Federal Reserve Mortgage Debt Outstanding (Financial Accounts Z.1)

It is the balance-sheet view of the American mortgage market: quarterly levels of total US mortgage debt, split by who holds it and what kind of property sits underneath, maintained by the Board of Governors of the Federal Reserve System inside the Financial Accounts of the United States (Z.1) - the flow-of-funds accounting framework outlined in flow of funds, explained. Filed under Commercial & Residential Mortgage Finance, the record reaches back to 1945, which turns the all-holders line into a single-chart history of the market: 14,336, in millions of dollars, at 1945 Q4 against 6,108,362 at 2026 Q1 - a roughly 426-fold rise from $14.3 billion to about $6.1 trillion.

Two layers ship together. The quarterly series carry the levels themselves, one series per holder-category-by-property-type combination. The crosswalk carries continuity: every line of the retired Mortgage Debt Outstanding (MDO) table - whose final release came in March 2020 - mapped to the Z.1 table and line where the series now lives, chiefly tables L.217 through L.221, alongside its FL-coded quarterly identifier. Roughly 100 mapped lines, each backed by a few hundred observations, mean any research citing the old MDO rows keeps resolving instead of dead-ending at a frozen publication. Get a sample of this dataset scoped to the holder groups and year windows you model.

One row per retired MDO line in the crosswalk, one row per quarter in the levels - flat enough to read without a parser. Rows below come straight from the August 2026 research pass:

# Crosswalk: retired MDO line -> Financial Accounts (Z.1) destination
MDO_LINE  MDO_DESCRIPTION                 Z1_TABLE_LINE     Z1_SERIES
1         All holders                     L.217 / Line 1    FL893065005.Q
2         One- to four-family residences  L.218 / Line 1    FL893065105.Q

# Quarterly level series: depository institutions, millions of dollars
OBSERVATION_DATE    SERIES_VALUE
1945-10-01          14336
2026-01-01          6108362

The first block is the bridge itself: legacy row 1 - all holders - lands at table L.217 line 1 under series FL893065005.Q, while row 2, one- to four-family residences, resolves to L.218. The second block is what the bridge buys you: the same measure observed eighty years apart, $14.3 billion compounding into roughly $6.1 trillion. Because the series code travels with every observation, joining the crosswalk to the levels needs nothing more than the code and the calendar quarter - no fuzzy matching, no reconstruction project. Live rows for whichever holder groups and windows you name arrive in this identical shape - request a sample and the same schema comes back cut to your list.

Four columns carry the entire crosswalk. MDO Line and MDO Description identify the retired row - its number and its label naming holder group and property type, such as All holders. Z1 Table / Line states the destination, the table-and-line address where the series lives today. Z1 Series carries the FL-coded quarterly identifier, suffixed .Q, that keys the levels themselves.

That is the whole dictionary, and the compactness is the point: identity in, address out, series code forward. Everything analytical - holder shares, property-type mixes, decade-over-decade growth - derives from combinations of these four once the levels attach. Definitions above were checked against the custodian's own documentation during the August 2026 research pass rather than guessed from headers.

Geography - United States national totals only. No state, metro or regional breakdown exists anywhere in the structure; geography enters by pairing this record with sets that carry it, such as CFPB HMDA application data or the state-level FHFA NMDB aggregate statistics.

Temporal - quarterly observations from 1945 through the most recent Z.1 release, verified end-to-end on the depository-institutions series. The legacy MDO side stopped moving years earlier: its final release came in March 2020, which is precisely why the crosswalk exists.

Granularity - one series per holder-category-by-property-type combination, levels stated in millions of dollars, roughly 100 mapped lines with a few hundred observations apiece. Small enough to reason about whole; long enough to span every housing cycle on record.

API, files, or your warehouse. Daily, weekly, or hourly.

Your cadence is decoupled from how the underlying releases roll out: most teams load the full history once - a few hundred observations per series - then keep a warehouse table current so each new quarter's rows diff cleanly against the last. Deliveries arrive normalised to the field dictionary above, keyed on mdo_line, z1_table_line and z1_series, in JSON, CSV, spreadsheet-ready Excel tables or XML. Versioning is standard: when a restatement or definitional note lands, superseded values are preserved rather than overwritten, so a holder-share series built last quarter reproduces exactly - and the diff between versions becomes analysable in its own right. A sample scoped to your holder categories ships first either way.

Six jobs the record settles outright:

• Credit-cycle and macro modelling - quarterly total-debt levels with eighty years of history supply the stock variable no origination-flow dataset offers; see quant backtesting.
• Holder-structure analysis - bank-versus-agency-pool-versus-insurer shares across decades quantify where credit intermediation migrated, the arithmetic underneath nonbank-lending debates.
• Whole-market denominators - the all-holders total bounds every lender-segment sizing with an official figure; see market sizing.
• Legacy research reconciliation - work citing the retired MDO table keeps resolving through the crosswalk, line by line, instead of stalling at a frozen release.
• Citable research and journalism - the authoritative long-run count of US mortgage debt, quotable with the worksheets behind it; see citation-grade research.
• Securitization-shift studies - the pools-and-trusts category read against depository holdings tracks the movement of mortgage risk off balance sheets and into securities.

The common thread: each job needs the whole market's stock with its holder structure intact, not a channel's share of it.

Investors and quant researchers (relevance 3/3) get a credit-cycle indicator with deeper history than anything else in the slice; see investors quants use cases. Journalists, academics and students (3/3) get citable official counts reaching to 1945; see journalists academics use cases. Market researchers and consultants (3/3) get the whole-market denominator under housing-finance sizings; see market researchers use cases. Data scientists and ML engineers (relevance 2/3) get four typed fields and stable series codes that join without a mapping project of their own; see data scientists use cases.

Persona fit has edges, stated plainly: teams needing borrower attributes, property geography or loan-level performance need other records - this one is deliberately aggregate-altitude, national and anonymous by construction. Pair it with Ginnie Mae Disclosure Data and Reports when the unit of analysis drops to the loan.

Within commercial and residential mortgage finance, this record owns the whole-market balance sheet by holder: the only set in the slice counting who holds US mortgage debt across every category, quarterly, back to 1945. The FHFA National Mortgage Database aggregate statistics own the borrower-side view - weighted origination, balance and performance tables from a nationally representative five percent sample - but reach only to 1998. Ginnie Mae Disclosure Data and Reports resolves individual loans and pools inside the government-insured book, and the altitude trade is worked through directly in Ginnie Mae Disclosure vs Federal Reserve Mortgage Debt Outstanding. The FHFA House Price Index prices the collateral beneath the debt, FRED Mortgage and Housing Data supplies the weekly rate environment around it, and the longer argument about what the holder split means sits in mortgage debt outstanding by holder type. If the question is how big is US mortgage debt and who holds it, this is the record that answers it.

Three things worth settling upfront. First, name your holder groups and windows: roughly 100 mapped lines across five holder families and four property types reward a scoping conversation, and a sample cut to your categories beats the full set every time. Second, respect the national altitude - no regional or borrower detail exists at this level, so plan the geographic join against a companion record rather than hoping for a column that was never there. Third, mind the 2023 seam: a December 19, 2023 note flags four series whose values absorb mortgages held in FDIC receivership following September 2023 reclassifications, so long-run holder-share constructions spanning the change should state their treatment explicitly. We walk through that seam with the sample rather than let a definitional shift ambush a model.