Your Fat Tail Skunked My Morning Coffee
Flight to safety versus flight to liquidity.
July 27, 2010
Extreme shocks have the unfortunate effect of pushing all asset correlations to one – so much for diversification. Of course, investors who hold government bonds make out like bandits – except bondholders don’t normally invest with capital gains in mind. Instead, they want yield – which gets defenestrated in times of crisis.
Otherwise market-abiding assets seem to rub shoulders uncorrelatedly in ordinary times. But in extraordinary times, they eye each other differently. Is there a relation between their everyday behaviour and their divergences under pressure.
In their paper: “Comovements of Different Asset Classes During Market Stress” Jan Piplack, at the aUtrecht School of Economics and Stefan Straetmans at Maastricht University start with the “the conditional probability that [two] assets simultaneously boom or bust given that at least one asset exhibits extreme behavior.” The assets they examine are stocks, bonds, T-bills and gold.
“[T]here is no clear evidence that all asset returns follow the same distribution − even less so for the crisis situations we are interested in here.”
So asset classes wend their own way, day in and day out – until the cataclysm.
In that situation, “extreme losses and gains for stocks and gold are generally much higher than for bonds and T-bills. Even excluding the most extreme stock returns in October 1987 would not change this result. Moreover, for stocks and gold the historical extremes point toward tail asymmetries. The extreme negative returns are much larger in absolute value tha[n] the respective positive returns. For bonds and T-bills this is not so clear cut and tends to be the other way around.”
Key here is that linear correlations between asset classes are not useful in a crisis. Instead, the authors try to apply extreme value theory, which matches up the performance along the tails of the return distribution – the fat (or thin) outliers that can overwhelm the median compounded annual return – amongst asset classes. In those instances, market crashes trash all but cash.
“The probability of having an extreme gain or loss in one asset category suddenly becomes much higher once another ‘domino stone’ has fallen.” Piplack and Straetmans argue.
“[G]old looks as a reasonable hedge against all other asset classes considered. When we have a look at the estimates for stocks and bonds, a co-crash is more likely than flight to quality from stocks into bonds…. As for stock-T-bill co-crashes the probabilities are higher but still lower than for stock-bond co-crashes. So, capital leaving the stock market does not seem to cause a run on T-bills, i.e., the flight to liquidity hypothesis. Bonds and T-bills strongly co-move in the lower tails.”
For extreme risk protection, keep your head about you … and watch your tail.