Case Study 2 — When the Help Costs More Than It Helped

Type: Two labeled composites built from documented industry patterns, plus one real public regulatory episode Chapter: 33 — First-Time Homebuyers Focus: Three distinct ways assistance can leave a borrower worse off — recapture arriving at the same moment as negative equity, a rate premium that outruns the subsidy, and assistance that was never really a gift at all

Why this case exists

Case Study 1 argued that assistance reaches too few of the people it was built for. That is true and it is not the whole picture. Assistance is a trade, and like every trade in this book it has a break-even, a set of conditions under which it is the wrong choice, and a specific way it goes bad.

A loan officer who can only say "there's a program that'll cover your down payment" is selling. A loan officer who can say "here is what it costs you if you leave in two years, here is what it costs you if you stay for thirty, and here is the number of months at which the trade flips" is advising. This case study builds the three arithmetics you need to do the second thing.

Composites are labeled as such. No real borrower's file appears here, and no statistic is asserted that was not verified.

Part 1 — Composite File A: the recapture and the negative equity arrive together

[Composite, constructed from documented patterns. All figures illustrative.]

The file as originated. A single borrower buys at \$228,000 with FHA 203(b) financing and a \$12,000 forgivable county second at 0%, forgiven 20% per year over five years.

Check it: \$228,000 × 0.035 = \$7,980.00; \$228,000 − \$7,980.00 = \$220,020.00; the LTV is \$220,020.00 ÷ \$228,000 = 96.50%. \$220,020.00 × 0.0175 = \$3,850.35, so the total loan is \$223,870.35, and the annual premium factor on that total is \$223,870.35 × 0.0055 ÷ 12 = \$102.61. CLTV is (\$220,020.00 + \$12,000.00) ÷ \$228,000 = \$232,020.00 ÷ \$228,000 = 101.76%.

The event. At month 27, the borrower's employer relocates them. This is not a hardship, a default, or a bad decision. It is an ordinary thing that happens to working people.

What selling looks like. Assume the market rose 2% over those 27 months — a modest but real gain, not a crash:

COMPOSITE FILE A — the sale at month 27                     [composite; illustrative]

  Original price                                                  $228,000.00
  Sale price at +2%                                               $232,560.00
  Less selling costs at 7%                                       ($16,279.20)
  ────────────────────────────────────────────────────────────────────────────
  Net proceeds                                                    $216,280.80

  Payoff, first mortgage after 27 payments (approx.)             ($218,003.00)
  ────────────────────────────────────────────────────────────────────────────
  Shortfall before the second lien                                 ($1,722.20)

  DPA recapture: $12,000 less two years' forgiveness at $2,400     ($7,200.00)
  ────────────────────────────────────────────────────────────────────────────
  CASH REQUIRED TO SELL                                            ($8,922.20)

Now the part that matters, and it is not the part borrowers expect.

Run the arithmetic on the assistance alone. The borrower received \$12,000 at closing and **repaid \$7,200** at sale. On the DPA by itself, they are ahead by \$4,800. The forgivable second did exactly what its term sheet said it would do.

Compare it against the alternatives that were theoretically available on the same \$12,000:

The forgivable second was the cheapest structure available, by a wide margin, and it was chosen correctly.

So what went wrong? Nothing about the assistance. What hurt the borrower was the position the assistance made possible: a 101.76% combined loan-to-value on day one, which means that for several years the transaction costs of selling exceed the equity, and that the recapture arrives stacked on top of a shortfall rather than deducted from a gain.

And the actual failure was informational. Nobody ran this table at application. The borrower was told, accurately, that the second was "forgivable" and that they would not have a payment on it. The word forgivable did the rest of the work, and it is a word that borrowers hear as free.

The corrective is one page and twenty minutes. At application, on every assistance file, produce the sale-at-month-thirty table and hand it to the borrower. Say the number out loud: "If you have to move in two or three years, you will likely bring money to closing, and on these numbers it could be around nine thousand dollars. That's the real trade. It goes away if you stay."

Part 2 — Composite File B: assistance priced into the rate

[Composite, constructed to illustrate a documented structure. All figures illustrative.]

Here is the version where the assistance genuinely costs more than it saved.

Some — not all — housing finance agency first mortgages carry a rate above what the same borrower could obtain on a standard product from the same lender. The reason is structural rather than sinister: the assistance has to be funded from somewhere, and one of the places it is funded from is the yield on the first mortgage. Other agency products price at or below market. You have to run the comparison; you cannot assume either direction.

The comparison. Same borrower, same house, a total loan of \$220,000 either way.

The arithmetic: \$1,463.66 − \$1,390.55 = **\$73.11 per month**, which is what the \$8,000 costs.

COMPOSITE FILE B — the break-even on assistance      [composite; illustrative]

  Assistance received at closing                              $8,000.00
  Monthly rate premium                                           $73.11

  BREAK-EVEN: $8,000.00 / $73.11 = 109.4 months = about 9 years 1 month

  Cumulative cost of the premium:
    at 27 months     27 x $73.11 =    $1,973.97     borrower is AHEAD $6,026.03
    at 60 months     60 x $73.11 =    $4,386.60     borrower is AHEAD $3,613.40
    at 84 months     84 x $73.11 =    $6,141.24     borrower is AHEAD $1,858.76
    at 110 months                                   BREAK-EVEN
    at 360 months   360 x $73.11 =   $26,319.60     borrower is BEHIND $18,319.60

  A borrower who keeps this loan for its full term pays $26,319.60
  for $8,000 of help.

How to read that. The trade is good for a borrower who will move or refinance within about nine years, and it is bad for a borrower who stays. Which is to say it is a points break-even problem — the same calculation Chapter 13 runs on discount points and Chapter 29 runs on pricing — wearing different clothes.

Two honest complications, and you should raise both:

The borrower usually cannot choose. They do not have \$8,000. The comparison above is between a loan they can close and a loan they cannot, which makes "the standard product is cheaper after nine years" true and irrelevant. Say it anyway, because it sets up the next sentence.

The exit is a refinance, and the exit has a toll. The borrower who does well on this trade is the one who refinances into a market-rate loan once they have equity. But a refinance frequently triggers recapture on the DPA second (§33.4) and kills a mortgage credit certificate unless it is reissued (§33.6). So the escape hatch has a price, and the price should be on the same page as the break-even.

What a good loan officer says at application: "This program gets you into the house. It also costs you about seventy-three dollars a month in rate, forever, unless you refinance. In about nine years the rate premium adds up to more than the eight thousand dollars they gave you. So the plan is: take the program now, and let's talk in three or four years about refinancing out of it — and when we do, we have to check what that does to the second lien first."

That is a three-sentence answer that treats the borrower as somebody capable of planning. Most of them are.

Part 3 — The real one: seller-funded down payment assistance

This is public record and it is the reason several rules in §33.5 exist.

The structure. Through the late 1990s and 2000s, a set of nonprofit organizations operated a mechanism for providing down payment assistance on FHA loans. The buyer received a "gift" from the nonprofit sufficient to cover their required investment. The seller made a "donation" to the same nonprofit, generally equal to the gift plus a fee. The nonprofit was the intermediary, so on paper the gift did not come from the seller — which is what made it appear to satisfy FHA's rule against the borrower's required investment coming from an interested party.

What actually happened to the price. Sellers did not absorb the donation. They raised the price.

SELLER-FUNDED DPA — the mechanism             [composite; illustrative arithmetic]

  Arm's-length value of the home                                  $180,000
  Contract price written to fund the "gift"                       $187,000

  Buyer's required investment, 3.5% of $187,000        $6,545 -- paid by the "gift"
  Fee retained by the conduit                            $500 -- paid by the seller
  Seller nets: $187,000 - $6,545 - $500 =                             $179,955
                                          (essentially the $180,000 they wanted)

  Buyer's base loan: $187,000 - $6,545 =                              $180,455
  Financed upfront premium at 1.75%:                                   $3,157.96
  Buyer's total loan:                                                 $183,612.96

  Against the home's arm's-length value of $180,000, that is 102.01%.

  The buyer put nothing down, borrowed more than the house was worth in an
  arm's-length sale, and paid for the "gift" over thirty years at interest.

And the appraisal supported it. This is the part that makes the mechanism self-sustaining and genuinely dangerous. Once enough sales in a neighborhood are written this way, the comparable sales an appraiser relies on are themselves inflated by the same amount. The appraisal comes back at \$187,000 because other homes really did sell for \$187,000. The distortion validates itself.

What the record shows. The Internal Revenue Service issued guidance in 2006 concluding that organizations operating this way did not qualify for tax-exempt charitable status, on the reasoning that the funds circulated back from sellers rather than functioning as charity. HUD raised concerns over a period of years about the performance of loans made with seller-funded assistance, and government reviews of FHA's portfolio found materially higher default rates on those loans than on otherwise comparable ones. Congress ultimately prohibited the practice for FHA-insured loans in the Housing and Economic Recovery Act of 2008.

(No default-rate figure is quoted here because none was verified for this book. The direction of the finding — materially higher — is documented; the magnitude should be looked up in the primary source before anyone repeats a number.)

What it cost the borrowers. Everything the chapter warns about, at once: no equity, an inflated basis, a loan larger than the property's arm's-length value, mortgage insurance premiums calculated on that larger loan, and a payment based on a price nobody would have paid in a clean transaction. When values fell, these borrowers were among the first and furthest underwater. The "gift" had been their own money, borrowed at 6-point-something percent for thirty years, with a fee taken out of it.

The rule that survives. Down payment assistance funded, directly or indirectly, by the seller or any other party with a financial interest in the transaction is not an acceptable source of the FHA minimum required investment. That is why the source of the funds is a documented condition on every assistance file, and why "where did this money come from, and can I prove it" is a question you answer before the underwriter asks it.

The lesson

Assistance is a trade with a break-even, and the loan officer's job is to compute it rather than to celebrate it.

The three composites give three different break-evens:

• File A — the forgivable second was the right structure, and the borrower still needed \$8,922.20 to sell at month 27, because a 101.76% CLTV takes years to unwind. The failure was that nobody produced the table at application.
• File B — the assistance was real and the rate premium that funded it was also real, and they cross at about month 110. The right answer was still to take the program, with a documented plan to refinance out of it and a warning about what that does to the second lien.
• Part 3 — when the assistance is funded by the person on the other side of the transaction, it is not assistance. It is a price increase with a bow on it, and Congress said so.

None of this is an argument against down payment assistance, and a reader who takes it that way has read it wrong. Assistance is frequently the only path into ownership for a household without family wealth, and Case Study 1 argues it is underused. The argument here is narrower and harder: "forgivable" is not "free," "assistance" is not "gift," and the number that tells the borrower which is which is one you can compute in twenty minutes and are obligated to show them.

Discussion questions

1. In Composite File A the borrower is ahead \$4,800 on the assistance itself and still needs \$8,922.20 to sell. Explain to a borrower, in three sentences and without the word "equity," why both of those things are true at the same time.

2. Composite File B's break-even is about 109 months. Chapter 13 computes a discount-point break-even of 60.3 months on the Linden Street file. Compare the two decisions: what is structurally the same, and what is structurally different about who is able to choose?

3. Recompute File A's month-27 sale assuming the market was flat rather than up 2%. How much cash does the borrower need? Now recompute assuming the DPA had been a deferred second instead of forgivable. Which change moves the number more?

4. In Part 3, the appraisal supported the inflated price because the comparable sales were themselves inflated. Chapter 18 covers appraisal. Was the appraiser wrong? What, if anything, could an individual appraiser have done — and what does your answer suggest about relying on appraisals as a fraud control?

5. A borrower asks you directly whether they should take a program that carries a half-point rate premium. They do not have the down payment without it. Write the answer you would actually give, including at least two numbers, without making a recommendation you cannot support.

6. Part 3 describes a structure that satisfied the letter of the rule (the gift did not come from the seller; it came from a nonprofit) while defeating its purpose. Chapter 27 covers fraud. Where, exactly, is the line between aggressive structuring and misrepresentation here — and what would you have needed to know, at the time, to be sure which side you were on?