Case Study 22.2 — The Transition: What Happened When TRID Went Live

Type: Two parts. Part 1 is the documented public record of the industry's transition to TRID (Tier 1 for dates and rulemaking; Tier 2 for described patterns, with no invented statistics). Part 2 is a clearly labeled composite — a lender's implementation failure assembled from documented industry patterns, not an account of any real company. Every figure in Part 2 is constructed.

Case Study 22.1 covered how the forms were designed. This one covers what happened when four hundred thousand people had to start using them on the same Saturday.

Part 1 — The public record

The delay

The final rule was issued in November 2013 with an effective date of August 1, 2015 — an unusually long runway, deliberately, because every loan origination system, document provider, settlement agent platform, and investor delivery process in the country had to change.

In June 2015, weeks before go-live, the Bureau announced a short delay. The effective date moved to October 3, 2015. The stated reason was procedural rather than substantive, and the practical effect was that a large industry which had spent two years preparing for one date spent two months preparing for another.

Two things about that delay are worth noticing as a professional.

First, an industry-wide compliance deadline is not like a personal deadline. Thousands of independent organizations had to be simultaneously ready, and readiness was not distributed evenly — the largest lenders had dedicated TRID programs running for eighteen months, and the smallest had a vendor promise and hope.

Second, the delay did not create additional readiness in most shops. It created two extra months of uncertainty. Organizations that were ready stayed ready; organizations that were behind mostly stayed behind.

What went wrong first: the calendar

The most immediate, most visible effect of TRID was that loans took longer to close.

The mechanism is not mysterious and this chapter has already explained it. The three-business-day Closing Disclosure receipt rule inserted a hard, non-negotiable interval between "clear to close" and "consummation" that had not previously existed in that form. On top of it, the Closing Disclosure had to be produced by the lender rather than by the settlement agent — a reversal of decades of practice — which meant lender and settlement agent had to reconcile figures earlier and more formally than before.

Industry reporting through late 2015 and into 2016 consistently described average time-to-close lengthening after implementation, before settling back as processes matured. The direction of the pattern is well documented; specific monthly averages vary by source and by product mix, so treat any single figure you see quoted as an industry benchmark rather than a fact, and check its definition before you repeat it.

For a loan officer, the operational lesson survived the transition and is still true: the file must be clear to close roughly a week before the closing date, not the day before. The Linden Street file is a demonstration — clear to close on day 47, Closing Disclosure on day 48, funding on day 51, with zero spare business days in between.

What went wrong second: the secondary market

The harder problem showed up in the money.

A lender that originates a loan usually sells it within weeks (Chapter 1, §1.3). The buyer — an aggregator, an investor, an agency — reviews the file before purchase. Beginning in late 2015, purchasers began identifying TRID defects in loan files: forms with incorrect placement of a charge, disclosures whose timing could not be documented, tolerance analyses that did not hold up, fields completed in a way that did not match the model form.

Some of these defects were substantive. Many were technical. The problem was that in the first months after implementation, nobody was certain which was which, and a purchaser with any doubt about assignee liability had an obvious response: decline the loan, or buy it at a discount.

That is a serious event for a lender. A loan that cannot be sold sits on the warehouse line, funded with borrowed money, generating carrying cost and consuming capacity that would otherwise fund new originations. A small backlog of unsalable loans is an inconvenience. A large one is existential.

The industry escalated, and in late December 2015 the Bureau's director sent a widely publicized letter to the Mortgage Bankers Association addressing the concern. The letter's substance, as reported and discussed across the industry at the time, was that the Bureau's early examinations would be diagnostic and corrective in orientation — focused on good-faith efforts to come into compliance — and it addressed the scope of liability that industry participants feared attached to minor technical errors. It did not repeal anything. It did materially calm the secondary market.

What happened next: the rule got amended

TRID has not stood still.

In 2017 the Bureau issued a substantial set of amendments, generally called TRID 2.0, resolving dozens of interpretive questions the first two years had surfaced and adding clarity to areas the original rule handled ambiguously.

In 2018 the Bureau closed the "black hole." Under the original rule, a creditor could not issue a revised Loan Estimate on or after the date the Closing Disclosure was provided, and could use a Closing Disclosure to reset tolerances only within a narrow window before consummation. Between those two constraints sat a gap in which a legitimate changed circumstance could arise and the creditor had no mechanism to disclose it — so the creditor simply absorbed the cost. The amendment removed the timing restriction, permitting a creditor to use an initial or corrected Closing Disclosure to reset tolerances when the timing requirements are met.

In 2020 the Bureau published an assessment of the rule, as its statute requires for significant rules after five years. It described improvements in consumers' ability to locate and understand key loan information, against substantial one-time implementation costs.

The professional point: the version of TRID you learn is not the version that will be in force for your whole career. Learn the structure — two forms, two clocks, three buckets, three restart triggers — and take the current values, deadlines, and interpretive details from your compliance department every time they matter.

Part 2 — A composite implementation failure

This is a constructed composite built from documented industry patterns. It is not an account of any real lender, and every figure in it is illustrative. It is included because the mechanism it describes is real and recurs, and because it is the kind of failure that produces no symptom until the moment it produces a large one.

The setup

A mid-size correspondent lender, roughly 900 closed units a year across four states, implemented TRID on time with a vendor loan origination system and a document provider. Implementation went well by the standards of the period. Closings slipped for about six weeks and then recovered. The tolerance analysis was automated: the system compared the operative Loan Estimate to the Closing Disclosure, bucketed every charge, ran the zero-tolerance and ten-percent tests, and flagged exceptions for a compliance analyst.

Nine months in, a routine post-close quality control sample turned up a file where a title fee had increased 6% and been passed to the borrower without a flag. The analyst pulled the file to confirm that 6% is comfortably inside a ten-percent bucket, and then noticed that the file contained no written list of service providers.

The mechanism

Regulation Z's structure here is precise, and the composite lender's system did not model it.

A creditor that permits the consumer to shop for a settlement service must provide a written list of service providers. If the consumer selects a provider from that list, the charge sits in the ten-percent cumulative bucket. If the consumer selects a provider not on the list, the charge has unlimited tolerance.

But the list is a condition of the whole arrangement. If the creditor permits shopping and fails to provide the written list, the charges for those services are treated as charges for services the consumer could not shop for — which means zero tolerance. Verify the current rule with your compliance department; the structure is what matters here.

The composite lender's document template had a conditional flag that suppressed the written list when a particular product code was used. The flag had been set during implementation testing and never reset. It affected one product on one channel. Every affected file still showed Section C charges, still ran a tolerance analysis, and still passed — because the automated test was running the ten-percent math on charges that, in the absence of the list, carried a zero tolerance.

The system was working perfectly. It was answering the wrong question.

The arithmetic

THE COMPOSITE FAILURE                            [constructed teaching example]

  Files closed on the affected product/channel over 19 weeks         212
  Files where a Section C charge increased at all                    138
  Files where the increase was zero or negative                       74

  Correct tolerance for those 138 files              ZERO (no written list delivered)
  Tolerance the system applied                       10% cumulative
  Exceptions flagged by the system                   0

  Average zero-tolerance exposure per affected file              $187.40
  Total refunds owed          138 x $187.40  =                $25,861.20
  Corrected Closing Disclosures required                             138

  Of the 138, files still within 60 calendar days of consummation
  at the moment of discovery                                          97
  Files already past 60 days                                          41

The refunds were the smallest number in the incident. The real cost was the remediation: a file-by-file re-analysis of every loan on that product and channel for the full period, 138 corrected Closing Disclosures produced and delivered with proof of delivery, refunds issued and tracked to confirmation, outside counsel engaged on the 41 files that could no longer be cured inside the period the rule provides, self-reporting, a system fix, a re-test, and a control added to detect the condition going forward. Multiples of \$25,861.20, and months of a compliance team's capacity.

Why nobody caught it earlier

Four reasons, and all four are general.

The failure was silent. Nothing was out of balance. No form failed to generate. No borrower complained, because no borrower knew the list existed or that its absence changed anything.

The control tested the output, not the input. The tolerance engine verified that the arithmetic was right. It did not verify that the bucket assignment was right, and the bucket assignment depended on a document nobody was checking for.

The exception was a product code, not a person. One channel, one product. It never appeared in a sample large enough to notice until a random pull happened to land on it.

Everyone assumed someone else owned the list. Loan officers assumed the system produced it. Processors assumed the disclosure desk sent it. The disclosure desk assumed the template included it. It did, on every product but one.

The lesson

A compliance control that checks the math but not the classification is not a control. The hardest part of tolerance analysis is not the arithmetic — the arithmetic is a subtraction and a percentage. It is knowing which bucket a charge belongs in, and that answer frequently lives in a document somewhere else in the file.

Three things follow for a loan officer, none of which require you to be a compliance officer.

Know what should be in the disclosure package, and look at one occasionally. You send them every week. Open one all the way through, once a month, and confirm the written list of service providers is in it. This is a five-minute habit that would have caught the composite failure in week one.

When something in your file is unusual — a product you rarely use, a channel, an exception — assume the automation has not been tested on it. Systems are validated against the common case.

Ask the question that has no symptom. "Did the borrower get the written list, and do we have proof of it?" is not a question anything in your workflow will prompt you to ask. Which is exactly why it is the one worth asking.

Discussion questions

1. The secondary market's reaction in late 2015 — declining or discounting loans with technical TRID defects — was arguably an overreaction. Was it? Whose money was at risk, and what would you have done in the purchaser's position?

2. The composite lender's automated control passed 138 files that should have failed. Design a second control that would have caught it. What does your control cost, and what would it miss?

3. The "black hole" existed for roughly three years before it was amended away. During that period, creditors absorbed costs the rule gave them no mechanism to disclose. Is that a defect in the rule or an acceptable cost of a bright-line timing regime? Argue both.

4. Part 1 describes a regulator responding to industry distress with a letter about examination posture rather than a rule change. What are the advantages of that tool, and what are its risks for a lender that relies on it?

5. In the composite, 41 files were past 60 days from consummation at discovery. Describe what you believe the lender's obligations were at that point, and identify precisely which part of your answer you would refuse to give without your compliance department.

6. The chapter argues that "compliance is not paperwork, it's the license." Using the composite, identify the specific moment at which a paperwork problem became a licensing problem — and the specific, cheap habit that would have prevented it.