Employment costs
rose in Q3 2026
That is not the same as a pay raise.
BLS reports 0.8% quarterly employment-cost growth. Separate wages, benefits and purchasing power before using the figure in a budget.
The pay number and the employment-cost number are not the same
The latest Employment Cost Index presents a useful tension for American workers and employers: compensation continued to become more expensive, yet wage purchasing power did not keep up with consumer prices. In the Bureau of Labor Statistics release issued October 30, total civilian compensation rose 0.8 percent from June to September 2026 after seasonal adjustment. Wages and salaries rose 0.7 percent, while benefits rose 0.9 percent. The difference matters because an employer's rising cost is not necessarily money that an employee can spend on rent or groceries. A benefit contribution, an overtime premium and an increase in regular pay enter different parts of the compensation account.
The twelve-month figures make the distinction sharper. Civilian compensation increased 3.4 percent, wages and salaries 3.1 percent, and benefits 3.9 percent, all measured without seasonal adjustment. BLS reports that inflation-adjusted civilian wages declined 0.6 percent over the same year. These are national index changes, not a promise that every employee received a 3.1 percent raise or lost exactly the same purchasing power. They are best read as three linked questions: what did labor cost employers, which component changed, and what could wages buy after prices changed?
Why the index holds the job mix steady
Imagine a business reporting the same hourly pay rates for two occupations in consecutive quarters, but hiring more people into the higher-paid occupation. Its average hourly payroll could rise even if no job's pay rate changed. The ECI is designed to separate that compositional effect from changes in the cost of labor. BLS uses a fixed-weight approach across occupation and industry categories, rather than allowing every shift in the number of high-paid and low-paid jobs to drive the headline measure. The point is not that the labor market stops changing. It is that a stable comparison basket helps identify one particular kind of change.
That makes the ECI useful alongside, rather than interchangeable with, monthly average earnings or an employer's own payroll. Payroll can change because of staffing, hours, overtime, job mix, bonuses and rates. A fixed-composition labor-cost index asks a narrower question. An index reading of 3.4 percent is therefore not a national pay level, a median salary, or the percentage of workers receiving raises. The current index base is December 2005 equals 100. An index level above 100 expresses cumulative change from that base, not dollars per hour. Confusing the level with the growth rate would turn a comparison tool into a misleading wage claim.
Read a quarter as a quarter, and a year as a year
The 0.8 percent headline covers three months and is seasonally adjusted. The 3.4 percent annual figure covers twelve months and is not seasonally adjusted. They are not competing estimates of the same period. Seasonal adjustment removes recurring patterns so adjacent quarters can be compared more meaningfully. The annual comparison asks how September's cost relates to the preceding September. A reader checking whether compensation growth slowed should keep the period, adjustment and worker group constant instead of comparing whichever pair of numbers looks largest.
On that consistent basis, civilian compensation rose 0.9 percent in the June quarter and 0.8 percent in the September quarter. Civilian wage growth moved from 0.9 to 0.7 percent, and benefit growth from 1.0 to 0.9 percent. The twelve-month compensation increase was 3.4 percent in both June and September. Thus a modestly lower quarterly increase can coexist with an unchanged annual rate. It is also different from a fall in the cost level: positive growth means compensation costs continued to rise. Multiplying a single quarter by four would be a hypothetical pace, not the actual annual result that BLS reported.
What counts as a benefit may surprise a payroll reader
The health-benefit figure is a cost signal, not a coverage verdict
Private-industry health-benefit costs increased 6.4 percent over the year through September, compared with 6.0 percent in June and 6.1 percent a year earlier, according to the release's summary table. That is a separate series from the 3.9 percent increase in all private-industry benefit costs. Health insurance is one part of the broader package, so its growth rate should not be substituted for the entire benefits measure. Nor does the number describe the increase in every worker's premium contribution, deductible or out-of-pocket medical bill.
An employer can face higher health-plan costs without employees experiencing an equivalent improvement in access or generosity. Conversely, a worker's household medical expenses can change for reasons that do not appear as the same movement in employer costs. To assess a particular job offer or renewal, the relevant documents are the actual plan terms and the employer and employee contributions. The national index gives context for a budget conversation; it does not replace the contract, benefit summary or enrollment information. The October release does not isolate how much of the increase came from prices, utilization, plan design or the characteristics of covered workers.
Private employment and state-local employment move differently
Private-industry compensation rose 0.8 percent in the September quarter, with wages also up 0.8 percent and benefits up 0.9 percent. State and local government compensation rose 0.6 percent, with wages up 0.6 percent and benefits up 0.7 percent. Those separate results matter to a school district, municipal agency or private employer comparing its budget with the headline civilian series. The combined civilian result is not a description of every ownership group, and the private-sector wage figure is not identical to the civilian wage figure.
Over twelve months, state and local government compensation increased 3.5 percent, wages 3.2 percent and benefits 3.8 percent. Private-industry compensation increased 3.4 percent, wages 3.1 percent and benefits 3.9 percent. The differences are relatively small in the published rounded figures, and they do not rank the adequacy or fairness of individual compensation packages. Public budgets, occupations, contract calendars and benefit structures differ. A local government should not treat a national private-industry index as a direct estimate of its own staffing bill without examining which series fits the question and which expenses its budget includes.
Who is outside the civilian measure
The word 'civilian' in this survey does not mean every person working outside the armed forces. It combines private-industry and state-local government workers within the survey's coverage. Federal and quasi-federal workers and military personnel are excluded. The agricultural sector, private-household workers and the self-employed are also excluded from the private-industry coverage. So are people who set their own pay, such as proprietors and certain owners or partners. Volunteers, unpaid workers and people working overseas are outside the civilian measure described by BLS.
These boundaries are essential when using photographs, examples or comparisons. A farmer's income, a federal employee's pay adjustment and a small proprietor's profits cannot be inferred from the civilian ECI. A picture of a workplace illustrates an activity, not that the photographed establishment participated in the confidential survey or experienced the reported increase. In the same way, the presence of a job in an industry does not tell readers how a particular worker is classified. The ECI is a population estimate built from a survey design, not a public roster of employers with their individual pay decisions attached.
Real wages: why a positive raise can still buy less
Current-dollar wage growth and constant-dollar wage growth answer different questions. The first measures nominal labor costs. The second adjusts for consumer-price changes to describe purchasing power. BLS reports a 0.6 percent annual decline in constant-dollar private-industry wages and a 0.5 percent decline in state-local government wages. Civilian wages also declined 0.6 percent in constant dollars. A positive nominal wage increase therefore does not by itself establish that workers gained ground against inflation over this period.
The real-wage figures are not a personalized household inflation calculation. Rent, transportation, health expenses and other purchases carry different weights in different families' budgets, and work hours and taxes also affect spendable income. A reader can use the national result to frame the question, then compare their own pay and recurring expenses over the same dates. What the index does establish is that nominal growth and real growth must not be treated as synonyms. It does not establish that every household's standard of living moved by exactly the same percentage, or that benefits can be converted into immediately available cash.
Union comparisons show a difference, not a controlled experiment
The sample is large; the estimates still have uncertainty
For this reference period, BLS describes approximately 28,000 occupational observations selected from about 6,600 private-industry establishments, and roughly 7,300 observations from about 1,400 state-local government establishments that provided data at the initial interview. Those numbers describe the survey's observation structure; they are not counts of all workers in the economy. Sampling makes a national measure feasible, but estimates from one sample can differ from what a complete population measurement would show.
BLS publishes standard errors for relevant unadjusted series to help users judge precision. Its reliability guidance explicitly excludes seasonally adjusted and constant-dollar series from that standard-error publication. A tenth of a percentage point in rounded headline figures should therefore not automatically become a dramatic claim of an economic turning point. Sampling errors are also distinct from nonresponse, incorrect reporting or processing problems, which standard errors do not fully measure. The appropriate response is not to dismiss the survey, but to match the strength of the conclusion to the available evidence and avoid precision that the release does not support.
January brings a measurement change, not the end of a benefit
The next ECI release is scheduled for January 29, 2027, covering December 2026. BLS says that release will introduce fixed employment weights based on 2025 Occupational Employment and Wage Statistics data, replacing the 2021 weights used from December 2022 through September 2026. The employment counts are benchmarked to industry totals from the Quarterly Census of Employment and Wages. Updating the weights changes the reference composition used in the index so it reflects a more recent distribution of jobs; it does not mean that the September release was a simple average of current payrolls.
BLS will also remove workers' compensation costs from the ECI calculations and the National Compensation Survey collection. The agency cites a substantially lower item response rate than for other benefits and says the component represents about 1 percent of total compensation. This is a statistical collection and index change. It is not an announcement abolishing workers' compensation insurance, eliminating an employer's legal obligations or canceling an employee's claim. Readers comparing releases across the transition should read the updated methodological notes rather than treating every difference as a new change in employer behavior.
A practical way to use the release without asking it to do too much
For an employer building a labor budget, begin with the correct ownership and compensation component, then separate rate changes from headcount and hours. For an employee reviewing an offer, distinguish base pay, premiums, bonuses, employer benefit contributions and the costs the employee must pay. For a reader assessing the economy, compare quarterly figures on the same adjusted basis and annual figures over the same twelve months. These uses are related, but no single headline number answers all three. The ECI is strongest when the question concerns changes in the price of labor while holding broad employment composition steady.
If the question instead concerns the dollar cost per hour of wages or benefits, BLS publishes Employer Costs for Employee Compensation. That companion measure uses current employment weights, unlike the fixed weights of the ECI, so year-to-year movements in its cost levels need not match ECI growth. Choosing the correct measure is more useful than forcing them to agree. The September report's central message is consequently precise: labor costs kept rising, benefits outpaced wages in the civilian annual comparison, and wage purchasing power fell on the published constant-dollar measure. The next release will add both a new quarter and a methodological transition that deserves to be read on its own terms.
For the household-spending side of the distinction, see our analysis of September PCE and purchasing power. ECI tracks employers’ labor costs; the PCE accounts answer a different income-and-spending question.
Comments
Post a Comment