Every time someone sends you money over payment apps and services such as Zelle, Stripe, PayPal, or Google Pay, it’s of course nice to receive it almost immediately. At first sight, it improves convenience and doesn’t hurt anybody — money merely moves from one bank to another, without draining deposits, which stay within the overall banking system. Not surprisingly, with no apparent risk or obvious displacement of deposits in sight, the instant payments space is not usually under the spotlight for regulation.
But with the increasing demand for instant payments, banks must ensure they have enough liquidity at all times, which opens the door to unintended costs and risks. An important aspect here is the extent to which a bank is able to match its instant payment obligations with receipts from other banks, or net payments, typically on a daily basis. Without instant payments, banks can defer the settlement of transfer requests until the end of the day or even several days later. Under instant payments, however, such an ability to delay outgoing payments and net them against incoming payments is gone, thereby complicating how banks manage their balance sheets.
In a recent paper titled “The Effect of Instant Payments on the Banking System: Liquidity Transformation and Risk-Taking,” Wharton finance professor Yao Zeng and his co-authors showed how the growth of instant payments forces banks to park more of their deposits in high-liquidity securities such as government bonds and cash and, perhaps more surprisingly, taking more risks in lending at the same time. Zeng’s co-authors are MIT doctoral student Ding Ding, Banco Central do Brasil (Brazil’s central bank) senior economist Rodrigo Gonzalez, and Columbia University finance professor Yiming Ma.
According to the paper, when banks place more of their deposits in liquid assets, they have less of that money available for lending. In that sense, banks become “narrower.” That subtle displacement also has a more perverse incentive, especially for smaller and less profitable banks: With more of their deposits in liquid assets, they enjoy the buffers and are hence more tempted to take on excessive risk in lending, which could imperil the larger economic system. “There hasn’t been much understanding about the costs or risks that instant payments would pose on the banking and economic systems,” Zeng said.
The paper warned that banks will face more pressure as stablecoins and tokenization gain in popularity, as they economically carry functions similar to instant payments. “Our findings have important implications for the adoption and design of new payment technologies, including CBDCs, stablecoins, and tokenization,” the authors wrote.
The authors made their case after analyzing data from Brazilian banks’ use of Pix, one of the most widely used instant payment systems. They ran that data through with a theoretical model and a novel instrument they constructed to identify the constraints that instant payments inflict on banks.
The study’s dataset spans five years from January 2018 to March 2023, while its primary analysis focused on the period from Pix’s implementation in November 2020 through March 2023. That period included the Covid pandemic, and the macroeconomic conditions prevailing at the time could also have influenced the liquidity shifts at banks, Zeng pointed out.
“The increase in risk-taking is concentrated among less well-capitalized banks, where the incentives to ‘reach for yield’ are strongest.”— Yao Zeng
What the Pix Lens Revealed
The study first compared the top 25% of banks in Brazil whose customers use Pix the most, or the banks most exposed to instant payments, which it called “high-Pix banks,” to those whose customers use Pix less. The study found significant, large, and visible balance sheet changes at high-Pix banks over a relatively short period.
The study found that Pix placed significant pressure on bank liquidity by eroding the deposit netting ability. For the median bank in the average month, about 71% of total payment flows would have been nettable if they did not have to make instant payments.
With such pressures, the liquid holdings of high-Pix banks rose substantially from about 8% in September 2020, right before the introduction of Pix, to more than 14% by May 2022. Over the same period, total liquid assets including cash, that is, central bank reserves, plus government bonds at those banks, rose from roughly 14% to more than 20% of assets, indicating a substantial shift toward safer, more liquid balance sheets.
On the liability side, checking deposits at high-Pix banks increased from about 10% of assets in September 2020 to around 15% by mid-2022. That reflected the increasing preference among bank customers to keep funds in their checking deposit accounts that can be used for instant payments.
The reduced deposit netting ability led to “narrower” banks that have more liquid assets and less illiquid assets such as loans to businesses, for instance. Over the same period, illiquid holdings of high-Pix banks declined from 86% to below 80%.
“Despite becoming narrower and holding more liquid buffers, risky loan lending became noticeably riskier at these high-Pix banks,” Zeng said. The study showed that within illiquid loans, the share of prime loans declined from about 72% in 2020 to roughly 60% by mid-2022. At the same time, the share of subprime loans rose from around 22% to about 35%, reflecting a tilt in lending to riskier borrowers.
From Safe Assets to Risky Lending
Why would a bank with safe, liquid assets make riskier loans? “When a bank holds a larger buffer of safe assets such as reserves or government bonds to handle everyday payment needs, it is also better protected against losses in risky lending. This cushion makes the bank less vulnerable if some loans go bad,” Zeng said. “But precisely because of that protection, the bank has less to lose from taking risks. As a result, it becomes more willing to lend to riskier borrowers in search of higher returns.”
The tilt towards high-risk lending was clearly visible in the study’s findings. A one-standard-deviation increase in Pix usage resulted in a 21.8 percentage points decrease in the ratio of prime loans. At the same time, the ratio of sub-prime loans rose by 18.6 percentage points. That suggests “a remarkable increase in riskier lending,” Zeng said.
That effect depends strongly on how well capitalized a bank is. For banks with weaker capital positions, the shift toward riskier lending is large and pronounced, Zeng noted. In contrast, for banks with strong capital buffers (or those above the median level), the effect of Pix usage on lending risk was almost entirely offset, and their loan portfolios changed little in response. In other words, “the increase in risk-taking is concentrated among less well-capitalized banks, where the incentives to ‘reach for yield’ are strongest,” he added.
“There hasn’t been much understanding about the costs or risks that instant payments would pose on the banking and economic systems.”— Yao Zeng
Orange Alert for Stablecoins
The study’s findings have broader implications on stablecoins, tokenization, and regulation, the paper noted. Every time a user redeems a stablecoin, the issuer typically has to send funds back through the banking system, often by drawing on bank deposits or selling safe assets such as government bonds, Zeng explained.
“If redemptions happen quickly and at scale, they can generate the same kind of immediate and unpredictable outflows that we study in the context of Pix,” Zeng said. “As deposit-backed stablecoins become potentially mainstream, our results suggest that banks and intermediaries will respond by holding more liquid assets and adjusting their lending with potential consequences for credit supply and risk-taking.”
The GENIUS Act of July 2025 facilitated the rapid rise of deposit-backed stablecoins and tokenized deposits. Although these new forms of money do not directly crowd out bank deposits, they may still constrain banks’ liquidity transformation and increase bank risk, the paper noted.
Tensions in the Financial System
The growth of instant payment systems has an impact on how the Federal Reserve goes about monetary policy design. Zeng pointed to conflicting pressures on this front. On the one hand, the Fed has been trying to reduce the size of its balance sheet through quantitative tightening (QT), effectively draining reserves from the banking system. At the same time, there is a parallel push, including through instant payment systems like FedNow and broader payment innovations, to make payments faster and more immediate. FedNow is used mostly by institutions, corporations, and the government, although individuals could also use it.
“Our findings suggest that these two forces may work in opposite directions: Faster payments increase banks’ demand for liquidity, while QT reduces the supply of liquidity,” Zeng said. “Even though both policies are well motivated on their own, taken together, they may create unintended tensions in the system.
The broader implication is that payment design, financial innovation, and monetary policy are more tightly linked than they may appear, Zeng noted. “Changes that improve payment speed and convenience can reshape banks’ balance sheets and their demand for liquidity, with downstream effects on lending, risk-taking, and financial stability.” The takeaway from that is that policymakers working on payments and those managing the central bank balance sheet should think jointly about these interactions, rather than in isolation, he added.
