Supervising and Regulating the Use of Credit Repository Data Raphael W. Bostic Federal Reserve Board June 18, 2001
Introduction Federal Reserve oversees activities of institutions that use data at credit repositories Our concerns: –Data are accurate –Data are complete
Regulatory Structure What data can be kept –Prohibitions against collecting certain data –Examples: gender, race, marital status
Regulatory Structure What data can be kept How long data can be kept –Allows for opportunity to recover
Regulatory Structure What data can be kept How long data can be kept Who can access the data –Grounds for obtaining personal information –Legitimate business purpose required
Regulatory Structure What data can be kept How long data can be kept Who can access the data How are disputes handled –Critical for addressing accuracy concerns
Institutional Structure Repositories have an incentive to make their data complete –Profit motive: lenders want models that work best
Institutional Structure No regulatory requirement to report credit information A self-enforcement mechanism –Lenders do not have to report –Repositories do not have to provide access
Introduction to Regulation and Supervision Credit scoring and safety and soundness Credit scoring and fair lending enforcement
Credit Scoring and Safety and Soundness Credit scoring uses –Rank applicants according to credit quality –Estimate probability of default for individual applicants at portfolio level
Credit Scoring and Safety and Soundness If the scoring model does not work, an institution can face significant, and unforeseen, risks Could potentially results in institutional stress
Credit Scoring and Safety and Soundness Supervisors must monitor the performance of scoring models - “Model validation” In the US, model validation is done by specialists at the banking regulatory agencies
Credit Scoring and Safety and Soundness Validation objectives –Evaluate model stability, reliability, and accuracy –Conduct benchmark analysis/back-testing
Credit Scoring and Safety and Soundness Validation process –Evaluate sample design –Evaluate model design –Assess internal validation procedures
Credit Scoring and Fair Lending Ensure that discrimination against protected classes does not occur –Overt discrimination –Disparate treatment –Disparate impact
Credit Scoring and Fair Lending In the US, fair lending oversight is conducted by examiners of the banking regulatory agencies
Credit Scoring and Fair Lending Model validation –Was the model developed in accordance with fair lending laws?
Credit Scoring and Supervision and Regulation Final point –“I don’t know” is not an acceptable answer!