Performance Persistence And Determinants Of Indian Fund Of Mutual Fund PDF Download

This thesis contributes to existing literature by analysing the role of performance streaks in the US mutual fund industry. Existing research suggests that performance streaks, i.e. multiple consecutive months of positive or negative performance, are an important determinant of mutual fund flows. My dataset comprises monthly returns and net-flows from US equity mutual funds from 1996 through 2015. My first analysis shows that streaks are not an indication of performance persistence and should not be used in investment decisions. Next, I develop two forecasting models using streaks based on several different performance metrics, such as excess returns and CAPM-alphas. The first one is a probit model that forecasts future investor sentiment, measured by the sign of future net-flows. This model is very robust to different time period specifications. The second one is a multiple linear regression model that forecasts actual future net- flows. The performance of this model strongly depends on the time period specified, as it performs poorly following the financial crisis. In both models the best-performing specification uses streaks based on CAPM-alphas. However, a Shapley decomposition reveals that streaks are, despite being statistically significant, the least-important predictors of future net-flows. Instead, lagged net-flows are the most-important determinants of future net-flows. The results of this thesis suggest that active streaks tip the scales when investors decide between two or more funds with a comparable track record. Hence, the results presented are ambiguous regarding investor rationality.