Credit crunch? An empirical test of cyclical credit policy

In this paper we test the hypothesis that credit policies are pro-cyclical. Our approach is based on a stochastic frontier analysis of borrower data, as in Chen and Wang (2008). We extend the applicability of the approach, and propose a novel test specification which is informative of many types of pro-cyclicality. The analysis of representative samples of household borrowers during a huge cycle and its aftermath yields evidence of time-varying credit policy. We find that the focus of credit policy changed from collateral to current income during the cycle. Instead of a credit crunch, ie, an overall tightening of credit during the economic and financial contraction, we find a tightening of credit limits with respect to a minority of borrowers and an easing for the majority. In the course of the post-crisis period, credit policy became more lenient. Both the level of credit limits and the 'tailoring' of limits to group-specific characteristics of households increased. A reduction in the idiosyncratic variance of limits suggest that banks have become more consistent in their credit policies.