Abstract
I exploit a series of natural experiments to provide quantitative estimates of diseconomies of scale in fund management. My evidence shows that such diseconomies exist, are quantitatively an order of magnitude higher than previously believed, and are large enough to match many workhorse models of mutual funds, like Berk and Green's (2004) model. The regression discontinuity setting, where variation in capital flow is generated by investors' attention to the Wall Street Journal "Category Kings" ranking of funds, also enables me to show that trading costs and non-scalability of investment strategy are two important sources of diseconomies of scale.