Posted on March 10, 2021
Despite some recent modest revisions to data from the Bureau of Economic Analysis (BEA), we are still getting different signals on the pace of recovery in economywide wages and salaries from BEA, the Bureau of Labor Statistics (BLS), and the Treasury Department’s tax withholding data. On Friday of last week, BLS released the data from the February establishment report, and while the report showed an increase in jobs compared to January (and on a year-over-year basis, a smaller decline, so also an improvement), the recovery in economywide wages and salaries stopped for the first time since the worst of the recession in April (see chart below); indeed, private-sector wages and salaries as measured by BLS actually declined in February (both on a month-to-month and year-over-year basis) because of a decline in average hours worked per employee (see chart below). We don’t have wage and salary data yet for February from the National Income and Product Accounts (NIPAs) as reported by BEA, but that NIPA measure of wage and salary growth has been running above the BLS measure throughout the recovery. And tax withholding jumped in January and February well above the recent growth rate of wages from both BLS and BEA by our measure of withholding, which removes the estimated effects of tax law changes and standardizes the number and makeup of business days across months.
The wage and salary data from the establishment employment report for February seem a bit perplexing. Jobs in the private sector–the broadest measure available from the establishment survey–rose but average weekly hours worked per employee fell by a bigger percentage, both on a month-to-month and year-over-year basis. Average hourly earnings remained flat compared to the amount from February of a year ago, so they didn’t figure into the movement in total wages and salaries. That all means that total wages and salaries fell in February, the first such decline since the bottom of the cycle in April. Specifically, total private sector wages as measured in the establishment report were 0.3 percent above year-ago amounts in January, but were 0.8 percent below year-ago amounts in February. (The number of jobs times the average number of hours worked per week times the average wage per hour yields the total amount of wages and salaries earned in a week.) It appears that more people were hired in February, perhaps many part time, and others already working had their hours cut. It’s an unusual combination, given that if that occurred in any one firm, it could instead just keep the existing workers working the same number of hours and bring on (or bring back) fewer new workers. There could certainly be some mix effects going on across industries that help explain the movement, such as the new hires being in positions with a below average number of hours worked per week, but that alone would still increase total wages absent other workers having their hours cut.
In addition, economywide wages and salaries through January as reported by BEA in the NIPAs are running at a faster clip than BLS reports for the private sector. We discussed this in a previous post, and BEA revisions since that post have narrowed the gap just a bit.
Furthermore, tax withholding in February was a strong 4.7 percent above the amount from February 2020, as we measure it to exclude the estimated effects of tax legislation affecting withholding and to standardize the number and makeup of business days across months. That is quite a different signal than a year-over-year decline in the BLS data for wages and salaries in February. That movement in withholding could be indicating a strengthening economy or could be caused by measurement issues. Although withholding is most importantly determined by the total amount of wages in the economy, movements in withholding do not always line up with movements in total wages as measured by BEA and BLS. The income tax component of withholding (as opposed to the payroll tax component) tends to be focused on middle- and higher-income taxpayers, and the nature of the recession and recovery appears to be causing those workers to be doing much better than lower-income, and hence lower-taxed, workers. Also, the fact that we didn’t observe withholding growth exceeding BLS- or BEA-measured wage growth through the course of 2020, but now we do, could suggest that our adjustments to remove the revenue-reducing effects of legislation were too small; the expiration of most such provisions at the end of 2020 has removed the importance of that component of our withholding measure and now we could be getting a cleaner measure of withholding growth. Perhaps withholding growth on a constant law basis was running ahead of BLS- and even BEA-measured wage growth all along. Unfortunately, it isn’t until August and early September that we get the best first quarter data on economywide wages, from the Quarterly Census of Employment and Wages, which then get incorporated into the NIPAs.