Case Study 22.1 — Know Before You Owe: How Two Forms Got Designed

DataField.Dev

Case Study 22.1 — Know Before You Owe: How Two Forms Got Designed

Type: Real public case — a federal rulemaking. Tier 1 for the statutory history, the dates, and the published forms. No statistics are invented; where a figure is not part of the public record, the case says so.

Background

For roughly forty years, a borrower buying a house in the United States received disclosures under two federal statutes that had been written independently, administered by different agencies, and never reconciled.

The Truth in Lending Act of 1968 was a credit-cost statute. Its purpose was to make the cost of borrowing comparable across lenders by forcing everyone to compute and disclose the same quantities: the annual percentage rate, the finance charge, the amount financed, and the total of payments. It was administered through Regulation Z, written by the Federal Reserve Board.

The Real Estate Settlement Procedures Act of 1974 was a settlement-cost statute. Its purpose was different: to stop kickbacks and referral fees in the settlement services industry, and to make the charges at a real estate closing visible in advance. It was administered through Regulation X, written by the Department of Housing and Urban Development.

Both statutes required a disclosure near the beginning of a transaction and another near the end. That produced four forms — the Good Faith Estimate and the initial Truth in Lending disclosure at the front, the HUD-1 Settlement Statement and the final Truth in Lending disclosure at the back.

The problem was not that there were four. The problem was that they did not correspond.

The Good Faith Estimate grouped settlement charges into blocks that the HUD-1 did not use, so a borrower could not line up an estimate against an actual.
The Truth in Lending forms used vocabulary — "amount financed," "finance charge" — that appeared nowhere in the RESPA forms, was used by no one at the closing table, and describes quantities that are counterintuitive on first encounter. ("Amount financed" is smaller than the loan amount. Most borrowers reasonably assume it is the loan amount.)
The two front-end forms were triggered at different moments by different rules.
Nothing on any of the four forms invited the borrower to compare it to any of the other three.

The consequence was predictable and, by the mid-2000s, well documented in complaint data and in industry practice: borrowers signed loans they could not describe. They knew a monthly payment. Many did not know whether their rate could change, whether their payment included taxes and insurance, whether there was a prepayment penalty, or how much cash they would need until they arrived at closing.

Then a very large number of those loans defaulted.

The issue

The Dodd-Frank Wall Street Reform and Consumer Protection Act, enacted in July 2010, created the Consumer Financial Protection Bureau and gave it an explicit assignment: combine the TILA and RESPA mortgage disclosures into a single, integrated set of forms. Congress did not merely permit the integration; it directed it, on a deadline, and it amended both underlying statutes to make the integration legally possible.

That framing matters. The Bureau was not asked to decide whether four forms should become two. It was asked to decide what the two forms should look like — which is a design problem, not a legal one, and the Bureau treated it as such.

The project was named Know Before You Owe, and the Bureau ran it in public, beginning in 2011, before the agency was fully stood up.

What the Bureau actually did

The methodology is the reason this case is worth studying, because it is not how financial regulation usually gets written.

It published prototypes and invited the public to compare them. Rather than drafting a form internally and publishing it once for comment, the Bureau posted competing prototype designs online and asked consumers, loan officers, brokers, settlement agents, and anyone else to look at both and say which communicated better and why. Successive rounds published revised designs that responded to what the previous round produced.

It ran qualitative testing with real people, in person, across the country. The Bureau retained an outside communications-research firm to design and conduct rounds of one-on-one testing in cities in different regions, with both consumers and industry participants, using realistic loan scenarios. A tester would put a form in front of a person and ask them to find something — the interest rate, whether the payment could change, the cash they would need at closing — and record whether they could, how long it took, and what they said while they looked. That is usability testing, and it is standard practice in software and almost unheard of in disclosure rulemaking.

It tested the forms against the forms they would replace. Before finalizing, the Bureau conducted a larger quantitative validation study comparing consumers' performance on the integrated forms against their performance on the existing GFE, TIL, and HUD-1. The point was to establish that the new forms were not merely different but measurably better at the specific tasks borrowers need to perform.

It published the research. The testing reports were released alongside the rulemaking, which means the design rationale for individual boxes on the form is part of the public record rather than an internal deliberation.

The timeline of the rulemaking itself:

KNOW BEFORE YOU OWE — THE PUBLIC RECORD                        [dates are public record]

  July 2010      Dodd-Frank Act enacted; directs integration of the TILA and RESPA
                 mortgage disclosures and amends both statutes to permit it.
  2011           Know Before You Owe launched. Prototype forms published online in
                 successive rounds; in-person qualitative testing begins.
  July 2012      Proposed rule published, with the testing reports.
  Nov. 2013      Final rule issued.
  Aug. 1, 2015   Original effective date.
  Oct. 3, 2015   Actual effective date, after a short announced delay.
  2017 / 2018    Amendments, including the change removing the timing restriction on
                 using a Closing Disclosure to reset tolerances (the "black hole").
  2020           The Bureau published an assessment of the rule under its statutory
                 obligation to review significant rules after five years.

What the design decisions show

Look at the finished forms with the testing in mind and the choices stop being arbitrary.

The second column on the Loan Terms table. "Can this amount increase after closing?" answered YES or NO for the loan amount, the interest rate, the monthly principal and interest, the prepayment penalty, and the balloon payment. Under the old regime, whether a rate was adjustable was disclosed — somewhere, in language. The new form converts a product feature into a yes/no question a person can answer under stress in a parking lot. Nearly every catastrophic surprise in the 2005-2008 period was a borrower who did not know one of those five answers.

The sentence at the top of the Loan Estimate. "Save this Loan Estimate to compare with your Closing Disclosure." It is on the form because the whole design theory of the rule is that the borrower performs a comparison, and a comparison requires the borrower to still have the first document.

The identical structure of the two forms. Same three boxes on page 1, in the same order, with the same labels. Same lettered sections A through J on page 2. The old GFE and HUD-1 could not be laid side by side; the new pair is designed for exactly that. Chapter 22's §22.8 is about the conversation this design makes possible.

The Calculating Cash to Close table on Closing Disclosure page 3. A three-column comparison — Loan Estimate, Final, Did this change? — that performs the borrower's comparison for them, line by line, with an explanation attached to every change. This is the single most consumer-protective piece of design in the rule, and it exists because testing showed that consumers who were told to compare two documents frequently did not.

Plain-language definitions printed on the form itself. "Annual Percentage Rate (APR). Your costs over the loan term expressed as a rate. This is not your interest rate." That last sentence is on the form because testers kept reading the APR as the interest rate. It is a designed response to a documented failure.

Outcome

The rule took effect on October 3, 2015. Two forms replaced four. Every closed-end consumer mortgage secured by real property — with a short list of exclusions including home equity lines of credit, reverse mortgages, and loans on dwellings not attached to real property — now runs on the Loan Estimate and the Closing Disclosure, and the model forms live in Appendix H to Regulation Z as H-24 and H-25.

The implementation was not smooth, and Case Study 22.2 takes that up. The Bureau's own later assessment of the rule described improvements in consumers' ability to locate and understand key loan information, alongside substantial one-time implementation costs to industry — which is roughly what you would expect from a rule that redesigned every mortgage disclosure system in the country at once.

What did not change is worth naming too. TRID redesigned the disclosure. It did not change what may be charged, who may be paid, or what a lender must verify. A borrower can receive a perfectly compliant Loan Estimate for a loan that is wrong for them. The forms make the loan legible; they do not make it good. That job still belongs to a person, and Chapter 13 is about that person being you.

The lesson

Three, and the first is the one that changes how you work.

The forms were designed around a specific human being doing a specific task under specific conditions, and that person is not a compliance analyst. They are two people in a parking lot with a phone, between shifts, three days before they are supposed to get keys. Every box on those five pages exists because somebody watched a person like that try to find something and fail. When you explain a Closing Disclosure, explain it the way the form was designed — by task, in the order the borrower cares about, from the numbers that determine whether they can go to closing.

Disclosure timing is a consumer protection, not a workflow. The three-business-day waiting period exists so the borrower has time to read, ask, and if necessary walk away. It was not put there to give the lender a buffer. A loan officer who treats it as slack in the schedule has taken something from the borrower without noticing.

Design is regulation. The most consequential decisions in this rule were not legal. They were choices about column headings, question phrasing, and what to print next to the APR. A rule that had been written the usual way — legally correct, tested by nobody — would have produced two forms as unusable as the four they replaced.

Discussion questions

The Bureau tested prototypes with consumers before writing the rule. Name two other mortgage disclosures or documents in this book that would benefit from the same treatment, and say what task you would ask a tester to perform.
"Can this amount increase after closing?" answers five questions with YES or NO. Identify a sixth question you believe belongs in that box, and argue for it — then argue the strongest case against adding it.
The rule made the disclosures legible without changing what may be charged. Is that the right division of labor for a consumer-protection statute? What does it assume about borrowers?
The Closing Disclosure prints "This is not your interest rate" next to the APR because testers misread it. What does it tell you about the APR as a disclosure that the form has to say this?
The Bureau ran the design process in public, with published prototypes and published testing. What did that cost, and what did it buy? Would you make the same trade in your own shop when rolling out a new borrower-facing document?
A colleague argues that TRID's real effect was to add cost and delay without changing borrower behavior. Using only what is in this case study, construct the strongest version of that argument — and then the strongest response.