Employers added 225,00 jobs last month as the unemployment rate ticked up slightly to 3.6 percent, largely due to more people entering the job market, yet another sign that there’s still room-to-run in this long labor-market expansion. Wage growth, a perennial soft spot in recent jobs reports, ticked up slightly to a yearly rate of 3.1 percent, around where it has been for much of the past year. That’s ahead of inflation, last seen running at 2.3 percent, but the fact that the wages have not accelerated suggests some degree of slack remains in the job market (other wage and compensation series show roughly similar stability).
Our monthly smoother pulls out trends in job growth by averaging monthly gains over 3, 6, and 12 months. The pattern it shows is interesting and revealing. Over the past 12 months, job gains average 171,000 per month. Yet that average has accelerated over the past 3 months. Typically, as the job market closes in on full capacity, job gains tend to decelerate, much the way you have to pour more slowly as you reach the brim of a glass to avoid spillage (which, in this analogy, is inflation). Instead, we’re seeing no such deceleration, another sign of room-to-run.
In a similar vein, the closely watched employment rate for prime-age workers (25-54) continues to rise, and at 80.6 percent now stands above its 2007 peak of 80.3 percent. However, that’s more of function of job gains for women than for men. Prime-age men’s employment rate is still 1.4 percentage points short of its 2007 peak, while women have surpass their peak by almost 2 points. This partially reflects job gains is services versus recent job losses in manufacturing.
Factory employment fell again last month, down 12,000. Over the past 12 months, factory jobs are up just 26,000, one-tenth their gains over the prior 12 months (267,000). This clearly relates to Trump’s trade war, and while the recent “phase one” agreement with China may improve conditions in the sector–though I doubt it will have much impact–it will take time for trade flows to recover. Note also that blue-collar weekly earnings in the sector are up just 1.3 percent over the past year, a full point below inflation, meaning weekly paychecks for blue-collar factory workers are falling in real terms.
Today’s report includes the BLS’s annual benchmark revision to the payroll jobs data. In order to adjust the jobs data to more closely reflect a true census of the underlying jobs count, once a year the Bureau adjusts the level of jobs in the previous March up or down by factor based on more complete data. That factor this year was -514,000, a larger than average downward revision (the average revision, without regard to its sign, is 0.2% of payrolls; this one was 0.3%). The revision is “wedged” into the jobs data at a rate of -43,000 per month between April 2018 and March 2019. The negative revision for retail trade was particularly large, at -159,000, or 1 percent, likely a symptom of the accelerating loss of brick-and-mortar retail outlets at the hands of online competition.
The figure shows the difference between the level of payrolls before and after the revision. The new results do not change the fact that the historically long jobs recovery has been solid in terms of job quantity (job quality remains a significant problem). But the new trend is notably less robust than was previously recognized.
The wage-growth story remains much the same as it has been in recent months: stable gains but, despite the tight job market, no acceleration. The figures show annual, nominal wage gains for all and middle-wage private sector workers (the dark lines are 6-month trends). In both cases, we see clear evidence of slowing gains. Both series are beating inflation, so hourly wages are growing in real terms, but the pause in their upward trajectory is evidence that there’s still slack in the job market. Other wage series show similar, though less stark, stabilization in recent months.
Another critique of recent wage trends is that while they’re clearly being nudged up by the tight labor market, the trends are not as positive as you’d expect given the lowest unemployment rate in 50 years. One way to investigate this claim is to construct a statistical model, including labor market slack, to predict wage growth. If the predictions map closely onto the actual series, then perhaps wage growth is about where you’d expect, i.e., not too low, even given the tight job market.
The “full smpl” line in the figure below shows the results of such a model for mid-wage workers. The line cuts right through the actual trend in hourly wage growth, suggesting there’s no gap between expected and actual wage gains.
However, this isn’t quite the right way to do test this question. If the relationship between unemployment and wage gains has diminished over time, that change gets built into model estimates like this one. The way to account for that potential problem is to run the model through an earlier year and predict “out-of-sample.” The “smpl thru 2010” line shows the result from this approach. Sure enough, it predicts wage growth closer to 4 percent than the current growth rate of X percent. In other words, at least by this simple model, it’s not unreasonable to expect faster wage gains than we’re seeing.
See the data note below for details and caveats.
Summing up, labor demand remains admirably strong in the US job market, which shows few signs of age. And equally importantly, labor supply is responding to the demand, as the job market continues to pull people in. On the down side, the trade war has clearly damaged export-oriented sectors, especially manufacturing, both on the job and wage side. Moreover, even with unemployment persistently near a 50-year low, wage growth, at least in these data, has stopped climbing. This, along with low, steady inflation data, clearly implies there’s still slack left in the job market, with no rationale at all for the central bank to tap the brakes on growth.
Data note on wage model: The model’s dependent variable is year-over-year quarterly hourly wage growth for production, non-supervisory workers. Regressors include a constant, the unemployment rate minus the CBO estimate of the natural rate, two lags of the DV, and “expected trend wage growth” taken from a recent Goldman-Sachs analysis. They define this variable as follows: “Trend wage growth is estimated as the sum of the Fed’s measure of inflation expectations and a simple average of the backward-looking productivity growth trend and the Survey of Professional Forecasters’ estimate of productivity growth over the next 10 years.” The full sample goes for 1992q1 through 2019q4. The “out-of-sample” model runs through 2010.
Some analysts have correctly noted that unemployment doesn’t capture slack as well as the prime-age employment rate, especially when it comes to correlating with wage growth. If I substitute the prime-age employment rate into the model, the difference between the two predictions is negligible. My point here is simply that those who think wage growth should be faster at 3.5 percent unemployment are not necessarily wrong.