Case Study 1 — When the CFPB Stopped Using DTI

A real, public regulatory case. Effective dates and threshold values in this area have been amended more than once; the account below describes the arc, and specific current values must be verified at the CFPB.

Background

Chapter 4 §4.6 argued that debt-to-income is a good ratio with structural blind spots. That is a practitioner's opinion. It is also, more or less, the conclusion the federal regulator reached about its own rule — and the story of how it got there is the most useful thing a new loan officer can read about DTI.

The Dodd-Frank Act required creditors to make a reasonable, good-faith determination that a borrower can repay. The CFPB implemented this as the Ability-to-Repay rule, and paired it with a category of loans — Qualified Mortgages — that receive a presumption of compliance.

The question was: what makes a loan a Qualified Mortgage?

The answer the CFPB originally adopted for the "General QM" category, effective January 2014, was a hard debt-to-income limit of 43%, computed under a prescriptive appendix (known as Appendix Q) that specified how income and debt had to be calculated.

The reasoning was defensible. DTI correlates with default. A bright-line limit is easy to administer, easy to audit, and hard to game. It gives lenders certainty about which loans carry the compliance presumption.

The operating issue

Three problems emerged, and each one is a version of §4.6.

First, a bright line at 43% is arbitrary at the margin. A file at 42.9% and a file at 43.1% are not meaningfully different in risk, and nothing about the underlying households changes as the computation crosses the threshold. But the legal treatment of the loan changes completely.

Second, DTI alone is a poor predictor compared to DTI in combination. A 45% ratio with a 780 credit score, twelve months of reserves, and twenty years of documented employment is a different proposition from a 45% ratio with a 640 score and no reserves. Automated underwriting systems had been evaluating files holistically for years, weighting compensating factors — which is exactly what §4.5 says. A single-variable cap discards that.

Third, Appendix Q did not match how the industry actually computes income. The prescriptive appendix diverged in places from agency guidelines, which meant a file could be underwritten as acceptable by Fannie Mae's rules and land above 43% under Appendix Q's arithmetic — the same borrower, the same documents, two different numbers.

There was an escape hatch, and it swallowed the rule. Loans eligible for purchase by Fannie Mae or Freddie Mac were treated as Qualified Mortgages regardless of the 43% limit — the "GSE Patch," a temporary category set to expire when the enterprises left conservatorship or on a scheduled date. Because the GSEs' automated systems approved plenty of files above 43%, a very large share of the market qualified through the patch rather than through the 43% test.

The rule's headline standard was, in practice, substantially bypassed by its own exception.

What happened

Facing the patch's scheduled expiration, the CFPB reconsidered the General QM definition and, in a final rule issued in December 2020, removed the 43% DTI limit and Appendix Q entirely, replacing them with a price-based test.

The new approach: a loan qualifies as a General QM if its annual percentage rate does not exceed the average prime offer rate for a comparable transaction by more than a specified amount — for first-lien loans, a threshold in the range of a couple of percentage points, with wider allowances for smaller loan amounts. (Verify current thresholds at the CFPB; they are set by loan size and have been amended.) The rule retains requirements to verify and consider income, assets, debts, and DTI — the creditor must still compute and consider the ratio — but the ratio is no longer a bright-line eligibility cap.

The mandatory compliance date was extended once amid the pandemic; the CFPB also created a Seasoned QM category under which a loan that performs for a defined period can attain QM status.

What it shows

1. The regulator concluded that price is a better single summary of risk than DTI. That is a remarkable statement and worth sitting with. The reasoning is that a loan's price already embeds the market's assessment of everything — credit score, loan-to-value, occupancy, documentation type, and DTI — because lenders price all of it. A price-based test therefore aggregates the compensating factors that a single-variable cap ignores.

2. It did not conclude DTI is useless. The rule still requires the creditor to compute and consider it. This is exactly the framing §4.5 recommends: DTI is an input to a judgment, not the judgment. Never tell a borrower they are declined "because of a number," and never tell them they are approved because of one.

3. Bright lines get bypassed rather than enforced. The GSE Patch was written as a temporary accommodation and became the primary path. This is a general lesson about rules: when a bright line is set in a place the market cannot live with, the market does not comply — it finds the exception. Watch for this pattern; you will see it again in overlays (Chapter 14) and in appraisal waivers (Chapter 18).

4. Prescriptive calculation rules that diverge from practice create two numbers for the same borrower. Appendix Q's divergence from agency income calculation meant the same documents produced different qualifying income depending on which rulebook applied. A loan officer's practical defense against this class of problem is the one this chapter teaches throughout: know which basis you are computing on, and say so.

The outcome for the practitioner

Stop quoting 43% as a limit. It is the single most persistent obsolete number in mortgage conversation. It was a General QM threshold, it was substantially bypassed while it existed, and it was removed. If you hear a colleague tell a borrower "you can't go over 43," they are working from a rulebook that has changed.

Know what actually constrains your file. In descending order of practical force: your investor's requirements, your employer's overlays, the automated underwriting system's recommendation, and the program's stated benchmarks. Chapter 14 and Chapter 15 take these apart.

Keep computing DTI carefully anyway. It is still required, still considered, and still the number an underwriter looks at first. The lesson of this case is not that the ratio does not matter. It is that the ratio is evidence, and the file is the case.

Discussion questions

1. The CFPB replaced a risk measure (DTI) with a market measure (price relative to average prime offer rate). Argue for that substitution, then argue against it. Which argument do you find stronger, and what would change your mind?

2. The GSE Patch was intended as temporary and became the main path to QM status. Describe the incentives that produced that outcome. Was anyone behaving improperly?

3. Appendix Q produced a different qualifying income than agency guidelines for the same documents. Using §4.2's "96-cent problem" as a model, explain why two rulebooks for one quantity is a structural hazard rather than a technicality.

4. This case argues that price "already embeds" the compensating factors DTI ignores. Identify a borrower characteristic that affects ability to repay and that price does not embed. (Hint: reread §4.6.)

5. A colleague says, "So DTI doesn't matter anymore." Write the two-sentence correction.