Stablecoins as Structural T-Bill Buyers: The Invisible Layer of Offshore Dollar Plumbing

Stablecoin issuers have quietly become structural buyers of US Treasury bills at the short end of the curve. At end-March 2026, Tether held $117 billion of US Treasury bills directly (BDO attestation for Q1 2026), and USDT and USDC together about $142 billion: more than a third of the $388 billion of bills held by all foreign official institutions combined, according to the Federal Reserve’s Z.1 Financial Accounts (end-2025). That demand has turned a market widely framed as speculative into an infrastructural layer of offshore dollar plumbing. The argument of this study: stablecoins are not a crypto asset. They are a new form of dollar-denominated shadow money whose growth presses on short-end T-bill yields, accelerates dollarization outside the traditional banking circuit, and introduces systemic vulnerabilities that the runs of May 2022 and March 2023 began to expose.

What this study covers when it speaks of stablecoins

The term stablecoin, in current crypto usage, refers to a digital asset whose unit price is pegged to a reference currency, most often the US dollar, by a dedicated mechanism. The family contains three subcategories with distinct properties: stablecoins backed by reserves of real fiat-denominated assets, stablecoins overcollateralized by other crypto-assets such as DAI, and so-called algorithmic stablecoins, where stability rests on an arbitrage system with a volatile companion token. The third category was effectively disqualified by the May 2022 collapse of TerraUSD. The present analysis focuses on the first, which carries the bulk of capitalization and all the macro-financial implications of the topic.

The observed perimeter centers on two issuers: Tether (USDT) and Circle (USDC), which together accounted for 83% of dollar stablecoin capitalization at end-March 2026 ($184 billion and $77 billion out of $313 billion, according to DefiLlama). Total capitalization was already above $300 billion in February 2026, against $27 billion at the start of 2021. The trajectory represents an elevenfold increase in five years, across a rate environment that has been successively zero, sharply rising, then stabilized at a historically elevated level. Growth did stall during the hiking cycle: capitalization fell from $187 billion in April 2022 to $123 billion in August 2023 (DefiLlama) and only regained its peak in November 2024. Elevated rates did erode the appeal of unremunerated dollar instruments for a while, without breaking the long-run trend.

The resilience reflects the use case for stablecoins, which is not reducible to crypto speculation. According to Chainalysis Geography of Cryptocurrency reports for 2024 and 2025, the bulk of on-chain USDT volumes corresponds to cross-border transfers — commercial settlements, informal remittances, hedging against domestic inflation in soft-currency jurisdictions. The stablecoin functions as a working digital dollar outside banking rails. The point is central to the macro-financial reading developed below: net stablecoin demand does not move mechanically with crypto risk appetite.

A complementary analytical layer is the digital dollarization layer approached from the perspective of decentralized infrastructure and protocol architecture. The present study departs from that view by anchoring its center of gravity outside the crypto system, in the consequences for global dollar plumbing. Reserve and ledger sit on two different layers, and it is the second that the infrastructure question around distributed ledgers addresses.

Reserve composition: what the T-bill layer actually holds

The macro-financial angle of the topic is won or lost in the reading of the reserves backing these tokens. Circle has published a monthly transparency report since November 2022 detailing USDC’s reserve composition line by line. The April 2026 version shows concentration on US Treasury bills with an average maturity of about thirty days, overnight repo collateralized exclusively by Treasuries, and a minority share of bank deposits held primarily at BNY Mellon. The granularity allows reconstruction of the reserve’s duration profile and its cash-to-securities ratio. Duration profile and cash-to-securities ratio are the very levers set by the composition and liquidity rules for stablecoin reserves.

The history of that publication is instructive. Before the move to line-by-line granularity in November 2022, Circle’s communication was more synthetic and did not isolate individual bank exposures. The concentration of USDC reserves at Silicon Valley Bank, revealed in March 2023 under stress conditions, accelerated the evolution toward finer disclosure. Since then, the monthly report names depository banks, the external manager (BlackRock for the T-bill portion), and weighted average maturity. The sector turned the episode into a de facto market standard.

Quarterly attestations published by Tether are less detailed. The Q1 2026 report, certified by BDO Italia, indicates similar concentration on 1-3 month T-bills but retains a residual “other exposures” line that has historically been more opaque, including in earlier editions bitcoin-collateralized loans and positions in selected corporate bonds. According to the same report, directly held US T-bills represented $117 billion of the Tether reserve at end of March 2026, out of $141 billion of direct and indirect exposure, which alone constitutes one of the largest individual positions on that market.

The convergence is documentable: the two principal issuers concentrate their reserves on the short end of the Treasury curve, with weighted average duration below two months. The concentration is not accidental. It reflects a simple economic arbitrage (short T-bills offer the best compromise between yield, near-instantaneous liquidity and negligible price risk) coupled with a regulatory constraint that the MiCA regulation in the European Union (applicable to stablecoins since June 30, 2024) and the GENIUS Act in the United States (signed into law on July 18, 2025) have formalized.

This convergence places stablecoin issuers structurally close to US government money market funds, which similarly invest exclusively in short Treasuries and Treasury-collateralized repo. The comparison is more than analogical: it carries regulatory and systemic implications that US authorities have made explicit since 2023. For the granular asset breakdown, maturity buckets and comparative market shares, see inside the T-bill backing mechanics, which decomposes Circle and Tether reports line by line.

The actual weight on the short Treasury bill market

The 1, 3 and 6 month T-bill segment is, by construction, one of the most liquid debt markets in the world. Total Treasury bill outstandings stood at $6,547 billion at end-2025 according to the Federal Reserve’s Z.1 Financial Accounts ($6,819 billion at end-March 2026). The distribution of holders is concentrated: at end-2025, again per Z.1, US money market funds held $2,687 billion of T-bills, depository institutions about $400 billion, and the rest of the world $1,448 billion, of which $388 billion for foreign official institutions.

Stablecoins, at about $142 billion at end-March 2026, thus amount to more than a third of all bills held by foreign official institutions combined ($388 billion at end-2025). A related perspective: our reading of persistent dollar strength against the Fed-cuts paradox.

The absolute gap with money market funds remains an order of magnitude: $2,687 billion against about $142 billion. And it is not closing. Between end-2022 and end-2025, money market fund bill holdings rose from $602 billion to $2,687 billion (Z.1), a 4.5-fold increase, while total stablecoin capitalization went from $137 billion to $307 billion (DefiLlama), a 2.2-fold increase.

The yield effect is more delicate to isolate. Several academic studies have attempted the measurement. The reference work is Bank for International Settlements Working Paper No. 1270, “Stablecoins and safe asset prices” (Rashad Ahmed and Iñaki Aldasoro, May 2025, daily data from January 2021 to March 2026 in its current version). It isolates flows through idiosyncratic variation at the token-blockchain level and finds that a $3.5 billion (2-standard-deviation) inflow lowers 3-month T-bill yields by 0.71 basis points on impact and by up to 4 basis points within 10 days, with limited spillovers to longer tenors. The effect increases under Treasury market stress and with the scale of the sector.

The effect is asymmetric. Stablecoin demand for T-bills is largely price-inelastic: an issuer places capital inflows almost instantaneously into eligible collateral, with little tactical room for maneuver. The inelasticity makes downward pressure on yields mechanical during growth phases, and upward pressure during contraction phases equally mechanical. The first version of the BIS paper (May 2025) estimated outflow effects at two to three times those of inflows. The March 2023 episode shows the issuer side of the mechanism: on March 10, Circle disclosed that $3.3 billion of USDC reserves sat at Silicon Valley Bank, just as redemption requests were mounting.

The stablecoin layer within offshore dollar plumbing

Offshore dollar plumbing refers to the ecosystem of dollar-denominated deposits and claims held outside the US banking system. The layer, whose contours are imprecisely measured, is historically dominated by the eurodollar market — interbank dollar deposits and loans operated by non-US banks, tracked by the BIS in its global liquidity indicators. Alongside that classical system, a stablecoin layer is now developing, whose cross-border use is described in the Chainalysis Geography of Cryptocurrency reports.

The geography of usage is documented. Chainalysis points to Southeast Asia (Vietnam, the Philippines, Indonesia) and Latin America (Argentina, Brazil). Turkey, Nigeria and parts of East Africa complete the map. In these jurisdictions, stablecoin usage is not speculative: it serves as a working dollar store of value in contexts of elevated domestic inflation, capital controls or banking instability.

The distinction with the classical eurodollar system is economic in nature. Eurodollars operate on a fractional reserve banking model: a dollar deposit in a bank outside the United States serves as the base for a dollar-denominated loan, creating endogenous money. Fiat-backed stablecoins operate on a full-reserve model: each token issued corresponds, in theory and according to attestation reports, to one dollar-equivalent immobilized in a liquid asset. The absence of internal leverage changes the systemic profile. The stablecoin layer, at equal volume, mechanically generates fewer credit cycles than the eurodollar layer. The difference is partly attenuated by emerging practices of yield-bearing stablecoins and DeFi-replicated stablecoins with leverage, whose share remains minor but is growing according to DefiLlama 2025 data.

The historical comparison with the rise of the eurodollar market in the 1960s and 1970s offers context. Eurodollar deposits grew from a few billion at the start of the 1960s to over $1 trillion by the late 1970s, reshaping the offshore financing structure of multinational corporations and creating regulatory blind spots that took two decades for the BIS and central banks to map. The stablecoin trajectory, on a different scale and with different drivers, displays the same property of regulatory exteriority during its formative phase. Faced with that exteriority, several central banks have opened a file of their own: the state of play on central bank digital currencies.

For broader monetary plumbing context (global financial conditions, QT-era dynamics, market impact channels) see the broader monetary plumbing context, which frames the operating environment in which any new dollar liability layer takes shape.

Digital shadow money: a distinct risk profile

The shadow money characterization of stablecoins is by now an established academic thesis. Gary Gorton, Yale professor, and Jeffery Zhang published in 2023 in the University of Chicago Law Review an article titled “Taming Wildcat Stablecoins” that formalizes the argument: stablecoins share the economic features of private money issued by US wildcat banks of the nineteenth century, par-redeemable liabilities into sovereign money, backed by assets whose quality is not guaranteed by a public authority, and susceptible to runs when issuer solvency is questioned. The historical lens has predictive utility: wildcat banks produced, between 1837 and 1863, recurring sequences of defaults and holder losses, until the National Banking Act standardized monetary issuance.

The analogy has been empirically tested twice in recent years. The May 2022 collapse of TerraUSD wiped out close to $18 billion in capitalization within four days. During the March 2023 USDC depeg, the token fell as low as 87 cents on March 11 and only regained parity two days later, following the public intervention on Silicon Valley Bank. The two episodes do not rest on the same mechanism: Terra suffered an endogenous algorithmic failure tied to the reflexive dynamics of its backing model, while USDC suffered a transmitted banking shock from its reserve concentration at SVB. But both produced the same secondary market dynamic: rapid price decoupling, rush to redemption, partial freezing of redemption channels for USDC during the hours when the US banking market was closed. The comparative reading is developed in what the 2022 and 2023 stablecoin runs revealed, which traces the minute-by-minute chronology of both episodes.

The absence of a lender of last resort is the property that most distinctly separates the stablecoin layer from the regulated commercial banking system. US banks benefit, through the Fed, from access to the discount window and, in case of systemic stress, to specific facilities such as the BTFP created in March 2023 during the regional banking crisis. US money market funds benefit indirectly from guarantees through the MMLF and MMIFF facilities activated in 2008 and 2020. Stablecoin issuers have no equivalent. In a generalized stress event on the layer, forced liquidation of T-bill reserves would occur at market prices, with no public backstop. More context: DeFi and traditional finance, two rails compared.

The absence defines the residual risk profile. It has two implications. The first is microeconomic: backing quality becomes the principal discipline mechanism for issuers, which explains the convergence toward short T-bills as reference asset and the rising pressure for reserve transparency. The second is systemic: a major stress event on a dominant issuer could trigger significant forced selling of T-bills, whose effect on the short segment is not trivial to isolate from other market movements.

The regulatory frameworks under construction aim precisely at mitigating these vulnerabilities. The European Union’s MiCA regulation, applicable to the stablecoin chapter since June 30, 2024, imposes strict reserve, liquidity and disclosure rules on issuers of reference stablecoins (Asset-Referenced Tokens and E-Money Tokens). The GENIUS Act adopted by the US Congress in 2025 introduces an equivalent federal framework, with supervision combining the OCC for bank issuers and a specific status for non-bank issuers. The two frameworks converge on three principles: full reserve, eligible liquid assets restricted to T-bills, Treasury repo and insured bank deposits, and regular publication of attestations.

In our reading, the cross-border footprint of stablecoins remains the dimension least covered by current frameworks. Issuers serve users across jurisdictions whose regulators have no jurisdiction over the issuer, creating a coordination gap that the Financial Stability Board has begun to address but has not yet closed.

Three documentable macro-financial implications

Marginal pressure on short Treasury bill yields

The effect of stablecoin demand on 1-3 month T-bill yields is modest at the individual issuer scale, but aggregated it becomes measurable. According to BIS Working Paper No. 1270 (Ahmed and Aldasoro), a $3.5 billion inflow lowers 3-month yields by up to 4 basis points within 10 days. The effect is measured at that horizon: it does not add up mechanically over the sector’s growth since 2021. It remains below that of a monetary policy decision, a 25 basis point fed funds move, but it is measurable on the short compartment. A broader view: How stablecoins and CBDCs compare.

Acceleration of dollarization outside the banking circuit

Dollarization refers to the use of foreign currency, here the dollar, in the monetary functions of a domestic economy — store of value, unit of account, medium of exchange. Historical episodes of dollarization (Argentina, Lebanon, Zimbabwe) rested on the use of physical banknotes or dollar-denominated bank accounts in jurisdictions that permitted them. The stablecoin layer bypasses the banking circuit by allowing dollar-denominated digital holding and transfer without a bank account. According to Chainalysis 2024 reports, stablecoin adoption in soft-currency jurisdictions follows dynamics distinct from classical dollarization, with a lower adoption threshold (a smartphone and an internet connection are sufficient) and higher usage volatility than dollar-denominated bank deposits. For the broader macro frame in which the dynamic sits, see the real return record of US 3-month T-bills, which contextualizes the short-end dollar instrument over a multi-decade horizon.

A category of systemic risk with a distinct profile

The systemic risk specific to the stablecoin layer combines three unusual characteristics. The first is concentration: USDT and USDC together accounted for 83% of dollar stablecoin capitalization at end-March 2026, creating a very high idiosyncratic issuer risk. The second is partial opacity: despite improvements in transparency reporting, Tether’s quarterly reports do not provide the daily granularity needed for stress monitoring. The third is the absence of a lender of last resort, already documented. The Financial Stability Board published its final recommendations on global stablecoin arrangements in July 2023, calling for internationally coordinated supervision.

An open architectural tension with the assigned sub-pillar

The present study is classified within the crypto infrastructure sub-pillar, which reflects the technical origin of stablecoins. But the macro-financial angle developed here belongs structurally to the sub-pillar dedicated to the global dollar system and offshore monetary plumbing. The tension is readable and deliberate. It reflects the current state of the topic’s evolution: stablecoins were born as crypto infrastructure, but their implications now reach far beyond that perimeter. Resolution of the architectural tension is a long-horizon editorial choice — either creating a dedicated digital monies sub-pillar within the monetary pillar, or maintaining the current attachment with explicit derivation. The present study leaves the question open, while acknowledging the dual framing. Whichever sub-pillar prevails, the wider frame remains the crypto-assets pillar, from liquidity cycles and real rates to financial integration.

Synthesis: what this invisible plumbing changes for reading the dollar system

The thesis defended in this study reduces to three empirical propositions. First, the stablecoin layer has become, at about $142 billion of T-bills held at end-March 2026, a structural actor on the short Treasury bill market, at more than a third of the bill portfolio of all foreign official institutions combined and exerting documentable marginal pressure on 1-3 month yields. Second, the layer functions as a new form of offshore dollar plumbing, with an economic profile distinct from classical eurodollars by the absence of internal leverage and the documented presence of non-speculative cross-border use. Third, it introduces into the global financial system a category of systemic risk previously unobserved, characterized by issuer concentration, partial reserve opacity, and the absence of a lender of last resort.

The reading completes naturally when placed against global monetary regimes and liquidity conditions in which the layer is embedded. It does not substitute for them: it adds an analytical layer to a macro-financial frame whose principal components (major central bank policies, dollar liquidity conditions, yield curve cycles) remain the primary determinants of the short Treasury segment.

Frequently asked questions

Difference between fiat-backed and algorithmic stablecoins

Fiat-backed stablecoins such as USDT and USDC rest on a reserve of real fiat-denominated assets (short Treasuries, Treasury repo, bank deposits) meant to fully cover circulation. Algorithmic stablecoins, of which TerraUSD was the archetype, rest on an arbitrage system with a volatile companion token without external asset backing. The May 2022 Terra collapse effectively disqualified the second model in its purest form. The present study deals exclusively with the first category.

Stablecoin holdings of US Treasury bills as of early 2026

At end-March 2026, Tether held $117 billion of US T-bills directly (BDO attestation for Q1 2026), out of $141 billion of direct and indirect exposure. USDT and USDC together hold about $142 billion directly, or about 2% of total Treasury bill outstandings ($6,819 billion at end-March 2026 according to the Federal Reserve’s Z.1 Financial Accounts).

Relationship between stablecoins and US money market funds

Fiat-backed stablecoin issuers and US government money market funds invest in the same underlying assets, short T-bills and Treasury repo, and perform a similar economic function of transforming liquid liabilities into short-dated paper. According to the Federal Reserve’s Z.1 Financial Accounts, MMFs held $2,687 billion of T-bills at end-2025, against about $142 billion for stablecoins, a ratio of roughly 19 to 1. The gap has not narrowed since end-2022: money market fund bill holdings rose 4.5-fold, stablecoin capitalization 2.2-fold.

Effect of stablecoins on short Treasury bill yields

BIS Working Paper No. 1270 (Ahmed and Aldasoro, 2025) estimates that a $3.5 billion stablecoin inflow lowers 3-month T-bill yields by 0.71 basis points on impact and by up to 4 basis points within 10 days, with a stronger effect under Treasury market stress. The effect is below that of monetary policy decisions, but it is econometrically detectable.

Last updated — 7 October 2026