This article is educational and not lending, legal, or tax advice; timeline figures reflect ICE Mortgage Technology data through March 2026, condominium project standards reflect Fannie Mae Lender Letter LL-2026-03 as implemented August 3, 2026, and lender extension pricing varies by institution and loan program.
TL;DR — Quick Verdict
- The average purchase loan closed in 36.8 days in March 2026 — the fastest since ICE began tracking in 2019 — yet underwriting conditions still push individual files 10 to 30 days past that mark.
- Only 11 of those 36.8 days are spent moving from application to rate lock; the remaining 26 days run from lock to closing, which is the window where extension fees get triggered.
- Rate lock extensions cost roughly 0.125% to 0.375% of the loan amount per 15-day period — $500 to $1,500 on a $400,000 loan, based on published lender pricing.
- Only three changes force a new three-business-day Closing Disclosure waiting period under CFPB rules: an inaccurate APR, a loan product change, or an added prepayment penalty. Everything else can be corrected at the table.
- Comparison result: a 0.375% extension beats re-locking on a two-year horizon once market rates have risen about a quarter point; at the cheaper 0.125% end of the range, a tenth of a point is enough.
- Recommendation: build a 45-day lock rather than 30 for any condo, self-employed, or gift-funded file, and get every conditional-approval item to the underwriter within 48 hours.
Thirty-six point eight days. That is how fast the average purchase mortgage closed in March 2026, according to Intercontinental Exchange’s May 2026 Mortgage Monitor — the quickest pace since the company started measuring the metric in 2019, and down from 37.3 days a year earlier. The national average, however, describes nobody’s actual file. Behind that number sits a distribution where clean W-2 borrowers close in 21 days and condo buyers waiting on an HOA questionnaire close in 60, if they close at all.
The gap between those two outcomes is underwriting review. It is the least visible stage of the mortgage process and the one that generates almost every expensive delay — rate lock extension fees at Rocket Mortgage and UWM, per diem penalties written into the purchase contract, and in the worst case a denial after weeks of work. This article breaks down what underwriting actually examines, which conditions reliably stall files, what a delay costs in dollars, and where the 2026 rule changes have shifted the risk. Figures come from ICE Mortgage Technology, the Consumer Financial Protection Bureau, Fannie Mae, Freddie Mac, and Federal Reserve Bank of St. Louis research on Home Mortgage Disclosure Act data.
What the 36.8-Day Average Actually Contains
Break the closing timeline into its two measured segments and the risk profile changes completely. ICE reports that the typical purchase loan moved from application to rate lock in 11 days, then from rate lock to closing in an additional 26 days. Across all origination types — purchase and refinance combined — the average closing time was 38.2 days, the third-fastest on record.
That second segment carries the money. Once a rate is locked, the clock runs against a hard expiration date, and every underwriting condition that surfaces during those 26 days consumes lock time the borrower has already paid for. A 30-day lock issued at application leaves roughly four days of slack against the national average. A single re-appraisal request or a delayed employment verification erases it.
Source: Intercontinental Exchange, May 2026 ICE Mortgage Monitor (verify at ice.com). Figures reflect loans closed in March 2026 and remain ICE’s most recent published closing-timeline reading as of August 2026.
Anyone reading these averages should treat them as a floor, not a forecast. The broader mechanics of closing timelines and what affects them show wide variance by loan type, with government-backed and condo files consistently trailing conventional single-family transactions.
What Underwriting Review Actually Examines
Underwriting is verification, not evaluation. By the time a file reaches an underwriter, an automated system — Fannie Mae’s Desktop Underwriter or Freddie Mac’s Loan Product Advisor — has already returned a recommendation. The human underwriter’s job is to confirm that every input feeding that recommendation is documented and true.
Consider a concrete file. A borrower earning $145,000 as a salaried employee plus $28,000 in annual bonus income applies for a $520,000 loan on a $650,000 house. Desktop Underwriter returns an Approve/Eligible. The underwriter then has to prove the income: two years of W-2s to establish the bonus history, a year-to-date pay stub, and a verbal verification of employment dated within ten business days of closing. If the bonus arrived only once in two years, the underwriter strips it out, the debt-to-income ratio jumps from 38% to 44%, and the file needs restructuring.
That single adjustment can add a week. Research published in the Federal Reserve Bank of St. Louis Review in May 2026, analyzing more than 30 million home purchase applications in HMDA data from 2018 to 2024, found that denial rates jump 15 to 17 percentage points once debt-to-income crosses the 50% mark — the functional boundary of the market — while the 43% qualified mortgage threshold proved largely non-binding in practice. The same research documented aggregate denial rates rising from 12.2% to 15.7% during the 2022–2023 tightening cycle, driven specifically through the debt-to-income channel.
Conditions arrive in tiers. Prior-to-document conditions must clear before closing papers are drawn. Prior-to-funding conditions clear after signing but before money moves. Understanding mortgage denial causes and reapplication costs matters here because an unclearable prior-to-document condition is functionally a denial with extra steps.
The Five Conditions That Reliably Stall Files
Delay concentrates in a short list of predictable places. Each of the following adds measurable days, and each has a preemptive fix.
Modeled from lender condition-clearing practice and Fannie Mae Selling Guide project review requirements (verify at fanniemae.com). Day ranges are estimates, not measured averages; provider-level condition-clearing data is not publicly reported.
Condominium files changed materially this year, and the change is no longer pending — it is live. Fannie Mae issued Lender Letter LL-2026-03 on March 18, 2026, aligned with Freddie Mac and coordinated with the Federal Housing Finance Agency. Its central provision retires the Limited Review process outright. For loan applications dated on or after August 3, 2026, established projects that once qualified for the streamlined path must be reviewed under Full Review, or under a Waiver of Project Review where the project is eligible. The waiver was widened in the same letter to cover new and established projects of ten or fewer units, with the caveat that five- to ten-unit projects must not sit inside a master association or larger development.
The practical effect is a hard split. Small buildings got easier; anything above ten units now routes through Full Review, where the HOA financial review in underwriting examines reserves, delinquency rates, and deferred maintenance line by line. Two related provisions bear on timing. Also effective for applications dated on or after August 3, 2026, a lender relying on a reserve study must verify that the project’s budget includes the study’s highest recommended allocation, and the baseline funding method — the approach that lets a reserve balance approach zero without crossing it — is no longer acceptable. Separately, the minimum replacement reserve allocation rises from 10% to 15% of annual budgeted assessment income under Full Review for applications dated on or after January 4, 2027. Associations that have not restructured their budgets before that date will fail the test, and buyers in those projects will find conventional financing unavailable rather than merely slow.
Appraisal timing compounds the problem. A low valuation does not merely delay — it restructures the entire transaction, and the mechanics of home appraisal cost, process, and low appraisals determine whether the gap gets bridged by seller concession, buyer cash, or contract termination.
Rate Lock Extension vs. Re-Locking at Market: Which Is Better When Underwriting Runs Long?
Suppose underwriting is eight days behind and the lock expires Friday. Two options exist: pay the lender to extend the existing lock, or let it expire and re-lock at whatever the market offers.
Published lender pricing places extension fees in the range of 0.125% to 0.375% of the loan amount per 15-day extension period, though individual lenders quote in basis points and some waive the fee entirely when the delay is attributable to the lender’s own processing. Aggregated, provider-specific extension pricing is not published by any primary source, so treat this as a defensible industry range rather than a fixed schedule and request the actual quote in writing.
Run the math on a $400,000 loan locked at 6.67% — roughly the Freddie Mac national average for a 30-year fixed in mid-August 2026 — at a monthly principal and interest payment of $2,573. A 0.375% extension costs $1,500. If market rates have moved a quarter point to 6.92%, re-locking costs roughly $67 more per month, or $799 per year, and the extension recovers its cost in about 23 months. If rates moved half a point to 7.17%, the monthly difference is roughly $134 and the payback period drops to about 11 months. But if market rates have fallen a quarter point to 6.42%, re-locking saves approximately $66 monthly and letting the lock expire is the correct move — the $1,500 fee buys nothing.
The breakeven depends on which end of the fee range applies, and the distinction is easy to get wrong. A 0.375% extension on this loan needs roughly a quarter point of upward movement to repay itself inside two years. At a tenth of a point the monthly difference is only about $27, and recovery stretches past four and a half years. A 0.125% extension at $500 clears the same two-year test on a tenth of a point with room to spare. Over a full thirty-year hold the arithmetic changes again — the fee is paid once while the rate premium never stops — so a borrower who expects to keep the loan to term should extend on almost any upward move.
Verdict
Extend when market rates have risen and you expect to hold the loan; re-lock when they have fallen, or when a quoted 0.375% fee is chasing a rate movement of only a few basis points on a loan you plan to refinance out of quickly. Before accepting either, ask the loan officer directly whether the delay originated with the lender’s underwriting queue — many lenders grant one free 7-to-15-day extension in that circumstance, and the request must be made in writing. On a $400,000 loan, that single question is worth $500 to $1,500.
What the Closing Disclosure Rules Do and Do Not Delay
Widespread belief holds that any change to the Closing Disclosure restarts a three-day countdown. The regulation says otherwise, and the distinction is worth several thousand dollars in avoided extension fees.
Under 12 CFR §1026.19(f)(2)(ii), the Consumer Financial Protection Bureau specifies exactly three changes that require the borrower to receive a corrected Closing Disclosure at least three business days before consummation: a change causing the annual percentage rate to become inaccurate, a change to the disclosed loan product information, or the addition of a prepayment penalty. Every other change may be corrected at or before closing without a new waiting period.
Loan product changes mean structural changes — fixed to adjustable, or one ARM structure to another. A revised seller credit, a corrected transfer tax, or an adjusted escrow figure does not trigger the waiting period. Borrowers who understand this can push back when a closing agent asserts that a minor line-item correction requires a three-day delay.
The annual percentage rate trigger carries a tolerance. Regulation Z permits limited variance before a disclosed rate becomes legally inaccurate, and the CFPB has clarified that the “No Wait for Lower Mortgage Rate” provision of the Economic Growth, Regulatory Relief, and Consumer Protection Act did not eliminate the redisclosure requirement when a decrease renders the prior figure inaccurate. Reading a loan estimate’s key numbers against the final disclosure is how borrowers catch the changes that actually matter, and the closing disclosure line items explained in detail reveal which figures are locked and which remain fluid.
These rules may not hold their current shape indefinitely. On July 9, 2026, the CFPB published a Request for Information on Promoting Access to Mortgage Credit (91 Fed. Reg. 42,382), issued in response to Executive Order 14393 and directed almost entirely at the TILA-RESPA Integrated Disclosure rule. Its questions cover the rule’s timing and tolerance requirements, including whether consumers’ ability to waive the three-business-day Closing Disclosure waiting period should be streamlined and whether the separate rescission window on refinances remains necessary. The comment period closed August 10, 2026. No rule has been proposed, and nothing in this section changes until one is — but borrowers closing in 2027 should expect the timing mechanics described here to be under active revision.
What Most People Get Wrong About Underwriting Delays
Five errors account for a disproportionate share of blown closing dates. Each is avoidable.
Mistake 1: Treating conditional approval as approval
Conditional approval means the underwriter has identified everything still missing. Borrowers who read it as a green light stop gathering documents. Consequence: the file sits idle for days while the borrower assumes work is happening. Correct action: request the full condition list in writing the day approval issues and return every item within 48 hours.
Mistake 2: Moving money between accounts during underwriting
Transferring $30,000 from savings to checking to consolidate the down payment creates an unsourced deposit in the receiving account. Consequence: the underwriter demands a paper trail for both sides of the transfer, adding three to seven days. Correct action: consolidate funds before application, then leave every account untouched.
Mistake 3: Opening credit before funding
Lenders pull a refreshed credit report shortly before closing. A new auto loan or furniture financing changes the debt-to-income ratio. Consequence: given the 15-to-17-percentage-point denial cliff at 50% debt-to-income documented in the St. Louis Fed’s HMDA analysis, a marginal file can fail outright. Correct action: no new accounts, no large purchases, no cosigning until funds disburse.
Mistake 4: Assuming a 30-day lock fits every file
With purchase loans averaging 36.8 days from application to closing and 26 of those days falling after the lock, a 30-day lock on a condo or self-employed file is structurally too short. Consequence: an extension fee of $500 to $1,500 that a 45-day lock would have avoided for a smaller upfront cost. Correct action: match lock duration to file complexity, not to the contract date.
Mistake 5: Ignoring the seller’s per diem clause
Many purchase contracts impose a daily penalty on the buyer for failing to close on time. Consequence: a ten-day underwriting delay at $150 per day adds $1,500 on top of any extension fee. Correct action: read the delay provisions before signing and negotiate a financing-contingency extension mechanism rather than a flat penalty. Where cash is short, understanding seller concession limits toward closing costs can create room to absorb the hit.
Who Should Build Extra Time — and Who Should Not Bother
Not every borrower needs a padded timeline. The decision follows file characteristics, not anxiety.
Build a 45-day or longer lock if any of the following apply: the property is a condominium or co-op with more than ten units, which since August 2026 means a mandatory Full Review; any borrower is self-employed or receives more than 25% of income from commission or bonus; the down payment includes gift funds; the transaction involves a trust or power of attorney; the property requires repairs before closing; or the loan is FHA or VA with a case number pending.
A 30-day lock is adequate when the borrower is salaried with two years at one employer, the property is a single-family detached home with no HOA, all funds have been seasoned in one account for 60 days or more, and the debt-to-income ratio sits below 40%. That profile clears underwriting well inside the 36.8-day average.
Is paying for a longer lock worth it? On a $400,000 loan, upgrading from a 30-day to a 45-day lock typically costs a fraction of a discount point priced into the rate — meaningfully less than a 0.375% extension at $1,500. The asymmetry favors the longer lock for anyone whose file carries a single complicating factor. For borrowers weighing that upfront cost against other line items, the origination fee components and negotiability and the broader set of closing costs that are negotiable often contain more recoverable dollars than the lock pricing itself.
One caution for refinance borrowers: delays interact with prepaid interest. A closing pushed from the 28th to the 3rd of the following month changes the prepaid insurance, tax, and interest at closing and can shift the first payment date by a full month. Similarly, a delay spanning a tax or insurance due date alters the escrow account calculation and payment changes the borrower will face at the table.
Frequently Asked Questions
How long does underwriting review itself take?
The underwriting stage is not separately measured in public data. What is measured is the full timeline: ICE Mortgage Technology reports the average purchase loan closed in 36.8 days in March 2026, with 26 of those days falling between rate lock and closing. Initial underwriting review commonly returns a conditional approval within that window, but condition-clearing rounds — not the first review — consume most of the elapsed time.
Can the lender charge me for a delay it caused?
Legally, yes — rate lock extension fees are contractual, not regulated. Practically, many lenders waive one 7-to-15-day extension when the delay traces to their own processing queue. Request the waiver in writing, cite the specific dates the file sat pending, and ask for the loan officer’s manager if refused. On a $400,000 loan, a waived 0.375% extension is worth $1,500.
Does a corrected Closing Disclosure always delay closing by three days?
No. Under 12 CFR §1026.19(f)(2)(ii), only three changes trigger a new three-business-day waiting period: the annual percentage rate becoming inaccurate, a change in disclosed loan product information, or the addition of a prepayment penalty. The Consumer Financial Protection Bureau confirms that all other corrections may be delivered at or before consummation without restarting the clock. The Bureau opened a Request for Information on possible revisions to these timing rules in July 2026, but no change has been proposed.
Why are condo purchases delayed more often than single-family?
Condo files require a project-level review in addition to borrower and property review, and the HOA or management company controls documents the lender cannot obtain independently. That burden grew in 2026. Under Fannie Mae Lender Letter LL-2026-03, the streamlined Limited Review process was retired for all loan applications dated on or after August 3, 2026, pushing established projects above ten units into Full Review. A further increase in the minimum reserve allocation, from 10% to 15% of annual budgeted assessment income, applies to Full Review applications dated on or after January 4, 2027.
How We Researched This Article
Timeline figures come from Intercontinental Exchange’s May 2026 ICE Mortgage Monitor report, released May 11, 2026, which reports the average purchase loan closed in 36.8 days in March 2026 against 37.3 days a year earlier, with an 11-day application-to-lock segment and a 26-day lock-to-closing segment, and an all-origination average of 38.2 days. ICE derives these figures from its own origination platform data covering a substantial share of U.S. mortgage applications. The June, July, and August 2026 Mortgage Monitor releases addressed equity withdrawal, generational rate-lock share, and home price and equity trends respectively, and did not publish an updated closing-timeline reading, so the March 2026 figures remain current. The full report is available from ICE Mortgage Technology.
Disclosure timing rules were verified against the Consumer Financial Protection Bureau’s TILA-RESPA Integrated Disclosure FAQs, which cite 12 CFR §1026.19(f)(2)(ii) directly, and cross-checked against the Bureau’s October 2020 data point publication on how mortgages change before origination. The pending Request for Information is docketed at 91 Fed. Reg. 42,382 (July 9, 2026).
Denial-rate and debt-to-income findings come from research published in the Federal Reserve Bank of St. Louis Review in May 2026, analyzing more than 30 million home purchase applications reported under the Home Mortgage Disclosure Act between 2018 and 2024. Condominium project standards were taken directly from Fannie Mae Lender Letter LL-2026-03, dated March 18, 2026, which states its own effective dates for each provision, and cross-checked against the Fannie Mae Selling Guide project standards chapter.
Two categories required fallback treatment. Rate lock extension pricing is not published by any primary regulator or government source; the 0.125% to 0.375% per 15-day range reflects published lender and trade sources, and individual quotes vary by institution, program, and delay cause. Condition-clearing day ranges in the stall-trigger table are modeled from documented lender practice rather than measured from a reported dataset — no public source tracks average days per condition type. Both are labeled as estimates in context.
Dollar illustrations are original calculations applying stated rates and fee percentages to a $400,000 loan balance, with amortization computed on a 30-year fixed structure, principal and interest only. The 6.67% baseline reflects the Freddie Mac Primary Mortgage Market Survey reading for the week ended August 13, 2026. Research was last conducted August 2026. All figures were verified against named primary sources before publication.