Does Past Success Guarantee Future Results? The Yield Curve

Laura Taylor: Hello and welcome to the St. Louis Fed’s Timely Topics podcast. Today, I’m here with Chris Neely, Vice President and Economist at the St. Louis Fed, and we’re here to talk about the yield curve. Now, that’s a topic that’s certainly been getting a lot of attention lately, so we’re going to dive right in. Chris, welcome to Timely Topics.

Chris Neely: Thanks, Laura.

Taylor: So, okay. Chris, the yield curve, talk to me about the yield curve and why it’s making headlines.

Neely: Well, although it isn’t exactly new, the most alarming recent feature of financial markets may be the inverted yield curve. That is short rates are now higher than long rates and they have been for some months. This is concerning because past yield curve inversions have reliably predicted recessions, that is sustained downturns in economic activity as defined by the National Bureau of Economic Research. This concerns the FOMC because it has a congressional dual mandate to maintain stable prices and maximum sustainable employment.

Taylor: Oh, boy. All right. We already went into the R-word, recession. So, tell me, Chris, what is a yield curve and what does it mean when a yield curve, like you said, is inverted.

Neely: Well, a yield curve is just a picture of interest rates, or yields, on bank deposits and bonds with similar risk characteristics at a point in time. That is, a yield curve describes how short, medium, and long rates relate to each other at the same time. Short rates are usually lower than long rates. In other words, one-year yields are usually higher than five-year yields, which are usually higher than 10-year yields. So, the yield curve usually slopes upwards if you graph those yields by maturity. When short rates are about equal to long rates, that’s called a flat yield curve.

Taylor: Okay.

Neely: An inverted yield curve is one in which short rates are actually higher than long yields. In other words, an inverted yield curve means that the yield curve is sloping down instead of up.

Taylor: Okay. So, you know, if we’re, kind of, imagining a graph in our heads right now, we can actually see that curve start to go down. Okay. So, if an inverted yield curve has reliably or even historically predicted recessions in the past, you know, why? Why would an inverted yield curve predict recessions?

Neely: Well, there are several stories about why an inverted yield curve might predict recession. Two of the most stories are that Fed tightening raises short rates and this action makes it more expensive to borrow for investment and consumption and, thereby, slows the economy. A second and related story is that interest rates should be related to expected economic activity, and when medium and long rates are relatively low compared to short rates, that indicates that in the future there is going to be low desired investment and lower growth. But, you know, I should note that there’s nothing special about a yield curve that happens to be slightly downward-sloping as opposed to flat. Both predict reduced economic activity in the future.

Taylor: Okay. So, flat and downward? So, how specifically do you measure the slope of the yield curve? So, you know, I mean, I can get out my ruler, but I have a feeling that’s not quite how it’s done.

Neely: Well, that’s a good point. So, there are many measures of the slope of the yield curve, and economists and forecasters do disagree about the best one to use. Some people at the Board of Governors, for example, like to use the near term forward spread, which is the difference between 18-month and three-month interest rates. I follow some economists at the New York Fed and use the spread between the 10-year yield and the three-month Treasury yield. If one looks at a graph of the 10-year yield minus the three-month yield, that is that 10-year/three-month yield spread, one sees that this spread is almost always positive. That is, long rates are almost always higher than short rates. But when the slope of the yield curve declines, that is when the three-month interest rate rises relative to the 10-year yield or equivalently the 10-year/three-month yield spread declines, a recession becomes statistically more likely. A negative spread, that is an inverted yield curve, has preceded each recession since the 1950s.

Taylor: Oh, wow.

Neely: Statistical models that allow the spread to predict recessions in a formal way show that the spread does predict recessions well in the sense that implied recession probabilities rise before each recession. The model isn’t perfect, however. There have been false positives. For example, there were high probabilities of recession when no recession occurred in the late 1960s and late 1990s. But the nominal yield spread is currently quite low and predicts a 65 percent probability of recession in the next 12 months, and this recession probability would be unprecedently high for a false positive. The near term forward spread from the Board of Governors currently estimates about a 50 percent chance of recession in 12 months.

Taylor: Oh, boy. Well, that’s certainly concerning. I’m not sure there’s much positive news there, but, you know, let me ask you this. So, you mentioned two, kind of, false positives. So, is there any reason to doubt the current predictions of the yield spread? So, I think we’ve all heard or read in the headlines that a recession is coming, you know, quote, unquote, in the next six months, but it also hasn’t happened yet. So, could it be wrong? That it’s not predicting that?

Neely: So, yeah. There actually is reason to doubt that the nominal yield spread has its usual predictive power right now. So, the normal practice has been to examine nominal spreads, but there are also good arguments for looking at real, that is inflation-adjusted spreads. In particular, real interest rates should be more closely related than nominal rates, both to the stance of monetary policy and to expected future growth. So, if one does this and one corrects the 10-year/three-month spread for expected inflation at those horizons, using inflation expectations from survey measures, the real interest rate spread also implies an elevated, but lower probability of recession in 12 months. That is, it implies a probability of recession of about 40 percent.

Taylor: Okay. So, lower?

Neely: So, I should note that this 40 percent probability is the highest probability in the history of the series, however, exceeding that even given in an actual recession. So, one interpretation of this discrepancy between the nominal and real yield spread is that the probability of recession in 12 months is somewhat lower than usually claimed. But another possibility is that the real spread variables don’t predict historical recessions quite as well as the nominal spread for unknown reasons.

Taylor: Okay. So, there is some difference between the predictions, if I can call them that, based on nominal and real spreads. So, what’s driving this difference?

Neely: Well, mathematically, the difference between the nominal and real spreads is equal to the difference between 10-year and one-year expected inflation. So, this difference—and by the way, I should note that I’m using one-year expected inflation to stand in for expected inflation over three months. So, this difference in expected inflation over long versus short horizons must drive the difference in how these spread data predict recessions. Since the early 1960s, this difference between the one-year and 10-year expected inflation has been pretty small, usually less than one percent. But it did increase significantly in about 1975 during the first oil shock and in about 1981 during the great disinflation. During these two periods, which each coincided with recessions, 10-year expected inflation was much lower than one-year expected inflation, and the 10-year/three-month real yield spread rose relative to the nominal version, which made the real spread a weaker predictor of those deep recessions. In other words, the nominal spread predicted the 1975 and 1981 recessions well, but the real spread did not predict them as well. So, this difference in historical data means that the real spread’s predictions are not as sensitive to fluctuations in the real spread. That is, when the real spread changes, it doesn’t translate into an increase or decreased probability of recession as much as when the nominal spread changes.

Taylor: I see. Yeah. So, okay. So, what I’m hearing is that the nominal and real yield curves are both flashing recessions signals. So, does it make sense to look at other information? So, what do other recession, you know, potential recession predictors say about the current outlook? And, Chris, I’m looking for some good news here.

Neely: Well, let’s say I have some more mixed news.

Taylor: Okay.

Neely: You know, I’d say it definitely makes sense to look at a variety of indicators because forecasting relations can and do break down. So, past success doesn’t necessarily guarantee future results because the economy is a complicated thing. So, information from other sources can give us greater confidence in or suggest skepticism about the yield curve forecast, for example. So, right now, there are other variables that do imply elevated probabilities of recession, such as measures from loan officers’ willingness to lend or measures of corporate borrowing. But not all potential recession indicators predict the high probability of recession. Many of the indicators that key on real activity, such as initial and continuing employment claims, for example, currently predict fairly low probabilities of recession. The labor market is currently strong, but labor market variables are lagging indicators and don’t predict recession well at all. So, it’s not really clear how seriously we should take those.

Taylor: Okay. Yeah. Interesting.

Neely: But another thing we can do is we can examine professional forecasts, which provides another source of information. So, forecasters, kind of, do this job for us and gather many types of data and synthesize them into a single number. In addition to using a wider range of data, experienced forecasters can make judgmental adjustments to the outcome of statistical procedures to account for unusual conditions, and there are very frequently unusual conditions in the economy. So, Consensus Economics is a firm that surveys economic forecasters. In July of this year, this firm predicted a 59 percent probability of recession in the next 12 months, which is fairly close to recent probabilities from the nominal yield spread model. So, I guess, in summary, I’d say that the inverted yield curve is consistent with restrictive monetary policy. The probability of recession is fairly high in the next 12 months, but it’s certainly not a sure thing. Labor market variables and good shipments, which tend to be lagging indicators of economic activity, imply the lowest probabilities of recession, but then, they usually do lag the rest of the economy or lag other predictors.

Taylor: Oh, wow. Okay. Well, certainly, we have so much information on both sides of this right now. So, Chris, just so interesting about the yield curve. I don’t know about you, but I’m rooting for the yield curve to be wrong this time around. Thank you so much for being on our Timely Topics podcast today.

Neely: Well, thank you, Laura.

Taylor: If our listeners are interested in learning more about Chris’ work or the work of others at the St. Louis Fed, visit stlouisfed.org today. Thank you.