Case Study 2 — The Overlay as Competitive Weapon and Fair-Lending Exposure

DataField.Dev

Case Study 2 — The Overlay as Competitive Weapon and Fair-Lending Exposure

What this case is about: the same rule, seen from two directions. To a sales manager, an overlay is a product decision — something you either have or do not have, and the lender who does not have it wins the file. To a compliance officer, the identical rule is a policy applied to a protected class of applicants whose effects have to be measured. Both readings are correct. This case study is about what a loan officer does when they are standing between them.

Case Study 1 explained why overlays exist. This one is about the part nobody resolves: overlays are legal, rational, and unpublished, and those three properties together create a problem the industry has not solved.

Sourcing note. The legal framework (ECOA and Regulation B, the Fair Housing Act, HMDA and Regulation C, and the discriminatory-effects doctrine) is Tier 1 and real. The market patterns described are Tier 2 — real, documented in trade and policy literature, with magnitudes that vary by source and method. The two lender scenarios are explicitly labeled composites built from patterns that recur across the industry; they are not any particular institution. No statistic in this case study is invented, because none is stated. Where you want a number, go to the HMDA public data, the agencies, or a named research institution, and cite it with a date.

Part A — The competitive story

The setup [COMPOSITE — constructed from a recurring industry pattern]

Two lenders. Both sell conventional loans to the same agency. Both apply the same published guide.

Lender A carries an overlay: a minimum representative credit score several points above the agency's floor on any loan above 90% loan-to-value, plus a required minimum of two months of reserves at those loan-to-values. The overlay came out of a credit-policy review after a cluster of early payment defaults in that segment three years ago. It is documented, it is applied consistently, and nobody at Lender A thinks it is unreasonable.

Lender B does not carry it. Lender B is smaller, sells to a different aggregator, retains no servicing, and has decided that this segment is where it can win business against larger competitors.

A borrower walks into Lender A. Score just under the overlay floor. Ninety-five percent loan-to-value. Reserves: one and a half months. Agency-eligible. Lender A declines.

What happens next depends entirely on the loan officer

Three loan officers, same file, same decline.

The first tells the borrower they "don't qualify for a conventional loan." This is false. What is true is that they do not qualify at this lender. The borrower goes home and stops looking. This happens constantly and it is the most common failure in the entire chapter.

The second tells the borrower the truth — "this is our rule, not the agency's; another lender may say yes" — and refers them out. They lose the commission and keep the relationship. In two years that borrower refinances, and their sister buys a house.

The third asked the question before the file was ever submitted. On the day of the pre-approval they checked the overlay matrix, saw the score-and-reserve constraint, and had a different conversation up front: here is what our shop can do, here is what it cannot, and here are the two things that would change the answer — a rapid rescore if there is anything to correct on the report (Chapter 10), or eight weeks of documented saving to clear the reserve requirement.

Notice that only the third loan officer produced a plan. The other two produced outcomes.

The competitive dynamic, stated plainly

Overlays are the largest source of genuine product differentiation in conforming lending. Every lender is selling essentially the same agency loan under essentially the same guide. What actually differs between them, file by file, is:

WHAT ACTUALLY DIFFERS BETWEEN TWO CONFORMING LENDERS
                                            [constructed teaching summary]

  IDENTICAL                        DIFFERENT
  ---------------------------------------------------------------------
  the agency guide                 the overlays
  the eligibility matrix           the exception process and who owns it
  the note, the security           turn times and operational capacity
    instrument, the disclosures    which aggregator or agency they sell to
  federal law                      whether servicing is retained
  the property                     mortgage insurer relationships
  ---------------------------------------------------------------------
  A borrower shopping "rate" is comparing the left column.
  A borrower whose file is tight is being decided by the right column.

This is why a broker's structural advantage is real and not merely marketing (Chapter 1, Chapter 31): access to multiple overlay sets is access to multiple answers. It is also why a retail loan officer who knows their own shop's matrix cold is worth substantially more than one who does not — they can tell a borrower on day one whether this is the right building.

Part B — The fair-lending problem

Now run the identical facts through a different lens.

An overlay is a credit standard set by the lender, applied to applicants. Everything a lender does with applicants is subject to the Equal Credit Opportunity Act and Regulation B, and everything it does with housing credit is subject to the Fair Housing Act. Chapter 25 covers this material properly; what belongs here is the specific intersection with §14.7.

Three exposures, in increasing order of subtlety.

1. Inconsistent application. An overlay waived for one applicant and enforced against another, where the difference is not a documented, consistently applied credit standard, is the clearest version of the problem. Exceptions granted by relationship — because the loan officer is a top producer, because the borrower knows the branch manager, because somebody called somebody — are exactly this. It is why §14.9 insists that an exception be requested in writing, decided through a defined process, and recorded. The written record is not bureaucracy. It is the evidence that the standard was the standard.

2. Discouragement. Regulation B addresses conduct that would discourage a reasonable person from making or pursuing an application. A loan officer who says "there's no point applying, we'd never approve that" has made a credit decision without an application, without an underwriter, and without producing an adverse action notice the applicant could act on. Take the application. Let the process produce the answer in writing. This is stated in §14.7 and it is worth stating twice.

3. Effects. A facially neutral rule, applied uniformly, with no discriminatory intent whatsoever, can still produce outcomes that differ substantially across protected classes — and the discriminatory-effects (disparate impact) doctrine addresses exactly that situation. The regulatory standard here has been written, amended, and rewritten over the last decade and is one of the most contested areas in housing-finance law; verify the current rule and the current agency guidance rather than relying on any textbook's summary, including this one.

The reason this matters for overlays specifically is arithmetic that nobody disputes: credit scores, down-payment capacity, and reserve balances are not distributed evenly across the population. An overlay keyed to any of them will not fall evenly either. That is not an accusation against any lender. It is a description of what a score floor does.

The 2020 episode [documented market event; specifics Tier 2]

The clearest recent illustration is public and recent enough that many working loan officers lived through it.

In the spring of 2020, the arrival of the pandemic produced an acute disruption in mortgage liquidity and an immediate, enormous uncertainty about forbearance: a lender that closed a loan and could not sell it, or sold it and had to advance payments on a borrower who never made the first one, faced losses it could not size. Lenders responded within days — not months — by imposing temporary overlays. Minimum score requirements were raised. Certain products were suspended outright. Larger down payments were required on some programs. Some lenders stopped taking certain government-program applications altogether for a period.

Every one of those decisions was a rational response to a real and immediate risk, made under genuine time pressure, by people acting in good faith. And the population of would-be borrowers who were eligible under the published program rules but could not find a lender that week was disproportionately made up of exactly the households with the least cushion.

Both sentences in that paragraph are true simultaneously. That is what makes this a case study rather than a rule.

Part C — Where the chapter's advice runs out

§14.7 gives a loan officer a good question — is that the agency's rule, or ours? — and a good move: if it is ours, another lender may say yes. That advice is correct and it is not sufficient, for three reasons worth being honest about.

The advice only helps a borrower who already reached a loan officer who gives it. A borrower declined by the only lender they called, by a loan officer who said "you don't qualify," gets nothing from any of this. The self-correcting market described in Part A corrects only for the people inside it.

Overlay-shopping has a hard boundary and it is not a fuzzy one. Moving a file to a lender without a particular overlay is ordinary brokerage. Moving a file to escape a finding — a discrepancy, a misrepresentation, something an underwriter caught — is fraud (Chapter 27). The facts travel with the file. Always. If you would not want the second underwriter to know what the first one found, you are not shopping an overlay; you are concealing a defect.

Some declines are the right answer. A chapter about how to get files approved has an obligation to say this. Overlays exist because certain loans defaulted at rates their originators did not expect, and some of those loans should not have been made. A borrower with one and a half months of reserves buying at 95% loan-to-value is genuinely exposed, and the fact that some lender somewhere will do the loan does not by itself mean the loan is a good idea for that household. §14.6's whole point is that you are able to see this before the underwriter tells you. Seeing it and saying nothing because a different lender will close it is not skill. It is the thing this book has spent fourteen chapters arguing against.

Discussion questions

Lender A's overlay came from a documented cluster of early payment defaults. Lender B does not have the overlay. Is Lender B being reckless, or is it correctly pricing a risk Lender A has overreacted to? What evidence would you need to decide, and does a loan officer ever get access to that evidence?
Write the exact sentence you would say to a borrower declined by your employer for an overlay. It must be true, it must not blame your employer, it must not raise false hope, and it must leave the borrower with a specific next step. Then write the sentence you would say if the decline were an agency guideline instead, and identify every way the two differ.
Overlays are not published publicly. Argue for and against requiring lenders to publish them. Address, on the "for" side, what borrowers could do with the information; and on the "against" side, what a published overlay matrix would tell competitors and what a lender might do in response.
Part B says a facially neutral overlay can produce uneven effects because scores, down payments, and reserves are unevenly distributed. Suppose you are a credit-policy officer who has just been shown that data for your own overlay. Name three things you could do next, and rank them by how much each one actually helps an applicant.
The 2020 episode describes decisions made in days, in good faith, under real risk, with uneven effects. Design a process — not a rule — that a lender could adopt in advance so that the next emergency produces better-documented and more reviewable decisions. What would the process cost, and who inside the company would resist it?
Part C claims that some declines are the right answer, and that closing a file elsewhere because you can is not skill. Take the opposite position as strongly as you can: the borrower is an adult, the loan is agency-eligible, and it is not a loan officer's job to substitute their judgment for the borrower's. Where does that argument hold, and where does it break?