Today’s ADP report showed a loss of 371K jobs roughly in line with other economic indicators. Together with the New Claims data it suggests that the BLS payroll report on Friday will show a loss of around 365K jobs.

The historical ADP data tracks BLS pretty well but this after substantial revisions to both series.



ADP only counts private sector jobs and has historically been about 15K jobs lower than the BLS, but also about 10% less volatile. Take together this would imply that BLS should  clock in at 396K.


However, there are a few reasons to think that might not be quite right. First, government employment is likely to be less volatile than trend given the stimulus package’s attempt to stabilize state and local employment.

Also, the New Claims for Unemployment Insurance data predict a somewhat lower BLS Payroll number.


To get this I averaged the new claims for employment insurance from the last four weeks of each month to get a monthly average for new claims. I plotted this monthly average against the payroll data.

I used a second degree polynomial because of the same floor problem that existed with the unemployment rate. New Claims just don’t go below 200K even if the economy is smoking.

This rough and ready method predicts a loss of just under 340K on the BLS payroll series. Taking these two methods together gives me around 365K as a very rough forecast.

As a note the the scatter data suggests that to get no change in employment you need 388K new claims for unemployment insurance, roughly equal to Brad Delong’s eyeball estimate.