Yield Curve Inversion as a Signal of Recession

Using the yield curve as a tool to forecast recessions is ... popular again ... but still probably for no good reason.

First off, recessions aren't really predictable at all. This is what makes them a problem. 

FWIW: forecasting recessions is arguably the single most studied problem in human history. If it had an easy answer, don't you think we'd just tell you?

Second, the yield curve is kind of mysterious. Because it isn't a number. It's a curve. How do you summarize it quickly? I know what it means to say that a number X causes a number Y, or that when X goes up by 1%, Y goes up by 3%. I don't know what it means to say a curve (or the shape of a curve) causes a number. Huh?

***

A lot of people don't even know what a yield curve is, in the sense of how you'd put one together if you were doing it by hand. 

Here goes. First, you'd need to have securities from which you can measure interest rates. Second, you need those securities to have different maturities. Third, they need to be securities on the same sort of thing, because rates can differ across different securities depending on risk, preferences, senior vs. junior status, bankruptcy features, covenants, and so on.

When you think about this, you can't really form a yield curve from say, car loans. Yes, they do have different maturities. But they also have rates that depend on buyer features (like income) and seller features (like how badly they want to get the car off the lot).

Ditto for mortgages. Ditto for corporate bonds, unless you found a corporation that issued debt at a lot of different maturities, and was representative of the whole country.

You're really just looking at government bonds then. So if you were to put together one for the U.S., you'd need to get rates on all the bonds of all maturities. At least this route is possible.

Then you plot, for a point in time, all those rates on the vertical axis, against their maturity on the horizontal axis. If you do you usually get something that's upward sloping.

Inversion is when it's downward sloping.And people on the internet usually start squawking about inversion when any part of the yield curve is downward sloping, even if overall the whole thing seems to be generally upward sloping.

***

A better way to summarize this information is with a gap, usually between the rate on a long maturity bond and a short maturity bond. Usually the latter is lower. An inversion would be when it's higher.

You can actually get this for free from FRED using the keywords fred, yield, and curve in Google. It will be listed as something like: 10-Year Treasury Constant Maturity Minus 2-Year Treasury Constant Maturity.

BTW: you can "get" this because it's a number. How would you "get" a curve to download?

If you look, sure enough, it "predicts" recessions.

But what sort of prediction is it? First off, it screams recession over on the left, but nowhere else. The early 80's were bad, but comparable to 2007-9 where it didn't scream recession. Second, the time before a recession is not the same, and it's not that close either. Third, it forecast the 2020 recession, which everyone is pretty sure that was about a pandemic and lockdown. So are we to believe it forecast those too?

It's better to view the gap as reflecting Fed policy. The Fed has much more control over short rates than long rates. And the Fed tends to start "tapping the brakes"  by raising rates late in an expansion. So, if they can't budge the long rate, and they start raising the short rates, the gap will go towards zero ... late in an expansion.

That explanation is a good one. The thing is ... the yield curve doesn't add anything on top of that. So what's the point?

Cynically, I think the point is that it's one more piece of data that doomsayers can crow about. It's up to you to decide if it's one more valuable piece of data.