Statistical Implications of Inflation Targeting

19 Taking the Agenda Forward

Carol Carson, Claudia Dziobek, and Charles Enoch
Published Date:
September 2002
  • ShareShare
Show Summary Details
Carson, Carol S., Enoch, and Charles and Dziobek Claudia, 

This chapter seeks to distill the main conclusions of this volume and from discussions during the seminar. One conclusion of the seminar is that statistical issues that arise for countries with explicit inflation-targeting regimes are largely the same as those of other major central banks, such as the European Central Bank (ECB) or the Federal Reserve Board, that view price stability as a high priority. It appears that the conclusions of the seminar are relevant for many IMF member countries. The papers and the discussion brought out four topic areas: (1) information requirements and more systematic ways of handling information, (2) forward-looking indicators and target variables, (3) institutional arrangements to produce and use data, and (4) an agenda for the IMF looking forward.

Information Requirements and More Systematic Ways of Handling Information

The information used for traditional monetary policy remains largely relevant in an inflation-targeting setting. However, an inflation target focuses attention on published data on prices, which in turn entails a demand from central banks for more details and more indicators of price data, and a better grasp of economic sentiment and expectations to forecast inflation and take policy action. Inflation targeting also provides more discipline for research and for data presentation because central bankers require more knowledge about the time lags for policy actions and more details on the transmission mechanism in order to be able to predict inflation 18 to 24 months ahead.

Several authors refer to this discipline and added focus as a new way of handling information. Handling of information also implies a demand for more details about raw data (to allow “slicing and dicing”), more detailed information on survey distributions, and more detailed metadata (that is, information about the methodology used in compiling data). For example, research by the Bank of Canada shows that, during periods of high and variable inflation, the distribution of inflation forecasts was bell-shaped and very broad. Since inflation targeting, the distribution has collapsed very much on the rate of wage increases. Central banks use this kind of information to model wage and price setting as it changes over time. The new way of handling information also influences thinking about information used internally and what information should be communicated to the public.

While transparency itself is not an end, it is an important means to foster credibility. A number of tools are available, including providing minutes of policymaking meetings to the public, predetermining dates for policy action, issuing press releases, writing letters to the government or to parliament to explain deviations from targets, and issuing periodic inflation reports detailing the information and analysis that contributed to policy decisions. Important messages to get across to the public are the lags involved in policy action and their results, the combined role of scientific assessments and judgment, and the idea that there are risks.

Forward-Looking Indicators and Target Variables

Forward-Looking Indicators

Forward-looking indicators are central to policymaking in an inflation-targeting environment. These include surveys on price expectations and market-based indicators, such as interest rates and yields. Inflation expectation surveys include household surveys and surveys of professional forecasters. On household surveys, the discussion suggested that while household views are very important to understanding and modeling of the transmission process of monetary policy, there are some practical difficulties involved in collecting meaningful information—for instance, in providing appropriate incentives to households in responding to survey questionnaires. Focusing on wage expectations rather than on inflation per se may be a useful alternative or an additional way of capturing household sentiment. It was also noted that, while household surveys may not be accurate mirrors of actual inflation, they do influence economic decisions and thus actual outcomes. Concerning professional forecasts, some central banks conduct surveys of professional forecasters and disseminate the outcome on their website. This can be set up so as to provide incentives for forecasters and to foster the development of market expectation data—for example, by highlighting the accuracy of individual forecasts over time.

Interest rates and yield curves, already well-established data requirements for monetary policy, are another important source of forward-looking indicators. Many inflation-targeting countries invest resources to extract information from these data sources, both to obtain better data and to model the information that can be extracted from these data.

Nonfinancial asset prices and indicators of accumulated wealth are useful complements to forward-looking indicators, though such indicators are not well developed in most countries. Such data are important because household behavior is driven by wealth effects as well as by income flows. For the statistical community, the tracking of asset prices and of wealth poses challenges and potentially very useful contributions to central banks.

Target Variables

Choosing the appropriate target is the central issue of inflation targeting. Most often, a consumer price index (CPI) is chosen. The rationale for this choice is that CPI is well known, transparent, credible, available, monthly, timely, rarely revised, of high quality, and relevant in the economy. The relevance is increased by the fact that a CPI is often combined with measuring expectations. For example, indexed bonds often use a CPI as the reference.

While the CPI is most often the target, central banks look at more specialized subindicators—for example, core indicators of inflation. Responding to the needs of central banks, national statistical offices often publish on such core indicators, leaving out, for instance, tax effects or some housing costs. Where the various measures diverge, there may be risks to the credibility of the policy regime, which therefore requires additional research and explanation to the public.

There may be trade-offs between accuracy and timeliness; a few countries, such as Australia, find quarterly CPI data adequate for monetary policy purposes. Many other countries and the ECB, on the other hand, are very interested in more timely information. Some countries with a history of high inflation, such as Brazil, have weekly and even daily inflation indicators, but this may no longer be warranted. Moreover, for policy purposes, weekly inflation indicators do not correspond to the needs of monetary policy, which works through a much longer time horizon. Moreover, to the extent that higher frequency data imply greater variability, it may be more difficult to distinguish between the signal and the noise of a price change.

Institutional Arrangements to Produce and Use Data

The added emphasis on data and data quality in an inflation-targeting regime throws a new spotlight on national statistical offices as providers of much of the data. Key elements for the credibility of an inflation-targeting regime are an independent central bank and independence for the statistical office that provides data. For central banks, a useful distinction can be made between goal independence (the right to set the goal of policy) and instrument independence (to set the policies once the target is set). Concerning the first, country practices differ. But once the goal is established, the independence of the central bank in choosing the policies to achieve the target is of prime importance. For statistical offices, independence covers, among other factors, a clear legal mandate to collect data, independence in choice of sources, protection of the staff from outside interference, and independence in the dissemination of its work.

Beyond independence, the importance of communication between data providers and data users highlights the importance of investing in statistical capacity to produce the data on which policy is focused. With regard to statistical capacity, very much depends on the country-specific circumstances. Often there are also important issues about the funding of statistical agencies. While it is part of the political reality that it is often easy to cut the budget of the statistical offices, especially in the context of an inflation-targeting regime, cutting might jeopardize the regime and its credibility, and it could ultimately prove very costly.

The collection of data from market sources—for example, data on government bond markets—may entail some new arrangements. In some countries, public debt management is spun off to independent agencies or outsourced to the private sector, requiring new arrangements for the transmittal of relevant data for policy analysis. In some countries market data derive directly from the markets—for instance, the stock exchange. Such data may be high frequency and timely, but construction of data and dissemination may be a “black box” that the authorities have little ability to influence.

Agenda for the IMF: Looking Forward

As focus shifts to include forward-looking and market-based indicators, it needs to be considered whether the IMF’s International Financial Statistics (IFS) may need some updating to incorporate more of these indicators. Some caution is perhaps called for, given the limited experience with some of the more experimental data on expectations. Similarly, the IMF should consider whether it would be useful to incorporate some of the variables brought to the fore under an inflation-targeting regime into the Special Data Dissemination Standard (SDDS). There may be a case also to use the Dissemination Standards Bulletin Board (DSBB) to put the information in the public domain, some of which—core inflation, for example—may go beyond what the SDDS currently calls for.

Much of the discussion focused on countries with developed financial markets—these indeed represent the majority of those countries that have so far adopted inflation targeting. However, the prerequisites for moving to inflation targeting apply to a much broader set of countries. For these countries the statistical requirements may be particularly acute. The IMF’s Statistical Department has an ongoing program to provide technical assistance to enhance statistical capacity. As noted throughout the seminar, such capacity is particularly critical for countries with inflation-targeting regimes. There is also scope for research on how to apply the requirements of such regimes to countries with less-developed financial markets. A more specific suggestion for further research was made in the context of forward-looking indicators. Industrial countries extract valuable information from highly developed government bond markets, a source of information not available in many emerging countries. Further research work to identify appropriate market indicators in emerging market countries in order to generate indicators of inflation expectations would certainly be useful.

Finally, the discussion on data requirements for an inflation-targeting regime has implications for the data that the IMF itself should request from, and disseminate on, a country in its regular Article IV surveillance exercises. This indeed was the issue that prompted the seminar and this book. Data needs are evolving, including those for the Article IV surveillance exercise. A number of participants at the seminar—whether conducting the Article IV consultations from the IMF side or participating from the side of the member countries with inflation-targeting regimes—indicated that in the future different (or additional) datasets would underpin their discussions. With many of these countries already publishing their Article IV reports, these discussions should give extensive insights into how the statistical requirements for inflation targeting will be applied in practice.

    Other Resources Citing This Publication