Recession Probability Increases—but to just 22%--Despite Strong Employment Report
This morning’s Employment Situation report was quite a bit stronger than expected, with nonfarm payrolls expanding by 223 thousand and the unemployment rate ticking down to 3.5%.
Among the industries reporting the strongest employment growth were Leisure & Hospitality (adding 67 thousand jobs), Health Care & Social Assistance (74 thousand(, and Construction (28 thousand). State Government Education showed a decline of 24 thousand jobs.
Employment growth is perhaps the single most important macroeconomic force guarding against recession, so it was definitely surprising when my recession forecasting model showed an increase in probability of a recession sometime in the next 12 months.
My model still doesn’t predict a recession: the probability sits as just 22% following this morning’s employment report, meaning that the available data still aren’t convincing. The estimated recession probability increased from just 0.2% in April, peaking at 53% in September before dropping back down to 8% in November.
Still, three factors cause me to be pessimistic about the recession outlook going forward:
First, the recession probability has increased from 8% two months ago to 13% last month to 22% this month. Historically, my model has shown the estimated recession probability increasing during the months before the recession (except in 2020, when the onset of the covid recession was a complete surprise).
Second, the Near-Term Forward Spread turned negative just before Thanksgiving. I don’t put much stock in the recent yield curve inversion, but I find the NTFS more convincing as a recession predictor.
My third reason for concern is in this morning’s Employment Situation report itself. Since the onset of the covid pandemic I’ve been paying attention to the difference between seasonally-adjusted and unadjusted figures, because the covid disruptions have played havoc with the seasonal adjustment factors. This morning’s unadjusted figures were much less positive than the seasonally-adjusted numbers: for example, total nonfarm employment was down 244 thousand on an unadjusted basis compared to its +233 adjusted figure. I just honestly don’t know how important that is: after all, the purpose of the seasonal adjustments is to give us a more accurate reading, and what I’m saying is that I can’t tell if it did. But it’s a concern.