More evidence that efficient markets are a theoretical ideal and not yet a practical reality. From George Korniotis of the Fed and Alok Kumar of U of Texas:
In this study, we propose a demographics-based proxy for smartness and show that the portfolio distortions of “smart” investors reflect an informational advantage that generate high risk-adjusted returns. In contrast, the distortions of “dumb” investors arise from psychological biases because they experience low risk-adjusted performance...when portolio distortions are large, smart investors outperform the passive benchmarks by about 2 percent and the smart-dumb performance differential is over 5 percent.
Here, "psychological biases" are the types of investment behavior that any proponent of efficient markets would rail against -- holding concentrated portfolios, trading actively, or overweighting stocks that are geographically closer to where the investor lives. But these same "distortions" -- when a smart person does them -- can generate high returns.
Korniotis and Kumar can't measure smartness directly so they create a proxy for it based on demographic attributes like wealth, income, age, and education. Despite what a rough measure this is, their results hold up -- though at relatively lower signficance levels -- even after checking for alternate explanations like risk levels, stock selection, geographic location, or better access to insider information.
The following chart nicely illustrates the smartness advantage. It shows how the average portfolio a "smart" person would hold compares with the average dumb portfolio. The contents of the portfolio aren't adjusted over time, which explains the reversals after about the 40th month. Presumably, actively trading investors would have been out of those positions long before the three years was up: