There Are Early Signs of a Value-at-Risk Shock

(Bloomberg Opinion) -- Financial assets, pretty much all of them, are in increasing danger of some nasty shocks and not only because many are horribly expensive.

The problem, as we started to see last week, is a changing relationship between government bonds and equities. Should that change become more entrenched — and this week's movements suggest it will — many investors will be able to hold fewer of either. That's because of the risk-management model that just about everyone uses and which is written into the way banks and big investment firms are regulated and capitalized.

The model in question is called Value-at-Risk, more commonly known as VaR. The best we can hope for is that this de-risking happens slowly; the worst would be mechanistic selling not dissimilar to the forced selling in the 1987 crash. What made a bad problem much worse in October of that year was the spectacularly misnamed portfolio insurance, a product that forced investors to sell ever greater quantities of stocks in a vain attempt to replicate the behavior of put options.

At least in 1987 bond markets flew. This time around, the change in the relationship between between bonds and equities lies at the heart of the problem.

Simply put, VaR works out the most money a portfolio could lose on a given day or week with a certain degree of confidence (a 95% degree of confidence to be exact). If you have two assets in a portfolio, VaR looks at the volatility of each of them and the correlation between them. Although most people tend to concentrate on volatility, correlations are also crucial. If gains in one asset offset losses in another, the overall risk of a portfolio goes down. For more than two decades investors have thus been able to hold more equities and bonds than they would have otherwise because when either fell, the other has tended to go up. The problem comes when that assumption no longer holds.

Until the late 1990s, equities and longer-dated bonds generally moved in the same direction. The evidence suggests that their relationship changed largely because inflationary worries morphed into disinflationary worries. There are several factors that suggest we may be returning to the days of stocks and bonds falling and rising in lockstep. 

The first three problems are to do with the riskiness of bonds. First, bond yields are so low that the only way in which investors can get a sufficient kick out of them is to hold larger amounts. Second, coupons are now so low (or even negative) that mathematically bonds' interest-rate sensitivity is higher: For a given change in yields, prices would move more than in the past. Third, whatever central bankers might say, inflation is a clear and present danger. This concatenation of problems is partly why longer-dated bonds have been thumped in recent months.

The fourth problem is VaR. If I'm right and worries about inflation change correlations, investors' portfolios are much riskier than they thought. In some recent sell-offs, both bonds and equities have fallen sharply. You should, of course, expect some bad days if you're only assuming the VaR model works 95% of the time. But it isn't working 95% of the time right now and the behavior of portfolios is becoming very worrying.

In extremis, risk managers and their VaR models will mechanistically force the selling of both bonds and equities. That would in turn mean higher volatility and an even more pronounced change in correlations, thereby necessitating yet more selling. Hence the 1987 comparison. We're already seeing the beginnings of such a VaR shock and it may have much further to run. That's one reason I don't buy the argument that higher bond yields are good for equities.

I don't want to overdo the 1987 parallel. One lesson that people learned then is that you need to protect against extreme events. Regulated institutions run “slides” of their portfolios showing what losses would look like if markets moved given percentages in either direction. More sophisticated institutions will also be loading up on options struck at prices far away from current levels, so-called deeply-out-of-the-money options. These cost next to nothing and provide protection against extreme moves.

Still, for every buyer there must be a seller and the ones selling are generally speculators, so the exposure is really only shifted. And these options help only against very extreme events. Buying options with strike prices closer to current prices would protect against more modest moves but they're very expensive. So unless institutions have strong views they won't do it.

The bigger point, however, is this: The disinflationary forces that helped propel assets higher are turning into inflationary ones. If that leads to a shift in bond-equity correlations, as seems to be happening, institutions big and small will have to stump up more capital or reduce risk across the board. There is no third way. When asset prices are falling and portfolios are losing money, the preferred strategy, you may not be surprised to hear, is to sell.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Richard Cookson was head of research and fund manager at Rubicon Fund Management. He was previously chief investment officer at Citi Private Bank and head of asset-allocation research at HSBC.

©2021 Bloomberg L.P.