Who actually owns the US national debt?

How to read this

The stacked area shows who holds the US national debt, bucket by bucket, every month from Invalid Date to today. Toggle between “Share” (each band as a percent of the total) and “Absolute” ($ Trillions) — the first answers composition, the second answers scale. Toggle to the “By Country” view for the F1-style bar race of the top dozen foreign holders. Drag the scrubber to walk through history, hover to read exact values, click any annotation pin to jump.

What this is showing right now

Why the picture has changed

For most of the post-2000 era, foreign central banks were the marginal buyer of US debt. China climbed to a peak of $1.317T in November 2013, then began a long, slow retreat that continues today — much of it diversification into gold and other reserves, some of it reallocation through Belgium and the Caribbean financial centres. Japan briefly retook the number-one spot in 2019 and has held it since.

The Federal Reserve’s SOMA portfolio is the second biggest holder. It exploded in March 2020 (COVID QE doubled the stack), peaked in mid-2022 at roughly $5.7T of Treasuries, and has been gradually drawn down by the QT program since.

Watch the five domestic bands — money market funds, US banks, households, pensions, and state & local governments. As foreign appetite has flattened and the Fed has shrunk, those domestic sectors have had to absorb the new issuance. Money market funds in particular ramped sharply once the Fed’s reverse-repo facility began draining in 2023, and households (a Z.1 catch-all that also captures the basis-trade hedge funds nominally classified there) have steadily climbed. That domestic absorption is the single most important structural shift in the modern Treasury market.

How the domestic split is built

The height of the stack is anchored to GFDEBTN — the Treasury’s published figure for total federal debt outstanding — so the bands always add up to the official number. Foreign holdings (TIC) and the Federal Reserve’s SOMA portfolio sit underneath as direct, source-reported values, and what’s left over is the domestic residual: everyone else who owns a Treasury. We split that residual across the five Z.1 sectors (money market funds, US banks, households, pensions, and state & local government) in proportion to each sector’s reported holdings in the Z.1 Flow of Funds, so the shape of the domestic stack tracks Z.1 while the total stays nailed to Treasury accounting. The chart timeline starts at TIC’s first month (Invalid Date) so the early china-vs-japan bar race is visible, but the Z.1 sector series typically don’t have a first observation until 2003 or later. For those pre-Z.1 months the domestic residual is held in a single neutral “Other Domestic” band rather than being split — the underlying sector breakdown simply isn’t known yet, and dumping the entire residual into Households would have falsely attributed several trillion dollars to a sector Z.1 had never measured. We don’t plot Z.1 directly because Z.1 reports many more sectors than we render — mutual funds, ETFs, insurance companies, GSEs and broker-dealers all hold Treasuries too — and because the two releases follow slightly different timetables; that combined gap is what the residual split absorbs (mostly into the Households & other band, which Z.1 itself uses as a catch-all). As of Jan 1970 the reconciliation gap between the chart’s five domestic bands and the raw Z.1 sum was +$0B (about 0.0% of total debt) — visible per band as the dim subline in the chart tooltip.

Sources & caveats

• Federal Reserve (SOMA Treasuries) — FRED WSHOTSL, weekly H.4.1 Wednesday level, resampled to last week-of-month. SOMA’s Treasury leg only — the MBS leg ( WSHOMCB) and agency debt are excluded so the band shows the Fed’s Treasury book against GFDEBTN cleanly.
• Foreign Official + Foreign Private + per-country — Treasury’s TIC Major Foreign Holders tables, monthly with a ~60-day publication lag.
• Total federal debt — FRED GFDEBTN, quarterly, forward-filled to monthly grid.
• Domestic split via Z.1 Flow of Funds. Money market funds, US banks, households, pensions, and state & local governments come from the Federal Reserve’s Z.1 release (Table L.210, Treasury Securities by Holder Sector — quarterly, forward-filled to monthly). Each band is sized as that sector’s share of the historical “total domestic ex-Fed ex-foreign” residual, so the bands continue to sum to GFDEBTN. Pensions combine private pension funds, federal government retirement, and state & local government employee DB plans.
• Caveat — Households is a Z.1 catch-all. The Z.1 households sector absorbs unclassified holdings as a residual, which historically includes mutual funds, ETFs, insurance companies, GSEs, broker-dealers, and large basis-trade hedge funds. Treat the “Households & other” band as the broader domestic non-bank / non-pension chunk rather than literal household savers.
• Caveat — pre-Z.1 “Other Domestic” band. The chart starts at TIC’s first month (Invalid Date), but the Z.1 sector series typically don’t have an observation until 2003 or later. For those early months we hold the entire domestic residual in a single neutral grey “Other Domestic” band at the top of the stack — the per-sector breakdown simply isn’t known yet. Once any Z.1 sector has data the band drops to zero and the five Z.1-derived bands take over.
• Caveat — Belgium & Cayman Islands. A large fraction of holdings reported under Belgium and Cayman are not local — Belgium is the home of Euroclear, the world’s biggest custodian, and Cayman hosts the legal shells of large basis-trade hedge funds. The country leaderboard is a custody snapshot, not a beneficial-ownership one.

Related charts

• The Firehose — where every Fed-created dollar ends up.
• Eurodollar Black Box — offshore-USD claims and funding stress.
• Net Liquidity Stack — Fed BS minus TGA minus RRP, vs the S&P 500.