Circle, Tether and the fight for bank deposits

Stablecoins are digital tokens designed to track a currency, here the US dollar. With Circle’s USDC and Tether’s USDT, reserve investment connects the money in bank accounts with the cost and stability of funding for US banks.

Documentary investigation and analysis ; 8 November 2026. Reserve data as of 30 September 2026. Numerical examples identified as illustrations were constructed by l0g.

Imagine a bank losing $100. Its customer uses the money to obtain stablecoins: tokens designed to hold a value of one dollar and move across a blockchain. When the payment settles, another bank receives the funds in the issuer’s account. The customer now owns tokens; a company owns the bank deposit.

This simplified transaction captures the challenge for banks. An institution can lose a customer relationship, a payment account and a cheap source of funding while the dollars continue circulating through the banking system. The next stage depends on where the issuer places them: in a deposit, in government debt or in a secured loan to a financial intermediary. ^B04

A bank has a familiar way to compete for that money: pay more interest. Research published by the Bank for International Settlements (BIS) finds that stablecoin demand responds to US deposit rates. The rate offered to the customer, however, is only the beginning of the story. The rest unfolds inside Circle’s and Tether’s reserves, then on the balance sheets of the institutions receiving the funds. ^B01

A new claim on the dollar

Consider a new issue at par, with no fees, funded within the US banking system. In this illustrative example, an eligible buyer transfers $100 from bank A. The issuer receives it at bank B and creates 100 tokens. The buyer’s deposit falls by $100 and the issuer’s rises by the same amount. Within this closed example, total bank deposits are unchanged. Their ownership has shifted.

Settlement between A and B also moves their balances at the Federal Reserve (Fed), the US central bank. A transfers $100 of settlement reserves; B receives $100. Those interbank reserves settle the payment. The assets the issuer keeps to back its tokens are the stablecoin’s reserves. The same word describes two different layers of the financial system. ^R03

The buyer now holds a different claim. A deposit is money owed by a bank. A stablecoin rests on the issuer’s obligations and the applicable redemption terms. Direct redemption may depend on eligibility and contractual conditions; holders can also use a platform or sell their tokens. Tracking the effect on bank funding therefore requires separating token issuance and redemption from trading between users. ^{R04 R05}

Suppose Alice buys 100 existing tokens from Ben. Her payment may simply credit Ben’s bank account. The number of tokens outstanding stays the same. Treating the trade as another $100 of stablecoin reserves would confuse the transfer of an asset with its creation. For the same reason, blockchain transaction volume answers a different question from the amount of bank deposits displaced.

Deposit rates change the cost of holding tokens

For someone keeping dollars readily available, a better bank rate raises the opportunity cost of holding a stablecoin: the interest forgone by choosing the token. How much that matters depends on the purpose of the balance. Money earmarked for an imminent payment bears that cost for hours or days; a cash reserve bears it for months. The same rate increase can be decisive for one user and barely relevant to another.

Ahmed and Aldasoro study the period from January 2019 to June 2025. Their paper, published by the BIS on 13 October 2026 following an initial version in March, links higher deposit rates to slower subsequent stablecoin growth, particularly for USDC. The estimated magnitude varies by method. The analysis concerns demand for an asset; it identifies neither the individual depositors involved nor an aggregate loss of credit. ^{B01 B08}

The economic direction is useful: a bank can improve the relative appeal of its account. Translating a statistical coefficient into a commercial decision takes more work. A response estimated from small weekly changes in a particular setting provides a research benchmark. A bank would still need to identify the customers sensitive to rates, how much money they might move and how competitors would respond.

The practical question is how much a dollar of retained funding is worth. Imagine a bank with $1 billion in deposits, of which $100 million is at risk of leaving. An increase of 0.10 percentage points, or ten basis points, across the whole deposit base costs $1 million a year, assuming balances remain constant. Replacing the departing $100 million with funding priced one percentage point above the original funding cost produces the same annual increase in expense.

This is arithmetic built from illustrative assumptions, with no observed rate or forecast of withdrawals. It assumes the general rate increase retains all $100 million at risk, replacement funding is available and other costs stay the same. It nevertheless exposes the commercial difficulty: a bank may have to pay more to many customers who had no intention of leaving. A targeted offer changes the calculation; its feasibility and effectiveness would need to be established.

Competition delivers a gain to the saver and a higher interest bill to the bank. The bank can absorb the expense in its margin, seek other revenue, change its services or adjust lending terms. Its market and balance sheet shape the response. In a historical review published in May 2026, Federal Reserve economists describe earlier banking adaptations to money market funds and new payment technologies through changes in products and strategy. ^R01

Where the reserves go

Circle’s disclosures let us follow the next stage. On 30 September 2026, assets backing USDC totalled $73.957 billion. Of that, $62.467 billion consisted of shares in the Circle Reserve Fund, a money market fund. The reserve attestation looks through the fund to its holdings: $36.654 billion in overnight loans secured by Treasury securities, $25.362 billion in Treasuries, plus cash and settlement adjustments. The remaining reserve assets were held outside the fund. ^{I01 I02}

A money market fund adds a legal intermediary. Understanding where the money goes economically requires examining what the fund owns. Adding the value of its shares to the value of its portfolio would count the same investment twice.

The secured loans are known as reverse repos from the cash lender’s perspective. The lender advances money for a short period and receives securities as collateral, with a reverse transaction agreed. Circle’s attestation describes overnight lending to global financial institutions, overcollateralised by US Treasury securities. A dollar can therefore finance an intermediary through a short-term secured claim whose terms differ from an ordinary deposit. ^I01

Tether’s structure is different. Its report for 30 September 2026 lists $191.451 billion in assets, including $122.840 billion in US Treasury bills and $16.696 billion in reverse repos. Cash and bank deposits amount to $20.427 million. The scope is all assets of Tether International backing its fiat-denominated tokens, including surplus assets. That differs from the USDC reserve perimeter. The report expressly identifies Treasury collateral for overnight repos; the term-repo note describes collateral and credit-quality requirements more generally. ^I03

A small bank-deposit line tells us what the issuer owned at a particular moment. Establishing a withdrawal from the banking system would also require knowing who sold it the securities, where the payment arrived and what the recipient did with the proceeds. A portfolio statement leaves that part of the journey unresolved.

The Treasury seller matters

Return to the $100 sitting in the issuer’s US bank account. The issuer now uses it to buy a Treasury bill. Three transactions that look similar on a screen have different effects within the US banking system.

A non-bank investor sells an existing security and receives payment into a bank account. The issuer’s deposit falls by $100; the seller’s rises by $100. At this stage, total deposits are unchanged. The seller might then buy another asset or spend the proceeds. Each subsequent transaction needs its own analysis.

A bank sells a security from its own portfolio. It receives settlement in its capacity as a bank, rather than crediting a selling customer’s account. Across the banking sector, securities and deposits both fall by $100. Existing loans can remain unchanged at the moment of the transaction. The balance sheet has contracted through a change in securities holdings, a different process from a loan repayment.

Finally, the issuer buys a newly issued Treasury security, paying from its deposit. The proceeds reach the Treasury’s account at the Fed, the Treasury General Account, or TGA. In this isolated scenario, bank deposits and settlement reserves both fall by $100. When the Treasury spends the same $100 on a private recipient paid through a US bank account, deposits and reserves return to their starting levels in this example. The timing of government financing and spending becomes important. ^{R03 M05}

These are accounting illustrations, with other conditions held constant. They identify counterparties before drawing macroeconomic conclusions. Actual portfolios combine trades, maturities, redemptions, funds and repos. A single figure for stablecoin issuance can therefore be consistent with several different changes in bank balance sheets.

The receiving bank changes too

Banking terminology distinguishes retail deposits, held by individuals, from wholesale deposits placed by companies and institutions. Their characteristics are part of assessing a bank’s soundness.

Bank A knows the customer it has lost. Bank B receives an issuer’s funds, which may move as tokens are issued or redeemed and reserves are invested. The amount can be identical while the cash-management behaviour changes. A large account concentrates decisions previously spread across many holders. For the receiving bank, managing liquidity becomes critical. ^B04

Lee and Tou examine this channel in a New York Fed paper published in February 2026. They link blockchain transactions to the interbank payments of an anonymous issuer. After March 2023, three new partner banks held a larger share of their assets in reserves and a smaller share in loans relative to control banks. The sample is narrow, and the authors cannot estimate the effect on aggregate credit. ^M01

A bank creates a deposit when it makes a loan. It must then settle its customer’s payments and manage the funding, liquidity and capital of its balance sheet. Stable deposits help it meet those demands. ^R06

Maturity transformation matters too. A mortgage or business loan commits a bank’s funds over a long period. A highly mobile deposit can require a larger liquidity cushion. If the bank devotes more of its balance sheet to immediately available assets, loans can account for a smaller share even as the balance sheet grows. A lower loan ratio and a decline in the total dollar amount of loans are separate findings.

An unchanged aggregate deposit total would thus provide incomplete reassurance about financing for the economy. Extrapolating the experience of three banks into a national forecast would also stretch the evidence. The more specific finding is useful: the characteristics of incoming funding can influence the assets a bank chooses to hold.

A separate ECB paper, published in March 2026, studies deposits and lending using bank data and a measure of attention to stablecoins derived from Google searches. It provides evidence consistent with this funding risk, subject to its identification assumptions. Online attention remains a proxy for actual adoption. ^B05

On October 29, 2026, BIS General Manager Pablo Hernández de Cos highlighted a redemption risk: an issuer facing customer withdrawals could draw down bank deposits or sell reserve assets under pressure. Reserve composition and banking links therefore matter when holders want their dollars back as well. ^R07

Yield, payments and access to dollars

Our analysis of who receives stablecoin reserve income follows the contracts linking issuers, platforms and token holders.

A deposit rate matters most when the user can actually choose that account. For someone abroad, access to transferable dollars, the platforms used by trading partners or a particular payment channel may outweigh a few additional basis points at a US bank. A research note published by the Fed emphasises that both the geography of demand and the assets converted into stablecoins affect the outcome for US deposits. ^B04

The income earned on reserves also needs to be distinguished from the return received by a token holder. A platform may pay rewards funded by reserve income or by its own market activities. BIS Bulletin 125 describes both arrangements. Their funding and risks differ. A bank’s rate therefore competes with a package of services, holding conditions and possible rewards. ^B03

Regulation shapes that contest. The final rule issued by the Office of the Comptroller of the Currency (OCC), published on 23 October 2026, is scheduled to take effect on 18 January 2027. It retains the prohibition on issuer-paid yield within the statute’s scope and drops the proposed presumption covering certain indirect payments, while preserving anti-evasion authority. Arrangements involving intermediaries remain subject to assessment on their facts. Any comparison of returns needs to account for both the applicable rules and their implementation dates. ^{M06 M09}

A model published by White House economists on April 8, 2026 examines a ban on stablecoin yield. Its baseline estimates a small increase in bank lending. That result depends partly on ample Federal Reserve reserves and issuers’ reserve allocation. The exercise connects a regulatory choice with credit in a calibrated framework, using substitution behaviour drawn from money market funds. Its intervention and assumptions need to accompany the result before it is extrapolated to stablecoin growth. ^R08

Bargaining power shifts

Three observations would be especially revealing as this market develops. The first is the rate actually received by customers choosing between deposits and stablecoins, after fees and conditions. The second is the identity of the banks receiving the funds and how much of their funding is concentrated among a few depositors. The third is what those banks hold against it: settlement reserves, liquid securities or loans to the economy.

Growth at Circle or Tether alone leaves these questions open. More tokens outstanding could reflect a US bank customer switching products, an overseas business seeking dollar liquidity or additional payment needs. Each case has a different chain of transmission.

Banks have a plausible lever for defending part of their deposit base: pay more for it. The commercial effectiveness and cost need to be assessed institution by institution. Stablecoins introduce another contest over reserve investment and the services that keep funds moving. The outcome could be bank funding that is more expensive, more concentrated or associated with a greater preference for liquidity, with substantial differences between banks.

The useful starting point is to follow the dollar to its next holder and to the asset it finances. Those two destinations help determine how the role of banks changes.

How to read this investigation

Portfolio values are stocks measured on a particular date. Research coefficients are estimates from a historical sample. The $100 payment routes and funding-cost calculation are teaching examples. The references below give the primary sources, their scope and limitations. The econometric studies cited here have not been replicated for this investigation.

Methods and illustrations

The 100 dollar paths represent isolated transactions, all else equal, at par and without fees, within US banks. A secondary-market purchase differs from a new token issue. Deposit changes are kept separate from settlement reserves and the issuer’s assets. Each participant’s accounts balance; subsequent Treasury spending is an additional stage.

The cost model is fictional and annualized on a constant balance. An increase of 10 basis points on 1 billion dollars costs 1 000 000 000 × 10 / 10 000 = 1 000 000 dollars a year. Replacing 100 million at an additional 100 basis points costs 100 000 000 × 100 / 10 000 = 1 000 000 dollars a year. The curve applies that second formula to hypothetical spreads. Successful retention, available replacement funding and unchanged other costs are assumptions whose feasibility requires investigation.

The cover illustration is conceptual artwork generated with an AI tool. The passive diagrams explain mechanisms or scenarios; this investigation conducted no bank-withdrawal or conversion experiment.

Sources and documents

1. B04 · Board of Governors of the Federal Reserve System, FEDS Notes · Accessed: 2026-11-08Banks in the Age of Stablecoins: Some Possible Implications for Deposits, Credit, and Financial Intermediation
   Passages and scope

   Document: 2025-12-17

   Sections 2.1–2.3 (sources of demand, reserve allocation, deposit composition); section 3.2 (scenario assumptions)

   A Federal Reserve researcher’s conceptual analysis with conditional scenarios. The note supplies mechanisms and assumptions rather than an observed causal estimate of lost lending.

2. B01 · Bank for International Settlements · Accessed: 2026-11-08Are stablecoins and bank deposits substitutes? : BIS Working Papers No 1382
   Passages and scope

   Document: 2026-10-13

   Printed pp. 1, 6–15, 17, 23, 26–29, 32–33; Table 1 p. 8; equation 5 p. 9; equation 6 p. 10; Table 2 p. 17; Table OA.A.1 p. 23

   Weekly asset-demand estimates, sensitive to method and identification. The sample contains one sustained high-rate episode; aggregate lending is outside the measured outcomes.

3. R03 · Bank for International Settlements · Accessed: 2026-11-08Anchoring trust in money: innovation beyond stablecoins : Annual Economic Report 2026, Chapter III
   Passages and scope

   Document: 2026

   Macro-financial implications of stablecoins; Scenarios for stablecoin reserve assets; Graph 4

   Conceptual reserve-allocation scenarios. Settlement reserves at the central bank and assets backing tokens are distinct balance-sheet items.

4. R04 · Circle · Accessed: 2026-11-08USDC Terms
   Passages and scope

   Document: 2026-10-15

   Preamble; section 2 Scope of USDC and Key Terms; section 4 Eligibility; Claim on funds

   Circle’s contractual terms for the stated jurisdiction, including the non-EEA scope. Redemption access depends on eligibility and the applicable arrangement.

5. R05 · Tether · Accessed: 2026-11-08How to redeem Tether tokens to fiat currency
   Passages and scope

   Document: Publication date not established

   Requirements for redemption: verified account and minimum amount

   Provider terms for direct redemption, distinct from a sale between token holders. Publication date has not been established.

6. B08 · Bank for International Settlements · Accessed: 2026-11-08Are stablecoins and bank deposits substitutes? : official BIS publication page
   Passages and scope

   Document: 2026-10-13

   Date Published, authors, abstract, PDF download

   Metadata and a rounded summary of the same research as B01, rather than independent confirmation.

7. R01 · Board of Governors of the Federal Reserve System · Accessed: 2026-11-08Banks in the Age of Stablecoins: Lessons from Their Historical Responses to Financial Innovations
   Passages and scope

   Document: 2026-05-01

   Introduction; sections on money market funds and payment platforms; conclusions

   Historical research by Federal Reserve economists. Earlier banking responses provide context with their own market and regulatory conditions.

8. I01 · Circle Internet Group, Inc.; Deloitte & Touche LLP · Accessed: 2026-11-08USDC Reserve Report : September 4 and September 30, 2026; Independent Accountants’ Report
   Passages and scope

   Document: Publication date not established

   PDF pp. 1–2: opinion, scope and reserve/circulation totals; pp. 4–5: September 30 reserve components; p. 6 notes 3–4: repo and settlement mechanics

   Assurance on Circle’s assertions at the specified dates. Reserve stocks are dated snapshots with a scope distinct from Tether’s.

9. I02 · Circle Internet Group, Inc.; SEC EDGAR · Accessed: 2026-11-08Circle Internet Group, Inc. : Form 10-Q for the quarter ended September 30, 2026
   Passages and scope

   Document: 2026-11-04

   Printed pp. 59–60 Composition of USDC Reserves; p. 63 Item 3 Interest rate risk; pp. 44–45 interest-rate fluctuations; Note 2 reserve accounting

   Circle’s regulatory disclosure. Fund shares and their underlying portfolio are two levels of the same investment and must be distinguished in aggregation.

10. I03 · Tether International, S.A. de C.V.; BDO Advisory Services S.r.l. · Accessed: 2026-11-08Tether International, S.A. de C.V. : Financial Figures and Reserves Report as of September 30, 2026, with ISAE 3000R assurance report
    Passages and scope

    Document: 2026-10-30

    PDF pp. 2–4 assurance scope; PDF p. 6 / internal p. 1 entity and all-assets scope; PDF p. 8 / internal p. 3 totals; PDF p. 9 / internal p. 4 asset composition; PDF p. 10 / internal p. 5 overnight repo collateral

    All assets allocated to Tether International’s fiat-denominated tokens, including surplus. The entity and coverage differ from the USDC reserve; bank-account locations are not fully disclosed.

11. M05 · Federal Reserve Bank of St. Louis · Accessed: 2026-11-08The Fed’s Balance Sheet and Ample Reserves
    Passages and scope

    Document: 2026-02-18

    Paragraphs defining TGA/reserve balances; section Why Will There Be Movement in Currency, the TGA, and Reserves?

    An educational description of the Federal Reserve’s balance sheet and ample-reserves framework. Treasury settlement and subsequent spending are separate stages.

12. M01 · Federal Reserve Bank of New York · Accessed: 2026-11-08Stablecoin Disintermediation
    Passages and scope

    Document: 2026-02

    Sections 3.2–3.5, printed pp.15–19; Section 4.2, pp.23–24; Tables 4–5, pp.41–42; Section 5 robustness

    One anonymous issuer and three new partner banks. The researchers’ design leaves the effect on aggregate bank credit unresolved.

13. R06 · Bank of England · Accessed: 2026-11-08Money creation in the modern economy
    Passages and scope

    Document: 2014-03-14

    Quarterly Bulletin 2014 Q 1, pp.14–27; sections on loan creation and limits to broad-money creation

    A general Bank of England explanation of money creation and banking constraints. The numerical transaction examples in this article are l0g constructions.

14. B05 · European Central Bank · Accessed: 2026-11-08Stablecoins and monetary policy transmission : ECB Working Paper Series No 3199
    Passages and scope

    Document: 2026-03-03

    Section 3.1 pp. 10–13; Figure 4 p. 13; Section 3.2 pp. 14–17; Table 1 p. 17; balance-sheet accounting p. 10 and footnote 13

    German Google Trends searches proxy stablecoin attention or adoption. Findings depend on identification assumptions and confidential bank data; the authors’ views remain distinct from an ECB policy position.

15. R07 · Pablo Hernández de Cos / Bank for International Settlements · Accessed: 2026-11-08Deposit insurance in a changing financial landscape
    Passages and scope

    Document: 2026-10-29

    New forms of money: redemptions, bank withdrawals and reserve-asset sales.

    An attributed policy speech describing conditional redemption risks. It provides no new numerical estimate of lost bank credit.

16. B03 · Bank for International Settlements · Accessed: 2026-11-08Stablecoin remuneration on centralised exchanges : BIS Bulletin No 125
    Passages and scope

    Document: 2026-06-19

    Printed pp. 1–4; reserve/activity remuneration p. 2; Graphs 1–2 p. 3; Graph 3 p. 4; references p. 6

    Historical data from two major exchanges selected for data availability. Present reward terms and yields require a separate current assessment.

17. M06 · Office of the Comptroller of the Currency · Accessed: 2026-11-08GENIUS Act Regulations: Final Rule : OCC Bulletin2026-50
    Passages and scope

    Document: 2026-10-21

    Summary, Highlights, Background

    The announcement precedes Federal Register publication on October 23. The final rule is scheduled to take effect on January 18, 2027.

18. M09 · Office of the Comptroller of the Currency / Federal Register / Government Publishing Office · Accessed: 2026-11-08Implementing the Guiding and Establishing National Innovation for U.S. Stablecoins Act for the Issuance of Stablecoins by Entities Subject to the Jurisdiction of the Office of the Comptroller of the Currency : Final rule
    Passages and scope

    Document: 2026-10-23

    Federal Register 91, pp.67000–67194, doc 2026-21706; Yield discussion; OCC prepublication PDF pp.95–110; final 12 CFR 15.10(c)(4)

    A final rule with future effect at the research cutoff. The issuer prohibition and anti-evasion powers remain; intermediary arrangements require a facts-based assessment.

19. R08 · Council of Economic Advisers / The White House · Accessed: 2026-11-08Effects of Stablecoin Yield Prohibition on Bank Lending
    Passages and scope

    Document: 2026-04-08

    Full report, pp.4–5: intervention and static model; pp.11–12: calibration; pp.13–18: results and limitations.

    A calibrated yield-prohibition model using domestic demand, ample reserves and substitution behaviour drawn from money market funds. Its policy intervention differs from the BIS deposit-rate study.