In the run-up to the 2024 U.S. presidential election, few economic issues loomed larger than inflation. Grocery prices, rent, and gas costs featured heavily in campaign messaging, and voters told pollsters that the cost of living topped their concerns. But when analysts try to disentangle what voters were actually responding to, a puzzle emerges: were people angry about rising prices themselves, or about the fact that their paychecks weren’t keeping up?
Two economists set out to test this question using granular local data, and their answer complicates the standard story. In a working paper circulated by the National Bureau of Economic Research, Juan Felipe Riaño of Georgetown University and Francesco Trebbi of the University of California, Berkeley’s Haas School of Business report that declining real wages, not inflation on its own, tracks with Republican gains across U.S. counties. Once you account for how much purchasing power households actually lost, higher local inflation is associated with counties shifting less toward the Republican ticket, not more.
The question behind the numbers
Economists have long argued that voters use elections to reward or punish incumbents for economic conditions. This “pocketbook voting” idea has a straightforward prediction: when household finances worsen, the party in power loses ground. Between 2021 and 2024, the U.S. experienced its most sustained inflation episode in four decades, with cumulative price increases exceeding 20 percent and real wages falling behind. Since Democrats held the White House, the pocketbook theory predicts that voters should have swung Republican in places hit hardest.
But there are two ways to think about “hardest hit.” One is the gap between wage growth and price growth, meaning how much real purchasing power households actually lost. The other is the raw pace of price increases, which might carry political weight independent of wages because prices are visible, salient, and psychologically painful. Riaño and Trebbi designed their analysis to tell these two channels apart.
Building a local measure of the cost of living
The standard U.S. inflation measure, the Consumer Price Index, is only calculated for a limited set of metropolitan areas. Most rural and exurban counties simply aren’t covered. To get around this, the researchers turned to the Economic Policy Institute’s Family Budget Calculator, which estimates the annual cost of a “modest but adequate” standard of living for a two-parent, two-child family in every U.S. county. The calculator breaks costs into seven categories: food, housing, childcare, transportation, healthcare, other necessities, and taxes.
The researchers paired these local cost estimates with county-level wage data from the Bureau of Labor Statistics, county-level election returns from the MIT Election Data and Science Lab, and demographic controls from the Census Bureau. Their final sample covered 3,102 counties observed in both the 2020 and 2024 presidential elections.
To validate their price measure, they compared it against the CPI in the metropolitan areas where both exist. In those overlapping regions, the two measures correlated strongly, giving them confidence that their county-level index captured real price signals rather than statistical noise.
Separating wages from prices
The analytical trick at the heart of the paper is a reparameterization. Rather than treating wage growth and price growth as two independent variables, the researchers constructed a “wage-price gap” equal to nominal wage growth minus inflation. This gap measures the change in purchasing power directly. They then ran regressions using both the gap and inflation as predictors, with state fixed effects and controls for urbanization, density, age, education, GDP, and demographic composition.
This setup lets them ask two distinct questions. First, does real wage growth predict how counties voted? Second, does inflation carry any additional predictive weight beyond its effect on real wages?
The pattern that emerged pointed in two directions at once. Counties where real wages fell more sharply relative to the state average shifted more strongly toward the Republican candidate, consistent with the classical retrospective voting story. But the coefficient on inflation, holding the wage-price gap constant, was negative: counties with higher residual inflation actually shifted less toward Republicans.
Why higher inflation would go with weaker Republican gains
This second finding runs against the intuition that voters punish incumbents for rising prices. Riaño and Trebbi offer an interpretation grounded in what “higher inflation, same wage-price gap” implies about local economies. If the gap is fixed, then higher prices mean higher wages too. Both are rising faster together, which is the signature of a hot local labor market with strong demand: employers competing for workers, wages being renegotiated, and prices adjusting quickly.
The authors argue that these demand-heavy counties tend to be higher-income and more urbanized, and they have been trending Democratic for reasons that predate the 2021-2024 inflation episode. In other words, the negative coefficient on residual inflation appears to capture a compositional pattern: the counties experiencing the fastest nominal growth are the same ones already shifting toward Democrats as part of a longer-running realignment along income and urbanization lines.
The researchers checked whether this could simply be the continuation of pre-existing partisan trends. They controlled directly for Republican vote-share changes going back as far as 2008. The lagged trends themselves showed up as strong predictors of 2024 voting, confirming that realignment is real. But adding these controls barely moved the inflation coefficient, suggesting the 2021-2024 relationship isn’t just a mechanical extension of earlier trends.
Turnout tells a different story
When the researchers looked at voter turnout instead of vote share, a cleaner pocketbook pattern emerged. Counties where real wages grew faster saw larger increases in presidential turnout. This is consistent with a resource-based view of political participation: when households feel financially stable, they’re more likely to engage with the electoral process.
The turnout finding also helps explain why the wage-price gap alone doesn’t predict partisan swings very well. Real wage losses appear to affect whether people vote more than they affect which party they choose, at least in this cycle.
The presidential-congressional split
The researchers replicated their analysis at the congressional district level using U.S. House election returns. The results were noisier, but one clear pattern held: inflation showed no robust association with anti-incumbent voting in House races. Even when the researchers constructed a “challenger party” outcome to test the incumbent-punishment mechanism directly, the effect washed out.
The authors interpret this as evidence that voters hold the executive branch, not their local congressional representative, responsible for national economic conditions. Inflation and real wage losses appear to translate into presidential voting swings but not congressional ones, at least in the aggregate.
What the study can and can’t say
Riaño and Trebbi are explicit about the limits of their design. Their estimates come from comparing counties within states, which absorbs the aggregate national effect of the inflation shock. They cannot say how the national vote share would have moved absent the price surge, only how counties moved relative to each other.
They also cannot fully separate two possible explanations for the negative inflation coefficient. One is compositional: high-inflation counties happen to be Democratic-trending places for reasons independent of prices. The other is behavioral: in booming local economies, the favorable elements (tight labor markets, rising wages) may be more salient to voters than the unfavorable ones (rising prices). Distinguishing these would require either exogenous variation in local prices or individual-level data linking experienced inflation to vote choice.
A separate caveat concerns the price measure itself. The EPI Family Budget Calculator estimates the cost of a defined living standard rather than pricing a fixed basket. If the definition of “adequate” housing or childcare shifts over time, some of the measured cost growth reflects standards updating rather than pure price changes. The researchers argue that over a three-year window this is unlikely to matter much, and that any such shifts are largely absorbed by state fixed effects since EPI’s reference standards are set at national or broad-area levels.
There’s also a tension the authors flag between their findings and survey evidence. Households consistently tell pollsters that inflation is harmful and blame incumbents for it. Yet the county-level data show high-inflation places shifting Democratic. The resolution, the researchers suggest, lies in which counties experience the highest residual inflation: these are systematically the higher-income, more urbanized, Democratic-trending areas whose overall economic experience during the period was better than the price signal alone might suggest.
