ArticlePDF Available

Abstract and Figures

This paper calculates indices of central bank autonomy (CBA) for 163 central banks as of end-2003, and comparable indices for a subgroup of 68 central banks as of the end of the 1980s. The results confirm strong improvements in both economic and political CBA over the past couple of decades, although more progress is needed to boost political autonomy of the central banks in emerging market and developing countries. Our analysis confirms that greater CBA has on average helped to maintain low inflation levels. The paper identifies four broad principles of CBA that have been shared by the majority of countries. Significant differences exist in the area of banking supervision where many central banks have retained a key role. Finally, we discuss the sequencing of reforms to separate the conduct of monetary and fiscal policies. IMF Staff Papers (2009) 56, 263–296. doi:10.1057/imfsp.2008.25; published online 23 September 2008
Content may be subject to copyright.
SPECIAL PAPERS IN INTERNATIONAL ECONOMICS
No. 19, May 1996
THE POLITICAL ECONOMY OF
CENTRAL-BANK INDEPENDENCE
SYLVESTER C. W. EIJFFINGER
AND
JAKOB DE HAAN
INTERNATIONAL FINANCE SECTION
DEPARTMENT OF ECONOMICS
PRINCETON UNIVERSITY
PRINCETON
, NEW JERSEY
SPECIAL PAPERS
IN INTERNATIONAL ECONOMICS
S
PECIAL PAPERS IN INTERNATIONAL ECONOMICS are
published by the International Finance Section of the De-
partment of Economics of Princeton University. Although
the Section sponsors the publications, the authors are free
to develop their topics as they wish. The Section welcomes
the submission of manuscripts for publication in this and
its other series. Please see the Notice to Contributors at
the back of this Special Paper.
The authors of this Special Paper are Sylvester C.W.
Eijffinger and Jakob De Haan. Professor Eijffinger is
Professor of Economics at the College of Europe (Bruges),
Humboldt University of Berlin, and Tilburg University, and
a Fellow of the Center for Economic Research in Tilburg.
He has been a Visiting Scholar at the Deutsche
Bundesbank, the Bank of Japan, the Banque de France,
the Bank of England, and the Federal Reserve System.
Professor De Haan is Jean Monnet Professor of European
Economic Integration at the University of Groningen. He
has written extensively about the political economy of
monetary and fiscal policy.
PETER B. KENEN, Director
International Finance Section
SPECIAL PAPERS IN INTERNATIONAL ECONOMICS
No. 19, May 1996
THE POLITICAL ECONOMY OF
CENTRAL-BANK INDEPENDENCE
SYLVESTER C. W. EIJFFINGER
AND
JAKOB DE HAAN
INTERNATIONAL FINANCE SECTION
DEPARTMENT OF ECONOMICS
PRINCETON UNIVERSITY
PRINCETON
, NEW JERSEY
INTERNATIONAL FINANCE SECTION
EDITORIAL STAFF
Peter B. Kenen, Director (on leave)
Kenneth S. Rogoff, Acting Director
Margaret B. Riccardi, Editor
Lillian Spais, Editorial Aide
Lalitha H. Chandra, Subscriptions and Orders
Library of Congress Cataloging-in-Publication Data
Eijffinger, Sylvester C. W.
The political economy of central-bank independence / Sylvester C.W. Eijffinger and
Jakob De Haan.
p. cm.—(Special papers in international economics ; no. 19)
Includes bibliographical references.
ISBN 0-88165-308-X (pbk.) : $11.00
1. Banks and banking, Central. I. Haan, Jakob De. II. Title.
HG1811.E37 1996
332.11—dc20 96-14334
CIP
Copyright © 1996 by International Finance Section, Department of Economics, Princeton
University.
All rights reserved. Except for brief quotations embodied in critical articles and reviews,
no part of this publication may be reproduced in any form or by any means, including
photocopy, without written permission from the publisher.
Printed in the United States of America by Princeton University Printing Services at
Princeton, New Jersey
International Standard Serial Number: 0081-3559
International Standard Book Number: 0-88165-308-X
Library of Congress Catalog Card Number: 96-14334
“THE ONLY GOOD CENTRAL BANK IS ONE THAT CAN SAY
NOTOPOLITICIANS
(The Economist, February 10, 1990, p. 10)
PREFACE
In recent years, academics and policymakers have shown increasing
interest in the independence of central banks with respect to the
formulation of monetary policy. In the European Union, this interest
was realized in the Treaty on European Union (Maastricht Treaty),
according to which the European Central Bank will have complete
autonomy in conducting the common monetary policy of the European
Union. Hungary, the Czech Republic, and several other countries in
Central Europe have decided on autonomy for their central banks. In
most of the Anglo-Saxon countries, the issue continues to be discussed.
Public debate in the United Kingdom seems to lean toward more
independence for the Bank of England; the Congress in the United
States continues to question the autonomy of the Federal Reserve.
This paper analyzes from various perspectives the advantages and
disadvantages of central-bank independence and discusses the theoretical
and empirical arguments in favor of autonomy. It reviews and criticizes
generally accepted indices of central-bank independence, investigates the
determinants of independence, and, ultimately, tries to decide whether
or not an independent central bank is, in practice, desirable.
We wish to acknowledge a number of colleagues for their many helpful
comments and suggestions on previous drafts of this paper. We are
especially grateful to Onno De Beaufort Wijnholds, Helge Berger, Alex
Cukierman, Paul De Grauwe, Gert-Jan Van ’t Hag, Marco Hoeberichts,
Lex Hoogduin, André Icard, Otmar Issing, Flip De Kam, Mervyn King,
David Laidler, Manfred Neumann, Ad Van Riet, Eric Schaling, Helmut
Schlesinger, Pierre Siklos, Dave Smant, Carl Walsh, Nout Wellink, Tony
Yates, Jean Zwahlen, and an anonymous referee. The views expressed in
the paper remain solely the responsibility of the authors, however, and
should not be interpreted as reflecting the opinions of these scholars and
policymakers or their institutions.
Tilburg/Groningen Sylvester C.W. Eijffinger
May 1996 Jakob De Haan
CONTENTS
1 INTRODUCTION 1
2
THEORETICAL CONSIDERATIONS ON CENTRAL-BANK
INDEPENDENCE
4
Inflation 4
Inflation Variability 12
The Level and Variability of Economic Growth 13
Objections to Central-Bank Independence 15
3
MEASURES OF CENTRAL-BANK INDEPENDENCE 22
Legal Measures of Central-Bank Independence 22
A Comparison of Legal-Independence Measures 24
Nonlegal Measures of Central-Bank Independence 26
4
EMPIRICAL EVIDENCE ON THE CONSEQUENCES OF
CENTRAL
-BANK INDEPENDENCE 29
The Level and Variability of Inflation 29
Economic Growth and Disinflation Costs 34
Other Variables 38
5
THE DETERMINANTS OF CENTRAL-BANK INDEPENDENCE 41
The Equilibrium Level of Unemployment 41
Government Debt 44
Political Instability 44
The Supervision of Financial Instruments 46
Financial Opposition to Inflation 49
Public Opposition to Inflation 51
Other Determinants 52
6
CONCLUDING COMMENTS 54
APPENDIX A: LEGAL MEASURES OF CENTRAL-BANK
INDEPENDENCE
56
APPENDIX B: EMPIRICAL RESEARCH ON THE
CONSEQUENCES OF CENTRAL
-BANK INDEPENDENCE 63
REFERENCES 70
TABLES
1 Alternative Approaches to Central-Bank Independence
and Accountability 19
2 Legal Indices of Central-Bank Independence 23
3 Rank-Correlation Coefficients of Indices of Central-Bank
Independence 25
4 Aspects of Central-Bank Independence: A Comparison
of Five Indicators 26
5 The Turnover Rate of Central-Bank Governors, 1950–1989 28
6 Inflation and Aspects of Central-Bank Independence 33
7 Average Inflation in Six Industrial Countries under
“Left-Wing” and “Right-Wing” Governments 35
8 Empirical Studies on the Determinants of Central-Bank
Independence 42
9 Central Banks and the Supervision on Financial Institutions 47
A1 Cukierman’s Legal Variables: The Dutch Case 60
B1 Empirical Studies on the Consequences of Central-Bank
Independence 64
B2 Empirical Studies on the Relation between Central-Bank
Independence and Inflation 66
B3 Empirical Studies on the Relation between Central-Bank
Independence and Economic Growth 68
B4 Empirical Studies on the Relation between Central-Bank
Independence and Other Economic Variables 69
1 INTRODUCTION
It is often argued that a high level of central-bank independence
coupled with an explicit mandate that the bank aim for price stability
are important institutional devices for maintaining that stability. In-
deed, a number of countries have recently increased the independence
of their central banks in order to raise their commitment to price
stability. According to Cukierman (1995), they have done so for various
reasons. First, the breakdown of institutions designed to safeguard
price stability—the Bretton Woods system and the European Monetary
System (EMS), for example—has led countries to search for alterna-
tives. Second, the relative autonomy of the Bundesbank is often seen as
evidence that central-bank independence can function as an effective
device for assuring price stability (Germany has one of the best
post–World War II inflation records among the industrial countries).
Third, the Treaty on European Union (Maastricht Treaty) requires an
independent central bank as a precondition for membership in the
Economic and Monetary Union (EMU); price stability will be the
major objective of the future European System of Central Banks
(ESCB), which will consist of the European Central Bank (ECB) and
the national central banks of all the member states of the European
Union (EU). Fourth, after recent periods of successful stabilization,
policymakers in many Latin American countries are looking for institu-
tional arrangements that can reduce the likelihood of a return to high
and persistent inflation. Fifth, the creation of independent central
banks in many former socialist countries is part of a more general
attempt of these countries to create the institutional framework needed
for the orderly functioning of a market economy. The extensive recent
literature suggesting that inflation and central-bank independence are
negatively related has also, no doubt, prompted governments to consider
enhancing the autonomy of their central banks. This paper critically
reviews that debate.
Most authors provide no clear definition of central-bank indepen-
dence. According to Friedman (1962), central-bank autonomy refers to
a relation between the central bank and the government that is com-
parable to the relation between the judiciary and the government. The
judiciary can rule only on the basis of laws provided by the legislature,
and it can be forced to rule differently only through a change in the
1
law. Central-bank independence relates to three areas in which the
influence of government must be either excluded or drastically cur-
tailed (Hasse, 1990): independence in personnel matters, financial
independence, and independence with respect to policy. Personnel
independence refers to the influence the government has in appoint-
ment procedures. It is not feasible to exclude government influence
completely in appointments to a public institution as important as a
central bank. The level of this influence, however, may be discerned by
criteria such as government representation in the governing body of the
central bank and government influence in appointment procedures,
terms of office, and dismissal of the governing board of the bank.
Financial independence refers to the ability given to the government
to finance government expenditure either directly or indirectly through
central-bank credits. Direct access to central-bank credits implies that
monetary policy is subordinated to fiscal policy. Indirect access may
result if the central bank is cashier to the government or if it handles
the management of government debt.
Policy independence refers to the maneuvering room given to the
central bank in the formulation and execution of monetary policy. As
pointed out by Debelle and Fischer (1995) and Fischer (1995), it may
be useful to distinguish between independence with respect to goals
and independence with respect to instruments. Two related issues are
important with respect to goals: the scope the central bank has to
exercise its own discretion and the presence or absence of monetary
stability as the central bank’s primary goal. If the central bank has been
assigned various goals, such as low inflation and low unemployment, it
has been accorded the greatest possible scope for discretion. In that
case, the central bank is independent with respect to goals, because it
is free to set the final goals of monetary policy. It may, for example,
decide that price stability is less important than output stability and act
accordingly. If it is given either general or specific objectives with
respect to price stability, however, the central banks’s discretionary
powers will be restricted.
To defend its goals, however, a central bank must wield effective policy
instruments. A bank is independent with respect to instruments if it is
free to choose the means by which to achieve its goals. It is not indepen-
dent if it requires government approval to use policy instruments.
1
The
Reserve Bank of New Zealand, for which the goal is precisely described
in a contract with the government, is not independent with respect to
1
If the central bank is obliged to finance budget deficits, moreover, it also lacks instru-
ment independence. In this regard, financial independence and instrument independence
2
goals; it is independent with respect to instruments, however, because
it chooses the methods by which to achieve its goal.
This paper uses the distinction between these aspects of independence
are related; instrument independence is, however, much broader, because it includes also
the power to determine interest rates.
in reviewing the literature on central-bank autonomy. It considers four
issues. Chapter 2 begins with a review of the theoretical case for
central-bank independence. The literature has advanced various argu-
ments to explain why countries with relatively independent central
banks may have a better inflation performance than countries in which
politicians have control over the central banks. These arguments often
refer to one or more specific aspects of central-bank independence.
Although central-bank autonomy may improve inflation performance, it
may also yield undesirable consequences in terms of lower and more
volatile economic growth rates.
Chapter 3 discusses the ways in which central-bank independence has
been measured and reviews four widely used indices of central-bank
autonomy. These measures have been developed by Alesina (1988,
1989), Grilli, Masciandaro, and Tabellini (1991), Cukierman (1992),
and Eijffinger and Schaling (1992, 1993a). Although the measures all
focus on legal aspects of central-bank independence, they diverge in
their rankings of central banks. The measures place different weights
on the various aspects of central-bank independence, as outlined above.
Chapter 4 reviews empirical studies on the link between central-bank
autonomy and economic performance. It begins by discussing the relation
between central-bank independence and the level and variability of infla-
tion, reviews the link between independence and the level and variability
of economic growth, and considers whether central-bank independence
reduces disinflation costs. The chapter concludes with a brief review of
studies discussing the link between central-bank independence and
other variables such as interest rates and government budget deficits.
Chapter 5 questions why central-bank independence varies across
countries—that is, what the determinants are of central-bank indepen-
dence. This issue has only recently been put on the research agenda.
Chapter 6 concludes the paper.
3
2 THEORETICAL CONSIDERATIONS ON CENTRAL-BANK
INDEPENDENCE
Inflation
Many observers believe that countries with independent central banks
have lower levels of inflation than countries in which central banks are
under the direct control of the government. Why would central-bank
independence, ceteris paribus, yield lower rates of inflation? The
literature provides three sorts of answers to this question: those based
on public-choice arguments, those based on the analysis of Sargent and
Wallace (1981), and those based on the time-inconsistency problem of
monetary policy.
According to the “older” public-choice view, monetary authorities are
exposed to strong political pressures to behave in accordance with the
government’s preferences.
2
Monetary tightening aggravates the budget-
ary position of the government. The reduction in tax income brought
about by a temporary slowdown of economic activity, possibly lower
receipts from “seigniorage,” and the short-run increase in the interest
burden on public debt all worsen the deficit. The government may
therefore prefer “easy money.” Indeed, some evidence exists that even
the relatively independent U.S. Federal Reserve caters to the desires of
the president, the Congress, or both. This evidence is based either on
close inspection of the contacts between the polity and the central bank
or on tests to determine whether monetary policy turns expansive before
elections
3
—as predicted by Nordhaus’s (1975) theory of the political
business cycle (Allen, 1986)—or diverges under administrations with
2
As Buchanan and Wagner (1977, pp. 117-118) put it: “A monetary decision maker is
in a position only one stage removed from that of the directly elected politician. He will
normally have been appointed to office by a politician subject to electoral testing, and he
may even serve at the pleasure of the latter. It is scarcely to be expected that persons who
are chosen as monetary decision makers will be the sort that are likely to take policy
stances sharply contrary to those desired by their political associates, especially since these
stances would also run counter to strong public opinion and media pressures. . . . ‘Easy
money’ is also ‘easy’ for the monetary manager....
3
See, for example, Akhtar and Howe (1991) and Havrilesky (1993). Havrilesky (p. 30)
even argues that “the contemporary view is that the [U.S.] Administration, while granting
significant leeway to the Fed, when necessary obtains the monetary policy actions that it
desires.”
4
different political orientation—as predicted by Hibbs’s (1977) partisan
theory (Alesina, 1988). At this stage, it suffices to conclude that the
more independent a central bank is, the less it will be under the spell
of political influences. The argument of Buchanan and Wagner relates
primarily to independence with respect to personnel and policy.
4
The
more influential the government is in appointing board members, the
more likely it will be that the central bank pursues the kinds of policies
desired by government.
A second argument to explain why central-bank independence may
tear on inflation was first put forward by Sargent and Wallace (1981),
who distinguish between fiscal authorities and monetary authorities. If
fiscal policy is dominant, that is, if the monetary authorities cannot
influence the size of the government’s budget deficit, money supply
becomes endogenous. If the public is no longer able or willing to
absorb additional government debt, it follows from the government
budget constraint that monetary authorities will be forced to finance
the deficit by creating money. If, however, monetary policy is domi-
nant, the fiscal authorities will be forced to reduce the deficit (or
repudiate part of the debt). The more independent the central bank is,
the less the monetary authorities can be forced to finance deficits by
creating money. This argument relates to financial independence.
A third, and, indeed, the most prominent, argument for central-bank
independence is based on the time-inconsistency problem (Kydland
and Prescott, 1977; Calvo, 1978; Barro and Gordon, 1983). Dynamic
inconsistency arises when the best plan made in the present for some
future period is no longer optimal when that period actually starts.
Various models have been based on this dynamic-inconsistency ap-
proach (Rogoff, 1985; Cukierman, 1992; Eijffinger and Schaling,
1993b; Schaling, 1995). In these models, the government and the
public are drawn into some setting of the prisoner’s dilemma. The
4
Neumann (1991, p. 103) emphasizes personnel independence with respect to the
governing board of the central bank: “The conditions of contract and of office would
have to be set such that the appointee frees him- or herself from all former political ties
or dependencies and accepts the central bank’s objective of safeguarding the value of the
currency as his or her professional leitmotif. We may call this a ‘Thomas Becket’ effect.”
Waller (1992b) develops a model for appointments to the central bank in the context of
a two-party political system, in which the victor of the last election is allowed to
nominate candidates, but the losing party is given the right to confirm the nominees. An
interesting outcome of the model is that if society wants to minimize partisan monetary
policy, it should increase the length of office of central-bank policy-board members
relative to the length of the electoral interval.
5
models differ in their assumptions with regard to government incen-
tives. Following McCallum (1995a), their central insights may be
explained as follows. It is assumed that policymakers seek to minimize
the loss function
where 0 < w and k > 1, whereas output is driven by
(1)
L(π
t
) wπ
2
(y
t
ky
n
)
2
,
where π is inflation, π
e
is expected inflation, y
t
is output, y
n
is the
(2)
y
t
y
n
β(π
t
π
e
t
u
t
),
natural output, and u
t
is a random shock. We assume, here, that
deviations of employment from its natural level are positively related to
unanticipated inflation. This follows from the existence of nominal-
wage contracts in conjunction with a real wage that is normally above
the market-clearing real wage. Policymakers have an objective function
that assigns a positive weight to employment stimulation (for reelection
considerations, for example, or for partisan reasons) and a negative
weight to inflation. Policymakers minimize the loss function (equation
[1]) on a period-by-period basis, taking the inflation expectations as
given. This gives
With rational expectations, inflation is then
(3)
π
t
β(k 1)y
n
w β
2
β
2
w β
2
π
e
t
β
2
w β
2
u
t
.
If policymakers were to follow a rule taking into account private
(4)
π
t
β(k 1)y
n
w
β
2
w β
2
u
t
.
rational-expectational behavior, inflation would be
Because the same level of output pertains in both cases, the latter
(5)
π
t
β
2
w β
2
u
t
.
outcome is clearly superior. No matter what exactly causes the dynamic-
inconsistency problem, the resulting rate of inflation is, in all cases,
suboptimal.
5
5
Other sources of the time-inconsistency problem originate with the public finances.
The dynamic inconsistency of monetary policy may first arise because of the incentives for
the government to inflate change before and after the public has settled for a nominal
interest rate, taking into account its expected rate of inflation. Before the public commits
6
Devices have therefore been suggested in the literature to reduce
the inflationary bias. Barro and Gordon (1983) conclude that the best
solution for the time-inconsistency problem consists of the introduction
of fixed rules in monetary policy, that is, the authorities commit them-
selves to certain policy rules. Once uncertainty is introduced and the
level of output is affected by shocks, the case becomes one for a
feedback rule, in which monetary policy optimally responds to shocks.
The problem with rules, however, is the absence of a higher authority
to enforce a commitment. The handing-over of authority to the central
bank by the political authorities may help, because it can be regarded
as an act of partial commitment (Rogoff, 1985; Neumann, 1991; Cukier-
man, 1992, chap. 18). By delegating some of their authority to a
relatively apolitical institution, politicians accept certain restrictions on
their future freedom of action.
6
The degree of central-bank independence plays a meaningful role
only if the central bank puts a different emphasis on alternative policy
objectives than the government. The literature points to two main
differences (Cukierman, 1992, chap. 18). One relates to possible
differences between the rate of time preference of political authorities
and that of central banks. For various reasons, central banks are often
more conservative and take a longer view of the policy process than do
politicians. The other difference concerns the subjective weights in the
objective function of the central bank and that of the government. It is
often assumed that central bankers are more concerned about inflation
than about policy goals such as the achievement of high employment
levels and adequate government revenues. If monetary policy is set at
the discretion of a conservative central banker, a lower average time-
itself, the central bank has an incentive to abstain from making inflation. After positions
in government bonds have been taken, policymakers have an incentive to create inflation
(Cukierman, 1992). Another source of the inconsistency problem also originates in the
finances of government and may be referred to as the “revenue” or “seigniorage” motive
for monetary expansion (Barro, 1983). The dynamic inconsistency of monetary policy arises
in this regard because of incentives for the government to inflate change before and after
the public has chosen the level of real money balances. The conclusion that the resulting
rate is suboptimal generally also holds for models with incomplete information. Cukierman
(1992, chap. 18), for instance, provides a model in which the public is not fully informed
about the shifting objectives of the political authorities and in which there is no perfect
control of information.
6
An alternative solution to the time-inconsistency problem is reputation building
(Canzoneri, 1985). Fratianni and Huang (1994) show, however, that the case of asym-
metric information gives no assurance that reputation may work for the central bank in
the Barro-Gordon model.
7
consistent inflation rate will result.
7
The foregoing analysis makes it
clear that this argument for central-bank independence is primarily
related to policy independence.
The best way to illustrate the argument is to present a “stripped”
version of Rogoff’s (1985) model. In Rogoff’s model, society can
sometimes improve its position by appointing a central banker who
does not share the social-objective function but, instead, places a
higher weight on price stability relative to output stabilization. In the
simplified version, output is given by equation (1), in which the natural
level of output is put at 0 and the parameters at 1. The timing of
events in the Rogoff model is as follows: first, inflation expectations
(π
t
e
) are set (nominal-wage contracts are signed); then, the shock (u
t
)
occurs; and finally, the central banker sets the inflation rate (π
t
).
Society’s loss function is given by
where the weight on output stabilization χ > 0 and yˆ > 0, so that the
(6)
L
t
1
2
π
2
t
χ
2
(y
t
ˆy
t
)
2
,
desired level of output (yˆ) is above the natural level. Rogoff shows that
it is optimal for society to choose an independent (conservative) central
banker who assigns a higher weight to price stability in his loss function:
where ε, the additional weight on the inflation goal, lies between zero
(7)
I
t
1 ε
2
π
2
t
χ
2
(y
t
ˆy
t
)
2
,
and infinity (0 < ε < ).
Substituting and taking first-order conditions with respect to π
t
and
solving for rational expectations, we obtain
Policy rule (8) shows that the introduction of a conservative central
(8)
π
t
χ
1
ε
ˆy
χ
1
ε χ
µ
t
.
7
Waller (1992a) analyzes the appointment of a conservative central banker in a model
that distinguishes between sectors that differ in their degree of competitiveness of the
labor market. The main result of his paper is that, although agents in both sectors have
the same preferences with respect to inflation and output stability, nominal-wage rigidity
in the nonclassical labor market causes output in this sector to be more variable in
equilibrium than in the classical sector. Consequently, if the classical sector were allowed
to choose the “conservative” central banker, it would choose a more vigorous inflation
fighter than the nonclassical sector would choose.
8
banker (ε > 0) leads to a lower inflationary bias and a lower variance of
inflation. The variance of output is, however, an increasing function of
the conservatism of the central banker. There is a trade-off between
credibility and flexibility in the Rogoff model. It can be shown that the
optimal value for ε, in terms of social-loss function (6), is positive but
finite. This implies that it is optimal for society to appoint a conserva-
tive central banker.
Rogoff makes the crucial assumption that the central banker is
completely independent and cannot be overridden ex post when the
inflationary expectations (π
t
e
) have been set and the policy is to be
carried out. This may lead to large losses for society when extreme
productivity shocks (u
t
) occur. Lohmann (1992) introduces the possibil-
ity of overriding the central banker at a strictly positive but finite cost.
Society’s loss function therefore changes to
where δ is a dummy that takes on the value of 1 when the central bank
(9)
L
t
1
2
π
2
t
χ
2
(y
t
ˆy
t
)
2
δc ,
is overridden and is 0 otherwise, and c is a cost that society incurs
when the central bank is overridden. The central bank’s loss function
(7) stays the same.
The timing of events in the Lohmann model is as follows: In the first
stage, the central banker’s additional weight (ε) on the inflation goal is
chosen, as is the cost (c) of overriding the central banker. The inflation
expectations are then set. In the third stage, the productivity shock
realizes, after which the central banker sets the inflation rate, which is
either accepted or not. If it is not accepted, society overrides the
central banker, incurs the cost (c), and resets the inflation rate. Finally,
inflation and output realize.
In equilibrium, the central banker will not be overridden. In the
case of an extreme productivity shock, he or she will set the inflation
rate so that society will be indifferent between overriding or not.
Rogoff’s model is a special case of Lohmann’s argument, where c = .
Lohmann shows that the optimal central-bank institution is character-
ized by 0 < ε
*
< and0<c
*
< .
An important result from Rogoff’s model is that the reduction in the
equilibrium inflation rate that results from the appointment of a
conservative and independent central banker generally comes at the
expense of greater output variability from supply shocks, because the
central banker offsets output shocks to a lesser extent than would the
9
government.
8
Nevertheless, gains from lower inflation exceed losses
from decreased stability. On net, therefore, society is made better off by
appointing a conservative central banker. It is not optimal in the Rogoff
model, however, to appoint a central banker whose only concern is low
and stable inflation.
Rogoff’s model has been criticized by McCallum (1995a), who
contends that it is inappropriate simply to presume that the central
bank behaves in a discretionary manner. It is more useful, he argues,
to set the constant term and inflationary-expectations (π
t
e
) coefficient in
equation (3) equal to zero, thereby eliminating the inflationary bias
while retaining the desirable countercyclical response to the shock (u
t
).
All that is needed to avoid the inflationary bias is for the central bank
to recognize the futility of continually exploiting temporarily given
expectations while planning not to do so in the future, and to recognize
that the bank’s objectives will be more fully achieved, on average, if it
abstains from attempts to exploit these transient expectations. McCallum
(1995b, p. 18) argues that “there is nothing tangible to prevent an
actual central bank from behaving in this ‘committed’ or ‘rule-like’
fashion, so . . . some forward-looking banks will in fact do so. Analyti-
cal results that presume non-committed or discretionary behaviour may
therefore be misleading.” Although McCallum has a point, the problem
is that a highly dependent central bank may not be able to behave in
such a manner.
In addition to a legislative strategy, which will create, by law, an
independent central bank and will mandate it, also by law, to direct its
policies toward achieving price stability, other mechanisms have been
suggested to overcome the incentive problems of monetary policy. The
so-called contracting strategy regards the design of monetary institu-
tions as one that involves structuring a contract between the central
bank and the government. The optimal contract is an application of
ideas from the literature on principal agents. In this application, the
government is viewed as the principal, and the central bank is viewed
as the agent. The principal signs a contract with the agent, according to
which the bank is subject to an ex post penalty schedule that is linear
in inflation. The nature of the contract will affect the incentives facing
8
Recently, Alesina and Gatti (1995) have introduced another source of output
variability in a model of the Rogoff type, namely, variability introduced in the system by
the uncertainty about the future course of policy. This uncertainty results from uncertain
electoral outcomes in the case in which there are two contending parties with different
preferences regarding inflation and unemployment. In this circumstance, the overall effect
of central-bank independence on output variability is ambiguous.
10
the bank and will, thereby, affect monetary policy (Walsh, 1993).
Persson and Tabellini (1993) suggest a targeting strategy, in which the
political principals of the central bank impose an explicit inflation
target and make the central-bank leadership explicitly accountable for
its success in meeting this target. Such a system has existed since 1989
in New Zealand, where the governor of the reserve bank may, under
certain circumstances, be dismissed if the inflation rate exceeds 2 per-
cent (see below for further details).
It is interesting to note that the analysis of Persson and Tabellini
(1993) suggests that the optimal contract with a central bank implies no
loss in terms of stabilization policy. As pointed out above, this result
contrasts with the outcomes of most models in which monetary policy
is delegated to an independent central bank, and credibility is increased
at the expense of an optimal output-stabilization policy. Walsh (1993,
1995b) and Persson and Tabellini (1993) show that the optimal central-
bank contract may serve to eliminate the inflation bias while still
preserving the advantages of stabilization. This conclusion holds even if
the central bank has private information.
9
It is thus clear that the
contracting strategy is related to independence with respect to instru-
ments but not with respect to goals.
The contracting strategy has also been criticized. For one thing,
although the concept of social planners may be useful as a benchmark,
these planners do not exist in practice. Hence, the government has to
be relied on to impose the optimal incentive schedule on the central
bank ex post. The government is also subject to an inflationary bias and
usually to a greater extent than is the central bank (Cukierman, 1995).
McCallum (1995a) argues that the optimal contract will not be credible
if the government cannot commit to the optimal penalty schedule
before various types of nominal contracts are concluded. A contract
does not overcome the motivation for dynamic inconsistency; it merely
relocates it.
Svensson (1995) has recently shown that when the objective function
of the central banker differs from that of society with respect to the
desired level of inflation (rather than the relative preference for price
9
Walsh (1995b) also considers the situation in which candidates to head the central
bank differ in their competency, the central bank’s stance on monetary policy is not
observable, and the informational content of a publicly observable signal about an
aggregate supply shock is affected both by the central bank’s competency and by the
bank’s implementation of given policies. In Walsh’s model, the principal can induce the
central bank to behave as demanded by using a contract that resembles an inflation-
targeting rule with a reporting requirement.
11
stability), delegation of authority to a central banker with the “right”
desired inflation level or target achieves the same results as the optimal
contract. This implies that the socially optimal level of welfare can be
achieved through delegation of authority to a central banker with a
suitable desired level of inflation, rather than through an incentive
contract for the bank. As pointed out by Cukierman (1995), the big
advantage of the first institution is that it does not have to rely on the
ex post implementation of the optimal contract by governments ridden
by inflation bias. It would therefore appear that Svensson’s result implies
that it is possible to reach the social optimum simply by delegating
authority to an appropriately chosen type of central banker. A practical
difficulty that may prevent the implementation of such an institution is
that the political principals may not be able to identify ex ante the
levels of inflation potential candidates for central-bank leadership may
desire. Svensson suggests that this problem may be circumvented by
giving the bank independence only with regard to instruments, but not
to goals, so that the target or “desired” rate of inflation in the bank’s
loss function is mandated by government.
Inflation Variability
The preceding analysis suggests that central-bank independence may
reduce pre-election manipulation of monetary policy. If that is the
case, central-bank independence may also result in more stable money
growth and, therefore, less variability in inflation.
Another, related, argument also explains why central-bank indepen-
dence may lead to less variability in inflation. Politicians not only strive
to remain in office as long as possible, they are also partisan and wish
to deliver benefits to their constituencies (Hibbs, 1977). Some evidence
indicates that the pattern of unemployment and inflation is systemati-
cally related to the political orientation of governments. Whereas
“right-wing” governments are generally thought to give a high priority
to lower inflation, “left-wing” governments are often supposed to be
more concerned about unemployment. Alesina (1988) reports that the
unemployment rate in the United States is generally higher under
Republican administrations than under Democratic administrations,
whereas the inflation rate is lower under Republican administrations.
Similar results have been reported by Havrilesky (1987) and Tabellini
and La Via (1989). Existing evidence lends support to the view that the
redistributional consequences of inflation provide an incentive for the
political Left to endorse expansionary policies and for the Right to fight
inflation (Alesina, 1989). This implies that inflation variability may be
12
high if the government changes regularly, especially if the monetary
authorities are dominated by elected politicians. A relatively indepen-
dent central bank, however, will not change its policy after a new
government has been elected. Central-bank autonomy may therefore
reduce variability in inflation (Alesina, 1988).
Milton Friedman (1977) gives another reason why central-bank
independence may affect inflation variability. Friedman wanted to
explain why a positive correlation exists between the level of inflation
and the variability of inflation across countries and over time for any
given country. In Friedman’s analysis, a government may temporarily
pursue a set of policy goals (output, employment) that leads to high
inflation; this, in turn, elicits strong political pressure to reduce the
debasing of the currency. Chowdhury (1991) recently reexamined the
relation between the level and the variability of inflation for a sample
of sixty-six countries over the 1955–85 period. His results indicate the
presence of a significant positive relation between the rate of inflation
and its variability.
The Level and Variability of Economic Growth
Two opposing views have been expressed in the literature with respect
to the effect of central-bank independence on the level of economic
growth. Some authors have argued that the real interest rate depends
on money growth; they assume that the Fisher hypothesis is nullified
by the Mundell-Tobin effect.
10
A low level of inflation that is caused
by a restrictive monetary policy results in high real interest rates, which
may have detrimental effects on the level of investment and, hence, on
economic growth (Alesina and Summers, 1993). There seems to be
some evidence in support of the first part of the argument: countries
with low levels of inflation have high ex post real interest rates (De
Haan and Sturm, 1994a).
Other arguments, however, suggest that central-bank independence
may further economic growth. As outlined above, an independent
central bank may be less prone to political pressures and may therefore
behave more predictably. This may enhance economic stability and re-
duce risk premia in interest rates, thereby stimulating economic growth
(Alesina and Summers, 1993). In addition, central-bank independence
10
A rise in expected inflation will lead, according to Mundell (1963), to the substitu-
tion of long-term financial assets for liquid assets and, according to Tobin (1965), to the
substitution of physical-capital goods for liquid assets, thus lowering the marginal
efficiency of capital and, thereby also, the expected (ex ante) real interest rates.
13
may moderate inflation. High levels of inflation may obstruct the price
mechanism, and it is likely that this will hinder economic growth. Many
economists, especially those involved in central banking, believe that
even moderate rates of inflation impose significant economic costs on
society (Fischer, 1993).
11
Recently, Grimes (1991) and Fischer (1993)
have provided evidence to support the view that inflation harms eco-
nomic growth.
12
One channel through which this effect may operate is
increased inflation uncertainty. As noted above, a strong link exists
between the level and the variability of inflation. Strong variation may
lead to high inflation uncertainty, which, in turn, may damage economic
growth. If central-bank independence reduces inflation variability and
promotes less inflation uncertainty, the economy may prosper. Empirical
studies on the links between inflation variability, inflation uncertainty,
and economic growth, however, provide only mixed support for this
point of view. Logue and Sweeney (1981), using annual data for twenty-
four countries, find no evidence for a significant negative impact of
inflation variability on real growth. A similar conclusion is reached by
Jansen (1989). Engle (1983) finds little evidence for a link between
inflation uncertainty and the relatively high rates of inflation experienced
by the United States in the 1970s. Cukierman and Wachtel (1979),
however, report a positive correlation between the rate of inflation and
the dispersion of inflation forecasts gathered from the Michigan and
Livingston inflation surveys. Evans (1991) has also published evidence
consistent with the point of view that uncertainty about the long-term
prospects for inflation is strongly linked to the actual rate of inflation.
Various theoretical positions have been delineated concerning the
impact of central-bank independence on the variability of economic
growth. One of these states that if the central bank introduces restrictive
measures to combat inflation, it is likely to provoke recessions. In this
view, inflation has become too high because the monetary authorities
were too lax in previous periods. An independent central bank striving
for price stability will not so easily let inflation run out of control and
therefore will not follow such a stop-and-go policy. Fluctuations in real
output will consequently be smaller (Alesina and Summers, 1993).
Rogoff (1985), however, and Eijffinger and Schaling (1993b) conclude
that when the central bank gives priority to price stability, the variability
11
Fischer (1994) points out that the relation between inflation and economic growth
may be nonlinear. Furthermore, the link between inflation and growth for low levels of
inflation (1 to 3 percent) is difficult to determine empirically.
12
See also Karras (1993), however, who argues that the correlation reported by Grimes
(1991) is a consequence of the cyclical character of both variables.
14
of income will be greater than when the central bank also strives for
stabilization of the economy.
It will have become clear by now that only empirical research can
decide which view corresponds most closely to the data. Chapter 4
takes up this issue.
Objections to Central-Bank Independence
Various theoretical arguments have been given in support of central-
bank autonomy. Chapter 4 will show that performance with regard to
inflation is better, on average, in countries that have a relatively inde-
pendent central bank than in countries in which the government more
directly controls the central bank. Furthermore, various indications
suggest that central-bank independence does not imply sacrifices in
terms of lower output growth or higher unemployment.
Two objections have been raised against central-bank independence:
the lack of democratic accountability and the potential damage to
policy coordination (Goodhart, 1994). The final sections of this chapter
will deal with these issues.
Accountability. The issue of the way in which central-bank indepen-
dence relates to democratic accountability is discussed mainly in the
Anglo-Saxon countries (Fischer, 1994; Eijffinger, 1994). Some authors
have argued that monetary policy is just like other instruments of
economic policy, such as fiscal policy, and so should be determined
entirely by democratically elected representatives. Such a view implies,
however, a too direct involvement of politicians with monetary policy.
Nevertheless, in every democratic society, monetary policy has ultimately
to be under the control of democratically elected politicians; one way or
another, the central bank must be accountable. The parliament or, in the
United States, the Congress, is responsible for central-bank legislation.
In other words, the “rules of the game” (that is, the objectives of
monetary policy) are settled by the legislatures in accordance with
normal democratic procedures. The “game” (monetary policy), however,
is delegated to the central banks. Because the parliament, or Congress,
can alter legislation, the central bank remains under the ultimate control
of the legislative body. Furthermore, in case the specified objective is
not realized, the central bank, or the politician who bears final responsi-
bility through his or her power to overrule the bank’s policy, can be held
accountable.
Central-bank independence and democratic accountability may be
implemented in various ways. Each country organizes things differently.
Three relatively independent central banks, the Deutsche Bundesbank,
15
the Nederlandsche Bank, and the Reserve Bank of New Zealand,
exemplify the variations in approach. Five aspects of the division of
responsibilities between the government and the central bank illustrate
these differences (Roll et al., 1993):
(1) The ultimate objective(s) of monetary policy. The Reserve Bank
of New Zealand has only one formal objective: price stability. Thus, the
central bank is not independent with respect to its goals. The Bundesbank
has a similar prime objective, which is, however, less specific—formally
referred to as “defense of the value of the currency” (Casear, 1981;
Kennedy, 1991). In addition, the Bundesbank has the obligation to offer
general support to the government’s economic policy in instances in
which support does not prejudice the primary objective of price stability
(Bundesbank Law 1957, section 12). This subsidiary statutory objective,
however, is de facto unimportant. The objective of the Nederlandsche
Bank is to regulate the value of the guilder in order to enhance welfare
(Dutch Bank Law, section 9.1). This objective is nowadays interpreted
as a stable exchange rate for the guilder vis-á-vis the deutsche mark.
(2) Precision of target specification. The Reserve Bank of New Zealand
has to agree with the government on a tight target range for inflation for
a three-year period. The Bundesbank has no obligation to agree to,
obey, or announce any such targets. Since 1974, the Bundesbank has
announced the targeted rate (or zone) for money growth, which implies
an inflation target. The German government has been responsible for
decisions about the exchange rate. This has been a reason for many
conflicts between the Bundesbank and the government (Marsh, 1992).
(3) Statutory basis for independence. The Reserve Bank of New Zea-
land must agree with the government about a target for inflation but is
free to choose its instruments (Debelle and Fischer, 1995). The Bundes-
bank is completely independent of any instruction from the government.
It may consult the government, but it has no obligation to agree.
Under section 13 of the Bundesbank Law 1957, government repre-
sentatives have the right to attend meetings of the Zentralbankrat
(Central Bank Council), but not to vote. The Dutch Bank Law of 1948
contains no specific articles on the statutory basis for the independence
of the Nederlandsche Bank.
(4) Overriding the central bank. In New Zealand, the governor of the
central bank can be dismissed if he fails to deliver the inflation target
(obligation ad hominem). The contract ensures this by some clearly
identified escape clauses, such as a rise in indirect taxes or a change in
exchange-rate regime. In Germany, the government can suspend
decisions of the Bundesbank for a maximum of two weeks (Bundesbank
16
Law 1957, section 13), a temporary veto that has seldom been formally
deployed (Berger, 1995). Only through a change in the relevant legisla-
tion by a simple majority in parliament can the Bundesbank be overruled
by the government. The Zentralbankrat is responsible for monetary
policy (collective responsibility). The Netherlands has a unique central-
bank legislation. According to section 26 of Dutch Bank Law, the
minister of finance has the right to give an “instruction” to the bank with
regard to monetary policy.
13
The right to give instructions makes the
minister responsible for monetary policy vis-á-vis parliament.
14
(5) Appointment of bank officials. In New Zealand, both the minister
of finance and the board of the central bank must ratify the appointment
of the governor (double veto). Board appointments are made by the
finance minister, and the deputy governor is appointed by the board, on
recommendation of the governor. In Germany, the Zentralbankrat is the
governing board of the Bundesbank. Apart from the so-called Direktor-
ium (Directorate), the presidents of the nine Landeszentralbanken
(regional central banks) are members of the Zentralbankrat. The
Direktorium is comprised of the president, the vice-president, and
nowadays, a maximum of six other members, who are appointed by the
president of the Federal Republic on nomination of the federal govern-
ment.
15
The Zentralbankrat is consulted in this process. The presidents
of the Landeszentralbanken are nominated by the Bundesrat (the upper
federal chamber), based on recommendations from the governments of
the Länder (states). The Zentralbankrat is then again consulted. In the
Netherlands, the president and the director-secretary of the Neder-
landsche Bank are appointed by the minister of finance, on the basis of
a recommendation list containing only two names, which have been
selected in a combined meeting of the governing board and the super-
visory board of the bank (Dutch Bank Law, section 23). The other
members of the governing board are also appointed by the minister, on
13
This right is a kind of ultimum remedium and has never been applied. Zijlstra (1992),
who was president of the Nederlandsche Bank between 1967 and 1981, recounts in his
memoirs that Prime Minister Den Uyl (1974 to 1977) considered using this instrument
after the bank had introduced credit restrictions in 1977.
14
This construction is no longer allowed under the Maastricht Treaty. In the third
phase of EMU, which according to the Treaty should start no later than 1999, the right
of the minister of finance to give instructions to the central bank must be abolished.
15
Before unification, each of the eleven western Länder had its own central bank; their
presidents were members of the Zentralbankrat, as were the members of the Direktorium,
which could maximally consist of ten persons, including the president and the vice-
president of the Bundesbank. After unification, the number of Länder representatives was
reduced to nine, and the maximum total for the Direktorium, to eight (Smith, 1994).
17
the basis of a recommendation list containing three names, again
selected by the governing and supervisory boards. The royal commis-
sioner is responsible for supervision on behalf of the government; he is
also appointed by the minister (Dutch Bank Law, sections 30–31).
Table 1 summarizes the preceding analysis and indicates that central-
bank independence in a democratic society may be implemented in
different ways. According to the Maastricht Treaty, the ECB will
become responsible for monetary policy within EMU. An important
objection that has been raised to the ECB, however, is its lack of
accountability (Gormley and De Haan, 1996). Indeed, the statutes of
the ECB suggest that the democratic accountability of the ECB is
poorly arranged, compared with the accountability of the central banks
of the countries examined in this survey. This is true even in compari-
son with the Bundesbank, because the mandate of the ECB can only
be changed through an amendment of the Treaty, which requires
unanimity. By contrast, the Bundesbank must always take into account
the possibility of a change of the law. Because of this, the Bundesbank
will, in the long run, follow a policy that is in line with the preferences
of democratically elected politicians. In the Netherlands, political
approval is arranged differently, but the Nederlandsche Bank also
pursues policies that generally enjoy broad political and popular support.
Coordination of Policies (This section draws heavily from Pollard,
1993). In addition to the lack of democratic accountability, potential
problems in the coordination of economic policies have been put
forward as an important argument against central-bank independence.
Although most of the theoretical models discussed above make no clear
distinction between monetary and fiscal policy, some theoretical studies
concentrate on the conflicts that may arise when the government
controls fiscal policy and the central bank controls monetary policy.
Policymakers choose their own priorities concerning the goals for the
economy, and the government and the central bank may cooperate or
choose not to cooperate in implementing their policies. Andersen and
Schneider (1986) distinguish three models of the economy that address
this issue. In the first, Keynesian, model, even anticipated policy will
affect the level of output and inflation. In the second, “Keynesian–New
Classical” model, anticipated monetary policy is neutral; it can affect
only inflation. In the third, “New Classical,” model, anticipated mone-
tary policy and fiscal policy can affect inflation but cannot affect
output, and both the government and the central bank establish targets
for inflation and output. Andersen and Schneider compare the economic
outcomes under cooperation with those under noncooperation. Although
18
the equilibrium level of output and the rate of inflation vary depending
TABLE 1
A
LTERNATIVE APPROACHES TO CENTRAL-BANK INDEPENDENCE
AND
ACCOUNTABILITY
Deutsche
Bundesbank
Reserve Bank of
New Zealand
Nederlandsche
Bank
Policy objective
Price stability
Primary
objective
Sole
objective
If welfare
enhancing
Support of gov’t
economic policy
Secondary
objective
Government override Only implicit
(new law)
Provision in
current law
Right to give
instruction
Policy targets
Obligatory
No Yes No
As agreed with
government No Yes No
Escape clauses No Yes No
Time horizon No Three years No
Responsibility
Vested in
Central-bank
council
Governor of the
central bank
Minister
of finance
Monitoring Only implicit Dismissal of gov-
ernor for failure
By royal
commissioner
S
OURCES: Roll et al., “Independent and Accountable” (1993), and the Dutch Central
Bank Law of 1948.
on the model used, the cooperative solution in all three models is
Pareto superior to the noncooperative solution. This result is invariant,
moreover, to the structure of noncooperation, be it Nash or Stackelberg.
Andersen and Schneider (1986, p. 188) conclude that “two indepen-
dent policymakers do not automatically guarantee a policy outcome
which is preferred to other outcomes under different institutional
solutions.” Similar conclusions have been drawn by other authors
(Hughes Hallett and Petit, 1990; Blake and Westaway, 1993).
Several comments are in order. First, many of these models take no
account of a third “player,” that is, trade unions or the general public.
16
As we have already seen, the perception the public has of the credibility
16
This is not true for the model of Blake and Westaway (1993), which is similar to that
of Barro and Gordon (1983). The conclusion these authors reach is that “it is unlikely to
be sensible to appoint a monetary authority with an ability to make credible policy com-
19
of announced policies will affect macroeconomic outcomes. Second,
most of these studies do not examine the sustainability of fiscal policy.
As noted above, Sargent and Wallace (1981) have analyzed this issue,
showing that if the government embarks on a path of unsustainable
deficits, the central bank may eventually be forced to inflate to cover the
deficit. If the public realizes that government debt is on such a path, it
will expect inflation to increase, an expectation that may cause inflation
to increase well before a debt limit is reached. Third, uncertainty about
the macroeconomic models used by the policymakers may affect
conclusions with respect to the usefulness of cooperation. Frankel and
Rockett (1988) argue that, for the case in which the policymakers
cooperate, model uncertainty may eventually yield negative outcomes.
Fourth, many of the models referred to above equate central-bank
independence with noncooperation between the fiscal and monetary
authorities in policy implementation. This definition differs from the
concept underlying the empirical indices for independence that are
discussed in Chapter 3. As pointed out by Pollard (1993), differences in
the definitions of independence may partly explain the diverging results
of the theoretical models discussed above and the empirical studies
that will be reviewed in Chapter 4.
Debelle (1993) deals with some of the shortcomings of this literature
by differentiating between fiscal and monetary authorities in a model in
which he also distinguishes private-sector agents (labor and firms).
17
He shows that, in addition to affecting central-bank independence, the
objectives of the fiscal authorities also affect the inflation rate. Central-
bank independence in this model, as in many others, is defined as the
weight the central bank places on inflation relative to output (that is,
how “conservative” the central bank is). Central-bank autonomy may
reduce inflation, but it may also lead to lower social welfare, depending
upon society’s loss function.
18
In other words, the optimal degree of
mitments if at the same time it is following objectives which differ markedly from those
of government itself” (Blake and Westaway, p. 79).
17
The model draws on Alesina and Tabellini (1987). A similar model is presented by
Debelle and Fischer (1995).
18
Output is produced by labor, for which the nominal wage is predetermined; firms
maximize profits and can hire the amount of labor they demand at the predetermined
nominal wage. Social welfare is assumed to depend on inflation, the difference between
the actual and the natural rate of production, and the difference between the target and
actual level of government spending. The fiscal authorities have a similar loss function
with different weights. Government spending is not included in the loss function of the
monetary authorities. The simplest version of the model assumes that government
20
conservatism of the central bank depends on the society’s aversion to
inflation and output fluctuations.
spending can be financed only by seigniorage. It is clear why central-bank autonomy may
result in lower social welfare, because a more independent central bank will yield not
only lower inflation, but also a lower level of output and a lower level of government
spending.
21
3 MEASURES OF CENTRAL-BANK INDEPENDENCE
It is difficult to measure the degree of legal independence central
banks have, let alone the degree of their actual independence from
government. Cukierman (1992) has pointed out that actual, as opposed
to formal, independence hinges not only on legislation, but on a myriad
of other factors as well, such as informal arrangements with the gov-
ernment, the quality of bank personnel, and the personal characteristics
of key individuals at the bank. Because factors such as these are virtually
impossible to quantify, most research has focused on legal indepen-
dence; in addition, it is mainly restricted to the industrial countries. The
three sections below discuss four widely used legal measures of central-
bank independence, present a critical comparison of these indicators,
and review some nonlegal indicators of central-bank independence.
Legal Measures of Central-Bank Independence
Table 2 presents four measures of central-bank independence, as
developed by Alesina (1988, 1989), Grilli, Masciandaro, and Tabellini
(1991), Eijffinger and Schaling (1992, 1993a), and Cukierman (1992),
respectively. The higher the score is for the various indices, the more
independent will be the central bank. The measures of Alesina and
Eijffinger-Schaling range from 1 to 4, and 1 to 5, respectively. The
index used by Grilli, Masciandaro, and Tabellini is the sum of their
indicators for political and economic independence (see below for
further details) and ranges from 3 to 13. The value for their index of
political independence ranges from 0 to 6 and is shown in parentheses.
The index of Cukierman varies from 0 to 1.
Although the indicators are all based on a similar approach, they
sometimes show very different outcomes. According to the measure used
by Grilli, Masciandaro, and Tabellini, for example, the Greek central
bank has little autonomy; according to Cukierman’s (1992) index, it is
relatively independent. The remainder of this chapter briefly reviews
these indicators; Appendix A provides more detailed information.
The pioneering attempt by Bade and Parkin (1988) to codify the legal
independence of central banks has been extended by Alesina (1988,
1989). This index asks whether the central bank has final authority over
monetary policy, whether government officials sit on the governing
22
board of the bank, and whether more than half of the board members
TABLE 2
L
EGAL INDICES OF CENTRAL-BANK INDEPENDENCE
Country Alesina
Grilli, Masciandaro,
and Tabellini
Eijffinger-
Schaling
Cukierman
(LVAU)
Australia 1 9 (3) 1 0.31
Austria 9 (3) 3
a
0.58
Belgium 2 7 (1) 3 0.19
Canada 2 11 (4) 1 0.46
Denmark 2 8 (3) 4
a
0.47
Finland 2 3
a
0.27
France 2 7 (2) 2 0.28
Germany 4 13 (6) 5 0.66
Greece 4 (2) 0.51
Iceland 0.36
Ireland 7 (3) 0.39
Italy 1.5 5 (4) 2 0.22
Japan 3 6 (1) 3 0.16
Netherlands 2 10 (6) 4 0.42
New Zealand 1 3 (0) 3
a
0.27
Norway 2 2
a
0.14
Portugal 3 (1) 2
a
Spain 1 5 (2) 3
a
0.21
Sweden 2 2 0.27
Switzerland 4 12 (5) 5 0.68
United Kingdom 2 6 (1) 2 0.31
United States 3 12 (5) 3 0.51
N
OTE: The Grilli, Masciandaro, and Tabellini measure is the sum of the
indices for political and economic independence. Their index for political
independence alone is shown in parentheses. LVAU is the unweighted legal-
independence index.
a
Extensions are based on Eijffinger and Van Keulen, “Central Bank Inde-
pendence” (1995). Except for Denmark, the ranking of these seven countries
refers to central-bank laws adjusted during the last ten years.
are appointed by the government.
Grilli, Masciandaro, and Tabellini (1991) present indices of political
and economic independence. Their political-independence indicator
focuses on appointment procedures for board members, the length of
members’ terms to office, and the existence of the statutory requirement
to pursue monetary stability. Their economic-independence indicator
considers the extent to which the central bank is free from government
influence in implementing monetary policy. Generally, the total score for
both political and economic independence is employed as an indicator
for legal independence.
23
Eijffinger and Schaling (1992, 1993a) construct an index based on the
location of final responsibility for monetary policy, the absence or pre-
sence of a government official on the board of the central bank, and the
percentage of board appointees made by the government. Central-bank
laws in which the central bank is the final authority get a double score.
Cukierman (1992) and Cukierman, Webb, and Neyapti (1992) provide
an index that is aggregated from sixteen legal characteristics of central-
bank charters grouped into four clusters: the appointment, dismissal, and
legal term of office of the governor of the central bank; the institutional
location of the final authority for monetary policy and the procedures for
the resolution of conflicts between the government and the bank; the
importance of price stability in comparison to other objectives; and the
stringency and universality of limitations on the ability of the govern-
ment to borrow from the central bank.
A Comparison of Legal-Independence Measures
Although all four measures are similar in principle, they yield quite
different outcomes. This impression is confirmed by Table 3, which
shows Kendall’s rank-correlation coefficients of the various measures,
with the Spearman rank correlation shown in parentheses. Note,
especially, the low correlation of the measure of Grilli, Masciandaro, and
Tabellini to the Cukierman and the Eijffinger-Schaling indices.
At least two explanations can be given for these diverging outcomes.
First, the interpretation of the relevant bank laws differs. In general, one
can say that different rankings will occur for those countries with which
the author is most familiar. For instance, Alesina (1988, 1989) disagrees
with the Bade–Parkin ranking for Italy. This does not lead to a higher
ranking for the Banca d’Italia, however, but to a lower ranking. Eijffinger
and Schaling (1992) conclude that Alesina (1988, 1989) implicitly
includes a fourth criterion for Italy—namely, is the central bank obliged
to accommodate the government budget deficit? Alesina does not,
however, apply this criterion to the other countries in his sample.
In discussing the index of Grilli, Masciandaro, and Tabellini, Malinvaud
(1991) argues that the Banque de France has been given a higher degree
of independence than it actually merits, because the governor can be
removed at any time by decision of the French government. In a similar
vein, we have some doubts about Cukierman’s (1992) interpretation of
the Dutch central-bank law, with which we are most familiar. As
explained in Appendix A, we believe that the Nederlandsche Bank is
24
considerably more independent than Cukierman’s coding suggests.
19
TABLE 3
R
ANK-CORRELATION COEFFICIENTS OF INDICES OF CENTRAL-BANK INDEPENDENCE
Alesina
Grilli, Masciandaro,
and Tabellini
Eijffinger-
Schaling
Cukierman
(LVAU)
Alesina 1 0.58 (0.69) 0.71 (0.78) 0.38 (0.44)
Grilli, Masciandaro,
and Tabellini 1 0.36 (0.48) 0.52 (0.63)
Eijffinger-Schaling 1 0.20 (0.35)
Cukierman (LVAU) 1
N
OTE: Kendall rank-correlation coefficients are used, with Spearman rank-correlation
coefficients in parentheses. The Grilli, Masciandaro, and Tabellini measure is the sum of
the indices for political and economic independence. LVAU is the unweighted legal-
independence index.
A second reason for the diverging outcomes of various indicators is
that the measures focus on different aspects of central-bank indepen-
dence. Eijffinger and Schaling (1993a) criticize as “watered down” the
measure Grilli, Masciandaro, and Tabellini (1991) use, because the
rather large number of criteria that these authors apply erodes the
weight of the most important criteria, which are the central-bank
objectives and the appointing procedures. Cukierman’s indicator may
also be criticized in this respect. His aggregation procedures suggest
that the criteria we believe to be the most important in determining
central-bank autonomy (the variables in clusters 1 and 3) receive a
relatively low weight.
More generally, these indices may be compared to the four aspects
of central-bank independence outlined in Chapter 1. Table 4 shows
how the four previous indicators plus that of Bade and Parkin (1988)
focus on different aspects of central-bank independence. Because the
measures differ and show a low correlation, it is doubtful that a reliable
indicator for central-bank independence can be constructed from an
average of various independence measures—as, for instance, Alesina
and Summers (1993) and Fratianni and Huang (1994) have done.
19
The Dutch Bank Law is apparently difficult to interpret. Roll et al. (1993, p. 27)
are clearly wrong when they state that there is a “17-member Council that advises [the
finance] minister of guidelines that [the] Bank should follow in policy.”
25
Nonlegal Measures of Central-Bank Independence
TABLE 4
A
SPECTS OF CENTRAL-BANK INDEPENDENCE:ACOMPARISON OF FIVE INDICATORS
Measure Bade-Parkin Alesina
Grilli,
Masciandaro,
and Tabellini
Eijffinger-
Schaling
Cukiermans
(LVAU)
Maximum total score 4.00 4.00 16.00 5.00 1.00
Personnel independence 2.67 2.00 6.00 2.50 0.20
Financial independence 1.00 5.00 0.50
Policy independence 1.33 1.00 5.00 2.50 0.30
Goal independence 2.00 0.15
Instrument independence 3.00 0.15
Cukierman (1992) develops a measure for central-bank independence
on the basis of answers to a questionnaire under “qualified individuals
in various central banks.” He gives both an unweighted (QVAU) and a
weighted (QVAW) variant of this indicator. The questionnaire exam-
ined five issues: (1) legal aspects of independence; (2) actual practice
when it differs from the stipulation of the law; (3) monetary-policy
instruments and the agencies controlling them; (4) intermediate targets
and indicators, and (5) final objectives of monetary policy and their
relative importance. Unfortunately, the responding number of central
bankers was rather limited. Furthermore, an obvious methodological
drawback of the questionnaire is that central bankers may benefit from
providing a too positive impression of their independence. It is there-
fore doubtful that the personnel of central banks are the most appro-
priate recipients for a questionnaire on central-bank autonomy.
20
To
the best of our knowledge, however, a similar survey has never been
organized among other financial-market participants.
Cukierman (1992) and Cukierman, Webb, and Neyapti (1992) have
also developed a yardstick for central-bank autonomy based on the
actual average term of office of central-bank governors in a number of
countries from 1950 to 1989. This indicator is based on the presump-
tion that a higher turnover of central-bank governors indicates a lower
level of independence, at least above some threshold, and that even if
20
Indeed, one can argue that the difference between the legal-independence measure
and the indicator based on the questionnaire gives some impression of the degree to
which central bankers overrate their independence. For example, the score for Cukier-
man’s unweighted legal-independence index (LVAU) for Italy is 0.22, whereas the score
for QVAU is 0.76.
26
the central-bank law is quite explicit, it may not be operational if a
different tradition has precedence. A striking example is Argentina,
where the legal term of office of the central-bank governor is four
years, but where there is also an informal tradition that the governor
will resign whenever there is a change of government, or even a new
finance minister. Consequently, the actual average term of office of
Argentinean central-bank governors during the 1980s amounted to only
ten months. This example suggests that the turnover rate of central-
bank governors may be a good indicator for the degree of central-bank
autonomy.
21
Table 5 presents the average turnover rate of central-
bank governors for fifty-five industrial and developing countries during
the forty years ending in 1989. Two conclusions may be drawn from
the table. First, the turnover rate differs greatly across countries,
varying from 0.03, for Iceland, to 0.93, for Argentina. Second, the
average and standard deviations of the turnover rate for developing
countries are much higher than the corresponding measures for indus-
trial countries. The average turnover rate for the developing countries
is 0.28; the average for the industrial countries is 0.13. The highest
turnover rate in the industrial countries (excluding Turkey) is 0.2, for
Spain and Japan. This measure of autonomy therefore hardly discrimi-
nates between the central banks of the industrial countries.
Cukierman and Webb (1995) have gone one step further. They argue
that the frequency of transfers of central-bank governors reflects both
the frequency of political change (shifts in regime, for example, or in
the head of government) and the percentage of political changes that
are followed by changes in the governorship of the central bank. They
therefore develop an indicator of the political vulnerability of the
central bank, which is defined as the percentage of political transitions
that are followed within 6 months by the replacement of the central-
bank governor. For the period from 1950 to 1989, they calculate that
the average index of political vulnerability is 0.24 (0.10 for industrial
countries; 0.34 for developing countries). Again, one should be careful
in interpreting this index. De Haan (1995a) shows that the score of
0.10 for the Netherlands is the result of pure coincidence.
It follows from the foregoing analysis that existing indices of central-
bank independence are often incomplete and noisy indicators of actual
21
A long term in office may also indicate a low level of independence, because a
relatively subservient governor will tend to stay longer in office than will a governor who
stands up to the executive branch. Cukierman (1992) argues that this may be true for
countries with exceptionally low turnover rates, such as Denmark, Iceland, and the
United Kingdom.
27
independence. This does not mean that they are uninformative, but it
TABLE 5
T
HE TURNOVER RATE OF CENTRAL-BANK GOVERNORS, 1950–1989
Industrial Countries Developing Countries Developing Countries (contd.)
Belgium 0.13 Argentina 0.93 Philippines 0.13
Canada 0.10 Bahamas 0.19 Singapore 0.37
Denmark 0.05 Barbados 0.11 South Africa 0.10
Finland 0.13 Botswana 0.41 South Korea 0.43
France 0.15 Chili 0.45 Tanzania 0.13
Germany 0.10 Colombia 0.20 Thailand 0.20
Greece 0.18 Costa Rica 0.58 Uganda 0.34
Iceland 0.03 Egypt 0.31 Uruguay 0.38
Ireland 0.15 Ethiopia 0.20 Venezuela 0.30
Italy 0.08 Ghana 0.28 Zaire 0.23
Japan 0.20 Honduras 0.13 Zambia 0.38
Luxembourg 0.08 India 0.33 Zimbabwe 0.15
Netherlands 0.05 Israel 0.14
New Zealand 0.15 Kenya 0.17
Norway 0.08 Lebanon 0.19
Spain 0.20 Malaysia 0.13
Sweden 0.15 Malta 0.28
Switzerland 0.13 Mexico 0.15
Turkey 0.40 Nigeria 0.19
United Kingdom 0.10 Panama 0.24
United States 0.13 Peru 0.33
Average 0.13 Average 0.28
Standard deviation 0.08 Standard deviation 0.17
S
OURCE: Cukierman, Central Bank Strategy (1992). The (previously) Communist coun-
tries are not included.
does imply, as pointed out by Cukierman (1995), that their use should
be supplemented by judgment of the problem under consideration. In
particular, certain indices are more appropriate for some purposes than
for others. Measures of legal independence, for example, may be a
better proxy for independence in industrial countries than in develop-
ing countries. In some cases, various proxies that capture different
aspects of central-bank independence may be usefully combined to get
a more complete picture. Legal measures of central-bank independence
are more likely to be exogenous with respect to the economy, but
because they vary little over time, they are generally poor explainers of
developments in economic variables within countries. Most empirical
studies on the consequences of central-bank independence are there-
fore cross-sectional.
28
4 EMPIRICAL EVIDENCE ON THE CONSEQUENCES OF
CENTRAL-BANK INDEPENDENCE
This chapter examines the empirical evidence for a link between
central-bank independence and various economic variables such as
inflation and economic growth. Appendix B summarizes all the studies
of which we are aware that use one or more of the independence
indicators discussed in Chapter 3. One conclusion that follows from
this summary is that most of the studies are confined to the industrial
countries. Another insight is that many authors consider only one
measure of central-bank independence, so that the conclusions they
draw may be “measure specific.” Because the independence indicators
used focus on different aspects of central-bank independence, it is
important to use various indicators, even if the sample includes only
industrial countries. Some studies compare the findings when several
indicators are used, thereby examining the robustness of the empirical
results. As will be shown below, some outcomes do, indeed, depend on
which indicator is used for central-bank independence.
The Level and Variability of Inflation
The well-known inverse relation between central-bank independence
and the level of inflation is supported by most empirical studies (see
Table B2 for a summary of the conclusions reached). An exception is
Cargill (1995), who argues that this statistical association is not robust
and depends on the countries and periods included and on the regres-
sion specification. His argument is unconvincing, however, because he
uses only one measure of central-bank independence—and also because
he presents the outcomes of various model specifications without
analyzing which specification is to be preferred from an econometric
perspective. Furthermore, one would expect different results under
fixed and under floating exchange-rate regimes (see also Cukierman,
Rodriguez, and Webb, 1995). Under the Bretton Woods system of
fixed exchange rates, countries were committed to an exchange-rate
target and had little room to conduct an autonomous domestic mone-
tary policy. Thus, the relation between central-bank independence and
inflation is likely to be much less straightforward for the period before
1973. Regression analyses by Grilli, Masciandaro, and Tabellini (1991)
29
and De Haan and Sturm (1994a) support this view.
22
Indeed, one can
argue that if no evidence of a relation between independence and
inflation is found in the Bretton Woods period, although such a link is
found for the post–Bretton Woods era, the argument that central-bank
independence is a primary determinant of a country’s inflation perfor-
mance will be strengthened (Pollard, 1993). Still, many authors lack
precision in distinguishing between exchange-rate regimes.
Despite the overwhelming evidence in support of a negative relation
between central-bank independence and inflation, it should be noted
that a negative correlation does not necessarily imply causation. The
correlation between the variables can perhaps be explained by a third
factor (for example, the culture and tradition of monetary stability in a
country) that explains both an independent central bank and low
inflation.
23
Similarly, there may be a two-way causality between infla-
tion and central-bank independence. It is likely that less independence
contributes to higher inflation. But high inflation may also affect
independence. As will be explained in more detail in Chapter 5, it can
be argued that high inflation leads to more or to less central-bank
independence. On the one hand, high inflation may lead to political
pressure for low inflation; on the other, it may encourage processes
that make it easier for the government to influence monetary policy,
thereby reducing actual independence. Most studies do not address this
issue of two-way causality. Cukierman (1992) and Cukierman, Webb,
and Neyapti (1992) deal with this issue by using two-stage least-squares
and instrumental variables. They conclude that there is a vicious circle
between inflation and low levels of independence. When sufficiently
sustained, inflation erodes central-bank independence. Then, low
independence contributes to higher inflation. De Haan and Van ’t Hag
(1995) conclude, however, that high levels of inflation in the past led
eventually to more central-bank independence.
22
Similarly, one would expect that the Exchange Rate Mechanism (ERM) of the EMS
might affect outcomes. Ungerer (1990) characterizes the first phase of the EMS (1979 to
1982) as a period of “initial orientation” full of frequent and, sometimes, large realign-
ments of central rates. From 1982 onward, however, the EMS entered a second phase of
“consolidation” (1982 to 1987), and after the accord of Basle-Nyborg, it moved into a
third phase of “reexamination” (1987 to the present). The negative correlation between
central-bank independence and inflation is thus expected to be less clear-cut during the
second and third subperiods than during the first, because the priority EMS countries
initially gave to exchange-rate stability was not given subsequently. Regression analyses
by Eijffinger and Schaling (1993b) and De Haan and Eijffinger (1994) support this view.
23
The standard example is the case of Germany, where the hyperinflation in the
1920s led to a culture and tradition of monetary stability (Bresciani–Turroni, 1953).
30
There is evidence that inflation and political instability are positively
associated. Some recent studies show that even if some measure of
political instability is included in the regression, proxies for central-
bank independence, such as the turnover rate of central-bank gover-
nors or the political vulnerability of the central bank, remain signifi-
cantly related to inflation (De Haan and Siermann, 1994; Cukierman
and Webb, 1995).
As mentioned, most of the studies summarized in Appendix B are
confined to industrial countries, although developing countries are
included in the studies of Cukierman (1992), Cukierman, Webb, and
Neyapti (1992), Cukierman et al. (1993), De Haan and Siermann
(1994), and Cukierman and Webb (1995). For the developing countries
considered, these authors find no significant link between inflation and
the legal independence of the central banks. They believe that this is
because the developing countries have “less regard for the law”; indus-
trial countries, by contrast, exhibit an inverse relation between the
legal-independence measure and inflation. If the turnover rate of
central-bank governors is used as a measure of actual independence in
the developing countries, however, a significant, negative relationship
appears for the developing countries as well.
24
Similar results are
found by Cukierman and Webb (1995), using the political vulnerability
of the central bank as an indicator for central-bank independence.
Most empirical studies consist of simple cross-sectional estimates in
which the average inflation rate is “explained” by some measure of
central-bank independence. Some authors, however, include additional
explanatory variables. Grilli, Masciandaro, and Tabellini (1991) and De
Haan and Sturm (1994a), for example, include some indicators for
political instability. This does not rob the coefficients of the indepen-
dence indicators of their significance. Similarly, Havrilesky and Granato
(1993) and Bleaney (1996) take wage-bargaining structures into account,
and Al-Marhubi and Willett (1995), employ indicators for openness,
degree of exchange-rate fixity, and budget deficits. Once again, the
coefficients of the various indicators for central-bank independence
remain significant.
Although many empirical studies examine the relation between
central-bank independence and inflation, only two studies try to differ-
entiate between the various aspects of central-bank autonomy. Debelle
and Fischer (1995) decompose the independence measure of Grilli,
24
Although there are only twenty-two observations available for the 1980s, the
independence index based on the questionnaire with central bankers also has the
predicted sign and is statistically very significant.
31
Masciandaro, and Tabellini (1991) into independence with respect to
goals, personnel, and instruments.
25
They conclude that lack of goal
independence (that is, a mandate to maintain price stability) and
instrument independence are most closely tied to inflation perfor-
mance, whereas personnel independence is not significantly related to
inflation. Similarly, De Haan (1995b) has decomposed the legal-inde-
pendence measure of Cukierman (1992) and relates its components to
inflation. Using pooled time-series and cross-sectional data for a sample
of twenty-one industrial countries, he concludes that only indepen-
dence with respect to instruments matters for inflation performance.
26
Table 6 reproduces some results from these studies.
As explained above, most empirical studies on the relation between
central-bank independence and inflation consist of cross-sectional
regressions. Other studies, however, follow a somewhat different
approach. Capie, Mills, and Wood (1994) investigate the relation
between the level of inflation and central-bank independence for twelve
countries: Austria(-Hungary), Belgium, Brazil, Canada, England (the
United Kingdom), France, (West-) Germany, India, Italy, Japan, New
Zealand, Spain, Sweden, and the United States. Based on the degree of
policy influence, beginning between 1871 and 1916 and ending in 1987,
central banks in these countries are classified as “independent,” “depen-
dent,” or “unclassified.” During all periods—before World War I, during
the interwar period, and during and after the Bretton Woods system—
the nations with independent central banks have continuously been in
the group of countries with low inflation. Sometimes this group has also
included countries with dependent central banks. Capie, Mills, and
Wood conclude, therefore, that independence is a sufficient, but not a
necessary, condition for low inflation.
25
Goal independence is measured as the presence of a statutory requirement that the
central bank pursue monetary stability among its goals. Personnel independence is
measured as the Grilli–Masciandaro–Tabellini index of political independence, excluding
the statutory requirement. Instrument independence is the Grilli–Masciandaro–Tabellini
index of economic independence minus the bank-supervision criterion.
26
The proxy for personnel independence is the sum of all variables in the first cluster
of variables distinguished by Cukierman (1992)—and explained in some detail in
Chapter 3. The proxy for instrument independence is the sum of the variables in the
second cluster, except for whether the central bank has an active role in the formulation
of the government’s budget, which has little to do with central-bank independence. Goal
independence is Cukierman’s (1992) score for the third cluster; according to Cukierman
(1992, p. 377), “it proxies the . . . independence of the CB to elevate the target of price
stability above other objectives. In Rogoff’s terminology, it measures how strong is the
‘conservative bias’ of the CB as embodied in the law.” Financial independence is proxied
by the sum of most variables in the fourth cluster, as discerned by Cukierman (1992).
32
Johnson and Siklos (1994) use reaction functions of central banks,
TABLE 6
R
EGRESSIONS OF INFLATION ON ASPECTS OF CENTRAL-BANK INDEPENDENCE
Study
Goal
Independence
Instrument
Independence
Personnel
Independence
Financial
Independence
Debelle and
Fischer (1995)
−1.76
(1.02)
−1.02
(2.04)
a
−0.41
(0.91)
−2.28
(1.42)
−1.02
(2.04)
a
——
De Haan (1995b) 2.33
(1.12)
−2.27
(−2.01)
a
−0.16
(−0.26)
−0.58
(−0.95)
2.13
(1.10)
−3.02
(−3.83)
b
−0.36
(−0.58)
N
OTE: t-statistics are in parentheses.
a
Denotes significance at the 5 percent level.
b
Denotes significance at the 1 percent level.
with the money-market interest rate as the policy variable. If central-
bank independence can be measured by the change in interest rates,
little difference can be discerned across the central banks considered.
The analysis of Johnson and Siklos covers seventeen industrial coun-
tries from 1960 to 1990.
Eijffinger, Van Rooij, and Schaling (1994) apply a panel-data ap-
proach to the reaction functions of the central banks of ten industrial
countries for the 1977–90 period, using present and past inflation and
real economic growth as the explanatory variables of changes in money-
market rates. They find a country-specific factor that they interpret as
the degree of empirical central-bank independence. Regression analysis
of average inflation on their empirical index (EMP) of central-bank
independence confirms that having an independent central bank will
lead to lower inflation.
The main conclusion from empirical studies on the relation between
central-bank autonomy and inflation is that central-bank independence
is inversely related to the level of inflation in both the industrial and
the developing countries. The independence measures used to reach
this conclusion, however, differ between the two groups of countries.
Although legal independence is a good proxy for actual autonomy in
industrial countries, proxies such as the turnover rate of the central-
bank governorship or the political vulnerability of the central bank
should be used for developing countries.
33
What is the empirical relation between central-bank independence
and the variability of inflation? Inflation variability is positively corre-
lated with the level of inflation (Chowdhury, 1991). Consequently, if a
high degree of central-bank independence results in lower levels of
inflation, greater independence may also lead to less variability of
inflation. Indeed, as shown in Appendix B, many authors conclude that
the variability of inflation—generally measured as the standard devia-
tion of inflation—shows an inverse relation to central-bank indepen-
dence. This concurs with Rogoff’s (1985) model.
There is another way to consider the impact of central-bank inde-
pendence on the variability of inflation. As shown in Chapter 2, parti-
san considerations will lead to inflation variability if the government
changes regularly and if the monetary authorities are dominated by
elected politicians. In contrast, a relatively independent central bank
will not change its policy after a new government has been elected.
Central-bank independence may therefore reduce inflation variability
over longer periods of time. Table 7 updates and extends a table
provided by Alesina (1988), in which he analyzes differences in the
influence central-bank independence has on inflation under “right-
wing” and “left-wing” governments. Only those countries that during
the 1980s witnessed a change in government in which a right-wing
government replaced a left-wing government, or vice versa, and in
which a meaningful comparison is possible, are included in Table 7.
The first three rows give an update of the data provided by Alesina
(1988), who concludes that inflation shows little variation between
governments with different political orientations in countries with
relatively independent central banks. The lower part of the table
presents our extension to Australia, New Zealand, and Norway. It
follows that Alesina’s conclusion depends strongly on the specific
countries included in the analysis. When Australia, New Zealand, and
Norway are also included, the results are less clear-cut. First, the
inflation rate is lower under “left-wing” governments in these three
countries than under “right-wing” governments. Second, the inflation
differentials seem hardly to be related to central-bank independence.
Norway, for instance, has a relatively dependent central bank, accord-
ing to the index Cukierman (1992) uses, but its inflation differentials
are similar to those of Australia and New Zealand.
Economic Growth and Disinflation Costs
The foregoing analysis put forward two reasons why central-bank
independence may stimulate economic growth in the long run: less
34
TABLE 7
A
VERAGE INFLATION IN SIX INDUSTRIAL COUNTRIES UNDER “LEFT-WING AND “RIGHT-WING”GOVERNMENTS
Country
Government,
Period,
and Percent Inflation
Percent Differ-
ence between
Governments
Measures of Central-
Bank Independence
GMT LVAU
Australia Frazer (Liberal)
(1976–83): 10.3
Hawk (Labor)
(1984–90): 7.2 3.1 9 0.31
Germany Social Democrats
(1975–82): 4.8
Christian Democrats
(1983–92): 2.4 2.4 13 0.66
New Zealand Muldoon (National Party)
(1976–84): 13.6
Lange/Palmer/Moore (Labor)
(1985–90): 10.5 3.1 9 0.27
Norway Willoch (Conservative)
(1982–86): 7.9
Brundlandt (Social Democrat)
(1987–92): 4.9 3.0 0.14
United Kingdom Labor
(1975–79): 15.4
Conservatives
(1980–92): 6.8 8.6 6 0.31
United States Carter (Democrat)
(1977–80): 8.4
Reagan (Republican)
(1981–92): 4.7
3.7 12 0.51
N
OTE: GMT = Grilli, Masciandaro, and Tabellini. LVAU = Cukierman’s unweighted legal-independence index; the higher the
score, the more independent the central bank is.
uncertainty about inflation and better functioning of the price mecha-
nism. Empirical research by Grimes (1991), Fischer (1993), and Barro
(1995) suggests that inflation reduces economic growth.
27
This may be
explained by the positive correlation between the level and variability
of inflation. Greater variation in the rate of inflation may imply increas-
ing uncertainty about inflation and may thereby lead to lower economic
growth. This relation between inflation variability and economic
growth, however, is not supported by most studies. Logue and Sweeney
(1981) find no significant influence of inflation variability on real
growth rates. Jansen (1989) draws the same conclusion.
Various studies have asked directly whether central-bank indepen-
dence is related to economic growth. A summary of these studies
(Appendix B) shows that various authors conclude that central-bank
independence is not related to economic growth (or to unemployment).
Despite the association of a high degree of central-bank independence
with lower inflation in the long run, a policy of disinflation is apparently
not associated with high costs or great benefits in terms of long-term
economic growth. Indeed, it is tempting to conclude that the absence
of a long-term trade-off between inflation and growth implies that the
establishment of central-bank independence is a free lunch. Recall,
however, that price stability is generally regarded as an essential
condition for sustainable economic growth and that central-bank
independence should accordingly lead to a higher level of economic
growth. From this point of view, lack of a significant, positive relation
between growth and independence would be disappointing. Two
studies exist, however, that report a positive relation between central-
bank autonomy and economic growth. De Long and Summers (1992)
find a positive relation between central-bank independence and gross
domestic product (GDP) per worker for their sample of industrial
countries. Cukierman et al. (1993) find that economic growth is not
correlated with central-bank independence in the industrial countries,
even after corrections are made for other factors that may influence
growth. They find that it is positively correlated with growth in the
developing countries, however, if the frequency of changes of central-
bank presidents is used as a proxy for independence.
A next question, of course, is whether a relationship exists between
central-bank independence and the variation of economic growth.
Alesina and Summers (1993) argue that an autonomous central bank
will be less inclined to conduct a stop-and-go policy, which may limit
27
A recent study by Karras (1993), however, contradicts this conclusion.
36
fluctuations in economic growth. Most researchers find that a higher
degree of central-bank independence is generally not associated with
greater variation of real economic growth rates (see Appendix B).
The greater credibility attributed to independent central banks is
often thought to reduce the costs of subsequent policies designed to cut
inflation. Increased central-bank independence tends to shift the short-
term Phillips curve inward to the origin. Walsh (1995a) has pointed
out, however, that central-bank independence may also affect the slope
of the Phillips curve. If, for instance, independent central banks foster
an economic environment that produces nominal-wage contracts of
longer duration or with less indexation because inflation is less variable,
nominal rigidities in the economy will increase, thereby flattening the
slope of the Phillips curve. The effect on the slope of the short-term
trade-off between unemployment and inflation will raise the real
economic costs of a policy to diminish inflation. This can, in turn,
reduce, and potentially offset, the reduced costs of disinflation attributed
to the gain in credibility that comes with increased independence.
Another reason why there may be a positive relation between central-
bank independence and the costs of disinflation has been put forward
by Gärtner (1995). In order to keep disinflation costs at a minimum in
a framework of staggered wage contracts and relative-wage consider-
ations, disinflation must start slowly and accelerate only as the bulk of
wage contracts has been renegotiated (see also Taylor, 1983). Because
more independent central banks are likely to disinflate faster, they will
face higher disinflation costs.
The net effect of independence on the costs of disinflation is an
empirical question. De Haan, Knot, and Sturm (1993) ask whether
central-bank independence reduces disinflation costs (measured as the
cumulated unemployment rate over the 1980–89 period relative to its
average level for 1973–79); they find no supporting evidence that it
does. Four more recent studies also analyze the effects of central-bank
independence on the costs of disinflation. Debelle and Fischer (1995)
compare the output costs of recessions in Germany and the United
States and find that the costs are similar. The sacrifice ratio (output
lost because of the inflation reduction) in Germany is larger than in the
United States for all recent recessions, despite the widely assumed
“credibility bonus” of the Bundesbank. Debelle and Fischer also report
a positive and significant relation between the Grilli-Masciandaro-
Tabellini index of central-bank independence and output losses. Similar
results are reported by Gärtner (1995) and Andreas Fischer (1996),
37
who use other indicators as well for central-bank independence.
28
Posen (1994) also concludes that the costs of disinflation are not lower
in countries with independent central banks, even when differences in
contracting behavior are taken into account. All this evidence implies
that output losses suffered during recessions have, on average, been
larger as the independence of the central bank increases. As Debelle
and Fischer argue, this suggests that no credibility bonus exists in the
labor markets for more independent central banks: the banks have to
prove their toughness by continually being tough. Similar results have
been reported by Walsh (1995a) for various EU member states. European
Union countries with greater central-bank independence also appear to
face higher costs of disinflation. Walsh has pointed out that this positive
correlation could arise because inflation is more costly to reduce at lower
levels of inflation, and central-bank independence is associated with lower
levels of inflation. Walsh also reports evidence, however, that even after
controlling for average inflation in a cross-sectional regression of EU
countries, the relation between the trade-off parameter and central-bank
independence is positive and significant. According to Walsh, this
evidence suggests that the greater central-bank independence required
under the Maastricht Treaty may lead to a rise within EMU of the costs
associated with policies designed to reduce inflation. Still, one should
bear in mind that the causality may run in the other direction. Perhaps
countries with flat short-term aggregate supply curves will be more likely
to establish independent central banks. Walsh has noticed that flat supply
curves make disinflation more costly, but they also raise the temptation
to stimulate the economy and thus increase the inflationary bias of
discretionary policy. This issue will be taken up in Chapter 5.
Other Variables
It follows from the preceding analysis that greater central-bank inde-
pendence is associated with lower inflation rates. Through the Mundell-
Tobin effect, this may result in higher (ex post) real interest rates. De
Haan and Sturm (1994a) find some limited support for the Mundell-
Tobin effect: low inflation countries usually have high (ex post) real
interest rates. It can also be argued, however, that more independence
dampens inflation uncertainty and, through the Mascaro–Meltzer (1983)
28
Gärtner’s results differ from Taylor’s (1983), however, because they show no
relation between central-bank independence and the speed of disinflation. Fischer
reports that independent central banks deflate more slowly.
38
effect, brings down (ex post) real interest rates.
29
These opposite effects
on the real interest rate might also explain why the net effect on
economic growth turns out to be insignificant. Alesina and Summers
(1993) examine the link between central-bank independence and the (ex
post) real interest rate and find no clear relationship. Nevertheless, these
authors discover a negative correlation between central-bank inde-
pendence and the variability of (ex post) real interest rates.
Is there, finally, some relation between central-bank independence
and (the monetary accommodation of) government budget deficits? One
would expect that an independent central bank is in a better position to
resist the pressure of its government to accommodate budget deficits by
means of monetary financing. Moreover, the government has a strong
incentive, when budget deficits are financed on the capital market, to
reduce the deficit so as to minimize future interest payments.
Parkin (1987) concludes that Germany and Switzerland, the two
countries in his sample with the most independent central banks, appear
to have almost no government deficits in the period under consideration.
Masciandaro and Tabellini (1988) rate Australia, Canada, Japan, New
Zealand, and the United States, taking the budget deficit as a ratio of
gross national product (GNP) for the period from 1970 to 1985. They
find that New Zealand, with (until recently) the least independent
central bank, has the highest average deficit over this period, whereas
the United States, with (according to Masciandaro and Tabellini) the
most independent central bank, has a deficit equal to that of the
remaining three countries.
Grilli, Masciandaro, and Tabellini (1991) also find for their measure
a negative correlation between the deficit and the degree of indepen-
dence from 1950 to 1989, although it is not significant. Their results are
supported by De Haan and Sturm (1994a).
30
Pollard (1993) finds a
negative correlation between central-bank independence and the deficit-
to-GDP ratio that also appears not to be significant. Quite strangely,
however, Pollard discovers a significant, negative relation between inde-
pendence and the variance of the budget deficit as a percentage of GDP.
29
According to Mascaro and Meltzer (1983), monetary and inflationary uncertainty,
measured by the variability of (unexpected) money growth and inflation, respectively, will
result in a risk premium that risk-averse investors will demand in order to compensate for
uncertainty and, thereby, in a higher (ex ante) real interest rate (see also Bomhoff, 1983).
30
Note that in case of the Grilli-Masciandaro-Tabellini index, there is the danger of
circular reasoning because their (modified) index comprises at least four elements of
monetary accommodation of government deficits. The empirical evidence found by Grilli,
Masciandaro, and Tabellini and De Haan and Sturm is therefore not at all surprising.
39
Our prudent conclusion is that an independent central bank cannot
restrain its government from creating budget deficits, but that it may
have some restrictive influence on the fiscal policies pursued by its
government.
What should we make of the preceding review of empirical studies on
the relation between central-bank independence and macroeconomic
variables such as inflation and growth? Is it true, as Grilli, Masciandaro,
and Tabellini (1991, p. 375) claim, that “having an independent bank is
almost like having a free lunch; there are benefits but no apparent costs
in terms of macroeconomic performance”? Although overwhelming evi-
dence exists that central-bank independence and inflation are negatively
related, one should be careful in jumping to this conclusion. As has been
pointed out, only limited support exists for the view that central-bank
independence stimulates economic growth and that it does not reduce
disinflation costs. Furthermore, central-bank independence may be
endogenous, in the sense that countries with a commitment to price
stability may have a greater propensity for central-bank independence.
If true, the mere establishment of a central bank with a commitment to
price stability will not bring inflation benefits to a country. The following
chapter analyzes the determinants of central-bank independence.
40
5 THE DETERMINANTS OF CENTRAL-BANK
INDEPENDENCE
The previous chapters have shown that the degree of central-bank
independence varies considerably among the industrial countries. The
question is which factors ultimately determine the degree of central-
bank independence. It is remarkable that a literature dealing with this
issue has hardly developed. Before discussing some of the determinants
of central-bank independence in greater detail, we shall first review
recently developed theory.
Cukierman (1994) presumes that the delegation of monetary policy to
(partly) independent central banks is used as a “(partial) commitment
device.” By specifying the objectives of the central bank more or less
tightly and by giving the bank broader or narrower powers, politicians
determine the extent of their commitment to a policy rule. Such policy
action leads to greater credibility of monetary policy, which, in turn, is
reflected in lower inflationary expectations and, thereby, lower (capital-
market) interest rates and more moderate wage demands. From the
politician’s point of view, the costs of an independent central bank
consist mainly of the loss of flexibility in monetary policymaking. The
balance between flexibility and credibility determines the optimal
degree of central-bank autonomy in a country. Based on these or other
theoretical considerations, various economic and political determinants
of central-bank independence have been formulated. These determinants
may be categorized as follows (note that they are not mutually exclusive
and may partly overlap). They are (1) the equilibrium or natural rate of
unemployment; (2) the stock of government debt; (3) political instability;
(4) the supervision of financial institutions; (5) financial opposition to
inflation; (6) public opposition to inflation; and (7) other determinants.
Table 8 summarizes empirical studies on the determinants of central-
bank independence. The second column shows the measure(s) of
independence used. The third and fourth columns present the sample
of countries and the estimation period, respectively. The last column
contains the economic and political variables examined in the studies.
The Equilibrium Level of Unemployment
The first determinant of central-bank independence may be the average
employment-motivated inflationary bias in a country. This inflationary
41
TABLE 8
E
MPIRICAL STUDIES ON THE DETERMINANTS OF CENTRAL-BANK INDEPENDENCE
Study Measure(s) Used Countries Estimation Period Variables Examined
Cukierman (1992) LVAU; LVAW 14 middle-income 1972–79;
1980–89
Political instability (party and regime)
Posen (1993a) LVAU 17 industrial 1950–89 Financial opposition to inflation
De Haan and
Siermann (1994)
TOR 43 developing 1950–89;
subperiods
Political instability (party and regime)
Moser (1994) Average of
GMT and LVAW
22 industrial 1967–90 Political system index (PSI); standard devia-
tion of output growth
Cukierman and
Webb (1995)
Political vul-
nerability
64 industrial and
developing 1950–89
Four types of political instability (high- and
low-level)
De Haan and
Van ’t Hag (1995)
GMT; LVAU;
SUMLV
19 (16) industrial
21 (18) industrial
17 industrial
16 (13) industrial
1980–88
1980–89
1950–98
1900–40
NAIRU; government debt ratio; frequency
of (significant) government changes;
banking supervision; universal banking;
very long-term inflation
Eijffinger and
Schaling (1995)
AL; ES; GMT;
LVAU (latent-
variables method)
19 industrial 1960–93
(for NAIRU:
1960–88)
NAIRU; relative number of years of socia-
list (left-wing) government; variance of
output growth; compensation of employees
paid by resident producers
N
OTE: AL = Alesina; BP = Bade-Parkin; ES = Eijffinger-Schaling; GMT = Grilli, Masciandaro, and Tabellini; LVAU = Cukierman’s
unweighted legal-independence index; LVAW = Cukierman’s weighted legal-independence index; QVAW = Cukierman’s weighted index based
on a questionnaire; SUMLV = Cukierman’s sum of sixteen legal variables; TOR = the turnover rate of central-bank governors.
bias can be approximated empirically by the equilibrium or natural rate
of unemployment.
31
Cukierman (1994) shows that the larger the aver-
age employment-motivated inflationary bias in a country, the higher
will be the costs for the government to override the central bank and
the more independent the central bank will be.
32
Because in the case
of nominal-wage contracts, unexpected inflation has positive effects on
the levels of both production and employment, a higher equilibrium or
natural rate of unemployment implies that surprise inflation is more
valuable for the government.
33
De Haan and Van ’t Hag (1995) have tested this hypothesis, using
two measures of Cukierman (LVAU and SUMLV)
34
and the index of
Grilli, Masciandaro, and Tabellini (GMT). Proxies for inflationary bias
are the equilibrium rate of unemployment, as estimated by Layard,
Nickell, and Jackman (1991) for nineteen industrial countries, and the
difference between the actual and the equilibrium rate of unemployment
during the 1980s. In simple cross-country regressions with each measure
of central-bank independence as a dependent variable, the coefficients
of both proxies prove to be insignificant. Eijffinger and Schaling (1995)
employ a latent-variables method to distinguish between the actual
(legal) and optimal degree of central-bank independence in these coun-
tries. As measures for actual central-bank autonomy, the indices of
Alesina (AL), Grilli, Masciandaro, and Tabellini (GMT), Eijffinger and
Schaling (ES), and Cukierman (LVAU) are chosen. Eijffinger and
Schaling also find an insignificant coefficient for the natural rate of
31
In this case, the natural rate of unemployment is referred to as the
“nonaccelerating inflation rate of unemployment” (NAIRU). This implies that the desired
unemployment rate is being held constant and, thus, that the inflationary bias is driven
by the difference between the desired and natural unemployment rate.
32
Similarly, Eijffinger and Schaling (1995) suggest that the higher the natural rate of
unemployment, the higher will be the optimal degree of central-bank independence. The
intuition behind this proposition is as follows. A higher natural rate of unemployment
leads to a higher time-consistent rate of inflation and, consequently, to an increase in
society’s credibility problem. Hence, with an unaltered relative weight placed on inflation
stabilization, as opposed to unemployment stabilization, the monetary authorities’ commit-
ment to fighting inflation will now be too low to be effective.
33
Schaling (1995) gives an analysis with an endogenous NAIRU and its implications
for the optimal degree of central-bank independence.
34
The index SUMLV measures the total score of sixteen legal variables of Cukierman
(1992) with respect to the appointment, dismissal, and term of office of the central-bank
president; the solution for conflicts between the government and the central bank; the
policy goals of the central bank; and the legal limitations for the government to borrow
from the central bank.
43
unemployment. We may therefore conclude that empirical studies
provide no support for any relation between the equilibrium or natural
rate of unemployment and the degree of central-bank independence.
Government Debt
The stock of government debt is another potential determinant of
central-bank independence. The larger the sum the government wants
to borrow on the capital market, the more weight will be placed on lower
inflationary expectations and, thus, on lower nominal capital-market
interest rates. The benefits of a once-and-for-all reduction of the real
value of government debt by unexpected inflation do not outweigh (in
this case) the costs of permanently higher interest payments as a
consequence of lower credibility. Cukierman (1994) argues that the larger
the debt, the more likely it is that politicians will delegate authority to
the central bank and the more independent the central bank will be. This
hypothesis is empirically investigated by De Haan and Van ’t Hag (1995)
for several measures of independence (LVAU, SUMLV, and GMT) for
the 1980–89 period. Using gross government debt as a percentage of
GDP in their regression analysis, these authors find no significant
coefficient for the debt ratio.
Political Instability
The influence of political instability on central-bank independence is,
at first sight, less obvious than the impact of the other factors discussed
above.
35
It can be argued that when politicians in office are faced
with a greater probability that they will be removed from office, they
have a stronger interest in delegating authority to the central bank (an
apolitical institution) in order to restrict the range of policy actions
available to the opposition should the latter come into office. This
implies that greater political instability leads to a more independent
central bank. Conversely, it can be argued that the incumbent politi-
cians will fortify their hold on the central bank if there is a greater
probability of government change and will eventually overrule central-
bank decisionmaking. The short-term benefits of surprise inflation may
thereby exceed their long-term costs. It follows that greater political
instability will result in a more dependent central bank.
35
For the effect of political instability on variables such as (the increase of) the stock
of government debt and seigniorage, see Persson and Svensson (1989), Alesina and
Tabellini (1990), Tabellini and Alesina (1990), Cukierman, Edwards, and Tabellini (1992)
and De Haan and Sturm (1994b).
44
Cukierman (1992) argues that it is possible to combine both hypothe-
ses into a single, internally consistent theory. In countries with a
sufficiently high degree of national consensus, greater political instability
may be associated with increased independence of the central bank,
whereas the reverse may apply for countries with a relatively low level
of national consensus. Cukierman has tested this combined hypothesis,
using two indices of political instability constructed by Haggard et al.
(1991) for fourteen middle-income countries during the 1970s and
1980s. The first index, party political instability, measures the degree
of political instability under a given regime and refers to a relatively high
level of national consensus. The second index, regime political instability,
reflects the degree of political instability in the case of a relatively low
level of national consensus. Regression analysis by Cukierman for legal-
independence measures during the 1972–79 and 1980–89 periods
shows that the indices have the expected signs. This result may be
questioned, however, because legal measures of central-bank indepen-
dence may not be a very good proxy for actual central-bank indepen-
dence in developing countries. Two studies have recently employed
nonlegal measures of central-bank independence.
Cukierman and Webb (1995) use a measure of political vulnerability,
the percentage of times that political transition is followed by a change
of central-bank governor, as a dependent variable, and four types of
political instability as explanatory variables for a mixture of industrial
and developing countries during the 1950–89 period. Only high-level
political instability (a change in regime) and the dummy for developing
countries prove to be significant.
De Haan