Utilisation of scanner data in the measurement of inflation – international experiences and methodological challenges

Abstract

According to the Organisation for Economic Co-operation and Development’s definition, scanner data are detailed data on consumer goods obtained by scanning their bar codes at points of sale. The advantages of this kind of data include completeness at the lowest level of aggregation, relatively low cost of acquisition and the fact that they enable a multiplicity of observations. Nevertheless, scanner data also have drawbacks and limitations. The aim of the paper is to identify the problems and methodological challenges related to the acquisition, processing and aggregation of scanner data, which are then used to estimate the Consumer Price Index (CPI). One of the most important decisions in the whole process is the right choice of an index formula dedicated to elementary (homogeneous) product groups. The essence of the problem, along with some recommendations, has been presented on the example of two sets of data from Allegro e-commerce platform for the period of 4 Dec, 2015 to 28 Dec, 2018, obtained through a special tool – TradeWatch. The evaluation of the sensitiveness of the results of the price dynamics measurement according to the chosen index formula has been carried out on the basis of two groups of products: men’s sports watches and office chairs. The most important observations are as follows: firstly, the differences between bilateral price indices and their chained versions are likely to be significant because of the dynamic character of scanner data sets; secondly, the differences between multilateral price indices might amount to several percentage points for a yearly time window; thirdly, the differences between the values of the GEKS and CCDI indices are slight, and the Geary-Khamis index for the full-time window ceases to differ significantly from the real time index just after a few months; and fourthly, the prices of products sold via electronic platforms as well as their quantities and sales volumes might differ depending on which particular day of a week they were sold, and even at which hour.