Chapter 31: Retail, Broker, Correspondent: Business Models, Warehouse Lending, and Where You Fit

"Everybody in this business does the same work. The only real question is whose money is on the table at two o'clock on closing day, and who has to answer for it in three years." — constructed; a warehouse banker's summary of the industry

Overview

Three loan officers take the same application on the same Tuesday, from the same two borrowers, on the same house at 4412 Linden Street. Same \$385,000 price, same 5% down, same \$365,750 loan, same 706 representative score, same 95% loan-to-value, same 6.625% note rate with a half point.

Three completely different transactions.

In the first, the borrowers' money and the lender's money meet at a title company and the lender is the company whose logo is on the loan officer's business card. In the second, the logo on the card belongs to a firm that has no money at all and never will — the funds are wired by a lender the borrowers have not spoken to, whose name will appear on the note they sign. In the third, the logo on the card belongs to a company that does fund the loan, in its own name, with money it borrowed that morning from a bank you have never heard of and will repay in about three weeks.

The borrowers cannot tell these apart. They will sign nearly identical documents, make nearly identical payments, and remember one loan officer. But the three structures determine which guidelines the file must satisfy, what happens when it does not, how fast underwriting moves, what appears on the Closing Disclosure, who owns the loan officer's license, and — for the company — whether one bad file is an annoyance or a wound.

Chapter 1 §1.6 sketched these three models in a paragraph each and promised you this chapter. Chapter 28 showed you where the money ultimately comes from and Chapter 29 showed you how it gets priced. This chapter is the missing middle: the plumbing between the investor who supplies the capital and the loan officer who takes the application. It is also, quietly, the most career-relevant chapter in the book, because almost every loan officer chooses a channel without knowing there was a choice, and one of the choices decides whether you own a credential or merely borrow one.

In this chapter, you will learn to:

• Distinguish the three origination channels by whose money funds the loan and whose name is on the note
• Explain how a warehouse line of credit works and why it makes an unsaleable loan existential
• Describe table funding and the mini-correspondent structure and the question a regulator asks about them
• Explain why a depository originator is registered and a non-bank originator is licensed, and what that does to a career
• Compare what the borrower experiences in each channel, from Loan Estimate to servicing transfer
• Identify the structural revenue and disclosure differences among the models
• Evaluate an offer in any channel using questions that actually predict your production

Learning Paths

🎓 Exam — §31.1, §31.5, and §31.6. The SAFE test asks who the creditor is, what table funding means, and — reliably — the licensed-versus-registered distinction. Know that a mortgage broker is not a lender and that a correspondent is. 🏠 New LO — §31.6, §31.7, and §31.9. Read §31.9 before your next interview, not after. 🤝 Partner — §31.7 and §31.10. If you refer clients, you should be able to explain to them why the name on the note is not the name on the business card, without making it sound like a problem. 📊 Operations — §31.4 above all. Dwell time on a warehouse line is a capacity constraint, which means post-closing turn time is a revenue line, not a housekeeping matter.

31.1 Three ways to originate the same loan

Start with the word the industry uses for this and almost never explains: channel.

A channel is the structural route by which an application becomes a funded loan. It is not a department, a product, or a sales strategy. It is an answer to four questions that have to be answered before anyone can wire money:

1. Who takes the application?
2. Who underwrites it?
3. Whose money funds it at the closing table?
4. Whose name is on the note?

There are three ordinary answers in American residential lending, and Chapter 1 named them: retail, broker (served by the wholesale channel), and correspondent. Everything else — direct-to-consumer call centers, joint ventures with builders, bank branch referral programs, mini-correspondents — is a variation on one of those three or a hybrid of two.

THE THREE CHANNELS — the four questions                    [constructed teaching example]

                      RETAIL              BROKER              CORRESPONDENT
  ───────────────────────────────────────────────────────────────────────────────
  Takes the           the lender's        the brokerage's     the correspondent's
  application         own employee        own employee        own employee
  ───────────────────────────────────────────────────────────────────────────────
  Underwrites         the lender          the WHOLESALE       the correspondent
                                          lender              (to the investor's
                                                              guidelines)
  ───────────────────────────────────────────────────────────────────────────────
  Funds at the        the lender's        the WHOLESALE       the correspondent's
  closing table       own capital or      lender's money      WAREHOUSE LINE
                      warehouse line
  ───────────────────────────────────────────────────────────────────────────────
  Name on the         the lender          the WHOLESALE       the correspondent
  note                                    lender
  ───────────────────────────────────────────────────────────────────────────────
  Sells the loan      yes, after          it is already the   yes, after closing
  afterward?          closing             lender's loan
  ───────────────────────────────────────────────────────────────────────────────

Read down the "broker" column and notice something a lot of new loan officers get wrong: the brokerage never owns the loan. Not for a minute, not for a day. In a conventional wholesale transaction the loan closes in the wholesale lender's name, funded with the wholesale lender's money, on the wholesale lender's approval. The brokerage's product is the file — an assembled, documented, saleable application — and its customer is a lender, not a borrower. (There is a structure in which a loan does close in the broker's name and get funded at the table by someone else. That is table funding, and §31.5 handles it, because it is where the compliance questions live.)

Now the vocabulary that runs the rest of the chapter.

Wholesale lending is lending to consumers through originators who are not your employees. The wholesale lender publishes rate sheets and guidelines to those originators, underwrites what they submit, funds and closes in its own name, and never advertises to the borrower. It has a sales force too — but its sales force sells to brokers.

That salesperson is the account executive, universally the AE. The AE is the broker's representative inside the wholesale lender: the person who gets a scenario answered, an exception requested, a pricing concession considered, a file escalated when it has sat for three days. A good AE is the difference between a broker who closes hard files and one who only closes easy ones. The AE is not the borrower's advocate and has never met them.

Third-party originator, or TPO, is the lender's word for anyone who delivers it loans without being its employee — brokers and correspondents both. Large lenders run a "TPO division" containing a wholesale desk (for brokers) and a correspondent desk (for correspondents), and you will see the acronym on portals, agreements, and org charts constantly. It is a useful word because it names the thing all non-retail origination has in common: the lender is buying files from somebody it does not employ, and therefore cannot supervise the way it supervises staff. Every approval requirement, audit right, indemnification clause, and repurchase provision in a TPO agreement exists because of that sentence.

📞 On the Phone

Borrower: "Wait — so are you the bank? Whose money is this?"

The evasive answer: "We're the lender." Sometimes true, sometimes not, and if it is not, they will find out at the closing table when a company they have never heard of is named on the note, at the exact moment they are least able to absorb new information.

The answer that works, retail or correspondent: "We are. My company funds the loan at closing and the note will have our name on it. Most lenders then sell the loan to an investor afterward — that's how the money keeps circulating — and if that happens you'll get a letter telling you where to send the payment. It's routine, it doesn't change your rate or your terms, and I'll walk you through it when it happens."

The answer that works, broker: "No, and that's actually the advantage. We're a mortgage brokerage. We're approved with a number of lenders, and I shop your file across them — because a file like yours prices differently at different lenders and one of them may have a guideline that fits when another doesn't. Whichever one we choose funds the loan and their name goes on the note. You'll see their name on your disclosures, and I'll tell you who it is as soon as we pick."

Say it early — at application, on the first disclosure conversation, not at the closing table. A borrower who learns this in week one finds it interesting. A borrower who learns it in the signing room finds it alarming, and they are not wrong to.

Here is the claim the rest of the chapter defends: the channels differ enormously in economics, risk, and speed, and hardly at all in what the borrower gets. The work is the same work. A file at 95% loan-to-value with a 706 score and \$2,380.00 of variable income needs the same documentation in all three. The differences show up in what happens when the file does not fit, in how fast the answer comes, in what the closing costs look like on paper, and in who is holding \$365,750 of somebody else's money if the loan turns out to be unsaleable.

31.2 Retail lending

Retail lending is origination by the creditor's own employees. The company that advertises, the company that takes the application, the company that underwrites, the company that funds, and the company on the note are one company. Most loan officers start here, and a majority of the industry's originators work here at any given time.

The shapes retail takes

Retail is not one thing. It comes in at least four recognizable shapes, and they are different jobs.

Distributed retail is the classic model: a loan officer in a local office, with a personal referral network of real estate agents, builders, financial advisors, and past clients. You generate your own business. The company provides the license sponsorship, the products, the pricing, the processing, the underwriting, the compliance apparatus, and the brand. Your production is yours. This is the model the Linden Street file assumes throughout the book — the call at 8:40 on a Wednesday came from an agent who had closed four files with the loan officer personally, and no company generated that.

Direct-to-consumer (call center) is centralized: leads are purchased or generated by marketing, routed to loan officers by a dialer or a queue, and worked at high volume with short cycle times. The loan officer does not own the relationship; the company does. This model is extremely efficient at refinances, where the product is nearly commoditized, and structurally weaker at purchases, where somebody has to be reachable on a Saturday when the appraisal comes in short. Chapter 37 takes apart what happens to a direct-to-consumer shop when a refinance boom ends.

Bank or credit union retail places mortgage originators inside a depository's branch network. Some of the business arrives by referral from tellers, personal bankers, and commercial lenders — which is a real and valuable lead source most independent originators do not have — and some is self-generated. §31.6 covers what this does to your license, which is the most important thing about it.

Joint ventures and in-house lenders are retail operations affiliated with a real estate brokerage, a homebuilder, or a title company. The referral flow is structural rather than personal: buyers arrive already in a transaction. These arrangements are legal and common, and they are also exactly the arrangements RESPA scrutinizes most closely, because a referral relationship between settlement service providers with common ownership is the fact pattern Section 8 was written about. An affiliated business arrangement disclosure is required, the borrower may not be required to use the affiliate, and the fees must be reasonable for services actually performed. Chapter 24 owns the rules; know that if you take a job in one of these, the compliance stakes are not theoretical.

How retail works, mechanically

RETAIL — one company, one hallway                          [constructed teaching example]

   BORROWER
      │  application
      ↓
   ┌──────────────────────────────────────────────────────────────────┐
   │  YOUR EMPLOYER (the creditor)                                    │
   │                                                                  │
   │   loan officer ──► processor ──► UNDERWRITER ──► closer          │
   │        │                              │             │            │
   │        │                       approves/denies      │            │
   │        │                       to AGENCY GUIDELINES │            │
   │        │                       + THIS COMPANY'S     │            │
   │        │                         OVERLAYS           │            │
   │        └──────────── one rate sheet, one price ─────┘            │
   │                                                                  │
   │   funds at closing ── from its own capital or its warehouse line │
   │   name on the note ── this company                               │
   └──────────────────────────────────┬───────────────────────────────┘
                                      │  sells the closed loan (Ch. 28)
                                      ↓
                          AGGREGATOR / AGENCY / INVESTOR

Everything is inside one boundary. That is the model's entire advantage and its entire limitation.

The advantage is proximity. You and the underwriter work for the same company, are measured by overlapping metrics, and can talk. When a file has an unusual income structure — a shift differential averaged over twenty-four months, a commission that is paid quarterly and lands lumpy — you can have a conversation before you submit rather than an argument after. When a condition is written badly, you can get it rewritten. When a file needs an exception, there is an internal path to a person with authority to grant one, and you can learn who that person is and what they respond to. None of this is available to a broker, and it is worth more than most new loan officers realize.

Retail also usually comes with the largest support apparatus: salaried processing, in-house compliance review, marketing, a brand borrowers have heard of, benefits, and — critically — somebody else paying for your licensing, your errors-and-omissions coverage, and your technology. If you are new, that scaffolding is not a perk. It is the difference between learning the job and drowning in it.

The limitation is that there is exactly one answer. One product menu. One set of overlays. One rate sheet. One turn time. If your employer does not offer a bank-statement program, you do not have one. If your employer's overlay requires a 640 minimum score on a government loan when the agency floor is lower, then for you the floor is 640. If underwriting is running nine days in a purchase market, it is running nine days for every file you have.

⚠️ Where Deals Die

The retail loan officer who has no Plan B, and does not find out until day 23.

The pattern: you take an application that looks fine, you order the appraisal, the borrower pays \$650 for it, you submit — and underwriting declines it on an overlay. Not an agency guideline. A rule your own company wrote, on top of the agency rule, for its own risk reasons (Chapter 14 draws that distinction). The file would be approvable somewhere. You just cannot get it there.

By then the borrower has spent the appraisal fee, the sellers have taken the house off the market, and the financing contingency is close to expiring. On the Linden Street file's calendar, day 23 is when the file went to underwriting; day 28 is the conditional approval. A decline on day 28 leaves seventeen days on a fifty-one-day file to start over somewhere else, and no lender starts a new file on day 28 and closes it on day 45.

The discipline has three parts. First, know your own overlays cold — not the agency guidelines, your overlays, on the products you sell most. They are written down; ask for the matrix. Second, run the unusual features past an underwriter or your manager before the borrower spends money on an appraisal, not after. Third, know one or two people at other shops you can hand a file to with a phone call. Referring a file you cannot do is not disloyalty. It is what makes the agent who sent it to you send you the next one.

What retail actually earns

A retail lender's revenue on a loan has three sources, and they are worth naming because loan officers who cannot name them tend to believe their employer is getting rich on the interest, which Chapter 1 already disproved.

1. Origination charges paid by the borrower at closing. On the Linden Street file, the origination charge is **\$3,657.50** — exactly 1.000% of the \$365,750 loan.
2. Discount points, when the borrower buys the rate down. Here, 0.500 point = \$1,828.75. Together with the origination charge that is \$5,486.25 of Section A charges on the Closing Disclosure.
3. The gain on sale in the secondary market — the difference between what the loan is worth to an investor and its face amount — plus the value of the servicing rights, if the lender keeps them. Chapter 28 owns servicing value; Chapter 29 owns the pricing that produces the gain.

The third one is usually the largest and the borrower never sees it, because it is not a charge to the borrower. It is the price an investor pays for a stream of payments at 6.625% in a market that would otherwise buy 6.750%. A retail lender that also underwrites and funds and closes has a large fixed cost structure sitting against those three revenue lines, and this is why retail lenders in slow markets close branches. Their break-even is measured in loans per month per office.

31.3 Wholesale and the mortgage broker

A mortgage broker takes applications and places them with wholesale lenders, who underwrite, fund, and close in their own names. The brokerage is licensed, regulated, examined, and liable — and it is not a lender. It never advances a dollar.

That last point is the source of everything else about the model. A brokerage has no warehouse line, no capital requirement of the kind a lender faces, no loans on its balance sheet, and no exposure to the market value of a closed loan. It also has no control over the underwriting decision, the closing timeline, or the docs.

How a brokerage actually operates

A brokerage is approved with some number of wholesale lenders — a handful for a small shop, dozens for a large one. Each approval is a real onboarding: a signed broker agreement, corporate and license documentation, financial statements, a quality control plan, background checks on principals, and access credentials to that lender's portal. Each lender then arrives with its own:

• guidelines and overlays, which differ, sometimes materially, on exactly the features that make a file hard
• rate sheet, priced daily and independently, with its own adjustments
• compensation plan for the brokerage, set in advance with that lender
• lock desk, lock terms, extension policy, and extension pricing
• submission package, portal, disclosure process, and condition format
• turn times, which move with that lender's volume and are outside your control
• AE, who is your route to all of the above

So a broker's day contains a task no retail loan officer ever performs: choosing the lender. You price the file across several rate sheets, check the features against several guideline sets, weigh turn times against the contract's closing date, and pick. On a clean file this takes fifteen minutes and the answer is mostly about price. On a hard file it is the entire job.

BROKER — the brokerage is a customer, not a colleague      [constructed teaching example]

   BORROWER
      │  application
      ↓
   ┌──────────────────────────────┐
   │  THE BROKERAGE               │   no money. no underwriting. no closing docs.
   │   loan officer ─► processor  │   its product is an assembled FILE.
   └───────┬──────────────────────┘
           │  prices the file across approved lenders, then submits to ONE
           │
     ┌─────┴─────┬───────────┬───────────┐
     ↓           ↓           ↓           ↓
  ┌──────┐   ┌──────┐    ┌──────┐    ┌──────┐
  │ WHOL.│   │ WHOL.│    │ WHOL.│    │ WHOL.│    each: own guidelines, own
  │ LDR A│   │ LDR B│    │ LDR C│    │ LDR D│    overlays, own rate sheet,
  └──┬───┘   └──────┘    └──────┘    └──────┘    own turn times, own AE
     │  chosen
     ↓
   UNDERWRITES ─► DRAWS DOCS ─► FUNDS ─► CLOSES IN ITS OWN NAME ─► owns the loan
     ▲
     │  your only route in: the ACCOUNT EXECUTIVE

The advantage is real, and it is choice

Do not let anyone — including a retail sales manager recruiting you — tell you the broker channel is a lesser version of lending. Its advantage is genuine and it is structural.

Overlays are not uniform. Chapter 14 established that agency guidelines are the floor and each lender layers its own rules on top. Those layers differ. One wholesale lender declines a condominium project because of its own litigation policy; another reviews it and approves. One requires two months of reserves on a 95% loan-to-value purchase where the automated findings ask for none; another follows the findings. One will not count a shift differential without a two-year history in the same role; another will follow the agency rule. When a file dies at lender A on a rule that lender B does not have, a broker moves the file and closes it. A retail loan officer, facing the same rule, has an exception request and a prayer.

Pricing is not uniform either. Wholesale lenders price independently, every morning, according to their own execution, their own hedging position, their own appetite for volume that week, and their own margin decisions (Chapter 29 explains the margin component). The spread between the best and worst price for a given file on a given day is real and it moves. What it is not is a fixed, reliable channel advantage you can quote to a borrower. Anyone who tells you "brokers are always cheaper" or "retail is always cheaper" is selling something. The honest practice is to price the actual file with three lenders and look, every time, because the answer changes with the file and the day.

Product breadth. A brokerage approved with a dozen lenders can offer product a single retail menu cannot: non-QM and bank-statement programs (Chapter 34), unusual property types, jumbo structures with different reserve or ratio treatment, renovation lending, and specialty government programs. For a loan officer whose market includes self-employed borrowers — the Fulton Avenue file is the archetype, an S-corporation owner whose accountant's number and whose underwriter's number are not the same number — breadth is not a luxury. It is the practice.

The cost is real too, and it is control

Here is the sentence to remember: a broker is a customer of the underwriting department, not a colleague of it.

Everything follows from that. You cannot walk down the hall. You cannot ask the underwriter what they meant by condition 6 — you submit a question through a portal and wait, or you call your AE and they ask. When a lender's turn times go from two days to nine days in a hot market, your files go with them and you find out at the same time your borrower does. You have no standing in that company's queue, no internal advocate above the AE, and no ability to escalate on the strength of your relationship with anyone who decides.

The complexity is a real operating cost, too. Multiple portals with multiple passwords. Multiple disclosure processes. Multiple condition formats. Multiple lock desks with different cutoff times and different extension pricing — which matters, because a lock is a contract with one specific lender and does not travel if you move the file. A file moved from lender A to lender B on day 30 is a new lock at today's market, not a transferred one. On a file where the lock was already three days short of the contract's closing date, that is not a small thing.

And there is an exposure brokers are frequently surprised by: the early payoff provision. Broker agreements commonly require the brokerage to refund its compensation if the loan pays off within a period set in the agreement after closing — a period usually measured in months. The wholesale lender paid for a loan expecting to earn on it; if the borrower refinances immediately, the lender's economics evaporate and the contract shifts that loss back. A brokerage that lets a rate-shopping borrower refinance out of a loan it placed ninety days ago can find itself writing a check for compensation it has already spent. Read the agreement. Every one is different.

📞 On the Phone

The call to the AE, day 26, file submitted day 23, underwriting is running seven days and the contract closes in nineteen.

The call that fails: "Where is my approval? This is ridiculous. Your turn times are killing me." True feelings, zero leverage. The AE cannot make an underwriter work faster by absorbing your frustration, and you have just spent a relationship you will need in three weeks.

The call that works: "I've got 2214 in your queue, submitted the twenty-third. Purchase, contract closes on the fifteenth, and the lock runs out four days after that. I'm not asking you to jump the line — I want to know two things. One: is seven days still the number, or is it eight now, so I can tell the agent something true today. Two: if it slips past the tenth, what's the escalation path and what do you need from me to use it? I'll have anything you want in an hour."

Three things happened there. You gave the AE a loan number, a submission date, and a hard date — the three facts they need to do anything at all. You asked for information you can act on rather than an outcome they cannot promise. And you pre-committed to being fast, which is the only currency you have with a lender you do not work for.

The version of this call available to a retail loan officer is different — you walk over, or you message the underwriter directly. Neither version is better. They are different jobs, and the broker's version requires more preparation and better manners.

Being fair about the trade

The trade is legible once you state it plainly. The broker trades control for choice. The retail loan officer trades choice for control. Neither is a better deal in the abstract, and which one is better for you depends almost entirely on what your book of business looks like.

If most of your files are clean W-2 purchases with good scores and comfortable ratios, choice buys you very little — those files fit everywhere — and control buys you a lot, because your competitive edge is certainty of closing. If a meaningful share of your files have something wrong with them — self-employment, a thin file, an unusual property, a score at the edge, a debt-to-income ratio at the top of the range — choice is worth more than control, because the alternative to choice is a decline.

31.4 Correspondent lending and the warehouse line

A correspondent lender closes loans in its own name, with its own borrowed money, to guidelines published by the investor that has agreed to buy them — and then sells them, typically within weeks.

From the borrower's side of the desk it looks exactly like retail. Same company name on the application, the Loan Estimate, the note, and the deed of trust. From the capital-markets side it is a different animal: the correspondent is running a small securities-adjacent business, with a funding facility, a pipeline that must be hedged or committed, delivery deadlines, and a balance sheet that can be damaged.

The delivery decision

A correspondent sells its closed loans, and it decides how:

• Best efforts — the correspondent commits to deliver a specific loan if it closes, at a price locked when it commits. If the loan falls apart, nothing is owed. Safer, and priced accordingly.
• Mandatory — the correspondent commits to deliver a specific dollar amount by a specific date whether or not the underlying loans close. Better price, real risk: a shortfall must be covered by buying loans or paying a pair-off fee. This is where pull-through (Chapter 29) stops being a statistic and starts being a liability.
• Servicing-released — the loan and the right to service it are sold together, and the borrower begins paying the buyer almost immediately.
• Servicing-retained — the correspondent sells the loan but keeps the servicing rights and the borrower relationship. Chapter 28 explains why a mortgage servicing right is an asset with its own market and why keeping it requires capital most small correspondents do not have.

Chapters 28 and 29 own those choices. What this chapter owns is the thing that makes all of them urgent: the money the correspondent used to close the loan was borrowed, and the clock is running from the moment it was wired.

What a warehouse line of credit actually is

A warehouse line of credit is a secured, revolving, short-term credit facility that a lender draws on to fund loans at the closing table and repays out of the proceeds when the loan is sold. Chapter 1 named it. Here is how it works.

The provider is a warehouse bank — usually a commercial bank with a specialty lending group, sometimes an investment bank. It is not the investor who eventually buys the loan, and it does not want to own mortgages. It wants a short, well-collateralized loan and a fee.

Getting a line is an underwriting process aimed at the company, not the loan:

• audited financial statements and minimum tangible net worth
• minimum liquidity, often a specific unrestricted-cash figure
• a leverage limit — how much line against how much equity
• experienced management, a documented quality control program, agency and investor approvals
• and, frequently, personal or corporate guarantees

Once the line exists, each loan is a draw:

1. The correspondent requests funding for a specific loan, delivering the funding package.
2. The warehouse bank advances funds — not the whole loan amount. The line advances a percentage and the correspondent puts in the rest from its own cash. That percentage is the advance rate; the difference is the haircut. Advance rates and haircuts are negotiated, vary by loan type and by borrower quality of the lender, and are confidential commercial terms. Never quote one you have not seen.
3. The note is endorsed and the collateral — the original note, usually held by a document custodian — secures the advance.
4. Interest accrues daily on the advance. So do facility fees, non-usage fees, and per-loan transaction fees. Collectively this is the cost of carry.
5. When the loan is sold, the note ships to the investor under a bailee letter — a document that tells the investor the warehouse bank still has a security interest in the note until it is paid, and directs the purchase proceeds to the warehouse bank. This is why the money comes back to the right place.
6. The advance is repaid, the haircut comes back, and whatever is left is the correspondent's.

THE WAREHOUSE CYCLE — one loan, about three weeks          [constructed teaching example]

   DAY 0 (closing)                                    DAY ~18 (purchase)
   ┌──────────────┐                                   ┌──────────────┐
   │  WAREHOUSE   │── advances (loan amount minus ───► │  CLOSING     │
   │     BANK     │    the haircut)                    │  TABLE       │
   └──────▲───────┘                                    └──────┬───────┘
          │                                                   │ funds the loan
          │  repaid from the sale proceeds,                    ↓
          │  under a BAILEE LETTER                        BORROWER gets the house
          │                                                   │
   ┌──────┴───────┐                                           │  note + file
   │  INVESTOR /  │ ◄───── loan delivered, reviewed, ─────────┘
   │  AGGREGATOR  │        purchased
   └──────────────┘
          ▲
          │  meanwhile, every single day:
          │  INTEREST ACCRUES ON THE ADVANCE.
          │  The correspondent's own cash sits in the haircut, earning nothing.

And the constraints that make it dangerous:

• Aging. Every line limits how long a loan may sit before it must be curtailed — paid down or off with the correspondent's own money. An aged loan is a warehouse bank's early warning that something is wrong with either the loan or the lender, and aged loans get attention.
• Covenants. Minimum net worth, minimum liquidity, maximum leverage, and often profitability tests. Breach one and the bank can reduce the line, raise the price, or terminate.
• Advance rate changes. The bank can lower the advance rate — which is a capital call by another name, because a lower advance rate means the correspondent must fund more of every loan itself.
• The demand feature. Warehouse facilities are short-term and frequently terminable on brief notice. A lender whose lines are pulled cannot fund tomorrow's closings. Not "will struggle to" — cannot. Closing tables go empty the next morning.

🧮 Run the Numbers

What the warehouse line costs on the Linden Street loan, and what happens if nobody buys it.

[constructed teaching example — advance rates, warehouse pricing, and investor bids are negotiated, confidential, and move daily. These figures illustrate the STRUCTURE. Verify any actual terms in the actual facility agreement.]

Assume: loan \$365,750.00 · advance rate 97.5% · warehouse rate 7.50% on a 360-day basis · the loan is purchased by the investor 18 days after closing · the investor pays 101.750.

1. Funding day.

Loan amount	\$365,750.00
Warehouse advances (97.5%)	\$356,606.25
Correspondent's own cash (the 2.5% haircut)	\$9,143.75

2. Cost of carry. Daily interest on the advance:

$$\$356{,}606.25 \times \frac{0.0750}{360} = \$74.29 \text{ per day}$$

Eighteen days: $18 \times \$74.29 = \$1{,}337.22$. (Every figure below carries the daily rate rounded to the cent, so you can check each step on a calculator. Carrying it unrounded moves the totals by a few cents — pick one convention and hold it.)

3. Sale day. The investor pays 101.750 of the loan amount:

$$\$365{,}750.00 \times 1.01750 = \$372{,}150.63$$

Sale proceeds	\$372,150.63
Less: repay the advance	(\$356,606.25)
Less: 18 days of carry	(\$1,337.22)
Returned to the correspondent	\$14,207.16
Of which its own haircut coming back	(\$9,143.75)
Net gain on the sale	\$5,063.41

Check it the other way: the loan sold 1.750 points over par, which is \$6,400.63, less \$1,337.22 of carry = **\$5,063.41**. It reconciles.

4. Now break it. Suppose a post-closing audit finds a defect and no investor will buy the loan at the agreed price. It sits. At 60 days the carry alone is $60 \times \$74.29 = \$4{,}457.40$ — which has consumed almost 70% of what the loan was going to earn, and the loan is still on the line.

Suppose it finally clears at a scratch-and-dent bid of 94.000:

$$\$365{,}750.00 \times 0.94000 = \$343{,}805.00$$

The loan cost \$365,750.00** of somebody's money and returned **\$343,805.00 — a hole of \$21,945.00**, before any carry at all. The advance was \$356,606.25, so the correspondent must find \$12,801.25** plus accrued interest out of its own pocket just to retire the draw, and the \$9,143.75 it put in on funding day is gone.

\$21,945.00 ÷ \$5,063.41 = 4.33. One unsaleable loan erases the net gain on more than four clean ones.

That ratio is the whole reason guidelines are treated as absolute in this industry. It is why an underwriter will not "just this once." It is why the eleven conditions on the Linden Street approval were eleven and not eight. And it is the mechanical version of Chapter 1's claim that the lender is not primarily in the interest business: the first month's interest on this loan is \$2,019.24, which is less than half of what selling it earned in three weeks — and less than half of what carrying it for sixty days would have cost.

Dwell time is capacity, which means it is revenue

Here is the part operations people love and salespeople never think about.

A warehouse line has a size. The number of loans a correspondent can have outstanding at once is the line divided by the advance per loan. But the number of loans it can fund in a year is that capacity multiplied by how many times the line turns over — and turnover is governed by dwell time, the average number of days between funding a loan and being paid for it.

DWELL TIME IS CAPACITY                                     [constructed teaching example]
  A $50,000,000 line. Loans averaging $365,750 at a 97.5% advance = $356,606.25 each.
  Loan counts are whole loans, rounded DOWN. Turns are carried to two decimals.

  Loans outstanding at once:   $50,000,000 / $356,606.25   =  140 loans

  ── 18-day average dwell ───────────────────────────────────────────────
     turns per year   365 / 18   = 20.28
     loans per year   140 x 20.28 = 2,839      volume  $1,038,364,250

  ── 30-day average dwell ───────────────────────────────────────────────
     turns per year   365 / 30   = 12.17
     loans per year   140 x 12.17 = 1,703      volume  $  622,872,250

  ── the difference ─────────────────────────────────────────────────────
     twelve days of dwell time  =  $415,492,000 of annual capacity
                                =  1,136 loans that cannot be made

Twelve days. Not twelve days of underwriting — twelve days of post-closing: trailing documents, the recorded instrument, the final title policy, the audit exception, the investor's purchase review. Work that happens after the borrower has the keys and after the loan officer has moved on, and that determines how many loans the company can fund next quarter.

This is also where the book's fifth theme — every day costs money — acquires a second clock. On the Linden Street file, the fifteen-day lock extension cost \$914.38**, which is **\$60.96 per day of pre-closing delay. Illustratively, the warehouse carry on the same loan is \$74.29 per day of post-closing delay. They are not alternatives. They are consecutive. A file that runs late before closing and then sits after closing pays both.

31.5 Table funding and mini-correspondent

Two structures sit between the three clean models, and both exist because the boundaries have consequences.

Table funding

Table funding is a settlement at which the loan is funded by a contemporaneous advance from one party and simultaneously assigned to that party, while closing in the name of another. The loan closes in name A; name B's money arrives at the table; the loan is assigned to B in the same transaction.

Regulation X names table funding specifically and treats a table-funded transaction as a loan origination, not a secondary-market transaction — which matters because RESPA's rules about settlement services and referrals apply to originations. Regulation Z's loan-originator provisions also address the structure: a person who closes a table-funded loan in its own name is treated, for purposes of those provisions, as a loan originator rather than as the creditor.

Why does anyone do it? Historically, because it let an originator without capital close in its own name, which looked and felt like being a lender and — this is the part that mattered — changed how its compensation appeared to the borrower. It also concentrates the closing in one place: one wire, one settlement, one signing.

Mini-correspondent

A mini-correspondent is a small originator, very often a former mortgage broker, that has taken on the correspondent form: it closes in its own name, usually funding through a warehouse line, and sells the loan immediately — frequently to the same investor whose money is behind the warehouse line.

The structure is entirely legitimate when it is real. Plenty of well-run brokerages have grown into genuine correspondents: they raised capital, obtained a real warehouse facility, hired underwriting or delegated underwriting authority, built a post-closing and quality-control function, and accepted repurchase exposure. That progression is the ordinary way small lenders are born.

It becomes a problem when it is a costume. If the "correspondent" does not underwrite, does not draw its own documents, does not bear post-closing risk, has no meaningful capital at stake, and funds through a facility provided by the same party that buys every loan — then nothing about the economic substance has changed from brokering. What has changed is the paperwork, and specifically the disclosure of compensation, which §31.8 explains.

The Consumer Financial Protection Bureau issued guidance on mini-correspondent arrangements in 2014 setting out the questions it would ask in deciding whether an entity was genuinely acting as a creditor or was a broker in correspondent clothing. Verify the current guidance before relying on this summary, but the questions have a durable shape:

• Does the entity have a real, adequate warehouse facility, and who provides it?
• Who makes the underwriting decision, and does the entity have delegated authority or is every file underwritten by the buyer?
• Who draws the closing documents and who is named as creditor on them?
• Does the entity bear risk after closing — repurchase, indemnification, early payment default?
• Does it perform post-closing quality control?
• Is it selling essentially every loan to the same party that funds it?

⚖️ Compliance Check

Getting the creditor question wrong is not a technicality.

Whether an entity is the creditor or a loan originator determines which disclosures it must give, who must be identified where on the Loan Estimate and the Closing Disclosure, how compensation is disclosed, which licensing regime applies, who reports under the Home Mortgage Disclosure Act, and who is liable when something is wrong. Under Regulation Z the creditor is generally the person to whom the obligation is initially payable on the face of the note — but the loan originator provisions contain specific treatment for table-funded transactions, and the answer is not always the name printed at the top of the page.

If you are considering a move to a mini-correspondent, or your brokerage is considering the conversion, the questions to ask are the ones above and they are not rude: who actually underwrites, who actually funds, who bears the risk after closing. If the answer to all three is "the investor," the structure may not be what it is labeled.

Requirements change, guidance is periodically revised, and state law adds its own layer — several states impose their own net worth, bonding, and licensing requirements on lenders that do not apply to brokers. Verify current federal and state requirements with your compliance department and your state regulator before you act on any of this.

🔍 Check Your Understanding

1. A brokerage submits a file to a wholesale lender, which underwrites it, funds it, and closes it in the wholesale lender's name. At any point, does the brokerage own the loan?
2. A correspondent funds a \$365,750 loan on a line with a 97.5% advance rate. How much of its own cash goes into that loan on funding day, and when does it come back?
3. Why does a warehouse bank care how long a loan has been on the line, given that it earns interest the whole time?

(2 is the one people miss: \$9,143.75, and it comes back only when the loan is purchased. It is not a fee — it is the correspondent's capital, tied up and earning nothing. Multiply by the number of loans outstanding and you have the reason correspondents care about post-closing speed. On 3: the interest is not the point. An aged loan usually means the loan cannot be sold, which means the collateral securing the advance may not be worth the advance.)

31.6 Depository vs. non-bank

This is the section Chapter 1 and Chapter 3 both pointed to, and it cuts across all three channels. A retail lender can be a bank or a non-bank. So can a wholesale lender. So can a correspondent. The question is not how the loan is originated. It is what kind of institution your employer is — and it has more effect on your career than the channel does.

A depository lender is a bank, savings institution, or credit union that takes federally insured deposits. A non-bank lender — the industry says independent mortgage bank, or IMB — does not take deposits and exists to originate, sell, and often service mortgage loans.

The business-model difference

DEPOSITORY vs. NON-BANK — where the money and the rules come from

                      DEPOSITORY                    NON-BANK (IMB)
  ────────────────────────────────────────────────────────────────────────────
  Funding for the     insured deposits, its own     WAREHOUSE LINES.
  loan                capital, and warehouse        Borrowed, short-term,
                      capacity                      revolving, and cancellable.
  ────────────────────────────────────────────────────────────────────────────
  Can it keep the     YES — portfolio lending is    Essentially no. It must
  loan?               a real option                 sell to keep operating.
  ────────────────────────────────────────────────────────────────────────────
  Prudential          a federal banking agency:     none. There is no capital
  supervision         capital rules, safety-and-    adequacy examiner.
                      soundness examination,
                      deposit insurance
  ────────────────────────────────────────────────────────────────────────────
  Consumer-compliance its prudential regulator      state regulators (through
  supervision         below the statutory asset     NMLS and multistate exams)
                      threshold; the CFPB above it  and the CFPB
  ────────────────────────────────────────────────────────────────────────────
  YOUR CREDENTIAL     REGISTERED with the NMLS.     STATE-LICENSED. Education,
                      Unique identifier,            the SAFE MLO test, a
                      fingerprints, background      background and credit
                      check. No test, no            review, continuing
                      pre-licensing education,      education, renewals,
                      no continuing education,      surety bonding, per state.
                      no state license.
  ────────────────────────────────────────────────────────────────────────────
  IF YOU LEAVE        the registration does not     THE LICENSE IS YOURS. It
                      go with you. To originate     moves with you, subject to
                      for a non-bank you must       sponsorship transfer and
                      get licensed from scratch.    each state's process.
  ────────────────────────────────────────────────────────────────────────────

Why the divide exists at all

Chapter 3 owns the licensing mechanics — what registration requires, what licensing requires, how sponsorship works, what continuing education looks like. What is worth understanding here is why Congress drew the line where it did, because once you see the reason, the rule stops being arbitrary.

Before 2008, depository institutions were already the most heavily supervised financial firms in the country. A national bank had a resident examination team, capital requirements, liquidity requirements, deposit insurance and the standards that come with it, and an established federal supervisor with authority over its lending practices. Whatever else was true of that system, there was a regulator in the building.

Independent mortgage companies had none of that. Many were supervised in some fashion by state regulators, unevenly, with wildly varying standards, no common database, and no way for one state to know that a loan officer it was about to license had been thrown out of another state last year. An originator could fail in one state on Friday and open in the next state on Monday.

The S.A.F.E. Act's answer was to build the missing infrastructure where it was missing: a nationwide registry with a unique identifier for every originator, and a licensing regime — testing, education, background and financial review, bonding, continuing education, renewal — imposed on originators at non-depositories. For originators already inside a federally supervised depository, Congress required registration: get in the database, get an identifier, get fingerprinted, be findable — and relied on the existing prudential supervision for the rest.

That is the business-model reason. The divide is not a judgment that bank originators are better trained. It is a judgment about which institutions already had a supervisor. It follows the institution, not the person, which is why the same human being doing identical work is licensed on one side of the street and registered on the other.

The line has an edge case worth knowing: registration generally covers originators employed by a depository institution or by a subsidiary that is owned and controlled by one and regulated by a federal banking agency. An affiliate that does not meet that description is a non-bank for this purpose, and its originators are licensed. Some banks own mortgage subsidiaries in each category. Ask which one you would be joining, by name, and verify the answer in the NMLS Consumer Access database rather than taking it from a recruiter.

🎓 NMLS Exam Watch

The licensed-versus-registered distinction is one of the most reliably tested items on the SAFE MLO test, and the stem is usually built to make you answer from the activity rather than the employer.

A typical trap: "An individual takes residential mortgage loan applications for compensation at a federally insured credit union. Which is required?" The activity is identical to what a licensed originator does — so candidates reach for "a state license." The answer is registration with the NMLS: unique identifier, fingerprints, background check, no state license and no MLO test.

Two more the exam likes:

• Registration is not portable. The unique identifier follows the individual, but the authority to originate does not. Leaving a bank for an independent mortgage company means completing pre-licensing education, passing the test, and obtaining a state license before taking an application.
• A mortgage broker is not a lender. Any stem describing an entity that "funds" or "makes" the loan is not describing a broker, no matter what the entity is called in the question.

Chapter 3 is the full treatment. Requirements and thresholds change; verify current requirements with the NMLS and your state regulator.

What it means for where you work

The choice is usually presented to a new originator as bank versus mortgage company, as though the difference were culture and dress code. Here is what it actually decides.

Product. A depository can hold a loan on its own balance sheet. That single capability makes possible things a non-bank structurally cannot do at all: a portfolio jumbo with underwriting that answers to nobody's selling guide, a loan on an unusual property, a relationship price for a customer with substantial deposits, a construction loan that stays in-house. If your market has those borrowers, that is a real weapon. Against it: depositories typically carry heavier overlays, change products slowly, and are more conservative about anything that draws examiner attention.

Speed and technology. Non-banks are usually faster to adopt, faster to change pricing, and faster to launch or kill a product, because there is less to ask and less to lose. Their entire enterprise is originating mortgages, so the mortgage operation gets the investment. In a depository, mortgage lending competes for resources with every other line of business and is often not the one that wins.

Stability. A depository funds itself with insured deposits and does not go out of business because the warehouse line was cut. A non-bank does. Non-banks have no deposits, no discount window, and financing that can be reduced or withdrawn in exactly the market conditions that also reduce their revenue. Every rate shock in this industry's history has been followed by non-bank consolidation. When you are evaluating an independent mortgage company, its warehouse capacity and its capital are not trivia — they are whether it will be here in eighteen months.

Compensation. Structures differ, and the bible of this book is honest about it: bank programs often include salary or draw, benefits, and referral flow, and typically pay less per loan; independent mortgage companies typically pay more per loan and provide less. Neither of those is a rule and both vary enormously by company and market. Do not accept anyone's characterization, including this one, without reading the actual plan document.

And the credential. This is the sleeper, and it is why Chapter 1 called it the most consequential career decision most new originators make without knowing they are making it. A registered originator holds no portable credential. A licensed one does. Three years at a bank produce experience, a book of business, and a payroll history — but not a license. Three years at a non-bank produce all of that plus a credential that belongs to you, that you can move, and that another employer cannot take away. If you are early in a career and unsure where it goes, that asymmetry is worth weighing. If you know you want to be inside a depository for the long run, it does not matter at all.

31.7 What the borrower experiences differently

Now change seats. Everything above is the industry's view. What does the household buying 4412 Linden Street actually notice?

Less than you would think, and it shows up in five specific places.

1. The name on the disclosures. In retail and correspondent transactions, one company's name appears on the Loan Estimate, the Closing Disclosure, the note, and the security instrument, and it is the company on the business card. In a broker transaction the Loan Estimate identifies the mortgage broker, and the creditor is the wholesale lender — a name the borrower has not heard, which may not even be selected when the first conversation happens. This is not a defect. It is a disclosure requirement doing its job. But it is a conversation, and a borrower who has it in week one is fine and a borrower who has it at the signing table is not.

2. Who they call when something is wrong. In every channel, they call you. That is the point of the job. The difference is what you can do next. Retail: you call the underwriter or walk to their desk. Correspondent: the same. Broker: you call the AE, who calls the underwriter, and the answer comes back through two people instead of none. That extra hop is usually hours and occasionally days, and managing it is a broker's core operational skill.

3. Turn times, which they experience as your competence. A borrower cannot see whose queue their file is in. They experience an eight-day underwriting turn as you being slow. This is the most underappreciated fact about channel choice: the channel's operational performance is attributed to you personally. Chapter 39 builds a pipeline discipline around communicating during dead time for exactly this reason.

4. Whether the loan is portable inside a problem. Here the broker channel has a real, concrete advantage the borrower can feel. When a file dies at lender A on day 30 over an overlay, a broker can move it to lender B — a new lock at today's market and a new underwrite, but a live transaction. When a file dies at a retail lender on day 30, the borrower starts over somewhere else, from application, with a new appraisal fee. Both outcomes are bad. Only one of them has a path.

5. The first servicing transfer, which arrives at different speeds. The clean version: retail and correspondent lenders that sell servicing-released hand the borrower to the buyer quickly — a goodbye letter and a hello letter, sometimes within weeks of closing. Lenders that retain servicing keep the relationship for years. In a broker transaction, whatever the wholesale lender does governs. Chapter 23 covers the transfer notices, and it is worth telling every borrower at application that a transfer is normal, does not change their rate, terms, or payment, and that they should never send a payment anywhere on the strength of a letter alone without verifying it.

📄 Read the File

```text FIGURE 31.1 — "The same costs, one extra column" [the Linden Street file, re-originated through the broker channel — constructed teaching example] THE DOCUMENT Closing Disclosure, page 2, Loan Costs, Section A (Origination Charges), with the "Paid by Others" column shown. Three constructed presentations of the SAME transaction: the file as actually originated (retail or correspondent), and two broker versions. THE CONTEXT $365,750 loan, 6.625% with 0.500 discount point, closing day 51. The borrowers are first-time buyers who have already asked twice why the closing costs "keep changing."

               SECTION A — ORIGINATION CHARGES        Retail /   Broker,    Broker,
                                                      corresp.   lender-    borrower-
                                                                 paid comp  paid comp
               ─────────────────────────────────────────────────────────────────────
               Lender origination / admin charge      3,657.50   1,095.00   1,095.00
               0.500% of Loan Amount (Points)         1,828.75   1,828.75   1,828.75
               Mortgage broker fee (borrower-paid)          --         --   3,657.50
               ─────────────────────────────────────────────────────────────────────
               SECTION A, PAID BY BORROWER            5,486.25   2,923.75   6,581.25
               ─────────────────────────────────────────────────────────────────────
               Broker compensation, PAID BY OTHERS          --   8,229.38         --
                 (2.250% of loan amount, marked "(L)" -- paid by the lender)
               ─────────────────────────────────────────────────────────────────────
               Note rate                               6.625%    above      at/below
                                                                 6.625%     6.625%

WHAT IT SHOWS Three arrangements of the same loan. The middle column's borrower pays $2,562.50 LESS in Section A than the retail borrower -- and the right column's borrower pays $1,095.00 MORE. The broker's compensation is visible in both broker columns: itemized as a fee in one, disclosed in the "Paid by Others" column in the other. The retail column shows no loan originator compensation anywhere, because none is required to be itemized. WHAT IT DOESN'T It does not show the note rate that produces each result, and that is the whole game. Lender-paid compensation is not free; it is bought with rate. Section A alone therefore CANNOT rank these three. It does not show what the retail creditor earns on the secondary-market sale, which is real revenue and appears on no disclosure the borrower ever sees. And it says nothing about which loan will actually close on time. THE DECISION Never compare offers on Section A. Compare at a common rate, or compare total cost over a stated holding period at each offer's own rate, and show the borrower the arithmetic. If they ask why the broker's compensation is printed and yours is not, tell them the truth: the rules require disclosure of a broker's compensation and not of a creditor's employee's, and it does not mean one is more expensive. THE LESSON The disclosure asymmetry between channels is real, and it measures VISIBILITY, not COST. A borrower who chooses on the visible number is choosing on the wrong number, and the loan officer who lets them is either not paying attention or counting on it. ```

Constructed. The retail column's \$3,657.50 origination charge and \$1,828.75 in points are the Linden Street file's frozen figures; the broker columns are illustrative arrangements built to the same loan amount. Actual charges, compensation plans, and disclosure placement vary — verify current requirements with your compliance department.

31.8 Compensation and disclosure differences

Two separate subjects get tangled here constantly, so separate them first.

Chapter 26 owns the loan originator compensation rule, and its central prohibitions apply to individual originators in every channel: compensation may not vary based on a term of the transaction, an originator may not be paid by both the consumer and another person on the same transaction, and steering a consumer to a transaction that pays the originator more is prohibited. Retail, broker, correspondent, bank, non-bank — the rule does not care.

This chapter owns the structural question: where the money each company earns actually comes from, and which parts of it a borrower can see.

Where the revenue comes from, by model

Read the middle column. A brokerage's revenue is compensation on closed loans and essentially nothing else. It has no gain on sale because it never owns a loan, no servicing because it never funds one, and no exposure to a loan's market value for the same reason. That is a genuinely different business: lower risk, lower capital, thinner and more concentrated revenue, and total dependence on volume and on the lenders it is approved with.

Read the outer columns. A retail or correspondent lender's largest revenue line is usually the one the borrower never sees, because it is not a charge to the borrower — it is a price paid by an investor. That is also the revenue line that disappears when a loan cannot be sold, which is §31.4's entire lesson.

Where the disclosure differs

The compensation of a mortgage broker is disclosed to the borrower. The compensation of a creditor's own loan originator is not.

• Borrower-paid broker compensation appears as an itemized charge in the origination charges section of the Loan Estimate and Closing Disclosure. The borrower sees a number with the word "broker" next to it.
• Lender-paid broker compensation appears on the Closing Disclosure in the "Paid by Others" column, generally with a designation identifying the lender as payer. The borrower still sees a number with the word "broker" next to it — one they are not paying directly.
• A retail or correspondent creditor's employee originator is compensated out of the company's general revenue. Nothing on the borrower's disclosures itemizes it.

This asymmetry is a real feature of the disclosure regime and it produces a predictable misunderstanding: a borrower comparing a broker's Closing Disclosure to a retail lender's sees a compensation figure on one and none on the other, and concludes the broker is more expensive. That conclusion does not follow. Both companies are paid. One is required to show it.

The honest way to handle this, from either side, is the same as the honest way to handle every pricing conversation in this book: compare at a common rate, or compare total cost. Chapter 22 explains why the annual percentage rate exists and what it does and does not capture; Chapter 29 explains how a rate and a price are two views of the same thing. A loan officer who cannot conduct this comparison honestly is going to lose files to competitors who present numbers misleadingly, and the answer is to get better at the arithmetic, not to join them.

🧮 Run the Numbers

What this one loan pays, three ways — company and loan officer.

[constructed teaching example. Compensation plans vary enormously by company, channel, market, and year. These are NOT benchmarks and must not be used as any kind of expectation. The point is the SHAPE, not the level.]

Loan amount \$365,750.00 throughout.

Retail. Loan officer paid 120 basis points of loan amount (a fixed percentage of the amount of credit extended — the structure Chapter 26 explains is permitted):

$$\$365{,}750.00 \times 0.0120 = \$4{,}389.00$$

The employer receives the \$3,657.50 origination charge, the \$1,828.75 in points, and whatever the loan sells for above par — and pays out of that the loan officer, the processor, the underwriter, the closer, the branch, the technology, and the compliance function.

Broker. The brokerage's compensation with the chosen wholesale lender is 2.250%, lender-paid:

$$\$365{,}750.00 \times 0.0225 = \$8{,}229.38$$

That is the firm's revenue on this loan, not the originator's. If the originator is an employee on a 55/45 split:

Loan officer, 55%	\$4,526.16
Brokerage, 45%	\$3,703.22
Total	\$8,229.38

Out of its \$3,703.22, the brokerage pays processing, rent, its origination system, marketing, licensing and bonding for the firm and every originator in it, and its own compliance obligations. There is no second revenue line behind it.

Correspondent. Loan officer at the same 120 basis points = \$4,389.00. The company's economics differ from retail only at the far end: instead of a wholesale lender capturing the secondary-market gain, the correspondent does — \$5,063.41 net of warehouse carry on the illustrative assumptions in §31.4 — and it accepts the funding cost, the capital tied up in the haircut, and the repurchase risk that come with it.

The interpretation, which is the reason for the exercise: the loan officer's number is the least different thing about the three models. \$4,389.00, \$4,526.16, \$4,389.00 — the same order of magnitude, on the same loan, for the same work. What differs by an order of magnitude is who carries the risk and who keeps the residual. When someone tells you a channel "pays better," ask whether they are describing the originator's split or the firm's revenue, because those are different sentences and only one of them is about your paycheck.

31.9 Choosing where to work

Most loan officers pick an employer for reasons that turn out not to predict anything: the recruiter was persuasive, the split sounded higher, the office was closer, a friend worked there. Here is a better method.

Ask the questions that predict production

Take these to an interview. Ask them plainly; a good manager will respect it and a bad one will tell you something useful by being annoyed.

Match the model to your actual book

The right channel depends on the files you actually get, not the ones you imagine.

Retail fits you if you are new and need the scaffolding; your business is mostly clean purchase files where certainty of closing beats price; you want a brand borrowers recognize; you value being able to walk to the underwriter; or you want the referral flow that a depository's branch network or a builder joint venture provides.

Brokering fits you if a meaningful share of your files have something structurally difficult about them; you serve self-employed borrowers, unusual properties, or credit profiles at the edges; you are comfortable running your own operational discipline; and you would rather have four answers and no control than one answer and a hallway. It also fits people who genuinely want to run a small business, because a brokerage is one.

Correspondent fits you if you want retail's simplicity for the borrower with more control over product and price than a retail branch has; or if you are building toward ownership, because the correspondent form is where a growing origination business ends up. Understand that you are joining a company whose survival depends on funding facilities and secondary execution — ask about both.

The thing that actually matters more than the channel

After eighteen years I will tell you plainly: the channel matters less than the operations team behind you. A great processor, an underwriter who reads the file before writing conditions, a closer who catches a bad figure at four o'clock on a Thursday, and a manager who will get on the phone when a file is dying — that combination outproduces any structural advantage in this chapter. I have watched loan officers leave excellent shops for a better split and never recover their volume, because they traded a team that closed 96% of what it took in for a percentage on loans that stopped closing.

Ask the structural questions. Then go find out who you would actually be working with, and weight that heaviest.

31.10 The same file, originated three ways

Take the Linden Street loan — \$365,750.00** at **6.625%** with **0.500 point (\$1,828.75), an origination charge of \$3,657.50, borrower-paid mortgage insurance at a 0.58% annual factor producing \$176.78 a month, closing on day 51 — and originate it three ways.

Everything the borrower cares about is identical. The payment is \$3,033.72 in all three. The housing ratio is 28.89%, the total debt ratio 42.66%, cash to close \$25,376.34, reserves after closing \$12,623.66 — 4.16 months. The eleven conditions are the same eleven conditions. The day-44 furniture debt would have blown the approval in every version.

Now the differences.

THE SAME FILE, THREE WAYS — whose money crosses the table   [the Linden Street file;
                                                             channel variants constructed]

  ── RETAIL ─────────────────────────────────────────────────────────────────────
     BORROWER ──application──► YOUR EMPLOYER ──underwrites──► approves
                                    │
                                    └── wires $365,750 ──► CLOSING TABLE
                                        (own capital or its warehouse line)
                                    NOTE: your employer's name
                                    AFTER: sells the loan (Ch. 28)

  ── BROKER ─────────────────────────────────────────────────────────────────────
     BORROWER ──application──► YOUR BROKERAGE ──submits──► WHOLESALE LENDER
                                    │                            │ underwrites
                                    │                            │ draws docs
                                    │        ◄── conditions ─────┤
                                    │                            └── wires $365,750
                                    │                                ──► CLOSING TABLE
                                    NOTE: the WHOLESALE LENDER's name
                                    AFTER: it already owns the loan; sells or keeps

  ── CORRESPONDENT ──────────────────────────────────────────────────────────────
     BORROWER ──application──► YOUR EMPLOYER ──underwrites to the investor's
                                    │           guidelines──► approves
                                    │
     WAREHOUSE BANK ──advances $356,606.25──►  │
                                    └── wires $365,750 ──► CLOSING TABLE
                                        ($9,143.75 of it your employer's own cash)
                                    NOTE: your employer's name
                                    AFTER: sells to the investor in ~18 days and
                                           repays the warehouse advance

And the six questions that actually distinguish them:

(All compensation figures constructed and illustrative — see §31.8. They are not benchmarks.)

Three more rows that do not fit the brief but matter operationally:

The one row that does not change

Look at the fifth row of the first table again. In every column, the first answer is you.

The borrower does not call the underwriter, the AE, the warehouse bank, the investor, or the servicer. Not at first. When the appraisal comes in, when the condition list arrives, when the goodbye letter shows up in February from a company they have never heard of — they call the person whose name they know. That is true in all three channels and it does not depend on whose money was on the table.

Which means the structural question this chapter has spent ten sections on determines your economics, your product menu, your speed, and your credential — and does not determine your value. Your value is built out of the same materials in every channel: telling a borrower on day one that a 706 score at 95% loan-to-value does not price like the number on the website, holding a calendar that nobody else is holding, and being the one who calls back.

🗂️ The Loan File

Chapter 31 contribution: the channel map — the same file, three ways, and what each would have changed.

Everything the borrower experienced on this file is channel-independent: the \$3,033.72 payment, the 42.66% back-end ratio, the eleven conditions, the day-44 crisis, the fifty-one days. What the channel would have changed is this:

What this settles: why the file behaved the way it did operationally. The eleven-day dead window and the three-days-short lock are calendar failures in any channel — but the tools available to fix them are channel-specific, and now you can name yours.

What it does not settle: which channel this borrower would have been best served by. That depends on facts this chapter cannot supply — what this file's pricing actually looked like at three wholesale lenders on day 12, and whether any of them had an overlay that the employer did not. Exercise 31.34 asks you to reason it out anyway, which is the honest version of the question.

Open questions carried forward:

• Q31.1. If the day-44 debt had not been curable, which channel would have had a second place to send this file, and at what cost in days? (Chapter 39, pipeline)
• Q31.2. The borrowers were shopping and had been quoted a lower rate elsewhere. Does the competitor's channel explain any of that difference? (Chapter 40 resolves the comparison in full.)

Your task. In Appendix C's workbook, complete the channel map for your own situation: name your employer's model, whether you are licensed or registered, who underwrites your files, how many lenders you can place a file with, and what your escalation path is when a file is late — with the name of the person at the end of it. If you cannot fill in the last one, that is the finding.

Conclusion

Three channels, one job.

Retail puts the whole transaction inside one company: your employer takes the application, underwrites it, funds it, closes it in its own name, and sells it. You get proximity to the decision-maker and a support structure, and you get exactly one set of guidelines, one price, and one turn time. When a file does not fit, you have an exception request and nothing else.

Brokering separates the origination from the lending. Your firm assembles the file and places it with one of several wholesale lenders, who underwrites, funds, and closes in its own name. You get genuine choice — different overlays, different prices, a second place to send a file that died at the first — and you pay for it in complexity and in control, because you are a customer of the underwriting department rather than a colleague of it. Neither of those is a footnote; they are the trade, and which side of it is better depends on what your files look like.

Correspondent lending looks like retail to the borrower and like a capital-markets business from the inside. Your employer funds the loan with borrowed money on a warehouse line, closes in its own name, and sells within weeks. That borrowed money is why guidelines are absolute: on the illustrative figures in §31.4, a loan that cannot be sold at the expected price puts a \$21,945.00 hole in a business that nets \$5,063.41 on a good one. It is also why post-closing speed is a capacity constraint, and why twelve days of dwell time is worth \$415,492,000 of annual volume on a \$50,000,000 line.

Cutting across all three: depository or non-bank. A bank or credit union funds itself with insured deposits, can hold a loan, is examined by a prudential regulator — and its originators are registered, with an identifier and a background check and no license. An independent mortgage bank funds itself with borrowed money that can be withdrawn, must sell what it originates, is supervised by the states and the Bureau — and its originators are licensed, with education, testing, bonding, continuing education, and a credential that belongs to them and moves when they move. That divide exists because in 2008 Congress built supervision where supervision was missing and relied on it where it already existed. It is the most consequential thing about your employer and almost nobody explains it at the interview.

And the row that never changes: whatever the structure, the borrower calls you.

Next: Part VI ends here. Part VII turns to the borrowers who do not fit the standard file — beginning with the self-employed, where the Fulton Avenue file's two numbers, the accountant's \$9,500 a month and the underwriter's \$8,916.67, finally get reconciled line by line.

Key Terms

Retail lending — origination in which the creditor's own employees take the application, and the creditor underwrites, funds, and closes the loan in its own name. (Ch.31)

Wholesale lending — lending to consumers through originators who are not the lender's employees; the wholesale lender publishes guidelines and pricing to third-party originators, underwrites what they submit, and funds and closes in its own name. (Ch.31)

Account executive (AE) — the wholesale lender's salesperson, whose customer is the broker rather than the borrower; the broker's route to scenarios, exceptions, pricing, and escalation. (Ch.31)

Third-party originator (TPO) — from a lender's perspective, any originator that delivers it loans without being its employee: brokers and correspondents alike. (Ch.31)

Correspondent lending — origination in which the lender underwrites and closes in its own name, funding with its own borrowed capital, then sells the closed loan to an investor whose guidelines it underwrote to. (Ch.31)

Advance rate / haircut — the percentage of a loan amount a warehouse line will advance, and the remainder the lender must fund from its own cash. (Ch.31)

Cost of carry — the interest and fees a lender pays on a warehouse advance for every day between funding a loan and being paid for it. (Ch.31)

Dwell time — the average number of days a funded loan sits on a warehouse line before the investor purchases it; the constraint that converts a line's size into an annual funding capacity. (Ch.31)

Bailee letter — the document that ships with a note to an investor, preserving the warehouse bank's security interest until the purchase proceeds are applied to the advance. (Ch.31)

Table funding — a settlement at which a loan closes in one party's name while being funded by a contemporaneous advance from another party, to whom the loan is simultaneously assigned. (Ch.31)

Mini-correspondent — a small originator, often a converted brokerage, that closes in its own name using a warehouse line and sells loans immediately, frequently to the party financing the line. (Ch.31)

Depository lender — a bank, savings institution, or credit union that takes federally insured deposits, may hold loans in portfolio, and is supervised by a federal banking agency; its originators are registered rather than licensed. (Ch.31)

Non-bank lender (independent mortgage bank, IMB) — a lender that takes no deposits, funds through warehouse lines, and must sell what it originates; its originators are state-licensed. (Ch.31)

Defined in Chapter 1 and used here in full: retail lender, mortgage broker, correspondent lender, warehouse line of credit. Defined in Chapter 3: licensed versus registered originator. Defined in Chapter 14: guideline versus overlay.

Spaced Review

1. (This chapter.) A borrower asks why the company on their note is not the company on their loan officer's business card. In two sentences, name the channel this describes and explain the arrangement without making it sound like a problem.

2. (Chapter 28 + this chapter.) Trace the Linden Street loan from the closing table to an investor twice: once as a retail loan sold servicing-retained, and once as a correspondent loan sold servicing-released. Name every party that touches it and say, in each version, when the borrower's payment address changes.

3. (Chapter 29 + this chapter.) A broker prices the same file with three wholesale lenders on the same morning and gets three different rates at the same price. Using Chapter 29's components of a rate, name three legitimate reasons the numbers differ that have nothing to do with the borrower.

4. (This chapter.) On the illustrative figures in §31.4, the warehouse carry on this loan is \$74.29 per day and the lock extension cost \$60.96 per day. Explain why those two numbers are consecutive rather than alternative, and name the event that separates them.

5. (Chapters 28, 29, and this chapter.) A correspondent commits a loan for mandatory delivery, and the borrower's transaction falls apart eight days before the delivery date. Explain what the correspondent now owes and to whom, using pull-through (Chapter 29) and the warehouse mechanics of §31.4. Then say what a best-efforts commitment would have changed.