Internet Appendix for “Asymmetric Information about Collateral Values”

In this section I present a theoretical model of the competition between differentially informed lenders to provide mortgage financing. This model formalizes the empirical predictions discussed in Section II of the published article. The model builds on the contributions in von Thadden (2004) and Hauswald and Marquez (2006), as well as Engelbrecht-Wiggans, Milgrom and Weber’s (1983) analysis of first-price sealed-bid common value auctions with differentially informed bidders. I first characterize the equilibrium interest rate offers of the integrated and non-integrated lenders. I then simulate the model to generate empirical predictions about each lender’s equilibrium collateral quality, and the observed interest rates charged by the non-integrated lenders. Houses: Houses cost $1, and can be either of high quality (θ = h) or low quality (θ = l). High-quality houses will be worth H > 1 with certainty next period. Low-quality houses will be worth L = 0. Final house value is observable, but house type θ is unknown ex-ante. The fraction of houses that is high quality, q, is common knowledge.