Hong and Sraer's Explanation of the Low Vol Anomaly

In 1977 Ed Miller proposed a simple model where greater dispersion in in beliefs generated a greater price for stocks because those holders of a stock are in the 95th percentile of valuation, and given a constant mean, a higher variance implies a higher 95th percentile. Key to this model is the short sales constraint, because otherwise sellers would see these assets as overpriced and short them. Another key is limited rationality, because people should simply not include high beta assets in their portfolio; the market portfolio is dominated by one that excludes high volatility assets.

Harrison Hong and David Sraer have a new version of their Speculative Betas paper (see here).   It is basically the Miller model though it's more dynamic, and they emphasize the 'new' finding that their model shows that when there is greater disagreement, there will be a greater low volatility premium.  I don't think that's really new, in that it follows pretty straightforwardly from the Milller model.

It's an alternative to the 'constrained leverage' model of Frazzini and Pederson. However, like Frazzini and Pederson I don't see how it's consistent with the below average returns. Lower than CAPM is different than lower than average. Further, it requires irrationality by those who don't have opinions on stocks, because it seems obvious that investors without a view should simply avoid high volatility stocks in their index funds. Thus, in both these models, low volatility investing should be much more popular than it is.

I think low volatility investing is a fringe strategy still because of tracking error, the fact that one underperforms 'the market' too much too often. Sure, over time it's a higher Sharpe ratio, but the real objective is a relative return, or an Information Ratio. This is an equilibrium, requiring no ad hoc constraints or massive irrationality if one presumes a relative status utility function, as I propose in my book The Missing Risk Premium.

When I documented the low return to highly volatile stocks, the main reason professors found it unconvincing was because it implied irrationality. I didn't think of the relative utility solution, and that basically only leaves theories with ad hoc constraints and massive irrationality. That was certain irrelevance circa 1993, but after 20 years several trends in the zeitgeist have changed a lot, in large part due to the increase in behavioral finance, Freakonomics, and the simple persistence of the low volatility cross-sectional fact.  But I guess I have the intellectual equivalent of Stockholm syndrome, as the irrationality obstacle was simply burned into my brain. I don't like results that imply massive arbitrage to this day.