The asymmetry of interest rate duration risk

A common way to think about bond yields is to view them as a cushion that protects bondholders from the potential negative effects of duration risk. The yield cushion provides protection from capital losses if yields were to rise. (details here)

As bond yields have now collapsed to very low levels, that protection from duration risk has vanished.

For example, the chart below shows the duration vs. yield trade-off for the most widely referenced global bond index. Yields have declined, while duration has increased, leaving investors faced with more interest rate risk for less return.

Normally investors would demand higher yields in return for taking more duration risk. However, as yield chasing capital has flooded into longer maturity bonds to eke out a bit more return, compensation for interest rate duration risk has eroded away.

Of course, duration risk works both ways. If bond yields fall a lot, duration exposure can deliver large capital gains, which is exactly what had been happening across global bond markets. (details here).

But with bond yields fast approaching zero, and indeed already negative in many places, duration exposure becomes unfavourably asymmetric i.e. the loss potential becomes disproportionally large relative to the upside.

The chart below puts this asymmetry into context for conventional duration based global bond investments and brings to mind the ubiquitous but frequently ignored disclaimer that past performance is not indicative of future returns.

Year to date, the global bond index below has delivered an unusually large return of 7.4%, which has largely been driven by the interest rate duration exposure inherent in the bonds that comprise the index. This is the same type of duration exposure that has largely driven the performance of conventional bond portfolios this year.

However, what duration gives can easily be taken away.

If yields were to revert modestly, say just back to where they were at the start of this year, duration exposure would impose a loss of 6.2%. But to generate a gain of that magnitude going forward, the index yield would need to drop to an eye wateringly extreme level of just 0.3%.

This is the asymmetry that conventional duration-based bond investments are currently faced with and because of this, the lower that yields go the more vulnerable investors become to future losses.

An unavoidable reality is that bond prices can’t keep going higher unless yields keep going lower, which is why, unlike with equities, the upside potential in bonds is inherently capped.

Even if stocks are expensive, there is no real limit to how high they can go. Bond prices on the other hand are bounded by the fact that hold to maturity returns are limited to interest payments received over the life of the bond and any capital gains before then are limited by how low yields can go.

This unfavourable asymmetry of risk vs return is inescapable as bond yields keep going lower because when the starting point of yields is already extremely low to begin with, there is naturally less room for them to drop further and this is what makes chasing bonds at record low yields (i.e. record high prices) very different to chasing stocks at high valuations.

Stocks can keep going up indefinitely, while bonds can’t.

As the experience of Japan and Europe shows, there is strong resistance to bond yields going far below zero, because a hold to maturity investor buying bonds with negative yields would be locking in a guaranteed loss.

We can think of ultra-low bond yields as pushing against an elastic band that sits at the zero-yield level. Yes, yields can push below zero but as they do, that elastic bands gets stretched further and further, raising the risk of a vicious snap back at some point.

Conventional high-quality bond portfolios are widely assumed to be ‘safe’ investments and in terms of credit default risk that assumption holds. However, they are now highly risky in terms of the risk of near-term capital losses due to the unfavourable asymmetry of duration risk.

It seems that investors rushing to the perceived safety of ‘high quality’ bonds, have careened headlong into the poorly compensated and treacherous territory of interest rate risk. As equity investors know well, the price at which you buy matters as much as what you buy. ‘High quality’ bonds are not immune from losses if you buy at too high a price and that’s doubly true when faced with such unfavourable asymmetry.

With yields already so low, the outlook for conventional duration focused bond portfolio returns is now heavily dependent on whether yields can keep going lower.

Of course, it’s possible yields could keep falling. One group of highly credentialed bond experts argues compellingly that yields will head lower as central banks keep cutting interest rates in response to weakening economic growth. This is currently the overwhelming consensus view that’s driving record inflows to bonds and skewing pricing to extreme levels.

But it’s far from certain how far that goes. Another group of equally well qualified bond experts argues with equal conviction that the economic outlook isn’t so bad, central banks won’t cut rates much and yields will rise.

Our view? Given all the variables, assumptions, feedback loops and subjective judgements involved we struggle to see how anyone can have high conviction about such a blunt directional market call. (details here)

In the absence of any directional view on bond yields, duration exposure is often held solely as a way to diversify portfolios and protect against downside equity risks, but even in this use case, ultra-low yields have now made duration a very expensive form of portfolio insurance. (details here)

Ultra-Low yields necessitate more nuanced thinking about duration. When bond yields get really low, long held assumptions need to be questioned and what worked in the past may not work so well in the future.

This is not a blunt question about which way bond yields will go but rather a more nuanced consideration of compensation for risk and whether things will behave the way they are expected to.