theory: Hypothesis Testing with-
out Theory, American Economic Review, 74, 74249.
Woodford M. 2008. How Important is Money in the Conduct of Monetary Pol-
icy?, Journal of Money, Credit and Banking, 40, 15611598.
22
Appendix not Intended for Publication
A1. Robustness: Results for M1, M2 and M3 aggregates
Table A11Coecients of the regressions on ltered data (19802007)
on m R on m
PreEMU PostEMU PreEMU PostEMU
M1 0.63 0.08 0.57 0.25
M2 1.34 0.03 1.38 0.12
M3 1.16 0.10 1.08 0.01
Table A12Standard deviation of the growth of various monetary aggregates (19802007)
PreEMU PostEMU
PostEMU
PreEMU
M1 0.69 1.31 1.89
M2 0.52 0.60 1.17
M3 0.53 0.61 1.15
A2. Models details
A. Nonlinear equilibrium conditions
This section reports the rstorder conditions for the agents optimizing problems
and the other relationships that dene the equilibrium of the baseline model.
Eective capital:
K
t
= u
t

K
t1
Capital accumulation:

K
t
= (1 )

K
t1
+
i,t
_
1 S
_
I
t
I
t1
__
I
t
Marginal utility of consumption:

t
=

b,t
C
t
hC
t1

h
b,t+1
C
t+1
hC
t
Consumption Euler equation:

t
= R
t
E
t
_

t+1
P
t
P
t+1
_
23
Money demand equation:

t
=
b,t

m,t

1
z,t
_
M
t+s
P
t+s
_

+E
t
_

t+1
P
t
P
t+1
_
Investment equation:
1 = Q
t

i,t
_
1 S
_
I
t
I
t1
_

I
t
I
t1
S

_
I
t
I
t1
__
+E
t
_

t+1

t
Q
t+1

i,t+1
_
I
t+1
I
t
_
2
S

_
I
t+1
I
t
_
_
Tobins Q:
Q
t
= E
t
_

t+1

t
_
R
k
t+1
P
t+1
u
t+1
(u
t+1
) + (1 ) Q
t+1
__
Capital utilisation:
R
k
t
= P
t

(u
t
)
Production function:
Y
i,t
= K
i,t

[Z
t
N
i,t
]
1
Z
t
F
Labor demand:
W
t
= (1 ) Z
t
_
K
t
Z
t
N
t
_

MC
t
Capital renting:
R
k
t
=
_
K
t
Z
t
N
t
_
1
MC
t
Price setting:
E
t

s=0
(
p
)
s

t+s

t
Y

t,t+s
_
P

t

p
t,t+s

p,t+s
MC
t+s

= 0
Aggregate price index:
P
t
=
_
(1
p
) (P

t
)
1/(p,t1)
+
p
_

1
p

p
t1
P
t1
_
1/(p,t1)
_
(
p,t
1)
Wage setting:
E
t

s=0
(
w
)
s

t+s
N

t,t+s
_
W

t
P
t+s

w
t,t+s

b,t+s

w,t+s
_
N

t,t+s
_

t+s
_
= 0
24
Aggregate wage index:
W
t
=
_
(1
w
) (W

t
)
1/(w,t1)
+
w
_

1
p

w
t1
W
t1
_
1/(w,t1)
_
(w,t1)
Government spending:
G
t
=
_
1
1

g,t
_
Y
t
Monetary policy rule:

t
=

t1
_
_

_
Y
t
Y
t1

z
_

y
_
(1

r,t
Resource constraint:
Y
t
= C
t
+I
t
+G
t
+(u
t
)

K
t1

p,t
Y
t
=
_
u
t

K
t1
_

[Z
t
N
t
]
1
Z
t
F
B. Stationary equilibrium
To nd the steadystate, we express the model in stationary form. Thus, for the
nonstationary variables, let lowercase denote their value relative to the technology
process Z
t
:
y
t
Y
t
/Z
t
k
t
K
t
/Z
t

k
t


K
t
/Z
t
i
t
I
t
/Z
t
c
t
C
t
/Z
t
g
t
G
t
/Z
t

t

t
Z
t
w
t
W
t
/ (Z
t
P
t
) w

t
W

t
/ (Z
t
P
t
) m
t
M
t
/ (Z
t
P
t
)
where we note that the marginal utility of consumption
t
will shrink as the economy
grows, and we express the wage in real terms. Also, denote the real rental rate of
capital and real marginal cost by
r
k
t
R
k
t
/P
t
and mc
t
MC
t
/P
t
,
and the optimal relative price as
p

t
P

t
/P
t
.
Then we can rewrite the model in terms of stationary variables as follows.
Eective capital:
k
t
=
u
t

k
t1

z,t
25
Capital accumulation:

k
t
= (1 )

k
t1

z,t
+
i,t
_
1 S
_
I
t
I
t1

z,t
__
x
t
Marginal utility of consumption:

t
=

b,t
c
t
h
c
t1

z,t
E
t
_

_
h
b,t+1

z,t+1
_
c
t+1
h
c
t

z,t+1
_
_

_
Consumption Euler equation:

t
= R
t
E
t
_

t+1

z,t+1

t+1
_
Money demand equation:

t
=
b,t

m,t
m

t
+E
t
_

t+1

z,t+1

t+1
_
Investment equation:
1 = q
t

i,t
_
1 S
_
i
t
i
t1

z,t
_

i
t
i
t1

z,t
S

_
i
t
i
t1

z,t
__
+E
t
_

t+1

z,t+1
q
t+1

i,t+1
_
i
t+1
i
t

z,t+1
_
2
S

_
i
t+1
i
t

z,t+1
_
_
Tobins Q:
q
t
= E
t
_

t+1

z,t+1
_
r
k
t+1
u
t+1
(u
t+1
) + (1 ) Q
t+1
_
_
Capital utilisation:
r
k
t
=

(u
t
)
Production function:
y
i,t
= k

i,t
N
1
i,t
F
Labor demand:
w
t
= (1 )
_
k
t
N
t
_

mc
t
Capital renting:
r
k
t
=
_
k
t
N
t
_
1
mc
t
26
Price setting:
E
t

s=0
(
p
)
s

t+s

t
y

t,t+s
_
p

t
P
t
P
t+s

p
t,t+s

p,t+s
mc
t+s
_
= 0
Aggregate price index:
1 =
_
(1
p
) (p

t
)
1/(
p,t
1)
+
p
_

1
p

p
t1
1

t
_
1/(p,t1)
_
(p,t1)
Wage setting:
E
t

s=0
(
w
)
s

t+s
N

t,t+s
_
w

t
P
t
P
t+s
Z
t
Z
t+s

w
t,t+s

b,t+s

w,t+s
N

t,t+s

t+s
_
= 0
Aggregate wage index:
w
t
=
_
(1
w
) (w

t
)
1/(
w,t
1)
+
w
_

1
p

w
t1
w
t1

z,t
_
1/(w,t1)
_
(w,t1)
Government spending:
g
t
=
_
1
1

g,t
_
y
t
Monetary policy rule:

t
=

t1
_
_

_

z,t
y
t

z
y
t1
_

y
_
(1

r,t
Resource constraint:
y
t
= c
t
+x
t
+g
t
+(u
t
)

k
t1
/
z,t

p,t
y
t
=
_
u
t

k
t1
_

N
1
t
F
C. Steady state
We use the stationary version of the model to nd the steady state, and we let
variables without a time subscript denote steadystate values. First, the expression
for Tobins Q implies that the rental rate of capital is
r
k
=

z

(1 )
27
and the price-setting equation gives marginal cost as
mc =
1

p
.
The capital/labor ratio can then be retrieved using the capital renting equation:
k
N
=
_

mc
r
k
_
1/(1)
,
and the wage is given by the labor demand equation as
w = (1 ) mc
_
k
N
_

.
The production function gives the output/labor ratio as
y
N
=
_
k
N
_

F
N
,
and the xed cost F is set to obtain zero prots at the steady state, implying
F
N
=
_
k
N
_

w r
k
k
N
.
The output/labor ratio is then given by
y
N
= w +r
k
k
N
=
r
k

k
N
.
Finally, to determine the investment/output ratio, use the expressions for eective
capital and physical capital accumulation to get
i
k
=
_
1
1

z
_

z
,
implying that
i
y
=
i
k
k
N
N
y
=
_
1
1

z
_

z
r
k
.
Given the government spending/output ratio g/y, the consumption/output ratio
is then given by the resource constraint as
c
y
= 1
i
y

g
y
.
In addition, we have:
R =

z

.
28
D. Loglinearised version
We loglinearise the stationary model around the steady state. Let
t
denote the
log deviation of the variable
t
from its steadystate leve

l :

t
log
_

_
.
The log-linearised model is then given by the following system of equations for the
endogenous variables.
Eective capital:

k
t
+
z,t
= u
t
+

k
t1
Capital accumulation:

k
t
=
1

z
_

k
t1

z,t
_
+
_
1
1

z
_
(
t
+
i,t
)
Marginal utility of consumption:

t
=
h
z
(
z
h) (
z
h)
c
t1


2
z
+h
2

(
z
h) (
z
h)
c
t
+
h
z
(
z
h) (
z
h)
E
t
c
t+1

h
z
(
z
h) (
z
h)

z,t
+
h
z
(
z
h) (
z
h)
E
t

z,t+1
+

z

z
h

b,t

z
h
E
t

b,t+1
Consumption Euler equation:

t
= E
t

t+1
+
_

R
t
E
t

t+1
_
E
t

z,t+1
Money demand equation:
m
t
=
1

(
b,t
+
m,t
)
1

1
(R1)

R
t
Investment equation:

t
=
1
1 +
(
t1

z,t
) +

1 +
E
t
(
t+1
+
z,t+1
) +
1

2
z
(1 +)
( q
t
+
i,t
)
Tobins Q:
q
t
=
(1 )

z
E
t
q
t+1
+
_
1
(1 )

z
_
E
t
r
k
t+1
( r
t
E
t

t+1
)
29
Capital utilisation:
u
t
=
1
u

u
r
k
t
Production function:
y
t
=
Y +F
Y
_

k
t
+ (1 ) n
t
_
Labor demand:
w
t
= mc
t
+

k
t
n
t
Capital renting:
r
k
t
= mc
t
(1 )

k
t
+ (1 ) n
t
Phillips curve:

t
=

p
1 +
p

t1
+

1 +
p
E
t

t+1
+
(1
p
) (1
p
)

p
_
1 +
p
_ ( mc
t
+
p,t
)
Wage curve:
w
t
=
1
1 +
w
t1
+

1 +
E
t
w
t+1

(1
w
) (1
w
)

w
(1 +)
_
1 +

w

w
1
_ ( mrs
t
+
w,t
)
+

w
1 +

t1

1 +
w
1 +

t
+

1 +
E
t

t+1

1
1 +

z,t
+

1 +
E
t

z,t+1
Marginal rate of substitution:
mrs
t
= w
t

_
n
t

t
+
b,t
_
Government spending:
g
t
= y
t
+
1 g/y
g/y

g,t
Monetary policy rule:

t
=

t1
(1

)
_


t
+
y
( y
t
y
t1
+
z,t
)

+
r,t
Resource constraint:
y
t
=
c
y
c
t
+
i
y

t
+
g
y
g
t
+
r
k
k
y
u
t
30
A3. Scatter plots for the period 20082011
We extend our sample to include the years 2008 to 2011 and apply the low
frequency approach. Figure A31 displays the resulting scatters of points for the
current crisis period. While the slopes were at or even negative until 2007, we nd
that they revert positive over the past four years. In addition, one can even see a
concentration of points around the diagonal. Such ndings reveal a return of the
quantity theory of money.
0
2
4
6
8
0 2 4 6 8
Money growth
2010-2011
2008-2009
0
2
4
6
8
0 2 4 6 8
Money growth
2010-2011 2008-2009
Figure A31. Scatter plots of ltered money growth, ination and interest rate
31

You might also like