Keith Hall is a senior research fellow at the Mercatus Center at George Mason University, and former commissioner of the Bureau of Labor Statistics.
Two things were notable about Friday's job report. First, the unemployment rate made a surprising, statistically significant decline from 8.1 percent to 7.8 percent, and this drop wasn't a result of further decline in labor force participation. On its own, this is a signal of an improving labor market and, since this occurred just over a month from a presidential election, it got a great deal of attention—some of it bad as there was considerable but unfounded speculation that the numbers must have somehow been manipulated by the administration.
Second, most of the other data in the report was less positive. In particular, data from the much larger survey in the release indicated that were only 114,000 new payroll jobs created in September. This is more consistent with the continued sub-par economic performance we've been experiencing over the past six or seven months.
The monthly jobs report presents the results from two different surveys. One is a survey of businesses and uses payroll tax records—hence, it is often called the "payroll survey." The other is a survey of households conducted in person or by phone.
Data from the two different surveys often give slightly inconsistent views of the labor market, something that has been occurring for several months now, as seen in the accompanying chart below. Using the best single, real-time indicator from each survey—the payroll job growth data from the establishment survey and the employment ratio from the household survey—we can see this discrepancy.
We briefly started the year with good labor market news. We had two months of solid payroll job growth and the employment ratio (the share of the population that is employed) rose to its highest level in two and a half years. Then, both series weakened and other economic data helped confirm that economic growth had slowed. Though it has fluctuated some, payroll job growth has remained mostly tepid and we have now averaged 111,000 per month over the past seven months.
The household survey was slightly less consistent. In May, the employment ratio made a bit of a recovery, rising to 58.6 percent before falling back to just 58.3 percent, near its lowest level of the recession, by August. Then, we had Friday's release that showed a remarkable jump of 0.4 percentage point to its highest level in over two years. This was the largest one-month rise in the ratio in over 25 years. To put this in perspective, this last happened in April of 1984 when the unemployment rate fell from 7.7 percent to 7.4 percent, when we were averaging 350,000 new payroll jobs per month and economic growth was 7.2 percent.
So what is the bottom line here? First, the labor market is in terrible shape. Whether September's jobs numbers were mediocre or showed signs of improvement makes little difference for the millions of people without work. We need years of strong economic growth to even begin to get back to a healthy economy. Second, there is broad agreement that the payroll survey is a more reliable indicator of employment trends. When I look at a new jobs report, I first look at the number of payroll jobs created for the month. If the household survey is accurately signaling an improving labor market, then I would expect that additional economic data over the next month will confirm this.