Federal Reserve Bank of New York

Staff Reports

Investors Appetite for Money-Like Assets:

The Money Market Fund Industry after the
2014 Regulatory Reform
Marco Cipriani
Gabriele La Spada
Philip Mulder

Staff Report No. 816

June 2017

This paper presents preliminary findings and is being distributed to economists

and other interested readers solely to stimulate discussion and elicit comments.
The views expressed in this paper are those of the authors and do not necessarily
reflect the position of the Federal Reserve Bank of New York or the Federal
Reserve System. Any errors or omissions are the responsibility of the authors.
Investors Appetite for Money-Like Assets: The Money Market Fund Industry after the
2014 Regulatory Reform
Marco Cipriani, Gabriele La Spada, and Philip Mulder
Federal Reserve Bank of New York Staff Reports, no. 816
June 2017
JEL classification: G21

Abstract

We document the reaction of money market fund (MMF) investors and portfolio managers to a
new SEC regulation that came into effect in October 2016. This regulation forces all prime and
municipal MMFs to adopt a system of redemption gates and fees and institutional prime and muni
MMFs to also operate under a floating net asset value (NAV). First, we show that in anticipation
of the new regulatory framework, investors flowed from prime and muni into government MMFs
and especially toward the riskier type of government MMF, agency MMFs, consistent with their
likely higher risk appetite profile. Second, the flows from prime and muni MMFs into
government MMFs mostly occurred within fund families, supporting the hypothesis that the
flows were due to the regulatory changes. This contrasts with past outflows from prime and muni
MMFs into government MMFs, such as those seen during the 2008 crisis; in those cases,
investors often left not only their prime and muni fund but also the fund family. We relate such
differences in investors behavior to their appetite for money-like assets, whose supply was
impacted by the new regulation, as opposed to traditional flight-to-safety motives. Third, the
outflow from prime and muni MMFs was stronger for institutional investors, consistent with the
fact that these investors are more elastic to industry developments and have been subject to a
stricter regulation than retail investors. Finally, as a result of the outflows from prime and muni
funds, MMF credit to the private sector has been significantly reduced, whereas credit to
government-sponsored enterprises (and in particular Federal Home Loan Banks) has increased
substantially.

Key words: money market funds, regulation

_________________

Cipriani, La Spada, Mulder: Federal Reserve Bank of New York (emails:

marco.cipriani@ny.frb.org, gabriele.laspada@ny.frb.org, philip.mulder@ny.frb.org). The authors
thank Adam Spiegel and Joseph Step for outstanding research assistance during this project. The
views expressed in this paper are those of the authors and do not necessarily reflect the position
of the Federal Reserve Bank of New York or the Federal Reserve System.
1 Introduction

After the 2008 run on the money market fund (MMF) industry, academics and policy
makers have spent much effort to try to understand the sources of fragility in the
industry and make it more resilient. These efforts have led the SEC to approve a
new regulation in July 2014, which came into effect in October 2016.
U.S. money market funds are open-ended mutual funds that invest in money market
instruments. MMFs are pivotal players in the financial markets: as of the end of 2014,
they had roughly $3 trillion in assets under management and held approximately 35%
of the global outstanding volume of commercial papers (see ICI, 2015). In particular,
they are a critical source of short-term financing for financial institutions: in May
2012, they provided roughly 35% of such funding, with 73% of prime MMF assets
consisting of debt instruments issued by large global banks (see Hanson, Scharfstein,
and Sunderam, 2015). Similarly to other mutual funds, MMFs are paid fees as a
fixed percentage of their assets under management (AUM) and are therefore subject
to the tournament-like incentives generated by a positive flow-performance relation
(see La Spada, 2017). In contrast to other mutual funds, however, until the new
SEC regulation came into effect in October 2016, all MMFs aimed to keep the NAV
of their assets at $1 per share; they have done so by valuing assets at amortized
cost and providing daily dividends as securities progress toward their maturity date.
Since their deposits are not insured by the government and are daily redeemable,
this stable NAV feature makes MMFs susceptible to runs. If a fund breaks the
buck, i.e., its NAV drops below $1, investors will likely redeem their investment en
masse (i.e., run on the fund) to preserve the value of their capital. This happened on
September 16, 2008, when Reserve Primary Fund, the oldest MMF, broke the buck
after writing off debt issued by Lehman Brothers.
The interaction between risk-taking incentives and exposure to runs made MMFs
a key ingredient of the recent financial crisis. Indeed, in September 2008, the run
on Reserve Primary Fund quickly spread to other prime MMFs, triggering investors
redemptions of more than $300 billion within a few days after Lehmans default. This
caused a severe shortage of short-term credit to the banking sector (see Kacperczyk
and Schnabl, 2013). In the summer of 2011, a slow-motion run hit the prime
MMF sector as fears about European sovereign debt problems mounted, causing
redemptions of more than $170 billion in approximately two months and disrupting
the ability of both European and non-European firms to raise financing in the money

1
markets (see Chernenko and Sunderam, 2014).
MMFs are regulated under Rule 2a-7 of the Investment Company Act of 1940. This
regulation restricts fund holdings to short-term, high-quality debt securities. For
example, prime MMFs can only hold commercial papers that carry either the high-
est or second-highest rating from at least two of the nationally recognized credit
rating agencies. During the January 2002-August 2008 period, prime MMFs were
not permitted to hold more than 5% of investments in second tier (A2-P2) paper
or more than a 5% exposure to any single issuer (other than the U.S. government
and agencies). Also, the weighted average maturity of the portfolio was capped to
90 days.1 The MMF industry is divided in three main sectors based on funds port-
folio composition: 1) prime MMFs mainly invest in private unsecured and secured
debt in addition to Treasuries and Agency debt; 2) muni MMFs mainly invest in
municipal and local authorities debt; 3) government MMFs mainly invest in Trea-
suries and Agency debt and can only lend to the private sector through repurchase
agreements (repos) collateralized by Treasuries or Agency debt. Government MMFs
can be further divided in two subgroups: Treasury MMFs, which can only invest in
Treasuries and repos collateralized by Treasuries; and Agency MMFs, which can also
invest in Agency debt and repos collateralized by Agency debt. MMFs can also be
divided into institutional and retail funds based on the profile of their investors. In
September 2008, prime&muni MMFs were at the center of the run on the industry,
whereas government MMFs actually experienced an inflow of investors; in particular,
within the prime&muni sector, institutional MMFs were most affected, whereas the
outflow from the retail ones was smaller and slower.
On July 23, 2014, the SEC approved a new set of rules for MMFs (SEC Release No.
IC-31166) focusing on the prime&muni segment of the industry. The main pillar
of these rules is that from October 2016, institutional prime&muni MMFs must
sell and redeem shares based on the current market-based value of the securities in
their underlying portfolios. That is, they have to move away from a stable NAV to
1
In 2010, after the turmoil generated by the collapse of Reserve Primary Fund, the SEC adopted
amendments to Rule 2a-7, requiring prime MMFs to invest in even higher-quality assets with shorter
maturities. E.g., the weighted average maturity was capped to 60 days (SEC Release No. IC-29132).
Funds were also required to have enhanced reserves of cash and readily liquidated securities to
meet redemption requests, and could invest only 3% (down from 5%) of total assets in second tier
securities. These first regulatory changes, while making MMF portfolios safer, did not alter the
economic structure of MMFs and, in particular, did not alter their runnability or the money-like
quality of a MMF investment. For this reason, the 2010 regulatory changes did not create large
outflows of investors from any sector of the MMF industry and are not the subject of this analysis.

2
a floating NAV. The purpose of this regulatory change is to eliminate (or at least
mitigate) the risk of runs by making investments in MMFs less money-like (and more
similar to investments in traditional mutual funds). In addition, all prime&muni
MMFs will have discretion to impose gates on redemptions or charge redemption
fees of up to 2% in times of stress.2
The new regulation came into effect in October 2016. In this paper, we study the
evolution of the MMF industry ahead of the implementation of the new regulation.
We do so by studying MMF portfolio data from MMF regulatory filings with the
SEC (form N-MFP). We find that from November 2015 to October 2016, there have
been large outflows from prime&muni MMFs to government MMFs. Moreover, the
outflows from prime&muni into government MMFs have occurred mostly within fund
families: investors have switched from one MMF type to the other but have kept
their investment in the same fund complex; this supports the hypothesis that the
observed outflows from prime&muni into government MMFs are due to the regulatory
changes. These within-industry investors flows are quantitatively very significant: in
the 12 months we consider, roughly $1.2 trillion have flowed from the prime&muni
segment into the government one; to put this number in perspective, the overall
outflow from prime&muni funds during the 2008 crisis was only $300 billion. This
reshuffling of investors within the MMF industry indicates that investors care about
the money-likeness of their investment: they flow from funds that are no longer
money-like (prime&muni funds, which now operate with a floating NAV and can
impose redemption gates and fees) to funds that are still money-like (government
funds, which still operate with a fixed NAV and do not have a system of redemption
gates and fees).
We also document that investors in prime&muni MMFs have mainly flowed toward
the riskier segment of the government MMF sector: agency MMFs. This is consistent
with the likely higher risk-appetite profile of investors coming from the prime&muni
segment of the industry. Finally, the observed outflows from prime&muni MMFs
are stronger for institutional investors, consistent with the fact that they are more
elastic with respect to developments in the industry and have also been subject to a
stricter version of the regulation.
Section 2 describes the N-MFP form. Section 3 describes investors flows within the
MMF industry. Section 4 describes the changes in MMF portfolios after the new
2
Gates and fees can be imposed when a funds liquid assets fall below 30 percent of its total
assets.

3
regulation by looking at private versus public investment.

2 The N-MFP Form

The Form N-MFP is a publicly available regulatory filing that every MMF is required
to submit to the SEC each month. Each filing contains information on a funds bal-
ance sheet, share classes, security-level portfolio holdings, performance, and investor
flows. Funds reports all of this information as of the end of the month and submit
their filings to the SEC within the first five business days of the next month. The
SEC makes all N-MFP submissions publicly available. The form was created in May
2010 along with a set of MMF reforms adopted in the immediate aftermath of the
financial crisis.3 The first N-MFP filings were submitted in December 2010 and have
continued every month since. Funds occasionally submit their forms late or make
small corrections by amending filings from previous months.
We download, parse, and clean information from the Form N-MFP to construct our
monthly panel dataset of MMFs. A funds N-MFP filing specifies whether the fund
is a feeder or a master fund, whether it is liquidating or merging with another fund,
and whether it is a prime fund, a municipal fund, an agency fund, or a treasury fund.
The filing reports the funds month-end dollar weighted average portfolio maturity,
total net assets, and the annualized gross yield for the last seven days of the month.
The fund also reports its fixed NAV, at which shares are redeemed and subscribed,
and its shadow price, defined as the NAV calculated using market prices rather
than amortized costs and rounded to the fourth decimal place, with and without
capital support agreements.
One fund can have multiple share classes, that is, types of shares that differ in terms
of fees, minimum investment, and other characteristics. For each of its share classes,
the fund reports the net assets, the aggregate monthly redemptions and subscriptions
by shareholders, and the annualized net yield for the last seven days of the month.
Generally, different share classes are offered to different types of investors, such as
institutional or retail investors. In the empirical analysis, where we compare the
different behavior of retail and institutional investors, we identify institutional and
retail share classes based on the classification reported by iMoneyNet (a private
provider of MMF data).
3
See footnote 1.

4
Finally, each fund reports detailed information on the securities in its portfolio.
For each security, the fund specifies the name of the issuer, the title of the issue
and its CUSIP, the general category of the investment (e.g., variable rate demand
note, government agency debt, certificate of deposit, asset backed commercial paper),
whether the security is a repurchase agreement and, if so, the value and type of its
collateral, the maturity date of the security, the principal value of the security, the
market value of the security (with and without capital support from the funds
sponsor), the value of the security at amortized cost, and the share of the funds
total net assets that are invested in the security. Moreover, for each security in a
funds portfolio in each month, we obtain (long-term) credit rating information on
the issuer from Moodys, S&P, and Fitch for that month using IHS Markit data;
from January 2015 through February 2017, we have credit ratings for more than 84
percent of the total value of all portfolio items.

3 Investors Flows

3.1 The Response of MMF Investors to the New Regulation

This section studies the response of MMF investors to the 2014 SEC regulation.
Figure 1 and Table 1 show that from January 2015 to February 2017, the total net
assets (TNA)4 of the whole MMF industry remain roughly constant at around $3
trillion.
Within the industry, however, the relative size of the MMF categories changes dra-
matically. The TNA of prime&muni funds decrease by $1,315 billion (i.e., by ap-
proximately 65%), while the TNA of government funds increase by $1,191 billion
(i.e., by more than 115%). As a result, the share of government funds in the MMF
industry goes from 33.3% in January 2015 to 75.4% in February 2017. The bulk
of these flows (about 60%) occurs between June and October 2016, that is, before
the SEC regulation comes into effect. Indeed, the first major flow from prime&muni
into government funds directly attributed to the new SEC regulations occurred in
4
A funds TNA is the total value of its portfolio of securities minus its debt liabilities. Note
that although MMFs can be thought of as deposit-like institutions, from an accounting point of
view, their shares are classified as an equity instrument and not included among the liabilities.
MMFs usually issue very little debt: indeed, between January 2015 and February 2017, total MMF
liabilities were only 1.3% of their total assets. For this reason, the industrys TNA are very close
to the industrys total assets under management (AUM), which measure the industry size.

5
 TNA January 2015 February 2017
Total $3,057 bn $2,931 bn -$126 bn
Prime&Muni $2,037 bn $722 bn -$1,315 bn
Government $1,019 bn $2,210 bn +$1,191 bn
Government Share 33.3% 75.4% +42.1 pp

Table 1. Total Net Assets by Fund Category.

December 2015, when Fidelity converted $130 billion of its prime MMFs into gov-
ernment MMFs. From November 2016 to February 2017, after the adoption of the
new SEC regulation, there has been a very modest net flow back into prime&muni
funds.5
When looking at flows between the different types of MMFs, we study the changes
in their TNA (as opposed to looking at redemptions and subscriptions) because we
want to capture those instances in which a fund reclassified itself from prime&muni
to government. Such reclassifications have indeed occurred after the SEC regulation
came into effect; for instance, as we mentioned above, in December 2015, Fidelity
converted a large portion, roughly 34%, of its prime MMFs into government MMFs.
A fund reclassification from prime to government will show up in our data as an
increase in the familys government TNA and a decrease in the familys prime TNA.
In contrast, if we focused on investors redemptions and subscriptions, we would miss
the movement of investors money from a prime&muni to a government MMF, as
the reclassification would not require investors to redeem their shares.6
The response of MMF investors to the new regulation can be seen as driven by their
desire to hold money-like assets in their portfolio. Indeed, from an investors per-
spective, the introduction of redemption gates and liquidity fees and the adoption of
a floating NAV makes an investment in a prime&muni MMF less similar to a regular
bank deposit. In response to these regulatory changes, investors move their assets
into government funds, which instead preserve the money-like features historically
associated with the MMF industry.
In our data, we do not directly observe individual investors flows; we only observe
5
Between November 2016 and February 2017, prime&muni TNA increased by $24.4 bn, while
government TNA decreased by $7.9 bn, increasing the prime&muni share of the MMF industry
from 23.9% to 24.6%.
6
However, even if we look at redemptions and subscriptions in prime&muni and government
MMFs, we obtain similar results to those described above.

6
Figure 1. MMF Total Net Assets by Fund Category: Government vs. Prime&Muni.
Solid black line: share of government MMFs in percentages (right y-axis).

fund TNA at month-end and aggregate redemptions and subscriptions within-month.

The growth of government MMFs and the decline of prime&muni MMFs may be un-
related to the SEC regulation. For example, changes in macroeconomic conditions
and available investment opportunities might have lead investors in prime&muni
funds to leave the MMF industry altogether, while outside investors decide to place
their cash into government funds; alternatively, some government funds or fund fam-
ilies specializing in government funds may have had a better advertising strategy
than prime&muni funds in 2016. If this were the case, the flows from prime&muni
into government MMFs would be equally likely to occur across families as within
families: for instance, if families specializing in government MMFs have better ad-
vertising strategies, they are likely to attract investors from prime&muni MMFs of
other families. In contrast, if the growth of government MMFs comes from the de-
cline of prime&muni MMFs due to the new SEC regulation, our data are more likely
to show that the changes in the relative importance of these MMF categories happen
through within-family flows; indeed, since fund families have a strong incentive to
retain their clients, one would expect that investors desiring to exit prime&muni
MMFs and invest in government MMFs would do so by moving their money within
the same family.

7
To test whether flows from prime&muni to government MMFs occurred across or
within families, we run the following monthly regression at the family level:

GovtT N Ait = i + t + 0 P rimeT N Ait + 1 P rimeT N Ait1 + it , (1)

where GovtT N Ai is the monthly change in the TNA of family is government

MMFs; P rimeT N Ai is the monthly change in the TNA of family is prime&muni
MMFs; and i and t are family and month fixed effects. We also include the one-
month lagged change in prime&muni TNA to account for the possibility that within-
family flows from prime&muni into government MMFs are not contemporaneous.
The sample is from November 2015 to October 2016. Standard errors are clustered at
the family level. The results of the regression are in Table 2. The estimated slope on
contemporaneous outflows from prime&muni MMFs is -0.81, which means that there
is almost an exact 1-to-1 relationship between the outflow from prime&muni MMFs
and the inflow to government MMFs within each family (indeed, the coefficient is not
statistically different from -1); the lagged coefficient on P rimeT N Ai is statistically
insignificant, although slightly negative (-0.02), showing that almost all of the flows
from prime&muni into government MMFs are contemporaneous. The results show
that before the SEC reform came into effect, investors who did not want to keep their
assets in the newly regulated prime&muni MMFs simply shifted their investment
to government MMFs within the same family. Note that the relationship between
outflows from prime&muni MMFs and inflows to government MMFs is tight. Indeed,
Figuresand plot GovtT N A against P rimeT N A between November 2015 and
October 2016 both in levels and on a logarithmic scale, together with the best-fit
regression line;7 as the figure shows, almost all the families lie close the regression
line.
The regression reported on the first column of Table 2 is run on all family and periods,
irrespective of whether in a given month the family experienced an outflow or an
inflow from its prime&muni MMFs. The regression results do not substantial change
if we run our regression on outflows only (see column (3)). Moreover, if we estimate
a separate slope for outflows and inflows to prime& muni MMFs, we observe that the
slope for outflows is very similar to that of our original regression (-0.89 vs. -0.81).
7
Using a logarithmic scale allows us to visualize more clearly the behavior of all the families
in the sample, given than changes in TNA vary greatly across families, ranging from a few dozen
million to several hundred billion dollars. In the chart, the prime&muni outflows are given a positive
sign to allow us to use logs.

8
Figure 2. Within-family flows: government vs. prime&muni. Time period: Novem-
ber 2015-October 2016. x-axis: total outflow from a familys prime&muni MMFs;
y-axis: total inflow to a familys government MMFs. The dashed line is the regression
line of government inflows on prime&muni outflows (slope: 1.00 (0.06), R2 =0.92).

In contrast, the coefficient on inflows is of the opposite sign (1.00), significant, and
statistically indistinguishable from 1. In other words, in those months and families in
which prime&muni MMFs increased their TNA, we observe a 1-to-1 increase also in
the TNA of government MMFs. This indicates that when there was an inflow into a
familys prime&muni MMFs, the family experienced an overall increase in its TNA;
in contrast, when there were outflows from its prime&muni MMFs, the familys total
TNA did not change as investors flowed from one type of MMF to the other within
the family.
It is instructive to have a control sample and compare the regression results described
above with those obtained from the same regressions run on November 2014-October
2015 data, that is, more than a year before the SEC regulation came into effect. The
results are reported in columns (5)-(8) of Table 2. In the November 2014-October
2015 regressions, the slope coefficient is positive (e.g., in the baseline regression
the coefficient is 0.15 and statistically significant), suggesting a positive relationship
between a familys government and prime&muni TNA, even after conditioning on
month and family effects. In other words, it is only in the year before the regulation
came into effect that investors moved their assets from prime&muni to government
MMFs within the same family.
One possible concern is that the results described above may be driven by the be-
havior of outliers. Indeed, as shown in Table 3, there is a lot of heterogeneity across

9
 GovT N Ait
11/201510/2016 11/201410/2015
(1) (2) (3) (4) (5) (6) (7) (8)
P rimeT N Ait -0.81*** -0.81*** 0.15** 0.15**
(0.14) (0.15) (0.06) (0.06)
P rimeT N Aitt1 -0.02 0.05
(0.07) (0.08)
P rimeT N Ait < 0 -0.91*** -0.89*** 0.07 0.00
(0.12) (0.12) (0.09) (0.07)
P rimeT N Ait 0 1.00** 0.27*
(0.40) (0.14)
R2 0.68 0.68 0.79 0.74 0.03 0.03 0.00 0.04
N 1008 1008 466 1008 1140 1140 505 1140

Table 2. Within-family flows: government vs. prime&muni. The sample is a bal-

anced panel of within-family monthly flows (that is, it only includes families that
are active in both the prime&muni and the government segment in every month);
GovT N Ait is the monthly change in a familys goverhment TNA; P rimeT N Ait is
the monthly change in a familys prime&muni TNA; the regressions includes month
and family fixed effects; standard errors (in parentheses) are clustered at the fam-
ily level. Column (1) to (4) are estimated on November 2015-October 2016 data.
Columns (1), (2) and (4) are run on all flows; column (3) is run only on month-
families with an outflow from prime&muni funds. Columns (5) to (8) report the
results of the same regressions using November 2014-October 2015 data. ***, **, *
represent 1%, 5%, and 10% statistical significance, respectively.

10
Figure 3. Within-family flows: government vs. prime&muni; log-log scale. Time
period: November 2015-October 2016. x-axis: log of total outflow from the familys
prime&muni MMFs; y-axis: log of total inflows to the familys government MMFs.
The dashed line is the regression line of government inflows on prime&muni outflows
(slope: 0.97 (0.05), R2 =0.85).

fund families. In particular, although the average TNA across families is $34 bn, 50
percent of family have TNA of $2.2 bn or less; indeed, the standard deviation of TNA
across families is $80 bn, and the distribution of family sizes is heavily skewed to the
right. These numbers suggest that there are a few families with very large TNA rel-
ative to the rest of the industry. In principle, it may be that only very large families
have a sufficiently strong franchise value or reputation to retain their prime&muni
investors in the transition to government funds. Insofar as these families are those
that have exhibited the largest outflows from prime&muni funds, they may also have
driven our estimated coefficients.
To address this concern, Table 4 implements robustness checks of the results from
Table 2. Columns (1)-(3) replicate the monthly regressions from Table 2 but exclud-
ing all those families whose prime&muni outflow in any month is in the top 5% of the
cross-sectional distribution. The results largely align with those of Table 2, showing
that the impact of changes in prime&muni TNA on changes in government TNA is
not driven by the behavior of the largest families.

11
 Prime&Muni Funds Government Funds Entire Family
Mean TNA $24.5 bn $18.6 bn $34.6 bn
SD TNA $46.3 bn $38.0 bn $80.2 bn
Median TNA $3.8 bn $2.5 bn $2.2 bn
Range TNA $258 bn $355 bn $538 bn
Mean Change -$1.0 bn $0.6 bn $0.5 bn
SD Change $4.5 bn $3.0 bn $25.7 bn

Table 3. Summary statistics of TNA and net flows within fund families between
November 2015 and October 2016. Note that the statistics on prime&muni and
government funds exclude families with no funds in those categories.

GovtT N A
(1) (2) (3)
P rimeT N A -0.74***
(0.06)
P rimeT N A < 0 -0.93*** -0.85***
(0.06) (0.05)
P rimeT N A 0 0.59*
(0.31)
R2 0.70 0.82 0.77
N 756 305 756

Table 4. Within-family flows: government vs. prime&muni; robustness checks. Re-

gression in column (1)-(3) replicate regressions in column (1)-(3) of Table 2 excluding
all those families whose prime&muni outflow in any month is in the top 5% of the
cross-sectional distribution for that month. Time period: November 2015-October
2016. Standard errors (in parentheses) are clustered at the family level. ***, **, *
represent 1%, 5%, and 10% statistical significance, respectively.

12
3.2 Institutional vs. Retail Investors

The new SEC regulation impacts institutional and retail investors in prime&muni
MMFs differently. The regulation requires both institutional and retail funds to
adopt liquidity fees and redemption gates; only institutional MMFs, however, must
switch to a floating NAV. This difference means that an investment in prime&muni
MMFs is further away from a money-like investment for institutional investors than
for retail investors. Moreover, the experience of the 2008 run on MMFs has shown
that, in general, institutional investors are much more responsive (e.g., to economic
news) than retail investors. For both reasons, we should expect that, although both
institutional and retail investors transfer their funds from prime&muni MMFs into
government MMFs, institutional investors do so to a greater extent. This is indeed
what we observe in the data.
Figure 4 and Table 5 show TNA by fund category from January 2015 to February
2017 separately for institutional and retail share classes. The TNA of institutional
prime&muni MMFs decrease by roughly 86% (i.e., by $937 bn), while the TNA of
retail prime&muni MMFs decrease by only 43% (i.e., by $287 bn). As a result,
the share of government funds in the institutional segment has increased by over 49
percentage points (pp), from 41.6% to 91.1%, whereas that in the retail segment has
increased by only 38.3 pp, from 22.4% to 60.7%.8
Indeed, prime&muni MMFs declined from 58.4% of all institutional TNA in January
2015 to 8.9% in February 2017; the decline is more muted for retail TNA, from
77.6% in January 2015 to 39.3% in February 2017. Indeed, in Table 6, we regress
the monthly share of prime&muni TNA in a familys institutional and retail share
classes9 on a time dummy equal to 1 after the regulation came into effect (October
2016) and its interaction with an institutional dummy equal to 1 for institutional
share classes. As controls, we add family-share class type fixed effects. That is, on a
balanced panel of families with prime&muni MMFs from October 2015 to February
2017, we run the following regression:

Share(P rime&M uni)ikt = i,k + 0 Octt + 1 Octt Instk + ikt ,

8
We obtain similar results if we look at net flows into government and out of prime&muni for
retail and government share classes separately. See Appendix.
9
In other words, for each family-month, there are two observations: the ratio of prime&muni
TNA over all institutional TNA and the ratio of prime&muni TNA over all retail TNA.

13
Figure 4. MMF Total Net Assets by Fund Category and Investor Type: Institutional
(left) vs. Retail (right). Solid black line: share of government MMFs in percentages
(right y-axis).

Institutional Retail
Jan. 2015 Feb. 2017 Jan. 2015 Feb. 2017
Total $1,860 bn $1,694 bn -$166 bn $854 bn $957 bn +$103 bn
Prime&Muni $1087 bn $150 bn -$937 bn $663 bn $376 bn -$287 bn
Government $774 bn $1544 bn +$770 bn $191 bn $581 bn +$390 bn
Prime&Muni 58.4% 8.9% -49.5 pp 77.6% 39.3% -38.3 pp

Table 5. Total Net Assets by Fund Category and Investor Type: Institutional vs.
Retail.

where i is the fund family, k is the share class type (institutional or retail), and
t is the month. Standard errors are clustered at the family level. The coefficient
on the regulation time dummy is negative and significant, indicating that the share
of prime&muni funds declined in the retail MMF segment after the regulation; in
particular, after controlling for family-share class type fixed effects, the share of
prime&muni MMFs among retail share classes decreased by 32 pp after October
2016. However, the interaction coefficient with the institutional dummy is also neg-
ative and significant, indicating that such a reduction was even more pronounced
for institutional MMFs; in fact, the share of prime&muni MMFs among institutional
share classes decreased by an additional 13 pp after October 2016.

14
 ShareP rimeikt
Octt -0.32***
(0.04)
Octt Instk -0.13***
(0.05)
R2 0.34
N 1649

Table 6. Institutional vs. Retail: family-level monthly regression. The regression

equation is Share(P rime&M uni)ikt = i,k + 0 Octt + 1 Octt Instk + ikt , where
i is the family, k is the share-class type (institutional or retail), t is the month, and
Share(P rime&M uni)ikt is the share of prime&muni TNA in family is share classes
of type k in month t, Octt is a dummy equal to 1 from October 2016, and Instk is
a dummy equal to 1 for institutional share classes. i,k are family-share class type
fixed effects. The sample is a balanced panel of families with prime&muni MMFs
from October 2015 to February 2017. Standard errors (in parentheses) are clustered
at the family level. ***, **, * represent 1%, 5%, and 10% statistical significance,
respectively.

3.3 The Risk Appetite of Ultimate Investors

Prime&muni MMFs can hold riskier assets (e.g., commercial papers and certificates
of deposit) than government MMFs and therefore offer higher yields to their in-
vestors. For example, from 2011 to 2015, the TNA-weighted average gross yield of
prime&muni MMFs was 23 basis points, while that of government MMFs was only 11
basis points. Hence, investors that previously invested in prime&muni MMFs have
arguably a greater risk-appetite than traditional investors in government MMFs and
should be more likely to flow into higher-yield government funds. Namely, we should
expect larger inflows to agency MMFs than to treasury MMFs, since agency funds
can also buy Agency debt and repos backed by Agency debt. From 2011 to 2015,
agency MMFs earned on average 5.5 basis points more than treasury MMFs. As
expected, the data show that inflows are concentrated among agency MMFs.
Figure 5 and Table 7 show the TNA of both agency and treasury MMFs from January
2015 to February 2017: agency funds grow by roughly $1,060 billion (i.e., roughly
194%), while treasury funds grow only by $131 billion (i.e., less than 28%). The
overall surge in the share of government funds in the MMF industry is almost entirely
accounted by agency funds, whose share goes from 17.9% in January 2015 to 54.8%
in February 2017. In contrast, the share of treasury MMFs increases by only 5 pp,

15
Figure 5. MMF Total Net Assets by Fund Category: Treasury, Agency, and
Prime&Muni.

TNA January 2015 February 2017

Total $3,057 bn $2,931 bn -$126 bn
Agency $546 bn $1,605 bn +$1059 bn
Treasury $473 bn $604 bn +$131 bn
Agency Share 17.9% 54.8% +36.9 pp
Treasury Share 15.5% 20.6% +5.1 pp

Table 7. Total Net Assets by Fund Category: Agency vs. Treasury

from 15.5% to 20.6%.

To quantify the risk appetite of the investors that previously placed their cash in
prime&muni MMFs, we run the following monthly regression at the family level::

GovtT N Aikt = ik + t + 1 P rimeT N Ait + 2 Agcyk P rimeT N Ait + ikt

where k is either agency or treasury; Agcyk is a dummy equal to 1 if k is

agency; GovtT N Aikt is the change in the TNA of family is type-k govern-
ment MMFs;P rimeT N Ait is the corresponding change in the TNA of family is

16
Figure 6. Within-family flows: agency vs. prime&muni, and treasury vs.
prime&muni. Time period: November 2015-October 2016. x-axis: total outflow
from the familys prime&muni MMFs; y-axis: total inflows to the familys agency
and treasury MMFs (separately). The dashed lines are the regression lines from the
regression of inflows on outflows allowing a different slope for inflows in agency and
treasury MMFs (blue line: slope: 0.10 (0.04), R2 = 0.28; red line: slope: 0.90 (0.08),
R2 = 0.94).

prime&muni MMFs; ik are family-fund type fixed effects; and t are month fixed
effects. The coefficients of interest are 1 and 2 : 1 represents the fraction of out-
flows from prime&muni MMFs that flows into treasury MMFs, and 1 +2 represents
the fraction of outflows from prime&muni MMFs that flows into agency MMFs. The
results of the regression are in column (1) of Table 8: a one-dollar outflow from a fam-
ilys prime&muni funds generates a 78 cent inflow to the familys agency funds (sig-
nificant at the 1% level) and only a 3 cent inflow to the familys treasury funds (statis-
tically insignificant). Figure 6 plots GovtT N Aik against P rimeT N Ai separating
inflows to agency MMFs from inflows to treasury MMFs, together with the regression
lines of treasuries and agency inflows on prime&muni outflows. For both agencies
and treausuries inflows, the relationship with prime&muni outflows is very tight: all
points in the chart lie close their regression lines; however, the predicted inflow into
a familys government MMFs following an outflow from the familys prime&muni
MMFs is much larger for the agency segment of government MMFs.

17
 GovT N Aikt
(1) (2) (3)
Agcyk P rimeT N Ait -0.75***
(0.17)
P rimeT N Ait -0.03
(0.02)
Agcyk P rimeT N Ait < 0 -0.85*** -0.81***
(0.15) (0.16)
P rimeT N A < 0 -0.03 -0.04
(0.02) (0.03)
Agcyk P rimeT N Ait 0 0.61**
(0.30)
P rimeT N Ait 0 0.20*
(0.11)

R2 0.67 0.77 0.72

N 2016 932 2016

Table 8. Within-family flows: agency vs. prime&muni, and treasury vs.

prime&muni. The sample is a balanced panel of within-family monthly flows (that
is, it only includes families that are active in every month) from November 2015 to
October 2016; k is either agency or treasury; Agcyk is a dummy equal to 1 if
k is agency; GovT N Aikt is the monthly change in a familys type-k goverhment
TNA; P rimeT N Ait is the monthly change in a familys prime&muni TNA; the
regressions includes month and family fixed effects; standard errors (in parentheses)
are clustered at the family level. Columns (1) and (3) are run on all flows; column
(2) is run on month-families with an outflow from prime&muni funds. ***, **, *
represent 1%, 5%, and 10% statistical significance, respectively.

18
3.4 Past Episodes of Outflows from Prime&Muni MMFs

It is instructive to compare portfolio flows after the SEC regulation came into effect
in October 2016 with what happened during two past episodes of turmoil in the
MMF industry: the period around September 2008 after the Primary Reserve Fund
broke the buck (the so-called 2008 MMF Run), and the second half of 2011 at the
height of the European debt crisis (the so-called Silent Run). Whereas portfolio
flows over the 2015-2016 were the results of new regulation, both the 2008 MMF Run
and the Silent Run were driven by investors concerns over the safety of prime&muni
MMFs (after Lehmans bankruptcy and after the European debt crisis respectively).
Figure 7 and Table 9 show MMF TNA from January 2008 to January 2011. From
August 2008 to October 2008, investors redeemed $464 bn from prime&muni MMFs
and invested $486 bn in government MMFs. Outflows from prime&muni into gov-
ernment funds subsided after October 2008: the share of government MMF assets
jumped to 38% at the end of September 2008, reached a peak of 40% in October, and
very gradually reverted to its pre-run level over the following three years. Moreover,
institutional investors accounted for almost all of the net outflows from prime&muni
MMFs. We detect a similar pattern if we look at the Silent Run of 2011. Figure 8
and Table 10 show MMF TNA from January 2011 to January 2013. From May to
December 2011, when investors were most skeptical of the viability of European debt,
investors only redeemed $228 bn from prime&muni MMFs and invested $190 bn in
government MMFs. The share of government MMF assets jumped by only a couple
of percentage points during the Summer of 2011 and remained stable afterwards.
These numbers pale with respect to what we observed in anticipation to the new
SEC regulation becoming effective in October 2016. As we described in Section 3.1,
between January 2015 and February 2017, investors roughly moved $1tn from the
prime&muni MMF sector to the government MMF sector. As we discuss in Section
3.1.1, although outflows from prime&muni funds were more pronounced for institu-
tional asset classes, both institutional and retail classes were significantly affected.
As a result, the overall share of government MMFs jumped from 33% to 75%. The
contrast between recent investors behavior and that in 2008 and 2011 is even more
surprising given that, for example, the 2008 MMF run happened in the midst of the
financial crisis and after Lehman bankruptcy; in contrast, the recent outflows from
prime&muni MMFs occurred in a time of relative calm in financial markets.
A possible interpretation of the much larger impact of the SEC regulation on MMF

19
flows is that the new rule affected investors desire to hold prime&muni MMF in a
more fundamental way than credit-risk concerns. By altering the system of float-
ing NAV and by imposing a system of gates and fees, the new regulation made
prime&muni MMF less similar to money instruments. Since one of the main rea-
sons to invest in MMF is the similarity of the MMF investment to money, the new
regulation made the typical MMF investor to switch his or her investment toward
government funds. The interpretation given above is consistent with the fact that,
as documented in Section 3.1, between 2015 and 2017, flows from prime&muni into
government MMFs were mainly intra-family flows: if the reason behind such flows is
investors desire to preserve the money feature of their investment, there is no reason
for them to move to a different family of funds (and disrupt the investor-sponsor
relationship).
In contrast, when flows from prime&muni funds into government funds are due to
credit-risk concerns (as was the case in 2008 and 2011), there is less reason for
investors to remaining within the same family: the preservation of the investor-
sponsor relationship should be weighted against the fact that MMF families may
differ in their riskiness. Accordingly, we observe that, both during the 2008 MMF
Run and the 2011 Silent Run, the within-family relationship between prime&muni
outflows and government inflows was weaker.

20
Figure 7. The 2008 MMF Run. (Above) MMF Total Net Assets by Fund Category:
Government vs. Prime&Muni. (Below) MMF Total Net Assets by Fund Category
and Investor Type: Institutional (left) vs. Retail (right). Solid black line: share of
government MMFs in percentages (right y-axis).

Aug. 08 Sep. 08 Oct. 08 Nov. 08 Dec 08 Jan 09 Jan 11

Total $3,478 bn $3,331 bn $3,501 bn $3,615 bn $3,754 bn $3,825 bn $2,700 bn
Prime&Muni $2,556 bn $2,067 bn $2,092 bn $2,157 bn $2,269 bn $2,341 bn $1,916 bn
Government $923 bn $1,264 bn $1,409 bn $1,458 bn $1,458 bn $1,484 bn $784 bn
Gov. Share 26.5% 37.9% 40.2% 40.3% 39.6% 38.8% 29.0%

Table 9. MMF Total Net Assets Around the 2008 Financial Crisis: Prime&Muni Vs.
Government.

21
Figure 8. The 2011 Silent Run. (Above) MMF Total Net Assets by Fund Category:
Government vs. Prime&Muni. (Below) MMF Total Net Assets by Fund Category
and Investor Type: Institutional (left) vs. Retail (right). Solid black line: share of
government MMFs in percentages (right y-axis).

May 11 Jun. 11 Jul. 11 Aug. 11 Sep. 11 Oct. 11 Jan 13

Total $2,702 bn $2,661 bn $2,541 bn $2,618 bn $2,604 bn $2,583 bn $2,666 bn
Prime&Muni $1,947 bn $1,856 bn $1,790 bn $1,780 bn $1,738 bn $1,714 bn $1,765 bn
Government $755 bn $805 bn $751 bn $838 bn $866 bn $869 bn $901 bn
Gov. Share 28.0% 30.2% 29.6% 32.0% 33.3% 33.7% 33.8%

Table 10. MMF Total Net Assets Around the 2011 European Debt Crisis:
Prime&Muni Vs. Government.

As figure 9 shows, in 2008, the slope of the best-fit line of within-family flows into
government MMFs versus outflows from prime&muni MMFs is only 0.53; this con-
trasts with the almost 1-to-1 relation we observed around the SEC regulation (see
Figure 2). Furthermore, as Figure 10 shows, there is little difference between agency
and treasury MMFs; the best-fit line coefficients are 0.17 and 0.37 respectively. If

22
anything, during the 2008 MMF Run, the relationship between prime&muni outflows
and government inflows is stronger for treasury MMFs than agency MMFs; this is
consistent with the fact that prime investors were flying to safety rather than seek-
ing to preserve the money-likeness of their investment. Similar results are obtained
when looking at the Silent Run of 2011 (see Figures 11 and 12), for which the slopes
of the best-fit lines are 0.35 for treasury MMFs and 0.25 for agency MMFs.

Figure 9. Within-family flows: government vs. prime&muni. Time period: August

2008-October 2008. x-axis: total outflow from the familys prime&muni MMFs;
y-axis: total inflows to the familys government MMFs. The dashed line is the
regression line of government inflows on prime&muni outflows (slope: 0.53, p = 0.09,
R2 = 0.21).

23
Figure 10. Within-family flows: agency vs. prime&muni, and treasury vs.
prime&muni. Time period: August 2008-October 2008. x-axis: total outflow from
the familys prime&muni MMFs; y-axis: total inflows to the familys agency and
treasury MMFs (separately). The dashed lines are the regression lines of inflows on
outflows allowing a different slope for inflows in agency and treasury MMFs (blue
line: slope: 0.37, p = 0.05, R2 = 0.29; red line: slope: 0.17, p = 0.21, R2 = 0.10).

Figure 11. Within-family flows: government vs. prime&muni. Time period: May
2011-December 2011. x-axis: total outflow from the familys prime&muni MMFs;
y-axis: total inflows to the familys government MMFs. The dashed line is the
regression line of government inflows on prime&muni outflows (slope: 0.61, p = 0.00,
R2 = 0.51).

24
Figure 12. Within-family flows: agency vs. prime&muni, and treasury vs.
prime&muni. Time period: May 2011-December 2011. x-axis: total outflow from
the familys prime&muni MMFs; y-axis: total inflows to the familys agency and
treasury MMFs (separately). The dashed lines are the regression lines of inflows on
outflows allowing a different slope for inflows in agency and treasury MMFs (blue
line: slope: 0.36, p = 0.00, R2 = 0.57; red line: slope: 0.25, p = 0.01, R2 = 0.30).

4 The Impact on MMF Portfolios: Lending to

Public vs. Private Borrowers

Has the SEC regulation changed the volume of MMF lending by borrower type? Bor-
rowers from US MMFs can be divided in four main categories: the US government,
government agencies, the NY Fed via the RRP program, and private borrowers.10
It it is interesting to study whether the new regulation reduced the credit flowing
to the private sector from the MMF industry. Note that this does not immediately
follow from the fact that after the regulation was implemented, the size of the prime
segment shrank: indeed, agency and treasury MMFs can supply credit to the private
sector through repos collateralized by treasuries and agency debt. This financing
however is limited by the avaliablity of eligible collateral and by the fact that not all
private institutions have the capability of entering into repo agreements (e.g., non
10
For the purpose of this section, we include lending to local authorities by muni MMFs as private
lending. Since lending to local authorities is a very small fraction of overall MMF lending, our choice
does not materially impacts the results of this section.

25
financial corporations).
Figure 13 shows the 3-month backward moving average of MMF portfolio compo-
sition by borrower type from January 2015 to February 2017.11 Tables 11 and 12
shows the shares and dollar volumes in January and November 2016. In the whole
industry, lending to private borrowers decreases from roughly 64% of overall MMF
lending in January 2015 to roughly 40% in February 2017; in dollar value, this de-
cline corresponds to a reduction of roughly $700 bn, from approximately $1,900 bn
to approximately $1,200 bn. In contrast, treasuries, agency debt, and repos with the
NY Fed increase by roughly $300 bn, $250 bn, and $20bn, respectively; as a fraction
of the whole industrys portfolio, treasuries increase from 14% to 27%, agency debt
from 18% to 24%, and repos with the NY Fed from 3.4% to 5.9%.
These changes in the composition of the industry portfolio reflect investors migra-
tion from prime&muni into government funds. Government MMFs are much more
restricted in their investment opportunities than prime MMFs: they can only buy
private debt in the form of repurchase agreements. This restriction limits their abil-
ity to absorb the credit previously provided by prime MMFs to the private sector via
CDs, CPs, and ABCPs, as well as via repos. When we look at the portfolio compo-
sition of each MMF category separately, we observe patterns that differ from that
of the industrys portfolio. The portfolio composition of both treasury and prime
MMFs has remained fairly stable from January to November 2016: both categories
have slightly reduced the fraction of private debt in their portfolios and increased
the fraction of treasuries and repos with the NY Fed. In contrast, the portfolio com-
position of agency funds has changed significantly. Even though agency MMFs have
increased their dollar holdings of both treasuries and agency debt, they have done
so in a very unequal way: the percentage of their portfolio invested in treasuries has
increased by more than 12 pp, from 5% to almost 18%, while that invested in agency
debt has decreased by 14 pp, from 56% to 42%. On the other hand, the fraction of
their portfolios invested in private debt has remained fairly stable at around 30%.
These changes in the portfolio composition of agency MMFs are driven by the com-
bination of two factors: the increase in the size of this segment of the MMF industry
and the limited supply of agency debt relative to treasuries. In fact, the decrease in
11
We take a 3-month moving average because European banks, when trying to meet their regu-
latory requirements, heavily cut their borrowing from US MMFs at quarter-ends (Munyan, 2015);
this generates a strong exogenous cyclicality in the borrower-composition of MMF portfolios. Since
in this paper we are not interested in capturing this regulation-driven behavior at quarter-ends, we
smooth the data over 3-month rolling windows.

26
 Fed Repos Treasuries Non-FHLB Agy FHLB Agy Private
All MMFs
Jan. 2015 5.7% 14.7% 6.5% 9.1% 64.1%
Feb. 2017 6.6% 26.6% 6.5% 18.1% 42.3%
+0.9 pp +11.9 pp +0.0 pp +9.0 pp -21.8 pp
Prime&Muni
Jan. 2015 2.8% 4.8% 2.8% 4.8% 84.7%
Feb. 2017 5.1% 5.8% 0.9% 1.8% 86.5%
+2.3 pp +1.0 pp -1.9 pp -3.0 pp +1.8 pp
Agency
Jan. 2015 10.5% 5.1% 25.7% 32.6% 26.2%
Feb. 2017 7.0% 18.1% 11.4% 32.2% 31.4%
-3.5 pp +13.0 pp -14.3 pp -0.4 pp +5.2 pp
Treasury
Jan. 2015 12.6% 68.7% 0.5% 0.1% 18.1%
Feb. 2017 7.4% 73.3% 0.0% 0.0% 19.3%
-5.2 pp +4.6 pp -0.5 pp -0.1 pp +1.2 pp

Table 11. MMF Portoflio by Borrower Type and Fund Category: Percentage Com-
position.

the relative weight of agency debt and the increase in that of treasuries starts around
August 2016, when the issuance of treasuries increases relatively more than that of
agency debt.
It is interesting to remark that in an effort to maintain their yield while facing inflows
from the prime&muni sector (and given the limited supply of non-FHLB agency
debt), agency MMFs have increased their lending to FHLB. Indeed, as Figure 14
shows, their lending to FHLB as a share of overall agency lending has increased
steadily since the beginning of 2015, from roughly 60 percent to over 75 percent.
From a policy perspective, such a sizable increase in MMF lending to FHLBs is
particularly important: FHLBs can lend to U.S. commercial banks and thrifts and
hence work as a bypass for MMF credit to some U.S. private borrowers that cannot
meet their funding by borrowing directly from the (now much smaller) prime&muni
MMF segment.

27
Figure 13. MMF portfolio composition in percentages by borrower type and fund
category. Top left: all MMFs; top right: prime&muni MMFs; bottom left: agency
MMFs; bottom right: treasury MMFs.

28
 Fed Repos Treasuries Non-FHLB Agy FHLB Agy Private
All MMFs
Jan. 2015 $173 bn $447 bn $198 bn $278 bn $1957 bn
Feb. 2017 $194 bn $781 bn $190 bn $531 bn $1244 bn
+$21 bn +$334 bn -$8 bn +$253 bn -$713 bn
Prime&Muni
Jan. 2015 $57 bn $98 bn $58 bn $98 bn $1,730 bn
Feb. 2017 $36 bn $42 bn $6 bn $13 bn $622 bn
-$21 bn -$56 bn -$52 bn -$85 bn -$1,108 bn
Agency
Jan. 2015 $57 bn $28 bn $140 bn $177 bn $142 bn
Feb. 2017 $112 bn $291 bn $183 bn $518 bn $505 bn
+$55 bn +$263 bn +$43 bn +$341 bn +$363 bn
Treasury
Jan. 2015 $59 bn $321 bn $1 bn $2 bn $85 bn
Feb. 2017 $46 bn $449 bn $0 bn $0 bn $118 bn
-$13 bn +$128 bn -$1 bn -$2 bn +$33 bn

Table 12. MMF Portoflio by Borrower Type and Fund Category: Dollar Values.

Figure 14. Non-FHLB and FHLB Agency Debt (USD billions) held by the MMF
industry. The solid line (FHLB Share) shows the share of FHLB debt as a fraction
of all Agency debt.

29
References

Chernenko, S. and A. Sunderam, 2014. Frictions in shadow banking: Evidence from

the lending behavior of money market mutual fund. Review of Financial Studies
27.6, 1717-1750.
Hanson, S. G., D. S. Scharfstein, and A. Sunderam, 2015. An evaluation of money
market fund reform proposals. IMF Economic Review 63.4, 984-1023.
Investment Company Institute, 2010, The Case for Stable NAV Money Market
Funds, Investment Company Institute, Web, last accessed 14 Apr. 2017, https://w
ww.ici.org/mmfs/positions/stable nav/10 mmfs caseforstablenav.
Investment Company Institute, Preserving Money Market Funds For Investors, For
America, Web, last accessed 4 Sept. 2014, http://www.preservemoneymarketfunds
.org.
Kacperczyk, M. and P. Schnabl, 2013. How Safe Are Money Market Funds? The
Quarterly Journal of Economics 128, 10731122.
La Spada, G., 2017. Competition, Reach for Yield, and Money Market Funds.
Forthcoming in the Journal of Financial Economics.
McCabe, P., M. Cipriani, M. Holscher, A. Martin, 2013. The Minimum Balance
at Risk: a Proposal to Mitigate Systemic Risks Posed by Money Market Funds.
Brookings Papers on Economic Activity, Spring 2013.
Munyan, B., 2015. Regulatory Arbitrage in Repo Markets. OFR Working Paper.
Plantier, C. and J. Steenstrak, 2015, Investment Company Factbook, 55th ed., In-
vestment Company Institute, https://www.ici.org/pdf/2015 factbook.pdf.
SEC, 2010, Money Market Fund Reform, Release No. IC-29132, https://www.sec.gov
/rules/final/2010/ic-29132.pdf.
SEC, 2011, Unofficial Transcript: Roundtable on Money Market Funds and Systemic
Risk, https://www.sec.gov/spotlight/mmf-risk/mmf-risk-transcript-051011.htm.
SEC, 2014, Money Market Fund Reform; Amendments to Form PF, Release No.
IC-31166, https://www.sec.gov/rules/final/2014/33-9616.pdf.

30