From Forbes, a piece highlighting Baker and Wurgler's paper (that's Baker in the picture):
Here's a provocative concept. Stodgy stocks offer better returns than highfliers.
That's the conclusion of Malcolm Baker of Harvard Business School, Brendan Bradley of Acadian Asset Management in Boston and Jeffrey Wurgler of New York University's Stern School of Business. The trio studied returns for the entire stock market and the 1,000 largest stocks from 1968 to 2008.
Their surprising conclusion: $10,000 invested in the riskiest 20% of all stocks, and with the portfolio refreshed monthly, would have shrunk 5.5% annually to less than $1,000 after inflation over the 41 years studied. The same amount invested in the least risky 20% would have grown to an inflation-adjusted $100,000, representing an annualized 5.8% return, or 1.6 percentage points above the S&P 500.
The piece highlights the preference towards lottery ticket stocks as driving down their returns. They say this is inspired by Kahneman and Tversky's, presumably his prospect theory, the same theory that Taleb claims is sympatico with his Black Swan theory that improbable events are underpriced. I don't blame Taleb for this confusion, but note that prospect theory is not a good explanation if it explains why risky assets are both over and undervalued.
The piece goes on to quote Baker:
"Living through a year when you're underperforming, it's hard to hang on to assets," says Baker, who is also a consultant to Acadian, with $50 billion under management. "Investors feel they are being left behind."
I created a low vol fund, the FalkenFund, to demonstrate this in 1996. After a couple years of good returns, I took it everywhere to try to get real money behind it, but by then the high risk stocks were skyrocketing and my strategy was lagging. Several times I remember showing it to people and they would say their quants backtested my idea and it did well, but it would have lost money over the past year, which made it impossible to sell.
Such popular pieces would have been useful when I was getting sued for using this idea in 2006, which was supposedly confidential information I could not use in any way. I remember spending several hours--at $500 an hour--going over this concept with my befuddled IP lawyer. Alas, it was rather pointless, because my ex-employer's demand I not use low volatility in any way morphed into 'low volatility in a way to be determined after of discovery', which was the same thing in practice, because it basically made everything I did potentially suspect. The whole point of constructing a rational defense against a party with bad faith is futile.