# OT - yield curve inversion

There has been a lot of talk of yield curve inversion lately.
Sounds interesting, how do I do that?
Which immediately leads to some confusion: which term gap? and for how long? what does that portend, and when?

I found this little interview snippet quoted below from the fellow who did the original research, Cam Harvey.

Executive summary: stick with the 10 year minus three month. You can use 5 year minus 3 month if you like but it’s pretty much the same result.
If that gap stays negative for a month or so (not just a brief blip), expect a recession starting 12-18 months later.

inversion, there are always arguments about whether it’s the 5-year/2-year
slope, 10-year/3-month, 10-year/2-year, you know, pick your slope.
What have you found?

A: I think we need to go with the original research. In 1986, I looked at two parts
of the yield curve, the 5-year note minus the 3-month bill, and the 10-year bond
minus the 3-month bill. The crucial thing is to use a very short-term interest rate.
Those two yield spreads are highly correlated. I get nervous when I see people
talking about the 10-year minus the 2-year. I got all these emails last week about
the 5-year minus the 3-year inverting, saying “Happy Yield-Curve Inversion Day!”
An investor can data mine to find a piece of the yield curve—maybe it’s the
11½-year minus the 7¾-year—that has an inversion. That doesn’t mean anything.
I think we need to go with the original un-data-mined theory, which is based on
a longer-term rate and a 3-month rate, and the inversion importantly needs to
last for a quarter. If it’s a day, so what? GDP is measured quarterly, so we need to
measure this quarterly also.

The update: perhaps surprisingly, it hasn’t gone negative yet this cycle.
Last time it turned negative in any lasting way was around late May or early June 2019.
It very nicely predicted the pandemic recession, which is pretty impressive.
It’s still batting 1000 out of sample since 1988.
Predicted every recession, and hasn’t raised any false alarms.

In a way it’s a shame that knowing the date of a future ecession is not the same at all as knowing the date of the next bear market.

Jim

40 Likes

https://fred.stlouisfed.org/series/T10Y2Y – currently negative
https://fred.stlouisfed.org/series/T10Y3M – currently positive, heading toward zero.

Click on “Max” to see the 10Y-2Y series back to 1975. Too bad it doesn’t go back to 1970 in order to see the nasty 1973-74 bear market.

https://fred.stlouisfed.org/series/DGS3MO

https://fred.stlouisfed.org/series/DGS2

https://fred.stlouisfed.org/series/DGS10

Wendy

5 Likes

Last time it turned negative in any lasting way was around late May or early June 2019.
It very nicely predicted the pandemic recession, which is pretty impressive.

By my tracking it turned negative on 5/27/19 and back to positive on 10/14/19. It flickered negative again in Feb. 2020.

That’s hardly a valid prediction of the Feb.-Apr. 2020 pandemic recession which came well before the expected 12-18 month lead time. I suppose your characterization of pretty impressive is tongue in cheek.

Elan

1 Like

https://fred.stlouisfed.org/series/T10Y2Y – currently negative
https://fred.stlouisfed.org/series/T10Y3M – currently positive, heading toward zero.

I guess the key point of my post is that the person being interviewed, the godfather of the metric, says look at the second one NOT the first one.
He emphasized that the underlying theory requires the that short end be very short, or you’re measuring something else.
In short, pay no attention to the ten-minus-two.

As for whether the 10-year-minus-3-month is headed towards zero–well, it’s lower than it was, for sure.
But the graph does not have momentum like a truck full of bricks.
Maybe it will keep falling, but on the other hand maybe not.

If it were easy to predict such things, there would be no point using it as an indicator.
It ain’t negative for multiple weeks till it’s negative for multiple weeks.
Unless and until that happens, the bear market signal isn’t there.

In short, I think your “heading towards zero” comment is, as yet, just a speculation.
Maybe an astute one–I tend to agree–but still, a speculation.

Jim

10 Likes

That’s hardly a valid prediction of the Feb.-Apr. 2020 pandemic recession which came well before the expected 12-18 month lead time.
I suppose your characterization of pretty impressive is tongue in cheek.

Not at all.
The lag is quite variable–the 12-18 month part is just the most common range, not the key insight.

I think it’s astoundingly impressive.
To date, the signal has had precisely zero false positives and zero false negatives in 34 years out of sample.
That’s perfection: it literally could not be any better.

Nothing else comes close.
The next best would likely be a “coincident indicator” that gives you no advance warning at all.

The biggest real world problem with the signal in terms of utility is that it predicts US recessions, not bear markets.
To profit from it directly would require something like trading GDP futures.

Jim

7 Likes

<In short, pay no attention to the ten-minus-two.>

Thanks, I know how to read a chart. When I see 100% correlation, I pay attention.

Wendy

2 Likes

Thanks, I know how to read a chart. When I see 100% correlation, I pay attention.

I get that, but two comments;

• One has 34 years of out of sample validation, the other doesn’t.
• At the moment one is signalling an upcoming recession and the other isn’t. So there definitely isn’t 100% correlation.

Maybe you know how to read a chart, but I know how predictive models are built. : )
If your cookies or IP address profile you or locate you in a way that can be validated, it’s probably using a patent with my name on it.

Jim

27 Likes

Both the stock market and yield inversion are leading indicators of recession. Sometimes the stock market tops before the inversion. For example, in 1973 there was a January market top and a July yield inversion.

The quarterly average was important when yield inversion was originally formulated. (It is not a one-day signal, but instead uses a quarterly average.)

Bear markets are worse when there is a recession. So, the yield inversion might provide insight into the size of the drawdown. Sorting the last table in the post linked below, by months:

``````              Mkt
observation  CAGR1  market top   days   months
31-Aug-69   -15%   29-Nov-68    -275     -9
31-Jul-73   -26%   11-Jan-73    -201     -7
31-Aug-00   -27%   24-Mar-00    -160     -5
31-Dec-80    -3%   28-Nov-80     -33     -1
31-Jul-89     3%    4-Jun-90     308     10
31-Aug-06    15%    9-Oct-07     404     13
31-Jan-79    25%   28-Nov-80     667     22
31-Dec-76   missed
25-Aug-87   missed
``````

CAGR1 is for the 12 months after the yield inversion event.
months is number of months from yield inversion to market top.
(months is negative when yield inversion is after the market top.)

If there is a yield inversion this quarter, the market might fall another 15% to 30% over 12 months. Yield inversion looks likely given the FED’s determination to stop inflation. Iinflation is a worldwide event with macro underpinnings (war, pandemic), and so is not going to be easy to tame this time around.

Inverted Yield Curves and Stock Returns
by Campbell Harvey, Published Jul 19, 2019

"The paper defines yield inversion events using a quarterly average of 10 year minus 3 month Treasury interest rates. There have been 8 events in the last 50 years, with the latest [in 2019]…
4 bull market tops were 1 to 9 months before yield inversions.
3 bull market tops were 10 to 22 months after yield inversions.
2 bull market tops do not line up with any yield inversions.
https://discussion.fool.com/i-repeated-the-analysis-presented-in…

11 Likes

Not at all.
The lag is quite variable–the 12-18 month part is just the most common range, not the key insight.

I think it’s astoundingly impressive.

Yeah, okay, the main objection I have is not the twelve months but the prediction itself in this case. Many factors can be encompassed by this yield curve measure, which ultimately correlate with recessions, such as fed policy, long term inflationary expectations which guide long term interest rates, etc. One thing no one could predict, not the Fed, not the most astute observers in May 2019, was that six months later a deadly virus would emerge in a wet market in Wuhan and engulf the world in a pandemic that would bring life to a halt. An astoundingly impressive prediction it is not. An astounding coincidence it is.

We’ll never know if, in the absence of the pandemic, there would have been a recession in 2020 anyhow, but the recession that did occur was not predicted by the monetary indicators. I don’t care what anybody may think to the contrary.

Elan

6 Likes

We’ll never know if, in the absence of the pandemic, there would have been a recession in 2020 anyhow,
but the recession that did occur was not predicted by the monetary indicators.
I don’t care what anybody may think to the contrary.

Your position has a bit of an internal quirk–
While asserting that one can’t know what would have happened absent the pandemic, you go on to assert with certainty what would have happened–no recession : )

I think it’s much more parsimonious assume that a US recession was likely going to happen around 2020-2021 anyway.
In my mind the only uncertainty was the timing.
In 2019 there were already lots of cracks visible in the system, with end-of-days style signs of extremes everywhere at the end of the longest economic expansion in memory.
And, yes, the bond market was yelling at us to pay attention.
10 year treasury yields had already fallen from over 3.2% to under 1.5% before the pandemic rout started.
If the pandemic had any effect on that probably inevitable recession, it was probably to fine tune the timing.
It came on more suddenly, and was much weirder.
But most importantly it was probably very much shorter because of the ocean of cash added to the mix.

Ascribing specific causes to market and economic moves is a very doubtful business.
I bet you can’t spot Sept 11 on a graph of the S&P closes on trading days in 2001 without the dates marked.
(I looked at this graph yesterday, then just tried it today, and was still wrong)
In this case, all we know for sure is that the prediction from the yield curve was right, out of sample, once again.

Jim

22 Likes

Please, Motley Fool, do not close down these remaining message boards. A thread like this is the reason I’ve been here for 20 years. I don’t know of any other online community where I can enjoy such great discussions about economics and investing.

Thanks, all you great posters!

22 Likes

Agree 100% percent Ges. After a period where stock investing looked like a sorcery, I finally saw some green, and some of the MI signals talked about by Jim and Zeelotes seemed to do exactly what they said they would do.

Thanks again,
ges

1 Like

Thanks Jim. I note that you very wisely stated that while a yield curve inversion indicates recession, it may not indicate the date of bear market…

Naïve question, but is it possible (hopeful of course):), that the bear market may precede the recession, or occur in the anticipation of a recession, such that the market falls, then recession becomes official, and then market recovers and runs in the positive direction…of course, this is purely momentum investing, but still would be nice to know if there has been precedents for such events, where a bear market precedes recession, and then the recession occurs, and counter-intuitively triggers a market rally (a true one rather than bear market rallies)

For example, since the talk was about recession for more than a few months now…, and both the S&P500 and QQQ had fallen by 20-30% (and recovered a bit recently), is it possible that the worst has come and gone… Would be nice if true and hopefully this is more than just wishful thinking ):

1 Like

Your position has a bit of an internal quirk–
While asserting that one can’t know what would have happened absent the pandemic, you go on to assert with certainty what would have happened–no recession : )

I didn’t say that. I said that it was utterly impossible for the monetary indicator to predict the recession that did happen.

Elan

1 Like

is it possible (hopeful of course):), that the bear market may precede the recession

Recessions are a lagging indicator, and are not called until several months after the actual recession. The US might be in a recession now, we will know in about a year. Most bear markets end before the recession ends. The pattern is:
bear market starts (market top)
recession starts
bear market ends (market bottom)
recession ends

But the recession dates are not known until several months after the event. The current recession might be over. There might be some indicator that the recession is over, but the official dates are not much use in identifying stock market bottoms.

“Seven of the last ten recessions started soon after a stock market top, with an average lead of 7 months (range 1 to 13 months). Stock market tops did not lead three recessions (1953, 1960, 1980 recessions). There were four stock market tops that did not lead recessions (1961, 1966, 1976, 1987 market tops).”
https://discussion.fool.com/bear-markets-and-recessions-32819220…

5 Likes

If your cookies or IP address profile you or locate you in a way that can be validated, it’s probably using a patent with my name on it.

Wow … that’s impressive indeed. However, I’m loathe to thank you for your contribution to this outcome.

Tom

4 Likes

However, I’m loathe to thank you for your contribution to this outcome.

Yes, well, I’m a little depressed myself that my net contribution to society could have been better : )
I justified it at the time with the reasoning that hey, you’re going to get an ad anyway, you might as well get one that you might conceivably relate to.
I was also in the mailing list business (physical and email), which goes hand in hand with profiling.
(most valuable search keyword in the late 1990s: mesothelioma, purchased by the asbestos law suit ambulance chasers)

Having seen how the sausages are made, the main result is that I was left with a desire not to be profiled myself.
I don’t have a smartphone, and don’t have accounts with Facebook, Google, Instagram, Uber, or anything similar. I don’t use Chrome.
Even the dumb phone I carry is in someone else’s name (by accident originally, but I decided to keep it that way).
My iPad has almost no apps, and I don’t use iCloud.
It’s super easy to profile me using my posts here, but otherwise pretty difficult.

Jim

16 Likes

Sadly, that hasn’t stopped those folks from creating profiles of you anyway. Esp. if you have ever emailed someone with, e.g, a gmail account. My phone is too much my vade mecum to revert to dumb phone-ness but I also don’t have accounts with the obvious social media. It is a tough tradeoff.

Rgds,
HH/Sean

2 Likes

Sadly, that hasn’t stopped those folks from creating profiles of you anyway. Esp. if you have ever emailed someone with, e.g, a gmail account

Oh, I’m pretty up with how it works.
I’ve never sent a gmail message nor had an account with which to do so, nor would I let anyone log into gmail on one of my devices.
My browser deletes all cookies after each session.
They have no doubt been fingerprinting my iPad, (and trying to do so on my PC), but unfortunately there’s not very much I can do about that.
Consequently I would never use my iPad to search for anything personal.

My biggest beef is mandatory government web sites which can not be used without allowing profiling by commercial firm.
Same with mandatory cashless transactions that require giving all the transaction data to commercial profilers.
If I choose to make the deal with the devil to trade my information for freebies, as pretty much everyone else does, that’s my decision.
And if the government wants to track me or my money movements, well, they’re the government. They have the power, and (surprisingly) I generally trust them.
But no government should make private company profiling mandatory.
It is the responsibility of every government to ensure there is a means of payment (cash, or CBDC)
for every transaction which does not require one to allow tracking and profiling by an entity OTHER than the government.

The Europeans are real keeners on data privacy–in the press releases, anyway.
The FT did a study of European government portals recently, and IIRC there were not dozens, but hundreds of private sector tracking devices encountered.
If Amazon or Walmart or MonsterVibrators.com or whoever wants to use Google Analytics to track and profile me and my location, that’s cool.
If I want to use those web sites, it’s the cost of playing. I can always just not use them, in theory.
But the drivers’ licence bureau? The passport office?
They found private sector tracking widgets on sections of government help sites specific to certain diseases.

Jim

7 Likes

Thank you for recommending this post to our Best of feature.