Expectations and credibility in modern central banking: A practitioner’s view”
Speech by José Manuel González-Páramo, Member of the Executive Board of the ECB
Conference on “Inflation Targeting, Central Bank Independence and Transparency”
Cambridge, 15 June 2007
Ladies and gentlemen, dear colleagues,
It is a pleasure for me to be here at the University of Cambridge today to participate in this conference on the occasion of the tenth anniversary of the granting of operational independence to the Bank of England. I would like to thank the organisers – Professors Honkapohja and Holly – for inviting me to join in this event and I congratulate them for having put together a set of outstanding papers on issues of great relevance for the practice of monetary policy.
I must confess that my first reaction when I read the invitation to the conference was of surprise at seeing that it is only ten years since the Bank of England was granted independence. We have become so accustomed to the social benefits arising from the institutional framework predominant today among the main central banks – the most important being lower inflation rates – that it is almost an effort to remember the time when monetary policy was left to the vagaries of political pressures.
Independence, credibility and the role of expectations
Indeed, the principle of central bank independence in the pursuit of the goals of monetary policy has been codified in the legal systems of many countries. Perhaps even more importantly, there is evidence that the importance of this principle seems to be increasingly well understood by society at large. According to all surveys among euro area citizens, an overwhelming majority of respondents support the pursuit of price stability as a goal of the European Central Bank (ECB) and back the ECB’s independence in order to guarantee the achievement of this goal.
I am absolutely certain that the ECB will also be able to celebrate its tenth anniversary as an independent central bank in one year’s time. Any noises to the contrary that you may have heard or read recently only testify to the liveliness and excitement of political campaigns in Europe and, like so many things which are said at the time of elections, should not be taken too seriously.
But I am not here to talk about politics. Indeed, one of the great advantages of central bank independence is precisely that we can ignore the political furor and concentrate on the welfare-enhancing objective of keeping inflation under control. At the same time, I could not agree more with the view that central bankers must strive to fully comply with the standards of transparency and accountability that democratic societies rightly demand from independent public agencies.
Today I would like to talk about two issues of critical importance in modern central banking: the role of expectations and the importance of credibility. Let me warn you that I will talk on these issues from a practitioner’s point of view. In particular, I would like to share some considerations and also some concerns about the use of expectations in the day-to-day business of a central bank.
Expectations play a central role in modern economics. The notion that the future – or even the present for that matter – is known only with uncertainty, while most economic decisions taken by agents depend on their expectations regarding future events or the behaviour of other agents, is an important ingredient of the main building blocks of modern economic theories and also has crucial implications for the design and conduct of policies. For instance, expectations are at the core of the natural rate hypothesis of Milton Friedman and Edmund Phelps. This hypothesis – which is no longer controversial and is now supported by a large body of empirical evidence – challenged the belief underlying the so-called Keynesian policies that policy-makers could permanently and systematically exploit a trade-off between inflation and unemployment.
It is difficult to think of many other theoretical contributions that have had a greater influence on policy-making practice than the natural rate hypothesis. By postulating in 1968 that attempts to systematically stimulate the economy through monetary injections would simply end in tears and higher inflation in the long run, Friedman and Phelps provided a clear answer to a key organisational question for any society: what should central banks do? The answer – now a tenet of central banking theory and practice – is that central banks are at their best when they focus on delivering price stability at policy-relevant horizons and refrain from attempts to “fine-tune” the economy.
Why are central banks so concerned with expectations?
The need to reckon with expectations has been taken up by the rational expectations hypothesis, which is a key ingredient of most macroeconomic models nowadays. Indeed, expectations and the modelling of the process by which they are formed pervades modern theoretical and empirical models. But, perhaps more surprisingly, measures of the expectations of various groups of agents – households, firms, financial markets, professional forecasters, public institutions, international organisations, etc. – regarding future developments in a number of variables seem to have become pervasive in central bank communication also.
Let me illustrate this point by referring to the ECB’s publications. A quick look at the ECB Monthly Bulletin shows that it has regularly reported on expectations drawn from surveys among firms or analysts, financial market data or projection reports from professional forecasters and international organisations on a very broad range of variables, including manufacturing and services activity, employment, investment, short-term interest rates, inflation and corporate earnings as well as public budget balance and debt ratios. And, of course, the ECB Monthly Bulletin has also reported on the outcome of the Eurosystem staff macroeconomic projections, which are in turn reliant on exogenous assumptions about future interest rates and commodity prices derived from financial market data.
Measures of expectations recurr as frequently or more often in the communication of other central banks, and there is certainly no need for me to dwell on the importance of inflation projections for central banks pursuing inflation targeting strategies.
Given the amount of time and resources that central banks spend on collecting and analysing expectations data obtained from other sources as well as on compiling their own, it may be worth asking ourselves why central banks are so concerned with expectations?
A first reason is that expectations play an important role in the transmission of monetary policy. Consider, for instance, the term structure of interest rates. Central banks have a direct influence only on short-term interest rates through their monetary policy instruments – typically an overnight call rate. However, consumption and investment decisions, and thus medium-term price developments, are to a large extent influenced by longer-term interest rates, which in turn depend on private sector expectations regarding future central bank decisions and future inflation.
More generally, if we think about the main channels of monetary policy transmission, it is clear that expectations are central. For instance, using a present-value model of asset pricing, it is easy to see how changes in policy rates may be transmitted to the economy through movements in asset prices (e.g. exchange rates, bond and equity prices and housing prices) via changes in both discount factors and expected returns. Expectations play an even more obvious role in the confidence channel, since this hinges entirely on how changes in policy rates affect expectations regarding future economic developments and how confidently those expectations are held.
Thus, the final impact of a policy move depends to a very large extent on its impact on expectations, which increases the significance of credibility issues for monetary policy. In this respect, as Mervin King (2005) has put it, “the real influence of monetary policy is less the effect of any individual monthly decision on interest rates and more the ability of the framework of policy to condition inflation expectations”. This is a statement that we can all agree with.
A further motive why expectations matter to central banks is that economic theory and the experience of the last few decades suggest that a monetary authority that is concerned with price stability should systematically monitor private sector expectations regarding future economic developments, particularly inflation, and should react – not necessarily through changes in policy rates - when those expectations signal the emergence of risks to price stability.
Expectations, credibility and communication
Another key reason is that measures of expectations are an important source of information about the extent to which a central bank’s commitment to price stability is perceived as credible by the general public. Indeed, anchoring inflation expectations requires that the central bank be regarded as credible. Economic agents should be confident that the central bank will react to the various shocks that affect the economy so as to maintain price stability, i.e. that we have some sort of “reaction function”.
The literature on policy rules suggests that, in reacting, central banks should adhere to the Taylor principle, which requires that the nominal interest rate respond to the inflation rate by more than one for one. If it does so, private sector inflation expectations will remain anchored to levels consistent with the maintenance of price stability, which itself greatly facilitates the job of keeping inflation low. We were warned this morning, though, that, under some uncertainty conditions, announcing a quantitative definition of price stability and adhering to the Taylor principle may not be sufficient to anchor inflation expectations and, in such cases, the central bank may need to work harder on communication and clarify how it intends to achieve price stability. 
To put this discussion into perspective, let us glimpse at the history of the euro area bloc in recent decades. In the 1970s, when most European countries experienced high and volatile inflation, annual consumer price inflation averaged about 9%. In the 1980s, when central bankers started to react and the disinflation process was set in motion, the inflation rate declined but was still quite high, at around 6% per annum. Conversely, since the introduction of the euro in January 1999 (a period in which oil prices have risen from around USD 10 per barrel to a peak of almost USD 80 per barrel) consumer price inflation has averaged about 2% per annum, which is very close to the ECB’s quantitative definition of price stability. Not surprisingly, recent research by ECB staff has shown that the central bank’s definition of price stability does a relatively good job of forecasting average inflation in the euro area!  What is perhaps more interesting is that, over the same period, long-term inflation expectations recorded in the Consensus Economics survey among professional economists have never exceeded 2%.
These data seem to validate many of the points made above regarding credibility. Credible commitment to an explicit inflation objective helps to anchor inflation expectations to the desired level of inflation, and this anchoring itself contributes to delivering price stability. It is clear then that, as long as they remain well-anchored at the desired level, expectations can greatly facilitate the tasks of a central banker. Small deviations in private sector expectations from the central bank definition of price stability may not give rise to concern (for instance, they may reflect imperfect or incomplete information regarding the current state of the economy on the part of private agents). However, policy-makers should react firmly when these deviations become significant, as this may signal incipient inflation or deflation “scares”.
Over-reliance on expectations: Some drawbacks
While acknowledging that it us useful for central banks to take private sector expectations into account, there are a number of reasons why we should regard the use of inflation expectations in the same way as we regard the taking of medicine. It can be very good for our health and help us to stay well, but we need to use it with care. Let me now turn to some of the motives why I think we should use private sector expectations with care, especially those about future inflation and policy rates.
A major reason is the risk of policy complacency. The data for the last few years suggest that, thanks to our credible commitment to delivering price stability, we have achieved quite a favourable equilibrium in which well-anchored long-term inflation expectations and low and stable inflation mutually reinforce each other. Whether or not this equilibrium is sustainable very much depends on agents continuing to trust in our commitment. If long-term inflation expectations – as measured from (premia-adjusted) break-even inflation rates or survey data – remain pegged to “below but close to 2%”, which is the definition of price stability of the ECB, we can continue to assume that this trust will endure.
However, a recent study by Cecchetti et al. (2007) for the United States has warned us against relying excesively on measures of inflation expectations as indicators of policy effectiveness. The authors argue that long-run inflation expectations had predictive power for inflation trends at the time of the Great Inflation, but this is no longer the case in the current environment of monetary stability. If anything, they conclude that the causal relationship between inflation expectations and actual trends in inflation may be now inverted, with expectations following rather than leading changes in inflation trends.
Given the potential policy implications of these findings, it is important to dedicate more effort to investigating this hypothesis in greater detail. Indeed, if they are confirmed by further analysis, we should be aware that, even when long-term inflation expectations seem to be stable at low levels, we cannot take it for granted that agents trust in our commitment, and we should bear in mind that in the event of shocks large enough to affect inflation trends, the expectations of private agents may become unanchored.
In recent years, many studies have examined the conditions under which inflationary expectations may unanchor, i.e. drift away from the objective of the central bank and become an independent source of instability for the economy.
Some studies – which are based on the rational expectations hypothesis – see such an outcome as the result of insufficient trust on the part of the public in the central bank’s commitment to price stability. These studies typically lay the blame for such a lack of trust on the past record of policy responses to inflation. More precisely, if the central bank fails to systematically act in accordance with its objective of low-inflation with sufficient force (i.e. a systematic neglect of the Taylor principle in the context of interest rate rules), this is likely to undermine private agents’ confidence in how committed policy-makers are to price stability. The Great Inflation of the late 1960s and 1970s is often quoted as an example of a major policy error which occurred because central bankers failed to react strongly enough to inflation developments. 
A second strand of studies – to which the organiser of this conference, Professor Honkapohja, and his associates have contributed a great deal – leaves rational expectations aside and instead emphasises that agents are constantly learning about the economic environment. According to this literature, instability in inflation expectations may not necessarily stem from a systematic lack of strength in the policy reaction to observed inflation, but may simply reflect the fact that past shocks set in motion a learning process. This process could eventually lead to the alignment of private sector inflation expectations with the central bank’s objective, but there could be instances – perhaps partly reflecting an inappropriate monetary policy – when this learning process does not lead to a convergence. 
Recent studies have also argued that some agents may only have access to limited information in forming their expectations. This hypothesis may explain not only discrepancies in forecasts between the private sector and the central banks, but also among private agents.  In particular, this strand of research may potentially identify the reason why not all private sector expectations are equally reliable. For example, in the United States expectations formed by professional forecasters tend to systematically outperform those obtained from household surveys.
The ECB does not compile data on consumer expectations regarding future inflation in the euro area and, therefore, cannot judge the ability of the average euro area consumer to forecast inflation. However, we do have survey data on qualitative inflation perceptions and these typically show that consumers do not always find it easy to correctly estimate the strength of current inflationary pressures. It would be unfair to conclude from this evidence that consumers also would prove to be poor forecasters. I have raised the issue only because I think that limited information – together with some psychological explanations of the type favoured by behavioural economists – may also help us to understand why perceived inflation occasionally deviates from observed inflation.
To be honest, I have also brought up this matter because I would like to direct your attention to the need to dedicate more research efforts to an issue that may potentially have implications for the credibility of central banks. Indeed, not much thought has been given by academic researchers to the question of why public perceptions of inflation sometimes deviate from the officially measured rate and what central banks should do in these cases. You may think that this is only a problem for the euro area, confined to the period immediately following the introduction of the new banknotes and coins, but there is some evidence that this problem is also common to other developed economies which have not undergone a currency changeover.
Private expectations regarding policy rates: Predictability and credibility
Going back to the subject of expectations, I have thus far focused on inflation expectations. Before concluding, however, I should like to mention another type of expectation which is of interest to central banks, namely expectations regarding future monetary policy rates. Such expectations provide benchmarks against which to assess various aspects of crucial importance for a central bank’s success, such as its transparency, its predictability, and the effectiveness of its communication strategy, among others. As a result, expectations regarding future policy rates, whether taken from surveys or financial market data, are an essential source of information for assessing the degree of understanding of a central bank’s monetary policy strategy and conduct by market participants and external observers.
There is a more practical reason why central bank economists carefully examine market expectations about future policy rates, i.e. that these expectations are often used to produce conditional macroeconomic projections. Whereas in the past central bank forecasters tended to assume constant interest rates in their work, the prevailing convention at present – also followed in the Eurosystem staff projections since mid-2006 - is to extract the assumed path of future interest rates from financial contracts. And, in a move that goes beyond standard practice, three central banks following inflation targeting strategies have more recently decided to “reveal the secrets of the temple” – if I can borrow Rudebusch and Williams’ terminology - and have explicitly announced their own conditional policy rate projections. 
The latter approach is believed to yield advantages in terms of increased transparency and the removal of inconsistencies between the macroeconomic projections and the underlying “exogenous” assumptions for monetary policy. However, announcing one’s own conditional policy rate projections may be problematic since policy-makers do not always find it easy to agree on the future path of policy rates over the forecast horizon. Besides, there is a risk that market participants may not fully appreciate the “conditional” nature of the announcement and may mistakenly interpret it as a firm commitment to future actions. In addition, they may stop doing their homework and rely excessively on the announcements of the central bank instead of making up their own minds about the future. We do not know whether these concerns are justified as the experience with releasing central banks’ own conditional policy rate projections is rather limited both temporally and in terms of country coverage.
The use of market participants’ expected policy rates as an information variable is not entirely uncontroversial, as it may give rise to what Blinder (2006) labels the “dog chasing its tail” problem. This issue has been examined in more detail by Bernanke and Woodford (1997) who demonstrate that a central bank “following” market expectations of its own future behaviour may incur serious dynamic instability or equilibrium indeterminacy.
More generally, in recent years, Alan Blinder has voiced a number of concerns about the excessive reliance by central banks on market interest rate expectations and the risk of having gained independence from politics, only to surrender it to financial markets.
These concerns arise from the fact that modern central banks tend to see predictability as facilitating and enhancing the transmission of monetary policy impulses. In order to increase predictability, central banks use a variety of communication channels (speeches, statements, press conferences, publications, etc.) and signalling devices. In some cases, central banks may provide market participants with more or less explicit indications of the expected direction, timing and pace of future monetary policy decisions at varying horizons.
The risk is, however, that — once specific market expectations regarding future monetary policy are in place — the policy-maker may be forced by unexpected events to deviate from such expectations in order to achieve the goal of delivering price stability. Policy-makers striving to achieve predictability may then face a trade-off between (i) “surprising” the market and (ii) failing to reach their own objective. It should be fairly obvious that – faced with risks to price stability – a central bank should not refrain from surprising the market in order to achieve its goal and, ultimately, safeguard its credibility.
However, the problem is also that the public debate on monetary policy is often based on rather “loose” and confusing definitions. Thus, while discussing policy “surprises”, market participants and the financial press tend to refer to short-term predictability, transparency and credibility as if they were more or less the same concept. The fear that market participants may confuse lack of short-term predictability with lack of transparency or, more worryingly, lack of credibility, may sometimes make central banks reluctant to disappoint market expectations.
Short-termism and reliability of private sector expectations
There is another risk of relying excessively on market interest rate expectations, namely that of their volatility and their short-term horizon. As I mentioned earlier, central banks’ independence has freed monetary policy from the vagaries of political ado. There is a famous anecdote told by Milton Friedman that you may already be familiar with. Faced with the forthcoming re-election, Richard Nixon asked Friedman to convince the President of the Federal Reserve System, Arthur Burns, to increase the money supply in order to stimulate the economy. When Friedman protested that this would only lead to higher inflation in the future, Nixon replied: “we can worry about inflation after we get re-elected”. 
This anecdote says a lot about the short-termism of political life, but as Alan Blinder (2006) warns us, “the next election is usually much further away than the close of the trading day” and financial market participants tend to have a much shorter horizon than probably any other category of agents in our economies, including politicians up for re-election.
An additional reason for caution is that market expectations do not always prove reliable. In general, the assumption that financial market data can provide unbiased predictions is mainly based on the market efficiency hypothesis. However, the empirical evidence is not always supportive of such an assumption. A particularly relevant example for the subject at hand is that of the expectation hypothesis of the term structure (stating that the long-run interest rate is determined by the market’s expectation for the short-term interest rate plus a constant risk premium) for which the empirical evidence is rather mixed.
In the same vein, more recently, we have had “bad” experiences with the futures market data on oil and other commodity prices, as these data have severely underpredicted the rise in actual prices observed in recent years. Given the importance of commodity prices in equations describing the inflation dynamics, the systematic underprediction of oil prices may eventually turn into significant forecast errors for inflation.
We could go further and consider more “pathological” cases in which expectations formation mechanisms may pose serious challenges for a central bank. In particular, we could consider the role of expectations in the formation of asset prices, and the possibility that self-fulfilling dynamics might generate outright bubbles, or at least protracted periods of atypical price configurations in certain segments of the financial market.
To sum up and also to clarify, the point I want to make today is not that we should not “trust” the information embedded in market expectations, but simply that we should handle it with care. Also, while it is important to strive for predictability, since this enhances the effectiveness of monetary policy, central bankers should stress the “conditional” nature of the guidance provided to the markets and should also limit the horizon of reference to the very short term, without pre-committing to a specific path of policy rates that they may be obliged to renege on in the future.
Finally, central bankers should make sure that they lead rather than follow the markets. This can be achieved by putting a premium on communication and transparency, since these help to guide the expectations of market participants and bring them into line with the central bank’s objectives and actions. Effective communication about the central bank’s objectives, strategy, assessment of the economic situation and the outlook for price stability, as well as of the associated risks, together with transparency regarding policy decisions and their rationale, greatly reduce the likelihood that policy-makers will ever be forced to surprise economic agents.
Let me conclude my speech with a quote from the Governor of the Bank of England. When referring to the challenges for monetary policy over the next decade he states: “The anchoring of inflation expectations has been central to the stability enjoyed by the UK economy over the past decade. The key lesson from economic theory is not to take those expectations for granted — they depend on the actions of the MPC…”. 
These sentences refer to the United Kingdom and the Bank of England’s Monetary Policy Committee, but I think they would be equally valid for the euro area and the Governing Council of the European Central Bank.
Thank you very much for listening.
Bernanke, B. and M. Woodford (1997), “Inflation forecasts and monetary policy”, Journal of Money, Credit and Banking 29(4), pp. 653-684.
Blinder, A.S. (2006), “Monetary policy today: Sixteen questions and about twelve answers” in Fernández de Lis, S. and F. Restoy (Eds.), Central Banks in the 21st Century, Banco de España, pp. 31-72.
Cecchetti, S.G., Hooper, P., Kasman, B.C., Schoenholtz, K.L. and M.W. Watson (2007), “Understanding the evolving inflation process”, US Monetary Policy Forum 2007.
Clarida, R., Galí, J. and M. Gertler (2000), “Monetary policy rules and macroeconomic stability: Evidence and some theory”, Quarterly Journal of Economics 115(1), pp. 147-180.
Diron, M. and B. Mojon (2005), “Forecasting the central bank’s inflation objective is a good rule of thumb”, ECB Working Paper No. 564.
Eusepi, S. and B. Preston (2007), “Central bank communication and expectations stabilization”, mimeo.
Evans, G.W. and S. Honkapohja (2003), “Adaptive Learning and Monetary Policy”. CEPR Working Paper No. 3962.
King, M. (2005), “Monetary policy: Practice ahead of theory”, Mais Lecture.
King, M. (2007), “The MPC ten years on”, Lecture to the Society of Business Economists.
Mankiw, M., Reis, R. and J. Wolfers (2004), “Disagreement about inflation expectations”, NBER Macroeconomics Annual 2003, Cambridge, MIT Press, 18, pp. 209-248.
Rudebusch, G.D. and J.C. Williams (2007) “Revealing the Secrets of the Temple: The Value of Publishing Central Bank Interest Rate Projections”, Federal Reserve Bank of San Francisco Working Paper 2006-31.
Taylor, J.B. (2001) “An interview with Milton Friedman”, Macroeconomic Dynamics 5, pp. 101-131.
 This was argued in the paper by Eusepi and Preston (2007).
 Diron and Mojon (2005) compare the predictive performance of benchmark inflation forecasts (based on both simple autoregressive and random walk models and published projections) with that obtained by fixing the forecasts to the central bank’s “inflation target”.
 See for instance Clarida, Galí and Gertler (2000).
 Evans and Honkapohja (2003), for example, indicate that some monetary policy rules that would lead to a unique equilibrium under rational expectations conditions would not necessarily be stable under learning conditions.
 Mankiw, Reis and Wolfers (2004) analyse survey data on US inflation expectations from three different sources (The Michigan Survey of Consumer Attitudes, The Livingston Survey and the Survey of Professional Forecasters) and document significant “disagreement” about the expected future inflation path both across and within the different categories of surveyed agents.
 See Rudebusch and Williams (2007).
 The anecdote is recalled in the interview by Taylor (2001).
 King (2007).