A re-examination of credit rationing in the Stiglitz and Weiss model

Abstract

To explain the widely observed phenomenon of credit rationing, Stiglitz and Weiss (1981) propose a theory of random rationing under imperfect information. With a simple model plausibly expanding the Stiglitz and Weiss setting, we argue that, random rationing occurs only in some extreme cases and hence is not likely to be a prevalent phenomenon. We start by illustrating that the Stiglitz and Weiss (1981) model and hence random rationing are quite sensitive to the assumption of the ranking of projects. Given that the ranking is according to the Mean-preserving Spread, there is adverse selection but no moral hazard. In the absence of moral hazard, random rationing is almost impossible to occur. Then by presuming the coexistence of adverse selection and moral hazard, we derive two required conditions for the occurrence of random rationing. First, random rationing occurs only if collateral has an overall deadweight cost other than the negative adverse selection effect. As collateral is a widely observed debt feature in practice, such an overall deadweight cost should not be the case for the majority of borrowers. Second, the occurrence of random rationing entails that the potential negative effects of the loan rate, collateral, loan size and any restrictive debt covenant simultaneously overweigh their positive effects exactly at the current contracting level. In this case, the zero-profit curve of the lender degenerates to a single point and borrowers face a take-it-or-leave-it offer. We conjecture that such a required condition leaves little space for the significance of random rationing.