How package size changes affect food inflation: evidence from scanner data

In recent years, increased attention has been paid to how changes in product size-especially downsizing-affect measured inflation. Despite anecdotal examples and widespread media coverage, there is limited empirical evidence on the broader role of package size reductions in inflation measurement. This paper quantifies how changes in package size among packaged food products influence measured inflation and investigates the mechanisms behind these size changes. Using U.S. retail scanner data from 2012 to 2019, we document that the average package size (measured in grams per UPC) declined by 14.6% over the period. We show that these size reductions are not the result of downsizing in the conventional sense-i.e., shrinking of the same product-but are instead driven almost entirely by product turnover: newly introduced products tend to be smaller than the products they replace. To isolate the impact of package size on food inflation, we compute two hedonic price indices: one that adjusts for size and other attributes and a counterpart that omits size. We find that when size is omitted from the hedonic model, annual inflation is understated each year. As a result, over the entire period, accumulated inflation would be 3.7 percentage points lower if size changes were not accounted for. Finally, we explore heterogeneity in size reductions. We find they are more pronounced among products purchased by high-income households, in states without unit pricing regulations, in sugary products, and in non-bulky goods. This suggests that product shrinkage may result from consumer inattention and market segmentation strategies. We benchmark our findings against official inflation statistics and methodological practices to highlight key similarities and differences.