Answers to Selected Exercises

Worked solutions to the daggered (†) and odd-numbered exercises from each chapter. Try every problem before reading its solution.

Chapter 1

Exercise 1.2 †

The security instrument is recorded; the note is not.

The note is a private contract between borrower and lender evidencing a debt. Nothing about a private promise to repay money requires public notice, and recording it would serve no function.

The security instrument creates a claim against a specific parcel of land. The purpose of a county land record is to allow anyone — a future buyer, a future lender, a title examiner — to determine what is attached to that parcel. Recording puts the world on constructive notice of the lien and establishes its priority relative to other liens, generally by the date and time of recording.

The practical consequence: the debt can be sold repeatedly without touching the county record, while the lien stays exactly where it was first recorded. Case Study 2 shows what happened when that separation was industrialized.

Exercise 1.5 †

Three sentences at a first-time-buyer level:

"They do the same job — they're both the document that pledges your house as security for the loan. Which one you sign depends on your state, not on us or on your lender. The practical difference is what happens in the rare case of a foreclosure: a deed of trust brings in a neutral third party called a trustee, which usually means the process happens outside of court, while a mortgage state generally requires the lender to file a lawsuit."

Grading note: an answer that says one is "better" for the borrower is wrong. Non-judicial foreclosure is faster, which cuts both ways, and both routes carry statutory notice requirements.

Exercise 1.8 †

The chapter's point: decision authority and borrower contact are almost disjoint. The underwriter decides and does not talk; the loan officer talks and cannot decide.

(Note: the exercise asks for five roles; the closing agent is a sixth party listed in §1.5 and is correctly included. Accept either five or six.)

Exercise 1.11 †

Model reply, 96 words:

"Nothing has gone wrong. Mortgage loans are almost always sold after closing — the company that made your loan sells the debt, and a servicer collects your payment. Your rate, payment, and terms do not change. What you should do before sending money anywhere: confirm the transfer against the two notices you should have received, one from your current servicer and one from the new one, and call the number on your existing statement rather than a number in the letter. If both notices line up, pay the new servicer on the date they specify. Forward me the letter and I'll confirm."

Grading note: the answer must include verify independently — servicing-transfer letters are a known fraud vector, and Chapter 27 covers it.

Exercise 1.14 †

• Whose money crosses the closing table: the wholesale lender's.
• Whose name is on the note: the wholesale lender's.
• Who the borrower calls about escrow in six months: the servicer — which may be the wholesale lender, or an entity that bought the servicing, and is not the broker.

The teaching point: the broker's name appears nowhere on the borrower's loan documents, and the broker has no ongoing relationship with the loan. The borrower's relationship is with the loan officer; the borrower's loan is with someone else entirely.

Exercise 1.17 †

Restated: the lender's revenue on a loan comes mostly from selling it, not from holding it.

The single best supporting fact: the lender funds at closing with borrowed money — a warehouse line — which it must repay in days or weeks. It cannot hold the loan even if it wanted to, so its economics are the sale price plus fees, not thirty years of interest.

Illustrative arithmetic from §1.3: a \$365,750 loan sold at a price of 101.500 produces \$371,236.25, a gain of **\$5,486.25** in weeks, versus a first-month interest accrual of \$2,019.24 against the cost of borrowed funds.

Exercise 1.18 †

First month's interest:

$$\$365{,}750 \times \frac{0.06625}{12} = \$365{,}750 \times 0.005520833\overline{3} = \$2{,}019.24$$

First month's principal:

$$\$2{,}341.94 - \$2{,}019.24 = \$322.70$$

So 13.8% of the first payment reduces the balance, and 86.2% is interest. Chapter 4 explains why, and what that curve looks like over 360 payments.

(A one-cent rounding difference from computing the unrounded payment of \$2,341.9373 is expected and acceptable.)

Exercise 1.20 †

$$\text{back-end DTI} = \frac{\$3{,}033.72 + \$1{,}446.00}{\$10{,}500.00} = \frac{\$4{,}479.72}{\$10{,}500.00} = 42.66\%$$

Which to lead with on a call: the housing ratio (\$3,033.72 ÷ \$10,500.00 = 28.89%), because it answers the question the borrower is actually asking — "what will this house cost me every month, and is that reasonable against my paycheck?" The back-end ratio answers the underwriter's question. Lead with the borrower's question; bring the second one in when you explain what could still change the answer.

Accept any answer that distinguishes the borrower's question from the underwriter's and chooses accordingly.

Exercise 1.22 †

Documented out-of-pocket, from those two items:

Two costs that cannot be quantified: the house itself (they do not get another chance at this one); and the market they re-enter — rates, prices, and inventory in the week they restart are whatever they are. Credit for other defensible unquantifiables: moving arrangements already made, a lease already terminated, time off work, and the effect on the borrowers' willingness to trust the next loan officer.

Exercise 1.24 †

Model sentence:

"I can give you a real number, but I need two more things first — your middle credit score and how much you're putting down — because those two move the price more than anything else. Without them the honest answer is a range: for a file like yours today, somewhere between about six and a half and seven and a quarter, and I'd rather narrow that in ten minutes than guess and be wrong."

Grading note: the answer must (a) not refuse, (b) not invent a single number, (c) name what is missing and why it matters, and (d) commit to a time. Refusing to quote at all is scored as a wrong answer in this book; it loses the borrower to someone less careful.

Exercise 1.26 †

"The note only covers principal and interest — the part that pays back the loan itself. Your actual monthly payment also includes property taxes, homeowners insurance, and mortgage insurance, which the servicer collects along with it and pays on your behalf."

The document that shows the full figure is the Closing Disclosure, which sets out the estimated total monthly payment including escrow. (Chapter 22.) A Loan Estimate shows it earlier.

\$2,341.94 P&I + \$385.00 taxes + \$130.00 insurance + \$176.78 mortgage insurance = \$3,033.72.

Exercise 1.29 †

Model email, 94 words:

Subject: Linden St — pre-approval by 2:00

"Yes. I need twenty minutes with them on the phone and their authorization to pull credit. If we can talk by 10:00, you'll have a letter before 2:00 that the listing agent will take seriously — credit pulled, income and assets discussed, and an amount I can actually support.

If we can't reach them in time I'll tell you that rather than send something soft. A letter that falls apart in underwriting costs your clients more than a delayed offer does.

Send me their numbers and I'll call."

Grading note: it commits to a time, names exactly what is required, promises only what a credit-pulled conversation can support, and pre-frames the failure case. Chapter 8 develops the twenty minutes.

Exercise 1.32 †

There is no single correct answer; the exercise is scored on the supported/assumed split.

A well-done version marks essentially everything assumed on day 0 — the income, the assets, the credit, the value of the property, the program, and the payment — and marks as supported only: the property address, the contract price the buyers intend to offer, the referral source, and the deadline. Students who mark more than four items "supported" have not understood the exercise.

The paragraph should conclude that the debt will most likely be owned by an investor through an agency security, and the payment will go to a servicer, neither of which can be named on day 0. Saying "I don't know yet, and here is what would tell me" is the target answer.

Chapter 2

Exercise 2.2 †

A balloon mortgage requires the entire remaining principal at the end of the term.

It was not considered a defect because nobody expected the borrower to produce it. The operating assumption was renewal: at maturity the loan would be refinanced into another short-term loan, indefinitely, so long as the borrower kept paying and the lender kept wanting the paper. The borrower's equity came from the large down payment (typically around 50%) and from appreciation, not from amortization.

The defect was invisible because it was an assumption, not a term: continuous availability of refinancing. It was correct every time it was tested until 1930, when it failed for everyone simultaneously.

Exercise 2.4 †

The FHA insures. It does not lend, and it does not buy loans.

It offers lenders protection against loss on loans meeting its standards, in exchange for a premium paid by the borrower. That structure is what let it dictate the product: a lender who would not voluntarily make a long-term, high-loan-to-value loan would make one if the government absorbed the loss, and to get that protection the loan had to be long-term, fully amortizing, fixed-rate, within stated LTV limits, on a property appraised to FHA standards, to a borrower meeting FHA underwriting criteria.

What it does not do: originate, purchase, securitize, or service.

Exercise 2.7 †

The pattern, and the answer to most exam questions in this family: none of them lends money to a homebuyer.

Exercise 2.11 †

The assumption: refinancing will be available at maturity on terms similar to today's.

The mechanism: a borrower reaches the end of a five-year interest-only balloon having made every payment. They owe the full principal. They ask for the customary renewal. The lender declines — because the lender is under its own funding pressure, or because the property no longer supports the loan at current values. The borrower cannot produce the principal from savings; that was never the plan and the loan was not designed for it. Foreclosure follows, on a borrower who is employed and current.

The teaching point: the failure did not require the borrower to do anything wrong. It required only that a third party's willingness to renew disappear at the wrong moment.

Exercise 2.13 †

For an agent, phrase it as a resale problem:

"A loan can only be sold to somebody who wasn't in the room when it was made. If every lender uses its own application, its own appraisal form, its own rules about income, and its own note, then anyone buying that loan has to re-underwrite it from scratch — which costs so much that nobody does it, so there is no buyer, so the lender has to hold the loan with its own money, so it runs out of money and stops lending. Standardizing the forms and the rules is what makes a stranger willing to buy sight unseen. It isn't paperwork; it's what creates the buyer."

Accept any answer that identifies buyer confidence without individual inspection as the function.

Exercise 2.15 †

Covenants becoming unenforceable in 1948 removed one private mechanism. The FHA's underwriting posture — an institutional mechanism embedded in the standards that determined which loans were insurable and therefore marketable — persisted afterward.

This matters for disparate impact doctrine because it establishes the pattern the doctrine addresses: the exclusion outlived the explicitly discriminatory instrument by operating through facially neutral criteria (neighborhood stability, valuation trend, marketability). A legal theory requiring proof of discriminatory intent would have been unenforceable against a standards manual that spoke in the language of property values.

Chapter 25 develops the modern doctrine; the historical point is that the doctrine's shape is a response to this specific pattern.

Exercise 2.17 †

"A bank can fail without a single borrower missing a payment, because a bank has two sides. On one side it holds loans at a fixed rate. On the other it owes depositors, who can leave tomorrow. In the 1970s the thrifts held mortgages made years earlier at 6 or 7 percent, and then rates rose so far that they had to pay depositors 12 percent to keep the money from walking out the door. Every loan on the books was performing. The institution was losing five cents on every dollar, every year, and it could not fix it — the mortgages were fixed for thirty years and could not be repriced. The borrowers were fine. The bank was insolvent."

Exercise 2.20 †

Assumptions embedded in current practice — any defensible answer with a test. Strong responses include:

• That agency guidelines will remain available on current terms. Test: what would this file look like if the conforming limit fell, or if the enterprises exited conservatorship on different terms?
• That appraisal waivers correctly identify low-risk valuations. Test: what is the loss experience on waived files in a falling market? Nobody has a large sample from one.
• That a borrower's employment continues. Test: this is why two-year history and continuity documentation exist — the assumption is at least named.
• That digital verification data is accurate. Test: what happens when the payroll aggregator is wrong, and who bears it?

Grading note: an answer that names an assumption without proposing a test has done half the exercise.

Exercise 2.21 †

\$10,000 house, 50% down, \$5,000 loan, five-year interest-only balloon at 6%:

Compare the Linden Street loan: after 60 payments the balance has fallen from \$365,750.00 to \$342,870.17**, building **\$22,879.83 of equity from amortization alone.

Exercise 2.23 †

Breakeven deposit rate: 7.00% — the yield on the asset book. Anything above it loses money.

Is that achievable? No, and that is the point. In an environment where money market funds pay 12%, a thrift offering 7% loses its deposits, which is worse than losing money on them — an institution that cannot fund its assets must sell them, and a book of 7% thirty-year mortgages in a 12% market sells at a large discount, crystallizing the loss immediately.

There is no operating decision available that fixes this. That is why the crisis required a legislative resolution.

Exercise 2.25 †

Model reply, 108 words:

"Fair question, and the honest answer is that the rules changed because of something specific. In 2006 you could get a mortgage without showing anyone a paystub. A lot of people ended up in loans they could never have afforded, and a lot of them lost the house. Every document I'm asking you for traces back to that. I'm not going to pretend it isn't a nuisance — it is — but it's a nuisance with a reason.

Worth knowing about your parents' loan, too: it probably came due in five years and had to be renewed. If the bank had said no, they'd have owed the whole balance. Ours can't do that."

Grading note: it explains rather than apologizes, and it reframes the nostalgic comparison accurately rather than conceding it.

Exercise 2.29 †

The line doing the most work: the down payment. Term and amortization change the borrower's experience of the loan enormously, but the 50%-versus-5% line is the one that decides whether the transaction happens at all. These borrowers have \$38,000; a 1928 structure requires \$192,500.

Accept "loan-to-value" or "mortgage insurance" as the answer with a good argument — MI is what makes the 5% possible — but the down payment is the binding constraint.

Chapter 3

Exercise 3.2 †

1. Pre-licensing education — 20 hours, NMLS-approved
2. Pass the SAFE MLO test — 75% or better
3. Background check — fingerprints and FBI criminal history
4. Credit report review — for financial responsibility
5. Surety bond or recovery fund coverage, in an amount set by the state

Plus two conditions that govern whether the license functions: character and general fitness (a standard, not a test) and active sponsorship (a license authorizes nothing until an employer sponsors it).

Exercise 3.4 †

The category that is the same in both PE and CE: federal law, at 3 hours. Every other category shrinks — ethics 3→2, non-traditional 2→2 (unchanged in hours but not in share), electives 12→1.

(Grading note: non-traditional lending is also 2 hours in both. Accept either answer; the stronger answer names both and observes that electives collapse from twelve to one, which is where the compression actually happens.)

Exercise 3.6 †

The seven-year bar: a felony conviction during the seven years preceding the application disqualifies. This bar expires.

The permanent bar: a felony conviction at any time, with no lookback limit, involving fraud, dishonesty, breach of trust, or money laundering disqualifies permanently.

The distinction the exam tests: the first has a clock; the second does not. A 1996 embezzlement conviction bars licensure in 2026.

Exercise 3.9 †

Sponsorship is an employer's attestation in NMLS that an originator works for and is supervised by them.

An unsponsored license authorizes no origination activity. It is a valid credential — the education, the test, the background check, and the bond are all satisfied — and it permits nothing until an employer activates the sponsorship record. Sponsorship ends the day employment ends.

Exercise 3.11 †

The two shared rows are the point: both are in the registry and both are background-checked. Only one holds a credential.

Exercise 3.13 †

Model reply, 92 words:

"You don't need it now — that's true. The problem is the day you want to leave. Every non-depository lender in town requires a state license, and you can't get one in a week: twenty hours of education and a national exam you've never sat. That's four to ten weeks of not originating, right when you're trying to start somewhere new.

Nothing stops you from getting licensed while you're registered. A few weekends and a few hundred dollars, and you'd never have to think about it again. Ask if they'll pay for it."

Grading note: the persuasive move is naming the specific future moment rather than arguing about professional standards. Answers that lecture about competence score lower.

Exercise 3.16 †

The definition: a mortgage loan originator is a person who, for compensation or gain (or in the expectation of either), takes a residential mortgage loan application or offers or negotiates terms of a residential mortgage loan. Either activity suffices. Job title is irrelevant.

Exercise 3.18 †

What must appear: your NMLS unique identifier and your employer's, per the advertising requirements. A post that solicits mortgage business is advertising, regardless of whether it is on a "personal" profile.

Who is responsible: you are, and so is your employer. "Marketing posted it" is not a defense — the requirement attaches to the advertisement, and enforcement findings in this area routinely involve originators whose social content was produced by someone else.

The practical rule: if it solicits mortgage business, it carries your identifier. Verify your employer's specific policy with compliance; requirements also vary by state.

Exercise 3.19 †

Using the midpoints of §3.5's approximate ranges — federal law 23.5%, uniform state content 11%, general mortgage knowledge 21.5%, origination activities 26%, ethics 17% (sums to 99%; normalize or round):

Study first: federal law and ethics. Together they are roughly 40% of the score and they are the most learnable content on the exam — the answers are facts in statutes rather than judgments. Front-loading them banks points early and leaves the more interpretive origination-activities material for when your baseline is already passing.

Note: verify the current content outline at NMLS. An allocation built on stale weights misdirects your time, which is the exact error this exercise is designed to prevent.

Exercise 3.21 †

A model first ten rows. Yours will differ; the requirement is that at least four come from Chapters 1–2.

  Ch.3   20 hrs PE = 3 fed / 3 ethics / 2 non-trad / 12 elective
  Ch.3   8 hrs CE  = 3 fed / 2 ethics / 2 non-trad / 1 elective
  Ch.3   75% to pass · 120 questions (115 scored) · 190 minutes
  Ch.3   30-day retake wait; 180 days after three consecutive failures
  Ch.3   7-year felony lookback; PERMANENT for fraud/dishonesty/breach/laundering
  Ch.3   Renewal window: November 1 – December 31
  Ch.1   Note = evidence of debt, NOT recorded
  Ch.1   Security instrument = creates the lien, IS recorded
  Ch.1   Deed of trust = 3 parties (trustor, trustee, beneficiary)
  Ch.2   S.A.F.E. Act = HERA 2008.  CFPB = Dodd-Frank 2010.
  Ch.2   FHA 1934 · Fannie 1938 · VA 1944 · Ginnie 1968 · Freddie 1970

Exercise 3.23 †

After three consecutive failures the waiting period is 180 days.

Cumulative timeline, assuming the minimum waits and immediate retesting:

Against a 90-day conditional hire, the candidate misses the deadline after the third failure — by day 60 it is already arithmetically impossible to test again before day 240. The employment consequence is decided two attempts before the candidate feels it.

The practical lesson: do not schedule the first attempt until practice scores are comfortably above 75%. The retake structure punishes early attempts far more than it punishes waiting.

Exercise 3.25 †

Advise them to do first: contact the taxing authority and enter a documented payment arrangement, or pay it. A lien under an active, current installment agreement reads completely differently from one being ignored, and the difference is a phone call.

Then: disclose all three items fully. The bankruptcy and the collections are survivable. A concealed anything is not.

Exercise 3.28 †

"It's a fair objection, and the answer isn't that people with credit problems are dishonest. Regulators aren't scoring you — there's no minimum FICO for this license. They're looking for a pattern, and specifically for whether you're addressing obligations or ignoring them. A discharged bankruptcy from years ago with clean history since tells them something good. An unpaid tax lien you haven't called anyone about tells them something else.

The reason it's relevant to the job: you'll spend your career handling other people's financial documents, sometimes their funds, and constantly their trust. The regulator has to decide whether you'll operate honestly, fairly, and efficiently, and they only have evidence — so they use the financial record you've actually produced."

Grading note: the answer must distinguish pattern from score and must engage with the objection rather than dismissing it.

Exercise 3.30 †

Today, in order:

1. Register for CE immediately — before anything else on the list. Choose a provider with fast NMLS reporting and confirm the reporting turnaround in writing.
2. Complete the eight hours today or tomorrow. They are eight hours. Clear the calendar.
3. Verify the completion posted to your NMLS record. Not the provider's confirmation — the NMLS record. Provider reporting is not instantaneous and the gap is what actually catches people.
4. Submit the renewal request and pay the fee. Confirm the state has received it.
5. Tell your manager and your processor today what your status is, so that contingency coverage for the January closings can be arranged now rather than on January 2.
6. Only then return to the pipeline.

Next year's policy, two sentences:

"CE is completed by July 31 every year, calendared in January as a recurring commitment. The renewal request goes in during the first week of November."

Grading note: full credit requires step 3. Candidates who stop at "complete the course" have missed the actual failure mode.

Exercise 3.33 †

What a borrower can see on Consumer Access: the NMLS unique identifier; the individual's name; current employer and employment history; every state license held, with status and issue date; whether the person is state-licensed or federally registered; and any regulatory actions in any jurisdiction.

Model two sentences:

"My NMLS number is on every document I send you — it's on the Loan Estimate you'll get this week. Type it into NMLS Consumer Access; it's a free public site, no account, and it'll show you every license I hold, where I've worked, and anything any regulator has ever filed against me."

Grading note: the answer must hand them the method. "I'm licensed and regulated" is an assertion and scores zero on this exercise — that is the entire point of the prohibition in the prompt.

Chapter 4

Exercise 4.2 †

\$412,000 at 7.375%, 360 months.

Exercise 4.4 †

40-year payment on \$365,750 at 6.625% (480 months): **\$2,173.96**.

The sentence for a borrower:

"It saves you about a hundred sixty-eight dollars a month, and it costs you about two hundred thousand dollars in extra interest for ten more years of payments. If a hundred sixty-eight dollars is the difference between yes and no on this house, that's a real conversation — but I'd rather look at the down payment or the rate first, because both do more for less."

(Note for instructors: 40-year terms are not Qualified Mortgages. The arithmetic is the point.)

Exercise 4.6 †

\$247,000 at 7.000%, 30 years. Factor from §4.10: **\$6.6530** per \$1,000.

$$247 \times \$6.6530 = \$1{,}643.29$$

Exact: \$1,643.30. Error: one cent.

The factor method is accurate to the penny on this loan and within a dollar or two on almost any loan, which is why it is worth memorizing five rows.

Exercise 4.7 †

\$200,000 at 6.000%, 360 months. Monthly rate 0.005. Payment **\$1,199.10**.

Principal share of the first payment: \$199.10 ÷ \$1,199.10 = 16.6%.

(Interest rounded to the cent each month — the servicer method used throughout this book.)

Exercise 4.9 †

Interest was 83.7% of five years of payments.

Exercise 4.11 †

• (a) The interest portion of payment 2 is lower than it would have been, because payment 1 reduced the balance by \$322.70 **plus** the extra \$200 — a balance \$200 smaller earns \$1.10 less interest at this rate, and every subsequent month compounds that.
• (b) The payoff date moves earlier. Each extra dollar of principal removes not only itself but all the future interest it would have carried.
• (c) Total interest falls substantially — far more than \$200 × 360, because of (a).

The information needed to compute the payoff date exactly: whether the extra \$200 is applied to principal and whether the servicer re-amortizes or keeps the scheduled payment. On a standard fixed-rate loan the payment stays the same and the term shortens; the payoff month is then found by running the schedule with the higher payment until the balance clears. (You also need to confirm the note has no prepayment penalty — this one does not.)

Exercise 4.13 †

The two failure modes:

1. The borrower budgets on \$2,341.94 and discovers at closing that the real figure is \$3,033.72 — 29.5% higher. That is a household planning error you caused.
2. The borrower hears it, mistrusts it, and assumes you are shading numbers to win the business. Some borrowers already know PITI exists.

What they should have said:

"Your payment all in — principal, interest, taxes, insurance, and mortgage insurance — is about three thousand thirty-five a month. The loan part of that is twenty-three forty-two; the rest is taxes, insurance, and the mortgage insurance that comes with five percent down."

Exercise 4.15 †

Qualify on **\$385.00**. A borrower qualified on the seller's \$175.00 has a payment that is \$210 per month wrong from the first payment, and a housing ratio understated by two full points.

How to find the right number: ask the county assessor's office, or the title company, what the property will be assessed at after transfer and at what rate. Where reassessment practice is unclear, use the price × the local rate and say you are estimating conservatively.

Exercise 4.17 †

Loan \$332,500, price \$350,000, appraisal \$341,000.

Exercise 4.19 †

First \$240,000; HELOC limit \$75,000, balance \$12,000; value \$400,000.

The HELOC's undrawn \$63,000 moves the ratio by 15.75 percentage points. That is the whole reason HCLTV exists.

Exercise 4.21 †

Statute: the Homeowners Protection Act. Applies to borrower-paid private mortgage insurance on conventional loans. FHA's MIP is governed separately by HUD.

Exercise 4.23 †

The underwriter will use 34.69%. With eight payments remaining the auto loan falls inside the ten-month convention and may be excluded, subject to program specifics and to the payment not being large enough to affect near-term ability to pay.

5.44 percentage points, for free. Which is why you read the credit report for remaining terms, not just for payments.

Exercise 4.25 †

Three questions:

1. "What does the automated underwriting system actually say?" A ratio is not a decision. If the findings are Approve/Eligible at 46%, the file is approved at 46%.
2. "Is that our overlay or the agency's guideline?" An employer's stricter rule is not the investor's rule, and another lender may not have it. Chapter 14.
3. "Which of those debts is nearly paid off, and what is the actual remaining term on each?" The ten-month rule (see 4.23) is the cheapest four points available.

Credit also for: "Is the income calculation right?" (Chapter 11 — variable income is frequently understated), "Are there reserves or other compensating factors we haven't documented?" and "Whose credit report is dragging the representative score, and can it be rescored?"

Exercise 4.26 †

Which to worry about: the Linden Street file, decisively — even though its ratios are more comfortable. A household absorbing a 64% increase in its largest monthly obligation is doing something it has no track record of doing. The second borrower has already demonstrated they can carry \$2,650 and is being asked for \$250 more.

The deeper point: DTI ranks these two files in the opposite order from payment shock, and payment shock is the one the household experiences. Neither ratio is wrong; they answer different questions.

Exercise 4.29 †

Model script:

"You do qualify up there — that's a real number, not a soft one, and if you find the right house at four-sixty I'll close it. Let me show you the other side of it, though.

At four-sixty, your all-in payment is around thirty-six hundred a month. At three-eighty-five it's about three thousand. You're paying eighteen-fifty now, so one of those is a sixty percent jump and the other is closer to ninety-five.

Both get approved. The difference is what happens the month the water heater goes. Where do you want to be?"

Grading note: full credit requires all three elements — confirm the qualification without hedging, give both payment figures, and hand the decision back. Scripts that talk the borrower down score lower than scripts that inform them, because the objective is an informed borrower rather than a smaller loan.

Exercise 4.31 †

$$\frac{\$3{,}400}{\$54} = \mathbf{63.0 \text{ months}} = \mathbf{5.2 \text{ years}}$$

The question you must ask: how long until you sell or refinance, whichever comes first? Not how long they intend to own the house — the loan ends at either event.

Two circumstances that make it a mistake even with a favorable answer:

1. Rates fall. A drop of a point in year three triggers a refinance; the monthly saving stops and the \$3,400 stays spent.
2. The cash is needed elsewhere. \$3,400 spent on points is \$3,400 not in reserves. On a file with four months of reserves, converting cash into a 5.2-year payback is a real reduction in the household's ability to absorb a shock — and reserves are also a compensating factor in underwriting.

Credit also for: the borrower takes the standard deduction, so the tax effect they were promised does not exist for them.

Exercise 4.33 †

The exclusions come from Regulation Z §1026.4(c)(7) (real-estate-related fees, when bona fide and reasonable) and §1026.4(e) (itemized recording fees). The pattern worth remembering: fees paid to third parties for services about the property are generally excluded; charges that are the cost of the credit itself are included. The tax service fee is the one that surprises people — it is included.

Exercise 4.35 †

1. It assumes the loan is held to maturity. The borrower who sells in year six paid all \$6,095.34 of prepaid finance charges and received one-fifth of the term. Their effective cost far exceeded 7.253%. Misleads: any borrower with a short horizon — a relocating professional, a first-time buyer expecting to move up, anyone in a market where rates are expected to fall.

2. It compares poorly across loan types. An ARM's APR must assume future index values, and those assumptions will not be what happens. Misleads: a borrower comparing a 5/1 ARM's APR to a 30-year fixed's APR as though both were computed rather than one being projected.

Credit also for: the exclusions mean two loans with identical APRs can require materially different cash at closing; and MIP duration assumptions on FHA loans are invisible in the single figure.

Exercise 4.37 †

The borrowers are \$1,400 short.

Lever 1 — move the closing date later in the month. Per-diem interest is \$66.3861. Moving from October 24 (8 days prepaid, \$531.09) to October 30 (2 days, \$132.77) reduces cash needed by \$398.32. Moving to the last business day available saves slightly more. Costs nothing, requires the seller's and closing agent's cooperation, and does not save a payment — it defers collecting interest.

Lever 2 — take a lender credit by accepting a higher rate. From §4.7's grid, moving from 6.625% to 6.875% converts a \$1,828.75 point charge into a **\$1,371.56 credit — a swing of \$3,200.31** in cash at closing. The cost is \$60.78 more per month (\$2,402.72 vs. \$2,341.94) for as long as they hold the loan.

Together the two levers cover the shortfall more than twice over. The judgment call is that lever 2 is permanent and lever 1 is free — so try lever 1 first, and use lever 2 only for what remains. Credit also for: reducing the down payment is not generally available here, since 5% is already the program minimum, and asking the seller for a larger credit is capped by interested-party-contribution limits (Chapter 20).

Exercise 4.38 †

Part 1 — the twelve-row schedule. Month-12 balance must be \$361,757.88. Total principal paid over twelve payments: \$3,992.12**. Total interest: **\$24,111.16.

Part 2 — the file rebuilt at 10% down:

The line that should surprise you: the mortgage insurance. It nearly halves — from \$176.78 to \$92.40 — because the MI factor itself drops with the LTV, not merely because the loan is smaller. \$84.38 of the \$207.64 monthly saving, about 41%, comes from the MI factor rather than from the reduced principal.

Students who expected the saving to be roughly proportional to the loan reduction (5.26% smaller loan → ~5% smaller payment) will find the actual saving of 6.8% larger than expected, and the MI factor is why.

The trade: \$19,250 of additional cash — which these borrowers do not have, since their total verified funds are \$38,000 and cash to close at 5% down is already \$25,376.34 — buys \$207.64 a month. A 92.7-month payback, and it would leave them with no reserves at all. Correct answer: they cannot do this, and the arithmetic is why. That is the point of the exercise.

(Illustrative MI factors. Verify current rate cards with the mortgage insurers.)

Chapter 5

Exercise 5.2 †

Conventional = not insured or guaranteed by a government agency. Conforming = meets Fannie Mae's or Freddie Mac's purchase requirements, including a maximum loan amount. The two are independent.

The teaching point: "conventional" says nothing about size or down payment.

Exercise 5.4 †

Premium structure: an upfront premium (UFMIP), commonly 1.75% of the base loan amount, typically financed into the loan; plus an annual MIP collected monthly, whose factor varies by loan-to-value and loan amount, and whose duration depends on the LTV at origination (11 years at 90% or less; life of the loan above 90%).

Illustrative — verify current figures with HUD.

Exercise 5.6 †

The two USDA tests:

1. Geographic eligibility — the property must be in a designated eligible area, checkable on USDA's property eligibility map.
2. Household income limit — commonly stated as 115% of area median income.

The second one counts income of people not on the loan. It is a household limit. A borrower who qualifies comfortably on their own income can be ineligible because an adult child living in the household earns money. Ask on day one.

Exercise 5.8 †

Exercise 5.11 †

Model reply, 84 words:

"Fair enough, and worth untangling one thing — 'conventional' just means the loan isn't backed by a government agency. It doesn't mean better, and a lot of very ordinary borrowers use FHA or VA because those programs fit them better.

What I'd rather do than pick a label is price two or three of them side by side on your actual numbers, and let you look at the real difference. Sometimes conventional wins and sometimes it doesn't, and it's usually not obvious until you see it."

Exercise 5.13 †

"I wish I could show you the rule, and I can't — jumbo guidelines aren't published the way agency guidelines are. They're an investor's internal matrix, and I'm allowed to tell you what it says but not to hand it over. That's genuinely frustrating and I'm not going to pretend otherwise.

Here's what I can do. This is one investor's answer, not the market's. I have three others I can submit to, and their rules on this particular point differ. Give me two days."

Grading note: the answer must (a) acknowledge the limitation honestly, (b) not pretend it is the borrower's fault or the market's verdict, and (c) name the next concrete step.

Exercise 5.15 †

To the borrower:

"That's advice I hear a lot, and it's mostly left over from how the program worked twenty years ago. There's a kernel of truth — VA has property condition requirements and the appraisal comes through the VA, so it can add a step. What it doesn't do is make your offer fail.

Here's what it's worth on your file: no down payment, no monthly mortgage insurance ever, and depending on your disability rating, possibly no funding fee. On a file like this that's tens of thousands of dollars. I'd want a very concrete reason before walking away from that."

To the agent:

"Can I show you something on the VA side? I know the reputation, and I want to walk you through what actually happens on a VA appraisal now — including Tidewater, which is the part that scares people and is more manageable than it sounds. It'll make you better at advising the next veteran who calls you, and it means we don't leave money on their table."

Grading note: the agent version must be collaborative rather than corrective. The agent is not the enemy; they are repeating something nobody has ever checked for them. Chapter 38's lunch-and-learn material is the scaled version of this conversation.

Exercise 5.17 †

An ARM's low initial payment does not help a borrower qualify because the Ability-to-Repay rule requires the creditor to qualify the borrower at the greater of the fully indexed rate or the introductory rate. On the chapter's illustration the borrower is qualified at 7.00% (\$2,433.34), not at the 5.875% initial rate (\$2,163.55).

The historical failure: the 2/28 and 3/27 hybrid ARMs of the 2000s, which were underwritten at the teaser rate. Borrowers qualified for a payment they would make for two years and then could not make. The plan was to refinance before the adjustment, which worked until house prices fell and refinancing became unavailable for everyone simultaneously. Chapter 2 §2.6.

The consequence for practice: the main reason borrowers ask for ARMs — "I can afford the payment at 5.875%" — is no longer a reason available to them.

Exercise 5.19 †

Additional down payment to reach the limit exactly: \$10,500 — bringing total cash down to **\$53,500** and the loan to exactly \$806,500. That is a 6.22% down payment.

Note what this costs beyond the cash: \$10,500 out of reserves, which is itself a compensating factor in underwriting (Chapter 14).

Exercise 5.21 †

From 5.20: base loan \$299,150.00 (\$310,000 × 96.5%), UFMIP \$5,235.13, total loan **\$304,385.13**.

Will it terminate? No. The LTV at origination — base loan ÷ price = \$299,150 ÷ \$310,000 = 96.50% — is above 90%, which places the loan in the life-of-loan category. The category is set at origination and is never revisited, so paying the balance down does not change it.

Total MIP over 360 payments: \$50,223.60.

Illustrative factors — verify with HUD.

Exercise 5.23 †

Note that the caps apply to the initial rate, not to the fully indexed rate — 5.875% + 5.00% = 10.875%.

Exercise 5.25 †

Which figure is the borrower more likely to have been shown? The monthly one — \$173.26 versus \$176.78, which makes FHA look \$3.52 a month better. Total lifetime cost of mortgage insurance appears on no standard disclosure, is not quoted by pricing engines, and is almost never presented.

That asymmetry is the exercise's point: the figure that decides the comparison is the one nobody shows.

Exercise 5.27 †

Grading note: (e) is the one students get wrong by assuming "rural" means farmland. (c) is the one they get wrong by reaching for non-QM before attempting an agency analysis.

Exercise 5.29 †

The table:

The three sentences:

"On the two things people ask about — the cash and the payment — FHA is better: about fifty-eight hundred less down and eighteen dollars a month cheaper. The catch is the mortgage insurance: on conventional it comes off automatically in about eleven and a half years, and on FHA at this down payment it never comes off, which over the full term is about thirty-eight thousand dollars more. So this is really one question."

The question to ask: "How long do you expect to keep this loan — not the house, the loan? If you'd refinance the moment rates drop a point, the thirty-eight thousand is mostly theoretical. If this is the loan you plan to pay off, it isn't."

Exercise 5.31 †

What you do, in order:

1. Ask directly and open-endedly. "Help me understand the plan — you're keeping the current house and this one will be home? Walk me through how that works with your commute." Then listen.
2. Write down the answer, in the file, in the borrower's words.
3. Ask the follow-ups the underwriter will ask. Will the current property be rented? Is there a lease? Is there a job change or a family reason? Is the current home being listed?
4. Explain the occupancy question honestly and early. "Primary residence pricing is better and the down payment is lower, and the reason is that lenders lose less on primary residences. So they check. If this is really an investment property or a second home, I'd rather structure it that way from the start than have it come apart in underwriting."
5. Price it both ways if there is genuine ambiguity, so the borrower can see what the correct structure costs.

What you do not do: suggest how to characterize it; help construct an explanation; proceed on an occupancy statement you have reason to doubt without documenting the conversation; or accept "it's primary" as an answer when the facts point elsewhere and you have not asked why.

Most of the time there is an innocent explanation — a new job, a parent to care for, a spouse's relocation. The reason to ask is that when there is not one, you want the file to show you asked. Chapter 27.

Exercise 5.34 †

The eliminations, one sentence each:

Facts needed to choose between conventional 95% and FHA 96.5%:

Five facts we have, four we do not — and all four missing ones are about the borrowers rather than the loan. That asymmetry is the point of the exercise and the reason Chapter 8 comes before Chapter 13.

Chapter 6

Worked solutions to the daggered (†) and odd-numbered exercises. Where an exercise asks for judgment or a piece of writing, the "solution" is a model answer plus the criteria a grader should apply — not the only acceptable response.

Exercise 6.1

Grading note: the point of the exercise is the exit event, not the activity. "Processing ends when all documents are collected" is wrong — documents may be collected on day 9 and the file may not be submitted until day 23. A stage is defined by the event that moves the file out of it.

Exercise 6.2 †

The six items: (1) the consumer's name; (2) the consumer's income; (3) the consumer's Social Security number, to obtain a credit report; (4) the property address; (5) an estimate of the value of the property; (6) the mortgage loan amount sought.

Two things that happen the moment all six exist:

1. The Loan Estimate obligation attaches — the LE must be delivered or placed in the mail within three business days, with a second deadline tied to consummation.
2. The fee restriction attaches — before the consumer has received the LE and indicated intent to proceed, the lender generally may not impose fees other than a bona fide and reasonable credit report fee.

Either of these is also acceptable: an ECOA/Regulation B notice-of-action-taken obligation now runs, and the file is subject to record-retention and (where applicable) HMDA reporting considerations.

Verify current requirements with compliance and the regulator.

Exercise 6.3

(a) Pre-qualification — an estimate of borrowing capacity built from what the borrower has told you, with nothing checked. (b) Pre-approval — a written statement of qualification built from documents you have obtained and reviewed, subject to a property and to underwriting conditions.

The distinguishing word: verified.

Grading note: a student who says "a pre-approval is stronger" has not answered the question. The difference is not strength; it is the basis. A pre-approval on bad documents is weaker than a careful pre-qualification, and the label does not fix that.

Exercise 6.5

• Clear to close — the underwriter signs off on every prior-to-document condition and authorizes the closing department to prepare documents.
• Closing (consummation) — the borrowers sign the note and security instrument; conducted by a title company, escrow company, or attorney depending on state practice.
• Funding — the lender disburses the loan proceeds; the settlement agent then disburses to the parties.

Recording is a fourth event, performed by the closing agent, and it is what creates the public lien Chapter 1 described.

Exercise 6.7

A milestone is a defined, dated event in the loan origination system that a file has either reached or not — application taken, submitted, approved with conditions, clear to close, docs out, funded.

Why the definition determines the value of every turn-time number: turn time is the interval between two milestones. If two companies define "application taken" differently — one at the six-item threshold, one at the point a signed 1003 is returned — then their "days to close" numbers measure different intervals and cannot be compared, even though both are honestly reported. §6.7's "moving the start line" distortion is this problem exploited rather than merely suffered.

Exercise 6.9

Three facts needed:

1. The date the contract was executed — not the date the offer was written. On Linden Street the offer was written the night of day 0 naming a day-45 closing; execution came day 4, so 41 days remained.
2. The closing date named in the contract, as a date, not as a duration.
3. Every other dated deadline in the contract — financing contingency, inspection, appraisal contingency — because those expire before the closing date and are what actually put earnest money at risk (Chapter 20).

Most often overlooked: the gap between the offer date and the execution date. It is invisible, nobody announces it, and it is pure loss from your calendar.

Exercise 6.10 †

Model rebuttal:

Five days at the front of a file is not five days you can make up later, because the stages after it are not compressible by effort. The appraisal takes as long as it takes; the title examiner's queue is the title examiner's queue; the underwriter's first look is a published standard you do not control. Ordering on day 12 does not delay one stage by five days — it moves every subsequent date by five days, and the only stage with any give in it is the condition loop, which is exactly the stage you most want margin in because it is the one with no bounded duration. So the five days do not get made up; they get spent out of the only reserve the file has. And they get spent invisibly, because the damage does not surface for a month, at which point nobody connects a missed closing date to an order that went out on day 12 instead of day 7.

Grading note: an answer that argues from urgency ("we should always move fast") has missed the exercise. The argument has to be structural — where in the pipeline the slack actually lives.

Exercise 6.11

Two defensible answers, both true:

• First-look turn time: 4 days (submitted day 20 → suspended day 24). This is what an operations report would show.
• Submission to decision, end to end: 10 days (day 20 → day 30), which is what the file actually experienced.
• A third is also defensible: 4 days + 4 days = 8 days of underwriting work, with 2 days of re-submission time excluded as the loan officer's own.

Which to report to an agent: 10 days. The agent is not grading the underwriting department; they are asking how long the file took. Reporting the 4-day number to an agent is technically true and functionally misleading — and when the file's contract-to-close number lands, the agent will notice the gap and will discount everything you told them.

Exercise 6.13

The structural reason: every other stage exits on an event that a defined party is obligated to produce. Underwriting exits on a decision the underwriter owes. Processing exits on a submission the processor owes. Stage 5 exits on clear to close, which requires an unknown number of round trips between three sets of parties, each with its own queue, and nothing in the structure caps the number of trips. A second round of conditions can generate a third.

One change that would bound it: require the underwriter to review returned conditions within a published, short standard (for example, same business day for a resubmission of previously identified items), and require that any new condition arising from a resubmission be approved by an underwriting manager. The first bounds the queue time per round trip; the second bounds the number of round trips. Other defensible answers: a hard rule that all conditions be issued in a single list at first decision, or a "condition triage" call between LO, processor, and underwriter within 24 hours of the conditional approval.

Exercise 6.15

Two examples other than new credit, with the document each one moves:

1. Documentation goes stale. Paystubs and bank statements have a shelf life measured to the note date; a file that sits long enough must be re-documented. The documents that move: the paystubs, the bank statements, and potentially the credit report itself — which, on a file where anything has changed, is not a neutral re-pull.
2. Employment changes. A promotion, a transfer to commission-only compensation, or a layoff between approval and funding changes qualifying income. The document that moves: the verbal verification of employment, which is a prior-to-funding condition specifically because the lender does not trust the world to hold still.

Also acceptable: an insurance market that suspends binding after a catastrophe (the binder), a rate lock approaching expiration (the lock confirmation and, eventually, an extension), a property damaged between appraisal and closing (a re-inspection), or a co-borrower's death or divorce.

Exercise 6.16 †

Step 1 — monthly taxes.

$$\$3{,}300.00 \div 12 = \$275.00$$

Step 2 — monthly homeowners insurance.

$$\$1{,}320.00 \div 12 = \$110.00$$

Step 3 — monthly mortgage insurance. The factor is annual and applies to the loan amount:

$$\$279{,}000 \times 0.0030 = \$837.00 \text{ per year} \qquad \$837.00 \div 12 = \$69.75$$

Step 4 — full PITI + MI.

Check: 1,809.59 + 275.00 = 2,084.59; + 110.00 = 2,194.59; + 69.75 = 2,264.34. ✓

Exercise 6.17

Housing (front-end) ratio:

$$\frac{\$2{,}264.34}{\$7{,}800.00} = 0.2903 = \mathbf{29.03\%}$$

Total obligations:

$$\$2{,}264.34 + \$640.00 = \$2{,}904.34$$

Total debt (back-end) ratio:

$$\frac{\$2{,}904.34}{\$7{,}800.00} = 0.372351 = \mathbf{37.24\%}$$

In plain language: about twenty-nine cents of every pre-tax dollar this household earns goes to the house, and about thirty-seven cents is committed to the house plus everything else they owe — before taxes are withheld, before groceries, before anything they choose.

Exercise 6.18 †

New total obligations:

$$\$2{,}904.34 + \$310.00 = \$3{,}214.34$$

New back-end ratio:

$$\frac{\$3{,}214.34}{\$7{,}800.00} = 0.412095 = \mathbf{41.21\%}$$

The ratio rose 3.97 percentage points (37.24% → 41.21%) and is still below 43%.

Why the approval may be blown anyway. Condition 11 in FIGURE 6.1 does not read "DTI must be below 43%." It reads "DTI may not exceed the approved ratio." The approval was issued against a specific file with a specific ratio, and the automated underwriting recommendation the lender is relying on was returned against that data. New debt means the findings no longer match the file, which means the findings must be re-run — and a re-run can return a different recommendation, different documentation requirements, or a different eligibility answer entirely, none of which is guaranteed to be as favorable.

Two further points worth stating:

• The debt itself is not the only problem. The inquiry and the new account are also new facts, and a lender is entitled to ask what else changed.
• A file that has to be re-underwritten in its closing week is a file that has lost its place in every queue.

Exercise 6.19

A model sketch. Every assumption must be stated — that is what is being graded.

Earliest defensible closing: day 34 — and only with every assumption holding. Note what is not compressible in that table: days 6→18 (third parties), day 19→24 (the underwriting standard), and day 31→34 (the disclosure waiting period). Roughly twenty of the twenty-nine days are somebody else's clock or the regulation's.

Grading note: any answer is acceptable if the assumptions are stated and the arithmetic is consistent. An answer with no stated assumptions is wrong regardless of the number.

Exercise 6.20 †

Model ninety-minute plan, in order:

1. Read all eleven conditions before doing anything (5 min). Do not start clearing the first one you can clear. You are looking for the one that will take longest, which is almost never the one at the top.
2. Sort by source into three lists (5 min): six borrower items, two third-party items, three lender items.
3. Sort by timing (5 min): nine PTD, two PTF. The two PTF items go straight onto the calendar for the closing week and off today's list. This is the step that prevents wasted effort.
4. Call the title company about condition 7 today (15 min). Third-party items have the longest tail and the least leverage; they must start first. This is the condition that will govern the file's date.
5. Send the written VOE follow-up for condition 3 (10 min) — also third-party, also long-tailed.
6. Confirm conditions 9, 10, and 11 are queued internally (10 min). You cannot work them; you can confirm nobody forgot them.
7. Draft the borrower message (30 min) — six items, plain language, one due date. (Exercise 6.29.)
8. Ask operations for the earliest date this file can be cleared to close if everything lands (10 min). This is the step almost nobody takes and it is the one that converts a condition list into a plan.

Grading criteria: third-party items first; PTF items removed from the active list; the borrower message drafted but not sent until the plan is complete; and an explicit CTC target set on day 28 rather than allowed to emerge.

Exercise 6.21

Borrower-supplied (6): conditions 1, 2, 4, 5, 6, 8. Leverage: real — you can call, text, and if necessary drive there; escalate by raising the stakes honestly ("this is the item holding your closing date"), not by nagging. Third-party (2): conditions 3, 7. Leverage: influence only — someone else's queue governs; escalate to the vendor's manager, and where a relationship exists, to your operations team's contact. Lender-internal (3): conditions 9, 10, 11. Leverage: none — you can only confirm they are queued; escalate to the underwriting or closing manager if a date approaches with an item unassigned.

Exercise 6.22 †

The two: condition 10 (verbal verification of employment) and condition 11 (pre-closing credit refresh / undisclosed-debt report).

Why: both are prior-to-funding items that must be current as of the note date. A verbal VOE obtained four weeks early proves employment four weeks ago, which is not what the guideline requires and not what the investor is relying on. A credit refresh pulled early is a photograph of a moment that will have passed. Their whole purpose is to be late — they exist because the lender does not trust the world to hold still between approval and funding.

What the loan officer should do about them: three things, and "nothing" is only defensible if all three are done.

1. Take them off the active worklist, so effort goes where it can produce movement.
2. Put them on the calendar with a date, so nobody discovers on the closing morning that the VVOE was never ordered.
3. Manage the risk they exist to detect. This is the real answer. Condition 11 is a tripwire; you cannot clear it early, but you can reduce the chance it trips — by telling the borrowers, more than once and in language they will remember, not to open credit, finance anything, change jobs, or move money. On the Linden Street file that conversation was worth the closing date and a back-end ratio that went to 48.48%.

A caution for graders, because students reliably over-attribute here. The \$914.38 lock extension is not on this list. That charge was incurred on day 42, two days before the credit refresh fired, and it would have been incurred whether or not anyone bought furniture — the 30-day lock taken on day 12 expired three days before the file's own scheduled closing. A student who credits the "no new credit" conversation with saving \$914.38 has merged two independent failures that happened to share a window. The furniture cost the closing date; the stall and the lock term cost the money.

Exercise 6.23

Model questions, in order, with what each changes:

1. "Is the lien released, or is it recorded and awaiting a release?" — This establishes whether the problem is a document or a negotiation. A recorded release that has not yet been reflected is a days problem; an unreleased lien with a disputing contractor is a weeks problem and may be a different-house problem.
2. "Who has to act next, and do you have their contact information?" — Establishes whether the title company is waiting on the prior owner, the contractor, the seller's attorney, or a recorder's office. Each has a different escalation path, and only one of them is anybody's to push.
3. "What is your realistic date, and what would have to happen for it to be sooner?" — Gets a date you can put in front of the agent and the borrowers, and identifies the single lever, if any.

Grading note: the order matters. Asking for a date first gets an answer that means nothing, because the person giving it has not yet told you what the problem is.

Exercise 6.25

The three files, and the action before 9:00 a.m.:

1. L-2201 — 14 days in conditions on a 30-day-old file. The FLAG says "AGING," which describes the symptom, not the problem. Action: call and ask which specific condition is blocked and by whom. Not "any update?"
2. L-2226 — 9 days in stage and no borrower contact in 4 days. Two flags compounding: a file in conditions that nobody has spoken to has almost certainly stopped. Action: call the borrower, not the processor.
3. L-2247 — took an application yesterday and is not locked. The FLAG says so, but the report does not treat it as urgent, and it is: this is an unhedged pricing exposure on a live file (Chapter 30). Action: have the lock conversation today.

A fourth is also defensible: L-2231 (Harlow Street) has been at pre-approval for 12 days and is flagged "shopping." Not a process failure — but it is in the total, and it will not close, which is exactly §6.9's warning about adding a pre-approval to a dollar column.

Exercise 6.27

Benign reading: the file was conditionally approved today, and 0 days in stage simply means it just arrived at that milestone. Age 34 is unremarkable for a purchase file.

Alarming reading: the file has been in the "Approved" milestone for some time but the milestone was re-stamped today — for example because it was resubmitted and re-decisioned, resetting the in-stage counter and hiding however long it had actually been sitting. Age 34 with a fresh milestone can mean a file that has gone around the condition loop more than once.

The distinguishing question: what was the date of the original conditional approval, and how many times has this file been decisioned? Reports that show only the current milestone date cannot answer it; the loan origination system's milestone history can.

Exercise 6.28 †

The likely inference: the approval was issued well before the closing date. A competent lender issues prior-to-funding conditions on every file, because dated items — the verbal verification of employment, the pre-closing credit refresh — exist on every file. If none appear on the list, the most likely explanation is that the approval was issued far enough from the closing that the PTF items have not been added yet; some lenders add them at the clear-to-close stage rather than at first decision.

What to verify before relying on it: ask the underwriter or your operations team directly whether this lender issues PTF conditions at first decision or later, and whether the ones you expect — verbal VOE, credit refresh, undisclosed-debt report — are standard on this program. Two other explanations must be ruled out: that the list is incomplete, and that this lender simply does not require a pre-closing refresh, which would be worth knowing for a different reason.

Grading note: the wrong answer, and a tempting one, is "this file has fewer conditions and is therefore in better shape." A shorter list is not a cleaner file; it is frequently an earlier one.

Exercise 6.29

Model day-5 email. Criteria: under 250 words; four or more dated or approximate milestones; the credit instruction in memorable language; no promised closing date.

Subject: Your loan — what happens over the next six weeks

Congratulations on the accepted contract. Here is the map, so nothing that happens is a surprise.

This week: you'll receive your Loan Estimate — that's the official cost sheet, and I'll walk you through it line by line. I'm ordering the appraisal and the title work in the next day or two.

Weeks two and three: the appraisal and title work come back. These are the two things nobody in this transaction controls, and they're the most common reason a file moves. I'll tell you the day each one lands.

Week four: the file goes to underwriting. About a week later we'll get an approval with a list of items — usually eight to twelve. That list is normal. It is not a sign of a problem, and most of the items are small.

The last week: you'll get a Closing Disclosure a few days before signing. The figures on it should match what I've told you. If anything is different, call me before you sign.

One thing I need from you, and it matters more than anything else on this list. Between now and the day we fund: do not open any credit, finance anything, change jobs, or move money between accounts. Not a car, not a couch, not a store card at checkout. We re-check your credit days before closing, and a new payment can undo everything above. If you're not sure, call me first — it's free and it takes two minutes.

Exercise 6.30 †

Model day-36 email, under 120 words, containing all four elements.

Subject: Linden Street — status, day 36

Quick update so you have it before your client asks.

Where it is: nine of eleven conditions cleared as of day 33.

What's left: the employment verification and the credit refresh. Both are pulled the week of closing by design — that's normal, not a problem.

What's next: I'm pushing for a clear to close ahead of the date rather than on it. I'll call you the day it happens.

One thing you can help with: remind them not to buy anything on credit until we fund. Not a car, not a couch, not a store card. It's the single most common way a done deal comes apart.

Word count: 103. Grading criteria: where it is · who owes the next action · when it changes · one specific thing they can do. An answer missing the fourth element is incomplete even if it is accurate — that element is what converts a status report into a risk control.

Exercise 6.31

Model memo shape (a full draft is the student's work; grade the structure):

• The proposal: a mandatory scripted "no new credit" contact at three fixed points — application, conditional approval, and clear to close — logged in the loan origination system as a milestone so it can be audited.
• The cost: roughly three minutes per file per touch, plus a one-time script and an LOS field. No vendor spend.
• Who has to do it: the loan officer at application and CTC; the processor at conditional approval. Operations has to add the field.
• How you would know in ninety days: count pre-closing credit refreshes that surface new debt, as a share of files refreshed, before and after. Also track closings missed for a DTI condition. Both numbers already exist in the system.

Grading note: a proposal without a measurement is a preference. The ninety-day test is the part being graded.

Exercise 6.33

There is no single right answer; there is a defensible one and several indefensible ones.

What you owe the truth: the number is not false, but it is measured from a start line the consumer never experienced and on a population that excludes the files that took longest. Marketing it as "24-day closings" invites a reader to understand it as contract-to-close, which it is not.

What you owe your employer: raise it internally first, with the specific defect named, and propose the alternative rather than simply refusing. Managers respond much better to "here is a number I can defend" than to "I won't."

What you owe your referral partners: the number they will actually experience. An agent who is promised 24 and lives 47 does not conclude that the measurement methodology differed. They conclude you are unreliable, and they are not wrong to.

What you owe your license: advertising in mortgage lending is regulated, and misleading claims about terms or performance carry real exposure. Chapter 26 covers advertising rules; the short version is that "technically true and predictably misunderstood" is not a safe harbor.

What to put on the flyer: a contract-to-close number you hit nine times out of ten, stated as such — for example, "most of my purchase files close within X days of contract" — or, better, no number and a specific commitment you fully control: "you get a call from me the day your file is cleared to close, and the day before that if it isn't going to be."

Exercise 6.34 †

B. The six items are name, income, Social Security number, property address, an estimate of the value of the property, and the mortgage loan amount sought.

Why the distractors work: A describes ordinary practice and is what most candidates picture, but a signature is not part of the definition. C describes what a loan officer needs to pull credit, not what constitutes an application. D is the pre-2015 catch-all restated as a rule — it is exactly the discretion the current definition removed, which is why it is the most tempting wrong answer for anyone who learned the older framework.

Exercise 6.35

B — a pre-approval. Verified information (credit report, paystubs, bank statements) was reviewed.

A is wrong because nothing was merely stated. C is wrong because a commitment to lend is an underwriting decision on a complete file with an identified property; this letter is expressly subject to both. D is wrong because a conditional approval is issued by an underwriter on a submitted file, not by a loan originator on a pre-approval.

Exercise 6.36 †

B — prior to funding. The defining feature is the timing: after documents are drawn, before disbursement.

A items gate the clear to close and therefore precede document preparation. C items are investor-level requirements that arise when the closed loan is delivered and sold; they do not delay the closing. D is not standard terminology for this and is included as a plausible-sounding distractor.

Exercise 6.37

B. The system evaluates the data entered against published guidelines and returns a recommendation plus the documentation required to rely on it.

A is the error the chapter spends a section on — a recommendation is not an approval and must not be communicated as one. C is wrong: a human underwriter still verifies that the data matches the documents, and reps and warrants depend on it. D is wrong: automated underwriting is routinely run before the appraisal exists, as it was on day 6 of the Linden Street file.

Exercise 6.38 †

B. A change in the annual percentage rate beyond the applicable tolerance is one of the specific triggers that requires a corrected Closing Disclosure and a new waiting period, along with a change in loan product and the addition of a prepayment penalty.

A describes the treatment of most other changes — corrected disclosure, no new waiting period — and is the answer a candidate gives if they remember only half the rule. C is wrong: the trigger is the APR, not the payment. D is wrong: the waiver is narrow, limited to a bona fide personal financial emergency, and is not available at anyone's discretion. Chapter 22 works the mechanics; verify current requirements with compliance.

Exercise 6.39

Grade the calendar against the frozen days in the chapter's Loan File checkpoint. The three added columns are the point of the exercise:

• Lock expiration — day 42, and this is the column that carries the whole exercise. The lock was taken on day 12 for 30 days: $12 + 30 = 42$. The strongest answers notice, without being prompted, that day 42 falls three days before the contract closing date of day 45 — meaning the lock was short from the moment it was taken, and an extension was arithmetically certain before the file had a single problem. A 15-day extension was purchased on day 42 at 0.250 point (\$914.38), carrying the lock to day 57. Chapter 30 owns the pricing judgment.
• Contract closing date — day 45, and the student should notice it was named from the offer date (day 0), not from execution (day 4). Set beside the lock expiration, this is the exercise's whole point: two dates on the same file, three days apart, and nobody compared them.
• Days since last borrower contact — not populatable from the chapter, and saying so is the correct answer. That is the exercise's real lesson: the most predictive column on a pipeline report is the one nobody records.

A student who invents contact dates to fill the column has made the error the exercise is testing for.

Exercise 6.40 †

Model counter-argument (two paragraphs), then the judgment.

Paragraph 1 — the seller's calendar is not yours. A closing date is a contract term agreed by two parties, and moving it forward requires the seller to be ready to vacate a week early. Sellers have movers booked, leases starting, and frequently a purchase of their own on the other end. A loan officer who plans around an accelerated closing is planning around something they cannot deliver, and a team that pushes for it can damage the relationship with the listing side for no gain.

Paragraph 2 — the exposure window cannot be closed, only moved. The pre-closing credit refresh must be pulled near the note date; that is what makes it useful. Pulling it on day 37 instead of day 44 does not eliminate the window in which a borrower can open credit — it relocates it. The borrowers who financed furniture on day 41 might have financed it on day 34 instead; people buy furniture when they are about to have a house, and the trigger is the closing, not the calendar. On this reading, the file was not saved by scheduling; it was exposed by a conversation that did not happen often enough.

The judgment — and it splits. The counter-argument is strong against one of the two failures and powerless against the other, which is what makes this exercise worth assigning.

Against the furniture, paragraph 2 lands. The refresh must be pulled near the note date, so accelerating the closing relocates the exposure window rather than closing it, and people buy furniture because a closing is imminent — the trigger travels with the date. On that failure the chapter's own remedy concedes the point: "say the furniture sentence more than once" addresses the mechanism, while "drive to clear to close" only shortens the exposure.

Against the lock, paragraph 2 has nothing to say. The lock expired on day 42. A closing on day 38 is inside day 42, so the \$914.38 extension would simply never have been purchased — no behavioral assumption required, no argument about what borrowers do, just two numbers. Scheduling does not relocate that cost; it eliminates it.

So the honest conclusion is that the two rules in §6.5 are not competing answers to one problem but correct answers to two different ones. Drive to clear to close is dispositive against the lock and merely helpful against the furniture. Say the furniture sentence more than once is dispositive against the furniture and irrelevant to the lock. A file needs both, and there is a third that this file needed more than either: compare the lock expiration to the contract date on the day you lock.

Grading note: a student who argues only paragraph 1 has produced a scheduling objection. A student who finds paragraph 2 has understood what the tripwire is for. A student who notices that paragraph 2 cannot touch the lock has done the best available work on this question — and it requires only that they check whether 38 is less than 42, which is exactly the kind of arithmetic nobody on the file performed.

Chapter 7

Worked solutions to the daggered (†) and odd-numbered exercises. Every constructed rate is stated before it is used; students should be reminded that the point of §7.1 is to replace all of them with measured numbers.

Exercise 7.1

A lead source is an identifiable origin of prospective borrowers — an agent partner, a past client, a purchased list, a walk-in, a social post — tracked separately so that cost per closed loan, conversion, ramp time, and durability can each be measured for it. Lead conversion is the share of prospects from a source who reach a defined next stage.

A conversion rate is meaningless without stable stage definitions because the denominator and numerator are both definitional choices. If "application" means a signed 1003 in January and a verbal intent in March, the March rate is not comparable to January's, and the trend line is measuring vocabulary rather than performance. The chapter's specific example is pull-through: shops define "application" differently enough that cross-company comparison of pull-through is close to meaningless. Fix the definitions once, write them down, and never change one mid-year.

Exercise 7.3 †

The four stages and rates (all [constructed teaching example]):

End to end:

$$0.40 \times 0.50 \times 0.40 \times 0.80 = 0.064 = \mathbf{6.4\%}$$

Contacts per closed loan: $1 \div 0.064 = \mathbf{15.625}$.

Exercise 7.5

A drip campaign is a pre-scheduled automated sequence of messages sent to a segment of a database over time.

Does well: education and market notes at scale, anniversary and event messages, staying present in the memory of several hundred people at a cost per contact near zero.

Cannot do: the annual review call. The chapter is explicit that the entire value of that call is that a human being who remembers you dialed the phone and quoted a fact from their file. Automate it and you have removed the only thing that made it work — and past clients can tell.

Exercise 7.7 †

An affiliated business arrangement is permitted under RESPA when all three are true:

1. The relationship is disclosed to the consumer at or before the time of referral;
2. The consumer is not required to use the affiliate; and
3. The referring party receives only a return on its ownership interest — nothing else of value.

Required use is the recurring wrong answer. Exam stems frequently describe a builder incentive available only through the affiliated lender, coupled with a statement that the buyer must use it; that fact pattern breaks condition 2 and is the trap.

Exercise 7.9

The lead-source report — twelve months, by source, showing conversations, applications, closings, volume, direct cost, and cost per closing. The chapter says it takes about an hour to produce and calls it the highest-return hour of the year.

The predicted direction of error: the originator's belief will overweight the channel that felt like work. In Figure 7.1 the originator believed their business was "mostly agent referrals" and was half right — agents produced 12 of 24 closings — but had no idea that purchased leads consumed 60% of the marketing budget (\$5,400 of \$9,030) to produce 8% of the closings, at \$2,700.00 each against \$154.17 for an agent closing.

Exercise 7.11

(See 7.9 — same report, same hour, same predicted direction of error. Where 7.9 asks what the document is, this item asks the student to name it unprompted.) Credit any answer that identifies a lead-source report run from the CRM or LOS, notes that it requires populated source fields to be true, and predicts the error direction as overweighting the effortful channel.

Exercise 7.13

Section 7.3's constructed base rate is 5 producing partners out of 30 agents met — 16.7%, over eighteen months.

The originator has run a sample of six against a base rate of one in six. The expected number of conversions from six meetings is $6 \times 0.167 = 1.0$, and the probability of drawing zero from six attempts at that rate is substantial — roughly $(1 - 0.167)^6 \approx 0.33$, or about one time in three. Getting zero from six is an ordinary outcome, not evidence of anything.

A statistically honest version of the conclusion would require a sample large enough to distinguish the originator's true rate from 16.7%, plus a full eighteen-month ramp, since the chapter's rate is measured over that horizon and the originator has run four months. They have not learned that agent business does not work for them. They have learned that they stopped.

Exercise 7.15

Model answer:

"I'm not going to — and it's not me being difficult. A rate sheet without the four facts that price a file is worse than nothing: it'll show you a number that isn't available to your buyer, and then I'm the person who quoted it. What I'll do instead is this — send me any buyer, any time, and inside a day you'll get a real number for that specific file, and a letter you can hand a listing agent without worrying. And once a month I'll send you thirty seconds on what actually moved and what it does to a \$400,000 buyer. That's more useful to you than a sheet you can't quote from."

Grade on: refusing without making the agent feel managed; naming why (the four pricing facts — representative score, LTV, occupancy/property type, lock period, per Ch.1's discipline); replacing the refusal with something more valuable; and not mentioning a rate anywhere.

Exercise 7.17 †

The risk: company-provided leads produce income immediately and are worth nothing the day the originator changes employers — and, as Case Study 7.2 shows, the risk can also arrive without a job change, when the employer redirects the marketing budget.

The cost of realizing it, using the chapter's constructed figures: the originator must rebuild from zero at 54 hours per producing agent partner, with a twelve-to-eighteen-month ramp before the first dollar. Five partners is $5 \times 54 = 270$ hours, which at the chapter's 15 hours a month of partner development is eighteen months — during which the originator must live on whatever remains. A mature partner is worth roughly \$19,200 a year, so the delayed asset is $5 \times \$19{,}200 = \$96{,}000$ a year of production that does not exist during the rebuild.

What they should be doing: not stopping the leads. Capturing them. Every closed borrower goes into a database with populated trigger fields, and a fixed block of hours each week goes to partner development — the 4.9 hours in §7.1's time budget that contains the entire future of the business. Case Study 7.2's Originator A had 88 households in an LOS over two years and let every one go; at §7.4's 0.11 factor those alone would have produced nearly ten closings a year at no acquisition cost.

Exercise 7.19 †

(a) End-to-end rate.

$$0.35 \times 0.55 \times 0.45 \times 0.78 = 0.0675675 = \mathbf{6.757\%}$$

(b) Required monthly volume at each stage, working backward from 3 closings:

Check against the end-to-end rate: $3 \div 0.0675675 = 44.40$. ✓

(c) Per business day, at 21 business days a month:

$$44.400 \div 21 = \mathbf{2.11 \text{ new contacts per business day}}$$

Exercise 7.21 †

At the stated 0.09 factor:

$$180 \times 0.09 = \mathbf{16.2 \text{ closings per year}}$$ $$16.2 \times \$3{,}250 = \mathbf{\$52{,}650}$$

At a true factor of 0.05:

$$180 \times 0.05 = \mathbf{9.0 \text{ closings per year}}$$ $$9.0 \times \$3{,}250 = \mathbf{\$29{,}250}$$

Difference: 7.2 closings and \$23,400 a year — from the same database, the same effort, and the same number of households. This is the chapter's central caution about §7.4: the compounding is real, but the factor is earned by the quality of the contact rather than assumed, and it is the number a student must measure rather than adopt.

Exercise 7.23

Compensation per closed loan:

$$0.0125 \times \$275{,}000 = \$3{,}437.50$$

Closings required:

$$\$120{,}000 \div \$3{,}437.50 = 34.909 \to \mathbf{35 \text{ closings}}$$

Annual contacts at the constructed 6.4% end-to-end rate:

$$35 \div 0.064 = \mathbf{546.9 \to \text{about } 547 \text{ contacts a year}}$$

At roughly 250 business days, that is 2.19 new contacts per business day — about 46% more daily prospecting than the 1.5-a-day figure in §7.1, for a 46% higher closing target. Note that the relationship is linear in contacts but the lag does not scale: the first closing still lands around day 139 regardless of the target.

Exercise 7.25 †

(a) Gross compensation per closed loan.

$$0.0090 \times \$310{,}000 = \mathbf{\$2{,}790.00}$$

(b) Break-even conversion.

$$\frac{\$55.00}{\$2{,}790.00} = 0.019713 = \mathbf{1.97\%}$$

(c) At a measured 1.20% conversion.

$$\text{leads per closing} = 1 \div 0.0120 = 83.33$$ $$\text{lead cost per closing} = 83.33 \times \$55.00 = \mathbf{\$4{,}583.33}$$ $$\text{margin} = \$2{,}790.00 - \$4{,}583.33 = \mathbf{-\$1{,}793.33 \text{ per closed loan}}$$

(d) The decision. Stop, at the end of the bounded test. The measured conversion is well below break-even, so each closed loan costs \$1,793.33 more in lead spend than it produces in compensation — before a single hour of the originator's time is counted. This is not a channel to optimize; at 1.20% against a 1.97% break-even it is a channel to exit, and the honest next question is whether the hours would have produced a producing agent partner instead.

Exercise 7.27 †

Annual hours:

$$18 \text{ hrs/month} \times 12 = 216 \text{ hours}$$

Producing partners forgone, at 54 hours each:

$$216 \div 54 = \mathbf{4.0 \text{ producing agent partners}}$$

Dollar opportunity cost, at \$14,000 a year per mature partner:

$$4.0 \times \$14{,}000 = \mathbf{\$56{,}000 \text{ a year, at maturity}}$$

Discounted to a 25% realization:

$$\$56{,}000 \times 0.25 = \mathbf{\$14{,}000 \text{ a year}}$$

Does the conclusion change? No, and the student should say why: even at a quarter realization — which concedes that most of those hours would not actually have gone to partner development, that most partners would not mature, and that none would reach full production — the forgone annual value exceeds anything 18 hours a month of lead follow-up is likely to have produced. What does change is the confidence with which the claim can be made, and the honest version of the argument states the discount rather than hiding it.

Exercise 7.29 †

(a) Annual gross compensation.

$$0.0110 \times \$285{,}000 = \$3{,}135.00 \text{ per closed loan}$$ $$3 \times \$3{,}135.00 = \mathbf{\$9{,}405.00 \text{ per year}}$$

(b) Five-year gross.

$$5 \times \$9{,}405.00 = \mathbf{\$47{,}025.00}$$

(c) Total investment and return multiple.

$$\$480.00 + (36 \times \$100.00) = \$480.00 + \$3{,}600.00 = \mathbf{\$4{,}080.00}$$ $$\$47{,}025.00 \div \$4{,}080.00 = \mathbf{11.5\times}$$

(d) Acquisition cost per closed loan.

Closings over five years: $3 \times 5 = 15$.

$$\$4{,}080.00 \div 15 = \mathbf{\$272.00 \text{ per closed loan}}$$

Against \$2,666.67 for a purchased lead at a 1.5% conversion, this relationship delivers closings about $\$2{,}666.67 \div \$272.00 = \mathbf{9.8\times}$ cheaper — and unlike the leads, it compounds, because each of those fifteen borrowers enters a database.

Note this is a weaker relationship than the Linden Street agent (3 closings a year rather than 5, \$285,000 rather than \$320,000, no second-order referrals counted) and it still beats purchased leads by nearly ten times. That is the point of the exercise.

Exercise 7.31 †

Model answer (86 words):

Hi — we haven't met. I'm a loan officer at [company], NMLS #[ID], and I'm not writing to ask for business; you've clearly got a lender and I'd be surprised if you didn't.

I'm trying to get better at the part of this job that costs agents deals, and you close enough volume that you'd know. Could I buy you fifteen minutes of coffee and ask one question — what's the last transaction that fell apart on financing, and what actually happened?

Any morning next week.

Grade on: (i) no referral request; (ii) no rate, no product claim, no "twenty-four seven"; (iii) a reason to reply that is about her expertise, not his availability; (iv) a specific, small ask with a time bound; (v) the NMLS ID present; (vi) whether the student would actually send it. Reject anything containing "I'd love the opportunity to earn your business."

Exercise 7.33

Model opening paragraph:

I'm working on your client's purchase financing and wanted to send you the income analysis before the underwriter sees it, so nothing surprises either of us. What I've computed is not net income and it isn't cash flow — it's "qualifying income," a defined figure produced by the lender's own worksheet, which starts at the return you prepared and then applies the investor's add-backs and deductions. It answers a narrower question than the one you were answering. My figure is \$8,916.67 a month. The worksheet is attached, line by line, so you can see exactly which adjustments produced the difference — and section 4 lists the specific items on next year's return that would move it, in both directions.

Grade on: never stating or implying the CPA is wrong; naming explicitly that the two figures answer different questions; showing the worksheet rather than asserting a number; and including the forward-looking item, which is the part that actually earns the referral. Deduct for any sentence offering tax advice.

Exercise 7.35 †

(a) Defensible proportionate share.

The loan officer occupies roughly 30% of the creative (the agent's photo, logo, and four listings occupy ~70%).

$$\$1{,}800.00 \times 30\% = \mathbf{\$540.00 \text{ per month}}$$

(b) The difference.

A 50/50 split bills the loan officer $\$1{,}800.00 \div 2 = \$900.00$.

$$\$900.00 - \$540.00 = \mathbf{\$360.00 \text{ per month}}$$ $$\$360.00 \times 12 = \mathbf{\$4{,}320.00 \text{ per year}}$$

(c) The analysis and the action.

Statute: RESPA Section 8 — 8(a) on things of value pursuant to an agreement or understanding for the referral of settlement service business, with the 8(c) goods-and-services exception being the only shelter available, and it does not cover the \$360 excess because no service explains it.

Direction: value is moving from the loan officer (a lender) to the agent (a referral source). The excess is the loan officer paying part of the agent's advertising costs.

What you do next: propose the measured split — \$540 and \$1,260 — in writing, with the measurement attached. If the agent will not accept it, decline the arrangement. Either way, take the proposal and the creative to compliance before the first invoice, not after. Do not agree verbally and "sort out the paperwork later," which is how Figure 7.2's eleven-month pattern begins. And in all cases, verify current requirements with your compliance department.

Exercise 7.37

Statute: RESPA Section 8. Direction: the thing of value is moving toward the loan officer — this is the reverse-direction case, and it is the chapter's most reliable blind spot. Who is at risk: both parties, but the loan officer specifically, because Section 8 prohibits accepting as well as giving, and a loan officer's individual license and individual liability are in the frame.

Model response (two sentences):

"I appreciate it, and I can't take it — if you're paying for space that features me, that's value moving from you to me, and RESPA doesn't care that neither of us meant anything by it. Let me pay my measured share of it instead; send me the creative and I'll work out what percentage is actually mine and get it approved this week."

Grade on: declining without moralizing; naming the statute without lecturing; offering a compliant alternative in the same breath, so the relationship survives; and escalating to approval rather than deciding alone.

Exercise 7.39

Considerations that apply, from §7.8: (i) it is an advertisement, subject to review, retention, and disclosure requirements; (ii) NMLS unique identifier must appear (§3.8); (iii) Regulation Z triggering terms — stating a rate, payment, term, down payment, or number of payments triggers additional disclosures, and most employers therefore prohibit it outright; (iv) no prohibited or comparative claims; (v) record retention — it must go into the employer's archive and deleting it later does not un-publish it; (vi) housing advertising is a restricted category on major platforms, so no demographic or tight geographic targeting; (vii) employer policy and state rules govern, and both may be stricter.

Model caption:

Rates moved about an eighth this week. On a typical purchase in this market that's roughly the price of a streaming subscription each month — it is not a reason to rush and it is not a reason to wait. If you want to know what it actually does to your numbers, send me your scenario and I'll run it. [Name], [Company], NMLS #[ID]. Equal Housing Lender.

Grade on: no rate, no payment figure, no term, no comparative claim, NMLS ID present, and a call to action that moves the conversation off the platform and into a system.

Exercise 7.41

(a), (b), and (c).

Section 8(a) prohibits both giving and accepting a thing of value pursuant to an agreement or understanding for the referral of settlement service business — so (a) and (b) are both prohibited, and students who select only (a) have the chapter's most common blind spot. Section 8(b) prohibits splitting or accepting a portion of a charge for a settlement service where no service was actually performed — (c).

(d) is not prohibited. Payments by an employer to its own employees are an express Section 8(c) exception.

Exercise 7.43

(b) The originator's NMLS unique identifier. This is a S.A.F.E. Act requirement, covered in §3.8, and it does not stop applying because the medium is a social profile, a text message, or a flyer.

Exercise 7.45 †

(a) Annual value using the student's own plan. The method, not the number, is what is graded. A worked example at 115 basis points on a \$300,000 average loan:

$$0.0115 \times \$300{,}000 = \$3{,}450.00 \text{ per closed loan}$$ $$5 \times \$3{,}450.00 = \mathbf{\$17{,}250.00 \text{ per year, direct}}$$

Adding the chapter's second-order assumption — each closed borrower produces 0.4 referrals over three years, half of which close, so $5 \times 0.4 \times 0.5 = 1.0$ additional closing a year:

$$\$17{,}250.00 + \$3{,}450.00 = \mathbf{\$20{,}700.00 \text{ per year, all in}}$$

Require the student to state their plan's basis points and average loan amount explicitly. A student who cannot state their own compensation terms has found the exercise's real lesson.

(b) Share of a 24-loan year.

$$5 \div 24 = \mathbf{20.8\%}$$

(c) The plan for partner number six. Grade on specificity, not ambition. A satisfactory answer names a profile (an agent closing buyer-side transactions in the same price band, at a brokerage where the incumbent lender is known to be slow), a first action with a verb and an artifact ("send the §7.3 message Tuesday morning, then attend her Thursday office meeting"), and a date. An unsatisfactory answer says "network more." The chapter's arithmetic is the standard: 54 hours per producing partner, twelve to eighteen months to a first dollar, and the work has to start before it is needed.

Exercise 7.47

Cost difference. A purchased lead at the chapter's constructed \$40 price and 1.5% conversion costs \$2,666.67 per closed loan. The Linden Street agent relationship delivers closings at \$132.28.

$$\$2{,}666.67 - \$132.28 = \mathbf{\$2{,}534.39 \text{ more per closed loan}}$$ $$\$2{,}666.67 \div \$132.28 = \mathbf{20.2\times}$$

Three things about the transaction that would likely have been different (accept any three well-reasoned; these are the strongest):

1. The pre-approval would not have carried the same weight with the listing agent. The letter's credibility on this file rests partly on four prior closings between the buyer's agent and the loan officer. A lead-sourced borrower arrives with no such history.
2. The 8:40 call would not have happened at all. The borrowers would have contacted a lender when they decided to, not when a professional who knew the deadline decided for them — and the 2:00 p.m. requirement would have surfaced late or been missed.
3. The borrowers would almost certainly have been shopping several lenders simultaneously, because that is what a shared lead is, which changes every conversation about rate, structure, and documentation from advisory to competitive.
4. (Also creditable) The six-day slip from day 45 to day 51 would have been survivable in a different way. With an agent who has closed four prior files with you, that slip is a conversation; with a lead-sourced borrower and an agent who has never met you, it is evidence.

Chapter 8

Worked solutions to the daggered (†) and odd-numbered exercises. All figures are constructed for teaching.

Exercise 8.1

Affordability is whether a household can actually carry a housing payment alongside everything else it must pay, with margin left for the ordinary emergencies of owning a building. It is not a ratio, it has no threshold, and no guideline defines it.

Purchasing power is the maximum loan amount — and therefore maximum purchase price — that a borrower's documented income and documented debts will support under a given program's ratio limits at a given rate.

An automated underwriting system reports on purchasing power only. Affordability appears nowhere in a loan file, is computed by no system in the transaction, and is the question the borrower is actually asking. That asymmetry is the reason §8.2 exists.

Exercise 8.3

Three respects, each with the document (or its absence) that creates the difference:

1. Credit. A pre-approval rests on a tri-merge credit report the lender pulled, with a representative score known and dated. A pre-qualification often has no credit report at all — in which case the score, one of the four facts that price a loan, is a guess and so is the payment.
2. Income and assets. A pre-approval rests on pay statements, Forms W-2, and account statements the lender has received and read. A pre-qualification rests on what the borrower said. The absence of the document is the difference.
3. Automated underwriting. A pre-approval has been submitted to an automated underwriting system and the findings are retained in the file. A pre-qualification has not been submitted to anything.

Credit for also noting: neither term has a legal definition, so the heading on the letterhead proves nothing — the recitals in the body are the only evidence of which one you are holding.

Exercise 8.5

The six items are:

1. the consumer's name
2. the consumer's monthly income
3. the consumer's Social Security number (to obtain a credit report)
4. the property address
5. an estimate of the value of the property
6. the loan amount sought

The item most often absent during a pre-approval is the property address, because in the ordinary case the borrower has not chosen a house yet.

Operationally: until all six have been submitted, the Loan Estimate timing requirement has not been triggered. This is why many lenders' written procedures require a pre-approval letter to be issued without naming a subject property. It is not a loophole — it is the structure of the rule, and following your company's procedure without understanding it is how originators get surprised. Chapter 22 works the timing rules. Verify current requirements with compliance.

Exercise 8.7

A budget-first conversation is one in which the borrower's target monthly payment is established before any purchase price is discussed, and the price is then derived from the payment — reversing the industry's default order, in which a price is named first and the payment arrives later as a surprise.

The opening question, word for word:

"What number, coming out of your checking account on the first of every month, would not scare you?"

"What's your budget?" is not that question for three reasons. A budget is a price, and a price is an abstraction the borrower cannot feel. It invites a number the borrower assembled in the car on the way over, usually anchored to whatever a listing site told them. And it produces a single figure with no component structure, so nothing about it can be checked.

The correct technique after asking is to wait. The first number people say is usually the number they think you want. The second one is real.

Exercise 8.9

Five follow-up questions, each with its stated reason:

1. "Is that salary, hourly, or something else?" — "Because an underwriter counts a salary and an hourly wage differently, and I want to use the version that's actually documentable."
2. "If hourly: what's your rate, and how many hours are you guaranteed?" — "Forty hours is an assumption until somebody writes it down; I need the base I can stand behind."
3. "Is any of that overtime, shift differential, bonus, or commission — and if so, how much, for each of the last two years separately?" — "Variable pay gets averaged, and I need two years to average, plus I need to see which direction it's moving."
4. "How long have you been doing this, and how long at this employer?" — "They're looking for income that's likely to continue, and time in the line of work is most of that argument."
5. "Is there any income that doesn't come from an employer — child support, a pension, disability, a second job, rental income?" — "Some of it counts and some of it doesn't, and I'd rather find out today than on day thirty."

Full credit requires the reasons as well as the questions. The point of §8.3 is that borrowers answer better when they know why — and that the reason is always the same underlying fact: the person deciding has never met them and will only ever see documents.

Exercise 8.11

The answer to give a borrower:

"They average two years of commission because two data points are what they have. Yours went from \$19,800 to \$23,400 — it went up. That's why they'll let me use the average, which is \$43,200 ÷ 24 = \$1,800 a month. If it had gone the other way, averaging would mean using a number higher than what you're currently earning to qualify you for a thirty-year obligation, and they won't do that. On a declining trend the general rule is that the lower, more recent figure is used instead — which is a much bigger haircut."

Two things the answer must include to be complete:

• The direction of conservatism runs one way. Rising income is averaged down; falling income is taken at the low point. Both choices protect the same party.
• Nothing about this is a judgment on the borrower. It is a rule about what two data points can prove. Say that out loud; borrowers hear "your income doesn't count" as an accusation unless you head it off.

Cross-reference: the Fulton Avenue file (Chapters 11 and 32) is the same principle applied to self-employment, where a 2.3% year-over-year decline forces the lower figure. Verify the current treatment in the applicable guide.

Exercise 8.13

The reason: under Regulation Z's integrated disclosure rules, an application is the submission of six items, one of which is the property address. Once a creditor holds all six, the Loan Estimate timing requirement runs. A pre-approval issued before a house has been chosen is normally missing that one item — so a procedure that forbids naming a subject property keeps the pre-approval process cleanly on one side of a bright line, rather than depending on individual originators to notice when they have crossed it.

The rule: Regulation Z, implementing the Truth in Lending Act, as amended by the TILA-RESPA Integrated Disclosure rule. Chapter 22.

The practical drawback: listing agents like to see the address of their property on the letter, and a letter without one reads, to some agents, as less specific and therefore less serious. The answer is not to break the procedure. It is to make the recitals strong enough that the letter is obviously better than the alternative, and to call the listing agent yourself if the offer is competitive. A thirty-second phone call outperforms a line of text.

Credit for also noting the side benefit: a letter without a subject property and written for the offer amount discloses nothing about the buyer's ceiling.

Exercise 8.15

Model answer:

"It's true, and that's what makes it dangerous. The borrowers hear a permission slip and the agent hears a search parameter. By that afternoon the saved search has been updated and every house at \$385,000 looks like the compromise house. They write at \$455,000 in a competitive market, win, and spend thirty years at the top of a ratio with about \$349 a month of margin — until the water heater, which is not an emergency, it's a Tuesday. Nobody lied. Nobody violated anything. The file closed, the commission was paid, and the household is fragile, and it all happened because a number was said out loud in the wrong order."

The sentence to say instead:

"You told me \$2,700 a month was the number that wouldn't scare you. At today's rate, with taxes and insurance and mortgage insurance included, that lands you right around \$340,000. The file supports more than that if you need it, but let's start from your number, not from the ceiling."

The graded elements: state their payment first, derive the price from it, and mention the ceiling only as a ceiling.

Exercise 8.17

(a) $\$2{,}166.00 \div \$1{,}425.00 = 1.52$ — the proposed payment is 1.52 times current rent.

(b) An increase of 52.0%.

(c) $\$2{,}166.00 - \$1{,}425.00 = \$741.00$ per month. $\$741.00 \times 12 = \$8{,}892.00$ per year.

(d) $\$741.00 \times 4 = \$2{,}964.00$ accumulated over four months.

(e) Beyond testing whether the number is livable, the \$2,964.00 is reserves — verified liquid assets remaining after closing, which underwriters treat as a compensating factor and which are stated in months of PITI. The practice payment is the only technique in this chapter that improves the file and tests the household with the same dollar.

Exercise 8.19

(a) Housing ratio and back-end ratio.

$$\text{housing} = \frac{\$2{,}520.00}{\$8{,}400.00} = 30.00\%$$

$$\text{back-end} = \frac{\$2{,}520.00 + \$760.00}{\$8{,}400.00} = \frac{\$3{,}280.00}{\$8{,}400.00} = 39.05\%$$

Both comfortable. On the qualifying question, this file is not close to a limit.

(b) What is left.

$$\$6{,}300.00 - \$2{,}520.00 - \$760.00 = \$3{,}020.00$$ $$\$3{,}020.00 - \$2{,}180.00 = \mathbf{\$840.00}$$

(c) After maintenance.

$$\$310{,}000 \times 1\% = \$3{,}100 \text{ per year} \div 12 = \$258.33 \text{ per month}$$ $$\$840.00 - \$258.33 = \mathbf{\$581.67}$$

(d) Two sentences. Model answer:

"On the qualifying side you're at 30% and 39%, which is comfortable — the loan isn't the constraint here. On your side, after the payment, the debts, everything you told me you spend, and setting aside a percent a year for the house itself, you're at about \$582 a month, and that's the number I'd actually want you looking at."

Note what the model answer does not do: it does not tell them whether \$581.67 is enough. That is not the loan officer's call, and saying so is part of the answer.

Exercise 8.21

(a) Each component of the \$3,033.72 payment:

$\$2{,}341.94 \div \$3{,}033.72 = 77.20\%$ · $\$385.00 \div \$3{,}033.72 = 12.69\%$ · $\$130.00 \div \$3{,}033.72 = 4.28\%$ · $\$176.78 \div \$3{,}033.72 = 5.83\%$. Sum: $77.20 + 12.69 + 4.28 + 5.83 = 100.00\%$ ✓

(b) The escrow portion — \$515.00, or 16.97% — changes without anyone in the transaction deciding to change it. The mechanism is the annual escrow analysis: the servicer recalculates the monthly escrow deposit against the actual tax and insurance bills. Property tax reassessment is governed by state and local law and in many jurisdictions a sale is a reassessment event; homeowners insurance premiums are reset at renewal. Chapter 23 covers escrow analysis. The mortgage insurance component follows the program's own termination rules; the principal and interest — \$2,341.94 — is fixed for all 360 months.

(c) Model sentence:

"About seventeen percent of this payment is money the lender collects and passes through for your taxes and insurance, and both of those get recalculated once a year — so unlike your rent, this number can move a little even though the loan itself never does."

Exercise 8.23

(a) $\$145{,}000 - \$126{,}000 = \mathbf{\$19{,}000}$ per year, or $\$19{,}000 \div 12 = \mathbf{\$1{,}583.33}$ per month.

(b) At this file's back-end ratio of 42.66%:

$$\$1{,}583.33 \times 0.4266 = \mathbf{\$675.45} \text{ per month of additional obligation}$$

(c) At \$6.403117 of principal and interest per \$1,000 borrowed:

$$\frac{\$675.45}{\$6.403117} \times \$1{,}000 = \mathbf{\$105{,}487} \text{ of additional loan}$$

For scale: \$100,000 of loan at this rate costs \$640.31 a month.

(d) Model answer:

"Chapter 1 described a letter that had to be reduced by \$40,000 after verification, because the borrower quoted a gross figure that included a bonus with no two-year history. This file is the same mechanism at twice the size, and nobody lied at any point — both borrowers accurately described what landed in their accounts last year. A letter written on the household's own honest answer would have overstated their purchasing power by roughly a hundred thousand dollars of loan, and the correction would have arrived around day twenty, after the offer was accepted and the earnest money was deposited."

Exercise 8.25

The letter should contain, at minimum: issue date; a specific expiration date; both applicants; occupancy (primary residence); program and term (conventional 30-year fixed); down payment (10%); purchase price supported (\$268,000) and the corresponding loan amount; a WHAT WE REVIEWED block naming the credit report and its date, the two pay statements and two Forms W-2 per applicant, the single bank statement, and the automated underwriting submission; a WHAT THIS LETTER IS NOT block stating that it is not a commitment to lend, not a rate lock, not a guarantee, not a pre-qualification, and not a verification; a CONDITIONS block; and the originator's and company's identifying information.

The corresponding loan amount: $\$268{,}000 \times 0.90 = \$241{,}200$.

The audit. Sentences a careful auditor deletes from a typical first draft:

A first draft that contains all seven deletes seven sentences. The exercise is graded on whether the student can name the document behind each surviving sentence, not on the letter's prose.

Exercise 8.27

Model answer, four sentences, in order:

"I've got everything back and I want to give you the real number first: the price that works is two fifteen, not two forty. [the fact, first sentence] At two forty the payment and the ratios don't land anywhere a lender will go, and at two fifteen they do. [the number] That is a ceiling, not a no — you are approvable today at two fifteen with the assistance program. [what it is not] Send me the payoff statement on the car this afternoon and I'll have two versions of this for you before noon tomorrow. [next step with a time]"

Graded elements: the fact is in the first sentence with no preamble; it is stated as a number rather than a category; the boundary is stated before the borrower has to ask; and the next step has a verb and a deadline. An apology longer than five words loses credit — heavy apologizing signals that the news is worse than it is.

Exercise 8.29

Model plan:

WHERE YOU ARE TODAY
  Savings $4,200 against about $9,500 needed for this price range.
  Back-end ratio 47% at the price you're looking at.
  One unresolved collection from 18 months ago.

WHAT HAS TO CHANGE
  The collection needs to be resolved or documented.
  Savings needs to reach roughly $9,500.
  The ratio needs room -- either less debt or a lower price.

THE THREE THINGS
  1. Request the collection agency's written validation of the account.
  2. Set an automatic transfer of $450 on the 1st and the 15th.
  3. Send me the payoff figures on both installment loans so we can price
     which one buys the most ratio.

THE DATE
  I will call you on [specific date, roughly 90 days out]. It is in my
  calendar and it is in the email I am sending you today.

The sentence you may not include: anything of the form "this should get your score to 640." You may not promise a credit score outcome — not a number and not a timeline. You may explain how utilization reporting works and how derogatory items age; you may not assert a result. You must also never refer a borrower to an operation that charges advance fees to dispute accurate information.

Exercise 8.31

(a) Number, reason, time:

"I can write \$385,000 right now — that one I can back up completely. I can't write \$430,000 tonight, because the second borrower's income is a third commission and I haven't seen the statements. Have them send me the last two years' W-2s and the most recent pay statement, and I'll tell you by nine tonight what the real ceiling is."

(b) The second answer, after "it's just a formality":

"It isn't, and that's actually the reason I'm useful to you. That letter is the thing the listing agent's client relies on when they take the house off the market. If I write a number I can't support and it comes back \$45,000 lower after verification, your client is the one holding a contract they can't perform on. I'm not being difficult about the letter — I'm being careful about your client's earnest money."

(c) What is at risk for her client: if the financing contingency has expired by the time the verified number arrives, the earnest money is exposed. Also lost: the appraisal fee and inspection fee already spent, the house itself, and any rate advantage, since the borrower restarts in whatever market exists that week. Chapter 20 covers what a financing contingency actually protects.

(d) If she takes the file elsewhere, you send the borrowers' documents to nobody, you note the outcome in the file, and you keep the relationship civil. That is not automatically the wrong outcome: you cannot be the originator who says yes to everything and the originator whose letters are worth something, and agents sort themselves accordingly over about eighteen months. The book's fourth theme — the relationship outlasts the transaction — is not a consolation prize here; it is the actual business case. Chapter 7 covers referral partners and Chapter 38 the arithmetic.

Exercise 8.33

(a) Model answer:

"That's a completely reasonable reaction to that number, and I want to be clear that I'm not trying to talk you out of it. Here's where I do want to push a little: what alarmed you was the residual figure at this price. There's a version of this at a lower price where that number looks different, and it might be worth twenty minutes to see it before you decide to wait a year. And if you still want to wait after that, we'll build you a plan and I'll call you in six months."

(b) What you may not say. You may not tell them not to apply, and you may not handle the inquiry in a way designed to make them go away. If they want to apply after hearing the arithmetic, you take the application. The boundary is legal, not merely ethical: the Equal Credit Opportunity Act and Regulation B address the discouragement of applications, and a good-faith motive ("I was saving them the hard inquiry") is not a defense to a pattern. The formal decline machinery — including adverse action notice requirements — exists precisely so that a denial is documented and reviewable rather than made informally on a phone call. Chapter 25; verify your company's procedure with compliance.

(c) The file gets a factual note: what was computed, what was said, what the borrower decided, and the date. No characterization of the borrower. The follow-up system gets the four-part plan and a specific call-back date — not "check in sometime," a date. Chapter 7 covers the CRM; §8.9 covers what belongs in the note.

Exercise 8.35

B. A pre-approval is based on documentation the lender has obtained and reviewed, including a credit report.

• A is wrong: a pre-approval commits the lender to nothing, which is why the letter says so.
• C is wrong: locking is a separate act. On the Linden Street file the letter is day 1 and the lock is day 12.
• D is wrong: no federal law requires a pre-approval before an offer. It is a market convention imposed by listing agents and sellers, which is why the 8:40 call in Chapter 1 exists at all.

Exercise 8.37

C. A statement of the purchasing power the lender's documentation supports, subject to stated conditions.

• A is the most common misconception, held by borrowers and by a surprising number of agents.
• B confuses the letter with a lock agreement.
• D is wrong: the letter is not a required disclosure under RESPA or anything else. It has no prescribed form at all — which is the point of §8.6, and the reason its discipline has to be voluntary.

Exercise 8.39

The DO NOT HAVE line, at minimum, contains all of these as of 1:40 p.m. on day 1:

1. no third-party verification of employment for either borrower
2. no verification of deposit
3. no tax transcripts
4. no gift letter and no documentation of the donor's ability to give
5. no confirmation that Borrower 1's shift differential and overtime will continue
6. no confirmation that Borrower 2's commission structure is unchanged
7. no underwriter review of the income calculation
8. no appraisal, no property review, no title report
9. no rate lock
10. no program decision — conventional and FHA are both live

Which item would do the most damage to the letter? The best answer is the income calculation surviving verification (items 1, 5, 6, and 7 together) — because the letter's central assertion is a purchase price, the purchase price rests on \$10,500.00 of qualifying income, and roughly 22.7% of that income is variable: \$580.00 of differential and overtime plus \$1,800.00 of commission is \$2,380.00, and $\$2{,}380.00 \div \$10{,}500.00 = 22.67\%$. Nothing else on the list moves the supported price as far.

A defensible alternative answer is the appraisal, on the grounds that it can invalidate the structure entirely — but note the difference: the appraisal is a risk the letter discloses as a condition, while the income calculation is the thing the letter asserts. Full credit for distinguishing the two.

Chapter 9

Worked solutions to the daggered (†) and odd-numbered exercises. Figures follow the frozen Linden Street values: income \$10,500.00/month, debts \$1,446.00/month, PITI plus mortgage insurance \$3,033.72, total obligations \$4,479.72, housing ratio 28.89%, back-end 42.66%.

Exercise 9.1

The six items, with what each one is:

Three about the borrower, two about the house, one about the deal. Mnemonic: name, income, number, address, value, amount.

Note what is not on the list: a form, a signature, a fee, a purchase contract, an appraisal, a credit report, employment verification, or a date of birth. Note also that item 5 requires only an estimate — a list price or the borrower's own guess satisfies it.

Exercise 9.3

The seventh item was "any other information deemed necessary by the loan originator."

It allowed a creditor to postpone the existence of an application indefinitely by declaring that it also needed something else — a contract, a paystub, a completed form. Because the creditor defined the item, the creditor controlled the trigger, and therefore controlled when its own disclosure deadline began.

The TILA-RESPA integrated disclosure rule removed it for covered transactions. A creditor may still require anything it likes before it will underwrite, quote, or commit. It may not require anything beyond the six before an application exists.

Exercise 9.5

Components:

• Borrower Information — one complete set per borrower, signed by that borrower. Sections 1 through 9.
• Additional Borrower — the same content for each additional borrower.
• Lender Loan Information — completed by the lender, not signed by the borrower. Property and loan detail, title information, mortgage terms, and the qualifying arithmetic (minimum required funds or cash back).
• Continuation Sheet and, where state law makes it relevant, the Unmarried Addendum.

The lender-completed component is the answer to the second half of the question. Practical consequence: a borrower who reports "an error on the 1003" has frequently found the lender's own arithmetic rather than a misstatement of their own, and the fix belongs to the lender.

Exercise 9.7

Application date — the date the creditor received the sixth of the six items and an application therefore existed; the date the disclosure deadlines are counted from, and not necessarily the date printed on a signed form.

Intent to proceed — the consumer's affirmative communication, after receiving the Loan Estimate, that they wish to continue; silence does not constitute it, and until it is given the creditor generally may not impose fees other than a bona fide and reasonable credit report fee.

E-consent — the borrower's consent under the ESIGN Act to receive required disclosures electronically, valid only if given in a manner reasonably demonstrating the borrower can access the documents in the form in which they will be delivered.

Borrower authorization — the borrower's signed permission for the lender to verify employment, income, assets, and credit and to obtain tax transcripts; a lender and investor requirement rather than a federal disclosure, and needed from each borrower individually.

Exercise 9.9

(a) Items held: name (1), Social Security number (3), income (2) — the paystub establishes \$7,200 a month. Three of six.

Missing: the property address (4), an estimate of the value of the property (5), and the mortgage loan amount sought (6). "Looking in the \$400,000 range" is a shopping range, not a value estimate for an identified property and not a loan amount.

(b) No application exists.

(c) Nothing is due. But note the position you are in: you hold all three consumer-side items, and the remaining three all arrive together the moment this borrower identifies a house. Behave as though you are one email away, because you are — see 9.10.

Exercise 9.11

(a) Items held: both borrowers' names (1), both incomes (2), the property address (4), and the list price as an estimate of value (5). Four of six.

Missing: the Social Security numbers (3) and the mortgage loan amount sought (6).

(b) No application exists.

(c) Nothing is due yet. Two observations. First, you have never spoken to these people and you already hold four of the six items, supplied by a third party — which is the whole point of §9.1. Second, the two missing items are the two you will collect in the first ninety seconds of the first call, one of them (the Social Security number) because you cannot pull credit without it. Set up the file assuming the application will exist by the end of that call.

Exercise 9.12

All six items are on the form and have been submitted to the creditor. An application exists.

A signature is not one of the six items and is not part of the definition. The Loan Estimate is due no later than the third business day after receipt.

The refusal to sign is a real and separate problem — the file cannot be underwritten, sold, or delivered without executed documents, and the acknowledgments in Section 6 are the borrower's certification that the contents are true. It is also information: a person who completes every field and then declines to sign is telling you something, and the right response is a question ("what part of this are you not comfortable with?"), not pressure.

But the refusal does not un-receive the information, and it does not stop the clock. Disclose.

Exercise 9.13

Yes. The creditor has received an application.

To the colleague who wrote the policy: the definition of "application" for these transactions is the six items, full stop. The older definition included a seventh — "any other information deemed necessary by the loan originator" — and the integrated disclosure rule deleted it precisely because it let creditors control their own deadline. A company policy is not a source of law; it is a source of exposure when it conflicts with one.

The policy is fixable without losing anything the company actually wants. Keep the contract requirement as a workflow rule: no appraisal ordered, no submission to underwriting, no commitment of any kind without the contract. Separate it from the disclosure rule: the Loan Estimate goes out on the six items, every time, regardless of what else is missing.

Exercise 9.15

An estimate of the value of the property is required, but it does not have to be accurate, informed, or supported. The list price, the contract price, or the borrower's own guess all satisfy item 5. "Whatever it appraises for" is not an estimate, so on those facts you hold five items and no application exists.

What you say next: "What's it listed at?" — and the answer is an estimate.

Which is the honest point of this exercise. The gap between five items and six is one ordinary question, and treating that gap as a compliance strategy is both fragile and beneath the job. Use it the other way around: know that you are one question from an application so that you disclose on purpose rather than discover it in an audit.

Exercise 9.16

The common structure: replace a total with components, and replace a category with a concrete test. "Do you own property" is a category and people self-classify wrongly; "is your name on a deed" is a test.

Exercise 9.17

• Start date, seven months ago → Section 1b, the current employment block.
• Nine years as a physical therapist → Section 1b's years in the line of work field.
• The previous clinic → Section 1d, previous employment, which is what covers the two-year history the current employer cannot.

Years in the line of work is the field that rescues the file. Seven months at one employer looks thin in isolation. Nine years doing the same work, with documented prior employment covering the gap, answers the underwriter's actual question — which is never "how loyal is this person" but "is this income likely to continue." Chapter 11 works the calculation; your job on day 5 is to make sure both facts are captured, because if only the seven months is on the form, only the seven months gets evaluated.

Exercise 9.19

Across all five: the borrower is not confessing. They are volunteering something that had not occurred to them was relevant, which is exactly the outcome a slow, one-at-a-time reading of Section 5 produces on day 5 rather than on day 40.

Exercise 9.21

Two questions, in this order:

1. "Where will you be the rest of the time — do you own or rent somewhere else?"
2. "Is anyone else going to be living there, and is anyone going to be paying you rent?"

And if the answers are ambiguous, a third: "How far is it from where you work?"

What you are doing is classifying, not coaching. Occupancy is a fact with three possible values — primary residence, second home, or investment property — and each carries different pricing, a different down payment, and different eligibility. You are establishing which one is true so that the file is built correctly on day 5, when it is free.

What you must not do is explain the pricing consequences first and then ask. That sequence tells the borrower which answer is cheaper before they answer, and a loan officer who does it habitually is building an occupancy-misrepresentation problem into their own pipeline. If the answer is "second home," say so plainly and restructure the file. Chapter 27 covers the detection side.

Exercise 9.22

DAY 5 REQUEST — teacher + self-employed electrician    [worked solution; illustrative]

  BORROWER 1 — high school teacher (salaried) + summer camp job
    [ ] Paystubs covering the last 30 days, teaching position
    [ ] W-2s for the last two years — BOTH employers
    [ ] Most recent paystub or 1099 from the camp, if you have one
        (why: a second job is usable income, but only with a two-year
         history — the W-2s are what make it count)
    [ ] Your teaching contract or employment letter for the coming year,
        if the school issues one

  BORROWER 2 — self-employed electrician, single-member business
    [ ] Personal federal tax returns, last 2 years, ALL pages and schedules
    [ ] Business returns, last 2 years, if the business files separately
    [ ] Year-to-date profit and loss statement and balance sheet
    [ ] Business license, or a letter from your CPA confirming the business
        and how long it has operated
        (why: your income is computed from the returns, not from what the
         business takes in — see Chapter 32. This is the longest lead-time
         item on the list; start here.)

  BOTH
    [ ] Photo ID, front and back, unexpired
    [ ] Last two statements, ALL PAGES, every account we're using to close
        (why: page 3 of 5 that says "intentionally left blank" is still
         page 3 of 5; the underwriter counts pages)
    [ ] Fully executed contract with every addendum and every signature
    [ ] Earnest money check image + the statement page showing it clear
    [ ] Landlord's name, address, and phone
    [ ] Homeowners insurance agent's name and number

  DUE: Friday — five days. If it's all in by Friday, the appraisal is
  ordered Monday and we are ahead of the contract instead of chasing it.
  If the tax returns are the holdup, send everything else and tell me;
  I'd rather start on the rest.

Two things the solution is graded on. The self-employment documents are called out as the long-lead item, because they are, and a borrower who does not know that will send them last. The deadline has a stated consequence and an escape hatch that keeps the rest of the list moving.

Exercise 9.24

Problems, and what to compare each against:

Aggregate effect of the omissions alone: \$1,240 + \$362 = \$1,602 a month of undisclosed obligation, plus an undisclosed property. On any file, that is the difference between an approval and a decline, and none of it required asking the borrower a single question — items 2, 3, 4, 5, and 10 are all visible by reading the application against a credit report you already have.

Exercise 9.25

Omitting Borrower 2's auto payment of \$429.00:

• Monthly debts: \$1,446.00 − \$429.00 = \$1,017.00
• Total obligations: \$3,033.72 + \$1,017.00 = \$4,050.72
• Housing ratio: unchanged at \$3,033.72 ÷ \$10,500.00 = 28.89% (the housing ratio has no non-housing debt in it)
• Back-end: \$4,050.72 ÷ \$10,500.00 = 38.58%

Correct back-end is 42.66%. The omission understates the ratio by 4.08 percentage points.

The guideline concept is the exclusion of installment debt with a small number of payments remaining. Conventional guidelines do permit excluding an installment obligation when very few payments remain — the common threshold is ten months or fewer, subject to conditions, and Chapter 14 states it properly. Nineteen payments remaining does not qualify, and no amount of "it's almost paid off" makes it qualify. The credit report will disclose the remaining term, so the omission is not merely wrong; it is wrong in a way that is guaranteed to be discovered.

Exercise 9.27

THE DAY 5 RECONCILIATION — worked checklist            [worked solution]

  #   1003 FIELD                  COMPARE TO              LOOKING FOR
  ──────────────────────────────────────────────────────────────────────────
   1  Legal name, exact spelling  Photo ID; credit        Middle initial, suffix,
                                  report header           maiden vs. married name
   2  Social Security number      Credit report header    Transposed digits
   3  Date of birth               Photo ID                Typo; wrong century
   4  Current address + months    Credit report           An address on the report
                                  address history         that isn't on the 1003
   5  Prior address              Credit report            Blank when months at
                                                          current is under 24
   6  Housing: rent/own/none      Credit report           A mortgage tradeline on
                                                          a borrower who "rents"
   7  Employer legal name         Paystub header          A d/b/a vs. the entity;
                                                          a staffing agency
   8  Employment start date       Paystub YTD             YTD that can't be earned
                                                          in the elapsed time
   9  Base income                 Paystub rate x hours    Base that doesn't
                                                          annualize
  10  Variable income by type     Paystub YTD; W-2s       OT booked as base; an
                                                          average with no history
  11  Each monthly debt           Credit report           Omitted installment;
                                                          changed payment; a debt
                                                          with >10 payments left
                                                          treated as excludable
  12  Real Estate Owned           Credit report           Mortgage tradelines with
                                                          no property in Section 3
  13  Subject property address    Purchase contract       Unit numbers, directional
                                                          prefixes, digits, spelling
  14  Purchase price              Contract + ALL addenda  A price amended in an
                                                          addendum nobody read
  15  Seller credits              Contract + addenda      A credit on the contract
                                                          that isn't on the 1003
  16  Loan amount                 Price minus down pmt    An amount that doesn't
                                                          foot
  17  Occupancy                   Contract; commute       A "primary" 90 minutes
                                                          from work
  18  Property type / units       Contract                Condo called a townhome
  19  Declarations vs. everything Credit report;          A "no" contradicted by a
                                  Section 3               tradeline or a property
  20  Gift disclosed              Section 4d              A gift in the notes and
                                                          not on the form

Twenty lines, twenty minutes, once. Note how many of them require no contact with the borrower at all — items 1 through 12 and 19 are done with documents already in your possession.

Exercise 9.29

The rule. Regulation B, implementing the Equal Credit Opportunity Act, prohibits a creditor from making any oral or written statement to applicants or prospective applicants that would discourage, on a prohibited basis, a reasonable person from making or pursuing an application.

Why good intentions do not help. Three reasons, and they compound.

First, discouragement is assessed by what was said and what happened, not by what was meant. No regulator, and no plaintiff, will ever be able to see the loan officer's motive; they will see a prospective applicant who was told not to bother and who did not apply.

Second, it is a pattern problem. One well-meant discouragement is invisible. A career of them, applied to the borrowers a loan officer intuitively reads as marginal, produces a distribution of applications that does not match the distribution of people who walked in the door — and aggregated across a market, well-meant discouragement is indistinguishable from redlining. That is precisely what the rule exists to prevent.

Third, and most concretely for the borrower: the loan officer substituted a guess for an underwriting decision. A borrower who applies and is declined receives an adverse action notice with the specific principal reasons. Those reasons are the only actionable thing in the entire transaction — they tell the borrower what has to change and roughly how long it will take. A borrower who is talked out of applying receives nothing, learns nothing, and waits a year on the strength of a stranger's hunch that may have been wrong.

What should have happened. Take the application. Be candid about the odds — candor about likelihood is not discouragement, and "here is what I think will happen and here is why" is exactly the conversation Chapter 8 is about. Run it honestly. If it is declined, deliver the notice, call before the letter arrives, and give the borrower the three specific things that have to change.

Exercise 9.30

To: Branch Manager Re: File — 18 days open, borrower non-responsive

I'm not going to mark this withdrawn, and I want to put the reason in writing so we're aligned.

Withdrawn means the applicant withdrew. This borrower hasn't. If I code it that way, three things happen, and none of them are good for us. We misclassify what our institution reports. We deny the applicant a notice they're entitled to, along with the specific reasons — which is the only part of a decline that's actually useful to them. And we create a record that says the customer walked away when in fact we did, which is exactly the kind of thing that looks worst when someone reads it back to us two years from now.

The correct options are two. We can deny the application and send an adverse action notice with the specific principal reasons within the required time. Or — better here, because we don't actually have grounds for a credit decision — we can send a written notice of incompleteness naming exactly what's missing, giving a reasonable period to supply it, and stating that if we don't hear back we'll take no further action. If they don't respond, the file closes for incompleteness, which is a real and reportable outcome and is the accurate one.

What I'm doing today. Compliance is drafting the incompleteness notice with the three outstanding items named specifically. Before it goes, I'm making one more attempt by phone and one by text with a single item in it, because in my experience a borrower who has gone quiet is usually stuck on one document and embarrassed about it rather than gone. I'll document both attempts in the file either way.

If you want the pipeline number to look different, I'd rather fix it by closing the file correctly than by coding it wrong.

The point of the exercise is that the manager's instruction is almost certainly not malicious — it is pipeline hygiene. The answer is not to accuse anyone; it is to name the three specific consequences and offer the correct alternative in the same message.

Exercise 9.31

"Hi — it's me. I wanted you to hear this from me before a letter shows up, because a letter is going to arrive in the next few days and it's going to sound worse than this conversation.

We couldn't get it approved. The reason is the debt ratio — with the car, the student loans, and the payment on that house, you land at about fifty-one percent, and the program tops out at forty-five. That's it. It isn't your credit, it isn't your job, and it isn't anything you did wrong.

Here's the honest version of what that means. This isn't a five-year problem. The car has eleven payments left. When it's gone, you're at roughly forty-three percent on the same income and the same house — and that's the whole difference. So this is a next-year conversation, not a someday conversation.

The letter will list the reasons formally; keep it. I'm going to email you the three things, written down, tonight. And if you want, I'll call you in nine months and we'll look at it again — no obligation, and I'll tell you honestly if it still doesn't work."

Graded on four elements: the reason in plain numbers, whether it is changeable and roughly when, explicit removal of blame, and a specific next action with a date. Note that the call comes before the notice, not instead of it — the notice is a legal requirement and goes out regardless.

Exercise 9.33

C — a bona fide and reasonable fee to obtain the consumer's credit report.

It is the only listed exception to the rule that a creditor may not impose fees in connection with an application before the consumer has received the Loan Estimate and indicated an intent to proceed. A is wrong (an application fee is exactly what the rule blocks), B is wrong (the appraisal deposit is the most common real-world violation of this rule), and D is wrong on its face — there is no lock to extend at application.

Exercise 9.35

Why intent is irrelevant. The rule's trigger is a receipt event: has the creditor received these six items. Receipt is an objective, reconstructable fact — it appears in emails, texts, call logs, system audit trails, and the memories of third parties like real estate agents. Intent is neither objective nor reconstructable, and a rule that turned on it would be unenforceable, because every late disclosure would be defended by an assertion nobody could test. Making the trigger objective is what makes the deadline real.

There is also a consumer-protection logic: the Loan Estimate exists so a borrower can compare offers. A borrower's need to compare does not depend on whether a loan officer meant to start anything.

One practice that reduces accidental triggers: write pre-approval letters to a maximum purchase price with no property named. The property address is normally the last of the six to arrive; keeping it out of the file keeps the file a creditworthiness assessment rather than an application. (Other acceptable answers: a standing rule that any file holding five items gets disclosed anyway; a written intake script that records the date and source of each of the six as it arrives; training referral partners not to send property details with borrower financials in the same message.)

Exercise 9.36

LINDEN STREET — TRIGGER LOG                            [worked solution]

  ITEM                        DAY   SOURCE
  ────────────────────────────────────────────────────────────────────
  1  Name                      1    Borrowers, discovery call
  2  Income                    1    Borrowers, stated on the call
  3  Social Security number    1    Credit authorization, discovery call
  4  Property address          5    Borrowers, subject property named in
                                    the application
  5  Estimate of value         5    $385,000 contract price
  6  Loan amount sought        5    $365,750
  ────────────────────────────────────────────────────────────────────
  APPLICATION DATE .................... day 5
  LOAN ESTIMATE DUE ................... 3rd business day after day 5
  LOAN ESTIMATE DELIVERED ............. day 6

The file note:

Day 1 was not an application. The pre-approval issued on day 1 was written to a maximum purchase price with no subject property identified, and the borrowers had not identified a property to us; five of the six items were held as of day 1, and the property address, estimate of value, and loan amount sought were all first received on day 5 when the executed contract was provided and the application was completed. Application dated day 5; Loan Estimate delivered day 6.

Two things the note does that matter more than its wording. It states the affirmative reason the earlier date does not apply rather than merely asserting a date, and it identifies which items were and were not held, so that a reader years from now can evaluate the judgment instead of having to trust it.

Exercise 9.37

Subject: Everything I need — one list, due Thursday

Congratulations again. Here's the whole list in one place. Nothing else
is coming after this unless the underwriter asks for something specific.

BORROWER 1 (nurse)
  - Paystubs covering the last 30 days
  - W-2s for the last two years
    (why: the shift differential and overtime are real income, but they
     only count with a two-year history — the W-2s are what make them
     count)

BORROWER 2 (sales)
  - Paystubs covering the last 30 days
  - W-2s for the last two years
    (why: same reason — your commission is averaged over 24 months)

BOTH OF YOU
  - Photo ID, front and back
  - Last two statements, ALL PAGES, for both savings accounts
    (why: page 4 of 5 that says "intentionally left blank" is still
     page 4 of 5, and a missing page reads as a hidden page)
  - The earnest money check image and the statement page showing it clear
  - Your landlord's name, address, and phone
  - Your insurance agent's name and number

THE $10,000 FROM YOUR PARENTS
  - Signed gift letter — I'll send the form today
  - A statement from their account showing the funds
  - DO NOT MOVE THE MONEY YET. Call me first. There's a right way to
    transfer it and a way that costs us two weeks.

Everything goes in the secure portal, not email — link below, and I'll
text it to you too. Please don't email a paystub; it isn't safe.

DUE IN FOUR DAYS. The appraisal and title go out either way at the end
of this week — what these four days buy is a complete file going into
underwriting instead of a partial one, which is the difference between
one round of questions and three. If one item is going to be late, send
the rest and tell me which one — I'd rather start on nine than wait
for ten.

Roughly 280 words. Graded on: one message, grouped by borrower, a reason on more than half the items, a deadline with a stated consequence, an escape hatch for a single late item, and the secure-delivery instruction. The gift instruction is the highest-value sentence in the message and it is the one most often omitted, because the loan officer assumes nobody would move \$10,000 without asking. They do it constantly.

Chapter 10

Worked solutions to the daggered (†) and odd-numbered exercises. Percentages to two decimals, dollars to the cent.

Exercise 10.1

Equifax, Experian, and TransUnion.

They are three separate, competing, privately owned companies. There is no central credit database and they do not share data with one another. Each buys information from furnishers — banks, card issuers, auto lenders, servicers, collection agencies — and furnishing is voluntary, so a creditor may report to one, two, three, or none of them.

The result is three separately assembled files on the same person, containing different accounts with different balances as of different report dates, each scored by that bureau's implementation of the model. Three data sets produce three scores. The Linden Street borrowers' spreads — thirteen points on one, fourteen on the other — are ordinary. A spread of forty points usually means one bureau is missing an entire account or carrying something that does not belong.

Exercise 10.3

A charge-off is what the original creditor did with its accounting. After prolonged nonpayment — commonly around 180 days — it writes the balance off its books as a loss. The debt is not forgiven, it remains collectible, and it remains reportable.

A collection is what a third party is doing about the debt. The account has been placed with or sold to a collection agency, which reports it as its own separate tradeline.

Yes — one debt can produce both. The original creditor's tradeline shows charged off with a balance, and the collection agency's tradeline shows the same debt again. On the report it looks like two obligations.

How you tell: read the original creditor field on the collection tradeline and match it against the charged-off account — same creditor, same or similar balance, consistent dates. Then order a credit supplement to confirm before you count anything twice in the ratio or tell a borrower to pay twice.

Exercise 10.5

A credit reporting agency — in mortgage usage, a reseller. Not a bureau, and not Fannie Mae.

What the reseller does that a single bureau does not:

1. Pulls all three bureau files simultaneously under one request
2. Reconciles duplicates — the same auto loan reported by all three becomes one merged line
3. Formats to residential mortgage standards so the report is usable by an underwriter and by the automated systems
4. Returns all three scores per borrower, each labeled with its bureau and model version
5. Provides the downstream services in §10.8 — supplements and rapid rescores

A bureau returns its own file and its own score. Only the reseller produces the merged document.

Exercise 10.7

• Rapid rescore — you must have documentation from the creditor that a condition has actually changed: a paid-in-full letter, a zero-balance statement, a corrected limit, a letter confirming an account is not the borrower's. A screenshot of an app is not documentation.
• Credit supplement — you must have a specific question the merged report does not answer: the current balance, the actual minimum payment, a payoff, a missing credit limit, twelve months of rent.
• Dispute — you must have information the consumer believes is inaccurate or incomplete. Note the subject of that sentence: it is the consumer's tool, on a statutory clock, and you do not file it for them.

Exercise 10.8

Any three of the following, none of which involves a missed payment:

• The household pays cash and has never borrowed
• The borrower recently arrived in the country and has no domestic credit history
• The borrower is young — a 24-year-old with one nine-month-old account has nothing to score
• The borrower spent years abroad and their file went dormant
• The borrower rebuilt after a bankruptcy without borrowing, so the file is thin because they were careful
• The borrower's obligations — rent, utilities, insurance, tuition, childcare — are all with parties that do not furnish to the bureaus

The point to state explicitly: a scoring model with nothing to measure returns nothing. That is not the same as returning a low number, and treating the two the same way is the most common error in this part of the business.

Exercise 10.9

Step 1 — reduce each borrower to one score. Step 2 — reduce the loan to one score.

Note (c): the duplicate rule. With two identical scores and one different, use the duplicated score — 640 — not 640's neighbor and not an average.

Note (e): 625 clears the common conventional floor of 620 but sits near the bottom of the pricing grid, and the 802 does nothing for the file. This is the scenario that produces the hardest phone call in the chapter.

Note (f): a third borrower can only make the file's score worse or leave it unchanged. It can never improve it.

Exercise 10.11

Three errors in one sentence:

1. You never average. Averaging is not the rule at any step. 724 is a number that does not appear anywhere on this file.
2. The words "742 and 706" skipped step 1. Those are already the middles; the colleague got step 1 right by luck of phrasing but has described a one-step process, which means they will get the next file wrong when a borrower returns only two scores or two identical ones.
3. The direction of the error is the dangerous one. 724 is higher than the true score, so every quote built on it is better than the file can deliver. The error surfaces at the worst possible moment — when pricing is locked or when the Loan Estimate has to be redisclosed.

The file's representative score is 706 — the lower of the two middles.

The professional habit that prevents this: quote conservatively at the lower score. If the pricing turns out to use something better, you deliver good news. The reverse conversation costs you the borrower.

Exercise 10.13

A model answer at 38 words:

"You both have good credit. When two people borrow together, the loan is priced on the lower of the two middle scores — that's the rule, not my choice. Yours is 742, theirs is 706, so the file prices at 706."

What makes it work: it opens by validating both borrowers, it names the rule as a rule rather than a judgment, it uses no jargon beyond "middle score," and it never assigns responsibility to the borrower with the lower number. Never say "because of your co-borrower." Say "the file."

Exercise 10.14

Method A = original term − months since opened. Method B = balance ÷ payment, as a check. Method B should always read slightly lower than method A, because part of every payment is interest.

(a) Method A: 72 − 62 = 10 payments remaining. Method B: \$4,980 ÷ \$511 = 9.75. Lower than 10, by a quarter of a payment. Consistent.

(b) Method A: 48 − 26 = 22 payments remaining. Method B: \$7,510 ÷ \$362 = 20.75. Lower than 22, by about 1.25 payments. Consistent.

(c) Method A: 60 − 30 = 30 payments remaining. Method B: \$9,440 ÷ \$295 = 32.00. Higher than 30. Inconsistent — see Exercise 10.15.

Exercise 10.15

Tradeline (c).

Why the arithmetic cannot be right: method B divides the balance by the payment and ignores interest entirely. Since part of every payment is interest, a balance always takes more payments to retire than balance ÷ payment suggests — so method B is a lower bound on the remaining term. When method B produces a figure above method A's remaining term, the two fields are describing loans that cannot both exist.

Three conditions that could produce it:

1. The original term is misreported — the loan is longer than 60 months, or was refinanced or re-contracted and the furnisher kept the old term.
2. Payments were deferred, skipped, or modified — a forbearance, a skip-a-payment promotion, or a loan modification extends the payoff without changing the reported term.
3. The balance or the payment is stale or wrong — the furnisher reported an old balance, or the payment field carries a promotional or interest-only amount rather than the fully amortizing one.

What you do: order a credit supplement, or get a payoff statement or amortization schedule from the servicer. Do not guess, and do not put an excluded debt in front of an underwriter on the strength of subtraction alone.

Exercise 10.16

The rule: Chapter 4's ten-month rule. With exactly ten payments remaining, this installment debt is a candidate for treatment under it — subject to the program-specific conditions Chapter 4 sets out.

The document the underwriter will require: a payoff statement or the servicer's amortization schedule confirming the number of payments remaining — or, at minimum, a credit supplement verifying the terms and balance.

Why the credit report alone is not enough:

1. The report does not print the remaining term. You derived it by subtraction from two other fields, either of which can be wrong or stale.
2. The balance is as of the last furnish date, not today.
3. The reported original term can be inaccurate for the reasons in Exercise 10.15.
4. The rule removes a real monthly obligation from the ratio, which materially changes the approval. An underwriter certifying a file to an investor does not accept a derived figure for something that consequential.

And the timing trap: ten payments today is nine at closing on a file that runs long, and eleven if you counted the date opened wrong by a month. Verify the direction of the error before you rely on it.

Exercise 10.17

The utilization the model is most likely reading:

\$2,780 ÷ \$3,010 = 92.36%

With no assigned credit limit reported, scoring models generally fall back to the high credit — the highest balance ever reported — as the denominator. That is almost always the least favorable available figure, because a borrower who has ever run the card near its true limit produces a high credit close to their balance.

If the account's real limit is, say, \$6,000, the true utilization is 46.33% — but nothing on the report says so, so the model cannot know it.

What you do: order a credit supplement requesting the assigned credit limit.

How long: typically days.

Who pays: the lender, as a file cost. Not the borrower.

This is the cleanest opportunity in the chapter: no borrower cash, no dispute, no risk, and if the limit comes back higher than the high credit, utilization is recomputed downward on the next scoring. If the change is material, it is also a clean candidate for a rapid rescore.

Exercise 10.19

Restated for an oldest-first grid. Reading the same string left to right as oldest-to-newest, the delinquencies now sit at the beginning of the two-year window rather than near the end. The "1" that was seven months ago is now seventeen months ago, and the "1 2" sequence that was sixteen and seventeen months ago is now seven and eight months ago — and critically, the sequence now reads as a 30-day late followed by a 60-day late, meaning the account rolled deeper rather than curing.

That is a materially worse file: a 60-day late seven or eight months ago is recent and severe. Under the first reading, the most recent event was a single 30-day late seven months ago that cured immediately. Same twenty-four characters, two different borrowers.

The habit, in one sentence: read the grid's direction indicator on every report, from every vendor, every time — before you read the grid.

Say it as a rule rather than a caution, because the failure mode here is not carelessness. It is confidence. A loan officer who assumes the wrong direction will tell a borrower something specific and wrong with total conviction, and neither of them will find out until the underwriter reads it correctly.

Exercise 10.20

(a) Per account and aggregate.

(b) Every account below 30%.

• Account 1 target: \$3,000 × 0.30 = \$900 → pay \$2,340 − \$900 = \$1,440
• Account 2 at 15.60% → \$0
• Account 3 target: \$5,000 × 0.30 = \$1,500 → pay \$4,900 − \$1,500 = \$3,400
• Account 4 at 7.67% → \$0

Total paydown: \$1,440 + \$3,400 = \$4,840.

New balances: \$8,135 − \$4,840 = \$3,295. New aggregate: \$3,295 ÷ \$14,500 = 22.72%.

(c) With \$2,000.

New balances: \$8,135 − \$2,000 = \$6,135. New aggregate: \$6,135 ÷ \$14,500 = 42.31%.

Why allocation does not change the aggregate: the aggregate is total balances over total limits. Spending \$2,000 reduces the numerator by \$2,000 no matter which account it lands on, and the denominator does not move at all. The aggregate is 42.31% whether the whole \$2,000 goes to account 3, is split across all four, or is applied to account 1.

Why you still care where it goes: scoring models read individual accounts as well as the aggregate, and a single account at 98% is its own negative. The strongest use of \$2,000 here is probably \$1,440 to account 1 — which takes it fully under 30% and eliminates one high-utilization account entirely — with the remaining \$560 to account 3, bringing it to \$4,340 ÷ \$5,000 = 86.80%. That is not obviously better than putting all \$2,000 on account 3 (\$2,900 ÷ \$5,000 = 58.00%), and reasonable practitioners differ. What you may not do is tell the borrower which allocation is worth more points. Nobody knows.

Exercise 10.21

The Linden Street bank card: \$3,850 balance against a \$4,500 limit.

(a) That card. \$3,850 − \$1,500 = \$2,350. \$2,350 ÷ \$4,500 = 52.22%, down from 85.56%.

(b) The aggregate. \$8,400 − \$1,500 = \$6,900. \$6,900 ÷ \$17,200 = 40.12%, down from 48.84%.

Worth noting: \$1,500 buys a meaningful move on the aggregate and still leaves that card above 50%. The \$2,500 required to take it under 30% is only \$1,000 more, which is the argument for doing the larger number if the cash exists — and §10.7 is the argument for checking first whether it does.

Exercise 10.22

If they pay today (the 21st): the statement already closed on the 18th. The furnisher reported the balance as of that date. The \$3,000 payment posts to an account whose reported balance for this cycle is already fixed, and the scoring models will keep reading the old, higher balance until the next cycle closes on the 18th of next month and the furnisher reports again — commonly a few days after that. Elapsed time before the score can reflect the payment: roughly four weeks.

If they wait until the 16th of next month: the payment posts before the cycle closes on the 18th, the lower balance is what the furnisher reports, and the reduction reaches the bureaus on the ordinary schedule.

Which you recommend on a 45-day contract signed yesterday: pay today.

The reasoning is timing, not scoring theory. Waiting until the 16th means the new balance does not report until roughly the 20th–24th, which on a contract signed yesterday is day 25 or later — with underwriting submission, conditions, and a lock all downstream of it. Paying today puts the money in place immediately, and the moment the statement closes on the 18th you can order a rapid rescore with the creditor's documentation rather than waiting for the ordinary reporting cycle.

Paying today and rescoring after the 18th is faster than waiting to pay and letting it report naturally — and it is the whole reason the rapid rescore exists.

The additional discipline: get each card's statement closing date in writing from the borrower at application, before you plan any of this. You cannot time a payment to a date you do not know.

Exercise 10.23

Before the paydown:

\$22,400.00 − \$16,900.00 = \$5,500.00 remaining. \$5,500.00 ÷ \$2,050.00 = 2.68 months of reserves.

After a \$2,600 paydown:

\$5,500.00 − \$2,600.00 = \$2,900.00 remaining. \$2,900.00 ÷ \$2,050.00 = 1.41 months of reserves.

The question you must answer first: does this file need those reserve months more than it needs the utilization improvement?

Reserves are a compensating factor an underwriter can see and use. Going from 2.68 months to 1.41 months on a file that is already thin is a real reduction in the strength of the approval, and the utilization improvement is worth an unknown number of points that may or may not cross a pricing band. On a file with a comfortable ratio, ample reserves, and a score sitting a few points below a band, the paydown is obviously right. On a thin-reserve file, it can be obviously wrong.

Also required before recommending it: confirm which dollars are available. Gift funds and other sourced assets carry permitted-use rules (Chapter 12), and not every verified dollar can be spent on a debt paydown.

Exercise 10.24

Exercise 10.25

Order: (f), then (d), then (a) — on documentation grounds.

(f) Bring the past-due account current — first. It is the only item on the list that represents an ongoing harm rather than a historical one, and its documentation is the simplest thing in this chapter: the creditor's confirmation that the account is current, obtainable within the week. It also removes a live condition an underwriter would issue anyway.

(d) Get the missing limit reported — second. Documentation is a credit supplement, ordered by you, returned in days, costing the borrower nothing and requiring no decision from them. It cannot make anything worse. The only reason it is not first is that it does not stop an active problem.

(a) The paydown — last. Its documentation cannot exist until the money has moved and the creditor will issue a statement or letter reflecting the new balance. That means borrower action, borrower cash, a decision about reserves (Exercise 10.23), and then a wait for the statement — and only then can a rapid rescore be ordered.

Note the discipline the question is teaching: this ordering says nothing about which action is worth the most points, because you do not know that and cannot find out. Ordering by documentation available is the only defensible ordering on a forty-day clock. A defensible alternative sequence is (d), (f), (a) — supplement first because it is the fastest and free — and either answer is fine if the reasoning is documentation and time, not expected points.

Exercise 10.26

A model answer at 74 words:

"He's not making it up — that does happen sometimes, when an item genuinely can't be verified. But the furnisher almost always can verify an accurate item, so accurate things come back unchanged. Here's the part that worries me: while a dispute is open, the tradeline gets flagged, the underwriting system stops, and clearing it takes about a month. On a 45-day contract that's the whole file. Send me anything you think is actually wrong and I'll look today."

Why it works: it concedes the brother-in-law's factual kernel rather than attacking him, it corrects the misreading (unverifiable items come off; accurate items get verified), it names the concrete transaction risk with a timeline, and it ends by giving the borrower a legitimate channel for the same impulse.

Exercise 10.27

What is happening: almost certainly one debt appearing twice — the hospital system's charged-off account, and the collection agency's tradeline for the same debt after placement or sale. The matching \$1,180 balances are the tell.

What you order: a credit supplement on both tradelines, requesting the original creditor on the collection and the current status and balance on both. If the collection's original creditor is the hospital system and the amounts reconcile, it is one obligation.

What you tell the borrower:

"There are two lines on here for what I think is the same hospital bill — the hospital's own record and the collection agency that took it over. I'm confirming that before either of us treats it as two debts, because you should never pay the same bill twice and we shouldn't count it twice either. I'll have an answer in a few days. Don't send anyone money in the meantime — call me first."

Three things that last sentence does: it prevents the borrower from paying a collector who calls next week, it prevents them from paying the wrong party, and it keeps them from taking any action that could alter the report mid-file.

Exercise 10.28

How both documents can be correct. They are reporting from different sources. Beginning in 2017, under a joint initiative the three nationwide bureaus adopted in connection with settlements with state attorneys general, the bureaus imposed enhanced identification and update-frequency standards on public record data. Most civil judgments and tax liens could not meet those standards and were removed from consumer credit reports — judgments beginning in mid-2017, with essentially all remaining tax liens removed by 2018.

The judgment never went away. It was entered by a court, it is in the county's public records, and it attaches to the borrower and to property they own. The credit report simply stopped carrying it. The title search reads the land and court records directly, which is why the title commitment found it in week three.

What you should have done at application: taken the URLA declarations seriously. Chapter 9's application asks the borrower directly about outstanding judgments, delinquent federal debt, and party status in a lawsuit — and those questions exist precisely because the credit report no longer answers them. Ask them out loud rather than clicking through them, and follow up on any yes immediately, before the title work rather than after.

The sentence to internalize: "the credit is clean" and "there are no liens" are two different statements, and only one of them is supported by the document in front of you.

Exercise 10.29

Sentence one (accurate reporting period): "A Chapter 7 bankruptcy may generally be reported for up to ten years from the date of the order for relief or adjudication, so this one has roughly five years left."

Sentence two (no removal date promised): "I can tell you the outer limit the law allows; I can't tell you the exact date any particular bureau will drop it, so let's plan the file around it still being there."

Why the second constraint exists: the ten-year figure is the maximum period the Fair Credit Reporting Act permits, not a scheduled deletion date, and the bureaus' actual removal practices are their own — they have, for instance, voluntarily removed Chapter 13 filings earlier than the statutory maximum, which is a practice rather than a requirement and could change. Promising a borrower a date is a prediction about a private company's operational behavior that you cannot control, cannot verify, and gain nothing from making. It is the same discipline as never promising a score outcome: state the rule, refuse the date.

Exercise 10.30

A model letter. The wording is yours; the five requirements are not.

Subject: Your credit report — what I found, and one decision for you

Thanks for the time this morning. Here is everything I learned from the credit report, in plain language, so you have it in writing.

Your scores. Each of the three credit bureaus returns its own score, so there are three per person. We use the middle one for each of you, and then the lower of those two for the loan. That comes out to 742 for one of you and 706 for the other, so the file is priced at 706. That is the rule for every lender, not a preference of mine, and it has nothing to do with which of you earns what.

What's on the report. It is a clean report. No late payments in two years, no collections, no public records. Two car loans and the student loans, all current.

The one soft spot, and your one decision. You have four credit cards carrying \$8,400 against \$17,200 of total limits — about 49% used. Two of the cards are above 84% of their limits. How much of a limit is in use is the largest part of the score model that can actually change quickly, usually within one billing cycle.

So here is the option, with real numbers:

	Cost	Cards would be	Your reserves after closing
Do nothing	\$0 | 48.84% used | \$12,623.66 = 4.16 months of payments
Bring every card under 50%	\$2,450 | 34.59% used | \$10,173.66 = 3.35 months
Bring every card under 30%	\$4,040 | 25.35% used | \$8,583.66 = 2.83 months

What I can and can't tell you about that. I cannot tell you how many points it would move. Nobody can — not me, not the bureaus, not any company that says otherwise. What I can tell you is that it moves the biggest changeable part of the model, that it works within about a month, and that it costs you cushion after closing that the underwriter also looks at. It is a real trade in both directions, and it is your call, not mine. I'm happy to walk through it on the phone.

One thing I need you to do, which is actually the most important part of this email. Between now and the day you get keys, your credit gets checked again by the lender right before closing. Anything new shows up. So please:

• No new accounts of any kind. No car, no furniture, no appliances, no store card at the register.
• Don't cosign for anyone.
• Don't close any cards you already have.
• Don't file any credit disputes, and don't hire a company to do it. That freezes the file for about a month.
• Don't change jobs, and don't move large sums between accounts, without calling me first.

If you want to buy something, call me. Two minutes and I'll tell you whether it's safe. I would much rather take ten of those calls than have a different conversation eight days before closing.

What I need from you, by Thursday at noon: the statement closing date for each of the four credit cards. Just the date each statement cuts — it's on the top of the statement or in the app. I need it before we decide anything about the balances, because when a payment lands matters as much as how much it is.

Grading, in order of importance:

1. Does it contain a promise? Check the last third especially. First drafts almost always slip one in as encouragement — "that should get you where you need to be," "that'll help a lot," "we'll be in great shape." Any of those fails the exercise regardless of everything else.
2. Does it name 706 and how it was derived, without blaming a borrower? The correct construction is "the file is priced at 706." The failing construction is any sentence that makes the 706 belong to a person.
3. Are the figures right and complete? \$8,400 / \$17,200 / 48.84%, and reserve months computed against the \$3,033.72 payment: \$12,623.66 ÷ \$3,033.72 = 4.16; \$10,173.66 ÷ \$3,033.72 = 3.35; \$8,583.66 ÷ \$3,033.72 = 2.83.
4. Is the no-new-credit instruction present, specific, and in list form? "Be careful with your credit" is not an instruction.
5. Is there a specific request with a deadline? The statement closing dates are the right ask, because nothing else in the plan can happen without them. A request for "anything you can send" is a fail.

Then apply the exercise's own last instruction: strike every sentence that would embarrass the writer if read aloud in a deposition. Students are usually surprised by how little survives that test in a first draft, and by how little is lost.

Exercise 10.31

The ninety-second no-new-credit script. A model version — the content is the requirement, the wording is yours:

"Last thing, and this is the one that actually causes problems, so I want ninety seconds.

Between today and the day you get keys, your credit gets checked again. Not by me — by the lender, right before closing. Anything new shows up.

So until we close: no new accounts. No car. No furniture. No appliances, even the ones the store swears are interest-free. No store card at the register for ten percent off. Don't cosign for anybody. Don't close any cards you already have. Don't file any credit disputes and don't hire a company to do it — that one freezes the file for a month. Don't change jobs without calling me. And don't move large sums of money between accounts without telling me where it came from.

I know some of that sounds paranoid. Here's the real reason: a five-thousand-dollar furniture purchase can add six hundred dollars a month to your obligations, and six hundred dollars a month is enough to break the approval you already have. It happens constantly and it is completely avoidable.

So: if you want to buy something, call me first. Two minutes, and I'll tell you if it's safe. I'd rather answer ten calls than have this conversation on day forty-four.

I'm going to email all of this so you have it in writing. Forward it to anyone who might buy you a housewarming gift on a store card."

Then time yourself. Read aloud, that is roughly ninety to a hundred seconds. If your version runs over two minutes, cut the explanation — never the list.

Exercise 10.32

A model reply:

"No — and good catch asking.

The credit report shows what the servicer last furnished, which on income-driven plans is frequently stale or zero. What we use isn't what the bureau shows and isn't what the borrower remembers; it's what the current plan documents say, and the treatment rule for income-driven repayment is program-specific.

Please get me: (1) the servicer's current statement or the plan approval letter showing the payment amount and the recertification date, for every loan, and (2) if the statement is ambiguous, a credit supplement from our credit vendor verifying the current required payment.

Then Chapter 4's rule decides what actually enters the ratio — including what happens when the documented payment is \$0.00, which is not the same question as what the report shows. Don't put a zero in the file until we have the documents and I've confirmed the treatment."

What the exercise is testing: three things. That you go to the servicer, not the bureau, for a student loan payment. That you name the document rather than asking for reassurance. And that you know which chapter owns the rule — the credit report supplies the number, Chapter 4 decides what is done with it. Those are different jobs and confusing them is how a file gets submitted with a figure the underwriter will not accept.

Exercise 10.33

(a) What you say.

"I believe you, and I'm sorry — that's a genuinely unfair outcome for one missed due date after nine clean years. Here's where I am: that late is accurate, and I can't help you dispute something that's accurate. Not because I'm being cautious — it's a rule I'm bound by, and separately it wouldn't work, because the creditor will verify it and it'll stay. What I can do is show you exactly what would move, and we can look at whether any of it is worth doing."

(b) What you do. Redirect to the things that are both legitimate and available: compute utilization, identify any genuine errors on the report, look for a missing credit limit, check whether a documented paydown timed to the statement date could plausibly reach the band, and price the file honestly at where it is now so the borrower can make a real decision.

(c) What you document. A dated note in the file recording that the borrower requested assistance disputing an item you determined to be accurate, that you declined, that you explained why, and that you offered legitimate alternatives. Also note that you did not refer them to any credit repair organization. If they raise it again, note that too.

(d) The two rules from §10.10 that close the question.

1. You may never advise a borrower to dispute accurate information. Advising a consumer to make an untrue or misleading statement to a consumer reporting agency is specifically prohibited by federal law, and there is no "but it was only one late" exception.
2. You may never promise a score outcome. Even if the dispute somehow succeeded, you have no basis for telling this borrower it would move them two points — and their entire premise is a point estimate nobody can supply.

What it costs you if they leave. Be honest about this rather than pretending it away: it may cost you the loan. A borrower who is two points from a band and has just been told no will sometimes call someone who says yes. You lose the origination, the referrals from it, and the closing.

What you keep is your license and the file note that shows what you did. And there is a real, unsentimental business case: the loan officer who says yes to this request is the one whose borrower's file freezes in week four, whose lock expires, and whose agent stops calling. Case Study 10.2 is what saying yes looks like eight weeks later.

Exercise 10.35

B — 699.

Step 1: Borrower A sorts to 690 / 715 / 742 → 715. Borrower B sorts to 680 / 699 / 705 → 699. Step 2: the lowest of {715, 699} = 699.

The distractors and their errors:

• A (680) — took the single lowest number printed anywhere on the page. A step-1 failure.
• C (707) — averaged the two middles: (715 + 699) ÷ 2 = 707. Never average.
• D (715) — took the higher borrower's middle. Step 1 correct, step 2 backwards.

Exercise 10.37

C — the automated findings flag the dispute and the lender must resolve it before relying on them.

Why the others fail:

• A — a dispute has no effect on the debt-to-income calculation. The obligation exists and is counted.
• B — scores are not recalculated to exclude disputed items. The item remains in the file.
• D — this is the belief that produces Case Study 10.2. Disputes are between the consumer and the bureau in law, and they stop a mortgage file in practice.

The mechanism to be able to state: the disputed tradeline carries a comment code, the automated findings flag it, the lender must resolve the flag, some furnishers will not supplement an account under active dispute, and clearing it requires the consumer to withdraw with the bureau and the furnisher to update — which can consume most of a thirty-day cycle.

Exercise 10.39

B.

Why B is right: it names the dependency (model version), gives the accurate practical answer for mortgage lending (the older classic versions generally do not disregard a collection merely because it has been paid — that behavior arrived in newer versions such as FICO 9), gives the borrower an actionable rule (pay it if the program requires it), and refuses to promise a point outcome.

Why the others fail:

• A — false. Paid collections are not universally removed, and the model behavior differs by version. Note separately that the bureaus' 2022 changes removed paid medical collections specifically, which is a different and narrower statement.
• C — false in the other direction; newer models do treat paid collections differently.
• D — a promised point outcome. This is the answer that ends careers, and it is on the exam for that reason.

Also worth noting for the nine-year-old item: it is approaching or past the seven-year reporting window for most adverse items and arguably should not be on the report at all. Confirm what is actually reported and when the reporting period began before advising anyone to pay it.

Exercise 10.41

(a) With nine payments remaining on the \$429.00 auto.

Excluding it, the numerator becomes:

\$4,479.72 − \$429.00 = \$4,050.72

\$4,050.72 ÷ \$10,500.00 = 0.385783 = 38.58%, down from 42.66%. A four-point improvement, from a number the credit report never printed.

(b) With the day-41 furniture account added.

\$4,479.72 + \$611.00 = \$5,090.72

\$5,090.72 ÷ \$10,500.00 = 0.484830 = 48.48%, up from 42.66%. Nearly six points worse.

(c) Which you can influence.

You cannot influence (a). The remaining term is a fact about a loan signed years ago. What you can do is read it, on day 1, and know it — because on a file that runs long, 19 becomes 12 becomes 10, and a loan officer who never computed it in the first place will never notice when it crosses. Read every installment tradeline and flag anything that will be at ten or fewer before the projected closing.

You can absolutely influence (b), and it is the single highest-leverage thing in this chapter. The furniture account is not a credit-analysis problem; it is a communication problem, and the entire prevention is the ninety-second script in Exercise 10.31 delivered on the day of application and put in writing.

The asymmetry is the lesson. The number you cannot change is worth four points and you find it with subtraction. The number you can change is worth six points and you prevent it with a conversation. Most loan officers spend their effort on the first one.

Exercise 10.43

A model checklist — fifteen lines, and the word "score" does not appear until line six.

CREDIT INTAKE — read in this order, compute as you go

  1  HEADER — names, addresses, SSN match indicators, date of pull. Does this
     report belong to your borrowers? Note the date; it has a shelf life.
  2  ALERTS — fraud alerts, security freezes, address discrepancies, consumer
     statements. Anything here stops everything else until it is handled.
  3  PUBLIC RECORDS — bankruptcies. If any, note the date; the waiting-period
     question goes to the program (Ch. 14 / 16), not to you.
  4  COLLECTIONS — read the ORIGINAL CREDITOR field on every one. Match against
     charged-off tradelines: one debt can appear twice.
  5  TRADELINES, INSTALLMENT — for each: payment, balance, date opened, terms.
     COMPUTE remaining term = terms − months since opened. Check: balance ÷
     payment should be slightly LOWER. Circle anything ≤ 10, or ≤ 10 by closing.
  6  TRADELINES, REVOLVING — for each: balance, limit, minimum payment.
     COMPUTE utilization per account. Flag any account with NO reported limit
     — that is a supplement, today.
  7  TOTAL the revolving: balances, limits, minimums. COMPUTE aggregate
     utilization. Check the minimums against what you put in the ratio.
  8  PAYMENT GRIDS — read the DIRECTION INDICATOR first, then the grid. Read
     every mortgage and rental tradeline's grid before any other.
  9  INQUIRIES — count them, read the dates and the requesting parties.
 10  SCORES — three per borrower. Circle each middle. Write the REPRESENTATIVE
     SCORE at the top of page one. Quote from that number and nothing else.
 11  REASON CODES — read all four. Note how many name the same underlying fact.
 12  TOTAL monthly obligations from the tradelines. Reconcile to the 1003.
 13  Write the three sentences you will say on the phone today.
 14  Deliver the no-new-credit / no-disputes instruction. Then email it.
 15  Date and initial the page. It is now the first verified document in the file.

Why the ordering matters: it forces the reader to have opinions about the file before they have an opinion about the number. A checklist that starts with the score produces a loan officer who reads the score and skims the rest — which is precisely the habit this chapter exists to prevent.

Chapter 11

Worked solutions to the daggered (†) and odd-numbered exercises. All guideline thresholds are stated as commonly applied; on a live file, confirm the current requirement in the applicable guide.

Exercise 11.1

The three questions, in order:

1. Is it stable? Does the income have a history long enough for a pattern to be distinguishable from an event?
2. Is it likely to continue? Forward-looking, commonly stated as a three-year expectation.
3. Can it be documented? By a third party the underwriter will believe.

One income type that most often fails each:

• Question 1 — a second job started ten months ago. Documentable, continuing, and with no two-year history.
• Question 2 — overtime with a long history where the employer states on the VOE that it is ending, or child support with 26 payments remaining.
• Question 3 — cash tips that never touch a W-2, or monthly help from a family member.

Exercise 11.2 †

(a) Correct monthly income. Bi-weekly is 26 pay periods:

$$\$2{,}700.00 \times 26 = \$70{,}200.00 \qquad \$70{,}200.00 \div 12 = \mathbf{\$5{,}850.00}$$

(b) The ×2 figure. \$2,700.00 × 2 = **\$5,400.00**.

(c) The error. \$5,850.00 − \$5,400.00 = \$450.00 per month understated.

$$\frac{\$450.00}{\$5{,}850.00} = \mathbf{7.69\%}$$

The same percentage arises directly from the periods: multiplying by 2 counts 24 periods instead of 26, and 2 ÷ 26 = 7.69%. Because the error is proportional, it is 7.69% on every bi-weekly file regardless of the pay rate — which is exactly why it survives undetected.

Exercise 11.3

Qualifying income is \$71,000 ÷ 12 = \$5,916.67 per month — the current salary. The W-2's \$61,400 corroborates employment and the story; it does not set the figure.

The reason the same file averages overtime but not salary is question 1. Base pay has a rate; variable pay only has a history. A salary is a contractual amount the employer is obligated to pay going forward, so the current rate is itself the forward statement. Overtime is not promised by anything — the only evidence that it will occur next year is that it occurred in the last two, and the honest way to extrapolate from a fluctuating series is to average it.

Third sentence: an underwriter who averaged salary would be qualifying the borrower on a wage the employer has already stopped paying, which is the same error, pointed the other way, as averaging a declining bonus.

Exercise 11.4 †

(a) At 36 contracted hours:

$$\$28.75 \times 36 = \$1{,}035.00 \text{ per week} \qquad \$1{,}035.00 \times 52 = \$53{,}820.00 \qquad \$53{,}820.00 \div 12 = \mathbf{\$4{,}485.00}$$

(b) At 40 hours:

$$\$28.75 \times 2{,}080 = \$59{,}800.00 \qquad \$59{,}800.00 \div 12 = \mathbf{\$4{,}983.33}$$

Difference: \$4,983.33 − \$4,485.00 = **\$498.33 per month** — nearly \$6,000 a year of qualifying income riding on one field of one form.

(c) The Verification of Employment controls the hours; the paystub corroborates them. If the VOE leaves the hours blank, do not assume forty. Go back to the employer for the contracted or guaranteed schedule, and in the meantime reconcile against the paystub's regular hours per period (80 hours per bi-weekly period implies 40; 72 implies 36). Never build a pre-approval on an assumed schedule — this is the single most common source of a qualifying income figure that quietly does not exist.

Exercise 11.5 †

(a) The 24-month average:

$$\frac{\$4{,}860 + \$6{,}300}{24} = \frac{\$11{,}160}{24} = \mathbf{\$465.00 \text{ per month}}$$

(b) Year-over-year change:

$$\frac{\$6{,}300 - \$4{,}860}{\$4{,}860} = \frac{\$1{,}440}{\$4{,}860} = \mathbf{+29.63\%}$$

(c) The trend is rising, so the 24-month average is usable. Note what that means for the borrower: the most recent year alone was \$6,300 ÷ 12 = **\$525.00 a month, and the rule hands them \$465.00. The average is the lower figure, which is precisely why it is the permitted one. Rising variable income is good news the borrower does not get to spend.

Exercise 11.7 †

(a) By pay periods elapsed (the correct method for a bi-weekly earner):

$$\frac{\$52{,}530.00}{18} = \$2{,}918.33 \text{ per period} \qquad \$2{,}918.33 \times 26 = \$75{,}876.67 \text{ per year}$$

$$\$75{,}876.67 \div 12 = \mathbf{\$6{,}323.06 \text{ per month}}$$

(b) By calendar months elapsed:

$$\frac{\$52{,}530.00}{8} = \mathbf{\$6{,}566.25 \text{ per month}} \qquad \times 12 = \$78{,}795.00 \text{ per year}$$

(c) The difference: \$6,566.25 − \$6,323.06 = \$243.19 per month**, or \$78,795.00 − \$75,876.67 = **\$2,918.33 a year — exactly one pay period, which is the tell.

The methods disagree because they divide by different fractions of the year. Eighteen of twenty-six pay periods is 18 ÷ 26 = 69.23% of the year. August 30 is 8 ÷ 12 = 66.67% of the year. Method B divides the same earnings by the smaller fraction, so it produces the larger annual figure.

(d) Use the pay-period method for any earner whose pay periods do not align with month ends — every weekly and bi-weekly borrower. The calendar-month method is appropriate for a semi-monthly or monthly earner, where the period boundaries and the month boundaries are the same thing.

Exercise 11.9 †

(a) 24-month average:

$$\frac{\$31{,}200 + \$38{,}400}{24} = \frac{\$69{,}600}{24} = \mathbf{\$2{,}900.00 \text{ per month}}$$

(b) Most recent year: \$38,400 ÷ 12 = **\$3,200.00 per month. The trend is rising, so the 24-month average of \$2,900.00 governs** — again, the lower of the two.

(c) Total qualifying income: \$6,100.00 base + \$2,900.00 commission = \$9,000.00 per month.

(d) Commission as a share of this borrower's income:

$$\frac{\$2{,}900.00}{\$9{,}000.00} = \mathbf{32.22\%}$$

That exceeds the threshold commonly discussed at 25%, so expect additional documentation requirements — historically including personal tax returns and an analysis of unreimbursed business expenses. Confirm the current treatment for your program; it has changed and it differs by agency.

The denominator is the borrower's own income rather than the household's because the test is asking a question about this earner's compensation structure — how much of their pay is contingent on production. Using household income would let a second earner's salary dilute the measurement and mask exactly the risk the test exists to detect. On a two-earner file, the same \$2,900 against a \$14,000 household total would compute to 20.71% and would wrongly appear to clear.

Exercise 11.11 †

(a) At the correct income:

$$\frac{\$4{,}479.72}{\$10{,}500.00} = \mathbf{42.66\%}$$

**(b) At \$10,060.00** (Borrower 1's base miscalculated as \$5,280.00 instead of \$5,720.00):

$$\frac{\$4{,}479.72}{\$10{,}060.00} = \mathbf{44.53\%}$$

(c) A change of 1.87 percentage points from one multiplication.

An arithmetic error is more dangerous than an unknown guideline because a guideline you do not know announces itself — the underwriter writes a condition, you look it up, you learn it once. A multiplication error produces a plausible number that flows silently into the pre-approval, the quote, the disclosures, and the offer, and it is only discovered when someone re-performs the calculation, which on a bad file is post-closing quality control.

Exercise 11.13 †

(a) Adjusted gross rents: \$2,200.00 × 0.75 = **\$1,650.00**.

(b) Net rental income: \$1,650.00 − \$1,975.00 = −\$325.00 per month.

(c) A negative result is not zero income. It appears on the application as a \$325.00 monthly liability, added to the borrower's debts.

For a borrower with \$8,000.00 of other monthly income and \$2,900.00 of other obligations:

Four full points of debt-to-income from a property the borrower described as "income."

Exercise 11.15 †

(a) No gross-up. Income = \$1,950.00 + \$2,300.00 = \$4,250.00.

$$\text{DTI} = \frac{\$1{,}900.00}{\$4{,}250.00} = \mathbf{44.71\%}$$

(b) Illustrative 15% gross-up on the non-taxable portion only:

$$\$1{,}950.00 \times 1.15 = \$2{,}242.50 \qquad \text{income} = \$2{,}242.50 + \$2{,}300.00 = \$4{,}542.50$$

$$\text{DTI} = \frac{\$1{,}900.00}{\$4{,}542.50} = \mathbf{41.83\%}$$

(c) Illustrative 25% gross-up:

$$\$1{,}950.00 \times 1.25 = \$2{,}437.50 \qquad \text{income} = \$2{,}437.50 + \$2{,}300.00 = \$4{,}737.50$$

$$\text{DTI} = \frac{\$1{,}900.00}{\$4{,}737.50} = \mathbf{40.11\%}$$

(d) Nothing about the household changed. Same benefits, same pension, same house, same payment — and a debt-to-income ratio of 44.71%, 41.83%, or 40.11% depending entirely on a percentage set by a program. For a file sitting near a threshold, the gross-up percentage is the approval.

The implication is operational and absolute: look the percentage up for the specific program, on this file, today, and note in the file where you looked it up. Do not carry a number in your head, do not quote one on a first call, and do not let one from a prior program follow you into a new one. Note also that the pension is fully taxable and receives no adjustment even though it appears on the same statement.

Exercise 11.17

(a) Satisfied: question 1 (stability) — eleven months of documented receipt from a state disbursement unit is strong evidence of a pattern, and commonly exceeds the roughly six months frequently required. Question 3 (documentation) is also satisfied: a court order plus an official payment record is exactly the third-party evidence required.

(b) Failed: question 2 (continuance). With 26 payments remaining against a commonly applied three-year forward expectation of roughly 36 months, the income falls short by about 10 months.

(c) The \$650.00 is generally **not countable**, which removes \$650.00 from the denominator of every ratio. On a file with \$5,400.00 of other income and \$2,300.00 of obligations, the back-end ratio moves from \$2,300.00 ÷ \$6,050.00 = 38.02% to \$2,300.00 ÷ \$5,400.00 = 42.59% — a four-and-a-half-point swing.

Verify the specific continuance requirement in the applicable guide (Selling Guide, Seller/Servicer Guide, HUD Handbook 4000.1, or the VA/USDA handbook) and against your employer's overlays, because the required window and how it is measured vary. And tell the borrower on the first call, not after the pre-approval — this is the single most commonly miscounted income type in first-time-buyer files.

Exercise 11.18 †

(a) What a 24-month average would produce:

$$\frac{\$8{,}400 + \$6{,}200}{24} = \frac{\$14{,}600}{24} = \$608.33 \text{ per month}$$

(b) Why that number is not the answer. There is no twenty-four-month history to average. The job began fourteen months ago, so ten of the twenty-four months in the denominator are months in which this income did not exist. Dividing real earnings by a period that includes non-existent months does not produce a conservative figure; it produces a meaningless one.

Note the tempting alternative and why it also fails: averaging over the actual twenty months of earnings gives \$14,600 ÷ 20 = **\$730.00 a month**, which is arithmetically defensible and still not usable, because the objection was never the divisor. It was the history.

(c) The failed question and the cure. This fails question 1 (stability). Questions 2 and 3 are both satisfied — the borrower still holds the job, and paystubs and a W-2 document it.

The cure is time, and it is measurable: a part-time or second job commonly requires a two-year uninterrupted history, so this borrower is roughly ten months from eligibility. Nothing the loan officer does accelerates it. Confirm the specific requirement for the program and read the automated findings, which occasionally permit a shorter history with additional documentation.

What it costs the file. For a borrower with \$5,800.00 of other monthly income and \$2,400.00 of obligations:

Nearly four points of debt-to-income, from income the borrower is genuinely earning. Say so on the first call — this is the exclusion borrowers find hardest to accept, and it is much easier to accept before an offer has been written.

Exercise 11.19

Use the new salary: \$81,000 ÷ 12 = \$6,750.00 per month. Base is taken at the current rate; the prior employer's \$72,000 does not average in.

Document: an offer or employment letter confirming the salary and start date; paystubs from the new employer covering the required period; a written VOE from the new employer; and two years of W-2s (which will show the prior employer, and that is fine). Because the move is within the same line of work with the same duties, the two-year history carries across employers and the stability question is satisfied.

The follow-up question that would change the answer: "Is any part of the \$81,000 not base salary?" If the new package is, say, a \$56,000 base plus a \$25,000 target bonus, the borrower has converted \$25,000 of documented base into variable income with no history at the new employer — qualifying income would be \$56,000 ÷ 12 = \$4,666.67, more than \$2,000 a month below the figure above, on a compensation package that is nominally identical. A secondary question worth asking: whether the new employer imposes a probationary period.

Exercise 11.21 †

(a) Length of the gap. From March 3 to July 15:

Approximately 134 days, or about 4.4 months. (If the new job's first day is July 15 itself, the borrower was out of work for about 133 of those days; either reading lands in the same place.)

(b) A gap of roughly thirty days or more commonly requires a written letter of explanation. At four and a half months the file is approaching the range where an extended gap can also raise the question of whether the borrower has been back at work long enough for the income to be considered stable. Both the trigger and the treatment vary by program and by the automated findings; verify.

(c) Two sentences for the letter of explanation (the borrower writes them; you tell them what has to be in them — a specific reason and a specific date):

"I left [prior employer] on March 3 because my employer closed the facility where I worked and my position was eliminated. I began working at [new employer] on July 15 in the same field and at a higher rate of pay, and I have been continuously employed there since."

A document that strengthens it: a separation or layoff notice, a WARN-type notice, an enrollment record if the gap was for school, or an unemployment benefit determination establishing the dates. Attach it with the letter. A vague letter invites a second condition; a specific letter with a supporting document ends the conversation.

Exercise 11.22

(a) Base monthly income from the document:

$$\$26.00 \times 2{,}080 = \$54{,}080.00 \qquad \$54{,}080.00 \div 12 = \mathbf{\$4{,}506.67}$$

(b) The rate the borrower's claim implies:

$$\$5{,}200.00 \times 12 = \$62{,}400.00 \qquad \$62{,}400.00 \div 2{,}080 = \mathbf{\$30.00 \text{ per hour}}$$

The borrower described a \$30.00 rate. The paystub says \$26.00. That is the first finding, and it is a \$693.33 per month discrepancy.

(c) The year-to-date hours. Eighteen periods at 80 hours would be 1,440 regular hours. The stub shows 1,080.

$$1{,}440 - 1{,}080 = 360 \text{ hours} \qquad 360 \div 40 = \mathbf{9 \text{ weeks}}$$

Roughly nine weeks of regular pay is missing from the year to date. That is an employment gap, a leave, a reduction in hours, or a mid-year start — and until you know which, you do not know this borrower's income.

Confirm it independently by annualizing the YTD:

$$\frac{\$30{,}420.00}{18} = \$1{,}690.00 \times 26 = \$43{,}940.00 \qquad \div 12 = \mathbf{\$3{,}661.67 \text{ per month}}$$

That is far below both the borrower's \$5,200.00 and the computed base of \$4,506.67 — consistent with the nine missing weeks, and it corroborates the finding rather than contradicting it.

(d) The two conditions to expect: (1) a written Verification of Employment stating the current rate, the contracted hours, dates of employment, and the prior two years' earnings; and (2) a letter of explanation for the year-to-date shortfall, with supporting documentation.

The two questions to ask the borrower first: "Between January and now, was there any period you weren't working or were working reduced hours?" and "When you said \$5,200 a month, were you including overtime, or is that what you understand your base to be?" Ask both before submission. Both answers are almost always innocent, and both are conditions if the underwriter finds them first.

Exercise 11.23 †

(a) With the figures reversed — \$23,400 two years ago, \$19,800 last year:

$$\frac{\$23{,}400 - \$19{,}800}{\$23{,}400} = \frac{\$3{,}600}{\$23{,}400} = \mathbf{-15.38\%}$$

Declining. The 24-month average (still \$43,200 ÷ 24 = \$1,800.00) is not usable; the most recent year governs.

(b) Commission: \$19,800 ÷ 12 = **\$1,650.00**.

• Borrower 2: \$2,400.00 + \$1,650.00 = \$4,050.00
• Borrower 1 (unchanged): \$6,300.00
• **Household total: \$10,350.00** (down \$150.00 from \$10,500.00)

(c) The ratios:

$$\text{housing} = \frac{\$3{,}033.72}{\$10{,}350.00} = \mathbf{29.31\%} \qquad \text{back-end} = \frac{\$4{,}479.72}{\$10{,}350.00} = \mathbf{43.28\%}$$

Up from 28.89% and 42.66%.

(d) What you would have said differently on the first call. You would have quoted and pre-approved on **\$10,350.00 from the very beginning** — never on \$10,500.00 — and you would have told the borrowers plainly that because last year's commission came in below the prior year, the file counts the lower year and not the average. The rest of the conversation is arithmetic: at 43.28% the file is materially tighter, the room to absorb a new debt before closing is much smaller, and the day-44 furniture account this file is famous for would be even more destructive. Deliver that on day one. You can always give better news later; you cannot un-give a pre-approval.

Exercise 11.25

A — \$5,500.00. Semi-monthly is 24 pay periods:

$$\$2{,}750.00 \times 24 \div 12 = \$66{,}000.00 \div 12 = \$5{,}500.00$$

which is the same as \$2,750.00 × 2, because semi-monthly is twice a month.

B (\$5,958.33) is the bi-weekly calculation (× 26 ÷ 12) applied to a semi-monthly earner — the error running in the opposite direction from the usual one, and it overstates income by \$458.33 a month. C and D are arithmetic noise. The distinction the item tests: bi-weekly is 26, semi-monthly is 24, and only the second one may be multiplied by two.

Exercise 11.27

A tax return transcript shows line items from the return as the taxpayer filed it. A wage and income transcript shows the information returns filed about the taxpayer by third parties — W-2s, 1099s, and similar forms reported to the IRS by employers and payers.

The practical difference: the wage and income transcript corroborates a specific W-2 the borrower handed you, line by line, from the IRS's own copy. The return transcript only shows what the borrower reported.

What a wage and income transcript can reveal that nobody mentioned: income sources the borrower did not disclose — a 1099 from a side business, a second employer, a 1099-R distribution, or a 1099 from a partnership. Some of those are simply additional income. Some of them are a self-employment analysis nobody planned for (Chapter 32), a liability, or a red flag (Chapter 27). All of them are better discovered by you, on day ten, than by an underwriter on day thirty.

Exercise 11.28 †

(a) The 24-month average:

$$\frac{\$28{,}600 + \$30{,}200}{24} = \frac{\$58{,}800}{24} = \mathbf{\$2{,}450.00 \text{ per month}}$$

(b) The divisor is 24 because the borrower will make 24 mortgage payments over those two years, not 14. Seasonal income is earned in a concentrated period and spent across the whole year; dividing by the months worked would produce a monthly figure the borrower has never actually had available on a year-round basis, and would qualify them for a payment they cannot make in February.

(c) The two documents answering continuance: the Verification of Employment carrying the employer's stated intent to rehire for the coming season, and the W-2 history establishing that the borrower has in fact worked the same seasonal line of work across prior seasons. Without the rehire statement, the W-2s are a record of jobs that ended. Note also that the trend test still applies — this stream rose (\$28,600 → \$30,200, +5.59%), so the average is usable.

Exercise 11.29

The arithmetic the message has to carry:

$$\text{base} = \frac{\$70{,}000}{12} = \$5{,}833.33 \qquad \text{bonus} = \frac{\$20{,}000 + \$28{,}000}{24} = \frac{\$48{,}000}{24} = \$2{,}000.00$$

$$\text{qualifying income} = \mathbf{\$7{,}833.33 \text{ per month}}$$

The borrower's "ninety-eight thousand" is \$98,000 ÷ 12 = **\$8,166.67. The file will use \$333.34 a month less**, and the bonus is counted at \$2,000.00 rather than at last year's \$2,333.33.

A model message:

Good news first — your bonus counts, both years of it, and that is not automatic. Here is exactly how the file will read it, so nothing surprises you later.

Your salary counts at today's rate: \$70,000 ÷ 12 = **\$5,833.33 a month.**

Bonus income gets averaged over the last twenty-four months rather than taken at the most recent year, because it is not a fixed part of your pay. You received \$20,000 and then \$28,000, so: (\$20,000 + \$28,000) ÷ 24 = \$2,000.00 a month.

**Total qualifying income: \$7,833.33 a month.** That is about \$333 a month less than the \$98,000 figure suggests, because the average includes the smaller first year.

What that means for you: it is the number I will use for your pre-approval and the number the underwriter will use, so the letter I give your agent will hold. Your bonus trend is rising, which is exactly what an underwriter wants to see. Call me and I will walk you through the price range this supports — I would rather show you a number that closes than one that looks better today.

Why it works: it opens with what counted rather than what did not, it shows the arithmetic so the borrower can check it, it names the reason without blaming a rulebook, and it ends with a concrete next step. It never says "unfortunately," and it never says "guidelines."

Exercise 11.31

The four components, rebuilt from the source facts:

It foots: \$5,720.00 + \$580.00 + \$2,400.00 + \$1,800.00 = \$10,500.00. Borrower 1 contributes \$6,300.00; Borrower 2 contributes \$4,200.00.

The four sentences (two per borrower):

To Borrower 1: "Your hourly rate times a full-time year is \$5,720 a month, and that counts at today's rate — a raise would count immediately." · "Your nights-and-weekends differential and your overtime come to \$580 a month, which is a two-year average rather than what you earned last year, because that kind of pay isn't guaranteed and the file averages it."

To Borrower 2: "Your salary is \$2,400 a month and counts at today's number." · "Your commissions average \$1,800 a month across the last two years — last year alone was better, at \$1,950, but the file uses the two-year average, so the rising trend works in your favor by making the average usable rather than by raising the number."

If any of those takes more than about twenty seconds to say, you are explaining the rulebook instead of the borrower's own pay.

Exercise 11.33 †

(a) What you must establish before you can answer at all:

1. Which program the file will be placed in — availability of an offer-letter path varies by agency, program, and lender overlay, and it changes.
2. Whether the offer is non-contingent — no pending background check, license, drug screen, relocation approval, or contingency of any kind.
3. The start date relative to closing — here, 21 days after closing. Programs that permit this typically define a window, and whether 21 days falls inside it is the threshold question.
4. The compensation structure — \$6,900.00 is stated as base. If any part is bonus or commission, that part has no history at the new employer and generally cannot be counted at all (§11.9).
5. Whether the current job continues until closing, and what income the file would qualify on if the offer-letter path is unavailable.
6. Verified reserves — enough to cover payments from closing until the new income begins.

(b) Two additional file requirements to expect: verified reserves sufficient to cover the payment obligation across the gap between closing and the first paycheck; and a post-closing or pre-funding re-verification once employment actually begins — often a paystub from the new employer plus a verbal VOE. Expect the lender's overlay to be stricter than the agency guideline here; this is a path lenders are cautious about.

(c) The one-sentence answer on the phone today:

"There is a path for this on some programs, so send me the offer letter today and give me until tomorrow morning — I'll tell you whether it works on the program we're using, and if it doesn't, I'll tell you exactly what would."

That sentence is honest, it is fast, it commits to a time, and it promises nothing. Never tell a borrower this is possible before you have confirmed it for the specific program — an offer-letter qualification that turns out to be unavailable at day thirty is a transaction, not an inconvenience.

Chapter 12

Worked solutions to the daggered (†) and odd-numbered exercises. Arithmetic is shown. Linden Street figures are the book's frozen constructed figures.

Exercise 12.1

Sourcing is documented origin — third-party evidence of where a specific sum came from. Seasoning is time in the account.

Seasoned but unsourceable: \$6,000 in currency deposited eleven months ago. It has been there long enough to fall outside any documentation window, and no third party can ever certify where currency has been.

Unseasoned but fully sourced: a commission check deposited yesterday with the commission statement and the check copy attached. Zero seasoning, complete documentation, not a problem.

The pair is the point: the tests are independent, and seasoning is a boundary of inquiry rather than a certificate of legitimacy.

Exercise 12.3

The average balance for the preceding two months. A borrower-supplied statement shows a moment; a VOD shows a moment and its recent history, side by side, certified by the institution. An account whose current balance is a multiple of its own two-month average is announcing that money arrived recently — without anyone having to read a transaction line or ask an accusatory question.

Exercise 12.5

Reserves are verified liquid assets remaining after closing, expressed in months of PITI (including MI and HOA where applicable).

The unit is months rather than dollars because dollars say nothing about a particular household. \$12,000 is eleven months of housing payment for one borrower and three for another. Months answer the only question reserves exist to answer: how long could this family keep paying if the income stopped?

Exercise 12.7

Vested is the portion of a retirement account the employee actually owns. Employee contributions generally vest immediately. Employer contributions frequently vest on a schedule tied to years of service; until they vest they belong to the employer.

Read the vested balance, never the account balance. Then ask the second question, which is whether the borrower may withdraw it at all — the plan's terms of withdrawal answers that, and many plans do not permit in-service withdrawals.

Exercise 12.9 †

Note the shape of the marking. Only the items whose description identifies a payer are safe. Everything the bank labels generically — "DEPOSIT," "MOBILE DEPOSIT," "CASH DEPOSIT" — is a condition, regardless of how legitimate the money is. That is the same mechanism as Figure 12.1.

Exercise 12.11 †

$$\frac{\$31{,}400.00}{\$6{,}180.00} = 5.08$$

The current balance is 5.08 times the two-month average.

What it tells you: a large amount of money — on the order of \$25,000 — arrived recently, inside the documentation window.

What it does not tell you: what it was, who sent it, whether it is the borrower's, or whether it is a loan. A VOD reports balances, not sources.

First two actions, before submission: (1) obtain complete statements for the full window and identify the specific deposit or deposits producing the gap; (2) call the borrower with an open question — "do you know what came in recently?" — and then request the documents that source it. Do not submit and wait for the condition; you already know it is coming.

Exercise 12.13 †

Established by the statement: a savings account ending ****4419 existed over the period; the beginning balance was \$2,799.38 and the ending balance \$7,850.00; four transactions occurred; a credit of exactly \$4,900.00 posted on the last day of the period; and the balance nearly tripled as a result.

Not established: who paid the \$4,900; whether it was earned, given, lent, or repaid; what account or institution it came from; whether any obligation attaches to it; and whether the borrower is the sole owner of the funds.

Would it still be a large deposit by direct deposit? Yes. \$4,900.00 is 46.67% of \$10,500.00 of monthly qualifying income regardless of how it arrived. Size is a fact about the amount.

Would it still be a condition? In practice, generally no — a direct deposit carries the employer's name into the statement description, so the statement sources itself and the underwriter has the answer on the page.

The difference between the two questions is the difference between the amount and the evidence. "Large deposit" describes the money. "Condition" describes what the file can prove about it. A loan officer who conflates them will argue with underwriters about the wrong thing for an entire career.

Exercise 12.15 †

Exercise 12.16 †

$$\text{per diem} = \frac{\$365{,}750.00 \times 0.06625}{365} = \frac{\$24{,}230.9375}{365} = \$66.3861$$

Moving from a 22-day position to an 8-day position — closing roughly two weeks later in the month — reduces prepaid interest by 14 × \$66.3861 = **\$929.41** (\$929.40 from the rounded figures above; the penny is a rounding artifact and worth noticing, because a Closing Disclosure will not have one).

What actually happens, and what does not. The first payment date does not move — an October closing produces a December 1 first payment either way — so the borrower genuinely brings less cash to the table. But this is a timing shift, not a discount: the interest not prepaid is simply interest that accrues later inside the payment stream. Two things move against it: the rate lock has an expiration and pushing a closing can force an extension that costs real money, and a later closing means another period of the borrower's current housing cost plus contract risk. Never sell a later closing as free money.

Exercise 12.17 †

Base case.

$$\$38{,}000.00 - \$25{,}376.34 = \$12{,}623.66 \qquad \frac{\$12{,}623.66}{\$3{,}033.72} = 4.16 \text{ months}$$

(a) \$4,900.00 unsourced.** Verified assets \$38,000.00 − \$4,900.00 = \$33,100.00. \$33,100.00 − \$25,376.34 = \$7,723.66** ÷ \$3,033.72 = 2.55 months.

(b) Decline owner's title, survey, and pest inspection (\$875.00 + \$450.00 + \$125.00 = \$1,450.00). Cash to close \$25,376.34 − \$1,450.00 = \$23,926.34. \$38,000.00 − \$23,926.34 = \$14,073.66** ÷ \$3,033.72 = 4.64 months. (And say the other half out loud: owner's title insurance is the only thing in that column protecting the buyer. Chapter 21.)

(c) Pay off a \$5,200.00 obligation from reserves.** \$12,623.66 − \$5,200.00 = **\$7,423.66 ÷ \$3,033.72 = 2.45 months.

(d) 10% down. Down payment = \$38,500.00. Cash to close = \$14,126.34 + \$38,500.00 − \$5,000.00 − \$3,000.00 = **\$44,626.34 — against \$38,000.00 of verified assets. They cannot do it.** They are \$6,626.34 short, with zero reserves even if they could.

Why the "costs unchanged" assumption is not quite true: origination and discount points are percentages of the loan amount, which falls from \$365,750.00 to \$346,500.00 — origination drops to \$3,465.00 and discount to \$1,732.50, a combined \$288.75 less — and the per-diem falls with the loan balance too. The refinements shrink the shortfall by a few hundred dollars. They do not change the answer.

Exercise 12.19 †

$$\text{vested} = \$41{,}000.00 + (0.60 \times \$21{,}000.00) = \$41{,}000.00 + \$12{,}600.00 = \mathbf{\$53{,}600.00}$$

Two documents before any of it counts: the retirement account statement showing the vested balance, and the plan's terms of withdrawal, establishing whether the borrower may access funds while still employed.

Why an inaccessible vested balance may still matter: some programs will count a discounted portion of a retirement account toward reserves even where it cannot be withdrawn for closing, because it represents genuine net worth. Whether yours does, and at what discount, is a guide question — verify.

Exercise 12.21

$$\$33{,}100.00 - \$25{,}376.34 = \$7{,}723.66 \qquad \frac{\$7{,}723.66}{\$3{,}033.72} = 2.55 \text{ months}$$

Against the base case of 4.16 months, one unsourced \$4,900 deposit cost this file **\$4,900.00 and 1.62 months of reserves** (\$4,900.00 ÷ \$3,033.72 = 1.62; the rounded figures subtract to 1.61).

State it the way it should be stated to a borrower: a document you can get in fifteen minutes is worth a month and a half of your mortgage payment.

Exercise 12.22 †

Ranked by paper trail, best to worst:

(c) Wire from the donor's account directly to the closing agent. Documents: gift letter, donor's bank statement, outgoing wire confirmation, closing agent's receipt. The money never enters the borrower's account, so there is no large deposit to source at the borrower's bank.

(b) Personal check deposited by the borrower. Documents: gift letter, donor's bank statement showing the debit, a copy of the check, and the borrower's statement showing the credit. Four documents and a large-deposit condition in the borrower's account.

(a) Cashier's check. Documents: gift letter, the instrument itself, the donor's statement showing the funds leaving the donor's account to purchase it, and the borrower's deposit. Ranked last of the three because the instrument is traceable but does not identify a source account on its face — you still need everything you needed for (b), plus an extra step.

The instruction to the donor: "Please don't move the money yet. On closing day we'll send you wire instructions from the title company, and the funds go straight from your account to theirs — that way nobody has to document anything twice."

Exercise 12.23

Route 1 — 401(k) loan. Documents: terms of withdrawal, the loan agreement or plan confirmation showing the amount and repayment terms, and the deposit. Effect on DTI: many programs exclude the repayment on the reasoning that the borrower repays themselves — confirm the guide and the lender overlay before relying on it. The risk you must state: if the borrower leaves that employer, the outstanding balance commonly becomes due or is treated as a taxable distribution.

Route 2 — securities sale. Documents: the brokerage statement, the trade confirmation, and the deposit. Effect on DTI: none; nothing is borrowed. The risk you must state: the value moves between today and settlement, and the proceeds are subject to whatever tax treatment applies to the gain — which is the borrower's tax advisor's question, not yours.

Verify before recommending: the discount applied to each asset type, whether the 401(k) repayment is excluded, whether the securities require seasoning after deposit, and how many days each institution takes to actually release funds.

Exercise 12.25

The three questions, with documents:

1. Does the borrower have access? At 55%, 45% of that account is somebody else's. Documents: operating agreement or ownership schedule, and commonly the other members' written consent to the withdrawal.
2. Will the withdrawal harm the business? Documents: a cash-flow analysis, and frequently a letter from the borrower's tax preparer. Requirements vary by program and by manual versus automated underwriting — verify.
3. Is the money already counted? Documents: whatever income worksheet the file used.

Question 3 is the one that can reopen a calculation from another chapter. If the dollars were already reflected in the self-employment income analysis (Chapters 11 and 32), counting them again as assets is a double-count, and an underwriter who finds it reopens the income calculation — a far bigger problem than the asset question you were trying to solve.

Exercise 12.26 †

Cover note to the underwriter:

Re: condition 7 of 11 — \$4,900.00 deposit, account *4419. Attached: (1) employer quarterly commission statement, gross \$6,900.00 less \$2,000.00 withholding = net \$4,900.00, matching the deposit exactly; (2) copy of the deposited check, front and back; (3) borrower letter of explanation. Source is commission income already documented in the file and already included in the 24-month average supporting qualifying income — *it is not additional income and has not been counted twice. Please advise if anything further is needed.

Attachments list: quarterly commission statement · deposited check, front and back · signed letter of explanation.

Borrower letter of explanation (under 80 words):

The \$4,900.00 deposit to my savings account was my quarterly sales commission from my employer, paid by company check and deposited by me. It is not a loan and no repayment is owed to anyone. The commission statement showing gross earnings of \$6,900.00 and net pay of \$4,900.00 is attached, along with a copy of the check.

(signature, date)

Exercise 12.27

The call. Slow down, be warm, and take the burden off the donor:

"I'm sorry to be asking you this, and I want to explain why so it doesn't feel strange. The lender has to confirm that the gift is really a gift and not a loan, and the way they do that is by seeing that the money was already yours. It's one page — the most recent statement for the account the money is coming from. You can black out the account number except the last four digits. I'll send a stamped envelope or come by and pick it up, whichever is easier. Nobody at my company will see your balances except the underwriter on this file, and it goes nowhere else."

What you will do with it: transmit it securely, store it in the loan file under your company's GLBA safeguards, and use it for the single purpose of clearing the condition.

What you will not do: forward it to the real estate agent, discuss the donor's balances with the borrower beyond confirming the gift is supported, keep a personal copy, or send it by unencrypted email because it was easier.

Exercise 12.29 †

Gift letter.

GIFT LETTER
Date:               [date]
Donor:              [name], aunt of [borrower]
Donor address:      [street, city, state, ZIP]
Donor telephone:    [number]
Relationship:       aunt of the borrower
Recipient:          [borrower name]
GIFT AMOUNT:        $12,500.00
Subject property:   118 Callow Road
Source of funds:    donor's [institution] account ending [****]
Transfer method:    wire from the donor's account directly to the closing agent

"The funds described above are a bona fide gift. NO REPAYMENT OF THIS GIFT,
 IN ANY FORM, IS EXPECTED OR IMPLIED, now or in the future."

_________________________          _________________________
donor signature / date              borrower signature / date

Note to the file: This letter does not prove the donor has the money — the donor's bank statement does. It does not prove the funds ever moved — the wire confirmation and the closing agent's receipt do. (Confirm the current program's required gift letter content and donor sourcing requirements before submission.)

Exercise 12.31

When you're pre-approved, the lender has verified a specific set of accounts and a specific set of balances. Underwriting will look at sixty to ninety days of statements before closing and ask about every deposit that isn't a paycheck — on every file, for every buyer. That's normal.

Here's the part that surprises people: moving money between your own accounts creates the same question as new money. From the underwriter's side, a transfer with no matching withdrawal documented looks identical to a deposit from nowhere, so they have to ask for statements on both accounts, both ends, sometimes from a bank you've already closed out.

So between now and closing: don't move money between accounts, don't close anything, and don't deposit anything other than your paycheck without calling your loan officer first. It isn't that any of it is wrong. It's that each move is a document somebody has to go find, usually on the week you can least afford it.

(About 150 words, no jargon, forwardable as written.)

Exercise 12.33

The next question is an open one, and then silence: "Tell me more about the arrangement with your brother — what did the two of you actually agree to?" Not "is it a loan?", which invites a yes/no answer the borrower may already sense is the wrong one.

The range of correct outcomes. If the answer is that no repayment is expected, it is a gift: gift letter, donor documentation, and the transfer trail, and the file proceeds. If repayment is expected in any form — even informally, even "eventually," even with no agreement — it is a loan: it must be disclosed, the payment considered in the ratio, and the file restructured or requalified. A third, common outcome is that the borrower genuinely does not know, in which case the brother's understanding is the fact that matters and somebody has to ask him.

What you may never do. You may not tell the borrower which answer produces the outcome they want. You may not process a gift letter you have reason to believe is inaccurate. You may not decide for yourself what you think really happened and document that. And "will probably pay him back eventually" is not something you can un-hear — it is now a fact in your file.

Exercise 12.34 †

What you say:

"Don't do that — and I want to be really clear I'm not saying there's anything wrong with your money. Splitting up cash deposits specifically to stay under the reporting limit is its own separate problem under federal law, and it's a problem even when the money is completely clean. Put it in all at once. The bank files a routine report on large cash deposits; that report is not about you and it's not a problem for us. What we're waiting on isn't the report — it's time."

Why. The Bank Secrecy Act requires financial institutions to report currency transactions over \$10,000 (31 U.S.C. § 5313; 31 CFR 1010.311), and 31 U.S.C. § 5324 separately makes it unlawful to structure transactions to evade that requirement. The offense is the evasion; it stands independent of the legitimacy of the funds. A loan officer who suggests it is assisting in structuring.

The legitimate path and its cost. Deposit the \$9,400 in one transaction. Let it sit until it falls outside the documentation window the lender examines — commonly 60 or 90 days, verify per program. Then re-verify with fresh statements. Realistically that is two to three months, and in a purchase market that is a real cost: the house they wanted may be gone.

The two sentences that keep it from feeling like judgment:

"Saving that much cash takes discipline, and nothing about this is a knock on you — it's a limitation of what a bank statement is able to show. Let's put it in this week and build the timeline around it, so you're shopping with money that's already documented instead of finding this out under contract."

Exercise 12.35

What is wrong with it. The listing agent is paid at this closing. A "gift" from a party with an interest in the sale is not a gift — it is a price concession routed through a bank account. It makes the sales price a fiction, and every ratio computed from it — loan-to-value above all — is overstated by the amount of the "gift." It is generally prohibited on every program you are likely to run, and it is the exact structure Case Study 12.1 describes Congress eventually prohibiting by statute.

What you say to the agent. Plainly, without accusation — most agents offering this genuinely do not know: "I can't use that, and I want to save you the trouble of trying. A gift can't come from anybody who gets paid at this closing — the underwriter treats it as a seller concession, which has its own limits and can't be applied to the down payment. If the sellers want to help, let's talk about a seller credit inside the contract and see how far it gets us."

What you say to your manager. That the offer was made, that you declined it, what you offered instead, and that you are documenting the conversation. Not because the agent did anything criminal — they very likely did not — but because a contemporaneous note is worth a great deal if the same question resurfaces later in the file.

Exercise 12.37

The deficiency: the letter omits the statement that no repayment is expected.

Why it is the one that matters most: every other element is descriptive — the amount, the property, and the relationship all identify a transaction that the file could establish other ways. The no-repayment statement is the only element that establishes what the money legally is. Without it, the document proves that \$15,000 changed hands between two identified people, which is equally consistent with a gift and with a loan — and a loan carries a payment that belongs in the ratio. Everything else in a gift letter is identification. That one clause is the substance.

(Note also that acceptability of a fiancé as a donor varies by program — verify. But that is an eligibility question, not a defect in the document.)

Exercise 12.39

A Verification of Deposit is sent by the lender (or its processor) to the depository institution, and is completed and returned by the institution on its own authority.

What distinguishes the returned information is that it (a) never passes through the borrower's hands, and (b) includes the average balance for the preceding two months alongside the current balance — a historical figure a single statement does not provide. A borrower-supplied statement is a photograph of a moment; a VOD is a certified statement of a balance and its recent trajectory.

Exercise 12.41 †

New documents required if the \$10,000 gift is deposited into the joint savings account on day 30 rather than wired to the closing agent:

• the gift letter (already in the file)
• the donors' bank statement showing the funds available and the debit
• a copy of the check or the wire confirmation into the borrowers' account
• the borrowers' updated savings statement showing the \$10,000 credit — a statement that did not exist at submission
• a large-deposit explanation tying them together, because a \$10,000 credit is 95.24% of monthly qualifying income

New savings balance at verification: \$28,000.00 + \$10,000.00 = \$38,000.00 in the accounts.

Does the reserve figure change? No. Total verified assets are still \$38,000.00, cash to close is still \$25,376.34, and reserves are still \$12,623.66 = 4.16 months. What changes is the cleanliness, not the number. With the gift commingled, the \$12,623.66 remaining is a mixture, and if the program excludes gift funds from reserves the underwriter now has to reason about which dollars survived rather than simply observing that the gift was consumed at the table. You have converted a fact into an argument, and arguments take days.

What you tell the donor on day 6: "Don't send the money to your daughter. On closing day the title company will send you wire instructions, and it goes straight from your account to theirs. It's one wire and one confirmation instead of four documents and a condition."

Chapter 13

Worked solutions to the daggered (†) and odd-numbered exercises. All figures are illustrative and constructed; verify current pricing, mortgage insurance factors, and guideline values at their sources.

Exercise 13.1

The six decisions, in order: program; down payment; term; rate and point position; mortgage insurance structure; concessions and buydowns.

Decisions 1–3 (program, down payment, term) mostly determine whether the loan is possible — they set eligibility, the loan amount, and the qualifying payment. Decisions 4–6 (rate/point position, MI structure, concessions) mostly determine what it costs. The trap is that the first group has few options and receives most of a new originator's attention, while the second group has many options and receives almost none.

Exercise 13.3

The four questions: (1) How much cash can they actually put in, and does the difference between programs buy convenience or feasibility? (2) What does the credit look like — representative score and derogatory history? (3) What is the property, and how will they occupy it? (4) How long will they keep this loan?

The gate before all four — Question Zero: is there an entitlement or an eligibility? Has either borrower served (VA)? Is the property in a USDA-eligible area with household income under the cap? Ask about military service on every application, of every applicant, every time — borrowers routinely do not volunteer service they assume does not count.

Exercise 13.5

Total cost of credit is the sum of everything a borrower pays to obtain and carry a loan over the period they actually hold it: points and origination charges, all interest, all mortgage insurance, and any financed one-time charge.

Two things it includes that the note rate does not: discount points and other upfront charges; and mortgage insurance (including a financed upfront premium such as FHA's UFMIP). Either of these can move the comparison by tens of thousands of dollars without touching the rate.

What it excludes and why: the down payment. A down payment is not spent — it is converted into equity in the property. It reduces liquidity, not net worth, so counting it as a cost of borrowing would systematically favor whichever structure requires the least of it, which is precisely the error this chapter exists to prevent.

Exercise 13.7

Who funds it: whoever agrees to — most commonly the seller or the builder as a concession, sometimes the lender. The borrower may fund one only where the applicable guide permits it; verify.

Where it lives: in an escrow or buydown account established at closing and administered by the servicer, which draws from it each month to make up the difference between the note-rate payment and the subsidized payment. The funds are delivered in full at closing — this is a cost to the payer on day one, not an obligation spread over two years.

Unspent balance on an early payoff: if the loan is refinanced or paid off in month 14, the remaining balance is generally applied to the loan rather than retained by the lender. Verify the specific treatment against the note, the buydown agreement, and the servicer's policy — this is a real protection and worth telling the borrower about, but the documents govern.

Exercise 13.8 †

Conventional 5% down: \$260,000 × 0.05 = **\$13,000, plus \$5,900 of costs and prepaids = \$18,900** required. Against \$19,000 verified, that leaves \$100.00.

FHA 3.5% down: \$260,000 × 0.035 = **\$9,100, plus \$5,900 = **\$15,000 required. Against \$19,000 verified, that leaves **\$4,000.00**.

The relief is \$13,000 − \$9,100 = \$3,900.

The answer: functionally feasibility, and this is the gray zone the test has. On paper the conventional structure funds — by one hundred dollars. In practice a structure that leaves \$100.00 in the borrower's account is not a structure any responsible originator should treat as available. It survives no appraisal-required repair, no earnest-money timing problem, no per-diem interest change from a two-day closing delay, and it will almost certainly draw a reserve requirement from the AUS findings or a lender overlay. Call it feasibility, recommend FHA, and say out loud why: "technically you can do the conventional loan and it would leave you a hundred dollars, which is not a cushion, it's a rounding error."

The fact that would flip it: any additional verified funds — a documented gift, a seller concession covering closing costs, or a lender credit taken from a higher row on the rate sheet — that restores a real reserve position. Roughly \$4,000 to \$6,000 of additional available cash converts this from feasibility back to convenience, at which point the credit score, the horizon, and the mortgage insurance cancellation terms decide it.

Exercise 13.9

A defensible rewrite:

"FHA usually requires less cash at closing and prices its mortgage insurance without regard to credit score, which makes it the cheaper loan for a borrower whose score is low or whose down payment is the binding constraint. For a borrower with a strong score who can fund a conventional down payment and intends to hold the loan, FHA is frequently more expensive over the term, because its mortgage insurance does not cancel."

The original sentence fails on three counts: it names a program rather than a borrower; it does not say cheaper measured how (monthly payment, cash to close, or total cost of credit — the answer differs); and "first-time buyer" is not a credit or cash characteristic at all.

Exercise 13.11

Four sentences, with arithmetic:

"Twenty percent on a \$340,000 house is \$68,000, and you have \$41,000 — so that version of this purchase is about twenty-seven thousand dollars away before we even count closing costs. Ten percent is \$34,000, which you can technically do, but with roughly six thousand in costs on top you'd close with about a thousand dollars left in the bank, and I don't recommend anyone buy a house with a thousand dollars left. Five percent is \$17,000, which leaves you around eighteen thousand after costs — call it five or six months of the new payment sitting in your account. The advice you heard isn't wrong; it's just written for somebody with more cash than you have, and the actual question is how much cushion you want on the other side of closing."

Note what the answer does not do: it does not tell them the advice is stupid, and it does not decide for them. It converts a slogan into three numbers and hands back the choice between the 5% and 10% columns.

Exercise 13.13

Why not cash back: on a purchase, a lender credit offsets closing costs and prepaid items. It is not a source of funds to the borrower — paying it out as cash would make it, in substance, financing that has not been underwritten or disclosed as such, and program rules do not permit it.

The practical problem: the credit is generated by the rate, so its size is set by which row of the rate sheet you pick — not by what the borrower's costs happen to be. If you price a file at 7.000% for a \$2,743.12 credit and the borrower's total costs and prepaids are \$2,100, there is \$643.12 with nowhere to go. The excess is generally lost, and the correct fix is to reprice to a lower row (6.875%, credit \$1,371.56) rather than to leave the borrower paying a permanently higher rate for money they never received. Know the borrower's actual closing costs before you quote a credit.

Exercise 13.14 †

(a) Loan amount and LTV. Down payment: \$285,000 × 0.05 = \$14,250. Loan: \$285,000 − \$14,250 = \$270,750.00. LTV: \$270,750 ÷ \$285,000 = 95.00%.

(b) Principal and interest at 6.750%, 30-year fixed, on \$270,750: **\$1,756.08**. (Payment factor at 6.750%/360 months is approximately \$6.48599 per \$1,000: 6.48599 × 270.75 = \$1,756.08.)

(c) Monthly mortgage insurance. \$270,750 × 0.0058 = \$1,570.35 per year ÷ 12 = \$130.86.

(d) Total housing payment. \$1,756.08 + \$285.00 + \$105.00 + \$130.86 = \$2,276.94.

(e) Housing ratio. \$2,276.94 ÷ \$7,200.00 = 31.62%.

(f) Back-end debt-to-income. (\$2,276.94 + \$640.00) ÷ \$7,200.00 = \$2,916.94 ÷ \$7,200.00 = 40.51%.

Exercise 13.15

What you need to know first: does the borrower have cash to cover closing costs independently?

Because the arithmetic answers itself once you look at it. Closing costs and prepaids are \$8,400; the concession is \$6,000. The concession does not cover the costs. There is nothing left over to spend on any kind of buydown, and the borrower still has to produce \$2,400 out of pocket on top of their down payment.

So the answer is the same at both horizons — closing costs — and the horizon question does not arise. You cannot spend the same \$6,000 twice.

This is the point of the exercise. The buydown-versus-costs debate only becomes live when the concession exceeds the costs it is needed for, or when the borrower has enough independent cash to pay their own costs and free the concession for something else. Ask about the cash before you ask about the horizon. If the borrower did have the \$8,400 in hand, then the eighteen-month horizon points to a temporary buydown (front-loaded relief, all of it inside the horizon) and the twenty-year horizon points to a permanent buydown or a price reduction (smaller monthly relief that never stops).

Exercise 13.17 †

(a) 1.000 point on \$300,000 = 1% × \$300,000 = \$3,000.00.

(b) \$1,945.80 − \$1,896.21 = \$49.59 per month.

(c) \$3,000.00 ÷ \$49.59 = 60.5 months, or 5.04 years.

(d) At month 36: Recovered: \$49.59 × 36 = **\$1,785.24. Unrecovered: \$3,000.00 − \$1,785.24 = \$1,214.76 lost**.

(e) Recommendation: do not buy the point. Their stated horizon of about three years is roughly half the break-even, and the expected cost of buying it is \$1,214.76 — real money, spent for nothing. Sit at par at 6.750%, or, given that a transfer is likely, look at the credit side of the grid instead: a lender credit repays faster than a point recovers, and a borrower leaving in three years may be better off taking money at closing and paying the higher rate for only 36 months.

The one fact that would reverse it: a materially longer horizon — anything past about 60.5 months. Note that "we might not get transferred after all" is not the same as a longer horizon, and should be priced as uncertainty rather than as a decision.

Exercise 13.18 †

(a) Effective rates. Year 1: 5.000% (7.000% − 2). Year 2: 6.000% (7.000% − 1). Year 3 and after: 7.000%, the note rate, which never changed.

(b) Monthly subsidy. Year 1: \$1,995.91 − \$1,610.46 = \$385.45. Year 2: \$1,995.91 − \$1,798.65 = \$197.26.

(c) Total escrowed at closing. Year 1: \$385.45 × 12 = **\$4,625.40. Year 2: \$197.26 × 12 = **\$2,367.12. Total: \$4,625.40 + \$2,367.12 = \$6,992.52.

(d) As a percentage of the loan. \$6,992.52 ÷ \$300,000 = 2.331% — about 2.33 points, which is a very large concession and worth stating to the buyer in exactly those terms so they understand the size of what is being given.

(e) The qualifying payment is \$1,995.91, the P&I at the note rate of 7.000%. The subsidy is temporary and the obligation is not, so the underwriter runs the ratios at the payment the borrower will be making in year three. A 2-1 buydown does not qualify anyone for anything they could not otherwise qualify for.

Exercise 13.19 †

(a) Fully indexed rate. Index 4.25% + margin 2.50% = 6.750%.

(b) Qualifying rate and payment. Under Ability-to-Repay, the greater of the fully indexed rate (6.750%) and the introductory rate (6.000%) is 6.750%, so the qualifying P&I is \$2,594.40 — not the \$2,398.20 the borrower would actually pay. The ARM qualifies this borrower at a higher payment than a fixed loan at 6.000% would.

(c) Rate ceilings. Caps of 5/1/5: the first adjustment may move the rate by up to 5.000 percentage points, to 11.000%. The lifetime cap is 5.000 percentage points over the initial rate, which is also 11.000%. On this cap structure the first adjustment can take the loan all the way to its lifetime maximum in a single step — a structure worth pointing at explicitly, because borrowers assume the caps are a staircase.

(d) Worst-case increase over the initial payment. \$3,809.29 − \$2,398.20 = \$1,411.09 per month.

(e) Two sentences, worst case first:

"Before I tell you what this saves, here's the ceiling: the note lets this payment go to \$3,809.29 — that's \$1,411.09 a month more than you'd start at, and because of how the caps are written, it could get there at the very first adjustment rather than in stages. Against that, it starts \$196.20 a month below the fixed loan, it has to be approved using a payment of \$2,594.40 anyway, and the only question that matters is whether you'll still own this loan in seven years."

(Initial saving vs. a 6.750% fixed: \$2,594.40 − \$2,398.20 = \$196.20.)

Exercise 13.20 †

Step 1 — fundability.

Conventional 5%: \$248,000 × 0.05 = **\$12,400 + \$6,400 costs = **\$18,800 required. Against \$14,200 verified: **short \$4,600. NOT fundable.**

FHA 3.5%: \$248,000 × 0.035 = **\$8,680 + \$6,400 costs = **\$15,080 required. Against \$14,200 verified: **short \$880. NOT fundable as stated either.**

Step 2 — what that means. Neither column funds. The analysis is not a program question yet; it is a cash question, and saying so plainly is the correct first move. Do not present a program recommendation on top of a shortfall.

Step 3 — the structural fixes, in order of what is inside your control:

• A lender credit. This is the answer available entirely within the loan. Moving up the rate sheet produces a credit that covers the \$880 gap and can build a reserve position on top of it, at the cost of a permanently higher rate. Given a 648 score and a thin cash position, this is very likely the right instrument.
• A seller concession toward closing costs, negotiated in the contract.
• A documented gift, if a family member is available and willing.
• Any state or county down-payment assistance for which the borrower qualifies — the Harlow Street pattern.

Step 4 — the program recommendation, contingent on the cash being solved: FHA. Question 2 decides it independently of the cash. A 648 representative score sits below where conventional mortgage insurance prices competitively and at or below many lenders' conventional overlay floors, and FHA's premium does not vary with score at all. Even if the borrower found \$4,600 tomorrow, FHA would remain the indicated program on credit.

Two facts needed before finalizing: (1) the horizon — a long horizon makes the life-of-loan MIP expensive and argues for a plan to refinance to conventional once equity and score permit; (2) whether a seller concession is achievable in this market, which determines whether the fix is a concession or a lender credit.

What would change the answer: a materially higher representative score combined with enough additional verified funds to make the conventional column fund with real reserves. Both, not either.

Exercise 13.21

Recommend the 7.000% row — the largest lender credit, \$2,743.12.

At 7.000% the payment is \$61.09 a month above par. Over their stated 26-month horizon:

• Extra payments: \$61.09 × 26 = **\$1,588.34**
• Credit received at closing: \$2,743.12
• Net ahead of par: \$2,743.12 − \$1,588.34 = \$1,154.78

Compare the 6.875% row: \$30.47 × 26 = \$792.22 of extra payments against a \$1,371.56 credit, for \$579.34 net ahead. Both beat par; the deeper credit wins at this horizon because the credit repays in about 44.9 months and 26 is comfortably short of it.

Buying points here would be actively harmful: the half point costs \$1,828.75 and returns \$30.31 × 26 = \$788.06, a **\$1,040.69 loss**.

The caveat to state out loud: the credit must have costs to absorb it. If this borrower's total closing costs and prepaids are less than \$2,743.12, reprice down a row rather than leaving money that cannot be applied.

Exercise 13.23

What you would need, for each column:

Temporary buydown. The note-rate P&I on \$380,000 and the P&I at two rates below it (note − 2.000% and note − 1.000%). Escrow = (year-1 monthly difference × 12) + (year-2 monthly difference × 12). Then check the answer against the \$12,000 available — a 2-1 on a \$380,000 loan will generally cost somewhat less than \$12,000, leaving a remainder to apply elsewhere, while a 3-2-1 will cost more. Also required: confirmation that the program permits a temporary buydown from this funding source, and the qualifying payment (the note rate).

Permanent buydown. The rate sheet's cost in points for each row below par, and the P&I at each. \$12,000 ÷ \$380,000 = 3.158 points — likely deeper than a typical sheet publishes, so the practical question is how far down the grid \$12,000 reaches and what the incremental break-even is at each step.

Price reduction. A revised price of \$400,000 − \$12,000 = \$388,000, then recompute the down payment (\$12,000 × the down-payment percentage less), the loan amount, the P&I, and the mortgage insurance — all of which fall. Also required: seller agreement and a contract amendment, and awareness that the recorded price affects comparable sales.

The single fact that ranks them: the borrower's horizon. Short horizon ranks the temporary buydown first, because all of its relief lands inside the horizon. Long horizon ranks the permanent buydown and the price reduction first, because their smaller monthly relief never stops.

Exercise 13.25

Delete the 10%-down and 20%-down columns.

10% down: \$390,000 × 0.10 = **\$39,000 — more than the borrower's entire \$31,000 before a dollar of closing costs. 20% down: \$390,000 × 0.20 = **\$78,000 — not remotely close.

Both surviving columns fund. Conventional 5%: \$19,500 down plus roughly \$6,200 of costs and prepaids ≈ \$25,700, leaving about \$5,300. FHA 3.5%: \$13,650 down plus roughly the same costs ≈ \$19,850, leaving about \$11,150.

The sentence to add to the page:

"A 10% and a 20% down payment are shown here only so you can see where they sit: at \$39,000 and \$78,000 they are above your verified funds of \$31,000, so they are not choices on this purchase. Nothing below this line assumes them."

Better still, remove the columns entirely and put that sentence where they were. A structure the borrower cannot fund is not a structure, and leaving it on the page invites them to feel they gave something up.

Exercise 13.27

Two things wrong with "2-1 BUYDOWN — QUALIFY AT 4.625%!":

1. The borrower is not qualified at 4.625%. They are qualified at the note rate, which is unchanged by the buydown. The subsidy is temporary and the thirty-year obligation is not, so the underwriter runs the ratios at the payment that begins in year three. The flyer states the opposite of the rule.

2. "Buydown" here implies a rate the borrower holds, and they do not hold it. The note rate never changes. 4.625% is a subsidized payment for twelve months, funded out of an escrow account somebody else deposited at closing — not an interest rate the borrower has been given.

What they are actually qualified at: the P&I at the note rate, plus taxes, insurance, and mortgage insurance, run against their income and debts. If the note rate is 6.625% on the Linden Street loan, that is \$2,341.94 of P&I and a total housing payment of \$3,033.72 — not the \$1,880.47 and \$2,572.25 they will pay in year one.

Say it to the buyer without insulting the builder: "That first-year number is real and it's a genuine concession. The part the flyer leaves out is that you get approved on the year-three payment, which is \$499.54 higher — let me show you both, and then let me show you two other things that same money could have done."

Exercise 13.29 †

This is a writing task; the grader is the rubric. A model page and presentation:

THE PAGE. Two columns, side by side, on one sheet. Identical assumptions stated at the top: same price, same closing date, same taxes and insurance, same lock period. Four rows of numbers in the same units for both columns: total monthly payment; total cash to close; reserves remaining after closing, stated in months; total cost of credit over a stated horizon. Beneath them, one line in the same type size as everything else: "This comparison assumes you keep this loan [N] years. If that is wrong, tell me — it changes the answer."

THE PRESENTATION. Walk the rows in that order. Then:

"Option B has the lower payment by \$X a month and needs \$Y less at closing, and it leaves you Z months of cushion instead of W. Option A costs \$V less over the horizon you described, because its mortgage insurance comes off and B's does not. My recommendation is Option A, for that one reason. What would change my recommendation is if you told me you might not be here past [N] years, or if the smaller cushion in Option A worries you — that second one is a judgment about how you sleep, not an arithmetic error, and it is a completely legitimate reason to take B. It's your decision. Take tonight if you want it."

What demonstrates to a compliance reviewer that you did not steer — three sentences:

1. Both structures appear on one page in the same units and the same type size, including every number that argues against the recommendation.
2. The recommendation is stated with its reason and with an explicit reversal condition, which shows the recommendation was derived from the borrower's stated facts rather than from anything else.
3. The decision is documented as the borrower's, in a written confirmation sent the same day that names the option declined and the number that made it close.

What must be absent, and this is the whole exercise: any mention, weighting, or trace of the compensation difference. It is not a permissible input, it does not appear on the page, and if it influenced the ordering of the columns or the choice of horizon assumption, the page is a violation no matter how good the arithmetic is.

Exercise 13.31

Both — and it starts as a craft problem that becomes a compliance problem.

As craft, it is straightforwardly wrong. "Always show the 2-1 buydown first" is a rule about the originator's convenience, not about the borrower's file. It skips §13.8's three-column comparison and therefore skips the only analysis that determines whether a temporary buydown is the right shape of relief for this household. On a long-horizon borrower it quietly costs them thousands of dollars, and Case Study 13.2 is what that looks like from the inside.

It becomes a compliance problem the moment the ordering is driven by economics. If the 2-1 buydown is shown first because it closes faster, or because the affiliated lender's economics are better, or because the branch's compensation differs across structures, then an instruction to present it first is an instruction to tilt the field — and Regulation Z's anti-steering provision is aimed at exactly that tilt. Note that the manager's stated reason, "it's the easiest yes," is a statement about conversion rate, which is a statement about the originator's interest.

It is also a training failure, and worth naming as one. A new originator given this instruction will not learn to run the comparison, will not develop the judgment to know when the buydown is right, and will be defenseless the first time a borrower asks a good question.

What to say: "I'll show it — it's a real concession and sometimes it's the best answer. I'm going to show it next to a permanent buydown and a price reduction, because that takes fifteen minutes and I don't know which one fits until I've asked how long they're staying." That is not insubordination. It is the job.

Exercise 13.33

B — 6.750%. Fully indexed rate = index 4.00% + margin 2.75% = 6.750%. Under Ability-to-Repay the qualifying payment is computed at the greater of the introductory rate (5.500%) and the fully indexed rate (6.750%), on a fully amortizing schedule. A is the introductory rate and the most common wrong answer; C is the index alone; D confuses the qualifying rule with the cap structure.

Exercise 13.35

C — \$4,200.** A point is 1% of the **loan amount**: 1.500 × 1% × \$280,000 = 0.015 × \$280,000 = \$4,200**. A is 0.5 points, B is 1.0 point, D is off by a factor of ten. The trap is a candidate who computes on the purchase price or the down payment.

Exercise 13.37

B — FHA charges an upfront premium that may be financed into the loan.

A is false and is the most useful distractor in the set: FHA's premium does not vary with credit score, which is precisely why FHA competes better as scores fall. C is false — above 90% LTV on a 30-year term the annual premium currently runs the life of the loan; the 78%-of-original-value automatic termination is the conventional rule under the Homeowners Protection Act. D is false; FHA requires mortgage insurance regardless of LTV. Verify current MIP factors and duration rules in HUD Handbook 4000.1.

Exercise 13.39 †

Linden Street, with verified funds of \$31,500.00 instead of \$38,000.00.

(a) Fundability.

Arithmetic: \$31,500.00 − \$25,376.34 = \$6,123.66; ÷ \$3,033.72 = 2.02 months. Removing the \$1,828.75 point drops cash to close to \$23,547.59, leaving \$7,952.41 against a par payment of \$3,064.03 (\$2,372.25 + \$691.78) = **2.60 months**. FHA: \$31,500.00 − \$19,601.34 = \$11,898.66; ÷ \$3,015.84 = 3.95 months.

(b) Feasibility or convenience? Still convenience — both programs fund. But the margin has collapsed, and the honest characterization changes: at \$38,000 the choice was between a comfortable cushion and a slightly more comfortable one; at \$31,500 it is between a thin cushion and an adequate one.

(c) The recommendation, in the chapter's five-step order.

1. Present every fundable structure. Three now, not two: conventional with the point, conventional at par, and FHA. The 10% column is deleted and the deletion is explained. Adding the par column is the single most important change to the analysis — the first move when cash gets tight is to stop spending it on discretionary items, and a discount point is discretionary.
2. Identical assumptions across all three.
3. Quantify in the same four units. The reserve line now does most of the work: 2.02 / 2.60 / 3.95 months.
4. Name the load-bearing assumption. Still the horizon — and now a second one: their tolerance for a cushion under three months.
5. Recommend: conventional at par, 6.750%, no point. Reasoning: dropping the point buys back 0.58 months of reserves for free, and conventional at par still costs \$16,299.34 less than FHA over thirty years (\$512,478.86 against \$528,778.20). Then name the reversal condition and mean it: if 2.60 months feels too thin to them, FHA at 3.95 months is a legitimate choice and the \$16,299.34 is what it costs. That is a real trade, not a mistake.

(Conventional at par total cost of credit: \$0 points + \$488,260.00 interest (\$2,372.25 × 360 − \$365,750.00) + \$24,218.86 MI = \$512,478.86. FHA unchanged at \$528,778.20.)

(d) What this proves, in one sentence:

Program selection is a function of the borrower's balance sheet, not of the programs — change one number on a bank statement and the same house, the same rate sheet, and the same borrowers produce a different recommendation, a different rate, and a genuinely close call where there had been none.

(e) The workbook entry under the revised facts:

STRUCTURE DECISIONS - 4412 Linden Street - revised funds $31,500.00
  1. Program ............ Conventional      (both fund; 706 score prices well)
  2. Down payment ....... 5% = $19,250.00   (10% needs $38,500 - unavailable)
  3. Term ............... 30-year fixed     (qualifying ratios; 42.66% back-end)
  4. Rate / points ...... 6.750% PAR        (point dropped: buys 0.58 mo reserves)
  5. MI structure ....... BPMI monthly      (terminates payment ~137 vs. never)
  6. Concessions ........ none              (no seller contribution negotiated)
  DECIDED BY: the borrowers, on the comparison, day 15.
  RECOMMENDATION REVERSED IF: reserves under 3 months are unacceptable to them.

Chapter 14

Worked solutions to the daggered (†) and odd-numbered exercises. Where a solution turns on a guideline value, the answer given is the reasoning, and the student is expected to have looked the value up and recorded the date — that is the graded behavior.

Exercise 14.1

The Fannie Mae Selling Guide and the Freddie Mac Single-Family Seller/Servicer Guide. Each is a set of terms on which that specific purchaser will buy a loan — they are contracts, not statutes. Meeting them makes a loan saleable to that buyer; it does not make the loan legal, and it does not oblige any lender to approve it.

Exercise 14.3

The sentence is necessarily false whenever the obstacle was an eligibility failure. Compensating factors operate only on creditworthiness — gate two. If the property type, the project, the occupancy, the loan purpose, the product, or the loan amount put the file outside what the investor buys, no documented strength reaches it. The only true version of the sentence is "compensating factors got that file through a creditworthiness concern."

Exercise 14.5

1. The event type — Chapter 7, Chapter 13, foreclosure, deed-in-lieu, short sale, and mortgage charge-off are different events with different periods.
2. The measuring date — a specific, documented date, not the year the trouble began.
3. The ending date — commonly the disbursement or note date of the new loan, not the application date.
4. The exception path — documented extenuating circumstances, which shortens most of these and which is a defined standard, not a sympathy standard.

Exercise 14.7

Layered risk means risk factors compound rather than add. The distinguishing test: a student who says "several problems" is describing a count; layered risk is about interaction. Low equity means selling the house does not fix a problem. A tight ratio means a smaller shock is enough to cause one. Thin reserves mean nothing absorbs the shock. Each factor increases the consequence of every other factor, so the combination is categorically different from any single member of it — not four small problems but one larger one.

Exercise 14.9

Any three of: misrepresentation, misstatement, and omission; specified data inaccuracies; clear title and first-lien enforceability; compliance with the enterprise's charter requirements; compliance with certain laws, including high-cost and responsible-lending requirements; unacceptable mortgage products.

The one that should be first on a loan officer's list is misrepresentation, because it is the only one on the list that ordinary origination conduct can create. A loan can perform flawlessly for a decade and the exposure is still live. This is the specific reason an underwriter who is relaxed about a judgment call becomes immovable about a document.

Exercise 14.11 †

The graded skill is the reasoning, not the label. Boundary cases should be argued, not guessed.

The pattern to make explicit: items with a category in them (property, project, occupancy, purpose, product, amount, borrower status, a waiting period not yet satisfied) are eligibility. Items with a degree in them (score above the floor, ratio below the cap, reserves above the requirement, history) are creditworthiness.

Exercise 14.13 †

Three sentences, roughly:

What was wasted: two pages of writing, the underwriter's reading time, and — far more expensively — the days between the decline and the moment somebody realized the letter could not work, because a property-type exclusion is an eligibility failure and compensating factors have no bearing on eligibility. What should have been done in the same time: one call to establish whether the exclusion is the agency's or the lender's, and if it is the lender's, one call to a second source that does not carry it. The general rule: before you argue a file, classify the decline — if it is gate one, the only moves are a different investor, a different product, or a different property, and every minute spent on borrower quality is a minute the borrower does not have.

Credit students who also note that the borrower should have been told within the hour.

Exercise 14.14 †

Flags

1. Loan-to-value 90.00% — thin equity; mortgage insurance required, so a separate MI approval.
2. Representative score 688 — adequate, not strong.
3. Back-end ratio 37.85% — moderate. Not alarming on its own.
4. First-time buyer — no mortgage payment history to observe.
5. Payment shock 1.59× / +58.74% — \$1,425.00 to \$2,262.00, an extra \$837.00 every month.
6. 22 months with the current employer — short tenure at this job.
7. One 30-day consumer late 19 months ago — minor, but it is a data point, not nothing.

Offsets, each paired to the flag it actually engages

Unengaged flags. Two, and students should find both:

• Flag 4, first-time buyer. There is no verified housing history in this file. The Linden Street file had 36 months of documented on-time rent; this one has a rent amount and nothing proving it was paid. Getting a verification of rent is the single cheapest thing anyone could do to this file.
• Flag 1 combined with flag 2 — 90% loan-to-value with a 688 score has no offset here, exactly as 95%/706 had none on Linden Street. The only cures are more down payment or a better score.

The required closing sentence: the flag with no offset is the loan-to-value/score pairing, and the flag with an available-but-missing offset is the first-time-buyer flag.

Exercise 14.15

$\text{Housing ratio} = \$2{,}262.00 \div \$7{,}800.00 = 29.00\%$ $\text{Back-end} = (\$2{,}262.00 + \$690.00) \div \$7{,}800.00 = \$2{,}952.00 \div \$7{,}800.00 = 37.85\%$ $\text{Reserves} = \$9{,}800.00 \div \$2{,}262.00 = 4.33 \text{ months}$ $\text{Payment shock} = \$2{,}262.00 \div \$1{,}425.00 = 1.59\times$ $\text{Increase} = (\$2{,}262.00 - \$1{,}425.00) \div \$1{,}425.00 = \$837.00 \div \$1{,}425.00 = 58.74\%$ Full credit requires the numerator and denominator shown for each, per §2.1 of the house style.

Exercise 14.16 †

There is no single correct change; there is a correct argument. The strongest answers change loan-to-value or add the verification of rent, and the best answers explain why one of those beats the more obvious candidates.

Why loan-to-value. Dropping from 90% to a lower tier does four things at once: it removes the thin-equity flag, it changes or removes the mortgage insurance requirement and therefore removes the third approval entirely, it lowers the payment (which lowers both ratios and reduces the payment shock), and it improves pricing. One change, four layers touched. That is the compounding argument applied correctly — you are not subtracting a unit of risk, you are dividing the stack.

Why the verification of rent instead. It is nearly free. It costs a form and a few days, requires no money from the borrower, and it converts the completely unoffset first-time-buyer flag into an offset one. On a cost-per-layer basis it is the best move in the file.

Why the obvious answers are weaker. Paying off the \$690.00 of other debt improves the back-end ratio — which was already the least alarming number in the file at 37.85% — and consumes reserves, which were doing real work against payment shock. That is the §14.5 trade made backwards: spending the strongest offset to improve the weakest flag. Students who propose it should be asked to compute the resulting reserves before defending it.

Exercise 14.17

Three factors at 40%:

$\text{additive} = 1.00 + 3(0.40) = 2.20 \qquad \text{compound} = 1.40^3 = 2.74$ Four factors at 25%:

$\text{additive} = 1.00 + 4(0.25) = 2.00 \qquad \text{compound} = 1.25^4 = 2.44$ (For reference, the chapter's four factors at 40%: additive 2.60, compound 3.84.)

What it tells you. Three severe layers (2.74) beat four mild ones (2.44) — and the gap between additive and compound thinking is also larger for the severe set (0.54 versus 0.44). Severity compounds harder than count. The practical consequence, which is the point of the exercise, is that a file with one badly weak characteristic and one moderate one can be riskier than a file with four mildly imperfect ones, even though the second file has a longer list of flags. A loan officer counting items on a list is measuring the wrong thing.

Exercise 14.19 †

The meta-lesson to draw out: in half these cases the statement itself does not tell you, and the speaker frequently does not know either. The diagnostic question is not skepticism about the underwriter; it is a request for information the underwriter has and you do not.

Exercise 14.21 †

Four things, in order:

1. Get the decline in writing and confirm the specific rule. Not "insufficient reserves" but the requirement, the threshold, and the source. You cannot request an exception to a rule you have paraphrased.
2. Find out who owns the exception and what they need to see. Ask the underwriter, or the underwriting manager, or credit policy. Learn the process before you use it — who decides, what form, what turnaround, and what they have granted before.
3. Assemble the memo before you make any promises — the six-part structure from §14.9, with a document named for every offset and the residual risk stated honestly in section 5.
4. Then call the borrower.

Steps 1 through 3 happen before the borrower call. That is the graded part of the answer. The reason is not that the borrower cannot handle bad news; it is that a call which says "we were declined and I'm working on it, I'll know by Thursday" is a call you can only make once you know what you are working on and when you will know. A call made before step 1 produces a borrower who calls you every two hours because you gave them nothing to hold.

Full credit also notes: if the exception is denied, the borrower is entitled to a decision and a notice through the proper process, and the loan officer's next move is to tell them plainly that this is the lender's rule rather than the agency's, and that another lender may reach a different answer.

Exercise 14.23

Everything about the instruction is wrong. The application must be complete and accurate; omitting a source of income to avoid explaining it is a misstatement in a file that will be sold with a representation that it is accurate — and misrepresentation is a life-of-loan exclusion that never ages off (§14.10). It is also, potentially, a violation of law rather than merely of policy, and Chapter 27 covers the criminal exposure. There is no "the file works without it" defense, because the representation is about what the file says, not about whether the loan would have been approved anyway.

What you do: decline the instruction, document the income in the file, and if the instruction was serious rather than careless, escalate it through your compliance channel. Students should also note the smaller point — a second job is not a problem to be hidden. It is either usable income or it is not, and the guide has an answer either way.

Exercise 14.25

Five questions, in order:

1. Which chapter — 7 or 13? They are different events with different periods.
2. Was it discharged, or dismissed? Different dates, different consequences, and borrowers use the words interchangeably.
3. What is the exact date of the discharge or dismissal order? Not the filing date, not the year.
4. Has there been more than one filing in the past seven years? Multiple filings carry a longer period.
5. Was there a mortgage in it, and what happened to the property afterward? If a mortgage debt was included and the property was foreclosed later, there is a specific rule about which clock governs and it must be looked up.

Then: ask for the document, and say nothing about timing until you have it.

Exercise 14.26 †

The two points behind the table. First, in every case the authoritative record is a recorded or court-issued document, not the credit report — Chapter 10 established that dates on old derogatory tradelines are frequently unreliable, and this is where that unreliability is most expensive. Second, in (c), (d), and (e) the operative date is the date title transferred, which is very often months or even years after the borrower moved out. Borrowers date these events from when they left the house. That gap has cost real files.

Exercise 14.27

The problem: the waiting period commonly runs to the disbursement or note date, not the application date. A closing on the 9th does not clear a period satisfied on the 14th, so as scheduled this loan cannot fund. This is not a condition anyone can clear and not an exception anyone can grant; the calendar is the guideline.

The two sentences to the agent, today:

"I need the closing moved to the 15th or later — not for a document, for a date. There's a requirement on this file that is satisfied on the 14th and there is no version of this loan that funds before it, so if the seller can't move six days, we need to know that this week rather than the day before closing."

Full credit requires that the call happens today, that the loan officer does not describe it as a paperwork delay, and that the loan officer proposes the specific new date rather than asking for "more time."

Exercise 14.29

Graded on process, not on the values retrieved. A complete answer records, for each guide: the value, the topic identifier or section, the effective date shown, and the date the student looked. The final sentence must state whether the two agencies agree.

The learning outcome is the discovery that (a) the guides are free and take about ninety seconds to search, (b) the value has qualifiers attached that nobody quotes, and (c) the two guides frequently state the same idea differently — which is the whole reason §14.1 insists there are two rulebooks rather than one thing called "agency guidelines."

Exercise 14.30 †

Compensating factor inventory — the Linden Street file, submitted manually

Struck out — requirements offered as strengths:

• ~~"Both borrowers have documented, stable W-2 employment."~~ Three and four years of stable employment is close to what the guide already required to count this income at all. It is a qualifying condition, not a compensating one. It may be mentioned as context; it may not be counted as an offset. (Students who keep the four-year tenure specifically, arguing it exceeds the requirement, are making a defensible argument — accept it if they say why.)
• ~~"The gift is properly sourced and documented."~~ Correct sourcing is a requirement (Ch. 12). A properly documented gift is not a strength; an improperly documented one is a defect.
• ~~"They have good credit scores."~~ Already counted. The score is the credit evaluation.

The discipline being taught: every line must survive three questions — is it documented, is it aimed at a specific flag, and is it something the file was already required to have? A factor that fails the third question is double-counting, and offering one tells the underwriter you do not understand the requirement.

Exercise 14.31

Ranking, most to least useful:

1. (a) Eleven months of verified reserves. Documented, quantified, and aimed at almost every weakness a file can have. This is the strongest single item on the list.
2. (b) Payment rising only \$80 a month. Minimal payment shock, and it is measurable. Especially strong when paired with evidence the borrower has been saving the difference.
3. (c) Nine years with the same employer. Real and documentable, and it exceeds any ordinary history requirement — which is what keeps it from being double-counting.
4. (e) "The house appraised for \$12,000 over contract." Worth nothing as a compensating factor. Loan-to-value is computed on the lesser of price or appraised value (Ch. 4), so an appraisal over contract changes neither the loan-to-value nor the equity in the transaction. It is a fine thing to tell the borrower and an irrelevant thing to tell the underwriter.
5. (d) "They're going to get a bonus in March." Worth nothing. Future income is not qualifying income. It is not documented, not verified, and not certain, and offering it invites the underwriter to wonder what else in the file is aspirational.

The two at the bottom fail for different reasons and students should say which is which: (e) fails because it is not relevant to any weakness; (d) fails because it is not documentable.

Exercise 14.32 †

First, the arithmetic the student must do before writing.

$\text{Total obligations} = \$2{,}530.00 + \$1{,}780.00 = \$4{,}310.00$ $\text{Back-end ratio} = \$4{,}310.00 \div \$9{,}240.00 = 46.65\%$ $\text{Overlay cap of }45\%: \quad 0.45 \times \$9{,}240.00 = \$4{,}158.00$ $\text{Over by } \$4{,}310.00 - \$4{,}158.00 = \$152.00 \text{ per month, or } 1.65 \text{ points}$ Supporting figures the memo needs:

$\text{Housing ratio} = \$2{,}530.00 \div \$9{,}240.00 = 27.38\%$ $\text{Reserves} = \$27{,}900.00 \div \$2{,}530.00 = 11.03 \text{ months}$ $\text{Payment shock} = \$2{,}530.00 \div \$2{,}410.00 = 1.05\times, \text{ an increase of } \$120.00 = 4.98\%$ A model memo (approximately 330 words):

1. The ask. Requesting an exception to our 45% back-end DTI overlay to allow 46.65% on this file — 1.65 points over, \$152.00 a month.

2. Which rule. This is our overlay, not an agency guideline. The agency maximum for this transaction is higher; the 45% cap applies here because the representative score of 731 is under our 740 threshold.

3. The weakness. The back-end ratio is 46.65% — \$4,310.00 of obligations against \$9,240.00 of qualifying income, of which \$1,780.00 is non-housing debt. That is a real burden and the largest number on the file.

4. The offsets. - 11.03 months of reserves after closing, \$27,900.00 verified (bank statements, pp. 2–9). - Payment shock of 1.05× — \$120.00 a month above current rent of \$2,410.00 (verification of rent; lease). The household is already carrying essentially this housing payment. - 84 months of mortgage history on the prior home with no lates (credit report, tradeline 4; payoff statement). This is not a first-time buyer with a rent estimate. - 11 years with the same employer, W-2 salary, no variable component (written VOE dated ___). - Housing ratio 27.38%, well under any benchmark — the ratio pressure is consumer debt, not the house. - 85% loan-to-value, meaningful equity relative to the high-LTV segment the overlay was written for.

5. What the offsets do not cover. \$1,780.00 a month of consumer debt is a genuine ongoing obligation and the offsets do not reduce it. If any of it is revolving rather than installment, the ratio could deteriorate after closing without any new borrowing at all. I have not verified the amortization schedule on that debt and would not represent that it retires soon.

6. The ask again. Requesting credit policy's decision by Thursday the . Closing is scheduled for the ___ and the rate lock expires the . If the answer is no, I need to tell the borrower and the listing agent the same day.

Grade on section 5. A memo without it is incomplete regardless of how good sections 1–4 are. Grade also on section 3 appearing before section 4, and on every offset carrying a document reference.

Exercise 14.33

Any rewrite that names the rule, the number, and the ask. For example:

"Requesting an exception to the 45% DTI overlay to allow 46.65% on loan #____. This is our overlay, not an agency requirement. The weakness and the offsets are below, with document references."

What was removed and why: "I'm hoping" (not an ask), "some consideration" (not a number), "the borrowers are great" (not documentable, and it is the reader's job to decide that), and "I've known the agent for years" (irrelevant at best, and at worst it is asking for a decision on a basis that would be a fair-lending problem if the shop actually granted exceptions that way).

Exercise 14.35 †

What happens next. The delinquency triggers a file review. A reviewer with hindsight and no deadline re-underwrites the loan and finds that the overtime income was averaged over the wrong period and that the correct figure would have pushed the back-end ratio above the applicable maximum. That is a breach of the representation that the loan met the guidelines: the loan, as delivered, was not what the lender said it was. The investor may issue a repurchase demand, or in some circumstances negotiate an indemnification or make-whole payment. Chapter 23 works the mechanics.

Who bears the cost. The lender, not the loan officer and not the borrower. The lender buys the loan back at par — unpaid principal balance plus accrued interest and costs — on an asset that is currently sixty days delinquent and worth considerably less than par. The gap between those two numbers is the loss.

Does twenty-eight months help? Possibly, and the honest answer is it depends on the framework and on the facts. Under the representation and warranty framework, selling reps are relieved after a defined period of consecutive on-time payments — commonly cited as thirty-six months, with an alternative path that tolerates a bounded number of early delinquencies. Twenty-eight months is short of thirty-six, and the loan is now delinquent, so the payment-performance path is not available. Relief through a satisfactory quality-control review might have applied if the review had occurred and had been satisfactory — but this review found a defect, which is the opposite.

The part students should reach on their own: if the finding had instead been a misstatement rather than a calculation error, the analysis would not depend on twenty-eight months at all, because misrepresentation is a life-of-loan exclusion. The distinction between "we got the arithmetic wrong" and "the file said something untrue" is the most consequential distinction in §14.10.

Exercise 14.37

Something close to:

"Because they can argue about a judgment call and win, and they can't argue about a document at all. When this loan gets sold, our company promises in writing that the file met the guidelines. If it didn't, we can be made to buy it back — years later, usually after it's already gone bad, at full price. A judgment call they made in good faith is defensible. A missing page isn't a judgment; it's either there or it isn't, and 'the borrower told me' has never been a defense. So they'll flex on interpretation and they will not flex on evidence, and once you see it that way you'll stop taking it personally and start sending them the page."

Grade on whether the student explains the asymmetry — judgment is defensible, missing evidence is not — rather than merely restating that reps and warrants exist.

Exercise 14.39

What has changed: possibly nothing, and possibly something material, and the loan officer cannot tell from a casual remark. Two facts are potentially in play — whether the occupancy is still what the file represents (it very likely still is; family members living in an owner-occupied primary residence does not change occupancy), and whether "paying us some rent" describes rental income, a household contribution, or nothing at all.

What you are obligated to do: find out what is actually true, and make sure the file reflects it. That means asking directly, plainly, and without drama, and then documenting whatever the answer is. You may not know something material about the file and let the file say otherwise — that is exactly the situation §14.10 describes, and misrepresentation does not age off.

What you must not do: you must not coach the borrower toward an answer, and you must not decide on your own that it is too small to mention and move on. Those are the two failure modes, and the second one is far more common than the first because it feels like kindness.

How to handle the conversation: treat it as ordinary, because it is.

"Good — one thing I want to get right so nothing surprises us at the end. When you say they'd pay you some rent, is that a formal arrangement with an amount, or is it more that they'd chip in? I ask because those are two different things in the file and I'd rather know now than have it come up later. Either answer is completely fine."

Grade on: the student asks, the student does not lead the witness, the student documents the answer, and the student does not treat the borrower as a suspect. This is the chapter's borrower-dignity requirement and its documentation requirement in the same ninety seconds.

Exercise 14.40 †

The arithmetic first. A new obligation of \$611.00 a month:

$\$1{,}446.00 + \$611.00 = \$2{,}057.00 \text{ in other monthly debts}$ $\$3{,}033.72 + \$2{,}057.00 = \$5{,}090.72 \text{ total obligations}$ $\$5{,}090.72 \div \$10{,}500.00 = 48.48\% \quad \text{(up from 42.66\% — a 5.82-point jump)}$ Which rows changed?

The answer to "which changed most": row 3, by a wide margin and unambiguously.

The answer to the analytical question — and this is the graded part: four of the seven rows did not change numerically and got worse anyway. Reserves in months are identical. Payment shock is identical. The variable-income share is identical. The loan-to-value is identical. Every one of those offsets is now doing more work against a household with less room.

What day 44 proves. A file with layers and a file with margin are different things, and the ratios do not distinguish them. This file had 4.16 months of reserves, clean credit, verified rent, and a 28.89% housing ratio — genuinely good numbers — and one ordinary consumer decision, made by two people who had no idea it mattered, took the back-end ratio from comfortable to over. Layers do not just make a file riskier; they make it fragile, which is a different property. A file with margin absorbs a surprise. A file with layers converts a surprise into a crisis.

The resolution belongs to Chapter 19. The lesson belongs here: this is why you tell borrowers, on the day you take the application, not to open new credit — and why "don't buy furniture" is not a superstition but arithmetic.

Chapter 15

Worked solutions to the daggered (†) and odd-numbered exercises. Guideline figures used here are illustrative for the exercise; agency limits, ratio maximums, exclusion thresholds, and waiver criteria change and must be verified at the source.

Exercise 15.1

Automated underwriting system — software that evaluates a loan application plus a credit report against a published rulebook and a proprietary risk model, returning a recommendation, an eligibility assessment, and a documentation list.

Findings report — the document the system produces: recommendation, the loan data used, the underwriting analysis, the risk and eligibility assessments, verification messages, and observations.

Verification message — one numbered item in that report specifying a document that must be provided to support the recommendation.

Re-run — a subsequent submission on the same casefile; it changes the recommendation only if the inputs changed.

Exercise 15.3

Approve → Accept. Refer → Caution.

Exercise 15.5

The six parts: (1) summary and recommendation; (2) loan data — what you typed; (3) underwriting analysis — income, ratios, assets, reserves; (4) risk and eligibility; (5) verification messages; (6) observations and lender notes.

Reading order: 2 and 3 first, because they are the only pages you control and they are an audit of your own data entry. 5 second, because it is the only page that creates work and the slow items need to be ordered today. 4 third, read for the reason rather than the result. 1 last, because the recommendation is a function of pages 2 and 3 — if those are wrong, page 1 is wrong, and reading it first only tells you what to feel about a number you have not checked.

Exercise 15.7 †

Model answer, roughly:

The agency that will buy the loan maintains enough data about American residential property — prior appraisals submitted to its systems, public records, and its own valuation models — that on some transactions it will tell the lender it does not require a new appraisal, and will accept the value stated in the application. The offer is made per casefile inside the findings, its eligibility criteria are set by the agency and revised frequently, and it is a statement about what the agency requires in order to buy the loan — not an opinion that the property is worth the contract price, and not a substitute for a home inspection.

Nothing in that paragraph depends on which name is current.

Exercise 15.9 †

Knowable: the direction of most factors (a higher score, lower ratios, more reserves, and a lower LTV all help), and every published eligibility parameter, which can be looked up in the guides. Also knowable from experience: which factors tend to be binding on which kinds of files.

Not knowable: the weights, the interaction terms, the thresholds, and the exact trade-off rate between any two factors. The GSEs and HUD do not publish model internals. The claim "20 points of score is worth about a point and a half of DTI" asserts a precise exchange rate that is not published anywhere, and even if it were approximately right on one segment of files it would not be stable across the population.

How to talk about it: describe direction and structure, never a rate. "More reserves help, and on a file like this one I'd expect them to matter" is honest. "Get me 20 more points and I can carry 1.5% more DTI" is not, and it will eventually be quoted back to you by a borrower who reorganized their finances on it.

Exercise 15.11

Model answer:

"We have an Approve/Eligible, which is exactly what we wanted and is not the same as approved. What it means is that the automated system read the file we submitted and said this loan fits Fannie Mae's guidelines if we document a specific list of items — which I've already sent you. Our underwriter approves the loan once those documents are in and they support what we said, and I expect that around day 23."

Exercise 15.12 †

Model answer, under forty-five seconds:

"It's not a decline — nobody has declined anything. FHA's scoring system looked at the file and said it isn't going to sign off on its own, so a human underwriter reviews it instead, against HUD's written rules and the strengths in the file. That's slower and it's more paperwork and it works. I'll know within a couple of days whether it's a real problem or just a longer road. What I need from you is not to tell the sellers this is dead, because it isn't."

The distinction being tested: a Refer is a routing decision about who underwrites the file, not a credit decision about whether it can be approved.

Exercise 15.13

Agree in part. Running a Refer file through the other agency's system is legitimate: both rulebooks are published, both agencies buy loans, and a lender approved to sell to Freddie Mac is entitled to underwrite to Freddie Mac's guide. What makes it legitimate is that the data does not change — the same facts are submitted to a second published rulebook.

What would make it improper: changing the facts. Entering income that cannot be documented, omitting a known debt, recoding occupancy or property type, or adjusting figures until something approves is misrepresentation regardless of which system you submit it to (Chapter 27).

Two practical constraints: your employer may not deliver to both agencies or may have a policy on which system runs first, and the lender's overlays apply to whichever answer you get.

Exercise 15.15

Housing ratio: \$1,980.00 ÷ \$7,200.00 = 27.50% Total obligations: \$1,980.00 + \$690.00 = \$2,670.00 Back-end ratio: \$2,670.00 ÷ \$7,200.00 = 37.08%

In plain language: about twenty-eight cents of every pre-tax dollar goes to the house, and about thirty-seven cents of every pre-tax dollar is already committed before groceries.

Exercise 15.16 †

(a) Front: \$1,721.57 ÷ \$4,150.00 = 41.48% (41.4836%). Back: (\$1,721.57 + \$395.00) ÷ \$4,150.00 = \$2,116.57 ÷ \$4,150.00 = 51.00% (51.0017%).

(b) Maximum housing at 31%: \$4,150.00 × 0.31 = **\$1,286.50. Maximum total obligations at 43%: \$4,150.00 × 0.43 = **\$1,784.50.

(c) Housing exceeds the benchmark by \$1,721.57 − \$1,286.50 = \$435.07/month. Total obligations exceed it by \$2,116.57 − \$1,784.50 = \$332.07/month.

(d) The benchmark asks two independent yes-or-no questions and this borrower fails both. The scorecard evaluates the characteristics together — the credit history behind the 641, the modest \$395.00 of existing debt, employment, the assistance structure — so strength in one dimension can genuinely offset a high ratio. A payment \$435.07 smaller is roughly a quarter less house, which in most markets is a different neighborhood.

Note for the instructor: the \$1,721.57 payment is built from components in Chapter 16 (P&I, annual MIP, taxes, and insurance). Students who work backwards from the published ratios will land within a few cents of it; accept anything that reproduces 41.48% and 51.00% at two decimals, but the canonical figure is \$1,721.57.

Exercise 15.17 †

Baseline: \$3,360.00 ÷ \$6,400.00 = 52.50%.

Most DTI improvement per dollar of borrower cash: (b), decisively. It is free. An installment debt with a small number of payments remaining — currently ten or fewer under Fannie Mae's guide, and not if the payment would materially affect the borrower's ability to pay in the months right after closing; verify the current rule — may generally be excluded. This is not a change the loan officer makes to the borrower's life; it is the correct application of a published rule that should have been applied the first time. It buys 6.17 percentage points of DTI and touches reserves not at all.

Note the shape of the answer: the best move on this file was never a restructure. It was reading the guide.

Exercise 15.18 †

(a) Income entered \$12,300.00 Housing: \$3,033.72 ÷ \$12,300.00 = 24.66% (true 28.89%) Back-end: \$4,479.72 ÷ \$12,300.00 = 36.42% (true 42.66%)

(b) Income entered \$10,050.00 Housing: \$3,033.72 ÷ \$10,050.00 = 30.19% (true 28.89%) Back-end: \$4,479.72 ÷ \$10,050.00 = 44.57% (true 42.66%)

(c) \$318.00 student loan omitted Obligations: \$4,479.72 − \$318.00 = \$4,161.72 Housing: unchanged at 28.89% Back-end: \$4,161.72 ÷ \$10,500.00 = 39.64% (true 42.66%)

(d) Who finds it, when, and what it costs

• (a) The underwriter, on about day 23, recalculating income from the W-2s and written verifications. On this file the loan survives because 42.66% is still inside the guideline — luck, not design. On a file whose true back-end is 51%, the cost is an approval issued on day 6, an appraisal fee, a title order, a borrower who gave notice to a landlord, and a decline around day 25.
• (b) Nobody, unless the loan officer reconciles the report against their own worksheet — because the error makes the file look worse, and a worse-looking file that still approves raises no flags. On a marginal file this typo produces a Refer, and the borrower is told they do not qualify when they do.
• (c) The AUS itself, at submission or on the next run, because it reconciles entered liabilities against the credit report and flags the mismatch. The dangerous version of this error is a debt that is not on the credit report — a family loan, a lease, a court-ordered obligation — where nothing catches it until the underwriter reads a bank statement and sees the payment leaving every month.

Exercise 15.19 †

(a) Loan amount over the limit. Failed half: the loan (Ineligible). LTV is fine: \$806,000 ÷ \$1,007,500 = 80.00%. Fix: reduce the loan to \$766,550. Down payment becomes \$1,007,500 − \$766,550 = \$240,950, versus \$201,500 today — **additional cash \$39,450.00**. Alternatives: a non-conforming (jumbo) product, or a first-plus-second structure if one is available and the CLTV works.

(b) LTV above the second-home maximum. Failed half: the loan. \$414,000 ÷ \$450,000 = 92.00%, against an assumed 90% maximum. Fix: loan \$405,000 at 90.00%, down payment \$45,000 versus \$36,000 — **additional cash \$9,000.00. Note what is not a fix: recoding occupancy to primary residence. That is fraud (Chapter 27).

(c) CLTV, not LTV. Failed half: the loan. First mortgage LTV is 80.00% and perfectly fine; the combined ratio is (\$240,000 + \$45,000) ÷ \$300,000 = 95.00%, against a 90% CLTV cap. Fix: reduce the second lien so total liens are at most \$270,000 — a **\$30,000 second instead of \$45,000, a reduction of **\$15,000 the borrower must replace from another source. Alternatively find a first-mortgage product whose CLTV cap accommodates the assistance program. Teaching point: read which ratio broke. A loan officer who sees "LTV" in the message and checks only the first mortgage will conclude the findings are wrong.

(d) Ineligible project. Failed half: the loan — property/project eligibility. No arithmetic available. Fixes, in order of likelihood: determine what specifically failed and whether it is curable through a different project review path; a different loan product or a lender with a portfolio option; FHA project approval if the project is or can be approved; or a different property. Get the actual deficiency from the findings and the project documents before promising anything.

(e) Program condition not met. Failed half: the loan. \$252,200 ÷ \$260,000 = 97.00% — inside the LTV cap, but the 97% product requires at least one first-time homebuyer and neither borrower qualifies. Fix: 5% down. Loan \$247,000 at 95.00% LTV, down payment \$13,000 versus \$7,800 — additional cash \$5,200.00. Alternatives: a 97% product without the first-time-buyer condition if one is available to your shop, or FHA at 96.5%.

Exercise 15.21 †

Model document request. Today is day 4 of a 40-day contract.

Items whose turn time you do not control: 3, 5, 8 (in part), 10 (in part), 11, 12 — call it six of thirteen, and they include every one of the slowest items on the list.

What that tells you: roughly half the critical path is outside your hands, so it must leave on day 4 and day 5 rather than after the borrower's documents arrive. The instinct to "collect the borrower's stuff first, then order the third-party items" is exactly backwards and is the single most common reason a 40-day contract needs an extension.

Also note item 6 and item 8 interact: the retirement loan is funding part of the down payment, so the deposit will land in the depository account and will need to be traced. Ask for both at once.

Exercise 15.22 †

The errors (seven, of at least six required):

1. Occupancy — "Second home," should be primary residence. Changes LTV maximums, reserve requirements, and pricing, and would make the recommendation about a different loan entirely. Deliberate misstatement here is occupancy fraud (Chapter 27).
2. Property type — "1 unit, detached, site-built," should be an attached townhome in a planned unit development. Affects project/PUD review requirements, MI, and pricing.
3. Estimated value \$321,500 entered above the \$318,000 sales price with no appraisal. LTV runs on the lesser of price or value. True LTV: \$286,200 ÷ \$318,000 = 90.00%, not the 89.02% shown. The error will be corrected the moment an appraisal arrives, and it makes the file look better than it is in the meantime.
4. Income entered \$5,590.00 instead of \$5,950.00 — a transposition.
5. Student loan \$189.00/month omitted from the liabilities. It is on the credit report, so the system will likely flag a mismatch — but it was still submitted wrong.
6. Family loan \$150.00/month omitted. Disclosed by the borrower at application and not on the credit report, so nothing will catch it automatically. This is the dangerous one.
7. 401(k) balance of \$41,000 entered as a depository account. Inflates total verified assets from \$35,500 to \$76,500 and misstates reserves. Retirement assets are treated differently from liquid depository funds; entering one as the other is a documentation problem waiting to happen.

As entered: Other obligations \$362.00 + \$95.00 = \$457.00; total obligations \$1,914.00 + \$457.00 = \$2,371.00 Front: \$1,914.00 ÷ \$5,590.00 = 34.24% Back: \$2,371.00 ÷ \$5,590.00 = 42.42%

As they should be: Other obligations \$362.00 + \$189.00 + \$95.00 + \$150.00 = \$796.00; total obligations \$1,914.00 + \$796.00 = \$2,710.00 Front: \$1,914.00 ÷ \$5,950.00 = 32.17% Back: \$2,710.00 ÷ \$5,950.00 = 45.55%

The instructive part: the two errors run in opposite directions on the front-end ratio. The understated income makes the housing ratio look worse (34.24% versus a true 32.17%), while the omitted debts make the back-end look better (42.42% versus a true 45.55%). Neither ratio on the submitted file is right, and neither error would necessarily look wrong to someone glancing at the report. Only a line-by-line reconciliation against the fact sheet finds them — which is the entire argument of §15.9.

Exercise 15.23

Three things available only from the liabilities block:

1. Which debts the system actually counted, so you can identify one you entered that it dropped or one it picked up from the credit report that you did not enter.
2. The exact payment it used for each, which is where guideline judgments about revolving minimums, income-driven student loan payments, and lease payments become visible. If it used a payment different from yours, one of you applied the rule wrong.
3. The number of payments remaining on installment debts, which is the input to the exclusion question — and therefore the cheapest DTI relief available on many files (see 15.17).

Exercise 15.25 †

Model email, roughly 300 words:

Subject: Great news, and here's exactly what we need

Good news — the automated underwriting came back Approve/Eligible, which is the answer we wanted. One honest note about what that means: it is the system confirming that your loan fits the guidelines if we document a specific list of items. It's not the final approval, which comes from our underwriter after these documents are in. Nothing here is a surprise or a problem; this is the normal next step, and the list below came directly out of the report.

This week (by Friday): 1. Your most recent paystub, plus your last two W-2s — each of you 2. Confirmation of your four monthly payments: the two car loans, the student loans, and your credit card minimums. If there's anything you pay monthly that wouldn't show on a credit report, tell me now. 3. Your last two months of statements on both savings accounts — every page, including the blank ones

Early next week: 4. Your rental payment history for the last three years — cancelled checks, bank records, or a letter from your landlord 5. The gift letter for the \$10,000; I'm attaching the form. Your parents sign it. 6. Something showing the funds are available on their side — a bank statement is fine, and they can black out the account number.

Already handled by me: the appraisal and title are ordered tomorrow, and I've sent employment verifications to both of your employers today — those take the longest, which is why they go first. The insurance binder we'll do in about two weeks.

Send anything as you get it; don't wait to have it all. Call me with questions.

Exercise 15.27

Key elements the answer must contain: an accurate description of what the day-6 findings were (a recommendation that the loan fit the guidelines if documented, not an approval); an accurate statement of what changed between then and the decision; an accurate statement of who declined (the lender, not the software and not the agency); a reference to the adverse action notice and the specific reasons in it; and no blame directed at the borrower, the underwriter, or the system. If the loan officer contributed to the misunderstanding by saying "approved" on day 6, the letter should say so plainly.

Exercise 15.29

What is wrong with it, in three layers.

Factually: the AUS evaluates the income you enter. Entering \$400 of overtime that is not yet documented does not create \$400 of qualifying income; it creates a recommendation about a borrower who does not exist (§15.9).

Procedurally: qualifying income requires documented history and a reasonable expectation of continuance (Chapter 11). "We'll get the VOE later to back it up" inverts the order — you would be asserting a fact and then hoping evidence appears. If it does not, you have a submission history showing income that was never supported.

Legally: submitting data you know to be unsupported, to obtain a recommendation you would not otherwise get, on a loan that will be sold to an investor, is misrepresentation. Chapter 27 covers the exposure, which is criminal as well as professional, and it attaches to the person who entered the data.

What to say: "I'm not going to enter income we can't document. But let's find out whether it's real — if the overtime has a two-year history, we can order the VOE now, and if it comes back supporting an average, that's legitimate income and I'll enter it and re-run. Give me two days."

What to do instead: order the written verification, calculate the averaged figure the guide permits if the history exists, and re-run on documented facts. If it does not exist, work the DTI from the other side — the debt schedule, the structure, the program.

Exercise 15.30 †

The options. (1) Say nothing and hope the appraisal's property description does not trigger a review. (2) Correct the data, notify the processor and underwriter immediately, and re-run.

The honest one is (2), and it is also the only one that works, because a condominium is going to be identified as a condominium by the appraisal, the title commitment, and the insurance requirements. The question is not whether it is discovered; it is whether it is discovered from you on day 30 or from a document on day 38.

Who is told first: your processor and the underwriter, immediately and in writing, because they own the timeline consequences and the project review that now has to happen. Then your manager, because a re-run this late may affect the lock and the closing date. Then — with a plan already in hand, not before — the borrower and the agent.

What actually changes: project review requirements, possibly the mortgage insurance, possibly pricing, and possibly eligibility. The re-run may return Ineligible if the project does not meet review requirements, which is exactly the information everyone needs eleven days out rather than three.

The professional point: the cost of this error is entirely a function of how fast you report it. Nobody in this business has ever been fired for finding their own mistake on day 30. People are fired for the version where it surfaces on day 49.

Exercise 15.31

The honest conversation has three parts: (1) the file is eligible under the agency guideline; (2) your employer applies an additional requirement — an overlay — that this file does not meet, and your employer is the one making the loan; (3) other lenders set overlays differently, and a broker or another lender may be able to do this loan.

What makes a version of it dishonest: implying the borrower does not qualify at all when they qualify under the guideline and fail only your employer's overlay; implying the agency or the automated system declined them; or discouraging them from applying elsewhere. Regulation B's prohibition on discouraging applicants and the general obligation not to mislead both live here, and Chapter 25 covers them. Telling a borrower where else to look costs you a file and builds the referral business Chapter 38 quantifies.

Exercise 15.33

(a) 2. Approve addresses the borrower; Ineligible addresses the loan. Distinction tested: the recommendation has two independent halves.

(b) Loan Product Advisor (Freddie Mac). Distinction tested: DU says Refer, LPA says Caution.

(c) 2. Manual underwriting under HUD Handbook 4000.1. Distinction tested: a Refer is a routing decision, not a credit denial, and only a lender denies a loan.

(d) 3. No one — a lender approves loans. Distinction tested: an AUS returns a recommendation against published guidelines; the lender decides, subject to its overlays and its underwriter's review.

(e) 2. An identical recommendation. Distinction tested: the system is deterministic on a given engine version, so a re-run without a data change is a re-run without an effect.

Exercise 15.34 †

(a) Day-6 reconciliation

Supporting arithmetic: income \$6,300.00 + \$4,200.00 = \$10,500.00. Housing \$2,341.94 + \$385.00 + \$130.00 + \$176.78 = \$3,033.72. Obligations \$3,033.72 + \$1,446.00 = \$4,479.72, where \$487.00 + \$429.00 + \$318.00 + \$212.00 = \$1,446.00. Assets \$28,000.00 + \$10,000.00 = \$38,000.00. Reserves \$38,000.00 − \$25,376.34 = \$12,623.66, and \$12,623.66 ÷ \$3,033.72 = 4.16 months.

(b) Day 44 \$4,479.72 + \$611.00 = \$5,090.72 \$5,090.72 ÷ \$10,500.00 = 48.48%

(c) Day 47 \$5,090.72 − \$611.00 = \$4,479.72 \$4,479.72 ÷ \$10,500.00 = 42.66% — exactly where the file started.

(The account was paid on day 46, a Sunday; the file was re-run on day 47, Monday, once the payoff documentation was in hand. The fact changes first, the documentation second, the re-run last.)

(d) Reserves after the payoff \$12,623.66 − \$5,200.00 = \$7,423.66 \$7,423.66 ÷ \$3,033.72 = 2.45 months, down from 4.16.

Would you make the trade if the ratio problem had been smaller? Not automatically. Spending \$5,200 to buy 5.82 points of DTI was mandatory here, because 48.48% broke the approval's DTI condition. If the same \$5,200 bought two points on a file already comfortably approved, giving up 1.71 months of reserves — themselves an input the model uses and a compensating factor an underwriter values — would be a bad trade. Reserves are not spare money. They are part of the file.

(e) The two new verification messages you would expect, in the style of §15.6:

  CREDIT AND LIABILITIES
  15  Obtain evidence that the retail installment account reported at
      $611.00 per month has been paid in full and closed. A zero balance
      alone is not sufficient.
  16  Document the source of the $5,200.00 used to pay the account,
      including evidence that the funds were not borrowed.

A practical note worth raising with students: the payoff happened online on a Sunday, so on Monday morning no refreshed credit report will yet show a zero balance. The evidence that clears message 15 is the creditor's own payoff confirmation and account screen, not a new tri-merge. Students who answer "pull credit again" have identified the right requirement and the wrong document, and on a file closing Friday the difference is the whole week.

(f) What the re-run did not re-verify. Plenty — and one answer worth naming is income and employment. The re-run used the same income data submitted on day 6; it did not confirm that both borrowers are still employed. That confirmation is a separate requirement with its own timing window before the note date, and on a file where the borrowers have just demonstrated that they make financial moves without telling their loan officer, it is exactly the assumption you would want checked. A re-run records the changes you told it about. It does not go looking for the ones you did not.

(g) What a day-44 re-run would have returned.

48.48% and the same problem. On day 44 the furniture account existed, the \$611.00 payment was real, and nothing had been paid. Re-submitting the file that Friday would have returned obligations of \$5,090.72 and a back-end ratio of 48.48% — because on Friday afternoon, 48.48% was true.

That is §15.8's entire argument in one number. The system is deterministic and it is honest: it returns the file's actual condition, and the only way to change what it returns is to change the file's actual condition. The three-day sequence that saved this loan runs in exactly one direction:

   day 44 (Fri)  the fact is discovered            → 48.48%
   day 46 (Sun)  THE FACT IS CHANGED (debt paid)
   day 47 (Mon)  the change is DOCUMENTED
   day 47 (Mon)  the file is RE-RUN to record it   → 42.66%

Reverse any two of those steps and the file does not close. A loan officer who re-runs before the fact changes learns nothing; one who changes the fact but cannot document it has an approval built on an assertion; one who does both and forgets to re-run leaves the file sitting on a blown DTI condition that nobody has cleared.

Chapter 16

Worked solutions to the daggered (†) and odd-numbered exercises. Every FHA factor used below is illustrative; verify current figures with HUD.

Exercise 16.1

FHA insures approved lenders against loss on qualifying mortgages. FHA does not lend money to consumers, does not buy closed loans, and does not guarantee securities. Claims are paid from the Mutual Mortgage Insurance Fund.

The three-party distinction worth stating: the mortgagee funds the loan, FHA insures the mortgagee against credit loss, and Ginnie Mae guarantees timely payment to the investor who buys the resulting security.

Exercise 16.3

Handbook 4000.1 is HUD's consolidated single-family policy rulebook. A mortgagee letter is a numbered policy communication that announces or amends policy between handbook revisions.

When they disagree, the mortgagee letter governs — it is the newer instrument, and the handbook has simply not caught up. The FHA case number date matters because mortgagee letters are typically made effective for case numbers assigned on or after a stated date. That means the version of policy governing a loan can be fixed weeks before the borrower's application is complete and has nothing to do with the closing date.

Exercise 16.4 †

Adjusted value is the lesser of the purchase price and the appraised value.

When the appraisal comes in at or above contract, adjusted value equals the price and the definition is invisible — 3.5% of the price is what everybody expected.

When the appraisal comes in low, the definition bites twice, and in the direction nobody anticipates:

1. The MRI is computed on the lower value, so the base loan falls.
2. But the borrower still owes the contract price to the seller.

So the borrower's required cash is not the MRI — it is price minus base loan, and it rises.

Worked, using exercise 17's file with a \$241,000 appraisal on a \$248,000 contract:

  Adjusted value ............. lesser of $248,000 and $241,000 = $241,000
  MRI 3.5% ................... $241,000 x 0.035 = $8,435.00
  Base loan .................. $241,000 - $8,435 = $232,565.00
  Cash to the down payment ... $248,000 - $232,565 = $15,435.00
  Increase over the original . $15,435.00 - $8,680.00 = $6,755.00

The MRI went down by \$245.00 and the borrower's cash requirement went *up* by \$6,755.00. That is the sentence to say out loud on the phone.

Exercise 16.5

CAIVRS = Credit Alert Verification Reporting System, maintained by HUD.

What lands a person in it: default on a federally insured or guaranteed mortgage (FHA, VA, USDA), default on a federal student loan, default on an SBA loan, certain federal judgment liens, and other delinquent federal debt reported by the participating agency.

What a hit does: it renders the borrower ineligible for a federally related loan until the debt is resolved or the applicable exclusion period has elapsed and the record has been updated. It is a stop, not a risk factor — it cannot be offset by reserves, a compensating factor, or a letter of explanation.

Exercise 16.7

The rule: for terms greater than fifteen years, annual MIP is collected for 11 years where the loan-to-value at origination is 90% or less, and for the life of the loan where it is above 90%.

What surprises borrowers: the category is set once at closing and never revisited. Paying the balance down, the home appreciating, or obtaining a new appraisal changes nothing. A borrower who puts the minimum 3.5% down lands at 96.50% and can never reach the eleven-year band on that loan.

Exercise 16.9 †

Error one — about underwriting. FHA is not lenient about credit history. It has waiting periods after bankruptcy, foreclosure, and short sale; it requires derogatory accounts to be addressed; and it will downgrade a file to manual underwriting for a recent mortgage delinquency. Calling it "the bad-credit loan" describes a program that does not exist.

Error two — about pricing. What is actually true is that FHA's mortgage insurance premium is not priced off the credit score, while conventional MI is priced off it steeply. That is insensitivity in the premium, not leniency in the underwriting, and it is a completely different mechanism.

Corrected version (under 40 words): "FHA's mortgage insurance costs the same whether your score is 640 or 780. Conventional insurance gets much more expensive as the score drops. That pricing difference — not looser underwriting — is what decides which program is cheaper."

Exercise 16.11

Under the illustrative structure in §16.3, a minimum decision credit score of 574 falls in the 500–579 band, requiring a 10% minimum required investment rather than 3.5%.

The separate question you must answer first: where your lender's overlay stops. FHA policy permitting a 574 score is not the same statement as your shop being willing to originate one. Many lenders stop at 600, 620, or 640. Find out in writing before you tell this borrower they are approved for anything — "FHA allows it" is not an approval, and a borrower who has been told otherwise will make plans.

Exercise 16.12 †

Why the HPA does not help. The Homeowners Protection Act governs borrower-paid conventional mortgage insurance — request at 80% of original value, automatic termination at 78%. It does not apply to FHA-insured loans. There is no FHA equivalent for loans originated above 90% LTV: no request form, no threshold, no automatic termination.

The only two events that end the annual MIP obligation on such a loan:

1. The loan ceases to exist — it is paid off, whether by sale of the property or by a payoff from other funds.
2. The loan is refinanced out of FHA — in practice, into a conventional loan once the borrower has genuine equity supported by a current appraisal.

And the trap: an FHA streamline refinance is usually not an exit, because it requires no new appraisal and computes LTV on the original value — so the original duration band typically repeats on the new loan.

Exercise 16.13

Three plausible concerns, of which one is legitimate:

• "FHA buyers are weak buyers." Not legitimate as stated. A minimum decision credit score of 700 with an Approve/Eligible is a strong file regardless of the program on the top of the page. This is a proxy assumption and the honest response is to send a real pre-approval letter with real verification behind it.
• "FHA appraisals come in low." Not legitimate. An FHA appraisal develops value using the same methodology as any other appraisal. The value is the value.
• "The FHA appraisal will require repairs, and we'll have to do them before closing." Legitimate. This is the real concern and the one worth addressing directly, because it is true: the FHA appraisal certifies the property against HUD's minimum property requirements, and a "subject to" finding puts a repair and a re-inspection onto a closing calendar the seller did not budget for.

The professional move is to name the third one yourself before the listing agent does, and to say what you will do about it — flag likely items early, and get the re-inspection lead time from the appraiser's office up front.

Exercise 16.15 †

The circumstance: the borrower is relocating for employment to a new principal residence more than 100 miles from the current one. That is one of the narrow exceptions to FHA's general one-FHA-loan-at-a-time rule.

The threshold: 100 miles.

Two cautions. First, the exception categories are specific and there are others (an increase in family size, vacating a jointly owned property following a divorce), each with its own documentation requirements. Second, the same 100-mile figure appears in a different rule — whether rental income from the vacated residence may be used — and conflating the two is the common error. Verify both with HUD, including how the distance is measured.

Exercise 16.16 †

An Approve/Eligible can still end up measured against the manual benchmark because the underwriter downgrades the file to manual underwriting. Recurring triggers include:

1. A mortgage payment delinquency within the last 12 months.
2. Disputed derogatory accounts above the applicable threshold.
3. Undisclosed debt discovered after the AUS run — the furniture-account problem, in FHA form.
4. Information the underwriter obtains that was not in the data the scorecard evaluated, making the Accept unreliable. This is the catch-all and it is real.

[Verify the current enumerated trigger list in 4000.1.]

The principle: an AUS recommendation is a recommendation about the data it was given. Change the data and you may change the path — and on FHA, changing the path changes the rules, not just the paperwork.

Exercise 16.17 †

[All factors illustrative.]

  (a) MRI 3.5% ............. $248,000 x 0.035 ................. = $  8,680.00
  (b) Base loan ............ $248,000 - $8,680 ................ = $239,320.00
  (c) LTV .................. $239,320 / $248,000 .............. = 96.50%
  (d) UFMIP 1.75% .......... $239,320 x 0.0175 ................ = $  4,188.10
  (e) Total loan ........... $239,320 + $4,188.10 ............. = $243,508.10
  (f) P&I .................. $243,508.10 at 6.375%, 360 mo .... = $  1,519.17
  (g) Monthly MIP .......... $243,508.10 x 0.0055 / 12 ........ = $    111.61
                             ($1,339.29 per year)
  (h) PITI + MIP ........... $1,519.17 + $111.61 + $260 + $95 . = $  1,985.78
  (i) Housing ratio ........ $1,985.78 / $5,600.00 ............ = 35.46%
  (j) Back-end ratio ....... $2,525.78 / $5,600.00 ............ = 45.10%
                             ($1,985.78 + $540.00 = $2,525.78)
  (k) MIP duration ......... LTV 96.50% > 90% ................. = LIFE OF THE LOAN

Do the ratios clear 31/43? No — 35.46% and 45.10% both exceed the benchmark.

Does that matter? Only if the file is manually underwritten. With an Approve/Eligible from TOTAL, the manual-underwriting ratio table does not govern this file, and the ratios above are unremarkable. If the file were downgraded to manual, those same ratios would need documented compensating factors under the current published table.

Exercise 16.19 †

  PITI + MIP = $1,163.42 + $85.60 + $180.00 + $88.00 ........... = $1,517.02

  Housing ratio  = $1,517.02 / $4,850.00 ....................... = 31.28%
  Back-end ratio = ($1,517.02 + $610.00) / $4,850.00
                 = $2,127.02 / $4,850.00 ....................... = 43.86%

Both figures sit just above the 31/43 benchmark — 31.28% and 43.86%. On a manually underwritten file those twenty-eight and eighty-six basis points are the whole conversation and require documented compensating factors under the current table.

What would need to be true to submit with confidence:

1. An Approve/Eligible from TOTAL, which takes the manual table out of play entirely.
2. No downgrade trigger anywhere in the file — no recent mortgage lates, no disputed derogatory accounts, no undisclosed debt.
3. If it is going manual: documented compensating factors in the file, not asserted in a cover letter — verified reserves, a payment-shock comparison against current housing expense, or additional income not used to qualify.
4. And separately from any of that: a borrower who has seen the \$1,517.02 written down and has said out loud that they can live with it.

Exercise 16.21 †

[All factors illustrative. Annual MIP factor held constant at 0.55% in both scenarios to isolate the duration effect.]

                                    SCENARIO A (3.5%)      SCENARIO B (10%)
  Down payment .................... $ 10,500.00            $ 30,000.00
  Base loan ....................... $289,500.00            $270,000.00
  LTV at origination .............. 96.50%                 90.00%
  UFMIP 1.75% ..................... $  5,066.25            $  4,725.00
  Total loan ...................... $294,566.25            $274,725.00
  Annual MIP (0.55%) .............. $  1,620.11/yr         $  1,510.99/yr
  DURATION CATEGORY ............... LIFE OF LOAN           11 YEARS
  TOTAL ANNUAL MIP PAID ........... $ 48,603.43            $ 16,620.86
                                    ($1,620.11 x 30)       ($1,510.99 x 11)

  DIFFERENCE IN TOTAL MIP ......... $48,603.43 - $16,620.86 = $31,982.57
  ADDITIONAL CASH REQUIRED ........ $30,000.00 - $10,500.00 = $19,500.00

The two-sentence recommendation: "Putting ten percent down instead of three and a half costs you nineteen thousand five hundred dollars more at the table, and it saves you just under thirty-two thousand dollars in mortgage insurance — because at ten percent down the insurance stops after eleven years, and at three and a half percent it never stops. If you have the money and you intend to keep this house past year eleven, the larger down payment wins, and it wins by more than the monthly payment difference will ever show you."

(Note the teaching point: the monthly payment comparison alone badly understates the value of crossing 90%, because what you buy by crossing it is not a smaller payment — it is an end date.)

Exercise 16.23

Model memo:

  TO: Processing
  RE: [borrower] - program selection

  1. Running FHA first. Rep score 638; conventional MI at that score and this LTV is
     either unavailable or priced at a multiple of the FHA factor, and the 47% back-end
     is unlikely to clear conventional AUS. FHA MIP is not priced off score.
  2. Running conventional in parallel as a check, not as a fallback. If we get an
     Approve/Eligible on conventional, I want to see the MI quote before we choose.
  3. Borrower has 5% down. On FHA that puts base LTV at 95%, still above 90%, so MIP
     is life-of-loan either way. Flagging that for the borrower conversation now, not
     at CD.

The judgment worth noting: 5% down on FHA does not buy the eleven-year band — the borrower would need 10%. Telling them that early is the difference between an informed choice and a complaint in year six.

Exercise 16.24 †

Linden Street FHA at 10% down [illustrative; the property and rate are the chapter's frozen figures]:

  Down payment 10% ............ $385,000 x 0.10 ................ = $ 38,500.00
  Base loan ................... $385,000 - $38,500 ............. = $346,500.00
  LTV at origination .......... $346,500 / $385,000 ............ = 90.00%
  UFMIP 1.75% ................. $346,500 x 0.0175 .............. = $  6,063.75
  Total loan .................. $346,500 + $6,063.75 ........... = $352,563.75
  P&I at 6.250%, 360 months ................................... = $  2,170.80
  Annual MIP 0.55% ............ $352,563.75 x 0.0055 = $1,939.10/yr
                                                     = $161.59/mo
  Taxes + insurance ........................................... = $    515.00
  PITI + MIP .................. $2,170.80 + $161.59 + $515.00 .. = $  2,847.39

  DURATION CATEGORY ........... 90.00% is NOT above 90% ........ = 11 YEARS
  Total annual MIP ............ $1,939.10 x 11 ................. = $ 21,330.11
      (against $62,374.40 on the 3.5% structure - a $41,044.29 difference)

Why the question is academic: the Linden Street borrowers have \$38,000.00 in total verified assets. The down payment alone is **\$38,500.00** — \$500.00 more than everything they have, before a single closing cost, prepaid item, or dollar of reserves. The structure that would have solved the MIP problem is the one they cannot reach, which is the ordinary condition of the borrowers this chapter is about.

Exercise 16.25

The one question you must ask first: "Is your aunt connected to this transaction in any way — is she the seller, the agent, the builder, or related to any of them?" If the answer is yes, no amount of documentation fixes it and you need to restructure before anyone gathers paperwork.

Assuming the answer is no, the documents to clear the condition:

• A signed gift letter stating the amount, the date, the donor's name, address, and phone, the relationship to the borrower, and — explicitly — that no repayment is expected.
• Evidence of the donor's ability to give: a statement or documentation showing the funds in the donor's account.
• Evidence of the transfer: the donor's withdrawal or wire confirmation and the borrower's corresponding deposit, so the money can be traced end to end.
• The borrower's bank statement showing the deposit, matching the amount and date.

[Verify the current documentation requirements in 4000.1 — they are specific and they are stated.]

Exercise 16.26 †

Eleven days to closing, three repair items, and a re-inspection that has to be scheduled. The sequence matters more than any individual step.

  DAY  OWNER              ACTION
  ─────────────────────────────────────────────────────────────────────────────
  0    loan officer       Call the appraiser's office FIRST and get the
       (today)            RE-INSPECTION LEAD TIME. Everything below is planned
                          backward from that number. Do not skip this to "get
                          started" - if the lead time is six days, the schedule
                          is different from the one you would have written.
  0    loan officer       Conference the buyer's agent and listing agent
                          together. Name the three items. Get a written answer
                          to one question: WHO does the work and BY WHEN?
  0    loan officer       Email both agents and the borrower a one-page summary
                          with the target dates. Nothing verbal survives.
  1    listing agent /    Contractor scheduled and confirmed in writing.
       seller             If the seller refuses, escalate TODAY - see below.
  1    loan officer       Notify processing and underwriting that a 1004D is
                          coming; confirm exactly what form and what evidence
                          the underwriter wants (photos, receipts, or both).
  2-5  seller's           Work performed. Ask for dated photos as it is done,
       contractor         not after.
  5    loan officer       Order the re-inspection the moment the work is
                          reported complete - not the next morning.
  6-8  appraiser          Re-inspection performed; 1004D certification issued.
  8    processor          1004D delivered to underwriting; condition cleared.
  9-11 closer             Closing package, CD timing, closing.
  ─────────────────────────────────────────────────────────────────────────────

The single step most likely to slip: the seller agreeing to do the work, and doing it on time. The seller still owns the property, the repairs benefit a buyer they have no further interest in pleasing, and nothing in the contract necessarily obligates them. This is the step that is outside your control and outside your borrower's control, and it is therefore the one you escalate first and hardest.

Second most likely: the re-inspection lead time, which is why it is step zero rather than step five. A loan officer who spends four days getting repairs done and then discovers the appraiser cannot return for a week has lost the closing date to a phone call they could have made on day one.

What a strong answer also includes:

• A contingency: if the seller will not do the work, the options are the buyer performing it with written seller consent (uncommon but real), a price adjustment plus an amendment, or an extension of the closing date — and each of those needs the agents, not you, to execute.
• Telling the borrower the truth about the date early, before movers and time off work become non-refundable. That is the item most students omit and it is the one the borrower will remember.

Exercise 16.27

Model call, and what you leave out:

"I need to tell you about something that came back on the file, and I want to be straight with you because it's serious and it's fixable more often than people think. There's a federal database called CAIVRS that every government-backed loan gets checked against, and it flagged a prior government loan — most often that's an old FHA mortgage, a student loan, or an SBA loan that went to default. It doesn't show up on your credit report, so I'd expect this to be news to you.

Here's what I need. Think back over every government-backed loan you've ever had, including ones tied to a former spouse or a business, and including anything from a long time ago. If a property you were on went to foreclosure — even if a divorce decree said it wasn't your responsibility — that's the likely source, because a decree divides responsibility between the two of you but does not release you from the debt with the lender.

This stops the file until it's documented and resolved, and I don't yet know how long that takes for your specific situation. I'm not going to guess at a timeline. I'll find out today and call you back with a real one."

What you do not say: that it is probably a mistake; that you can "get around it"; that it will not affect the closing date; that it is definitely the ex-spouse's fault; or any speculation about what the underlying debt is before you have documentation. You also do not use the word "default" as an accusation. The borrower is almost certainly hearing this for the first time.

Exercise 16.29 †

Four errors. Two of them come from the same misconception.

  LINE                CORRECT VALUE / RULE
  ─────────────────────────────────────────────────────────────────────────────
  UFMIP               WRONG. Computed on the PRICE ($265,000 x 0.0175 =
                      $4,637.50) instead of the BASE LOAN.
                      Correct: $255,725 x 0.0175 = $4,475.19

  Total loan          Follows from the error above.
                      Correct: $255,725 + $4,475.19 = $260,200.19

  LTV                 WRONG. Computed as total loan / price.
                      Correct: BASE loan / adjusted value
                             = $255,725 / $265,000 = 96.50%

  MIP duration        WRONG, twice over. LTV at origination is 96.50%, which is
                      above 90%, so the category is LIFE OF THE LOAN, not 11
                      years. And the parenthetical reasoning is the deeper error:
                      reaching an LTV under 90% "by year 6" is irrelevant, because
                      the category is set at origination and never revisited.

  Monthly MIP         Arithmetically consistent with the wrong total loan.
                      Correct: $260,200.19 x 0.0055 / 12 = $119.26
  ─────────────────────────────────────────────────────────────────────────────

The single misconception producing two of the errors: treating the financed UFMIP as part of the loan-to-value calculation. It is not — program LTV is computed on the base loan amount. That one mistake produced both the 98.25% LTV figure and, indirectly, the confusion about which duration band applies. (The UFMIP base error is a separate, purely arithmetic mistake.)

Exercise 16.30 †

Part (a) — the \$12,000 seller credit email

The problem: on a \$310,000 FHA purchase the minimum required investment is \$310,000 × 0.035 = **\$10,850.00, and the seller may not fund one cent of it**, directly or indirectly. The agent's email describes the credit as covering "your down payment and most of your closing costs," which is the part that has to be corrected before anyone relies on it.

What is still usable: a seller may contribute toward closing costs, prepaid items, and discount points, up to the permitted interested-party contribution limit — a percentage of the sales price that is more generous on FHA than on conventional, and that you should verify currently with HUD. A \$12,000 credit is 3.87% of \$310,000, which is likely inside the FHA limit as a closing-cost credit. It simply cannot be applied to the \$10,850.00.

Model reply:

"This is good news and I want to make sure we use it correctly, because there's one rule that catches people here. On an FHA loan the seller can help with closing costs, prepaid items like taxes and insurance, and points — but the seller is not allowed to pay any part of your down payment. That's a federal rule, not our lender's policy, and there's no workaround.

So the \$10,850.00 down payment still needs to come from you or from an acceptable source like a gift from family. The \$12,000 can go to work on your closing costs and prepaids, which on a file this size is likely most or all of them — I'll price it out today and send you the exact numbers. Can you let your agent know so the credit gets written into the contract as a closing-cost credit rather than a down-payment credit? The wording matters."

Grader's note: a student who simply says "that's not allowed" has answered half the question. The item is testing whether they can salvage the usable part and fix the contract language, which is where the problem actually gets solved.

Part (b) — the Approve/Eligible with a recent mortgage late

What it means procedurally: a mortgage payment delinquency within the last 12 months is a recurring downgrade trigger. An Approve/Eligible in the file does not settle the question, because the underwriter may set the Accept aside and underwrite the file manually — at which point the 31%/43% ratio table and its compensating-factor requirements begin to apply for the first time.

What you do today, not at submission:

1. Compute the ratios against the manual benchmark now, so you know whether a downgrade is survivable or fatal. If they are 38/48, you have a problem to solve, not a surprise to receive.
2. Gather the compensating factors in documented form now — reserves, payment shock, additional income not used to qualify.
3. Get the explanation and supporting documentation for the late payment into the file proactively, rather than as a response to a condition three weeks later.
4. Tell the borrower that the recent late may change the path, and what that would mean. Do not let them believe the AUS approval is the end of the credit conversation.

The general habit: when you can see a downgrade trigger in the file, underwrite it both ways before you submit.

Exercise 16.31

Model paragraph. Constraints: under 120 words, no percentage signs, no acronym used before it is expanded, both exits stated.

"One thing I want in writing so it isn't a surprise later. On the loan we discussed, the monthly mortgage insurance premium — MIP — stays on the loan for as long as you have it. It does not come off when you reach twenty percent equity, and it does not come off when the home goes up in value. That is set by the size of the down payment on the day you close, and it cannot be changed afterward. There are exactly two ways it ever ends: you pay the loan off, or you refinance into a conventional loan once you have enough equity. Neither is a decision for today. I just want you to know it now."

(108 words. No percentage signs. "MIP" is expanded on first use. Both exits stated.)

Grader's note: mark down any version that says "it comes off eventually," that uses a percentage sign, or that implies an FHA streamline refinance is one of the exits — it usually is not, because there is no new appraisal and the loan-to-value is computed on the original value.

Exercise 16.32 †

Part (a) — the comparison memo

Model memo. Grader's note: the memo must lead with the three headline numbers, state a recommendation, and contain an honest sentence about uncertainty.

  LINDEN STREET - PROGRAM COMPARISON            prepared day 2
  [All FHA and MI factors illustrative; verify current figures with HUD.]

                                CONVENTIONAL 95%       FHA 96.5%
  Down payment .................. $ 19,250.00          $ 13,475.00
  Base loan ..................... $365,750.00          $371,525.00
  UFMIP 1.75% financed .......... --                   $  6,501.69
  Total loan .................... $365,750.00          $378,026.69
  LTV (base / price) ............ 95.00%               96.50%
  Rate .......................... 6.625%               6.250%
  P&I ........................... $  2,341.94          $  2,327.58
  Monthly MI / MIP .............. $    176.78          $    173.26
  Taxes + insurance ............. $    515.00          $    515.00
  PITI + MI ..................... $  3,033.72          $  3,015.84
  Back-end ratio ................ 42.66%               42.49%
  MI ENDS ....................... payment 137          NEVER
  TOTAL MI OVER THE TERM ........ $ 24,218.86          $ 62,374.40

  THE THREE NUMBERS
  1. FHA is $17.88 per month cheaper.
  2. FHA needs $5,775.00 less at the closing table.
  3. FHA costs $38,155.54 more in mortgage insurance, because it never ends.

  Also worth seeing: the FHA upfront premium of $6,501.69 is $726.69 LARGER than
  the $5,775.00 of cash it saved you.

  RECOMMENDATION: conventional, 95%. You have the $19,250.00, and after closing
  you still hold $12,623.66 - about 4.16 months of payments - in reserve. FHA's
  down-payment relief would be a convenience for you, not a necessity, and
  $5,775.00 of convenience is not worth $38,155.54.

  HONESTLY, THE UNCERTAINTY: this recommendation assumes you keep the loan a long
  time. Over the first ten years the two structures are within roughly $500 of
  each other. If you refinance or move by year seven, the difference above mostly
  evaporates. I can't know which will happen, so I'm recommending the structure
  that is better in the case we can't rule out.

Part (b) — the three relationship questions

Model wording. The point of the item is that these are questions about people, asked in plain language, at application — not conditions cleared in underwriting.

1. Source and interest: "Where is every dollar of your down payment coming from? And is any part of it — any part at all — coming from the seller, the real estate agents, the builder, or anyone connected to this sale?"
2. Identity of interest: "Do you know the seller? Are you related to them, do you work for them, or do you have any kind of business relationship with them?"
3. Federal debt: "Have you ever had a government-backed loan — FHA, VA, USDA, a federal student loan, an SBA loan — that went to default, foreclosure, or collection? Even a long time ago, and even if it was somebody else's responsibility?"

Grader's note: look for a fourth question in strong answers — "Is anyone expecting to be paid back?" — which catches the gift-that-is-really-a-loan. Also look for whether the student wrote the questions in spoken language. A student who writes "Please disclose any identity-of-interest relationship with the seller" has not done the exercise; no borrower knows what that means.

Exercise 16.33 †

Part (a) — the branch manager's instruction

The commercial problem. "Lead with FHA on anyone under 720" is a rule about a category of borrower, not about a file. Section 16.10 showed that a 706-score borrower with cash is meaningfully better off conventional — \$38,155.54 better off on the Linden Street file — and 706 is under 720. The instruction would therefore route borrowers into a more expensive product, and those borrowers eventually find out, and they tell their agents.

The regulatory problem. A standing instruction to steer a class of borrowers toward a particular product, defined by a credit-score cutoff, is a fair-lending exposure before it is anything else. Credit score correlates with protected characteristics in ways that are documented, and a policy producing a disparate distribution of products across protected classes is a problem whether or not anyone intended it. Chapter 25 treats this at length. "Lead with" is also not a neutral word: at the point where a borrower is presented with one option and not the other, it is steering.

The professional problem. The instruction substitutes a category for an analysis. The four questions in §16.10 — feasibility, MI availability and price, approvability, holding period — cannot be answered by a score alone, and the manager's rule answers all four with one number.

The response in the meeting (respectful, specific, and not a lecture):

"I want to make sure I understand the goal — is it turn times, or margin, or both? Because I think we can get at both without a score cutoff. My concern with 'lead with FHA under 720' is that a 706 with five percent down and cash in the bank is usually cheaper conventional, and if we lead them the other way they find out later. And a standing rule keyed to a score is the kind of thing compliance will want to look at for fair-lending reasons, which I'd rather we raise ourselves than have raised for us. Could we instead standardize the comparison — require both structures priced and presented on every file under 720 — and see whether FHA still wins on the ones you're thinking of? Then we're faster and we're documented."

Part (b) — the 745-score borrower who insists on FHA

What you do: run the comparison, present it in writing, explain it once clearly — and then, if they still want FHA and they are eligible for it, originate the FHA loan.

That is the answer, and students who resist it should be pushed on why. The line is this:

• Advising is making sure the borrower has the information: both structures priced, the total cost stated in dollars, the duration difference stated plainly, and the chance to ask questions.
• Overriding is refusing to originate the product the borrower has chosen, or continuing to relitigate the decision after they have made it, or quietly making the FHA path slower or harder.

A borrower is entitled to choose a product that costs them more. People choose products for reasons that are not on a spreadsheet — a family member's advice, a bad experience with a prior lender, or a preference for a government program they trust. Those reasons are theirs.

What you document: that you presented both structures, in writing, with the figures; the date; and the borrower's decision. Not because you expect a dispute — because if a servicing conversation in year six goes badly, the file should show that this borrower knew.

The one thing you must not do: decide privately that FHA is "close enough" and stop mentioning conventional. The obligation is discharged by disclosure, not by winning.

Grader's note: an answer that concludes "so I would keep pushing until they agreed" has the ethics backwards. So does an answer that never presented the comparison at all.

Exercise 16.34

In order:

1. Verify the correct factor from the source — the current mortgagee letter, not the training deck and not a colleague. Confirm the case-number date and which version of policy governs this specific loan. Know the exact right number before you say anything to anyone.
2. Recompute the file — the payment, the ratios, and whether the higher payment changes the approval. A \$41 increase in payment is a ratio change; on a marginal file it can be an approval change, and you need to know that before the borrower asks.
3. Tell your manager and processing immediately, so the file's disclosures and the underwriting condition set can be corrected on the same day, and so anyone else quoting from the same deck stops today.
4. Tell the borrower yourself, promptly, in your own words — before they receive a revised disclosure showing a number they were not expecting. "I got this wrong and here is the correct figure and here is what it changes" is a survivable sentence on day 30. It is not survivable on day 48.
5. Confirm the disclosure obligations with compliance. A change in payment flows through to the Loan Estimate and, later, the Closing Disclosure, and there are timing rules. Do not guess at them.
6. Fix the source. Get the outdated training deck corrected or withdrawn, because the next person who reads it will make the same error.

Who you tell first: your manager and processing, in the same breath — because the file has to be corrected and the correction has a compliance dimension you do not own. But you tell the borrower yourself, and you tell them the same day. The order is about the mechanics; the principle is that they hear it from you rather than from a document.

Exercise 16.35

(c) the life of the loan.

92% is above 90%, so the loan falls in the life-of-loan band. The distractors are all conventional-mortgage-insurance answers dressed up as FHA: (b) is the Homeowners Protection Act's automatic termination threshold, which does not apply to FHA loans at all, and (d) is FHA's own pre-2013 rule — which is why experienced practitioners pick it. (a) is the correct answer for a different loan, one at 90% or less.

Teaching note: students who answer (a) have understood that there are two bands but have not absorbed which side of the line 92% falls on. Students who answer (b) or (d) have not absorbed the rule at all.

Exercise 16.36 †

(c) UFMIP may be financed into the loan and is excluded from the LTV used for program eligibility.

• (a) is wrong — UFMIP is financeable and is financed on the large majority of files.
• (b) is the trap, and it is the most commonly chosen wrong answer. Program LTV is computed on the base loan amount. On Linden Street that is \$371,525 ÷ \$385,000 = 96.50%, not \$378,026.69 ÷ \$385,000 = 98.19%.
• (d) is wrong — a partial refund is available on a declining schedule when refinancing into a new FHA loan within a defined period, not a full refund on payoff. (Verify the current schedule with HUD.)

Exercise 16.37

(b) an identity of interest transaction.

An employer-to-employee sale is a business relationship between buyer and seller, which is exactly what identity of interest means. FHA's default treatment restricts the maximum loan-to-value — but note that a purchase from an employer under a relocation program is one of the situations covered by the defined exceptions, so the standard LTV may be restored. (Verify the current exception categories with HUD.)

• (a) describes what the seller may contribute toward costs — a different rule entirely.
• (c) is not an FHA term.
• (d) is wrong and is the answer students pick when they confuse "restricted" with "prohibited." The transaction is permitted; the loan-to-value is what is at issue.

Exercise 16.38

(b) The benchmark ratios for manually underwritten files, which may be exceeded with documented compensating factors.

• (a) is the industry's most common misstatement and the reason approvable borrowers get declined.
• (c) confuses the manual path with the TOTAL path; the scorecard does not require these ratios.
• (d) is invented — nothing about UFMIP varies with the qualifying ratios.

Exercise 16.39 †

The FHA column, from the chapter's frozen figures:

  Purchase price / adjusted value ......... $385,000.00
  MRI 3.5% ................................ $ 13,475.00
  Base loan ............................... $371,525.00
  LTV at origination ...................... 96.50%
  UFMIP 1.75%, financed ................... $  6,501.69
  Total loan .............................. $378,026.69
  Rate, 30-year fixed ..................... 6.250%
  P&I ..................................... $  2,327.58
  Annual MIP 0.55% ........................ $    173.26 / month
  Taxes + insurance ....................... $    515.00
  PITI + MIP .............................. $  3,015.84
  Back-end ratio .......................... 42.49%
  Total MI over the term .................. $ 62,374.40

  THE ADDED ROW
  Date the mortgage insurance ends:
      Conventional 95% .................... payment 137
      FHA 96.5% ........................... NEVER

The twenty-five-word explanation (a model; grade on plainness, not on matching this):

"On the conventional loan the monthly insurance stops after about eleven and a half years. On the FHA loan it never stops."

(23 words, no percentage sign, no jargon. If a student's version uses "loan-to-value," "LTV," "equity," or a percentage, it has not met the constraint.)

Exercise 16.40 †

Which of the four questions changes: question 1 — feasibility.

With \$38,000.00 in assets the conventional cash requirement of **\$25,376.34 was comfortable and left \$12,623.66** = 4.16 months of reserves. With **\$22,000.00, the borrowers are \$3,376.34 short of the conventional structure entirely. FHA's \$5,775.00 of down-payment relief would bring the requirement to roughly \$19,601.34 (holding all other costs constant — an approximation), leaving about \$2,398.66**, which is **under one month** of the \$3,015.84 payment.

The two sentences:

"At twenty-two thousand, conventional isn't a choice anymore — you're about thirty-four hundred short of the cash it takes to close it, so FHA is the structure that gets you into this house. I want you to know what that costs, because the mortgage insurance on the FHA loan never comes off and over the life of the loan that's about thirty-eight thousand dollars more than the conventional loan would have been — and I also want to talk about the fact that you'd be closing with less than one month's payment left in the bank, which worries me more than the thirty-eight thousand does."

The teaching point: when question 1 changes its answer, questions 2 through 4 stop being a comparison and become disclosure. You are no longer helping the borrower choose. You are telling them, honestly, what the only available option costs — and, in this version, raising the reserve problem that is now the more urgent of the two.

Chapter 17

Worked solutions to the daggered (†) and odd-numbered exercises. All dollar figures are constructed teaching values; every funding fee rate, per-square-foot factor, county loan limit, and USDA figure is illustrative and must be verified at the agency.

Exercise 17.1

FHA insures. It collects an upfront premium and an annual premium from the borrower and stands behind the loan, so the lender's protection is bought with a recurring borrower charge.

The VA guarantees. It collects a one-time funding fee and promises to absorb a defined portion of the lender's loss, so the lender's exposure is bounded without any recurring charge to the borrower.

Where it shows up: on the monthly payment. An FHA borrower pays an annual mortgage insurance premium in twelve monthly installments — on the Linden Street FHA option, \$173.26 a month, and at a loan-to-value above 90% it never terminates. A VA borrower pays nothing monthly at all. On the Linden Street VA counterfactual that difference is the whole reason a loan \$27,527.50 larger carries a payment \$65.18 smaller.

Exercise 17.3

A COE answers two questions: (1) is this applicant eligible for the VA home loan benefit, and (2) how much entitlement is available to them. It also carries the VA's determination of funding fee exemption status and any prior loans still charging entitlement.

Three questions it does not answer, but is routinely read as answering:

1. "Is this borrower approved?" No. It says nothing about credit, income, assets, or reserves. An eligible applicant can be declined on any of them.
2. "How much can they borrow?" No. Remaining entitlement is an input to §17.3's arithmetic. It is neither a loan amount nor a purchase price, and reading it as one understates the borrower's buying power by a factor of roughly four.
3. "Will a lender make this loan?" No. Overlays, product availability, and the lender's own VA approval status are separate questions (Chapter 14, §17.2).

Exercise 17.5

The three dimensions: (a) transaction type — purchase, cash-out refinance, and IRRRL each carry different rates; (b) down payment tier — the fee falls as the down payment rises; (c) first use versus subsequent use of entitlement, with subsequent use carrying a higher rate at the zero-down tier.

Why the book does not print the schedule. The schedule has been revised repeatedly, including a 2020 restructuring that also equalized rates previously differing between regular military and Guard/Reserve applicants. A textbook that printed the current numbers would be wrong within a year and would teach the reader a habit — quoting a fee from memory — that eventually produces a wrong Loan Estimate and an unhappy borrower. What is durable is the structure: three dimensions, financeable, and waivable by exemption.

Exercise 17.7

Four things that could be true about a borrower who "used it in 2003":

1. The prior loan was paid off and the property sold, but restoration was never applied for. Entitlement is restorable on application — and application is the step that gets skipped.
2. The prior loan was paid off with the property retained. A one-time restoration may be available.
3. The prior loan is still outstanding (they kept the house as a rental). Entitlement is partly charged; §17.3's two lines apply and there is very likely still a zero-down purchase available.
4. The prior loan ended in a claim — foreclosure or a compromise sale. Entitlement is reduced until the government's loss is repaid, but the borrower is not barred; the reduced figure still goes into the same arithmetic.

The single document: the Certificate of Eligibility. Order it before you discuss any of the four.

Exercise 17.9 †

What "the VA has no minimum credit score" means: it is literally true. The VA publishes no minimum representative score. The VA's own underwriting standard is a judgment about satisfactory credit supported by the residual income requirement (§17.5).

What it does not mean: that your borrower can get a VA loan at 618 from your employer. Your 640 overlay is a lender rule, not an agency rule (Chapter 14), and it is fully enforceable against this file. Telling the borrower "there's no minimum score on VA loans" without the next sentence is a promise you cannot keep.

Two concrete actions:

1. Work the score. Pull the tradeline detail and look for utilization-driven gains. Chapter 10's rapid-rescore analysis applies unchanged; 618 to 640 is 22 points, which is frequently a balance-reporting problem rather than a credit problem.
2. Place the file elsewhere. Another lender's VA overlay may be lower, and if your shop is a broker or the borrower is willing to work with a referral, the same loan may be approvable today. Note that this is a program-preserving referral — you are moving the file to keep the VA loan, which is the opposite of talking the borrower into FHA (§17.8).

Say both out loud, in that order, and give the borrower the choice.

Exercise 17.11

(a) Entitlement. It remains charged until the loan is paid in full. A non-veteran assumer has no entitlement to substitute for the seller's, so the veteran's entitlement stays attached to a loan on a house they no longer own — potentially for decades.

(b) Liability. Without a VA-approved assumption and a release of liability, the original veteran may remain liable on the debt. An "assumption" the parties arrange privately is not a release. This is the half of the answer that has real financial consequences and it is the half that gets forgotten.

(c) The next purchase. They run §17.3's partial-entitlement arithmetic. Remaining entitlement is 25% of the applicable county limit less the amount charged to the assumed loan, and the maximum zero-down loan is four times that. Above it, a down payment of 25% of the excess.

The practitioner point: tell every VA client who is selling, before they list, that an assumption by a non-veteran ties up their benefit and may leave them liable. A 2.75% assumable note is genuinely valuable and worth selling — but the veteran needs to know what they are trading.

Exercise 17.13

Three common NOV requirements, and what each actually is:

Note that none of the three is a document request. That is the point of the exercise. A stip that says "provide a bank statement" is a Tuesday. A stip that says "repair the south elevation" is a schedule, a cost, and a negotiation with a seller who believed they were finished — and it arrives late, because the NOV arrives late.

Exercise 17.14 †

Applicable county limit \$766,550** (illustrative). Prior VA loan **\$248,000. Purchase \$610,000.

Verification 1 — the shortcut. \$766,550 − \$248,000 = \$518,550 ✓ — the maximum zero-down loan, without touching a percentage.

Verification 2 — the 25% guaranty check. Guaranty \$129,637.50 + down payment \$22,862.50 = \$152,500.00**, and 25% × \$610,000 = \$152,500.00** ✓.

A financed funding fee at the subsequent-use rate goes on top of the \$587,137.50; look the rate up.

Exercise 17.15

The VA requires no down payment at all. With full entitlement there is no VA loan limit — the VA guarantees 25% of the loan regardless of amount (verify the current rule; this follows legislation effective in 2020).

What actually determines whether the loan can be made: two things, neither of them the VA.

1. The lender's overlay. Many lenders cap zero-down VA loans at some amount and require a down payment above it. That cap differs by lender, which makes this a shoppable file (Chapter 14).
2. The borrower's own qualification — credit, ratios, and above all the residual income requirement (§17.5), which on a \$610,000 loan will be measured against the higher side of the published loan-size breakpoint.

The teaching point: on a full-entitlement VA file, "how much can they borrow?" is a question about your employer and your borrower, not about the VA.

Exercise 17.16 †

Prior VA loan \$412,000** ended in foreclosure; **\$103,000 of entitlement remains charged. County limit \$766,550**. Purchase **\$340,000.

Shortcut check: \$766,550 − \$412,000 = **\$354,550** ✓ (and note 25% × \$412,000 = \$103,000, so the charged figure is internally consistent).

Yes — zero down. The purchase is \$14,550 below the maximum zero-down loan. A financed funding fee at the subsequent-use rate goes on top, and the file still has to clear credit, ratios, residual income, and any seasoning or re-establishment requirements applicable after a prior foreclosure (Chapter 10, Chapter 14) — none of which is what the borrower called about.

The two sentences:

"Your entitlement isn't gone — part of it is still charged from the old loan, and what's left supports a zero-down purchase up to about \$354,550, which is more than this house."

"There's still credit history to work through and I'm not promising an approval today, but the thing you were afraid of — that the foreclosure ended the benefit — isn't what happened."

Exercise 17.17 †

Residual income per person: \$2,506.00 ÷ 4 = **\$626.50**.

Back-end ratio: (\$1,975.00 + \$742.00) ÷ \$6,800.00 = \$2,717.00 ÷ \$6,800.00 = 39.96%.

Whether \$2,506.00 passes depends on the VA's current published minimum for a four-person household in this property's region at this loan size. Look it up; do not carry a number in your head.

Exercise 17.19 †

File A: \$4,800.00 ÷ \$12,000.00 = 40.00%. File B: \$1,440.00 ÷ \$3,600.00 = 40.00%.

Identical. Every automated underwriting system, pricing engine, and guideline matrix in this book reads them as the same file.

The four facts you need, in priority order:

1. Household size. File A at 40% leaves \$7,200.00 of gross before taxes; File B leaves \$2,160.00. If A is two people and B is six, the per-person gap is not close, and nothing in the ratio shows it. This is the residual income question and it is first for a reason.
2. The dollar residual after taxes and the utility line. Not gross remainder — actual remainder. Run §17.5's worksheet on both.
3. Income stability and continuity (Chapter 11). A 40% ratio on income with a documented three-year history is a different risk from a 40% ratio on income that began nine months ago.
4. Reserves after closing, in months of PITI (Chapter 12). The cushion that absorbs a furnace, a car, or a lost shift. On the Linden Street file it is what paid off the day-44 furniture account.

The point of the exercise: the ratio was the first number you computed and the least informative one you will use.

Exercise 17.21

From §17.4's table, on the \$385,000 Linden Street VA counterfactual:

Monthly saving: \$2,968.54 − \$2,831.03 = \$137.51. Payback: \$19,250.00 ÷ \$137.51 = 140.0 months — eleven years and eight months.

The two sentences:

"You can put five percent down, and it saves you \$137.51 a month — but it takes a hundred and forty months, almost twelve years, before that saving gives you back the \$19,250."

"On a conventional loan I'd push harder, because a bigger down payment buys away the monthly mortgage insurance. On a VA loan there's no monthly insurance to buy away, so unless you have a specific reason to want a smaller balance, I'd keep the cash."

Exercise 17.22 †

The case for VA:

• No monthly charge of any kind — no mortgage insurance, no annual fee. Over thirty years this is usually the largest single number in the comparison.
• No household income ceiling, no geographic constraint, and no VA loan limit with full entitlement.
• Funding fee is one-time, financeable, and waivable entirely if the veteran is exempt.

The case for USDA:

• No entitlement arithmetic — a veteran with entitlement tied up in an existing VA loan may need a down payment under §17.3 and needs none here.
• Eligible closing costs may be financed to the extent the appraised value exceeds the purchase price — a cash-to-close tool that does not exist on VA.
• No funding fee at the VA's subsequent-use rate.

Recommendation: VA. The absence of a recurring monthly charge is decisive, and it compounds for the life of the loan.

The circumstance that reverses it: the veteran has partial entitlement and the §17.3 arithmetic produces a required down payment they do not have. USDA then delivers zero down where VA cannot, and the annual fee is the price of the transaction happening at all. (A weaker second case: your lender's VA overlay is materially worse for this file than its USDA overlay — but that is a shopping problem, and the better fix is a different lender, not a different program.)

Exercise 17.23

The response: "Neither of those is the question. Let me show you what each one costs you, and then you tell me which we run."

The figures that belong in it, from Chapter 13's comparison and this chapter's counterfactual:

\$47.30 a month cheaper, \$13,475.00 less cash at the table, and \$62,374.40 of mortgage insurance that never happens. Turn time is a real consideration and belongs in the conversation — but it belongs after those numbers, and if the honest answer is that VA will take a week longer, say so and let the borrower decide with both facts. A borrower who chooses speed knowing what it costs has made a decision. A borrower who chooses speed because nobody told them has been sold something.

Exercise 17.25

To the underwriter:

"Condition response: the COE dated [date] states the applicant is not exempt from the funding fee, and we are not asserting an exemption. The fee is disclosed and financed as reflected on the Loan Estimate. The applicant has a disability claim pending with the VA; no rating has been issued. Please confirm the condition may be cleared as 'not applicable — applicant is not exempt per COE.'"

To the borrower:

"Right now the VA's record says you're not exempt, so we've priced the funding fee into the loan and nothing about your approval changes. Here's the part I want you to write down: if your claim is granted later with an effective date before our closing date, a refund of that fee may be available. Call me when the rating letter arrives — even if it's two years from now — and I'll walk you through it."

What you do not do: assert an exemption you cannot document, hold the file waiting for a rating that has been pending eleven months, or let the borrower believe the fee is provisional.

Exercise 17.26 †

Nine days on the contract, eleven on the lock. Pre-1978 construction, so this is a lead-based-paint requirement and the work has to be done by someone who can do it correctly.

The step most likely to slip: step 7, and specifically the weather. Exterior paint has a temperature and moisture window that no amount of urgency overrides. That is why step 10 is dated day 2 rather than day 8 — you request the extension while it is a routine ask, not while it is an emergency.

The secondary slip risk is step 6. Ordering the re-inspection after the work is complete adds three to five days for no reason. Order it in advance, contingent on completion.

Exercise 17.27

At least four errors in "COE received — borrower approved for zero down up to the remaining entitlement amount."

1. "Approved." A COE establishes eligibility, not approval. Nothing about credit, income, assets, or the property has been evaluated. Writing "approved" in file notes on the strength of a COE is the kind of sentence that gets quoted back to you.
2. "Up to the remaining entitlement amount." Backward by a factor of four. The maximum zero-down loan is 4 × remaining entitlement, not the entitlement itself. As written, this note understates the borrower's zero-down capacity by seventy-five percent.
3. "Zero down" stated flatly. Above 4 × remaining entitlement, a down payment of 25% of the excess is required. The note omits the condition entirely.
4. It omits the funding fee. A subsequent use of entitlement carries a higher fee at the zero-down tier, and that fee changes the loan amount and the payment.

The correction: "COE received; eligibility established. One prior VA loan outstanding; entitlement partially charged. Remaining entitlement \$X. Maximum zero-down loan \$4X; above that, down payment = 25% of the excess. Subsequent-use funding fee applies. Not an approval — credit, income, assets, and property still open."

Exercise 17.29

The question, rewritten (any version keeping every working clause is acceptable):

"One eligibility question I ask everyone, because it's worth money and people miss it. Has either of you ever served in the military — active duty, National Guard, Reserves, any length of time, any era? And are either of you the spouse or the surviving spouse of someone who served?"

What each clause is for:

The follow-up to "My husband was in the Navy. He passed in 2019":

"I'm sorry. I ask because surviving spouses can be eligible for the VA home loan benefit, and a lot of people are never told. May I request the certificate? If it comes back eligible, it removes your down payment and your mortgage insurance — and if it doesn't, we're exactly where we were five minutes ago."

Then move to the next question. Do not linger, do not ask what happened, and do not make the borrower manage your reaction.

Exercise 17.30 †

(One acceptable version — roughly 200 words.)

Thanks for taking my call. I want to put three things in writing so your sellers can decide with facts.

Sellers are not required to pay a VA buyer's closing costs. The VA limits what a veteran may be charged and caps seller concessions as a percentage of value, but a cap is not a mandate. Costs are negotiated here the same as on any other contract, and my buyer is not asking your sellers for anything the contract does not already say.

The VA escape clause makes the bad scenario predictable. If the appraisal comes in below the contract price, my buyer may withdraw and recover their deposit — or pay the difference and proceed. Your sellers will know which within days of the value, in writing, rather than watching a financing contingency drift.

One thing I will not oversell. VA property requirements are real, and if the appraiser flags deteriorated paint or a mechanical issue, it becomes a repair and a re-inspection. In our market that runs one to two weeks. I would rather you hear that from me now than on day thirty.

My mobile is below. Call me directly, any day.

Exercise 17.31

What is legitimate in it. Turn times are real. Underwriter familiarity is real. A loan officer who knows their shop's FHA desk is faster than its VA desk knows something operationally useful, and Chapter 19 spends a whole chapter on why calendar risk is not a soft consideration.

What is not legitimate. The instruction routes an eligible borrower away from a benefit on the basis of the lender's convenience, and the borrower is never told. That has three problems:

1. It costs the borrower money — a monthly mortgage insurance premium the VA loan would not have had, potentially for the life of the loan at FHA's higher loan-to-values.
2. It is a pattern, and patterns are what examiners look at. Regulation B prohibits discouraging an applicant, and discouragement includes what is said before an application exists. Chapter 25.
3. It substitutes the loan officer's judgment for the borrower's on a decision that is entirely the borrower's to make.

What you do Monday. Present both programs with the numbers, including the honest turn-time difference, and let the borrower choose. If they choose FHA for speed knowing what it costs, document it. If the manager repeats the instruction as a directive rather than a suggestion, that is a compliance conversation, not a production conversation — take it to compliance, in writing, and keep the copy.

Exercise 17.33

The options, honestly:

1. Refer the veteran to a lender approved for VA. Costs you the file and the commission. Preserves the borrower's benefit entirely. Frequently returns to you as a referral relationship, because the borrower knows exactly what you did.
2. Originate an FHA or conventional loan for them without disclosing the comparison. Keeps the file. This is the option §17.8 and Chapter 25 are about, and it is not available to you.
3. Present the comparison honestly, tell them your employer cannot do VA, and let them decide. They may still choose to stay with you for reasons of their own — timing, trust, a relationship. If they do, the decision is documented and theirs.
4. Push your employer to obtain VA approval. Slow, and not this file's answer, but it is the answer for the next fifty.

The choice: option 3, falling back to option 1 if they want the VA loan — which they usually will, once they see the numbers. Option 3 is better than option 1 alone because it gives the borrower the information and the choice rather than assuming what they want.

What it costs you: probably this commission. What it buys: a referral source who tells the story, and a file you will never have to explain to an examiner. Chapter 38 makes the arithmetic of that trade explicit.

Exercise 17.34 †

Under VA rules, an IRRRL requires: prior occupancy and a funding fee (at the reduced IRRRL rate). It does not require an appraisal, income documentation, or a credit underwriting package. It does require an existing VA loan to refinance, and current occupancy is specifically not required.

The item most candidates get wrong is prior versus current occupancy. A veteran who bought at one duty station, was reassigned, and now rents the property out may still IRRRL it — because the certification is that they previously occupied it.

Why lender practice differs: what the VA does not require, lenders frequently add back as overlays (Chapter 14). Many require a credit report, a minimum representative score, and a clean twelve-month mortgage payment history; some require an appraisal or an automated valuation. None of that is a VA rule, which is exactly why the same IRRRL can be declined at one lender and approved at another — and why an exam answer and a Monday-morning answer can legitimately differ.

Layered on top are the statutory guardrails from 2018 legislation: seasoning, a net tangible benefit test, and fee recoupment (commonly stated as a 36-month window). Verify current standards.

Exercise 17.35

Effect on the funding fee: the veteran is exempt. No funding fee is charged.

Where the determination appears: on the Certificate of Eligibility, on its funding fee status line. That is the document the underwriter will condition for and the one the file relies on.

If the rating is granted after closing with a retroactive effective date preceding the closing date: a refund of the funding fee may be available. Pursue it through your lender's VA channels. This matters more than it sounds like it does — disability claims routinely take many months, and ratings are frequently made effective as of an earlier date, so the veteran who paid a fee in March may have been exempt in March and not learn it until November. They will not know to call you. Calling them is one of the highest-value things a loan officer can do with fifteen minutes.

Exercise 17.37

(One acceptable version.)

"It would have been a little cheaper — about \$65 a month, and you'd have kept your \$19,250 in the bank instead of putting it into the house, because a VA loan has no down payment and no monthly mortgage insurance."

"It wasn't available to you, so it isn't a missed opportunity; it's just a different program for different circumstances, and what we built is the right loan for the facts we actually have."

Note what the second sentence does not do: apologize, speculate, or imply that the borrowers lost something. They were not eligible. Nothing was lost.

Exercise 17.38 †

The Linden Street VA counterfactual, assuming the veteran is exempt from the funding fee.

Against the conventional loan as closed (PITI \$3,033.72): a monthly difference of \$3,033.72 − \$2,916.90 = \$116.82.

Over the first sixty payments: \$116.82 × 60 = **\$7,009.20**.

Cross-check against the non-exempt counterfactual. The financed \$8,277.50 funding fee accounts for \$8,277.50 × 0.00623871 = **\$51.64 of the monthly payment. \$2,968.54 − \$51.64 = \$2,916.90** ✓ — the same PITI, arrived at two ways.

What it implies about the intake question. The exemption is worth \$51.64 a month for as long as the loan lives — roughly \$18,590 over a full thirty-year term. The only thing standing between a borrower and that number is whether somebody asked, out loud, whether they receive VA compensation for a service-connected disability. That question takes eleven seconds and it is the single highest dollars-per-second activity described anywhere in this book.

Chapter 18

Worked solutions to the daggered (†) and odd-numbered exercises. Arithmetic is shown. Where an exercise asks for judgment, a defensible answer is given along with what a marker should be looking for.

Exercise 18.1 †

Appraisal — an independent, supported opinion of a property's market value as of a stated effective date, developed and reported by a licensed or certified appraiser under professional standards.

Appraised value — the figure that opinion concludes at, used together with the purchase price to set the loan-to-value basis, the lender taking the lesser of the two.

Appraisal gap — the additional cash a buyer must produce when the appraised value falls below the contract price, equal to the maximum loan-to-value ratio multiplied by the shortfall.

(The constraint against "worth" is not a word game. It forces the student to state that an appraisal is an opinion held by a specific person as of a specific date, which is the thing borrowers most consistently fail to understand.)

Exercise 18.3

The four states of the reconciliation box:

1. "as is"
2. subject to completion per plans and specifications (new construction or a renovation not yet finished)
3. subject to the following repairs or alterations
4. subject to the following required inspection (roof, septic, well, structural engineer, pest)

Only "as is" means collateral is finished. The other three each put work between the report and a closing: (2) the construction must be completed and certified; (3) the repairs must be negotiated, performed, paid for, and certified; (4) a specialist must inspect and report, after which the appraiser may or may not be able to conclude, and repairs may follow.

In each of cases 2 and 3, the certification comes on Form 1004D, generally from the original appraiser, and carries a re-inspection fee. §18.9's timeline shows the real cost: 23 days, of which 7 were the repair itself and 7 were an argument about who paid for it.

Exercise 18.5

Sales comparison approach — value derived from the adjusted sale prices of comparable properties that recently sold. This is the approach that decides the number on an owner-occupied single-family purchase, because closed sales of substitutes are the most direct available evidence of what the market actually pays, and because the principle of substitution says a rational buyer will not pay more than the cost of an equally desirable alternative.

Cost approach — site value plus the cost to build the improvements new, less depreciation. Genuinely important on (a) new construction, where the cost is known and comparable sales may be thin, and (b) unusual or limited-market properties where few true substitutes have sold.

Income approach — market rent capitalized through a gross rent multiplier. Genuinely important on (a) two-to-four-unit properties and (b) investment property, where the buyer is in fact purchasing a cash flow. Generally not developed on a one-unit owner-occupied purchase.

Exercise 18.7 †

The comparable has something the subject does not, so the comparable is superior. You subtract, and you subtract from the comparable — the subject has no price yet, which is what you are solving for. The mnemonic is CBS: Comp Better, Subtract (and its mirror, CIA: comp inferior, add).

At the \$25 per square foot below-grade finish adjustment used in Figure 18.1, the adjustment is 400 × \$25 = **−\$10,000**.

Effect on the subject's indicated value: it goes down. The comparable sold for a price that included a finished basement; strip that value out and what remains is the evidence of what a house like the subject sells for. Students who answer that the subject's value rises have reversed the whole method, and this is the single most common reversal in the chapter.

Exercise 18.9

Two legitimate reasons the value lands at contract price so often:

1. The appraiser is expected to analyze the contract, and a recent arm's-length agreement between a willing buyer and a willing seller on the subject property itself is market evidence — often the single best piece of evidence in the file.
2. Appraisal produces a supported range, not a point. If the contract price falls inside that range, concluding at it is the ordinary and correct outcome, not a compromise.

The inference that runs the other way: when a value comes in below contract, the appraiser had the contract price in front of them and knew exactly what the transaction needed. Concluding below it was the harder path and they took it. That does not make the report right, but it means you should read it as evidence rather than as an obstacle — which is the posture §18.8 and §18.10 both depend on.

Exercise 18.11

The appraiser reports seller-paid financial assistance because a buyer who needs money back at closing may have agreed to a higher price to get it, in which case the recorded sale price overstates what the market actually paid. It is a check on whether the price is real.

\$3,000 ÷ \$385,000 = 0.779%, call it 0.78%. Small, ordinary, and typical of the market — it is very unlikely to have inflated the price, and the appraiser would report it and move on.

When would you expect a comment? There is no bright line, and inventing one would be exactly the error §7.1 of this book's standards warns about. The honest answer is that the appraiser comments when the concession is out of line with what the comparable sales carried. That is a market question, not a percentage. As a working reflex: a credit large enough that a buyer might plausibly have traded price for it is large enough to expect a comment, and if the comparables all carried similar concessions, no adjustment follows because the market itself is transacting that way.

Exercise 18.12 †

Four things the borrower is not getting:

1. An independent opinion of whether the price is supportable. On a purchase, value acceptance uses the contract price as the value. Nobody checks it.
2. Any observation of the property's condition. Nobody looked at the house on the lender's behalf, or on anyone's.
3. A trigger for the appraisal contingency, if the contract carries one (Ch. 20). No appraised value means nothing to invoke — a protection the buyer negotiated for and then gave up at a step that felt administrative.
4. A report. There is no appraisal to receive, keep, or read later, and nothing to deliver under the ECOA valuations rule, because no valuation of that kind was developed.

The one you must put in writing every time: the home inspection recommendation. "No appraisal required" is heard as "the lender says the house is fine," and that inference has to be broken in writing. (A strong answer also puts item 1 in writing when the borrower bought in a competitive situation — see Case Study 18.2.)

Exercise 18.13 †

(a) As planned. Down payment 25% × \$540,000 = **\$135,000. Loan \$540,000 − \$135,000 = \$405,000.**

(b) After the appraisal. LTV is computed on the lesser of price or value, so on \$505,000. Maximum loan 75% × \$505,000 = **\$378,750. The price has not changed, so the required down payment is \$540,000 − \$378,750 = \$161,250.**

(c) The gap. \$161,250 − \$135,000 = \$26,250.

(d) Why it is smaller. The buyer is less leveraged, so the loan absorbs less of the shortfall. At 80% leverage the loan gives up 80 cents of each shortfall dollar; at 75% it gives up 75 cents. The rule predicts it directly:

cash gap = max LTV × shortfall = 75% × \$35,000 = \$26,250 ✓

The general statement: a larger down payment makes a low appraisal cheaper, and a smaller one makes it more expensive. That is the opposite of most students' intuition and it is the whole point of the leverage table in §18.7.

Exercise 18.15

Comparable A — \$432,000, closed 2 months ago, 1,980 sq ft, 4/2.5, 2-car, C3, no deck.

  time      2 mo x 0.30% = 0.60% x $432,000 = $2,592  ->  +$2,600
  GLA       2,050 - 1,980 = +70 sf x $55            ->  +$3,850
  ---------------------------------------------------------------
  NET                                                   +$6,450
  ADJUSTED  $432,000 + $6,450                        =  $438,450
  net %     $6,450 / $432,000                        =    +1.49%
  gross %   $6,450 / $432,000                        =     1.49%

Comparable B — \$455,000, closed 4 months ago, 2,190 sq ft, 4/3.5, 3-car, C3, 250 sf deck.

  time      4 mo x 0.30% = 1.20% x $455,000 = $5,460  ->  +$5,500
  GLA       2,050 - 2,190 = -140 sf x $55            ->  -$7,700
  bath      2.5 vs 3.5, one full bath                ->  -$6,000
  garage    2-car vs 3-car, one bay x $5,000         ->  -$5,000
  deck      comp has 250 sf, subject none            ->  -$4,000
  ---------------------------------------------------------------
  NET       +5,500 -7,700 -6,000 -5,000 -4,000       =  -$17,200
  ADJUSTED  $455,000 - $17,200                       =  $437,800
  net %     -$17,200 / $455,000                      =    -3.78%
  gross %   ($5,500+7,700+6,000+5,000+4,000)/455,000 =     6.20%

Comparable C — \$409,000, closed 6 months ago, 1,910 sq ft, 4/2.5, 2-car, C4, no deck.

  time      6 mo x 0.30% = 1.80% x $409,000 = $7,362  ->  +$7,400
  GLA       2,050 - 1,910 = +140 sf x $55            ->  +$7,700
  condition C4 -> C3                                 ->  +$9,000
  ---------------------------------------------------------------
  NET                                                   +$24,100
  ADJUSTED  $409,000 + $24,100                       =  $433,100
  net %     $24,100 / $409,000                       =    +5.89%
  gross %   $24,100 / $409,000                       =     5.89%

(b) The indicated range: \$433,100 to \$438,450.

(c) Most weight to Comparable A. It required the smallest adjustments by a wide margin — 1.49% gross against 6.20% and 5.89% — which means the fewest judgment calls stand between the comparable and the subject. It also matches the subject on room count, bath count, garage, condition, and deck, so essentially the only difference being priced is 70 square feet.

Model reconciliation sentence: "Greatest weight is given to Comparable A, which is the most similar to the subject in room count, bath count, garage capacity, and condition, and which required the smallest gross adjustment of the three at 1.49%; Comparables B and C bracket the subject in gross living area and support the indicated value."

(d) Yes, \$437,500 is supported. It sits inside the indicated range and is bracketed — below A and B, above C — and it is within \$950 of the least-adjusted indication, which is 0.22% of the price and well inside the precision of the method. What would change the answer: if Comparable A's adjustments were large rather than small, if two of the three comparables indicated meaningfully below \$437,500, or if the appraiser could not support the \$55 per square foot and \$9,000 condition adjustments from market data.

Exercise 18.16 †

(a) As printed (GLA adjustment reversed to +\$7,700):

  $455,000 + 5,500 + 7,700 - 6,000 - 5,000 - 4,000 = $453,200
  net adjustment as printed  -$1,800   (-0.40% of sale price)

As corrected:

  $455,000 + 5,500 - 7,700 - 6,000 - 5,000 - 4,000 = $437,800
  net adjustment corrected  -$17,200   (-3.78% of sale price)

**(b) The error moves Comparable B's indication by \$15,400**, which is **twice** the \$7,700 adjustment — you have to remove the wrongly-added \$7,700 *and* apply the correct \$7,700 deduction. Direction errors always cost double, which is why they are worth hunting for.

Note the detail that makes this a good exam question: the gross adjustment percentage is unchanged at 6.20%, because gross uses absolute values. A screen that looks only at gross adjustment will not catch a reversed sign. Only reading the grid catches it.

(c) Yes — §18.8's third category, an error in analysis. And the uncomfortable part: the correction hurts the borrower. Comparable B's indication falls from \$453,200 to \$437,800, so a reconciliation that leaned on the printed figure is now supported at a lower number.

You file it anyway. A reconsideration of value is a request for accuracy, not an advocacy tool, and a loan officer who reports only the errors that help their file has converted the process into precisely the thing appraiser independence exists to prevent. In practice this error is also likely to be caught by the lender's collateral review tools or the underwriter, at which point selective reporting has cost you your credibility on every future ROV as well.

Exercise 18.17

(a) Maximum loan 95% × \$378,000 = **\$359,100.**

(b) Required down payment \$385,000 − \$359,100 = \$25,900. Cash gap \$25,900 − \$19,250 = **\$6,650** — which is 95% × \$7,000, as the rule predicts. ✓

(c) Using the 30-year payment factor implied by the frozen file (\$2,341.94 ÷ \$365,750 = 0.00640311 at 6.625%):

  P&I     $359,100 x 0.00640311                       = $2,299.36
  MI      $359,100 x 0.58% = $2,082.78 / 12           =   $173.57
  taxes                                                  $385.00
  insurance                                              $130.00
  ------------------------------------------------------------------
  PITI + MI                                            $2,987.93

(Down from \$3,033.72 — the loan is smaller, so the payment is smaller.)

(d) Headline: reserves \$12,623.66 − \$6,650.00 = \$5,973.66, which is \$5,973.66 ÷ \$2,987.93 = 2.00 months.

More precisely, three costs computed on the loan amount fall: origination 1% from \$3,657.50 to \$3,591.00 (−\$66.50); the half point from \$1,828.75 to \$1,795.50 (−\$33.25); and eight days of prepaid interest at a per-diem of \$359,100 × 6.625% ÷ 365 = \$65.1791, so \$521.43 rather than \$531.09 (−\$9.66). Total credit \$109.41**, so the true additional cash is \$6,540.59 and reserves land at \$6,083.07 = 2.04 months.**

(e) Yes, advise closing — with the arithmetic stated. The back-end ratio improves to (\$2,987.93 + \$1,446.00) ÷ \$10,500.00 = 42.23%, and roughly two months of reserves is thin but real, unlike the \$372,000 counterfactual in the chapter where reserves fall to about \$477. The sentence to say out loud: "You can do this, and it costs you two months of cushion instead of four — so before you say yes, tell me what happens in your household if the water heater goes in March."

Exercise 18.19 †

(a) Maximum loan 80% × \$598,000 = **\$478,400. Required down payment \$620,000 − \$478,400 = \$141,600. Gap \$141,600 − \$124,000 = \$17,600** — check: 80% × \$22,000 = \$17,600. ✓

(b) Loan \$620,000 − \$124,000 = \$496,000. LTV = \$496,000 ÷ \$598,000 = 82.94% (LTV is computed on the lesser figure, the appraised value).

(c) \$496,000 × 0.32% = \$1,587.20 per year ÷ 12 = \$132.27 per month.

(d) The loan lands in the 80.01%–85.00% mortgage insurance band. There is no cheap partial step here. The only boundary below the current LTV is 80.00%, and reaching it requires a loan of 80% × \$598,000 = \$478,400 — which is the full \$17,600. Contrast Cypress Court, where the LTV landed at 85.54%, barely inside the 85.01%–90.00% band, so \$2,750 of additional cash crossed a boundary.

The transferable lesson: before you recommend the sixth path, look at where the LTV lands relative to the band boundaries. Sometimes a small amount of cash crosses one and sometimes there is no boundary to cross, and you cannot tell without computing it.

Exercise 18.21

  New price                                            $530,000
  Loan (unchanged; 80% of the $505,000 value)          $404,000
  Buyer's down payment  $530,000 - $404,000            $126,000
  Increase over the planned $108,000                   +$18,000
  Seller gives up       $540,000 - $530,000            +$10,000
  ----------------------------------------------------------------
  Total                 $18,000 + $10,000               $28,000  ✓
  Buyer's share         $18,000 / $28,000                64.29%
  Seller's share        $10,000 / $28,000                35.71%

Advice: it depends on one fact the problem does not give you — how much liquid cash the buyer has after closing costs, prepaids, and reserves. \$126,000 down is \$18,000 more than planned, and whether that is a stretch or a rounding error is the entire question. If reserves survive, accepting a 64/36 split to keep a house they want is a defensible trade; if it empties them, the counter to make is \$526,000 (the even split of the gap), with the arithmetic attached so the seller's agent can see it is even.

The second thing worth naming: this is a negotiation, and the buyer's willingness to walk is their only leverage. A loan officer who has already told both agents "the lender is indifferent" has strengthened that leverage without saying anything untrue.

Exercise 18.23

The monthly cost of option 6, as usually quoted: mortgage insurance \$108.00 plus additional principal and interest on \$28,000 of \$179.29 = \$287.29 per month more than option 1.

The honest restatement: part of that \$179.29 is principal the borrower gets back. In the first month, interest on the extra \$28,000 is \$28,000 × 6.625% ÷ 12 = \$154.58, and the remaining \$24.71** is principal. So the true carrying cost of keeping \$28,000 in the bank is about \$154.58 of interest plus \$108.00 of mortgage insurance = \$262.58 a month**, and the mortgage insurance ends at the termination point under the Homeowners Protection Act (Ch. 5) while the interest continues.

Say it that way to a borrower and they can actually decide. Say it as "\$287 a month" and they are comparing a payment to a payment, which is not the trade.

What removes option 6 entirely: no debt-to-income room. The larger payment has to fit inside the back-end ratio the underwriter already approved.

The thirty-second check: take the approved back-end ratio, add \$287.29 ÷ gross monthly income expressed in percentage points, and compare to the approved maximum in the findings. On a household earning \$12,000 a month that is 2.39 percentage points — a file approved at 44% becomes 46.4%, and option 6 does not exist. On a household earning \$20,000 a month it is 1.44 points and probably survives. Run it before you offer it.

Exercise 18.25 †

What you submit, in order:

1. First, verify it yourself. Pull the county property record card and confirm it actually says 2,742 square feet. Twenty minutes of your own work before you file anything — a "factual error" that turns out not to be one costs you credibility you will need later.
2. Assemble the documents: the county property record card, the recorded plat, and the builder's floor plan with exterior dimensions. Three independent sources beat one.
3. Write a one-page cover note stating the discrepancy factually: the report states X, these attached sources state Y, difference Z square feet.
4. Submit it the same day, through the lender's appraisal desk or the AMC, which routes it to the assigned appraiser. Never directly to the appraiser (§18.2).
5. Tell the borrower and the agents what you filed and what it is worth — on Cypress Court, if accepted alone at the report's own \$55 per square foot, +\$7,260 of value moves the gap from \$28,000 to \$22,192. It helps; it does not fix it.

Three things you must not include, and why:

• A target value or a request to reach the contract price. This converts a factual submission into an attempt to influence value, which is the thing appraiser independence prohibits, and it gives the appraiser a reason to decline the whole request.
• The borrower's or the agent's opinion of value, a comparative market analysis, active listings, or a consumer website's estimate. None of these are facts about the property, and including them signals that the request is about the outcome rather than the evidence.
• Any characterization of the appraiser — their competence, their familiarity with the area, their motives. It is unprofessional, unprovable, and it lives permanently in a loan file that can be produced.

(A fourth, less obvious one: do not include your closing date or your lock expiration as an argument. The appraiser's obligation is to the accuracy of the report, not to your calendar, and saying so out loud reveals what the request is really about.)

One nuance worth flagging for markers: showing the arithmetic that follows from the appraiser's own stated adjustment schedule is accepted practice at many lenders — it is the appraiser's number, not yours. Asking for the resulting value conclusion is not. Some lenders' ROV forms prohibit any value arithmetic at all. Follow your lender's form.

Exercise 18.27

The form: 1004D, in its Appraisal Update capacity (the same form's other half is the completion certification — one form, two jobs, and the most reliable confusion in this chapter).

Who performs it: normally the original appraiser, which means their availability is now on your critical path.

What it does: states whether the property has declined in value since the effective date of the original appraisal, with support.

What it does not do: it does not produce a new opinion of value, does not re-appraise the property, and does not raise or lower the original number. It answers one question — has it declined? — and nothing else.

The one question to ask your underwriter first: "Do you need an update or a new full appraisal?" That depends on how old the report is against the investor's limits and on your lender's own overlay (Ch. 14), and ordering the wrong one costs a week and a fee. Ask before you order. Ask the same day.

Exercise 18.29

(a) In the next two hours, in order:

1. Stop. Do not forward the report onward with your own commentary attached.
2. Read the sentence carefully and record it verbatim, with the section of the report and the date, in the file.
3. Escalate to your compliance or fair-lending function the same day, through your lender's defined channel.
4. Follow their instruction on whether the report goes back through the appraisal desk for revision, and on what, if anything, is said to the borrower and when.

(b) What you do not do: contact the appraiser yourself; characterize the appraiser's motives in writing or out loud; decide the sentence is harmless because the value supports the contract; delay because the file is closing Friday; or handle it informally with a phone call to someone you know.

(c) Inside your company: compliance / fair lending, and the appraisal desk at compliance's direction. Outside: potentially the state appraiser regulatory agency — the reporting obligation that attaches when there is a reasonable basis to believe an appraiser has failed to comply with applicable law or has engaged in unethical or unprofessional conduct runs to the regulator, and your compliance department directs how and whether that reporting happens. The borrower separately retains their own right to complain to the appropriate federal regulator and to the state board.

(d) Why "the value is fine, so it doesn't matter" is the wrong analysis. Three reasons. The prohibited consideration is a problem regardless of the number — the rule is about what may be considered, not about the outcome. The report is a document, delivered to the applicant under the valuations rule and retained in the file, and it will be read by people other than you. And a pattern is only visible in aggregate: you are looking at one report and you are not in a position to know whether this is one sentence or a practice. Deciding it is harmless is exactly the decision you are not qualified to make.

Exercise 18.31 †

A model answer. 194 words.

Subject: Cypress Court appraisal — the numbers and your options

Following up on our call so you have this in writing.

The appraisal came in at \$505,000**. Your contract is **\$540,000.

• The lender lends against the lower of the two. At 80%, the most it will lend is \$404,000, not \$432,000.
• The price has not changed, so your down payment goes from \$108,000 to \$136,000.
• The gap is \$28,000. That is the whole problem.

Your options:

1. **Bring the \$28,000.** Down payment \$136,000.
2. **Ask the sellers to reduce to \$505,000.** Your down payment would *fall* to \$101,000.
3. Split it. At \$526,000 you each absorb \$14,000; your down payment would be \$122,000.
4. Reconsideration of value. I found a square-footage error and I am filing it today. If it is accepted and nothing else changes, the gap goes to about \$22,200. It helps. It does not fix it, and I cannot promise the appraiser will accept it.
5. Terminate under your appraisal contingency. Your earnest money returns. The appraisal and inspection fees do not.

Your appraisal contingency expires [date]. Your rate lock expires [date]. The contingency is first.

Call me tonight, any time.

What a marker should look for: the number in the first line; all five options, not three; the ROV priced and explicitly not promised; both deadlines named with the order stated; no jargon; no editorializing about the appraiser; and an invitation to call that has no time limit on it.

Exercise 18.33 †

A model answer. 142 words.

I can't, and I want to be precise about why, because it isn't a technicality.

The rule doesn't prohibit sending the appraiser facts. It prohibits anyone with an interest in the loan closing from communicating with the appraiser about value, or selecting or influencing the selection of the appraiser. Sending a comparable directly, from me, on a report that has already been delivered, is a communication about value from the production side. The information being public doesn't change who is sending it or why.

What I'm doing instead: I've asked the agent for the MLS sheet and the closing date on Rosewood, and it goes into a formal reconsideration of value through our appraisal desk, along with a documented square-footage error I found on the county record card. That's the channel, and it's already moving.

If the instruction were repeated as a direction rather than a suggestion: put your refusal and the reason in writing, escalate to compliance the same day, and keep the record. A direction to violate appraiser independence is not a management decision you are obliged to follow, and the mandatory-reporting provisions run to the regulator, not to your branch. Markers should reward students who name the escalation without theatrics — the point is that the answer is documented, not that it is confrontational.

Exercise 18.35

I, II, and V.

• I. The fully executed purchase contract — permitted and expected. On a purchase the appraiser is generally expected to analyze the contract; withholding it is the error, not sending it.
• II. A list of permitted improvements with dates and invoices — permitted. Factual property data is exactly what the appraiser should have.
• III. A range of values that would allow the loan to close — prohibited. This is a communication about value from the production side and is the core of what appraiser independence forbids.
• IV. A note that the assignment is one of several the appraiser may receive this month — prohibited. Conditioning or implying future work on this report is an attempt to influence, and it is the precise mechanism Case Study 18.1 describes.
• V. Homeowners association documents for a condominium project — permitted. Factual, and the appraiser needs it.

The nuance in the word "directly." Even for the permitted items, the ordinary route is through the appraisal desk or the AMC, and many lenders require it. The question is testing which content is permissible; the practical answer on a real file is permissible content, correct channel, always.

Exercise 18.36 †

Higher-priced mortgage loans, under Regulation Z's appraisal requirements: a written appraisal performed by a certified or licensed appraiser, including a physical interior inspection of the property, with a copy provided to the consumer.

The second appraisal is required in certain resale ("flip") situations — where the seller acquired the property within a specified recent period and is reselling it at a significantly higher price. This is the anti-flipping mechanism the exam wants recognized, and the second appraisal generally may not be charged to the consumer. Verify the current time periods, price-increase thresholds, and exemptions; they are revised.

Exercise 18.22 †

(a) Which options survive on \$118,000 of liquid funds.

  Option 1  buyer pays the gap    needs $136,000   short $18,000   OUT
  Option 2  seller to $505,000    needs $101,000   leaves $17,000  IN
  Option 3  split the gap, $526k  needs $122,000   short  $4,000   OUT
  Option 3b "split the difference", $522,500
                                  needs $118,500   short    $500   OUT
  Option 4  reconsideration of value                               always available
  Option 5  terminate                                              always available
  Option 6  keep $108,000 down, loan $432,000, LTV 85.54%, add MI  IN, if DTI allows

(b) The highest price this buyer can close at. Down payment = price − \$404,000, so price ≤ \$404,000 + \$118,000 = \$522,000** — and that leaves **\$0 for closing costs, prepaids, and reserves, which is not a closeable file. Stating a realistic ceiling requires an assumption: allow, say, \$14,000 for costs and prepaids and two months of PITI in reserves, and the workable ceiling falls well below \$510,000. Markers should reward students who state their assumption rather than students who produce \$522,000 with a straight face.

(c) Model recommendation, 108 words:

My recommendation is to ask the sellers to come to \$505,000 — the appraised value. Your down payment would be \$101,000, which is \$7,000 less than you planned, and it leaves you roughly \$17,000 for closing costs, prepaids, and reserves. I am filing the reconsideration of value today in parallel, on the documented square-footage error, but I am not counting on it.

If the sellers will only come partway, the fallback is to keep your \$108,000 down and borrow \$432,000 — that is 85.54% of the appraised value, so it adds about \$108 a month of mortgage insurance. I need to confirm your ratios support it before I offer it.

The risk being accepted: option 6 adds mortgage insurance and a larger payment, and it is only available if the back-end ratio has room — which has to be confirmed against the findings, not assumed. The second risk, worth naming out loud, is that option 2 depends entirely on the sellers, and if they refuse there is no version of this transaction the buyer can fund alone.

Exercise 18.28 †

Three errors.

Error 1 — the gross living area adjustment is reversed. The comparable has 1,940 square feet and the subject has 1,860, so the comparable is superior by 80 square feet and the adjustment must be subtracted: 80 × \$55 = **−\$4,400, not +\$4,400. (Comp better, subtract.)

Error 2 — the condition adjustment is reversed. The comparable is C2 and the subject is C3, so again the comparable is superior: **−\$7,000**, not +\$7,000.

Error 3 — the net adjustment does not foot. The line items as printed sum to \$4,400 + \$7,000 + \$2,400 = **\$13,800, not the +\$11,400** shown, and the adjusted sale price of \$413,400 was derived from that unsupported total. A grid that does not add up is not a small problem; it is the report failing at the one thing anybody can check.

(The time adjustment is correct: 2 months × 0.30% = 0.60% of \$402,000 = \$2,412, rounded to \$2,400.)

The corrected comparable:

  Sale price                                            $402,000
  Date of sale / time     -2 months                      +$2,400
  Gross living area       1,940 vs 1,860 = -80 sf x $55  -$4,400
  Condition               C2 vs C3                       -$7,000
  ----------------------------------------------------------------
  NET ADJUSTMENT                                         -$9,000
  ADJUSTED SALE PRICE                                   $393,000
  net adjustment as % of sale price                        -2.24%
  gross adjustment as % of sale price                       3.43%

All three are legitimate bases for a reconsideration of value — errors 1 and 2 fall under §18.8's third category (an error in the analysis) and error 3 is a straightforward arithmetic error in the report.

And the point of the exercise: correcting them moves this comparable's indication from \$413,400 to \$393,000 — **\$20,400 lower. The corrections hurt** the file. You report them anyway, for the reasons set out in the answer to 18.16, and you should expect the lender's collateral review tools to flag at least the footing error regardless of what you do.

Chapter 19

Worked solutions to the daggered (†) and odd-numbered exercises. Linden Street figures used throughout: income \$10,500.00/mo; PITI + MI \$3,033.72; other debts \$1,446.00; obligations \$4,479.72; back-end 42.66%; loan \$365,750.00 at 6.625%; P&I \$2,341.94; assets \$38,000.00; cash to close \$25,376.34; reserves after closing \$12,623.66.

Exercise 19.1

A conditional approval is a decision: the underwriter has evaluated the file and will lend on stated terms once a list of items is delivered. A suspended file has received no decision — underwriting cannot decision it as submitted. A denial is a decision, and it is no.

Suspense is the one that is not a decision. That matters practically (nothing to appeal, only something to fix and resubmit) and legally (the Equal Credit Opportunity Act's adverse action requirements attach to a decision on a completed application, not to a request for more information — Chapter 25).

Exercise 19.2 †

PTD (prior-to-doc) — must be satisfied before closing documents are drawn. Covers anything that could change a number on the Closing Disclosure or change the credit decision. PTF (prior-to-funding) — must be satisfied before funds are released; may be satisfied after documents are drawn. Covers items whose value is that they speak to the note date.

Exercise 19.3

A condition owner is the single named person responsible for seeing that a condition gets cleared — one person, with a phone number.

"The processing team" is not one because a team cannot be called, cannot be asked "where are we," and cannot be held to a date. More precisely: a team is a place where a condition can be believed to be handled by everyone and worked by no one. Ownerless conditions do not age slowly. They age at exactly the same rate as difficult ones, without producing anything.

Exercise 19.5 †

A documenting condition asks you to produce evidence of something the file already asserts. Nobody doubts it; it costs time, never approval. "Provide the most recent 30 days' paystubs for both borrowers."

An interrogating condition asks whether something is true, and there is a version of the answer that changes the loan. "Provide a letter of explanation and source documentation for the \$4,900 deposit; if borrowed, provide the terms and include the payment in the ratios."

The five signal words: if · unless · must not exceed · subject to · may not. A conditional clause in a condition means an answer exists that costs you the file as structured. Read those first, work them first, and never delegate one without understanding the exposure.

(Limit worth noting — see Case Study 19.2: a condition can carry zero credit risk and enormous schedule risk. Triage twice: once by credit risk, once by whether you control the counterparty.)

Exercise 19.7

Because a number supplied by an interested party verifies nothing. The purpose of the call is to confirm employment through a channel the borrower does not control; using the number the borrower gave you converts an independent verification into a self-certification with extra steps.

Acceptable sources: a published directory, the employer's main published line, or a third-party verification service. The log records the person contacted, their title, the date, the number called, and the lender employee who called.

Exercise 19.9 †

Classification, owner, and clearing document

Start order, and the defense of the top three

1. Condition 5 (project documentation). Longest and least controllable turnaround in the list; an HOA management company has no deadline of yours. It is also the only item that can produce a project-level finding — a defect that no amount of borrower cooperation fixes.
2. Condition 6 (prior-employer VOE). Third party, and the one most likely to fail silently. Order it and then confirm receipt with a human. Case Study 19.2 is this condition.
3. Condition 7 (judgment / identity). Third party, curative, and prior-to-doc — it changes what the security instrument attaches to, so nothing gets drawn while it is open.

Then 3 and 2 (interrogating, borrower-controlled), then 4 and 1, then the two PTFs scheduled for a midweek morning in the week of closing.

Note the pattern: the three you start first are the three you cannot chase. Credit risk sets the order in which you read; counterparty control sets the order in which you start.

Exercise 19.11 †

The question to send:

"Condition 1 — please confirm scope: (a) whose paystub, borrower, co-borrower, or both; (b) covering what period; and (c) does the year-to-date need to reconcile to the written VOE already in the file? I will send exactly what you specify today."

Why guessing costs more: a wrong guess is a round trip through the underwriting queue — your turn time, twice — and you learn nothing that a one-line question would not have told you in an hour.

Why sending everything costs more: it is also a round trip if the responsive document is buried in an unindexed upload, and it volunteers documents nobody asked for. Every extra bank statement is a new deposit to source; every extra paystub is a new figure that must reconcile. You have converted a documenting condition into two interrogating ones for free.

Exercise 19.13

In order:

1. Read the alert carefully and confirm it is inquiry-only: no new tradeline, no balance change on an existing account, no new public record.
2. Draft a single-question request and send it to the borrower: did this inquiry result in any new credit account, loan, lease, or financing — yes or no? Ask for a one-line signed, dated statement.
3. Submit the statement against the condition, indexed to the condition number, with the refresh report.
4. If the answer is yes, stop and treat it as a new liability: get the agreement or the creditor's statement, recompute the ratios, and notify the underwriter before you notify anybody else.

What you do not do: treat the inquiry as a debt and add an estimated payment to the ratios, and do not send an alarmed message that invites a long explanatory letter. An inquiry is not a debt. Over-conditioning on one frightens borrowers, produces defensive letters that volunteer new problems, and occasionally costs a closing over a credit pull the borrower never authorized and knows nothing about.

Exercise 19.14 †

Ranked, most costly first:

1. (b) A verbal VOE that discovers a job change. Days to weeks, and it may not be recoverable at all. A new employer means a new verification, possibly a probationary period, possibly a different pay structure, and a re-underwrite of qualifying income. This is the only item on the list that can end the loan rather than delay it.
2. (d) A title bring-down that finds a new lien. Days to weeks, and the counterparty is a stranger: a creditor who must be paid or must agree to release. You control none of it, and the file cannot record until it clears.
3. (c) A bank statement revealing a new \$4,000 deposit. Two to five days. It is an interrogating item — the answer could be a liability — but it is borrower-controlled and usually resolves with a statement and four sentences.
4. (a) A wrong mortgagee clause on an insurance binder. One to three days. Annoying, universal, entirely fixable, and the only thing standing between it and same-day resolution is whether somebody calls the carrier.

The ranking principle: cost rises with (i) how many counterparties must act and (ii) whether the finding is a fact about the borrower rather than a defect in a document. Document defects are cheap. Facts are not.

Exercise 19.15

Three approaches:

1. Find the surviving HR path. Acquisitions almost always leave a shared-services or payroll line, and the acquirer's payroll vendor often holds the records. Call the acquiring company's main published number and ask for employment verifications.
2. Third-party verification service. Many employers, including post-acquisition ones, report to commercial verification databases. A database hit is frequently acceptable and is instant.
3. Request an exception to alternative documentation — most recent paystub plus W-2 plus a signed 4506-C or transcript, in lieu of the verbal, with the underwriter's approval on the record.

Try 1 first, because it is free, fast, and preserves the condition as written — an exception you did not have to ask for is worth more than one you were granted. Start 2 in parallel the same hour; it costs nothing to run both. Save 3 for the day you can demonstrate 1 and 2 failed, because an exception request that arrives before you have exhausted the ordinary paths gets declined on process grounds.

Exercise 19.17 †

(a) Before and after

Housing ratio (unchanged by a non-housing debt):

$$\frac{\$2{,}268.00}{\$8{,}400.00} = 27.00\%$$

Before — obligations \$2,268.00 + \$714.00 = \$2,982.00:

$$\frac{\$2{,}982.00}{\$8{,}400.00} = 35.50\%$$

After — obligations \$2,982.00 + \$389.00 = \$3,371.00:

$$\frac{\$3{,}371.00}{\$8{,}400.00} = 40.13\%$$

(b) The move

$$40.13\% - 35.50\% = 4.63 \text{ percentage points}$$ $$\frac{\$389.00}{\$8{,}400.00} = 4.63\%$$

They are the same number, and they always are. A new payment moves the back-end ratio by exactly that payment divided by gross monthly income — nothing else in the calculation changes. This is the fastest mental check in origination: divide the payment by the income and you have the damage before you open a calculator.

(c) Headroom to 45.00%

$$0.4500 \times \$8{,}400.00 = \$3{,}780.00$$ $$\$3{,}780.00 - \$2{,}982.00 = \boxed{\$798.00}$$

The borrower had \$798.00 a month of room and used \$389.00 of it, leaving \$409.00. This file survives on the ratio — which does not mean it closes on time, because the approval was issued at 35.50% and must be re-decisioned regardless.

Exercise 19.19 †

Option A — retire the furniture account.

Obligations return from \$5,090.72 to \$4,479.72:

$$\frac{\$4{,}479.72}{\$10{,}500.00} = 42.66\% \qquad 48.48\% - 42.66\% = \mathbf{5.82 \text{ points}}$$

Option B — apply the same \$5,200 to principal.

Scale the payment by the loan amount, since rate and term do not change:

$$\frac{\$2{,}341.94}{\$365{,}750.00} = 0.00640312 \text{ per dollar borrowed}$$ $$\$5{,}200.00 \times 0.00640312 = \$33.30 \text{ per month}$$

(Check: the new loan of \$360,550.00 × 0.00640312 = \$2,308.64, and \$2,341.94 − \$2,308.64 = \$33.30.)

Obligations fall to \$5,090.72 − \$33.30 = \$5,057.42:

$$\frac{\$5{,}057.42}{\$10{,}500.00} = 48.17\% \qquad 48.48\% - 48.17\% = \mathbf{0.32 \text{ points}}$$

The comparison

$$\frac{5.82}{0.32} \approx 18\times$$

And Option B is worse than that ratio suggests, because the \$611.00 obligation is still there and the borrowers now need \$5,200 more cash at the table, from the same reserves.

The principle: per dollar, retiring a short-term installment debt is the most powerful ratio tool a borrower has at the end of a file, and adding to the down payment is one of the weakest — because a payoff removes a whole payment while principal reduction removes only the small slice of one payment that \$5,200 of a \$365,750 loan represents.

Exercise 19.21 †

$$0.250\% \times \$365{,}750 = \$914.375 \rightarrow \mathbf{\$914.38}$$ $$0.125\% \times \$365{,}750 = \$457.1875 \rightarrow \mathbf{\$457.19}$$ $$0.500\% \times \$365{,}750 = \mathbf{\$1{,}828.75}$$

Expressed in months of the \$60.14 payment difference between 6.625% and 6.375% (Chapter 1):

Interpretation. A quarter-point extension, bought because a finished file waited nine days for a date, costs the same as fifteen months of the rate advantage borrowers spend weeks shopping for. Read that alongside Chapter 1's argument: the rate is the thing people call about, and the calendar is where the money actually goes. (On the Linden Street file the lender absorbed the \$914.38 as a tolerance cure, so the borrowers never saw it — which makes it easier, not harder, to forget that somebody paid it.)

Exercise 19.23 †

March 28

RE: Loan #____ / [property address]
Underwriting condition ___ — source of $3,150.00 deposit, checking ...8802

The $3,150.00 credited to our checking account ending 8802 on March 12 is the
proceeds of the sale of my 2016 motorcycle to a co-worker. The sale is documented
on the attached signed bill of sale dated March 10 and the attached copy of the
cashier's check dated March 12. The funds are proceeds of the sale of personal
property. They are not borrowed, and no repayment is owed to any person.

______________________          Date: March 28
Borrower

Attachments: (1) signed bill of sale, 3/10  (2) copy of cashier's check, 3/12

Attachments named: the bill of sale and the cashier's check. Ideally also the signed-over title or the registration transfer, if the lender's condition asks for evidence of ownership; ask before sending it rather than volunteering it.

Why it works: four sentences. It answers the exact question, identifies the account and the date, names both attachments, states affirmatively that the funds are not borrowed (which is the condition's if clause, closed off), is dated and signed, and says nothing else. It creates no new condition because it contains no new fact.

Exercise 19.25

Subject: Six documents for underwriting — loan L-2214

We have the approval. Underwriting needs six things from you. Each one is a
single document; nothing here is a problem.

  1. The signed application and disclosure package — sign and return the
     e-sign link I sent this morning.
  2. Your two most recent paystubs, each of you.
  3. The signed 4506-C forms — same e-sign link, last two pages.
  4. A short letter about the $4,900 deposit from September 2, plus that
     commission statement. I'm sending you the wording separately; four
     sentences is right.
  5. The gift letter — I've attached it. Your parents sign it, and we also
     need a screenshot or statement page from their account showing the
     $10,000 leaving, and yours showing it arriving.
  6. Your homeowners insurance. Give the attached one-page sheet to your
     insurance agent — it has the exact wording they need. Don't retype it.

If everything comes back by Friday we are in great shape. Call me with any
question, including the ones that feel too small to ask.

Note what each item is: a document, not a concept. "Proof of the gift" produces a week of confusion; "the gift letter plus a screenshot showing the \$10,000 leaving your parents' account and arriving in yours" produces two attachments that evening.

Exercise 19.26 †

EXCEPTION REQUEST — Loan L-2214 · 4412 Linden Street

1  THE REQUIREMENT
   Lender overlay: 3 months' reserves required at LTV > 90% on primary
   residence purchases. [Cite your lender's overlay number and text verbatim.]

2  THE FILE'S FACTS
   Conventional 30-yr fixed, $365,750, LTV 95.00%, representative score 706.
   AUS Approve/Eligible. Qualifying income $10,500.00/mo. Housing $3,033.72.
   Total obligations $4,479.72. Housing ratio 28.89%, total debt 42.66%.
   Verified assets $38,000.00 less cash to close $25,376.34 less a $5,200.00
   documented payoff of a retail installment account = reserves of $7,423.66,
   which is 2.45 months of PITI.

3  THE ASK
   Exception to permit closing with 2.45 months' reserves in lieu of 3.00.

4  THE OFFSET
   - Housing ratio 28.89%, more than four points below the file's total-debt
     ratio and well inside any front-end benchmark.
   - 36 months of verified rental history at $1,850.00/month with no lates.
   - No lates in 24 months on any tradeline; no collections; no public records.
   - Both borrowers 3+ years in position; income is W-2 at both employers.
   - The reserve shortfall arises from a payoff that IMPROVED the file: it
     removed a $611.00 obligation and returned the back-end ratio from 48.48%
     to the 42.66% at which this loan was approved.

5  THE BUSINESS CASE
   Purchase, contract closing 10/24, lock extended to day 57. Referral source
   has closed four prior files with this branch.