This paper develops a new public database to estimate inflation heterogeneity across socio-demographic groups in the United States in real time. These distributional CPIs (D-CPIs) are fully consistent with the methodology of the official CPI and are available from 2002 to the present day by household income, age, race and other characteristics. Using this database, I establish three results showing that D-CPIs have important implications for the measurement of long-run trends in inequality and poverty, as well as of real wage dynamics during crises. First, "real" inequality across household income quintiles increased about 45% faster with DCPIs between 2002 and 2019, compared to the official CPI. While the pre-tax income gap between the top and bottom household income quintiles increased by 15.6% during this period according to the official CPI, it increased by 22.6% with D-CPIs. Similarly large adjustments apply to consumption inequality and inequality in pre-tax and post-tax national incomes. Second, today 2.3 million people are below the "real" poverty line using D-CPIs but above the poverty threshold using the official CPI. Third, during the inflation burst following the Covid-19 pandemic, inflation was higher for the middle class, compared to low-income and high-income households. This pattern is driven by gas and vehicles and implies that the compression of "real" wages was about 25% faster with D-CPIs than with the official CPI. Similar patterns of inflation heterogeneity hold in extensions allowing for geographic heterogeneity in inflation, non-homothetic price indices, and expanding the analysis back to 1983.
Xavier Jaravel
5 June 2025 Paper Number POIDWP121
Download PDF - Distributional consumer price indices