Friday's "Goldilocks" unemployment report needs to be taken with a grain of salt.
Unemployment produces a mixed bag of mostly lagging indicators of various predictive qualities. Coincident and lagging jobs data is inconclusive.
In the past, the unemployment rate was consistently lowest just months before recessions.
The state of the economy may be much less rosy, especially taken together with the direction of other key indicators.
On Friday, the Bureau of Labor Statistics released the much watched employment data for September. The unemployment rate dropped to 3.5% - the lowest since 1969. However, non-farm payrolls rose by lower than the expected 135,000 jobs.
At first, this seems a bit counter-intuitive. How can we add fewer jobs, but get a lower unemployment rate? And why did economists think payrolls would improve, but the unemployment rate wouldn't? How can the low unemployment rate on the one hand tell us we are in an economy that's on fire, but on the other hand, we're creating far fewer jobs than expected?
To find the answers, it is important to look much closer on how this report is produced. The unemployment data actually consists of two different surveys. The household and the establishment survey. Both surveys use entirely different sources of data and have different sample sizes, with different margins of errors, assumptions and inputs.
The Household Survey
Let's take a look at this month's household survey to understand moves in the unemployment rate. Since our population is constantly growing, job growth has to make up for this population growth to keep employment steady. Anything less and the unemployment rate rises and vice versa. The labor force is projected to grow 0.5% annually due to population growth. The Bureau of Labor Statistics works with the U.S. Census Bureau to estimate changes in the U.S. population. The data is based on the Current Population Survey. Adjustments to the population numbers are made accordingly. On top of that, all numbers are seasonally adjusted. In September, 117,000 new workers joined labor force, according to the household survey. That many new workers in September is a relatively small amount. In August, five times as many workers were added to the labor force.
Now various dynamics are at work. Most importantly, workers added to the labor force need jobs. If we have a month where the net addition of jobs exceed the increase in the labor force, the unemployment rate drops and vice versa. This happened in September - which was not hard to achieve, given that the labor force grew by only 117,000 workers. In August, 571,000 workers joined the labor force with about the same number of net job adds, holding the rate constant. Now imagine next month we had the same number of new workers, but this month's low net increase in jobs of 391,000. The unemployment rate would shoot back up - a very plausible scenario. A month of data is not very meaningful, given that last month we added five times as many jobs and crated significantly more jobs - holding the unemployment rate constant. This can reverse very quickly.
It is important to note that the household survey includes a much larger set of workers, but a smaller sample size with a very large margin of error. Compared to the establishment survey, it includes agricultural, self-employed, unpaid workers, etc.
The Establishment Survey
Part two of the employment report, the establishment survey, paints a slightly different picture. Economists forecast a fairly rose future at a monthly increase of 145,000 total non-farm payrolls. However, looking at the year-over-year percentage change of the payroll data, one can easily see that the labor market turns on a dime. Below are the last few economic cycles. In each case, it took only a few months from strong employment to recession (recessions in gray).
Payroll data typically declines a bit ahead of a recession compared to the unemployment rate, as can be seen from the charts above. Because of that, payroll data is considered a coincident economic indicator. If the payroll data continues to disappoint, a recession is not far out.
The Unemployment rate and the 1960s
Now that the unemployment rate has hit 3.5%, we hear comparisons to the '60s. That should be a caution.
Unemployment rate during the 1960s
In August 1968, the unemployment rate fell to 3.5%. Three months later the Dow started dropping from 985 to 683.
The unemployment rate as an indicator
Historically, it quickly becomes clear that the unemployment rate is a lagging indicator. It traditionally is at its lowest often just months before a recession. By this time, the stock market has long corrected based on other economic data and declining earnings. The chart below demonstrates the lagging nature of the unemployment rate (recessions in gray).
In some cases, unemployment data can provide some leading indication of a recession. The initial claims of unemployment often trend up earlier, but this is not consistent and can have many false alarms. Once these claims rise enough to provide a clear picture, the economy is already in a recession and the market has long corrected. We now bottomed out for a year, and it would be hard to believe we can maintain this much longer.
Initial claims of unemployment
Other data, such as the labor force participation rate, are not very useful in making economic predictions. They have significant month-to-month fluctuations and show no clear trends.
Average hourly earnings data may provide some insight; however, not enough historical data is available. The recent downtrend is certainly noticeable.
The unemployment rate dropped to 3.5% mainly due a smaller increase in the labor force relative to newly created jobs. In fact the household survey shows fewer new jobs than last month. Disappointing payroll data is concerning, given its more coincident nature as an indicator.
When unemployment data turns sour, the economy is already in deep trouble. As I've shown above, the onset of a recession is almost always less than a year away once unemployment indicators start to reverse. If we do start to see significant trends, we can be assured that we are extremely close to a recession since these reversals are typically sharp. By then, the market has most likely already started its descent since there are plenty of leading indicators that will have made that very clear before we get there. This past week the manufacturing and services PMIs have given us a taste.
Disclosure: I am/we are long PSQ. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.