Case Study 34.2 — The \$38,364 Assumption: A Borrower Placed in Non-QM Who Qualified Agency

Type: composite, constructed from documented industry patterns and from the file structures this book has already established. No real borrower, lender, or transaction is depicted. Every figure is illustrative and every program parameter is constructed. Tier 3 throughout.

Why a composite rather than a named case. The error this study describes — placing an agency- qualifying borrower into non-QM — is real, common, and almost never litigated, because the borrower never finds out. There is no enforcement action to cite because in the overwhelming majority of these files nobody broke a rule. Somebody skipped an analysis. A composite is the honest way to teach it, and it is the format §9 of this book's style guide reserves for exactly this situation.

Part 1 — The file that should have been conventional

The borrower

A fifty-percent owner of a two-person marketing agency organized as a limited liability company and taxed as a partnership. Two full years of returns. A 738 representative score. Buying a \$430,000** primary residence with **20% down (\$86,000), a \$344,000 loan, and \$640.00 of other monthly debts.

They walked into their first appointment holding a piece of paper their accountant had given them and a sentence they had already decided was true: "Our K-1 showed a loss last year. I know I can't get a normal loan."

What the first loan officer did

Looked at the K-1s. Confirmed the sentence. Quoted a 24-month bank statement loan.

That is the entire decision, and it took about four minutes. It was not made in bad faith. The borrower had pre-announced the conclusion, the K-1 appeared to confirm it, and the loan officer had a program that solved the stated problem. The bank statement analysis produced a very comfortable number — business deposits averaging \$58,400.00 per month after exclusions, a 50% expense factor, and a documented 50% ownership share, giving \$14,600.00 per month of qualifying income. Nobody questions a file that qualifies with that much room.

The loan closed at 9.125% with 1.750 points and 1.000% origination.

What the analysis would have found

Chapter 32 owns the cash flow worksheet; this is only its output. Applied to the same two returns:

24-month average \$8,695.83** per month. Most recent year alone **\$8,133.33. Income declined 12.15% year over year, so the underwriter uses the lower figure: \$8,133.33 per month.

Notice what happened to the loss. The K-1 reports a \$4,600 loss in year 2, and that is a true statement about the partnership's ordinary business income. It is also nearly irrelevant to this borrower's capacity, because the partner took \$78,000 in guaranteed payments before the partnership computed that loss, and because \$26,400 of the year's deductions were depreciation — a non-cash charge on equipment already paid for. The business bought equipment and paid its owner more. Neither of those facts makes the household poorer.

Run the conventional file at 6.875% (illustrative):

A 42.72% back-end ratio at 80% loan-to-value with a 738 score is an ordinary conventional file. Not a stretch, not an exception, not a compensating-factors argument. Ordinary.

What the assumption cost

THE COST OF NOT RUNNING THE ANALYSIS                 [composite - all figures
                                                      illustrative]
  Loan amount, both scenarios                              $344,000.00

                                  CONVENTIONAL        BANK STATEMENT
                                       6.875%                 9.125%
  ------------------------------------------------------------------------
  Monthly principal and interest      $2,259.84              $2,798.90
                                             difference:      +$539.06 /mo

  Origination, 1.000%                 $3,440.00              $3,440.00
  Points                                  0.000                  1.750
  Points in dollars                       $0.00              $6,020.00
  ------------------------------------------------------------------------
  Lender charges at closing           $3,440.00              $9,460.00
                                             difference:    +$6,020.00

  FIVE-YEAR COST OF THE ASSUMPTION
    Payment difference, 60 months     $539.06 x 60  =       $32,343.60
    Lender charges difference                       =        $6,020.00
    ------------------------------------------------------------------
    TOTAL                                           =       $38,363.60

  Over the full 360 payments the payment difference alone is $194,061.60,
  though no borrower in this position stays thirty years - which is exactly
  why the five-year number is the honest one to quote.

Thirty-eight thousand three hundred sixty-three dollars over five years, produced by four minutes of not looking. And because the loan closed, funded, and performed, no rule was broken, no complaint was filed, and the loan officer received a thank-you note.

How it came out

Two years later the borrower called a different loan officer about a refinance. That loan officer asked for the tax returns — as a matter of routine, because it is what you do — ran the analysis, and told the borrower they had qualified conventionally the entire time.

There is no remedy in this story. The borrower cannot recover the difference. What they can do, and what they did, is move every piece of business they influence — their own refinance, their business partner, their partner's brother, and a run of self-employed referrals from a network of people who all believe they cannot qualify — to the second loan officer.

That is Theme 4 operating in the direction nobody talks about. The relationship outlasts the transaction, including the transactions you handled badly.

Part 2 — The file that genuinely needed it

If Part 1 were the whole story the lesson would be "agency always works," and that is false and would get a reader in trouble in the other direction. So here is the same industry, the same documentation problem, and the opposite answer.

The borrower. Left a salaried role in the same field fourteen months ago and now runs their own shop. Prior W-2 income of \$96,000 per year, documented. One tax return exists, covering a partial year. Twelve months of clean business bank statements. Buying at \$580,000 with 20% down — a \$464,000 loan.

Why agency is unavailable. Not income. History. Agency guidelines generally require a two-year history of self-employment, with a narrow documented exception path in limited circumstances — verify the current Selling Guide, because the exception language has been revised. This borrower has fourteen months and one partial return. The automated findings ask for two years of returns and there is only one. There is no analysis that fixes this, because the missing item is time.

What was placed. A 12-month bank statement loan at 9.500% (illustrative; 12-month documentation typically prices worse than 24-month). Deposits after exclusions averaged \$29,400.00 per month; at a 50% expense factor and 100% ownership, qualifying income of \$14,700.00** per month. Principal and interest on \$464,000.00 at 9.500% is \$3,901.56**.

What made the placement professional rather than merely profitable. Four things, and all four were in the file:

1. The agency attempt was run and documented. A dated note: findings requested two years of returns; one exists; file is ineligible on self-employment history, not on income.
2. The cost was quantified for the borrower in writing — the §34.9 comparison, in dollars per month, before they committed.
3. The exit was named with a date. In ten months the borrower will have two full years of self-employment and a second return. The refinance conversation is calendared, not hoped for.
4. The absence of a prepayment penalty was confirmed and explained. This is an owner-occupied consumer loan; Regulation Z does not permit a prepayment penalty on a non-QM covered transaction. The exit is clean, and the borrower was told so rather than left to wonder.

That is what a correct non-QM placement looks like. It is not a different product from Part 1. It is the same product with an analysis behind it.

What these two files show together

The identical hour produced opposite answers. That is the argument of §34.10 in its cleanest form: the agency-first rule is not a preference for agency loans. It is a requirement to know which file you are holding before you price it, and the only way to know is to run it.

The limits of this case study

Three honest caveats, because a composite that overclaims is worse than no composite.

Not every declining-income file survives the analysis. The Part 1 borrower's income fell 12.15% and the underwriter used the lower year. A steeper decline, or a decline with no add-backs behind it, produces a number that genuinely does not qualify. The analysis is not a rescue technique; it is a measurement, and measurements sometimes come back short.

The bank statement number was not fraudulent. \$14,600.00 per month is what the program's arithmetic produced from real deposits. The problem was not that the number was false. It was that the cheaper number was never computed.

Automated findings, overlays, and reserves all matter and are not modeled here. A 42.72% back-end ratio is approvable in the abstract; whether this file receives an Approve/Eligible depends on the full picture, which Chapter 15 owns and this study simplifies.

Discussion questions

1. The Part 1 loan officer made a decision in four minutes based on a document that said what the borrower had already told them. Name the cognitive shortcut at work, and describe a specific process control — something you could put in a checklist — that would have caught it.

2. The borrower opened the conversation with "I know I can't get a normal loan." How much of the responsibility for this outcome belongs to the borrower's own framing? Does that change what the loan officer owed them?

3. The five-year cost was \$38,363.60 and no rule was broken. Should something in the rulebook reach this? Argue both sides, and consider what a rule requiring a documented agency attempt would cost in cases where the attempt is obviously futile.

4. In Part 2 the barrier was time, not income. List three other barriers that no amount of analysis can fix, and for each state what you tell the borrower to do while they wait.

5. Compare the four things that made Part 2's placement professional against a non-QM file you could imagine closing next month. Which of the four is easiest to skip under deadline pressure, and what does skipping it actually risk?

6. The second loan officer in Part 1 gained a stream of referrals by delivering bad news about a competitor's work. Where is the line between honestly informing a borrower and disparaging another originator, and what would you actually say?