Submitted by Guilherme Cordeiro (gvd.cordeiro@gmail.com) on 2013-01-10T13:03:53Z No. of bitstreams: 1 dissertacao_vies_sobrevivencia.pdf: 427527 bytes, checksum: 0950068bee911ee3c656393f8a193fdd (MD5) / Approved for entry into archive by Eliene Soares da Silva (eliene.silva@fgv.br) on 2013-01-10T14:19:53Z (GMT) No. of bitstreams: 1 dissertacao_vies_sobrevivencia.pdf: 427527 bytes, checksum: 0950068bee911ee3c656393f8a193fdd (MD5) / Made available in DSpace on 2013-01-10T14:23:07Z (GMT). No. of bitstreams: 1 dissertacao_vies_sobrevivencia.pdf: 427527 bytes, checksum: 0950068bee911ee3c656393f8a193fdd (MD5) Previous issue date: 2012-12-14 / Este trabalho tem o objetivo de analisar os efeitos da estratégia das gestoras de recursos em eliminar seus fundos com piores resultados a fim de preservar o melhor histórico de retornos possível para seu conjunto de fundos. Tal prática leva a superestimação dos retornos do portfolio existente de fundos, gerando uma diferença entre a performance do portfolio sobrevivente daquele composto por todos os fundos, o que se denomina viés de sobrevivência. O estudo se concentrou apenas nos fundos de ações brasileiros. Utilizou uma amostra total de 1.106 fundos, entre liquidados, incorporados e ativos. O período de estudo ficou compreendido entre junho de 2004 e junho de 2012. / The goal of this paper is to analyze the effects of the asset managers strategy of eliminating the funds with poor performance in order to preserve the best performance as possible of the remain funds. This practice leads to overestimating the return of a group of funds, generating a difference between the unbiased and the biased portfolio that is called survivorship bias. This study is focused on Brazilian equities funds. The total number of funds within the sample is 1.106, including liquidated, incorporate and active funds. The performance of funds was analyzed from June 2004 through June 2012. Regressions were realized by the Least Square method. For the equal weighted portfolios, the results signalized a positive bias for the end of period survivors (roughly 0,10% per month). The opposite was obtained to the full period survivors, indicating theirs worst performance compared to the unbiased portfolio. For the value weighted portfolios, despite being statistically significant, the bias is small (0,03% per month) with the end of period funds. The full period survivors, when compared with the unbiased portfolio, showed an insignificant result.