Firm size, credit scoring accuracy and banks' production of soft information

Abstract

Research on SME bank financing generally assumes that smaller firms are more opaque from a lender's perspective. We propose that the discriminatory power of credit scoring models can be thought of as a proxy for firm opaqueness, given that when these models perform poorly, lenders must invest in the production of 'soft information' to supplement the financial data used in these models. Measuring the discriminatory power of probit default models across quintiles of the Irish SME size distribution, we show that our proxy for firm opaqueness increases monotonically as firms get smaller. This finding supports an assumption that is the starting point to a wide strand of literature on SME bank financing. Our findings can also be interpreted as providing an insight to the literature on the determinants of banks' choice of lending technology. While smaller banks may, as found in a substantial previous literature, produce larger amounts of 'soft information' due to their organizational advantages, they may also do so out of <italic>necessity</italic>: hard-information-based default modelling is less effective among smaller firms, thereby forcing banks that lend to these borrowers to invest more in relationship banking technologies to retain competitiveness.