Moody's Talks - Inside Economics - Weighing Recession Probabilities

Episode Date: March 27, 2026

Hostilities with Iran are entering their second month, and the damage to financial markets and the economy is mounting. The Inside Economics team and colleague, Shandor Whitcher, take up the question of what it all means for the prospects of recession. Shandor tells us about his prescient random forest model of the probability of recession starting in the next year, and it’s not encouraging. Odds are still less than half, but not by much, and the direction of travel is disconcerting. Guest: Shandor Whitcher Participate in the weekly Survey of Business Confidence: https://www.economy.com/business-confidence/participate/ Email us at [e-mail] for more info about the Moody's Summit '26 Conference in San Diego Hosts: Mark Zandi – Chief Economist, Moody’s Analytics, Cris deRitis – Deputy Chief Economist, Moody’s Analytics, and Marisa DiNatale – Senior Director - Head of Global Forecasting, Moody’s Analytics Follow Mark Zandi on 'X' and BlueSky @MarkZandi, Cris deRitis on LinkedIn, and Marisa DiNatale on LinkedIn Questions or Comments, please email us at [e-mail]. We would love to hear from you.  To stay informed and follow the insights of Moody's Analytics economists, visit Economic View. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Welcome to Inside Economics. I'm Mark Zandi, the chief economist of Moody's Analytics, and I'm joined by my two trusty co-host, Marissa Dina Talley, Chris DRED. Hey, guys. Hey, Mark. Hi, again. This is podcast Take Two. We recorded a podcast right before this on the commercial real estate market, so folks can listen to that. I don't know when we're going to post that, but really good conversation, I thought, with two of our colleagues, Kevin Fagan and Tom LaSalvia, really very, very good. But I'm just going to dive right in, guys, right?

No chit-chat today because we're short on time. But I do want to introduce one of our other colleagues, Shandor Wichert. Hey, Shandor. Hi, Mark. Happy to be here. Good to have you. You know, I love your name, Shandor. Thank you.

I mean, I really love your name. Well, like, is it like, it sounds like. It's a Hungarian name. I was going to say, I was going to say, you sound like you could be in a vampire movie or something. Well, my lineage traces back to the mountains of Transylvania. So that's. Is that right? Yeah, truly.

Well, wasn't there a TV show or series called Witcher or something? Yeah, that's, that, that was, that, that was good for the brand, for sure. It certainly helped. And that was a really good series. I, you know, I thought it was, it's kind of a sci-fi slash, not horror. It wasn't horrible. Fantasy, fantasy, that's the right word. Fantasy.

Yeah, yeah, quite good. Well, it's good to have you on board because. It's a good series of books, too. Sorry. Is it a good series of books? I'm told. Oh, okay, because I'm just about ready to finish. I like sci-fi.

I'm reading Hyperion. Have you right Hyperion? I have indeed. Yeah, it's pretty good. So I'm looking for, I'm almost done. So I've got to get a new book. But it's great to have you on because we're going to talk about recession risks. That seems like it's coming to the fore.

And you've built a really cool random forest model to boil down all the economic data into one statistic, the probability recession in the next 12 months. But we'll come back to that. I want to come back to that. And I want to go to Chris and Morrison. I'm going to start. We haven't done this in a while, but I'm going to start this way. Here we are. This is Friday, March 27th.

We're now four weeks into the war with Iran. Now it feels like things are starting to go a bit off the rails and markets. Stock markets down, I don't know, 7, 8% since the war started. Interest rates are up a lot. The 10-year treasury yields up a half a point. 30-year fixed is now 6.5%. We were at 6. Of course, obviously, gasoline prices are up a buck for a gallon or regular unleaded.

You know, things are starting, the economic damage is starting to accumulate amount. And so with that as, you know, context, here I am asking the question. We haven't asked, again, I haven't asked this question in a while, but what do you think the problem? And Shannon, we're going to come back and see what the model says. Before we go to the model, I'm going to go to the economist. What do you think the probability is now a recession starting some point in the next 12 months? Chris? In 12 months, 42%.

that's a pretty high degree of precision I didn't go to the decimal place come on 42% and that's actually 42% for you based on my memory that's pretty high it's pretty high that's pretty high

yeah labor market is uh you know not not doing so well there it's a struggling here and all the risk right there's clearly a number of shocks. The only reason why I don't actually go higher is because we have a lot of stimulus still that's providing a lot of support.

So I'm actually more worried about next year. If you asked me about 2027, when all the stimulus wears off, and if we're still coping with somewhat higher oil prices, they don't have to be $100, but still, if they're in the 90s, then I worry. But in the short term, it's intense, but we do have some headwind or some tailwinds to offset some of the current headwinds.

So you referenced the fiscal stimulus. So you're referring to the one big beautiful bill act. Well, yes. Yeah, so we had tax cuts. We had defense spending, accelerated depreciation for business investment. So you do have a number of things here that are helping, are, you know, contributing to economic growth. but those will eventually wear off. Yeah, so what you're saying is if you did the mind game of,

suppose we didn't have that fiscal stimulus, that deficit finance, those deficit finance tax cuts for business and individuals, your recession probabilities would be presumably even higher. 50% plus? Plus, yeah. Probably another attempt to hire, yeah. Okay. Yeah, because the oil, I mean, this is a pretty significant oil price shock,

and it's not going away. of receding quickly here. The only reason why consumers are able to manage is because they've had this tax cut to a large degree that kind of absorbs that additional cost. Right. Mercia, where do you stand on this? I was going to say 45%. But I think, yeah, I think a week ago, we talked kind of off the podcast and I was at 40. the fact that this is still ongoing and not looking like there's a very clear end in sight,

I'm raising it to 45%. And that's pretty high for you as well, right? Yeah. Have you been higher than that over the past number of years? Have you ever – I guess at one point you might have been. Probably was around 50, right? At some point when recession – When the Fed was jacking up interest rates.

Right. Yeah. But it's high. It's high. Or the tariffs. Yeah, the tariffs. Yeah. I think even if this gets, even if Iran gets resolved, quote, air quotes with my hands here,

we're still looking at elevated oil prices for a long time. The job market is not creating any jobs. I think confidence is slipping fast across the board. I just don't see a lot of interest rates are. quite high. I mean, the mortgage rate is 664 as of this morning when I looked. Is it really? Yeah. Okay. Yep. Wow. You know, 10-year treasury is 4-4. And if you look at the probabilities of rate movements by the Fed over the course of the next year, you now see more likely to hike than to cut through the end of the year. So we're looking at potentially higher interest rates,

probably not much change. We'll see the direction of travel, but, you know, it doesn't, it doesn't look like we're going to get relief on that front anytime soon. And I think that all combines for very high probability of recession. Chandra, I promise I'm going to come back to you, but the next, I also want to see, and I have not looked at prediction markets or the good judgment project. Have you guys, Chris, Marissa, looked at those markets at all? Not recently, no. Yeah, just curious. Okay, Shandor, you're up. And maybe you can take a, just take one step back and spend a couple minutes describing this leading indicator that you've constructed. And then we'll get to, we'll get to the bottom line. What is the model saying?

Yeah, absolutely. So we rolled this out last year, kind of trying to address all these cross. costs currents, right? These, you know, the bevy of mixed signals going on in the economy and, you know, sort of distill all that down into one kind of composite estimate of recession risk over the next 12 months. And so, you know, we've had existing econometric models before, but, you know, trying to just sort of build out our offerings. We wanted to take a machine learning approach and try to a handful of different techniques and eventually settled on a random forest technique, which is basically a sort of you would imagine a series of decision trees. So, you know, they sort of, you begin

splitting as you go forward. So, you know, perhaps a decision tree shows up and it says, is the yield curve inverted? Yes or no. It would then split from there to say, you know, is employment rising? Yes or no. And from there, perhaps, you know, has, has consumer sentiment fallen? Yes or no. So eventually you have many of these trees that are trained on slightly different subsets of the data. And eventually they all could have cast a single deciding vote for each one, which are then aggregated up to establish the percentage of the probability of recession. So it's really kind of what share of trees have cast their vote in favor of recession in a given period. And so we use a variety of economic indicators in that employment and labor market really carries the bulk of the weight, followed by financial, then composite indicators. So the composite leading indicator by the conference board, my apologies, how I get with names.

it really is one of the big deciding factors here, the key input. And then you have housing, real estate, and energy markets. So we do include oil prices in that, which will be exciting to see in March, how that impacts the forecast. That's if oil prices rise, all else equal increases the probability of recession in the next one month. You know, that's part of the problem with these is there's no coefficient. So it's, you know, I can see that it has historically been associated with providing a more accurate prediction. But I don't know the directionality. So that's, you know, that's part of the fun part of tracking this thing, right?

Is like I get to, you know, as the data come out this month and this is a- Every month, I ask you a bunch of these questions and you come back and say, yeah, Mark, I can't really tell you. I promise if I could get it to tell me, I'd tell you. All I know, Mark, is it works. Somehow, some way it works. I'm not really sure. Yeah. It's certainly moved in an intuitive directional since.

Exactly. Since we've been following it. Yeah. And yeah, that's the primary thing. And it's kind of, you know, tracked with what you'd expect throughout. So, yeah, the model, you update this because employment is a key variable into the model. We wait to the employment report each month to update. So we're going to update this again next Friday when the job.

jobs data comes out for the month of March. But you have now run this through the month of February. And what was the probability recession in the next 12 months, according to the model in February of 2026? Yeah. So immediately after the jobs release, I ran the model and it jumped to just over 48 percent, which is, you know, quite scary, right? It's a sizable jump. You go over 15. You historically, it's nailed every recession. There's no false positives here. No going over 50 and no recession not ensued. Yeah, we have it adjusted to where when it crosses that 50th percentile, that's when it's been associated with a recession every time.

And with more data, as more data have come out, because when you run the jobs, when you run it right after the jobs data, really all it has is the jobs data and financial data. So you're still leaving a significant share of the pie of kind of inputs out of the estimate. So we ran it again earlier this week, and that that was kind of revised down to 40%. So I was pleased last week on the pod when you preached a bit of humility about that. You know, it's kind of, you know, you're running this with only partial data. And when the jobs of data is, you know, so discouraging, right?

It's kind of, it's going to respond to that in the absence of anything else. So you're saying now when we ran it back after the job numbers came out for the month of March, about a month ago now, three weeks ago. It said 48% are a little over 48%. Now with this run that you did a couple days ago, a few days ago, we're now at what, 40%? Just about, yeah. Just about 40%.

Okay. And so pretty close to Chris and Marissa, you know, in the same ballpark. Okay, so we're between 40 and 45%. So we're going to rerun the model. a week from now when we get the data, the March employment data. But we do know all the other indicators that we're going to have through the month of March,

they're signaling, it's sound like it's probably going to go up. Oil prices are a lot higher. All the financial market indicators, stock prices are down, interest rates are up, spreads have become more difficult. Everything would suggest that the probability is going to rise

when we get the month of March data. That's what it's set up. That's certainly my intuition, yes. Yeah, right. Not necessarily over 50%, but, you know, directionally. Yeah, and from a kind of a lower baseline of maybe 40% in February, right? So that's, yeah, it gives you. Got it, got it.

It's pretty remarkable. It's pretty remarkable because the yield curve is still positively sloped, right? That's traditionally been a very significant, indicator of recession. So to have such a high probability. What's that? What's the two year? I know the 10 years at 4, 4, 4, 4, 5.

What's the two year at? The two year. 3.98. 3.98. Okay. So it's still, it's pretty flat. It's pretty flat. It's still positively slope, though.

Yeah. If we actually were to go negative, right, then I think Schender's model would be, you know, 80%. Yeah. Yeah. Hey, here's a question for you. And this is for anybody, but maybe I'll direct it to Chris first. It feels like we're the most pessimistic economists out there.

At least it feels like I am. And by the way, I have not given you my probabilities yet. I'm going to, you got to wait because I'm not going to tell you what they are until. Well, I think you revealed them on CNBC. Oh, did I? And based on CMBC, yes, we are the, we seem to be the most pessimistic bunch. We're the most pessimistic. Everyone's moving up, right?

Yeah. No one's moving their probability is down. It's just a question of degree. Yeah. I find that perplexing, though. I mean, because, you know, the last, I'm wondering whether economists got chastened by all the bum recession calls made back when the Fed jacked up interest rates. in 2022, 2022,

2022. Remember back then? Widespread expectation that we were going into recession, Bloomberg said 100% probability. And we stuck to our guns and said, no recession, no recession.

But all those folks, I think it just feels like to me that they're not gun-shy. You know, they don't want to, they don't want to go out there again and say recession. And, you know,

obviously that would be, you know, if we don't go into recession, then it does damage to their, to their reputation, to their brand. I mean, is that, am I just reading too much into things? Or do you think that's partly what's going on here?

I think, I think there's some recency bias, right? Recency bias. Like, so tariffs, you know, day one, big tariff announcement, everyone panics and says recession is imminent, but then has facts or the situation changes, things get revised. And I think it also speaks to some of the, we've spoken to this on previous podcast,

some of the unpredictability of this administration, right? Right. Things could turn around pretty quickly, right? Things could, the president could pivot fairly rapidly, and that could lead to a much different outcome. But, yeah, I think there's some of that. And I think you're right as well. Just, you know, better not to stick your neck out, perhaps.

Stick with the herd, and then no one's going to remember if your call's wrong. Right, right. You know what that sort of like consensus probability of recession is? I don't. I think Bloomberg, in fact, does create a consensus, but I don't know what it is.

I just, my anecdotal reading of what other economists are saying, they're all, none of them are at 40, 45%, or what I said on C&BC, which was based on the model, you know, almost 50%, almost 50%. So, yeah, I think there is a fair amount of reticence out there.

What, what, since we're so close, what would it take for us to push our baseline? We think probability of recession are very high, but our baseline is not a recession, right? I mean, in fact, we're assuming that the president stands down and declares victory here pretty soon, that all prices start to come in. They won't get back to where they, they won't get back to where they were quickly, but they'll come back in. Stock market will anticipate the lower old prices and rise again, and interest rates will come in again and will be okay. We'll be diminished by what's happened. The economy isn't going to be as strong in 2026 as it would have been, if not for, what's going on with Iran, but we'll still have a reasonably good year. You know,

you know, GDP growth that's in the low twos, enough job growth to maintain relatively stable unemployment, you know, it all feel pretty good. What, Chris, I'll kind of start with you, what would it take in your mind for us to actually change the forecast and adopt recession scenario as our baseline, our most likely scenario. What would it take? I think we're very labor market oriented, so persistent job declines, right, for several months, not just one month with revision, but, you know, something that's persistently showing losses of jobs, rise in unemployment insurance claims, layoffs, right? I think that that would be the trigger, certainly.

Yeah. Yeah. Yeah. Yeah. I think if we get, I mean, I even think if we're talking this time next month, and this is still going on in Iran, there's not a resolution. Maybe things have even escalated. I mean, the president is now sending troops over there, talking about ending it, but sending troops, right? And we get a decline in employment again.

for the month of March, and maybe we get a bad CPI reading, too. That would make me feel like recession probabilities were over 50%. I mean, our forecast philosophy has been, is that if we make a major change in the underlying assumptions that drive the forecast or make a major change in the forecast, and I would put adopting a recession as a major change in the forecast, we need to be confident in that change. And when I say that, the kind of the rule of thumb is that we have to have a, we have to attach a probability to that change that's two-thirds or more. So there's got to be only a one-third chance that we're wrong, two-third chance that we're right in making that change. that's a pretty high threshold, you know, to get over to make a change.

So it feels like we're still not, you know, quite there. Yeah, or even it feels a little, that feels a couple, three months away. That's what it feels like. It could be a couple months away. I think maybe not next month. Like I'd probably push it above 50, but I wouldn't get to two thirds, right? But by May, if we're still kind of in the same place and going in the same direction, then I think we might be there.

Certainly be marking down the growth forecast and hiking the inflation forecast even next month. So we're moving in that direction of slower growth. It's just when does it tip? What if there's a stock market crash? Right. They're down 1520 and it stays down. So we anticipate, before we actually see the jobs reports or anything like that, does that change your? I don't know. The stock market's up and down all over the place. It's like, you know, we haven't really seen anything that's really sustained so far, right?

Yeah, let's say it sustains for a couple weeks, right? Does that, is that enough or do you? I don't think it's enough. I think I'd need to see actual hard data deteriorate significantly. Okay. You know, it may also, the thing that we might be want to focus on is like the unemployment rate, going back to the SOM rule. I mean, obviously there's issues with that and it's not foolproof. The SOM rule being that if unemployment rises by, on a three-month moving average basis, more than a half a point from the low unemployment rate in the preceding year. I know I'm just saying a lot, but what it means is unemployment's rising pretty significantly. then you're past the point of no return.

At that point, consumers are going to pull back on their spending. Businesses see that pull back on their hiring. You see more job loss and you're often running into a recession. We're not there yet. But even before all of this was going on, before the war on Iran, no, we were seeing those dynamics starting to unfold. Unemployment is starting to move higher. By the way, the other thing on that is that the client, that increase in unemployment, that we observed was happening in the same time that labor force participation was declining.

So that reinforces the signal here, you know, from the rising unemployment. But that may be the threshold that it might be, you know, once, you know, if we get a big bump in unemployment in March with the jobs data we get next week and we get another bump in the following month, that might trigger the somer role and that might be the catalyst for, you know, feeling confident enough to make that change, to make the change for... One of the regional Fed banks has put, and I'm forgetting apologies, which one, has put together a SOM rule that adjusts for participation. Oh, all right.

So it takes into account the supply side of the labor market, because of course, you know, the SOM rule has been, we're watching it over the last two years, but we're also seeing major shifts in labor force supply. So I forget which one it is, but I'll look at it. it. And I haven't looked at it recently to see when you adjust for supply what that looks like. But she kind of watch that. Yeah. Yeah. And the other point I'd make is once you say, okay, we're in recession, then the question is, well, when did it start? When did the recession start? And actually, historically, if you look, the best measure of start and end dates of recessions,

If you go to the National Bureau of Economic Reserve, this business cycle dating committee that says recession, no recession, picks exactly when the month it started and the month it ended, it highly correlated with payroll and employment. And if you look, I think we lost jobs in, we lost jobs in February. We gained jobs in January. I'm speaking from memory, so I might not have it quite right. But I think we also lost jobs in November and December. So you could potentially date the recession all the way back to November of 2025, November of 2025. Well, you had huge job losses in October, right? But did we have, was it October?

Maybe it was October. That was the doge. The doge cuts. Right. Yeah, it showed up. Right. So I don't know. It feels like we're getting awfully closer to a big change in the forecast.

But you could continue to have relatively strong GDP. That's the weird part of it. How do you... Yeah, okay. That's a great point. Yeah, what do you do with that? Yeah. That's the wrinkle in all of this, right? You could have GDP expansion with labor market recession, right?

In this weird new environment that we have, so... Well, actually, you make a great point. I didn't even dawn on me, but, like, because I'm so, as you said, we're so labor market focused. Could you have a recession if GDP is still positive? That's... It's the question. It's the question du jour. Yeah.

What do you think, Shandre, have you on any of this? Yeah, I would say the key variable I didn't hear anybody mention is initial claims. So, you know, even as you have the unemployment rate rising, you're still not seeing layoffs tick up. It's certainly something I've been focusing on and being sure to include when I present to clients. You know, I guess what? What we're saying, well, what I'd be saying is if, you know, we do see more, we start to see some out, more outright job laws. That would be consistent with some layoffs.

So we'll start to see some layoffs. But I'm increasingly, skeptical of the UI claims data. I'm just, and we've talked about this in the past, but given changes in eligibility roles, I mean, each state has its own UI program. they set their own eligibility rules. And they've gotten a lot stiffer since the pandemic. It's harder to qualify for UI. And there's also the concern that people just go get gig jobs.

You know, they lose their payroll job. They go get a gig job and therefore don't file for UI. And then there's also the argument that at least so far, most of the layouts have been in kind of technology, financial services, professional service is kind of a higher wage jobs. And it just doesn't, given the benefits are so much lower than they have been in the past, it just doesn't make sense for a lot of these workers, at least quickly for them to go out and file for UI.

It's just not, it doesn't work for them. So I am increasingly of the mind that UI might not be the barometer that it has been historically, you know, and I start to focus more on the overall job, payroll number than I would with with UI claims. Just something that I think we need to take into account. Any other indicators you guys are watching? We mentioned

UI, mentioned the sum rule, you mentioned stock prices, unemployment, anything else we should be watching? I know we're not playing the stats game, but I came prepared with a number. So my number is negative 18.4%.

negative 18.4%. It has something to do with recessions? It does indeed. Very topical. Is it related to the model, your random forest model that you built? It's an input. Is it labor market related?

It's related to a variety of things. Is it financial market and economic indicator? That's the one. It is the peak to trough decline so far. as of the most recent print. So the only other time you've seen a larger peak to trough decline was 25.6%. And you've never had to decline this large peak to trough in the series without it leading to a recession eventually. I'm sorry, Shandor, I missed that.

What indicator is that? The composite leading indicator from the conference board. Oh, I see. which is an input into the random forest model. It is. I was, I thought we were playing the game, so I had to drum something up. Oh, and Chris got it.

Chris did get it. I did get it. Yeah. They got it. And that indicator includes lots of other stuff, yield curve, stock market, housing permits, housing permits, a lot of manufacturing related kind of activity. And that's been pretty weak.

for a long time. It's been declining for nearly four years. Declining for four years. It was one of the kind of one of the never failed indicators that failed in that kind of 2022 to now period, but it's not as though it ever kind of rebounded, right? It just has yet to be proven right almost, you know? Right. Right. Right. Well, that was a good statistic. But we are going to continue to play the game.

Matt, we were talking about this in the last podcast, asking participants or listeners if they wanted to continue on with the game. And resoundingly, the answer is yes. We might make, we might make a few operational changes, but we're going to continue to do the set scene going forward. Okay, guys, we're going to keep this short. And we've had, again, a bunch of technical difficulties. So I'm afraid we're going to lose somebody if we don't call it a podcast.

But is there anything else before we leave? I do want to mention, again, the banking. summit. We have a, we're all meeting in San Diego in early May. I think, is it May 6th, I believe, is economic day. That's the Inside Economics Day. That's right. Inside economics day, right. A lot of us, including everyone on this call, on this conversation, will be at in San Diego and participating in that function. So hope to see you there. And guys, anything else? Oh, I'm, I haven't quite gotten the 50% on the, on the, on, the probability recession, but I'm pretty darn close. I'm sticking with that 48% that Shandor

posted back, you know, a few weeks ago. And I think we'll be back close to 48%, but very close to recession. But anything else before we call it a podcast, Shandor, anything? Yeah, thanks for having me. No. Chris? Jobs Friday next week. Jobs Friday next week. Email us at Inside Economics at Moody's.com. If you have questions, comments about the podcast or you want to weigh in on the stats

game inside economics at Moody's.com. Or the pasta. Still raging. Sounds good. Okay, good. Any other advertisements? Shandor, anything you want to pitch?

If we're talking pasta, we've gotten really into protein pasta in this house and it's just a fantastic addition. It's like all gain and no cost. It's a Pareto optimal option. Whoa, there's a lot to unpack there. We're going to have to get on.

There's a lot that we've got to explore, but we will. And with that, dear listener, I hope you have a good weekend and we'll talk to you next week. Take care now.