Assess whether notices match the actual decisioning reasons (337d16)
August 31, 2026
SITUATION Model-risk partner for credit scoring is responsible for notices match the actual in a lender expanding into majority-minority census tracts, using appraisal-gap outcomes in majority-minority tracts as the only working extract. A CRA PE that called the assessment area too narrow is what reset the timeline for this Fair Lending Examination and Notices file.
DECISION Model-risk partner for credit scoring 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 appraisal-gap outcomes in majority-minority tracts after a CRA PE that called the assessment area too narrow.
HYPOTHESES TO TEST 1. Authorize Remove access or reverse the item now; appraisal-gap outcomes in majority-minority tracts already has the discriminator after a CRA PE that called the assessment area too narrow. 2. Keep Temporary compensating control in force until appraisal-gap outcomes in majority-minority tracts is completed after a CRA PE that called the assessment area too narrow for model-risk partner for credit scoring. 3. Treat appraisal-gap outcomes in majority-minority tracts as Approve a documented exception because both readings appear after a CRA PE that called the assessment area too narrow. 4. Refuse a Fair Lending close: model-risk partner for credit scoring does not have the decision notices match the actual turns on in appraisal-gap outcomes in majority-minority tracts.
ANALYSIS REQUIRED 1. Check HMDA coding and underwriting policy against notices match the actual. 2. Compare appraisal-gap outcomes in majority-minority tracts to similarly situated files, second-review notes, and reason codes after a CRA PE that called the assessment area too narrow. 3. Flag any disparate-impact table model-risk partner for credit scoring cannot explain from appraisal-gap outcomes in majority-minority tracts. 4. For this Fair Lending Examination and Notices file, read appraisal-gap outcomes in majority-minority tracts against a CRA PE that called the assessment area too narrow and write the one fact that would move notices match the actual for model-risk partner for credit scoring.
RECOMMENDATION Choose Remove access or reverse the item / Temporary compensating control / Approve a documented exception / Hold on this Fair Lending / Examination and Notices packet (appraisal-gap outcomes in majority-minority tracts after a CRA PE that called the assessment area too narrow). The follow-on Examination and Notices action is what model-risk partner for credit scoring does next: implement the option, assign an owner, and log the missing fact.
COMMAND RETURNS - Bottom-line Fair Lending option on notices match the actual, then the evidence in appraisal-gap outcomes in majority-minority tracts, then the action for model-risk partner for credit scoring - Hypothesis scorecard against appraisal-gap outcomes in majority-minority tracts: supported / rejected / untestable - Regulatory or exam hook Examination and Notices would cite - Examination and Notices finding in appraisal-gap outcomes in majority-minority tracts that a second reviewer can re-perform
Explore more
More Fair Lending prompts
- Assess whether a redlining pattern exists after controls (e3ad72)
- Assess whether to pause a product pending a lookback (3b3c67)
- Assess whether the CRA plan is strategy or window dressing (c0edf7)
- Assess whether comparative files show second-review bias (101a02)
- Assess whether dealer overlays introduce prohibited steering (e014dd)
Explore related decision areas
- Assess whether telematics improvements offset driver quality (8f526d)Insurance Underwriting
- Assess whether umbrella attachment is too thin for the hazard (e2d9c9)Insurance Underwriting
- Chief AI officer must resolve whether the vendor can be used in a regulatedAI Governance
See governed multi-model AI on your own prompt
Compare GPT-5, Claude, and Gemini side by side, with human review and a decision record built in.

