We’re in a weird period for bonds right now.
Interest rates are at all-time lows, bond yields are at all-time lows, and it’s looking more and more likely we may have to deal with a positive stock/bond correlation at some point soon.
So should we still be investing in them?
The purpose of bonds is to act as a diversifier to equities, and (hopefully) act as a hedge – i.e. go up when stocks down. They’re there to make drawdowns less painful, and to give you something to sell so you’re able to buy equities when they’re cheap and you need to rebalance.
But are there any situations where bonds don’t act as a diversifier? And if so, what are they?
This is a useful question to think about. If it turns out that bonds don’t perform well as a diversifier in a specific situation – say, during periods of sub-1% interest rates, or perhaps when bonds become positively correlated with stocks – then this will likely affect how we construct portfolios. We might need to consider alternative diversifiers in our portfolio if we can’t rely on bonds to do the job.
So I thought I’d go back through history, look at all the past market drawdowns, across several different equity markets, and see how bonds performed during those periods. There’s enough historical data to capture plenty of different regimes – low interest rates, high interest rates, rising rates, falling rates, positive correlations, and negative correlations.
- Global equities vs global bonds hedged to GBP
- US equities vs US bonds
- Japanese equities vs Japanese bonds
Global equities vs global bonds (hedged to GBP)
We’ll start off with the most relevant market for UK based passive investors – the market-cap weighted global index.
As this blog is aimed at UK investors, we’ll be comparing the global market’s performance with a global bond index hedged to sterling.
Because we’re currently in a low interest-rate environment, I thought it might also be useful to have a look at what the interest rates were at the start of each market drawdown. We can then figure out if bonds were a worse diversifier when interest rates were low at the start of the crash.
Finally, I’ve also included data on the level of the stock/bond correlation at the start of the drawdown. (I’ve included this in the tables, rather than the charts). This could be useful in gauging whether a high stock/bond correlation means bonds are more likely to fall alongside equities during an equity market selloff.
For selecting drawdowns, I’ve used the usual maximum drawdown methodology for finding the largest peak-to-trough falls. Any drawdowns over 10% are included. But if there were additional drawdowns over 10% during the recovery from the trough to the previous high, these are also included. This makes sure we capture every drawdown greater than 10% (as the usual maximum drawdown calculations don’t include any drawdowns experienced in the recovery).
Now let’s have a look at some charts.
Here’s how global bonds hedged to sterling have fared during past global market drawdowns:
Here’s the data in tabular form, along with information on the levels of the stock/bond correlation at the start of each drawdown:
Sources: Equities and bonds: Bloomberg, UK interest rates: Bank of England. Equities are the MSCI World Index (total return) priced in GBP. Bonds are the FTSE World Government Bond Index hedged to GBP. The stock/bond correlation is the preceding 5-year correlation based on monthly data.
A few thoughts:
- We have data going back to 1985 for global stocks/hedged bonds, and for every selloff over 10% since that time, bonds have provided excellent protection against equities. Not only have bonds fallen less than equities, they’ve provided positive returns over every single drawdown period.
- Since interest rates have been sub-1%, we’ve had four drawdowns over 10% (2010, 2011, 2018, 2019). During each of those drawdowns, bonds still provided positive returns in each of them, and outperformed stocks by an average of 16%. Low interest rates don’t seem to have prevented bonds acting as a hedge to equities.
- If there’s a relationship between the stock/bond correlation at the start of the drawdown and the performance of bonds during the drawdown, I can’t see it. Bonds still provided positive returns during equity selloffs when the correlation was positive going into the selloff. In fact, bonds provided an average return of +11% during equity drawdowns over 10% when the correlation was positive going into the drawdown, and provided an average outperformance of 32% vs stocks.
- Bonds did seem to generate higher returns during selloffs when interest rates were higher (although maybe that’s just me looking to confirm my biases). But the relationship isn’t clear cut. Some large drawdowns during periods of high rates saw bonds provide large positive returns – the selloff in 1991/1992 saw bonds rise 22% with rates starting at 11%. But the selloff in the early 1990s saw bonds only rise 6%, despite rates starting at 15% and the selloff being even more severe (-35% vs -14%).
So overall, bonds are looking very good so far.
Sadly I only have data for global bonds going back to 1985. (If anyone knows where to get global bond data hedged to GBP pre-1985, please let me know!)
This is annoying, because bonds have been in a huge bull market since the 1980s thanks to the outrageously high yields on offer to bond investors in the 80s (NB: bonds have not been in a bull market due to falling interest rates, as I wrote about here). So most of our data, as it starts in 1985, covers periods with relatively high interest rates – we only have four periods where rates are sub-1% at the start of the drawdown. Although bonds still performed admirably during those four periods, it would be useful to find some more examples of how bonds perform vs equity selloffs during periods of low interest rates.
US equities vs US bonds
Using the same methodology as described for the global stocks vs global bonds example, here’s the data for the US:
And the table:
Sources: Stocks: Bloomberg, bonds: Shiller data, interest rates: FRED. Equities are the S&P 500 Index (total return). Bonds are the US 10-year government bond return. The stock/bond correlation is the preceding 5-year correlation based on monthly data. All performance data is in USD.
There’s loads of data to unpick here, but here are a few highlights:
- There have been 20 drawdowns over 10% since 1927. Bonds provided positive returns in 17 of them. In the 3 drawdowns where bonds were also negative, they still massively outperformed stocks, by an average of 20%.
- Interest rates were at 1.5% or lower between 1937 and 1948. This period saw 5 drawdowns, with bonds providing positive returns in 4 of them (and a return of -0.5% in the fifth). This included one 50% drawdown when rates were at 1.5% at the start of the drawdown – bonds still returned +4% over the drawdown period. It also included three drawdowns of over 20% for stocks, with bonds returning +0%, -1%, and +1%. Low interest rates again don’t appear to dampen their ability to diversify stocks.
- Bonds have spent most of their time since 1927 being positively correlated to stocks. Of the 19 drawdowns we have correlation data for, bonds were positively correlated going into 11 of them. Despite this, they’ve still provided excellent protection during crashes. Overall, they still provided an average return of +3% when the correlation was positive, for an average outperformance vs equities during drawdowns of 23%. Special mention here should be given to the three drawdowns where the correlations were particularly high – between 0.37 and 0.47 – going into the selloffs (1968-1970, 1972-1974, and 1976-1978). Despite such high correlations, and despite the selloffs being severe (-29%, -37%, -14%), bonds still provided returns of -5%, 7%, and 1% respectively – leading to an average outperformance of 28%. During the 1972-1974 crash, the correlation was 0.47 going in, stocks fell 37%, and bonds still returned 7%. Positive correlations really don’t seem to be anything to worry about.
- Again, the ability of bonds to hedge crashes doesn’t seem to be reliant on higher interest rates. The largest returns for bonds were in the dot-com crash and the 2008 crash, where interest rates were at pretty average levels of 6% and 5% respectively.
We saw bonds had done well hedging crashes for UK investors in the global market, and it looks like bonds have done similarly well in the US – and for a very long time.
Finally, I thought it would be worth taking a look at Japan.
“Now show Japan” is very often shouted across the internet to anyone who’s trying to prove any kind of investment point. Japan has such a unique market dynamic that many investment theories which hold up outside Japan are proved totally useless in the Japanese market.
It’s also useful for us to look at Japan because, although we in the UK have only recently had the pleasure of dreadful interest rates, Japan have had sub-1% interest rates since 1995. That gives us another useful dataset to find out whether low interest rates impair the ability of bonds to hedge equities.
Japanese equities vs Japanese bonds
Straight to the good stuff:
Sources: Stocks, bonds, and interest rates: Bloomberg. Equities are the Nikkei 225 Index (total return). Bonds are the FTSE Japanese Government Bond Index. The stock/bond correlation is the preceding 5-year correlation based on monthly data. All performance data is in JPY.
First of all, wow.
I knew it would’ve sucked to be a Japanese equity investor, but that chart really puts it into perspective. I can’t imagine losing 78% of my portfolio then, during the recovery, watching it fall over 50% again. If that wouldn’t put you off investing for life then I don’t know what would.
If you’d invested at the market peak in 1989, you would’ve only broken even in December of 2020. That’s 31 years of sitting on a loss. I wrote about all the reasons it’s not a good idea to invest in a single country’s market here (including the FTSE 100 or S&P 500), and Japan is a great example of why.
Thinking about the ability of bonds to hedge equities, obviously the Japan data is massively skewed by the monster drawdown between 1989 and 2003. This was the reason Japan adopted its zero/negative interest rate policy so much earlier than the rest of the world. But still, there are a few interesting things we can see from the data:
- Japanese bonds provided positive returns in every market drawdown. You would have doubled your money in bonds if you’d held them through the 78% drawdown. Partly that’s a function of how long the drawdown lasted, but still, you can’t complain about +109% when markets have fallen 78%.
- Interest rates were at sub-1% in four of the six drawdowns. This included one drawdown of over 50% (the 2008 crash), where rates were 0.5% going into the selloff. Bonds still provided a return of 6% over the drawdown – an outperformance of 64%. Over all four of the sub-1% interest rate drawdowns, bonds provided an average return of 5%, and outperformed stocks by 32%.
- Two of the drawdowns occurred during periods of negative interest rates. Despite rates being at -0.10% going into the selloffs, bonds still provided positive returns, and outperformed equities by 18% and 12%. So even negative rates aren’t a barrier for stocks generating positive performance and strong outperformance during selloffs.
- Correlations were negative going into most of the selloffs, although interestingly the correlation was slightly positive going into the huge 1989 drawdown – and bonds didn’t do too badly during that one.
Despite Japan’s reputation for being the market where investing theories go to die (or, at least be disproved), the idea that bonds are able to hedge equities when interest rates are low holds up especially well in Japan.
Overall, bonds have done an excellent job at hedging equities.
Far more often than not, bonds rose when stocks fell, and provided meaningful outperformance during equity market drawdowns.
This was still true during the periods where interest rates were particularly low. Of all drawdowns where interest rates were sub-1%, bonds still not only significantly outperformed stocks, but they provided positive returns in almost every instance. This was true across all three markets. Even during selloffs in periods of negative interest rates in Japan, bonds still provided strong outperformance with positive returns.
As I was putting this analysis together, I was surprised at how well bonds had done during low interest rates. I’d have thought that the reduced levels of income from bonds in low yielding environments would reduce their ability to generate positive total returns during selloffs. But it looks like this was compensated for by price appreciation through people engaging in a ‘flight to safety’ – people just like buying bonds when equities fall.
I was also surprised at how useless the correlation metric was – correlations seem to be a pretty terrible indicator of bonds’ ability to hedge equities. At the end of the day, nobody really cares about long-term correlations – all anyone really cares about is whether bonds rise when stocks fall enough for it to hurt. And the answer is yes, they do – regardless of what the correlation was when the selloff started.
Hopefully this has provided a bit of hope for bond investors. While the returns on bonds are much lower than they’ve been in the past due to low interest rates (I wrote about this here, in ‘How to predict bond returns’), if history is anything to go by, bonds are still very likely to serve their function as a hedge against equities when markets fall.
So to answer the question of whether bonds have failed in their ability to diversify equity exposure during selloffs – I’d say the answer is a resounding no. I’m still more than happy to hold them, and still believe they’re the best hedge against equities you can find.