Guide · Markets
Employment Cost Index (ECI): The Fed's Wage Gauge
The employment cost index measures what US employers pay in wages and benefits for a fixed mix of jobs, the cleanest gauge of labour-cost growth. Here is how it is built, how to read a release, and what the 0.8% rise for the third quarter of 2026 said about wages and inflation.
The employment cost index, or ECI, tracks how much it costs US employers to pay their workers: wages and salaries, plus benefits such as health insurance, retirement contributions, paid leave and legally required payments like Social Security. The Bureau of Labor Statistics publishes it once a quarter, about a month after the quarter ends. The figures for the three months to September 2026 came out on Friday, October 30, at 8:30 a.m. Eastern.
Its main advantage is what it holds still. The average hourly earnings figure in the monthly jobs report divides total pay by total hours for whoever happens to be employed, so it rises when low-paid workers lose their jobs even if nobody gets a raise; that is what happened in the spring of 2020. The ECI instead prices a fixed basket of jobs across occupations and industries, surveyed at the same employers each quarter, so a change in who is working does not move it. That is why Federal Reserve officials treat it as the most reliable read on underlying pay growth, even though it arrives only four times a year.
A release has more numbers than the headline. The headline is the three-month change in total compensation for civilian workers, seasonally adjusted. Next to it come wages and salaries, which make up roughly 70% of compensation, and benefits, each split between private industry and state and local government. The twelve-month changes are not seasonally adjusted, and the BLS also shows them in constant dollars, deflated by consumer prices, which tells you whether pay is keeping up with inflation. Analysts often single out private-sector wages excluding incentive-paid occupations, which strips out commissions and bonuses that swing with sales.
The September 2026 report is a good example of why the details matter. Total compensation rose 0.8% in the quarter, down from 0.9% in the three months to June; wages and salaries rose 0.7%, from 0.9%, and benefits 0.9%, from 1.0%. Over twelve months compensation was up 3.4%, wages 3.1% and benefits 3.9%, with private-sector health benefits up 6.4%. In constant dollars, though, compensation was 0.4% lower than a year earlier and wages 0.6% lower. Employers' costs were still rising, while workers' pay was falling behind prices.
That combination is what makes the ECI useful for reading inflation. A common rule of thumb is that pay can grow at roughly the inflation target plus productivity growth without pushing prices up, which with productivity of 1.5% to 2% puts the comfortable range around 3.5% to 4% a year. A quarterly rise of 0.8% compounds to about 3.2% annualised, inside that range. With core PCE inflation at 3.1% in September and the Fed having held rates at 3.75% to 4% two days earlier, the report suggested that the inflation of 2026 was being driven more by energy and goods than by a wage spiral, which matters for how long high rates need to last.
Because it comes out so rarely, the ECI can move Treasury yields when it surprises, mostly at the short end of the curve, which follows Fed expectations. On release day, compare the quarterly change with the forecast, then look at private wages rather than the total, and at benefits separately, since health-insurance costs can lift the index without putting more money in workers' pockets. One technical change is coming: starting with the December 2026 data, due on January 29, 2027, the BLS will update the index's employment weights and stop counting workers' compensation insurance costs, so quarter-to-quarter comparisons across that release deserve a footnote.