It might not feel like it, but the last three years have actually been quite good for equity investors, because the All-Share index has risen more than 20 per cent during this time. What it's not been such a good time for is riskier stocks. Of the six FTSE sectors with the highest betas during this time, three have underperformed the market (banks, miners and general financials) and only two have delivered great returns: autos and industrial engineers. By contrast, five of the six sectors with the lowest betas have beaten the market.
Indeed, across the 31 main FTSE sectors, the correlation between beta and return in the last three years has been a mere 0.17 - which is statistically indistinguishable from zero.
Even in a reasonably good time for the market, therefore, investors have not been greatly rewarded for taking on risk.
This is, of course, not a new finding. In fact, it's one of the oldest in financial economics. Back in 1972, Michael Jensen, Fisher Black and Myron Scholes found that low-beta shares in the US did better than they should have between 1931 and 1965 while high-beta shares did worse. The security market line - which relates returns to beta - was flatter than it should be. This is the defensive anomaly, the finding that low-risk stocks do better than they should.
This idea, devised in 1979 by the late Amos Tversky and Nobel laureate Daniel Kahneman, says that when people are faced with a loss they often gamble in an attempt to get even. Losing race-goers often back outsiders in the last race, and gamblers in casinos take more desperate bets later in the night.
Developments in neuroscience have corroborated this. Researchers have found that our brain processes profits and losses completely differently. While conventional economics says profit and loss differ merely in sign, neurology says otherwise.
All this matters because it undermines the key assumption behind the conventional idea that high risk must, on average, bring high reward. This assumption is that people dislike risk and so must be compensated for taking it on, in the form of higher expected returns. But prospect theory says this isn't always true. After losses, people might want to take risk as it offers them the best chance of getting even. And if people like risk, then risky stocks should offer lower returns, not higher ones - just as bookies and casinos give people bad average returns because folk like a flutter.
Messrs Wang, Yu and Yan have found that this is true for US shares. They have found that, between 1967 and 2011 in the US, "among the firms where average investors face capital losses, there is a robust significant inverted risk-return relation". This is exactly consistent with prospect theory.
It's quite possible that this also explains the pattern of UK returns in the last three years. In 2009, many investors were sitting on big losses, especially on high-beta stocks. An attempt to 'get even' and recoup those losses - the disposition effect - caused them to hold onto banks and miners, which in turn caused these to be overpriced, with the result that their returns since then have been disappointing.
This might imply that we could be about to see a more 'normal' relationship between risk and return. Now that fewer investors are sitting on large capital losses, there should be less risk-chasing behaviour. This could mean that high-beta stocks such as banks and miners are less overpriced now than they were in 2009. If so, they are more likely to outperform if the market does rise.
This, though, is not the only point. What we're seeing here is that economics can be a useful science.
It's sometimes said that economics isn't a 'proper' science because it doesn't yield good predictions.
Now, while it's true that economics doesn't help us forecast the future - as this is inherently unknowable - it does make testable predictions. Prospect theory predicts that the risk-return relationship will be negative after investors have incurred large losses. And this prediction is vindicated by the data. In this sense, behavioural economics at least is a genuine science. And, I'll add, a more useful activity than mere forecasting.