We distinguish the impact of information technology adoption on information processing costs and agency costs by conducting a randomized control trial with a bank that adopts a new credit-scoring tool. The availability of scores significantly increases credit committees' effort and output on difficult- to-evaluate loan applications. Output increases almost as much in a treatment where the committee receives no new information, but anticipates the score becoming available after it evaluates an application, which suggests that scores reduce incentive problems inside the credit committee. We also show that scores improve efficiency by decentralizing decision-making and equalizing marginal returns across loans.
Committee Actions: Committee approves/rejects. Committee Effort: Evaluation time. Loan Outcomes: whether or not the loan was approved, amount approved and whether it matched the application amount, maturity approved and whether it matched the application maturity, disbursed amount, default after 6 months, default after 12 months, defaulted amount at 6 months and 12 months. Loan Prospecting and Branch Output: number of loans, amount of loans, fraction of defaults, fraction of amount that defaults.
Committee Actions: When scores are added as an input in the decision process, the number of cases in which committees cannot decide is reduced by 41.8%. The difference is positive but not significant. Observing a score decreases the probability of referring the application to the manager by 2.3 percentage points, a 48% decline reduction. Scores reduce the probability of collecting additional information by 1.7 percentage points, a 27% decline relative to the baseline. Committee Effort: Committees spend on average 16.2% more time per application when scores are available. Yet scores do not shift the entire distribution of evaluation times; credit scores increase the evaluation time on applications that take longer than the median time to evaluate. Put together, scores reduce the cost of deciding for any given default probability, and the reduction is larger for larger loan amounts, where the committee members have more at stake. Loan Outcomes: The effect of scores on the probability that the loan is issued is close to zero and not statistically significant. This implies that the addition of scores to the loan production process does not affect the overall extensive margin of lending. Scores don't have a significant effect on the average level of any of the measured outputs relating to loans - loan size, probability that loan amount and application are different, absolute value of the loan amount adjustment and default probability. Loan Prospecting and branch output: No statistically significant change in the score or requested loan amount of approved loans. Overall, scores do not affect total output in the short-run.