Reduced Form Coefficients, Structural Coefficients, and Mortgage Redlining

Empirical studies of redlining have long been plagued with the identification problem. Observed behavior in alleged redlined areas is ultimately a combination of both demand and supply influences coming from both mortgage and housing markets. How then can inferences from reduced form experiments be utilized to provide unambiguous structural information concerning purely lender behavior? This problem is analyzed in detail with the help of an implicitly specified skeletal model of the mortgage and housing markets. The results suggest that structural inferences from reduced form information are, indeed controversial at best. Although this fact is fairly widely known at present, many research techniques still utilize an underlying reduced form methodology. It is hoped that the methodology employed in this paper highlights the identification problem in a manner that will generate greater insight into this issue. Some remedies are suggested in line with the general econometric literature dealing with structural identification.