We are living in a truly fiscally profligate era. From 2022-2025, Treasuries held outside the U.S. government but counting the Fed (Z.1 Table F.3.2) increased from $21.5 trillion to $28.5 trillion, absent any major recession or crisis.

So the question is, who has been choking down all of this issuance? The results may surprise you, but the tl;dr is that the marginal buyer is now financed with runnable liabilities. Here’s a chart, from the Fed’s Z.1 data:

Relative to long-run averages (since 1945), four sectors are holding significantly more Treasuries and three are holding less. These trends largely reflect structural changes in U.S. capital and money markets over the last 80 years, as well as shifts in the balance of trade.

The larger holders are foreign accounts, households, money market funds and standard mutual funds/ETFs. Banks, the Fed, and insurance companies & pensions (real money) have reduced holdings.

The Reducers

Bank holdings have been steadily falling, from a peak of 50% of outstanding in 1945 to 7% today. Commercial banks’ position in the Treasury market in 1945 is largely a story of war debt being financed by depositors, and the decline from 50% is just the normalization of a wartime economy to a peacetime economy. I should add that banks’ exposure to government liabilities is higher than their Treasury holdings, as they also hold central bank reserves as settlement money.

The Fed only holds 13.5% of outstanding Treasuries, below its long-run average of 17%. When issuance jumped during COVID to pay for various stimulus programs, Fed QE exceeded its prior pro-rata share of government issuance, and holdings jumped to 26% of issuance in 2021. But since 2023, when outstanding Treasuries have increased by +$7T, the Fed has been shrinking its balance sheet. When we add Fed runoff to the central government issuance figure, the amount of Treasuries absorbed by non-governmental sectors over this time period increases to ~$8.3 trillion.

Insurance companies & pensions’ holdings have declined from 15.6% in 1945 to 6% today. Although their share of cash Treasury holdings has fallen, the real money/ALM sector continues to have very large exposures to the Treasury market through long positions in futures, likely due to capital efficiency (which is a form of leverage). These investors are providing the structural longs in Treasury futures that allows hedge funds to arbitrage with the cash-futures basis trade.

The Gainers

Foreign accounts are now the largest holders of Treasuries, at 32.5%. They started at 1.2% in 1945 and peaked at just over 55% in 2008. Foreign accounts are still adding to their stock of Treasuries, but they are not keeping pace with new issuance.

Foreign holders first became a significant factor in 1971, jumping from 8.3% to 17.7% in a single year. The spike there coincides with the collapse in Bretton Woods, as foreign official accounts bought dollars to defend their currency pegs.

Money market funds’ holdings climbed from 4.9% in 2022 to 12.3% today, or ~$1T to ~$3.5T. $2T was in bills. This represents a significant portion of the increase in bills outstanding over the same time period, from $3.5T to $6.5T. This increase in MMF holdings of bills was largely funded by a drawdown of the Fed’s O/N RRP from 2023 onwards (~$2.5T), which is now empty. It remains to be seen how MMFs will fund further large increases in bill issuance, now that the O/N RRP is empty.

Money market funds began investing in the Treasury market in 1975. Over their existence, they have held 3.9% of the Treasury market, on average.

“Households’” ownership has increased from 2022-2025, absorbing $1.4T. I put “households” in quotes, because the z1 data consolidates hedge fund holdings of any asset into the household line.

Given the size of the Cayman Islands hedge funds cash-futures basis trade in the rest-of-world holdings, I assume domestic funds have the same trade on and represent some of the increase in the household line. Looking at the flow data (F.3.2t), the impact is even more pronounced, as the household sector absorbed >$2.5T of net issuance from a flow perspective. This means that this sector has been buying duration into a rate selloff, which would make sense if the marginal buyer is hedged on the duration side (short Treasury futures) and only arbitraging the basis. Although I imagine there were some stressful moments at pod shops amid long-end repricing over the last few years.

Implications

The biggest conclusion that I draw from this analysis is that the marginal buyer of Treasuries in today’s market is financing them with runnable liabilities. In 1945, a large portion of Treasuries were financed by banks, who in turn financed their assets with FDIC-insured sticky bank deposits. The marginal buyers’ liabilities are now less sticky.

Money market funds issue NAV shares, which can break the buck through a bill fire sale or be gated if investors suddenly want to convert their shares into deposits. Hedge funds (domestic and foreign) are financing their Treasury holdings in the repo market, so any type of friction or haircut increase in the Treasury repo market could endanger these buyers’ presence in the market.

To be sure, there are still large pools of sticky Treasury investors. In a payments crisis the Fed is likely to expand its holdings by providing settlement money—bank reserves—and foreign real money accounts and central banks are somewhat price insensitive.

Regardless, the marginal buyer still finances Treasuries with runnable liabilities. What could cause a run? It’s impossible to know the exact trigger, but we do know the usual culprits: a run on repo, something that scares people about the bill market (debt ceiling?), or something of that nature. But the fragility is there.