.png)
357 – Statistical Methodologies in Consulting
Enhancing the Quality of Price Index Estimates Combining Updated Weights: A More Representative Sample Design and a Different Aggregation Structure
Daniele Toninelli
University of Bergamo
Zdenek Patak
Statistics Canada
Martin Beaulieu
Statistics Canada
Consumer Price Index (CPI) and Service Producer Price Index (SPPI) calculation proceeds in stages. In the first stage, elementary aggregates of homogeneous sets of products are estimated. These elementary indices are then aggregated to obtain higher-level indices using, as weights, the elementary expenditure aggregates (CPI) or fiscal data from Business Register (SPPI). As regards SPPI, the robustness of the sampling design and the reliability of the index is often linked to an extended use of aging samples and basket weights in the aggregation process. By means of a simulation study based on data collected on business services and wholesale and retail trade, this paper aims at understanding how often weights should be updated to reduce the bias of the index. As regards CPI, in order to better represent the market shares of the various kind of stores selling same types of products, stratification by outlet type could be done. This raises questions as to what level would be more appropriate to define as the elementary aggregate. Our simulation study aims at assessing the impact on the index of the switch from the Jevons to the Laspeyres index at a given level of the aggregation.