Measuring inflation with heterogeneous preferences, taste shifts and product innovation: methodological challenges and evidence from microdata

This paper provides an extensive literature review and analyses some open issues in the measurement of inflation that can only be explored in depth using micro price data. It builds on the analysis done in the context of the ECB’s strategy review, which pointed at directions for improvement of the Harmonised Index of Consumer Prices (HICP), including better quantification of potential biases. Two such biases are the substitution bias and the quality adjustment bias. Most analyses of substitution bias rest on the concept of the cost of living, positing that preferences are stable, homogeneous and homothetic. Consumer behaviour is characterised by preference shifts and heterogeneity, which influence the measurement of the cost of living and substitution bias. Climate change may make the impact of preference shifts particularly relevant as it causes the introduction of new varieties of “green” goods and services (zero-kilometre food, sustainable tourism) and a shift from “brown” to “green” products. Furthermore, PRISMA data show that consumption baskets and thus inflation vary across income classes (e.g. higher-income households tend to buy more expensive goods), pointing to non-homotheticity of preferences. When preferences are heterogeneous and/or non-homothetic, it is important to monitor different experiences of inflation across classes of consumers/citizens. This is particularly important when very large relative price changes affect items that enter the consumption baskets of the rich and the poor, the young and the old, in very different proportions. Another open area of analysis concerns the impact of quality adjustment on measured inflation. Evidence based on web-scraped prices shows that the various implicit quality adjustment methods can produce widely varying inflation trends when product churn is fast. In the euro area specifically, using different quality adjustment methods can be an overlooked source of divergent inflation trends in sub-categories, and, if pervasive, shows up in overall measured inflation divergence across countries. JEL Classification: E31