BLS acknowledges that there was an error in calculation, due to classification of furloughed workers as being "out of office" rather than unemployed. They estimate the real numbers are about 3 percent higher, and that this will likely show up in the adjusted numbers now and moving forward, just as March and April were adjusted downward.
It is wishful thinking to believe the unemployment rate is 13.3%. Let’s do some simple math. In February 2020 (the peak), 158.8m people were employed and 5.8m were unemployed, implying a workforce then of 164.5m and an unemployment rate of 3.6%. Fast forward a few months. Friday’s report showed 137.2m employed. The difference is 27.3m (164.5m − 137.2m). Divide the difference by the February workforce and we get 16.6% (27.3/164.5), given the reasonable assumption that the workforce has not dramatically changed in three months. Two other items. First, the unemployment rate is based on a sample of households (survey) which can by noisy in a time of stress. Second, while at the onset of the economic shutdown it had been useful to look at initial claims (which exceed 40m since the crisis began), by now some people might be filing twice. The continuing claims for the week ending May 23 are 21.5m. We have had new claims since then. In addition, the claims should be considered a lower bound on the number unemployed - but they are hard data not survey-based. The stock market reacted with exuberance to the surprise “13.3%.” I recommend caution. Over the next few weeks, I will detail the risks to the recovery in these posts.