How Stablecoin Demand Is Changing U.S. Treasury Bills

Buying a stablecoin does more than create a digital unit displaying $1 on a screen. The issuer receives an actual dollar and places it in a reserve asset. That reserve is often a short-dated U.S. Treasury bill, a very short-term cash transaction backed by Treasuries, or a bank deposit.

Demand for a payment instrument inside the crypto economy can therefore become a purchase in a traditional money market. The most important innovation is not only the token. It is the balance-sheet bridge that can connect demand for digital dollars anywhere in the world to the U.S. government's short-term borrowing cost.

The short answer

Stablecoin issuers are not large enough to direct the entire U.S. Treasury market. A February 2026 presentation prepared for the U.S. Treasury said they held less than 1% of Treasuries outstanding. Their reserves, however, are concentrated in one- and three-month instruments. The sector can therefore remain small in the overall market while acting as a measurable marginal buyer of three-month bills.

A BIS paper revised in June 2026 found a temporary decline of several basis points in three-month bill yields during the weeks after stablecoin inflows. The estimated move became materially larger when Treasury-market intermediation was under strain and did not carry into longer-dated bonds.

This does not mean that stablecoin growth automatically lowers every interest rate. The net effect depends on where the funding came from, which asset receives the reserve, how many bills the Treasury supplies, and how much trading the market can absorb at that moment. The same channel can reverse during redemptions. Safe reserve assets reduce credit risk, but they do not eliminate the risk of delayed access to cash or forced transactions.

Start with plain-language definitions

A stablecoin is not the same thing as a bank deposit. It is not a Treasury bill or a direct liability of the central bank either. The holder has a digital claim on an issuer and its redemption arrangement. In its Q2 2026 filing, Circle explicitly said its obligations to stablecoin holders were not covered by U.S. deposit insurance.

The distinction matters. A safe asset inside the reserve does not mean that the token holder owns that asset directly or can necessarily turn the token into cash at the same instant under every condition.

How does the $1 behind a token move?

The simplest example has four steps.

1. A customer sends the issuer $100.
2. The issuer gives the customer stablecoins worth $100 in total.
3. The issuer records two balance-sheet entries: a $100 redemption liability to the customer and a $100 reserve asset backing it.
4. When the customer returns the tokens, the issuer pays $100, cancels the tokens and reduces the reserve by the same amount.

The third step matters for markets. The issuer can leave the money at a bank, buy a three-month Treasury bill, enter a reverse repo backed by Treasuries, or place the cash in a government money market fund that owns those instruments.

The ordinary inflow looks like this:

New dollar -> stablecoin issuer -> reserve account -> Treasury bill, repo, money market fund or bank deposit

Redemption reverses the path:

Returned stablecoin -> dollar payment to customer -> cash use, receipt from a maturing bill, or sale of a reserve asset

Routine redemptions may be easy to meet when the issuer holds enough cash and bills mature at regular intervals. The timing changes if many holders want to leave at once. The issuer can lose the freedom to wait for a convenient date and may need to unwind repo or sell an asset to obtain cash.

Why issuers prefer short-dated Treasury bills

Three reasons stand out.

First, credit risk is low. A Treasury bill is an obligation of the U.S. federal government and trades in a deep market. Second, its maturity is short. An instrument that repays in three months is less sensitive to changing interest rates than a long-term bond. Third, it produces income. A banknote or idle cash does not pay interest, while a bill or repo does.

That income is central to the stablecoin business model. Circle reported $667.7 million of reserve income and $701.3 million of total revenue and reserve income in Q2 2026. Reserve income accounted for 95.2% of that combined top-line item. This is not a profit margin; Circle also made substantial payments to distribution and transaction partners. The underlying economics are still clear: the user gets stable value and payment utility, while the issuer and its distribution network capture most of the reserve yield.

This creates an opportunity cost for the holder. If a three-month bill yields 4% and the token itself pays no direct interest, the user gives up that return in exchange for liquidity and utility. An exchange or another intermediary may offer a reward, but that is not native interest on the token and introduces third-party risk.

The U.S. GENIUS Act of 2025 establishes at least 1-to-1 identifiable reserves for permitted payment stablecoins. Permitted assets include cash, demand deposits, Treasury securities with no more than 93 days to maturity, specified repo transactions, and government money market funds invested only in eligible assets. The law also calls for monthly reserve-composition reporting and public redemption policies.

The statute and full implementation are not at the same stage. On August 19, 2026, the Office of the Comptroller of the Currency said it was targeting November for its final rule. The new framework can therefore be expected to strengthen demand for short-dated public assets if adoption grows, but it would be inaccurate to assume that every operational detail is already final and in force.

Why more buying lowers bill yields

A Treasury bill pays a fixed amount at maturity. An investor who pays a higher price today for the same future payment accepts a lower return. That is why a bill's price and yield move in opposite directions.

The sequence is simple:

• More buyers enter the market.
• The bill's price rises.
• Its repayment at maturity does not change.
• The gap between today's higher price and the fixed future payment narrows.
• The annualized yield falls.

The moves can look small. Four basis points is only 0.04 percentage point. Yet money markets roll very large sums over short periods, so a few basis points can change which instruments funds, banks, corporate treasurers and arbitrageurs prefer.

Two mistakes should be avoided. Four basis points does not mean a 4% return. It is also inappropriate to turn the BIS estimate into a fixed effect for every $1 billion of inflow. The paper explicitly warns that the relationship may not stay linear as the sector grows and market conditions change.

What does the evidence actually show?

The BIS study uses daily data from January 2021 through March 2026. It compares five-day changes across seven large dollar stablecoins with changes in the three-month Treasury bill yield. It does not rely only on two lines moving together. The analysis attempts to control for crypto prices, money market fund flows, short-term Treasury fund flows, bill supply, monetary-policy surprises, the dollar, equity markets and Treasury-market liquidity.

That separation is necessary. Higher rates can reduce stablecoin demand at the same time that stablecoin outflows can affect rates. Looking at correlation alone cannot identify the direction of the arrow. In fact, the paper's uncontrolled estimate is implausibly large, close to the effect of a central-bank rate cut. The researchers address this problem with an instrument built from movements that are specific to individual token and blockchain pairs.

The preferred estimates are:

• A $3.5 billion inflow over five trading days lowers the three-month bill yield by 0.71 basis point on impact.
• The effect reaches about 4 basis points within 10 days.
• The deepest point is close to 5 basis points around day 13.
• The response stays near 3 to 4 basis points from day 10 through day 25 and then fades.
• A meaningful response also appears in the one-month bill.
• The six-month, two-year and 10-year yields show limited or no response.

This maturity pattern supports the proposed mechanism. Finding the result only where issuers tend to invest is more consistent with direct reserve demand than with a general macroeconomic shock. It is still not perfect causal proof. The authors explicitly list the short data history, incomplete maturity reporting, especially for Tether, and possible remaining model error as limitations.

Small in the overall market, influential at the short end

Two facts can look contradictory:

• Stablecoin issuers hold less than 1% of all U.S. Treasuries.
• A several-basis-point effect can still be measured in three-month bills.

There is no contradiction because the denominators differ. The first measure covers the enormous market for bills, two-year notes, 10-year bonds and longer maturities. The second measures a new flow in the narrow part of the curve where issuers concentrate their reserves.

In the Treasury advisory presentation, T-bills were 53% of Tether and Circle assets and their bill holdings had risen by $70 billion since 2022. BIS estimated that the sector had $153 billion of direct and indirect T-bill exposure at end-2025 and bought $33 billion during that year. The dates and methods differ, but both sources point to the same structure: limited share of the whole market, concentrated demand at the short end.

Trading capacity also matters. On an ordinary day, other investors may refuse a lower yield, sell their bills or move to another instrument. Their response contains the price move. When intermediary balance sheets are strained or Treasury liquidity is poor, fewer flexible sellers are available. The BIS estimate of an 8 to 10 basis point effect in stress states is consistent with that mechanism.

Strongest counterargument: is the money actually new?

Stablecoin market growth and new Treasury demand are not the same number. The net effect cannot be calculated without knowing what the token buyer used to fund the purchase.

The Federal Reserve's March 2026 analysis stresses this point. The asset held by the issuer matters, but so does the next decision made by the investor displaced from the bill. When a bill moves to an issuer, the previous owner receives cash and can put it back into a deposit, fund or another bill.

This counterargument does not erase the first-round purchase. It defines the boundary of the conclusion. The issuer's transaction is real, but the amount of additional system-wide demand remains unknown without the full source-of-funds chain.

Circle: transparent fund structure and reserve income

Circle's status as a public company makes it possible to follow the USDC reserve through detailed corporate filings. USDC in circulation was $73.3 billion on June 30, 2026. Circle reported the following composition:

• $61.9 billion in the Circle Reserve Fund, a government money market fund;
• $11.4 billion as cash across several banks.

The calculation puts 84.4% of the disclosed composition in the fund and 15.6% in bank cash. BlackRock manages the fund, which holds Treasuries maturing within three months, overnight reverse repo backed by Treasuries, and cash. It would therefore be wrong to count the entire $61.9 billion as direct Treasury bills.

The structure reveals two separate forms of risk. Short-dated government instruments limit credit and maturity risk. At the same time, the reserve depends on a fund manager, custodian, banks and payment infrastructure. Circle also says its reserve cash greatly exceeds deposit-insurance limits and that its obligation to the stablecoin holder is not an insured deposit.

Reserve income makes the interest-rate sensitivity visible. More USDC in circulation creates a larger interest-earning reserve. Lower short-term rates reduce the return on the same reserve. These are two basic drivers of Circle's revenue. But higher rates can also reduce USDC demand by increasing the return that holders give up. Circle itself says the relationship between rates and USDC circulation is complex and unproven.

Tether: greater scale, different reserve mix

Tether's reserve report for June 30, 2026 showed approximately $184.6 billion in gross contractual token redemption value. Its principal disclosed asset categories included:

• $115.0 billion of direct U.S. Treasury bills;
• $18.6 billion of overnight reverse repo;
• $7.0 billion of term reverse repo;
• $140.6 billion in the cash-equivalent and other short-term subtotal;
• additional holdings of gold, Bitcoin, public equities, other investments and secured loans.

Direct Treasury bills represented 61.2% of assets, while the short-term subtotal represented 74.9%. Total assets exceeded liabilities by $4.1 billion, equal to 2.2% of reported liabilities.

These ratios are informative, but they are not a stress test. The Q2 document is a point-in-time reserve report on which BDO provided reasonable assurance. It is not a full set of financial statements, its valuations assume normal market conditions, and BDO did not provide assurance on management's going-concern assessment. On August 13, Tether separately announced that KPMG had issued an unqualified opinion on its financial statements for 2025. That strengthens the historical reporting context but does not change the point-in-time nature of the June 30, 2026 reserve mix.

Adding Circle and Tether's same-date scale produces about $257.9 billion. The sum shows why the two issuers matter. It does not make their reserve quality or disclosure equally detailed, and it does not support a universal assumption that one token equals one directly held Treasury bill.

Rates also move stablecoins

The arrow does not run only from stablecoins to bills. Short-term interest rates also affect stablecoin demand.

When rates rise, a money market fund or Treasury bill earns more. The opportunity cost of holding a non-interest-bearing token increases. At the same time, tighter monetary policy can weaken crypto prices and trading appetite. If the token is used less, circulation can contract, the issuer can reduce its reserve, and bill demand can weaken.

A separate BIS study using weekly data from January 2019 through July 2024 found that prime money market funds grew after contractionary U.S. monetary-policy shocks while stablecoin market capitalisation fell. It also found no consistent aggregate safe-haven inflow into stablecoins during major crypto shocks.

The older sample does not prove that future behavior cannot change. The 2025 law, payment use and offshore dollar demand may transform the market. The finding still supplies an important warning: stablecoin demand should not be treated as safe-haven dollar demand that automatically rises during every form of stress.

Why safe reserves do not make a token risk-free

At least five questions determine a stablecoin's resilience:

1. Asset quality: How likely is the reserve to repay?
2. Liquidity: How quickly can the asset become cash without a large loss?
3. Access: Are the bank, fund, custodian and payment rails operating when needed?
4. Conversion rights: Who can redeem directly for $1, at what time, with which fee and minimum?
5. Network capacity: Do transactions become slower or more expensive when the digital network is busy?

A short-dated Treasury bill can perform well on the first two questions. It cannot solve the last three by itself.

A June 2026 Federal Reserve theoretical paper illustrates the distinction in another way. Digital money becomes more useful as more people adopt it, but network congestion can raise transaction fees and waiting costs. If one holder expects others to rush for the exit, that person may also redeem to avoid being late. In the model, this behavior can become self-reinforcing even when the reserves are assumed to be perfectly safe.

This is a possible mechanism, not a forecast. Actual outcomes depend on network design, fees, bank operating hours, market-maker liquidity and the issuer's redemption capacity.

What the 2023 USDC episode taught

When Silicon Valley Bank failed in March 2023, Circle disclosed that it could not access $3.3 billion of USDC reserves. The sum represented about 8% of reserves at the time. The problem was not a loss on Treasury bills; it was loss of access to bank deposits.

Primary conversion, in which an authorized customer returns the token directly to Circle for $1, was constrained over the weekend by banking hours. Holders who wanted to sell turned to exchanges and decentralized markets. USDC briefly traded as low as $0.86. The price returned to $1 after U.S. authorities announced protection for all SVB depositors and Circle resumed redemptions.

The event provides three lessons.

• High-quality reserves can still trigger a confidence problem if part of them becomes inaccessible.
• Closing the primary conversion rail does not close the secondary market; the price can fall and display the pressure.
• Stress in one stablecoin can spread to other digital currencies that use it as collateral or as a conversion asset.

No forced Treasury sale occurred in this episode. The Federal Reserve analysis discusses such a sale as a counterfactual that could have become necessary if pressure had continued after bank cash was depleted. The observed event and the possible extension should remain separate.

What can happen to Treasuries during a redemption wave?

During ordinary growth, an issuer can spread purchases over time, wait for a bill auction, stagger maturities or hold cash temporarily. A redemption wave reduces that flexibility.

Cash is the first line of defense. Bills that mature very soon are the second. A third option is raising short-term cash against bills as collateral. If those sources are insufficient, asset sales become more likely.

The current BIS paper says the inflow estimate, which comes from a period of broad market growth, is likely a lower bound for outflow effects. The reasoning is simple: a purchase can be delayed, while a redemption often cannot. A wave of sales can also coincide with already poor Treasury liquidity, making price impact nonlinear.

The paper does not justify a fixed multiplier. The current revision does not establish that an outflow is always two or three times as powerful as an inflow. The defensible statement is narrower: a fast redemption wave removes timing discretion, and the same-sized transaction can have a larger price effect when liquidity is weak.

What regulation needs to solve

Saying only that reserves should be safe is not enough. A good framework addresses four layers together:

• reserve credit quality and maturity;
• cash that is genuinely available on demand;
• the issuer's own capital that can absorb losses;
• continued operation of banks, custodians, networks and redemption infrastructure.

A June 2026 BIS model finds that usable capital and liquidity buffers can reduce both losses to holders and spillovers from forced sales. But a rigid rule that never allows a buffer to be used can force premature asset sales in a crisis and make the problem worse. How a buffer works under stress matters as much as its size in normal conditions.

The GENIUS Act creates a strong starting point for reserves, disclosure and redemption. The real test will be implementation: how often and at what level of detail assets are reported, how concentrated bank exposure is, how quickly different customers receive cash, and whether liquidity buffers can actually be used during a run.

Eight indicators to monitor

No single indicator is sufficient. Market value can rise while bill supply expands at the same pace, containing the yield effect. The token can remain at $1 while a direct redemption queue grows. The portfolio can look entirely short-dated while access depends on one bank or custodian.

What would weaken the thesis?

The Guide's conclusion would weaken materially if:

• complete balance-sheet data showed that most inflows came from existing Treasury and government money market fund investors;
• longer datasets failed to reproduce the BIS three-month bill result across different rate and liquidity regimes;
• issuers shifted reserves mainly to central-bank balances or bank deposits;
• greater bill supply and more elastic arbitrage absorbed stablecoin flows without moving prices;
• final rules, round-the-clock conversion and tested buffers materially reduced forced-sale risk;
• stablecoins failed to create durable payment and savings demand beyond crypto trading.

The short-end connection would strengthen if genuinely new offshore dollar demand gained share, standardised reserve reporting spread across issuers and stress episodes produced observable bill sales.

Conclusion

Reading stablecoins only through crypto prices is now incomplete. A dollar-backed token is also a private money-market balance sheet. It gives the user a digital payment instrument while producing demand for Treasury bills, repo, banks and money market funds behind the scenes.

The narrow conclusion supported by current evidence is that stablecoin issuers do not determine the full U.S. Treasury market, but they have become a measurable and growing buyer of one- and three-month instruments. The effect depends less on the headline size of the sector than on the reserve mix, source of the money, bill supply and market liquidity.

For the same reason, safe reserves do not make a stablecoin risk-free. Strong credit quality can coexist with bank-access problems, conversion delays, network congestion, custody dependence and self-reinforcing redemptions. The bridge that buys bills during growth can transmit selling pressure in the opposite direction during stress.

The most useful question is not simply, "Will stablecoins lower rates?" It is: Where did the money financing the new token come from, which reserve asset received it, and how quickly can that asset become cash if everyone asks for redemption at once?

Methodology and limitations

The main price estimate comes from a BIS paper using daily data from January 2021 through March 2026. The study controls for common macroeconomic and market factors and uses an instrument designed to isolate stablecoin-specific flows. The history is still short, issuer maturity disclosures are uneven, and remaining causality problems cannot be fully ruled out.

Circle and Tether figures come from company documents dated June 30, 2026. The cases demonstrate scale and differences in reserve composition; they do not support one portfolio assumption for the whole industry. The Federal Reserve's roughly $320 billion market estimate reflects earlier conditions in April 2026 and is therefore not combined with June issuer figures to calculate market share.

The Guide does not forecast sector size, Treasury yields or redemption losses. It does not extend the BIS inflow estimate linearly into trillion-dollar scenarios or use a numerical outflow multiplier. The regulation section reflects official texts and statements available through August 23, 2026; final implementation rules may change.

Sources and further reading

• BIS, Stablecoins and safe asset prices, June 2026 revision
• U.S. Treasury, Treasury Borrowing Advisory Committee investor-demand presentation, February 2026
• Federal Reserve, Financial Stability Report, May 2026
• Federal Reserve, Payment Stablecoins and Cross Border Payments, March 2026
• Public Law 119-27, the GENIUS Act
• OCC, GENIUS Act implementation next steps, August 19, 2026
• Circle, Q2 2026 Form 10-Q
• Tether, June 30, 2026 reserves report and BDO assurance opinion
• Tether, announcement of KPMG's audit of 2025 financial statements
• Federal Reserve, In the Shadow of Bank Runs, December 2025
• Federal Reserve, The Fragility of Perfectly Safe Digital Money, June 2026
• BIS, Stablecoins, money market funds and monetary policy
• BIS, Making stablecoins stable(r): can regulation help?, June 2026

Not investment advice; for research and educational purposes.