Why the Job Market is Freezing When the Economy is Growing | Darius Dale

[00:00:00]So, the exogenous factor, which is the orthogonal catalyst in the all of our economies is AI. Obviously, AI has the potential to be incredibly disinflationary for two reasons. One, it's probably going to limit demand for labor, at least until we get to a some sort of like very, you know, non-dystopian booming economy where there's so much GDP being created that it actually starts increase the demand for labor. I'm sympathetic to that view because that's been the history of general purpose technologies. They always historically increase the demand for labor. The problem with AI is that it's kind of a unique general purpose technology in that it actually can do labor for us in a way that we've never really had. So, I don't know the answer to that, but my core thesis is that at a bare minimum, AI is probably going to restrain the demand for labor. And by restraining the demand for labor, you're ultimately putting downward pressure upon wages. At the same time, you're also putting upward pressure on productivity. So, downward pressure on wages plus upward pressure on productivity equals downward pressure on inflation. Our analysis suggests that the U.S. trend inflation rate in the U.S. economy in particular can decline by roughly 50 basis points if we're right on just that that dynamic in terms of the wage productivity dynamic. So, you could have a just a not a sea change, but a material change in the inflation dynamic in the economy that lowers the trend inflation rate, that lowers the trend inflation volatility, that ultimately pulls that stock-bond correlation back to negative in a way that can increase investor demand for the securities. And so, that is a risk factor, but I'm less concerned about that risk at the current junction.

[00:01:21]The reason I would say this is when you look at the market's pricing of neutral, which, you know, is a hotly debated subject in institutional finance, but how we determine the market pricing for the new Fed's neutral rate is when we're in a cutting cycle, we look at the minimum value on the overnight index swaps curve out 5 years to see what the market essentially saying, "Okay, this is the level the Fed is trying to get the rate to." Or they should get the rate to, right? Or if we're in a hiking cycle, we look at the maximum value on the overnight index swaps curve out 5 years. You know, that's the market's estimate of of of neutral. This is where the Fed is trying to get the rate to or should get the rate to. And when you look at the minimum value right now, assuming that we're still in a cutting cycle, which I could I could argue we're probably not and shouldn't be. Assuming we're even still in the cutting cycle, the market's current estimate of neutral is now above the effective fed funds rate, which ultimately means the market thinks the Fed is now getting accommodative because the market is increasing its estimate of neutral. So, why would the market be increasing its estimate of neutral? Well, one of two things could be happening. It thinks AI is a massive aggregate demand shock to the economy, $800 billion in CapEx in one year going to $1.2 trillion next year, maybe even higher. So, that could be true or it thinks AI is going to be so productivity enhancing that it pushes up the neutral rate in terms of the supply and demand of capital or both.

[00:02:31]Either way, those are two dynamics that are still negative for bond prices.

[00:02:34]>> Negative for bond prices, why? Make that connection for me.

[00:02:37]>> The aggregate demand shock will push up inflation.

[00:02:39]>> Oh, that's right because it's still in that kind of an inflationary enhancing >> Here's $800 billion or 2.5% of GDP just to buy, you know, chips and conductors and data centers, you know. So, that's a massive demand shock that could be inflationary in the interim while these things are being built out prior to, you know, kind of experiencing the economy-wide benefits of the technology.

[00:02:57]So, that's part of the reason why the neutral rate could be gravitating higher. Another part of the reason is that investors are looking around and actually starting to see the impact of this technology. I don't know if you guys have spent any time on Claude in the last few months, but holy cow.

[00:03:08]>> The world's changing very fast.

[00:03:09]>> The world's changing very fast. And so, I think the market is also looking around and saying, "No, this stuff is going to boost productivity by a tremendous amount." And if you have productivity go up by a tremendous amount, historically that's been very positive for the neutral rate because ultimately you're competing for capital.

[00:03:22]The bond Treasury bond market will be competing for capital in a higher return private sector.

[00:03:26]>> Earlier you had you had talked a little bit about just the impact of AI on labor demand. I'm curious, have you tried to quantify that? It sounded almost like you had that kind of wash out in that 50 basis points change to to the inflation rate. I thought I heard that. I was just curious cuz that's a fascinating research question to me. I think it's on a lot of people's mind. Like, what is ultimately going to be the impact of this technology?

[00:03:49]>> Yeah.

[00:03:49]>> I don't know if you can say any more to that or if that's a that's a paid service.

[00:03:52]>> No, no, no, [laughter] no, no, no. So, just getting back and answering your question, have we quantified this? Yes.

[00:03:57]I don't know that we quantified it well and quite frankly, I don't know that anybody's quantified it particularly well cuz we're so early in this, but two ways in which we tried to quantify this.

[00:04:04]One, we looked at the core private payrolls, the core part of the labor market that is what we would consider to be most at risk for being disrupted by AI. So, if you look at private payrolls and then you X out what we would call things that are government and government-adjacent. So, like the actual federal, state, and local governments, healthcare, social services, and education. So, you take those out of the private payrolls and what is left are, you know, the private payrolls that we think could be, you know, materially impacted by AI. That's about 54% of the labor market. That statistic peaked in nominal terms in March of 2024 in an economy that's been growing above trend and a labor market that has generally added jobs since then. So, the part of the labor market that is most exposed to AI peaked to over 2 years ago >> Yeah.

[00:04:46]>> in terms of total employment. So, that's somewhat alarming in the context of a statistic >> that goes straight up over time. It looks like a chart of the S&P. It just goes straight up over time and then now it's just going sideways and gradually going down.

[00:04:56]>> Right. Right. And that's really like 2024 feels like that's when AI really started to come on the scene.

[00:05:02]>> I think that's when people really started to say, "Okay, ChatGPT is somewhat viable." Yeah, yeah, yeah.

[00:05:07]Yeah, yeah, yeah.

[00:05:08]That's certainly the first time I used it. I had a great podcast on our former show, which I no longer have time to do, but I was on with Raoul Pal back in in early 2024 and he was the first person to really get me to like sit down and really kind of open up, you know, like spend a day on ChatGPT. See, you know, this is stuff is real. And obviously in our opinion, I think quality is and Jim and I both overtaken it, but uh that's neither here nor there. The second piece of analysis that we've done to quantify AI's impact on the labor market and and the spoiler is we think AI is having a negative impact on the labor market. So, there's three ways in which in fact uh the second uh before I get into the second way, the third way is that we see this low high or low fire dynamic in the labor market. We've never seen an economy that has been persistently above trend for as long as we've had in nominal GDP terms have such a low higher low quits rate dynamic. And we've been well below trend in those metrics for such a long time. Now, we haven't seen the layoffs and discharges rates spike, so it hasn't been like mass layoffs or anything, but companies just aren't hiring.

[00:06:01]>> Mhm.

[00:06:02]>> And as a function of companies just aren't hiring, people just aren't quitting cuz they don't feel confident that they can go get another job. And so we have this very like it's a very low dynamic dynamic labor market. There's no dynamism in the labor >> With some actual layoffs sprinkled in there. I mean, some notable layoffs, yeah.

[00:06:15]>> Meta, you know, all these big All the people who are selling us the AI are firing people. What does that tell you at home about AI's impact on the labor market? The people who are building and selling the AI keep firing people, not hiring people.

[00:06:28]>> Yeah. Come to Come to your own conclusion.

[00:06:30]>> Exactly. Yeah, exactly. Yeah, so going back to analysis number two, which I think is the most compelling analysis.

[00:06:35]When you look at the labor market relative to things that indicators that lead the labor market. So, corporate profits, retained earnings, equipment investment, and labor supply, the civilian labor force data. My opinion, based on our analysis, those are the four most worthwhile These are the four best leading indicators of of total employment. When you look at the current growth rate of private payrolls relative to how fast they should be growing based on the trends and the growth rates of corporate profits, based on the growth rate of of retained earnings, based on the growth rate of equipment investment, based on the growth rate of labor supply, the growth rate of private payrolls is dramatically underperforming each of these leading indicators.

[00:07:10]Dramatically underperforming. So, for example, I think NIPA corporate profits are up about 9 or 10% year-over-year.

[00:07:17]So, in that quintile band of corporate profit growth, you know, the private nonfarm payrolls, with data going back to the 1940s, private nonfarm payrolls on a median basis when you're in this quintile band of NIPA corporate profits growth has historically grown at about 2.1, 2.2%. It's growing at 0.4%. So, dramatically underperforming by statistically significant degree how fast it should be going based on this leading indicator of the labor market.

[00:07:39]And so, across those four leading indicators, which you which our analysis has found that those are the four best leading indicators for total employment, is dramatically underperforming in each scenario, including in the civilian labor force. I know there's a view out there, especially amongst policy makers, that the reduction in the crackdown on immigration is reducing labor supply and it really that breaks down the the break-even rate of total employment, which is true. That is true. We have a decline in the labor force. However, we are in a fifth percentile value in the civilian labor force, and even if this quintile, this low quintile, we're still underperforming in private non-farm payrolls growth by about I think 60 basis points. We should be growing at about 1% and we're still only growing at about 0.4% so there's even, you know, some meaningful enough deviation downside deviation in that component as well. So, when we look at across those three components, this underperformance, the low higher low fire low dynamism labor market, and ultimately the peaking of the part of the labor market that was most impacted by AI to over two years ago, I think that's enough evidence to suggest that the early read on AI's impact on labor market is at best neutral and at worst probably negative.