A paper by Guiso and Jappelli looked at Information Acquisition and Portfolio Performance. They develop a theoretical model that shows how an overconfident investor is actually made worse off, because they overestimate the precision of their signal. The really interesting part of their paper is a novel set of survey data from an Italian bank that shows people who are confident are generally worse investors than those who are ignorant. Investors experienced a lower Sharpe ratio the more they invest in investment information. It seems people use information to rationalize risk taking that is not commensurate with greater returns; a simple strategy of basic naïve asset allocation dominates the average active investor.

The Sharpe ratios use sample data that is perhaps unrepresentative, as they use a 14-year period to estimate the sample moments for bonds, stocks, and the risk free rate, which can be tricky. It would have been nice if they simplified their presentation of the data, and didn't have a theoretical model that pointed to another paper for key parts of the proof (not that the model looked really compelling). But I imagine the results are robust, because they comport with my observations. Clearly this is a tentative result, but thought provoking anyway.

Hat tip: Stumbling and Mumbling