Real wages

Have American wages kept up with prices — month to month through the recent inflation, and cumulatively over the past two decades?

The real wage is the nominal wage divided by the price level: what an hour of work buys, not what it pays. Standard theory ties long-run real wages to productivity, while in the short run nominal wages adjust slowly — so an inflation surprise cuts real wages first and wage catch-up comes later, if at all. The gap between the two matters because people often reason in nominal terms (“money illusion”): a 5% raise during 8% inflation feels like a gain but is a roughly 3% pay cut in purchasing power.

Average hourly earnings of private-sector employees (monthly, from Mar 2006) and the consumer price index for all urban consumers (CPI-U, monthly). Both are national, seasonally adjusted series; several monthly snapshots legitimately skip October 2025 because of missed federal releases, and every 12-month comparison here is calendar-keyed so those windows are dropped rather than misaligned.

Year-over-year percent change is computed for both series; monthly real wage growth uses the exact ratio form ((1 + g/100) / (1 + π/100) − 1) × 100 rather than the g − π shortcut, which overstates real gains when inflation is high. The cumulative index deflates each month’s nominal wage to Mar 2006 dollars and normalizes the series to 100 at that base, so any point reads directly as purchasing power relative to the start.

The answer is: mostly, eventually — with a long detour. Wages roughly tracked prices before the pandemic, real pay then spiked and slumped through 2020–21 as the workforce’s composition shifted, and the 2021–23 inflation put real wage growth below zero for about two years before nominal wage gains finally caught back up. The cumulative index shows the average real wage modestly above its 2006 starting point — growth, but far short of what sustained productivity gains might have delivered.

Every number above depends on which wage and price measures are used. CPI-U tends to run a few tenths above the PCE deflator, so “real” growth deflated by CPI is the more pessimistic reading; alternative wage series (median usual weekly earnings, the employment cost index) date the squeeze differently. The dashed boundary dates of the squeeze move by a month or two under small revisions to either input series — treat its length as “about two years,” not exactly 25 months.

• Composition effects. Average hourly earnings is a ratio of total pay to total hours, so the 2020 layoffs — concentrated among lower-paid workers — mechanically raised the average without anyone getting a raise, and rehiring later reversed it. The pandemic-era spike and part of the 2021 “decline” are workforce-mix artifacts, not wage changes.
• Average ≠ median. The mean is pulled by high earners; median and distributional measures can tell a different story about the typical worker.
• One national price index. CPI-U is a national urban average; households with different consumption baskets — renters, drivers, regions with fast shelter inflation — experienced different real wage paths.

The calculation lives in apps/web/src/app/research/real-wages/page.tsx and uses lib/econ; every number on this page is recomputed from the committed snapshots on each build. Data series and vintages are listed on the Data page.