The interest rate on the U.S. Treasury 10-year bond just fell below the rate on the 3-month bill in response to the Fed’s March announcement. This is called yield curve inversion as defined by Arturo Estrella and Frederic Mishkin. It implies a 25%-30% probability of a recession on a 12-month view. Their research can be found here.
As economic relationships go, the yield curve has a good track record. You can see the data below going back to 1982. Per the chart, using this series over recent history, the yield curve inverts before a recession reliably with no false positives. An impressive record. The blue line shows the spread between 10-year and 3-month interest rates. The black line is the zero bound. The shaded grey periods are historical recessions. Note that there is a lagged relationship here, recession historically occurs 6-18 months after inversion. So today’s yield curve suggests a fair chance of a 2019-2020 recession.
Risks Of Interpretation
Nonetheless, there are some risks with this approach. The first is we’re looking at a limited run of data. There are only a few decades in the sample and a handful of recessions. We’re making a forecast here based on less than ten recessionary events, per the initial research and subsequent out-of-sample data. Plus there are countless pieces of economic data out there. Combining two of them and creating a good recession forecast is possibly due to data mining. For example Tyler Vigen’s site illustrates the problem, showing how correlations can be found between many things that have little basis in reality, such as an apparently strong relationship between mozzarella cheese consumption and sociology degrees. So even though the yield curve relationship looks robust, it has been plucked from hundreds of other relationships that could exist, but don’t look as meaningful. The human brain is adept at creating patterns where none exist.
Also, a 25-30% chance of recession is not that high. Going back from 1960 to 2018 we have 59 years of data. We’ve had U.S. recessions during 16 of those years. So even before any more sophisticated forecasting methods, your chance of being in a recession in any given year are 27%. There’s some auto-correlation there too, as recession years come in clusters, but still, saying the chance of recession coming within a year or so is around one in four isn’t that different from what history tells us regardless of what the economy is doing. Of course, even at a 30% probability the chances are roughly twice as high that a recession does not occur.
YOU MAY ALSO LIKE
Also, the Federal Reserve (Fed) has a key target of avoiding a recession in order to maintain full employment. This one of their key policy targets. The Fed are quite capable of learning. The increasing emphasis on the yield curve has not gone unnoticed by the Fed. In fact, the initial research here was published by the New York Fed itself. So unlike in the past, the Fed may be able to take corrective action, which was exactly what Chairman Powell was looking to stress in the Fed’s March meeting release. The Fed’s challenge is obvious but not simple, a few quarters ago the markets were spooked by a potential stack of rate hikes in 2019 that could risk recession, so the Fed changed course. Yet, in doing so they have helped create an inverted yield curve, introducing another set of recessionary fears.
However, the risk here is the markets are capable of learning too, and there is some evidence that recessions are self-fulfilling, meaning that if enough decision makers expect a recession they may then take the very actions actions, such as temporarily cutting back on spending, that cause a recession to happen. In that light, yield curve inversion gaining more attention is bad news if it causes people to anticipate a recession, which then makes one more likely.
So yield curve inversion is not a positive sign for markets, but we may be overstating its importance. Also if the indicator is to be believed, we should watch out not just for inversion, but when the 3-month yield falls 1% or more below then 10-year yield, then our confidence in a recession around the corner should be quite a bit higher.
Articles educational only, not intended as investment advice.