In this paper we analyzed the performance of Brazilian multimarket funds. In order to do so, we used an alternative tool that is related to the study carried out by Amin and Kat (2003). This performance measure represents a non-parametric test that can be used to evaluate funds with non-normally distributed returns. Based on the bootstrapping technique, we created confidence intervals for the Amin and Kat (2003) test in order to ensure more reliability to the performance estimates of the funds, and this procedure could be a methodological contribution to new studies in this area. With data of 107 funds in the period between January, 2005 and August, 2011, and applying the multiple regression analysis, the main results showed that, for the net performance, there is no evidence of extraordinary value added by the funds. This result is in conformity with the market efficiency hypothesis, specifically in the semi-strong form. When it comes to gross performance (which refers to the performance measured before the deduction of management fees and performance fees), in economic scenes without financial restrictions, we found evidences of extraordinary value added by the managers of the funds in the sample. In other words, active management can generate extraordinary value in Brazil; however, in accordance with Jensen (1978), Fama (1991) and also with the already discussed by Castro and Minardi (2009), this extraordinary return is eroded by the costs that are necessary for its achievement. We also show that periods of financial crisis had a significant impact on the funds' performance. Such periods also showed a moderator effect between the fund's performance and its determinants, which is one of the main contributions of this paper. These findings may indicate relevant contributions to the theory about investment funds, as similar results about the moderator effect were not found in previous researches.
Crisis; Agency conflicts; Market efficiency; Bootstrap; Gross performance
