Computing Sub-national PPPs with CPI Data: An Empirical Analysis on Italian Data Using Country Product Dummy Models

The Italian National Statistical Office is implementing a new project for computing sub-national Purchasing Power Parities (PPPs) on a regular basis, which is based on an appropriate use of existing Consumer Price Index (CPI) data and new sources of data. Concerning the use of CPI data, in this paper the role of the Country Product Dummy (CPD) method for compiling sub-national PPPs at Basic–Heading (BH) level is analysed together with the specific issues that arise in this context, such as the need to take spatial autocorrelation among price relatives into account. The results of various experiments based on CPI data concerning 7 BHs and 19 Italian regional chief towns are presented and discussed with the aim of exploring the performance of various CPD models and analysing to what extent the type and characteristics of the data affect the estimates obtained. The statistical uncertainty associated with the ranks derived from the regional PPPs was then calculated by means of a simulation procedure. Our findings prove to be interesting and confirm that methods and CPI data for spatial comparisons are reciprocally influenced.