How a Stablecoin Purchase Reaches Banks and Firms
This chapter follows one specific question: how can an unremunerated stablecoin still raise banks’ funding costs and affect firms’ investment? The answer is a chain of balance-sheet and pricing decisions, not a claim that every token purchase removes an equal amount of bank credit. Tracing that chain makes it possible to see both the contractionary bank-lending mechanism and the adjustments that can keep its aggregate effect modest.
From a household’s liquid-asset choice to a bank’s funding cost
A household does not hold a stablecoin because it pays more interest than a bank deposit. In the model, stablecoins are unremunerated. They compete with deposits because households value liquid assets for services that a financial return does not fully capture: ease of transaction, perceived safety and availability when the asset is needed. Cash provides some of the same services. A stablecoin adds another digital option, potentially usable on a programmable ledger and accessible through hosted or self-custodied wallets.
That choice changes the composition of household wealth. Suppose a household exchanges an existing bank deposit for a stablecoin. The household has not stopped saving; it has changed the form in which part of that saving is held. The bank, however, has lost a source of funding. The issuer has gained a liability—a token owed to the holder—and must place the proceeds in reserve assets. The first-round accounting depends on those assets, but the household’s demand for an alternative liquid claim creates competitive pressure on the deposit market in either case.
Banks benefit from the liquidity services that households associate with deposits. In exchange for those services, households accept a deposit rate below the economy’s risk-free rate. The difference is a liquidity premium for the bank: deposits are cheaper than fully market-priced funding because they are useful to depositors. When stablecoins become available, a household is less willing to supply the same amount of deposits at the old rate. To keep deposits, a bank must offer more remuneration.
In the working paper’s language, the deposit spread moves against the bank. If the risk-free rate is unchanged while the deposit rate rises, the gap between the two narrows. The bank no longer receives as much of the convenience value created by deposit holdings. That is the first important link: stablecoins need not pay interest to affect interest rates. They can change the opportunity set facing depositors, which changes the price banks must pay for funding.
The pricing change comes from the household’s marginal choice, not from a rule that forces banks to match a stablecoin’s market value. When a household has only deposits and cash, an additional deposit can provide substantial liquidity service. Once a stablecoin is available, the household has another way to hold a readily transferable claim. The extra service supplied by one more deposit is then less valuable at the margin. A bank must compensate for that lower convenience by offering a higher rate. In aggregate, the deposit market settles at a different quantity and price even though no one has ordered a particular bank to surrender a particular loan.
This is also why the relevant comparison is a deposit spread, not simply the level of the deposit rate. A deposit rate can rise because the central bank has raised the risk-free rate, while the bank’s relative funding advantage remains unchanged. The stablecoin mechanism concerns the gap between what a bank pays depositors and what a comparable risk-free asset pays. As that gap narrows, the bank’s cost of financing loans rises relative to the return it can earn on those loans. The effect then enters loan pricing through the bank’s financing wedge, alongside capital requirements, default risk and the expected return on other assets.
The result is not mechanically proportional to stablecoin market capitalisation. A household can finance the purchase by selling cash rather than a deposit. The working paper’s nested liquidity structure allows deposits, stablecoins and physical cash to substitute imperfectly, so the source of the funds matters. If much of the new stablecoin balance comes from cash, the direct loss of bank funding is smaller. If it comes mostly from retail deposits, the pressure on banks is greater.
Nor does the bank face one uniform funding market. The BIS Annual Economic Report 2026 distinguishes retail deposits from wholesale deposits and stresses that a stablecoin issuer’s deposits may be concentrated in a few large institutions. Wholesale funding can be more sensitive to market conditions and is treated as less stable for liquidity purposes. A bank may therefore replace a lost retail deposit with another liability without restoring the same funding quality or the same distribution of liquidity across institutions.
The adjustment also depends on competition. Banks may raise deposit rates, cut operating costs, improve payment services or seek other funding. The annual report notes that financial intermediaries have adapted to competition from money-market funds and online payment platforms in the past. Those responses could preserve some lending and give users lower fees or better deposit returns. The quantitative model deliberately keeps the focus on the funding channel and abstracts from the full range of possible bank innovation, so its result should not be read as a prediction that banks will remain passive.
What the model does establish within its assumptions is a clear price mechanism. When stablecoins make deposit supply less willing at a given rate, the equilibrium deposit rate rises. That raises marginal bank funding costs and weakens the financing advantage that deposits once provided. The next question is how that change reaches the real economy.
From bank funding to firm investment—and why the effect is cushioned
Banks do not absorb a higher funding cost without changing the price of their assets. In the model, banks price loans while accounting for their deposit financing, capital requirement, default risk and other asset returns. A smaller deposit-financing wedge means that a firm must pay more for a loan at the margin. The firm then faces a higher cost of capital.
A firm’s decision is intertemporal. Borrowing today can finance investment in productive capital, but the loan must be serviced from future revenues. When the loan price rises, fewer projects clear the firm’s return threshold. The firm borrows less, accumulates less capital and produces less output. If the firm has long-term debt, existing contracts do not reprice all at once; the new funding condition affects new borrowing and gradually works through rollover and investment decisions.
The adjustment can be seen as a change in the price schedule offered to firms. A firm with a risky project does not compare a loan rate only with the project’s expected average return. It also considers how much debt it can service in a weak productivity state and how an additional loan changes the probability of default. A higher loan price makes the firm’s marginal borrowing less attractive and can alter its preferred leverage. The model includes this risk choice, so lower credit supply is connected to both investment and the distribution of losses between lenders and equity holders.
Long-term debt makes the channel slower than a deposit-market quote. A new deposit rate can change as soon as households alter their portfolios, but a firm’s existing debt and installed capital remain in place. The initial response therefore appears in new issuance, refinancing and investment plans. As capital depreciates and debt rolls over, the higher financing cost reaches production. This timing is one reason the same adoption path can look expansionary during a transition and contractionary after the full capital and credit adjustment.
This is why “stablecoins disintermediate banks” is an incomplete explanation. Disintermediation is not an endpoint. The relevant macroeconomic question is what happens after the bank loses or reprices a funding source. Does lending fall? Do firms substitute toward equity? Do non-bank lenders expand? Do banks sell other assets or find wholesale funding? The working paper follows the first of these routes in a general-equilibrium model and then measures the other balance-sheet movements that offset it.
One offset begins with the household’s portfolio. Stablecoin holdings rise through substitution from both deposits and cash. Because cash is also unremunerated, the model finds that the shift away from cash is slightly stronger than the shift away from deposits. That keeps more bank funding in place than a simple one-for-one “stablecoin equals lost deposit” story would suggest.
A second offset occurs on the bank’s asset side. As stablecoin issuers become buyers of Treasury bills, banks can reduce their own bill holdings. Those holdings provide liquidity services and collateral, so the adjustment changes the bank’s asset mix. But it also means that a smaller bank balance sheet is partly matched by fewer Treasury bills rather than by an equal fall in loans. The model consequently finds that total loans respond less than the size of the portfolio reallocation might imply.
A third offset occurs at the firm. When bank loans become more expensive or less available, firms can issue more equity. This cushions investment and production, although it changes who bears the risk and can alter leverage. In the model’s additional results, the bank-lending channel lowers firm leverage and the default rate because firms reduce debt and rely relatively more on equity. That is not a free substitute: equity holders absorb more of the business risk, and the availability of equity finance varies across firms.
The public sector provides another adjustment. The government can change the maturity mix of its debt in response to the relative cost of Treasury bills and longer-term bonds. That response belongs more directly to the fiscal-space channel, but it matters here because banks and stablecoin issuers share the market for short-term public debt. The financial system can reallocate claims across households, issuers, banks, firms and the government even when the consolidated balance sheet changes only modestly.
These offsets do not cancel the bank-lending mechanism; they determine how much of it survives aggregation. Consider two economies with the same stablecoin size. In the first, households fund most purchases by giving up deposits and firms depend heavily on bank loans. In the second, households give up more cash, firms have easy access to equity and banks can replace retail deposits with stable funding. The initial token liability is the same, but the effect on loans and output is not. This is why a market-capitalisation figure cannot serve as a sufficient measure of disintermediation.
The working paper separates the channel in its quantitative exercises by holding banks’ Treasury-bill position fixed when measuring the bank-lending effect, and by holding the deposit spread fixed when measuring the fiscal-space effect. These are analytical devices, not claims that the economy can literally switch one channel off. They let the authors compare the contribution of each route while keeping the rest of the model structure in place. The overall response then reflects their interaction, including the way a change in one market alters prices and choices in another.
The model’s baseline is calibrated to US macro-financial data and assumes that deposits are broadly interpreted as bank liabilities and money-market-fund shares. It includes bank capital requirements, bank failure risk, firm default and long-term loans, but it does not allow banks to fully replace a contraction in deposit financing through wholesale markets. The authors state that relaxing this assumption weakens the quantitative effects in their setting without changing the qualitative direction. That qualification matters: the mechanism is a supported result of the model, while the size of the real-world effect depends on funding substitution and bank behaviour.
The same distinction applies to the role of users. Higher deposit rates can benefit households that continue to hold deposits. Banks may also improve payment services in response to competition, creating value that is not captured by a narrow output comparison. The working paper implicitly holds bank payment-service efficiency constant, while the annual report highlights adaptation as a possible counterweight. Evidence that stablecoins increase competition is therefore not evidence that every borrower benefits; it says that some gains and losses may be redistributed among depositors, banks, equity holders and firms.
The sequence should be read as a mechanism, not as a universal transaction path. The issuer’s reserve choice determines whether funds move toward Treasury bills or back into bank deposits. The household’s original asset determines how much retail bank funding is displaced. The firm’s access to equity and the bank’s access to replacement funding determine how much of the initial pressure reaches loans and investment.
This also explains why a large change in ownership can produce a small change in aggregate output. At adoption levels considered in the model, stablecoins become sizable participants in the Treasury-bill market, but remain a relatively small share of households’ total liquid assets. Banks, firms and the government adjust their portfolios. The consolidated financial sector’s balance sheet moves less than the individual balance sheets, even though the distribution of liquidity, interest income and risk changes materially.
The limits are as important as the chain. The model assumes depositors are protected by bailout guarantees, treats stablecoins as nominally safe, and holds adoption as an input rather than explaining why users choose one issuer or ledger over another. It does not fully model wholesale-market replacement, operational outages, cross-chain failures or a run. The annual report adds that non-bank financial institutions may take a larger role in credit, but their funding can be more procyclical and they do not create credit as elastically as banks. Those possibilities could either cushion or intensify the bank-lending effect.
The careful conclusion is therefore conditional. Stablecoin demand can make bank deposits more expensive and bank loans less attractive, creating a plausible route to lower firm investment. Cash substitution, equity finance, asset sales, bank adaptation and alternative lenders can weaken that route. The result is a real mechanism with an uncertain magnitude, not a binary verdict on whether stablecoins help or harm banks.
The next question follows from the issuer’s asset side. If the issuer buys Treasury bills instead of leaving all proceeds in the banking system, the same private token creates demand for public debt—and that can give the government fiscal space to offset some of the credit pressure.