The Hidden Rot in the Jobs Numbers
Hours worked are declining, resulting in the equivalent of a net loss of 100,000 jobs since September.
Edward P. Lazear
March 16, 2014 6:29 p.m. ET
Most commentators viewed the February jobs report released on March 7 as good news, indicating that the labor market is on a favorable growth path. A more careful reading shows that employment actually fell—as it has in four out of the past six months and in more than one-third of the months during the past two years.
Although it is often overlooked, a key statistic for understanding the labor market is the length of the average workweek. Small changes in the average workweek imply large changes in total hours worked. The average workweek in the U.S. has fallen to 34.2 hours in February from 34.5 hours in September 2013, according to the Bureau of Labor Statistics. That decline, coupled with mediocre job creation, implies that the total hours of employment have decreased over the period.
Job creation rose from an initial 113,000 in January (later revised to 129,000) to 175,000 in February. The January number frightened many, while the February number was cheered—even though it was below the prior 12-month average of 189,000.
The labor market's strength and economic activity are better measured by the number of total hours worked than by the number of people employed. An employer who replaces 100 40-hour-per-week workers with 120 20-hour-per-week workers is contracting, not expanding operations. The same is true at the national level.
The total hours worked per week is obtained by multiplying the reported average workweek hours by the number of workers employed. The decline in the average workweek for all employees on private nonfarm payrolls by 3/10ths of an hour—offset partially by the increase in the number of people working—means that real labor usage on net, taking into account hours worked, fell by the equivalent of 100,000 jobs since September.
Here's a fuller explanation. The job-equivalence number is computed simply by taking the total decline in hours and dividing by the average workweek. For example, if the average worker was employed for 34.4 hours and total hours worked declined by 344 hours, the 344 hours would be the equivalent of losing 10 workers' worth of labor. Thus, although the U.S. economy added about 900,000 jobs since September, the shortened workweek is equivalent to losing about one million jobs during this same period. The difference between the loss of the equivalent of one million jobs and the gain of 900,000 new jobs yields a net effect of the equivalent of 100,000 lost jobs.
The decline of 1/10th of an hour in the average workweek—say, to 34.2 from 34.3, as occurred between January and February—is like losing about 340,000 private nonfarm jobs, which is approximately 80% greater than the average monthly job gain during the past year. The reverse is also true. In months when the average workweek rises, the jobs numbers understate the amount of labor growth. That did occur earlier in the recovery, with a general upward trend in the average workweek between October 2009 and February 2012.
What accounts for the declining average workweek? In some instances—but not this one—a minor drop could be the result of a statistical fluke caused by rounding. Because the Bureau of Labor Statistics only reports hours to the nearest 1/10th, a small movement, say, to 34.449 hours from 34.450 hours, would be reported as a reduction in hours worked to 34.4 from 34.5, vastly overstating the loss in worked time. But the six-month decline in the workweek, to 34.2 from 34.5 hours, cannot be the consequence of a rounding error.
Was it the harsh winter in much of the United States? One problem with that explanation is that the numbers are already seasonally adjusted.
Imperfections in the adjustment method can result in weather effects, but the magnitude is far from clear, especially given that parts of the West, Midwest and South experienced milder-than-normal weather, with fewer business-reducing storms. Also, the shortening of the workweek began before the winter set in, with declines in hours from September to October.
Another possibility for the declining average workweek is the Affordable Care Act. That law induces businesses with fewer than 50 full-time employees—full-time defined as 30 hours per week—to keep the number of hours low to avoid having to provide health insurance. The jury is still out on this explanation, but research by Luis Garicano, Claire LeLarge and John Van Reenen (National Bureau of Economic Research, February 2013) has shown that laws that can be evaded by keeping firms small or hours low can have significant effects on employment.
The improvement in average weekly hours worked was reason for celebration after the recovery began. The recent decline is cause for concern. It gives us a more accurate but dismal picture of the past two quarters.
Mr. Lazear, who was chairman of the president's Council of Economic Advisers from 2006-09, is a professor at Stanford University's Graduate School of Business and a fellow at the Hoover Institution.