Análise comparativa de modelos de mensuração de desempenho e sua influência na captação líquida de fundos de investimento em ações no Brasil

• Ano de defesa: 2017
• Autor(a) principal: Anderson Rocha de Jesus Fernandes
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Fundos de investimento; Modelos de precificação de ativos; Captação líquida; Performance; Fundos de investimento Brasil; Administração
• Link de acesso: http://hdl.handle.net/1843/BUBD-ANYQTQ

The purpose of this study is to analyze the relationship between third and fourth comoments (higher comoments) and risk factors and their adequacy in different models of mutual funds performance measurement, and its association with funds flows. The literature about asset pricing enumerates factors that may be relevant to the specification of stock and portfolios returns, including market return, size, book-to-market, and two others reported by Fama and French (2015): profitability and investment. It is also argued that characteristics of the distribution of probabilities of returns are relevant to their explanation. Thus, coskewness and cokurtosis can be priced in capital markets. These variables are used in this study in order to determine funds performance and flows. Sample is formed by funds that have stocks as the main asset in their portfolios. They are classified by ANBIMA as: Ibovespa Ativo, IBrX Ativo, Ações Livre (from April 2001 to April 2015), Dividendos, Setoriais, Small Caps and Sustenabilidade/Governança (from March 2008 to April 2015). Methodologically, this study is divided in two phases. The first one refers to the estimation of parameters that represent the sensitivity of funds returns to the factors, as well as the comparison of the CAPM, FFC, FF5 and FF5M models, with and without coskewness and cokurtosis, using fixed effects procedure, in order to argue about the model that best describes funds retunrs. The second one deals with the relationship between performance measured by the intercepts (significant alphas) of the regressions in the first phase and funds flows. Then, the models were reestimated using moving windows so that the alpha calculated on each of them represented funds performance in the next period. These alphas were then regressed along with total net assets and lagged flows. It was also estimated the relationship between flows and performance through cross section regressions with fund fees and ages as control variables. Results show that the coefficients of coskewness and cokurtosis have little relevance in determining performance and funds flow. Among risk factors, market, SMB, and MOM were the parameters that showed relevant parameters to the explanation of funds returns. The models in which they are contained, FFC and FF5M, were those that presented greater explanatory power to the specification of fund returns. In addition, there is also evidence of convexity in the relationship between performance and flows.