The Relevance of Soft Information for Predicting Small Business Credit Default: Evidence from a Social Bank

This paper analyzes whether soft information improves predictive accuracy of a bank’s internal rating by better forecasting future credit defaults. We use a unique hand-collected database of 389 small loans granted by a French social bank (credit cooperative) dealing with (very) small businesses. Almost all the loans in our sample have an amount of under €250,000. These small loans are typical candidates for small business credit scoring, which is only based on hard information. Our study emphasizes the relevance of including soft information – in addition to hard information – to improve credit default prediction. Our paper contributes to the topical debate on the design of credit rating systems tailored to suit the lending technologies of different banks. This issue is of particular importance for non-conventional relationship banks such as social banks and credit cooperatives.