Chapter 37 — Key Takeaways

The core claims

1. Purchase and refinance origination are two businesses sharing a license. A purchase loan arrives attached to a contract, a deadline, third parties, and somebody who vouched for you. A refinance arrives attached to nothing, competes against every advertised rate in the country continuously, and can be abandoned by the borrower at no cost at any moment. Nearly every operational difference between the two follows from that.

2. Refinance demand is a stock, not a flow, and stocks can be exhausted. The population of refinance candidates is the set of outstanding loans whose note rate exceeds today's rate by enough to justify the costs. Refinancing removes a household from that population permanently. When rates rise, the pool does not shrink proportionally — it empties, and stays empty for years.

3. The standard break-even formula is wrong three times, all in the same direction. Closing costs divided by payment reduction ignores the amortization reset, treats a term-extended payment reduction as a saving, and prices financed closing costs at face value. The errors compound.

4. A payment reduction is not a saving. Part of it is money the borrower has stopped paying toward their own principal — a transfer from their balance sheet to their checking account, made with borrowed money.

5. The mortgage insurance clock follows the loan, not the borrower. A refinance restarts it, and the cost of the reset is proportional to how far into the schedule the borrower already is. It is the single most frequently uncomputed figure in the transaction and it can exceed every closing cost on the Loan Estimate.

6. Net tangible benefit requirements exist because the gap between a lower payment and a lower cost was exploited. Program tests, investor overlays, and state statutes each impose one, none of them identical — and on the largest category of refinances, conventional rate-and-term, the only recoupment test that runs is the one the loan officer decides to run.

7. A purchase business can only be built before you need one. Referral relationships take quarters, and the month you need them is the month every other originator in your market is calling the same agents with the same story.


The formula that matters

The net position test. Pick a horizon $H$ in months. For each loan:

$$\text{net position} = \big(\text{cash paid through month } H\big) + \big(\text{balance owed at month } H\big)$$

Cash paid includes principal, interest, mortgage insurance, and any costs paid at the table. Lower wins. There is no adjustment factor.

What to use instead of the payment formula when you need a shortcut: compare total remaining cost over the term that ends on the same date. Existing loan's remaining payments plus remaining mortgage insurance, against the proposed loan's payments over the same months plus its mortgage insurance plus its costs plus any balance still outstanding at that date.

Three break-evens, three different questions. Out-of-pocket break-even answers "when have I got my cash back." Net-position break-even answers "when do I have more money." Balance crossover answers "when do I owe less." Say which one you are answering.


The four inputs, before you quote anything

  1. The actual note rate on the loan they have — from the note or the statement, not from memory
  2. All the costs of the new loan, including the ones hidden in a "no-cost" rate
  3. How long they will keep it — the only input you cannot look up, and the one that decides most files
  4. What happens to the term and the mortgage insurance

The three errors, priced

On a constructed file — \$278,074 remaining at 7.000% with 288 payments left, refinanced to 6.000% with \$6,000 of costs financed:

The pitch The truth
Break-even 20.5 months Net-position break-even month 32
"Saving \$292.74/month" | \$91.01 of it is principal they stopped paying
"\$6,000 of costs" | **\$12,952.80** paid over 360 months, 2.16×
Total effect \$38,319.12 more than doing nothing
The option nobody showed them (288-month term) \$38,145.60 less than doing nothing

One number typed into one field: \$76,464.72.


Key terms

purchase market · refinance market · rate-and-term refinance (limited cash-out / no cash-out) · cash-out refinance · streamline refinance · break-even (refinance) · net tangible benefit · churning · serial refinancing · home equity extraction · application volume · pivot


Rules of thumb worth keeping

  • Cash-out is defined by what the proceeds retire, not by whether a check is written.
  • Ask when the current loan closed, first, every time. Inside a year, the burden of proof is on the refinance.
  • Ask where they are on the mortgage insurance schedule. Ask it before you quote.
  • Order the value early, and tell the borrower in advance what value the analysis requires. A short appraisal in a refinance has no seller to renegotiate with.
  • Most lenders will write a custom term. If 288 months remain, ask for 288 months.
  • The "skipped payment" is not skipped. The interest for it is paid at the closing table.
  • Write the benefit down in one sentence, dated, at application, in a form a stranger could evaluate in three years.
  • Treat purchase-side development as a fixed weekly cost, paid in a boom especially — because when it feels affordable, it is already too late.

What you should be able to do Monday morning

Take an inbound refinance call, get the four inputs in under ten minutes, and tell the borrower — correctly, and out loud — whether the transaction is worth doing. Run the net position test at their stated horizon rather than dividing costs by a payment reduction. Compute what the mortgage insurance reset costs before anyone orders an appraisal. Present at least three terms, including one with a higher payment, and let the household choose. Write the net tangible benefit sentence into the file the same day. And block the hours on your calendar this week for purchase-side development that you will not want to spend, because the week you want to spend them is the week they are worthless.