Information Losses in a Dynamic Model of Credit

Abstract

This paper examines dynamic information losses associated with loan terminations. We assume that the aggregated returns of current borrowers contain information about the mean returns to future borrowers. In a competitive loan market, the value of this information is not fully internalized by individual borrowers and lenders, and loan decisions fail to be first best. Introducing heterogeneous borrowers, who know their own risk characteristics better than lenders, safer borrowers are less willing to borrow when risk premia rise. As they cease borrowing, the information generated in credit markets becomes noisier and this tends to increase risk premia. The model produces alternating and persistent periods of tight and loose credit. IN A DYNAMIC ECONOMY, every business venture explores new terrain. The success or failure of any business holds economic consequences for other possible ventures and reflects on their likelihood of success: At successively aggregated levels, this proposition remains true: information about the current success of export ventures is relevant to future export ventures; the rate of failure of businesses in Montana is useful information to future investors in Montana. This informational mechanism lies at the heart of the paradigm of competition: in the absence of barriers to entry, firms enter markets which show high returns and dissipate the producer's surplus. The entrant, who initially makes the risky decision to create the market, provides information about this market to other competitors, as a by-product of enterprise. The provision of this information represents an externality. Because it is an externality, firms do not necessarily optimally provide it. Moreover, this information is not provided only by entrants. As long as markets remain uncertain, and current returns are informative about future returns, all present firms provide information to future firms. This paper presents a model in which aggregate returns from a class of risky projects are important for decisions about future loans. Current credit market decisions depend on the ex ante distribution of project returns, and in turn current decisions affect the production of information about future returns. The dynamic implications of this learning process are traced out in a simple model of credit under uncertainty. Three important results are obtained: