Effects of Stablecoin Yield Prohibition on Bank Lending
In April 2026 CEA published Effects of Stablecoin Yield Prohibition on Bank Lending. CEA’s analysis determined that an effective ban on stablecoin yield would have little effect on bank lending. Several trade groups and advocacy organizations have published responses. This FAQ reviews the paper’s core findings, addresses objections raised by trade groups and advocacy organizations, and shows how the paper’s results hold up under the alternative assumptions proposed by these organizations.
Executive Summary:
The GENIUS Act, signed into law in July 2025, requires stablecoin issuers to maintain reserves backing outstanding stablecoins on at least a one-to-one basis. Reserves may only consist of certain specified assets, including US dollars, federal reserve notes, funds held at certain insured or regulated depository institutions, certain short-term Treasuries and Treasury-backed reverse repurchase agreements, and money market funds. It also prohibits stablecoin issuers from offering any form of interest or yield to stablecoin holders, but does not explicitly prohibit affiliate or third-party arrangements that might offer interest-bearing products. Some variants of the proposed CLARITY Act would close this channel. One rationale for prohibiting yield is that if stablecoins were to offer competitive returns, households may shift dollars out of traditional bank accounts and into tokens. Since stablecoin reserves are fully backed rather than fractionally lent, this could reduce bank lending. Some analyses estimate the effect on lending in the trillions of dollars (Nigrinis 2025). We build a simple model to evaluate these claims.
At baseline calibration of CEA’s model, eliminating stablecoin yield increases bank lending by $2.1 billion and has a net welfare cost of $800 million. That translates into an increase in lending of 0.02% and a cost-benefit ratio of 6.6. Large banks would conduct 76% of this additional lending, while community banks, which have assets below $10 billion, would lend the remaining 24%. In our baseline, that adds up to $500 million in additional lending from community banks, meaning their lending rising by 0.026%.
Even stacking every worst-case assumption, the model produces only $531 billion in additional aggregate lending, which corresponds to a 4.4% increase in bank loans as of 2025Q4. That figure requires the stablecoin market to grow to roughly six times its current size as a share of deposits, all reserves to be locked in unlendable cash rather than treasuries, and the Federal Reserve to abandon its current monetary framework. Even under those implausible conditions, community bank lending only rises by $129 billion, corresponding to an increase of 6.7%. The conditions for finding a positive welfare effect from prohibiting yield are similarly implausible. In short, a yield prohibition would do very little to protect bank lending, while forgoing the consumer benefits of competitive returns on stablecoin holdings.
Source: CEA, Effects of Stablecoin Yield Prohibition on Bank Lending (April 2026).
Adjusting assumptions
Each row of the paper’s Table 3 changes one assumption relative to the row above it, so the changes accumulate and the last row combines every departure from the baseline. The column at left describes what each mix of assumptions represents. Load any row into the model below to see the calculation behind it.
| What this mix of assumptions represents | Cash share θ | Stablecoin share of deposits | Elasticity σ | Reserves | ΔLoans ($B) | % of loans | Welfare ($B/yr) | Cost / benefit | Community ΔLoans ($B) | % of community loans | Load row into the model |
|---|---|---|---|---|---|---|---|---|---|---|---|
| BaselineToday’s market: about $300 billion of stablecoins, 1.7% of bank deposits. Issuers hold 12% of reserves as bank cash (Circle’s share) and the rest in Treasury bills. Yield sensitivity is estimated from money market fund flows. The Federal Reserve keeps its ample-reserves regime. | 0.12 | 1.7% | 7 | Abundant | 2.1 | 0.02% | -0.8 | 6.60 | 0.5 | 0.03% | |
| Six-fold shareThe stablecoin share of deposits rises six-fold, to 10%, which is more than $1.7 trillion at today’s deposit base. Reserve composition, yield sensitivity, and the Fed’s regime stay at baseline. | 0.12 | 10% | 7 | Abundant | 11.1 | 0.09% | -4.2 | 6.60 | 2.7 | 0.14% | |
| Six-fold share, high yield sensitivityThe six-fold share, with households at the top of the yield-sensitivity range (elasticity 9 rather than 7). A yield ban then moves $375 billion from stablecoins into deposits. | 0.12 | 10% | 9 | Abundant | 14.3 | 0.12% | -5.5 | 6.60 | 3.5 | 0.18% | |
| Break-even cash shareThe six-fold share and high sensitivity, with issuers holding roughly 60% of reserves as locked bank cash (printed as 0.63). This is the point at which the lending benefit just offsets the welfare cost to households, so the welfare column reads zero. | 0.63 | 10% | 9 | Abundant | 56.8 | 0.47% | 0.0 | 1.00 | 13.8 | 0.71% | |
| All reserves lockedThe six-fold share and high sensitivity, with every reserve dollar held as locked bank cash or at the Federal Reserve. Nothing recycles through Treasury bills. The Fed still runs ample reserves. | 1.00 | 10% | 9 | Abundant | 72.0 | 0.60% | 2.7 | 0.46 | 17.5 | 0.90% | |
| Everything stackedAll reserves locked, the six-fold share, and high sensitivity, with the Federal Reserve abandoning its ample-reserves framework. Under scarce reserves banks are short of reserves, so each dollar of locked cash that returns to them supports several dollars of new loans instead of sitting in liquidity buffers. | 1.00 | 10% | 9 | Scarce | 531 | 4.42% | 14.3 | 0.14 | 129 | 6.66% |
Source: CEA, Table 3.
InteractiveExplore the model
Adjust the parameters of the model
This is the paper’s model, running in your browser from the paper’s own equations and calibration. Change a parameter and the results update. The presets load the assumptions behind each row of Table 3 and the alternative assumptions raised by critics. The paper’s table was computed with rounded intermediate values, so the live model can differ from the printed figures in the last digit.
Parameters
Advanced calibration
Defaults are the paper’s Table 1 calibration (H.8, FDIC QBP, FRED, February 2026). Banks choose their excess reserves optimally; with the paper’s log specification the buffer absorption is μ = E / ((iL − iH) εL L), about 1.2 at calibration. Reset with the Baseline preset.
What a yield ban does
Questions and answersFrequently asked questions
I. Reviewing Effects of Stablecoin Yield Prohibition on Bank Lending
1.What is the policy background motivating CEA’s analysis?
The GENIUS Act bars issuers from paying yield to stablecoin holders. There is a debate as to whether the ban should extend to affiliate or third party-arrangements. Many supporters of the extension of the prohibition claim that without it, deposits leave banks and lending falls by more than a trillion dollars.
2.What question does CEA address in the paper?
CEA considers the extent to which banning yield on stablecoins would prevent deposits from leaving banks and, correspondingly, protect bank lending. It also calculates the change in the loan rate and the cost of the policy to households.
3.Do stablecoin purchases cause “deposit flight”?
No. When a household purchases a dollar of a GENIUS-act compliant stablecoin, the household gives up a bank deposit and receives a dollar of stablecoin. The stablecoin issuer, directly or through an intermediary, must then hold safe reserve assets of equal value, such as a bank deposit or Treasury bills.
Importantly, the household’s deposit is not destroyed. Rather, the claim on bank funding is typically transferred. The economic effect depends on what asset mix the issuer holds as its reserves, and how the transaction is settled.
If the stablecoin issuer holds a dollar’s worth of Treasury bills, the seller of the Treasury bill receives a dollar bank deposit in exchange. In that case, deposits remain in the banking system, although they may move across account holders and across banks. Bank lending need not fall in the aggregate simply because a household switched from a deposit to a stablecoin.
If instead the issuer holds the reserve in the form of a bank deposit, the household’s deposit is again not destroyed, rather, it is replaced by a deposit held by the stablecoin issuer or its custodian bank. Aggregate deposits may shift across institutions, but they do not mechanically disappear. Any effect on bank lending comes from changes in the composition and distribution of funding and from bank’s liquidity, capital, and regulatory constraints, rather than from simple destruction of deposits.
Deposits can leave the banking system when funds settle into liabilities outside the commercial banking system, notably the Treasury General Account at the Federal Reserve. However, these outflows are generally temporary: when Treasury spends those balances, reserves return to banks and commercial bank deposits are recreated.
Deposits can also leave the system if stablecoin reserves are held directly as Federal Reserve liabilities. However, any resulting reserve drain is ultimately under the Fed’s control through monetary policy.
The model therefore tracks how many dollars a yield ban moves into or out of stablecoins, how those flows are allocated across reserve assets, how much deposit funding shifts across banks, and how those changes affect lending through banks’ balance-sheet constraints.
4.What does CEA find?
Suppose the stablecoin market is $300 billion prior to the ban. The paper finds that banning yield would move $54 billion from stablecoins into traditional commercial bank deposits. However, only $6.5 billion of the dollars moved from stablecoins to banks would represent marginal lending capacity. Moreover, since banks must hold reserves to meet regulatory obligations and customer demands, lending only increases by around $2.1 billion. In other words, the ban increases total bank lending by a mere 0.02% while having a large effect on demand for stablecoins. The effect is only a bit more pronounced for community banks, where lending increases by 0.03%.
Perhaps surprisingly, given the small changes to lending and the loan rate, the policy imposes a non-trivial cost on those who swap stablecoins for bank deposits following the ban. In total, households lose around $800 million per year, net of the lending gain. That amounts to a cost-benefit ratio of 6.6.
5.Is the model limited to considering the current market?
No. The model runs on the stablecoin share of deposits, not the dollar size of the market. In the paper, CEA presents results for a share as small as 1.7 percent, which is roughly today’s share, and as large as 10 percent (more than $1.7 trillion), which is about six times today’s share. These results are presented in Table 3.
CEA has also created an online tool, which allows one to adjust the relative share of the stablecoin market, the share of reserves stablecoins hold in cash, the share of liquid assets banks hold against issuer deposits, households’ sensitivity to yields, the share of bill proceeds that never return as lendable, the post-policy yield on stablecoins, and the Federal Reserve’s operating regime.
6.What conditions would generate a large increase in bank lending following a ban on yield?
The model implies that a ban on stablecoin yield would only result in a large increase in bank lending if (1) the stablecoin share of deposits were much greater than they are today (e.g, six times today’s level), (2) households were extraordinarily sensitive to yield, (3) stablecoin issuers held a much higher ratio of reserves in cash than they do today (e.g., 100 percent, compared with Circle’s roughly 12% and Tether’s roughly 0% today), and (4) the Federal Reserve were to abandon its ample-reserves framework. Under the extreme case where all of these conditions hold, CEA finds that a ban on yield would increase bank lending by 4.4 percent. However, this result depends on all of these conditions holding concurrently: relaxing any one of them sees the estimate fall considerably. For example, maintaining ample reserves would see the estimate fall to 0.6 percent.
7.Why is the welfare effect negative?
A yield ban is similar to a tax levied on stablecoin holders, except that the proceeds of the effective tax are collected by issuers and intermediaries instead of the government. Stablecoin holders respond to the effective tax by shifting the composition of their portfolio away from stablecoins and toward bank deposits, which are less desirable to them than the yield-paying stablecoins they would have otherwise held. The offsetting benefit is the increase in lending, and because only the locked share of reserves affects lending, that benefit is small.
II. Objections Raised by Bank Trade Groups and Advocacy Organizations
8.“CEA studied the wrong question. The issue is what happens if yield is allowed and the market scales.”
Yield is currently paid, primarily through exchange rewards. Some have argued that exchanges and other intermediaries should be prohibited from paying rewards. The policy-relevant question at the moment, therefore, is what happens if yield is prohibited.
Although CEA explicitly considered the effect of prohibiting yield, the same calculations show the effect of permitting yield. Figure 1 plots the change in stablecoin holdings and bank deposits across the entire range of stablecoin yields, from zero yield (completely prohibited) to competitive yield (completely permitted). Figure 2 plots the change in bank lending across the same stablecoin yield range. Reading Figure 1 and 2 from left to right reveals the change in bank deposits and lending as more and more stablecoin yield is permitted, up to the competitive yield.
CEA also considered what happens if the market scales. The shaded regions in Figures 1 and 2 range from a market share of 1.7 percent (about $300 billion, current share) to 10 percent (more than $1.7 trillion, roughly six times current share). Table 3 presents estimates for a range of stablecoin market shares, up to 10 percent.
9.“CEA assumes stablecoin market share is 1.7 percent.”
CEA’s baseline estimate assumes the stablecoin market share is 1.7 percent. However, the shaded regions in Figures 1 and 2 present estimates for a range of market shares, up to and including 10 percent (roughly six times current share). Table 3 also presents estimates for a range of stablecoin market shares, up to 10 percent.
10.“CEA assumes stablecoin issuers hold 12 percent of their reserves in cash based on one issuer’s cash holdings at one point in time.”
CEA reports the share of reserves Circle holds in cash at 12 percent. However, it also presents Tether’s reserves at nearly 0 percent and Treasury’s 2025 finding that major issuers hold most reserves in Treasuries.
Circle’s share was chosen for the baseline estimate because it was the higher of the two issuers surveyed. However, CEA also reports estimates for higher cash shares, all the way up to 100 percent, in Table 3.
CEA has also created an online tool, which allows one to adjust the share of reserves stablecoin issuers hold in cash (among other parameters).
11.“It is unreasonable to assume 88 percent of reserves recirculate. Bank deposits would have to equal the value of all financial assets.”
Buying an asset does not destroy a deposit. When a dealer sells a bill and buys another asset, it pays the seller of that asset with the same deposit he or she received by selling the bill. The deposit merely changes hands. A deposit only leaves the banking system if it goes to the Federal Reserve’s balance sheet or to the Treasury General Account. CEA explicitly identifies the Fed channel in the paper and notes that it applies to money market funds generally, not to stablecoins in particular.
12.“CEA acknowledges the scarce-reserves case in an appendix but declines to quantify it.”
CEA quantifies an extreme scarce-reserves case in the executive summary, Table 3, and the appendix. In the extreme scarce-reserves case, where stablecoin market share is 10% and stablecoin issuers hold 100% of their reserves in cash, banning yield increases total bank lending by $531 billion and community bank lending by $129 billion.
CEA did not quantify less extreme scarce-reserves cases in the paper. For example, if stablecoin issuers hold just 12% of their reserves in cash, the ban on yield increases bank lending by around $15 billion, or 0.13 percent of loans.
CEA has created an online tool that allows one to adjust the relative share of the stablecoin market, the share of reserves stablecoin issuers hold in cash, the share of liquid assets banks hold against issuer deposits, households’ sensitivity to yields, the share of bill proceeds that never return as lendable, and the post-policy yield on stablecoins under both the Federal Reserve’s prevailing ample-reserves operating regime and the alternative scarce-reserves operating regime.
13.“CEA finds that moving from ample to scarce reserves increases the effect of banning yield on bank lending from $2.1 billion to $531 billion.”
CEA reports its sensitivity analysis in Table 3. The reported increase from $2.1 billion to $531 billion reflects a change in several parameters, not merely the change from ample to scarce reserves. The sensitivity of estimates to the Federal Reserve’s operating regime can be seen by comparing the last two rows of Table 3, which share all parameter values except the Federal Reserve’s operating regime. In those cases ($1.7T stablecoin market, 100% reserves held as cash, and extreme yield sensitivity), the ban on yield increases lending from $72 billion under an ample-reserves regime and $531 billion under a scarce-reserves regime.
CEA has created an online tool that allows one to select either the Federal Reserve’s current ample-reserves operating regime or the alternative scarce-reserves operating regime and then consider the effect of banning stablecoin yield under a range of assumptions for the relative share of the stablecoin market, the share of reserves stablecoin issuers hold in cash, the share of liquid assets banks hold against issuer deposits, households’ sensitivity to yields, the share of bill proceeds that never return as lendable, and the post-policy yield on stablecoins.
14.“The ample-reserves assumption is doing enormous work. Relaxing it results in a 250-fold increase in bank lending.”
CEA reports its sensitivity analysis in Table 3. The reported increase from $2.1 billion to $531 billion reflects a change in several parameters, not merely the change from ample to scarce reserves. Replacing the ample-reserves assumption, while keeping the rest of the baseline in place, with a scarce-reserves assumption increases the effect of banning stablecoin yield on bank lending from $2.1 billion to around $15 billion, or 0.13 percent of loans.
CEA has created an online tool that allows one to select either the Federal Reserve’s current ample-reserves operating regime or the alternative scarce-reserves operating regime and then consider the effect of banning stablecoin yield under a range of assumptions for the relative share of the stablecoin market, the share of reserves stablecoin issuers hold in cash, the share of liquid assets banks hold against issuer deposits, households’ sensitivity to yields, the share of bill proceeds that never return as lendable, and the post-policy yield on stablecoins.
The Federal Reserve currently operates an ample-reserves regime. It has not expressed an intention to return to a scarce-reserves regime.
15.“CEA only considers the effect of banning yield in a scarce-reserves regime under the assumption that 88% of reserves recycle, which is implausible given their composition.”
CEA presented just one estimate of the effects of banning yield in a scarce reserve regime. That scenario, which is presented in Table 3, assumed stable coin issuers hold 100 percent of their reserves in cash, implying that none of their reserves recycle.
CEA has created an online tool that allows one to vary the share of reserves stablecoin issuers hold in cash (among other parameters) under both ample- and scarce-reserve operating regimes.
16.“CEA does not analyze what happens at the bank level when deposits migrate.”
CEA’s model has two bank types with separate reserve ratios, separate liquidity buffers, and a separate deposit share. CEA derived the community bank share of any lending change from FDIC and Federal Reserve data and reported results for community banks in every row of Table 3. It then explained why that share is an upper bound: reserves are custodied at large institutions; stablecoin holders skew young, urban, and higher-income; and published tests find no statistically significant relationship between USDC growth and community bank deposits.
17.“CEA gestures at narrow banking but ignores that narrow banking does not turn deposits into credit.”
CEA clearly stated that reserves held in cash (conservatively assuming the most unfavorable possible regulatory treatment of these deposits) do not support lending. It considered the scenario in Case 2, stating that the “strict regulatory treatment of […] stablecoin deposits restricts [a bank’s] ability to lend against” it. In Table 3, CEA presented two estimates of the effect of banning yield on bank lending under the assumption that stablecoin issuers hold 100% of their reserves in cash.
CEA has created an online tool that allows one to vary the share of reserves stablecoin issuers hold in cash (among other parameters).
18.“CEA does not consider stablecoin issuers with master accounts, which would break the model.”
In the model, a master account balance is equivalent to holding cash: it does not support bank lending. CEA presents estimates of the effect of banning stablecoin yield on bank lending under a range of assumptions about the share of reserves stablecoin issuers hold in cash or, equivalently, a master account. Furthermore, CEA has created an online tool that allows one to vary the share of reserves stablecoin issuers hold in cash or, equivalently, a master account (among other parameters).
19.“Recycled deposits are wholesale balances at large banks, not core deposits that fund local lending.”
In Case 1, CEA acknowledged that a deposit at a receiving bank may carry higher runoff assumptions and support less lending. But the effect on bank lending is small. Applying the 5 to 15 percent haircut used in the Federal Reserve staff note raises the estimated effect of banning stablecoin yield on bank lending from $2.1 billion to between $2.9 billion and $4.4 billion (up to 0.04% of total bank lending).
20.“CEA mischaracterized studies that find trillion-dollar effects.”
CEA discussed two studies that find significantly outsized effects of banning stablecoin yield on bank lending.
Andrew Nigrinis applies the substitution rate used by Toni Whited, Yufeng Wu, Kairong Xiao in the context of central bank digital currencies to stablecoins without tracing where displaced dollars go. A central bank digital currency is a liability of the Federal Reserve that drains deposits by construction.
Jessie Jiaxu Wang assumes recycling rates of 50, 20, and zero percent and states that the Treasury bill channel depends on whether counterparties ultimately deposit proceeds back into the banking system.
Both studies rely on a partial equilibrium model that identifies the lending contraction at a bank losing deposits while holding other banks’ balance sheets fixed. However, a general equilibrium model is required to consider whether the bank that gains the reshuffled deposits expands lending symmetrically.
In CEA’s model, a $1.5 trillion lending effect requires four extreme assumptions at once: the Federal Reserve abandons its ample-reserves framework, issuers hold 100% of reserves as bank cash, households are more sensitive to yield than the paper’s highest estimate, and the stablecoin market reaches about $4 trillion, roughly a quarter of all bank deposits. Under the Federal Reserve’s current framework no market size produces $1.5 trillion; the lending effect peaks near $300 billion. The outsized estimates offered by Nigrinis and Wang therefore rest on assumptions that cannot all hold.
III. Granting Each Objection
21.Grant a $3.7 trillion market, 21 percent of today’s deposits.
The effect scales with the share. At today’s cash composition it is roughly $20 billion, under 0.2 percent of loans. With 100 percent locked cash and implausibly high yield sensitivity it is roughly $130 billion, about 1.1 percent.
22.Grant that issuers hold 100 percent of reserves as locked cash.
$72 billion at a nearly $2 trillion stablecoin market under ample reserves, 0.6 percent of loans, and $17.5 billion at community banks. No issuer is anywhere near 100 percent, and regulatory treatment that penalizes issuer deposits pushes the cash share down, not up.
23.Grant that half of Treasury bill proceeds never return as deposits.
The locked share rises from 12 percent to 56 percent and the baseline effect rises from $2.1 billion to about $10 billion, 0.08 percent of loans.
24.Grant the wholesale-deposit haircut.
$2.9 billion to $4.4 billion at baseline.
25.Grant scarce reserves.
About $15 billion at today’s share and composition, 0.13 percent of loans.
26.Grant higher yield sensitivity.
Raising the substitution elasticity from 7 to 9 at a six-fold share (nearly $2 trillion stablecoin market) moves the effect from $11.1 billion to $14.3 billion.
27.Grant every objection at once.
$531 billion, 4.4 percent of loans. That requires the Fed to abandon its operating framework, every issuer to hold every reserve dollar as locked cash, the stablecoin share of deposits to rise six-fold, and households to sit at the top of the sensitivity range, all at the same time.
28.What about the assumptions that cut the other way?
The model treats every stablecoin dollar as a domestic deposit substitute. More than 80 percent of stablecoin activity is outside the United States, and a U.S. ban on intermediary rewards would have little to no effect on these holders. The model also treats banks as price takers. With deposit market power, banks would lower deposit rates as the prohibition pushes households toward them, shrinking the inflow further. Both mean $2.1 billion is, if anything, too high.
29.Does prohibition ever pass a cost-benefit test in the model?
Only if issuers hold more than about 60 percent of reserves as locked bank deposits. For comparison, Circle holds 12 percent. Tether holds nearly 0 percent.
Section IVThe Critics
ABA Banking Journal
What does the American Bankers Association argue, and what does the paper say?
The ABA raises three objections, of which each misreads the analysis or criticizes it for not answering a different question.
Sayee Srinivasan and Yikai Wang of the American Bankers Association (ABA) argue that the 1.) CEA’s analysis of “prohibiting” stablecoin yield is misguided and that the correct analysis would have been understanding the effect of “allowing” it. The authors also claim that 2.) the paper “avoid[s] the much more consequential scenario: yield-paying payment stablecoins scaling quickly.” Furthermore, the ABA authors’ claim that 3.) the CEA “does not analyze what happens at the bank level” and mistakenly report CEA’s headline effect of the increase in bank lending due to banning stablecoin yield as $1.2 billion.
Responding to each of these critiques:
Yield is currently paid, primarily through exchange rewards. Some have argued that exchanges and other intermediaries should be prohibited from paying rewards. The policy-relevant question at the moment, therefore, is what happens if yield is prohibited.
Therefore, although CEA explicitly considered the effect of prohibiting yield, the same calculations show the effect of permitting yield. Figure 1 plots the change in stablecoin holdings and bank deposits across the entire range of stablecoin yields, from zero yield (completely prohibited) to competitive yield (completely permitted). Figure 2 plots the change in bank lending across the same stablecoin yield range. Reading Figure 1 and 2 from left to right reveals the change in bank deposits and lending as more and more stablecoin yield is permitted, up to the competitive yield. The framing objection, therefore, lack substantive content.
While CEA’s baseline estimate assumes the stablecoin market share is 1.7 percent of bank deposits (about $300B), CEA also analyzes scenarios where stablecoin markets grow much larger, up to nearly $2 trillion.
The shaded regions in Figures 1 and 2 present estimates for a range of market shares, up to and including 10 percent (roughly six times current share). Table 3 also presents estimates for a range of stablecoin market shares, up to 10 percent. For comparison, 10 percent of deposits is about $1.7 trillion at today’s deposit base, which is within the ABA’s own $1 to $2 trillion range. Given the 10 percent share, CEA estimates the banning yield would increase bank lending by $11 to $14 billion, roughly 0.1% of outstanding bank loans. Accounting for the ABA’s own model of stablecoin growth, therefore, results in a net lending effect of one tenth of one percent of total bank loans.
CEA’s model also specifically evaluates the effects on community banks and in both ample and scarce reserves regimes. The model has two bank types with separate reserve ratios, separate liquidity buffers, and a separate deposit share. CEA derived the community bank share of any lending change from FDIC and Federal Reserve data and reported results for community banks in every row of Table 3. It then explained why that share is an upper bound: reserves are custodied at large institutions; stablecoin holders skew young, urban, and higher-income; and published tests find no statistically significant relationship between USDC growth and community bank deposits.
Comment letter to the FDIC
What does the Independent Community Bankers of America argue, and what does the CEA paper say?
In a June 9, 2026 comment letter to the FDIC, Brian Laverdure and Amy Ledig of the Independent Community Bankers of America (ICBA) claimed CEA relies on “today’s immature stablecoin market,” that “CEA’s analysis assumes that stablecoin-linked deposits as a share of total bank deposits will remain 1.7%,” and that “CEA treats community banks as outsiders in the stablecoin market.” In footnote 37, Laverdure and Ledig says the Federal Reserve’s abundant reserves framework is a consequence of post-2008 quantitative easing and is not a permanent feature of the banking system,” suggests “CEA finds that lending effects jump from $2.1 billion to $531 billion” under a scarce-reserves framework, and notes that the “jump occurs under the assumption that 88% of reserves recycle, which is implausible given their composition”
CEA’s baseline estimate assumed the stablecoin market share is 1.7 percent. However, CEA considered and presented estimates under much higher shares. The shaded regions in Figures 1 and 2 present estimates for a range of market shares, up to and including 10 percent (roughly six times current share). Table 3 also presents estimates for a range of stablecoin market shares, up to 10 percent.
CEA did not treat community banks as outsiders in the stablecoin market. It merely noted that, since cryptocurrency holders skew young, urban, and higher-income and since larger banks host more stablecoin business, the baseline estimates for the reduction in lending by community banks produced by CEA could be an upper bound. Interestingly, ICBA’s own scenarios do treat community banks as outsiders in the stablecoin market, noting that community banks “are not participants in T-bill secondary markets or repo settlement circuits and have no institutional mechanism to receive these deposits.” CEA maintains that a community bank does not have to buy a Treasury bill from an issuer to receive a deposit; the dealer’s customers and their payees bank somewhere and there is no reason to assume they could not bank at a community bank.
CEA’s baseline estimate assumed the Federal Reserve maintains its ample-reserves operating regime because the Federal Reserve currently operates an ample-reserves operating regime and has not expressed an intention to return to a scarce-reserves regime.
CEA presented just one estimate of the effects of banning yield in a scarce-reserve regime. The reported increase from $2.1 billion to $531 billion reflects a change in several parameters, not merely the change from ample to scarce reserves. The sensitivity of estimates to the Federal Reserve’s operating regime can be seen by comparing the last two rows of Table 3, which share all parameter values except the Federal Reserve’s operating regime. In those cases, the ban on yield increases lending from $72 billion under an ample-reserves regime and $531 billion under a scarce-reserves regime.
That scarce-reserves scenario, which is presented in Table 3, assumed stablecoin issuers hold 100 percent of their reserves in cash, implying that none of their reserves recycle.
There are other problems with Laverdure and Ledig’s analysis. Notably, the reserve composition they cite—including secured loans, precious metals, and bitcoin—describes an offshore issuer’s portfolio; those are not permitted reserves under the GENIUS Act. The ECB working paper Laverdure and Ledig cite for nonlinearity studies euro area banks. CEA’s own six-fold adoption case keeps the lending effect under 1 percent.
Newsletter article
What does Better Markets argue, and what does the CEA paper say?
In an April 16, 2026 newsletter article, Phillip Basil, Christopher Appel, and Amanda Fischer of Better Markets write that 1.) the 12 percent cash share “is based on a single point-in-time piece of data from one current stablecoin issuer,” that 2.) “it is unreasonable to assume that the 88 percent of stablecoin reserve holdings in the form of Treasuries and repos will entirely recirculate through the banking system,” because otherwise 3.) “bank deposits would roughly equal the value of all financial assets,” and that “all deposits and loans are not created equal.”
Responding to these:
The two largest stablecoin issuers are Circle and Tether, which encompass approximately 85% of the stablecoins market cap. Tether’s main product, USDT is currently not GENIUS compliant, though Tether is partnering with Anchorage Digital Bank to issue its U.S. market stablecoin USA₮ (USAT). Tether holds a negligible share of its reserves in bank deposits (less than 4%) Circle has historically maintained a ratio of bank deposits to total assets between 10% and 15% since 2024. Therefore, the 12% figure the CEA is using is not merely a single point in time but a reasonable assumption about how Circle has managed its business so far. CEA also reported estimates for higher cash shares, all the way up to 100 percent, in Table 3.
The recirculation argument confuses buying an asset with destroying a deposit. Buying an asset does not destroy a deposit. When a dealer sells a bill and buys another asset, it pays the seller of that asset with the same deposit he or she received by selling the bill. The deposit merely changes hands. A deposit only leaves the banking system if it goes to the Federal Reserve’s balance sheet, or to the Treasury General Account. CEA explicitly identifies the Federal Reserve channel in the paper and notes that it applies to money market funds generally, not to stablecoins in particular.
Basil, Appel, and Fischer correctly argue that the stock of cash does not equate the stock of financial assets. For example, a person can buy a $500,000 house by contributing only a small percentage of the house value as collateral for the loan that pays for the house. In a similar way, financial institutions can leverage their holdings. Obviously, aggressive lending creates credit risk. Incorporating credit risk in the CEA model would tilt the welfare gains in favor of having a competitive stablecoin industry, because GENIUS compliant issuers carry a small credit risk by construction.
Comparing the stock of bank deposits to the stock of all financial assets says nothing about where the marginal dollar goes. CEA explicitly considers the point that deposits are not all alike is the Case 1. Applying the Federal Reserve staff note’s haircut increases the net effect of banning stablecoin yield on borrowing from $2.1 billion to at most $4.4 billion.
A Model Built to Mislead
What do Americans for Financial Reform argue, and what does the CEA paper say?
In a May 5, 2026 Americans for Financial Reform white paper titled A Model Built to Mislead, Lee Reiners and Mark Hays question several assumptions in CEA’s analysis. They say the “relevant question is not about today’s equilibrium, it is about what happens to bank deposits and lending if stablecoins scale” and note that “even the more conservative public forecasts contemplate substantial growth” in stablecoins. They assert that “deposits leaving [a] Bank are far more likely to originate from community banks.” They also claim the “`ample reserves’” framework […] assumption is doing enormous work;” and that CEA “briefly acknowledges this possibility in an appendix but declines to quantify it.” Reiners and Hays say “CEA further assumes that Treasuries and overnight reverse repurchase agreements (repos) have equivalent effects on bank deposits,” which they describe as “a significant oversimplification.” Lastly, they say CEA “fails to account for potential policy changes that could significantly amplify these effects,” most notably “granting stablecoin issuers access to Federal Reserve master accounts.”
CEA explicitly considered what happens if the market scales. The shaded regions in Figures 1 and 2 range from a market share of 1.7 percent (about $300 billion, current share) to 10 percent (more than $1.7 trillion, roughly six times current share). Table 3 presents estimates for a range of stablecoin market shares, up to 10 percent.
The public forecasts cited by Reiners and Hays identify deposit accounts that are not currently denominated in US Dollars as the largest source of growth for stablecoin. CEA did not model this margin, but noted that it cuts against the lending rationale for banning stablecoin yield: any fiscal benefit from foreign-financed Treasury demand that is lost due to the prohibition on paying stablecoin yield may offset whatever gain in domestic bank lending the prohibition achieves. Including the foreign-source demand for stablecoins, as Reiners and Hays implicitly suggest, would result in a significant overestimate of the effect of banning stablecoin yield on domestic bank lending.
The assertion that deposit outflows are far more likely to originate from community banks is unfounded. As CEA noted, reserves are custodied at large institutions; stablecoin holders skew young, urban, and higher-income; and published tests find no statistically significant relationship between USDC growth and community bank deposits.
CEA’s baseline estimate assumed the Federal Reserve maintains its ample-reserves operating regime because the Federal Reserve currently operates an ample-reserves operating regime and has not expressed an intention to return to a scarce-reserves regime.
CEA quantifies an extreme scarce-reserves case in the executive summary, Table 3, and the appendix. In the extreme scarce-reserves case, where stablecoin market share is 10% and stablecoin issuers hold 100% of their reserves in cash, banning yield increases total bank lending by $531 billion and community bank lending by $129 billion.
CEA did not quantify less extreme scarce-reserves cases in the paper. But the ample-reserves framework is not “doing enormous work.” For example, if stablecoin issuers hold just 12% of their reserves in cash, in a scarce reserves framework the ban on yield increases bank lending by around $15 billion, or 0.13 percent of loans.
CEA considers the potential effects of the repos channel in Case 3. Furthermore, the claim by Reiners and Hays that stablecoin reserves would flow into the ON RRP facility (by means of an allowed Money Market Fund) “where they face virtually no counterparty risk” bolsters CEA’s claims about the effect of banning stablecoin yield on the welfare of households. CEA did not consider counterparty risk in its model. If it had, and if households place funds in stablecoin to hedge against credit risk, the effective tax on stablecoins would have produced a larger reduction in households’ welfare.
In CEA’s model, a master account balance is equivalent to holding cash: it does not support bank lending. CEA presents estimates of the effect of banning stablecoin yield on bank lending under a range of assumptions about the share of reserves stablecoin issuers hold in cash or, equivalently, a master account. Reiners and Hays concedes that “the CEA’s paper contemplates a related scenario.”
Big Assumptions, Bigger Impacts; op-ed in Open Banker
What does the Consumer Bankers Association argue, and what does the CEA paper say?
In a report published by the Consumer Bankers Association on April 16, 2026 and an op-ed in Open Banker on May 14, 2026, Andrew Nigrinis claims CEA’s analysis rests on five assumptions: the market is “still small enough to make the issue marginal,” the stablecoin-to-deposit ratio “will remain low enough,” “large banks and community banks are basically the same,” “money staying in the banking system means lending is preserved,” and “the ample-reserves regime is the right baseline.” Nigrinis is the author of an October 2025 study claiming the ban on stablecoin yield would reduce bank lending by $1.5 trillion.
CEA does not assume the stablecoin market will remain small. Although CEA’s baseline estimate assumed the stablecoin market share is 1.7 percent, the shaded regions in Figures 1 and 2 present estimates for a range of market shares, up to and including 10 percent (roughly six times current share). Table 3 also presents estimates for a range of stablecoin market shares, up to 10 percent. On its own, increasing the stablecoin market share does not significantly increase the reduction in bank lending following the ban on stablecoin yield.
CEA did not assume community and large banks are basically the same. CEA’s model has two bank types with separate reserve ratios, separate liquidity buffers, and a separate deposit share. Moreover, CEA’s assumption that the proportion of deposit flows to community banks is proportional to its share of total deposits likely generates a larger effect on net lending for community banks. As CEA noted, reserves are custodied at large institutions; stablecoin holders skew young, urban, and higher-income; and published tests find no statistically significant relationship between USDC growth and community bank deposits.
CEA did not claim that money staying in the banking system means lending is preserved. To the contrary, its model explicitly accounts for the effects of regulatory treatment on lending capacity. Because 12% of stablecoin issuers’ deposits are held as deposits, which CEA conservatively assumes are subject to the most stringent possible regulatory treatment, CEA discounts these as lost lending capacity even though it is money that remains in the banking system.
CEA’s baseline estimate assumed the Federal Reserve maintains its ample-reserves operating regime because the Federal Reserve currently operates an ample-reserves operating regime and has not expressed an intention to return to a scarce-reserves regime.
CEA quantifies an extreme scarce-reserves case in the executive summary, Table 3, and the appendix. In the extreme scarce-reserves case, where stablecoin market share is 10% and stablecoin issuers hold 100% of their reserves in cash, banning yield increases total bank lending by $531 billion and community bank lending by $129 billion.
As Nigrinis acknowledges, “the $531 billion result appears only in a scarce-reserves scenario that also relaxes several other assumptions at the same time, including scale and substitutability. The comparison therefore does not show the effect of reserve scarcity alone.” CEA did not quantify less extreme scarce-reserves cases in the paper. But Nigrinis is incorrect in claiming “the estimated lending effect rises sharply” when “the ample-reserves assumption is relaxed.” For example, if stablecoin issuers hold just 12% of their reserves in cash in a maximally scarce reserves regime, the ban on yield increases bank lending by around $15 billion, or 0.13 percent of loans.
What does the Progressive Policy Institute argue, and what does the CEA paper say?
In an April 14, 2026 response published by the Progressive Policy Institute, Paul Weinstein Jr. accuses CEA of “a cynical attempt to cloud the results” of existing studies. He claims “two things can be true: that a prohibition on yield would not significantly increase lending by banks, but allowing stablecoins to offer interest would significantly increase deposit outflows from banks.”
CEA discussed two studies that find significantly outsized effects of banning stablecoin yield on bank lending.
Andrew Nigrinis applies the substitution rate used by Toni Whited, Yufeng Wu, Kairong Xiao in the context of central bank digital currencies to stablecoins without tracing where displaced dollars go. A central bank digital currency is a liability of the Federal Reserve that drains deposits by construction.
Jessie Jiaxu Wang assumes recycling rates of 50, 20, and zero percent and states that the Treasury bill channel depends on whether counterparties ultimately deposit proceeds back into the banking system.
Both studies rely on a partial equilibrium model that identifies the lending contraction at a bank losing deposits while holding other banks’ balance sheets fixed. However, a general equilibrium model is required to consider whether the bank that gains the reshuffled deposits expands lending symmetrically.
In CEA’s model a $1.5 trillion lending effect requires four extreme assumptions at once: the Federal Reserve abandons its ample-reserves framework, issuers hold 100% of reserves as bank cash, households are more sensitive to yield than the paper’s highest estimate, and the stablecoin market reaches about $4 trillion, roughly a quarter of all bank deposits. Under the Federal Reserve’s current framework no market size produces $1.5 trillion; the lending effect peaks near $300 billion. The outsized estimates offered by Nigrinis and Wang therefore rest on assumptions that cannot all hold.
Contra Weinstein, CEA maintains that it cannot simultaneously be true that a prohibiting stablecoin yield would not significantly increase bank lending but permitting stablecoin yield would significantly increase bank deposit outflows. Yield is either permitted or prohibited. In a model designed to consider the effect of paying stablecoin yield on bank lending, the effect of permitting yield is equal to the effect of prohibiting yield with the sign reversed. Figure 1 plots the change in stablecoin holdings and bank deposits across the entire range of stablecoin yields, from zero yield (completely prohibited) to competitive yield (completely permitted). Figure 2 plots the change in bank lending across the same stablecoin yield range. Reading Figure 1 and 2 from left to right reveals the change in bank deposits and lending as more and more stablecoin yield is permitted, up to the competitive yield. Reading Figure 1 and 2 from right to left reveals the change in bank deposits and lending as more and more stablecoin yield is prohibited, up to the complete prohibition of yield.
Weinstein offers no alternative model or estimates to support the claim that the magnitude of the effect of prohibiting yield differs from that of permitting it. He also fails to identify a study producing differential effects.
What does Vance Ginn argue, and what does the CEA paper say?
In an April 22, 2026 op-ed published by RealClearMarkets, Vance Ginn claims CEA anchors its estimates on today’s stablecoin market of $300 billion, in spite of various other research which points to stablecoin growth between $1 and $2 trillion. Ginn also argues that deposits are more likely to flow out of community banks and into large banks, so the reserve recycling described in the CEA paper will more substantially impact “farmers or local manufacturers” who may lose “access to community bank credit.” He also claims stablecoin reserves will “tend to reappear as wholesale balances and custody flows, not as sticky core deposits that actually power local lending.”
CEA’s baseline estimate assumes the stablecoin market share is 1.7 percent, but it also considers scenarios where stablecoin growth results in a much bigger market share and the stablecoin market is nearly $2 trillion. The shaded regions in Figures 1 and 2 present estimates for a range of market shares, up to and including 10 percent (roughly six times current share). Table 3 also presents estimates for a range of stablecoin market shares, up to 10 percent. For scenarios that maintain a ratio of stablecoin reserves in bank deposits equal to the baseline scenario (12%), the increase in bank lending following a ban on stablecoin yield when stablecoins are 10% of the deposit base is between $11 to $14 billion. The larger estimate represents a mere 0.12% of outstanding bank loans. Under the same scenarios, the lending effect on community banks tops out at 3.5 billion, or 0.18% of their loan books.
The assertion that deposit outflows are far more likely to originate from community banks is unfounded. As CEA noted, reserves are custodied at large institutions; stablecoin holders skew young, urban, and higher-income; and published tests find no statistically significant relationship between USDC growth and community bank deposits.
CEA’s “Case 1” explicitly considered the effects of banning stablecoin yield when stablecoin reserves reappear as wholesale balances and custody flows, noting that the decrease in lending capacity comes almost exclusively from the different regulatory treatment of cash reserves (wholesale deposits) that require more high quality liquid assets to back them than other reserves. But, as CEA noted, only around 12% of issuer reserves is held as wholesale deposits, meaning that for every $1 in stablecoins purchased, only $0.12 becomes wholesale deposits.