06-reference/research

nonconforming lot width jumbo construction perm underwriting

2026-10-27·research-brief·source: deep-research·by Ray Data Co (deep-research synthesis)

The Lot's Nonconformity Is Dimensional and Belongs to the Site, Not a Nonconforming Use of the Improvements, So It Does Not Trigger the Agency Machinery Jumbo Lenders Copy; the One Public Prime-Jumbo Guideline Readable End to End Has No Lot-Width Overlay and Only Two LTV-Reduction Triggers, Neither of Them Zoning

The question

Verbatim: "Does the nonconforming 68-ft lot width (RS-75, 1212 S Suffolk Dr) create appraisal/LTV underwriting friction for a jumbo construction-to-perm loan?"

Context: this is the thread [[2026-08-04-flood-zone-ae-construction-loan-requirements]] opened and declined to chase, and that [[2026-08-14-construction-to-permanent-loans-florida]] re-flagged eight weeks later. The zoning-approval half is closed; this brief is only about how a lender and an appraiser treat the lot. It gates the Q1 2027 go/no-go on the financing leg.

CALIBRATION NOTICE. Jumbo is non-agency, so no selling guide settles this. What follows is: the agency convention (primary-source verified, Fannie Mae Selling Guide B4-1.3-04), one non-agency prime-jumbo guideline read end to end from its own PDF (verified, N=1, and structurally unavailable as the actual lender here), and a reasoning chain connecting them. Freddie Mac's parallel section could not be retrieved. No named Florida construction lender's overlay treatment of nonconforming lots is knowable from public sources, and none is asserted below.

What we already know (from the vault)

What the web says

Convergences and contradictions

Synthesis for RDCO

Answer to half (a), appraisal: the nonconformity is reportable, the report is a disclosure rather than a condition, and the only live exposure is comparable selection. The appraiser will state the zoning class and one of four compliance findings, and a 66.7-to-68 ft lot in a 75-ft-minimum district plausibly draws the "legal non-conforming (grandfathered)" box rather than "legal," which is a cosmetic difference in the file and not a trigger. The substantive obligation that attaches is the one in B4-1.3-04: the analysis must reflect any adverse effect on value. That effect is measured against comparable sales, and here is the asymmetry worth internalizing: this parcel's own legal description is "S 68 Ft Of Lot 34 Block 9," meaning the subdivision was platted and then split, so sub-75-ft frontages are likely common in Culbreath Bayou Unit No 4. If the comps share the characteristic, the measured adverse effect is near zero by construction, because the market already prices it. The adjustment that would actually move a number is a site adjustment, and site adjustments are driven by area, where this lot conforms at 8,653 sf. Two second-order appraisal notes. First, the mandatory collateral desktop analysis in §10.2 is where an unusual site becomes an LTV problem in practice, not through any zoning rule: a thin or mismatched comp set raises the odds of an "Indeterminate" CDA, which forces a second valuation and then uses the lower of the two. Second, the highest-and-best-use disclosure risk lands on the as-is appraisal of the existing 1962 1,842 sf house that supports the $415,622 payoff at initial closing, not on the subject-to-completion appraisal of the new build. A 1,842 sf 1962 house on a $765,816-just-value Culbreath Bayou lot is close to the textbook case where improvements do not represent highest and best use, and that must be "indicated on the appraisal report" when so. The new ~3,000 sf house is unambiguously highest and best use; the teardown is what cures the issue, not what creates it.

Answer to half (b), underwriting and LTV: no overlay was found, the agency convention it defaults to is permissive, and on this file the LTV test is not close enough for any plausible discount to bind. Take the program grid at face value as representative of the prime-jumbo class rather than of a specific lender: 80% max LTV to $3M, 43% DTI, two LTV reductions, both unrelated to zoning, explicitly non-stacking. Against the vault's own modeled numbers, roughly a $1.2M note on a $2.5M completed value is about 48% LTV. For an 80% ceiling to bind, the appraised value would have to come in near $1.5M, a roughly 40% shortfall. No frontage adjustment produces that. Even layering the two documented reductions would not: a declining-market finding takes 80% to 75%, which still needs a $1.6M value to bind. The honest conclusion is that lot width is not a financing risk on this file because this file is not LTV-constrained, and that reframes the risk register rather than merely clearing an item. The 8/14 brief's conclusion stands and is strengthened: the binding constraint is the 43% DTI cap against phData W-2 history, not collateral.

What that means for how we hold the Q1 2027 decision. This open thread should close as not a gate, and the engineering effort behind it should move to the two things that are gates. The first is income and document timing, where the 43% DTI cap and the re-verification posture at conversion live. The second is the completion-appraisal path, where the agency analogue in B5-3.1-02 already forces a new appraisal and full requalification if the 1004D comes in below the subject-to-completion number, per [[2026-10-22-construction-loan-structure-rate-lock-carry]]. One narrow residual stays open on the collateral side, and it is cheap to close: confirm the frontage is typical for the plat, because that is what makes the adverse-effect analysis trivial and keeps the comp set deep enough to avoid an indeterminate review. Three questions belong on the term sheet rather than in another research pass, since they need a lender rather than a source: whether the program imposes any site-dimension or legal-nonconforming overlay beyond the agency convention; whether a collateral desktop analysis or second appraisal is mandatory and whether lower-of-two applies; and whether land equity counts at appraised as-is value or at cost basis, which [[2026-08-14-construction-to-permanent-loans-florida]] left open and which moves required cash far more than anything in this brief.

Why this is in the vault

It closes, as a non-gate, the one open collateral question on the 1212 S Suffolk Dr financing leg that two prior briefs (2026-08-04 and 2026-08-14) each deferred, so the Q1 2027 go/no-go can stop carrying a lot-width contingency and reallocate that attention to the 43% DTI cap and the completion-appraisal requalification trigger. It also corrects a specific vault claim: that a national jumbo program's overlay matrix would penalize the 68-ft lot where a local bank would not.

Open follow-ups

Related

Sources