Are lay expectations of inflation based on recall of specific prices? If so, how and under what conditions?

Abstract

In 2019 when inflation was low and stable, we compared people’s direct estimates of inflation with indirect estimates obtained by averaging their estimates of price changes in all 12 product categories on which the consumer price index is based. Indirect estimates were much higher than direct ones and the two types of estimate were uncorrelated. This is consistent with a price-free model in which direct estimates are not based on recall of prices but are determined by other information such as media reports. In May 2022 when inflation was high, expected to rise in the short-term, but highly unpredictable in the longer term, we found that direct and indirect estimates were very similar and highly correlated. This is consistent with a price-recall model in which a representative set of prices is used to estimate overall inflation. Finally, in September 2022 when inflation was fairly stable except for certain product categories (food) where prices were rising rapidly, we found that results were consistent with a third approach, the price-salience model; overall estimates of inflation are selectively influenced by price rises in those product categories where they are particularly high. People’s strategy for estimating inflation appears adapted to the prevailing inflation environment.