Measuring wage growth: the Employment Cost Index

“Wages rose 4 percent last year” can mean several different things depending on which series is being quoted. The Employment Cost Index is the one designed to answer the question most people mean: how much more does the same job pay than it did a year ago?

What the ECI is

The Employment Cost Index is a quarterly measure, produced by BLS from the National Compensation Survey, of the change in employers’ cost of labor per hour worked. It is published for wages and salaries alone and for total compensation, which adds benefits such as health insurance, retirement contributions, paid leave and legally required insurance. It covers civilian workers, with separate indexes for private industry and for state and local government, and detail by occupational group, industry, region, and union status.

Its defining feature is that it holds the composition of employment fixed. The index weights each occupation-by-industry cell by a fixed set of employment weights, so a shift of workers from low-paid to high-paid jobs does not move it. What moves it is a change in what employers pay for a given job.

Why composition matters

Average hourly earnings, from the monthly payroll survey, divide total payroll by total hours. When low-wage workers lose jobs in a recession, average earnings jump although nobody got a raise; when they return, average earnings fall. The ECI is immune to that. Median weekly earnings from the Current Population Survey have the same composition problem in a different form. For the question “did pay for a job rise,” the ECI is the right instrument; for “did the typical worker’s paycheck rise,” the others are.

OEWS is not a wage growth series at all. BLS says it is not designed for comparisons over time, because its sample is pooled over three years and its classifications change; see how often OEWS is updated. A difference between two May releases for an occupation is not a growth rate.

How it is used

The ECI for private-industry wages and salaries is written into law as the default annual adjustment for military basic pay (37 U.S.C. 1009) and feeds the formula for General Schedule base pay adjustments (5 U.S.C. 5303), though the President and Congress have often set a different figure; see how federal GS pay works and how military pay is structured. The Federal Reserve and forecasters watch it as the cleanest measure of labor cost pressure. OEWS itself uses it to age earlier panels’ wages forward to the reference month.

Nominal and real

The ECI is nominal: it measures dollars, not purchasing power. Real wage growth is the ECI’s change less the change in a price index over the same period, conventionally the Consumer Price Index for All Urban Consumers (CPI-U). A 4 percent ECI increase with 3 percent CPI inflation is roughly a 1 percent real gain. Which price index is used changes the answer; the personal consumption expenditures index usually rises more slowly than the CPI-U, so real growth measured against it looks higher. Any real wage figure should name its deflator, and the trends pages planned for this site will.

Reading a release

BLS publishes the ECI about a month after each quarter ends. The headline numbers are the three-month change, seasonally adjusted, and the twelve-month change, not seasonally adjusted. Levels of the index (with a base of 100 in December 2005) are published too but are rarely quoted; the index is built for changes. The wages-and-salaries series moves more smoothly than total compensation, because benefit costs are lumpier.

Limits

The ECI does not say what any job pays in dollars; it says how fast that pay changed. For levels, the occupation and state tables here are the source. It does not cover federal government workers, the self-employed, agricultural workers or private households. And a fixed-weight index can diverge from workers’ actual experience when workers change jobs to get raises, which the index, by design, does not see.