INFORMED TRADING AND GAMBLING IN THE MUTUAL FUND INDUSTRY

Abstract

Cremers and Petajisto (2009) proposed a new measure of managerial skill for the mutual fund industry, which is the difference between the weight of each asset in the fund’s portfolio and the corresponding weight in the benchmark index disclosed by the fund’s prospectus. They coin this measure as Active Share. They show that Active Share is a good predictor of subsequent fund performance. However, in this thesis, I argue that high Active Share can result not just from a large degree of informed trading but also through gambling by the fund managers. I propose a decomposition of Active Share that effectively separates the truly informed fund managers from the gamblers. I show that only the informed component of Active Share predicts good subsequent performance of fund managers, outperforming the gambling funds by around 3.60% per year, and outperforming the closet indexers by around 2.40% per year. On the contrary, the gambling component results in poor subsequent performance of around -1.44% per year. High informed trading component also persist over time, where 62% of the mutual funds continue to have high informed trading component even after three years. These funds have a lower attrition rate than the rest at 2.05% after three years, compared to a maximum of 3.21%. An investor investing in funds with high informed trading component can expect to earn more than 2.30% in excess of the funds’ benchmark per year, or more than 7% within a period of three years. This paper hence suggests that within the actively-managed fund industry, there may be funds with superior stock picking skills that can be identified by their high informed trading component.