The determinants of fund flows associated with capital allocation decisions of investors has been the object of numerous studies. Some studies find that past performance represents one of the most important information used by investors in their decision-making while others report that factors, such as past flows, age, fund size, risk, return ratio and fund manager characteristics can also be important determinants. The study of mutual fund flows rests on the assumption that investors are sophisticated Bayesian learners able to infer managerial skills and integrate them in the asset allocation decisions. Yet, empirical evidence shows that investors lack the financial literacy sophistication suggested by these models. In addition, investors may not be able to understand the non-linear relationships between the variables having an impact on fund flows.

In a first paper presented by Stefan Ruenzi (University of Mannheim), the authors use machine leaning (ML) methods to study the predictability of mutual fund flows based on mutual fund characteristics and other variables. Using OLS regression and ML methods, the authors are able to identify non-linear interaction effects that have an impact on predicted fund flows. In addition, they aim at differentiating between high and low performing mutual funds based on machine learning implied monthly fund flow predictions. Empirically, their results indicate that past flows over the last month, followed by the average flows over the past six and twelve months, the Morningstar rating, and the total net assets are the most important variables predicting flows. 

The second paper presented by Felix Wilke (Nova School of Business and Economics) analyzes textual data from equity analyst reports to understand the extent to which investors use these reports to learn about fund’s managerial skill and future prospects as well as the relationship between fund size and returns (decreasing returns to scale). Using the LMcD dictionary to analyze reports that accompany analyst rating for worldwide equity funds, the authors provide evidence that analysts provide valuable information to investors. In addition, they confirm that Morningstar Analysts Ratings represent an important predictor of fund flows. Finally, their results suggest that analysts tend to quantitatively underestimate the impact of fund size on returns.