Spatial comparisons of consumer prices using big data: Empirical evidence from Italy

Abstract

Accurate measurements of price level differences across geographical areas within a country is essential to better assess regional disparities, thus enabling policy makers to adequately identify and address areas of intervention. The increasing availability of new source of data concerning household consumption expenditure (i.e. scanner data) may change the current approach for estimating sub-national spatial price indexes from both a methodological and an empirical point of view. Sub-national spatial price indexes are estimated by focusing on the stochastic approach since it allows us to incorporate modern econometric tools and deal with the issue of spatial dependence which is inherent in consumer prices. We use a scanner dataset set up for experimental CPI computations for Italy in 2017 which includes information on prices, quantities and quality characteristics of products at barcode level. This dataset refers to grocery products sold in a random sample of approximately 1,800 outlets across Italian provinces belonging to the most important retail chains (95\% of modern retail trade distribution), covering 55.4\% of total retail trade distribution for this product category. Estimates of provincial and regional spatial price indexes obtained using various index number formulae for specific product groups and for Food and Non-Food consumption aggregates are presented