Portfolio theory vindicated by crisis, says Markowitz
One of the founders of the theory of portfolio management has defended it against widespread claims that it failed during the crisis – in fact, the crisis vindicates the theory, argues Harry Markowitz, professor of finance at the University of California, San Diego.
Portfolio theory uses the variance of assets’ returns to construct portfolios that are less risky than their individual components, and introduced the concept of the “efficient frontier” at which the most return is earned on a chosen level of risk – mean-variance optimisation.
High levels of correlation during the crisis caused many observers to question the value of diversifying in this way but Markowitz (pictured), who was awarded the 1990 Nobel memorial prize in economic sciences, says much of the post-crisis criticism is wrong – and claims many critics have not read his work properly.
“Mean-variance optimisation theory worked in the crisis. In a crisis, the systematic risk factor moves downward so much it swamps the idiosyncratic factor – but do all assets go down by the same amount? No: emerging markets had beta higher than one, so fell more than the S&P 500,” he says.
“So I use 2008 as an example of why you should be using portfolio theory – somebody whose portfolio was high on the efficient frontier got hammered, but somebody low on the frontier didn’t. Either people should have been further down or they should be aware they are taking more risk.”
The theory has been criticised for its reliance on variance as the sole risk factor used in optimisation, and a perception that it depends on the assumption of Gaussian returns that dramatically underestimate the chance of extreme market moves. Markowitz is incensed by the last criticism in particular.
“Two things get me hyper: one is too much caffeine, and the other is when I’m asked for the one-thousand-and-first time why I’m assuming a Gaussian distribution for returns,” he says. “I never assumed Gaussian returns. I showed that mean-variance efficient portfolios came close to maximising expected utility for a wide variety of utility functions. I have made that argument over and over again until I’m blue in the face and it never goes away. If people assumed Gaussian distributions and thought they were getting all the risk then it’s because they didn’t read my work properly.”
Variance is still the best risk measure to use in asset allocation, according to Markowitz. He says the Basel Committee on Banking Supervision’s reliance on value-at-risk – and its proposed replacement, expected shortfall – to calculate trading book capital requirements is “misguided”.
“Optimising mean return subject to variance beats doing it with respect to all the other risk measures, including VAR and expected shortfall – and VAR is the worst of them. The Basel Committee are not the first people to do something wrong because they haven’t read my work. They are misguided and I know who misguided them – people who have PhDs in mathematics or physics, who tell them that is what the experts use. But the so-called experts haven’t read Markowitz – I get no respect,” he says.
This article was first published on Risk