This is the second post in a series on duration matching. The first post covered an introduction to duration, what duration matching is, and why we might want to use it.
The key idea of duration matching is that we can eliminate interest rate risk by matching our bonds’ duration to our investment horizon. If we have a short time horizon, we buy short duration bonds. If we have a long horizon, we buy long duration bonds.
Now, long duration bonds aren’t a particularly popular investment. Like inflation-linked bonds, they tend to get overlooked – and I can understand why.
Bonds are usually viewed as the ‘safe’ part of a portfolio. But because of their longer duration, long-term bonds do come with higher volatility. And that’s not what most fixed income investors are after. But for those long-term investors who adopt a duration matching approach to their bond allocation and understand that volatility isn’t the same thing as risk, long-duration bonds are actually less risky than short-duration bonds. Although they do have a higher volatility, they eliminate interest rate risk for the investor.
And they can not only eliminate interest rate risk, but they can do so while increasing diversification and increasing returns.
Let’s see how.
More diversification, more returns
Long-term bonds can not only help in reducing your interest rate risk, they can also help improve a portfolio’s risk-adjusted returns by 1) increasing returns from the higher yields of long-term bonds, and 2) providing a better diversifier to your equities than short-term bonds.
On point 1), intermediate-term bonds generally have lower expected returns than long-duration bonds, due to their lower yields. Investors in long-duration bonds are compensated for taking on the extra interest rate risk in the form of higher yields. Because of their higher yields, long-duration bonds should outperform short duration bonds in the long run.
This does depend on the shape of the yield curve, though, and there are some circumstances where, when the yield curve looks particularly flat, there’s not much extra yield to be found by reaching far out onto the yield curve. When the yield curve inverts (which is rarely), the yields on intermediate-term bonds can even be higher than the yields on long-term bonds.
This situations are rare, though, and any attempt to profit from them would require successful market timing, which we all know doesn’t work consistently. In a normal upward sloping yield curve environment, longer-dated bonds are likely to outperform over the long run.
On point 2), it’s worth making a distinction at this point between de-risking a portfolio and diversifying a portfolio.
It’s very possible to lower the volatility of a portfolio merely by allocating more to cash or very short-term bonds. That’s called de-risking.
Diversifying means adding something to the portfolio which goes up when your stocks go down. Cash and short-term bonds don’t really do that (at least not very much) – they merely reduce the impact of the drawdown by not falling as much, rather than producing meaningfully positive returns during crashes.
The traditional “bonds are for safety” approach (owning only short-term bonds) pushes all the risk onto a single source like the equity risk premium instead of diversifying it across the equity risk premium and the fixed income term/duration premium.
A traditional 60/40 portfolio with intermediate-terms bonds has roughly 90% of the portfolio’s risk coming from the equities.
This makes sense. The risk of stocks is roughly three times as high as bonds (15% vs 5%), and the allocation to stocks is 1.5x the bond allocation (60% vs 40%). These two factors combined result in the high contribution of stocks to portfolio risk.
So by replacing the bonds with longer duration bonds, this gives the potential to build a portfolio with higher expected return, similar overall portfolio variance, and higher diversification – only roughly 70% of the portfolio’s risk is now on equities.
The maths of diversification states the best diversifier is one which has the lowest correlation with the rest of the portfolio, and has the highest variance (aka “power”) to counteract the stocks. So by adding lower correlation and higher-variance longer-term bonds, you’re improving diversification. You’re spreading the risk exposure in the portfolio across different independent sources of risk.
Short-duration bond funds are less volatile than long-duration bond funds, so are less able to provide the oomf necessary to act as a strong diversifier.
So in theory long-duration bonds should provide us with higher returns, and lower risk.
But that’s in theory. Let’s see how it looks in practice.
Taking a look
The chart below shows the returns of a 60/40 portfolio at differing levels of bond duration:
Source: Bloomberg. Equities are the MSCI World TR (GBP). Long duration bonds are the Bloomberg UK Govt 15+ Years Float adjusted TR Index (the benchmark for Vanguard UK Long Duration Gilt Index Fund). Medium duration bonds are the FTSE Actuaries UK Gilts TR Index (the benchmark for IGLT). Short duration bonds are the FTSE Actuaries UK Gilts up to 5 Years Index (the benchmark for GIL5).
And some high-level stats:
Starting at the low-duration end, both short duration bonds and cash had almost identical returns, volatility and maximum drawdowns. Which is what we’d expect. Cash, with a duration of zero, shouldn’t be much different from short-term bonds with a duration of 2 years. There was very little added benefit from using short-term bonds over cash.
But by adding more volatile and lower-correlation long-duration bonds, the portfolio was able to increase returns with only slightly higher levels of risk – and without increasing the maximum drawdowns.
Now let’s dig into that drawdown point for a second, because it’s an important and interesting one.
Equity market drawdowns are the crucial time for bonds, because it’s during these periods when you want your diversifiers to be doing their job. Nobody really cares about correlations when equities are doing well. When equity markets sell off is when 1) we’re paying the most attention to our portfolio, and 2) we’re most likely to bail out of whatever’s performing poorly. So the better a diversifier is able to reduce our portfolio’s drawdowns, the more valuable that diversifier becomes.
When we vary the bond durations of our 60/40 portfolio, based on what we’ve seen so far we’d expect to see two things: 1) short-term bonds and cash providing similar levels of drawdown protection, and 2) long-duration bonds (hopefully) providing the biggest benefit.
Here’s the drawdown chart from our previous example:
Without looking at the legend, I’d have a hard time telling you which is which. All four options look pretty similar.
Here’s the table, which puts some numbers to it (I’ve also included the dot-com crash, which we don’t have short-duration bond data for):
So our first intuition looks correct. Cash and short-duration bonds performed extremely similarly during every crash – there was never more than 2% difference in the portfolio’s drawdown. They didn’t have a strong enough dose of volatility and negative correlation – both short-term bonds and cash both worked simply by brute-force risk dilution.
Our second intuition looks generally OK too. Longer-duration bonds provided superior diversification to cash/short-duration bonds during 4 of the 6 crashes (the 4 most recent ones), they were about the same during the 2008 housing crisis, and were marginally weaker during the dot-com bust.
Annoyingly the index for short-duration bonds only goes back to 2004ish, and the other bond indices back to 1999. But we can turn to our brethren in the US to get another 7 years of data, going back to 1992.
Using data from Portfolio Visualizer (sorry, it’s an American site), we can plot returns for 3 portfolios going back to the 90s.
Portfolio 1 is 60% US market, 40% cash
Portfolio 2 is 60% US market, 40% short-term treasuries
Portfolio 3 is 60% US market, 40% long-term treasuries
Source: Portfolio Visualizer
Again, there’s very little difference between cash and short-term bonds. And adding long-term bonds greatly enhanced returns while only slightly increasing risk – a 2% per year increase in return for a 0.2% increase in risk, and a lower maximum drawdown (-27% for long-duration bonds vs -32% for cash).
Drilling into the drawdown statistics, the picture in the US looks similar to what we found in the UK – long-term bonds reduced the portfolio’s drawdowns in 4 out of the 5 last market crashes:
Source: Portfolio Visualizer
And although we don’t have long-duration bond data in the US going back further than early 1990s, we do have 10-year US bond data going back to 1929. And by comparing the drawdowns of a 60/40 portfolio using 10-year bonds to a 60/40 portfolio using cash, we get a similar picture in favour of using longer-duration bonds:
The data going back to 1929 encompasses a whole bunch of different interest rate and correlation regimes, and it still shows a similar picture. The differences in drawdowns aren’t particularly large between using cash vs 10-year bonds as a diversifier.
And over that time, goes without saying that the 60/40 portfolio with 10-year bonds easily outperformed the 60/40 portfolio with cash (8.3% returns for the 60/40 with bonds, 7.7% for the 60/40 with cash). Using even longer-dated bonds than the 10-year in this example would’ve obviously provided even higher returns.
Before we wrap up, it’s worth remembering long-duration bonds aren’t perfect.
As standalone funds, they have higher volatilities than shorter-term bonds. We’ll get a bit more into the behavioural risks of owning long-term bond funds in a later post on the drawbacks of duration-matching. But it’s important to bear in mind that longer-term bonds may be more difficult to stick with if you focus on the risk/drawdown characteristics of individual holdings rather than the overall portfolio. Long-duration bonds might reduce the drawdowns of the overall portfolio, but the funds themselves have larger drawdowns than their shorter-term bond equivalents.
So they’re still not for those who don’t like seeing red numbers on their valuations.
They can also appear unattractive when their yields aren’t much better (or are even below) those on offer from intermediate/short-term bonds. Why take the risk of long-term bonds when you can get a similar yield from shorter-term, safer bonds?
Well, as we saw in the last post, if your bond duration is roughly equal to your investment horizon, then long-duration bonds aren’t riskier. In fact, by owning “safer” bonds with a duration shorter than your investment horizon, you’re actually increasing risk.
So I don’t think we should be afraid of long-duration bonds.
For those who utilise long-duration bonds, either as part of a specific duration-matching strategy, or simply as a diversifier in a long-term portfolio, they look like they can be a useful addition.
Not only can they substantially reduce your interest rate risk, but they come with the added benefits of being able to reduce drawdowns and increase long-term returns. Granted, the drawdown-reduction benefit over shorter-term bonds isn’t guaranteed, but history suggests the odds are good. In the instances where the drawdown-reduction benefit wasn’t better than shorter-term bonds, it was only marginally worse. And given their much higher long-term returns, it looks like a pretty good trade to me.