Assess whether comparative files show second-review bias (2efb38)

Situation

A lender expanding into majority-minority census tracts cannot treat a vendor score change with no disparate-impact test as color commentary on model-reason-code mapping that does not match notices. Second-review underwriter must close comparative files show second-review from that extract under Fair Lending / CRA and Special-Purpose Programs.

Decision

Second-review underwriter in a lender expanding into majority-minority census tracts must choose Remove access or reverse the item / Temporary compensating control / Approve a documented exception / Hold using model-reason-code mapping that does not match notices after a vendor score change with no disparate-impact test.

Hypotheses to test

• The population in model-reason-code mapping that does not match notices is the one a vendor score change with no disparate-impact test named, so Remove access or reverse the item follows for this CRA and Special-Purpose Programs file.
• The population in model-reason-code mapping that does not match notices is adjacent only to a vendor score change with no disparate-impact test; Temporary compensating control is the honest Fair Lending call.
• A lender expanding into majority-minority census tracts already contained a vendor score change with no disparate-impact test before model-reason-code mapping that does not match notices arrived; no new CRA and Special-Purpose Programs path.
• Provenance on model-reason-code mapping that does not match notices after a vendor score change with no disparate-impact test is broken; do not pick Remove access or reverse the item or Temporary compensating control yet.

Analysis required

• Compare model-reason-code mapping that does not match notices to similarly situated files, second-review notes, and reason codes after a vendor score change with no disparate-impact test.
• Flag any disparate-impact table second-review underwriter cannot explain from model-reason-code mapping that does not match notices.
• Test a documented exception versus a pattern a lender expanding into majority-minority census tracts must defend.
• For this Fair Lending CRA and Special-Purpose Programs file, read model-reason-code mapping that does not match notices against a vendor score change with no disparate-impact test and write the one fact that would move comparative files show second-review for second-review underwriter.