Appendix H — Worked Loan Scenarios

Twenty complete files, each one worked from the facts as they would actually arrive to a decision you could defend to an underwriter. Appendix A is the formula reference; this is its applied companion. Where a formula is needed here it is cited, not re-derived — the payment is §A.1, amortization is §A.2, the qualifying ratios are §A.5, mortgage insurance across the programs is §A.10.

Every one of these is a [constructed teaching example]. None of them is Linden Street, Cypress Court, Fulton Avenue, or Harlow Street; those four are worked in the chapters, and several scenarios here are deliberately harder versions of the same problem.

H.1 How to use this appendix

Work the file before you read the solution. That instruction is not a pedagogical courtesy, it is the entire design. Reading a worked loan is pleasant and teaches almost nothing, because the hard part of origination is never the arithmetic — it is knowing which arithmetic to do, in what order, and which number you are not allowed to trust. Cover the analysis, take the facts as they are printed, and write down three things: the qualifying income, the ratios, and the one fact you would verify before you said anything out loud to the borrower. Then read on.

What you are practicing is a reasoning order, not a calculation. Almost every wrong answer in this appendix comes from doing the right arithmetic on a number that had not been established yet. A payment computed from an income the underwriter will not count is a wrong payment. A ratio computed against a rent that has not been documented is a wrong ratio. Sequence is the skill.

The scenarios do not all end the same way, on purpose. Five of them end in "this loan does not work." Seven end in a restructure that looks nothing like the loan the borrower asked for. Five end with the answer that seemed obvious at the top of the page turning out to be the expensive one. If you find yourself expecting an approval every time, you are reading a book about origination rather than practicing it.

⚠️ EVERY GUIDELINE THRESHOLD IN THIS APPENDIX IS ILLUSTRATIVE AND PERISHABLE. Ratio benchmarks, reserve requirements, waiting periods, expense factors, DSCR minimums, income limits, loan limits, mortgage insurance factors, funding fee percentages, and every rate quoted are constructed for teaching. They are chosen to make the arithmetic instructive, not to report the market. Verify every one at the source before you quote it: the Fannie Mae Selling Guide, the Freddie Mac Seller/Servicer Guide, HUD Handbook 4000.1 and current Mortgagee Letters, the VA Lender's Handbook, USDA Rural Development notices, FHFA for loan limits, and — for anything non-agency — the specific investor's product matrix, in writing. What is durable in these twenty files is the structure of the reasoning. The numbers are scaffolding.

On the arithmetic. Payments are computed with the annuity formula in §A.1 and rounded to the cent. Amortization uses the servicer method (§A.2) — interest computed on the balance and rounded to the cent each month before subtracting. Totals of payments are the rounded payment multiplied by the term, and total interest is that total less the loan amount, which is the convention the whole book uses. Ratios are stated to two decimals, housing as PITI ÷ gross monthly income and back-end as (PITI + monthly debts) ÷ gross monthly income (§A.5).

H.2 The reasoning order every scenario follows

Before the first file, the sequence. An underwriter answers five questions (Appendix F §F.1); a loan officer working a file forward answers seven, and the order matters more than any single one of them.

THE ORDER A FILE IS ACTUALLY REASONED THROUGH

  1  WHAT INCOME MAY I COUNT?        stability, continuance, documentation.
                                     Not what they earn. What survives an underwriter.
  2  WHAT DEBTS MUST I COUNT?        the credit report, the application, the returns,
                                     the title search, and what the refresh will find.
  3  WHAT ASSETS ARE REALLY THERE?   sourced, seasoned, and still there after closing.
  4  WHAT IS THE PROPERTY?           value, type, occupancy, and -- for condos -- the
                                     PROJECT, which is a separate approval entirely.
  5  WHICH PROGRAMS ARE ELIGIBLE?    loan amount, LTV, occupancy, property, product.
                                     Eligibility fails separately from creditworthiness.
  6  WHAT IS THE PAYMENT AND THE     only now. Steps 1-5 are the inputs; running this
     RATIO?                          first is how loan officers promise things.
  7  WHAT BREAKS IT?                 name the weakness, then name the documented offset.
                                     If you cannot name the offset, you do not have one.

Almost every file below fails somewhere in steps 1 through 5, or survives on step 7. Very few of them are step-6 problems, and that proportion is roughly right for the job: the ratio is where trouble shows up, not where it comes from.

H.3 Scenario H-1 · The overtime that three methods value three different ways

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $198,000 · single-family · primary residence · under contract
  Borrower          one borrower, W-2, sterile processing technician, hospital
                    3 years 4 months, same employer · middle score 688
  Base pay          $26.50/hour, 40 guaranteed hours
  Overtime          W-2 year 1: $9,240 · W-2 year 2: $7,080
                    YTD, 8 months of paystubs: $5,760
  Debts             auto $392 · student loan $185 · revolving minimums $95
  Assets            $19,400 verified · no gift
  The problem       Another lender pre-approved them at $212,000. Mine says less.
                    The borrower wants to know who is right.

The analysis. Base first, because it is the only number here that is not a judgment: $26.50 \times 2{,}080 \div 12 = \$4{,}593.33$ (Appendix F §F.3.1 — the hours must be guaranteed). Then the overtime, three defensible ways:

The trend across the two completed years is down 23.38%, and the rule for variable income is use the conservative figure, and which one that is depends on the direction of the trend (Appendix F §F.3.2). Declining means the average is the optimistic number, so the answer is \$590.00 and the qualifying income is \$5,183.33. The other lender annualized eight months of a partial year that happens to be running hot.

At a 45% back-end cap that is \$2,332.50 of allowable obligations, less \$672.00 of debts, so \$1,660.50** of PITI. At \$720.00 it would have been \$1,719.00 — **\$58.50 more of payment, which at 6.750% with a 0.62% MI factor and this county's tax and insurance load is about \$7,200 of purchase price (roughly \$204,500 versus \$211,700).

The contract is \$198,000. Loan \$188,100 at 5% down, P&I \$1,220.01**, taxes \$198.00, insurance \$93.00, MI \$97.19 → PITI \$1,608.20. Housing 31.03%, back-end 43.99%**.

The answer. APPROVE. The file clears on the conservative income with room to spare, and it would have cleared on any of the three methods. Nothing about the outcome required winning the argument.

What makes this one hard. The temptation is to argue about the method, because the method is interesting and the borrower has been given a bigger number by someone else. But the method only mattered if they were shopping above roughly \$204,500 — and they were not. Compute the difference before you defend the position. The second difficulty is subtler: had they written at \$210,000 on the other lender's letter, this file would have failed at underwriting after the appraisal was paid for, and the loan officer who issued that letter would have been the last to know.

H.4 Scenario H-2 · The second year that fell twenty-six percent

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $420,000 · conventional · 20% down ($84,000) · primary
  Borrower          sole proprietor, marketing consultant, Schedule C, 6 years
                    middle score 771 · no derogatory credit of any kind
  Returns           YEAR 1  net profit $96,400 + depreciation $8,200
                            - meals exclusion $1,900          = $102,700
                    YEAR 2  net profit $68,300 + depreciation $9,400
                            - meals exclusion $2,100          =  $75,600
  YTD P&L           7 months, $41,300 net · unaudited, borrower-prepared
  Debts             $1,050/month · assets $131,000 verified
  What they say     "Last year was an outlier. Two clients left and I replaced them."

The analysis. The add-back logic is Appendix A §A.12 and Appendix F §F.5 and it is not the difficulty here. Run it and the two years produce \$102,700** and **\$75,600. The 24-month average is \$7,429.17**; the most recent year alone is **\$6,300.00; and the change is −26.39%.

This is the Fulton Avenue problem in a much harder form. Fulton Avenue declined 2.3%, and the underwriter's answer there was to drop from the average to the lower figure and proceed. At 26% the answer is not the same answer with a bigger number in it, because the question has changed. A 2% decline is noise in a stable business. A 26% decline is the file asking whether there is a stable business to underwrite at all — and the seven-month year-to-date, which annualizes to \$70,800 (\$5,900.00 a month), says the line is still going down.

P&I on \$336,000 at 6.625% is **\$2,151.44; with \$420.00 of taxes and \$140.00 of insurance the PITI is \$2,711.44. There is no MI at 80% LTV, the credit is spotless, and the borrower has \$131,000 in the bank. None of that reaches the problem.**

The answer. DECLINE. Not "decline at this price" — decline as a file today. At \$6,300.00 of income the program's ratio cap supports roughly \$1,785.00 of PITI, which is about a \$240,000 house at 20% down (P&I \$1,229.40 + \$240.00 + \$80.00 = \$1,549.40, back-end 41.26%) and would require \$48,000 rather than \$84,000 down. But recommending that is only honest if the income is countable, and while the trend is still falling an underwriter is entitled to conclude it is not. The professional answer is a dated plan: complete the year, file the return, and come back with a third data point.

What makes this one hard. Everything else in the file is excellent, and excellence in the other categories is exactly what makes a loan officer reach for a compensating factor that does not apply. Reserves offset thin cash. Credit depth offsets a derogatory. Neither offsets an income you cannot establish, because compensating factors adjust risk around a qualifying income — they do not create one. The second hard part is human: this borrower is not a weak borrower, and the conversation has to say so while still saying no.

H.5 Scenario H-3 · The bonus that has only been paid twice

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          shopping · pre-approval requested · primary residence
  Borrower          salaried $108,000/year · 6 years, same employer · score 744
  Bonus history     $16,000 paid February of last year
                    $11,000 paid April of this year
  Written VOE       "Annual incentive award. Discretionary. Based on company
                    performance. Not guaranteed and no continuance implied."
  Debts             $610/month · 10% down available · $52,000 verified
  What they want    "The bonus is $27,000 over two years, so that's $1,125 a month,
                    right? What can we buy?"

The analysis. The borrower's arithmetic is not wrong; it is answering a question nobody asked. \$27,000 ÷ 24 = **\$1,125.00 is what a 24-month average of a two-year bonus history looks like. But there is no two-year bonus history. There are two payments, fourteen months apart, and the second one is 31.25% smaller** than the first.

Watch what the divisor does, because this is the tell:

The divisor is a convention, not a measurement, and the convention presumes a history the file does not have. Add the VOE language and the direction of travel and the defensible answer is that this income is not established. Quote the file on base pay: \$9,000.00 a month.

Now price the error. At \$9,000.00 and a 45% cap the borrower supports **\$3,440.00 of PITI, which is about a \$457,000** purchase at 10% down. At \$10,125.00 they would support \$3,946.25 — about \$524,000. The pre-approval letter that counts the bonus sends this household shopping \$67,000 above what their file will actually close.

They buy at \$412,000. Loan \$370,800 at 6.750%: P&I \$2,405.00**, MI at a 0.30% factor **\$92.70, taxes \$464.00, insurance \$142.00 → PITI \$3,103.70. Back-end on base pay alone: (\$3,103.70 + \$610.00) ÷ \$9,000.00 = 41.26%.

The answer. APPROVE — on base pay, and the obvious answer was wrong. The bonus was never needed. It was only ever needed to justify a house the file could not carry.

⚠️ A pre-approval issued on income you have not established is a liability with a letterhead. The borrower does not experience it as a caveat; they experience it as a budget. By the time an underwriter removes the income, there is an executed contract, an appraisal invoice, a moving date, and a real estate agent who now believes you cannot count. Quote the number you can document, say out loud which income you excluded and why, and put it in the letter.

What makes this one hard. Two payments genuinely is a history — it is simply a history of something other than what the borrower thinks. Resisting the divisor is hard because 24 is the number the guideline names, and applying a named rule feels like following it. The transferable discipline is to ask what the rule presumes before you apply it: a 24-month average presumes 24 months.

H.6 Scenario H-4 · The rent that arrives as a liability

THE FILE AS IT ARRIVES                              [constructed teaching example]
  New purchase      $398,000 · conventional · 10% down · primary residence
  Borrowers         two, W-2, $8,900/month combined · scores 762 / 749
  Debts             auto $412 · student $268 · revolving $160  =  $840
  Departing home    keeping it as a rental. Note: $241,200 at 4.250%, 30-year
                    fixed, opened 6 years ago. P&I $1,186.56.
                    Taxes $270 · insurance $112 · HOA $85.
  The lease         executed, 12 months, $1,725/month, tenant takes possession
                    two weeks after closing. Security deposit not yet received.
  What they say     "The rental covers itself and then some, so this should
                    actually help us qualify."

The analysis. Rental income is a net concept and the borrower is reasoning in gross (§11.6). Two subtractions stand between \$1,725.00 and anything usable.

NET RENTAL INCOME -- DEPARTING RESIDENCE            [constructed teaching example]

  gross monthly lease rent                                       $1,725.00
  less vacancy and maintenance factor, illustrative 25%           - 431.25
  ---------------------------------------------------------------------
  adjusted gross rents                                           $1,293.75
  less the departing property's own PITIA
      P&I $1,186.56 + tax $270 + insurance $112 + HOA $85        -1,653.56
  =====================================================================
  NET RENTAL INCOME                                               -$359.81

  A negative result is not "zero." It is a $359.81 MONTHLY LIABILITY.

New loan \$358,200 at 6.750%: P&I **\$2,323.28, MI at 0.32% \$95.52**, taxes \$448.00, insurance \$140.00 → **PITI \$3,006.80**. Now the three versions of the same file:

Nine and a half points of ratio sit between the borrower's model and the underwriter's, and none of it is a disagreement about facts.

The answer. RESTRUCTURE. At 47.27% the loan does not work as presented. It works two ways: sell the departing residence (which also converts equity into a larger down payment and removes the liability entirely), or retire the \$412.00 auto loan and keep the rental — which costs cash the file needs and leaves the household carrying two mortgages against one income stream. Present both, in writing, with the arithmetic attached, and let the borrower choose.

What makes this one hard. The lease is real, the tenant is real, and the money will genuinely arrive — so the loan officer is not correcting a misunderstanding of fact but of accounting, which is much harder to deliver. Note also what has not been verified: the security deposit has not been received, and programs commonly require evidence of it, sometimes evidence of the tenant's first payment, and sometimes documented landlord experience before crediting a first-time landlord with rental income at all. Verify those before you count a dollar of it.

H.7 Scenario H-5 · The cheaper payment that also costs more

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $275,000 · primary residence · under contract
  Borrower          two borrowers, W-2, $6,400/month · representative score 662
  Debts             $585/month · $16,900 verified, no gift
  Taxes/insurance   $3,300/year taxes · $1,500/year homeowners
  Illustrative      conventional 95%: rate 6.875%, MI factor 1.03% at this
  pricing           score and LTV. FHA: rate 6.250%, UFMIP 1.75%,
                    annual MIP 0.55%. ALL CONSTRUCTED -- verify at the source.
  Horizon           "We'll be in this house six, maybe eight years."

The analysis. Two structures, and the arithmetic conventions are load-bearing: the annual MIP factor applies to the TOTAL loan (base plus financed UFMIP) while program LTV is measured on the BASE loan (§A.10).

FHA is \$154.16 a month cheaper**, needs **\$4,125.00 less down, and — at a 45% cap — is the only one of the two that qualifies. That is the file. But the total-cost sentence still has to be said, and said precisely:

Conventional PMI terminates automatically at 78% of original value — \$214,500 — which this schedule reaches at payment 140 (§A.4, Homeowners Protection Act). FHA MIP above 90% LTV runs for the life of the loan and the duration category is set at origination and never revisited. The figures above use the book's constant-factor convention (Appendix A §A.10); FHA's annual MIP actually recalculates on the declining balance, so treat the full-term figure as an approximation and say so.

The answer. APPROVE — FHA. It qualifies where conventional does not, it costs \$8,440.32 less in mortgage insurance over the horizon the borrower actually named, and it costs \$13,160.00 more if they stay thirty years. State all three sentences. Do not resolve the tension for them by predicting a refinance.

⚠️ Never sell a program on a refinance you cannot promise. "You'll just refinance out of the MIP in a few years" is a sentence that requires future rates, future value, future credit, and future income to cooperate. If the borrower's horizon genuinely is six to eight years, say that the arithmetic favors FHA on that horizon and that a longer stay reverses it. Then document the comparison in the file so the reasoning survives your memory of it.

What makes this one hard. The famous fact about FHA — that its insurance can run for the life of the loan — is true, and here it points the wrong way. The 662 score is doing the real work: it makes the conventional MI factor punitive enough that the "expensive" program is the cheaper one for the better part of a decade. A rule about a product is not a conclusion about a file.

H.8 Scenario H-6 · The fee nobody has to pay and the dollars that have to be left

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $318,000 · VA · zero down · primary residence · 1,640 sq ft
  Borrower          veteran, full entitlement, COE obtained
                    VA disability compensation confirmed on the COE ->
                    EXEMPT from the funding fee. Household of four.
  Income            W-2 $4,520/month + VA disability compensation $1,730/month
                    (non-taxable) = $6,250/month
  Debts             $720/month
  Taxes/insurance   $3,816/year taxes · $1,560/year homeowners · no HOA
  Rate              6.375%, illustrative · no monthly mortgage insurance on VA

The analysis. Two questions, and the second is the one almost nobody runs.

What the exemption is worth. A first-use funding fee at an illustrative 2.15% on \$318,000 is \$6,837.00**, which is ordinarily financed. Loan \$324,837.00 versus \$318,000.00 at 6.375% is \$2,026.56** versus **\$1,983.91 of P&I — \$42.65 a month**, or **\$15,354.00 across 360 payments, plus the \$6,837.00 of principal that never existed. Verify the current fee schedule and the exemption categories with the VA; both have been revised. And note the corollary: if entitlement to an exemption is established after closing and the fee was collected, a refund** is available (§17.4).

The ratio, and then the test that governs. PITI is \$1,983.91 + \$318.00 + \$130.00 = \$2,431.91**, with no monthly MI. Back-end: (\$2,431.91 + \$720.00) ÷ \$6,250.00 = 50.43%** — comfortably past the 41% guideline. On any other program in this book that is where the conversation gets difficult. On VA it is where the actual test begins.

RESIDUAL INCOME WORKSHEET                           [constructed teaching example]

  gross monthly income (all borrowers)                          $6,250.00
  less federal income tax        (illustrative)                 (  310.00)
  less state income tax          (illustrative)                 (   95.00)
  less Social Security and Medicare, 7.65% of the W-2 $4,520    (  345.78)
  ------------------------------------------------------------------------
  net take-home                                                 $5,499.22
  less proposed PITI                                            (2,431.91)
  less maintenance and utilities, 1,640 sq ft x $0.14/sq ft     (  229.60)
  less all other monthly obligations                            (  720.00)
  ========================================================================
  RESIDUAL INCOME, household of four                            $2,117.71
                                       per person, rounded:       $529.43

  The tax lines and the per-square-foot factor are ILLUSTRATIVE. The
  REQUIRED MINIMUM is published by the VA by region, household size, and
  loan-amount breakpoint, is revised, and is not reproduced in this book.
  Look it up for this region, this household size, this loan size.

Note the interaction the exemption creates: had the fee been financed, PITI would be \$2,474.56, the ratio 51.11%, and residual income \$2,075.06. The exemption improves both tests at once.

The answer. APPROVE — subject to the documented residual comparison. At a ratio this far above the guideline the underwriter must document that residual income exceeds the applicable published minimum by the required margin and justify the decision in writing; verify the current benchmark and margin with the VA. Do not eyeball the table and do not carry last year's figure in your head.

What makes this one hard. Two things, and both are habits rather than calculations. First, the non-taxable compensation: it may be eligible for a gross-up in the ratio (Appendix F §F.3.3, and the percentage is program-specific — verify), but the residual worksheet uses actual take-home, so grossing up there would be double counting the same tax advantage twice. Second, a 50.43% back-end reads as a decline to an originator trained on conventional files, and the reflex is to start restructuring before running the test that actually governs. On a VA file the ratio is the preliminary; residual income is the exam.

H.9 Scenario H-7 · Three hundred ninety dollars over a ceiling

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $234,500 · USDA guaranteed · zero down · address CONFIRMED
                    eligible on USDA's map on the day of application
  Borrowers         two, W-2, $62,880/year combined ($5,240/month)
                    scores 704 / 688 · debts $410/month
  Also in the home  an adult child, 19, part-time, $1,150/month
                    the borrower's mother, Social Security $1,410/month
  Two children      ages 9 and 12
  Assets            $3,100 verified
  Discovered        on day 26, when the underwriter asked who lives in the home

The analysis. USDA applies two different income calculations to the same transaction (§17.9), and this file passes one and fails the other.

Repayment income counts only the borrowers: \$5,240.00 a month, documented the ordinary way, and it is comfortably sufficient for a \$234,500 loan (P&I at an illustrative 6.500% is **\$1,482.20**).

Adjusted household income counts the annual income of every adult household member, on the loan or not, related or not, used to qualify or not — then allows specified deductions.

ADJUSTED HOUSEHOLD INCOME                           [constructed teaching example]

  borrowers                                                      $62,880
  adult child, part-time ($1,150 x 12)                            13,800
  mother, Social Security ($1,410 x 12)                           16,920
  ---------------------------------------------------------------------
  total household income                                         $93,600
  less dependent deduction, 2 children (illustrative $480 each)     (960)
  less elderly household deduction (illustrative)                   (400)
  =====================================================================
  ADJUSTED HOUSEHOLD INCOME                                      $92,240

  Published limit, this county, this household size (ILLUSTRATIVE
  -- verify the current figure with USDA Rural Development)      $91,850
  ---------------------------------------------------------------------
  OVER THE LIMIT BY                                                 $390

The answer. DECLINE. Three hundred ninety dollars, and the program is gone. There is no appeal, no compensating factor, and no exception request, because nothing was misapplied — this is the one program in the book where a household can be denied for earning too much, and the limit is a ceiling, not a benchmark.

The fallbacks do not rescue it either. FHA needs 3.5% of \$234,500 = **\$8,207.50 before closing costs; a 3%-down conventional needs \$7,035.00**. The household has **\$3,100**. The honest answer today is that there is no loan here, and the plan is a savings target with a date on it, a recheck of down-payment assistance eligibility (which carries its own income limits — verify), or a lower price band.

⚠️ You may not reshape the household to fit the limit, and you must not hint at it. Omitting a resident adult from the household calculation is a misrepresentation on a federally guaranteed loan. If a household member genuinely moves out and establishes a separate residence, that is a change of fact to be documented in the ordinary course — not a strategy to suggest, and never a sentence you say first. Chapter 27 covers where this line is and what sits on the other side of it.

What makes this one hard. The second calculation is invisible unless you ask, and the question that finds it takes eleven seconds: "Who else will be living in the home — everyone, including adult children, parents, and anyone else, whether or not they're on the loan?" This file was discovered on day 26 with an executed contract and a paid appraisal. The arithmetic was never difficult. The intake was.

H.10 Scenario H-8 · Fourteen hundred fifty dollars into the next category

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $960,000 · primary residence · 20% down planned ($192,000)
  Loan              $768,000
  Borrowers         two, W-2 and salaried, scores 771 / 764 · debts $1,140
  Liquid assets     $248,000 verified after the down payment source is traced
  Taxes/insurance   $12,000/year taxes · $3,840/year homeowners
  The limit         assume a one-unit conforming loan limit of $766,550 for this
                    county [ILLUSTRATIVE -- verify the current figure at FHFA;
                    it is revised annually and high-cost counties carry a
                    higher ceiling with a high-balance tier in between]
  Illustrative      conforming 6.875% · jumbo 7.250% [CONSTRUCTED]
  pricing
  The ask           "It's fifteen hundred bucks. Just do it as a jumbo."

The analysis. The loan is \$1,450.00 over the illustrative limit. That is the entire problem, and it is worth what a small used car is worth:

| | Conforming \$766,550 | Jumbo \$768,000 | |---|---|---| | Rate [constructed] | 6.875% | 7.250% | | P&I | \$5,035.69 | \$5,239.11 | | PITI (taxes \$1,000 + insurance \$320) | \$6,355.69** | **\$6,559.11 | | Monthly difference | — | +\$203.42 | | Over 60 months | — | +\$12,205.20 | | Over the full term | — | +\$73,231.20 |

But the price is not the binding constraint. Reserves are.

The answer. RESTRUCTURE — and the obvious answer was wrong. Adding \$1,450.00 to the down payment moves the loan to exactly the limit, and a loan at or below the limit conforms. That single adjustment buys a lower rate, a \$203.42 smaller payment, \$12,205.20 over five years — and, decisively, converts a file that fails the jumbo reserve requirement into one that clears the conforming requirement by a wide margin. "Just do it as a jumbo" would have declined this loan.

What makes this one hard. A guideline threshold is a cliff, not a slope, and the instinct near a cliff is to price the difference rather than to step back from the edge. The second difficulty is knowing there may be three categories rather than two: many high-cost counties carry a high-balance tier between the baseline limit and the ceiling, with its own pricing adjustments and its own guidelines. Whether this county has one is a lookup, not a memory — and if it does, the answer changes again.

H.11 Scenario H-9 · The unit is fine. The project is not.

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $284,900 · condominium unit · conventional · 20% down
                    ($56,980) · loan $227,920 at 6.750% · P&I $1,478.28
  Borrower          score 762 · $7,700/month · debts $690
  The payment       P&I $1,478.28 + taxes $285 + insurance $95 + HOA $385
                    = $2,243.28 · back-end $2,933.28 / $7,700 = 38.09%
                    reserves 11 months · Approve/Eligible on the findings
  The unit          renovated, occupied by the seller, appraises at contract
  The questionnaire returned day 22 by the management company:
                      96 units total
                      30 units investor-owned (31.25%)
                      one owner holds 12 units (12.50%)
                      HOA reserves $40,000 against a $486,000 annual budget
                      special assessment APPROVED, $8,400 per unit, facade
                      pending LITIGATION: association v. developer,
                        construction defects, water intrusion, active
  HOA dues          $385/month

The analysis. Nothing in the borrower's file is a problem. The findings are clean, the ratios are comfortable, the reserves are deep, and the appraisal supports the price. The loan is dead anyway, because a condominium loan requires two approvals and this is a failure of the second one (§5.9, §14.3, §21.10).

Each threshold is program-specific, revised, and layered with lender overlays — verify each in the Selling Guide, the Seller/Servicer Guide, or Handbook 4000.1 for FHA project approval. But note that the last row does not need a threshold. Litigation over construction defects goes to whether the collateral is what the appraisal says it is and who will pay to make it so, and it is the item most likely to close every agency door at once.

The answer. DECLINE. And say precisely what was declined: the project, not the borrower. A larger down payment does not fix it. A higher score does not fix it. Moving to another lender does not fix it if the failure is an agency rule rather than an overlay — which is exactly the question to ask, in those words (Appendix F §F.13). There are portfolio and non-warrantable condominium lenders who will consider it, at a higher rate and a larger down payment, and that is a legitimate conversation to have. It comes with a disclosure the borrower is owed: the same project problems will meet their buyer when they sell. A unit that only a portfolio lender will finance has a smaller resale market, and that is a cost of ownership, not a financing detail.

What makes this one hard. The reflex is to hunt for the borrower-side fix, because borrower-side fixes are what a loan officer controls. There is none here. The operational lesson is calendar discipline: order the project review the day the contract is executed, not when the file goes to underwriting. This questionnaire came back on day 22 and killed a loan that was otherwise ready — the same information on day 3 would have cost the buyer nothing but a decision.

H.12 Scenario H-10 · The value that arrived eighteen thousand dollars light

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $262,000 · FHA 203(b) · primary residence
  Borrower          score 651 · $5,900/month · debts $525 · $14,900 verified
  Planned           3.5% minimum required investment $9,170
                    base loan $252,830 · seller credit $6,000 toward costs
  The appraisal     returns at $244,000. $18,000 low. 6.87% under contract.
                    Comparables are defensible; two are superior in GLA and
                    were adjusted downward, which the report documents.
  Contract          FHA amendatory clause present and signed

The analysis. LTV is computed on the lesser of price or appraised value (§A.4), which on an FHA file is the adjusted value. Every downstream number moves:

**The gap is \$17,370.00**, and note that it is *not* \$18,000. The value shortfall is \$18,000, but the required investment fell by \$630.00 because 3.5% of a smaller number is smaller: $\$18{,}000 - \$630 = \$17{,}370$. Getting that reconciliation right is how you demonstrate to a skeptical agent that you have actually run the file rather than repeated the headline.

Three levers, in the order you pull them:

1. Reconsideration of value. Submit comparables the appraiser did not use, with data, through the proper channel — never a phone call to the appraiser and never a demand for a number (§18.2, appraiser independence). This one has a documented adjustment rationale and is unlikely to move.
2. The amendatory clause. On an FHA transaction the buyer has the contractual right to walk when the value comes in below the contract price without forfeiting earnest money (VA's escape clause is the parallel — verify the exact form language for the transaction in front of you). That right is the borrower's leverage in the renegotiation, and it is why the third lever usually works.
3. Renegotiate, then split the difference in cash.

The seller comes to \$249,000. The base loan does not change — it is governed by the appraised value, not the price — so the required cash is \$249,000 − \$235,460.00 = \$13,540.00. UFMIP of 1.75% is \$4,120.55 financed, total loan **\$239,580.55, P&I at an illustrative 6.375% is \$1,494.67**, annual MIP at 0.55% on the **total** loan is **\$109.81, plus \$285.00 taxes and \$118.00 insurance → **PITI \$2,007.48. The borrower has \$14,900 and now needs \$13,540, leaving \$1,360.00 — 0.68 months of reserves.

The answer. RESTRUCTURE. It closes. It should also make you uncomfortable, and you should say so out loud, because a household moving into a home with two-thirds of one payment in the bank is one water heater from a missed payment.

⚠️ An FHA case number attaches the appraisal to the PROPERTY, not to the lender. Moving the file to a different lender does not buy a new opinion of value for the applicable period; the appraisal follows the property (§18.10, §16.6). A borrower who has been told otherwise by someone helpful is about to spend three weeks discovering it. Say it on the day the value comes in.

What makes this one hard. The instinct is to fight the appraisal, because the appraisal is what changed. But the appraisal is usually the one number in the file that a third party has documented under a standard, and a defensible report rarely moves. The skill is to pivot within a day to the lever that actually has give — the contract — while the seller still has a motivated buyer and before the buyer has decided the whole transaction is cursed.

H.13 Scenario H-11 · A deposit with an innocent explanation and no paper

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $372,000 · conventional · 10% down · primary residence
  Borrowers         two, W-2, $10,400/month · scores 758 / 744 · debts $1,020
  Assets            $62,400 verified across two accounts
  On statement 1    a $9,000.00 deposit and, the next business day, a
                    $5,600.00 deposit -- $14,600 total, 19 days before
                    application, no payroll match, no prior pattern
  Explanation       "I sold my truck to a guy at work. He paid me cash."
  Earnest money     $4,000, already delivered, separate from the $62,400

The analysis. The consequence of an unsourced deposit is not a decline — it is subtraction (Appendix F §F.4.1). Run the file at all three possible outcomes before you ask the borrower for anything, so the request you make is proportionate to what is at stake.

Loan \$334,800 at 6.750%: P&I **\$2,171.51, MI at 0.32% \$89.28**, taxes \$372.00, insurance \$130.00 → **PITI \$2,762.79. Cash to close is \$37,200.00 down plus \$11,400.00 of costs and prepaids, less the \$4,000.00 earnest money already delivered = **\$44,600.00.

The borrower produces a state title-transfer record and a bill of sale showing a \$9,000.00 vehicle sale to a named buyer on the matching date. The remaining \$5,600.00, they explain, was paid in cash over the following two weeks. There is no document that can source it, so it comes out.

The answer. APPROVE — with \$5,600.00 backed out.** Reserves land at **\$12,200.00 = 4.42 months, the findings re-run clean, and the file closes. The unsourced money did not decline anything; it cost the borrower two months of cushion.

⚠️ Two deposits under \$10,000 on consecutive days is a pattern you must document and must never coach. Your obligation is to record facts without conclusions and to escalate under your firm's Bank Secrecy Act and anti-money-laundering program if the file warrants it (§12.10, §27.2, §27.11). Your obligation is emphatically not to explain to a borrower how a deposit should have been structured, timed, or described. Ask what happened, write down the answer in the borrower's words, attach what they give you, and let the process work. Most of these are exactly what they appear to be, and the innocent explanation is the common one.

What makes this one hard. The pull is toward a binary — either the money counts or the loan dies — and the reality is a dial. Running the three outcomes first changes the conversation from an interrogation into a request: "If we can document all of it, you keep about six and a half months of reserves; if we can only document part, you close with about four and a half. Either way you close. What do you have?" Note also that the \$4,000 earnest money must be sourced too, even though it left the account before the file existed, and that it is a credit at closing rather than part of the \$62,400 — the same trap Appendix F §F.4 flags on the anchor file.

H.14 Scenario H-12 · The debt that appeared three days before closing

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $364,000 · conventional · 5% down · primary residence
  Borrowers         two, W-2, $9,400/month · representative score 704
  Approved          day 27, back-end 43.06%, DTI condition capped at 45.00%
  Debt schedule     auto loan $402 (NINE payments remaining)
                    student loan $268 · revolving minimums $190
                    personal loan $250                        = $1,110
  Day 43 refresh    a new AUTOMOBILE LEASE, opened day 39.
                    $389/month. 36-month term.
  Closing           scheduled day 48

The analysis. The approved file: loan \$345,800 at 6.875%, P&I **\$2,271.66, MI at 0.58% \$167.14**, taxes \$364.00, insurance \$135.00 → **PITI \$2,937.80. Back-end (\$2,937.80 + \$1,110.00) ÷ \$9,400.00 = 43.06%**, inside the 45.00% cap the approval named.

Add the lease and it is 47.20%. The approval's DTI condition is blown and the findings must be re-run (§15.8, §19.10).

The obvious answer is pay it off. Here the obvious answer is wrong twice. First, this is a lease, not an installment loan: terminating one early generally means paying the remaining scheduled payments plus a disposition fee, and it does not necessarily remove the obligation from the file, because a lease that ends leaves the household needing a vehicle. Program treatment of leases differs specifically from installment debt — verify how yours handles it before you promise a payoff will work. Second, and more usefully, the fix was already sitting in the debt schedule:

The ten-month rule — an installment debt with approximately ten or fewer payments remaining may generally be excluded, subject to program specifics (§A.5, Appendix F §F.8) — applies to the existing auto loan and was never claimed, because at 43.06% nobody needed it. Documenting it costs a credit supplement or a creditor statement showing the remaining payment count. It costs the borrower nothing.

The answer. RESTRUCTURE — and the obvious answer was wrong. Document the nine remaining payments, re-run the findings at 42.92%, and close. Verify your program's specifics first: some treatments require the excluded payment not be large enough to materially affect ability to repay, and some require the debt to be current and non-deferred.

What makes this one hard. Under time pressure a loan officer reaches for the lever nearest the problem, and the lever nearest a new debt is paying off the new debt. The discipline is to re-read the whole debt schedule against the whole rulebook whenever the ratio moves, because the cheapest fix is frequently attached to a debt that has nothing to do with what changed. The other lesson is the one this book keeps returning to: the sentence "do not open any new credit, do not finance anything, do not co-sign, and do not change jobs" has to be said at application, said again at approval, and said a third time ten days out — and it still will not always work.

H.15 Scenario H-13 · The ratio that clears only on paper

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $228,000 · FHA 203(b) · 3.5% down · primary residence
  Borrower          single borrower, W-2, hospital scheduler, 5 years
                    score 628 · $5,780/month · debts $612
  Credit event      one 60-day late on an auto loan 14 months ago, following a
                    documented three-week hospitalization. Nothing since.
                    Automated finding DOWNGRADED to manual underwriting.
  Benchmark         31% housing / 43% total debt, manual [illustrative -- verify
                    in Handbook 4000.1]
  Housing history   $1,820/month rent, 34 months, verified, zero lates
  Assets            $14,080 verified
  Seller credit     $6,500, which covers all closing costs and prepaids, so
                    the cash to close is the $7,980 minimum investment

The analysis. Structure first: 3.5% minimum required investment \$7,980.00, base loan \$220,020.00**, LTV on the base loan **96.50%**, UFMIP 1.75% financed **\$3,850.35, total loan \$223,870.35**. P&I at an illustrative 6.375% is **\$1,396.66; annual MIP at 0.55% on the total loan is \$102.61**; taxes \$265.00; insurance \$115.00 → **PITI + MIP \$1,879.27.

Over on both, and over by very little on both — which is the worst place to be, because it is where a loan officer starts arguing about rounding instead of building an argument. A compensating factor is a documented strength that offsets a specific, named weakness (Appendix F §F.8.1). Write it as a pair, never as a list of adjectives:

THE EXCEPTION MEMO -- weakness, then offset      [constructed teaching example]

  WEAKNESS   Housing ratio 32.51% and back-end 43.10%, each above the manual
             benchmark of 31/43.
  OFFSET 1   MINIMAL PAYMENT SHOCK. Verified rent $1,820.00 for 34 months
             against a proposed $1,879.27. Ratio 1.03x -- an increase of
             +3.26%, or $59.27 a month. This household has already
             demonstrated it carries a payment of this size, every month,
             for nearly three years, with zero lates.
  OFFSET 2   RESERVES. $14,080 verified less $7,980 cash to close leaves
             $6,100.00 = 3.25 months of PITI, against a program that
             requires none for this transaction.
  WEAKNESS   One 60-day late, 14 months ago.
  OFFSET 3   THIRD-PARTY DOCUMENTATION of the cause -- hospital discharge
             summary and dated employer leave record -- plus 14 months of
             subsequent perfect payment history on all accounts.

The answer. APPROVE — manually underwritten, with the compensating factors documented in the file. And an honest disclosure to the borrower: this file is approvable, not automatic. A different underwriter at a different shop with a different overlay could reach a different conclusion on the same facts, and if it comes back declined the question to ask is "is that the agency's rule, or ours?" (Appendix F §F.13).

What makes this one hard. Payment shock is the strongest compensating factor in this file and it is the one loan officers forget to compute, because it lives in the rental history rather than in the ratios. \$59.27 is the entire increase in this household's monthly housing cost — a fact that makes the 43.10% look completely different, and a fact that no ratio can express. Contrast it with what is not a compensating factor and gets offered anyway: a clean recent credit record (already priced into the score), an appraisal above contract (the loan is made on the lesser of price or value), and "they're very responsible."

H.16 Scenario H-14 · The co-borrower who makes it worse

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $268,000 · conventional 95% · primary residence
  Borrower          one, W-2, $5,900/month · middle score 748 · debts $465
  Assets            $18,200 · gift of $4,000 from a parent, documented
  The offer         The borrower's parent offers to "go on the loan to help."
                    Parent: $3,100/month (pension + Social Security), middle
                    score 612, monthly debts $2,240 including the mortgage on
                    their own home.
  Illustrative      95% LTV, 748 score: 6.625%, MI factor 0.44%
  pricing           95% LTV, 612 score: 8.000%, MI factor 1.10% [CONSTRUCTED]

The analysis. Run the borrower alone first, because if the file works alone the rest of the conversation is about protecting it. Loan \$254,600 at 6.625%: P&I **\$1,630.23, MI \$93.35, taxes \$268.00, insurance \$120.00 → PITI \$2,111.58. Back-end (\$2,111.58 + \$465.00) ÷ \$5,900.00 = 43.67%**. It works.

Now add the parent, and watch both axes move the wrong way at once:

Adding income helped. Adding \$2,240.00 of debt against \$3,100.00 of income hurt more — the parent arrives with a 72.3% personal debt load — and the representative score for a multi-borrower conventional loan is the lower of the middle scores, which the pricing engine applies to the whole file regardless of who earned it.

The answer. RESTRUCTURE — take the co-borrower off, and the obvious answer was wrong. The file closes on the borrower alone at 43.67% with a 748 score. If the parent wants to help, the gift they have already made is the help, and a larger gift would help more than a signature.

What makes this one hard. Not the arithmetic — the conversation. A parent has just offered to put their name on a thirty-year obligation for their child and been told it would make things worse, which is a genuinely painful thing to hear delivered carelessly. Say what is true: their income is real, their credit is not the problem, and the specific arithmetic of this program penalizes the combination rather than the person. It is also worth saying when the opposite is true, because it often is: a non-occupant co-borrower with a strong score and little debt can rescue a thin-ratio file, which is precisely why the offer is made so often. The structure is sound. This instance of it is not.

H.17 Scenario H-15 · The gift that has to be traced twice

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $355,000 · conventional 95% · primary residence
  Borrowers         two, W-2, $9,600/month · representative score 707
  Debts             $780/month
  Assets            $9,800 of their own funds, seasoned
                    plus a $25,000 GIFT from the borrower's aunt
  How it arrived    (1) the aunt wired $25,000 from the operating account of an
                        LLC she owns
                    (2) it landed in the borrower's BROTHER's checking account,
                        because the brother had the wire instructions
                    (3) the brother transferred it to the borrowers four days
                        before the scheduled closing
  Closing           day 44 · the deposit hits day 40

The analysis. The gift itself is unremarkable — an aunt is an acceptable donor on a conventional one-unit primary residence, and on this transaction the entire down payment may be gifted (Appendix F §F.4.2, and verify the donor list for your program). The chain of custody is the problem, and it has been broken in two places.

THE GIFT CHAIN -- what must be documented          [constructed teaching example]

  WHAT A CLEAN CHAIN LOOKS LIKE
    donor's account  -->  borrower's account  -->  settlement agent
       withdrawal            deposit                 wire
    documented by: gift letter + donor withdrawal + borrower deposit

  WHAT THIS FILE HAS
    LLC account  -->  BROTHER's account  -->  borrowers  -->  settlement
        ?                    ?                    ?

  BREAK 1: the funds came from an ENTITY, not from the donor personally.
           Needed: evidence the aunt owns and controls the LLC, and a
           statement that the entity is not an interested party to this
           transaction. (An interested party's "gift" is a price
           concession, not a gift -- 20.7, 12.4.)
  BREAK 2: the funds passed through a THIRD PERSON. As documented, the
           brother is the transferor of record and is now, on the face of
           the paper, the donor. Needed: either his own gift letter and
           account documentation, or a documented correction of the path.
  ALSO:    "NO REPAYMENT IS EXPECTED" is the operative clause. A gift that
           must be repaid is a LOAN, its payment belongs in the ratio, and
           a gift letter saying otherwise is a false document.

Loan \$337,250 at an illustrative 6.750%: P&I **\$2,187.40, MI at 0.58% \$163.00**, taxes \$370.00, insurance \$125.00 → **PITI \$2,845.40. Cash to close is \$17,750.00 down + \$9,450.00 costs + \$3,350.00 prepaids − \$4,000.00 earnest = \$26,550.00**, against \$34,800.00 of assets, leaving \$8,250.00 = 2.90 months of reserves. The file is comfortable. Only the paper is not.

The answer. APPROVE — after six days and three documents. The aunt signs the gift letter and provides the LLC's organizational document plus her own statement of the withdrawal; the brother signs a short letter documenting that he transmitted the aunt's funds and contributed none of his own, with his statement showing the deposit and the matching outbound transfer; the borrowers' statement shows the receipt. Closing moves from day 44 to day 50.

What makes this one hard. Every person in this chain is telling the truth, and the paper still does not say what happened. That is the general case with gift funds: the failure is almost never fraud, it is helpfulness — somebody had the wire instructions, somebody was traveling, somebody moved money on a Friday to be useful. The prevention is a thirty-second script delivered before the money moves: "When the gift comes, it needs to go directly from your aunt's personal account to yours, or from her straight to the settlement agent. Not through anybody else, not from a business account, and not as cash. Call me before it moves and I'll tell you exactly what to send me."

H.18 Scenario H-16 · The date the borrower remembers is the wrong date

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Application       April, current year · $214,000 purchase · FHA intended
  Borrower          two, W-2, stable four years, $6,750/month, debts $580
                    scores 649 / 641 · $11,200 verified
  What they say     "We lost the house in 2019, but that was a long time ago
                    and we've been perfect since. We're past the three years."
  Credit report     the mortgage tradeline reports last paid MARCH 2019,
                    then 'foreclosure' with no completion date shown
  Public record     the trustee's deed / sheriff's deed of record shows the
                    foreclosure sale COMPLETED AUGUST 2023
  Since the sale    32 months of verified rent, zero lates, no new derogatory

The analysis. There is no payment arithmetic in this file. There is only a date, and the date is the entire loan.

The clock runs from the completion date of the foreclosure — not from the last payment, not from the notice of default, not from the day the household moved out (Appendix F §F.7). Borrowers reliably remember the move-out, because that is the day the event happened to them. Foreclosure timelines vary enormously by state and by circumstance, and four and a half years between the last payment and the completed sale is entirely ordinary.

The borrower's arithmetic — 2019 plus three years — put them eligible in 2022. They are four months away, not four years past.

The answer. DECLINE today — with a date, and a plan attached to it. Take no application you cannot close. Tell them the specific month, tell them where the date came from, and give them the four months as work: keep the rental history perfect (it is the strongest single predictor an underwriter has), add to the \$11,200, do not open anything, and reconvene in June to take a full application for an August closing. Verify the current period in Handbook 4000.1 before you commit to the month, and pull the public record rather than relying on the credit report, which shows no completion date at all.

⚠️ Getting this date wrong is expensive in both directions. Quote it long and you turn away an eligible borrower who buys from someone else. Quote it short and you take an application that cannot close, collect an appraisal fee, and decline the file after they have written an offer. The public record is free and takes minutes, and it is the only source here that carries the operative date.

What makes this one hard. A borrower who has been through a foreclosure has usually been told several different things by several confident people, and is braced to be told no again. The professional register matters: a waiting period is a rule with a date, not a judgment about character, and it should be delivered the way a calendar is delivered. The technical difficulty is that the credit report — the document a loan officer trusts most — is silent on the one fact that decides the file.

H.19 Scenario H-17 · The lock that expires on a Friday

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Loan              $347,000 · conventional · purchase · clear except one item
  Locked            45 days at 6.500% with 0.250 point. Lock expires day 45.
  Day 38            the appraisal returns "subject to" -- a section of roof
                    covering requires repair. Form 1004D re-inspection needed
                    after the work is completed.
  The roofer        first available: day 44. Re-inspection: day 48.
                    Realistic closing: day 52.
  The rate sheet    extensions: 15 days = 0.250 point · 30 days = 0.500 point
  today             expired locks reprice at WORST CASE -- the worse of the
                    original price or current market
  The market        6.500% now costs 1.000 point. It cost 0.250 a month ago.

The analysis. Three options, and the third is the one nobody prices.

Option 2 is \$2,602.50 worse than option 1 for an identical outcome, and worst-case repricing is why: a borrower who could re-lock at the better of the old rate and the new one would hold a valuable option the lender never sold (§29, §30).

Option 3 is the interesting one, because it is the only option with no invoice. \$867.50 ÷ \$57.36 = 15.1 months. If this household keeps the loan longer than about fifteen months — and the median borrower keeps a mortgage for years — the extension is the cheaper choice. Over five years the higher rate costs \$3,441.60**; over the full term, **\$20,649.60.

The answer. RESTRUCTURE — take the 15-day extension. Then answer the question the borrower has not asked yet.

⚠️ Who pays for the extension is a separate question from what it costs, and it is the honest one. This delay was caused by a property condition discovered on day 38 — not by the loan officer. Document the changed circumstance contemporaneously (§22.5, §22.9), because a fee that appears on a revised Loan Estimate without a documented changed circumstance is a tolerance problem, not a pricing decision. And when the delay is yours — a condition that sat in your inbox, a document you did not order, a lock you sized against your optimism rather than against the contract's closing date — you or your branch pay for it, and you say so first. Chapter 30 owns that argument and makes it sharply.

What makes this one hard. Under pressure the choice collapses into "pay the fee or don't," and the third option — accepting a higher rate at zero cost today — is invisible precisely because it has no invoice. It is also the most expensive of the three for anyone who keeps the loan. The second difficulty is calendar arithmetic that has nothing to do with money: the roofer's first availability is day 44, and no amount of lock strategy moves it. Measure the lock against the contract's closing date and the longest pole in the file, not against your optimism.

H.20 Scenario H-18 · The property that covers its payment and loses money

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $425,000 · two-unit · INVESTMENT · DSCR program
                    borrower will hold title in an LLC and guarantee personally
  Structure         25% down ($106,250) · loan $318,750 · 30-year fixed
                    rate 7.875% [CONSTRUCTED -- not a market quote]
  Carrying costs    taxes $5,940/year · landlord policy $2,280/year · no HOA
  The rents         Unit A: executed lease $1,650 · appraiser's market rent
                    opinion for Unit A (Form 1007/1025) $1,625
                    Unit B: VACANT · appraiser's market rent opinion $1,600
  Program rule      per unit, the LESSER of executed lease and market rent;
                    market rent alone where the unit is vacant
  The note          carries a 3-2-1 prepayment penalty
  The borrower      "It cash flows. The rents are more than the payment."

The analysis. DSCR = gross rental income ÷ PITIA, and both halves are program parameters, not facts (§34.6). Read the matrix before you compute anything.

DSCR WORKSHEET -- two-unit investment              [constructed teaching example]

  THE PAYMENT (PITIA)
    Principal and interest, $318,750 at 7.875%              $2,311.16
    Taxes ($5,940 / 12)                                       $495.00
    Insurance, landlord policy ($2,280 / 12)                  $190.00
    HOA / association                                           $0.00
    ----------------------------------------------------------------
    PITIA                                                   $2,996.16

  THE RENT
    Unit A: lesser of lease $1,650 and market $1,625     ->  $1,625.00
    Unit B: vacant, market rent opinion                  ->  $1,600.00
    ----------------------------------------------------------------
    Qualifying gross rent                                   $3,225.00

  THE RATIO      $3,225.00 / $2,996.16      =                    1.08

  Rent required for a 1.00 DSCR                              $2,996.16
  Rent required for a 1.25 DSCR                              $3,745.20
  Shortfall against a 1.25 program                             $520.20

Then run it the way an owner runs it, with constructed operating assumptions — real costs vary enormously by market, age of property, and management arrangement:

The answer. APPROVE — on a program with a 1.00 minimum, and the second table goes in the file. At a 1.25 minimum this declines by \$520.20 of rent. Get the threshold, the rent rule, and the payment rule in writing before you quote, because each one moves the ratio by more than a tenth of a point.

⚠️ This is where a prepayment penalty legitimately lives, and the distinction is precise. Regulation Z permits a prepayment penalty on a covered transaction only where the loan is a fixed-rate qualified mortgage that is not higher-priced. A consumer non-QM loan therefore generally cannot carry one. Penalties belong to business-purpose investor lending — like this file — which sits outside Regulation Z entirely (§34.8). Price it: on a 3-2-1 structure, a refinance in month 14 against a balance of \$315,544.42 triggers the year-two step of 2% = \$6,310.89. Say that number out loud at application, not at payoff. And note what the penalty depends on: occupancy is the bright line. If the facts do not support business purpose — if this becomes a home the borrower or a family member lives in — the loan is consumer credit, every consumer protection attaches, and a file underwritten with no income verification is badly non-compliant.

What makes this one hard. The ratio passed and the property loses \$685.77 a month, and both are true. The numerator is gross rent, which is documentable; nothing in the ratio has ever accounted for vacancy, management, turnover, or the roof. You are not the borrower's investment adviser and must not pretend to be. You can hand them the second table and say: "The program qualifies this at 1.08. Here's what it looks like with vacancy and management in it. If your numbers are better than mine, use yours." An experienced investor will thank you and buy anyway. A first-time investor may never have seen it.

H.21 Scenario H-19 · Deposits that are not income

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Purchase          $490,000 · primary residence · 20% down ($98,000)
  Loan              $392,000 · debts $1,240/month · score 763
  Borrower          100% owner of an S-corporation, 7 years, licensed trade
  What they were    "You're self-employed with write-offs, so you need a bank
  told              statement loan." No agency analysis had been run.
  Bank statements   12 months, business operating account
                    total deposits $268,400
                    excluded: $12,000 owner transfer from personal
                              $6,400 equipment loan draw (a liability)
                              $3,500 insurance reimbursement
                                                    = $21,900
  Tax returns       YEAR 1  W-2 to self $74,000 + K-1 $41,300
                            + depreciation $14,200 - meals $2,200 = $127,300
                    YEAR 2  W-2 to self $79,000 + K-1 $42,600
                            + depreciation $15,600 - meals $2,400 = $134,800
  Illustrative      agency 6.875% · non-QM bank statement 8.375% [CONSTRUCTED]
  pricing

The analysis. Run both. That is the whole scenario, and it is the agency-first rule stated as a procedure: attempt the agency file, document why it failed, and only then price non-QM (§34.10).

The agency file produces more income, not less, and the reason is sitting in the middle of the worksheet: \$29,800 of depreciation across two years is a deduction that reduced taxable income and never took a dollar out of the business, so it comes straight back. The folk wisdom that write-offs destroy self-employed borrowers is true of some write-offs and false of the largest one.

Then the price of not checking: \$404.32 a month**, **\$24,259.20 over five years. That is what the borrower would have paid for a product they did not need.

The answer. RESTRUCTURE — place it agency, and the obvious answer was wrong. The agency file clears at 41.73% with \$357.54 a month of room against a 45% cap. Put the bank statement worksheet in the file anyway, next to the cash-flow analysis, as evidence the comparison was run.

What makes this one hard. Somebody confident told this borrower they needed a bank statement loan, and confidence is persuasive. The professional habit is to treat "I need a non-QM loan" as a hypothesis rather than an instruction — and the corollary habit, on any file that genuinely does go non-QM, is to build the exclusion column yourself before the underwriter does. Four deposits totaling \$21,900 came out of the calculation above, not because anyone doubted them but because none of them was revenue: an owner transfer is the borrower's own money making a round trip, a loan draw is a liability, and an insurance reimbursement replaced a loss. Find those first, send one consolidated list, and stop the file from generating four separate conditions over three weeks.

H.22 Scenario H-20 · The refinance that pays off nothing

THE FILE AS IT ARRIVES                              [constructed teaching example]
  Property          owner-occupied, appraises at $392,000
  Current first     original $246,000 at 3.125%, 30-year fixed, 96 payments
                    made · P&I $1,053.80 · balance $201,003.00
                    264 payments remaining · LTV 51.28%
  The request       "$60,000 cash out. $41,000 to wipe out the credit cards --
                    the minimums are $1,190 a month -- and $19,000 for the
                    kitchen."
  Credit cards      $41,000 across four accounts, blended rate 21.9%
                    [constructed] · all current, never late
  Income/credit     $8,400/month · score 771 · no other debt
  Illustrative      cash-out first mortgage 7.125% · closed-end second
  pricing           lien 8.750%, 20-year term [BOTH CONSTRUCTED]

The analysis. The cash-out refinance is easy to approve and easy to sell, which is exactly why it needs the second table. New loan: \$201,003.00 + \$60,000.00 + \$8,900.00 of costs = **\$269,903.00, which is 68.85% of value and well inside an 80% cash-out cap (\$313,600.00). P&I at 7.125% is \$1,818.39**.

The headline is genuinely good. Today the household pays \$1,053.80 of P&I plus \$1,190.00 of card minimums = \$2,243.80**. Afterward it pays **\$1,818.39 — a saving of \$425.41 a month. Most borrowers stop reading here, and so do most loan officers.

Now the alternative nobody quoted: keep the 3.125% first mortgage and add a closed-end second lien for the same \$60,000.

Plan B is \$234.36 a month cheaper** and costs **\$240,262.00 less in interest (a smaller closing cost, too, which would widen the gap). Both plans deliver the same \$60,000 and eliminate the same \$1,190.00 of card minimums. The difference is entirely that Plan A refinances **\$201,003.00 of 3.125% money at 7.125%** in order to reach \$60,000 — the most expensive way to borrow sixty thousand dollars available in this file.

Two more facts belong in the conversation, and neither is a number the borrower asked for. First, \$41,000 of unsecured debt becomes secured by the home: the failure mode changes from a collection account to a foreclosure, and that is a real transfer of risk to the household even though it makes the payment smaller. Second, paying cards to zero without changing what produced them is a documented industry pattern, and the household that re-accumulates ends up carrying both. Say it plainly, once, without a statistic you cannot source.

The answer. DECLINE — as requested. The cash-out refinance does not get originated on this file. Present both structures side by side with the arithmetic attached, recommend the second lien, and let the borrower decide with the numbers in front of them.

⚠️ The structure you should recommend here is the one that pays you less, and that is the entire argument of this book's first theme. A \$269,903 first mortgage generates more compensation than a \$60,000 second, and the loan originator compensation rule exists because that incentive is real and structural (§26.5). The rule's requirement is not that you avoid the conversation; it is that you never steer a consumer to a transaction because it pays you more. Presenting options honestly is the job. Documenting that you presented them is how you prove you did the job. And a borrower who is handed \$240,262.00 in writing tends to remember who handed it to them — which is theme four, and it is how this business actually compounds.

What makes this one hard. Every number in the borrower's version of the story is true. The payment really does fall by \$425.41. The cards really are gone. Nothing has been misstated, and the file would close, fund, and perform. The difficulty is that the honest analysis requires a comparison the borrower did not ask for, against a product they have never heard of, delivered by the person who gets paid less if they take it.

H.23 What the twenty files have in common

Eight approvals, seven restructures, five declines. If that distribution surprises you, it is the most useful thing in this appendix — because the ratio of a real pipeline looks much more like this than like a book of solved problems, and the loan officers who last are the ones whose "no" arrives on day two rather than on day forty-four.

Five patterns account for nearly every wrong answer above.

One. The number was computed before it was established. H-3's bonus, H-4's rent, H-1's overtime, H-2's income — in each case the arithmetic was flawless and the input had not survived step 1 of §H.2. There is no way to recover a ratio built on an income an underwriter will not count.

Two. The threshold was a cliff and got treated as a slope. H-7's \$390 and H-8's \$1,450 are the same failure at opposite ends of the price scale. Near a published limit, the move is almost never to price the difference; it is to find out what it costs to get back on the right side of the line.

Three. The rule was applied without asking what it presumes. A 24-month average presumes 24 months (H-3). The declining-income rule presumes an income you can still establish (H-2). The ten-month rule was sitting unused in a debt schedule nobody re-read (H-12). Rules have preconditions, and the preconditions are where the judgment lives.

Four. The fix was looked for next to the problem. The lever nearest a new debt is paying off the new debt (H-12). The lever nearest a low appraisal is fighting the appraisal (H-10). Both times the give was somewhere else in the file.

Five. The obvious answer optimized the wrong quantity. FHA "costs more" — over thirty years, not over the six to eight this borrower named (H-5). More income "helps" — not when it arrives attached to more debt and a lower score (H-14). A cash-out refinance "lowers the payment" — while refinancing \$201,003.00 of 3.125% money at 7.125% (H-20).

⚠️ One last convention, because it will bite anyone who checks these files against a calculator. Every payment above is the annuity formula of §A.1 rounded to the cent, and every total of payments is that rounded payment multiplied by the term. A servicer-method schedule (§A.2) — interest rounded to the cent monthly — lands a few dollars away from the totals printed here, exactly as Appendix A describes on the anchor file, and a servicer adjusts the final payment to clear it. Quote the printed figure and state the convention. Do not re-derive it from full precision, and do not "correct" a total that was built the way the book builds totals.

And the habit under all of it, which is the same one Appendix F closes on: when a rule surprises you, do not repeat it — find it. When a borrower asks for a threshold, do not recall it — look it up in front of them and say so out loud. Every guideline figure in these twenty files is illustrative and some of them will be wrong by the time you read this. The reasoning is the durable part.