The Productivity Flu
Although this week brought what looked like bad news on productivity, the underlying story was much milder than the initial diagnosis suggested.
The government said that nonfarm productivity, which measures hourly output of workers, increased at a 1.8 percent annual rate last quarter instead of the 2.8 percent pace initially estimated. That was slightly worse than the downgrade to two percent economists had forecast.
Unit labor costs—how much it costs employers to produce one unit of production—were revised up to 4.4 percent from 2.8 percent, causing some folks to worry that a wage-push inflation spiral might get started.
But the revisions were not nearly as bad as they seemed because they were not really about workers suddenly becoming less productive or more costly.
The revisions were mostly the delayed echo of an earlier decision by government statisticians to mark down how much the economy produced in the fourth quarter. The Bureau of Economic Analysis (BEA) cut fourth-quarter real economic output growth to 0.7 percent from 1.4 percent. When you lower output without lowering the number of hours worked, productivity falls.
The logic is simple. Productivity is output per hour. If the government decides the economy produced less output than it first thought, while leaving labor input largely intact, productivity falls by definition. That is exactly what happened. The Bureau of Labor Statistics (BLS) said nonfarm business output growth in the fourth quarter was revised down to 1.5 percent from 2.6 percent, while hours were unrevised in the revision calculation. Once productivity is marked down, unit labor costs go up mechanically unless compensation falls enough to offset it. It did not. So what looked like a fresh labor-cost problem was, to a considerable degree, the arithmetic consequence of weaker measured output.
The real detective story is not the productivity release. It is the GDP revision underneath it. Why did output get marked down so sharply?
A Healthy Dose of Bad Data
The answer turns out to be more interesting than the usual macro gloom. BEA said the biggest downward revision on the consumer side came from services, especially health care. More specifically, it pointed to hospital and nursing home services as well as outpatient services. As well, there was some downward revision in the estimates for factory construction and software. In other words, the economy did not suddenly spring a leak everywhere at once. The markdown was concentrated in a few places, and one of the biggest was health care.
So why did health care output fall at the end of last year? It’s not because health care got worse or people were unhealthy. Quite the opposite. Demand for health care was lower than initially estimated. The Census Bureau’s Quarterly Services Survey shows health care and social assistance revenue rose just 0.5 percent in the fourth quarter after a 3.0 percent gain in the third quarter. BEA specifically cited new QSS data as the reason it revised down health-care services.
Demand was down because we were healthier than expected. Data from the Centers for Disease Control show severe respiratory illness hospitalizations were low late in the fourth quarter, consistent with a lighter illness period than one would expect in a normal winter ramp-up. Fewer sick people means fewer hospital visits, fewer outpatient visits, and less measured health-care output. So a mild holiday season for the flu and COVID turns out to be bad for GDP accounting, but it is not bad news for actual Americans.
What makes this especially interesting is that health-care employment did not collapse. Quite the opposite. Even on a not-seasonally-adjusted basis, health-care payrolls kept rising through the quarter. That is exactly what you would expect in a sector like health care. Hospitals and clinics do not instantly shed workers because one quarter brings a milder respiratory season. Workers are still on the payroll. They are still showing up. But if fewer people need treatment, then measured output per worker falls. That is a ready-made recipe for lower productivity and higher unit labor costs, even without any deterioration in the underlying labor market.
Put it all together and the story looks less like a broad macro breakdown than a statistical chain reaction. First, BEA revised output lower, with an important contribution from weaker-than-assumed health-care utilization. Then BLS took that weaker output and ran it through the productivity formula. The result was a lower productivity number and a higher labor-cost number that look scary in isolation but are much less alarming once you understand where they came from.
A chunk of the bad macro news was not a sign that American workers suddenly forgot how to produce. It was a sign that fourth-quarter output, especially in health care, had been overstated on the first pass. Sometimes the economy looks weaker because people are sicker. This time, part of the economy looked weaker because people were not as sick.
The irony is too good to ignore: one of the big drags on measured output may have been that the country was healthier than the statisticians initially assumed.
COMMENTS
Please let us know if you're having issues with commenting.