Inflation in the US has receded from its 2022 peak. This column uses payroll records from 2016 to 2025 to show that many workers have experienced a lasting reduction in their purchasing power, and argues that the lack of systematic wage indexation in the US is central to this story. When inflation surged, most firms continued to grant familiar raises of 2% to 4%, allowing purchasing power to fall. Once inflation slowed, those raises again produced modest real wage growth but did not compensate workers for what they had already lost.

Inflation has receded from its 2022 peak, and unemployment remains low. Yet through 2025, Americans’ assessments of the economy remained unusually bleak. Explaining why sentiment has remained weak despite improving headline indicators has therefore become an important policy question. 

Stantcheva (2024) shows that households associate inflation with lost purchasing power and economic insecurity, while Bolhuis et al. (2024) argue that higher borrowing costs help explain the weakness of consumer sentiment. The disconnect is easier to understand once we distinguish between the inflation rate and the price level. Lower inflation means prices are rising more slowly, but it does not reverse past increases. For workers to recover lost purchasing power, wages must catch up. For many, they have not.

In a recent paper (Hurst et al. 2026), we use ADP payroll records to follow the same workers and firms from 2016 through 2025. The administrative records measure contractual base wages and the timing of each wage change for all workers at each participating firm, allowing us to distinguish job-stayers from job-changers and standard annual raises from adjustments made outside the normal review cycle.

Real wage growth returned to normal – but earlier losses were not made up

Figure 1 shows the broader trajectory of real wages for US workers as a whole, combining workers who stayed with their employers and those who changed jobs. The ADP series closely tracks a comparable measure based on the US Census Bureau’s Current Population Survey data, and both tell the same story. Real wages fell by about 4% during the inflationary period and did not regain their 2020 level until late 2024. Even then, they remained well below the path implied by pre-pandemic wage growth. By December 2025, aggregate real wages were about 7% below where they would have been had the 2017–2019 trend continued.

Figure 1 US real wages, 2017–2025: ADP and CPS

Notes: Figure shows real wage indices using CPS and ADP panel micro data. The CPS wage series comes from the weighted median nominal wage growth for all matched CPS respondents as reported by the Atlanta Fed Wage Growth Tracker. The ADP wage series combines the median nominal wage growth of job-stayers and job-changers. An index is created for both series using the monthly nominal growth rates, normalized to 1 in January 2017, and deflated by the corresponding monthly CPI-U in January 2017 prices. The red dashed line extrapolates the 2017–2019 linear trend in the ADP real wage index through December 2025.

To understand the source of this aggregate shortfall, we first zoom in on workers who stayed with the same employer. Before the pandemic, job-stayers typically saw their real pay rise about 1% per year. However, among workers who remained with the same firm from December 2020 through December 2024, 43% ended with a lower real wage than they had at the start – about twice the share of workers over a comparable pre-pandemic period. Among those who lost purchasing power, the typical decline was about 7%.

Even these statistics understate the shortfall. Zero real-wage growth is not the right benchmark: job-stayers’ real pay normally rises with experience and seniority. A worker whose purchasing power was unchanged from 2020 to 2024 therefore ended roughly 4% below the path implied by normal career wage growth.

Figure 2 Cumulative real wage change for job stayers

(a) Probability density function 

(b) Cumulative distribution function

Notes: In both panels, the sample consists of workers continuously employed at the same firm at the beginning of the window and the end of the window. The pre-pandemic window covers December 2015 to December 2019; the post-pandemic window covers December 2020 to December 2024. Cumulative real wage growth is computed as the change in the worker’s nominal base wage over the window, deflated by cumulative growth in CPI inflation over the same period (2.1% per year from 2016-2019 and 4.9% per year from 2020-2024). The vertical dashed line denotes zero cumulative real wage growth. The x -axis reports the total cumulative real wage change over the four-year period.

Firms kept giving raises – but their standard raises changed very little

Firms tend to review pay once a year and give many workers the same round-number increase. Before the pandemic, 3% was the most prominent raise. Among workers receiving one annual raise, the majority received an increase within half a percentage point of their firm’s most common raise. Those norms shifted only modestly when inflation surged. Even as prices rose by more than 7% a year, 76% of workers were at firms whose standard raise remained between 2% and 4%.

These norms developed during decades of low and stable inflation, when a 3% raise usually delivered real-wage growth. The recent episode of an unexpected inflation shock exposed their weakness. Firms continued to adjust wages, but by too little, meaning that temporary inflation left a lasting hole in real pay.

Figure 3 Time series for firm-level wage norms

Notes: Inflation is measured as the CPI-U December-to-December percent change from the Bureau of Labor Statistics. The median modal wage change is the median across firms of each firm’s modal base wage change in that calendar year. In this calculation, firms are weighted by their number of employees.

Job changes and off-cycle raises helped – but they reached too few workers, too infrequently

Workers had two main ways to escape these sticky wage norms. The first was to receive an off-cycle raise outside the firm’s normal review process. Indeed, the share of workers receiving more than one wage adjustment in a year rose from roughly 17% before the pandemic to 27% during the inflationary period. The additional raises generally occurred outside the firm’s usual review month, were larger, and likely reflected promotions or responses to outside offers.

Changing employers provided a second escape route. Job-changers’ wage gains rose roughly with inflation in the year in which they switched jobs, allowing their real wages to keep growing while those of job-stayers fell.

Jordà and Nechio (2023) show that inflation expectations became more important in wage determination after the pandemic. Our evidence suggests that much of this adjustment occurred through job changes and off-cycle raises, rather than through a broad resetting of firms’ standard wage norms.

However, both escape routes were rare for any given worker. A worker might negotiate one large raise or switch jobs once over four years, but remain subject to the firm norm in other years. One adjustment often did not offset several years of raises that lagged prices. Moreover, these adjustments carry their own costs. Changing jobs requires searching and interviewing and may mean giving up valued co-workers or benefits. An off-cycle raise may require an outside offer or difficult negotiation. Workers who kept up often spent real effort simply to preserve their purchasing power (Guerreiro et al. 2026, Afrouzi et al. 2026).

Overall, the aggregate effect of these actions was limited. Even including job-changers, 37% of workers had a lower real wage in December 2024 than in December 2020. Fifty-eight per cent fell short of the growth they would have experienced had pre-pandemic trends continued.

Figure 4 Cumulative real wage growth: All workers

(a) Probability density function: 4-year changes, pre vs post 

(b) Cumulative distribution function: 4-year changes, pre vs post

Note: Compares the distribution of cumulative real wage growth over four-year windows beginning in December 2015 (pre-inflation period, ending December 2019) and December 2020 (inflation period, ending December 2024). The sample combines job-stayers and job-changers, weighting job-changers by a factor of seven to reflect their relative frequency in the workforce.

The losses were broad – but not evenly shared

At first, lower-wage workers fared better: they changed jobs more often and received stronger wage gains while real wages fell for much of the rest of the distribution. Wage gaps narrowed largely because middle- and higher-wage workers lost ground, not because everyone was doing well. By 2024, workers across the wage distribution were below the real wage paths implied by pre-pandemic growth.

Age differences were more persistent. Older workers switched jobs less often and gained less when they did, leaving them with larger cumulative losses. Overall, workers with fewer opportunities to escape their employers’ standard pay policies suffered the largest losses.

Figure 5 Real wage growth by initial income decile, inflation period

(a) One-year real wage growth

(b) Four-year cumulative real wage growth

Notes: Panel A of the figure shows the real wage growth between December 2020 and December 2021 (solid line) and the annual one-year real wage growth averaged over the December 2015 to December 2019 period, by initial income decile. Panel B shows the cumulative four-year real wage growth by decile over the December 2020 to December 2024 period and the December 2015 to December 2019 period, respectively. All lines pool together data from job-stayers and job-changers. Sample is restricted to individuals aged 25 to 50 in the initial period.

Why lower inflation did not feel like a recovery

Taken together, these findings show that the inflationary episode was not simply a temporary burst of rapidly rising prices. It left many workers with a lasting reduction in what their pay cheques could buy. This pattern offers one explanation for both the sharp decline in consumer confidence during the episode and the persistence of discontent after inflation subsided.

The lack of systematic wage indexation in the US is central to this story. When inflation surged, most firms continued to grant familiar raises of 2% to 4%, allowing purchasing power to fall. Once inflation slowed, those raises again produced modest real wage growth but did not compensate workers for what they had already lost.

An accounting exercise in our paper illustrates the quantitative importance of this mechanism. If firms’ modal raises had moved one-for-one with inflation above its pre-pandemic average – holding observed job switching and off-cycle raises fixed – this would have closed about 40% of the shortfall relative to the 2017–2019 trend. Relative to the longer 2000–2019 trend, the corresponding share would have been roughly 73%. This is an accounting counterfactual rather than a general-equilibrium policy experiment, but it shows how much aggregate real wage dynamics can hinge on firms’ standard raises.

Belgium provides a useful contrast. Belgium and its neighbours experienced similar inflation and sharp initial declines in consumer sentiment. But most Belgian wages are automatically indexed to inflation. Belgian real wages recovered much faster, and sentiment recovered with them. Where wages adjusted more slowly and incompletely, both real wages and sentiment remained depressed longer. The relative timing is consistent with sentiment recovering when real incomes recover, rather than simply when inflation declines.

The comparison suggests that households care not only about how quickly prices are rising today, but also about whether incomes have caught up with the earlier increase in the price level. Americans are not necessarily waiting for prices to return to 2020 levels. They are waiting for their pay cheques to catch up.

Source : VOXeu