Index methodology for transaction data in the Dutch CPI

This report documents the development of methods for measuring inflation using transaction data at Statistics Netherlands since the late 1990s, with particular emphasis on the currently used methods in the Dutch CPI.
The Consumer Price Index (CPI) is used as the primary measure of inflation. To calculate inflation, Statistics Netherlands (CBS) increasingly utilises transaction data. CBS receives these electronic data sources from retail chains mostly on a weekly basis. In addition to prices, these files contain sales quantities for all products scanned in stores. In principle, this rich data offers more opportunities for accurate inflation measurements than traditional price observation, which only collects prices for smaller product samples.

However, using both prices and quantities within a dynamic universe of emerging and disappearing products makes the search for an index method that can handle various forms of dynamics—such as clearance prices and seasonal patterns—far from trivial. Traditional bilateral methods are sensitive to drift, while new products are not incorporated in a timely manner.

Dynamics in prices, quantities, and product turnover can, however, be effectively processed using multilateral methods. These methods utilise product data from more than two periods used in bilateral methods. In the CPI, CBS employs the Geary-Khamis method alongside a technique to suppress drift in non-revisable CPI series. All products are included in the monthly inflation calculation, which contribute to aggregate price change according to their expenditure share. This method is applied across all transaction data within the CPI.

Downloads