Chapter 26: Compensation: How Loan Officers Get Paid — Basis Points, Comp Plans, and the LO Comp Rule

"Nobody in this business is neutral. The rules exist to make sure the places where you are not neutral are places where it cannot cost the borrower anything." — constructed; the working premise of this chapter

Overview

Here is a question almost no loan officer can answer on the first try: what did you earn on your last file, and what did that file cost your company to make?

Most originators know the first number approximately and the second not at all. That is a strange gap in a profession whose entire compensation is a percentage of somebody else's debt. It is also expensive. An originator who does not know their own per-file economics cannot tell whether a branch's 50/50 split is generous or predatory, cannot evaluate an offer letter, cannot say why the company will not let them cut their commission to save a deal, and cannot build an income model that survives the first quarter when rates move against them.

There is a second gap, and it is worse. Compensation in this business is not merely a business arrangement. It is regulated conduct, governed by a rule in Regulation Z that was written because of what originators used to do, and it is one of the few areas of mortgage law where the prohibited act is something a loan officer might do without ever feeling like a criminal: taking a little more money for placing a borrower at a slightly higher rate.

That is where this chapter starts — with the conduct, not the rule. Chapter 2 named the yield spread premium as one of the specific failures that produced the modern rulebook. Here we work it in dollars on the Linden Street file and show exactly how much a borrower paid so that an originator could earn a few thousand more. Then we take the rule apart: what "a term of a transaction" means, what a proxy is and how the two-part test actually runs, when you may be paid by the borrower and when by the lender, why you may generally not be paid by both, and what the anti-steering safe harbor requires you to put in front of a consumer.

Then we do the arithmetic nobody does. Basis points. Splits. Draws — recoverable and not, a distinction worth thousands of dollars in a slow first year. Tiers, and the pressure a volume cliff creates on the thirty-first of the month. W-2 versus 1099, and what each does to how your own income will be underwritten when you buy a house. A branch profit and loss statement built line by line on the Linden Street file, which will show something uncomfortable: that a loan can be profitable to the loan officer and unprofitable to the branch at the same time, from the same file, with both numbers correct.

Finally, an income model. Not a motivational one. A model with closings, average loan size, and basis points on one side and rent, taxes, and a bad quarter on the other.

In this chapter, you will learn to:

• Explain what the yield spread premium was and price its cost to a borrower in dollars
• Distinguish borrower-paid from lender-paid compensation and choose correctly
• Run the two-part proxy analysis on any factor a comp plan proposes to use
• State the dual compensation prohibition, its exception, and the anti-steering safe harbor
• Read your own compensation plan and find the four paragraphs that matter
• Compute compensation under straight, split, drawn, and tiered structures
• Build a per-file branch P&L and a personal income model that connects to your funnel

Learning Paths

🎓 Exam — §26.3, §26.4, and §26.5 are the testable core. Know the two prongs of the proxy test verbatim, know that loan amount is expressly permitted, and know that Regulation Z's "loan originator" is not the S.A.F.E. Act's. 🏠 New LO — §26.6 and §26.7. Read them before you sign an offer letter, not after. The word "recoverable" in §26.7 is worth more to you than anything else in this chapter. 🤝 Partner — §26.1 and §26.5. Referral partners should understand that your compensation does not change with the rate you quote, because that fact is the reason your advice is worth taking. 📊 Operations — §26.9 and §26.10. The per-file cost model is the whole section; everything a branch manager argues about resolves into it.

26.1 What the rule was written to stop

Before 2011, an originator could be paid more for putting a borrower in a worse loan. Not through fraud, not through a side agreement, not through anything hidden in a back office — through the published rate sheet, in the open, as the ordinary way the business worked.

The mechanism was called the yield spread premium, and understanding it takes ninety seconds.

A wholesale lender publishes a rate sheet. On that sheet, every rate carries a price. At one particular rate — par — the price is 100.000 and no money changes hands beyond the loan itself. Below par, the borrower pays discount points to buy the rate down. Above par, the loan is worth more than its face amount to the investor who will buy it, because it pays a higher coupon than the market requires. That excess value is real money, and somebody receives it.

Today that somebody is the consumer, in the form of a lender credit, or the creditor, as revenue. Before the rule, it could be the person sitting across from the borrower.

Read the Linden Street rate/point grid — the same grid Chapter 13 used to choose the structure — through a pre-2011 lens:

THE SHAPE OF A PRE-2011 WHOLESALE RATE SHEET     [reconstruction of a structure;
                                                  figures from this book's grid]
  A $365,750 loan, 30-year fixed, 30-day lock. Par is 6.750%.

  RATE     PRICE      WHAT IT MEANT THEN                WHAT IT MEANS NOW
  ────────────────────────────────────────────────────────────────────────────────
  6.500%    99.000    the rate was bought down;         the borrower pays
                      $3,657.50 came in                 $3,657.50 in points
  6.750%   100.000    par — nothing changes hands       par
  6.875%   100.375    lender paid the broker            creditor issues the
                      $1,371.56                         consumer a $1,371.56 credit
  7.000%   100.750    lender paid the broker            creditor issues the
                      $2,743.12 as YIELD SPREAD         consumer a $2,743.12
                      PREMIUM                           lender credit
  ────────────────────────────────────────────────────────────────────────────────
  The rate sheet did not change. The destination of the money did.

Now look at what that meant for a household.

🧮 Run the Numbers

What one notch of yield spread premium cost the borrower.

Take the Linden Street borrowers to par — 6.750% — and their principal and interest payment is \$2,372.25**. Take them to 7.000% instead, and it is **\$2,433.34.

	Rate	P&I	Price	Paid to the originator (pre-2011)
Par	6.750%	\$2,372.25 | 100.000 | \$0.00
One notch up	7.000%	\$2,433.34 | 100.750 | **\$2,743.12**
Difference	+0.250%	+\$61.09/month** | +0.750 | **+\$2,743.12

The borrower pays \$61.09 more every month for 360 months:

$$\$61.09 \times 360 = \$21{,}992.40$$

The originator received \$2,743.12**, once, at closing. The borrower paid back that \$2,743.12 in about forty-five months** — \$61.09 × 45 = \$2,749.05 — and then paid it again roughly seven more times over the life of the loan.

\$21,992.40 out of the household for \$2,743.12 into the originator's pocket. Eight dollars out for every one in.

And here is the part that made it lethal rather than merely expensive: nothing on the borrower's side of the table looked different. Same house, same closing, same stack of documents, a payment sixty-one dollars higher than it needed to be, and no line item anywhere that said this is why.

Two more features of the pre-rule world are worth naming, because both are now separately prohibited and the exam asks about both.

First, you could be paid twice on the same loan. A broker could charge the borrower a one percent origination fee and collect a yield spread premium from the lender for placing the loan above par. The borrower paid at closing and paid again monthly. That is the practice §26.4 addresses.

Second, there was no cap and no anchor. Because the compensation was set deal by deal, an originator could take a little on an easy file and a great deal on a borrower who did not shop — which meant the borrowers least equipped to compare offers systematically paid the most. This is where compensation and fair lending intersect, and Chapter 25 owns that intersection; note only that discretionary, per-transaction pricing has a documented history of producing disparate outcomes even where no originator intended one.

Disclosure was tried first. It did not work. Under the Real Estate Settlement Procedures Act, the Good Faith Estimate was redesigned to break out the credit or charge for the specific interest rate chosen, so that a yield spread premium would appear on the form. Borrowers still did not shop on it, still did not understand it, and the incentive it disclosed remained exactly as strong as it had been before anyone disclosed it. Chapter 24 covers RESPA's disclosure architecture; the lesson for us is narrower and harder: when the incentive is strong enough, telling the consumer about it is not a remedy. You have to remove the incentive.

Congress did. Title XIV of the Dodd-Frank Wall Street Reform and Consumer Protection Act — the Mortgage Reform and Anti-Predatory Lending Act — directed a prohibition on compensating a loan originator based on the terms of a residential mortgage loan. The Board of Governors of the Federal Reserve System issued the implementing rule, which took effect April 1, 2011 over litigation from industry trade associations that sought to enjoin it. Rulemaking authority transferred to the Consumer Financial Protection Bureau later that year, and the Bureau issued an expanded Loan Originator Rule in January 2013, generally effective January 10, 2014. The rule lives in Regulation Z at 12 CFR 1026.36, and practitioners call it the loan originator compensation rule, or just "LO Comp." (Verify current text and effective dates at the source; Regulation Z is amended regularly.)

What it did, in one sentence: it severed the connection between what the borrower pays and what the originator earns.

26.2 Borrower-paid vs. lender-paid compensation

There are exactly two places an originator's compensation can come from, and the rule cares intensely about which one it is.

Lender-paid compensation — LPC — means the creditor pays the loan originator organization. The rate is set in advance, in a written agreement between the creditor and the originator organization, and it applies uniformly to that creditor's loans for the compensation period. It cannot be adjusted deal by deal. The consumer does not write a check for it and will not find it as a line on the Loan Estimate — which does not mean it is free. The compensation is priced into the rate. A lender paying 250 basis points of compensation is a lender whose rate sheet reflects that it is paying 250 basis points of compensation.

Borrower-paid compensation — BPC — means the consumer pays the originator's compensation directly, as a disclosed charge, either out of pocket or from loan proceeds. Because the creditor is not funding the compensation through the rate, the pricing available on a borrower-paid transaction is generally better than on a lender-paid one at the same lender on the same day.

That trade-off is the whole substance of the choice:

Note the last row. It surprises people. Borrower-paid compensation is negotiated with the consumer, so it can differ from one transaction to another — but it still may not be set by reference to the interest rate, the points, the prepayment terms, or anything else that is a term of the transaction. The consumer's willingness to pay is not a loan term. The rate is.

The practical rule your compliance department will enforce: choose the compensation channel before you lock, document why, and do not switch. Switching after lock, when the market has moved, is exactly the fact pattern that makes compensation look like it varied with a term. Most shops prohibit it outright.

Now the part that matters most to a retail originator, and the part most frequently misunderstood.

⚠️ Where Deals Die

"The origination fee is my commission." It is not, and believing it will cost you a file and possibly a license.

On the Linden Street file, the borrowers paid a \$3,657.50** origination charge and **\$1,828.75 in discount points. Neither of those is the loan officer's compensation. Those are the creditor's revenue and the price of the rate, respectively. The loan officer is an employee of the creditor; what the loan officer earns is whatever their compensation plan says — and their plan says a fixed number of basis points of the loan amount.

Here is the test that makes it concrete. Suppose the borrowers had negotiated the origination charge to zero and taken a lender credit instead. The loan officer's compensation would not change by one cent. Suppose they had bought the rate down another full point. Same compensation. Suppose they had gone to 7.000% and taken the \$2,743.12 credit. Same compensation.

That invariance is the rule. It is also the single most useful sentence you will ever say to a borrower who suspects you are steering them, and §26.5 gives you the words.

The trap to watch: on this file the origination charge is \$3,657.50 and compensation at 100 basis points would also be \$3,657.50. Same number, entirely different money, arriving from different parties for different reasons. A number that appears twice in a file is the most reliable way to make an expensive mistake in this business. Label your quantities.

One further caution about lender-paid compensation that new originators discover the hard way. Because LPC is set for a period and applies uniformly, you cannot reduce your own compensation to save a transaction. The borrower is \$400 short at the closing table; you would happily give up \$400; you may not. Doing so would make your compensation vary based on this particular transaction, which is the thing the rule exists to prevent.

There is one narrow, heavily conditioned exception in the rule permitting a reduction in compensation to bear the cost of an increase in an actual settlement charge above the applicable tolerance — a tolerance cure, in the language of Chapter 22. It is narrow. It is not a general permission to discount. A pricing concession is not a tolerance cure, and the difference is a compliance question with a written answer at every lender in the country. Ask before you promise, not after.

26.3 The prohibition on varying with loan terms — and the proxy analysis

This is the technical center of the chapter and of the rule. Read it twice.

A loan originator's compensation may not be based on a term of a transaction.

Regulation Z defines a term of a transaction as any right or obligation of the parties to a credit transaction. That is broad on purpose. The interest rate is a term. Discount points are a term. The origination fee is a term. A prepayment penalty is a term. The presence or absence of an escrow account is a term. The loan's maturity is a term.

The prohibition reaches further than a single deal, too: compensation may not be based on the terms of multiple transactions by a single originator or by multiple originators. You cannot rebuild a prohibited incentive by averaging it. Paying a bonus on the average rate of a quarter's production is the same violation, spread out.

And then there is the carve-out that makes the whole system workable.

The amount of credit extended is expressly permitted

Compensation may be based on the amount of credit extended — the loan amount. The rule treats it as a permissible basis, expressly, and it may be structured as a fixed percentage of the amount of credit extended, which the rule permits to be subject to a minimum or maximum dollar amount.

This is why the entire industry is paid in basis points. It is the only volume-scaled basis the rule clearly blesses.

Read the permission carefully, though, because it is narrower than most originators assume. It is a permission for a fixed percentage — one number — possibly with a floor and a ceiling in dollars. A plan that pays 150 basis points on loans below \$200,000 and 100 basis points on loans above \$400,000 is not paying a fixed percentage; it is paying a percentage that itself varies with the loan size. That structure exists, it is common, and it is exactly the kind of design that gets a written compliance opinion rather than a shrug. The floor-and-ceiling structure — say, 125 basis points with a \$2,000 minimum and a \$9,000 maximum — is the version the rule most clearly contemplates.

The proxy analysis

Here is where the rule gets interesting, and where careless comp plans die.

A creditor cannot evade the prohibition by paying on something that is not a loan term but reliably tracks one. So Regulation Z defines a proxy. A factor that is not itself a term of a transaction is a proxy for a term if both of the following are true:

THE TWO-PART PROXY TEST                                  [Regulation Z, 12 CFR 1026.36]

  PRONG 1 — CONSISTENCY
  Does the factor CONSISTENTLY VARY with a term of a transaction
  over a SIGNIFICANT NUMBER OF TRANSACTIONS?
                              │
                        no ───┴─── yes
                         │          │
                    NOT A PROXY     ↓
                                PRONG 2 — CONTROL
                                Does the loan originator have the ability,
                                directly or indirectly, to ADD, DROP, or
                                CHANGE the factor in originating the
                                transaction?
                                          │
                                    no ───┴─── yes
                                     │          │
                                NOT A PROXY   PROXY  →  PROHIBITED

Both prongs. Not either. A factor that tracks the rate perfectly but that the originator cannot influence at all is not a proxy. A factor the originator controls completely but that has nothing to do with any loan term is not a proxy. It is the combination — a lever the originator can pull that moves a loan term — that the rule forbids.

Notice how well-designed that is. The test does not ask about intent. It does not ask whether the originator meant to steer. It asks whether the structure makes steering pay. That is the correct question, because compensation systems shape behavior in people who would sincerely deny being influenced by them.

Running the test on real factors

Work these. They are the ones that come up.

Loan amount. Permitted, expressly, as discussed above. It never reaches the proxy test because the rule carves it out. (And note the tension the rule accepts: bigger loans pay more, and an originator therefore has a mild interest in a bigger loan. The rule tolerates this because the alternative — flat fees on every loan — would make small loans uneconomic to originate, which harms exactly the borrowers who can least afford it. This is a deliberate trade, not an oversight.)

Interest rate, points, prepayment terms. Not proxy questions at all. These are terms. Prohibited directly.

Credit score. Prong 1: yes, decisively. Risk-based pricing means the representative score moves the rate, the price, or both, across essentially every transaction — that is what a loan-level price adjustment is, and Chapter 29 builds one. Prong 2 is where people argue, and where the Linden Street file settles the argument. The representative score on that file is 706, the lower of the two middle scores, even though Borrower 1's middle score is 742. An originator who is paid more on higher-score files has a direct financial interest in restructuring the application so that the 742 borrower stands alone. They can also pursue a rapid rescore, time the pull around a reporting date, or advise paying down revolving balances. The originator can change the factor. Both prongs satisfied. Most compliance departments treat credit score as a prohibited proxy and will not let a comp plan touch it.

Loan product type. Paying more on an adjustable-rate loan than a fixed, or more on a government loan than a conventional one. Prong 1: yes — product type determines the rate structure, the mortgage insurance, and often the fee structure. Prong 2: yes, emphatically — recommending a product is the core of what an originator does. Prohibited as a proxy, and several product features are arguably terms in their own right.

Whether the loan is held in portfolio or sold. The classic textbook proxy, and the rule's own commentary uses it. Portfolio loans and agency-eligible loans price differently; the originator can steer a borrower toward one or the other. Both prongs. Prohibited.

Geography. Paying originators in one state or one branch differently from another. Prong 1 is usually no: a property's location does not consistently vary with any particular loan term across a significant number of transactions. Prong 2 is usually no: the house is where it is, and the originator cannot move it. Generally permitted under this rule — with two loud caveats. If the creditor's pricing itself differs by region, prong 1 starts to bite. And fair lending has its own, entirely separate view of compensation that varies by geography, which Chapter 25 owns and which is not satisfied by clearing Regulation Z.

Referral source. Paying more on files from a particular partner. Under the proxy test this usually passes: the referral source does not consistently vary with a loan term. But this factor is governed by a different statute. RESPA Section 8 — Chapter 24's material — prohibits giving or accepting anything of value for the referral of settlement service business, and a compensation plan that pays more for loans from a settlement service provider is a Section 8 problem long before it is a Regulation Z problem. Clearing one rule does not clear the other.

Purchase vs. refinance. Genuinely contested, and a good place to see how a real compliance analysis works rather than a textbook one. Prong 1: does loan purpose consistently vary with a term? Cash-out refinances carry pricing adjustments; rate-and-term refinances often price close to purchases; the answer is "sometimes, and it depends on the creditor's own grid." Prong 2: can the originator change it? Rarely — a borrower buying a house is buying a house — but at the margin an originator advising on a cash-out refinance versus a second lien is choosing the transaction type. Some compliance departments permit different compensation on purchases and refinances; many do not. This is a written-opinion question at your shop, and the honest answer in a textbook is to tell you that it is one.

Number of loans closed, or total dollar volume. Permitted. An individual transaction's terms do not move because of how many loans you closed last month. This is the basis for tiered plans, which §26.7 works numerically.

Here is the whole analysis in one table. Keep it.

The last five rows deserve a sentence. Regulation Z's official commentary lists methods of compensation that are neither terms nor proxies, and it is a longer and more generous list than most originators realize: the amount of credit extended, an hourly rate for hours actually worked, whether the consumer is an existing or a new customer, a fixed payment for every loan, the percentage of applications that close, the quality and accuracy of the files submitted, the long-term performance of the originator's loans, and legitimate business expenses such as fixed overhead. There is also a separate, carefully bounded permission for profits-based compensation — bonuses and contributions paid out of a business unit's mortgage-related profits — subject to a limit tied to a percentage of the individual's total compensation, or available where the individual originated a small number of transactions in the preceding twelve months. Contributions to designated tax-advantaged retirement plans are treated separately again. (These thresholds are specific and are revised; verify the current text before relying on any of them.)

⚖️ Compliance Check

The proxy test is a design test, not a defense.

Loan officers encounter the proxy analysis after the fact — a comp plan lands in their inbox and they wonder whether it is lawful. That is the wrong moment. The test is written for the person designing the plan, and by the time you are reading it the design decisions have been made.

What you can do, and should:

1. Read every factor in your plan against both prongs. If any factor plausibly moves a loan term and you can plausibly move the factor, ask about it in writing.
2. Ask specifically about per-file adjustments. Chargebacks, pricing concessions, early payoff recoveries, and "manager discretion" are where prohibited variation hides.
3. Never propose your own reduction. "I'll cut my comp to make this work" is a sentence that sounds generous and is a rule violation at most lenders. Route it through your manager and compliance.
4. Understand the exposure. Violations of the compensation rule can carry civil liability under the Truth in Lending Act, supervisory and enforcement consequences from the Bureau or a state regulator, and — for an individual — the licensing consequences Chapter 3 describes. We do not quote penalty figures here because they are fact-specific and set case by case; the relevant fact is that this is a rule with individual consequences, not just corporate ones.

Requirements change and state law varies. Several states impose their own compensation restrictions on top of Regulation Z. Verify current requirements with your compliance department and your state regulator, and treat nothing in this chapter as legal advice.

🎓 NMLS Exam Watch

Three reliable traps in this material.

One: loan amount. Candidates who have memorized "compensation may not be based on loan terms" often mark loan amount as prohibited. It is expressly permitted. The distractor is usually phrased as "a fixed percentage of the loan amount," which is the permitted structure stated almost verbatim.

Two: the proxy test needs both prongs. Question stems give you a factor that varies with a term but that the originator cannot influence — the creditor's cost of funds, say, or the borrower's state of residence — and ask whether it is a proxy. It is not, because prong two fails. Read for both prongs every time.

Three, and this is the one that separates the top quartile: "loan originator" under Regulation Z is not "mortgage loan originator" under the S.A.F.E. Act. Different statutes, different purposes, imperfectly overlapping definitions, and candidates merge them constantly.

The S.A.F.E. Act definition — Chapter 3 owns it and you should reread it there — is conjunctive: an individual who takes a residential mortgage loan application and offers or negotiates terms, for compensation or gain. Both prongs. It defines individuals, because its purpose is licensing people.

Regulation Z's definition is disjunctive and much broader: a person who, for direct or indirect compensation or gain, takes an application, or offers, or arranges, or assists a consumer in obtaining or applying to obtain, or negotiates, or otherwise obtains or makes an extension of consumer credit for another person — and it expressly reaches referring a consumer to a loan originator or creditor, and advertising that one can perform origination services. It covers organizations as well as individuals, because its purpose is regulating compensation, and compensation flows to companies.

The consequences are real, not academic. A registered originator at a depository is exempt from S.A.F.E. Act licensing and fully covered by Regulation Z's compensation rule. Someone whose job is referrals may fall inside Regulation Z's definition without being an MLO for licensing purposes. If a question stem says "loan originator," read which statute it is asking about.

26.4 Dual compensation

The rule against being paid twice is short and has one important exception.

If a loan originator receives compensation directly from the consumer in connection with a transaction, no loan originator may receive compensation from any other person in connection with that same transaction. And the mirror image: if the originator is being paid by the creditor, the originator may not also collect compensation from the consumer.

This is the dual compensation prohibition, and it is the direct descendant of the pre-2011 practice named in §26.1 — the broker who charged a one percent origination fee and also collected a yield spread premium from the lender for placing the loan above par. The borrower paid at the table and paid again in the payment.

Three details matter in practice.

Payments from loan proceeds count as consumer-paid. A borrower who "finances" the origination charge into the loan has still paid it. The money's route does not change its source.

The consumer's payment cannot be quietly refunded by the creditor. A structure in which the consumer pays the originator and the creditor makes the consumer whole through the rate reassembles the prohibited arrangement out of permitted parts. Regulators look at substance.

Third-party credits require analysis, not assumption. The Linden Street file carries a \$3,000 seller credit toward closing costs. Whether a particular seller, builder, or real-estate-agent credit is treated as the consumer's payment for this purpose is a fact question your compliance department answers, and the answer can depend on how the credit is applied and disclosed. Do not reason your way to a conclusion in the car. Ask, in writing, before you structure around it.

The employee exception

Now the exception, which exists because without it borrower-paid transactions would be impossible.

Consider a mortgage brokerage doing a borrower-paid deal. The consumer pays the brokerage. If the prohibition applied without exception, the brokerage could not then pay its own loan officer, because the loan officer would be receiving compensation from a person other than the consumer — namely, the brokerage.

So the rule permits it: an individual loan originator may receive compensation from the loan originator organization that employs them, even where that organization is compensated directly by the consumer. The individual's compensation must still comply with everything else in this chapter — it cannot be based on a term or a proxy, and in practice it is the same basis-point plan the individual is paid on every other file.

Which produces the cleanest illustration of the rule's logic in the whole chapter:

DUAL COMPENSATION — WHO MAY PAY WHOM             [constructed teaching example]

  LENDER-PAID TRANSACTION
     consumer ──── pays nothing to the originator ─────────►  ✔
     creditor ──── pays the originator organization ───────►  ✔
     organization ─ pays its individual originator ────────►  ✔

  BORROWER-PAID TRANSACTION
     consumer ──── pays the originator organization ───────►  ✔
     creditor ──── pays the originator organization ───────►  ✘  PROHIBITED
     organization ─ pays its individual originator ────────►  ✔  (the exception)

  IN BOTH CASES
     the individual's compensation is the same basis points
     on the same loan amount, and does not move.

That last line is the point. The individual originator's pay is identical under both structures. Whatever the borrower-paid-versus-lender-paid analysis produces for the consumer, it produces nothing for the loan officer. Which is precisely why a loan officer can be trusted to run it honestly, and precisely why you should say so out loud when a borrower asks.

26.5 Anti-steering and the safe harbor

The compensation rule has a second, independent prohibition, and originators often do not know it exists.

A loan originator may not steer a consumer to consummate a transaction based on the fact that the originator will receive greater compensation from the creditor in that transaction than in other transactions the originator could have offered — unless the transaction is in the consumer's interest.

Two things about that sentence.

First, note the word steer and note that this is not the fair-lending steering prohibition. Same word, two different rules, two different statutes. Chapter 25 owns steering as a fair-lending matter — directing a borrower toward or away from a product or a neighborhood on the basis of a protected characteristic, and the disparate-impact analysis that follows. What §26.5 owns is narrower and selfish: steering driven by the originator's own compensation. A single act can violate both, and the defenses are unrelated. Clearing one does not clear the other.

Second, note the qualifier — unless the transaction is in the consumer's interest. The rule is not a prohibition on recommending the loan that happens to pay you more. It is a prohibition on recommending it because it pays you more. That is a state-of-mind standard, which is unsatisfying to prove and unsatisfying to defend against, which is why the rule provides a way out.

The safe harbor

An originator is deemed to comply with the anti-steering prohibition if the consumer is presented with loan options meeting the rule's specifications, for each type of transaction in which the consumer expressed an interest.

The rule identifies types of transaction by rate behavior: a loan whose annual percentage rate cannot increase after consummation (a fixed-rate loan), a loan whose annual percentage rate may increase after consummation (an adjustable-rate loan), and reverse mortgages as their own category. If the consumer expressed an interest in both fixed and adjustable financing, the options must be presented for both.

For each such type, the originator must obtain options from a significant number of the creditors with which the originator regularly does business, and must present the consumer with three specific loans:

THE ANTI-STEERING SAFE HARBOR — THE THREE OPTIONS   [Regulation Z, 12 CFR 1026.36(e)]

  For EACH type of transaction the consumer expressed an interest in:

  (A)  the loan with the LOWEST INTEREST RATE

  (B)  the loan with the LOWEST INTEREST RATE WITHOUT any of these features:
         · negative amortization
         · a prepayment penalty
         · interest-only payments
         · a balloon payment in the first seven years
         · a demand feature
         · shared equity or shared appreciation
       (for a reverse mortgage: without a prepayment penalty, shared equity,
        or shared appreciation)

  (C)  the loan with the LOWEST TOTAL DOLLAR AMOUNT of discount points,
       origination points, and origination fees

  PLUS: the originator must have a GOOD FAITH BELIEF that the consumer is
        likely to qualify for each option presented. And where more than
        three loans are presented for a type, the ones satisfying (A), (B),
        and (C) must be identified.

Look at what those three options are actually engineered to do. (A) is the cheapest monthly payment. (C) is the cheapest closing. (B) is the honest one — the cheapest rate available without the features that make a loan cheap today and dangerous later. A consumer shown all three has been shown the real trade space: rate against cash, and both against risk. That is the conversation Chapter 13 taught you to have, and the safe harbor is essentially a regulator writing down Chapter 13 and requiring it.

Two limits worth stating plainly.

The safe harbor is built for originators with a menu. Its machinery — options from a significant number of creditors you regularly do business with — is written for brokers. A retail originator employed by a single creditor cannot obtain options from other creditors and does not need to; the options come from that creditor's own product set, and the anti-steering prohibition still applies in full. Your compliance department has a documented procedure. Follow it, and document that you did.

The safe harbor protects against a steering claim. It does not make the recommendation right. Presenting three qualifying options and then talking the borrower into the one that suits you is compliant paperwork wrapped around the conduct the rule was written to stop.

📞 On the Phone

Borrower: "Can I ask something? How much do you make on our loan?"

The defensive answer: "That's between me and my employer." True, and you have just told a first-time buyer that there is something here you would rather they not see. They will spend the next three days on the internet.

The dishonest answer: "Nothing — the lender pays me." Also technically true in a lender-paid structure and deeply misleading, because the lender's payment is priced into their rate.

What actually works: "Good question, and you should ask every lender you talk to. I'm paid a fixed percentage of the loan amount, set by my company, and it's the same percentage on every loan I close. Here's what that means for you: I earn exactly the same on this file whether you take 6.625% and pay a half point, or 6.750% at par with no point, or 7.000% and take a credit back at closing. I've got no dog in that fight, which is why I'm going to tell you what I actually think. Do you want me to walk you through all three?"

Then walk them through all three.

Why the long version wins. You have (1) answered honestly, (2) explained the structural reason your advice is worth something, and (3) turned a suspicious question into the loan comparison you wanted to have anyway. And you have done it in about forty seconds. The originators who fear this question are the ones who have never worked out what the honest answer is.

One thing not to say: never quote a competitor's compensation, and never suggest that a competitor is steering. You do not know their plan. Talk about your own.

🔍 Check Your Understanding

1. A comp plan pays 100 basis points on conventional loans and 125 on government loans. Run both prongs of the proxy test and state the conclusion.
2. A borrower is \$400 short at closing. Your branch manager suggests you "give up a little comp." What is wrong with that, and what is the one narrow exception in the rule that is not this?
3. In a lender-paid transaction, who ultimately pays for the originator's compensation, and through what mechanism?

(2 is the one people get wrong. Reducing compensation on a single transaction makes compensation vary with that transaction — a term-based variation. The narrow exception in the rule concerns bearing the cost of an increase in an actual settlement charge above the applicable tolerance, which is a tolerance cure, not a pricing concession.)

26.6 Reading a comp plan

Most originators have never read their compensation plan closely. They read the basis-point number, they read the draw, and they sign.

The basis-point number is the least interesting thing in the document.

A compensation plan — usually an addendum to an offer letter, sometimes a standalone policy the employer can amend — answers seven questions. Find each one before you sign, and if you cannot find one, that is your first question.

1. What am I paid on? The base loan amount or the total loan amount? On a conventional loan there is no difference. On an FHA loan with a financed upfront mortgage insurance premium there is. The Harlow Street file has a base loan of \$207,475.00** and a total loan of **\$211,105.81 after the financed UFMIP. At 125 basis points:

Forty-five dollars is not the point. The point is that a plan paying on the total loan amount pays more on FHA than on conventional for the same house — and you should notice that, name it, and know that the rule permits it because financed UFMIP is part of the amount of credit extended, which is the one basis expressly carved out. Permitted and incentivizing at the same time. That is exactly why the anti-steering prohibition in §26.5 exists as a separate backstop: the compensation rule cannot neutralize every incentive, so a second rule governs what you do with the ones that remain.

2. When am I paid? On funding, on the investor's purchase of the loan, or on a payroll cycle following one of those? The difference is six weeks of cash flow, and in your first year six weeks of cash flow is the whole question.

3. What can be taken back? This is the paragraph nobody reads.

• Early payoff (EPO). If the borrower refinances or sells within a defined window — commonly measured in months from the first payment — the investor may claw back the premium it paid, and many plans pass some or all of that back to the originator. Ask whether yours does, and how it interacts with the compensation rule: a chargeback that reduces compensation on a specific transaction is exactly the shape the rule is suspicious of, which is why many compliance departments prohibit charging EPOs to individual originators at all.
• Early payment default (EPD). Similar mechanics, triggered by an early missed payment.
• Pricing concessions. Whether the branch or the originator absorbs a lock extension or a tolerance cure. See §26.2 and ask in writing.

4. Is there a floor or a ceiling? A minimum dollar amount per file makes small loans viable and is permitted. A maximum caps you on jumbo files. Both are common and both are permitted structures; neither should be a surprise.

5. Is the draw recoverable? §26.7 works this arithmetic in full. One word, thousands of dollars.

6. Is this a contract or a policy? Most compensation plans are policies the employer may amend prospectively. That is lawful and normal — the rule requires that changes apply going forward, not retroactively to loans already in the pipeline. What you are entitled to ask is how much notice you get and which pipeline the change applies to. A change that reprices loans already locked is a different animal, and you should ask about it before it happens rather than after.

7. What does the other half of the split buy? If you are on a split, get the list. Processing? Rent? Marketing? Leads? Licensing and continuing education? Errors-and-omissions coverage? A manager's time? Some splits buy a great deal. Some buy a chair.

📄 Read the File

text FIGURE 26.1 — "Seven paragraphs and a number" [constructed teaching example] THE DOCUMENT Loan Originator Compensation Plan Addendum, two pages, effective the first day of the following calendar quarter, signed by the originator and a company officer. Attached to an offer letter. THE CONTEXT A retail branch. The candidate has been told "125 basis points and a four thousand dollar draw" on the phone and is reading the actual document for the first time, three days before starting. WHAT IT SHOWS Compensation: 125.0 basis points of the TOTAL loan amount, minimum $1,500.00 per closed loan, maximum $9,000.00 per closed loan. Paid on the payroll cycle following the investor's purchase of the loan. Draw: $4,000.00 monthly, RECOVERABLE, offset against earned compensation, balance due on separation. Chargeback: full compensation recovered on any loan that pays off within 180 days of the first payment date. Plan may be amended by the company on 30 days' written notice, applying to loans locked on or after the effective date. Compensation does not vary by product, rate, points, or any term of any transaction. WHAT IT DOESN'T It does not say what happens to a loan already locked when the plan changes on day 20 of the notice period. It does not say who absorbs a lock extension. It does not define "total loan amount" for a loan with financed mortgage insurance. It does not say whether the 180-day chargeback survives the originator's departure. And it says nothing at all about the split, the overrides, or what the branch provides — because those live in a separate document nobody offered. THE DECISION Before signing: ask four questions in writing. (1) Does "total loan amount" include financed UFMIP? (2) Which pipeline does an amendment apply to — locked, or applied? (3) Who pays a lock extension? (4) Has compliance approved the 180-day chargeback as applied to individual originators, and will you put that in writing? Then read the branch agreement that was not attached. THE LESSON The number is the part of a compensation plan you can negotiate least and understand fastest. Everything that will actually surprise you is in the paragraphs about timing, recovery, and amendment.

Constructed. Real plans vary enormously by employer, channel, and state; the structure above is illustrative and is not a model document. Have your own reviewed by counsel if the stakes warrant.

26.7 Basis points, splits, draws, and tiers

Now the arithmetic. Everything in this section is illustrative — compensation levels vary enormously by channel, by market, by employer, and by year, and no figure here should be quoted as an industry benchmark.

The basis point

A basis point is one one-hundredth of one percent: $1 \text{ bp} = 0.01\% = 0.0001$. One hundred basis points is one percent. The unit exists because the differences that matter in this business are smaller than a percent and saying "zero point one two five percent" out loud eleven times a day is intolerable.

On the Linden Street loan of \$365,750.00:

$$1 \text{ bp} = \$365{,}750.00 \times 0.0001 = \$36.575$$

Note the third decimal. A basis point on this file is thirty-six and a half cents and a half cent more, which means every compensation figure in this chapter carries a rounding decision. Payroll systems round at the file level, so we round each file's compensation to the cent and then sum. Say so, show it, and never let a half cent disappear silently — a book that hides a half cent has taught you to hide one.

Straight plans

Two observations before we go on.

Plan A produces \$3,657.50, which is also the origination charge on this file. Coincidence. Different money, different payer, different reason, same number. §26.2 flagged this; it is worth flagging twice.

The steps between the plans are \$914.38 and \$914.37. Twenty-five basis points of \$365,750 is \$914.375, and the two roundings fall in opposite directions. Hold that number — it will reappear in §26.9 as something entirely different, and the fact that it does is the most useful thing in this chapter.

Splits

A commission split divides the compensation paid on a file between the originator and the branch or team. A 50/50 split on each of the three plans:

A split is not inherently worse than a straight plan. It is worse or better depending entirely on what the branch's half buys, which is question seven in §26.6. A 50/50 split where the branch supplies a dedicated processor, the lead flow, the office, the licensing, and a manager who underwrites your files before submission may be a better deal than 125 basis points straight where you supply all of it. Get the list.

Draws

A draw is an advance against future compensation. There are two kinds and the difference is the most consequential word in a new originator's offer letter.

• Recoverable: the draw is a loan. Every dollar advanced is offset against compensation you later earn, and an unrecovered balance is generally repayable on separation.
• Non-recoverable: the draw is a floor. In a month where you earn less than the draw, you keep the draw and the shortfall is forgiven.

Watch what that word does across a realistic first year.

🧮 Run the Numbers

A \$4,000 monthly draw, Plan A on a 50/50 split — \$1,828.75 per closed file.

A first-year originator ramps: one closing a month for three months, two in month four, three a month thereafter. Every file is a \$365,750 loan. Compensation is \$1,828.75 per file.

```text RECOVERABLE DRAW — the deficit [constructed teaching example]

MONTH FILES EARNED DRAW PAID MONTH NET CUMULATIVE OWED ───────────────────────────────────────────────────────────────────── 1 1 1,828.75 4,000.00 (2,171.25) 2,171.25 2 1 1,828.75 4,000.00 (2,171.25) 4,342.50 3 1 1,828.75 4,000.00 (2,171.25) 6,513.75 4 2 3,657.50 4,000.00 (342.50) 6,856.25 ← peak 5 3 5,486.25 4,000.00 1,486.25 5,370.00 6 3 5,486.25 4,000.00 1,486.25 3,883.75 7 3 5,486.25 4,000.00 1,486.25 2,397.50 8 3 5,486.25 4,000.00 1,486.25 911.25 9 3 5,486.25 4,000.00 1,486.25 0.00 10 3 5,486.25 4,000.00 1,486.25 0.00 11 3 5,486.25 4,000.00 1,486.25 0.00 12 3 5,486.25 4,000.00 1,486.25 0.00 ───────────────────────────────────────────────────────────────────── TOTALS 29 53,033.75 48,000.00 ```

Check the arithmetic. Twenty-nine files at \$1,828.75 is \$53,033.75. Twelve months of draw is \$48,000.00. The difference — \$5,033.75 — is what is paid above the draw across the year, all of it in months 9 through 12. The originator did not see one dollar above \$4,000 a month until the ninth month.

Now the same year with a non-recoverable draw. In months 1 through 4 the originator earned \$1,828.75, \$1,828.75, \$1,828.75, and \$3,657.50 — all below \$4,000 — and keeps the \$4,000 floor in each. From month 5 on, earnings exceed the draw and the two structures are identical.

$$4 \times \$4{,}000.00 + 8 \times \$5{,}486.25 = \$16{,}000.00 + \$43{,}890.00 = \$59{,}890.00$$

	Total paid, year one
Recoverable draw	\$53,033.75
Non-recoverable draw	\$59,890.00
The value of one word	\$6,856.25

And \$6,856.25 is exactly the peak cumulative balance in month 4. It has to be: the non-recoverable version forgives precisely the deficit the recoverable version recovers.

One more figure, because it is the one that reframes the job. Twenty-nine closed files at \$365,750.00 each is **\$10,606,750.00 of funded production. The originator's pay for producing ten and a half million dollars of mortgage debt, on this plan, in this year, was \$53,033.75**. Both numbers are true, and holding them together is the beginning of understanding your own business.

Tiers

A tiered plan pays a higher rate at higher production. Volume is a permitted basis — an individual transaction's terms do not change because of how many loans you closed — so tiers are common. The design question is whether the tier is marginal (each dollar of volume paid at its own band's rate) or retroactive (all volume paid at the highest band reached).

Take a constructed plan on monthly funded volume:

A TIERED COMPENSATION PLAN                        [constructed teaching example]
  up to $1,000,000 .................. 100 bps
  $1,000,000.01 to $2,000,000 ....... 115 bps
  above $2,000,000 .................. 130 bps

  A SIX-FILE MONTH: 6 × $365,750.00 = $2,194,500.00

  MARGINAL                                    RETROACTIVE
  ────────────────────────────────────        ─────────────────────────────
  1,000,000.00 @ 100 bps =  10,000.00         2,194,500.00 @ 130 bps
  1,000,000.00 @ 115 bps =  11,500.00                    =  28,528.50
    194,500.00 @ 130 bps =   2,528.50
                          ───────────                     ───────────
                            24,028.50                       28,528.50

  DIFFERENCE: $4,500.00 on the same six files.

Both structures are lawful. But look at what the retroactive version does to the sixth file.

At five closings — \$1,828,750.00 of volume — a retroactive plan pays 115 basis points: \$1,828,750.00 × 0.0115 = **\$21,030.63. The sixth closing takes the month over \$2,000,000 and repays everything at 130: \$28,528.50**. That one file is worth

$$\$28{,}528.50 - \$21{,}030.63 = \$7{,}497.87$$

— more than four times what a file is worth on a straight 100-basis-point plan, and it is worth that only if it funds by the last day of the month.

Nothing about that violates the compensation rule. Volume is a permitted basis and the plan does not touch a single loan term. But it is worth being honest about what it does to a human being on the twenty-ninth of the month with a file sitting at "clear to close, pending the verbal verification of employment." A cliff tier does not create a rule violation. It creates pressure to make one. The disciplined response is to know your own tier position by the twentieth, not the thirtieth, and to treat a file that is not ready as a file that funds next month. Chapter 39 owns pipeline discipline and it is the operational answer to this incentive.

26.8 W-2 vs. 1099 originators

Most originators in the United States are W-2 employees. Some are paid as 1099 independent contractors. The distinction changes almost everything about your economics and almost nothing about your regulatory obligations, and originators routinely get that backwards.

What it does not change: the compensation rule. Regulation Z regulates the compensation of individual loan originators without regard to how they are classified for tax purposes. A 1099 originator is a loan originator. Their compensation may not be based on a term or a proxy. They are inside the rule.

What it does not change: licensing. Chapter 3 owns this, and the short version is that S.A.F.E. Act licensing attaches to the individual and requires sponsorship by an employing entity that supervises them. Many states restrict or prohibit paying a licensed originator as an independent contractor, or impose supervision requirements that are difficult to satisfy at arm's length. State law varies enormously here — verify with your state regulator before you accept a 1099 arrangement, not after.

What it does change:

That last row is the one this book is uniquely positioned to explain, and it is the one that surprises originators when they go to buy their own house.

A W-2 originator paid on commission is a variable-income borrower — the same analysis Chapter 11 applied to Borrower 2 on the Linden Street file, whose commission income of \$19,800 and \$23,400 over two years averaged to \$1,800.00 a month. A 1099 originator is a self-employed borrower — the Fulton Avenue analysis, with a cash-flow worksheet, add-backs, and the rule that when income declined year over year the underwriter uses the lower figure.

That rule is not academic for a loan officer. Origination income is cyclical, and a rate cycle can cut it in half in eighteen months.

The loan officer applies for their own mortgage

An originator earned \$150,000** in a strong year and **\$118,000 in the year that followed, when rates rose and volume fell. They apply for a mortgage in the spring.

Income declined, so — exactly as on the Fulton Avenue file — the underwriter uses the lower figure. Qualifying income is **\$9,833.33**, not \$11,166.67.

$$\$11{,}166.67 - \$9{,}833.33 = \$1{,}333.34 \text{ of monthly income that does not count}$$

At a 43% back-end ratio, \$1,333.34 of income supports roughly \$573.34 of monthly obligation (\$1,333.34 × 0.43 = \$573.3362), which on a thirty-year loan is a meaningful amount of house.

The lesson runs both ways. It tells you what a declining-income borrower is up against, in a way you will not forget. And it tells you something about your own planning: the year after a good year is not the year to assume your income will be counted at last year's level. Chapter 11 owns the method; this is what it feels like from the other chair.

One more thing about classification, because it has a real and public history. Whether mortgage loan officers are exempt from the Fair Labor Standards Act's overtime requirements has been contested for years. The Department of Labor issued an opinion letter in 2006 concluding that typical loan officers fell within the administrative exemption, then withdrew that position in a 2010 Administrator's Interpretation concluding they generally did not. Industry challenged the reversal on procedural grounds, and in Perez v. Mortgage Bankers Association (2015) the Supreme Court held that an agency need not use notice-and-comment rulemaking to change an interpretive rule. The case is famous for its administrative-law holding rather than for settling the classification question, which remains fact-specific and litigated. (Public record; read the opinion and the current Department of Labor guidance rather than a summary — including this one.)

The practical takeaway for you is narrow: classification is not a preference and not a title. It is a legal determination based on what you actually do, and it has consequences for taxes, overtime, licensing, and how your own income will be documented. Ask your employer to explain their basis for it in writing, and if you are being offered 1099 status as a licensed originator, verify with your state regulator that the arrangement is permitted there.

26.9 The branch P&L and what your loans actually cost to make

We can now answer the question the chapter opened with. The Linden Street borrowers paid an origination charge of \$3,657.50. What did that loan cost to make?

Build it line by line. Every figure below is constructed. Real cost-to-originate varies enormously by channel, market, loan size, and year, and the industry's own published measures of it move by thousands of dollars from one quarter to the next. Do not quote these numbers. Run your own branch's.

Start with revenue, because two things about it are counterintuitive.

Discount points are not branch revenue. The borrowers paid \$1,828.75 for a half point to take the rate from 6.750% to 6.625%. That money bought a below-market coupon; it went into the loan's price when the loan was sold. It is not margin. An originator who counts points as branch revenue has double-counted the same dollars, and it is the most common error in this analysis.

The secondary execution is revenue, and it is allocated. The loan sells above par, and corporate credits some portion of that execution to the branch. On this constructed model, 150 basis points.

And then there is the six-day overrun.

BRANCH P&L — ONE FILE                             [constructed teaching example]
the Linden Street loan, $365,750.00, closed day 51
Loan officer compensated at 125.0 bps under Plan B. Every figure constructed.

  REVENUE                                              bps        dollars
  ─────────────────────────────────────────────────────────────────────────
  Origination charge, paid by the borrower            100.0      3,657.50
  Secondary-execution credit allocated to branch      150.0      5,486.25
                                                     ──────    ──────────
  Gross revenue                                       250.0      9,143.75
  Less: 15-day lock extension, branch-absorbed        (25.0)      (914.38)
                                                     ──────    ──────────
  NET REVENUE                                         225.0      8,229.37

  DIRECT COST OF PRODUCTION
  ─────────────────────────────────────────────────────────────────────────
  Loan officer compensation, 125.0 bps                125.0      4,571.88
  Processing, allocated                                          1,050.00
  Underwriting, allocated                                          785.00
  Closing, document preparation, funding                           540.00
  Technology: LOS, POS, pricing engine, CRM, e-sign                315.00
  Compliance, quality control, post-close audit                    265.00
  Verification and third-party costs not passed through            180.00
                                                               ──────────
  Subtotal, direct                                               7,706.88

  BRANCH AND CORPORATE OVERHEAD
  ─────────────────────────────────────────────────────────────────────────
  Occupancy: rent, utilities, insurance, telephone                 690.00
  Branch management and administrative support                     615.00
  Corporate allocation: capital markets, HR, legal                 480.00
                                                               ──────────
  Subtotal, overhead                                             1,785.00

                                                               ──────────
  TOTAL COST TO MAKE THIS LOAN                                   9,491.88
  NET REVENUE                                                    8,229.37
                                                               ──────────
  BRANCH RESULT ON THIS FILE                                    (1,262.51)

  Footing: 4,571.88 + 1,050.00 + 785.00 + 540.00 + 315.00 + 265.00 + 180.00
           = 7,706.88.  690.00 + 615.00 + 480.00 = 1,785.00.
           7,706.88 + 1,785.00 = 9,491.88.  8,229.37 - 9,491.88 = (1,262.51).

  Rounding, shown rather than hidden: 225.0 bps of $365,750.00 is $8,229.375,
  and 25.0 bps is $914.375. The extension rounds up to $914.38, so net revenue
  lands one cent under a clean 225.0 bps. That penny is real and is carried
  through every figure below.

Read the bottom line and then read the compensation line, and hold both.

The loan officer earned \$4,571.88 on this file. The branch lost \$1,262.51 on it. Both statements are computed from the same document and both are true. That is not a paradox and it is not an accounting trick. It is the structure of the business: the originator's compensation is a cost of production, incurred whether or not the file was profitable, and the originator's own economics are almost entirely decoupled from the branch's.

Now the number to hold from §26.7. The lock extension cost \$914.38. Twenty-five basis points of compensation is \$914.375. They are the same money. A quarter point of comp and fifteen days of lock extension are, on this file, indistinguishable amounts — which means that six days of delay cost the branch exactly what a compensation increase from 100 to 125 basis points would have cost it. Chapter 6 said the file took fifty-one days instead of forty-five. This is the invoice.

🧮 Run the Numbers

Four cells that explain most arguments between branch managers and loan officers.

Hold the cost structure constant. Non-compensation cost on this file is \$9,491.88 − \$4,571.88 = \$4,920.00. Vary two things: the originator's basis points, and whether the lock had to be extended.

	Loan officer at 100 bps	Loan officer at 125 bps
No extension (net revenue \$9,143.75) | **+\$566.25**	(\$348.13)
15-day extension (net revenue \$8,229.37) | (\$348.13)	(\$1,262.51)

Check every cell:

• 100 bps, clean: \$9,143.75 − (\$4,920.00 + \$3,657.50) = \$9,143.75 − \$8,577.50 = **+\$566.25**
• 125 bps, clean: \$9,143.75 − (\$4,920.00 + \$4,571.88) = \$9,143.75 − \$9,491.88 = **(\$348.13)**
• 100 bps, extended: \$8,229.37 − \$8,577.50 = (\$348.13)
• 125 bps, extended: \$8,229.37 − \$9,491.88 = (\$1,262.51)

Every step between cells is \$914.38, in either direction, because that is what twenty-five basis points of this loan is worth. The two off-diagonal cells are identical. Raising an originator's compensation by a quarter point and letting every file run six days late cost the branch precisely the same amount.

That equivalence is the single most useful thing in this chapter for anyone who manages a branch, and it is the answer to the perennial argument. The compensation number the branch fights about all year is the same size as the calendar it does not manage at all.

Fixed cost, variable cost, and why volume matters to a branch

Split the \$4,920.00 of non-compensation cost into what moves with each file and what does not.

Check: \$1,050.00 + \$785.00 + \$540.00 + \$180.00 + \$315.00 = \$2,870.00, and \$265.00 + \$690.00 + \$615.00 + \$480.00 = \$2,050.00; the two sum to \$4,920.00.

The fixed \$2,050.00 is an allocation, and an allocation assumes a file count. This model assumes 30 closings a month, so the fixed pool is 30 × \$2,050.00 = \$61,500.00 a month. Change the file count and the same file's result changes without anything about the file changing at all:

Check the 45-file row: \$61,500.00 ÷ 45 = \$1,366.67, which is \$683.33 less than \$2,050.00, and −\$1,262.51 + \$683.33 = −\$579.18.

Notice that even at sixty closings a month this file still loses money at 125 basis points with the extension. Volume improves it; volume does not fix it. Something else has to change.

The break-even loan amount

Here is the cleanest way to state the problem. Revenue on this model is 250 basis points gross less a 25-basis-point extension — 225 basis points — and compensation is 125 basis points. Both scale with the loan. So each file contributes 100 basis points of the loan amount toward the \$4,920.00 of non-compensation cost. Break-even is therefore:

$$\frac{\$4{,}920.00}{0.0100} = \$492{,}000.00$$

Verify it. At a \$492,000.00 loan: gross revenue \$12,300.00, less an extension of \$1,230.00, net \$11,070.00. Compensation \$6,150.00, plus \$4,920.00 of other cost, total \$11,070.00. Result: exactly zero.

The Linden Street loan is \$365,750.00** — **\$126,250.00 short of break-even on this constructed model. That is not a criticism of the file. It is a description of a real structural fact: at a given cost structure and compensation rate, there is a loan size below which a branch loses money on every file it makes, and small-balance lending is where that bites hardest. It is also, and this is the uncomfortable part, one of the structural reasons borrowers buying modest homes have historically been served worse than borrowers buying expensive ones — a fair-lending consequence of an arithmetic fact, which Chapter 25 addresses from the obligation side.

The levers are exactly four, and there are no others: more revenue per file (better execution, larger loans), lower compensation, lower cost per file (which mostly means volume against fixed cost, and fewer touches per file), and fewer wasted days. The loan officer controls the fourth one completely and the first one partly. That is the honest scope of your influence on your branch's economics, and it is larger than most originators think.

26.10 Building an income model you can live on

An income model has exactly three inputs:

$$\text{Annual compensation} = \text{closings} \times \text{average loan amount} \times \frac{\text{basis points}}{10{,}000}$$

You control one of them.

Basis points are set by your employer, constrained by the compensation rule and by what the market will bear, and they are the input originators spend the most energy on and can change the least — usually only by changing employers, which resets a pipeline and costs a quarter.

Average loan amount is set by your market and your borrowers. You can shift it at the margin by choosing where you prospect. You cannot decide that houses in your county cost more.

Closings is yours. And Chapter 7 established what closings actually are: the output of a funnel with a measured conversion rate — conversations become applications at a rate you can measure, and applications become closings at a pull-through rate you can measure. We use that conclusion here and do not rebuild it. What matters for the income model is that closings are not a goal; they are a consequence, and the only way to raise them is at the top.

Working the model backward

Start from what you need to live on, not from what you hope to earn. Here is the number of closings a year required to reach three gross-compensation targets, at each of the three plans from §26.7, on a \$365,750 average loan:

| Gross compensation target | 100 bps (\$3,657.50/file) | 125 bps (\$4,571.88/file) | 150 bps (\$5,486.25/file) | |---|---|---|---| | \$100,000 | 27.34 files | 21.87 files | 18.23 files | | \$150,000 | 41.01 files | 32.81 files | 27.34 files | | \$200,000 | 54.68 files | 43.75 files | 36.45 files |

Check one: \$150,000 ÷ \$4,571.88 = 32.81. Check the internal consistency: the 150-basis-point column is two-thirds of the 100-basis-point column throughout, because 150 basis points buys you the same money in two-thirds the files.

Take the middle cell — \$150,000 at 125 basis points, 32.81 files a year — and push it through the funnel. Suppose your own measured rates are a 70% pull-through from application to closing and a 25% conversion from qualified conversation to application. (Both figures are constructed for this example. Use your own; Chapter 7 explains how to measure them and why borrowed averages are worthless.)

THE MODEL, BACKWARD                               [constructed teaching example]

  TARGET                     $150,000 gross compensation
    ÷ $4,571.88 per file  →  32.81 closings a year   =  2.73 a month
    ÷ 0.70 pull-through   →  46.87 applications      =  3.91 a month
    ÷ 0.25 conversion     →  187.48 conversations    = 15.62 a month

  Which is roughly FOUR qualified conversations a week, every week,
  for fifty-two weeks — including the two you were sick and the one
  you spent at your daughter's graduation.

That final line is the entire value of building the model. "I want to make \$150,000" is a wish. "I need four real conversations a week" is a calendar entry, and a calendar entry is something you can either do or honestly admit you did not do.

From gross compensation to what actually arrives

Gross compensation is not income. Three things happen to it.

Unreimbursed business expense. Even W-2 originators typically pay for some or all of their own marketing, client events, association dues, licensing renewals and continuing education, and mileage. A constructed but not unrealistic year:

Tax. Assume a blended effective rate of 28% across federal, state, and payroll tax, purely for illustration. Your actual rate depends on filing status, state, deductions, and classification. Nothing in this book is tax advice; the point of the line is that the number exists, not what it is.

Timing. Compensation arrives after funding — and often after the investor's purchase — so the month you produce and the month you are paid are not the same month.

FROM $150,000 GROSS TO WHAT LANDS                 [constructed teaching example]

  Gross compensation                                       150,000.00
  Less unreimbursed business expense                       (10,000.00)
                                                          ───────────
  Net before tax                                            140,000.00
  Less tax at an assumed blended 28%                        (39,200.00)
                                                          ───────────
  TAKE-HOME                                                 100,800.00
  Per month                                                   8,400.00

  Check: 140,000.00 × 0.28 = 39,200.00.  100,800.00 ÷ 12 = 8,400.00.

So the honest statement of the model is this: to put \$8,400 a month in the account, this originator must close 32.81 files a year at 125 basis points on \$365,750 loans, which requires about four qualified conversations a week, sustained. Every part of that sentence is a number, and every number is checkable.

The three things that break an income model

One: it is built on last year's market. The average loan amount, the pull-through rate, and — most of all — the mix of purchase and refinance business are all functions of the rate environment. A model built during a refinance boom and carried into a purchase market will overstate income by an enormous margin, because refinance files arrive on their own and purchase files do not. Chapter 37 takes apart that pivot; Chapter 40 puts it into a career arc. For the model, the discipline is simple: rebuild it every year with your own actual numbers, and rebuild it immediately when the market moves.

Two: it treats a good month as the new baseline. Origination income is lumpy in exactly the way Borrower 2's quarterly commission is lumpy on the Linden Street file — and you already know what an underwriter does with lumpy income. They average it over twenty-four months, and if it declined they take the lower year. Do the same to yourself. Budget on the twenty-four-month average, not on August.

Three: it ignores the draw. An originator on a recoverable draw who budgets on earned compensation will spend money that is contractually owed back. Look again at the table in §26.7: that originator's cumulative obligation peaked at \$6,856.25 in month four, and it did not reach zero until month nine. Every dollar of the draw taken in months one through four was borrowed.

What the model is actually for

Not motivation. Two specific decisions.

Whether to take an offer. An offer of 150 basis points with no draw, no processor, and no leads is not obviously better than 100 basis points with a dedicated processor and a lead flow, and the model is how you find out. Run the same number of closings through both, then adjust the closings — because a dedicated processor raises your capacity and a lead source raises the top of your funnel. That is the comparison; the basis-point number alone is not.

Whether you are actually in trouble. An originator with a thin month feels like a failure. An originator with a model knows whether the thin month is a variance or a trend, because they know what the top of the funnel looked like ninety days ago. Compensation is a lagging indicator of work you did last quarter. The number that tells you whether you have a business is the one at the top of the funnel, and it is available today.

🗂️ The Loan File

Chapter 26 contribution: what you earned on this file, and what it cost to make.

The Linden Street loan closed on day 51 at \$365,750.00. Turn the compensation lens on it.

What the loan officer earned, three ways:

The steps between structures 1 and 2, and 2 and 3, are \$914.38 and \$914.37 — twenty-five basis points of this loan is \$914.375, and payroll must round.

What the file cost to make (constructed model, §26.9, at 125 basis points and 30 closings a month):

What this settles. Three things, and they are worth more than they look.

The originator's compensation on this file was identical under every rate the borrowers could have chosen — 6.625% with a half point, 6.750% at par, 7.000% with a credit. That invariance is what makes the Chapter 13 recommendation trustworthy, and it is not an accident of this file; it is the compensation rule doing exactly what it was built to do.

The borrower's \$3,657.50 origination charge and \$1,828.75 in points were not the loan officer's pay. They were the creditor's revenue and the price of the rate.

And the six days of overrun that Chapter 6 recorded as a calendar failure have a dollar figure: \$914.38, which is the same size as a twenty-five-basis-point change in the originator's entire compensation plan.

What it does not settle. Whether the branch's loss on this file is typical — it is a constructed model, and yours will differ. Whether the loan officer's structure was the right one — that is a career question and Chapter 40 owns it. And nothing at all about the competing quote the borrowers were shopping, which is still open.

Open questions carried forward:

• Q1. Under structure 3, in what month does the recoverable draw balance reach zero, and what does the originator actually deposit before then? (§26.7 gives you the method.)
• Q2. If this originator's plan paid on the total loan amount, what would they have earned on the FHA alternative Chapter 13 rejected — and what would that option have cost the borrowers? (Chapter 13 has both figures.)
• Q3. What is this originator's own qualifying income if next year's production falls by a third? (§26.8, and Chapter 11's method.)

Your task. In Appendix C's workbook, complete the compensation page. Record the three structures and the earned amount under each. Then build the branch P&L for the file using your own market's cost assumptions rather than this book's, and compute your own break-even loan amount. Finally, write one sentence answering: if my compensation were allowed to vary with the rate, which of Chapter 13's options would I have recommended, and how would I know? That sentence is the reason the rule exists.

Conclusion

Compensation in mortgage origination is not a private arrangement between an employee and an employer. It is regulated conduct, and it is regulated because of something specific that people did.

Before 2011, an originator could take a yield spread premium for placing a borrower above par. On the Linden Street file, one notch — 6.750% to 7.000% — would have paid the originator \$2,743.12 and cost the borrowers \$21,992.40 over the life of the loan. Disclosure was tried first and did not work, because the incentive survived being described. So Congress and the regulators removed it.

The rule that resulted is short and sharp. Compensation may not be based on a term of a transaction, or on a proxy for one — a factor that both consistently varies with a term and that the originator can change. Loan amount is expressly permitted, which is why the whole industry is paid in basis points. You may be paid by the consumer or by the creditor, generally not by both on the same transaction. And a separate anti-steering provision, with a safe harbor built on three specified loan options, governs what you may do with the incentives the compensation rule cannot reach.

Then there is the arithmetic, and it is not comforting. A basis point of a \$365,750 loan is \$36.575. Twenty-nine closed files on a modest split is ten and a half million dollars of production and \$53,033.75 of pay. The word "recoverable" in a draw paragraph was worth \$6,856.25 in a constructed first year. And on a constructed branch P&L, the originator earned \$4,571.88 on a file the branch lost \$1,262.51 on — with the six-day overrun accounting for \$914.38 of it, which is precisely what a quarter point of compensation costs.

That last equivalence is the chapter's argument in one line. The compensation rate you cannot change is the same size as the calendar you can.

Next: compensation creates incentive, and incentive is where fraud begins. Chapter 27 covers mortgage fraud from the detection side — what the schemes look like, how they are caught, what the red flags actually are on a file you are holding, and what the exposure is for an originator who looks away.

Key Terms

Loan originator compensation rule (LO Comp) — the Regulation Z provisions at 12 CFR 1026.36 governing how loan originators may be paid: no compensation based on a term of a transaction or a proxy for one, no dual compensation, and an anti-steering prohibition with a safe harbor. (Ch.26)

Basis point (bp) — one one-hundredth of one percent, or 0.0001. One hundred basis points equals one percent; on a \$365,750 loan, one basis point is \$36.575. (Ch.26)

Yield spread premium (YSP) — the pre-2011 payment a wholesale lender made to an originator for placing a borrower at a rate above par; now prohibited as compensation to the originator, though the above-par price itself still exists and belongs to the consumer or the creditor. (Ch.26)

Term of a transaction — under Regulation Z, any right or obligation of the parties to a credit transaction: rate, points, fees, prepayment terms, escrow, maturity. Compensation may not be based on one. (Ch.26)

Proxy (LO Comp) — a factor that is not itself a term but is treated as one because it both consistently varies with a term over a significant number of transactions and can be added, dropped, or changed by the originator. Both prongs are required. (Ch.26)

Lender-paid compensation (LPC) — compensation paid to the loan originator organization by the creditor, set in advance for a period, priced into the rate rather than shown as a charge to the consumer. (Ch.26)

Borrower-paid compensation (BPC) — compensation paid to the loan originator directly by the consumer, disclosed as a charge and paid at closing or from loan proceeds, with generally better pricing available in exchange. (Ch.26)

Dual compensation prohibition — the rule that an originator receiving compensation directly from the consumer may not also be compensated by any other person on the same transaction, subject to an exception permitting a loan originator organization to pay its own individual originators. (Ch.26)

Anti-steering (compensation rule) — the prohibition on steering a consumer to a transaction because it pays the originator more, unless the transaction is in the consumer's interest; distinct from steering as a fair-lending violation (Ch.25). (Ch.26)

Anti-steering safe harbor — deemed compliance achieved by presenting, for each type of transaction the consumer expressed interest in, the lowest-rate option, the lowest-rate option without risky features, and the option with the lowest total points and origination fees. (Ch.26)

Compensation plan — the document setting an originator's basis points, base, floor and ceiling, payment timing, chargebacks, draw terms, and amendment rights. (Ch.26)

Commission split — a division of the compensation paid on a file between the originator and the branch or team, in exchange for services the branch supplies. (Ch.26)

Draw — an advance against future compensation. Recoverable draws are offset against later earnings and repayable; non-recoverable draws function as a forgiven floor. (Ch.26)

W-2 originator / 1099 originator — an originator classified as an employee versus as an independent contractor. The classification changes taxes, benefits, expense treatment, and how the originator's own income is underwritten; it does not change coverage under the compensation rule, and state licensing law may restrict it. (Ch.26)

Branch profit and loss (P&L) — the statement of a branch's revenue and cost, per file or per period, that determines whether production is profitable to the company as distinct from profitable to the originator. (Ch.26)

Per-file cost — the total cost attributable to originating one loan, including compensation, fulfillment, technology, compliance, occupancy, and allocated overhead. (Ch.26)

Production goal — the number of closings required to reach a stated compensation target at a given average loan amount and basis-point rate; the output of an income model, not its input. (Ch.26)

Spaced Review

1. (Ch.26) A comp plan proposes to pay 10 additional basis points on any loan where the representative credit score is 740 or above. Run both prongs of the proxy test explicitly, using the Linden Street file's two middle scores of 742 and 706 to argue prong two. State your conclusion and what you would do about it.

2. (Ch.13 + Ch.26) Chapter 13 rejected the FHA alternative on the Linden Street file. On a plan paying 125 basis points of the total loan amount, the originator would have earned \$4,571.88 on the conventional loan of \$365,750.00 and \$4,725.33 on the FHA total loan of \$378,026.69 — a difference of \$153.45. Chapter 13 established that the FHA option would have cost these borrowers \$25,382.19 more in total cost of credit. Compute the ratio of the two, then answer: which rule prevents the originator from acting on that \$153.45, and which separate rule governs what happens if they recommend FHA anyway?

3. (Ch.24 + Ch.26) A branch manager proposes paying originators 15 extra basis points on files referred by one particular real estate brokerage, arguing that it survives the proxy test because referral source does not vary with any loan term. Assume the manager is right about the proxy test. Name the statute that still prohibits the arrangement and explain, in two sentences, why clearing Regulation Z did not clear it.

4. (Ch.26) An originator on a \$4,000 monthly recoverable draw closes one \$365,750 file a month for three months at 100 basis points on a 50/50 split. State the cumulative amount owed at the end of month three, and then explain what changes about that figure if the word "recoverable" is replaced with "non-recoverable."

5. (Ch.13 + Ch.24 + Ch.26) A borrower asks you to choose between 6.625% with a half point (\$1,828.75) and 6.750% at par. Explain, in the order you would actually say it on the phone: (a) what your compensation is under each, (b) what break-even analysis Chapter 13 would run, (c) which disclosure from Chapter 24 the borrower will use to check your arithmetic, and (d) why the answer to (a) is the reason they should believe (b).