Differences in consumption patterns between lower- and higher-income households suggest the potential for inflation inequality, but evidence on the scale and drivers of this disparity remains scarce. This column uses ‘distributional consumer price indices’ to reveal a clear and systematic gap in inflation rates across income percentiles in the US, reflecting the varying consumption patterns of households at different income levels. The findings suggest that a substantial number of individuals considered above the poverty line based on the official Consumer Price Index actually fall below it due to different inflation dynamics, and may be missing out on poverty alleviation programmes.
Rising inequality has become a pressing issue in both policy and academic debates (Atkinson and Morelli 2014, Meyer and Sullivan 2018). Could factors like inflation contribute to this growing divide? Since Engel’s seminal work (Engel 1857), we know that households with different incomes consume distinct bundles of goods and services. These consumption patterns suggest the potential for ‘inflation inequality’, where inflation affects income groups unevenly. Yet, evidence on the scale and drivers of this disparity remains scarce. How significant is inflation inequality, and what are its implications for policy? Recent research investigates this question in the US.
A new publicly available database for inflation inequality
Several years ago, a few papers documented higher rates of inflation for lower-income households for consumer packaged goods (e.g. Kaplan and Schulhofer-Wohl 2017, Jaravel 2019). These analyses covered food products, household supplies, and beauty and personal care products, representing about 10-15% of total household expenditures. The papers used detailed scanner data from private providers, rather than using the official price and expenditure data used by the U.S. Bureau of Labor Statistics to compute the official Consumer Price Index (CPI). Is the pattern of higher inflation for lower-income groups similar for the full consumption basket, using the same data as the official CPI?
In new research Jaravel (2024), I address this question by introducing a publicly accessible database that leverages high-frequency public data: monthly CPI price changes and annual expenditure shares from the Consumer Expenditure Survey. This methodology mirrors CPI construction, ensuring consistency with official inflation statistics while allowing disaggregation by socio-demographic groups (e.g. income percentiles, age, race, occupation). The resulting ‘distributional consumer price indices’ (D-CPIs) enable tracking of the distributional effects of inflation from 2002 onwards, with updates available within hours of monthly inflation data releases. This real-time tool offers valuable insights into the socio-economic dimensions of inflation.
All data are made available on the D-CPI project website. Researchers and policymakers can use the website to download price indices for selected socio-demographic groups or work directly with the micro data and generate price indices for additional groups, using any socio-demographic characteristic observed in the Consumer Expenditure Survey.
Results
The D-CPI database reveals a clear and systematic gap in inflation rates across income percentiles, reflecting the varying consumption patterns of households at different income levels. Figure 1 provides a detailed look at this disparity from January 2002 to November 2024, showing that cumulative inflation over this period was significantly higher for lower-income households. By November 2024, the inflation rate for the lowest income percentiles had reached approximately 90%, while it was only about 74% for the highest income percentiles, creating a 16 percentage point gap in cumulative inflation. From 2002 to 2024, the average annual inflation rate for the least affluent households was 2.96%, compared to 2.54% for the most affluent. This translates to an annual inflation gap of 41 basis points, with the least affluent households experiencing a noticeably higher rate of price increases year over year.
Figure 1 Long-run inflation inequality by income percentile
To measure the importance of this trend on inequality, it is instructive to compare household income growth across the income distribution using the official CPI index and the indices accounting for inflation inequality. Focusing on the period from 2002 to 2019 (i.e. stopping the analysis before the onset of the Covid-19 pandemic), Figure 2 shows that, according to the official CPI, household real income growth between 2002 and 2019 was higher at the top of the income distribution, ranging from 7.8% for the bottom income quintile to 24.6% in the top income quintile, and up to 26.5% for the top 5% of households. This gradient becomes considerably steeper with the income group-specific price indices. After accounting for inflation inequality, household real income growth is only 2.4% at the bottom of the distribution, i.e. earnings are almost stagnating, while income growth at the top is even faster, at 25.4% for the top quintile and 27.8% for the top 5%. Thus, according to the official metric, the income gap between the top and bottom quintiles increased by 15.6% between 2002 and 2019 (1.246/1.078). When accounting for inflation inequality, this gap grows significantly more, rising by 22.5% (1.254/1.024). This implies that the rate of increase in real income inequality is approximately 45% faster when inflation inequality is considered compared to using the official CPI. Similarly important adjustments are found when looking at consumption inequality or inequality in disposable income after taxes and transfers.
Figure 2 Implications for household real income growth, 2002 to 2019
Besides the measurement of inequality, the higher rates of inflation for lower-income groups matters for the indexation of the poverty line and the number of people in poverty. The official CPI fails to account for the fact that inflation is higher for individuals in poverty, meaning the poverty line should be indexed at a higher rate. Instead, D-CPIs can track the inflation rate experienced by individuals at the poverty line. Figure 3 compares the price index for the poverty line to the official CPI. Over time, a gap emerges gradually between the two, becoming particularly significant by the end of the period. This growing disparity highlights the limitations of the official CPI in accurately reflecting inflation experienced by those near or below the poverty line. Using the D-CPI index to calculate the number of people below the poverty line, it becomes clear that a substantial number of individuals are misclassified by the official CPI – they are considered above the poverty line despite being below it due to different inflation dynamics. By 2024, approximately 2.3 million people fall below the ‘real’ poverty line but remain above the official poverty threshold. This misclassification has important policy implications, as these individuals should have access to poverty alleviation programmes, such as Medicaid.
Figure 3 Cumulative Price Index by poverty status
D-CPIs also reveal substantially higher inflation rates for older households between 2002 and 2024, with potential implications for the indexation of Social Security payments. Extensions show that the inflation heterogeneity patterns by age and income described above were also similar in an earlier period, going back to 1983.
Conclusion and policy implications
The literature on the measurement of inequality has largely relied on common price indices. To date, Inflation heterogeneity has not been considered an important factor in understanding long-run inequality dynamics and the measurement of real incomes across household groups. Building a publicly available database, the paper demonstrates that inflation heterogeneity is, in fact, crucial for measuring long-run trends in income and consumption inequality.
Source : VOXeu
