Real Estate Data in Modern Mortgage Underwriting

For most of the past decade, the mortgage industry's approach to collateral data has been dominated by automated valuation models (AVMs). AVMs are useful, cheap, and fast, but they share a common weakness: they only answer one question, namely what is this property worth today. For risk teams operating in a higher-rate environment with thinner margins, that single answer leaves a lot of exposure unexamined. Property-level data — ownership history, assessed values, tax status, mortgage records, transaction history, and structural characteristics — fills the gaps AVMs cannot, and GSDSI's Real Estate Property Data product delivers those attributes across 155 million U.S. residential and commercial parcels.

Key Takeaways

  • AVMs answer only one question — today's value. Mortgage risk needs more: ownership, tax, lien, and transaction history.
  • 155M-record property data catches occupancy fraud at origination — a persistent source of loss severity per Fannie Mae's collateral-risk research.
  • Monthly property-data refreshes surface delinquent taxes, new liens, and quit-claim transfers weeks before credit bureau tradelines reflect the event.
  • Property data is public record and sits outside FCRA — a meaningful compliance advantage over credit-bureau attributes for this use case.

Collateral Validation at Origination

The first use case is collateral validation at origination. A lender pulling 155 million U.S. property records can cross-check the address, owner of record, lot size, and building characteristics on the application against the county-level source of truth. Mismatches catch everything from simple data-entry errors to deliberate misrepresentation. Occupancy fraud in particular — borrowers claiming a property is owner-occupied to access better rates when it is actually an investment property — remains a persistent source of loss severity on defaulted loans, and property-data cross-references are the single most effective pre-funding screen. The FBI's Mortgage Fraud Report has historically called out occupancy misrepresentation as one of the highest-frequency fraud patterns across both purchase and refinance channels.

Portfolio Monitoring After Closing

The second use case is portfolio monitoring. Once a loan is on the books, the collateral doesn't sit still. Property taxes can go delinquent, second liens can appear, the owner of record can change via quit-claim, and neighborhood transaction patterns can shift underwriting assumptions. A lender ingesting monthly refreshes of property data across their full servicing portfolio can surface these events weeks before they would show up in credit-bureau tradelines or borrower-initiated workouts. For HELOC and home-equity portfolios, where lien position matters enormously, this kind of monitoring has moved from nice-to-have to table stakes. The same dataset also feeds mortgage and refinance lead generation — the LTV and equity signals that identify a high-intent refi candidate are the same ones that flag a portfolio-monitoring risk.

Market-Level Risk Pricing

The third use case is market-level risk pricing. Traditional underwriting models price risk off borrower credit attributes and loan structure. Increasingly, sophisticated lenders are adding collateral-market signals to that baseline:

These signals refine loan-level pricing and capital allocation, which matters when basis-point differences determine whether a loan clears the hurdle rate. Lenders building this into their risk stack typically combine property-level data with the financial services solution portfolio their origination systems already integrate.

A Compliance Note on FCRA

A compliance note worth being explicit about: property data is public record. Because it describes real estate rather than consumers, it does not fall under the Fair Credit Reporting Act. Lenders can use it freely for underwriting, portfolio management, marketing, and risk analytics. That's a meaningful distinction from the consumer credit bureau products adjacent to these workflows, which carry FCRA obligations that shape how the data can be accessed and acted upon. Teams building out a modern property-data stack should confirm the source of each attribute in their vendor's documentation — genuine public-records sourcing versus reseller data with murky provenance is the difference between defensible and fragile. See also the companion piece on non-FCRA mortgage leads compliance in 2026 for the marketing-side analogue.

Frequently Asked Questions

Does property data fall under FCRA?
No. Property data describes real estate (parcels, deeds, tax assessments, ownership history) rather than consumers, and is sourced from public records. That places it outside the scope of the Fair Credit Reporting Act, which regulates consumer credit reporting agencies and the use of consumer credit information for FCRA-permissible purposes.
How does property data improve mortgage underwriting beyond an AVM?
AVMs answer one question — today's estimated value. Property data adds ownership of record, tax status, lien position, sale history, and structural attributes. That lets underwriters catch occupancy fraud, validate the collateral description on the application, and price risk against neighborhood-level transaction patterns, not just borrower credit attributes.
How frequently should a lender refresh property data for portfolio monitoring?
Monthly refresh cadence is the practical standard for active servicing portfolios. That's frequent enough to catch delinquent tax events, new second liens, and quit-claim transfers weeks before credit-bureau tradelines reflect them. Higher-touch portfolios (HELOCs in particular, where lien position is critical) sometimes move to a 2-week cadence.
What property attributes matter most for fraud detection at origination?
Owner of record, occupancy status (owner-occupied vs investment), address normalization against the parcel record, and recent transaction history. Mismatches between the application and the county-level record are the highest-precision fraud signal available pre-funding.