Arguments about alpha, beta and 'smart beta' are missing the point.
[This blog originally appeared on our sister site, IndexUniverse.eu.]
“Is ‘smart beta’ that smart?” asks Jeff Molitor, Vanguard’s European chief investment officer, in this week’s FTFM.
Molitor reminds investors that by choosing alternative index approaches, they are taking a decision to time market movements. And that’s a “loser’s game”, says the Vanguard CIO. “Smart beta should never be thought of as a perpetual motion alpha generator.”
It’s hard to disagree with Molitor’s suggestion that market timing is foolhardy. I’ve lost money doing it myself.
But—and this is not a criticism of Vanguard’s CIO—aren’t we getting confused here? We probably all thought we knew where we stood in classifying investment returns. Beta was the cheap, indexed bit of the investment portfolio, alpha was hedge funds and other active managers, the expensive part. Yet now we have “smart beta” apparently producing alpha? What’s going on?
Perhaps we’re all tying ourselves in mental knots by sticking to a discredited framework for measuring risk.
As soon as we refer to the first two letters of the Greek alphabet in describing investment returns, we’re accepting the assumptions of the 50-year-old Capital Asset Pricing Model. As Global Custodian’s editor Dominic Hobson recently reminded us, paraphrasing Keynes, bad ideas can have an incredible longevity. And as far as bad ideas go, CAPM certainly fits the bill.
It’s not the theory’s unrealistic assumptions that are the main problem (that investors can trade without transaction costs, that they are rational and risk-averse, and that they can lend and borrow unlimited amounts at the risk-free rate).
It’s when we get to measure risk—which CAPM defines as the standard deviation (or volatility) of returns—that we get into the deepest trouble.
CAPM assumes that asset class returns are normally distributed random variables. In other words, a graph showing the distribution of many periods of returns looks like the symmetric bell curve of the normal distribution. And the return in one period is not supposed to depend in any way on what happened before. If these assumptions reflected reality, volatility would be a suitable way to measure things.
Unfortunately, things are different. Investment returns have “fat tails” and a graph of their returns may have a characteristic skew, either upwards or downwards. There’s also ample evidence that what happened yesterday does influence what happens today. Benoit Mandelbrot’s ironically entitled book The (Mis)behaviour of Markets is perhaps the best summary of the case against CAPM.
These are not minor quibbles over some abstract theory. The broad acceptance of a risk framework based on CAPM has had devastating real-world consequences. Had regulators paid more attention to what people like Mandelbrot were saying, there’s no way such system-wide leverage would have been allowed to build up during the last 25 years, for a start. Reducing all the debt we created is proving mighty painful.
Belatedly, some of the necessary reforms to the financial system are now being put in place. Capital requirements are being raised to make banks safer, while “shadow banks” are under an increasingly harsh spotlight. Regulators’ prescribed risk measure for banks’ trading books is moving from value at risk (which relies heavily on standard theory) to expected shortfall. Putting a figure on how much you might lose, rather than on the probability of a “tail event” occurring, will concentrate many people’s minds.
But in the investment world it’s as though nothing has happened, particularly in the indexing business. Practitioners continue to refer to volatility, Sharpe ratios, alpha and beta as if CAPM can carry on regardless.
Is one reason why “smart beta” approaches appear smart the fact that the risk measure we’re using—volatility—is inadequate? Do much-hyped low-vol approaches “outperform” because we’re ignoring skewness in equity returns? How should such skewness be reflected in index design? Is sufficient attention being paid to “gap risk” when ETFs are put together? These are questions to which it would be good to hear better answers from those who claim they’re providing “beta”, smart or not.