- by New Deal democrat
There’s no major economic news today, so let me take this opportunity in view of last week’s update to the CPI, to update one of my favorite labor market indicators: real aggregate nonsupervisory payrolls.
As a brief refresher, over the past 60 years this has been a very good and reliable “real” economic indicator. Basically, so long as average American households’ income is rising in real terms, they can sustain and expand their consumption, which largely drives the economic expansion. When it declines, usually households have to pull in their horns, triggering Keynes’ “paradox of saving.” What is good for the individual household causes an aggregate decline in consumption, followed by an economic contraction.
Let’s start with the absolute value of this indicator through February:
Although there were several stalls or near-stalls last year, the general trend was increasing. The gap is due to the omission of CPI reports for the period of the government shutdown. Although there was an assist from the way the Census Bureau kludged shelter costs during the shutdown, evne if they had not done so, the subsequent months would still show an increase.
So the bottom line is that, before the Iran war this indicator did not suggest a recession was imminent.
But let me break down the components of this index to show why that is. Real aggregate nonsupervisory payrolls is calculated by taking average hourly nonsupervisory wages (blue in the graphs below) x the number of hours worked (green teal), and then normalizing by dividing by the CPI (red). Below I show each of them on a YoY% change basis.
First, here is the period from the inception of the data series in 1964 through 1986:
And here is the subsequent period up until the pandemic:
There was a consistent pattern of each of the three measures during this entire 50+ year period. The inflation rate increased in roughly the year prior to the onset of each recession. Almost simultaneously, the number of hours worked started decelerating precipitously in the months before the onset of the recessions, frequently turning negative shortly after the recession began. Meanwhile average hourly wages remained steady or even increased a little on a YoY basis going into the recessions. It was only after the onset of recessions that wage increases were trimmed back.
So, what we are looking for is an uptick in inflation, and a downtrend in hours worked. With that in mind, here is the post-pandemic record:
Hours worked did gradually decelerate through 2023 and 2024, even turning negative for several months in 2025 before rebounding slightly. if you go back to the pre-pandemic graphs, a 1% YoY increase in hours worked was pretty pathetic. But there was no precipitous decline post-pandemic. Meanwhile, assisted by the disinflation in shelter costs as measured by the CPI, YoY inflation was also gradually decreasing. Note that *this* measure was affected by the shelter kludge after the government shutdown; without it there would have been no further deceleration in the CPI thereafter.
Finally, we see that while wage growth has also been gradually decelerating, it remains higher than the YoY inflation rate. If you go back and look at the pre-pandemic cycles, inflation (red) almost always exceeded wage growth (blue) YoY in the months before the beginning of recessions, although there were several exceptions, notably 2000-01, but also several occasions when inflation grew more than wages for several years late in expansions. This was due to the general deceleration of wage growth as the huge Boomer generation, and women, entered the labor force, which together held down wage growth even as two-income *household* income increased.
Of course, it could be that “this time it’s different,” like 2000-01, although I would still expect to see aggregate hours decline YoY. Additionally, it could be that the Iran war will finally cause a spike in consumer inflation that will result in it being higher than YoY wage growth.
But the bottom line is, through February the consumer economy has surprisingly been aided by an increase in real aggregate nonsupervisory payrolls. And very low new weekly jobless claims suggest that has not changed. At least not yet.
