Guy Berger

1/ We get the Q1 2026 business employment dynamics data next Wednesday. We talk a lot about the "frozen", low turnover market, as measured by JOLTS & similar data. BUT: turnover induced by establishment openings/closings is actually up relative to pre-pandemic.

2/ A corollary - turnover *within existing/surviving establishments* is actually lower than it was pre-pandemic, i.e. JOLTS understates how frozen the labor market is at these establishments

On "the real white collar labor market disruption due to AI will happen after the next recession": 1/ I often see folks argue that a lot of latent/potential automation will kick in after the next recession. Companies will lay off workers during the downturn, then not rehire them.

3/ To pick a specific example: the Great Recession decimated US manufacturing & construction employment. By the late 2010s, neither had fully recovered. But *the labor market did*, and was in better shape by 2019 than it had been in 2007.

On "the real white collar labor market disruption due to AI will happen after the next recession": 1/ I often see folks argue that a lot of latent/potential automation will kick in after the next recession. Companies will lay off workers during the downturn, then not rehire them.

Claims: 1/ Layoffs are still really low, though the streak of "below 200K" is over after a tiny upward revision to last week

3/ I still think we're due for a decline in unemployment due to permanent layoff based on what happened last month, even if the decline in continuing claims has paused in October

Claims: 1/ Layoffs are still really low, though the streak of "below 200K" is over after a tiny upward revision to last week

I don’t think this is true, at all. If the Fed wants to lower inflation, it can do it in a pretty straightforwardly What *may* be true is various other policy choices (and possibly private sector developments) are making the short-term unemployment-inflation trade off worse

1/ We're seeing renewed pay growth compression among over-16 wage and salary workers... fastest wage growth at the 10th and 25th percentile, slowest wage at the 75th and 90th

2/ The picture is a little murkier once we restrict it to "25 and over" (younger workers are on average lower earning), but still the case that lower paid workers are experiencing faster wage growth than those at the 75th percentile

Indeed job postings update: 1/ Overall postings still up a little Y/Y, consistent with a small labor market recovery. Nothing more substantial than that.

2/ Software development job postings still growing at a very brisk past (>20% Y/Y) from a very deep deficit. At the current growth rate projected forward, we'll be back to pre-pandemic levels by early 2028.

Must-read piece by @annielowrey.bsky.social … One of the biggest shortcomings of “due to AI, GDP growth will be <really big number> but there will be mass unemployment” predictions is that they’re conspicuously silent on what we’ll actually do with all this new-found abundance

Indeed job postings: 1/ Overall US postings continue to slowly rise, consistent with gradual/mild improvement in the labor market.

2/ "Human resources" postings have accelerated further in recent weeks and we could be back at pre-pandemic levels by Halloween

Claims: 1/ initial claims extremely low yet again (though not falling further - CAN they even fall further)

2/ Continuing claims have also maybe stopped falling for the time being, at a meaningfully lower level than about 5-6 weeks ago.

Don’t mind me, I’m just the guy trying to download a 783MB IPUMS data extract on the hotel lobby wifi