Inflation Stopped Rising So Quickly. The Pay Cut Never Came Back.

A new University of Chicago working paper gives a remarkably good explanation for why people can be told inflation is under control while still feeling substantially poorer. The researchers analyzed ADP payroll records covering roughly 16 million U.S. workers per month from 2016 through 2025, and their central finding is painfully simple: companies kept giving raises designed for a 2% inflation world while prices were rising 7%, 8%, and 9%.

Before the pandemic, the typical firm's annual raise was around 3%. When inflation surged, companies barely changed that norm — even at the peak, most workers remained at firms where the standard raise was only 2% to 4%. The chart on page 25 makes this almost comically obvious: inflation shoots toward 7% while the typical company raise inches from roughly 3% to 3.5%. Workers absorbed the difference.

Among people who remained with the same employer from 2020 through 2024, 43% ended the period earning less in real terms than when they started, and among those who lost purchasing power, the average decline was almost 9%. Including people who changed jobs only improves the number to 37%. The paper estimates that roughly one in five American workers ended 2024 with real wages more than 9% below their 2020 level.

Lower inflation does not mean prices went back down. And getting another ordinary 3% raise after inflation returns to normal does not restore the purchasing power you lost while prices were rising much faster than your paycheck. The loss becomes permanent unless somebody eventually gives you a large enough raise to catch up.

Changing jobs was the escape hatch. For job-stayers, only about 27% of inflation flowed through into wage growth. For job-changers, it was 96%. That puts some economics behind the familiar experience of discovering that the best way to get a meaningful raise is to leave. Workers who stayed could sometimes escape through promotions or off-cycle adjustments, but those required negotiating, getting another offer, or otherwise forcing the issue — spending time and effort defending their purchasing power.

The age breakdown is especially ugly. Among workers 50 and older, 55% experienced a real wage decline. They changed jobs less often, got less benefit when they did, and were less likely to receive large internal raises. Younger workers had more ways out. Older workers absorbed the loss.

The authors also argue that when companies allow inflation to reduce real wages, lower labor costs benefit the employer. Corporate profits as a share of GDP rose from an average 11.4% before the pandemic to 13.1% during 2021–2025, the highest sustained level in roughly half a century. They do not claim sticky wages caused the entire increase — they call it a consistency check — but the magnitude is close enough to their estimated real-wage losses that the distributional story is hard to ignore. Workers received raises. Companies could truthfully say wages were increasing. Because those raises were smaller than inflation, a significant portion of workers effectively received a multi-year pay cut while corporate profitability rose.

Belgium automatically indexes wages to inflation. Real wages fell there too at first because indexing lags, then caught back up by mid-2023 as consumer confidence recovered. In Germany, Denmark, and the Netherlands, wages stayed depressed and confidence stayed low. People were not merely angry that prices went up — they actually lost purchasing power. And many never got it back.

Inflation is usually discussed as though the problem ends once the rate comes down. For millions of workers, it didn't. Their employer quietly reset their standard of living downward by continuing to hand out the same 3% raise it had always handed out. Inflation eventually ended. The pay cut remained.

Read Sticky Wage Norms and the Real Wage Cost of Unexpected Inflation.

This is an August 2026 working paper by Erik Hurst, Christina Patterson, Nela Richardson and Ye Liv Wang. It has not yet gone through the full journal publication process.