Seasonal Adjustements Likely Overstated Labor Force Growth, at Least in May

In an earlier blog post of mine, I predicted that the May and June jobs report would overstate job performance. The logic was that employment will not fall as much as it usually does this time of year (when schools close) because schools were already virtual. Hence, I argued, seasonal factors would overcompensate for a now weakened effect, and overstate job growth. I predicted, in particular, that seasonal adjustments would overstate relative job growth for potential child-rearers (who are predominantly women) and educational workers.

I took a victory lap in a follow up blog post, It Turns Out I was Right #1, when the May Report came out. Now that the June Report has come out, I am partially retracting that victory lap.

The two figures that I have been pointing to so far are 1) the level of education employment and 2) the fraction of all employed people who are women. My assumption that the seasonal adjustments would overstate job growth in the two reports, specifically for child-rearers and education workers, leads one to conclude that both measures will increase.

I am still clearly correct on the first count. The plot below shows the seasonally adjusted level of employment in “Education Services” from June 2020 to June 2021. It shows a clear (and I’d say illusory) decline as schools opened from August to October 2020. It also shows a clear (and illusory) bounce from April 2021 to June 2021.

On the second count, the data does not fully fall in line with my projections. The chart below shows that while relative female employment did rise over the two month period I talked about in the original post, it fell from May 2021 to June 2021.

In my victory-lap blog post I noted that overall employment growth for the month was significantly higher than employment growth excluding female employees and employees in the education and health services industry. I pointed to this as evidence that the overall job growth numbers were overstated. This time, for the month, the selective employment figure grew by more then the overall figures, which is evidence that my prediction was false. However, for the two month period overall, the selective figure did grow more than the wider employment figure, which supports my claims.

I predicted that seasonal adjustments would overstate job growth in May and June. I have evidence to suggest I was right for May and the two month period overall. However, no strong evidence to say I was right for the month of June.