This study compares different methods of calculating the core inflation rate, the latter being taken as the general price trend, i.e. the persistent component of measured inflation. This price trend is therefore assumed to be free of transitory price movements. This paper focuses on an empirical analysis of data taken from the consumer price index for (western) Germany. In addition, the different methods are examined critically from a theoretical perspective, focusing on two types of procedure: statistical methods and methods based on economic theory. The latter include a new procedure which is based on the P* approach. In order to assess the advantages and disadvantages of the various approaches, certain features of the core rates are tested, i.e. to establish whether they meet certain criteria, which would determine their suitability as indicators of the price trend. The principal finding of the study is that core inflation rates - irrespective of the method chosen - are not always able to meet the requirements. The methods encounter particular difficulties with regard to avoiding bias in relation to measured inflation, these problems being aggravated when adjustments are made to account for the effects of taxation. The predictive quality with regard to measured inflation also frequently leaves a great deal to be desired. In particular, however, the methods are unable to distinguish adequately between transitory and permanent components of the inflation rate. Statistical core inflation rates perform relatively well, while core rates based on economic theory suffer, in particular, from a constant need for revision. Even so, the newly introduced P* method is convincing, at least in the latter category. Given the relatively poor overall outcome, it would appear advisable not to use core inflation rates as the sole monetary policy indicators; however, they are a useful complement to measured inflation. Moreover, rather than focusing on one method only, it would be better to combine several selected methods.
