July Jobs Gain Masks Distortions; Half of Payroll Surge Seen as Temporary

The July jobs report arrived like a reprieve. After months of economic freefall, the headline numbers looked encouraging: 1.76 million new nonfarm payrolls, unemployment ticking down to 10.2 percent. Markets took the news as evidence that the labor market had found its footing again, that the worst might be behind us. But beneath those reassuring figures, economists were already spotting the cracks.

At least half of those new jobs appeared to be built on sand. The largest gains came from leisure and hospitality—592,000 positions added in a single month. Of those, 502,000 were in restaurants and bars, businesses that had shuttered when states locked down and were now reopening as restrictions eased. The timing mattered. The survey period captured only the first half of July, before many states reversed course and shut down indoor dining and bars again as COVID cases spiked. Those jobs, in other words, were likely already disappearing by the time the data was published.

Then there was the education anomaly. Local government education added 215,000 jobs in July; state government education added another 30,000. Normally, public-sector education employment falls in July before seasonal adjustments are made. But the pandemic had scrambled the calendar. Schools had closed earlier than usual in 2020, which meant the typical July decline happened ahead of schedule. When July arrived, the seasonal adjustment made it look like a surge. The Bureau of Labor Statistics acknowledged the distortion in its own release. Diane Swonk, chief economist at Grant Thornton, told CNBC that those gains would reverse course as school districts moved online for the fall. "That will suppress those numbers as we move into August," she said.

Danny Blanchflower, an economics professor at Dartmouth, calculated that without these classification errors and seasonal distortions, the unemployment rate would have been a full percentage point higher than reported. The true picture was grimmer than the headline suggested.

Look deeper still, and the damage became clearer. The number of people unemployed for more than 15 weeks jumped to 8 million from 3.3 million in June. As a share of all unemployed workers, the long-term jobless rose to 48.8 percent from 18.7 percent. This was not a temporary dislocation. This was structural damage—workers locked out of the labor market for months, their skills atrophying, their prospects dimming. Permanent job losses, by contrast, remained relatively flat, suggesting that the worst of the layoffs had already occurred.

The broader recovery narrative was losing coherence. Howard Marks, the investment strategist at Oaktree Capital, had defined a true V-shaped recovery as one where both sides of the V are symmetrical—where the economy recovers at the same pace it declined. The numbers did not support that story. In two months, the economy had shed 22 million jobs. In three months, it had recovered 42 percent of them. The math was lopsided.

Economists acknowledged they were operating without a map. Blanchflower told Bloomberg that the month-to-month volatility made it impossible to draw firm conclusions. "I have nothing in the past to refer to to help me with this," he said. The August report would matter enormously, but even that would only tell part of the story. The real question—whether a second wave of the virus would trigger a second wave of layoffs—remained unanswerable. For now, the labor market was a month-to-month proposition, and the ground beneath it was still shifting.