Policy Rate Uncertainty and Money Market Funds (MMF) Portfolio Allocations

Finance and Economics Discussion Series Federal Reserve Board, Washington, D.C. ISSN 1936-2854 (Print) ISSN 2767-3898 (Online) Policy Rate Uncertainty and Money Market Funds (MMF) Portfolio Allocations Samin Abdullah, Manjola Tase 2025-063 Please cite this paper as: Abdullah, Samin, and Manjola Tase (2025). “Policy Rate Uncertainty and Money Market Funds (MMF) Portfolio Allocations,” Finance and Economics Discussion Series 2025-063. Washington: Board of Governors of the Federal Reserve System, https://doi.org/10.17016/FEDS.2025.063. NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the Finance and Economics Discussion Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

Policy Rate Uncertainty and Money Market Funds (MMF) ∗ Portfolio Allocations Samin Abdullah† Manjola Tase‡ July 23, 2025 Abstract We find that an increase in policy rate uncertainty is associated with an increase in MMF portfolioallocationstowardsassetswithshorter-datedmaturities. Wealsofindthatthedirection ofuncertaintymatters: MMFportfoliomaturityismoresensitivetouncertaintywhenitrelates to changes in expectations for a larger increase or a smaller decrease in the policy rate than when it relates to changes in expectations for a smaller increase or a larger decrease in the policy rate. Furthermore, for MMF that are eligible to participate at the Federal Reserve’s Overnight Reverse Repurchase Agreement (ON RRP) facility, we find that when policy rate uncertainty increases, MMF adjust their portfolio composition by increasing their take-up at the facility. This suggests that the ON RRP facility helps smooth fluctuations in short-term funding markets. Keywords: money market funds, portfolio allocations, monetary policy expectations, uncertainty, Federal Reserve, ON RRP JEL Classification: G11, G23, E52 ∗The analysis and conclusions set forth are our own and do not necessarily reflect the views of the Board of Governors or the staff of the Federal Reserve System. We thank David Bowman, Chris Gust, Sebastian Infante, Zeynep Senyuz and seminar participants at the Federal Reserve Board for their helpful comments. We thank Erik Bostrom for sharing the mapping used to identify ON RRP-eligible counterparties in the N-MFP and iMoneyNet datasets. All remaining errors are our own. †Federal Reserve Board, E-mail: samin.abdullah@frb.gov. ‡Federal Reserve Board, E-mail: manjola.tase@frb.gov (corresponding author). 1

1 Introduction Money Market Funds (MMF) are an integral part of financial markets as they provide low risk, highly liquid shares to investors. As of December 2024, MMF held around $7.2 trillion in assets under management (AUM). MMF invest in short term, liquid securities including U.S. Treasury bills, commercial paper, certificate of deposits, and repurchase agreements. The interest rate on theseinstrumentsiscloselyrelatedtothemonetarypolicyrateanditsexpectedpath. Furthermore, eligible MMF play a role in the implementation of the monetary policy through their participation at the Federal Reserve’s Overnight Reverse Repurchase Agreement (ON RRP) facility which serves as a supplementary policy tool to help control the federal funds rate. In terms of MMF portfolio composition, the ON RRP is the closest substitute to private repo and Treasury bills and it plays an important role in MMF portfolio management. As MMF yields are closely related to the policy rate path and MMF play an important role in the Fed’s implementation framework through their participation at the ON RRP facility, it is important to understand MMF portfolio decisions as they relate to monetary policy. In this paper, we look at the effect of policy rate uncertainty on portfolio allocations. Our analysis covers the periodNovember2010toDecember2024. Weusetwodistinctmeasuresofinterestrateuncertainty: market- and survey-based measures. Our market-based measure is the swaption-implied volatility from derivatives based on the policy rate. Our survey-based measures are based on the forecast for the federal funds rate (FFR) - the Fed’s policy rate - from the Blue Chip Financial Survey. We construct three survey-based measures of policy rate uncertainty: range of the forecast (dispersion across forecasters), share of upward revisions to the forecast (the share of forecasters revising their forecast to a larger increase or a smaller decrease in the rate), and share of downward revisions to theforecast(theshareofforecastersrevisingtheirforecasttoasmallerincreaseoralargerdecrease in the rate). We find that our market-based and survey-based measures show similar dynamics. Furthermore, while the range of forecast and the implied volatility capture uncertainty in general, 1

our measures of uncertainty based on revisions to the forecast allows us to distinguish the direction of uncertainty. On MMF characteristics, we use two distinct datasets. The N-MFP dataset is a monthly dataset and includes the entire MMF universe. iMoneyNet is a weekly dataset, but it has a smaller coverage and it does not distinguish between take-up at the Federal Reserve’s overnight reverse purchase (ON RRP) facility and private repo. First, we find that an increase in policy rate uncertainty measured either by the swaptionimplied volatility or the range of forecasts is associated with a decrease in the maturity of MMF portfolio. Second, using survey-based measures of uncertainty constructed from revisions to the forecast, we find that the direction of uncertainty matters. MMF portfolio maturity is twice as sensitive to uncertainty as measured by upward revisions than to uncertainty as measured by downward revisions to the forecast of the policy rate. Finally, we look at the role of the ON RRP onMMFportfolioallocationsforONRRP-eligiblecounterparties. Welookatthedailyliquidasset (DLA) holdings which include ON RRP and other instruments that can readily be converted to cash within one business day. We find that overall, an increase in uncertainty is associated with an increase DLA as a share of AUM, consistent with our findings on the maturity of the portfolio. We also distinguish between ON RRP and DLA excluding ON RRP. The effect of an increase in policy rate uncertainty on ON RRP as a share of AUM fully captures the increase in MMF demand for shorttermassetswhenuncertaintyincrease. Conversely, whenlookingatDLAexcludingONRRP, the effect of policy rate uncertainty on these holdings reflect both demand and supply effects as the providers of these instruments would be likely responding to policy rate uncertainty. We find that when policy rate uncertainty increases, ON RRP-eligible funds shift their portfolio composition away from other DLA into ON RRP. This suggests that when policy rate uncertainty increases, the ON RRP facility helps ease upward pressure on demand in short-term funding markets at a time when the supply in such markets is likely to experience downward pressure, thus smoothing rate fluctuations in these markets. Furthermore, as the ON RRP is one of the liabilities in the 2

Fed’s balance sheet, these findings imply that changes in the policy rate uncertainty affect the composition of the Federal Reserve’s balance sheet. Broadly, the literature looking at MMF portfolio allocations and monetary policy often focuses on the policy rate itself. For example, Chodorow-Reich (2014) and Di Maggio and Kacperczyk (2017) specifically looked at how the low interest rate environment affected portfolios. They find that MMF reach for yield in order to retain their investors by giving non-negative net yields after fees. Xiao(2019)foundthatduringperiodsofmonetarypolicytightening, MMFareabletoattract more deposits, which allowed them to increase lending to the wider shadow banking sector. The paperclosesttooursisIm, Li, andWang(2023). Theyuseswaption-impliedvolatilityasameasure of interest rate uncertainty and also find that an increase in interest rate uncertainty is associated with a decrease in portfolio maturity. However, our paper differs from their paper along two key dimensions. First, using survey-based measures, we show that the direction of uncertainty matters. Second, while their paper looks at aggregate measures of portfolio duration, we also explore the role of the ON RRP. This paper is organized as follows. Section 2 discusses measures of uncertainty in the literature andthemeasuresweuseinourpaper. Section3describesthedata. Section4presentsourempirical strategy and discusses the results. Section 5 concludes. 2 Measuring uncertainty We use two types of measures of policy rate uncertainty: 1) market-based measures and 2) surveybased measures.1 Our market-based measure of policy rate uncertainty is based on the implied volatility from 1. Uncertaintymeasurescanbecategorizedintothreegroups: news-based,market-based,andsurvey-based. News based measures rely on textual analysis of news articles. Baker, Bloom, and Davis (2016) constructs an index based on the frequency of certain terms in major newspapers. Husted, Rogers, and Sun (2020) used a similar approach, but focused on monetary policy specifically, rather than all economic policy. Cascaldi-Garcia et al. (2023) provides a comprehensive survey of measures of uncertainty. 3

derivatives on the policy rate. These types of measures are widely used in the literature. For exampleDahlhausandSekhposyan(2018)usetherealizedvolatilityofthesixmonthsaheadFederal Funds futures as a measure of monetary policy uncertainty; Chang and Feunou (2013) use both the realized and implied volatility of futures on the 3-month Canadian dealer offered rate as the Bank of Canada’s policy uncertainty measure; Bauer, Lakdawala, and Mueller (2021) use the implied conditional variance of Eurodollar options as a measure of uncertainty. Our survey-based measure of policy rate uncertainty is based on the Blue Chip Financial Forecasts of the FFR with the range of the forecast as an intuitive proxy for uncertainty. For example, Doehr and Garc´ıa (2021) calculate uncertainty as the difference between the 90th and 10th percentile of the expected 3-month Treasury bill yield. We use the near term forecast for each month, which is the first projection quarter’s forecast, and construct the following measures in policy rate uncertainty: 1) range of forecasts for the FFR for the projection’s quarter forecast across all the forecasters; 2)shareofrespondentsmakinganupwardrevisiontotheirforecasttothetotalnumber of respondents; 3) share of respondents making a downward revision to their forecast to the total numberofrespondents. Anupwardrevisionmeansthattheforecasterexpectsthepolicyrateeither to increase by more or decrease by less than they had previously expected. Similarly, a downward revision means that the forecaster expects the policy rate either to decrease by more or increase by less than they had previously expected. That is, an upward (or a downward) revision does not necessarily correspond to an increasing (or decreasing) interest rate environment. Table 1 provides summary statistics for each of these three measures and the implied volatility measure. Table 2 shows the mean of each of three survey-based measures by sub-periods based on the policy rate environment. While during periods of interests rate easing, the share of downward revisions to the forecast is higher than the share of upward revisions to the forecast, both types of revisions are present across policy rate environments. As shown in Figure 1, the forecast range and the implied volatility measures of uncertainty show similar dynamics. However, our three measures 4

of uncertainty based on the Blue Chip survey have advantages relative the market based implied volatility measures as they allows us to capture uncertainty across different directions. 3 Data We use two distinct data sources for MMF portfolio characteristics: the SEC’s form N-MFP data andiMoneyNet’sMMFdata.2 Weusetwotypesofmeasuresofpolicyuncertainty: amarket-based measure from ICAP’s swaption-implied volatility, and survey-based measures that we construct for the Blue Chip Financial Forecast Survey responses.3 The N-MFP data sets provide monthly MMF level data on weighted average maturities (WAM) of portfolios, assets under management (AUM), fund type, usage of the Fed’s RRP facilities, etc. In our analysis, we aggregate the data groups based on the fund family level distinguishing between subsets of fund types (e.g. government, prime) and ON RRP-eligible and ON RRP non-eligible within each fund family. For example, if a fundfamilyhad3primeONRRPnon-eligiblefundsforonemonth, all3wouldbeconsolidatedinto one entity for that month. WAMs for each of these entities are calculated by taking the weighted average of each fund’s WAM, weighted by AUM. Tables 3 and 4 show summary statistics for the relevant variables at the fund and group level, respectively. The iMoneyNet data provides weekly MMF level data on WAMs, AUM, and fund type as of Tuesday. Same as with the N-MFP data, individual money market funds are aggregated to groups at the family-fund, type, and RRP eligibility level. While iMoneyNet provides us with more frequent data compared to the N-MFP data, it does not provide as detailed data on each fund’s portfolio. As such, we are unable to extract out a fund’s usage of the Fed’s RRP facility. Additionally, iMoneyNet covers a smaller part of the MMF universe with 331 funds included in 2. SEC’s form N-MFP data: https://www.sec.gov/dera/data/form-nmfp-data-sets. iMoneyNet: iMoneyNet, Inc., iMoneyNet Bulk Data - Offshore Analyzer and Gold Analyzer. 3. ICAP: TP ICAP. Swaptions and Interest Rate Caps and Floors Data. Blue Chip: Wolters Kluwer, Blue Chip Financial Forecasts. Wolters Kluwer, https://www.wolterskluwer.com/ en/solutions/blue-chip. 5

the iMoneyNet data compared to 776 in the N-MFP data. However, funds included in iMoneyNet are typically larger funds, so in terms of AUM, about 80% of AUM reported in the N-MFP is also being reported to iMoneyNet. Tables 5 and 6 show summary statistics for the relevant variables at the fund and group level, respectively. We use mapping by Bostrom (2025) to identify ON RRP eligible counterparties in the money funds datasets.4 Formarket-basedmeasuresofuncertaintyweusetheswaption-impliedvolatilityoftheone-year swap rate at a horizon six month ahead. Data is daily. We use the average of the last 5 days in the corresponding month to align with both the Blue Chip Survey data and the N-MFP data. We provide more details on the data timing below. For the weekly iMoneyNet data, we use the volatility of the day the MMF data was reported. For survey-based measures of uncertainty we use the Blue Chip Financial Forecasts of the FFR to construct measures of policy rate uncertainty based on dispersion of and revisions to the FFR forecast. The survey is at a monthly frequency. Each month’s FFR forecast has on average 44 forecasters. A note on the data timing: The Blue Chip Financial Survey for any given month is conducted during the last few days of the previous month. For example, the February survey is based on survey data collected during the last few days of January. N-MFP data for any month are as of the last day of that month. Hence, the information set window in the N-MFP data corresponds to the information set in the one-month forward Blue Chip data. As a result, in our regression analysis, we pair the month-end N-MFP data with the one-month forward Blue Chip data. By combining the MMF portfolio characteristics data and the policy uncertainty data, we construct a panel data set spanning from November 2010 to December 2024. 4. The list of eligible ON RRP counterparties is from the Federal Reserve Bank of New York’s (FRBNY) website is found in https://www.newyorkfed.org/markets/rrp counterparties. 6

4 Empirical Strategy and Results The first two parts of this section discuss the effect of policy rate uncertainty on MMF portfolio maturity and whether the direction of uncertainty matters. The third part looks at how ON RRP eligible counterparties adjust their portfolio allocations, distinguishing between their ON RRP holdings and their holdings of other daily liquid assets. 4.1 Does policy rate uncertainty affect MMF portfolio maturity? Figure 2 and Figure 3 plot the two measures of policy rate uncertainty - the survey-based FFR forecastrangeandthemarket-basedimpliedvolatility,respectively-andtheweightedaverageportfolio maturity (WAM) from N-MFP form fillings and iMoneyNet for the period between November 2010 to December 2024. Overall we see a negative relationship between policy rate uncertainty and the MMFs portfolio maturity. Note that policy uncertainty was low and little changed during the zero lower bound, but it did start to increase in 2015 even before the first policy rate increase out of the zero lower bound in December 2015. An argument for this negative correlation could be that shorter WAMs allow MMF to promptly change their portfolio allocations once the uncertainty about the path of the interest rate decreases. To test this hypothesis, we estimate a panel regression specification as in (1): WAM = β +β Uncertainty +β DaysTillFOMC +β (Uncertainty ×DaysTillFOMC ) i,t 0 1 t 2 t 3 t t +β (Uncertainty ×RRPstart ×RRPelig )+β (RRPstart ×RRPelig ) 4 t t i,t 5 t i,t +β RRPstart +β AUM +β NetTbillIssuance 6 t 7 i,t 8 t +β MMFreform +β (prime ×MMFreform )+α +(cid:15) 9 t 10 i t i i,t (1) where Uncertainty is one of the following two measures of policy rate uncertainty: 1) forecast t rangefromtheFFRforecastintheBlueChipssurvey;2)swaptionimpliedvolatility. Sinceswaption data is daily, we use the average of the last 5 days of the month to align with both the Blue Chip 7

Survey data and the N-MFP data. DaysTillFOMC is the number of days from month-end t t to the second day of the next FOMC meeting. RRPstart is equal to 1 if t is after the start of t the RRP operations (September 23, 2013), 0 otherwise. RRPelig is equal to 1 if entity i is an i,t ON RRP eligible counterparty on month-end t, 0 otherwise. prime is equal to 1 if entity i is a i prime fund, 0 otherwise. MMFreform is equal to 1 if month-end t is after the SEC adopted the t 2014-2016 MMF reforms (July 23, 2014), 0 otherwise5. NetTbillIssuance is net Treasury bills t issuance6 and controls for the supply of investment alternatives for MMF (Treasury bills directly and repo, indirectly). We also include fund group fixed effects α . The data is monthly for the i period November 2010 to December 2024. Tables 7 and 8 show the regression results when using the forecast range measure and the implied volatility measure, respectively. An increase in policy rate uncertainty is associated with a decrease in MMF WAM. This result is statistically robust across the two measures of policy rate uncertainty and across different specifications within each measure of uncertainty. The economic significance of the results is about the same. To illustrate, using specification (1) in Tables 7 and 8, a one standard deviation increase in uncertainty is associated on average with a 3.5 day or a 3.9 day decrease in WAMs respectively. Furthermore, the sensitivity of WAMs to policy rate uncertainty varies within the FOMC intermeeting period. As expected, the closer to the FOMC meeting (the closer in time to a possible change in the policy rate) the shorter the WAMs to allow MMF to adjust their portfolio accordingly. Thisresultisrobustacrossthesetwomeasuresofpolicyrateuncertaintyandacrossregression specifications. 5. In 2014, the Securities and Exchange Commission (SEC) adopted a series of amendments to Rule 2a-7 aimed atmakingMMFlesssusceptibletoruns. Theseamendmentswereenactedin2016andcontainedtwomajorreforms. OnewasswitchingfromastabletofloatingNAVforprimeinstitutionalMMF.Theotherallowedallnon-government MMFtoimposeliquidityfeesandredemptiongateswhenfacingarun. Gissler,Macchiavelli, andNarajabad(2023) andSundaresanandXiao(2018)findthattheregulationreformcausedMMFtoshifttheirportfoliostowardsFederal Home Loan Bank notes. Baghai, Giannetti, and Ja¨ger (2022) found prime MMF had a more performance sensitive investor base after the reform, causing them to reach for the yield and invest in riskier assets. 6. U.S. Treasury debt issuance data: https://fiscaldata.treasury.gov/datasets/treasury-securities-auctions-data/ treasury-securities-auctions-data 8

WealsoestimatesthisregressionusingweeklyMMFdatafromiMoneyNetandimpliedvolatility as our measure of policy rate uncertainty. The results are shown in Table 9 and they are consistent with the results from the monthly N-MFP datasets. Furthermore, the effect of uncertainty on WAMsisaboutthesameinthemonthlyandweeklydatasuggestingthatMMFadjustthematurity of their portfolio within a week in response to changes in policy rate uncertainty. 4.2 Does the direction of policy rate uncertainty matter? Weestimatetheregressionspecificationin(1)whereUncertainty isconstructedbasedonrevisions t to the forecast (share of upward revisions, share of downward revisions) from the FFR forecast in the Blue Chips survey. Tables 10 and 11 show the regression results using these two measures of uncertainty. An increase in policy rate uncertainty based on forecast revisions is associated with a decrease in MMF WAMs, similar to our findings when using the forecast range and swaption implied volatility as measures of policy rate uncertainty. However, the direction of uncertainty matters. First, looking at the estimated coefficient for uncertainty, we find MMF portfolio maturity is twice more sensitive to uncertainty related to upward revisions than to downward revisions to the forecast of the policy rate.7 Second, looking at the estimated coefficient for the interaction of the policy rate uncertainty and days till the FOMC meeting, we find different results depending on the measure of policy rate uncertainty. Specifically, we find the effect of the distance from the FOMC meeting on WAMs increasing with the increase in the share of downward revisions, but we find no statistically significant relationship when rate uncertainty is measured by the share of upward revisions to the forecast. A possible explanation is that, as shown in Table 2, downward revisions are more prevalent than upward revisions to the forecast during periods of policy rate easing. In a decreasing policy rate environment, if the revised 7. Note that as shown in Table 1, the FFRup and FFRdown series have about the same standard deviation, so we can compare the estimated coefficients directly. 9

expectations are for a larger decrease in the interest rate than previously expected (that is the share of downward revisions increases), it would be more profitable for the MMF to maintain a longer WAM the further away for the FOMC meeting to lock in the higher rates. 4.3 How do RRP-eligible MMF adjust their portfolio allocations? The Federal Reserve uses an overnight reverse repurchase agreement (ON RRP) facility as needed as a supplementary policy tool to help control the federal funds rate and keep it in the target range set by the FOMC. 8 When the Federal Reserve conducts an overnight RRP, it sells a security to an eligible counterparty and simultaneously agrees to buy the security back the next day.9 The ON RRP facility supports setting a floor on money market rates by providing a fixed rate on safe overnight investments to eligible MMF, government-sponsored enterprises (GSEs), and other entities that are ineligible to earn interest on reserve balances (IORB). The ON RRP offering rate (themaximuminterestratethattheFederalReserveiswillingtopayonONRRPoperations)plays a role for ON RRP counterparties that is similar to the role played by the interest rate on excess reservesfordepositoryinstitutions. Thatis, ingeneral, anycounterpartythatcanusetheONRRP facility should be unwilling to invest funds overnight with another counterparty at a rate below the ON RRP rate, just as any depository institution eligible to earn interest on reserves should be unwilling to invest funds overnight with another counterparty at a rate below the IORB. The Federal Reserve conducts ON RRP operations with many counterparties, covering a wide range of entities.10 MMF are the main participants at the Fed’s ON RRP operations (Figure 4). Furthermore, 8. The Federal Reserve conducted technical exercises using ON RRPs beginning in September 2013 in order to gain operational experience and garner information about how such operations might be used during the policy normalization process. In the Policy Normalization Principles and Plans announced on September 17, 2014, the FOMC indicated that it intended to use an overnight reverse repurchase agreement facility as needed as a supplementary policy tool to help control the FFR and keep it in the target range set by the FOMC. See https://www.federalreserve.gov/monetarypolicy/overnight-reverse-repurchase-agreements.htm. 9. ThistransactiondoesnotaffectthesizeoftheSystemOpenMarketAccount(SOMA)portfolio,butitchanges the composition of the liability side of the Federal Reserve’s balance sheet while the trade is outstanding. 10. Foralistofcounterpartiesandeligibilitycriteriaseehttps://www.newyorkfed.org/markets/rrp counterparties. 10

Figure 5 suggests that an increase in policy rate uncertainty, measured by the range of forecasts for the FFR, is associated with an increase in the share MMFs take-up at the ON RRP facility relative to their AUM. This is not surprising as ON RRP is the shortest maturity (overnight) and we established in the previous section that an increase in uncertainty is associated with a decrease in MMFs portfolio maturity. ON RRP is also one of the instruments included in Daily Liquid Assets (DLA). Broadly speaking, DLA include cash or securities that can readily be converted to cash within one business day.11 Table 12 provides summary statistics of MMF holdings across different categories within DLA. For non eligible funds, DLA consist mostly of Treasury securities with private repo coming in second, while for eligible funds, DLA is on average more uniformly spread across different categories. Furthermore, looking at measures of dispersion, such as the standard deviation or the difference between the 95th and the 5th percentile, the composition of DLAisrelativelystablefornoneligiblefunds, whileitvariesrelativelymoreforeligiblefunds. This suggests that eligible funds overall adjust their portfolio more than non eligible funds. We explore the role of the ON RRP on how RRP eligible funds adjust their portfolio allocations by breaking down DLA into two parts: 1) ON RRP, 2) DLA excluding ON RRP. We then estimate the sensitivity of the these two components to policy rate uncertainty as shown in specification (2). (Component/AUM) = β +β Uncertainty +β DaysTillFOMC i,t 0 1 t 2 t +β (Uncertainty ×DaysTillFOMC ) 3 t t (2) +β AUM +β NetTbillIssuance 4 i,t 5 t +β MinDLAincrease +α +(cid:15) 6 t i i,t where Component is one of the following: DLA, RRP, DLAexclRRP. Uncertainty is one of the following survey-based measures of policy rate uncertainty: FFRrange, FFRup, FFRdown. MinDLAincrease is a dummy variable, equal to 1 for the period after the minimum requirement t 11. MMF are are subject to minimum requirements about their DLA as a share their assets. Effective October 2, 2023, the minimum requirement for DLA as a share of assets increased from 10% to 25% of total assets. See https://www.sec.gov/files/rules/final/2023/33-11211.pdf. 11

for DLA as a share of assets was increased to 25% (October 2023). Other variables are defined as in regression (1) in the previous section. Data is at a monthly frequency for the period April 2016 (the first month N-MFP filings required DLA reporting) to December 2024. This breakdown allows us to capture the clean effect of policy rate uncertainty on demand for short-term assets. The estimated effect of policy rate uncertainty on ON RRP holdings provides a cleancaptureofchangesinMMFdemandforshort-termassetsinresponsetochangesinpolicyrate uncertainty as MMF can participate up to their counterparty limit at the ON RRP offering rate.12 Conversely, when looking at DLA excluding ON RRP, the effect of policy rate uncertainty on these holdings reflect both demand and supply effects as the providers of these instruments would likely be responding to policy rate uncertainty as well. In addition, looking at the private repo, which is the closest alternative to ON RRP, MMF could face constraints when trying to increase their repo holdings. For example, they might have counterparty constraints which limit their repo exposure to a specific counterparty. Or their counterparties might limit their offerings of repo based on their own counterparty limits. Regression results are shown in Tables 13, 14, 15 for each of the three survey-based measures of policy rate uncertainty: range of forecast, share of upward revisions to the forecast, and share of downward revisions to the forecast, respectively. Looking at column 1, we find a strong positive relationship between uncertainty and DLA/AUM. This is consistent with our results in the previous section that an increase in uncertainty is associated with a decrease in the maturity of the portfolio.13 Looking and columns 2 and 3, we find that changes in policy rate uncertainty are associated with opposite effects on ONRRP/AUM and DLAexclONRRP/AUM. Specifically, looking at column 2, we find a strong positive relationship between policy rate uncertainty and ONRRP/AUM across the three measures of policy rate uncertainty. As discussed above, this 12. Whiletherearebothcounterpartycapsandaggregatecaps,thecapsareunlikelytobebinding. Foroperational details, see https://www.newyorkfed.org/markets/opolicy/operating policy 151216.html. 13. Whenuncertaintyismeasuredasshareofdownwardrevisions,wedonotfindevidenceofarelationshipbetween DLA/AUManduncertainty. Thissuggestthat,inthiscase,thenegativerelationshipbetweenportfoliomaturitywe found in the previous section could be driven my adjustment portfolio holdings beyond the daily liquid asset. 12

estimate fully reflects the increase in MMF demand for liquid assets when policy rate uncertainty increases. Looking at column 3, we find the opposite effect: a negative relationship between DLA excluding ON RRP as a share of AUM and policy rate uncertainty. These findings suggests that the ON RRP facility also helps ease fluctuations in demand in short term funding market. For example, when policy rate uncertainty increases, eligible MMF move away from other DLA into RRP. This move eases upward pressure on demand for short-term investments by eligible funds at a time when demand by other participants for such investments increases and supply of such investment decreases. As a result, this smaller upward pressure on total demand combined with a decrease in supply results in a smaller increase in the price of these asset smoothing fluctuations the market for short-term investments. Another area where we see opposite effect on the two parts of DLA (DLAexclRRP and ONRRP) relates to the 2023 increase in the minimum requirement for DLA. We find that the 2023 increase in the minimum requirement for DLA as a share of AUM is associated with an increase DLAexclRRP/AUM and a decrease in ONRRP/AUM. A plausible explanation is that as changes in regulationare more of apermanent nature, MMF are likely torespond to the increase in minimum requirements for DLA by increasing the share of their DLA instruments which are part of their long term portfolio strategy, while the ON RRP provides an alternative investment when more attractive rates are not available. 5 Conclusion This paper provides insights into MMF allocations in very short-term assets in general and the ON RRP in particular in response to uncertainty about the monetary policy rate. We find that an increase in policy rate uncertainty is associated with an increase in MMFs holdings of overnight RRP. This move eases upward pressure on demand for short-term investments and smooth fluctuations in short-term funding market. Furthermore, this increase in ON RRP affects the composition 13

of the Fed’s balance sheet and could have implications for monetary policy decisions related to the Fed’s balance sheet. 14

References Baghai,RaminP.,MariassuntaGiannetti,andIvikaJ¨ager.2022.“LiabilityStructureandRiskTaking: Evidence from the Money Market Fund Industry.” Journal of Financial and Quantitative Analysis 57 (5): 1771–1804. Baker, Scott R., Nicholas Bloom, and Steven J. Davis. 2016. “Measuring Economic Policy Uncertainty.” The Quarterly Journal of Economics 131, no. 4 (July): 1593–1636. Bauer, Michael D, Aeimit Lakdawala, and Philippe Mueller. 2021. “Market-Based Monetary Policy Uncertainty.” The Economic Journal 132, no. 644 (November): 1290–1308. Bostrom, Erik. 2025. Insights from MMF Portfolio Allocations amid Balance Sheet Normalization. Finance and Economic Discussion Series. Federal Reserve Board of Governors, March. Cascaldi-Garcia,Danilo,CisilSarisoy,JuanM.Londono,BoSun,DeepaD.Datta,ThiagoFerreira, Olesya Grishchenko, et al. 2023. “What Is Certain about Uncertainty?” Journal of Economic Literature 61, no. 2 (June): 624–54. Chang, Bo Young, and Bruno Feunou. 2013. Measuring Uncertainty in Monetary Policy Using Implied Volatility and Realized Volatility. Staff Working Papers 13-37. Bank of Canada, October. Chodorow-Reich, Gabriel. 2014. Effects of Unconventional Monetary Policy on Financial Institutions. Working Paper, Working Paper Series 20230. National Bureau of Economic Research, June. Dahlhaus, Tatjana, and Tatevik Sekhposyan. 2018. Monetary Policy Uncertainty: A Tale of Two Tails. Staff Working Papers 18-50. Bank of Canada. Di Maggio, Marco, and Marcin Kacperczyk. 2017. “The unintended consequences of the zero lower bound policy.” Journal of Financial Economics 123 (1): 59–80. Doehr, Rachel, and Enrique Mart´ınez Garc´ıa. 2021. Monetary Policy Uncertainty and Economic Fluctuations at the Zero Lower Bound. Globalization Institute Working Papers 412. Federal Reserve Bank of Dallas, November. 15

Gissler, Stefan, Marco Macchiavelli, and Borghan Narajabad. 2023. Providing safety in a rush: How Did Shadow Banks Respond to a $1 Trillion Shock. Working Paper. https://ssrn.com/ abstract=4647599, March. Husted, Lucas, John Rogers, and Bo Sun. 2020. “Monetary Policy Uncertainty.” Journal of Monetary Economics 115:20–36. Im, Jay, Yi Li, and Ashley Wang. 2023. Investor Flows, Monetary Policy, and Portfolio Management of Money Market Funds. Working Paper. https://ssrn.com/abstract=4647599. Sundaresan, Suresh, and Kairong Xiao. 2018. Unintended Consequences of Post-Crisis Liquidity Regulation. Working Paper. Columbia Business School, November. Xiao, Kairong. 2019. “Monetary Transmission through Shadow Banks.” The Review of Financial Studies 33, no. 6 (October): 2379–2420. 16

Table 1: Summary Statistics - Measures of policy rate uncertainty Statistic Num of Obs Mean St. Dev. Min Median Max FFR Range (p.p.) 170 0.356 0.333 0.000 0.245 2.500 FFR Down 170 0.161 0.154 0.000 0.126 1.000 FFR Up 170 0.146 0.158 0.000 0.104 0.923 Implied Volatility 167 59.6 39.3 11.5 47.2 184.1 Note: FFR Range is the range of respondents’ Federal Funds Rate (FFR) forecast. FFR Down is the share of respondents who revised their FFR forecast downwards from the prior month’s survey. FFR Up is the share of respondents who revised their FFR forecast upwards from the prior month’s forecast. Implied volatility is the swaption-implied volatility of one-year swap rate at a horizon of six months ahead. We use the average of the last 5 days in the corresponding month to align with both the Blue Chip Survey data and the N-MFP data. Survey data is from Wolters Kluwer’s monthly Blue Chip Financial Forecasts. Swaption data is from ICAP. Data is for the period November 2010 to December 2024. 17

Table 2: Measures of policy rate uncertainty: mean by sub-periods based on the policy rate environment Sub-periods N FFR Range (p.p.) FFR Up FFR Down ZLB 1 (10/2010-11/2015) 61 0.192 0.093 0.187 Tightening 1 (12/2015- 7/2019) 44 0.357 0.180 0.177 Easing 1 (08/2019-02/2020) 7 0.557 0.150 0.256 ZLB 2 (03/2020-02/2022) 24 0.142 0.061 0.054 Tightening 2 (03/2022-08/2024) 30 0.720 0.273 0.125 Easing 2 (09/2024-12/2024) 4 1.05 0.124 0.328 Note: The sub-periods are defined by the path of the midpoint of the target range for fed funds rate (FFR). ZLB refers to when the midpoint was at the zero lower bound (ZLB), Tightening refers to when the midpoint was increasing or held constant following a prior increase, and Easing refers to when midpoint was decreasing or held constant following a prior decrease. Target range for the fed funds rate: https://www.federalreserve.gov/economy-at-a-glance-policy-rate.htm. Fed funds’ forecast data is from Wolters Kluwer’s monthly Blue Chip Financial Forecasts spanning between November 2010 to December 2024. FFR Range is the range of respondents’ Federal Funds Rate (FFR) forecast. FFR Up is the share of respondents who revised their FFR forecast upwards from the prior month’s forecast. FFR Down is the share of respondents who revised their FFR forecast downwards from the prior month’s survey. 18

Table 3: N-MFP Summary Statistics - At the fund level Statistic N Obs N Funds Mean St. Dev. 5th pct Median 95th pct WAM (Days) 71914 776 32.463 14.641 6 34 54 AUM ($ Billions) 71914 776 8.568 24.523 0.041 1.097 38.285 RRP Takeup ($ Billions) 13170 157 4.867 14.599 0 0.01 26.332 DLA/AUM 33806 535 0.438 0.302 0 0.396 0.979 RRP Takeup/AUM 13148 157 0.105 0.171 0 0.001 0.501 Note: Data is from monthly N-MFP form filings spanning between November 2010 to December 2024. Each entry is at the fund-month level. WAM is the weighted average maturity of the fund’s portfolio. AUM is the assets under management for the fund. RRP Takeup is the amount of takeup the fund has at the Fed’s ON RRP facility (ON RRP operations began September 2013). DLA/AUM is the fund’s daily liquid assets as a share of assets under management (N-MFP filings include DLA reporting starting April 2016). RRP Takeup/AUM is the fund’s ON RRP takeup as a share of assets under management. 19

Table 4: N-MFP Summary Statistics - At the group level Statistic N Obs N Groups Mean St. Dev. 5th pct Median 95th pct WAM (Days) 31317 435 32.834 14.054 7 34 53 AUM ($ Billions) 31317 435 20.473 55.859 0.099 2.254 102.457 RRP Takeup ($ Billions) 6572 87 8.047 29.329 0 0 40.522 DLA/AUM 16570 271 0.468 0.28 0.022 0.424 0.972 RRP Takeup/AUM 6572 87 0.084 0.149 0 0 0.445 Note: Data is from monthly N-MFP form filings spanning between November 2010 to December 2024. Individual money market funds are aggregated to groups at the family-fund, type, and RRP eligibility level. For example, if a fund family had 3 prime non-RRP eligible funds for one month, all 3 would be consolidated into one entity. WAM is the weighted average maturity of the entity’s portfolio. AUM is the assets under management for the entity. RRP Takeup is the amount of takeup the entity has at the Fed’s ON RRP facility. DLA/AUM is the entity’s daily liquid assets as a share of assets under management (N-MFP filings include DLA reporting starting April 2016). RRP Takeup/AUM is the entity’s ON RRP takeup as a share of assets under management. 20

Table 5: iMoneyNet Summary Statistics - At the fund level Statistic N Obs N Funds Mean St. Dev. 5th pct Median 95th pct WAM (Days) 210511 331 33.103 14.008 8 34 54 AUM ($ Billions) 210511 331 11.81 30.176 0.096 1.769 59.928 Note: Data is from weekly iMoneyNet dataset on MMF spanning between November 2010 to December 2024. Each entry is at the fund-week level. WAM is the weighted average maturity of the fund’s portfolio. AUM is the assets under management for the fund. 21

Table 6: iMoneyNet Summary Statistics - At the group level Statistic N Obs N Groups Mean St. Dev. 5th pct Median 95th pct WAM (Days) 75773 158 34.6 12.84 11 36 53 AUM ($ Billions) 75773 158 31.81 70.446 0.194 4.351 150.803 Note: Data is from weekly iMoneyNet dataset on MMF spanning between November 2010 to December 2024. Individual money market funds are aggregated to groups at the family-fund, type, and RRP eligibility level. For example, if a fund family had 3 prime non-RRP eligible funds for one month, all 3 would be consolidated into one entity. WAM is the weighted average maturity of the entity’s portfolio. AUM is the assets under management for the entity. 22

Table 7: MMF portfolio maturity and policy rate uncertainty (forecast range). Monthly frequency. (1) (2) (3) VARIABLES WAM WAM WAM F.FFRrange -10.67*** -9.023*** -8.187*** (0.425) (0.402) (0.519) DaysTillFOMC 0.0690*** 0.0663*** (0.00493) (0.00468) F.FFRrange*DaysTillFOMC -0.0760*** -0.0745*** (0.0103) (0.0101) F.FFRrange*RRPelig*RRPstart -2.636*** (0.825) RRPstart 1.012** (0.504) AUM 0.000200 0.000630 0.00277 (0.0106) (0.0105) (0.0108) NetTreasuryBillIssuance -0.000357 -4.99e-05 -7.04e-05 (0.000476) (0.000483) (0.000483) MMFreform -4.382*** -4.265*** -5.135*** (0.910) (0.908) (0.895) TypePrime*MMFreform -0.875 -0.899 -0.778 (1.220) (1.217) (1.215) Constant 39.44*** 37.62*** 37.40*** (0.516) (0.516) (0.565) Observations 30,555 30,555 30,555 R-squared 0.149 0.154 0.156 Number of group id 431 431 431 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: MMF data is from the N-MFP data sets. WAM is the weighted average maturity of the portfolio of MMF group i at time t. F.FFRrange is the range of the forecast for the fed funds rate for the first projection’s quarter forecast across the forecasters in the Blue Chip Financial Forecasts at time t+1. DaysTillFOMC is the number of days from month-end t to day 2 of the next FOMC meeting. RRPstart, RRPelig, TypePrime, MMFreform are dummy variables equal to 1 if after the establishment of the Fed’s ON RRP facility, RRP eligible, a prime fund, after the 2014-2016 MMF reform was approved (July 2014), respectively. AUM is assets under management. groups are at the MMF family-fund type-ON RRP eligibility level. A note on the data timing: The Blue Chip Financial Survey for any given month is conducted during the last few days of the previous month. N-MFP data for any month are as of the last day of that month. As a result, the information set window in the N-MFP data corresponds to the information set in the one-month forward Blue Chip data. As a result, in our regression analysis, we pair the month-end N-MFP data with the one-month forward Blue Chip data. Data is at a monthly frequency for the period November 2010 to December 2024. The results are from panel data regression including group fixed effects and group-clustered standard errors. 23

Table 8: MMF portfolio maturity and policy rate uncertainty (implied volatility). Monthly frequency. (1) (2) (3) VARIABLES WAM WAM WAM ImplVol -0.0986*** -0.101*** -0.0917*** (0.00520) (0.00585) (0.00691) DaysTillFOMC 0.0229*** 0.0232*** (0.00511) (0.00464) ImplVol*DaysTillFOMC 0.000178*** 0.000173*** (6.75e-05) (6.42e-05) ImplVol*RRPelig*RRPstart -0.0296*** (0.0101) RRPstart -0.0564 (0.532) AUM 0.0105 0.0109 0.0140 (0.0115) (0.0115) (0.0118) NetTreasuryBillIssuance 0.000252 0.000423 0.000422 (0.000466) (0.000456) (0.000456) MMFreform -4.270*** -4.371*** -4.452*** (0.911) (0.912) (0.902) TypePrime*MMFreform -0.813 -0.822 -0.716 (1.228) (1.228) (1.226) Constant 41.01*** 40.36*** 40.27*** (0.585) (0.626) (0.666) Observations 30,240 30,627 30,627 R-squared 0.156 0.156 0.158 Number of group id 433 435 435 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: MMF data is from the N-MFP data sets. WAM is the weighted average maturity of the portfolio of MMF group i at time t. ImplVol is the swaption-implied volatility of the one-year swap rate at a horizon of six months ahead. We use the average of the last 5 days in the corresponding month to align with both the Blue Chip Survey data and the N-MFP data. DaysTillFOMC is the number of days from t to day 2 of the next FOMC meeting. RRPstart, RRPelig, TypePrime, MMFreform are dummy variables equal to 1 if after the establishment of the Fed’s ON RRP facility, RRP eligible, a prime fund, after the 2014-2016 MMF reform was approved (July 2014), respectively. AUM is assets under management. groups are at the MMF family-fund type-ON RRP eligibility level. Data is at a monthly frequency for the period November 2010 to December 2024. The results are from panel data regression including group fixed effects and group-clustered standard errors. 24

Table 9: MMF portfolio maturity and policy rate uncertainty (implied volatility). Weekly frequency. (1) (2) (3) VARIABLES WAM WAM WAM ImplVol -0.0988*** -0.0991*** -0.0908*** (0.00600) (0.00597) (0.00724) DaysTillFOMC 0.00142 0.00143 (0.00218) (0.00219) ImplVol*DaysTillFOMC 1.49e-05 1.38e-05 (2.77e-05) (2.76e-05) ImplVol*RRPelig*RRPstart -0.0286** (0.0128) RRPstart -0.133 (0.801) AUM 0.00159 0.00159 0.00469 (0.00941) (0.00941) (0.00956) NetTreasuryBillIssuance 0.000482 0.000451 0.000371 (0.00175) (0.00175) (0.00175) MMFreform -5.878*** -5.875*** -5.921*** (1.265) (1.265) (1.245) TypePrime*MMFreform -0.951 -0.951 -0.852 (1.785) (1.785) (1.781) Constant 45.12*** 45.09*** 45.09*** (0.874) (0.876) (0.954) Observations 73,673 73,673 73,673 R-squared 0.204 0.204 0.206 Number of group id 158 158 158 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: MMF data is from iMoneyNet. WAM is the weighted average maturity of the portfolio of MMF group i at time t. ImplVol is the swaption-implied volatility of the one-year swap rate at a horizon of six months ahead. We use the average of the last 5 days in the corresponding month to align with both the Blue Chip Survey data and the N-MFP data. DaysTillFOMC is the number of days from month-end t to day 2 of the next FOMC meeting. RRPstart, RRPelig, TypePrime, MMFreform are dummy variables equal to 1 if after the establishment of the Fed’s ON RRP facility, RRP eligible, a prime fund, after the 2014-2016 MMF reform was approved (July 2014), respectively. AUM is assets under management. groups are at the MMF family-fund type-ON RRP eligibility level. Data is at a weekly frequency for the period November 2010 to December 2024. The results are from panel data regression including group fixed effects and group-clustered standard errors. 25

Table 10: MMFportfoliomaturityandpolicyrateuncertainty(upwardrevisionstotheforecast). Monthly frequency. (1) (2) (3) VARIABLES WAM WAM WAM F.FFRup -13.20*** -11.94*** -10.10*** (0.740) (0.980) (1.086) DaysTillFOMC 0.0460*** 0.0447*** (0.00485) (0.00460) F.FFRup*DaysTillFOMC -0.0365 -0.0347 (0.0233) (0.0229) F.FFRup*RRPelig*RRPstart -6.057*** (1.631) RRPstart 0.684 (0.510) AUM -0.00464 -0.00438 -0.00328 (0.0105) (0.0105) (0.0106) NetTreasuryBillIssuance -0.000699 -0.000282 -0.000344 (0.000484) (0.000490) (0.000487) MMFreform -5.766*** -5.753*** -6.320*** (0.916) (0.914) (0.901) TypePrime*MMFreform -0.847 -0.868 -0.782 (1.233) (1.231) (1.229) Constant 38.81*** 37.52*** 37.33*** (0.513) (0.530) (0.584) Observations 30,555 30,555 30,555 R-squared 0.099 0.102 0.104 Number of group id 431 431 431 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: MMF data is from the N-MFP data sets. WAM is the weighted average maturity of the portfolio of MMF group i at time t. F.FFRup is the number of forecasters making an upward revision to the forecast for the fed funds rate for the first projection’s quarter forecast in the Blue Chip Financial Forecasts at time t+1 divided by the total number of forecasters. DaysTillFOMC is the number of days from month-end t to day 2 of the next FOMC meeting. RRPstart, RRPelig, TypePrime, MMFreform are dummy variables equal to 1 if after the establishment of the RRP, RRP eligible, a prime fund, after the 2014-2016 MMF reform was approved (July 2014), respectively. AUM is assets under management. groups are at the MMF family-fund type-ON RRP eligibility level. A note on the data timing: The Blue Chip Financial Survey for any given month is conducted during the last few days of the previous month. N-MFP data for any month are as of the last day of that month. As a result, the information set window in the N-MFP data corresponds to the information set in the one-month forward Blue Chip data. As a result, in our regression analysis, we pair the month-end N-MFP data with the one-month forward Blue Chip data. Data is at a monthly frequency for the period November 2010 to December 2024. The results are from panel data regression including group fixed effects and group-clustered standard errors. 26

Table 11: MMF portfolio maturity and policy rate uncertainty (downward revisions to the forecast). Monthly frequency. (1) (2) (3) VARIABLES WAM WAM WAM F.FFRdown -1.334* -5.413*** -5.734*** (0.706) (1.052) (1.134) DaysTillFOMC 0.0356*** 0.0319*** (0.00568) (0.00548) F.FFRdown*DaysTillFOMC 0.118*** 0.126*** (0.0296) (0.0293) F.FFRdown*RRPelig*RRPstart 0.100 (1.422) RRPstart 1.146** (0.517) AUM -0.00547 -0.00515 -0.00414 (0.0107) (0.0107) (0.0110) NetTreasuryBillIssuance -0.000697 3.99e-06 1.20e-05 (0.000485) (0.000494) (0.000494) MMFreform -6.481*** -6.467*** -7.302*** (0.925) (0.924) (0.913) TypePrime*MMFreform -0.690 -0.705 -0.642 (1.243) (1.241) (1.239) Constant 37.65*** 36.80*** 36.58*** (0.512) (0.535) (0.581) Observations 30,555 30,555 30,555 R-squared 0.064 0.070 0.070 Number of group id 431 431 431 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: MMF data is from the N-MFP data sets. WAM is the weighted average maturity of the portfolio of MMF group i at time t. F.FFRdown is the number of forecasters making a downward revision to the forecast for the fed funds rate for the first projection’s quarter forecast in the Blue Chip Financial Forecasts at time t+1 divided by the total number of forecasters. DaysTillFOMC is the number of days from month-end t to day 2 of the next FOMC meeting. RRPstart, RRPelig, TypePrime, MMFreform are dummy variables equal to 1 if after the establishment of the RRP, RRP eligible, a prime fund, after the 2014-2016 MMF reform was approved (July 2014), respectively. AUM is assets under management. groups are at the MMF family-fund type-ON RRP eligibility level. A note on the data timing: The Blue Chip Financial Survey for any given month is conducted during the last few days of the previous month. N-MFP data for any month are as of the last day of that month. As a result, the information set window in the N-MFP data corresponds to the information set in the one-month forward Blue Chip data. As a result, in our regression analysis, we pair the month-end N-MFP data with the one-month forward Blue Chip data. Data is at a monthly frequency for the period November 2010 to December 2024. The results are from panel data regression including group fixed effects and group-clustered standard errors. 27

Table 12: Summary statistics on the share of each category within the Daily Liquid Asset Panel A. Non RRP Eligible Funds Category N Funds N Months Mean St.Dev. 5th pct Median 95th pct Tsy Securities 294 105 86.621 3.976 79.382 87.068 92.098 Tsy Repo excl. ON RRP 222 105 5.66 1.957 2.614 5.996 8.42 Agency Securities Repo 181 105 3.715 1.459 1.876 3.468 6.196 Other 378 105 4.003 1.563 2.575 3.769 6.008 Panel B. RRP Eligible Funds Category N Funds N Months Mean St.Dev. 5th pct Median 95th pct Tsy Securities 141 105 41.604 14.413 17.076 40.531 71.313 ON RRP 138 105 19.764 21.121 0.021 13.373 62.018 Tsy Repo excl. ON RRP 134 105 19.745 8.653 7.720 18.001 34.303 Agency Securities Repo 110 105 12.107 4.607 5.029 11.285 19.088 Other 119 105 6.78 4.912 2.733 5.328 10.978 Note: Data is from monthly N-MFP form filings spanning between April 2016 (N-MFP filings include DLA reporting starting April 2016) to December 2024. The table shows summary statistics on the share of each category within the Daily Liquid Asset. N Funds is the number of funds that reported having an asset in that category at any time during that entire time period. N Months is the number of months any asset of that category appeared in a filing. Tsy Securities are any securities issued by the U.S. Treasury. ON RRP is takeup at the Fed’s ON RRP facility. Tsy Repo excl. ON RRP is any repo collateralized by Treasury securities that doesn’t occur at the Fed’s ON RRP facility. Agency Securities Repo is any repo collateralized by U.S. Government Agency securities. Other is any daily liquid asset that doesn’t fall into any of the above categories. 28

Table 13: ON RRP vs. Other daily liquid assets: policy rate uncertainty = forecast range (1) (2) (3) VARIABLES DLA/AUM ONRRP/AUM DLAexclONRRP/AUM F.FFRrange 0.0497*** 0.0964*** -0.0467*** (0.00724) (0.0127) (0.0115) DaysTillFOMC 0.000168 -0.000765*** 0.000933*** (0.000121) (0.000126) (0.000165) F.FFRrange*DaysTillFOMC -0.000401** 0.00113*** -0.00153*** (0.000189) (0.000208) (0.000251) AUM 0.000938*** 0.000661*** 0.000277** (0.000202) (0.000191) (0.000134) NetTreasuryBillIssuance -5.05e-06 -2.45e-05*** 1.94e-05 (1.03e-05) (9.00e-06) (1.30e-05) MinDLAincrease 0.0669*** -0.0775*** 0.144*** (0.0201) (0.0166) (0.0216) Constant 0.418*** 0.00926 0.409*** (0.0149) (0.0154) (0.0113) Observations 5,038 5,038 5,038 R-squared 0.183 0.174 0.144 Number of group id 65 65 65 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: MMF data is from the N-MFP data sets. DLA/AUM is the share of daily liquid assets (DLA) to assets under management (AUM) of MMF group i at time t. ONRRP/AUM is the share of MMFs takeup at the Fed’s RRP facility to AUM of MMF i at time t F.FFRrange is the range of the forecast for the fed funds rate for the first projection’s quarter forecast across the forecasters in the Blue Chip Financial Forecasts at time t+1. DaysTillFOMC is the number of days from month-end t to day 2 of the next FOMC meeting. MinDLAincrease is a dummy variable, equal to 1 for the period after the minimum requirement for DLA as a share of assets was increased to 25% (October 2023). groups are at the MMF family-fund type-ON RRP eligibility level. A note on the data timing: The Blue Chip Financial Survey for any given month is conducted during the last few days of the previous month. N-MFP data for any month are as of the last day of that month. As a result, the information set window in the N-MFP data corresponds to the information set in the one-month forward Blue Chip data. As a result, in our regression analysis, we pair the month-end N-MFP data with the one-month forward Blue Chip data. Data is at a monthly frequency for the period April 2016 ((N-MFP filings include DLA reporting starting April 2016) to December 2024. The results are from panel data regression including group fixed effects and group-clustered standard errors. 29

Table14: ONRRPvs. Otherdailyliquidassets: policyrateuncertainty=upwardrevisionstotheforecast. Monthly frequency. (1) (2) (3) VARIABLES DLA/AUM ONRRP/AUM DLAexlONRRP/AUM F.FFRup 0.0283*** 0.0877*** -0.0594*** (0.00958) (0.0220) (0.0219) DaysTillFOMC -0.000222** -0.000990*** 0.000768*** (0.000105) (0.000148) (0.000160) F.FFRup*DaysTillFOMC 0.00116*** 0.00411*** -0.00295*** (0.000403) (0.000604) (0.000609) AUM 0.000964*** 0.000731*** 0.000234* (0.000204) (0.000201) (0.000139) NetTreasuryBillIssuance -3.30e-06 -2.47e-05*** 2.14e-05* (9.94e-06) (8.52e-06) (1.25e-05) MinDLAincrease 0.0723*** -0.0562*** 0.128*** (0.0206) (0.0156) (0.0211) Constant 0.433*** 0.0298* 0.403*** (0.0148) (0.0160) (0.0103) Observations 5,038 5,038 5,038 R-squared 0.177 0.142 0.134 Number of group id 65 65 65 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: MMF data is from the N-MFP data sets. DLA/AUM is the share of daily liquid assets (DLA) to assets under management (AUM) of MMF group i at time t. ONRRP/AUM is the share of MMFs takeup at the Fed’s RRP facility to AUM of MMF i at time t F.FFRup is the number of forecasters making as upward revision to the forecast for the fed funds rate for the first projection’s quarter forecast in the Blue Chip Financial Forecasts at time t+1 divided by the total number of forecasters. DaysTillFOMC is the number of days from month-end t to day 2 of the next FOMC meeting. MinDLAincrease is a dummy variable, equal to 1 for the period after the minimum requirement for DLA as a share of assets was increased to 25% (October 2023). groups are at the MMF family-fund type-ON RRP eligibility level. A note on the data timing: The Blue Chip Financial Survey for any given month is conducted during the last few days of the previous month. N-MFP data for any month are as of the last day of that month. As a result, the information set window in the N-MFP data corresponds to the information set in the one-month forward Blue Chip data. As a result, in our regression analysis, we pair the month-end N-MFP data with the one-month forward Blue Chip data. Data is at a monthly frequency for the period April 2016 (N-MFP filings include DLA reporting starting April 2016) to December 2024. The results are from panel data regression including group fixed effects and group-clustered standard errors. 30

Table 15: ON RRP vs. Other daily liquid assets: policy rate uncertainty = downward revisions to the forecast. Monthly frequency. (1) (2) (3) VARIABLES DLA/AUM ONRRP/AUM DLAexlONRRP/AUM F.FFRdown 0.128*** 0.361*** -0.232*** (0.0320) (0.0391) (0.0458) DaysTillFOMC 0.00105*** 0.00110*** -4.15e-05 (0.000133) (0.000134) (0.000151) F.FFRdown*DaysTillFOMC -0.00742*** -0.0116*** 0.00418*** (0.000947) (0.00123) (0.00117) AUM 0.000939*** 0.000742*** 0.000198 (0.000201) (0.000204) (0.000137) NetTreasuryBillIssuance -1.86e-05* -4.96e-05*** 3.10e-05** (1.01e-05) (9.57e-06) (1.31e-05) MinDLAincrease 0.0727*** -0.0717*** 0.144*** (0.0200) (0.0167) (0.0215) Constant 0.424*** 0.00801 0.416*** (0.0154) (0.0144) (0.0106) Observations 5,038 5,038 5,038 R-squared 0.190 0.088 0.119 Number of group id 65 65 65 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Note: MMF data is from the N-MFP data sets. DLA/AUM is the share of daily liquid assets (DLA) to assets under management (AUM) of MMF group i at time t. ONRRP/AUM is the share of MMFs takeup at the Fed’s RRP facility to AUM of MMF i at time t F.FFRdown is the number of forecasters making a downward revision to the forecast for the fed funds rate for the first projection’s quarter forecast in the Blue Chip Financial Forecasts at time t+1 divided by the total number of forecasters. DaysTillFOMC is the number of days from month-end t to day 2 of the next FOMC meeting. MinDLAincrease is a dummy variable, equal to 1 for the period after the minimum requirement for DLA as a share of assets was increased to 25% (October 2023). groups are at the MMF family-fund type-ON RRP eligibility level. A note on the data timing: The Blue Chip Financial Survey for any given month is conducted during the last few days of the previous month. N-MFP data for any month are as of the last day of that month. As a result, the information set window in the N-MFP data corresponds to the information set in the one-month forward Blue Chip data. As a result, in our regression analysis, we pair the month-end N-MFP data with the one-month forward Blue Chip data. Data is at a monthly frequency for the period April 2016 (N-MFP filings include DLA reporting starting April 2016) to December 2024. The results are from panel data regression including group fixed effects and group-clustered standard errors. 31

Figure 1: Measures of policy rate uncertainty: Forecast range and Implied volatility Sources: BlueChipFinancialForecastSurvey,ICAP Note: Thisfigureshowstwomeasuresofpolicyrateuncertainty. FFRRangeistherangeofrespondents’FederalFundsRate (FFR)forecast. DataisfromWoltersKluwer’smonthlyBlueChipFinancialForecasts. Impliedvolatilitytheswaption-implied volatilityoftheone-yearswaprateatahorizonofsixmonthsahead. DataisfromICAP.Weusetheaverageofthelast5days inthecorrespondingmonthtoaligntheBlueChipSurveydata. DataisfortheperiodNovember2010toDecember2024. 32

Figure 2: Forecast Range and WAMs Sources: FormN-MFP,BlueChipFinancialForecastSurvey Note: ThisfigureplotsthefedfundsforecastrangefromtheBlueChipFinancialForecasts(inblue)andtheweightedaverage portfoliomaturity(WAM)fortheMMFsfillingFormN-MFP(inred). DataismonthlyfortheperiodNovember2010to December2024. 33

Figure 3: Implied Volatility and WAMs Sources: ICAP,iMoneyNet Note: Thisfigureplotstheswaption-impliedvolatilityoftheone-yearswaprateatahorizonofsixmonthsaheadfromICAP (inblue)andtheweightedaverageportfoliomaturity(WAM)fortheMMFfromiMoneyNet(inred). Dataisweeklyforthe periodNovember2010toDecember2024. 34

Figure 4: ON RRP Takeup by Counterparty (Daily) Source: FederalReserveBankofNewYork Note: ThisfigureshowstakeupattheONRRPfacilitybrokendownbycounterpartytype. Thedecompositionisdailyand startsfromSeptember23,2013toDecember31,2024. Thered,yellow,green,andblueareasrepresenttakeupbybanks, primarydealers,governmentsponsoredentities(GSEs),andmoneymarketfunds(MMF),respectively. 35

Figure 5: MMF ON RRP Takeup and Policy Uncertainty (Monthly) Sources: FormN-MFP,BlueChipFinancialForecastSurvey Note: ThisfigureplotstheFedFundsForecastrangefromtheBlueChipFinancialForecastsinredandtheONRRPtakeup normalizedbyassetsundermanagement(AUM)inblue. BothONRRPtakeupandAUMarethetotalamountsacrossall moneymarketfunds. ThedataismonthlyfromSeptember2013toDecember2024. 36