Picture a break room in March, an illustrative example: a supervisor announces a raise, the number on the hourly line is higher than last March, and the room still argues about whether anyone is ahead. The argument cannot be settled from the hourly line alone. A nominal raise is a change in money paid. A real raise is that change after a price index has been applied. If prices rise by more than the wage, the paycheck number can go up while the basket it buys goes down. The fraction has two sides, and ignoring the price side is how a raise gets described as a gain when it functioned as a cut.

What BLS means by real earnings

Each month, alongside the consumer price indexes, BLS publishes real earnings from the Current Employment Statistics survey. The technical note is explicit about the deflator. Average hourly and weekly earnings for all private-sector employees are deflated with the CPI-U. Earnings for production and nonsupervisory employees are deflated with the CPI-W. Using one index on the other series is not a rounding choice. It is a different statistic.

The nominal piece is also narrower than the phrase "what workers make." CES average hourly earnings are total payroll divided by total hours for which employees were paid, including overtime and paid leave. They are an arithmetic mean across jobs in the private nonfarm sector, not a median, and not a typical worker's biography. Because each industry enters the average in proportion to its current employment or hours, a shift of jobs toward higher-paid or lower-paid industries moves the average even when no individual's rate changes. BLS states that directly. A booming month for a high-wage industry can lift "average hourly earnings" without a broad raise.

Those earnings also leave pieces out. They are not wage rates for a specific job. They exclude benefits, irregular bonuses and commissions, retroactive payments, and the employer's share of payroll taxes. Anyone who wants a compensation measure that holds the mix of occupations still and includes benefits is looking at a different BLS product, the Employment Cost Index. The ECI measures the change in the hourly labor cost to employers using a fixed basket of labor, so workers moving between occupations and industries do not masquerade as a change in the price of labor. It includes wages and salaries and the cost of benefits. It answers an employer-cost question. Real average hourly earnings answer a cash-earnings-versus-CPI question. Both are useful. They are not interchangeable.

A published episode, with the agency's own figures

BLS has already done the two-sided comparison for one widely discussed stretch, and the figures below are theirs, not a reconstruction. In Beyond the Numbers, the article "Did the COVID-19 pandemic affect real earnings?" reports that total private average hourly earnings rose from $28.55 in February 2020 to $34.56 in February 2024, a gain of 21.1 percent. Over those same four years the CPI-U rose 20.0 percent. Real hourly earnings therefore rose 0.9 percent, which the article describes as an annualized gain of about 0.2 percent. Nominal pay accelerated. Prices accelerated by nearly as much. The purchasing power of an average private-sector hour barely moved.

That episode is a lesson in the fraction, not a template for every later year. The article also notes that the CES average was affected by who held jobs during the pandemic, which is the mix problem in the technical note showing up in a real release. If employment shifts toward higher-paid jobs, the average can rise even while many continuing workers see something milder. Reading the 0.9 percent real gain as "every worker got ahead" overclaims a mean. Reading the 21.1 percent nominal gain as "workers got a large real raise" ignores the 20.0 percent CPI-U rise that BLS put beside it.

When the price side uses the wrong index

Your own raise is not the CES average, and the right deflator depends on the question. For a rough national comparison of cash pay to consumer prices, CPI-U all items is the series BLS itself uses on the all-employees earnings numbers. For a Social Security benefit, the law uses the CPI-W on a third-quarter to third-quarter basis, as the Social Security Administration describes on its COLA page. For a conversation about Federal Reserve policy, the yardstick in the Committee's strategy statement is the PCE price index, not the CPI. Those three sentences can be true in the same month and still produce three different "did pay beat prices" answers.

Component indexes matter when the household's basket is lopsided. A worker who spends an unusually large share of income on rent is not well described by an all-items deflator in a year when shelter and goods prices diverge. BLS says the CPI reflects an average household, and that a household with a heavier medical share can experience a faster personal rate of price change when medical prices lead. The grocery version of that point is why a grocery bill is not the official inflation rate. The shelter version is house prices, rent, and the same number trap. Using all-items CPI-U is honest when you label it as all-items CPI-U. It is less honest when you label it as "my inflation."

How to place one raise on both sides of the fraction

Stay with your own pay stub and a published index. Do not borrow the national average and pretend it is your raise.

  1. Pick a span with the same calendar footprint on both sides. A raise dated in July against a price change dated in December is an accounting error before it is an economic fact.
  2. Use the same pay concept at both ends: hourly rate, or weekly pay, or annual salary. Mixing an hourly rate with a weekly paycheck hides a change in hours. CES itself publishes hourly and weekly series because hours are a separate margin.
  3. Choose the deflator and write its name down. For a national cash-pay comparison, CPI-U all items, not seasonally adjusted, matches the spirit of the BLS inflation calculator. Say so in the sentence.
  4. Convert the older pay into later dollars with the ratio of index levels, or convert the newer pay into earlier dollars. One direction is enough. The steps are the ones in purchasing power versus nominal dollars.
  5. Subtract. If later pay is above the inflated earlier pay, the real wage rose over that span. If it is below, the nominal raise did not clear the price increase you chose.
  6. Only then talk about taxes and benefits, which this calculation has not touched. Income taxes are outside the CPI. Employer health contributions are outside average hourly earnings and inside the ECI. A real-earnings win on cash pay can coexist with a worse benefits package, and the reverse.

Hours deserve their own line. A higher hourly rate with fewer scheduled hours can lower weekly pay. A lower hourly rate with more overtime can raise it. Real average hourly earnings and real average weekly earnings are different BLS columns for that reason. A household argument that quotes one of them as if it were the other is a units error.

What "outruns" does not promise

Beating a price index over a year does not mean the household is comfortable, and failing to beat it does not mean the employer acted with a motive you can see from the index. The index has no information about your rent renewal, your childcare bill, or a change in household size. It also has no information about whether the raise was a promotion into a different job. CES average hourly earnings rise when the mix of jobs shifts. Your own history rises when you change jobs. Those are real events. They are not "wage growth" in the fixed-basket sense the ECI is built for.

You might be thinking that benefits make the cash comparison obsolete. They make it incomplete, which is a reason to add the ECI or your own benefits statement as a second page, not a reason to skip the cash page. Cash is what pays the consumption basket the CPI is describing. Benefits are a different price.

The longer habit of lining up whole decades, with taxes and hours included on purpose, is comparing salaries across decades fairly. If a tool offers to do the price side for you, check which index it baked in before you accept the real-wage sentence. That check is how inflation calculators actually work. The words "real exchange rate" and "purchasing power" in the glossary are the cross-border cousins of this same fraction: a nominal number on top, a price comparison underneath.

Read this as education

Real earnings are a published comparison between a pay series and a price series. They are not a finding about your employer, your city, or a fair wage in a legal sense. Releases are revised on the employment side more readily than the CPI-U is, and the latest technical note is the authority for which deflator BLS is using this month. Nothing on this page is a negotiation script or a forecast of next year's raise.

On your last raise, did the pay concept and the price index cover the same months, or only the same story?