Using billions of observations on the locations of bank branches and firms in China, we measure lender-borrower distance by geographic information system (GIS) and find a non-trivial amount of distant lending. Distant borrowers are more likely to be connected to banks’ local borrowers. We use novel data of monthly internal loan rating changes to directly measure soft information by tracing whether banks downgrade ratings before delinquency. For connected borrowers, banks have better soft information and predict delinquent events more accurately. This effect is more pronounced for distant borrowers. Consequently, connected borrowers’ delinquent rate is lower. Our findings show that the inter-firm network facilitates banks to collect soft i nformation and manage risks, especially for distant borrowers. 

Keywords: Big Data; Distance; Firm Network; Soft Information​