The ECB's monetary policy strategy and the quantitative definition of price stability

Letter of Dr. W. F. Duisenberg, President of the ECB to the Chairperson of the Committee on Economic and Monetary Affairs

13 December 2001

Mrs Christa Randzio-Plath Chairperson of the Committee on Economic and Monetary Affairs European Parliament Rue Wiertz 1047 Brussels

Subject: "Reply to the letter by Mrs. Randzio-Plath, Chairperson of the Committee on Economic and Monetary Affairs of the European Parliament, dated 15 November 2001"

Dear Mrs. Randzio-Plath,

In your letter of 15 November 2001, you asked me to elaborate further on the ECB's definition of price stability. In particular, you invited me to express my views on whether a point inflation target, with symmetrical margins, would be more effective, providing a more precise tool for evaluating the ECB's performance. I gather from your letter that you see this issue as being closely related to the adoption of an "inflation targeting framework", which is followed by a number of central banks around the world, and you ask me to elaborate on the pros and cons of such approach.

In answering your questions, I would like to separate three issues which are - to my mind - conceptually distinct: the first is the difference between a definition of price stability and an inflation target; the second is whether the final objective of monetary policy should be specified in the form of a point or a range; and the third concerns the framework which is adopted by a central bank in order to pursue its objective.

Starting with the first issue, the difference between a quantitative definition of price stability and an inflation target is that the latter does not necessarily define "price stability", but quantifies the objective with respect to price developments that monetary policy is aiming for. In fact, historically, there have been many instances when inflation targets were set at levels significantly away from "price stability", often in cases where countries starting from high levels of inflation set a path of inflation targets over time in order to bring inflation down to lower levels. Therefore, there is a conceptual difference between the ECB's definition of price stability and an inflation target. Similarities exist only when such inflation targets are aimed explicitly at achieving "price stability".

With regard to the second issue, whether a point target or a range would best serve the purpose of quantifying the primary objective, this is indeed an issue which is debated even among proponents of the inflation targeting framework and with regard to which actual practices also vary among central banks following that approach. The ECB's quantitative definition of price stability has established a range of inflation outcomes deemed compatible with price stability.

Our view is that a range definition honestly reflects the degree of uncertainty attached to the meaning and measurement of the concept of price stability. A range acknowledges this type of uncertainty about the precise objective openly. In particular, it allows uncertainty about and possible variations in the measurement bias in the HICP to be accommodated within the range, while ensuring - in all likelihood - that "true" price stability (zero inflation) is included in the range. Moreover, a range provides a "buffer" in the face of economic shocks and avoids the impression that monetary policy is equipped to (or might attempt to) fine-tune price developments with a high degree of precision.

At the same time, the ECB's definition is precise enough to provide a clear yardstick for accountability, i.e. to evaluate the ECB's performance over time. In this respect, in order to form an appropriate assessment of the success or failure of the ECB's monetary policy, the medium-term orientation of our policy should be borne in mind. This acknowledges, inter alia, the existence of short-term volatility in prices, resulting from non-monetary shocks to the price level that cannot be controlled by monetary policy. Nevertheless, when evaluated over a sufficiently extended period, it is clear that our policy can be considered successful only if price stability, according to our definition, has prevailed for most of the time and if variability of inflation has remained low.

Moreover, as I had already mentioned in my previous letter of 16 October 2001, our definition has served very well the purpose of anchoring medium-term inflation expectations (as evident from inflation expectations derived from market surveys or from the analysis of yields on long-term bonds), also in the face of substantial shocks to the price level over the past three years. It has thus reduced inflation uncertainty, thereby lowering the risk premia included in long-term interest rates.

There is yet another aspect of importance: a range definition helps avoid the impression that monetary policy should feed back mechanically from deviations of HICP inflation or a forecast of future HICP inflation at some fixed horizon from the particular "target point". Here, we come to the third issue, namely the framework according to which central banks should pursue their objective. This is an area where some differences exist between our approach and the approach which is often portrayed as characteristic of an "inflation targeting framework".

In this regard, let me first say that there is no widely agreed definition of a policy of inflation targeting.

One rather broad characterisation of an inflation targeting strategy is the prescription that monetary authorities should pursue a credible policy of low and stable inflation. This prescription is simple and convincing: monetary policy should pursue an overriding goal of price stability. Moreover, because of the lags in monetary transmission, monetary policy must be forward-looking. If this broad definition were adopted, virtually all stability-oriented central banks would be classified as wedded to an inflation targeting policy framework.

In this broad version, it is fair to say that the inflation targeting line of reasoning has made - together with other pre-existing traditions in monetary policy-making - an important contribution to the diffusion among central banks of a culture of stability.

Inflation targeting is generally also viewed as a strategy that places emphasis on the production and discussion of inflation forecasts within central banks. This is positive to the extent that it fosters a forward-looking orientation of policy and sharpens the assessment of economic prospects.

However, in a finer degree of detail and at a more operative level of policy design, inflation targeting is often seen as a framework that makes macroeconomic forecasts the main, or even unique and all-encompassing, tool of the policy-making process and the external communication of policy decisions. In fact, some observers understand inflation targeting to imply a simple policy rule whereby changes in interest rates should feedback from the deviation between a conditional inflation forecast (normally based on interest rates remaining unchanged at their pre-decision level) and the inflation objective at a specific time horizon (typically two years).

For the following reasons the ECB has not adopted this approach.

First, inflation forecasts, while useful ingredients of monetary policy strategies, do not represent sufficient statistics for the state of the economy at any horizon: the same inflation forecast figure can be associated with quite different states of the world, commanding quite different reactions on the part of the central bank. For this reason, the appropriate monetary policy to maintain price stability should always be made conditional on the circumstances and the nature and magnitude of the threat to price stability (for example, on whether the shock is temporary or permanent, on whether it has emerged on the supply or demand side, on whether it is of domestic or external origin). In fact, a policy of mechanical feedback from inflation forecasts may introduce unnecessary volatility in output and realised inflation. In our view, in response to certain shocks, a gradualist and measured monetary policy response aimed at price stability over the medium term is more appropriate than a response to an inflation forecast at a fixed horizon.

Second, the horizon on which policy-makers should focus will also depend on the economic circumstances prevailing. A fixed, e.g. two-year, or any other pre-specified, horizon is somewhat arbitrary, as the transmission process evolves over time with variable and uncertain lags. Limiting attention to a specific projection horizon may induce short-sighted (or delayed) reactions, the effects of which may have to be undone at a later date, with associated costs in terms of instability. Monetary policy needs to focus on the period covering the whole transmission process, bearing in mind that this may span an uncertain and protracted period of time.

Third, as we have already stressed on a number of occasions, inflation forecasts themselves may be subject to a number of problems. Econometric models underlying the projections, like any model of the economy, are subject to uncertainty and cannot provide a complete description of the economy. Forecasts are produced on the basis of expert assessments and a variety of econometric models using a broad range of data. How the various data, opinions and model analyses are combined can sometimes be not totally clear to users of the forecast. Moreover, the production of forecasts is time-consuming. Therefore, forecasts can hardly incorporate all the information in a timely manner and can quickly become out of date. Inflation forecasts (especially at horizons as long as two years) are surrounded by considerable uncertainty. The simple characterisation of inflation targeting does not offer guidance to policy-makers as to how to treat this uncertainty.

Last but not least, models used to produce forecasts do not usually give an important role to monetary developments. In the euro area context, however, all evidence tells us that money should be an essential piece of information in the context of a forward-looking monetary policy.

I should also note that proponents and practitioners of inflation targeting have increasingly recognised some of the problems associated with such simple, text-book versions of inflation targeting prescriptions. This is done, for example, by recognising that no single or "true" model of the economy exists and by emphasising the need to adopt a robust approach to policy-making which takes into account uncertainty about the interpretation of information. Furthermore, in practice, most inflation targeting central banks have relaxed the definition of the horizon for inflation targeting and have acknowledged that the feed-back from inflation forecasts cannot be mechanistic.

The ECB's strategy has honestly addressed the problems related to too simple a version of the inflation targeting approach and its two-pillar framework for monetary policy is a balanced response to many of them. In fact, a thorough assessment and cross-checking of the information coming from various indicators and different analytical frameworks are essential features of the ECB's strategy. ^[1] This approach helps the Governing Council to take robust decisions in assessing the importance of the various indicators and identifying the nature of the treat to price stability.

Within our strategy, we use macroeconomic projections as one important analytical tool to organise a large amount of information and help to create a consistent picture of possible future developments, without making them the sole, or main, conduit for analysis. In particular, we deem it important, within our framework, to clearly separate the production of forecasts, as carried out under the responsibility of the staff, from the monetary policy decisions taken under the responsibility of the Governing Council.

In addition, in recognition of the fundamentally monetary origins of inflation and reflecting the close relationship between the stock of money and the price level in the euro area over the medium term, our strategy assigns a prominent role to money in the formulation of monetary policy. The analysis of monetary developments represents a crucial element for our evaluation of the risks to price stability in the euro area.

Finally, in line with a medium-term orientation in the conduct of monetary policy, we have explicitly avoided a focus on a fixed point horizon. Rather, when deciding on the appropriate response to potential threats to price stability, we evaluate the effects of both the economic disturbances and our policy action over the whole transmission process.

We believe that the full-information and diversified approach we have adopted is conducive to policy decisions that reduce the risk of policy-related mistakes in an uncertain environment.

The aforementioned elements are reflected in our approach to the presentation of monetary policy decisions to the public. Transparency requires that our communication closely reflects our internal decision-making process. Adopting "too simple" a form of presentation would not honestly convey the complexity of the analysis we have to conduct.

Let me conclude by saying that differences in the practices of central banks oriented to price stability are in fact relatively limited. We share a number of key elements that guide the conduct of our monetary policy with many central banks labelled followers of the inflation targeting approach, namely a clear, quantitative definition of the overriding objective, a forward-looking orientation of our policy, the awareness of the need to take a broad range of information into account and to communicate with the public in a clear and transparent manner.

As the experience of major central banks in the world has shown, there is no unique way for a successful conduct of monetary policy. Different traditions, practices or frameworks have different merits and also depend on historical circumstances and institutional environments. We have chosen an approach that, we believe, is best suited to serve the interests of the citizens in the euro area.

With best regards,

[signed]

W.F. Duisenberg