Investment Funds: Performance using Carhart Model and Data Envelopment Analysis

Abstract This study evaluates the performance of Brazilian investment funds in stocks, comparing real returns and performance indicators, both parametric and non-parametric. The estimation of performance and classification of funds were realized with a basis in observation of real returns and the alphas obtained from Carhart’s parametric linear regression model. The indicators were compared with data envelopment analysis’ relative efficiency scores, the Banker, Charnes and Cooper non parametric model. The alphas were estimated considering the impacts of market factors, size, value, and momentum on the actual returns of each fund and the scores based on systemic risk, available resources, costs and total risk and, as a product of analysis, the return. As pointed out in the literature, a fall in fund performance was identified, as well as a high alpha correlation with actual returns, low correlation of scores with returns, and even lower correlation between alphas and scores. This indicates that the choice of the evaluation measure to be used is an important step in the selection of assets. Theoretical portfolios based on efficiency and alphas were also identified, identifying superior performance of the portfolios based on efficiency, none of which surpassed the market portfolio.