Fund Flows, Performance, and Exit Under Dynamic Unobservable Managing Ability

. We introduce continuous-time learning models of dynamic unobservable fund manager abilities under a nonlinear framework, with risk-neutral or risk-averse investors. In equilibrium, sensitivities of inferred manager abilities to fund returns’ innovation shocks are time nonmonotonic, inducing time-nonmonotonic flow-performance sensitivities and convexities. Funds’ exit probabilities change with time and fund sizes at any fund age. Our empirical evidence of nonmonotonic flow-performance sensitivities and convexities, and old funds’ increasing exit probability with fund age support our nonlinear framework, which current linear frameworks cannot explain. We also offer insights into the current empirical controversy of whether flow-performance relations are linear or convex.