Via the excellent Damien Boey at Credit Suisse:
Over the past few weeks, the bond rally has taken us to uncharted territory. It is not so much that global bond yields are pushing record lows. It is not so much that a significant portion of investable government debt around the world is now offering negative yields. Rather, the key development is that so-called term risk premia across the world are at their most negative in recorded history. Recent market action has made a complete mockery of the very concept of term risk premia as either a speed limit indicator, or valuation tool for bonds.
A case in point. US 10-year bond yields are still well off their all-time historical lows of 1.36%. There is still room for yields to fall if history is anything to go by. But according to the Fed’s best estimates of the term risk premium, the spread between bond yields and the market’s estimate of the long-term neutral rate, or the compensation for getting one’s rate forecast wrong is currently sitting at -100bps – the most negative in history. Investors are paying for the privilege of expressing dovish views, rather than being compensated for doing so. So from a long-term perspective, it makes sense to be thinking very negatively about bonds, even though yields are not strictly at historical lows. Indeed, cyclicality and tactical views aside, we are the most structurally negative on bonds we have ever been. Valuation is simply too rich, and unrewarding.
From an Australian perspective, the story is quite similar. We use a “fundamental” estimate of the long-term neutral rate, based on an average of trend nominal GDP growth and the cash rate needed to eliminate household negative gearing under the assumption of full pass through. On the assumption of 80% pass through, consistent with the recent commercial bank response to two RBA rate cuts, the neutral rate sits at 1.57%. And with current 10-year bond yields of 0.94%, this implies term risk premia of -63bps. This too is an all-time historical low. To justify current bond yields, one has to believe that 55% of every RBA rate cut from here will be passed through to end borrowers to drop the neutral rate to 0.94% – a very bearish assumption indeed when we consider how much interbank credit spreads have narrowed, and what the commercial banks have actually done.
Cynics might point to the fact that term risk premia estimates are very imprecise, because we have to estimate, rather than observe the long-term neutral rate to benchmark bond yields. Therefore, we should be careful of making big calls on bonds on the basis of imprecise technical analysis. For what it is worth we totally agree with this point of view. But in our view, estimation error or methodological issues, cannot explain the pricing dislocations we are currently seeing.
For example, we have successfully augmented Fed models to also account for sovereign debt risk in peripheral Europe. Essentially, we should expect to see term risk premia at more negative levels than normal, when sovereign debt risk in Europe causes pricing dislocations in bonds elsewhere. When asset allocators see that a large chunk of peripheral European debt has become uninvestable, their natural response is to spread their funds across the remaining investable universe in other markets. And as this investable universe is considerably smaller, we should expect bond supply shortages to artificially push up prices and push down yields. Indeed, the past few occasions when we have seen negative term risk premia in bonds have involved risks emanating out of Europe (eg European crisis, Brexit).
Even accounting for new factors like sovereign risk, we find that term risk premia in bonds across all the major markets are at their most negative in recorded history. Mathematically, the term risk premium, multiplied by bond duration gives us a “risk-neutral” estimate of year-ahead bond returns. Therefore, negative term risk premia foreshadow negative returns on bonds over the next year with wide rather uncertainty bands. But interestingly, bond returns are currently overshooting their “risk-neutral” estimate by 2 standard deviations. And history tells us that from this starting point, the most likely outcome is that bond returns undershoot the “risk-neutral” forecast in the period ahead by 2 standard deviations. If this occurs, bonds are in for one a gigantic sell off – comparable, to what we witnessed in 2016-17.
There are of course exceptions. The one, most notable exception, where bonds were overshooting by a significant margin, and kept overshooting was during the 2008 financial crisis.
This is helpful context. It means that to be positive on bonds from here, we have to believe that the global cyclical slowdown, amplified by trade war and geo-political uncertainty, will cause a 2008-like recession. Indeed, as it stands, to justify current bond valuations, you have to have the view that there is a sharp slowdown already underway, with some tail risk of a super recession (again, wide uncertainty bands). To imagine how this super recession might unfold, consider the effects of high uncertainty and rising volatility on global credit pricing. Consider how a major stock market correction might trigger a recession in business investment. Consider the collateral damage on other asset classes, given that correlation risk has become so high. But equally, consider how desperate times might call for desperate measures, such as co-ordinated central bank balance sheet expansion, outright foreign exchange intervention, and fiscal stimulus. And if you have the view that uncertainty can subside, because there are visible ways out of the current bind (albeit some, far-fetched at present), hope can become a strategy, and bonds become a sell.
Right now, our preferred measure of global uncertainty based on the VIX and newsflow mentions, is sitting at more than 2 standard deviations above average. So it makes sense that bonds would overshoot by 2 standard deviations. From this starting point, it would take a severe turn of events, several more policy mistakes and dramatic escalation of geopolitical tensions to cause uncertainty to rise even further and bonds to overshoot even further. Otherwise, the laws of mean reversion apply, and we should expect uncertainty to ratchet down from here.
The next thing to consider then is how a bond sell-off might impact other asset classes, especially because correlation risk is so high, and valuations are still quite full. Arguably, a bond sell off might be another risk-off catalyst, with the same consequences as the hard landing scenario alluded to earlier. But this is where multiple dispersion within equity markets is worthwhile noting. Equities might look expensive outright – but this masks the fact that cyclical half of the market is affordable, while the other half – the bond proxy half – is not. The greater likelihood is that a rise in bond yields, triggered by evidence that the outlook “is not so bad” causes a sector and factor rotation event within equities, rather than a wholesale rotation out of equities. Specifically, we should see cyclical value prevail over defensive quality, and reflation exposures (eg resources) outperform bond proxies.
A sagacious warning. It is always wise to take profits even if you think, as I do, that your bull market still has further to run.
He is also a former gold trader and economic commentator at The Sydney Morning Herald, The Age, the ABC and Business Spectator. He is the co-author of The Great Crash of 2008 with Ross Garnaut and was the editor of the second Garnaut Climate Change Review.