Survey_WB_2005

Abstract

In the spirit of the empirical growth literature, we are using regionally disaggregated data to examine the question of inflation diversity and convergence in the EMU. There are several reasons that make such an approach desirable. The most obvious one is, that it enables us to increase the number of observations and thus to obtain more precise statistical results. For the case of the EMU, e.g., any crosssectional examination with national data would be based on twelve observations only. However, when regional data are used, the number of available observations can be significantly increased. In our study, e.g., we are employing data from 77 regions. Another reason for employing intra-national data is that the extra (regional) dimension can help us understand aggregated inflation behavior as we will see below. Finally, as each country can be considered as a miniature monetary union, the use of regional data from well-established monetary unions can give us insights into future developments within the EMU. In this context, the study of U.S. cities is probably most helpful. A shortcoming of this approach is that regional data are not readily available and thus have to be collected in a time-consuming process. Furthermore, even if one is willing to carry this burden, one may not be successful because some countries’ statistical offices do not compile data at a regional level. Unfortunately, this is also true for some EMU countries that are therefore missing in our sample. Nevertheless, we managed to compile a relatively broad data base of regional CPI data that includes most major EMU countries. To get an idea of the scope of our regional price data,5 we start by giving a short description of it. An overview of the included countries and regions is given in table. As one can see, we are using data from six EMU countries comprising a total of 77 regions. In our estimation analysis, we arrange these countries into two different groups, denoted as European ‘core sample’ and European ‘extended sample’. These two groups differ with respect to the sample length and the coverage of included CPI subgroups. As table 2 shows, the European ‘core sample’ comprises data for German, Austrian, Finnish, Italian, Spanish and Portuguese regions and includes the total index and eleven subgroups.6 For all regions, the subgroups are constructed on the basis of an identical classification scheme, namely the COICOP (Classification of Individual Consumption by Purpose) scheme that was introduced in most EU countries in 1995.7 In our European ‘extended sample’, we extend the length of the sample period considerably (by five years). However, only total index data are available. All data are annually and are available in index form.