Bill Sharpe: "Smart beta makes me sick"
If you rely on “smart beta” strategies to achieve returns that you hope will beat the broad market, then you also need a response to the criticisms posed by Bill Sharpe, the Nobel laureate and Stanford economist. Sharpe uses unassailable logic, in my opinion, to demonstrate why smart-beta strategies must eventually do no better than the market.
“When I hear smart beta, it makes me sick,” he said.
Sharpe spoke at the CFA Institute Annual Conference in Seattle on May 5.
He also explained why a static 60/40 asset allocation has important limits. But let’s look at his criticisms of smart beta first.
Sharpe did not offer a precise definition of which strategies he would include in the definition of smart beta. But in an email exchange, he wrote that it would at least include portfolios that overweight small-capitalization or value stocks, as well as alternative weightings, including fundamental indexing.
The question Sharpe posed for a smart beta portfolio is whether it can be good for everyone. If so, then that would imply that he and his friends are dumb, because they hold the market portfolio. Other investors who underweight those stocks that are overweighted by the smart-beta strategies must be “really dumb,” he said.
The market must therefore be divided between smart, dumb and really dumb investors.
“If that’s your story,” Sharpe said, “then smart beta is a way to exploit stupidity.”
“If so, then I would suggest that before too long, the really dumb ones will at least begin to choose index funds,” he said. “Maybe some of the Index-fund people will try to move in your direction,” at which point the advantage of smart-beta funds would be eliminated.
“I think there are all kinds of confusion out there,” Sharpe said in regard to smart beta. “I don’t think it will work in the future.”
He said he has seen what appeared to be empirical evidence of alpha, only to see that alpha fail to show up in other countries or disappear after a few years. He acknowledged that some strategies, like fundamental indexing, rely on proprietary methods and cannot be replicated mechanically, and therefore he cannot say for sure whether they would fail such a test.