Performance Evaluation of Actively Managed Mutual Funds: With Focus on Active Share

Ulrik Galskov

Student thesis: Master thesis


This thesis uses two measures of active fund management, active share and tracking error, to evaluate whether the performance of actively managed mutual funds is related to the level of active management. By combining active share and tracking error, a methodology introduced by Cremers and Petajisto (2009), it is possible to sort active funds into five fund groups of active management, namely: i) Closet Index funds, ii) Factor Betting funds, iii) Moderately Active funds, iv) Concentrated funds, and v) Stock Picking funds. Based on a sample of 2,182 actively managed mutual funds with eight different investment areas in the period 02/28/1994 - 05/31/2015, it is found that the level of active management does matter in terms of performance. The performance evaluation on the total sampled funds showed that Closet index funds and Stock picking funds generated a superior benchmark-adjusted return before adjusting for fees, while the results were insignificant after fees. In order evaluate the performance of the sampled funds more accurately to account for potential different market characteristics, the funds have been further sorted according to specific investment areas. In this evaluation, the relationship between benchmark-adjusted performance and level of active management in the eight markets turned out to be ambiguous. Some of the highlights were that less actively managed funds, i.e. Closet Index funds, consistently underperformed their benchmark after adjusting for fees, whereas highly actively managed funds, i.e. Concentrated funds and Stock Picking funds, consistently outperformed their benchmarks before fees, while there was no statistical support after fees. In conclusion, as fund performance in practice is based on returns after fees, the thesis does not provide evidence that the two active measures, active share and tracking error in combination, can be used to identify high performing funds. Though, some of the findings suggest that the two active measures may be used to identify less actively managed funds that in some markets consistently underperform net of fees. Furthermore, it is found that the two measures, in particular active share, may have limitations in regards to the number of stocks held by the fund and markets with certain characteristics in terms of benchmark size. Consequently, evaluations of funds in small markets or funds with many stocks may be subject to a wrongly determined level of active management. The thesis suggests that practitioners going forward should take such circumstances into account by evaluating the level of active management based on a benchmark specific simulated mean active share rather than the industry standard threshold of 60% set by Petajisto (2013) for the US market.

EducationsMSc in Finance and Investments, (Graduate Programme) Final Thesis
Publication date2016
Number of pages123