AUS represents the algorithmic layer of mortgage underwriting. Decades ago, every loan was underwritten manually by a human underwriter reviewing the file from end to end. Today, most loans go through an automated underwriting system that evaluates the data against the GSE’s published guidelines (or FHA/VA equivalents) and returns a recommendation.
The two main AUS systems
- Desktop Underwriter (DU) — Fannie Mae’s AUS. Used for conventional conforming loans intended for sale to Fannie Mae. Evaluates DTI, LTV, credit score, asset reserves, employment stability, property type
- Loan Product Advisor (LPA) — Freddie Mac’s AUS, previously called Loan Prospector. Functionally similar to DU but with subtly different acceptance criteria, especially around assets and reserves
- FHA TOTAL Scorecard — runs alongside DU/LPA for FHA loans
- VA AUS — different system for VA-guaranteed loans, often run within Lender appraisal management systems
Common AUS recommendation outcomes
- Approve/Eligible — the loan can close as submitted, subject to standard conditions (appraisal, title, etc.)
- Approve/Ineligible — credit risk is acceptable but the loan doesn’t meet a specific guideline (often LTV or DTI). Often resolvable with additional documentation or restructuring
- Refer/Eligible — credit risk needs human underwriter review. Common for borrowers with thin credit files, recent credit events, or unusual income
- Refer/Caution — significant credit issues; manual underwrite likely with stricter scrutiny
What changed in AUS over the last decade
DU and LPA have both moved toward more flexible analysis of borrower assets, employment patterns, and credit history. Modern AUS handles gig economy income, asset depletion qualification, and trended credit data better than 2015-era systems did. The result: more borrowers qualify on AUS without requiring manual underwriting, which speeds up the process and lowers underwriting costs for lenders.
