Duke finance professor Cam Harvey, the father of the yield curve as a prescient predictor of future recession, weighs in on what the curve is saying about recession in the coming year. You will be surprised. Mark and Cris were.
For more on Cam Harvey, click here
Follow Mark Zandi @MarkZandi, Cris deRitis @MiddleWayEcon, and Marisa DiNatale on LinkedIn for additional insight
Mark Zandi: Welcome to Inside Economics. I'm Mark Zandi, the chief economist of Moody's Analytics, and this is a special bonus cut podcast. Chris, Chris deRitis is joining me. It's not a bonus because you're on Chris. Sorry to say that.
Cris deRitis: Oh, wow.
Mark Zandi: Yeah. Well, I mean-
Cris deRitis: How did it...
Mark Zandi: Really? You took that-
Cris deRitis: [inaudible 00:00:33] ego there, right?
Mark Zandi: No, no, no. Well, it's always special when you're on Chris, but you're always on. Yeah, that is true. We had an anniversary, the podcast.
Cris deRitis: We missed it. We totally missed-
Mark Zandi: Completely. No one said a thing to me. I think it was a couple podcasts ago, or maybe it was last week, I can't remember. We had our 100th podcast. Can you believe that? It's like two... We've been doing this for almost two years. It's like hard to imagine. So it is special talking to you, but I talk to you all the time. So no, this is a special bonus podcast because we have Campbell Harvey. Hey, Cam.
Cam Harvey: Thank you for inviting me.
Mark Zandi: Yeah. Cam, is the professor of finance at Duke University. And I've been reading your work for many years, and I think of you as the father of using the yield curve, the treasury yield curve as a predictor of recession. And so all roads lead back to you. And you told me that indeed they do because your PhD thesis was exactly on this topic, the yield curve as a predictor, future recession. When were you in grad school? When did you do get your PhD?
Cris deRitis: So my PhD is stated in 1986, but the idea goes back even before I got to the University of Chicago.
Mark Zandi: Oh, you got your PhD in Chicago. Oh, I didn't know that. And so where was the idea come from? What's the genesis of this as a... Oh, and let me stop for a second and to provide a little context, I just take it for granted that people know what we're talking about, but everyone, the strong consensus view of economists and in many others is that the economy's headed towards recession. And broadly, there's two reasons for that. One is to circumstance, we have high inflation, and the Federal Reserve is raising interest rates aggressively. And if you go back historically and look, when you're in a high inflation high rate world, you end up in recession a lot more often than not. In fact, you have to look pretty hard to find the time when you don't end in recession. And the second reason is that the leading indicators that economists tend to use to assess the probability that the economy is going to go into a near-term recession are not all of them, but many of them are signaling red, flashing red.
And the most prominent is the yield curve, the shape between or the difference between long-term interest rates, 10 say tenure, treasury yields, and short-term interest rates. And we're going to come back and discuss exactly how you measured the yield curve camp. But that when that curve, that curve is typically positively sloped, higher long-term rates than short rates. But sometimes it becomes inverted. Short rates rise above long rates. And when that happens historically, we end up in recession some period after that, 12, 18 months after that.
So with that as a preface, and it's very prescient. I mean, very prescient. If you go back and look at the recession since World War II, there's been 12 of them. And you look at the yield curve, it inverts before the recessions hit. And invariably, it depends on how you measure the old curve. You don't get false positives, you don't get the curve inverting and recession not happening. So it's a really prescient indicator. So economists look at that and they say, "Oh my gosh, if history is any guide whatsoever, we're going in." Okay. So first of all, let me ask you this, Cam, did I say anything there that you would take um bridge with or would like to elaborate on? But then let's go back to ground zero. Where did this idea come from? Where did this regularity that you observed, how did it pop in your mind and became ultimately your thesis?
Cris deRitis: So the story is, looking back on it, pretty bizarre. And let me go through what happened. So I was a first year master's student, and I had a couple of internship offers, and one was an all in Toronto that's my hometown, and one was at a major media company and the other was at the largest copper miner in the world at the time. And I chose to go to work for the corporate development division of this large copper miner. And the first day on the job, they gave me a task, and the task was to build a model to forecast real GDP growth. And at that in real time, I didn't think it was a big deal. So what do I know?
Mark Zandi: No big deal.
Cris deRitis: 23 years old.
Mark Zandi: I can do this.
Cris deRitis: And they give me a job. And looking back on it just doesn't make any sense. Like copper and GDP are so highly correlated that the decision to open a new mine or close a mine is critically dependent upon your view of the economy. Yet you've got this student intern in charge of developing a GDP forecasting model. But again, in real time, I thought it was no big deal. So I start and I realize that I'm at a significant disadvantage at the time, there were these econometric services you could subscribe to for tens of thousands of dollars. They've got these simultaneous equation models.
Mark Zandi: That would be us, Cam, by the way. That would be us. Yeah, go ahead.
Cris deRitis: Yeah. And there's just no way I could build a hundred equation model, assemble all the data and deliver a credible forecast. It was just impossible to think about it for one person versus these companies that have hundreds of economists working for them. So I had to take a different approach, and the first thing I thought about was looking at something simple. And what I wanted to look at were asset prices. So given my finance training, those asset prices should reflect expectations of what's going to happen in the future.
And I thought about stocks, I thought about bonds. I did a little research on stocks without looking at the data. And there was a prominent research stream at the time spearheaded by a Chicago professor, Eugene Fama that looked at the relation between the stock market and real activity and had to look at, looked at those papers and realized that this was not going to be the measure that I really wanted to look at. And there were multiple reasons for this. And one reason kind of a joke at the time was that the stock market had forecast nine of the last five are recessions and the reason-
Mark Zandi: I think that was a Samuelson quote, wasn't it? Wasn't that a Samuelson quip, I believe? Yeah, Paulson-
Cris deRitis: You're correct. Yeah. It was actually a quote in 1950s. I'm not really has changed. So there's many reasons why it could be an unreliable indicator. The essential intuition is that valuation of a stock is based upon the discounted present value of the cash flows. And those cash flows are driven by earnings, and earnings will be impacted by real activity. But there's a lot of other stuff going on that could make it really noisy, including you need to discount those cash flows by a rate that reflects both what's happening in the economy and just shifts and risk. The cash flow's not so obvious because it changes through time, and the duration of stocks is very long. So you put all this stuff together and it could be very unreliable as it was at the time. So I decided instead of looking at the stock market to look at the bond market, and the bond market had a number of advantages.
So number one, there was a fixed maturity. Number two, the cash flows were known, so the coupon is stated. And then number three, if we're looking at US treasury bonds, the risk is minimal in terms of default risk and things like that. So you put that all together along with the basic economic intuition that a nominal interest rate is made up of the real interest rate, expected inflation and a risk premium. Assume that risk premium is fairly low, and you've got this expected real rate, you've got expected inflation. The expected real rate, according to almost every standard economic theory is linked to expected real economic growth.
So I've got something, and the reason that I looked at a yield curve rather than just rates is very simple, that I didn't want to deal with the expected inflation component. So to take a difference, I could isolate that the expected growth. So that's kind of where I went. So I put this together and it looked very promising, and I was about to present it to the higher ups in the company. I show up and I'm told that the entire corporate development group has been laid off, including me.
Mark Zandi: Oh, geez. Oh, geez. What was the company's name?
Cris deRitis: Balkan Bridge Copper.
Mark Zandi: Oh, okay. Largest in the world-
Cris deRitis: You could look them up. They interestingly failed. Of course, if they're putting a 23-year-old kid in charge of a critical input.
Mark Zandi: You knew they were going down.
Cris deRitis: That was the indicator right there.
Mark Zandi: That's the indicator.
Cris deRitis: This is unbelievable, and this is really harsh that you're in the middle of your internship as a student and you're laid off. That usually doesn't happen at a firm of this prominence. And in addition, this is in Canada, the Canadians is supposed to be much nicer and to send the kid to the street. So I was not too pleased, obviously, but things kind of work out. So I had an extra four weeks, and I started to do additional research on the idea because I was excited about the idea. And then I went back for my second year and showed a few professors what I'd come up with, and they said, "Oh, wow, this is a really good idea. You Need to apply for a PhD."
Mark Zandi: Oh, wow.
Cris deRitis: And I had no idea. I was the first person in my family with a bachelor's degree, and maybe they got the idea that I could get a quick master's degree, but PhD, I had no idea.
Mark Zandi: From Chicago, no less was the-
Cris deRitis: Well, I didn't even know where to apply.
Mark Zandi: Novel laureates, right. Yeah.
Cris deRitis: So they helped me, I put my application together to various schools, and in hindsight, it was a strong application because what I did was I included my paper that I had written up. So my master's program they actually gave me, they combined a few courses and allowed me to just work on the paper. So I included the paper, I applied to these programs and they said, "Well, this person's doing research already and let's take a risk on him. So that's how I ended up in Chicago.
Mark Zandi: That's great, that's such a cool story.
Cris deRitis: But this was not easy. Even though I got to Chicago with my idea, Chicago's got very high standard. Indeed, three people on my committee went on to win Nobel Prizes so-
Mark Zandi: So Fama was...
Cris deRitis: Fama was my chair. Marton Miller-
Mark Zandi: Martin Miller. Oh, okay. Wow. That's incredible. You may be the only economist in the world of PhD with the three advisors as Nobel Laureates, that that could be a record.
Cris deRitis: Yeah, well...
Mark Zandi: Really-
Cris deRitis: At the time, it was interesting because the students kind of knew that they would win, but it took many years for them to win and-
Mark Zandi: Yeah. Oh, sure. Yeah. Right.
Cris deRitis: But nevertheless, it was-
Mark Zandi: Yeah.
Cris deRitis: They were very rigorous. And my paper, the quality data really, if you're looking at treasuries, is after the Fed treasury accord. So you really can't go back that far in history because the rates were so manipulated.
Mark Zandi: Which is in the '50s. I can't remember when was that accorded? It was 19-
Cris deRitis: 1953.
Mark Zandi: It's 53.
Cris deRitis: I could be wrong in around there.
Mark Zandi: I think that's right. Yeah.
Cris deRitis: So yeah, I've got an economic theory. So that's essential. So you just can't have an empirical finding and think you're going to get a dissertation from Chicago. So there is a theory, and then there is empirical result, which appeared quite striking. But nevertheless, think of it as being four out of four for recessions. And my committee saying, well, this could just be lucky. And they were impressed that I got the double dip recession and nobody else got that. So none of the big-
Mark Zandi: 1980, there was a recession then there was one in '82, and the curve signaled that it inverted before the "80s.
Cris deRitis: Yeah, yeah. Inverted. It went positive, then it inverted, then it went positive. And if you look at the yield curve and look at a real GDP growth, it's a mirror. Yeah. It was super impressive that it actually got that. But again, you get 4 0 4, it could be lucky. And I think a couple of things worked in my favor. The most important being that the idea had sound economic intuition and theoretical foundation. So when that occurs, then even if there's not that much data, people will kind of go along with it. And the other thing that they really liked was the fact that it was so simple and that it was competitive or beat these econometric services that cost tens of thousands of dollars to subscribe to. And the cost of my forecast was at the time 25 cents, which was the cost of the Wall Street Journal back then.
So that was good. And they signed off, and then we go to the outer sample period. So after you publish your dissertation, what happens, and usually there's two things that happen in the good scenario, the effect that you document gets weaker, and in the bad scenario, the effect completely goes away. But that's the way it works in science. But in my situation, the effect didn't go away. And the first real challenge I had was October 1987, so I'm a junior professor, and the stock market had crashed and economists believed that there'd be a recession in 1988. So as widespread agreement that real GDP growth was going to be negative and we're going to go down. And I remember being at a conference and all this doom and gloom, and I was the most junior person. And I said, "Well, I've got this model, this yield curve model that tends to do a good job."
Historically, I'm predicting real GDP growth. And I think real GDP growth is going to be 4.2% in 1988. And the reaction was almost laughter, what a joke. Like who is this kid? And the model is obviously a false model. And that was the first test. And in that growth was over 4% in 1988, and there was no recession. And then the next four inversions of the yield curve each were followed by a recession. So for the data that I looked at, so in sample four recessions out of sample four recessions, eight out of eight, and you might put an asterisk on the COVID recession because obviously-
Mark Zandi: Was going to ask you about that.
Cris deRitis: ... the yield curve didn't forecast COVID. But in real time in 2019, when the yield curve inverted at the end of June, there was widespread expectation that we were going to go into a recession. So our CFO survey at Duke University, 70% thought we're going into a recession. So we will never know the counterfactual. But nevertheless, the foundation was there, and I will count that as one of the eight out of eight.
Mark Zandi: Got it. Hey, so there's a lot to unpack there. Let me first though, begin with, there's lots of ways of measuring the yield curve, 10 year treasury yield versus the two-year treasury yield 10-year treasury yield versus the three-month treasury bill 10-year treasury yield. Usually, it's the 10-year yield as your long-term interest rate. And then there's a lot of short-term rates. The other would be the federal funds rate, the Federal Reserve controls. Which of those measures are your favorite, or do you have a favorite? You can't pick which one do you look at most regularly?
Cris deRitis: So I had to pick you back in 1986. So in 1986, I looked at the 10-year minus the three month. And the logic was I want some yields that, or some treasury bonds that are liquid. I chose the three month because I'm forecasting quarterly G D P, so it kind of makes sense to use a three-month rate. And I chose the 10 year because it was highly liquid, the most liquid and still is. So I looked at the 10-year minus the three month, and that one is the one I referred to as delivering no false signals.
Now, others have looked afterwards, let's say the 10-year minus the two year, and all of these yields are correlated. So if you look at that yield curve versus the 10-year minus three month, it's got high correlation. But the way I look at it is, okay, well, my original idea was 10 year minus free month. It's eight out of eight, and there's not really a good reason to switch it out. So if it was four out of eight, so it failed multiple times and forecasting, then that's a good reason to switch it out to something else.
Mark Zandi: No false. I can't recall. No false positives with the 10 year, three-month.
Cris deRitis: That's correct, eight out out of eight with no false signals. So that's important because you could be eight out of eight and have 20 false signals. This has got zero, and I didn't see a good reason to swap it out. And there are an infinite number of choices. So people say, "Well, the 10-year minus two year, but it could be the eight and a half year minus the one and a half year you could data minus to find something.
Mark Zandi: AI, yeah.
Cris deRitis: And in this case, as I said, there's no false signal. If you look at the 10-year minus two year there is. And in 1998, and I think you go with the original model until it fails and then you reexamine it.
Mark Zandi: One quick technical question on the three month, is that on an equivalent bond basis or not?
Cris deRitis: You need to be careful here with the historical data, the yields are quoted on a discount basis, discount basis. A lot of people make this mistake that if the treasury bill is the 12-month pressure bill is trading at $90, the discount yield is 10%, but we know the true yield is greater than 10. So you need to make conversions. There's all sorts of conventions that you have to-
Mark Zandi: So you convert, you do it on an equivalent bond basis. You look at 10-year treasury versus three month on an equivalent bond basis?
Cris deRitis: Yes.
Mark Zandi: Yeah, okay. Yeah. And Chris, I'm going to let you in just a second. I know-
Cam Harvey: Just a quick technical-
Mark Zandi: Okay, go ahead. Go ahead.
Cam Harvey: Do you consider any type of threshold in terms of the extent of the inversion, of the number of days? If it's one basis point, it's inverted, or-
Cris deRitis: Yes. So what I did in my dissertation, I looked at the average over the quarter.
Mark Zandi: Quarter.
Cris deRitis: If you at work for one day quarter or one week, that just doesn't count. Again, the measurement of GDP is a quarter, it's not a day. So take this into account. And so after a three-month period where you've got an inversion on average, then I declare that a code red event, in terms of my model.
Mark Zandi: So 10 year yield, three month on an equivalent bond basis for a three-month period, a quarter, that's the signal recession.
Cris deRitis: That is the signal. And that is what is correlated with economic growth. So my model shows that that spread is a strong predictor of real economic growth and-
Mark Zandi: How far ahead? How far ahead? What's the typical lead?
Cris deRitis: So the lead varies and it varies between let's say six months and 18 months.
Mark Zandi: Okay.
Cris deRitis: So let me tell you what the model does really well and what it doesn't do as well at, so obviously, given what we've already talked about, it's really good at predicting recessions given it's eight out of eight. And this is interesting, it's also very good at predicting the duration of recessions. So the length of the inversion is highly correlated with the length of the recession. And there's a third aspect that it doesn't do as well at, and that is the extent or the depth of the recession. So I've been criticized on social media, well, the Harvey model, it doesn't do very well getting the depth of recessions, and I'm thinking, well, if I get a forecast of the recession event accurately and then the duration, well that's pretty good for a single variable. There's just one variable. So yeah, it doesn't do everything, but at least historically, it's done really well.
Mark Zandi: Got it. Let me ask, I want to get back to why the curve is a good predictor. And then of course, obviously I want to go to is it a good predictor today of recession debt ahead? But before I do that, a couple other kind of nuts and bolts questions. One is, why isn't the yield curve useful or seemingly useful overseas? I mean, if we go look at yield curves and other developed economies, you don't see that kind of relationship. But what is your thinking around that? Why is the US yield curve such a good predictor, but others are not?
Cris deRitis: Yeah. So my early research looked at other countries, and you're correct that the other countries don't have this strong relation the US has, and an obvious reason for this is manipulation in the bond market. One country that was particularly interesting for me was my home country, Canada. So you think of Canada as well. It is highly tied to the us. So the business cycle in Canada just mirrors what happens in the us. So the Canadian yield curve should have very little information. So I wrote this paper where I tried to forecast the difference between Canadian economic growth and US economic growth. So the part that wasn't explained by what happens in the us, and it turned out that the difference between the Canadian and US yield curves was very powerful in predicting the difference. So for a country close to home, that indicator is quite important. But if you go to other countries, for example, Japan, there's no relation. And is it a surprise to you?
Mark Zandi: No.
Cris deRitis: Given what happens and-
Mark Zandi: Is the Vietnam-
Cris deRitis: Bond market. There's no surprise.
Mark Zandi: Right. Which gets to another quick question. Do you think the Fed's quantitative easing, quantitative tightening is messing with the curve in terms of its signaling? Because it's no longer totally a market driven measure, it's now affected by policy.
Cris deRitis: So yes. And I just want to emphasize something that the economic model I use is so simple that there is no Fed is really simple. And historically, the Fed has been very active in manipulating yields. They've Operation Twist is a good example, the original one. Indeed, I think that we talk about quantitative easing and quantitative tightening. In my opinion, given the massive size of the bond market today, that it was probably easier 30, 40 years ago for the Fed to manipulate the yield curve.
The market is just so large now, it's so difficult for the Fed to deal with it. So yes, there is a series of interventions that adds noise to this indicator, and there's very little I can do about that. The model is what it is. It's a simple model, and it gives us some information, which appears to be valuable. If I was in the business, again, if I was asked to develop a model for forecasting real GDP, just like I had the task as a student intern, I would look well beyond the yield curve. So the yield curve is important, but there's obvious other information that needs to be taken into account.
Mark Zandi: Yeah. Okay. So you're saying, and just for the listener, QE QT, that's the Federal Reserve buying treasury securities, mortgage securities, and then of course in QT, allowing those securities run off the balance sheet. So they are big, they've become, and that's since the financial crisis. So they've become really large players, the Fed in the bond market. And that is, as you're saying, having has to have some impact. The question is how big an impact? I mean, they've got nine in a half, 9 trillion on the balance sheet.
Now, before the financial crisis, it might have been a half a billion, I don't know. I mean, I don't know. I can't remember. No, it was four or 500 billion, now it's 9 trillion. So it must have some effect. Here's the other thing. The bond market has US bond market has... It feels like, correct me if I'm wrong, become more internationalized globalized over time as well. It used to be pretty much a domestic investors would buy treasury bonds and hold them. Now, if you look at the ownership, it's obviously all around the world. So what's going on overseas is also having an impact on here. Do you think that's also a messing with the recession signal of the curve?
Cris deRitis: I think that's probably second order.
Mark Zandi: Second order.
Cris deRitis: So I think Fed treasury is first, first order, and I say it's second order because of the influence of the US economy and the world economy. So the US economy is the most important driver of world economic growth. Yes, it's the size of the economy is smaller compared to the rest of the world compared to the past, but it's still a very important driver. So what other countries, in terms of their buying of US treasuries, is also correlated with expectations of what's going to happen in the US.
Mark Zandi: Okay. Okay. Very good. Chris, anything else you want to ask in with regard to the nuts and bolts of the yore before we move on to what's the intuition behind why it's such a good predictor? Anything else you wanted to bring up?
Cris deRitis: No, I think-
Mark Zandi: No. Okay.
Cris deRitis: Yeah, I think we got-
Mark Zandi: Just want to make sure I didn't miss anything. Let's turn to that question. And there's a couple three, and I'm sure there are more explanations for why, what's going on here that I guess the most obvious is the curve represents the collective wisdom of bond investors who are putting their money where their mouth is. So if a bond investor thinks, "Oh, this economy's going to go to hell and inflation's going to fall, I'm going to buy long-term bonds." And of course short rates are kind of pinned to where they are because of monetary policy. The Fed's got its foot on the brakes, it can't come down or at least can't come down as much. And you get that inversion. Is that your way of explaining why the curve is a good predictor or is it something else?
Cris deRitis: Yeah, did you've hit it exactly. It's a basic hedging argument. So you see there's a problem, a simple way to think about it. There's a flight to quality and that's the tenure. So prices bid up, yields come down, and that flattens or potentially inverts deal curve. So really straightforward.
Mark Zandi: Straightforward.
Cris deRitis: Yeah.
Mark Zandi: Yep. Okay. This is a Zandi, I think explanation and I want to try it out on you, and I've tried it out on Chris, and I think you're somewhat sympathetic to this, but let me play it, play it for you for a second. So when you have a positively shaped yield curve, and generally those are the good times, and when it's really positively shaped the boom times financial intermediaries, banks can make a boatload of money. Their net interest margin, the difference between their funding costs and their lending rates was very wide. They have a lot of incentive to go out and extend out a lot of credit. So you get a lot of credit flowing into the economy, to businesses and to households. Of course, in the boom times and the economy gets to full employment, inflationary pressure is developed, the fed steps on the brakes.
At some point it really steps. Hard curve goes flat, starts to invert these intermediaries. The banks can't make money. Their net interest margin goes negative or down or flat. Their funding costs are greater than the lending rates. And they stop lending, which is really first problematic because credit is necessary to keep the economy moving. But it's really problematic after a period of very strong credit growth because you've got a lot of businesses and households coming back and saying, "Hey, I need to refinance this debt. I can't pay you back." No one ever thought I would pay you back. At this point in time we got to refinance."
And the banks say, "Oh yeah, you can refinance, but now you got to pay me much higher rate. Or the lending terms are much higher, much more significant, the underwriting standards." And so businesses can't afford that. The lending rate's too high, the terms are too onerous, and they say, oh, I got to pull back on hiring. I got to pull back on investment. I can't expand, thus I go into recession. What do you think of that as an explanation for the intuition, monitor the curve. Does that make sense to you?
Cris deRitis: It does make sense indeed. The first paper I presented that the University of Chicago had a mechanism similar to that.
Mark Zandi: Damn. I thought this was a Zandi idea. This is now, gosh, I knew it.
Cris deRitis: So I never publish it. So you can have it, but it's a really interesting idea that as we become really flat, that actually puts pressure on the banks. So when we're really positively sloped, they're making a boat load of money more likely to make that loan to a corporation because they are making all of this money, and when it flattens, it's less likely. And this leads to companies making less investment, less employment, and that feeds into slower economic growth. So I think that that explains some of the mechanism, but it doesn't really explain like the cause. So what is the reason that we're flattening? So given that we're flat, it makes sense that banks become more cautious in terms of their lending, but how do you get to that flattening? So the mechanism that you describe is a credible mechanism that leads to perhaps extra predictive power for the yield curve. It's not clear though, that's the cost.
Mark Zandi: Yeah. I think of the credit cycle kind of driving the business cycle, or obviously it's causality is running in both directions here, but it key aspect of the business cycle in terms of boom bust is credit boom bust. And so the credit flows are driven by the shape of the curve and then net interest margin. And that Helps amplify the ups and downs in the business cycle. That's kind of the causal relationship that I have in mind.
Cris deRitis: And so I think you're right that we talk about causality, but everything really is connected. Everything isogenous here, and I think you point out something that's really important that I'm looking at the treasury yield curve, that if we were doing this job of forecasting real GDP, we would want to look at other information. And probably the number one place I would start is credit. Because if you look at credit spreads, they're also highly correlated with a future economic growth. So there's another variable that you could look at to bolster the accuracy of your forecast.
Mark Zandi: Okay. So collective wisdom of bond investors, maybe some aspect of the credit cycle is any other intuition behind it? Not that there needs to be, but any other kind of causal link or relationship that could help explain why the curve is such a good predictor of recession.
Cris deRitis: There are many different ways to go at this. And we actually tackled two of them the basic hedging argument, I think is the most powerful one. Most And just this idea that interest rates, they contain information about real economic activity that's expected. So again, the foundation is very intuitive and it's not really a surprise that this works. Let me also mention that while I documented the yield curve predicting real GDP, there was an earlier paper by somebody at the Fed in 1965, Ruben Kessel, and he had a long time series of deal curve, and he noticed that there was a cyclical behavior. He didn't link it to forecasting economic growth or anything like that, but he did notice that there were cycles, and that was influential paper for me, which obviously I cite in my dissertation.
Mark Zandi: Got it, got it. Okay. So here we are today. And I just looked the... And correct me if I'm wrong, Chris, the 10 year, three month, I think on an EBY basis upon basis inverted in October of last year. So here we go, November, December, January, February. We're four months in. That's three month moving averages. We are inverted. So the signal that you use is say saying recession anytime between mid this year and kind of early in 2024. Is that right? Is that what you're taking away?
Cris deRitis: That's correct. So yeah. Okay. We end the end of December, so-called code red in terms of this code indicator. So the indicator is forecasting a recession.
Mark Zandi: Got it. Okay. Let me ask you, this is the yield curve. Do you agree with that forecast? Do you think we're going into recession?
Cris deRitis: No.
Mark Zandi: Second half. You do not.
Cris deRitis: I do not.
Mark Zandi: Okay. Okay. Okay. This is blowing my mind. This is blowing my mind. Everything leading up to that said yes, you were going to answer, yes. Okay. All right, Cam, why is this time different? By the way, those deadly words this time is different. Why is this time different? And by the way, Cam, I am so wit on with the same page with you, but go ahead, go ahead. Yeah.
Cam Harvey: Let me just first establish something really important. Yeah, every time is different.
Mark Zandi: Okay. Okay. That's good point. Great point. Yes.
Cam Harvey: The yield curve model that I've got is a very simple model, and it is true that it's eight out of eight, but no false signals. It is naive to think that this model will never produce a false signal. And fair enough, I believe there's a number of reasons why it's producing a false signal at this time.
Mark Zandi: Okay.
Cam Harvey: And I can go through.
Mark Zandi: Yeah, we definitely got to go through them, man. I'm like dying. This is better than, yeah... This is like we got to sell tickets to this podcast. This is really cool. Oh yeah, go ahead. Go ahead.
Cam Harvey: So one thing that's unusual is the employment situation. And we know that employment is a lagging or maybe coincident indicator, and that's not what I'm talking about. I'm not talking about the fact that the unemployment rate is low, it's always low before a recession. It always increases. But what's unusual is the excess demand where we've got the ratio of job openings to unemployed is very high. And what that means is when we slow down, and I believe that the yield curve is accurately forecasting slower growth, just to be clear, when we slow down, there's a buffer and that means we're not going to see a spike. We will see an increase in unemployment, but not a spike in unemployment. And that's one of the first reasons to second guess the forecast.
Mark Zandi: Let me restate just so everyone can get their mind around it. You're saying this time is different because the labor market's different. The labor market is extraordinarily oversubscribed. There's just a 11 million unfilled positions out there that's a record number. And it also reflects kind of the idea that businesses know that their number one problem is retaining and holding onto workers. And in that kind of supercharged labor market, hard to see the kind of layoffs you would need to see the increase in unemployment you would need to see to go into recession, is that roughly right?
Cam Harvey: That is roughly right. And it's even beyond this.
Mark Zandi: Even beyond this.
Cam Harvey: If you dissect the type of unemployment that we've seen it's very interesting because what makes the headline are all the tech layoffs. And those are so different than the types of layoffs we had, for example, in the global financial crisis. So you lose your job at Lehman Brothers, where are you going to go? You going to go to Bear Stearns, you're going to go to one of the banks that are basically looking for a handout. You're facing a very long period of unemployment, as many did during the global financial crisis. These tech workers, you work for an A-level firm, whether it be Facebook, Twitter, alphabet, and you're laid off, those workers have a very low duration of unemployment because they are highly sought for just non-tech corporations.
Many companies would love to have one of those X workers at these A-level places in the technology sector. So if you empirically look at the data, and this is interesting that it appears as if the duration is a little longer, but I think it's purely by choice that, oh, well I'll take some time off. I know with the snap of the fingers I can get a job. Indeed. We're trying to get some of these workers at our MBA program at Fuqua at Duke. They're finally desired.
Mark Zandi: Moody’s too. Yeah, for sure.
Cam Harvey: So I think that that's another aspect. So the duration is lowered the structural makeup of what we've seen in terms of layoffs so I think of this as the first factor.
Mark Zandi: What's the second?
Cam Harvey: The second factor is fascinating to me.
Mark Zandi: And this is all fascinating to me, Cam.
Cam Harvey: This is something totally unexpected and it has to do with the yield curve. And before the global financial crisis, the yield curve was also strongly inverted for a long period of time. So the duration of the inversion was very, very long. Just like the recession was long. And I was screaming code red and nobody listened and-
Mark Zandi: I was with you, I was with you, I was with.
Cam Harvey: Maybe I wasn't screaming too drunk. So it definitely the case the Fed didn't listen and they were so late to the game and or think about it like a CEO or CFO during the global financial crisis, they had to make significant layoffs. Their firm could be in distress and they could credibly say we were blindsided by this. We had no idea this was going to happen. And my peers within the industry, they were also blindsided.
So it was a surprise. So today it's a different story. So after the global financial crisis recession, people started to see, to predict the power of the ill curve. It got a lot of media attention. This is not really my area of research anymore, but I get asked about the ill curve all the time. And I think given the publicity that the yield curve has got, that it's harder for a CEO to say if a recession occurred, well, we were completely surprised. It's hard for the CEO to make a major capital investment and borrow to finance that in the face of an inverted yield curve. Think about it, that making an investment or betting the firm in a situation with an inverted yield curve where you've got a record of eight out of eight and no false signals, you need to think twice about that or major hiring, no, you're not going to do that. You're going to wait.
And this is related to this idea of self-fulfilling our prophecy. So you get the inversion, people see it and say, "Oh, well that's bad news. I'm going to change my behavior. I'm not going to pull the trigger on this investment project. I'm not going to hire a hundred new employees. I'm going to wait and see." And that feeds into the slower economic growth. So that's the self-fulfilling prophecy, and it actually makes the yield curve causal.
So a negatively sloped yield curve could actually cause slower economic growth given that it's now a popular indicator. So what is this number two factor? Well, the yield curve has caused companies to be cautious and to exercise risk management and to take actions now that hopefully will protect them in the future for the extreme downside. So it's better to take actions now to slow economic growth and to reduce the probability that you need to take drastic actions in the depth of a serious recession. So that is that it is. So this is again, fascinating to me that there's a causal a link here. The predictive power is still there, but this risk management reduces the probability of a hard landing.
Mark Zandi: Interest. That is fascinating. So you're saying, "Look, the fact that everyone is focused on this as an indicator and is predicting recession means that they've become more cautious. That will allow the economy to kind of cool off and not experience the boom and bust that it typically does because it called off in anticipation of all of this." That is fascinating that the self-fulfilling aspect of it is actually going to reduce the odds that the economy actually goes into recession because people are responding earlier than they typically would.
Cam Harvey: Or at least a hard recession.
Mark Zandi: Hard recession.
Cam Harvey: It could be like a soft one, like 2001, which is not really a big deal. You might not even have a year over year it that's a-
Mark Zandi: That wouldn't have even been a recession. I don't think it without nine 11 right? Probably. It might not have been. Yeah. Yeah. So anyway, interesting. Okay. And now I'm cognizant a little bit of time, so I want to make sure I get through all the reasons. Is there a reason number three?
Cam Harvey: So there are multiple reasons but-
Mark Zandi: Oh, goodness.
Cam Harvey: Yeah. Let's just do one more reason.
Mark Zandi: Okay.
Cam Harvey: So if you look at the global financial crisis, recession housing was a big part of it. And if you look at equity to debt in the housing market, it looks sharply different than before the global financial crisis where there's much more equity. So even if housing is down, but even if it goes down further, that's not going to trigger a big problem nor, and reason number four, our financial system is sound in my opinion right now. So in the global financial crisis, that was actually the cause of the problem. And given that it's much more prudent today, I think it's less likely that accelerates any issues.
Mark Zandi: So the economy's on sounder of fundamental ground than it is typically before recession, therefore no recession. Yeah. Hey Chris, does this all sound familiar to you? I'm just asking Chris, this is sounds, Chris is, Chris is a true believer in your yield curve and saying he's relying very heavily on that as a predictor, future recession, anything to say with, I'm turning back to you, Chris, anything you'd like to say or push back on?
Cris deRitis: Oh gosh. So I'm fully on board with the idea that the next recession would be mild for the reasons you outlined here. And actually my question to you was around the self-fulfilling prophecy aspect. And I can see that certainly firms are acting much more conservatively today in anticipation. I think it's a very calibration though, right? Because they're pulling back on that investment. If they pull back too far of households back pulled back too far, then you will have the recess, right? So it's got a little bit of a dance there. But I do agree that there's very little evidence of access or that would lead to a very significant recession.
So I still use the yield curve certainly as a signal. I find it difficult to ignore completely. But I combined with some other factors, I still see that there is a significant risk. And we're right on that edge as you put it, that yeah, households are in pretty good shape. They have a lot of equity, but there are some cracks in consumer credit. Housing build home building is weak. So there are some areas where a little bit of a shock could certainly tip us into recession, in my opinion.
Mark Zandi: Yeah, I don't think we're that far apart here. So I totally agree there's risk and I believe that growth will slow. So I think the yield curve is accurately forecasting that. I just think the probability of a hard landing is pretty low given the economic scenario right now. That said, the big well card, in my opinion, is what the Fed is going to do, and the Fed could make the model nine out of nine.
Cris deRitis: No, it's
Mark Zandi: It's interesting. So do you think the Fed is responding to the yield curve as well?
Cris deRitis: No. So I think the Fed is using this very blunt instrument, the Fed funds rate, and this blunt instrument, given what they're doing, thinking that just raising the rate is going to erase inflation could drive us into a hard landing.
And we all know that the Fed was very late to the game where we had essentially zero interest rates for an extended period of time that didn't make any sense or you've got strong economic growth, you've got low unemployment, you've got record stock market, and the rates are very low or zero, even though inflation was increasing, they're late to the game and I believed that they will be late again. They will overshoot and there's evidence that they're already doing this, in my opinion. And I think that one of the major problems here, you mentioned housing, that is the problem, that it is fairly intuitive that housing inflation takes a while to make it into the CPI.
Mark Zandi: Yes.
Cris deRitis: If you think about, "Okay, rents, let's say go up by 10%." Well, if you've got a lease for the next 11 months, you don't feel that right until you have to renew in 11 months. So that's exactly what we had. If you look early on, you see the rents going up, the housing prices going up, and it takes a while to work its way through the CPI and then the Fed response. It's the same thing now that the rental component, the shelter component is 40% of the PCE Deflator, 33% of the CPI. And you can look at the data, you can see the rents coming down, you see the housing prices coming down, you see new construction going down, permits going down. All of this is consistent with this really important component coming down over the next six months. Yet the Fed is perhaps going to do 50 basis points next time, and that could be enough to push us into the recession.
Mark Zandi: It's very interesting when I asked you for the reasons why the yield curve is falsely predicting recession, I think that of broadly speaking, there's two sets of reasons. One are fundamental reasons, the labor market, the housing market, that the things that you mentioned. The other are what I would consider to be more technical measurement issues. I mentioned the QE QT, we talked about global investors increasingly in the marketplace.
The third I want to throw out there is that the Fed over time through business cycles have become increasingly clearer with regard to the path of future monetary policy that their flow, so-called forward guidance is becoming clearer and clearer and clearer. And now they're like crystal clearer is crystal clear as you can be. And that is also influencing the shape of the curve in the future more so than it has in the past. And it's interesting, you went to the fundamental reasons, you didn't go to the technical reasons, or is it that you don't think those technical reasons are, as you said, they're just second order kind of reasons. They're not by themselves sufficient to make the curve less predictive of future recessions. Does that make sense?
Cris deRitis: So I think I did use the word second order, and I also use the word noise.
Mark Zandi: Noise.
Cris deRitis: Because a lot of stuff happens that, again, this is a very simple indicator and other stuff will happen. And some of it maybe first order, some of its second order and it has the ability to mess up the productive power.
Mark Zandi: Yep. Interesting. Well, Cam, if we go into recession, how embarrassing would that be for you?
Cam Harvey: No, it's just, when he win tells you. Yeah. So I thought about this a little bit and again, this is not-
Mark Zandi: Of course that was a tongue in cheek question.
Cam Harvey: I got it. I didn't, yeah... It's just tongue in cheek. Yeah.
Cris deRitis: So I'm talking to you about the yield curves, predictive power. But this research is something that I did many, many years ago and I've moved on to a different area. But again, I think we need to look at this scientifically. So this is a model that I proposed and the model's done very well in terms of predictions and lack of false signals. And my job is not to just support the model. Yes, it's my model, but I'm a scientist. And any model is going to be eventually wrong. It has to be, it is a simplification of reality point. And that simplification is going to fail at some point. So I'm not going to be embarrassed. So if it works or it doesn't work, there's no embarrassment.
Mark Zandi: If you're right-
Cris deRitis: This is science.
Mark Zandi: If you're right, you will become the Oracle. You will become the Oracle you are now the Oracle of the yield curve. When do I listen to it and when don't I listen to it?
Cris deRitis: Yeah. You'll care about whisperer.
Mark Zandi: You'll be the yogurt whisper. That'll be a very good spot to be in.
Cris deRitis: Yeah, indeed. I did a podcast that they called exactly that.
Mark Zandi: The yogurt whisperer. Oh, actually, maybe that'd be a good one for the... Maybe we can't steal that, but that's a good one. Well, Cam, we took the hour and I really appreciate it. It was a fantastic conversation and really put things into clear relief, so thank you for that. And Chris, any parting words that you want to say? Well-
Cris deRitis: Stay tuned. How about stay tuned.
Mark Zandi: Stay tuned, stay tuned. Those are good parting words. Yeah. Well, Cam, thank you so much and please, we'll definitely have you back down the road here to see how after this thing all plays out. So thanks again. Appreciate it.
Cam Harvey: Well, thank you for inviting me, and indeed, I recommend this podcast to my students, so-
Mark Zandi: Oh.
Cam Harvey: Keep up the good work.
Mark Zandi: Thanks so much.
Cam Harvey: Thank you.
Mark Zandi: And you heard that, dear listener, we've got a fan. Hopefully, you're fans as well. And we'll talk to you soon. Take care now.