Routine empirical choices affect replicability and inference: evidence from mutual fund flows research

Kumar et al. (2015) report that mutual fund managers with foreign-sounding names attract less investor flow, a pattern which is consistent with taste-based discrimination. While I can reproduce their main finding using their sample, the result does not hold under independent sample construction, alternative name classifications and outlier-robust methods. My analysis finds that the original result is sensitive to a small number of extreme observations and classification decisions. This highlights how routine empirical choices can affect replicability and inference, and underscores the importance of economically motivated design strategies, particularly in filtering and classification.