Case Study 1 — Rewriting the 1003: The URLA Redesign

Type: Real, public industry event — an agency-directed redesign of the residential mortgage application and its underlying data standard.

Accuracy note. The facts below are drawn from the public record of the redesign as published by the Federal Housing Finance Agency, Fannie Mae, and Freddie Mac. Implementation dates for the redesigned application were announced and then revised more than once, and the form itself continues to be updated. Dates are given where they are well established in the public record; where a detail is a matter of degree or has changed, the text says so. Work from the current published form and the current agency announcements, not from this chapter.

Background: a form that outlived its design

For most of the modern history of American mortgage lending, every conforming loan in the country started on the same piece of paper. The Uniform Residential Loan Application — Fannie Mae Form 1003, Freddie Mac Form 65 — was the industry's universal intake document, and its universality was its whole point. A loan can only be sold into the secondary market if the buyer knows what is in it. A standard application was the front door of that standardization: same questions, same order, same meanings, every lender, every state.

The form that the industry used up until the recent transition was old. Not old in the sense of being out of date on the wall — old in the sense that its structure had been essentially fixed for decades while everything around it changed. It was designed to be filled out with a pen, at a desk, by a loan officer sitting across from a borrower. It squeezed two borrowers into shared columns to save paper. It used Roman-numeral sections. It asked several questions in language that had drifted away from ordinary English — the declaration asking whether the borrower was "a co-maker or endorser on a note" is the one every trainer used as an example, because approximately no consumer outside the industry knows what that means, and a question a consumer does not understand produces an answer nobody can rely on.

Underneath the paper problem was a data problem, and it was worse.

By the 2010s essentially no application was actually being handwritten. Applications were being keyed into loan origination systems, transmitted to automated underwriting systems, sold to aggregators, and reported to regulators — as data. But the form had never been designed as a data specification. It had been designed as a page. So a great deal of important information was captured in free-text fields, in margins, in "other" boxes, and in continuation sheets, and different lenders and different software vendors made different reasonable choices about where to put things. The result was that two files describing identical borrowers could arrive at an aggregator carrying meaningfully different data, and nobody was lying.

The issue: three pressures at once

Three separate forces converged on the same form.

First, data quality for the secondary market. Fannie Mae and Freddie Mac had spent years building uniform data standards for other parts of the process — appraisal data, loan delivery data, collateral data — under the umbrella of the Uniform Mortgage Data Program, directed by the Federal Housing Finance Agency in its role as the enterprises' regulator and conservator. The application was the conspicuous gap. The data the loan was sold on was standardized; the data the loan was taken on was not.

Second, expanded fair-lending data collection. The Home Mortgage Disclosure Act requires lenders to collect and report information about mortgage applicants so that lending patterns can be examined for discrimination and for unmet credit needs. Amendments made by the Dodd-Frank Act substantially expanded what has to be collected, and the implementing rule required, among other changes, that applicants be given the opportunity to self-identify using disaggregated subcategories of ethnicity and race rather than only broad categories, and that sex be collected. Most of the expanded data collection applied to data collected beginning January 1, 2018.

The old application's "Information for Government Monitoring Purposes" section could not hold that. It was a small block at the end of the form with a handful of checkboxes. The new requirements needed real estate on the page and a defined data structure behind it.

Third, consumer comprehension. The regulatory posture of the entire post-crisis era — visible in the Loan Estimate and Closing Disclosure, which replaced disclosures written by lawyers for lawyers with forms tested on actual consumers — was that disclosure documents should be written so that the person signing them can understand them. The application had never been through that treatment. It was the first document a borrower ever sees, and it was written in the vocabulary of the 1980s mortgage desk.

What was done

FHFA directed Fannie Mae and Freddie Mac to redesign the application jointly, and the enterprises did the work together — which matters, because a form that the two enterprises disagreed about would have destroyed the standardization that made it valuable in the first place. The project produced two deliverables, and understanding that there were two is the single most useful thing a loan officer can take from this case.

The redesigned form. A restructured application organized into components rather than one continuous document: a full Borrower Information set for each borrower, signed by that borrower; an Additional Borrower component; a Lender Loan Information component that the lender completes and the borrower does not sign; and addenda including a continuation sheet and, where state law makes it relevant, an unmarried addendum. Within the Borrower Information component, nine numbered sections replaced the old Roman-numeral layout, with military service and demographic information each promoted to a section of its own.

Questions were rewritten in plainer language. The old "co-maker or endorser on a note" became a question about whether the borrower is a co-signer or guarantor on debt. Income was broken into explicitly labeled components — base, overtime, bonus, commission, military entitlements, other — rather than a small grid that different lenders filled in differently. Gifts and grants got their own subsection with a defined list of source categories instead of a free-text line. Real estate already owned got a structured section that could actually hold a portfolio.

Some things moved. The old form's "Details of Transaction" block — the lender's arithmetic showing purchase price, costs, credits, and cash required — came off the borrower-signed pages and into the lender-completed component, which is where it always belonged, because the borrower was never the author of those numbers.

The data standard. Alongside the form, the enterprises published the Uniform Loan Application Dataset (ULAD), a mapping of every field on the application to the industry's MISMO data standard, and specifications tying it to the enterprises' automated underwriting systems and loan delivery datasets. This is the part that does not appear on any page and that changed the industry more than the layout did. The form became, for the first time, a defined dataset with a published mapping, so that "monthly base income" means the same field, in the same place, in the same format, from intake through delivery.

The transition. The redesign was announced years before it took effect, and the timeline was extended more than once as the industry, its software vendors, and the enterprises worked through implementation. It was extended again in 2020 in response to the operational disruption of the COVID-19 pandemic and the extraordinary origination volumes of that period. Following an optional-use period, the enterprises' announced date for required use of the redesigned application for new loan applications was March 1, 2021. Verify current requirements and any subsequent revisions with the enterprises directly.

A related later change. The question of a borrower's language preference was among the most debated elements of the project, and it was ultimately handled on a separate document rather than on the application itself: the Supplemental Consumer Information Form, which collects language preference along with information about homeownership education and housing counseling. FHFA directed its use for loans sold to the enterprises with application dates on or after March 1, 2023. Confirm the current requirement and the current form with your compliance department.

What it shows

A form is a data specification whether or not anyone designed it as one. The old 1003's real defect was not that it looked dated. It was that thousands of institutions were using it as a database schema without agreeing on what the fields meant. Every mismatch that produced was eventually paid for by someone: by an underwriter reconciling a file, by an aggregator repricing a pool, by a regulator drawing a conclusion from data that did not mean what it appeared to mean.

Comprehension is a data-quality control, not a courtesy. The rewrite of the declarations into plain language was not a public-relations exercise. A question a borrower does not understand produces an answer with no information in it, and that answer travels — into an automated underwriting run, into a sold loan, into a reported dataset. This is exactly the mechanism §9.3 and §9.4 of the chapter are about, applied at industry scale. The form got better at asking. The loan officer's job is to be better than the form.

Fair-lending data collection drove design. It is worth being explicit about this, because the demographic section is the one loan officers most often treat as an afterthought. The expansion of that section was one of the three forces that rebuilt the entire application. The data exists so that lending patterns can be examined for discrimination, and its usefulness for that purpose depends on it being collected consistently and accurately at the point of application — by you, in an hour, with a borrower who does not know why you are asking. Chapter 25 takes fair lending apart in full.

Industry-wide change is slow, expensive, and mostly invisible from the sales floor. The redesign took years, moved deadlines more than once, and required every loan origination system, every point-of-sale product, every document vendor, every automated underwriting interface, and every delivery pipeline in the country to change together. Most loan officers experienced it as "the screens look different now." That gap — between what a change costs the industry and what it looks like from a desk — is worth remembering the next time a guideline change seems arbitrary.

Outcome

The industry converged on the redesigned application and its dataset. Two practical consequences show up in daily work.

The first is that a two-borrower file is now a substantially larger document, because each borrower gets a full set of sections. Borrowers notice. The Linden Street borrowers will sign more pages on day 5 than their parents did on an entire loan, and a loan officer who says "it's longer because you each get your own complete application now, instead of sharing columns" has answered the question honestly in one sentence.

The second is that the data has to be right at intake, because it flows downstream as data rather than as a page somebody rekeys. When the application was a form that a processor retyped into a system, there was an accidental second set of eyes on every field. That step is gone. What you key is what the automated underwriting system evaluates, what the aggregator receives, and what gets reported. §9.8's reconciliation is the replacement for the accidental review the old workflow used to provide by mistake.

Lesson

The form is the industry's shared vocabulary, and shared vocabulary is what makes a loan sellable. Everything about the redesign — the components, the plain language, the structured income breakdown, the expanded demographic section, the published data mapping — serves one purpose: that a fact recorded on a kitchen table in one state means precisely the same thing to an underwriter, an aggregator, an investor, and an examiner somewhere else.

Which means that the loan officer keying that fact is the first link in a chain, and the only one who ever met the borrower.

Discussion Questions

1. The old application's declaration asked whether the borrower was "a co-maker or endorser on a note." The current one asks whether the borrower is a co-signer or guarantor on any debt not disclosed on the application. Identify three specific ways a borrower could answer the old question honestly and wrongly, and say what the new wording fixes and what it still does not.

2. The "Details of Transaction" block moved off the borrower-signed pages and onto the lender-completed component. Argue both sides: what is gained by removing the lender's arithmetic from the document the borrower signs, and what is lost?

3. Expanded fair-lending data collection was one of the three forces that rebuilt the form. A colleague tells you the demographic section is "just paperwork." Answer them in under ninety seconds, using only what is in this case study.

4. The redesign gave each borrower a full, separately signed set of sections instead of shared columns. Name two ways this improves data quality and one way it makes the application interview harder.

5. The case study argues that when applications were retyped from paper into a system, the retyping provided an accidental second review that no longer exists. Design a deliberate replacement for it that a two-person branch could actually perform on every file. What would you check, who would check it, and when?

6. The language-preference question was ultimately placed on a separate supplemental form rather than on the application itself. What arguments would you expect on each side of that decision, and what does the outcome suggest about how much weight the industry places on the application's role as a legal instrument versus its role as a consumer-facing document?