June 22, 2026

Retail Sweep Behavior Since the Elimination of Reserve Requirements¹

Mary-Frances Styczynski

Introduction

Retail sweeps emerged in the 1990s as a way for depository institutions (DIs) to reduce the cost of satisfying federally mandated reserve requirements.² Retail sweeps grew, even as the cost of satisfying reserve requirements was minimized with the beginning of interest on reserves in 2008. It wasn't until reserve requirements were effectively eliminated in March 2020 that retail sweep programs began to unwind. This note tracks the rise of sweep programs to the beginning of their unwind. The note goes on to predict how the further unwind of sweep programs may play out and how the changes affect the U.S. monetary aggregates.

Background

Reserve requirements are a monetary policy tool that the Federal Reserve used prior to March 2020 to establish demand for balances held at the Federal Reserve. Up until March 15, 2020, the Federal Reserve required DIs to maintain reserve requirements, or a proportion of their reservable liabilities in the form of vault cash or balances in an account at a Federal Reserve Bank. As vault cash and balances at the Federal Reserve earned no interest prior to 2008, DIs developed strategies over time to minimize their reserve requirements. A popular strategy to minimize reserve requirements emerged in the mid-1990s, called the retail sweep.

Retail sweep programs remained popular even after the Federal Reserve took steps to reduce the cost of holding reserve requirements. In 2008, the Federal Reserve began to pay interest on reserve balance requirements- the portion of reserve requirements not covered by vault cash. In theory, the payment of interest should have diminished the benefit to a DI of engaging in reserve avoidance. In practice, however, the Federal Reserve did not observe any meaningful change in sweep activity.

Mechanics of a retail sweep

In a retail sweep, a DI dynamically reclassifies funds from customers' accounts that are subject to reserve requirements (such as checking accounts) to those that are not subject to reserve requirements (like savings accounts), while leaving the customers' views of their liquidity at the DI completely unchanged.³ If operated properly, a retail sweep program should reduce a DI's reserve requirement without interfering with its customers' day-to-day activity.⁴ So, a customer's balance would remain unchanged while on the DI's books and records, the funds would have moved from checking to savings. The deposits on the DI's books and records would then be reported on the FR 2900 report form, which is collected by the Federal Reserve to compute reserve requirements and measure the monetary aggregates.⁵ The FR 2900 data would reflect the impact of retail sweeps rather than what the DIs communicate to their customers. Other regulatory reports, such as the quarterly reports of financial condition (Call Reports) would also be affected by retail sweeps.⁶

Retail Sweep Trends 1990s to March 2020

While there are no official statistics on retail sweeps, the Federal Reserve used to maintain estimates of initial amounts swept by DIs that reported the FR 2900 from January 1994 to April 2012. Over this period, the Federal Reserve informally gathered data on the type, amount, and start date of a DI's retail sweep program. I use this data to measure the number of DIs that started sweeping and the dollar value of the sweep for each year in this period. The data suggest that, over time, retail sweep programs grew in popularity (Figure 1) and DIs swept larger and larger amounts of deposits from checking accounts (also referred to as demand deposits) to savings (Figure 2). By the end of the data series in April 2012, over 3,000 DIs had started a retail sweep program, and cumulative, monthly average sweeps totaled $800 billion. To put this number in perspective, in April 2012, total demand deposits averaged around $2 trillion.⁷

These estimates are based on initial amounts swept at the start of a retail sweep program. The amount swept each month after the initial sweep is expected to grow over time as DIs deposit base expands. So, by March 2020, when reserve requirements were eliminated, the total value of DIs retail sweeps was likely much higher than $800 billion.

Retail Sweep Behavior Since April 2020

In March 2020, the Board of Governors effectively eliminated reserve requirements on all DIs by lowering reserve ratios applied to transaction accounts to zero.⁸ This signaled to the banking industry that reserve requirements are not an important tool for monetary policy implementation in an ample reserves regime. From a DI's perspective, the sole benefit from engaging in retail sweeps disappeared and DIs were left only with the costs of maintaining the reserve requirement avoidance infrastructure. Some DIs decided to end their retail sweep programs while others continued to rely on their sweep programs. In the next section, I attempt to quantify the amount of retail sweep programs that were ended.

Identifying Retail Sweepers and their Status

To assess the status of retail sweeping activity, I identify the DIs engaged in retail sweeping, and within that group, identify the DIs that ended their retail sweep programs since April 2020-the first full month when reserve requirements were set to zero. My identification process is as follows. I start with a list of DIs that file the FR 2900 (n ~ 1,000).⁹ I cross reference this list with a list of DIs that had initiated a sweep program at some point between January 1994 and April 2012. I also examine text remarks that are submitted by DIs with their FR 2900 reports to explain unusual movements in their data. The lifecycle of a retail sweep-while routine-does result in unusual movements in FR 2900 data that must be explained. I analyze text remarks received from April 2020 to February 2026 for mentions of retail sweep activity.¹⁰ Based on these steps, I identify 560 DIs that have been engaged in retail sweeping at some point from April 2020 to February 2026. This represents around 60 percent of the DIs that file the FR 2900. This identification approach likely undercounts the true population of DIs with retail sweep programs. Currently, there are over 8,000 DIs in operation in the U.S. and not all of them file the FR 2900. While my approach likely undercounts the total number of DIs sweeping, it captures the largest deposit holders in the United States that are likely to be sweeping a large volume of deposits.

Using the text remarks, I'm also able to identify DIs that have ended their retail sweep programs. Of the 560 DIs that I identify as being sweepers between April 2020 and February 2026, 175 DIs have discontinued their retail sweep programs over the same time period (Table 1). The decommissioning of retail sweep programs did not happen all at once (Figure 3). In 2021, there was a distinct increase in the number of DIs ending their retail sweep programs. Many of these DIs could have decommissioned their retail sweep programs at the same time as they were modifying their reporting infrastructure for the revised FR 2900 report form, which took effect April 6, 2021. The infrastructure supporting retail sweeps-whether designed internally or based on third-party software-is often embedded deeply in a DI's day-to-day processes and requires careful planning to turn off. Several DIs noted that they timed the ending of their sweep program to minimize operational risks and avoid complications with reporting. A number of DIs continue to engage in retail sweeps despite there being no economic incentive to do so. The reason for the persistent sweeping is not well understood. Retail sweeps may persist today because these programs are embedded deeply in DIs back-end processing and ending them would require time and resources to implement.

Table 1. Stock of Banks Engaged in Retail Sweeping

Leveraging the information in text remarks, I am able to estimate the amount of deposits that were reclassified from savings to demand deposits when a DI ended their retail sweep program. I find that between April 2020 and February 2026 about $1.5 trillion in savings deposits were reclassified as demand deposits due to the end of a retail sweep program (Table 2). Large DIs, those with deposits greater than or equal to $20 billion, account for most of the reclassified balances. This finding is consistent with large DIs having a larger deposit base from which to sweep.

Table 2. Summary Statistics for Depository Institutions that Discontinued Retail Sweep Programs

Table 3 presents the reclassification amounts by type of DI. DIs are grouped into three categories: commercial banks, which include U.S. branches and agencies of foreign banks; thrifts, which cover federal savings banks and savings and loan institutions; and credit unions. Commercial banks account for over 80 percent of the total number of DIs that ended their sweep programs and the total amount reclassified.

Table 3. Breakdown Depository Institutions that Ended Sweep Program by Type

Figure 4 plots the cumulative value of reclassified deposits by month. There are three distinctive increases in the time series: May/June 2022, November 2020, and November 2025. The reclassifications were broad based-more than one DI discontinued its retail sweep program at each point. However, all three events involved a large DI ending their sweep program.

Retail Sweeps in the Future

It is challenging to know if and when the remaining DIs will cease their retail sweep programs and by how much. I attempt to quantify the amount left to be reclassified by DI size based on data from Tables 1 and 2. For the lower range of this estimate, I multiply the number of DIs that continue to sweep by the value of deposits reclassified at the 25th percentile. The top of the range is generated the same way except I use the amount of funds reclassified for the DI at the 75th percentile. Based on this approach, the amount of funds to be reclassified because of the discontinuance of retail sweep programs are estimated to range between $526 billion and $2.4 trillion. The amount left to be reclassified is likely to be closer to the top of the range because a number of large DIs-those with deposit levels greater than $20 billion-have yet to discontinue their sweep programs. Overall, the estimates suggest that the volume of deposits to be reclassified from savings to demand deposits is likely to be significant.

Impact on the Monetary Aggregates

As discussed earlier, retail sweeping affects how DIs report their deposits on the FR 2900. The FR 2900 is a main input into the calculation of the U.S. monetary aggregates, which are made available to the public on the H.6 Statistical Release. For the aggregates, total M1 and M2 are unaffected by retail sweeps. However, the components of M1, demand deposits adjusted and other liquid deposits, would be affected by the reclassification.

Savings deposits are captured in the other liquid deposit component of M1. Demand deposits balances are included in either demand deposits adjusted or other liquid deposits based on the type of DI issuing them. If demand deposits are issued by commercial banks, then these deposits are included in the demand deposit adjusted component of M1. If they are issued by thrifts or credit unions, then they are aggregated with savings in the other liquid deposit component of M1. Given this configuration, the discontinuance of a retail sweep program will be visible at the component level if it involves a commercial bank.

Based on my estimates, commercial banks reclassified $1.2 trillion of savings deposits to demand deposits between April 2020 and February 2026. Over the same period, total demand deposits adjusted as reported on the H.6 statistical release increased by $4.8 trillion. Taken together, 25 percent of the rise in demand deposits adjusted can be explained by the unwind of retail sweep programs rather than meaningful economic activity. The impact of the reclassification on other liquid deposits is even larger. For data through February 2026, other liquid deposits decreased by $2.1 trillion and $1.2 trillion of this was due to the decommissioning of a retail sweep program.

As noted earlier, other data sources, such as the Call Reports, are affected by retail sweep activity. The unwinding of a retail sweep program will be visible in other statistical releases that rely on the Call Reports, such as the flow of funds (Z.1 statistical release). My estimates of the amount of funds reclassified can be used to differentiate between activity related to the ending of a retail sweep and customer activity.

Conclusion

Since April 2020, DIs have been reassessing the need to maintain their retail sweep programs-a popular tool for minimizing reserve requirements. Some DIs have ended their programs. Based on my estimates, 175 DIs have ended their retail sweep programs, reclassifying around $1.5 trillion from savings deposits to demand deposits. Many DIs continue to sweep despite the change in incentives. A back of the envelop calculation suggests that as much as $2.4 trillion may be left to reclassify. Tracking changes in retail sweeps is important because the Federal Reserve makes data on aggregate savings deposits and demand deposits available to the public through the publication of the monetary aggregates. Other data releases, such as the flow of funds, are also affected. This note helps shed some light on DI behavior that is driving movements in data reported to the public.