What Carlyle's Proprietary Data Signal About the Economy

Bloomberg

Watch What Carlyle's Proprietary Data Signal About the Economy - Bloomberg on Dailymotion

Transcript

00:00Jason, great to see you. And this is the holy grail, right? Even whether we have the data or not,

00:05what alternative data can you use to get a handle on this economy?

00:08So what did you look at to compile this picture of America?

00:14So we have 277 portfolio companies currently that employ about 730,000 people. So each month,

00:23and we've been doing this for about 15 years, we get a lot of data from those companies,

00:28business volumes, revenues, orders, backlogs, prices paid, prices received. Very often,

00:36it's by a number of different business lines from the company. So again, a large number of

00:40time series from each company. So we have a very large matrix that we then weight. So we have an

00:46algorithm that essentially weights those data to replicate time series of interest like GDP

00:52or payroll employment, industrial production, other series like that. And then each month,

00:59we get the data that come in, again, after going through this weighting algorithm. And we measure

01:03where it is in relation to the prior trend, and measure how discrepant that is relative to our

01:09expectations. And if it's statistically significant, we move the estimate up or down proportionally.

01:15And then, of course, we look at the evolution of official statistics. Very often, they're

01:22significantly revised. I think it's important that everyone recognize that these official

01:26statistics are not handed down on stone tablets. They're just estimates, too. They're using the

01:31same sort of sampling techniques. And then, you know, over a period of, say, six months,

01:36we can reweight our data if there's any signs that it's not quite tracking the official statistics

01:42statistics as well as it had in the past. Well, Jason, do you, I mean, I assume you backtest

01:47your data against the Bureau of Labor and Statistics. So I'm wondering how well does it

01:53track? And, or do you believe that maybe you're getting a better picture of the U.S. labor market

01:59than the BLS? Well, you know, I think that the, first of all, when you think about the weighting

02:05algorithm, essentially, we have enough data that we can perfectly replicate over the prior five

02:11years. So at a backward-looking basis, the sample is very much selected so that it can

02:18perfectly capture it. And then on a go-forward basis, you know, the correlations over the

02:23next three to six months range between, you know, 60 to 90 percent. So, again, pretty good

02:29on a forward-looking basis. I think the BLS has had a tough year, of course, when you think

02:36about the scale of the revisions. They cover what they think is about 38 percent of payroll

02:42employment in the United States. And I think it's always hard when you think about company

02:47births and deaths. Of course, there's questions about the response rate and who is responding.

02:53We saw that during the pandemic. And then, of course, they have an entirely separate series,

02:57that is the unemployment survey, that, you know, is a very small percentage of the population.

03:04Of course, it changes every month. But that's another area where, again, these are estimates.

03:11This is, there's other ways to go about collecting data. So to say it's better, I mean, no one has

03:18the sample or the breadth of the federal government when it comes to these things. But in terms of what's

03:24important for investors, yes, I think there are better real-time indicators. And I think

03:30the alternative data sources are something that a lot of people are really becoming much more attuned

03:35to. Well, for what it showed you specifically, Jason, your estimates are for this most recent

03:39jobs data that we didn't get would be that 17,000 jobs were created. You can look at that number in

03:44face value and say that is painful and significantly below numbers before. How do you interpret 17,000,

03:51especially as we ask questions of what the real break-even rate of jobs growth is considering

03:57how much immigration is changing? That's right. I really think that that's the critical question

04:03here as one evaluates this. If you just look at the raw numbers, you see an economy where payroll

04:09employment growth was averaging about 200,000 at the start of the year. And that falls down, again,

04:14in our estimates below 20,000. You would say, boy, this is an economy that is on the verge of recession

04:20or perhaps has already succumbed to recession. But I don't think that's the correct interpretation

04:24because it doesn't actually jibe with any of the other data that we're seeing. In fact,

04:30we saw a GDP growth in the quarter of about 2.7 percent, a rebound in consumption, particularly

04:36travel, tourism, live events, those experiences. And then, of course, business spending growing about

04:424.8 percent annualized rate according to our data. Much of that, again, the AI, this concentrated

04:50CapEx boom. So I do think that there's some decline in labor demand. We did see some evidence that when

04:59we, again, weight the sample for the broader U.S. corporate sector, that margins did contract relative

05:06to year-ago levels. I think there's less hiring perhaps because of tariffs, related uncertainty.

05:12But I do think most of this is explained on the supply side, which is to say, moving from a very

05:19large volume of immigration to this year, what could be something close to net zero immigration.

05:25The U.S. population, domestic population, is only expected to grow about 14 basis points this year.

05:33Wow. If you apply that to 160 million payroll employment, that implies full employment,

05:41monthly job growth of only about 18 or 19,000. It's pretty fascinating stuff. Really cool

05:49data. Jason, love to have you back and talk about it as this story evolves. Jason Thomas there of Carlisle.

Category

Transcript

Be the first to comment

Recommended