Site sourcing and screening — North Dakota

The QAP everyone finds first is not the QAP that governs an application filed today

This is worth stating before anything else, because it changes which document the rest of this phase is built on. The file NDHFA has published for years at a stable URL, titled the "2026 Allocation Plan," lists exactly one 9% application round with a closing date of September 30, 2025 -- a date that, as of today, is nearly a year in the past. That is not a typo: North Dakota, like a handful of other small-allocation states, approves next year's plan the year before, so the "2026" plan governs credits allocated during calendar 2026 based on an application that closed in September 2025. But the Industrial Commission has since moved on. Its own March 25, 2026 meeting minutes record approval of a full "2027 Low Income Housing Tax Credit Qualified Allocation Plan," following a draft published February 17, 2026, a public comment period that closed March 4, 2026, and a hearing held March 6, 2026 -- and that 2027 plan's own text sets its Round 1 application window opening September 1, 2026 and closing on the last business day of that month. As of today, that closing date is roughly a week away. A developer who found only the "2026" PDF through a search engine and built a submission calendar around it would be planning against a cycle that already closed.

One more thing worth flagging rather than explaining away: the "2026" PDF still hosted at NDHFA's own site carries a printed page-footer date of 06/10/2025 on every page, but the file's own metadata shows it was last modified February 26, 2026 -- eight months after that printed date, and after its own Round 1 had already closed. Nothing in the visible text suggests a substantive change (the content matches what the footer date implies), and this could simply be a routine server-side re-save. It is called out here only because the research standard for this library is to flag exactly this kind of unexplained gap between a document's stated date and its file history rather than assume it away, and because sibling states in this same library have had QAPs quietly amended in place without a renamed file.

What actually gets a site disqualified

The 2027 plan's site-suitability standard lives in Section 3(B), under "Ineligible Projects," not in the scoring section at all. It is short, it is discretionary, and its trigger phrase is a rejection ground rather than a checklist item to satisfy: "Agency may reject an application for detrimental characteristics on or adjacent to the proposed project site unless a satisfactory remediation plan and budget are provided." The list itself is explicitly non-exhaustive ("Unsuitable sites include, but are not limited to, those that..."), which means the four enumerated categories are a floor for what the Agency will flag, not a ceiling.

What is not on this list, or anywhere else in the plan's scoring section, is a points table for a site's proximity to transit, grocery stores, schools, medical care, or employment -- the kind of amenity-distance scoring that drives site selection in California's and Texas's competitive rounds. North Dakota's own "Geographic Location" heading, which sits inside Section 7 (Project Ratings) where a scoring table would normally be, carries no points at all: "No more than two projects will be approved in any one city during a competitive application round." It is a hard cap, not a score -- a good site in a city that already has two pending applications ahead of it in the same round is dead on arrival regardless of the site's own merits, and that fact will not show up in any location-scoring analysis because there isn't one.

One ceiling, one closing date, and a Native American set-aside that runs first

The 2027 plan's Round 1 allocates $3,701,073 in 9% credits statewide -- the same figure printed in the 2026 plan for its own Round 1, a coincidence worth independently verifying with NDHFA rather than assuming reflects any particular formula, since the plan itself does not explain how the figure is derived. Additional 9% rounds are possible but are published "at the sole discretion of the Agency" -- nothing in the current plan commits to a second round. Applications must score a minimum of 55 points to be eligible for further consideration at all, out of a maximum that adds up to roughly 100 across all scoring categories, which means the state's entire 9% program for the cycle is decided in a single competitive window.

Two structural facts change how a site should be screened before that window opens. First, Section 8(B)'s Native American Set-Aside requires the Agency to "award the highest-scoring qualified application in the first application round located within North Dakota Indian reservations or on tribal land (either held in trust or fee-simple), subject to the Maximum Credit Limit" -- a real, guaranteed set-aside for a site on a reservation or on the Trenton Indian Service Area, not merely a scoring preference. Second, 4% credits paired with tax-exempt bond financing run on an entirely separate, non-competitive track: proposals may be submitted "at any time between January 1st and August 31st," and the 2027 plan states plainly that "Applications for an allocation of noncompetitive 4% LIHTCs are not subject to a minimum score." A 4% bond deal in North Dakota is not fighting for a place in the same $3.7 million pool at all.

The Bakken makes the market study the real screening tool, not the map

North Dakota's population is heavily concentrated in four cities -- Fargo, Bismarck, Grand Forks, and Minot -- and thin almost everywhere else, but the exception that matters most for site screening is the oil-producing counties of the Bakken formation in the northwest: McKenzie, Williams, Mountrail, Dunn, and Stark. The U.S. Census Bureau's own account of the 2010-2019 period names these five counties as the five fastest-growing in the entire Northern Great Plains region, and separately reports that McKenzie County -- home to Watford City -- was "the fastest-growing county in the nation" over that span, driven by the oil extraction boom. Decennial Census counts put McKenzie County's population at 6,360 in 2010 and 14,704 in 2020, a 131% increase, and Williams County's at 22,398 in 2010 and 40,950 in 2020. The Census Bureau's own regional summary also notes these same oil counties "experienced growth in the beginning of the decade, lost population in the middle of the decade, and then grew again" -- boom, partial bust, and a second boom, inside a single ten-year window.

That volatility lands directly on the one diligence item every application needs regardless of score: the market study required under Section 5(J), which "must have been completed within six (6) months of Application Close Date" and must follow National Council of Housing Market Analysts (NCHMA) content standards. A six-month-old market study is a reasonable freshness window in Fargo or Bismarck, where demand moves slowly. In a Bakken county that has already gone through one boom-bust-boom cycle inside a decade, six months is a much thinner margin of safety, and a screening pass that stops at the QAP's physical unsuitable-site list without separately stress-testing the market study's demand assumptions against that county's own recent population swings is checking the wrong risk.

QCT and DDA status is federal, not an NDHFA layer -- and this plan does not publish its own list

Basis boost eligibility in Section 2(B) tracks the standard federal designations directly: a project "entirely located within a HUD-designated Difficult Development Area (DDA)" or "entirely located within a HUD-designated Qualified Census Tract (QCT)" qualifies for a 30% increase in non-acquisition eligible basis, available to tax-exempt bond deals as well as 9% deals. Nothing in either the 2026 or 2027 plan layers a state-specific geography on top of that federal designation, and nothing in the plan republishes HUD's own county- or tract-level QCT/DDA list. That determination has to be pulled from HUD's own SADDA lookup tool for the current calendar year -- this research did not independently verify which specific North Dakota tracts or non-metro counties currently carry a QCT or DDA designation, and that list changes annually, so it should be treated as a live lookup rather than a fact to hardcode.

Redevelopment scoring (Section 7.B) does reach QCT status indirectly -- a project "in a QCT or city revitalization area established by resolution or other legal action by the city" can earn 5 points if it also contributes to a locally-adopted, geographically-bounded revitalization plan, with a specific carve-out excluding plans "so broad as to encompass the entire community or so narrow as to encompass only the project's subject property." That is a real, if narrow, use of QCT status inside the scoring table -- but it is the only one, and it is not a location score in the sense CA or TX developers would recognize.

Where this leaves screening in practice

A North Dakota screen has fewer open datasets to check than California's or Texas's and a shorter physical-disqualifier list, but it trades that simplicity for two things that matter more here than almost anywhere else in this library: a hard annual ceiling small enough that the two-projects-per-city cap and the Native American set-aside can each single-handedly decide whether a given site is even worth submitting this cycle, and a market whose fastest-growing corner has already proven it can add and shed population inside a single reporting decade. EZFeasi has no North Dakota parcel, zoning, QCT/DDA, or award data loaded in any tool today -- this entire phase is greenfield product work.