Repeated Games, Information Asymmetries, Multiple Lending, Relationship lending

C72, C73, C92, G21

The widespread evidence of multiple bank lending relationships in credit markets suggests that rms are interested in setting up a diversity of banking links. However, it is hard to know from the empirical data whether a rm's observed number of lenders is symptomatic of nancial constraints or rather a well-designed strategy. We design an experimental credit to analyze the determinants of multiple bank lending relationships, both from the demand and the supply side. Our results show that borrowers prefer multiple lending when they are credit rationed and unable to stabilize their lending source, whatever their risk level. Moreover, rationed borrowers are less likely to repay and display a higher tendency to switch between lenders. At the same time, we observe that the determinants of lending change according to the type of information available on the loan applicants. If their quality is not observable, only credit history and relationship length matter, while the borrowers' behavior clearly impacts the lending decision if information is complete. Our ndings support the view that the number of banking relationships is mainly determined by the supply side.