In a crisis, when stocks crash, bonds go up.

At least, that’s the common story. But it actually isn’t true.

How’d did we get to the point where lots (and lots!) of people think it is true? It is mostly “recency bias” along with a helping of “overstating the case”.

Back in 2008, when everything went to hell in handbasket, the S&P 500 dropped -37.02%. At the same time, intermediate Treasuries went up 13.32%.

Exactly what the theory predicted! See! It happened!

Let’s rewind the clock a bit and look at a few other crashes…

Image for post
Image for post

This just looking at annual data. These are all the years since the 1920s when the S&P 500 dropped by at least 20%.

The first thing that jumps out is that 2008 and 2002 are outliers when it comes to bond returns during a crisis. The returns are massively higher than anything else. A shadow of doubt has already crept in…This is what I meant by “recency bias”.

(As an aside: is there some reason to think that the the behaviour in 2000 & 2008 is a “new normal? I suppose you could make an argument that modern central banks are more aggressive with cutting rates to cushion the effects of recessions and speed recoveries. Though, I can’t say that I’ve actually seen anyone try to make an argument along those lines as a reason to place an extra level of faith in Treasuries during a bad time.)

We know that there have been other recessions and bear markets. But they often don’t show up in annual data. Let’s take the Crash of 1962 — sometimes called the Flash Crash of 1962 or the Kennedy Slide of 1962 as an example.

During the crash markets dropped by 28%. Actual investors would have been extremely distressed living through those months.

The crash began in December 1961 and finished June 1962. That means early part of the crash — in 1961 — is “hidden” by the rest of 1961. The annual return for 1961 was 26.68%. And the rest of the crash — in 1962 — is obscured by the recovery over the next six months. The annual return for 1962 was -8.81%.

Not great but nothing to suggest that there was a sizeable bear market for 8 months. Yet 1962 saw the second largest drop in Dow Jones history (at that point). May 28, 1962, saw massive single day drops — the Dow Jones fell by 6% in a single day.

All of that is lost when you look at annual data. So let’s switch to monthly data and take another look.

There have been 9 bear markets since the 1950s. For each of them we can look at the monthly correlation between stocks and bonds. For our story to hold up — that bonds go up when stocks go down — we need the correlation to be negative.

Image for post
Image for post

We see more problems with the theory. There have been 9 bear markets.

  • In 3 of them, correlations were solidly negative (1957, 1987, 2007–2009)
  • In 2 of them, they were still negative but less so (1962, 2000–2002)
  • In 2 of them they were basically zero (i.e. uncorrelated) (1966, 1973–1974)
  • In 2 of them they were actually positively correlated (1969–1970, 1980–1982)

Correlations tell part of the story but they aren’t the whole story. We can also look at total returns during those bear markets.

Image for post
Image for post

(The “Stock Returns” are based on monthly returns, not daily returns, so they will differ slightly from numbers you see quoted elsewhere that use daily returns.)

We can see that our intuition of what correlation means doesn’t necessarily line up with returns.

  • 1980–1982 had a positive correlation, so we naively think that means “they both lost money”. But that’s not what actually happened.
  • 1987 has a very large negative correlation, so we naively think that means “bonds must have gained a lot!” but the actual gains were virtually non-existent.

But the biggest takeaway is just that the bond returns, when they exist at all, are generally extraordinarly anemic. Let’s switch from percentage to actual dollars to try to make this a bit more concrete.

Image for post
Image for post
Assuming an initial amount of $100,000 in bonds and $100,000 in stocks.

So…sure, in 1957 bonds had a very large negative correlation of -0.95 with stocks. But in the real world that means your bonds made a whopping $500 while your stocks lost over $12,000. The gain in bonds was only 3.5% of your loss in stocks.

I mean, yeah, technically bonds did go up when stocks went down. But I doubt anyone counts that as anything except a technical victory. You still lost tons of money.

The 1980–82 bear market is good case study of another problem with the claim. At first glance bonds appear to have “gone up” by a lot. The total return was +17.32%. But they didn’t actually “go up” by that much. Most of that return was from existing interest rates.

In October 1980, the month before the bear market began, the interest rate on 10-year Treasuries was 11.75%. Meaning even if the market does nothing you would get a return of 11.75%. If the total return of bonds was 17.32% then, as an approximation, we can say that the return was composed of the existing interest rate and increases based on “flight to safety”.

total return = interest rate + "going up"
17.32 = 11.75 + "going up"
"going up" = 17.32 - 11.75
"going up" = 5.57

As an approximation, it looks like bonds “went up” by around 5.5% during the bear market of 1980–82.

Suddenly those amazing bond returns in the 1980–82 crash don’t look so special anymore. The bond returns were great…but that was simply because prevailing interest rates at the time were great; it wasn’t because bonds “went up” during the crash.

The lesson is: when we look at bond returns during a crash we need to separate out “steady state returns” versus “returns due to the bear market”.

It seems clear that the claim that “bonds go up when stocks go down” isn’t well founded.

  • There are bear markets where bonds don’t go up.
  • Even when bonds go up, they might not go up by very much.
  • Sometimes what looks like “going up” is just be “staying relatively even and continuing to pay interest”.

That said, this doesn’t mean that bonds are somehow “risky” during a crash or bear market. They may not always be negatively correlated but even low correlations help in a bad time. And even when they are positively correlated bonds are still a low-volatility asset. They may go down…but they aren’t going to go down as much as stocks.

Look again at 1973–74. Bonds didn’t go up when stocks went down. Bonds went down by -$5,870 in our example above. But stocks went down by -$47,530 over the same period. Most people would have been pretty happy with bonds in that situation.

Instead of saying that “bonds go up when stocks go down” and pretending that bonds are a kind of magic prophylactic against bad times…we need to be realistic and say “bonds will stay mostly flat and I’ll continue to collect interest payments at the prevailing rates”.

In a crash, that ain’t too bad.

Written by

Learn how to enjoy early retirement in Vietnam. With charts and graphs.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store