Download as pdf or txt
Download as pdf or txt
You are on page 1of 37

Journal of Economic Literature 2008, 46:4, 910–945

http:www.aeaweb.org/articles.php?doi=10.1257/jel.46.4.910

Central Bank Communication and


Monetary Policy: A Survey
of Theory and Evidence
Alan S. Blinder, Michael Ehrmann, Marcel Fratzscher,
Jakob De Haan, and David-Jan Jansen*

Over the last two decades, communication has become an increasingly important
aspect of monetary policy. These real-world developments have spawned a huge new
scholarly literature on central bank communication—mostly empirical, and almost
all of it written in this decade. We survey this ever-growing literature. The evidence
suggests that communication can be an important and powerful part of the central
bank’s toolkit since it has the ability to move financial markets, to enhance the pre-
dictability of monetary policy decisions, and potentially to help achieve central banks’
macroeconomic objectives. However, the large variation in communication strategies
across central banks suggests that a consensus has yet to emerge on what constitutes
an optimal communication strategy.

1.  A Revolution in Thinking and Practice to this art and its proper execution is confined
to the initiated elite. The esoteric nature of the

P rior to the 1990s, central banks were art is moreover revealed by an inherent impos-
sibility to articulate its insights in explicit and
shrouded in mystery—and believed they
intelligible words and sentences.
should be. Conventional wisdom in central
banking circles held that monetary policy-
makers should say as little as possible, and say Fifteen years later, in his 1996 Robbins
it cryptically. In 1981, Karl Brunner (1981, lectures at the London School of Economics,
p. 5) wrote, with evident sarcasm: one of the authors of this paper (Alan S.
Blinder 1998, pp. 70–72) expressed a view of
Central Banking . . . thrives on a pervasive
impression that [it] . . . is an esoteric art. Access what central bank communications should
be—one that had been lurking around in the

*  Blinder: Princeton University. Ehrmann: Euro- Eric Swanson, Charles Wyplosz, and the editor (Roger
pean Central Bank. Fratzscher: European Central Bank. Gordon) and two anonymous referees of this journal for
De Haan: University of Groningen and CESifo. Jansen: valuable comments on earlier drafts. Views expressed in
De Nederlandsche Bank. The authors are grateful to this article do not necessarily coincide with those of the
Sylvester Eijffinger, Gabriel Fagan, Andreas Fischer, Otmar European Central Bank, de Nederlandsche Bank, or the
Issing, Frederic Mishkin, Glenn Rudebusch, Pierre Siklos, Eurosystem.

910
Blinder et al.: Central Bank Communication and Monetary Policy 911

underbrush but was far from mainstream at These new ideas have made a mark on
the time:1 central bank practice as well. At the Federal
Greater openness might actually improve the Reserve, for example, then Chairman Alan
efficiency of monetary policy . . . [because] Greenspan, who once prided himself on
expectations about future central bank behav- “mumbling with great incoherence,” was by
ior provide the essential link between short 2003 explicitly managing expectations by
rates and long rates. A more open central bank 
. . . naturally conditions expectations by pro-
telling everyone that the Fed would keep
viding the markets with more information the federal funds rate low “for a consider-
about its own view of the fundamental factors able period.” This guidance was only the lat-
guiding monetary policy . . . , thereby creating est step in what was, by then, a long march
a virtuous circle. By making itself more pre- toward greater transparency that began
dictable to the markets, the central bank makes in February 1994 when the Federal Open
market reactions to monetary policy more pre-
dictable to itself. And that makes it possible to Market Committee (FOMC) first started
do a better job of managing the economy. announcing its decisions on the federal
funds rate target. In May 1999, the FOMC
Five years later, Woodford (2001, pp. 307 began publishing an assessment of its “bias”
and 312) told an audience of central bank- with respect to future changes in monetary
ers assembled at the Federal Reserve’s 2001 policy in its statements. It also began issu-
Jackson Hole conference that: ing fuller statements, even when it was not
successful monetary policy is not so much a changing rates. About three years later,
matter of effective control of overnight inter- it began announcing FOMC votes—with
est rates . . . as of affecting . . . the evolution names attached—immediately after each
of market expectations . . . [Therefore,] trans- meeting. Starting in February 2005, the
parency is valuable for the effective conduct
of monetary policy . . . this view has become
FOMC expedited the release of its minutes
increasingly widespread among central bank- to make them available before the subse-
ers over the past decade. quent FOMC meeting. And most recently,
starting in November 2007, the Fed has
Notice the progression here: from increased the frequency and expanded the
Brunner’s 1981 lament about central bank- content and horizon of its publicly released
ers’ refusal to communicate, to Blinder’s forecasts.
1998 argument that more communication Other central banks have also become
would enhance the effectiveness of mon- remarkably more transparent in the last ten
etary policy, to Woodford’s 2001 claims that to fifteen years and are placing much greater
the essence of monetary policy is the art of weight on their communications. In fact, the
managing expectations and that this was Fed is more of a laggard than a leader in this
already received wisdom. Woodford prob- regard. The Reserve Bank of New Zealand
ably exaggerated that last point. But the view and the Bank of England were early and
that monetary policy is, at least in part, about enthusiastic converts to greater transpar-
managing expectations is by now standard ency, and Norges Bank (the central bank of
fare both in academia and in central bank- Norway) and Sveriges Riksbank (the central
ing circles. It is no exaggeration to call this a bank of Sweden) may now be in the van-
revolution in thinking. guard. Arguably, the European Central Bank
(ECB) has been more transparent than the
1 For example, the basic idea was stated in Marvin
Fed ever since it opened its doors in 1998.
Goodfriend (1991). We thank Michael Woodford for this More extensive central bank communication
reference. is truly a worldwide phenomenon.
912 Journal of Economic Literature, Vol. XLVI (December 2008)

One important driver of increased trans- an interviewer that there would be no fur-
parency is the notion that more independent ther central bank intervention to support the
central banks should be more accountable— euro. Those words led to an immediate depre-
that they have a duty to explain both their ciation of the euro and to heavy criticism of
actions and the thinking that underlies those Duisenberg. Similarly, when a supposedly
actions. But the intellectual arguments just off-the-record remark made in April 2006
mentioned also played a role. As it became by Fed Chairman Ben S. Bernanke, stating
increasingly clear that managing expectations that his recent Congressional testimony had
is a useful part of monetary policy, communi- been misinterpreted, was reported, markets
cation policy rose in stature from a nuisance reacted strongly—as investors concluded that
to a key instrument in the central banker’s Bernanke was “reversing himself” and saying
toolkit. In this survey, we concentrate on how that interest rates could easily go up.
central bank communication can be used to What constitutes “optimal” communica-
manage expectations both by what might be tion strategy is by no means clear. And these
called “creating news” and “reducing noise.” two examples illustrate that more talk is not
These real-world developments have always better.2 The key empirical question is
spawned a huge new scholarly literature on whether communication contributes to the
central bank communication—almost all of effectiveness of monetary policy by creating
it written in this decade. While this new lit- genuine news (e.g., by moving short-term
erature includes some theoretical contribu- interest rates in a desired way) or by reduc-
tions, most of it is empirical; and this survey ing noise (e.g., by lowering market uncer-
reflects that weighting. Studies of how central tainty). There are two main strands in the
bank communications create news focus on literature. The first line of research focuses
how, e.g., the central bank’s pronouncements on the impacts of central bank communica-
influence expectations and therefore move tions on financial markets. The basic idea is
asset prices. In extreme circumstances, com- that, if communications steer expectations
munication, used to anchor and guide mar- successfully, asset prices should react and
ket expectations, may even become the main policy decisions should become more pre-
tool of monetary policy. Studies of reducing dictable. Both appear to have happened.
noise focus, e.g., on how central bank talk The second line of research seeks to relate
increases the predictability of central bank differences in communication strategies
actions, which should in turn reduce vola- across central banks or across time to differ-
tility in financial markets. As William Poole ences in economic performance. For exam-
(2001) put it: “The presumption must be ple, does announcing a numerical inflation
that market participants make more efficient target help anchor the public’s long-run
decisions . . . when markets can correctly inflation expectations? The answer seems to
predict central bank actions” (p. 9). In both be a qualified yes.
cases, the central bank’s presumed objective This article reviews the impressive num-
is to raise the signal-to-noise ratio, and one ber of mostly empirical studies of central
major concern of this essay is how successful bank communication that have been writ-
that effort has been. ten in the last several years, mostly focusing
That said, communication is no panacea.
As with all human endeavors, there are pit-
2 However, in the Bernanke case, the Fed was going
falls and occasional errors. One famous
to raise rates further. So disabusing markets of the false
example came in October 2000 when then notion that the tightening cycle was finished probably did
ECB President Wim Duisenberg hinted to manage expectations in a constructive way.
Blinder et al.: Central Bank Communication and Monetary Policy 913

on the experience of advanced economies. where r t is the current overnight rate, r et11 is
We take stock of what we now know about today’s expectation of tomorrow’s overnight
how central bank communication can con- rate (and so on for t 1 2, t 1 3, . . . ), an is a
tribute to the effectiveness of monetary pol- term premium, and the error term indicates
icy, and we identify places where additional that the term premium might be stochastic.3
research is needed. Section 2 discusses Equation (1) makes it clear that intermediate
in more detail why central bank commu- and long-term rates should depend mostly
nication matters. Section 3 examines the on the public’s expectations of future cen-
practices of three major central banks: the tral bank policy. Today’s overnight interest
Federal Reserve, the ECB, and the Bank of rate barely matters. A particularly extreme
England. Section 4 reviews the first strand case arises when interest rates get close to
of empirical research mentioned above, and their zero lower bound. As long as the cur-
section 5 discusses the second. Finally, sec- rent overnight rate is stuck at or near zero,
tion 6 provides our answers to the question central bank communication about expected
of how central bank communication can future rates becomes the essence of mone-
contribute to the effectiveness of monetary tary policy (Bernanke, Vincent R. Reinhart,
policy and identifies avenues for future and Brian Sack 2004; Gauti B. Eggertsson
research. and Woodford 2003).
Let us now embed this idea in a simple
2.  Why Does Central Bank macroeconomic framework designed to illus-
Communication Matter? Theory trate the role of central bank communica-
tions, henceforth denoted by the vector st (for
Central bank communication can be “signals”).4 Imagine that r in (1) is the short
defined as the provision of information by rate and R is the long rate. Then aggregate
the central bank to the public regarding such demand depends on r, R, expected inflation
matters as the objectives of monetary policy, (p et ), and a host of other factors which need
the monetary policy strategy, the economic not be listed explicitly:
outlook, and the outlook for future policy
decisions. (2)  yt 5 D(r t 2 p et , Rt 2 p et , . . .) 1 e2t .
Nowadays, it is widely accepted that the
ability of a central bank to affect the economy The aggregate supply relation could
depends critically on its ability to influence (but need not) be something like the New
market expectations about the future path Keynesian Phillips curve:
of overnight interest rates, and not merely
on their current level. The reason is simple.
Few, if any, economic decisions hinge on the 3 The time subscript can be thought of as indexing days,
overnight bank rate. According to standard months, quarters, etc. The same interpretation holds. The
theories of the term structure, interest rates weaknesses of the expectations theory of the term struc-
ture are well known. We use it here only for illustrative
on longer-term instruments should reflect purposes.
the expected sequence of future overnight 4 We deliberately keep this model simple for exposi-

rates. So, for example, the n-day rate should tional purposes. It could be expanded in several direc-
tions. For example, in a New Keynesian setting, expected
be, approximately: output would appear on the right-hand side of equation
(2)—which would open up another channel by which cen-
(1)  Rt 5 an 1 tral bank communications could matter. One could also
add a more complex financial sector and/or more complex
interactions between the real and financial sectors. None
(1/n) (rt 1 r et11 1 r et12 1 . . . 1 r et1n21) 1 e1t, of this is necessary for current purposes.
914 Journal of Economic Literature, Vol. XLVI (December 2008)

(3)  pt 5 bE(pt 1 1) 1 g(yt 2 yt*) 1 e3t , Needless to say, these four conditions are
the norm, not the exception. The real world
where pt is inflation and yt and yt* are, is constantly changing, as Alan Greenspan
respectively, actual and potential real output. never tired of emphasizing.5 So learning,
The model could be closed by appending a including learning both by and about the
central bank reaction function (e.g., a “Taylor central bank, never ends. Furthermore, it
rule”): is virtually inevitable that the central bank
will know more about its own thinking than
(4)  r t 5 G(yt 2 yt*, pt , pt*, . . .) 1 e4t, the public does. In addition, contrary to the
impression given by simple Taylor rules,
where p* denotes the central bank’s inflation monetary policy decisions depend on much
target. more than current inflation and output gaps
Now imagine that the economic environ- (Svensson 2003). It is also extremely unlikely
ment is stationary (that is, equations (1)–(3) that the central bank would stick to an
do not change over time), that the central unchanged policy rule for long. For example,
bank is credibly committed to an unchang- Bernanke (2004) noted that “specifying a
ing policy rule (4), and that expectations are complete and explicit policy rule, from which
rational. In that unrealistic case, central the central bank would never deviate under
bank communication has no independent any circumstances, is impractical. The prob-
role to play. Any systematic pattern in the lem is that the number of contingencies to
way monetary policy is conducted would be which policy might respond is effectively infi-
correctly inferred from the central bank’s nite (and, indeed, many are unforeseeable).”
observed behavior (Woodford 2005). In par- Likewise, President Jean-Claude Trichet has
ticular, when it comes to predicting future repeatedly emphasized that the ECB takes
short-term rates, it would suffice to inter- its decisions one step at a time, rather than
pret incoming economic data in the light of following a rule.
the central bank’s (known) policy rule. Any Under conditions like that, as Bank of
explicit central bank communication would England Governor Mervyn King (2005) has
be redundant. Under Jon Faust and Lars E. observed, “Rational optimising behaviour
O. Svensson’s (2001, p. 373) definition of cen- is . . . too demanding, and actual decisions
tral bank transparency—that is, how easily may reflect the use of heuristics” (p. 13).
the public can deduce central-bank goals and Since central bank communication undoubt-
intentions from observable data—the central edly plays a role in shaping beliefs about
bank would be fully transparent without those heuristics, it also plays a potentially
uttering a word. important role in anchoring expectations.6
This extreme case points to four features Similarly, Bernanke (2004) used the recent
that have the potential to make central academic literature on adaptive learning to
bank communication matter: nonstationar- explain why communication affects mon-
ity (whether of the economy or the policy etary policy effectiveness. When the public
rule), the learning that is a natural concom- does not know, but instead must estimate,
itant of such an environment, and either
nonrational expectations or asymmetric
information between the public and the 5 See Blinder and Ricardo Reis (2005), especially pages

central bank. If one or more of these con- 15–24.


6 For example, King (2005), suggests that, under infla-
ditions hold, central bank communication tion targeting, a good heuristic would be “expect inflation
can matter. to be equal to target” (p. 12).
Blinder et al.: Central Bank Communication and Monetary Policy 915

the central bank’s reaction function, there employ standard statistical methods to learn
is no guarantee that the economy will con- about its parameters—which depend on the
verge to the rational expectations equilib- unobserved objectives and preferences of the
rium because the public’s learning process central bank. The learning process leads to
affects the economy’s behavior. The feedback different behavior than in the rational expec-
effect of learning on the economy can lead to tations equilibrium. For example, while peo-
unstable or indeterminate outcomes—which ple are learning, an increase in inflation may
effective communication by the central bank lead the public to revise its estimate of long-
can help to avoid (see, for example, Stefano run average inflation upward, which, in turn,
Eusepi and Bruce Preston 2007). raises actual inflation.7 As Bernanke (2004)
In addition, the central bank may have, or pointed out, such a situation opens up a clear
may be believed to have, superior informa- opportunity for the central bank to improve
tion on the economic outlook. Central banks economic performance by providing infor-
usually devote many more resources than mation about its long-run inflation objective.
private sector forecasters to forecasting and As is true in many contexts, an information
even to estimating the underlying unobserv- problem can be cured by providing more
able state of the economy. Various studies find information.
that financial markets react to information on We capture these ideas within our simple
the outlook that central banks provide (e.g., framework by replacing the assumption of
Malin Andersson, Hans Dillén and Peter rational (really, “model-consistent”) expecta-
Sellin 2006). Apparently, investors update tions by an explicit equation for interest rate
their own views in response to the informa- expectations such as:8
tion conveyed by the central bank. Donald
L. Kohn and Sack (2004) argue that private e
(5)  r t+j 5 Hj(yt , Rt, r t, . . ., st) 1 e5t ,
agents may attach special credence to the
economic pronouncements of their central
bank, especially if the bank has established where st is a vector of central bank signals,
its bona fides as an effective forecaster. They which might range from crystal clear (e.g.,
point out that the Federal Reserve has been announcing a numerical inflation target) to
broadly correct on the direction of the econ- cryptic (e.g., some of the Fed’s words.) Some
omy and prices over the past two decades, on of these communications, such as the infla-
occasion spotting trends and developments tion target, might be long-term and durable
before they were evident to market partici- while others, such as the daily reactions to
pants. In a well-known paper, Christina D. data releases, might be high-frequency and
Romer and David H. Romer (2000) provide fleeting—a distinction that will assume some
statistical evidence that Federal Reserve staff importance in our review of the empirical
forecasts of inflation were far more accurate evidence. There is no need to specify the
than private sector forecasts over a period of details of equation (5), which can stand for a
several decades. variety of possibilities for learning.
Central bank communication and learn-
ing are inextricably tied, despite a dearth
of scholarly attention to that obvious point. 7 Some other examples are Glenn D. Rudebusch and
There are exceptions, however. In Athanasios Williams (forthcoming) and Michele Berardi and John
Orphanides and John C. Williams (2005), the Duffy (2007).
8 We focus on interest rate expectations for simplicity.
public is assumed to know the form of the Expectations of inflation or even of output may be equally
equation describing inflation dynamics but to important.
916 Journal of Economic Literature, Vol. XLVI (December 2008)

In this schema, the total effect of any cen- 2.1 Is there a Downside to
tral bank action operates through at least Communication?
three distinct channels:
• the direct effect of the overnight rate on All that said, poorly designed or poorly
aggregate demand—Dr in equation (2)— executed communications clearly can do
which is probably quite small; more harm than good; and it is not obvious
• the direct effects of central bank signals that a central bank is always better off by say-
on expected future short rates: Hs in ing more. In practice, central banks do limit
equation (5), including any learning that their communications. In most cases, inter-
might take place; nal deliberations are kept secret. Only a few
• the effect of changes in the short rate central banks project the future path of their
on expectations of the entire sequence policy rate. (More on this later). And most
of future short rates, via equations (1) observe a blackout or “purdah” period before
and (5), and their consequent feedback each policy meeting, and in some instances
onto long rates, Rt, and therefore onto also before important testimonies or reports.
demand (DR). This channel will undoubt- The widespread existence of such practices
edly be influenced by the central bank’s illustrates the conviction of most central
signals, st. bankers that communication can, under
It should be clear from this trichotomy certain circumstances, be undesirable and
that any account of monetary policy that detrimental. In fact, communication during
ignores central bank communication is seri- the purdah period has been shown to lead to
ously deficient. Indeed, if the first channel is excessive market volatility (Michael Ehrmann
as unimportant as we suggest, then the com- and Marcel Fratzscher forthcoming-b).
munication channels constitute most of the The theoretical literature has not gener-
story—which is what Woodford meant and ated clear conclusions regarding the opti-
is why many economists these days charac- mal level of transparency (Petra M. Geraats
terize the job of monetary policy as one of 2002, Sylvester C. W. Eijffinger and Carin
managing expectations. van der Cruijsen 2007). The models differ
This modern view of monetary policy with respect to which aspects of central bank
leads directly to several empirical questions transparency they consider and their assump-
that form the central concerns of this sur- tions about how communications influence
vey. First, what does the vector st look like in the monetary transmission mechanism.
practice—and why might it vary across cen- Looking at real-world central bank behavior,
tral banks? Second, what evidence is there the range of views on what constitutes the
that central bank communications influence “optimal” degree (and types) of communica-
expectations directly, as posited by equa- tion has clearly evolved over time—mainly
tion (5)? Third, how do particular elements in the direction of greater openness. Are
of the vector st affect measurable variables there valid—and empirically relevant—
like interest rates, stock prices, and exchange arguments for limiting communication on
rates? Fourth, the framework suggests that monetary policy?9
skillful communications can (a) raise the
signal-to-noise ratio, (b) reduce financial
market volatility, and (c) lead to better mon- 9 We mention here, but do not discuss further, a few

etary policy outcomes (e.g., lower variances obvious ones: the need to preserve confidentiality, the fact
that financial stability sometimes limits central bank talk,
of inflation and output). Is there evidence and the obvious point that no central bank can divulge
that it does? what it does not know.
Blinder et al.: Central Bank Communication and Monetary Policy 917

One possible argument dates back to the decisions). This point has been emphasized
seminal paper by Alex Cukierman and Allan in the literature on coordination games initi-
H. Meltzer (1986).10 Their case for obfusca- ated by Stephen Morris and Hyun Song Shin
tion rested on two assumptions: that only (2002). Jeffery D. Amato, Morris, and Shin
unanticipated money matters, and that the (2002) argue that central bank communica-
central bank’s preferences are not precisely tion has a dual function: On the one hand,
known by the public. Under these assump- it provides signals about the private infor-
tions, some degree of opacity enhances the mation of central banks, and on the other
effectiveness of monetary policy because a hand, it serves as a coordination device for
fully transparent central bank cannot create the beliefs of financial market agents. They
surprises. However, two decades later, Pierre argue that central bank communication
Gosselin, Aileen Lotz, and Charles Wyplosz might be welfare-reducing if agents give too
(2007) pointed out that both the view that much weight to central bank communication
only unanticipated money matters and the as a focal point, and too little to their own
idea that the central bank conceals its prefer- information. The central bank might even
ences in order to pursue its own agenda seem coordinate the actions of markets away from
increasingly anachronistic. fundamentals.
Anne Sibert (2006) has recently raised But is this likely? Svensson (2006b) shows
doubts about the Cukierman and Meltzer that the validity of the argument requires
argument. Her two-period model of a non- that central bank communication has a much
transparent central bank focuses on the role lower signal-to-noise ratio than that of private
of private information. The central bank’s information. He argues that this assump-
welfare is increasing in unexpected inflation tion hardly ever holds in reality. Moreover,
(because it increases output) and decreasing Woodford (2005) notes that the Morris–Shin
in actual inflation. As is typical in models problem is even less likely to arise if the coor-
with such objective functions, an unobserved dination of private agents’ actions is a welfare
shock that is realized after the public’s objective per se. And Gosselin, Lotz, and
expectations are formed but before mon- Wyplosz (2007) point out that it is unreal-
etary policy is made offers the central bank istic to think that a central bank can with-
an opportunity to exploit a short-run Phillips hold information as Amato, Morris, and Shin
curve tradeoff. Nonetheless, one of Sibert’s (2002) suggest. For example, policymakers
main conclusions is that both the central tacitly reveal some of what they know merely
bank and society are always better off with by setting the interest rate. Furthermore,
increased transparency—mainly because it if we focus on providing information about
reduces the inflation bias. future monetary policy—as opposed to, say,
Surely there are limits to how much infor- forecasting the stock market or the exchange
mation can be digested effectively (Daniel rate—there is an even simpler and more
Kahneman 2003). So a central bank should compelling objection to the Morris–Shin
perhaps be wary of communicating about reasoning. Who, after all, knows more about
issues on which it receives noisy signals the central bank’s intentions than the central
itself—such as the evolution of the economy bank itself? Thus honest central bank talk is
(as opposed to, say, its upcoming interest rate almost certain to coordinate beliefs in the
right direction.
10 For related work see Michelle R. Garfinkel and
Finally, we should mention the “cacoph-
Seonghwan Oh (1995), Faust and Svensson (2001), and ony problem,” pointed out by Blinder (2004,
Henrik Jensen (2002). chapter 2). When monetary policy decisions
918 Journal of Economic Literature, Vol. XLVI (December 2008)

are taken and subsequently explained by a something does constrain future behavior—
committee rather than by a single individual, as in “giving a verbal commitment”—most
there is a danger that too many disparate central bank communication is not, or need
voices might confuse rather than enlighten not be, of this nature. In particular, the mere
the public—especially if the messages appear conveyance of information—such as about
to conflict. If done poorly, uncoordinated the policy decision, the inflation target, the
group communication might actually lower, forecast, etc.—does not commit the bank
rather than raise, the signal-to-noise ratio. to any future action or inaction (although it
But the appropriate remedy for this problem, might hint at such). Even the famous pub-
should it exist, is clarity, not silence. lished “forward tracks” of the Reserve Bank
of New Zealand (discussed later), which are
2.2 Communication is Not Precommitment
conditional forecasts of its own future behav-
Over the years, many central bankers and ior, are conditioned on many future variables.
economists have at times confused commu- That said, the conditional character of such
nication with commitment—or worried out forecasts may be difficult to convey (Otmar
loud that the public might confuse the two. Issing 2005).
For example, it has been argued that words Of course, there may be cases in which a
uttered today might restrict the freedom central bank wants to use words to commit
to maneuver tomorrow. For example, then itself in some way. For example, Bernanke
Chairman Paul Volcker defended the Fed’s et al. (1999) argued in favor of inflation tar-
refusal to announce its decisions immedi- geting on precisely these grounds—as a way
ately in 1984 as follows: to constrain central bank discretion. But
One danger in immediate release of the direc- that is the exception, not the rule. For the
tive is that certain assumptions might be made most part, the sorts of communications that
that we are committed to certain operations we deal with in this paper generally do not
that are, in fact, dependent on future events,
and these interpretations and expectations
imply any form of commitment. Since there
would tend to diminish our needed opera- is already a huge and well-known theoreti-
tional flexibility.11 cal literature on the role of commitment in
monetary policy, we will not deal with that
In a similar vein, Alan Greenspan opposed subject further.13
immediate disclosure of the FOMC’s deci- In sum, there are many theoretical reasons
sions in 1989 because “a public announce- why central bank communication should be
ment requirement also could impede timely expected to matter, and many of them imply
and appropriate adjustments to policy.”12 (Yet that skillful communication can improve
less than five years later, he voluntarily did macroeconomic outcomes. As against this,
precisely that.) the arguments against greater transparency
From today’s standpoint, the objections of seem to be thin gruel: the profession no lon-
Volcker and Greenspan to this minimalist dis- ger believes that only unanticipated money
closure proposal sound quaint—almost scho- matters; the Morris-Shin coordination “prob-
lastic. While there are cases in which saying lem” seems more likely to be an advantage
of central bank communication than a disad-
vantage; and communication need not imply
11  Quoted in Goodfriend (1986), pp. 76–77. Good-
friend’s paper was an early, and at the time highly contro-
versial, critique of the Federal Reserve’s secrecy—written 13 Among the many sources that could be cited, see
by a Fed employee. Richard Clarida, Jordi Gali, and Mark Gertler (1999) or
12 Quoted in Blinder (1998), pp. 74–75. Woodford (2003).
Blinder et al.: Central Bank Communication and Monetary Policy 919

(unwanted) commitment. We turn now from been directed) to provide their own quan-
theory to practice. tification, for at least two reasons. First,
numerical targets facilitate accountability,
enabling the performance of the central
3.  Central Bank Communication in
bank to be assessed against its mandated
Practice
yardstick (Jakob De Haan and Eijffinger
Many central banks with similar monetary 2000). Second, a quantitative objective (or
policy objectives nonetheless follow funda- objectives) helps to anchor the expectations
mentally different communication policies; of economic agents. In terms of our simple
and these policies have evolved over time. In modeling framework, agents’ expectation
our framework, this means that the vector formation in (5) is facilitated by knowing the
of communication signals, st, takes different targets y*t and p*t that enter the policy rule (4).
forms in different times and places. In this In turn, well-anchored inflation expectations
section, we illustrate the diversity in current help to stabilize actual inflation by removing
communication practices by examining the an important source of shocks. However, few
different types of signals that central banks if any central banks actually communicate a
send, concentrating mainly on three major precise policy rule.15 Instead, private agents
central banks: the Federal Reserve System, learn about the “rule” both by watching what
the Bank of England, and the European the central bank does and listening to what
Central Bank.14 We first split the vector of it says.
central bank signals, st, by content (section These accountability and anchoring argu-
3.1), and then by sender (section 3.2). ments figure prominently in the debate over
inflation targeting (IT) because better and
3.1 What to Communicate
more open communication is often taken to
Central banks communicate about at least be a defining virtue of IT. While the Bank
four different aspects of monetary policy: of England, for example, sets interest rates
their overall objectives and strategy, the independently, its inflation target comes
motives behind a particular policy decision, from the Chancellor. The ECB, in contrast,
the economic outlook, and future monetary was not given a quantitative objective by
policy decisions. Central banks’ objectives and the Maastricht Treaty, but provided one for
strategies tend to be more stable, so the cor- itself as an important part of its monetary
responding signals show less variability over policy strategy. Yet a third approach is fol-
time than signals about the other three items. lowed by the Federal Reserve, which has
two legislated objectives, namely price sta-
3.1.1 Objectives and Strategy
bility and full employment, neither of which
An independent central bank should be is quantitative as yet. This diversity of prac-
given a clearly defined mandate by its gov- tices among otherwise similar central banks
ernment. Generally, this is done by enunci- is striking, and we will later investigate the
ating central bank objectives, sometimes in extent to which these differences bear on
quantitative terms. Some central banks that economic outcomes.
are not given quantitative objectives by their
governments have nonetheless decided (or
15 Even formulating an objective function may be a
14 The diversity in communication practices across cen- daunting task for a central bank. Some of the difficulties
tral banks is also illustrated in Issing (2005), particularly in doing so are described by Frederic S. Mishkin (2004)
table A2. and Cukierman (2008).
920 Journal of Economic Literature, Vol. XLVI (December 2008)

3.1.2 Policy Decisions of releasing minutes, some central banks (such


as the ECB) hold press conferences immedi-
Most central banks nowadays inform the ately following their policy decisions. Press
public about their monetary policy decisions conferences may provide less detail than min-
on the day they are taken. However, this utes, but they are more timely and more flex-
was not always so. Prominently, the Federal ible, as they allow the media to ask questions.
Reserve only began announcing changes in We will return to this issue later.
its target federal funds rate on the day of
FOMC meetings in February 1994. Before 3.1.3 The Economic Outlook
that, markets had to infer the intended funds
rate from the type and size of open-market Another important aspect of a central
operations—until the decision was pub- bank’s communication strategy is the extent
lished after the subsequent FOMC meeting. and content of any forward-looking informa-
Prompt and clear announcement of monetary tion it provides. This information set includes
policy decisions clearly creates news, but it the central bank’s assessment (forecast) of
also reduces noise by eliminating any guess- future inflation and economic activity, and its
ing on the part of the public. So this type of own inclinations regarding future monetary
central bank communication evidently raises policy decisions. Central banks differ sharply
the signal-to-noise ratio. As we will see in the in whether and how they communicate such
next section, it also leads to improvements in information.
the efficiency of monetary policy. Inflation-targeting central banks typically
Practices differ enormously regarding what provide their assessment of expected future
central banks should or should not say in the inflation in periodic reports. In that context,
statement that accompanies the decision and, the Bank of England’s display of probabil-
presumably, explains it. In particular, central ity distributions through “fan charts” has
banks apparently disagree over how much many imitators. However, central banks that
should be disclosed about the decision-mak- are not inflation targeters also often release
ing process itself, e.g., through the release of (some aspects of) their inflation forecasts. In
minutes and voting records. The ECB does the case of the ECB, this is done through
not publish minutes, and insists that it makes the staff projections (now published four
monetary policy decisions by unanimity. The times a year), which serve as an input to the
Fed and the Bank of England do release min- Governing Council’s discussions, but need
utes (and both recently expedited the release), not be endorsed by it—a very different role
along with recorded votes. This informa- from inflation forecasts in an IT strategy. The
tion is particularly important for the Bank of Federal Reserve keeps its staff projections
England, whose Monetary Policy Committee secret; but it now publishes FOMC forecasts
(MPC) members are individually account- of inflation four times a year. The November
able, and therefore need to have their votes 2007 changes in its communication prac-
recorded and scrutinized. Interestingly, dis- tices increased both the frequency and
sents on the British MPC are much more length of its publicly released forecasts (see
frequent than they are on the FOMC, where Bernanke 2007). Although these changes
decisions are typically unanimous and dissent did not include the adoption of an explicit
connotes fundamental disagreement.16 Instead inflation target, the new three-year-ahead

16 On this point, see Blinder (2007), Henry W. Chappell, Meade and D. Nathan Sheets (2005), and Laurence H.
Rob Roy McGregor, and Todd Vermilyea (2005), Ellen E. Meyer (1998).
Blinder et al.: Central Bank Communication and Monetary Policy 921

forecast effectively reveals the inflation cal path of future policy rates that underlies
rate that ­policymakers believe is consistent their macroeconomic forecasts. Sweden and
with the Fed’s mandate to achieve “stable Iceland recently joined a small group that
prices.” includes New Zealand and Norway in doing
Until recently, the diversity across central so. Some observers view the central bank’s
banks was even wider when it came to the forecasting its own future behavior as the
outlook for economic activity. However, the last frontier of transparency, and none of the
Federal Reserve has now joined the Bank of three major central banks on which we have
England and the ECB in providing more fre- focused have yet been willing to go there.
quent official forecasts of output measures. The issue remains highly controversial.19
A number of central banks even publish esti- Both Mishkin (2004) and Charles A. E.
mates of the output gap. Given the difficul- Goodhart (2001) argue against announcing
ties in measuring and forecasting potential the path of the policy rate on the grounds
output,17 the latter option is practiced by that it may complicate the committee’s deci-
only a few central banks (including those of sion-making process. It may also complicate
New Zealand, Norway, the Czech Republic, communication with the public, which may
Sweden, and Hungary). not understand the conditional nature of the
projection. In practice, the main concern
3.1.4 The Path of Future Policy Rates holding back many central bankers is that
such communications might be mistaken for
When it comes to likely future policy deci- commitments. If the projected developments
sions, many central banks provide some sort do not materialize, the discrepancy between
of forward guidance, albeit in very differ- actual and previously projected policy might
ent ways. Some, such as the ECB, use indi- damage the central bank’s credibility (Issing
rect signals, often in the form of code words 2005). In addition, while forward guidance
like “vigilance” (David-Jan Jansen and De by the central bank is intended to correct
Haan 2007a). Other central banks are more faulty expectations, and thereby reduce mis-
explicit. The FOMC, for instance, some- allocations of resources, inaccurate forecasts
times (but not always) issues a statement might actually induce such misallocations,
with a forward-looking assessment of future e.g., if agents make economic decisions (such
monetary policy. These statements, which as taking on a mortgage) based on the central
began in earnest in May 1999, have evolved bank’s communication.
over time. They were originally phrased in To guard against these potential pitfalls, all
terms of the policy “bias,” then in terms of central banks that provide forward guidance
the “balance of risks” for the “foreseeable on interest rates emphasize that any forward-
future,” and so on. At times, especially dur- looking assessment is conditional on cur-
ing the 2003–05 period, the FOMC has rent information—and therefore subject to
been quite direct about its expected future change. For example, the Riksbank regularly
path of interest rates.18 stresses the conditionality of its projected
A few central banks even provide quan- repo rate path by repeating the mantra: “It is
titative guidance by publishing the numeri- a forecast, not a promise.”

17 See, e.g., Orphanides (2001).


18 For a detailed description of the evolution of the
FOMC’s forward-looking language, see Rudebusch and 19 The case in favor is made by Svensson (2006a) and
Williams (forthcoming). Woodford (2005).
922 Journal of Economic Literature, Vol. XLVI (December 2008)

3.2 How to Communicate banks provide detailed accounts and explana-


tions of the decisions in the minutes.21
Just as the content of signals differs mark- By contrast, the ECB not only releases
edly across central banks, so does the choice a press statement with the policy decision
of communication tools. Central banks can but also holds a press conference on the day
choose from a large menu of communica- of Governing Council meetings, includ-
tion instruments, and each central bank uses ing a question and answer (Q&A) session.22
its own mixture.20 This subsection provides a Compared to the approach of the other two
brief overview of one particularly important central banks, there are four main differences.
aspect of instrument selection, namely, the First, while providing background informa-
choice of sender (e.g., whether a signal is sent tion on the rationale for the decision, the ECB
by the committee or by an individual commit- press conference is generally less detailed
tee member), which in turn may influence the than the minutes of the Bank of England or
precision of the signal. When signals are sent the Federal Reserve. In particular, it does not
by or on behalf of the monetary policy com- provide any information on voting.23 Second,
mittee, the appropriate content, timing, and however, the press conference avoids the
channels must all be chosen. Communication substantial time delay of the minutes. Some
by individuals raises further issues—such as observers have argued that there is a trade-off
whether one member (e.g., the chairman or between the timeliness of this type of com-
governor) should serve as spokesperson for munication and its accuracy (Goodhart 2005),
the committee, reflecting a more collegial as minutes usually undergo a detailed sanitiz-
approach to communication, or each member ing process to put the best possible face on
should present his or her own views, represent- the committee’s views and intentions, which
ing an individualistic communication strategy. delays their release. Third, the Q&A session
allows the press to ask follow-up questions and
3.2.1 Communication by Committees thus can help clarify ­ambiguities (Ehrmann
and Fratzscher forthcoming-a). Fourth, press
The most natural occasion for communica- conferences are typically televised live, which
tion by an MPC as a whole arises on meet- gives the central bank an opportunity to reach
ing days, when decisions are announced. The out to the broader public.
timing of this communication and the amount Another natural communication oppor-
of detail provided differ substantially across tunity for a committee inheres in its legal
central banks. The Federal Reserve provides reporting requirements. For example, each
a short press release containing the decision, a of the three central banks is obliged to pro-
concise (and typically stylized) explanation of vide an annual report and to testify before
its underlying reasoning, and (at times) some its ­legislature. For the ECB and the Federal
forward guidance. The Bank of England’s
press statement announces the decision, but 21 In addition to the minutes, the Federal Reserve
normally provides an explanation only when eventually also releases the transcripts of FOMC meet-
interest rates are changed or when its decision ings, albeit only after a five-year lag.
22 The central banks of the Czech Republic, Japan,
was largely unexpected. Somewhat later, but
New Zealand, Norway, Poland, Sweden, and Switzerland
prior to the subsequent meeting, both central also hold regular press conferences.
23 For a stimulating debate on the ECB’s decisions not
to release either minutes or individual voting records, see
20 See Blinder et al. (2001) for a detailed, though by Willem H. Buiter (1999) and Issing (1999). One important
now somewhat dated, account and explanation of the vari- argument in favor of publishing minutes is that they pro-
ous instruments used by central banks. vide some information about the internal deliberations.
Blinder et al.: Central Bank Communication and Monetary Policy 923

Reserve, these hearings provide the commit- emphasizes that these distinct types of com-
tee’s views, whereas Bank of England testimo- mittees need different communication strate-
nies relate more to members’ personal views. gies. In the individualistic case, the diversity
Among the most important reporting of views on the committee should be appar-
vehicles are regular publications, such as the ent, as a way to help markets understand the
ECB’s Monthly Bulletin, which is published degree of uncertainty surrounding monetary
one week after each monetary policy meet- policy making. But in the collegial case, a sim-
ing and contains both the assessment of eco- ilar diversity of views, if made public, might
nomic developments and information on the undermine clarity and common understand-
analytical framework—e.g., models, methods ing. Therefore, communication should mainly
and indicators—used in its decision-making convey the committee’s views.
process. For the Bank of England, the most Since the importance of individual views
closely watched reports are its quarterly in the communication strategy of a particu-
Inflation Report, which sets out the detailed lar MPC will reflect the structure and func-
economic analysis that underlies the MPC’s tioning of the committee, it will vary both
decisions and presents the Bank’s assessment across banks and across time. Despite its
of the prospects for inflation over the follow- collegial structure, the Federal Reserve pur-
ing two years, and the Quarterly Bulletin, sues a somewhat individualistic communica-
with its commentary on market develop- tion strategy, which at times reveals highly
ments and monetary policy operations. The diverse opinions across FOMC members.
publication of the Inflation Report is also This diversity stands in sharp contrast to the
accompanied by an hour-long press confer- ECB, which has followed a far more collegial
ence. The Federal Reserve’s closest coun- communication strategy, often displaying a
terpart is its semiannual Monetary Policy much higher degree of consistency among the
Report to the Congress, presented with the statements of individual committee members
chairman’s testimony to Congress. (Ehrmann and Fratzscher 2007a).
One difference between communications
3.2.2 Communication by Individual by individual members and by entire com-
Committee Members mittees is the greater flexibility in timing of
the former. Communications by committees
Most central banks these days make deci- are generally prescheduled, and thus some-
sions by committee, reflecting an apparent what inflexible in timing. But changes in the
consensus that doing so leads to superior circumstances relevant to monetary policy
policy (Blinder 2004, chapter 2). But com- do not always coincide with meeting dates
mittees come in a wide variety of shapes and or testimonies. Furthermore, the central
sizes. Blinder (2004) distinguishes among bank might want to provide more guidance
three types of committees—individualis- to financial markets and the public in times
tic, genuinely collegial, and autocratically of great uncertainty (Jansen and De Haan
collegial—and characterizes the Bank of 2005). Occasional speeches and interviews
England’s MPC as individualistic, the ECB’s by individual committee members between
Governing Council as genuinely collegial, meetings offer a way to communicate changes
and the Federal Reserve’s FOMC under Alan in views rapidly, if so desired. But the large
Greenspan as autocratically collegial.24 He variation across central banks in the intensity
of intermeeting communication suggests that
24 The FOMC is clearly becoming more genuinely col- they differ greatly in how much importance
legial under Bernanke. they attach to timeliness.
924 Journal of Economic Literature, Vol. XLVI (December 2008)

This section has shown that central bank also, not by coincidence, much more litera-
practices differ enormously, both across cen- ture to survey.
tral banks and across time. However, there In section 4.1, we explain some of the
are clear trends toward more timely and more methodological approaches that have been
open communication. We have also highlighted used to identify and measure communication
the huge variety of signals that are subsumed signals. Section 4.2 then moves on to what
under the symbol st in our schema. This huge seems to be the logically first substantive
heterogeneity in central bank practices raises question: Do central bank signals success-
an obvious question: Are there better and fully steer market expectations about future
worse ways of communicating? This is clearly monetary policy and therefore render policy
an empirical question—and a large one. The more predictable? The evidence suggests
rest of this survey will suggest partial answers that they do, though imperfectly. Section 4.3
based on recent empirical evidence. takes the next step and inquires about the
impact of central bank communications on
interest rates and other asset prices—as well
4.  The Impact of Central Bank
as on their volatility. Some of these impacts
Communication on Financial Markets
appear to be sizable, others less so. Finally,
In section 2, we noted that central bank in section 4.4, we turn to the possibility that
communications (st) influence expectations some communications wind up being mis-
e
of future short-term rates (r t1j ), which, in communications, so that, e.g., relatively noisy
turn, influence long-term rates and other central bank signals might actually increase,
financial-market prices (Rt). These prices, in rather than decrease, uncertainty.
turn, influence such macro variables as infla-
4.1 Identifying and Measuring
tion and output (yt and pt). But there are at
Communication Events
least two crucial differences between the
earlier and later links in this causal chain. As we have noted, central bank communi-
Both relate to timing. First, while central cations differ in many ways and are sometimes
bank communications, just like monetary difficult to measure. So we focus first on rel-
policy, affect financial markets very quickly, atively well-defined, high-frequency signals,
interest rates and asset prices affect the such as announcements and speeches. It has
economy only gradually—with the prover- become standard practice to identify regu-
bial long and variable lags. Second, many lar prescheduled communications (such as
factors other than monetary policy influ- announcements of policy decisions) from cen-
ence macro variables such as yt and pt. But, tral bank websites and irregular statements
at least over the narrow time windows used (such as speeches and interviews) through
in many empirical studies, it is arguable financial newswire reports. However, it is not
that financial market variables are reacting always straightforward to determine exactly
only, or at least mostly, to central bank sig- when a communication “event” took place.
nals. The upshot is that it is a much easier For example, when a late-Thursday media
econometric task to estimate the effects of report on a Tuesday interview with a poli-
central bank signals using high-frequency cymaker causes financial markets to react
data from financial markets (the subject of on Friday morning, was the communication
this section) than using low-frequency data event on Tuesday, Thursday, or Friday?
on macroeconomic performance (the sub- A second challenge is how to extract
ject of the next section). So we begin with the intention or objective behind a policy
­financial market reactions—where there is ­statement—which is, after all, essential if we
Blinder et al.: Central Bank Communication and Monetary Policy 925

want to know whether central bank commu- conveys relevant information to market partic-
nications succeed. Different approaches to ipants. While this is an ­important first finding,
this issue have emerged. One line of research it leaves several open questions. Their approach
does not even attempt to divine directional says nothing about whether the policymaker
intent. Kohn and Sack (2004), Ellis Connolly intended to move asset prices in the way they
and Marion Kohler (2004), and Rachel Reeves did. It also does not tell us whether the rise in
and Michael Sawicki (2007) study instead the volatility reflects an increase in market uncer-
effects of central bank communication events tainty or a decrease in uncertainty that is out-
on the volatility of financial variables. The weighed by the immediate effect of the news.
basic idea is that, if communications affect In other words, the Kohn and Sack approach
the returns on financial assets, the volatility establishes that central bank communication
of these returns should be higher on days of creates news, but it cannot determine whether
central bank communications, ceteris pari- it reduces noise.
bus, because the signals contain news. Accordingly, a number of subsequent studies
Focusing on volatility makes it unneces- have attempted to quantify communication in
sary to assign a direction to each statement, order to assess both the direction and magni-
which is both a strength and a weakness. The tude of its effects on asset prices—and thus to
researcher just studies whether central bank determine to what extent communication has
communications create news, not whether its intended effects.25 To do so, all statements
they move markets in the “right” direction. must be classified according to their content
But there are other problems. One is that and/or likely intention, and then coded on a
many factors, some of them unobservable, numerical scale. Among the types of commu-
affect asset prices. So a rise in observed vola- nications that have been studied in the litera-
tility may reflect the reaction of financial ture are: statements on monetary policy and
markets to shocks other than central bank communication about the exchange rate (see
communication. A second problem is that Jansen and De Haan 2005 for speeches and
communication may be endogenous. A cen- interviews by members of the ECB Governing
tral bank may choose to communicate at a Council), policy inclinations and the economic
particular time because of a sudden change outlook (see Ehrmann and Fratzscher 2007a
in the economic outlook or some other news. for communication by committee members
In this case, asset prices will probably be at the Federal Reserve, the Bank of England,
more volatile on communication days, but not and the ECB), price stability, the real economy,
necessarily because of the statement (Reeves and monetary indicators (see Helge Berger, De
and Sawicki 2007). Such endogeneity is less Haan, and Jan-Egbert Sturm 2006 for the
of a problem when the dates of major com- introductory statement of the ECB’s press
munications—such as policy decisions and
testimonies—are known in advance. But it
25 This is similar to the so-called “narrative approach”
may be especially problematic for speeches
to modeling monetary policy making as, for instance,
or interviews of committee members, which applied by Romer and Romer (1989) to identify times
are flexible in both timing and content when the Fed tightened. Similarly, John F. Boschen and
(Ehrmann and Fratzscher 2007c). Leonard O. Mills (1995) generated a discrete measure of
the Fed’s policy stance taking five different values {–2, –1,
Kohn and Sack (2004) show that the vola- 0, 1, 2}, where –2 indicates a very tight policy stance, while
tility of various asset prices reacts significantly 2 indicates a very loose policy stance. Like the Romer
to statements by the FOMC and its mem- and Romer dates, their indicator is based on FOMC min-
utes. But it is a more informative measure of monetary
bers. They argue that the increased volatility policy, since it differentiates according to size as well as
is evidence that central bank communication direction.
926 Journal of Economic Literature, Vol. XLVI (December 2008)

conference and Stefan Gerlach 2007 for the One approach to these issues, first sug-
editorial in the ECB’s Monthly Bulletin). All gested by Refet S. Gürkaynak, Sack, and Eric
of these studies assign negative (positive) val- T. Swanson (2005) and Gürkaynak (2005),
ues to statements that are perceived as dov- and also implemented in Claus Brand, Daniel
ish (hawkish), and zero to those that appear Buncic, and Jarkko Turunen (2006), employs
to be neutral. Whereas some researchers indirect measures derived from financial
restrict the coding to directional indications market reactions. Gürkaynak, Sack, and
(e.g., Jansen and De Haan 2005, Ehrmann Swanson (2005) use a principal components
and Fratzscher 2007a), others assign a finer method that identifies two common factors
grid that is at least suggestive of magnitude that describe asset price movements around
(Carlo Rosa and Giovanni Verga 2007, Marie FOMC announcements. The factor that is
Musard-Gies 2006), e.g. by coding state- orthogonal to the surprise in the federal funds
ments on a scale from 22 to 12. target rate is defined to be the communica-
The strength of the coding approach is tion effect of the FOMC statement. Brand,
that it helps us understand whether com- Buncic, and Turunen (2006) make use of
munication succeeds or fails. But this the timing details of the ECB’s meeting day
comes at a price. First, the required classi- communications. The ECB announces its
fication is necessarily subjective, and there decisions at 1:45 pm, without any explanatory
may be misclassifications. This risk can be statements, and then explains the decisions
reduced through content analysis (Ole R. in detail in the press conference forty-five
Holsti 1969), such as when several research- minutes later. Because of that delay, the mar-
ers independently classify statements (as in ket reaction to the release of the decision can
Berger, De Haan, and Sturm 2006); but it be distinguished from the market reaction to
can never be eliminated. the forward-looking communication by using
Second, when statements are identified very high-frequency data. Doing so leads
through newswire reports (as in Jansen and to results that are similar to those obtained
De Haan 2005 and Ehrmann and Fratzscher with the principal components approach
2007a), communication is not measured at of Gürkaynak, Sack, and Swanson (2005).
the source, but rather via the media inter- Rosa (2008) approximates the unexpected
mediary. But these intermediaries could be components of both FOMC and ECB state-
selective or misleading in their reporting, or ments by estimating forecasting regressions
could release the story days after the inter- that explain the content of each statement
view. That said, if researchers are interested by the preceding statement and the slope
in testing market responses to communica- of the yield curve just prior to the release of
tion, it may make sense to focus on statements the statement. Interestingly, he finds stron-
that actually reach market participants, and ger market reactions to unexpected Fed talk
on the content as conveyed by the media. than to unexpected ECB talk.
Third, subjective communication indica-
tors are constructed ex post and may not 4.2 Does Central Bank Communication
accurately reflect how financial markets Enhance the Predictability of Monetary
understood the signals at the time. The lat- Policy?
ter is likely to be affected by, e.g., expecta-
tions about monetary policy at the time of Equation (5) in our conceptual schema
the statement. As we know, efficient financial posited effects of central bank signals on
markets should react only to the unexpected expectations of future monetary policy deci-
component of a statement. sions. It seems that successful central bank
Blinder et al.: Central Bank Communication and Monetary Policy 927

communication efforts should make policy months in advance has increased since the
more predictable and market expectations of late 1980s. Similarly, Swanson (2006) finds
future short rates more accurate. Is it true? that U.S. financial markets and private sector
Presumably, central banks are most con- forecasters have become both better able to
cerned with long-term predictability, which forecast the federal funds rate at horizons out
comes when the public develops a genuine to several months and less uncertain about
understanding of the way the central bank their interest rate forecasts ex ante, as mea-
thinks and operates. This is what King (2000) sured both by interest rate options and by the
means by saying provocatively that a central cross-sectional variance of interest rate fore-
bank should be “boring.” The idea is that the casts. Importantly, private sector forecasts
monetary policy reaction function should be of macroeconomic variables have not shown
so well understood that all the relevant news similar improvement, which points toward a
for financial markets would stem from devel- specific effect of communication about mon-
opments in the economy, not from the actions etary policy rather than a general decline
or words of the central bank. In a similar in macroeconomic volatility (the so-called
vein, Blinder (2004) suggests that “perhaps Great Moderation).
the best a central bank can do is to ‘teach’ Each of these authors argues that the
the markets its way of thinking” (p. 25). Federal Reserve’s practice of making same-
Unfortunately, long-term predictability day announcements of monetary policy deci-
is inherently difficult to gauge, so empirical sions since February 1994 was an important
studies typically resort to predictability at factor in reducing uncertainty. Furthermore,
short horizons, which is more measurable.26 Selva Demiralp and Oscar Jorda (2002) pro-
Often, financial market prices are used vide compelling evidence that announcing
for this purpose. And indeed, research has changes in the intended federal funds rate in
shown convincingly that the predictability real time enabled the Fed to move the actual
of the interest rate decisions of the major funds rate with smaller open market opera-
central banks has improved remarkably in tions—which suggests that greater transpar-
recent years—so much so that financial mar- ency also helped make policy implementation
kets’ expectations nowadays are generally more efficient.
well aligned with actual decisions (Kerstin Other types of central bank communi-
Bernoth and Jürgen von Hagen 2004). cation should also enhance predictability.
The case of the Federal Reserve, with its Regardless of whether policymakers have
continuously increasing level of transpar- superior information about the economy,
ency since the 1980s, has been studied most communicating their macro forecasts, and
extensively. Poole and Robert H. Rasche possibly also offering indications about the
(2003) provide evidence that the surprise future path of interest rates, should help
component of policy decisions decreased financial markets predict future central bank
considerably after February 1994. Joe Lange, actions better. A study by Andrea Fracasso,
Sack, and William Whitesell (2003) show Hans Genberg, and Wyplosz (2003) sheds
that the ability of Treasury bill yields to pre- light on this issue. They develop a number of
dict changes in the federal funds rate some indicators of the quality of inflation reports
for nineteen countries and find that higher
quality reports are associated with smaller
26 However, as more central banks begin to announce
policy surprises. In particular, three subjec-
projected interest-rate paths, it will become possible to
study the extent to which these forecasts are incorporated tive indicators—how convincing the report
in yield curves. is judged to be, how well it reflects the
928 Journal of Economic Literature, Vol. XLVI (December 2008)

e­ xpertise of the staff, and the quality of the attention: the qualitative guidance provided
writing style—increase the predictability of by the Federal Reserve’s (and other central
interest-rate decisions. These results suggest banks’) policy “bias” or “balance of risks.”
that clearer communication, which creates a As we will see in the next subsection, these
higher signal-to-noise ratio, reduces private statements strongly affect financial markets.
agents’ uncertainty. While its substantive Surprisingly, however, their predictive power
concern (financial market volatility) is quite for future monetary policy is mixed.
different, Jansen (2008) supports this sugges- In interpreting this evidence on the Fed,
tion. Using objective measures of readability it is important to distinguish between bias
to measure the clarity of the Federal Reserve statements made prior to and after May 1999.
chairman’s semiannual Humphrey–Hawkins Until that date, the FOMC’s policy directives
testimonies, he finds that greater clarity often were not public information. They were inter-
reduces the volatility of interest rates. nal declarations of intent, focused narrowly
Particularly strong effects can be expected on the inter-meeting period and released
if a central bank communicates its inten- only after the next FOMC meeting. By con-
tions about upcoming interest rate decisions. trast, the balance-of-risks assessments since
Unfortunately, we know of no research that May 1999 are external statements intended
analyses the effects of the publication of to provide markets with information about
quantitative forward guidance, as practiced possible future policy decisions.
for years by the Reserve Bank of New Zealand Consistent with their internal use, the pre-
(RNBZ) and, more recently, by Norges Bank 1999 bias statements have been found to be a
and Sveriges Riksbank. New Zealand has the statistically significant indicator of the likeli-
longest history of doing so by far, and David hood and direction of changes in the fed funds
Archer (2004) provides some evidence that target during the subsequent inter-meeting
market reactions to the publication of the period (John S. Lapp and Douglas K. Pearce
RBNZ’s forward interest rate tracks are lim- 2000), but not thereafter (Daniel L. Thornton
ited. He suggests that this might reflect the and David C. Wheelock 2000). By contrast,
market’s understanding of the conditional- Ehrmann and Fratzscher (2007d) focus on
ity of such publications: Market participants the period since May 1999 and do find con-
might adjust rates to match the Bank’s for- sistency with subsequent interest rate moves.
ward path only if they share the underlying Similarly, Michael R. Pakko (2005) finds that
assumptions about the evolution of the econ- the information content of these statements
omy. Beyond this, however, our knowledge is a statistically significant variable predicting
of the effects of central bank interest rate changes in the funds rate target in the context
projections on the market’s understanding of of an estimated Taylor rule.
monetary policy is minimal—which is inevi- Strikingly, however, Ehrmann and
table given the short time span and the small Fratzscher (2007d) find that policy decisions
number of central banks that have revealed after May 1999 have not been anticipated
such information. As more experience is better by financial markets than they were
accumulated, e.g., in Norway and Sweden, before. Figure 1 shows one case in point
this will be a high priority area for future by displaying the adjustment of the three-
research. Already, Norwegian data show that month T-bill rate around two pairs of FOMC
markets do not always uncritically adopt the meetings: one prior to 1999, and one after. In
central bank’s projection. both cases, there was no change at the first
Another type of forward-looking informa- of the two meetings (marked as day 0 on the
tion has been the subject of far more scholarly horizontal axis) and then a twenty-five basis
Blinder et al.: Central Bank Communication and Monetary Policy 929

5.95 5.7
Federal funds target Jan/Feb 2000 Feb/Mar 1997

5.85 5.6

5.75 5.5

5.65 5.4

5.55 5.3

5.45 5.2

5.35 5.1

5.25 5
– 4 –2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 2 4

Figure 1. Adjustment of Market Interest Rates under Alternative Disclosure Regimes:


March 25, 1997 versus February 2, 2000.

Note: Three-month money market rates for the 1997 tightening episode are shown on the right-hand scale whereas those
for the 2000 episode are depicted on the left-hand scale. Both tightening days are scaled so as to be shown on day 30 on
the horizontal axis. Day 0 refers to the corresponding previous FOMC meetings.

Source: Ehrmann and Fratzscher (2007b).

points increase at the second (marked as day day of Chairman Greenspan’s testimony
30). Furthermore, in both cases, the FOMC before the U.S. Senate (day 11 in the figure).
had actually adopted a tightening bias at the While the figure shows just two exception-
meeting on day 0. ally clear cases, Ehrmann and Fratzscher’s
In both instances, the decision to raise the statistical tests confirm this as a general ten-
funds rate was well predicted by the market: dency. Markets anticipated the Fed’s deci-
T-bill rates had already increased substan- sions rather well both before and after release
tially by the time the FOMC acted. But the of the “balance of risks,” but the mechanisms
timing was very different in the two cases. were different. Nowadays, they extract infor-
When the tightening bias was released on day mation from the statements, whereas before
0 (solid line), markets did most of the adjust- they relied more on other types of Federal
ing to the coming interest rate hike within Reserve communication between meetings,
one day. By contrast, when the bias was not such as speeches and testimonies by FOMC
released (dashed line), interest rates adjusted members.
much later. Interestingly, T-bill rates started A number of recent papers apply Pakko’s
anticipating the Fed’s next rate move on the (2005) approach to the case of the ECB,
930 Journal of Economic Literature, Vol. XLVI (December 2008)

examining to what extent its communications Reports seem to move financial markets
add information to Taylor-type rules. There is significantly.27 But just like Kohn and Sack,
general agreement that models based exclu- Reeves and Sawicki find that speeches do
sively on communication variables perform not.
worse than those that include relevant macro- However, other studies have found signifi-
economic variables, like inflation and output cant effects of speeches on financial ­markets,
gaps. However, whether communication can perhaps due to different procedures for
improve the fit of standard Taylor rules is in selecting communication events. Table 2,
more dispute. Jansen and De Haan (forth- taken from Ehrmann and Fratzscher (2007a),
coming) find that the ECB’s communications displays the estimated effects of interviews
do not add informational value, but Rosa and and speeches by central bank commit-
Verga (2007) and Friedrich Heinemann and tee members for the Federal Reserve, the
Katrin Ullrich (2007) find that the press con- Bank of England, and the ECB. It shows
ferences do. that speeches and interviews affect interest
rates through most of the maturity spectrum,
4.3 Do Financial Markets Respond ranging from very short maturities up to five
to (Which Form of) Central Bank years. Since these results are based on coded
Communication? communication events as explained above,
they offer directional information, too.
The evidence obtained from various event Ehrmann and Fratzscher (2007a) find that
studies demonstrates that central bank talk statements generally move financial markets
can move financial markets substantially. in the intended direction: Statements sug-
An early contribution to this literature is gesting tightening lead to higher rates, while
Graeme Guthrie and Julian Wright (2000), statements suggesting easing lead to lower
who, without distinguishing among different rates. They identify the largest effects for the
types of communication, provide compelling ECB; statements containing some reference
evidence that communication by the RBNZ to monetary policy inclinations move interest
has sizable and long-lasting effects on inter- rates an average of 1.5–2.5 basis points.
est rates across the maturity spectrum. Musard-Gies (2006) finds similar results
Communication by the Federal Reserve for the ECB, although the short end of the
also affects financial markets. As one indica- yield curve reacts more sharply to state-
tion of magnitudes, table 1 reproduces the ments than the long end. In stark contrast,
results of the seminal study by Kohn and statements pertaining to the economic out-
Sack (2004). It is apparent that both FOMC look are found to affect mainly U.S. markets,
statements and Greenspan testimonies move especially at the medium-to-long end of the
markets. The statements affect interest rates yield curve, but not (or much less) markets in
over short-to-medium-term horizons, while the United Kingdom or the euro area. Marek
testimonies by then Chairman Greenspan
affect the entire yield curve. In contrast,
Kohn and Sack find no significant effects of 27 Details of timing may matter here. The Inflation

other Greenspan speeches on financial mar- Reports are found to move markets only when the fre-
quency of the data is increased from daily (as in Kohn and
kets. Reeves and Sawicki (2007), following Sack) to intraday. This is presumably because, with rela-
the same approach, find similar evidence for tively low frequency data, the effects of other news gets
financial market effects of Bank of England mixed up with the communication effects. Using daily
data, Connolly and Kohler (2004) and Jaghit S. Chadha
communications. In particular, the release of and Charles Nolan (2001) also fail to identify effects of the
the MPC minutes and the Bank’s Inflation Inflation Report on financial markets.
Blinder et al.: Central Bank Communication and Monetary Policy 931

Table 1
Effects of Federal Reserve Communication on the Unconditional Variance of Asset Prices
Increase in unconditional variance due to
FOMC statements Greenspan testimony Greenspan speeches
Federal funds futures
  3 months ahead 24.1*** 10.0** 1.0
Treasury yields
  2-year 37.5** 41.4*** 4.3
  10-year 16.4 37.1*** 3.9
Treasury forward rates
  0 to 1 year ahead 28.9** 21.8** 2.1
  1 to 2 years ahead 49.7** 69.3*** 6.2
  2 to 3 years ahead 43.7 57.8*** 4.1
  3 to 4 years ahead 28.7 45.2** 1.8
Notes: Numbers report the increase in variance (corrected for the effect of macroeconomic and monetary policy surprises)
relative to that observed over the week preceding the communication events. Based on daily data.
  *** indicate significance at the 99 percent level.
    ** indicate significance at the 95 percent level.
     * indicate significance at the 90 percent level.
  Sample period: 1989–2003.

Source: Kohn and Sack (2004).

Table 2
Effects of Interviews and Speeches by Central Bank Committee Members on Interest Rates

Federal Reserve Bank of England ECB


3-month interest rates 0.97*** 0.46*** 2.05***
6-month interest rates 0.46 0.80*** 1.57***
1-year interest rates 0.88** 0.95** 2.47***
2-year interest rates 1.01* 0.13 2.48***
5-year interest rates 1.14* 0.15 1.96***
10-year interest rates 0.76 –0.21 0.61
20-year interest rates 0.64 0.02 0.44

Notes: The table shows the response of interest rates in basis points to speeches and interviews given by committee mem-
bers of the different central banks (FOMC members, MPC members and Governing Council members, respectively),
containing some non-neutral reference to monetary policy inclinations. Statements that are perceived as suggesting a
policy tightening inclination coded as +1, statements that are perceived as suggesting a policy easing inclination as 21.
Based on daily data.
  *** indicate significance at the 99 percent level.
    ** indicate significance at the 95 percent level.
     * indicate significance at the 90 percent level.
  Sample period: 1999 (1997 for Bank of England) to 2004
Source: Ehrmann and Fratzscher (2007a).
932 Journal of Economic Literature, Vol. XLVI (December 2008)

Rozkrut et al. (2007) repeat this analysis for are important market movers—a result cor-
the Czech Republic, Hungary, and Poland. roborated by Gürkaynak, Sack, and Swanson
They find that speeches about ­monetary (2005). In the principal components approach
policy move markets, although to differ- of the latter (described above), the “path fac-
ent degrees in each country, but speeches tor,” which is related to the forward-looking
about the economic outlook do not. Finally, communication in the statement, affects
Andersson, Dillen, and Sellin (2006) find that interest rates across the yield curve—and
speeches by the Riksbank’s Executive Board appears to be the dominant factor at the long
members are important market movers. end. Similarly, Brand, Buncic, and Turunen
Overall, the results from several countries (2006) find that the effects of the ECB’s
are qualitatively consistent, even though communications are particularly sizable at
heterogeneity in both communication prac- the long end of the yield curve.
tices and monetary policy strategies makes The fact that the ECB reveals its decisions
direct comparisons difficult. Central bank forty-five minutes before it provides expla-
statements and speeches do get the atten- nations in press conferences offers an inter-
tion of financial market participants, which esting opportunity to separate the effects of
finds its way into market prices. The quan- the two different types of communication
titative magnitudes naturally differ across events. Ehrmann and Fratzscher (forth-
central banks. For example, the comprehen- coming-a) find that press conferences have,
sive study of different communication tools on average, larger effects on asset prices
in six central banks by Connolly and Kohler than do announcements of policy decisions.
(2004) finds that monetary policy reports in Interestingly, these larger effects on interest
Australia, Canada, New Zealand, and the rates are accompanied by smaller effects on
United States provide information that sig- volatility, indicating that the signal to noise
nificantly affects markets’ expectations, and ratio of the press conference communication
thus interest rate futures.28 Parliamentary is high. Why? Their results suggest that the
hearings affect futures rates in Australia, Q&A session gives journalists the opportu-
New Zealand, the United Kingdom (albeit nity to digest the information provided, to
only marginally), and the United States, but compare it to their own prior information,
not in Canada or the euro area. However, and to ask clarifying questions where nec-
where they have effects, the impacts of hear- essary. In fact, there is evidence for such a
ings on interest rate expectations are the larg- clarification role. When macroeconomic
est among the various communication tools. uncertainty—proxied by the diversity of
There is also some evidence suggesting that expectations about data releases—is higher,
the financial markets respond more strongly the market’s response to the release of the
when communication is from the head of the monetary policy decision is more muted, sug-
central bank (Andersson, Dillen, and Sellin gesting that market participants wait for the
2006; Ehrmann and Fratzscher 2007a). clarification provided in the press conference
Central banks explain their policy deci- before they react.
sions in different ways. As noted above, Kohn Central banks often provide substantially
and Sack (2004) find that FOMC statements more detailed explanations of monetary pol-
icy decisions in the minutes of policy meet-
ings. For these to have meaningful news
28 For the United States, this effect also incorporates
content for financial markets, however, they
any effects of the Congressional testimony that coincides
with publication of the report, which might explain the must be released before the subsequent
relatively large effect that they find. meeting of the committee. In recent years,
Blinder et al.: Central Bank Communication and Monetary Policy 933

both the Federal Reserve and the Bank of In summary, while the evidence is mixed,
England have shortened the lag in releasing on balance it seems fair to say that, if the cen-
their minutes, moving it from after to before tral bank speaks clearly, its statements can
the subsequent meeting. As shown by Reeves (and apparently do) affect a variety of asset
and Sawicki (2007) and Reinhart and Sack prices in non-negligible ways. While financial
(2006), there are discernable financial mar- market reactions depend on the details of the
ket reactions only with more timely release. communications, especially their timeliness
and news content, central bank talk appar-
4.4 Communication about Exchange Rates
ently does contain news that is quickly incor-
Central bank communication is not con- porated into asset prices.
fined to monetary policy proper. While it is
4.5 Uncertainty in Central Bank
beyond the scope of this survey, some cen-
Communication
tral banks make occasional statements on
exchange rates, and this special type of cen- That said, we hasten to highlight an issue
tral bank communication has spawned a rap- that has received scant attention in the lit-
idly-growing literature of its own. We cannot erature on central bank communication,
summarize the literature on “oral interven- yet might be highly relevant to its success:
tions” here, but a few findings are worth not- namely, the role of uncertainty in central
ing because they dovetail nicely with those bank communication.
on interest rates: Uncertainty can arise from various
• Given the small scale of actual forex sources. First, inconsistent signals can arise
interventions relative to today’s massive when different members of a monetary pol-
and liquid markets29 —whether sterilized icy committee say different things—whether
or unsterilized, words may be more effec- intentionally (e.g., by conducting a debate in
tive than actions, perhaps because words public) or unintentionally (e.g., via uncoor-
serve as the focal point that coordinates dinated communication). As Blinder (2007)
the actions of market participants (Lucio notes, “A central bank that speaks with a
Sarno and Mark P. Taylor 2001).30 cacophony of voices may, in effect, have no
• Communication can complement inter- voice at all” (p. 114). On the other hand,
vention as well as substitute for it Bernanke (2004) argues that “the willingness
(Fratzscher 2008, forthcoming). of FOMC members to present their individ-
• Central bank talk about exchange rates, ual perspectives in speeches and other public
just like central bank talk about interest forums provides the public with useful infor-
rates, may have to substitute for inter- mation about the diversity of views and the
est rate cuts when the nominal inter- balance of opinion on the Committee.” Both
est rate reaches its zero lower bound views have validity. Whether communicating
(Lukas Burkhart and Andreas M. Fischer individual committee members’ views to the
forthcoming). public enlightens or confuses is ultimately
an empirical issue. And whether doing so is
advisable or inadvisable depends, inter alia,
29 See Kathryn M. E. Dominguez and Jeffrey A.
on whether the committee has group or indi-
Frankel (1993). vidual accountability.
30 For some empirical evidence, see Michel Beine, Gust As indicated by Bernanke’s comment,
Janssen, and Christelle Lecourt (forthcoming), Fratzscher FOMC members do indeed communicate
(2006), Jansen and De Haan (2007b), Dominguez and
Freyan Panthaki (2007), and Pierre L. Siklos and Martin with disparate voices (Ehrmann and Fratz-
T. Bohl (2008). scher 2007a), whereas the ECB ­generally
934 Journal of Economic Literature, Vol. XLVI (December 2008)

speaks more with a single voice.31 However, evidence that markets react more strongly
this has not always been so. Jansen and De to statements by the Riksbank’s governor.
Haan (2006) show that communication about Ehrmann and Fratzscher (2007a) show the
monetary policy inclinations by individual same for the Fed chairman. But they find
members of the Governing Council was rela- that, in the case of the ECB’s more collegial
tively high in the initial years of the ECB, but approach to communication, markets react
then declined over time. more equally to statements by all Governing
To what extent do consistency and other Council members.
characteristics of central bank communi- Central bank communication is also a two-
cation practices affect the predictability of way street: It must have both a transmitter
monetary policy? Ehrmann and Fratzscher and a receiver, and either could be the source
(2007b) find that more dispersed communi- of uncertainty or confusion. Moreover, on
cation on Federal Reserve monetary policy is the receiving end, the same message might
associated with less predictable policy deci- be interpreted differently by different lis-
sions at short- and medium-term horizons, teners who may have different expectations
and that the magnitude of this effect is size- or believe in different models. One good
able. In another recent study, Rozkrut et al. example is provided by Fracasso, Genberg,
(2007) corroborate this result for the central and Wyplosz (2003), who use survey data
bank of Hungary; but for the Czech National to analyze whether inflation reports of cen-
Bank, more dispersed communication actu- tral banks reach different audiences equally
ally enhances the predictability of policy well. They find that the same inflation report
decisions. is perceived differently by different respon-
There is also evidence that the voting dents, and that interest rate surprises tend to
records of the Bank of England’s MPC mem- increase with the divergence in perceptions.
bers help predict future policy changes (Petra Another example comes from De Haan,
Gerlach-Kristen 2004). Casting a minority Fabian Amtenbrink, and Sandra Waller
vote appears to be a bigger step, and there- (2004), who report substantial differences
fore one that carries more information, than between newspaper reports published the day
merely expressing a personal dissenting view after ECB policy decisions in the Financial
in public. Times (FT) and the Frankfurter Allgemeine
Markets, of course, adapt to a central Zeitung (FAZ). The British-based FT, which
bank’s communication style. When central is critical of the money growth pillar in the
banks give out relatively dispersed commu- ECB’s monetary policy strategy, tends to pay
nications, financial markets attempt to iden- relatively little attention to it. In sharp con-
tify pivotal committee members and attach trast, the home-town FAZ clearly supports
more weight to their statements. For exam- the idea that money should have a prominent
ple, Andersson, Dillen, and Sellin (2006) find role in the ECB’s strategy, and accordingly
gives that pillar substantial attention.
A similar phenomenon can arise when the
31 Issing (1999, pp. 508 and 510) has justified this on identical message is interpreted differently
the basis of the special circumstances in which the ECB at different points in time. As one example,
operates: “What matters for transparency is therefore
clarity as well as openness. For a new and supranational Berger, Ehrmann, and Fratzscher (2006)
institution like the ECB, it is particularly important that study the quantity and tone of reporting on
it sends clear and coherent messages to the markets and the ECB’s press conference in fifty-seven
the wider public . . . . Speaking with one voice—or at least
speaking ‘one language’—is of particular importance for international and national newspapers in the
transparency and clarity in the case of the Eurosystem.” euro area. They find that the press seems
Blinder et al.: Central Bank Communication and Monetary Policy 935

to take its cues from both prior market Before discussing the research findings,
expectations and the inflationary environ- a number of caveats are in order. Michael J.
ment. The tone of the reporting is generally Dueker and Fischer (2006) identify various
more ­critical when a policy decision takes problems encountered by studies comparing
markets by surprise. The decisions are also inflation targeters with nontargeters. First,
discussed less favorably if the most recent even if a central bank is recognized as an
euro area inflation figures exceed 2 percent, inflation targeter, the dating of the start of
thus violating the ECB’s definition of price the new policy regime is often contentious.
stability. For example, Bernanke et al. (1999) date the
start of inflation targeting (IT) regimes by the
announcement of the target, while Laurence
5.  The Impact of Central Bank
Ball and Niamh Sheridan (2005) date it by its
Communication on Inflation Performance
implementation. Second, the classification of
We turn now to the harder but perhaps inflation targeters is not always clear. Some
more important question of whether and central banks might declare themselves to be
how central bank communication affects the inflation targeters but, in practice, lack some
ultimate goal variables of monetary policy. As of the institutional features that are generally
we have noted, the combination of long lags associated with inflation targeting. Others
in monetary policy and the myriad influences might insist that they are not inflation tar-
on macro outcomes make it virtually impos- geters but share the characteristic practices
sible to isolate a link between a specific com- (Kenneth N. Kuttner 2004). Third, results
munication event (such as a speech) and, say, seem to depend crucially on the choice of
inflation. So this section takes a longer-term control group, which varies widely across
perspective by summarizing the literature on studies.
how (if at all) differences in communication Last, and certainly not least, there is a
strategies, either over time or across central potentially serious endogeneity problem. Just
banks, influence economic outcomes. Since ask yourself: Does the adoption of IT cause
there is almost no evidence on the effects of inflation to decline, or does high inflation
central bank communication strategies on cause a country to adopt IT? Another ver-
the real economy, we focus entirely on actual sion of this question is a bit more complex.
and expected inflation. The introduction of IT can be thought of as
a commitment device that changes the cen-
5.1 Anchoring Inflation Expectations
tral bank’s policy and, therefore, economic
Since a large number of central banks outcomes. But an alternative interpretation
have adopted formal inflation targeting, is that the formal adoption of IT is a mani-
and others (such as the Swiss National Bank festation of a country’s intent to focus more
and the ECB) provide quantitative defini- vigorously on reducing inflation. For exam-
tions of price stability, a substantial body of ple, Stephen G. Cecchetti and Ehrmann
research has emerged to assess the effects of (2002) argue that central bankers’ aversion
an explicit numerical inflation target on infla- to ­inflation variability increased during the
tion outcomes. The central, but not the only, 1990s in many countries, including both
questions are whether clear and consistent inflation targeters and nontargeters.
communication on the long-term inflation Subject to these caveats, what does the
objective (a) anchors the public’s long-run evidence show? A number of studies find
inflation expectations and (b) reduces infla- that IT has succeeded in anchoring inflation
tion forecast errors. expectations. Various approaches appear in
936 Journal of Economic Literature, Vol. XLVI (December 2008)

the ­literature. David R. Johnson (2003) esti- similar results for the United Kingdom prior
mates the determinants of expected inflation to the Bank of England’s independence, as
(p e) in the period prior to the announce- shown by the first set of results in table 4.
ment of an inflation target, and then uses the However, since the Bank of England became
estimated model to predict p e under the IT an independent, inflation-targeting central
regime. The difference between actual and bank in 1998, short-term nominal rates in
predicted p e is then taken to measure the the United Kingdom are still responsive to
effect of inflation targeting. He finds large macroeconomic news but long-term inflation
reductions in the level of expected inflation expectations are not—indicating that expec-
in the period after the announcement of tations have been successfully anchored.
inflation targets in Australia, Canada, New Interestingly, this evidence suggests that it is
Zealand, and Sweden, but not in the United not just the existence of an explicit inflation
Kingdom. objective that matters, but also the institu-
Another approach, adopted in Johnson tional setup of the central bank.
(2002) and in Andrew T. Levin, Fabio M. Finally, Jansen and De Haan (2007a) find
Natalucci, and Jeremy M. Piger (2004), com- evidence that communication by the ECB
pares targeting to nontargeting countries. regarding risks to price stability (measured
Controlling for country, year, trend-inflation, by the frequency and intensity of the key-
and business-cycle effects, Johnson detects a word “vigilance”) reduced high-frequency
reduction in inflationary expectations in the inflationary expectations slightly in the
IT countries but not in the control group. second half of 2005, when a change in the
Levin, Natalucci, and Piger provide evidence ECB’s policy stance became increasingly
that long-term inflation forecasts exhibit a likely.
highly significant correlation with a three- These studies all suggest notable effects of
year moving average of lagged inflation in inflation targeting on inflation expectations.
the control group, but not in the IT group, However, other authors find that expecta-
suggesting that the inflation-targeting cen- tions are equally well anchored in their con-
tral banks have successfully de-linked expec- trol groups, casting doubt on whether the
tations from realized inflation. effect identified in the previous studies is
Gürkaynak, Levin, and Swanson (2006) causal. Efrem Castelnuovo, Sergio Nicoletti-
take a different approach to the same ques- Altimari, and Diego Rodriguez Palenzuela
tion. They show that long-term inflation (2003) find that long-term inflation expec-
expectations derived from index-linked tations are well-anchored in all countries
bonds are less responsive to macroeconomic in their sample except Japan—regardless of
data releases and monetary policy announce- whether the central bank has an inflation tar-
ments in Sweden and the United Kingdom, get, a quantitative definition of price stability,
two inflation targeting countries, than they or no quantified target at all.
are in the United States, which has no A related strand of the literature focuses
explicit inflation objective. Table 3 reports on the uncertainty of inflation forecasts and
their results for the United States. The first inflation forecast errors rather than on the
column shows that one-year nominal interest forecast itself. Ippei Fujiwara (2005) dem-
rates respond to the surprise component of onstrates that the publication of the Bank of
macroeconomic releases. The second column Japan’s inflation and growth forecasts influ-
indicates that market expectations of infla- ences those of professional forecasters; in par-
tion over ten years are also affected signifi- ticular, the heterogeneity across ­forecasters
cantly. Gürkaynak, Levin, and Swanson find decreases after the Bank of Japan’s forecast
Blinder et al.: Central Bank Communication and Monetary Policy 937

Table 3
Responsiveness of U.S. Interest Rates and Inflation Expectations to Macroeconomic News,
1998–2005

One year nominal rate Inflation compensation


Capacity Utilization 1.57*** 0.52
Consumer Confidence 1.46*** 0.43
core CPI 0.98 1.42**
real GDP (advance) 2.17** 1.77**
Initial Jobless Claims 21.13*** 20.51**
NAPM/ISM Manufacturing 2.28*** 1.48**
New Home Sales 0.53 1.42***
Nonfarm Payrolls 4.44*** 0.52
Retail Sales 1.69*** 0.88
Unemployment Rate 20.97 20.11
Monetary Policy 0.23** 20.12

Notes: The coefficients denote the response of, respectively, one-year spot nominal rates and inflation compensation
(based on the different reactions between the nominal and real one-year forward rate from nine to ten years ahead) to
standardized macroeconomic surprises.
  *** indicate significance at the 99 percent level.
    ** indicate significance at the 95 percent level.
     * indicate significance at the 90 percent level.
Source: Gürkaynak, Levin, and Swanson (2006).

is published. Siklos (2002) reports that the have converged to levels of accuracy simi-
adoption of an inflation target leads to lower lar to those observed in the control group of
forecast errors in a panel study of twelve nontargeters.
countries. Overall, inflationary expectations appear to
But subsequent studies do not support his be generally well anchored, and inflation fore-
result. Johnson (2002) compares the stan- cast errors small, in IT countries. And studies
dard deviation of expected inflation and the of countries undergoing regime changes sug-
average absolute size of inflation forecast gest a causal link between adopting IT and
errors in a panel of eleven countries, five with anchoring inflation expectations. However,
inflation targets and six without, over 1984 cross-sectional comparisons yield more ambig-
to 2000. He finds that, when actual inflation uous results; the choice of the control group is
falls, uncertainty about future inflation falls apparently crucial. So communication of an
in both targeting and nontargeting countries. explicit inflation target is surely not the only
There is also little evidence that average way to anchor expectations.
absolute forecast errors are lower in target-
5.2 Inflation and its Dynamics
ing countries. These findings are in line with
those of Vittorio Corbo, Oscar Landerretche, Building greater trust by credibly com-
and Klaus Schmidt-Hebbel (2002), who ana- municating a long-term inflation objective
lyze how one-step-ahead inflation forecast may also affect inflation itself. Most obvi-
errors (constructed from estimated country ously, the introduction of IT might reduce
VARs) have evolved over time. They find that the average level of inflation—which was
countries that adopted inflation targeting certainly the intent of the inventors of IT
938 Journal of Economic Literature, Vol. XLVI (December 2008)

Table 4
Responsiveness of UK Interest Rates and Inflation Expectations to Macroeconomic News

1993–1997 1998–2005
One-year nominal Inflation One-year nominal Inflation
rate compensation rate compensation
Average Earnings 3.23*** 0.15 1.81*** 20.26
Real GDP (Preliminary) 1.75 1.80** 2.04*** 20.49
Manufacturing Production 0.76 0.33 1.26*** 20.04
Producer Price Index 2.13*** 2.22** 0.21 20.22
Core Retail Price Index 2.39*** 2.60*** 2.60*** 20.76
Retail Sales 2.17*** 20.19 1.58*** 21.18***
Monetary Policy 0.67*** 20.60*** 0.72*** 20.13

Notes: The coefficients denote the response of, respectively, one-year spot nominal rates and inflation compensation
(based on the different reactions between the nominal and real one-year forward rate from nine to ten years ahead) to
standardized macroeconomic surprises.
  *** indicate significance at the 99 percent level.
    ** indicate significance at the 95 percent level.
     * indicate significance at the 90 percent level.
Source: Gürkaynak, Levin, and Swanson (2006).

in New Zealand and most (if not all) of its tion levels drop after targeting began. But,
early adopters. IT might also affect the ­controlling for the initial level of inflation,
time-series properties of inflation. If agents the decline is similar for targeters and non-
believe that inflation will return to its tar- targeters. Ball and Sheridan suggest that
get level more quickly when it deviates from this finding reflects the endogeneity issue
that target, then inflation should be less mentioned earlier: Countries that adopted
persistent. King (2002) argues that U.K. IT had above-average inflation prior to
inflation has been lower, more stable, and adoption.
less persistent since inflation targeting was Luke Willard (2006), after dealing with
introduced. In support of this argument, the endogeneity problem in a variety of
Kuttner and Adam S. Posen (1999) report ways, supports Ball and Sheridan’s conclu-
that the introduction of inflation targeting sions. Similarly, Shu Lin and Haichun Ye
in Canada and the United Kingdom led to (2007) study twenty-two industrial coun-
lower inflation; but their findings for New tries over the period 1985–99 and address
Zealand are more mixed. the selection problem by using a variety of
By contrast, taking a cross-sectional per- recently developed propensity score match-
spective, Ball and Sheridan (2005) find no ing methods. Their main result is that infla-
empirical evidence that inflation targeting tion targeting has no significant beneficial
improves performance once you control effects on targeting countries’ inflation or
for regression to the mean. (High inflation inflation variability. However, Marco Vega
tends to come down.) IT leads neither to and Diego Winkelried (2005), using a simi-
lower levels of inflation, nor to less inflation lar methodology as Lin and Ye but for a
variability, nor to less persistence. It is true larger set of countries, do find that IT has
that IT-adopting countries saw their infla- helped reduce the level and volatility of
Blinder et al.: Central Bank Communication and Monetary Policy 939

inflation. Their results are robust to alter- of inflation targets. Table 5 reports ­estimates
native definitions of treatment and control of inflation persistence for a number of
groups. Analyzing ten OECD countries, countries, separated by monetary regimes
Alvaro Angeriz and Philip Arestis (2008) ­according to Benati’s classification. It is evi-
also find that IT has gone hand in hand with dent that inflation persistence is particularly
low inflation. However, IT was often intro- low under inflation targeting. But it is also
duced after inflation had already begun its low under the Bretton Woods regime.
downward trend. They also find that nontar- In conclusion, the evidence suggests that
geting countries have succeeded in achiev- adopting an inflation target may have benefi-
ing and maintaining low inflation, but their cial effects by lowering inflation, by delinking
control group consists of only two cases: the long-run inflation expectations from short-
United States and the European Economic run data, and by reducing inflation persis-
and Monetary Union. tence. However, these estimated benefits may
What is to be made of these disparate reflect a kind of selectivity bias: They seem to
results? Mishkin and Schmidt-Hebbel accrue primarily to countries that succeed in
(2007) emphasize that a judgment on the stabilizing inflation. There appears to be no
effectiveness of inflation targeting in bring- systematic difference in the economic per-
ing down inflation depends crucially on the formance of low-inflation countries with and
choice of the control group. In particular, without explicit inflation targets.
IT does not appear to lead to performance Accordingly, we conclude that inflation
superior to that of a group of success- targeting is one way, but certainly not the
ful nontargeters. The main benefit they only way, to control inflation and inflation-
see in inflation targeting is as a disciplin- ary expectations. One clear alternative is
ary device that helps potentially wayward establishing an anti-inflation track record
countries move closer to the performance that allows economic agents to make rea-
of this successful control group. Bernanke sonably accurate inferences about the cen-
et al. (1999) even argue that inflation tar- tral bank’s objectives and strategy. For
gets have often been adopted after central example, Linda S. Goldberg and Michael W.
banks have succeeded in lowering inflation. Klein (2005) analyze the response of euro-
The primary purpose is to lock in earlier area asset prices to U.S. inflation news, and
disinflationary gains. find that the ECB’s credibility in financial
Taking a broader perspective, Antonio markets improved after it went into its first
Fatás, Ilian Mihov, and Andrew K. Rose tightening cycle.
(2007) find for a sample of forty-two coun- One issue that remains open is the role of
tries over forty years that inflation is lower communication during a disinflation. One
in the presence of some quantitative tar- interesting comparison was made by Michael
get when other determinants of inflation D. Bordo et al. (2007), who show in three
are taken into account. According to their U.S. disinflation episodes that it is particu-
results, it is quantification that matters, not larly important for a central bank to com-
so much the exact target; inflation, exchange municate an aggressive policy stance, if it
rate, or monetary targets are all linked to starts with relatively low credibility. It would
lower inflation. be nice to know whether the introduction of
Turning to inflation persistence, Levin, an inflation target can facilitate the disinfla-
Natalucci, and Piger (2004) and Luca Benati tion process under these circumstances and
(2008) find it to be considerably lower— what alternatives a central bank should (and
sometimes even negative!—in the presence should not) consider.
940 Journal of Economic Literature, Vol. XLVI (December 2008)

Table 5
Measures of Inflation Persistence
Canada Sweden New Zealand United Kingdom

90 percent 90 percent 90 pecent 90 percent


confidence confidence confidence confidence
r interval r interval r interval r interval

Interwar period 0.58 [0.38; 0.81] 0.69 [0.42; 1.02] 0.98 [0.69; 1.06] ­— —
Bretton Woods 0.71 [0.54; 0.89] 0.29 [20.01; 0.59] 0.29 [0.02; 0.56] — —
1971 to inflation targeting 0.91 [0.72; 1.04] 0.53 [0.12; 1.02] 0.82 [0.67; 1.01] 0.93 [0.89; 0.93]
Inflation targeting 20.25 [20.73; 0.22] 0.37 [20.24; 1.05] 0.41 [0.14; 0.72] 20.19 [20.54; 0.15]

Notes: The parameter r is the sum of the autoregressive coefficients in a regression of inflation on its own lags. If the con-
fidence interval includes a value of 1, the null hypothesis of a unit root cannot be rejected.
Source: Benati (2008).

6.   Assessment and Issues for


Future Research countries. With only a few exceptions,
the empirical studies to date suggest that
To what extent, then, is central bank com- more and better central bank communi-
munication a useful instrument for policy- cation contributed to this improvement
makers? In 2001, Blinder et al. wrote: “To by “reducing noise.” We are tempted to
date, there is no research to report on” (p. 9). call this issue “closed” but, of course,
Since then, there has been a veritable explo- future research can always overturn a
sion of research on central bank communica- previous conclusion.
tion. While the evidence reviewed here is not • That said, the predictability of monetary
entirely one-sided, it seems safe to conclude policy appears to be degraded somewhat
that communication can be an important and when central banks speak with too many
powerful part of the central bank’s toolkit conflicting voices—as sometimes happens
since it has the ability to move financial mar- at the Federal Reserve, for example.
kets, to improve the predictability of monetary • What might be called “short-run” cen-
policy, and the potential to help the monetary tral bank communication—that is, dis-
authorities achieve macroeconomic objectives closing central bank views on, e.g., the
such as low and stable inflation. outlook for the economy and monetary
More specifically, our review of the rapidly policy—has a wide variety of impacts on
growing empirical literature, virtually all of financial markets, including on the levels
which was written in this decade, suggests and volatilities of various interest rates.
that: Official statements, reports, and ­minutes
• No consensus has yet emerged on what appear to have the clearest and most
communication policies constitute “best consistent empirical effects on financial
practice” for central banks. Practices, markets. The evidence on the impacts
in fact, differ substantially and are of speeches is more mixed. But it, too, is
evolving. mainly supportive of the idea that central
• The predictability of monetary policy bank communication “creates news.” The
decisions has improved notably in many distinctions among different forms of
Blinder et al.: Central Bank Communication and Monetary Policy 941

communication merit further empirical Another important, but barely explored,


evaluation (e.g., what works best?). issue is what constitutes “optimal” commu-
• The limited number of studies that try to nication policy and how that depends on the
assess the directional intent of the central institutional environment in which a central
bank’s messages generally find that mar- bank operates, the nature of its decision-mak-
kets move in the “right” direction—that ing process, and the structure of its monetary
is, what used to be called “announcement policy committee. For example, the Bank of
effects” help the central bank rather than England has a committee of nine members
hinder it. But there has been relatively lit- who are individually accountable, while
tle such research to date. We need more both the FOMC and the ECB Governing
evidence in order to be convinced on this Council are larger committees with group
point, plus further analysis of how, if at accountability, and the Reserve Bank of New
all, creating this type of directional news Zealand still has a single decisionmaker.
raises welfare. Central banks need to tailor their commu-
• Regarding what might be called “long- nication strategies to these and other institu-
run” central bank communication, the tional features. Research on that important
empirical evidence so far is largely lim- topic has barely begun.
ited to one set of questions: the effects Finally, virtually all the research to date
of inflation targeting (IT) on inflation has focused on central bank communication
outcomes. Those are good questions, but with the financial markets. It may be time
not the only ones. So research on the to pay some attention to communication
links between communication and other with the general public. Admittedly, study-
macro variables is essential. ing communication with the general public
• For a variety of reasons, isolating clear will pose new challenges to researchers—not
effects of IT turns out to be harder than least because financial market prices will be
might be expected. But there is clear evi- less relevant. But the issues are at least as
dence that IT helps anchor inflationary important. In the end, it is the general pub-
expectations. (However, it is not the only lic that gives central banks their democratic
way to do so.) The evidence that adopting legitimacy, and hence their independence.
IT leads to lower or less variable inflation
is far less compelling. On such questions, References
the choice of the control groups seems to
Amato, Jeffery D., Stephen Morris, and Hyun Song
be critical. Shin. 2002. “Communication and Monetary Policy.”
This list of research findings constitutes Oxford Review of Economic Policy, 18(4): 495–503.
a quantum leap over what we knew at Andersson, Malin, Hans Dillén, and Peter Sellin.
2006. “Monetary Policy Signaling and Movements
the start of the decade, which was almost in the Term Structure of Interest Rates.” Journal of
nothing. But there is a lot more to learn. Monetary Economics, 53(8): 1815–55.
For example, the publication of projected Angeriz, Alvaro, and Philip Arestis. 2008. “Assessing
Inflation Targeting through Intervention Analysis.”
paths for the central bank’s policy rate Oxford Economic Papers, 60(2): 293–317.
appears to be the “new frontier” in central Archer, David. 2004. “Communication with the Pub-
bank communication. But it has been prac- lic.” In Practical Experience with Inflation Target-
ing, 145–55. Prague: Czech National Bank.
ticed in so few countries for so few years Ball, Laurence, and Niamh Sheridan. 2005. “Does
that we have little empirical knowledge of Inflation Targeting Matter?” In The Inflation-Tar-
its effects as yet. As more data accumu- geting Debate, ed. Ben S. Bernanke and Michael
Woodford, 249–76. Chicago and London: University
lates, this should be a high-priority area of Chicago Press.
for future research. Beine, Michel, Gust Janssen, and Christelle Lecourt.
942 Journal of Economic Literature, Vol. XLVI (December 2008)

Forthcoming. “Should Central Bankers Talk to the Relation between Narrative and Money Market
Foreign Exchange Markets?” Journal of Interna- Indicators of Monetary Policy.” Economic Inquiry,
tional Money and Finance. 33(1): 24–44.
Benati, Luca. 2008. “Investigating Inflation Persis- Brand, Claus, Daniel Buncic, and Jarkko Turunen.
tence across Monetary Regimes.” Quarterly Journal 2006. “The Impact of ECB Monetary Policy Deci-
of Economics, 123(3): 1005–60. sions and Communication on the Yield Curve.”
Berardi, Michele, and John Duffy. 2007. “The Value European Central Bank Working Paper 657.
of Central Bank Transparency When Agents Are Brunner, Karl. 1981. “The Art of Central Banking.”
Learning.” European Journal of Political Economy, University of Rochester Center for Research in Gov-
23(1): 9–29. ernment Policy and Business Working Paper GPB
Berger, Helge, Jakob De Haan, and Jan-Egbert Sturm. 81-6.
2006. “Does Money Matter in the ECB Strategy? Buiter, Willem H. 1999. “Alice in Euroland.” Journal of
New Evidence Based on ECB Communication.” Common Market Studies, 37(2): 181–209.
CESifo Working Paper 1652. Burkhart, Lukas, and Andreas M. Fischer. Forth-
Berger, Helge, Michael Ehrmann, and Marcel Fratz- coming. “Communicating Policy Options at the
scher. 2006. “Monetary Policy in the Media.” Euro- Zero Bound.” Journal of International Money and
pean Central Bank Working Paper 679. Finance.
Bernanke, Ben S. 2004. “Fedspeak.” Remarks  presented Castelnuovo, Efrem, Sergio Nicoletti-Altimari, and
at the annual meeting of the American Economic Diego Rodriguez Palenzuela. 2003. “Definition of
Association, San Diego. http://www.federalreserve. Price Stability, Range and Point Inflation Targets:
gov/boarddocs/speeches/2004/200401032/default. The Anchoring of Long-Term Inflation Expecta-
htm. tions.” In Background Studies for the ECB’s Evalu-
Bernanke, Ben S. 2007. “Federal Reserve Communi- ation of Its Monetary Policy Strategy, ed. Otmar
cations.” Speech presented at the Cato Institute 25th Issing, 43–90. Frankfurt: European Central Bank.
Annual Monetary Conference, Washington, D.C. Cecchetti, Stephen G., and Michael Ehrmann. 2002.
http://www.federalreserve.gov/newsevents/speech/ “Does Inflation Targeting Increase Output Volatil-
bernanke20071114a.htm. ity? An International Comparison of Policymakers’
Bernanke, Ben S., Thomas Laubach, Frederic S. Mish- Preferences and Outcomes.” In Monetary Policy:
kin, and Adam S. Posen. 1999. Inflation Targeting: Rules and Transmission Mechanisms, ed. Norman
Lessons from the International Experience. Princ- Loayza and Klaus Schmidt-Hebbel, 247–74. San-
eton: Princeton University Press. tiago: Central Bank of Chile.
Bernanke, Ben S., Vincent R. Reinhart, and Brian Chadha, Jagjit S., and Charles Nolan. 2001. “Inflation
Sack. 2004. “Monetary Policy Alternatives at the Targeting, Transparency and Interest Rate Volatility:
Zero Bound: An Empirical Assessment.” Brookings Ditching ‘Monetary Mystique’ in the U.K.” Journal
Papers on Economic Activity, 2: 1–78. of Macroeconomics, 23(3): 349–66.
Bernoth, Kerstin, and Jürgen von Hagen. 2004. “The Chappell, Henry W., Rob Roy McGregor, and Todd
Euribor Futures Market: Efficiency and the Impact Vermilyea. 2005. Committee Decisions on Mon-
of ECB Policy Announcements.” International etary Policy: Evidence from Historical Records of
Finance, 7(1): 1–24. the Federal Open Market Committee. Cambridge
Blinder, Alan S. 1998. Central Banking in Theory and and London: MIT Press.
Practice. Cambridge and London: MIT Press. Clarida, Richard, Jordi Gali, and Mark Gertler. 1999.
Blinder, Alan S. 2004. The Quiet Revolution: Central “The Science of Monetary Policy: A New Keynesian
Banking Goes Modern. New Haven and London: Perspective.” Journal of Economic Literature, 37(4):
Yale University Press. 1661–1707.
Blinder, Alan S. 2007. “Monetary Policy by Commit- Connolly, Ellis, and Marion Kohler. 2004. “News and
tee: Why and How?” European Journal of Political Interest Rate Expectations: A Study of Six Central
Economy, 23(1): 106–23. Banks.” In The Future of Inflation Targeting, ed.
Blinder, Alan S., Charles A. E. Goodhart, Philipp Hil- Christopher Kent and Simon Guttmann, 108–34.
debrand, David Lipton, and Charles Wyplosz. 2001. Sydney: Reserve Bank of Australia.
How Do Central Banks Talk. Geneva: International Corbo, Vittorio, Oscar Landerretche, and Klaus
Center for Monetary and Banking Studies; London: Schmidt-Hebbel. 2002. “Does Inflation Targeting
Centre for Economic Policy Research. Make a Difference?” In Inflation Targeting: Design,
Blinder, Alan S., and Ricardo Reis. 2005. “Understand- Performance, Challenges, ed. Norman Loayza and
ing the Greenspan Standard.” In The Greenspan Raimundo Soto, 221–69. Santiago: Central Bank of
Era: Lessons for the Future, 11–96. Kansas City: Chile.
Federal Reserve Bank of Kansas City. Cukierman, Alex. 2008. “The Limits of Transparency.”
Bordo, Michael D., Christopher Erceg, Andrew T. Unpublished.
Levin, and Ryan Michaels. 2007. “Three Great Cukierman, Alex, and Allan H. Meltzer. 1986. “A The-
American Disinflations.” National Bureau of Eco- ory of Ambiguity, Credibility, and Inflation under
nomic Research Working Paper 12982. Discretion and Asymmetric Information.” Econo-
Boschen, John F., and Leonard O. Mills. 1995. “The metrica, 54(5): 1099–1128.
Blinder et al.: Central Bank Communication and Monetary Policy 943

De Haan, Jakob, Fabian Amtenbrink, and Sandra “Quantitative Goals for Monetary Policy.” Journal of
Waller. 2004. “The Transparency and Credibility of Money, Credit, and Banking, 39(5): 1163–76.
the European Central Bank.” Journal of Common Faust, Jon, and Lars E. O. Svensson. 2001. “Transpar-
Market Studies, 42(4): 775–94. ency and Credibility: Monetary Policy with Unob-
De Haan, Jakob, and Sylvester C. W. Eijffinger. 2000. servable Goals.” International Economic Review,
“The Democratic Accountability of the European 42(2): 369–97.
Central Bank: A Comment on Two Fairy-Tales.” Fracasso, Andrea, Hans Genberg, and Charles Wyplosz.
­Journal of Common Market Studies, 38(3): 393–407. 2003. How Do Central Banks Write? An Evaluation
Demiralp, Selva, and Oscar Jorda. 2002. “The of Inflation Targeting Central Banks. Geneva: Inter-
Announcement Effect: Evidence from Open Mar- national Center for Monetary and Banking Studies;
ket Desk Data.” Federal Reserve Bank of New York London: Centre for Economic Policy Research.
­Economic Policy Review, 8(1): 29–48. Fratzscher, Marcel. 2006. “On the Long-Term Effec-
Dominguez, Kathryn M. E., and Jeffrey A. Frankel. tiveness of Exchange Rate Communication and
1993. “Does Foreign-Exchange Intervention Mat- Interventions.” Journal of International Money and
ter? The Portfolio Effect.” American Economic Finance, 25(1): 146–67.
Review, 83(5): 1356–69. Fratzscher, Marcel. 2008. “Oral Interventions versus
Dominguez, Kathryn M. E., and Freyan Panthaki. Actual Interventions in FX Markets—An Event-
2007. “The Influence of Actual and Unrequited Study Approach.” Economic Journal, 118(530):
Interventions.” International Journal of Finance 1079–1106.
and Economics, 12(2): 171–200. Fratzscher, Marcel. Forthcoming. “Communication  and
Dueker, Michael J., and Andreas M. Fischer. 2006. Exchange Rate Policy.” Journal of Macroeconomics.
“Do Inflation Targeters Outperform Non-target- Fujiwara, Ippei. 2005. “Is the Central Bank’s Publica-
ers?” Federal Reserve Bank of St. Louis Review, tion of Economic Forecasts Influential?” Economics
88(5): 431–50. Letters, 89(3): 255–61.
Eggertsson, Gauti B., and Michael Woodford. 2003. Garfinkel, Michelle R., and Seonghwan Oh. 1995.
“The Zero Bound on Interest Rates and Optimal “When and How Much to Talk: Credibility and Flex-
Monetary Policy.” Brookings Papers on Economic ibility in Monetary Policy with Private Information.”
Activity, 1: 139–211. Journal of Monetary Economics, 35(2): 341–57.
Ehrmann, Michael, and Marcel Fratzscher. 2007a. Geraats, Petra M. 2002. “Central Bank Transparency.”
“Communication by Central Bank Committee Mem- Economic Journal, 112(483): F532–65.
bers: Different Strategies, Same Effectiveness?” Gerlach, Stefan. 2007. “Interest Rate Setting by the
Journal of Money, Credit, and Banking, 39(2–3): ECB, 1999–2006: Words and Deeds.” International
509–41. Journal of Central Banking, 3(3): 1–45.
Ehrmann, Michael, and Marcel Fratzscher. 2007b. Gerlach-Kristen, Petra. 2004. “Is the MPC’s Voting
“Social Value of Public Information—Testing the Record Informative about Future UK Monetary Pol-
Limits to Transparency.” European Central Bank icy?” Scandinavian Journal of Economics, 106(2):
Working Paper 821. 299–313.
Ehrmann, Michael, and Marcel Fratzscher. 2007c. Goldberg, Linda S., and Michael W. Klein. 2005.
“The Timing of Central Bank Communication.” “Establishing Credibility: Evolving Perceptions of
European Journal of Political Economy, 23(1): the European Central Bank.” National Bureau of
124–45. Economic Research Working Paper 11792.
Ehrmann, Michael, and Marcel Fratzscher. 2007d. Goodfriend, Marvin. 1986. “Monetary Mystique:
“Transparency, Disclosure, and the Federal Re- Secrecy and Central Banking.” Journal of Monetary
serve.” International Journal of Central Banking, Economics, 17(1): 63–92.
3(1): 179–225. Goodfriend, Marvin. 1991. “Interest Rates and the
Ehrmann, Michael, and Marcel Fratzscher. Forthcom- Conduct of Monetary Policy.” Carnegie–Rochester
ing-a. “Explaining Monetary Policy in Press Confer- Conference Series on Public Policy, 34: 7–30.
ences.” International Journal of Central Banking. Goodhart, Charles A. E. 2001. “Monetary Transmis-
Ehrmann, Michael, and Marcel Fratzscher. Forthcom- sion Lags and the Formulation of the Policy Deci-
ing-b. “Purdah—On the Rationale for Central Bank sion on Interest Rates.” Federal Reserve Bank of St.
Silence around Policy Meetings.” Journal of Money, Louis Review, 83(4): 165–81.
Credit, and Banking. Goodhart, Charles A. E. 2005. “Dear Jean-Claude.”
Eijffinger, Sylvester C. W., and Carin A. B. van der Central Banking, 16(1): 32–36.
Cruijsen. 2007. “The Economic Impact of Central Gosselin, Pierre, Aileen Lotz, and Charles Wyplosz.
Bank Transparency: A Survey.” Centre for Economic 2007. “Interest Rate Signals and Central Bank Trans-
Policy Research Discussion Paper 6070. parency.” Centre for Economic Policy Research Dis-
Eusepi, Stefano, and Bruce Preston. 2007. “Central cussion Paper 6454.
Bank Communication and Expectations Stabiliza- Gürkaynak, Refet S. 2005. “Using Federal Funds
tion.” National Bureau of Economic Research Work- Futures Contracts for Monetary Policy Analysis.”
ing Paper 13259. Board of Governors of the Federal Reserve System
Fatás, Antonio, Ilian Mihov, and Andrew K. Rose. 2007. Finance and Economics Discussion Series 2005-29.
944 Journal of Economic Literature, Vol. XLVI (December 2008)

Gürkaynak, Refet S., Andrew T. Levin, and Eric T. Targets on the Level of Expected Inflation in Five
Swanson. 2006. “Does Inflation Targeting Anchor Countries.” Review of Economics and Statistics,
Long-Run Inflation Expectations? Evidence from 85(4): 1076–81.
Long-Term Bond Yields in the U.S., U.K., and Swe- Kahneman, Daniel. 2003. “Maps of Bounded Rational-
den.” Federal Reserve Bank of San Francisco Work- ity: Psychology for Behavioral Economics.” Ameri-
ing Paper 2006-09. can Economic Review, 93(5): 1449–75.
Gürkaynak, Refet S., Brian Sack, and Eric T. Swanson. King, Mervyn. 2000. “Monetary Policy: Theory in Prac-
2005. “Do Actions Speak Louder Than Words? The tice.” Address to the joint luncheon of the American
Response of Asset Prices to Monetary Policy Actions Economic Association and the American Finance
and Statements.” International Journal of Central Association, Boston. http:/www.bankofengland.co.uk/
Banking, 1(1): 55–93. publications/speeches/2000/speech67.htm.
Guthrie, Graeme, and Julian Wright. 2000. “Open King, Mervyn. 2002. “The Inflation Target Ten Years
Mouth Operations.” Journal of Monetary Econom- On.” Bank of England Quarterly Bulletin, 42(4):
ics, 46(2): 489–516. 459–74.
Heinemann, Friedrich, and Katrin Ullrich. 2007. King, Mervyn. 2005. “Monetary Policy: Practice
“Does It Pay to Watch Central Bankers’ Lips? The Ahead of Theory.” Bank of England Quarterly Bul-
Information Content of ECB Wording.” Schweizeri- letin, 45(2): 226–36.
sche Zeitschrift fur Volkswirtschaft und Statistik/ Kohn, Donald L., and Brian Sack. 2004. “Central Bank
Swiss Journal of Economics and Statistics, 143(2): Talk: Does It Matter and Why?” In Macroeconom-
155–85. ics, Monetary Policy, and Financial Stability, 175–
Holsti, Ole R. 1969. Content Analysis for Social Sci- 206. Ottawa: Bank of Canada.
ences and Humanities. Reading: Addison-Wesley. Kuttner, Kenneth N. 2004. “A Snapshot of Infla-
Issing, Otmar. 1999. “The Eurosystem: Transparent tion Targeting in Its Adolescence.” In The Future
and Accountable or ‘Willem in Euroland.’” Journal of Inflation Targeting, ed. Christopher Kent and
of Common Market Studies, 37(3): 503–19. Simon Guttmann, 6–42. Sydney: Reserve Bank of
Issing, Otmar. 2005. “Communication, Transparency, Australia.
Accountability: Monetary Policy in the Twenty-First Kuttner, Kenneth N., and Adam S. Posen. 1999. “Does
Century.” Federal Reserve Bank of St. Louis Review, Talk Matter after All? Inflation Targeting and Cen-
87(2): 65–83. tral Bank Behavior.” Federal Reserve Bank of New
Jansen, David-Jan. 2008. “Does the Clarity of Central York Staff Report 88.
Bank Communication Affect Volatility in Financial Lange, Joe, Brian Sack, and William Whitesell. 2003.
Markets?” Unpublished. “Anticipations of Monetary Policy in Financial Mar-
Jansen, David-Jan, and Jakob De Haan. 2005. “Talk- kets.” Journal of Money, Credit, and Banking, 35(6):
ing Heads: The Effects of ECB Statements on the 889–909.
Euro–Dollar Exchange Rate.” Journal of Interna- Lapp, John S., and Douglas K. Pearce. 2000. “Does a
tional Money and Finance, 24(2): 343–61. Bias in FOMC Policy Directives Help Predict Inter-
Jansen, David-Jan, and Jakob De Haan. 2006. “Look meeting Policy Changes?” Journal of Money, Credit,
Who’s Talking: ECB Communication during the and Banking, 32(3): 435–41.
First Years of EMU.” International Journal of Levin, Andrew T., Fabio M. Natalucci, and Jeremy M.
Finance and Economics, 11(3): 219–28. Piger. 2004. “The Macroeconomic Effects of Infla-
Jansen, David-Jan, and Jakob De Haan. 2007a. “The tion Targeting.” Federal Reserve Bank of St. Louis
Importance of Being Vigilant: Has ECB Commu- Review, 86(4): 51–80.
nication Influenced Euro Area Inflation Expecta- Lin, Shu, and Haichun Ye. 2007. “Does Inflation Tar-
tions?” CESifo Working Paper 2134. geting Really Make a Difference? Evaluating the
Jansen, David-Jan, and Jakob De Haan. 2007b. “Were Treatment Effect of Inflation Targeting in Seven
Verbal Efforts to Support the Euro Effective? A Industrial Countries.” Journal of Monetary Eco-
High-Frequency Analysis of ECB Statements.” Euro- nomics, 54(8): 2521–33.
pean Journal of Political Economy, 23(1): 245–59. Meade, Ellen E., and D. Nathan Sheets. 2005.
Jansen, David-Jan, and Jakob De Haan. Forthcoming. “Regional Influences on FOMC Voting Patterns.”
“Has ECB Communication Been Helpful in Predict- Journal of Money, Credit, and Banking, 37(4):
ing Interest Rate Decisions? An Evaluation of the 661–77.
Early Years of the Economic and Monetary Union.” Meyer, Laurence H. 1998. “Come with Me to the
Applied Economics. FOMC.” Remarks made at Willamette University,
Jensen, Henrik. 2002. “Optimal Degrees of Trans- Salem, OR. http://www.federalreserve.gov/boarddocs/
parency in Monetary Policymaking.” Scandinavian speeches/1998/199804022.htm.
Journal of Economics, 104(3): 399–422. Mishkin, Frederic S. 2004. “Can Central Bank Trans-
Johnson, David R. 2002. “The Effect of Inflation parency Go Too Far?” In The Future of Inflation
Targeting on the Behavior of Expected Inflation: Targeting, ed. Christopher Kent and Simon Gutt-
Evidence from an 11 Country Panel.” Journal of mann, 48–65. Sydney: Reserve Bank of Australia.
Monetary Economics, 49(8): 1521–38. Mishkin, Frederic S., and Klaus Schmidt-Hebbel.
Johnson, David R. 2003. “The Effect of Inflation 2007. “Does Inflation Targeting Make a Difference?”
Blinder et al.: Central Bank Communication and Monetary Policy 945

National Bureau of Economic Research Working Rudebusch, Glenn D., and John C. Williams. Forth-
Paper 12876. coming. “Revealing the Secrets of the Temple: The
Morris, Stephen, and Hyun Song Shin. 2002. “Social Value of Publishing Central Bank Interest Rate Pro-
Value of Public Information.” American Economic jections.” In Asset Prices and Monetary Policy, ed.
Review, 92(5): 1521–34. John Y. Campbell. Chicago and London: University
Musard-Gies, Marie. 2006. “Do European Central of Chicago Press.
Bank’s Statements Steer Interest Rates in the Euro Sarno, Lucio, and Mark P. Taylor. 2001. “Official
Zone?” Manchester School, 74: 116–39. Intervention in the Foreign Exchange Market: Is It
Orphanides, Athanasios. 2001. “Monetary Policy Rules Effective and, If So, How Does It Work?” Journal of
Based on Real-Time Data.” American Economic Economic Literature, 39(3): 839–68.
Review, 91(4): 964–85. Sibert, Anne. 2006. “Is Central Bank Transparency
Orphanides, Athanasios, and John C. Williams. 2005. Desirable?” Centre for Economic Policy Research
“Imperfect Knowledge, Inflation Expectations, Discussion Paper 5641.
and Monetary Policy.” In The Inflation-Targeting Siklos, Pierre L. 2002. The Changing Face of Central
Debate, ed. Ben S. Bernanke and Michael Wood- Banking: Evolutionary Trends since World War II.
ford, 201–34. Chicago and London: University of Cambridge; New York and Melbourne: Cambridge
Chicago Press. University Press.
Pakko, Michael R. 2005. “On the Information Content Siklos, Pierre L., and Martin T. Bohl. 2008. “Policy
of Asymmetric FOMC Policy Statements: Evidence Words and Policy Deeds: The ECB and the Euro.”
from a Taylor-Rule Perspective.” Economic Inquiry, International Journal of Finance and Economics,
43(3): 558–69. 13(3): 247–65.
Poole, William. 2001. “Expectations.” Federal Reserve Svensson, Lars E. O. 2003. “What Is Wrong with Taylor
Bank of St. Louis Review, 83(2): 1–10. Rules? Using Judgment in Monetary Policy through
Poole, William, and Robert H. Rasche. 2003. “The Targeting Rules.” Journal of Economic Literature,
Impact of Changes in FOMC Disclosure Practices 41(2): 426–77.
on the Transparency of Monetary Policy: Are Mar- Svensson, Lars E. O. 2006a. “The Instrument-Rate
kets and the FOMC Better ‘Synched’?” Federal Projection under Inflation Targeting: The Norwe-
Reserve Bank of St. Louis Review, 85(1): 1–9. gian Example.” In Stability and Economic Growth:
Reeves, Rachel, and Michael Sawicki. 2007. “Do The Role of Central Banks, 175–98. Mexico City:
Financial Markets React to Bank of England Com- Bank of Mexico.
munication?” European Journal of Political Econ- Svensson, Lars E. O. 2006b. “Social Value of Public
omy, 23(1): 207–27. Information: Comment: Morris and Shin (2002) Is
Reinhart, Vincent R., and Brian Sack. 2006. “Grading Actually Pro-Transparency, Not Con.” American
the Federal Open Market Committee’s Communica- Economic Review, 96(1): 448–52.
tions.” Unpublished. Swanson, Eric T. 2006. “Have Increases in Federal
Romer, Christina D., and David H. Romer. 1989. Reserve Transparency Improved Private Sector
“Does Monetary Policy Matter? A New Test in the Interest Rate Forecasts?” Journal of Money, Credit,
Spirit of Friedman and Schwartz.” In NBER Mac- and Banking, 38(3): 791–819.
roeconomics Annual: 1989, ed. Olivier J. Blanchard Thornton, Daniel L., and David C. Wheelock. 2000. “A
and Stanley Fischer, 121–70. Cambridge, Mass. and History of the Asymmetric Policy Directive.” Fed-
London: MIT Press. eral Reserve Bank of St. Louis Review, 82(5): 1–16.
Romer, Christina D., and David H. Romer. 2000. Vega, Marco, and Diego Winkelried. 2005. “Infla-
“Federal Reserve Information and the Behavior of tion Targeting and Inflation Behavior: A Successful
Interest Rates.” American Economic Review, 90(3): Story?” International Journal of Central Banking,
429–57. 1(3): 153–75.
Rosa, Carlo. 2008. “Talking Less and Moving the Mar- Willard, Luke. 2006. “Does Inflation Targeting Mat-
ket More: Is this the Recipe for Monetary Policy ter? A Reassessment.” Center for Economic Policy
Effectiveness? Evidence from the ECB and the Studies Working Paper 120.
Fed.” Unpublished. http://carlorosa1.googlepages. Woodford, Michael. 2001. “Monetary Policy in the
com/research. Information Economy.” In Economic Policy for the
Rosa, Carlo, and Giovanni Verga. 2007. “On the Con- Information Economy, 297–370. Kansas City: Fed-
sistency and Effectiveness of Central Bank Com- eral Reserve Bank of Kansas City.
munication: Evidence from the ECB.” European Woodford, Michael. 2003. Interest and Prices: Foun-
Journal of Political Economy, 23(1): 146–75. dations of a Theory of Monetary Policy. Princeton
Rozkrut, Marek, Krzysztof Rybiński, Lucyna Sztaba, and Oxford: Princeton University Press.
and Radosław Szwaja. 2007. “Quest for Central Bank Woodford, Michael. 2005. “Central-Bank Communi-
Communication: Does It Pay to Be ‘Talkative’?” cation and Policy Effectiveness.” In The Greenspan
European Journal of Political Economy, 23(1): Era: Lessons for the Future, 399–474. Kansas City:
176–206. Federal Reserve Bank of Kansas City.

You might also like