This paper provides evidence on the effects of microfinance competition on moneylender interest rates and households' dependence on informal credit. The views among practitioners diverge sharply: proponents claim that the MFI competition reduces both the moneylender interest rate and households' reliance on informal credit, while critics argue the opposite. Taking advantage of recent econometric approaches to address selection on unobservables without imposing standard exclusion restrictions, we find that the MFI competition does not reduce moneylender interest rates, partially repudiating the proponents. There is no perceptible effect at low levels of the MFI coverage, but when the MFI coverage is high enough, the moneylender interest rate increases significantly. In contrast, a household's dependence on informal credit goes down after becoming an MFI member, which contradicts part of the critic's argument. The evidence is consistent with models where either the MFIs or the moneylenders engage in cream skimming, and fixed costs are important in informal lending.