The Bureau of Labor Statistics (BLS) reported that the jobless rate fell 1.4 percentage points to 13.3%. The May numbers bettered consensus expectations by the largest margin in history as analysts (including me) expected the jobless rate to rise to near 20%. The question on the mind of many analysts is how can the jobless rate decline when many millions more filed new jobless claims? Not sure there is a good explanation.
The household employment survey is normally conducted through a combination of in-person interviews and telephone interviews.
But for the safety of the Census workers, personal interviews were not conducted in May or April for that matter. As a result, the response rate to the household survey in May was exceptionally low at 67%, about 15 percentage points below the average rate prior to the pandemic.
The low response rate does raise questions about the accuracy of the numbers but does not mean the reported numbers are wrong. In the press release, BLS stated that they are investigating a “misclassification error” as a large number of workers temporarily unemployed because of the pandemic were wrongly classified as employed. According to BLS, the jobless rate would have been 3 percentage points higher if these workers would have been properly classified as unemployed. But even an error of 3 percentage points would have resulted in a jobless rate of 16%+, still far below consensus estimates of 19%.
Estimates of a much higher jobless rate were based on the jobless claims data. In mid-May, the four-week level of jobless claims was 5 million higher than the level in mid-April. Logically if millions more are filing new claims it means the jobless rate is rising, not falling, the exact opposite of what happened in May.
Jobless claims do reflect “hard data” as each new claim represents an individual stating that they are unemployed. However, not all jobless claims are legitimate, but one would think the lion share of 5 million in new claims is legit. It is worth noting that filing false claims can lead to serious penalties. The jobless rate, however, is based on the “honor system”, as people are asked to state their labor force status. There is no penalty for providing false information. And it would be impossible to prove anyway.
One possible explanation for the gap between the two data series is that people who work for businesses that received government support were told that they are back on the payroll (although not working) but still filed for jobless benefits. I have no proof that this took place. But it appears to be the only logical explanation. Perhaps in the coming days, there will be press reports of double-dipping; that is, people getting paid to not work and filing for jobless benefits at the same time.
The equity market reaction to the May data is not surprising, but seeing the equity market trade up so sharply even before the release of the data does raise questions over possible leaks. Here too I don't have any proof, but isn't it odd that the equity market was trading up so sharply before what everyone expected to be the highest jobless rate since the Great Depression.