Join our community of smart investors
OPINION

The elusive risk-return trade-off

The elusive risk-return trade-off
April 25, 2017
The elusive risk-return trade-off

The historical evidence for this is about as strong as we'll get in data as noisy as equity returns. Since 1966, the All-Share index has on average lost 0.6 per cent in real terms from May Day to Halloween, but gained an average of 8.5 per cent from Halloween to May Day - these numbers include dividends.

 

All-Share returns since 1966

 

This is no quirk of the UK. Ben Jacobsen of Tias Business School in Tilburg and Cherry Zhang of Nottingham University Business School have shown that a similar pattern exists for almost all national stock markets since records began.

Despite this, many of you are sceptical about the 'sell in May' rule. You've got good reason to be. The rule contradicts basic common sense.

This tells us that risky assets should offer decent returns. Shares should, therefore, be expected to do well over the summer. If you cleave strongly to this presumption, you might well be sceptical of the evidence that shares do badly between May Day and Halloween.

There is, though, another way of interpreting the clash between theory and evidence. Perhaps the theory is wrong. Perhaps there is in fact no trade-off between risk and return.

I know this seems radical. But in fact we have plenty of other evidence that risky assets do not earn high returns. For example:

■ Aim stocks are generally riskier than bigger, older shares. But they have fallen for most of the past 20 years.

Defensive stocks have outperformed high-beta ones.

■ Newly floated stocks are often riskier than better established ones in the sense that their lack of a track record makes them hard to value. But they tend to do badly.

■ Economists at Harvard University have found that stocks on the verge of bankruptcy "have delivered anomalously low returns", even though they are clearly riskier than others.

■ The correlation between the Vix index (sometimes called the 'fear gauge') and subsequent monthly changes in the S&P 500 is significantly negative; it's been minus 0.39 since 1990. Partly for this reason, MIT's Andrew Lo and Alan Moreira and Tyler Muir at Yale University have shown that you can beat the market over the long term by reducing equity exposure when expected volatility is high and raising it when it is low. This is the precise opposite of what should happen if higher risk means higher expected returns.

'Sell in May' fits this pattern. It's consistent with the claim of Eric Falkenstein at Pine River Capital Management that "there is generally no empirical risk-reward relation". Maybe, therefore, common sense is wrong. How can this be?

One possibility, says Mr Falkenstein, is that we care about outperforming other investors rather than about absolute returns. To the extent that this is the case, we're not so bothered by losses if others lose more. If this is the case, then all risk is idiosyncratic and so there's no general trade-off between risk and returns.

Personally, I'm not so sure about that. It might well be true of fund managers who are judged relative to their peers, but it's less obviously the case for retail investors.

There are other possible explanations, though. One is that our judgment of risk is clouded by systematic biases. For example, overconfidence about our ability to spot growth causes us to underestimate the risks of Aim or newly floated shares and so price them too highly. Or excess faith in management's ability to turn around struggling companies means these too are overpriced. In the same spirit, springtime exuberance causes us to underestimate how risky shares will be over the summer.

A further possibility is that it is macroeconomic and institutional forces that determine returns rather than attitudes to risk. Kingston University's Javier Lopez Bernardo says it's economic growth rather than risk premia that drive long-run equity returns. US researchers have shown that shifts in the balance of class power play a big role in share prices: the decline of worker militancy after the 1970s accounted for most of the rise in the US stock market since then. And then there's the effect of creative destruction. If this favours incumbent firms (say because they have 'economic moats' that allow them to fend off competition) the stock market will do well. But if it favours unquoted companies or ones that don't yet exist - as it did in the 1970s - the market will fall. Economic and institutional factors such as these are largely independent of risk.

If all this sounds like a clear case against a risk-return trade-off, it shouldn't. For one thing, I suspect the long-run returns on many value stocks is due to their riskiness; extra exposure to the risk of recession means they carry a risk premium that pays off in good times. And shorter-term variations in share prices are perhaps well explained (if only in hindsight) by variations in risk.

We shouldn't, therefore, wholly abandon the idea of a trade-off between risk and return. Instead, the only lesson I'd draw is that the world is a messier and more complex place than our intuitions tell us. Perhaps, therefore, we shouldn't be so strongly wedded in all cases to the idea that risky assets must outperform safer ones.