Conforming Loan Limits by Year History: Where the Tables Live

One federal publisher owns the series. The FHFA Data Portal — House Price Index, Conforming Limits & Mortgage Market Data carries a Conforming Loan Limit section that publishes county-level maximum loan values for each calendar year, reaching back to the pre-HERA era. Files ship as small per-year tables in XLSX, CSV and PDF, next to an interactive loan-limits map and a loan-limits dashboard that sit alongside the portal's HPI, Duty to Serve, Borrower Assistance and UAD dashboards.

The granularity is deliberately narrow: annual county loan-limit values, one value per county per year. Where the same portal's House Price Index master file consolidates roughly 184,000 index rows into about 17 MB, the limit lists rank among the smallest files it ships. Within Datadory's real estate operating companies slice — 22 pooled datasets, 9 of them primary — this portal record scores 9 out of 10 against a catalog-wide average quality score of 7.81.

The limits are the longest-running product on the portal. FHFA lists conforming loan limits "from the early 1970s by year" in its temporal coverage, while every neighbor on the same domain starts later: the House Price Index begins January 1975, National Mortgage Database aggregates begin in 1998, and UAD Appraisal Aggregate Statistics begin in 2013Q1.

That ordering matters when you scope a study. A purchasing-power series built on the limit tables can span five decades; the appraisal-based and database-backed series cannot follow it past their own start dates.

Each calendar year's file answers a single question: what is the maximum loan size eligible for purchase by Fannie Mae and Freddie Mac on a property in this county this year? The values are annual county loan-limit figures — not metro-wide approximations and not a single national ceiling — so the same year carries different ceilings in different counties. Loans above the county's number fall outside the conforming box; loans at or below it stay eligible for enterprise purchase, which is why origination teams re-check the tables every year rather than hard-coding last year's answer.

The glossary entry on conforming loan limits keeps the definition next to its evaluation checklist: demand county resolution over national baselines, verify the year span reaches the early 1970s, confirm new-year files appear promptly after announcement, check that the enterprise linkage is explicit — FHFA tags the data "fannie mae" and "freddie mac" — and pick from the portal's eight formats the one your pipeline ingests without reshaping.

Vintage discipline has a second edge at the older end of the history. Because the series runs by year back to the early 1970s, a study of 1990s originations can pull 1990s thresholds instead of borrowing modern ones — provided the panel keys each origination to its own vintage before matching geography, so no loan is measured against a threshold published after it closed.

CFPB Home Mortgage Disclosure Act (HMDA) Data classifies loans for you. Every reported application carries a conforming_loan_limit field beside county_code (the state-count FIPS code) and census tract, spanning filings from 2007 through the 2025 filing year released March 31, 2026. Nationwide first-lien owner-occupied records ran roughly 4.8 million to 8.3 million rows per year between 2007 and 2017, and the 2023 dynamic LAR zip alone compresses to about 631 MB across roughly 100 fields.

FHFA National Mortgage Database (NMDB) Aggregate Statistics adds market-level context from a nationally representative 5 percent sample: new-mortgage statistics annually 1998-2024, quarterly 1998Q1-2025Q3 and monthly January 1998-September 2025, with outstanding-balance tables through 2026Q1 and performance tables from 2002Q1 as of the June 26, 2026 release.

The limits even shape the portal's own indexes: FHFA's expanded-data House Price Index flavor folds in enterprise, FHA and recorder-deed transactions below the conforming loan limit, so the threshold defines which observations enter which index.

Classify loans against the right year. From 2007 forward, HMDA's conforming_loan_limit flag does the bucketing for you; for 1999-2006 vintages, apply the corresponding year's county threshold to original unpaid principal balance in the Freddie Mac tape. Then convert thresholds into purchasing power by indexing them to prices: divide the county-year limit by the FHFA House Price Index value for the same geography, remembering the index carries a 1990=100 base convention and reaches census-tract level in its annual all-transactions series. Finally, sanity-check your conforming share against NMDB aggregates, whose statistics are weighted to represent the full closed-end first-lien residential mortgage universe rather than a vendor subsample.

For the wider source landscape around these tables, read the real-estate-operating-companies data guide, and browse the real estate operating companies data hub for the full dataset roster. To see how the price side differs from the volume side, the side-by-side comparison of Census New Residential Construction and the FHFA Data Portal lays both records out attribute by attribute. The best real-estate-operating-companies datasets and free real-estate-operating-companies datasets rankings place the loan-limit source among its peers.