Case Study 11.1 — The Ability-to-Repay Rule and the End of Stated Income

Type: real, public regulatory history · Sources: Tier 1 (statute and regulation), Tier 2 (industry practice descriptions)

Background: the era in which income did not have to be proved

For roughly a decade before 2008, a substantial share of American residential mortgage lending was originated under documentation standards that did not require the lender to verify what the borrower earned.

The products had industry names, and the names were candid about what they were:

• Stated income / stated assets (SISA) — the borrower wrote down an income figure; the lender did not verify it.
• No income / no assets (NINA) — neither was collected.
• No income, no job, no assets (NINJA) — the outer edge of the practice.
• Low documentation or "lite doc" — a reduced set, often a paystub or bank statements without the corroborating tax record.

They were not, in origin, tools of fraud. The stated-income product was built for borrowers whose income was genuinely difficult to document with a paystub — the self-employed, the commissioned, the seasonal, and small-business owners — and it was priced for the additional risk. That is the honest version of the rationale, and it is worth stating fairly, because the failure mode is more instructive if you understand why intelligent people built the thing.

What happened next is documented public record. The products migrated outward from the borrowers they were designed for, to borrowers who could have documented their income and were quietly encouraged not to, and finally to borrowers whose stated income bore no relationship to anything. The colloquial industry name for the result — "liar loans" — arrived from inside the business, not from outside it. When home prices stopped rising, the loans defaulted at rates the models had not contemplated, and the securities built on them repriced catastrophically. The 2008 financial crisis has many causes and this is one of the best documented of them.

The issue: what does it mean to lend responsibly?

The regulatory question that came out of the wreckage was narrower and harder than "should stated-income lending exist."

It was this: before a lender makes a thirty-year loan secured by a family's home, what is it obligated to know?

Note what that question is not. It is not "may the lender take risk" — lenders take risk in every loan. It is not "must the loan be repaid" — some loans will not be. It is a question about the basis for the lender's belief at the moment of consummation, and about whether that belief must rest on evidence a third party produced.

Note also the structural incentive the pre-crisis system had created, because it explains why a rule was needed rather than merely better judgment. As Chapter 1 lays out, a lender that sells a loan within weeks of closing does not bear the loss when the loan defaults years later. When the party deciding whether a borrower can repay is not the party that finds out whether they did, "should we verify?" stops being answered by self-interest.

What was done: TILA, Dodd-Frank, and Regulation Z

The Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010 amended the Truth in Lending Act to require, for covered residential mortgage transactions, that a creditor make a reasonable and good-faith determination — before consummation — that the consumer has a reasonable ability to repay the loan according to its terms.

The Consumer Financial Protection Bureau implemented that mandate in Regulation Z (the Ability-to-Repay/Qualified Mortgage rule, 12 CFR 1026.43), which took effect January 10, 2014.

Two features of the rule matter for this chapter.

First, it enumerates what must be considered. A creditor must consider a specified list of underwriting factors, which includes the consumer's current or reasonably expected income or assets (other than the value of the dwelling securing the loan), employment status, the monthly payment on the covered transaction, payments on simultaneous loans and mortgage-related obligations, current debt obligations including alimony and child support, the monthly debt-to-income ratio or residual income, and credit history. Read that list against §11.1 of this chapter: it is the same three questions, written by a regulator.

Second — and this is the sentence that ended stated income — it requires verification. The consideration must rest on reasonably reliable third-party records. Not the borrower's assertion. Not the loan officer's confidence. Records, produced by someone with no stake in whether the loan closes. That is why Form 4506-C exists in your file, why a written Verification of Employment goes to the employer rather than to the borrower, and why the answer to "can't they just tell me what they make?" is a regulatory one and not a preference.

The rule also created the Qualified Mortgage (QM) category — loans meeting product-feature restrictions and points-and-fees limits, which receive a presumption of compliance with the ATR requirement. The definition of the general QM category has itself been revised: it originally incorporated a debt-to-income limit together with a prescribed appendix governing how income and debt were to be calculated, and the CFPB subsequently replaced that structure with a price-based approach, with compliance dates that shifted more than once. Confirm the current rule text before relying on any description of QM, including this one. What has not changed across every version is the verification requirement.

The outcome, and what it actually changed on the desk

Three consequences, all of them visible in your daily work.

Documentation became the product. The pre-crisis question "which documentation level do you want?" disappeared for covered transactions. There is now essentially one documentation level for a first-lien residential mortgage, and it is "verified." Alternative-documentation lending did not vanish — Chapter 34 covers the non-QM market, including bank-statement programs for self-employed borrowers — but it operates as an explicitly non-QM product with its own liability posture and its own pricing, not as a shortcut on an ordinary file.

"Reasonably expected" acquired real meaning. The statutory phrase is current or reasonably expected income. That is precisely question 2, and it is why an underwriter may decline to count a bonus a borrower genuinely received: the question is not whether it arrived, but whether it is reasonably expected to keep arriving.

Liability moved. A violation of the ATR requirement carries consequences for the creditor, including the possibility of the violation being raised defensively in a foreclosure. That risk is why your underwriter is unmoved by your certainty about a borrower, and why an income calculation is re-performed by post-closing quality control rather than accepted because you did it.

The lesson for a loan officer

The habit this history should install is not caution. It is sequencing.

The pre-crisis practice was not primarily a matter of dishonest people; it was a system in which the verification came after the promise, or did not come at all. Everything in Chapter 11 is designed to reverse that order — the written VOE requested on day 7, the year-to-date reconciled on day 5, the question about a possible job change asked at application and again at day thirty.

Which produces the sentence worth carrying: an unverified income figure is not a smaller version of a verified one. It is a different kind of thing, and it will not become the first kind by being repeated confidently.

Discussion questions

1. The ATR rule requires consideration of "current or reasonably expected" income. Argue that this phrase, rather than the verification requirement, is what actually killed stated-income lending. Then argue the opposite.

2. The stated-income product was designed for borrowers whose income is genuinely hard to document — the self-employed and the commissioned. Those borrowers still exist. Using §11.4's W-2-versus-1099 distinction, describe what a responsible modern version of that product must do that the pre-crisis version did not.

3. Chapter 1 argues that a lender who sells the loan within weeks does not bear the default. Explain how that structural fact makes a rule necessary rather than merely desirable, and identify one place in your own daily workflow where the same incentive gap exists on a smaller scale.

4. The rule requires "reasonably reliable third-party records." A borrower's W-2 is a third-party record — the employer produced it — and a tax transcript is another. Why does a careful lender want both? What class of problem does the second one catch?

5. A colleague says: "ATR is a lender-liability rule. It has nothing to do with me; I just take the application." Rebut that in four sentences, using something concrete from §11.9 or §11.10.

Tier 1: the Dodd-Frank Wall Street Reform and Consumer Protection Act; the Truth in Lending Act; Regulation Z, 12 CFR 1026.43 (Ability-to-Repay and Qualified Mortgage standards), effective January 10, 2014; the CFPB as implementing agency; the documented 2008 financial crisis. Tier 2: descriptions of pre-crisis product types and industry nomenclature, which were widely used but varied by lender. No enforcement figures, loss statistics, or market-share numbers are asserted here. Confirm the current text of 12 CFR 1026.43 and its commentary before applying any of it to a live file.