CORRECTION: Jobs data actually did peak in 2015, implications negative

I hate backtracking, but I need to make a correction to the analysis I just sent out. An alert reader, who also tracks the jobs data I am using, wrote me, noting that my data series seemed to chop off the last several months of 2015, making it look like the second derivative Non-Farm Payrolls was still positive. He says it is not positive at all. And when I looked at the data again just now, I realized he was right.

Here’s the original chart with the second derivative positive.

And here is an updated zoomed in version for the last two years showing the second derivate negative since early 2015.

Unfortunately, this chart looks bad enough that it pushes up my recession concern a good deal — from just a tail risk to a plausible outcome. We need to be watching this second derivative statistic — both for non-farm payrolls and for the more quickly updated jobless claim series.

Neither series is saying recession yet. But if the jobless claims series ticks above zero, i.e. to 300,000 by March and the non-farm payrolls second derivative is still negative, then we would have a fairly definitive recession call. The further into 2016 we go, the more difficult the comps become for initial claims as they go sub-300k in March.

Sorry to have to backtrack somewhat here, but the missing data are important. It moves recession from tail risk status to being just a 30 or 40,000 average initial claim rise away. If we have 300,000 jobless claims in March, the change in non-farm payrolls is falling, and the Fed is still on a rate hike path, recession could become the base case.

Thanks for the correction, Geoff. I still like your analysis, Tim. But now it looks like I am now closer to Raoul Pal and Warren Mosler than Menzie Chinn.