ABSTRACT
Controlling inflation is the primary objective of central banks. For this purpose, central banks need to know both the determinants of inflation and the basic features of its transmission mechanism. Given that both theoretical considerations and empirical evidence suggest that inflation expectations are a crucial element in these matters, the nature of inflation expectations must be carefully examined. Somewhat surprisingly, inflation expectations have been analysed relatively little. This is mainly because we have only a limited amount of data on “realized” inflation expectations. In most cases, inflation expectations have been derived not from observed survey or published forecast data but by using the orthogonality conditions connected with the Rational Expectations Hypothesis (REH) and employing the GMM estimator. This enables estimation and testing of key behavioural equations, but it does not really allow for an analysis of the determinants of inflation expectations. Thus the estimation results for Phillips curves are not very informative in terms of the inflation propagation mechanism and inflation expectations. Nor are they very informative for policy decisions. Take the simple question of how to reduce inflation. Conventional Phillips curve results just show the impact of the cyclical situation (e.g. output gap) on current inflation, while the role of inflation expectations stays in a “black box”, even though inflation expectations is clearly the most important variable.2 Phillips curves suffer from other problems as well. They usually fit the data poorly, and reasonable results can be obtained only by adding some auxiliary variables to the estimating equations (such as lags, in the case of the “hybrid” New Keynesian Phillips curve). Against this background, we clearly need a more general – and more data-consistent – representation of the model. The situation is quite different if we use data on “realized” inflation expectations. Then we can at least observe the independent role of inflation expectations. We can also see how inflation expectations react to other variables or policies or policy regimes. The main sources for expectations data are the regularly published macroeconomic forecasts of governments (finance ministries), research institutes and international organizations such as OECD. All these publish at least some form
of inflation forecast. The problem with most of the data is that they cannot be compared across countries, and it is difficult to construct a consistent aggregate euro area data from these series. Thus we are often limited to OECD data. Fortunately, the Consensus Forecast also provides survey-based inflation forecasts in the same format for all European countries. The problem is that the data only cover the period 1989-2003 which is very short for all analytical purposes.3
