Little To Celebrate In The Latest Employment Report
Today, the Bureau of Labor Statistics (BLS) released the unemployment rate for September. At 7.9% it was lower than the 8.2% the market expected, and better than last month’s 8.4%. At first glance, this seems to be good news and it will surely be touted as such by the Administration. But the report is far less positive than it seems.
Part of the reason the unemployment rate fell is that the pool of people looking for work shrunk. This is reflected in a big fall in the participation rate, from 61.7% last month to 61.4% in September, lower than the expected 61.8% and the same as in March 1976.
This is a big setback, not just for the economy today but also for its future. It is hard to envision a fast return to a vibrant economy when the number of “discouraged workers” – those who would like to work but have given up looking – stands at 552,000.
The number of non-farm jobs was also disappointing compared to expectations. Economists had anticipated that 894,000 non-farm jobs would be created in September, but the number came instead at a paltry 661,000. The big miss was due to the way the numbers are reported, a feature that may persist and lead to further surprises in upcoming economic releases.
The 661,000 number is “seasonally adjusted”, which means that it was altered in order to smooth out the normal ebbs and flows that come up during the year. Without such adjustments the data is very hard to read, because it looks like this:
Various factors during the year make the peaks and troughs predictable. Ski resorts closed in the summer, summer camps are closed in the winter, school-related businesses get busy in the fall, and so on. A lot of work goes into adjusting for those factors to smooth the series, which then looks like this:
This smoother set is the number that is released and the one that economists try to predict. While it is not the “real” number, in some sense it is a better number because it makes it possible to look behind the seasonal patterns to better judge the strength of the labor force.
The BLS employs very talented statisticians to come up with the proper adjustments. Their models, for example, incorporate a normal spike in school-related employment. However, many schools did not open or opened with reduced attendance due to the pandemic. This unique condition led to an overestimation of the number of people that would have a job, and set the stage for a disappointment. Adjusting the data is not easy and the models are constantly adjusted.
We created an indicator to measure how well the BLS seasonal adjustment factors work in smoothing out the cycles. We found the largest deviation in thirty years with respect to the “real” data, which means that measuring how well the U.S. labor force may or may not be recovering has become more difficult.
This, in turn, points to the possibility that the U.S. economy is undergoing a fundamental change, if the pandemic causes a permanent realignment of labor that boosts it in some fields and depresses it in others. It will be a while until we come up with seasonal adjustments that reflects this changing world.
The number of people working from home may turn out to be persistent, for example, leading people to move out of large cities and impacting impact workers involved in urban transit systems, just to name one. Disrupted global supply chains could end up increasing the number of workers employed in domestic production and decreasing it in import-export or distribution businesses. A growth in online learning could mean fewer custodians, and so on. The pandemic can have profound – but not necessarily negative – consequences on the U.S. and global economy, perhaps much more than on public health in the long term.
For the time being, the employment picture seems to have become more difficult to predict, which may contribute to higher levels of market volatility.