Here's something to worry you if you're the worrying sort:
You've probably heard that it's bad news when the yield curve inverts—that is, when long-term interest rates become lower than short-term rates. This is not a normal state of nature. However, the real danger signal is when the yield curve inverts seriously and then heads back up to positive territory. If recent history is any guide, it means a recession is no more than months away.
Well, it just happened after the biggest inversion since 1980. Let's hope that history isn't much of a guide, OK?
Of course the yield curve is un-inverting - everyone knows a federal funds rate cut is imminent in a week, so the bond market has been pricing that in and pre-emptively slicing yields on the 2-year Treasury note for the past month. But longer-term inflation forecasts haven't changed significantly, so the bond market hasn't been cutting yields on the 10-year note in lockstep with the 2-year - instead the decrease in yields on the 10-year has been modest enough that the 2-year has finally caught up to it.
I'm skeptical that the un-inversion of the curve is really that reliable a leading indicator of recessions in the US, especially with the Fed poised to get out ahead of things and cut rates right now, but I guess we'll find out.
I'm also skeptical. I think there are signs that this time is structurally different. Inflation situation was different, unemployment was different, there were major stressors on the financial system in those previous instances (of which there are only 3!!). It sure looks like those previous instances are mere coincidences.
Exactly. This number is usually in the positives, so how is it going to get there without flipping back over?
Well, as Samuelson said more than 60 years ago, "the stock market has called nine of the last five recessions." And yeah, bond market, so what; the remark has been widely applied for decades.
bw-exactly. The inversion is just a clear manifestation of the next cut rate. The only question is - is it 25 or 50 basis points. The chart suggests that the expectations are 50.
No way they give us 50 bp after that inflation report. But we will get 25, which will help.
Let's hope that history isn't much of a guide, OK?
I really don't think a recession can be avoided much longer, and so I'd just as soon get one soon. I don't think the economy will weaken rapidly enough to hurt Harris's chances—if anything falling inflation, falling mortgage rates and falling gas prices—all consistent with a weakening economy—will probably help her.
And if we are going to have one—and I'm modestly optimistic after that debate that she's going to win—better for her to inherit a recessionary economy from the getgo than get gifted with one in 2026 or 2027. The average downturn in the postwar period lasts less than twelve months. I think the (moderate) debt picture means this one's going to be relatively mild.
I really don't think a recession can be avoided much longer...
A recession may be coming soon, or not. But saying one can't be avoided "much longer" makes it sound like recessions happen on a schedule. They do happen periodically, but unpredictably.
Recessions in recent decades happen much less frequently than in earlier times. That's likely the result of better data and better policy. But it means we should be careful to set expectations today on what happened too far back in the past.
Still, if we look at the past 5 recessions (since 1981), here's some data points.
Growth Cycle Duration (between recessions)
Average: 8.5 years
Shortest: 6.75 years
Longest: 10.5 years
Current: 4.25 years and counting
Yield Curve Inversion Prediction (start of inversion to recession)
Average: 14 months
Shortest: 5 months
Longest: 23 months
Current: 29 months and counting
If you look at the average duration between recessions, a recession now would be remarkably soon. We wouldn't be "due" for another over the next 4 years.
If you look at the average duration from yield curve inversion, a recession is long overdue. It's never taken this long. We're more than double the usual duration.
I think the safest thing to say is that the usual indicators are not working as they used to. Perhaps we are in a new era.
But saying one can't be avoided "much longer" makes it sound like recessions happen on a schedule.
Not to me. I'm simply referring to the fact that current conditions (in my view anyway, your mileage may vary) suggests we are indeed approaching the end of the current expansion.
I guess my view is different. As I see it, the current expansion has a long run ahead, though it could end early through bad policy (something worse than what the Fed has done, or if we get a Maga trifecta).
My two cents is that the pandemic slowdown and extraordinary stimulus has distorted what's "normal," and economic indicators like inversion of the yield curve and Sahm rule are not working (at least for the time being).
My focus is more on the stock market, anyway. It looks like a good risk-on environment to me, especially with today's big reversal, a bullish signal. That doesn't jibe with a recession right around the corner. If there is one, it's more likely to be mild, and we may not hear about it till it's over.
Lots of retail stores closed this year. Also… restaurants like red lobster. Layoffs started, but no one noticed?
Even in the best of economic times, some businesses fail; conversely, even in the worst of economic times some businesses will prosper.
History is always a guide. Until it isn't. Ask the Boston Red Sox. Or the Chicago Cubs. The more useful metric would be the performance of the economy the last time we had inflation spurred by a world-wide pandemic followed by global supply chain disruptions. Oh, wait...
Looking at data is a good way to approach the problem of prediction. Get a sample of previous results. A sample size of about 50 ought to be sufficient to get a good estimate of mean values and standard deviations. Then plug those numbers into your prediction model.
How many previous samples did you say we have so far? What's that you say? Three? Hmmm ... I might be reluctant to jump to conclusions based on a sample of three data points over the past 35 years.
And the 08 "inversion" is barely a blip.
The infallible indicator here is a person named George Bush in the White House-- 1990, 2001, 2008. And this indicator shows an inversion-recession sequence at the rate of one per G. Bush term.
The upshot should be crystal clear: For the good of the country, George P. Bush needs to be kept out of national politics by hook or by crook. Forcible detention if necessary.
So it's bad if it inverts and bad if it un-inverts? I think this could use a little proofreading.
No, Kevin's text makes sense if you keep tabs on the investing/personal finance press. His point is that the conventional wisdom is that an inversion is bad, but that conventional wisdom doesn't really accurately capture what the experts have to say about it, as recessions tend to start well after the initial inversion of the curve. The *expert* view is that an inversion is a bad sign specifically because eventually the curve must un-invert, and once that finally happens a recession tends to be imminent within a few months.
There is still a large amount of debate about whether the experts are actually correct - maybe the curve crashing back to normal only makes there appear to be a strong correlation between un-inverting and recessions, which perhaps have some other cause entirely - but his point is pretty clear about the general consensus (even if that consensus is wrong).
https://fred.stlouisfed.org/graph/?g=1tE80
If you tracking rate inversions as a signal of an oncoming recession before election day, wouldn't you track the 10 year - 3 month? And don't you think these trackers mostly just reflect the Federal benchmark rate?
I have no idea what this all means. And reading the comments, I’m not convinced that anyone else does
Silly you, looking for "meaning" in predictions based solely on data.
As a general rule:
(1) You can use data analysis to try to discover data relationships.
(2) Then you formulate a theory as to why it happens that way.
(3) Then you test your theory by experimenting (or by continually observing nature) to get a sample of subsequent events to see if they conform to your theory.
(4) After all the above, if you have some good validation data showing your theory does, in fact, predict with some effectiveness, then you publish it as your theory, rather than just a compendium of data.
So far, in this endeavor, the economists have reached a point somewhere along the way in step 1. Their observations seem to show some correlation between variables (inversion/reversion and recession). Even that, of course, is suspect, given the paucity of sample size.
👍