Methodology, updates and corrections
Every figure on LIHTC Atlas comes from a public file or ruling we can name, and every computed figure follows a rule written down here. This page explains how each dataset is sourced, refreshed and computed, lists when the data last changed, and says how to report an error and what we have corrected.
Income limits (MTSP)
Source: HUD's FY2026 Multifamily Tax Subsidy Project income limits, MTSP-Data-FY26.xlsx for the 50% and 60% limits and MTSP-IncAvg-Data-FY26.xlsx for the 20% to 80% income averaging levels, both dated March 27, 2026 (HUD USER). One record per county, or per town in New England: 4,764 areas. Where a town carries a different FIPS code in the two files, the records are matched by state, county and town name. HERA Special limits are shown only where HUD lists them. We use HUD's numbers as published and do not round or adjust them.
Maximum rents
Gross rent limits follow IRS Publication 5913 and IRC 42(g)(2): 30% of the income limit for the unit's imputed household size (1 person for a studio, 1.5 people per bedroom), divided by 12. When the imputed size is not a whole number (1.5, 4.5, 7.5 people) we average the two neighboring limits. The result is rounded down to the dollar. The limit is gross rent, so it includes the utility allowance; the tenant's share is the limit minus the allowance. Income averaging rents use HUD's 20% to 80% limits the same way.
QCT and DDA checks
Source: HUD's 2027 designations (FR Doc. 2026-20390, October 6, 2026), from QCT2027.csv and "2027 DDAs: Data Used to Designate": 14,539 qualified census tracts, 2,657 metropolitan Small DDA ZCTAs and 286 nonmetropolitan DDA counties. They govern allocations and bond issues after December 31, 2026. HUD's map layers labeled "2026" hold the same tracts and areas, so they are not used separately.
The address lookup sends the typed address to the U.S. Census Bureau geocoder (current benchmark, 2020 tracts and ZCTAs) and matches the returned tract against the QCT list, the ZCTA against the Small DDA list and the county against the nonmetro DDA list. The address is not stored.
ZIP vs ZCTA. HUD draws metro DDAs by ZIP Code Tabulation Area, the Census Bureau's area approximation of a ZIP code. A ZCTA usually shares its number with the ZIP, but the boundaries differ and some ZIPs have no ZCTA. The tool shows the ZCTA the geocoder found and flags a mismatch with the typed ZIP. A point near a tract line can also land on the wrong side. Confirm any result you will rely on in an application with HUD's own map and the allocating agency.
Property records
Source: HUD's national LIHTC database, LIHTCPUB (revised May 28, 2026; projects placed in service through 2024), read with HUD's 2024 data dictionary (HUD USER). We count properties three ways, and label each:
- Every record in HUD's file: 55,345 projects.
- Still monitored: 42,673 properties, after dropping those the allocating agency flagged as no longer monitored and HUD's test and demo records. State pages and county totals use this count.
- On this site's city pages: 32,252 properties in 1,920 cities with at least five monitored properties, after merging HUD records that repeat one property at the same address and unit count.
Names. HUD's names are upper case and abbreviated. We convert them to title case and expand only abbreviations whose meaning is unambiguous in a property name (Apts to Apartments, Hsg to Housing, Ctr to Center, Ph II to Phase II, Sr before Apartments or Housing to Senior). "St", "Mt" and unknown acronyms are left as HUD wrote them, and the property page also shows HUD's own spelling. Addresses drop a trailing unit or suite number. Owner and contact fields are not published, and any email, web address or phone number in a published field is stripped; the build fails if one remains.
Compliance and extended use years. From HUD's placed-in-service year (Y): the 15-year compliance period runs from Y to about Y + 14, or one year later if the owner elected to start the credit period the next year (IRC 42(f)(1), 42(i)(1)). For credits allocated after 1989, the extended use agreement keeps units restricted for at least 30 years in all, so to Y + 29 at the earliest (IRC 42(h)(6)); many agencies require longer. These are estimates from HUD's record. The recorded agreement governs.
State credit ceilings and bond caps
For 2026, each state's 9% ceiling (population amount) is the greater of $3.416 times its population or $3,953,600, and its private activity bond cap is the greater of $135 times population or $397,625,000 (Rev. Proc. 2025-32 secs. 4.08 and 4.19; the 9% amounts already include the 12% increase from Pub. L. 119-21 sec. 70422(a)). Populations are from IRS Notice 2026-22. Insular areas smaller than the least populous state get a bond cap of population times $397,625,000 divided by Wyoming's population (IRC 146(d)(4)). We compute these and round to the dollar; an agency's own figure also adds carryforwards, returned credits and any national pool share, so it can be higher.
Applicable percentages
The monthly 70% and 30% present value rates come from the IRS revenue ruling for each month, Table 4, January 2024 (Rev. Rul. 2024-2) to November 2026 (Rev. Rul. 2026-21), 35 months, each linked to its ruling. The fixed 9% floor (IRC 42(b)(2)) and 4% floor (IRC 42(b)(3)) are shown beside them; the floors are higher than every computed rate in the table.
Equity pricing
Medians are the net equity price per dollar of federal credit, all deal types, from CohnReznick's Tax Credit Advisor Housing Tax Credit Monitor, the main public survey: 12 readings from Sep to Oct 2024 to Jun to Jul 2026. We list only figures printed in the monitor PDFs, each linked to its issue. Where an issue says only "the last 60 days", the period is placed by the order of surveys in the monitor's own trend chart and marked. Fund ranges are offering ranges from the same monitors, not closed deals. Deal-level and state-level pricing is not public, so we do not estimate it.
Calculators
Each tool shows its formula and sources on its page. Tool results are estimates for planning, not tax or legal advice. The tool math is checked against worked examples before each release.
Downloads
The CSV and JSON downloads are built from the same files as the pages, on every build:
| Dataset | Rows | As of |
|---|---|---|
| HUD FY2026 MTSP income limits (20% to 80% AMI, 1 to 8 persons) and maximum gross rents, per county or New England town | 4,764 | FY2026 |
| HUD 2027 Qualified Census Tracts and Difficult Development Areas (Small DDA ZCTAs, nonmetropolitan DDA counties) | 17,482 | 2027 |
| HUD LIHTC database properties still monitored, in cities with five or more (the properties on this site's city pages) | 32,252 | LIHTCPUB 2024 (revised 2026-05-28) |
From the downloads manifest, /data/downloads/manifest.json.
Refresh schedule
HUD publishes new MTSP limits each spring, new QCT and DDA lists each fall and a new LIHTCPUB release about once a year; the site is rebuilt from each new release, and the update log below records when. The IRS rate table adds each monthly revenue ruling after it is published. State agency and QAP links are rechecked when the site is rebuilt.
Update log
| Date | What changed |
|---|---|
| 2026-10-30 | Full rebuild of the site. Data files regenerated from the saved HUD files: 4,764 MTSP areas, 14,539 2027 QCTs, 2,657 Small DDA ZCTAs, 286 nonmetro DDA counties, 32,252 properties on city pages. New state, market, tool and guide pages; this methodology page added. |
| 2026-10-30 | IRS applicable percentages table checked through November 2026 (Rev. Rul. 2026-21); equity pricing medians checked against the CohnReznick monitors through September and October 2026. |
| 2026-10-30 | State agency, QAP and state credit links checked for all 56 jurisdictions. |
| 2026-10-24 | HUD files saved from huduser.gov: FY2026 MTSP limits (files dated 2026-03-27), the 2027 QCT and DDA lists (FR Doc. 2026-20390, October 6, 2026), and LIHTCPUB (revised 2026-05-28). Site launched on these files. |
Corrections policy
If you find an error, email hello@lihtc.loans with the page address and, if you can, the source that shows the right figure. We check every report against the primary source. When a page is wrong we fix it, change that page's "updated" date (and only that page's), and add the fix to the log below. When a report turns out to be a difference in source or method, we say so on the page if it would help other readers.
Corrections log
| Date | Page | Correction |
|---|---|---|
| 2026-10-30 | 25% bond test guide, 4% vs 9% guide, tax-exempt bond guide, construction loan guide, law changes page, bond test and capital stack tools, state pages | Some pages said a single post-2025 bond must finance 5% of aggregate basis by itself. The statute (IRC 42(h)(4)(B)(ii), added by Pub. L. 119-21 sec. 70422(b)) lets one or more post-2025 bonds count together toward the 5%. Every page now uses the same wording, and says the 50% test still applies as an alternative. |
| 2026-10-30 | Bond test tool | The worked example said a $9,000,000 issue frees $8,250,000 of volume cap. Against the $16,500,000 the 50% test needs, it frees $7,500,000 ($8,250,000 only if the bonds are sized at the 25% minimum). |
| 2026-10-30 | Qualified contract tool and Year 15 guide | The cost-of-living cap on adjusted investor equity was described as ignoring any year above 5%. IRC 42(h)(6)(G)(ii) counts such a year at 5%. |
| 2026-10-30 | Permanent loans guide | USDA Section 538 loan-to-value now shows 97% for nonprofit and public borrowers and 90% for others (7 CFR 3565.204), with a 25 to 40 year term (3565.209). |
| 2026-10-30 | Permanent loan, construction loan and RAD guides | The 25 bp FHA mortgage insurance premium now says it applies to applications submitted or amended on or after October 1, 2025 that were not yet initially endorsed (90 FR 45789). |
| 2026-10-30 | Law changes page | The CDBG new-construction change applies to amounts appropriated after July 11, 2026, not on enactment (Pub. L. 119-101 sec. 204). |
| 2026-10-30 | Equity pricing page | Fund IRR range corrected to 4.75% to 10.15% (October 2026 monitor); the $0.823 median is credited to both the September and October 2026 monitors, which both report it. |
| 2026-10-30 | Recapture guide | Removed a link to a recapture estimator that does not exist. |
| 2026-10-30 | Wyoming property records and the properties download | One HUD address field carried a management company email. Contact details are now stripped from every published property field, and the build fails if any remain. |
| 2026-10-30 | Property records | Dropped HUD test and demo records (for example "Test Property" in Austin, TX and "Inspec-Demo" in Jackson, MS) that appeared as real properties. |
| 2026-10-30 | Several guides and tool pages | Replaced dead or moved source links (the House law revision counsel code viewer, an IRS LIHTC page, HUD MAP Guide, HUD utility model, eCFR 24 CFR 100 Subpart E, a California copy of Rev. Proc. 94-57) with working primary sources. |
See also: data sources and dates and about LIHTC Atlas.