Working Paper: NBER ID: w19303
Authors: Daniel Paravisini; Antoinette Schoar
Abstract: We design a randomized controlled trial to evaluate the adoption of credit scoring with a bank that uses soft information in small businesses lending. We find that credit scores improve the productivity of credit committees, reduce managerial involvement in the loan approval process, and increase the profitability of lending. Credit committee members' effort and output also increase when they anticipate the score becoming available, indicating that scores improve incentives to use existing information. Our results imply that credit scores improve the efficiency and decentralize decision-making in loan production, which has implications for the optimal organization of banks.
Keywords: credit scoring; credit committees; randomized controlled trial; banking efficiency; lending decisions
JEL Codes: D23; G21; L23; O33
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| availability of a credit score (G21) | committee effort (D70) |
| anticipation of a score's availability (D84) | committee output (D72) |
| introduction of scores (Y20) | pool of borrowers or average loan characteristics (G51) |
| credit scores (G51) | capital allocation efficiency (D61) |