Why Hire Loan Officers? Examining Delegated Expertise∗

I develop and estimate a model of expertise that weighs the tradeoffs of delegating loan decisions to loan officers. While loan officers can screen soft information, their preferences may distort decisions away from the lender’s objective. The model features rich heterogeneity (e.g. risk aversion, willingness to exert effort, and overconfidence) and addresses possible endogeneity with a natural experiment. I can recover the loan officer’s primitives using loan decisions and resulting profits with data from a Chinese lender. I find that despite the costs, the average loan officer contributes three times his pay in additional annual profits compared to a counterfactual that maximizes profits conditional only on the borrower’s hard information. I further explore additional counterfactuals that eliminate some of the distortions due to preferences.