Case Study 40.2 — The Branch That Broke on a Good Year

Type: clearly labeled composite, constructed from documented industry patterns Sources: Tier 3 for every figure below. Tier 1 only for the market history in Case Study 40.1, which supplies the conditions this composite is placed inside.

⚠️ This branch does not exist. It is a composite, assembled from the failure pattern §40.8 names — a team hired at peak volume, an office leased on peak revenue, and a cost structure set on peak income — and given arithmetic so the pattern can be measured rather than merely described. Every dollar figure in this case study is constructed. Basis points, salaries, lease rates, average loan size, and contraction percentages vary enormously by market, channel, and employer. Do not quote any of them. The structure is the transferable part, and the structure is real.

Why this case study exists

Case Study 40.1 describes what happened to the industry. This one describes what it felt like from inside a P&L, and it exists because the failure it models is not caused by a bad decision.

Every decision this branch manager made was defensible on the day it was made, supported by the numbers then in front of them, and approved by people above them who had the same numbers. That is the point. The most common terminal error in this business does not announce itself as an error. It announces itself as growth, and the branch that makes it looks, for about eighteen months, like the best-run branch in the region.

Read it with Chapter 26 open — that is where the branch P&L was built — and with §40.8's closing question in mind: what am I assuming that I have not written down?

Part 1 — The branch at the peak

BRANCH P&L -- PEAK YEAR                       [constructed composite -- not real data]

  PRODUCTION
    Originators                                                          9
    Closed units                                                       420
    Average loan amount                                        $318,000.00
    Volume                                                  $133,560,000.00
    Refinance share of units                            78%  (328 refi / 92 purchase)

  REVENUE
    Branch revenue retained, 220 bps of volume                $2,938,320.00

  COSTS
    Originator compensation, 110 bps of volume                $1,469,160.00
    Operations payroll -- 11 staff, fully loaded $74,000 each   $814,000.00
      (processors, closers, a funder, a post-closer, an assistant manager)
    Occupancy  $25,000/mo                                       $300,000.00
    Technology, licensing, compliance, other                     $96,000.00
    ----------------------------------------------------------------------
    TOTAL COSTS                                               $2,679,160.00

  BRANCH PROFIT                                                 $259,160.00

Two figures decide everything that follows, and neither one appears on the statement above.

Contribution per closed unit. Revenue less originator compensation is 110 basis points, which on a \$318,000 average loan is **\$3,498.00 per file**. That is what each closing contributes toward everything else.

Fixed cost. Operations payroll, occupancy, technology, and overhead total \$1,210,000 a year, and they cost the same whether the branch closes 420 units or 120.

🧮 Run the Numbers

The number the peak year was hiding. [constructed composite]

$$\text{break-even units} = \frac{\$1{,}210{,}000}{\$3{,}498.00} = 345.9 \rightarrow \textbf{346 units}$$

The branch closed 420. Its record year cleared break-even by 74 units — 17.6%.

Read that again. In the best year in the branch's history, a 17.6% decline in units would have taken it to zero. Nobody computed this figure at the time, because the branch was profitable and nobody computes break-even on a profitable business.

Part 2 — The three decisions

Mid-boom, with volume running at record levels and the pipeline full, the branch made three commitments. Each was reviewed and each was approved.

Decision 1 — four more operations hires. Turn times had lengthened, files were sitting, and two originators had complained that conditions were taking a week to clear. Adding four staff took operations payroll from \$814,000 to \$1,110,000. Every one of those hires was justified by work that genuinely existed.

Decision 2 — a larger office on a seven-year lease. The existing space could not seat the current headcount, let alone the three additional originators being recruited. Occupancy went from \$25,000 to \$38,000 a month — \$300,000 to \$456,000 a year — on a seven-year term, because the seven-year rate was better than the three-year rate.

Decision 3 — three more originators, recruited with signing bonuses. Not modeled below, because originators are paid on commission and largely pay for themselves. Their cost appears indirectly, in the operations capacity and the office square footage that Decisions 1 and 2 were made to provide.

New fixed cost: \$1,662,000.

⚠️ Where Deals Die

The break-even nobody recomputed. At the new fixed cost, and at the same \$318,000 average loan:

$$\frac{\$1{,}662{,}000}{\$3{,}498.00} = 475.1 \rightarrow \textbf{476 units to break even}$$

The branch now needed more units than its record year had produced. Not more than a typical year — more than the best year in its history, achieved in the most favorable rate environment on record.

That single line of arithmetic was available on the day the lease was signed. It required the lease rate, the headcount, and one division. Nobody performed it, because the branch was making money and the question felt like pessimism rather than diligence.

Part 3 — The contraction, computed

Rates rise. Apply Case Study 40.1's structure: refinance candidates are largely eliminated, purchase volume declines but survives, and the average loan size rises as the mix shifts toward purchases in an appreciated market.

Constructed contraction assumptions: refinance units fall to 12% of the prior level; purchase units fall to 85%; the average loan rises to \$352,000.

BRANCH P&L -- CONTRACTION YEAR                [constructed composite -- not real data]

  PRODUCTION
    Refinance units   328 x 12%                                      39.36
    Purchase units     92 x 85%                                      78.20
    ----------------------------------------------------------------------
    TOTAL UNITS                                                     117.56    (-72.0%)
    Average loan amount                                        $352,000.00
    Volume                                                   $41,381,120.00

  REVENUE
    Branch revenue, 220 bps                                     $910,384.64

  COSTS
    Originator compensation, 110 bps                            $455,192.32
    Operations payroll -- 15 staff                            $1,110,000.00
    Occupancy  $38,000/mo                                       $456,000.00
    Technology, licensing, compliance, other                     $96,000.00
    ----------------------------------------------------------------------
    TOTAL COSTS                                               $2,117,192.32

  BRANCH RESULT                                             ($1,206,807.68)

Contribution per unit rose — \$352,000 × 110 bps = **\$3,872.00, because the average loan is larger. It did not matter. Break-even at that contribution is 429 units; the branch closed 117.56**.

The revenue decline is not the interesting number. The interesting number is that only \$455,192.32 of \$2,117,192.32 in costs — about 21% — declined with it.

Part 4 — Four counterfactuals

Run the same contraction against four different branches. Same market, same borrowers, same rates.

Scenario 3 mix: 231 purchase and 189 refinance units at the peak; 189 × 12% = 22.68 plus 231 × 85% = 196.35, giving 219.03 units. Scenario 4 reduces operations from 11 staff to 7 (\$518,000) and non-payroll fixed cost to \$300,000.

Three things fall out of that table, and the third is the one worth carrying.

**The expansion cost \$452,000 a year.** \$1,662,000 − \$1,210,000. That is the price of Decisions 1 and 2, charged annually, in a year when the branch could not pay it.

The mix was worth more than the discipline. Moving from 22% purchase to 55% purchase improved the result by \$392,891.84 — nearly as much as reversing the entire expansion, and it required no cost cut at all. It required work done two years earlier.

No single lever saved the branch. Not cost discipline. Not the purchase mix. Only the combination, and only with a genuine reduction on top of both. This is the honest lesson and it is more useful than "cut costs": a 72% decline in units cannot be managed by any one decision. What the five protections in §40.8 actually buy you is not immunity — it is a loss small enough to fund and a structure you can shrink toward viability before the cash runs out.

Part 5 — What the lease actually was

The four hires could be reversed, painfully and slowly, and eventually they were.

The lease could not. At \$456,000 a year with five years remaining, the branch was carrying **\$2,280,000** of committed occupancy against a business that had just generated \$910,384.64 of total revenue.

That is the difference between a reversible commitment and an irreversible one, and it is the distinction §40.8's list is really about. Everything on that list — the referral base, the database, the niche, the reserves, the survivable cost structure — is a way of ensuring that when conditions change, your commitments can change with them.

📞 On the Phone

The regional manager, reviewing the lease proposal at the peak: "The seven-year rate is forty basis points better than the three. Why would we take the shorter term?"

The answer that was not given: "Because the seven-year rate is better and the seven-year obligation is worse, and we are pricing one and not the other. Show me the P&L at half these units. If it works at half, sign the seven. If it only works at these units, we are not buying a discount — we are selling optionality, and we are selling it at the exact moment it is cheapest to us and most valuable to us later."

That is not a clever answer. It is one division and one question, and the whole case study turns on whether anybody in the room asks it.

Part 6 — What the manager could see, and when

This branch had every number it needed, in time.

None of that required a forecast. None of it required predicting the rate cycle, which nobody reliably does. Every line above is a description of the business as it already was. The failure was not a failure of prediction; it was a failure to describe.

That distinction is the reason §40.8 asks what am I assuming that I have not written down? rather than what do you think rates will do? The first question is answerable. The second is not, and building a business that requires the second answer is the actual error.

The lesson

A cost structure is a forecast, whether or not anyone wrote it down as one.

Eleven operations staff and a \$25,000 lease are a statement that the branch expects at least 346 units a year, indefinitely. Fifteen staff and a \$38,000 lease are a statement that it expects 476. Nobody at this branch believed they were forecasting. They believed they were serving demand that existed, which they were — and the forecast was embedded in the commitment regardless.

Three practices follow, and they are cheap:

1. Recompute break-even units every time a fixed cost changes, before the commitment, and write the number next to the signature.
2. State the volume level at which each commitment still works, and compare it to the worst year in the business's history rather than the best.
3. Track the share of costs that are variable. At 21%, a revenue decline arrives alone. That percentage is knowable in any month and is almost never on a management report.

And one more, which belongs to the originator rather than the manager: your own household is a branch. It has revenue that is per-unit and variable, costs that are mostly fixed, and commitments of varying reversibility. Every question in this case study applies to it at one-tenth the scale, and §40.1's five to seven months of reserves is the household version of the answer.

Discussion questions

1. Every decision in Part 2 was defensible on the day it was made, and the case study says so explicitly. Does that mean nobody was at fault? Distinguish between a decision that was wrong and a decision that was unexamined, and say which of the three this was.

2. The four hires were reversible and the lease was not. Rank the commitments an origination business typically makes — staff, lease, technology contracts, marketing spend, originator signing bonuses, personal household costs — from most to least reversible. Then state what you would be willing to pay, in worse pricing, for reversibility.

3. Scenario 3 shows that the peak mix was worth nearly as much as the entire expansion. But the mix could only be changed two years before it mattered. What does that imply about when a business should invest in purchase referral relationships — and what does it imply about the cost of doing so at the moment they feel least necessary?

4. Only Scenario 4 turns a profit, and it requires all three levers. Is the honest conclusion that a branch simply cannot survive a 72% unit decline? If so, what exactly are §40.8's five protections protecting — and is "a loss small enough to fund" a satisfying answer?

5. The regional manager asked why the branch would take the shorter lease term. Write out the full analysis you would bring to that meeting, using only the numbers in Part 1, and state how long it would take you to prepare.

6. Case Study 40.1 says the industry's manufacturing cost is fixed while its revenue is per-unit. That is not unique to mortgage lending. Name two other industries with the same shape, and say what they do about it that mortgage lenders generally do not.

7. Apply the whole case study to a household: an originator whose income doubled in the boom, who bought a house and a car against it. Write the version of Part 6's table — the numbers that were available and what they would have shown — for that household. Then answer §40.8's question about your own.