Measures of underlying inflation and their role in the conduct of monetary policy

Proceedings of the workshop of central bank model builders held at the BIS on 18-19 February 1999


On 18 and 19 February 1999 model builders from central banks met in Basel to discuss the issues of how best to measure underlying, or core, inflation and the implications of the use of alternative measures of core inflation for the conduct of monetary policy.

The concept of underlying inflation has always been central to the monetary policy strategies of central banks; the recent sharper focus on price stability, of which the growing number of countries adopting inflation targeting strategies is only one indication, has made it increasingly important to have an accurate and reliable measure of core inflation. This need arises from the notion that central banks should only resist persistent sources of inflationary pressures and not be concerned with short-term and reversible movements in prices and the inflation rate.

Yet there is no consensus on how to extract a solid measure of long-term price movements from headline inflation.

The nine papers presented at the conference follow three different approaches to this signal extraction problem.

According to the first one - which one can call the behavioural approach - estimates of core inflation are obtained by excluding from the headline measures the prices of certain items that are thought to be volatile enough to obscure long-term movements of inflation. The price index "excluding food and energy" is one well-known example. Several countries compute such indices and consider them in the setting of policy.

Another approach - the statistical approach - attempts to eliminate temporary fluctuations of inflation, or one-off changes in the price levels, by computing limited influence estimators, such as the median and/or the trimmed means. These measures are thought to have desirable properties to the extent they avoid the subjective decision to exclude particular prices from the aggregate price index and because they efficiently estimate long-term movements when the data are drawn from a leptocurtic distribution. The papers by Wynne, Apel and Jansson, and Bryan, Cecchetti and Wiggins discuss this approach.

However, despite the potential superiority of the "statistical" measures, central banks might find it difficult to use them primarily because they are not easy to explain to the public and because they are difficult to replicate. The papers by Johnson, Cockerell and Álvarez and Matea discuss such measurement issues and illustrate what central banks do in practice.

The final approach - the economic approach - tries to derive a measure of core inflation using the long-run neutrality assumption of monetary theory and to explore to what extent alternative measures of core inflation, once they are entered in a given feedback rule for monetary policy, produce different economic outcomes in terms of variability of real output and inflation and instrument instability. The papers by Aucremanne and Wouters, Fase and Folkertsma, and Cassino, Drew and McCaw explore these issues.

Papers presented