Norges Bank

How do central banks write?

An Evaluation of Inflation Reports by Inflation-Targeting Central Banks by Andrea Fracasso, Hans Genberg and Charles Wyplosz

Comments by Philip Lowe
Reserve Bank of Australia

It is with some trepidation that I agreed to participate in this session. I feel a bit like an errant schoolboy who has been called to the principal's office to explain how, after having skipped many of the required assignments through the year, he did so well on the end of year exams where it really mattered!

As the authors of this paper point out, Australia scores very well in terms of the predictability of monetary policy decisions. I might also add that it scores very well in terms of economic growth, and output and inflation variability. Over the past decade growth has averaged almost 4 per cent per year and swings in the business cycle have been moderate both by our history and by international standards. In addition, the average inflation rate over the past decade has also been right on the Bank's medium-term inflation target. I think that it is fair to say that this is a record that many would be happy to substitute for their own.

However when it comes to Australia's inflation report it is a different story. Few in this room would be happy to swap the marks given to them by the review team for the marks given to the Reserve Bank of Australia (RBA). According to the review team, the RBA's report contains the poorest presentation of strategy, the poorest discussion of the objectives of policy, the poorest presentation of information, the poorest Executive Summary and the poorest coverage of relevant information out of the 19 reports examined. And on a number of other important criteria, its scores are the second lowest. Moreover, the RBA's Statement of Monetary Policy does not contain tables of assumptions; it does not contain point forecasts of inflation at different horizons; and it does not contain fan charts. Yet despite all this, somehow things have worked out OK!

Given this assessment, I felt torn between two roads in preparing my comments. The first was to defend the honour of the RBA and present reasons why the scores given for various criteria were somehow mistaken. The second was to focus on the big issue raised by the paper, namely, what constitutes a good Inflation Report. I will do my best to stay on the latter (and hopefully higher) road, although I hope you will excuse me if occasionally I make a few detours.

I would like to touch on three interrelated issues.

The first is the role that an inflation report plays in a central bank's overall communication strategy. The second is the paper's discussion of the ‘must haves' in a report. And the third is the appropriate tone and target audience of the report.

The Role of the Inflation Report

As the paper correctly points out the inflation report is just one of the many vehicles that a central bank has at its disposal to get its message across. Other vehicles include speeches by senior officials, appearances before Parliament, media briefings, material on the web site, explanations of policy decisions, and research and analysis published by the Bank's staff. While almost all central banks use each of these vehicles, the relative importance of the different vehicles seems to differ substantially across countries.

My impression is that in a number of countries, inflation reports are by far the most important vehicle. These reports have been seen as a way of helping build credibility, and in some cases of reinforcing the operational independence of the central bank. Over the past decades, a number of countries that have undergone a substantial change in the monetary regime, including an increase in the independence of the central bank, have moved quickly to adopt quite detailed, and often quite technical inflation reports. Many of these countries have adopted what I will call the ‘Bank of England' model. This includes naming the report the ‘Inflation Report' (as opposed to a ‘Monetary Policy Report') and focusing the discussion in the report squarely on inflation, with detailed forecasts, as well as fan charts to describe the degree of uncertainty regarding the forecasts.

In a way, it is perhaps not surprising that this has been the case. The Bank of England model presents a solid platform for the central bank to demonstrate its practical and technical expertise. It establishes the central bank as an authority on the economy and reinforces the idea that it has a competent and well-trained staff. Moreover, by putting everything on show, the central bank is less likely to be accused of ‘cooking the forecasts'. Laying everything out for the world to see can also make it more difficult for the political process to influence, either directly or indirectly, interest rate decisions. In particular, by publishing a credible inflation forecast above target, the central bank can lessen some of the political opposition to an increase in interest rates. Where central banks are struggling to establish their newly won independence, this type of inflation report can be very useful.

But other models can work as well. While inflation reports can help build credibility, it is fair to say that ultimately credibility is built on outcomes, not processes. By outcomes I mean: (i) a record of low inflation; (ii) a record of acting in line with people's understanding of the policy framework; and (iii) and a record of not unnecessarily surprising the market.

Where past outcomes have been poor, or been made irrelevant by a sudden change in regime, it is not surprising that establishing credibility through the process of an inflation report has been important. There is simply little alternative. In this regard it is interesting to note that the 12 countries that call their report an Inflation Report have had an average inflation rate of 5 per cent over the past 5 years. For the other 7 inflation-targeting countries, the average inflation rate has been substantially lower at 3.2 per cent. This is not to suggest that calling one's report an Inflation Report causes higher inflation! Rather it raises the question of whether at least some of those who have adopted the orthodox model have done so because of a need to build credibility.

For central banks that already have a significant degree of credibility through performance and an ongoing dialogue with the community, the inflation report can take on a slightly different character. In particular, it can be less numeric and more qualitative. In effect, all the entrails do not need to be on display to gain credibility. What is important is the believability of the central bank's qualitative assessment of current and future conditions and the perception that the central bank will act appropriately. A good inflation report is helpful here, but it is no substitute for performance. Countries with good performance probably have more latitude in what constitutes a good inflation report, and I will argue in a moment that this latitude can be helpful.

In terms of this paper, the relevant point is that inflation reports cannot be judged in isolation from other communication by the central bank. While the authors acknowledge this point, they seem to quickly forget it, judging inflation reports as if they were the only form of communication by the central bank. In a sense I think the paper would have been of more value if it had tabulated what inflation reports do and do not contain, rather than making qualitative judgements on what is good and what is not. At a more technical level, I would be interested to know whether the econometric results in Table 4.2 that show that countries with ‘higher-quality' inflation reports have lower interest rate surprises stand up if Peru is left out of the sample.

The 'Must Haves'

I would now like to turn to what the authors' say are the three ‘must haves' in an inflation report.

The first must have is a ‘discussion of the objectives of policy, the decision-making process and how conflicting objectives are treated'.

Few would disagree with the idea that the central bank should talk openly about these issues. But do they really need to be in every issue of an inflation report? I suspect the right answer is no.

Including such a discussion might be justified on the grounds of tidy housekeeping, helping provide a one-stop shop on the monetary framework. As such, it might be helpful to a newly arrived citizen of Mars who knows nothing else about the country or its monetary policy. But, in practice, such a discussion is, in my opinion, not a ‘must have'. In Australia, few followers of monetary policy would have any doubt about the objectives of the Reserve Bank of Australia, about who makes the decisions, about when those decisions are made, and about how they are made. Despite the RBA providing considerable detail on these issues through a range of channels, Australia receives bottom of the class marks in these areas by the review team.

To repeat my earlier point, there are numerous other vehicles for this type of information to be communicated to the public. A central bank might find it useful to include such information in its inflation report, but it is not a must have.

The second ‘must have' is ‘an account of the analytical framework and information on which policies are based'. In this regard the paper rightly points out that the report needs to present a comprehensive set of underlying data. More controversially though, it also suggests that predictions of models used by the central bank be published, as well as any adjustments to those predictions made by the staff, the Governor or the Monetary Policy Committee.

The third ‘must have' is ‘a presentation of inflation forecasts and an evaluation of past forecasts and policy performance'. In elaborating what is required here the authors suggest that the central bank should provide its forecast of inflation at the relevant medium-term policy horizon. Furthermore, it should provide some measure of uncertainty, for example, through a fan chart.

Clearly, inflation reports need some discussion of the outlook for inflation and the general risks to that outlook. But I am not so sure that focussing the discussion of these issues around technical model forecasts and on forecasts at a specified horizon is particularly useful. In some respects, doing so might even make running good monetary policy even more difficult than it already is.

This focus on inflation forecasts at a set horizon comes partly from the view that what inflation targeting is about is setting the policy interest rate so that the inflation forecast at the policy horizon (say two years) is at the target rate. Given this view, what the central bank needs to do to ensure that it is credible is to lay out a convincing case that its inflation forecast for the policy horizon is sound. The argument goes that to do this it needs to display all the ingredients that go into its forecast, for if it does not, how will people know that the central bank is not artificially holding down its forecast so that it can run overly expansionary monetary policy to suit its political masters?

To my taste this seems too narrow a view of inflation targeting, and concerns about central banks cheating seem to be far away from current monetary policy concerns.

Inflation targeting should be about ensuring low and stable inflation over time. On most occasions focussing on a two-year horizon is probably a reasonable thing to do. But there are times when it may well not be. For example, during a period of rapid credit expansion and emerging imbalances in asset markets there may be a case for running monetary policy a bit tighter than suggested by a two-year inflation forecast. The rationale would be that by containing the development of financial imbalances, even at the cost of a temporary undershoot of the inflation target, the central bank might be able to avoid a potentially much larger and more damaging undershoot at some future point when the financial imbalances unwind. In effect such a policy, if it worked, would reduce the expected variance of inflation, and probably output as well, although both ex ante and ex poste the inflation rate would be below target.

Whether or not such a policy could work is subject to considerable debate. My personal view is that it is least worthy of serious consideration. While there are a number of serious obstacles, the main one is probably the political economy of such a response. In particular, it may prove very difficult to convince the public that it is a sensible thing to do.

The task, however, is made more difficult still by the central bank focussing its inflation report almost exclusively on a two-year inflation forecast and the technical apparatus underlying that forecast. Such an approach can entrench in the mind of the public the idea that inflation targeting is about ensuring that the two-year forecast is at the target. Having entrenched that view, it can become more difficult for the central bank to deviate from the rule, as any deviation could be seen as a lack of commitment to the regime. Ultimately, this approach could constrain the central bank's ability to run good monetary policy.

To repeat, inflation targeting should be about attaining low and stable inflation through time. It should not be seen as a mechanical rule-based approach to interest- rate decisions based on an explicit inflation forecast, but rather more as an organising principle for public communication on monetary policy. Under the later view, what is required in the inflation report is a credible discussion of current conditions and future prospects for the variables of interest to the central bank, and a credible discussion of why current settings of monetary policy are appropriate. Detailed numerical forecasts might be helpful in some cases, but again I think they fall short of being a must have.

Tone and Audience

The third issue is what is the appropriate tone for and audience of the report.

In our case the report is written to be accessible to a wide variety of economic decision makers, and those who influence those decision makers, including financial journalists. At least, in this regard we appear to have had some success, with the review team for this paper giving the RBA's report relatively high marks for accessibility by non-technical audiences.

Our report is not written to appeal to an academic audience. There are other channels to do this. So I can perfectly understand why the review panel of graduate students used in this study rated the RBA's report so poorly. But are they really the right group to be evaluating inflation reports? I suspect a better approach, but a significantly more difficult one, would have been to survey financial market participants and business and community leaders in each country about the quality of the central bank's communication. But perhaps the data on interest rates surprises already tell something about what the answers to such a survey would likely show!

Published 29 April 2005 15:06