When assumptions fail

An apology. If you found Monday’s piece a little hard to understand in places – that’s probably because I made quite an annoying and fundamental error of definition.

I roughly described a “60/40” portfolio of 60% equities and 40% bonds as a kind of “risk-parity” portfolio. This is wrong.

A risk-parity portfolio uses volatility as a measure of risk, and leverage as a tool. Through those means, it adjusts asset allocation, often with upwards of 50% bonds which are then leveraged to gain extra return.

Whereas a 60/40 portfolio (while based on the same underlying idea of bonds being less risky and anti-correlated which I was challenging on Monday) does not use volatility as an input to adjust asset allocation, rather it periodically rebalances to bring the balance back to a 60/40 split.

The arguments against bond safety and stock-bond anti-correlation still stand, but I must clarify that the section in the middle highlighting the risk of using volatility as an input for risk applies to risk-parity funds, and not 60/40 funds.

Both are common, and both are susceptible to the vulnerabilities of the assumptions I want to challenge, but they are different kinds of fund. I apologise for the error.

And I apologise again because I’m going to continue down a similar theme today, or at least follow on from Monday’s piece.

Because in March we had a test case for the equity-bond correlation. Did it work? Did bonds rise as equities fell, in the fastest bear market crash any of us can remember?

Weellll… not quite. For a bit, yes, and then no.

See here, it’s a chart of BND – the Vanguard bond market ETF, one of the main ones:

Source: BND on Koyfin

I have marked where the equity markets peaked – on Friday 21 February. Starting the following Monday, equity markets began their first down week – roughly 12%, which was astonishing at the time. The BND ETF did its job for a good couple of weeks after that, rising a few per cent by 9 March.

And then…

Source: BND on Koyfin

You can see that on 12 March alone it briefly went as low at $75, levels not seen since the summer of 2016. Here’s a chart up to today, showing BND and the S&P 500.

Source: Koyfin

Correlation? I think so. The bonds have outperformed slightly, but they have moved in line – suggesting that if the crash had continued, a 40% bond portion in a 60/40 portfolio, or a leveraged 70% bond allocation in a risk-parity portfolio, would have both suffered heavily, in line with equities.

And just look at their chart since 2009:

Source: Koyfin

So that goes some way to showing that perhaps the stock-bond anti-correlation assumption may not be as strong as we think.

There are a couple of reasons, specific to March 2020, which merit investigation.

Firstly, you have to of course deal with the panic. With markets falling all over, the race for cash meant that all assets, good or bad, were sold in order to secure portfolios against margin calls and further losses.

Then, there is the factor beloved by Chris Cole which I mentioned on Monday – which is that the use of volatility as an input to risk-parity funds means that when volatility spikes, these funds are forced to sell. That explains the ferocity of the crash in part, because selling creates volatility which forces more selling.

That’s a systemic cause for the ferocity of the crash, and the correlation of equities and bonds. And yes, this works on the way up too, explaining how we’re getting to such extreme valuations on the way up.

But Luke Gromen of Forest for the Trees has also offered a global macro explanation for the sell-off in bonds, especially in the US.

He points out that non-US governments now hold more US govt bonds than are held in the US itself. So when a sharp correction comes (he predicted this before it happened), those foreign governments will sell their US bonds in order to raise cash dollars, otherwise they might find themselves in an even worse hole – default.

Because foreign holders of dollar debt now outweigh US holders, in a crisis the sellers will outweigh the buyers, causing a crash in bond prices too. He believes this partly explains what happened in March.

It’s an interesting argument and I like the way it offers a current global macro case for why Chris Cole’s historical mistrust of the equity/bond anti-correlation assumption might be true today.

Cole argues that many fund managers, especially pension funds, place too much faith in bond anti-correlation with stocks, and there’s more. With rates too low for pension funds to comfortably achieve the 7% or 8% that they need to meet their liabilities (payments to retirees), they have ramped up their involvement in things like private equity and venture capital, which are essentially levered correlated assets.

This move depends on a stable investing environment to work. It may help squeeze out an extra per cent or two each year when things are going well, but they will massively exacerbate losses in a period of stress, which we know comes inevitably after periods of calm (Hyman Minsky’s point that periods of stability do not merely precede, but sow the seeds for a period of instability).

True diversification is not the layering of different but correlated assets – like equities, PE, VC, and levered bonds – he says. Truly uncorrelated assets like certain options, precious metals, or trend-following strategies are the ones that can offer true protection against the uncertainty of the future.

He argues that it’s critically important for your investments to be anti-correlated to the economic cycle because both your job and your mortgage can be thought of as correlated assets.

In a downturn such as the one we have, both jobs and mortgages come under increased threat, and if your portfolio is essentially cycle-correlated – ie, it rises and falls in unison, then it is likely to fall in value when you need it most.

It’s a concept I hadn’t thought of before, but your investments can act as a hedge against the impact of the economic cycle on the two main pillars of our lives, if correctly diversified.

And in a recent interview, Cole left us with a quote which I really love.

He looks at a 90-year period rather than a decade or two. And his focus is pointing out that investors today are mentally unprepared for deep periods of outperformance – such as the US saw in the 1930s, or Japan in the 90s and 00s.

He says that truly uncorrelated assets like gold, which benefit from the conditions in which equities and bonds fail (panic, central bank stimulus), “are not protection for a rainy day, but for a rainy decade”.

It’s worth considering that such things are possible, and whether we are really ready for them.

That’s why Nickolai Hubble spends much of his time writing research on investing outside of the financial system, or in truly antifragile, uncorrelated assets.

That’s why we’re repeatedly trying to get you signed up to his service, because we believe that the risks of a long period of outperformance are being underrated by most mainstream firms, commentators and investors.

His work is excellent, and I recommend it.

All the best for now, and sorry again for the errors on Monday.

Best wishes,

Kit Winder
Editor, UK Uncensored

You may like

In the news

Taking flight

A few months ago, you might’ve spotted (because I plugged it relentlessly for a fortnight) that I hosted an online investor summit about bitcoin and other cryptocurrencies. We were...
Load More