Measuring Monetary Policy Efficiency in European
Union Countries: The Pre-EMU Years
Stefan Krause∗
October 2004
Abstract
This paper proposes a method for measuring the contribution of improved monetary policy to the changes in macroeconomic performance identified with inflation and
output stability. Our technique involves estimating actual and optimal policy rules as
a function of the aggregate supply and demand shocks, with the purpose of examining
how much of the change in performance can be accounted for by changes in the volatility of the aggregate shocks and how much can be ascribed to improvements in policy
efficiency
We study the change from the 1980s to the 1990s in macroeconomic performance for
14 European Union countries. Our findings suggest that improved monetary policy has
played an important stabilization role in almost all European Union countries, while
a diminished exposure to aggregate shocks has largely contributed towards improved
performance in at least seven countries
JEL classification: E52, E58
∗
Department of Economics, Emory University. I would like to thank Fabio Canova, Stephen G. Cecchetti,
Michael Ehrmann, Paul Evans, Alfonso Flores-Lagunes, Nelson Mark, James Peck and Frank Smets for their
very useful suggestions, and the seminar participants at The Ohio State University, Federal Reserve Bank
of Boston, Emory University and Virginia Tech for their comments. A substantial part of the research was
undertaken while I was participating at the Graduate Research Programme at the European Central Bank.
For comments please contact me at skrause@emory.edu or 404-727-2944.
1
Introduction
Over the past years, macroeconomic performance has improved markedly in industrialized
and developing countries alike. Both the level and variability of inflation were lower in the
latter half of the 1990s than they were in the preceding ten years. Looking at a broad crosssection of 63 countries, median inflation dropped from 7.04% in 1985:I-1994:IV to 2.97% in
1995:I-1999:IV. The decrease in average inflation has been even sharper, falling from 83.19%
to 8.59%. Inflation rose in only 10 of the 63 countries examined, and in most of those the
increase was small - only in Ghana, Indonesia and Turkey did average inflation rise by more
than 2 percentage points.
Still, better macroeconomic performance usually means more than just lower inflation: it
also involves more stable inflation and real growth. Improved macroeconomic outcomes can
arise as the result of several factors. One possibility is that the world has become a more
stable place. If there are no shocks hitting an economy, it will surely be more stable. Alternatively, monetary policy makers may have become more skillful at implementing policies to
meet their stabilization objectives.
The objective of this paper is to develop a method for measuring the contribution of
improved monetary policy to the observed changes in macroeconomic performance. Specifically, we look at the changes in the variability of inflation and output for the 14 countries of
the European Union (excluding Luxemburg, which did not have an independent monetary
policy during the analyzed period), and compare the 1980s and the 1990s.1 We estimate a
simple macroeconomic model of inflation and output for each country specifying the dynamics of inflation and output as a function of the interest rate — our measure of central bank
policy — as well as additional exogenous variables. Using the estimated model, we are able
identify the monetary policy rules as a function of the aggregate shocks and the parameters
of the economy, for two sample periods, 1983 to 1990 and 1991 to 1998. This enables us to
compute the change in macroeconomic performance for each country using a weighted sum
1
See next section for details.
1
of inflation and output volatility, and examine how much of that change can be accounted
for by changes in the volatility of the aggregate shocks and how much can be ascribed to
improvements in policy efficiency.
Throughout the paper we assume that monetary policy is the main tool used for stabilization purposes. Even though we only consider changes in the proficiency of monetary
policy makers and the variability of aggregate shocks as sources of changes in macroeconomic
performance, there are clearly other factors that may have an effect in macroeconomic stability. For instance, changes in central bank independence, transparency and credibility can
affect the ability of policy makers to perform effectively.2 Furthermore, developments in the
information about the current state of the economy and reduced uncertainty regarding the
nature and effect of disturbances allow for improvements in the reaction of policy makers
to the economic conditions.3 Finally, fiscal, trade and labor market policies may have an
impact both on the structure of the economy and on monetary policy effectiveness.4 While
our techniques are not refined enough to distinguish among all of these possible causes of
the changes that we analyze, we consider them a necessary first step.
The remainder of the paper is divided into the following sections. In Section 2 we take
a preliminary look at the data on macroeconomic outcomes for the 14 countries of interest,
which allows us to establish whether there is a common behavior or trend of inflation and
output growth variability in European Union countries. Section 3 introduces the proposed
2
Empirical studies by Grilli, Masciandaro and Tabellini (1991), Cukierman, Webb and Neyapti (1992)
and Alesina and Summers (1993) find evidence of a negative correlation of central bank independence with
lower and more stable inflation, within industrialized countries.
Chortareas, Stasavage and Sterne (2002) examine the association between the cross-country differences
in macroeconomic outcomes and the degree of transparency exhibited by monetary policy. Their results
suggest that a high degree of transparency in economic forecasts is associated with a lower inflation in all
countries, except for those that directly target the exchange rate.
Finally, using cross-sectional data on a broad range of countries, Cecchetti and Krause (2002) find supporting evidence to the theoretical argument proposed by Faust and Svensson (2001) and others that more
credible central banks deliver superior macroeconomic performance.
3
See Rudebusch (2001) and the references therein.
4
For instance, a lower level of inventories held by firms may reduce the effects of supply disturbances to
the economy (Kahn, McConnell and Pérez-Quirós, 2002). In this paper we are only interested in the effect
that the actual shocks have in the economy and therefore such technology improvements would be shown as
a reduction in the variability of aggregate supply shocks.
2
method to analyze the changes in macroeconomic performance and determine the role of
monetary policy in the stabilization of inflation and output. The main tool used to compare
policy efficiency between the two periods of interest is to contrast the actual policy followed
by central banks with an optimal policy rule, which results from an optimization program.
The novelty of this approach consists of deriving and estimating the rules as policy responses
to aggregate shocks, instead of macroeconomic variables.
Section 4 presents and discusses the main results. Our findings suggest that improved
monetary policy has played an important stabilization role in almost all European Union
countries. At the same time, most countries also experienced reduced demand and supply
shock variability, making a substantial contribution towards improved performance in at
least seven countries. Section 5 concludes the paper.
2
Empirical Facts
Our first step is to take a first look at the data on macroeconomic performance over a period
of 16 years for Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy,
Netherlands, Portugal, Spain, Sweden and the United Kingdom. To this end, Figure 1 shows
inflation and output variability for two periods, 1983 to 1990 and 1991 to 1998. Inflation
variability is measured as the squared deviation from a target level of 2%, while output
variability is the squared deviation from a log-linear trend. We perform a more detailed
discussion of these target choices in Section 4. We choose 1983 as the starting year as a
result of data availability for the interest rate, while the choice of 1998 as the final year of
the sample is due to the fact that this is the last year before the European Monetary Union
comes into effect, discontinuing independent interest rate policy in 11 countries.
From the figure we can observe that, for the average country, the second period has been
characterized by lower inflation variability and higher output variability, as compared to the
previous one. Hence, we cannot conclude from casual observation that there has been an
3
unambiguous improvement in macroeconomic performance for most countries. Accounting
for performance changes requires us then to look at each country individually and also
examine how the preferences of the central bank towards inflation and output stabilization
have changed between one period and the next.
It is also worth noting that, while during the first period we observe a weak positive
relationship between output and inflation variability, the second period is characterized by
the presence of a trade-off between inflation and output variability. If in the 1990s monetary
policies did indeed come closer to the optimal in general, as we argue in this paper, this
latter result would be consistent with the notion, first discussed in Taylor (1979), that there
is an efficiency frontier along which policy makers can operate.
The information for each individual country is presented in Figure 2. We can see that, in
Denmark, Greece, Italy, the Netherlands and the UK, both output and inflation variability
fell, implying an unambiguous improvement in performance. For all countries except for
Germany, inflation variability fell between the first and the second period; we further note
that the eight countries that experienced higher output variability also exhibited a decrease
in inflation variability. This aspect can be linked to the increasing importance now placed
by central banks on explicit or implicit inflation targeting, as documented by Fry, Julius,
Mahadeva, Roger and Sterne (2000), among others. As a result, we expect to observe changes
in the relative preferences towards inflation and output stability between both periods of
interest.
In the next section we show how we obtain the measures of macroeconomic performance,
using the information in Figure 2, and describe the method that enables us to account for
the sources of performance changes. The procedure requires us first to identify the monetary
authority’s reaction to economic conditions and its relative preferences towards inflation and
output variability, and so we begin with this discussion below.
4
3
Method for Measuring Macro Performance and Policy Efficiency
As described in the previous section, we are interested in deriving a method to determine
the sources of the observed increase in macroeconomic performance and changes in policy
efficiency. For this purpose we assume that the monetary authority follows a linear policy
rule which can be expressed as a function of aggregate demand and aggregate supply shocks.
First, we construct a theoretical framework to derive the optimal and actual policy rules as
a function of the structural parameters of the economy and the aggregate disturbances; this
information also allows us to identify the relative preferences towards minimizing inflation
and output variability. We then introduce the measures of macroeconomic performance and
monetary policy efficiency, and finally, we focus on the procedure for estimating the relevant
parameters of the economy and the aggregate supply and demand shocks - for two different
periods - in order estimate the suggested measures.
3.1
The Role of Policy in Stabilization
We assume that the primary concern of monetary policy is to achieve stabilization of the
economy through the reduction in the variability of inflation and output growth over the
medium term. In doing this we abstract from other policy goals, such as stabilizing exchange
rates and interest rates, for we consider that these serve rather as intermediate goals towards
achieving domestic macroeconomic performance, measured by price and output stability. We
summarize the central bank’s objective through a standard quadratic loss function used in
most contemporary analyses of central bank policy:
L = Et [λ(π t − π Tt )2 + (1 − λ)(yt − ytT )2 ] ; 0 ≤ λ ≤ 1 ,
5
(1)
where Et is the expectation operator at time t; π is inflation; y is (log) aggregate output;
π T and y T are the target levels of inflation and output; and λ is the relative weight given to
squared deviations of inflation and output from their desired levels.5
Minimization of this loss requires knowledge of the determinants of deviations of inflation
and output from their respective targets. We assume that two random shocks push y and
π away from y T and π T . First, an aggregate demand shock (d) moves inflation and output
in the same direction, while an aggregate supply shock (s) moves inflation and output in
opposite directions. Since policy is only capable of moving inflation and output in the same
direction its effect is analogous to that of an aggregate demand shock.
We define aggregate demand (AD) as the negative relationship between (y − y T ) and
(π − π T ) that is shifted by the demand shock and the deviations of the policy instrument
from its equilibrium value (e
r):6
y − y T = −ω(π − π T ) − φ(e
r − d) ; ω > 0 , φ > 0 ,
(2)
where ω is the inverse of the slope of the aggregate demand function and −φ is the real
interest rate semi-elasticity.7 Analogously, aggregate supply (AS) is the positive relationship
between inflation deviations and output deviations that is shifted by the supply shock:
π − π T = γ(y − y T ) − s ; γ > 0 ,
(3)
where γ is the slope of the aggregate supply function. The aggregate disturbances d and s
have been normalized to yield the simple representation of the AD-AS model.
Combining (2) and (3) we obtain expressions for (y − y T ) and (π − π T ) as a function of
5
λ can be also interpreted as the policy maker’s aversion to inflation variability (see Cecchetti and
Ehrmann, 2001).
6
The equilibrium value of the interest rate is defined as the value needed such that output would equal
its potential (or target) level.
7
Romer (2000) provides a good description on how to derive an analogous version of this model. See also
Krause (2003) for a theoretical derivation using a rational expectations optimization process in the presence
of imperfect information.
6
the structural parameters, the aggregate shocks and the policy instrument:
y − yT =
−φ(e
r − d) + ωs
,
(1 + ωγ)
(4)
π − πT =
−φγ(e
r − d) − s
.
(1 + ωγ)
(5)
Minimizing the quadratic loss function, subject to the constraint imposed by the structure
of the economy, yields a simple linear policy rule of the form:
re = ad + bs ,
(6)
where, the optimal values for the coefficients a and b (whose derivation is presented in
Appendix 1) are given by:
a∗ = 1 ,
b∗ =
(7)
−λγ + (1 − λ)ω
.
φ[λγ 2 + (1 − λ)]
(8)
Thus, an optimal policy rule has two parts: first the authorities completely neutralize the
effect of demand shocks, and second they accommodate supply shocks depending on the
structural parameters (ω, γ, φ) and their preferences (λ).8
The optimal policy rule gives us a benchmark criterion for evaluating policy efficiency;
the closer the actual policy rule is to the optimal one in a given country, the higher the
degree of policy efficiency. Hence, we are interested in deriving the coefficients a and b of
the actual policy rule. While, the actual estimate of the parameter a will give us an idea of
how efficient is policy in neutralizing the effect of demand shocks, the estimate for b provides
us with useful information to derive estimates for the preference parameter λ, as we explain
8
This formulation of the policy rule as a function of the demand and supply shocks differs from the one
proposed in the Taylor (1993)-type rules, where the policy instrument is a function of the observable variables
in the economy. In the following Section 3.3 we will establish how to estimate the shocks, in order to make
the rule operational.
7
below.
The estimation procedure is as follows: Starting from the reduced form representation of
the economy, given by equations (4) and (5), we substitute the linear policy rule of equation
(6). By construction, we can define the aggregate supply and demand shocks in such a
way that they will be uncorrelated (σ d,s = 0). Hence, the observed variances of output and
inflation around their target levels can be given by the following expressions:
V ar(y) ≡ E(yt − ytT )2 = (1 + ωγ)−2 [φ2 (a − 1)2 σ 2d + (ω − φb)2 σ 2s ] ,
V ar(π) ≡ E(π t − π Tt )2 = (1 + ωγ)−2 [γ 2 φ2 (a − 1)2 σ 2d + (1 + γφb)2 σ 2s ] .
(9)
(10)
Combining equations (9) and (10) we can solve for the parameter b of the actual policy rule:
b=
(1 + ωγ)(σ 2π − γ 2 σ 2y ) − (1 − ωγ)σ 2s
.
2γφσ 2s
(11)
Given the estimate for b and equation (9) we obtain the squared deviation of a from its
optimal value, i.e.:
(a − 1)2 =
(1 + ωγ)2 σ 2y − (ω − φb)2 σ 2s
φ2 σ 2d
,
(12)
while the coefficient of aversion to inflation variability can be derived by combining equations
(8) and (11):
λ=
(ω − φb)
.
(ω − φb) + γ(1 + γφb)
(13)
The policy maker’s preferences will depend on the reaction of policy to supply shocks; if the
monetary authority only cares about reducing inflation variability (λ = 1), it will completely
offset the effect of supply shocks on inflation (since it implies that 1 + γφb = 0), and
conversely, for the case in which the only goal is output stability (λ = 0), the effects of
supply shocks on output growth will be neutralized (ω − φb = 0). The exercise of obtaining
the preferences through the use of the knowledge of the structure of the economy and the
policy rule is consistent with the discussion provided by Favero and Rovelli (2000); namely
8
that, given a policy environment, it is possible to reverse engineer the preferences of the
policy maker.
We note that there are several studies that estimate central bank preferences; to name
a few, Cecchetti and Ehrmann (2001) and Cecchetti, McConnell and Pérez-Quirós (2002)
measure λ for countries in the European Union, while Dennis (2001), Rudebusch (2001),
Favero and Rovelli (2003), and Castelnuovo and Surico (2004) do it for the case of the
United States. The main advantage of our procedure, as we show above, is that we do not
need to assume optimal policy behavior in order to estimate these preference parameters.
The policy rule parameters and the coefficient of aversion to inflation variability can be
used in the derivation of the measures of macroeconomic performance and policy efficiency.
We turn to this issue next.
3.2
Proposed Measures for Performance and Efficiency Changes9
In this subsection we define the theoretical measures of changes in performance and changes
in policy efficiency that we shall use in our empirical computations. To compute macroeconomic performance, we make use of the loss function in equation (1) in order to construct
a single measure of increased stability. The performance measure will then be the weighted
sum of the observed variances of inflation and output, given for each period i (= 1, 2) by:
Pi = λi V ar(π i ) + (1 − λi )V ar(yi ) ,
(14)
where the weights λ and 1−λ are derived using equation (13). The change in macroeconomic
performance is just the change in the measure from one period to the next, ∆P = P1 − P2 .
If ∆P is positive we interpret this as a performance gain.10
9
In this subsection we follow, to some extent, the discussion provided by Cecchetti, Flores-Lagunes and
Krause (2004).
10
Note that in computing ∆P we allow for changes in the preferences as well as changes in the variances. In
Section 4, however, we look at the estimates of the measures by considering both the changes in preferences
and by assuming that these preferences remain unchanged from one period to the next.
9
We evaluate monetary policy efficiency by looking at how close the actual performance
is to the performance achievable under optimal policy. Since optimal policy implies that the
effect of demand shocks on the economy is completely neutralized we are interested in the
following term
µi = (ai − 1)2 σ 2d ; i = 1, 2 .
(15)
A smaller value for µi will be associated with a more successful monetary policy. Let us define
V ar(yi |si = 0) and V ar(π i |si = 0) as the variability of output and inflation conditional on
no supply shocks in period i. Using equations (9) and (10) and the definition of µi above we
can define these variances as:
V ar(yi |si = 0) = (1 + ωγ)−2 φ2 µi ,
(16)
V ar(π i |si = 0) = (1 + ωγ)−2 γ 2 φ2 µi .
(17)
As a result, a measure for policy inefficiency that is comparable in units to the performance
measure in equation (14) is:
Ii = λi V ar(πi |si = 0) + (1 − λi )V ar(yi |si = 0) ,
Ii = (1 + ωγ)−2 (1 − λi + λi γ 2 )φ2 µi .
(18)
Since Ii will be smaller the closer actual outcomes are to the optimal, i.e., the closer the
actual policy rule is to the optimal one, our measure of the change in policy efficiency follows
immediately as the difference ∆I = I1 − I2 . We interpret positive values of ∆I as increases
in the efficiency of monetary policy. When ∆I is negative, it suggests that policy making
has deteriorated as the variances of inflation and output have moved further away from their
optimal values.
Finally, we calculate the proportion that can be accounted for by improved policy using
10
the following ratio:
Q=
∆I
.
|∆P |
(19)
Given that the absolute value of the performance gain is in the denominator, a positive
value of Q implies improved policy efficiency, whereas a negative Q implies that policy has
become less efficient. If we observe a macro performance gain at the same time as policy has
become more efficient and the variance of the aggregate shocks has become smaller, Q will
be between 0 and 1 and can be interpreted as the relative contribution of a more efficient
policy towards the achievement of a macro performance gain.
Implementing the procedure we have just described requires us to follow several steps.
First we must construct and estimate a dynamic model of inflation and output for both
periods of interest. The dynamic model is used to identify the structural parameters, the
aggregate supply and demand disturbances and their respective variances, which allow us to
derive the monetary policy rule and the optimal variances. These, in turn, will enable us to
compute ∆P , ∆I and Q.
3.3
Estimating the Structural Parameters and Aggregate Shocks
There are several methods available to estimate aggregate supply and demand disturbances.
One approach is to treat demand shocks as those that have transitory effects on output,
and identify supply shocks as those having permanent effects on both output and prices
(Blanchard and Quah, 1989). An alternative method consists in using orthogonal crosscorrelation functions of inflation and output (Uhlig, 1999, and Canova and DeNicoló, 2000).
In this paper, we choose to identify the disturbances directly through their effect on output
and inflation, with the use of the structural estimates.
11
Let us consider again the stylized model in equations (2) and (3):
π t − φ(e
rt − dt ) ,
yet = −ωe
yt − st ,
π
et = γe
(2’)
(3’)
e = π − π T for notational simplicity. As such,
where we have defined ye = y − y T and π
estimating the system only allows us to identify the parameter γ. Hence, in order to achieve
the identification of ω and φ we assume that the aggregate supply shock can be decomposed
into a domestic and a foreign component, namely:
st = ht − ψft ,
(20)
where h represents the domestic (home) component of the shock, while f represents the
foreign disturbance. The underlying assumption is that f affects domestic prices directly,
while its impact on output arises indirectly through its effect on inflation. To be consistent
with this description, we will use external price inflation as a proxy for f in the estimation,
as we detail below.
The stylized model in equations (2’)-(3’) can be reformulated to take into account the
dynamic behavior of the economy, a feature present in the data. To accomplish this, we
assume that the demand disturbance and the domestic component of the supply disturbance
have persistent effects on the economy and model dt and ht as AR(2) processes; i.e.:11
dt = kd,t + ϕ1 dt−1 + ϕ2 dt−2 ; Et (kd,t ) = 0 ,
(21)
ht = kh,t + χ1 ht−1 + χ2 ht−2 ; Et (kh,t ) = 0 .
(22)
11
The assumption about the autoregressive structure of the shocks is only crucial in terms of determining
the order of the Vector Autoregression in equations (29) and (30). Specifically, for the current specification
of the AD-AS model, an AR(n) process for the disturbances will result in the estimation of a n-order VAR.
12
Using equations (2’), (3’) and (20) we can represent the aggregate shocks as:
πt
yet + ωe
+ ret ,
φ
= st − ψfet = γe
yt − π
et − ψft .
dt =
(23)
ht
(24)
Substituting (23) into the right-hand side of (21) and the solution into (2’) yields:
π t − φ(e
rt + ϕ1 ret−1 + ϕ2 ret−2 ) + ϕ1 yet−1 + ϕ2 yet−2 + ωϕ1 π
et−1 + ωϕ2 π
et−2 + φkd,t . (25)
yet = −ωe
Analogously, substituting (24) into the right-hand side of (22) and the solution into (3’)
results in the following:
yt + ψ(ft + χ1 ft−1 + χ2 ft−2 ) − γχ1 yet−1 − γχ2 yet−2 + χ1 π
et−1 + χ2 π
et−2 − kh,t .
π
et = γe
(26)
The system of equations (25) and (26) represents a dynamic aggregate demand-aggregate
supply model. To make its estimation operational, we proxy the term ret + ϕ1 ret−1 + ϕ2 ret−2
with the lagged demeaned ex-post real interest rate (eιt−1 − π
et−1 ), where i is the short-term
nominal interest rate; while the proxy for the expression ft + χ1 ft−1 + χ2 ft−2 is one lag of
ext−1 ), where e is exchange rate devaluation and π x
demeaned external price inflation (e
et−1 + π
is foreign inflation. Taking this into account, we estimate the dynamic behavior of output
and inflation through the following:
π t − φ(eιt−1 − π
et−1 ) +
yet = −ωe
yt + ψ(e
et−1 +
π
et = γe
π
ext−1 )
+
2
X
l=1
2
X
l=1
α1l yet−l +
α2l yet−l +
2
X
l=1
2
X
l=1
α1(l+2) π
et−l + α15 Zt−1 + uyt ,
α2(l+2) π
et−l + α25 Zt−1 + uπt ,
(27)
(28)
where Z represents exogenous variables used as controls (see Appendix 2 for a detailed
description); and the other variables are defined as above. There are two crucial assumptions
for the estimation of the system. In the aggregate demand equation, the (lagged) real interest
13
rate has only a direct effect on output, and its outcome on prices arises through the effect
on output. Analogously, (lagged) external price inflation only affects domestic inflation
contemporaneously. While the first identification assumption is often found in the literature
(see, for example, Rudebusch and Svensson (1999) and the references in Taylor (2000)), the
use of the second one can be justified if, as mentioned above, a change in external prices
has its direct effect on domestic inflation immediately and only causes a change in domestic
output after a period.12 It is important to note that we are neither claiming that external
price inflation does not affect output, nor that a change in the real interest rate has no effect
on inflation. Instead, our assumptions are the interest rate affects first production and then
prices, and that external inflation affects domestic prices first and the adjustment in output
takes place later, both of which are consistent with recent empirical findings.
Since the real interest rate only enters the dynamic aggregate demand equation (27)
and the external price inflation only enters the dynamic aggregate supply equation (28)
we can identify the parameters ω, γ and φ through the estimation of the following vector
autoregression:13
yet = β 1r (eιt−1 − π
et−1 ) + β 1f (e
et−1 + π
ext−1 ) +
+β 15 Zt−1 +
yt
,
et−1 ) + β 2f (e
et−1 + π
ext−1 ) +
π
et = β 2r (eιt−1 − π
+β 25 Zt−1 +
πt
.
2
X
l=1
2
X
l=1
β 1l yet−l +
β 2l yet−l +
2
X
l=1
2
X
l=1
β 1(l+2) π
et−l
(29)
β 2(l+2) π
et−l
(30)
It is straightforward to show that the estimates for ω, γ and φ can be obtained from the
12
This would clearly be the case if the source of the external price change were an oil shock or a modification
in the terms of trade that the economy faces.
13
The VAR specification is similar in spirit to the one proposed by Mojon and Peersman (2001) for
measuring the effects of monetary policy in countries of the Euro Area. The most important differences are
that we do not include US real GDP and nominal interest rate as controls and that we estimate the same
model for Austria, Belgium and the Netherlands as for the other countries (excluding Germany).
14
VAR estimation as follows:
ω
b = −
γ
b =
β 1f
,
β 2f
β 2r
,
β 1r
b = −β 1r (1 + ω
φ
bγ
b) .
(31)
(32)
(33)
Finally, in order to compute the policy rule parameters in equations (11) and (12), we
need to obtain estimates of the variances of the aggregate shocks. We estimate the shocks
using equation (23) and combining equations (20) and (24) as follows:
bπ
et
yet + ω
+ ret ,
dbt =
b
φ
γ yet − π
et ,
sbt = b
(34)
(35)
where the variables have been normalized so that E(st , dt ) = 0. Given the estimates for the
aggregate shocks and the structural parameters - in addition to the data on inflation and
output variances - we can estimate the policy rule parameters in equations (11) and (12). In
the next section we present these results and proceed with the discussion of the performance
and policy efficiency measures.
4
Results
This section is organized into three parts. First, we present our estimates of the variances of
the aggregate shocks and the actual policy rule coefficients, and briefly discuss our findings.
We then compare the behavior of the actual policy rule, given by the path of the nominal
interest rate, with the optimal rule for the countries of interest. Finally, we report the results
for the measures of macroeconomic performance and policy efficiency, and the estimates for
the coefficient of aversion to inflation variability, and describe the role monetary policy and
15
aggregate shocks have played in accounting for the changes in macroeconomic performance.
4.1
Variability of the Aggregate Shocks and Policy Rule Parameters
We begin by estimating the model (29)-(30) for the periods 1983:I-1990:IV and 1991:I1998:IV for the 14 countries of our sample in order to obtain the parameters ω, γ and φ
for each period.14 For both the model specification and the derivation of the measures of
performance and efficiency change we assume that policy makers are interested in achieving
an inflation target of 2% and minimize the variability of output around its potential level,
and we measure potential output by applying the Hodrick-Prescott filter to industrial production. We note that, while the 2% target level for inflation can be viewed as a sensible
policy goal during the 1990s, it is less clear that this was the objective pursued by some
countries of the EU area during the 1980s (Greece, Portugal, Italy and Spain, for example).
However, we adopt the measure of inflation variability using this target level, since we believe a reduction in both average inflation and its variance, for a given variability of output,
should be associated with an improved macroeconomic outcome. Finally, 2% inflation is the
current upper bound in the European Monetary Union and it allows for a comparison across
countries.
Table 1 presents the estimates of the standard deviations of the aggregate shocks and
the policy rule parameters. Looking at the first two columns, we observe evidence of large
cross-country differences in the variability of demand and supply perturbations. Demand
shocks in Greece are nearly 10 times as large as they are in the Netherlands for both periods,
and are also sizable in Italy, Spain and Portugal, mainly during the 1980s. Similar differences
are present in the variability of supply shocks. We can also observe a general tendency for
supply and demand shocks to become less volatile during the 1990s; only Belgium, Finland
14
The estimates of the structural parameters and the description of the variables used and data sources
are presented in Appendix 2.
16
and the Netherlands experienced an increase in the variability of demand shocks, while only
Austria, Ireland and Sweden were exposed to a higher volatility of supply shocks.
As described in Section 2, all countries (with the exception of Germany) experienced
lower and more stable inflation in the 1990s, while output variability was higher in some and
lower in others. A reduction in the variability of aggregate shocks, accompanied by a shift in
preferences towards inflation stability, can give rise to this outcome. Still, we are interested
in determining whether or not improved policy making played a role in the documented
outcome, and so we proceed to this task in the next two subsections.
4.2
Comparing Actual and Optimal Policy Rules
As we mentioned in Section 3, a more efficient policy will be characterized by a policy rule
that closely resembles the optimal rule. Given our estimates for the actual and optimal rule
parameters and the estimates of the aggregate demand and supply shocks, we can compare
the behavior of the actual and optimal policy rule for the countries in our sample, which we
present in Figure 3.
The comparison between the actual and optimal policy rules gives rise to some important
observations. First, for most countries the optimal rule suggests that the actual policy followed by the central banks was tighter than needed for the early and mid 1980s, while higher
interest rates would have been advisable during the late 1980s and early 1990s. Second, the
optimal rule predicts well the interest rate spikes during the crisis of the European Monetary
System during late 1992 and early 1993 for Greece, Ireland and the UK, but fails to do so
for Denmark, Portugal and Spain.
Another key observation that arises from Figure 3 is that, in general, the actual policy
rule has come closer to the optimal rule after 1993. This is particularly the case for Austria,
Belgium, France, Ireland, Italy, Portugal, Spain and the UK. Consistent with our discussion
in Section 3, this fact yields evidence that monetary policy has indeed become more efficient
during the 1990s.
17
Still, a more detailed examination is needed to establish the role of monetary policy in
the observed performance changes, and we proceed with this analysis in the next subsection.
4.3
Accounting for Changes in Macroeconomic Performance
We begin by examining the changes in performance, and then we move to report the proportion of the change that can be accounted for by improvements in policy making. As
noted above, we estimate the dynamic AD-AS models and the measures of macroeconomic
performance and policy efficiency for the periods 1983:I-1990:IV and 1991:I-1998:IV. Table
2 reports our estimates of the inflation variability aversion coefficients, the standard deviations of inflation and output and the value of the loss function, Pi , for the 14 countries in
our sample. We also report the percentage change in P between the two periods for each of
the countries; using the comprehensive measure of performance, only Germany and Sweden
exhibited a decline in performance while the other 12 countries experienced improvements,
which ranged from 39% for Austria to 96% for Denmark.
As we noted in Section 3, the performance change measure involves both changes in the
variances of inflation and output, and changes in the policy maker’s preferences. As we can
observe in column 1, the coefficient of aversion to inflation variability has either increased or
decreased only slightly (in the case of Germany). In particular, for Portugal, Spain, Italy,
France, Sweden and Finland, this shift in preferences has been rather sharp, consistent with
the general decrease in inflation variability observed in the 1990s.
It is also useful to analyze how the measure of performance would be affected if the
preferences remained unaltered in both periods. Assuming that for the entire period the
weight given to inflation stability is the same as in the second period, the magnitudes of the
performance change remain mostly unaltered; exceptions to this are Germany and Sweden,
for which the performance loss becomes smaller; Austria, for which the performance gain
becomes close to zero, and Spain and Portugal, who exhibit a performance gain nearly twice
as large. However, if one assumes that the preferences for the first period are the ones that
18
prevail during the entire period, the results change dramatically: we would observe large
estimated performance losses for Sweden and Finland and moderate ones for Austria, Spain
and Ireland.15
Another robustness check is to fix the value of the preferences for inflation and output
stability, and analyze how the performance change for each country varies over a range
for values of λ. For all countries we consider a range for λ between 0.7 and 0.9, which is
consistent with the results in this paper, as well as in the studies of Cecchetti and Ehrmann
(2001) and Cecchetti, Flores-Lagunes and Krause (2004). The only exception is Greece, for
which we consider a range between 0.1 and 0.3 for λ, given that this country is not analyzed
in either of the two cited studies and that we find a value for λ to 0.1 for both periods in
the present research. We report these results in Table 3.
Contemplating these ranges for the central bank’s preferences, the results suggest that in
almost all of the countries the performance change remains relatively unaffected. The only
two exceptions are Austria and Finland, which exhibit a performance gain for certain values
for λ over the range, while for others we find a performance loss.
Table 4 reports the estimates of the performance change ∆P , improved policy efficiency,
∆I, and the proportion of performance change that is due to a change in the efficiency of
policy, Q. We see that all countries - except for Germany - experienced improvements in
the efficiency of monetary policy. These observed improvements for most countries are likely
linked to the desire to meet the qualification requirements for the European Monetary Union,
while the decline observed in Germany is almost surely the consequence of both unification
and the necessary adjustment prior to its entry into the EMU. We note, however, that the
deterioration in German performance and policy efficiency was fairly modest over this period.
We can also observe that in the case of Austria, policy efficiency gain was higher than the
macro performance improvement (which is due to the fact that monetary policy was also able
to offset the effect of an increase in the variance of supply shocks), while in Sweden, more
15
The results of exercise are not reported in the present document, but are available upon request.
19
efficient policy was present despite the measured macro performance loss. Looking at the
final column, for the remaining eleven countries more efficient policy accounted for between
41% (Greece) and 91% (Ireland and Spain) of the improvement in overall macroeconomic
performance.
Once again, we fix the value of the preferences for inflation and output stability for both
sub-periods and study what happens to our measures of policy efficiency and performance
change over a range for plausible values of λ. We can observe that, while the range of
the measures of efficiency change are relatively narrow and centered at the value of their
estimate, reported in Table 5, the range of values for the performance change measure is
large in some of the countries, in particular Finland, Greece, Ireland, Italy, Portugal and
Spain. Despite this fact, the ranges of the contribution of policy to the performance gain
still point to an important role of monetary policy in contributing to a macroeconomic
performance improvement in most of the European Union countries, as depicted in the last
column.
Summarizing, our main findings suggest that for Greece, Belgium, France, Italy, Denmark, Finland and the Netherlands both a reduction in the variance of supply shocks and
more efficient policy were roughly equally responsible for the macro performance improvement. For Portugal and the UK, the policy contribution accounted for approximately 3/4 of
the performance improvement, and finally, for Austria, Ireland and Spain more efficient policy played a far more important role than the reduced variability of shocks in macroeconomic
stabilization.
Finally, we are also interested in verifying how robust are these results to specifying
alternative targets for inflation and output.16 The rationale for this exercise is that the
2% inflation target is not considered a realistic goal for some countries during the 1980s.
Therefore, we examine each country’s average inflation for the 1980s and for the 1990s as
the policy target for inflation. Similarly, economic growth has been different in both decades,
16
We thank the referees and the editor for suggesting this analysis.
20
so we apply the Hodrick-Prescott filter separately for each subperiod, and use the trend as
our measure for the target for output.
We display the results of these alternative measures in Table 6. As expected, the standard deviations of both inflation and output are lower than the ones reported on Table 2.
These lower variances naturally result in smaller estimates for the volatility of both supply
and demand shocks. Looking at column 5, we also observe a decrease in the measure of
performance gain for all countries (except Austria, whose performance gain is slightly higher
than the estimate in Table 2). For half of the countries the change is modest; however,
for Germany and Sweden, the estimates suggest an even worse performance loss, while for
countries that had a relatively high average inflation rate during the 1980s (Greece, Italy,
Portugal, Spain and the U.K.), the measured performance gain goes down more sharply, but
still remains in a positive range between 26%-41%.
Turning our attention to the policy contribution, the results are quite consistent with the
ones obtained under the 2% inflation-target assumption. While for two countries, Denmark
and the Netherlands, improved policy now contributes less than 50% of the performance gain,
for France, Greece and Ireland policy efficiency now explains over 50% of the macroeconomic
improvement. The effects for the remaining countries remains largely unaltered, except for
Germany, for which the policy efficiency loss becomes more of a factor under the alternative
factors. We believe these findings provide some further validity of our main conclusions.
5
Conclusions
This paper proposes a general method for analyzing changes in macroeconomic performance
and identifying the relative contributions of improvements in the efficiency of monetary policy
and changes in the variability of aggregate shocks. We apply this technique to 14 European
Union countries in order to compare their macroeconomic performance in the 1980s with
that in the 1990s.
21
We find that, in general, the actual policy followed by the monetary authorities has come
closer to the optimal rule during the 1990s. Our results suggest that in all countries but
Germany monetary policy became more efficient in the last decade, as compared with the
1980s. The evidence regarding the importance of reduced variability in the aggregate shocks
as a source of performance improvement is somewhat mixed: for seven countries (Greece,
Belgium, France, Italy, Denmark and the Netherlands) it was nearly as important as more
efficient policy in stabilization, while for the remaining countries the change in the volatility
of the shocks either played a minor role (in the cases of Portugal, the UK, Spain, Ireland and
Austria) or contributed towards a deterioration of macroeconomic performance (for Sweden
and Germany).
The paper leaves some open questions relating to the reasons behind this improved monetary policy performance and the cross-country differences in the efficiency of policy. We
expect to address these issues in future research.
6
Appendix 1: Optimal Policy Rule Parameters
In order to derive the parameters a∗ and b∗ of the optimal policy rule, let us substitute
the expressions for the deviations of output and inflation from their respective target levels
(equations (4) and (5)) into the loss function (equation (1)), i.e.:
r − d) − s]2 + (1 − λ)[−φ(e
r − d) + ωs]2 } .
L = (1 + ωγ)−2 {λ[−φγ(e
(A1)
Minimizing (A1) with respect to the policy instrument, re, yields the following first-order
condition:
r − d) − s] + 2(1 − λ)φ[−φ(e
r − d) + ωs]} = 0 .
−(1 + ωγ)−2 {2λφγ[−φγ(e
22
(A2)
Simplifying the expression and arranging terms:
r = φ[λγ 2 + (1 − λ)]d + [−λγ + (1 − λ)ω]s .
φ[λγ 2 + (1 − λ)]e
(A3)
Solving for the optimal policy rule re∗ in terms of the aggregate demand and aggregate
supply shocks yields equation (6), with the parameters a∗ and b∗ given by:
a∗ = 1 ,
b∗ =
7
(7)
−λγ + (1 − λ)ω
.
φ[λγ 2 + (1 − λ)]
(8)
Appendix 2: Data sources and Model Specification
Data for Austria, Belgium, Denmark, Finland, France, Germany, Ireland, Italy, Netherlands,
Portugal, Spain, Sweden and the United Kingdom are from Datastream. For all countries,
the quarterly data is from 1983:I-1998:IV. Output (y) is given by seasonally adjusted industrial production. Inflation (π) is given by the annualized CPI inflation rate. The nominal
interest rate (i) is given by the 3-month money market rate except for Belgium, Ireland and
Sweden (3-month treasury bill rate), France (3-month call rate), Greece (overnight-interbank
rate) and Portugal (5-day money market rate). Devaluation (e) is given by the annualized
percentage change of the exchange rate to the Deutsch-Mark, except for Germany (DM/US$
exchange rate). External inflation (π x ) is given by annualized German CPI inflation, except
for Germany (annualized US CPI inflation). In all countries’ model specifications we also
included the IMF commodity price index and in all except for Germany we included German
output and interest rate.
Estimating the model (29)-(30) for the periods 1983:I-1990:IV and 1991:I-1998:IV for
the 14 countries of our sample yields the parameters ω, γ and φ for each period, which we
present in Table A1.
23
References
[1] Alesina, A. and L. H. Summers (1993). “Central Bank Independence and Macroeconomic Performance: Some Comparative Evidence”, Journal of Money, Credit and Banking, 25 (2), pp. 151-162.
[2] Blanchard, O. and D. Quah (1989). “The Dynamic Effects of Aggregate Demand and
Supply Disturbances”. American Economic Review, 79 (4), pp. 655-673.
[3] Canova, F. and G. DeNicoló (2000). “Monetary Disturbances Matter for Business Fluctuations in the G7”. International Finance Discussion Papers, Board of Governors of
the Federal Reserve System, No. 660.
[4] Castelnuovo, E. and P. Surico (2004). “Model Uncertainty, Optimal Monetary Policy
and the Preferences of the Fed”, Scottish Journal of Political Economy, 51, 105-126.
[5] Cecchetti, S. G. and M. Ehrmann (2001). “Does Inflation Targeting Increase Output
Volatility? An International Comparison of Policymakers’ Preferences and Outcomes”,
in K. Schmidt-Hebbel (ed.), Monetary Policy: Rules and Transmission Mechanisms,
Proceedings of the Third Annual Conference of the Central Bank of Chile.
[6] Cecchetti, S. G., A. Flores-Lagunes and S. Krause (2004). “Has Monetary Policy Become
More Efficient? A Cross-Country Analysis”, mimeo.
[7] Cecchetti, S. G. and S. Krause (2002). “Central Bank Structure, Policy Efficiency and
Macroeconomic Performance: Exploring Empirical Relationships”, Federal Reserve of
St. Louis Review, July/August 2002.
[8] Cecchetti, S. G., M. M. McConnell and G. Pérez-Quirós (2002). “Policymakers’ Revealed Preferences and the Output-Inflation Variability Trade-Off: Implications for the
European System of Central Banks”, The Manchester School, 70 (4), pp. 596-618.
24
[9] Chortareas, G., D. Stasaavage and G. Sterne (2002). “Does it Pay to be Transparent?
International Evidence from Central Banks”, Economic Review of the Federal Reserve
Bank of St. Louis, 79, July/August 2002, pp. 99-118.
[10] Cukierman, A., Webb, S. B. and B. Neyapti (1992). “Measuring the Independence of
Central Bank and its Effect on Policy Outcomes”, World Bank Economic Review, 6,
pp. 353-398.
[11] Dennis, R. (2001). “The Policy Preferences of the US Federal Reserve”, Working Papers in Applied Economic Theory, No. 2001-08, Federal Reserve Bank of San Francisco
(revised, June 2003).
[12] Faust J. and L. E. O. Svensson (2001). “Transparency and Credibility: Monetary Policy
with Unobservable Goals”, International Economic Review, 42 (2), pp. 369-397.
[13] Favero, C. and R. Rovelli (2000). “Modeling and Identifying Central Bank Preferences”,
Temi di Ricerca-Ente per gli studi monetari, bancari e finanziari Luigi Einaudi, No.19,
July 2000.
[14] Favero, C. and R. Rovelli (2003). “Macroeconomic Stability and the Preferences of the
Fed: A Formal Analysis 1961-1998”, Journal of Money, Credit, and Banking 35, pp.
545-556.
[15] Fry, M., D. A. Julius, L. Mahadeva, S. Rogers, and G. Sterne (2000). “Key Issues in the
Choice of Monetary Policy Framework”, in L. Mahadeva and G. Sterne (eds.), Monetary
Policy Framework in a Global Context, Routledge, London, United Kingdom.
[16] Grilli, V., Masciandaro, D. and G. Tabellini (1991). “Political and Monetary Institutions
and Public Financial Policies in the Industrial Countries”, Economic Policy, 13, pp. 341392.
25
[17] Kahn, J., M. M. McConnell and G. Pérez-Quirós (2002). “On the Causes of the Increased
Stability of the U.S. Economy” Economic Policy Review, 8 (May 2002), pp. 183-202.
[18] Krause, S. (2003). “Why Should Policy-makers Care About Inflation and Output Variability? The Role of Monetary Policy Stabilization under Asymmetric Information”,
Working Paper 03-13, Department of Economics, Emory University.
[19] Mojon, B. and G. Peersman (2001). “A VAR Description of the Effects of the Monetary
Policy in the Countries of the Euro Area”, ECB Working Paper No. 92.
[20] Romer, D. (2000). “Keynesian Macroeconomics Without the LM Curve”. Journal of
Economics Perspectives, 14 (2), pp. 149-169.
[21] Rudebusch, G. D. and L. E. O. Svensson (1999) “Policy Rules for Inflation Targeting”
in J. B. Taylor (ed.), Monetary Policy Rules, University of Chicago Press, Chicago, pp.
203-246.
[22] Rudebusch, G. D. (2001) “Is the Fed Too Timid? Monetary Policy in an Uncertain
World”, Review of Economics and Statistics, 83 (2), pp. 203-217.
[23] Taylor, J. B. (1979). “Estimation and Control of a Macroeconomic Model with Rational
Expectations”, Econometrica, 47, pp. 1267-1286.
[24] Taylor, J. B. (1993). “Discretion Versus Policy Rules in Practice”, Carnegie-Rochester
Conference Series on Public Policy, 39, pp. 195-214.
[25] Taylor, J. B. (2000). “Alternative Views of the Monetary Transmission Mechanism:
What Difference Do They Make for Monetary Policy?”, Oxford Review of Economic
Policy, 16 (4), pp. 60-73.
[26] Uhlig, H. (1999). “What Are the Effects of Monetary Policy: Results from an Agnostic
Identification Approach”. Tilburg University, CentER working paper 9928
.
26
Figure 1: Inflation and output variability
(1983-1990 and 1991-1998)
0.010
1983-1990
1991-1998
Output Variability
0.008
0.006
0.004
0.002
0.000
0.000
0.005
0.010
0.015
0.020
0.025
0.030
Inflation Variability
Figure 2: Change in inflation and output variability
(1983-1990 and 1991-1998)
0.005
0.0072
0.0071
0.004
0.002
0.001
0.000
-0.003
-0.004
Inflation variability
Output variability
-0.0171
-0.005
27
-0.0245
UK
-0.002
Sweden
Spain
Portugal
Nether.
Italy
Ireland
Greece
Germany
France
Finland
Denmark
Belgium
-0.001
Austria
Change in Variability
0.003
Figure 3: Actual and Optimal Policy Rule (1983:I-1998:IV)
Austria: Actual vs. Optimal Interest Rate
Belgium: Actual vs. Optimal Interest Rate
16
18
14
16
12
14
12
10
10
8
8
6
6
4
4
2
2
0
1983
0
1985
1987
1989
1991
Nominal Interest Rate
1993
1995
1983
1997
Optimal Interest Rate
1983
1985
1987
1989
1991
1993
1991
1993
1995
1997
Optimal Interest Rate
Finland: Actual vs. Optimal Interest Rate
1995
1997
1983
Optimal Interest Rate
1985
1987
1989
1991
Nominal Interest Rate
France: Actual vs. Optimal Interest Rate
1993
1995
1997
Optimal Interest Rate
Germany: Actual vs. Optimal Interest Rate
14
20
18
16
14
12
10
8
6
4
2
0
12
10
8
6
4
2
0
1985
1987
1989
1991
Nominal Interest Rate
1993
1995
1997
1983
Optimal Interest Rate
1985
1987
1989
1991
Nominal Interest Rate
Greece: Actual vs. Optimal Interest Rate
1993
1995
1997
Optimal Interest Rate
Ireland: Actual vs. Optimal Interest Rate
30
50
45
40
35
30
25
20
15
10
5
0
1983
1989
20
18
16
14
12
10
8
6
4
2
0
Nominal Interest Rate
1983
1987
Nominal Interest Rate
Denmark: Actual vs. Optimal Interest Rate
20
18
16
14
12
10
8
6
4
2
0
1985
25
20
15
10
5
0
1985
1987
1989
Nominal Interest Rate
1991
1993
1995
1983
1997
Optimal Interest Rate
1985
1987
1989
Nominal Interest Rate
28
1991
1993
1995
1997
Optimal Interest Rate
Figure 3: Actual and Optimal Policy Rule (1983:I-1998:IV)
Italy: Actual vs. Optimal Interest Rate
Netherlands: Actual vs. Optimal Interest Rate
30
12
25
10
20
8
15
6
10
4
5
2
0
1983
0
1985
1987
1989
1991
Nominal Interest Rate
1993
1995
1997
1983
Optimal Interest Rate
Portugal: Actual vs. Optimal Interest Rate
30
25
1991
1993
1995
1997
Optimal Interest Rate
20
20
15
15
10
10
5
5
0
0
1985
1987
1989
1991
Nominal Interest Rate
1993
1995
1997
1983
Optimal Interest Rate
1985
1987
1989
1991
Nominal Interest Rate
Sweden: Actual vs. Optimal Interest Rate
1993
1995
1997
Optimal Interest Rate
U.K.: Actual vs. Optimal Interest Rate
18
30
16
25
14
12
20
10
15
8
6
10
4
5
2
0
1983
1989
Spain: Actual vs. Optimal Interest Rate
30
1983
1987
Nominal Interest Rate
35
25
1985
0
1985
1987
1989
Nominal Interest Rate
1991
1993
1995
1983
1997
Optimal Interest Rate
1985
1987
1989
Nominal Interest Rate
29
1991
1993
1995
1997
Optimal Interest Rate
Table 1: Estimates of the Variability of Aggregate Shocks
And Policy Rule Parameters
Country / Period
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
U.K.
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
σd
σs
(a-1)2
b
10,000*µ
3.77%
2.55%
1.59%
2.12%
0.4761
0.2327
-0.0439
-0.3090
6.767
1.513
5.50%
5.67%
3.05%
2.49%
0.4450
0.2287
-0.0222
-1.6027
13.461
7.352
5.39%
5.12%
3.27%
1.41%
0.6444
0.1198
-0.0358
-2.4762
18.721
3.140
6.26%
9.01%
4.75%
4.66%
4.31%
3.93%
4.01%
1.61%
0.5568
0.3239
0.3785
0.3669
-0.0127
-2.0314
-0.0223
-1.4161
21.820
26.294
8.540
7.967
4.47%
4.14%
22.69%
20.62%
8.31%
7.77%
15.11%
3.95%
2.40%
2.30%
16.56%
11.62%
3.49%
4.09%
7.28%
2.79%
0.4421
0.6180
0.4478
0.1884
0.5482
0.1354
0.0906
0.4946
-0.8116
-0.5149
0.2545
-0.0453
-0.3780
-1.7223
-0.0183
-0.0241
8.834
10.592
230.544
80.105
37.857
8.174
20.685
7.717
1.90%
2.43%
1.71%
0.85%
0.1704
0.0945
-0.0635
-0.0887
0.615
0.558
14.51%
9.50%
17.01%
6.84%
0.4204
0.7884
0.6383
0.0453
88.511
71.153
15.04%
7.36%
7.83%
4.08%
0.1142
0.1949
-0.0933
-0.3242
25.832
10.558
10.00%
8.89%
6.31%
9.96%
0.2939
0.3241
-0.0879
-0.8282
29.390
25.614
6.62%
4.08%
4.60%
2.29%
0.6133
0.2734
-0.1069
-0.1374
26.878
4.551
30
Table 2: Coefficient of Aversion to Inflation Variability,
Value of Loss and Performance Change
Country / Period
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
U.K.
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
λ
σπ
σy
Value of
Loss
(10,000*PI)
Performance
gain (in %)
0.609
0.921
1.62%
1.23%
2.81%
4.33%
4.673
2.873
38.53%
0.651
0.941
2.93%
0.76%
3.51%
3.69%
9.869
1.340
86.42%
0.675
0.980
3.11%
0.39%
4.40%
4.15%
12.831
0.484
96.23%
0.410
0.966
4.18%
1.22%
3.10%
9.05%
12.838
4.179
67.45%
0.382
0.944
0.916
0.862
0.094
0.098
0.680
0.982
3.78%
0.77%
1.31%
1.61%
16.77%
10.47%
4.31%
0.80%
2.75%
3.12%
4.15%
4.00%
3.70%
2.34%
5.07%
8.25%
10.126
1.104
3.014
4.453
38.731
15.710
20.840
1.863
0.274
0.678
0.662
0.867
6.87%
2.65%
1.49%
0.76%
3.70%
2.71%
2.19%
2.01%
22.895
7.116
3.084
1.036
0.093
0.679
16.36%
4.79%
5.22%
6.41%
49.463
28.758
41.86%
0.195
0.778
6.72%
2.67%
3.44%
4.66%
18.326
10.389
43.31%
0.312
0.811
5.46%
3.10%
3.35%
9.05%
17.043
23.322
-36.84%
0.382
0.632
4.26%
1.90%
3.74%
1.90%
15.578
3.603
76.87%
31
89.10%
-47.77%
59.44%
91.06%
68.92%
66.41%
Table 3: Performance Change
over a range of values for inflation aversion
λ
Range of
Performance
gain (in %)
[0.7, 0.9]
[-59%, 42%]
[0.7, 0.9]
[54%, 91%]
[0.7, 0.9]
[0.7, 0.9]
[58%, 97%]
[-69%, 78%]
[0.7, 0.9]
[72%, 91%]
[0.7, 0.9]
[0.1, 0.3]
[-4%, -48%]
[59%, 61%]
[0.7, 0.9]
[61%, 93%]
[0.7, 0.9]
[67%, 84%]
[0.7, 0.9]
[56%, 71%]
[0.7, 0.9]
[46%, 87%]
[0.7, 0.9]
[38%, 79%]
[0.7, 0.9]
[-41%, -2%]
[0.7, 0.9]
[75%, 81%]
Range for
Country
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
U.K.
Table 4: Estimates of the measures of performance
and policy efficiency change
Country
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
U.K.
Change in policy
efficiency
(10,000*∆I)
Change in macro
performance
(10,000*∆P)
Contribution of
policy to change in
performance
(Q=∆I/∆P)
2.238
1.801
124.31%
3.701
8.529
43.39%
6.178
12.347
50.04%
4.678
8.659
54.02%
4.012
9.022
44.46%
-0.786
-1.439
-54.59%
9.387
23.021
40.78%
17.347
18.977
91.41%
7.453
15.779
47.23%
58.26%
1.193
2.048
15.463
20.705
74.68%
7.193
2.840
7.937
-6.279
90.62%
45.23%
8.564
11.975
71.52%
32
Table 5: Measures of Efficiency and Performance Change
over a range of values for inflation aversion
Country
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
U.K.
Range for
λ
Change in policy
inefficiency
(10,000*∆I)
Change in macro
performance
(10,000*∆P)
Contribution
of policy to
change in
performance
(Q=∆I/∆P)
[0.7, 0.9]
[0.458, 3.412]
[-2.489, 2.854]
[18%, 134%]
[0.7, 0.9]
[2.990, 5.130]
[5.208, 10.231]
[14%, 66%]
[0.7, 0.9]
[2.370, 7.293]
[7.312, 14.742]
[27%, 76%]
[0.7, 0.9]
[0.216, 5.243]
[-10.491, 10.902]
[8%, 63%]
[0.7, 0.9]
[1.048, 9.197]
[8.945, 12.129]
[9%, 56%]
[0.7, 0.9]
[-0.993, -0.489]
[-0.272, -1.512]
[-54%, -4%]
[0.1, 0.3]
[8.679, 10.370]
[14.523, 57.165]
[15%, 58%]
[0.7, 0.9]
[16.586, 19.271]
[11.890, 21.261]
[82%, 139%]
[0.7, 0.9]
[0.7, 0.9]
[3.170, 12.849]
[0.586, 1.268]
[12.679, 30.065]
[1.548, 3.070]
[12%, 49%]
[41%, 67%]
[0.7, 0.9]
[12.590, 17.055]
[18.773, 167.067]
[7%, 98%]
[0.7, 0.9]
[0.7, 0.9]
[10.486, 20.418]
[0.878, 3.161]
[5.666, 33.261]
[-7.078, -1.103]
[78%, 104%]
[12%, 154%]
[0.7, 0.9]
[6.388, 13.498]
[9.284, 14.109]
[69%, 86%]
33
Table 6: Estimates of Measures under Alternative
Targets for Inflation and Output
Country / Period
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
U.K.
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
σd
σs
σπ
σy
Performance gain
(in %)
Contribution of
policy to change in
performance
3.46%
2.25%
1.25%
1.54%
1.32%
1.08%
2.33%
2.97%
44.20%
139.28%
3.64%
4.57%
2.88%
1.82%
2.34%
0.74%
2.01%
2.49%
82.33%
37.18%
2.47%
3.80%
1.77%
0.54%
1.44%
0.38%
2.74%
2.68%
92.59%
35.49%
4.38%
5.11%
1.70%
2.04%
1.65%
1.18%
2.39%
3.46%
61.10%
89.39%
3.21%
4.30%
2.50%
1.23%
2.45%
0.75%
1.62%
2.08%
80.43%
51.24%
3.62%
2.60%
1.25%
2.08%
1.28%
1.51%
2.00%
2.61%
-58.70%
-87.70%
11.06%
9.15%
4.21%
5.51%
3.46%
4.96%
2.34%
1.41%
30.84%
75.07%
4.47%
4.06%
3.14%
4.09%
2.86%
0.74%
3.59%
2.89%
92.89%
56.03%
6.58%
2.92%
3.55%
1.52%
3.32%
1.48%
1.81%
2.31%
40.73%
28.62%
1.72%
1.79%
1.57%
0.63%
1.38%
0.58%
2.09%
1.80%
73.70%
56.49%
8.78%
7.93%
8.27%
3.79%
7.35%
2.12%
2.81%
4.12%
30.28%
76.90%
6.69%
4.42%
3.93%
2.11%
3.59%
1.55%
1.74%
2.92%
25.60%
75.32%
7.01%
5.62%
2.56%
4.89%
2.13%
2.96%
2.38%
3.57%
-78.67%
42.71%
3.31%
2.65%
1.85%
1.67%
2.02%
1.38%
1.91%
1.70%
40.46%
77.46%
34
Table A.1: Estimates of the Structural Parameters
Country / Period
Austria
Belgium
Denmark
Finland
France
Germany
Greece
Ireland
Italy
Netherlands
Portugal
Spain
Sweden
U.K.
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
ω
γ
φ
0.4136
1.6890
0.3038
0.3374
1.1826
3.1149
0.3355
1.4112
0.1919
0.6055
0.9734
0.7524
0.4144
2.8813
0.1780
2.2173
0.2295
1.7005
0.2194
0.2744
0.2695
0.3526
0.4106
0.4583
1.0539
0.9199
0.6743
0.8535
0.9700
0.9358
1.4519
1.2447
0.0860
0.1066
0.6962
1.9798
0.1739
0.3576
0.5085
0.4449
0.9023
1.1104
0.9285
0.4655
0.5052
0.1818
0.7208
1.8384
0.2602
0.2479
1.1904
1.0593
0.8325
0.9655
0.6470
1.0902
0.4789
0.2205
3.1577
2.9558
0.6828
1.2708
0.8917
0.5430
0.8576
0.8268
0.2277
0.9233
1.3639
0.4569
0.7581
1.2014
0.2727
0.8785
0.7569
0.8696
0.6403
0.8103
0.3433
0.5349
0.7448
0.3953
0.8212
0.8228
35