The Role of Money Market Mutual Funds in Mutual ... - SSRN papers

9 downloads 0 Views 628KB Size Report
This study examines cross-sectional differences among money market mutual funds (MMMFs) in the context of sponsoring fund families. The study finds that ...
Agapova – The Role of Money Market Mutual Funds in Mutual Fund Families

1

The Role of Money Market Mutual Funds in Mutual Fund Families Anna Agapova

This study examines cross-sectional differences among money market mutual funds (MMMFs) in the context of sponsoring fund families. The study finds that flows to family non-MMMFs are negatively related to family MMMF flows, and family non-MMMF cash flow volatility is positively related to family MMMF cash flow volatility. This suggests that families can offset cash transactions between MMMFs and non-MMMFs and that families place non-MMMFs cash in family MMMFs. Furthermore, loads positively affect cash flow volatility of MMMFs, suggesting that fund family investors also use family MMMFs as cash centers by utilizing free asset transfers within the family. Application of these strategies can translate into significant benefits for the fund family and its investors.

nMoney market mutual funds (MMMFs) have existed for more than three decades, with the first one introduced in 1972. By 1984, 305 MMMFs existed, totaling nearly $270 billion in assets. Over the twenty-year period through 2006, assets of MMMFs increased to $2.4 trillion across 848 funds, representing approximately 25% of the US open-end mutual fund assets.1 Investment Company Institute (ICI) report, December 2006

1

This paper derives from the first essay of my dissertation at Georgia State University. I thank an anonymous referee, Betty Simkins (Editor), Jason Greene, Vikas Agarwal, Conrad Ciccotello, Gerald Gay, and seminar participants at the 2006 Financial Management Association and 2009 Eastern Finance Association meetings for valuable comments and suggestions. All remaining errors are mine. Anna Agapova is an Assistant Professor of Finance at Florida Atlantic University in Boca Raton, FL.

Though MMMFs are the second largest segment of mutual fund industry assets after equity funds, the majority of the existing literature concentrates on equity funds. Many studies of open-end equity funds examine how these funds differ in the cross-section, focusing on the skills of fund managers and how funds are affected by characteristics of sponsoring fund families.2 In contrast, only a few studies have addressed cross-sectional differences among MMMFs, and none have addressed related MMMF fund family characteristics. This study contributes to the MMMF literature and the fund family strategy research by examining if MMMFs play a role of the internal cash center of fund families, whereby other family funds perform their liquidity transactions using in-house MMMFs. Such cash management strategies can save on transaction costs at the family level by avoiding transactions with external entities or by enjoying externalities associated with the MMMF managers’ expertise. Investors can take similar advantage of MMMFs by parking their money inside the fund family at lower costs than investing outside of the complex, due to free asset transfers within the family. Early research on MMMFs focuses on the ability of portfolio managers to predict interest rate fluctuations by examining the association between portfolios’ average maturities and interest rate changes (see Ferri and Oberhelman, 1981; Packer and Pencek, 1990; Domian, 1992; Seyfried and Packer, 2001). DeGennaro and Domian (1996) examine time-series differences in MMMFs’ average maturities and conclude that managers select their target levels for interest-rate risk. See, for example, Carhart (1997), Chevalier and Ellison (1997), Massa (2003), and Gaspar, Massa, and Matos (2006), among others. 2

1

2

Journal of Applied Finance – No. 1, 2011

Few cross-sectional studies analyze MMMF performance as maturity, yield, and expenses across MMMFs can be and flow characteristics. Domian and Reichenstein (1997) partially explained by family-specific characteristics, examine the factors that affect the cross-section of net returns including cash management strategies at the family level. and the persistence of relative returns through time. They find Application of these strategies can reduce operating costs that expense ratio is the most important factor in explaining and improve overall performance for families and investors. the difference between net returns and that MMMFs’ relative The remainder of this paper is organized as follows: returns show strong persistence. Christoffersen (2001) uses Section I develops the hypotheses; Section II describes the the MMMF setting to examine fee waving strategies used by data; Section III covers empirical analysis; and Section IV funds. To study the performance-flow relation of MMMFs concludes. and how it affects decisions to waive fees, Christoffersen (2001) employs Sirri and Tufano’s (1998) methodology to estimate a piecewise-linear fund flow function. She finds I. The Role of MMMFs in Fund Families that better performing funds attract more flows and that variation in fee waivers is significant and relates to the MMMFs offer investors a relatively homogeneous relative performance of MMMFs. product – short-term debt securities with relatively low Existing literature on fund family characteristics and risk. Regulations limit the choice of securities in money strategies is scarce. market portfolios. Massa (1998) This study finds evidence that fund families and Rule 2a-7 of the develops a model Investment Company of the mutual fund their investors can use family MMMFs as cash Act of 1940 limits industry structure centers. Families can use their MMMFs to offset MMMF instruments to that explains the role cash transactions of family non-MMMFs and to those with remaining of fund families. maturity of less than He argues that fund place cash of other family funds into in-house 397 days or with family strategies of MMMFs. Family investors can use MMMFs to average portfolio segmentation and maturity of less than product proliferation temporarily park assets by utilizing free asset 90 days. This rule can be used to transfers within the family. also specifies levels of exploit investors’ portfolio quality and heterogeneity. Mamaysky and Spiegel (2001) develop a diversification that MMMFs must maintain. model of mutual fund families that explains the existence Even though these restrictions on investment choices exist, of fund families and their strategies to satisfy investors’ the funds exhibit considerable cross-sectional variations in trading and hedging needs. Nanda, Wang, and Zheng (2004) characteristics such as maturity, return, and expenses.3 These show that a spillover effect of star performance exists, and differences arise from the varying investment objectives that families use the strategy of generating star performers MMMFs pursue and from the roles the funds play within to attract more assets. Gaspar et al. (2006) and Guedj and fund families.4 Papastaikoudi (2003) argue that families provide more support to the better performing and/or higher fee funds to maximize family level proceeds and the family objective A. Family Strategies with MMMFs function. This strategy of favoritism can be in the form of cross-fund subsidization by shifting performance (Gaspar Fund family literature shows that families engage in et al., 2006) or through limited resource allocation across different strategies to attract more of its investors’ assets. funds (Guedj and Papastaikoudi, 2003). Massa (2003) For example, the creation of star funds draws more money demonstrates that industry structure matters, showing that to other funds in the family (Nanda et al, 2004). Strategies of fund families compete through performance and product favoritism and scarce resource allocation have the same goal differentiation. of bringing in more money through family favored funds This study finds evidence that fund families and their (Gaspar et al., 2006, Guedj and Papastaikoudi, 2003). Massa investors can use family MMMFs as cash centers. Families can use their MMMFs to offset cash transactions of family 3Differences of MMMF characteristics are discussed in the following non-MMMFs and to place cash of other family funds into section of the paper. in-house MMMFs. Family investors can use MMMFs to temporarily park assets by utilizing free asset transfers 4CRSP reports six ICDI fund objective codes as well as 37 detailed Standard within the family. Differences in fund characteristics such & Poor’s objective codes for MMMFs.

3

Agapova – The Role of Money Market Mutual Funds in Mutual Fund Families

Figure 1. Flow of Cash in Mutual Fund Families A. Closed System – No Outside Transactions

B. Open System – Some Outside Transactions

C. Open System – All Outside Transactions

$50

MMMF

$100

Equity Fund A

$100 MM Securities

Equity Fund B

$100 MM Securities

$100

Bond Fund

MMMF

MMMF

Equity Fund B

Equity Fund B

$100

(2003) shows that fund families actively exploit product differentiation strategies to attract heterogeneous investors. Additionally, fund families can also pursue cost minimizing strategies. We expect that fund families use MMMFs as an element of overall family profit maximization, which can lead to variance in MMMF structures. More specifically, we expect fund families to benefit from using their MMMFs as cash management tools. Fund families can pursue multiple cost minimizing strategies with MMMFs. One strategy is to offset cash transactions within the fund family. Non-MMMFs hold some portion of their portfolios in cash (i.e., money market securities), while MMMFs hold their entire portfolios in various kinds of money market securities. If a non-MMMF, such as an equity fund, experiences cash inflows or if a MMMF experiences redemptions, the liquidity requirements of both funds could be met through financial markets transactions, with realizable costs. However, if the equity fund cash inflows and MMMF cash outflows correlate in time and amount to some extent, these funds can exchange securities within the family, thereby saving the transaction costs that they would have otherwise incurred in financial markets. In this scenario, the MMMF acts as a clearing house for cash transactions within the family. However, if no correlation exists between cash flows of the equity fund and the MMMF, this strategy is not feasible. As Figure 1 shows, this strategy can have various applications depending on the mutual fund family’s cash flow structure. If we consider a closed system with no transactions outside the fund family as shown in Figure 1.A, the fund family saves on transaction costs. In this scenario, investors make no adjustment to their assets invested with the family even when they change their positions within individual funds. As an investor withdraws $100 from MMMF and buys $100 of shares in equity fund B, the

$100

managers of the two funds can just exchange cash and money market securities to save on transaction costs. If the system is open, i.e., cash flows happen outside of the family, then two scenarios are possible: some outside transactions or all outside transactions. As shown in Figure 1.B, if the manager of equity fund B receives $100 of new cash from an investor while the manager of MMMF faces a simultaneous $150 withdrawal, a partial offsetting transaction within the fund family can be achieved. The fund managers can exchange $100 of cash and money market securities between the funds, with the MMMF manager selling the remaining $50 of money market securities in the open market to deliver the necessary cash for a withdrawal. This strategy saves the fund family on transaction costs on the $100 in trades that both funds would incur without the strategy. If, as shown in Figure 1.C, equity fund B receives $100 of new cash from an investor, but no cash leaves the MMMF, the equity fund manager exchanges cash for money market instruments in the open market, resulting in no cost savings to the fund family. In the second cost minimizing strategy, the equity fund can invest directly into the family MMMF whenever the cost of doing so is lower than the corresponding transaction cost in the open market.5 Two scenarios exist for this strategy. The first scenario occurs when separate MMMFs are offered by the family to satisfy the needs of investors and of other funds in the family. For example, in July 2004, Vanguard Funds created a special MMMF for cash management purposes at the family level and made the fund “available only to Vanguard funds and certain trusts and accounts managed An additional benefit of putting cash of non-MMMF into MMMF instead of buying money market securities directly in financial markets is the externality of MMMF managers’ expertise that non-MMMF managers may not have. 5

4

Journal of Applied Finance – No. 1, 2011

by Vanguard.”6 The second scenario occurs when the same family are offset using a MMMF or when separate MMMFs MMMF is used both by investors and by other funds in the are used by other family funds and investors. family.7 The implication of the cash management strategy of a I test the main hypotheses, whether MMMFs are run fund family is realized in lower transaction costs when the in conjunction with other funds in the same fund family, first strategy of offsetting cash transactions within the family by examining several predictions based on family cash is feasible and in both lower transaction costs and money management strategies with MMMFs. The offsetting market management expertise externalities when the second strategy is applicable if cash flows of MMMFs and other strategy is used to satisfy non-MMMF liquidity need using funds in the family are negatively correlated. The effect of family MMMFs. this strategy should be found in lower costs for the families that use MMMFs for the fund family liquidity needs. B. Family Investors and MMMFs The families that have funds with higher flow volatility and higher portfolio turnover face elevated need for cash management. If a fund family uses MMMFs as cash centers For investors, the role of MMMFs as cash centers is not as described in the second strategy above, i.e., if family explicitly defined by fund families. However, a family can funds place their cash directly in the family MMMFs, then make its MMMF more attractive to its investors for liquidity cash flow volatility transactions. Massa of other funds in The implication of the cash management strategy (2003) shows that the family will of a fund family is realized in lower transaction mutual fund families positively affect cash strategies that costs when the first strategy of offsetting cash employ flow volatility of rely on heterogeneity the MMMFs. When transactions within the family is feasible and in of investors by MMMFs actively both lower transaction costs and money market offering the ability manage increased to switch, at no cost, volatility of their management expertise externalities when the across different funds flows, they will have second strategy is used to satisfy non-MMMF of the same family. shorter maturities At the individual liquidity need using family MMMFs. and lower risk. Thus, fund level, Chordia if MMMFs are used as part of the family’s overall strategy (1996) shows theoretically and empirically that redemption aimed at cost-efficiency of the family structure, we should rates are higher in no-load funds and that fund cash holdings observe the effects of these cash management strategies decrease with load fees. Similarly, a higher demand for a through MMMFs’ cash flow volatility and maturity and the cash center can exist at the family level for a high-load fund performance of the family’s other funds. family since transfers are more likely to occur within the If a family offers a MMMF to investors and also uses it for family. This will be reflected in the use of MMMFs. Massa cash management, as in the second scenario of the second (2003) argues that investors who plan to reallocate their strategy, the magnitude of the relation between flow volatility assets more frequently tend to invest in funds with lower of MMMFs and the rest of the family should be higher, load fees that belong to larger families. We predict that since there is less room for cash flow shock absorption in families with higher average loads experience greater use of the family. This effect can be tested by comparing families MMMFs by investors who realize the option of free move. that have only one MMMF with those that have multiple A testable prediction is that families with higher load funds MMMFs. have higher volatility of MMMF cash flows. If investors The strategic use of MMMFs by fund families can create move their assets among funds using in-house MMMFs a potential for agency problems and a divergence of investor instead of outside accounts, volatility of family MMMF and family interests. This is the case when the same MMMF cash flows will be higher than for those families that do not is offered to investors and also used for cash management by restrain investors by charging loads. the family, since the family may optimize at the expense of the The role of MMMFs as cash centers for investors may not MMMF investors. However, potential for agency problems directly benefit the fund family and can burden the MMMF is less pronounced when matching cash transactions within a managers. However, this strategy ultimately can attract more assets from investors. Vanguard Market Liquidity Fund, Semiannual Report 2005.

6

Examples of Janus and Vanguard mutual fund families that illustrate use of cash management in fund families are given in Appendix A. 7

5

Agapova – The Role of Money Market Mutual Funds in Mutual Fund Families

Table I. Descriptive Statistics and Distribution of MMMFs Across Families Panel A. TNA is total net assets in millions of dollars; Maturity is the weighted average maturity in days, Expenses is the expense ratio in percentage; Return is the MMMF yearly gross return in percentage; Risk is the standard deviation of monthly returns in percentage; and N is the number of funds. Year

N

1992

704

1993

808

1994

714

1995

848

1996

1,008

1997

1,185

1998

1,217

1999

1,205

2000

1,225

2001

1,183

2002

1,173

2003

1,165

2004

1,443

2005

1,455

2006

1,413

1992-2006

16,746

 

 

Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev

TNA, $ mln

Maturity, days

Expenses, %

Return, %

Risk, %

606.05 1584.10 590.86 1487.16 536.69 1575.68 630.19 1813.53 662.38 1970.13 804.07 2314.35 998.05 2951.84 1163.54 3491.99 1247.83 3673.63 1550.93 4382.43 1532.93 4273.30 1424.90 3874.94 1194.72 3312.72 1254.32 3524.44 1360.98 4017.05 1095.93 3281.05

52.61 17.38 38.93 21.40 48.16 18.12 45.99 18.16 45.83 16.21 46.58 16.25 46.17 15.10 48.84 14.88 43.89 15.18 43.75 15.19 43.84 15.00 45.61 16.05 42.62 15.96 31.82 16.09 31.03 14.02 43.00 17.16

0.64 0.27 0.60 0.26 0.60 0.28 0.60 0.28 0.60 0.27 0.59 0.27 0.60 0.30 0.61 0.30 0.60 0.30 0.62 0.32 0.62 0.33 0.59 0.29 0.58 0.27 0.58 0.30 0.59 0.33 0.60 0.29

2.99 0.53 2.60 6.64 3.16 0.73 4.59 1.05 4.21 1.18 4.26 1.05 4.23 1.00 3.90 0.93 4.98 1.09 3.16 0.72 1.23 2.61 0.56 0.25 0.65 0.26 2.30 0.59 1.81 0.37 2.88 2.28

0.04 0.01 0.01 0.01 0.06 0.02 0.02 0.02 0.02 0.01 0.02 0.03 0.02 0.04 0.03 0.01 0.04 0.02 0.09 0.02 0.01 0.01 0.01 0.01 0.03 0.01 0.05 0.02 0.04 0.02 0.03 0.03 (Continued)

II. Data The data source for this study is the Center for Research in Security Prices (CRSP) survivor-bias-free US mutual fund database. The study period is from 1992 through 2006.8 Net asset value (NAV) equal to one is used to identify Though, the first MMMF was organized in 1972, the study period is from 1992 through 2006, because CRSP data have many missing observations prior to 1992, and also because significant development of the MMMFs market is referred to the late 1980s. 8

MMMFs.9 Fund observations with total net assets (TNA) less than $10 million are excluded, leaving 16,746 fund-year observations of MMMFs. Monthly data are used for flow analysis in Model (1) and yearly data are used for the rest of the Models (2) – (4). Table I Panel A presents descriptive Additional criteria are ICDI’s fund objective code and full investment in cash of portfolio holdings. 9

6

Journal of Applied Finance – No. 1, 2011

Table I. Descriptive Statistics and Distribution of MMMFs Across Families (Continued) Panel B. All Families refers to the total number of families in the industry, and Families w/MMMF indicates the number of families with MMMFs. The rest of the variables describe the number of MMMFs in a family. Year

All Families

Families w/MMMF

Mean

Std Dev

25th Pctl.

Median

75th Pctl.

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

365 374 385 392 401 430 443 464 521 509 479 504 509 499 494

177 182 163 180 183 192 194 190 209 198 175 131 181 173 164

4.0 4.4 4.4 4.7 5.5 6.2 6.3 6.3 5.9 6.0 6.7 8.9 8.0 8.4 8.6

5.8 8.9 9.2 9.2 11.0 12.0 11.6 11.6 10.3 10.3 11.1 14.4 13.2 14.4 14.4

1 1 1 1 1 1 1 1 1 1 1 2 1 2 1

2 2 2 2 2 3 3 3 3 3 3 4 3 3 4

5 5 5 5 6 6.5 7 6 6 6 7 9 8 9 9

statistics of the fund-year data for each year and over the entire period of the study. The data show there is substantial cross-sectional variation among MMMFs. For example, the standard deviation of the weighted average maturity is 17.2 days with a mean of 43 days for the whole sample period. Similarly, for the period 1992 through 2006, expenses have a standard deviation of 0.29%, with a mean of 0.60%, and the gross returns have a standard deviation of 2.28% with a mean of 2.88%. Over the period, weighted average maturity and expenses declined, though cross-sectional variation sustained. Table I Panel B presents statistics based on the number of fund families and the number of MMMFs within those families. Although the number of families with MMMFs is less than half of the number of all families, the average TNA of the families with MMMFs ($19,414.7 mil.) is significantly larger than the average TNA of the families without MMMFs ($1,168.5 mil.). Furthermore, the families with MMMFs represent more than 90% of the mutual fund industry asset share in December 2006. The average number of MMMFs within a family has increased from four to eight funds and the median number has increased from two to four funds during the study time period. The number of families that offer MMMFs varies over the years, with the peak occurring in 2000.10 Tests are conducted at the family level. First, all funds in the families with MMMFs are selected. The MMMF-only The total number of families also peaked in 2000.

10

families are dropped from the sample. Data are aggregated on the family level by finding value-weighted averages of all MMMFs and all other funds within the family accordingly. All family level non-MMMF measures used in the tests and descriptive statistics exclude family MMMFs. Mutual fund studies conducted at the fund level address multiple share classes by using only the share class with the highest TNA or by using the TNA-weighted average of all fund share classes (e.g., Gaspar et al., 2006). This study uses TNA-weighted averages of all family MMMFs and of all other funds in the family. Table II presents descriptive statistics of families with and without MMMFs. On average, families that do not have MMMFs are smaller, have higher risks, charge higher expenses and front and back loads, and are more concentrated in a single investment objective. Further, families without MMMFs on average have higher portfolio turnover, keep a higher portion of fund portfolio in cash, and have higher family cash flow volatility adjusted for family TNA. On average, monthly value weighted average returns of families without MMMFs are slightly lower than those of families with MMMFs.

III. Empirical Analysis A. Cash Management Analysis We start the analysis by testing relations between flows to MMMFs and flows to other funds in the family to see

7

Agapova – The Role of Money Market Mutual Funds in Mutual Fund Families

Table II. Descriptive Statistics of Families with and without MMMFs (1992-2006) N reports the number of year-family observations except for the Return rows where N reports the number of month-family observations and the rest of the columns report statistics of month-family observations. Return is the monthly value-weighted average net return of a family portfolio. riskFam is the standard deviation of the monthly weighted average net returns of the family. FamExpenses is the value weighted expense ratio for the family. TNAFam is the total net assets of the family measured in millions of dollars. FamFlowVol is the standard deviation of monthly net cash flows to the family measured in millions of dollars. AveFRNT_LD and AveREAR_LD are the value-weighted average front and back loads of the family respectively. CashFamPerc is the ratio of cash holdings of all funds, besides MMMFs, in the family relative to the TNA of the family. FamTurn is the family portfolio turnover. TNAMMMF/TNAFam is the liquidity ratio of the family (percentage). N

Mean

Std Dev

Median

75th Pctl.

Families without MMMFs Return (month) riskFam FamExpenses TNAFam FamFlowVol AveFRNT_LD AveREAR_LD CashFamPerc FamTurn

87,799 3,135 3,176 3,214 3,133 3,214 3,195 3,214 3,110

0.006 0.038 0.014 1,168.5 23.9 0.007 -0.036 0.071 0.933

0.045 0.026 0.005 4,633.2 76.5 0.016 1.280 0.121 1.719

-0.012 0.021 0.010 41.6 1.0 0 0 0.011 0.265

0.006 0.032 0.013 153.6 4.2 0 0 0.037 0.540

0.025 0.047 0.017 552.1 17.9 0 0.0013 0.081 1.033

Families With MMMFs Return (month) riskFam FamExpenses TNAFam FamFlowVol AveFRNT_LD AveREAR_LD CashFamPerc FamTurn TNAMMMF/TNAFam

45,183 2,474 2,466 2,477 2,471 2,470 2,467 2,477 2,304 2,477

0.007 0.028 0.001 19,414.7 145.6 0.002 0.000 0.054 0.103 31.64

0.036 0.018 0.003 67,760.7 401.9 0.007 0.002 0.065 0.613 29.11

-0.010 0.016 0.000 667.6 9.5 0 0 0.025 0.005 5.30

0.008 0.024 0.000 2,728.1 38.6 0 0 0.041 0.023 21.37

0.024 0.036 0.001 11,731.3 131.7 0.0009 0 0.067 0.087 55.82

whether the first strategy (offsetting cash transactions) can be used by families. Fixed effect regression analysis is employed on panel data using the following model with control for family and month fixed effects:11 MMMFflowi ,t = α + β1 Famflowi ,t + β 2 Famturni ,t + β3CashFami ,t +

β 4 log TNAmmmfi ,t + β5 log TNAfami ,t + β 6 retMMMFi ,t + β 7 riskMMMFi ,t + ε i ,t ,

(1)

to other funds in family i in month t, respectively. 12, 13 If cash flows for the MMMFs and the rest of family’s funds are negatively related, then cash flows between the funds can be partially offset inside the family without transaction costs as described in the first strategy. Control variables are: family portfolio turnover (Famturn), which is expected to have a positive effect on MMMF flows as a family’s need for cash transactions will increase; and cash holdings of non-MMMF 12

where MMMFflowi,t and Famflowi,t are flows to MMMFs and

Hausman test identifies fixed effect model as more appropriate than random effects model for the tests in this study.

Cash flows to MMMFs and to the rest of family are calculated using Sirri

and Tufano’s (1998) methodology: Flowi ,t = TNAi ,t − TNAi ,t −1 * (1 + Ri ,t ) , where TNA is the monthly total net assets at time t, and R is the weighted average return for the prior month. Christoffersen (2001) measures flows to MMMFs as a percentage change in assets, and she indicates that defining fund flows using Sirri and Tufano’s (1998) methodology does not change the results of her study. Although Christoffersen’s (2001) methodology may be justifiable for MMMFs, Sirri and Tufano’s (1998) methodology gives more consistent measure of fund flows for all types of fund investment objectives. 13

11

25th Pctl.

8

as percentage of TNA (CashFam), which is expected to have a positive effect on MMMF flows as an increase in family portfolios’ cash holdings will increase demand for MMMF services as a cash center. Similarly to Sirri and Tufano (1998), we include log of total net assets of MMMF (logTNAmmmf) and of the family (logTNAfam), family MMMFs’ monthly return (retMMMF), and risk (riskMMMF), measured as standard deviation of monthly returns in a calendar year, as control variables.14 To better understand effects of family structure on cash management strategies, tests are performed on the whole sample and also separately on sub-samples of families with more than one MMMF and of those with only one MMMF. Table III reports the analysis of the effect of family nonMMMF cash flows on MMMF flows. The results show that family non-MMMF flows are negatively related to MMMF flows for the whole sample, and also for both multi- and one-MMMF families at the less than 1% level. On average, as family non-MMMF flows increase by $1 MMMF flows decrease by 3.9 cents.15 These results indicate that families can use offsetting cash management strategies within families, as described in the first strategy. As expected, the level of cash in non-MMMFs positively affects cash flows to MMMFs of a family, with a significance level of less that 1%. A one percent increase in portfolio cash holdings of non-MMMFs increases MMMF flows by approximately $1 million. This result indicates that fund families actively use MMMFs as described in the second strategy. The results also show that MMMF size positively affects MMMF cash flows, and risk and return of MMMFs positively affect MMMF cash flows. Results are significant at less than the 1% level. These results on control variables hold for the whole sample and sub-sample of families with more than one MMMF. For families with only one MMMF, only MMMF return variable is significant at less than the 10% level.16 To test whether fund families use their MMMFs to manage cash of other funds in the family, as in the second strategy, we examine the relation between flow volatility of MMMFs and the flow volatility of the rest of the family. We further examine an effect of this strategy on MMMF maturity. To control for possible endogeniety of explanatory variables, Appendix Table summarizes definitions of the variables, the rationale for their inclusion, and their expected sign. 14

Also, as reported in Table II, CFCori,t , which measure correlation of flows between family MMMFs and non-MMMFs, has a mean of -0.076 with standard deviation of 0.378 across families. 15

While the regression R2 are relatively low, the regression models are significant in an F-test (Prob>F =0.000), and the variables of interest are statistically significant. 16

Journal of Applied Finance – No. 1, 2011

we employ two-stage least squares regression analysis of panel data on the following models with control for family and year fixed effects: MMMF cash flow volatility model: first stage: CFVolFami ,t = α + β1 Famreti ,t + β 2 Famriski ,t + β3 Famflowi ,t + β 4CFCori ,t + β5 AveFamCFCori ,t + β 6 Famturni ,t + β 7CashFami ,t + β8 log TNAmmmfi ,t + β9 log TNAfami ,t + β10 retMMMFi ,t + β11riskMMMFi ,t + ε i ,t ,

(2a)

second stage: CFVolMMMFi ,t = α + β1CFVolFami ,t + β 2CFCori ,t + β3 AveFamCFCori ,t +

β 4 Famturni ,t + β5CashFami ,t + β 6 log TNAmmfi ,t + β 7 log TNAfami ,t + β8 retMMMFi ,t + β9 riskMMMFi ,t + ε i ,t ,

(2b)

Maturity model: first stage: CFVolMMMFi ,t = α + β1retMMMFi ,t + β 2 riskMMMFi ,t + β3CFVolFami ,t +

β 4CFCori ,t + β5 AveFamCFCori ,t + β 6 Famturni ,t + β 7CashFami ,t + β8 log TNAmmfi ,t + β9 log TNAfami ,t + ε i ,t ,

second stage:

(3a)



Maturityi ,t = α + β1CFVolMMMFi ,t + β 2 riskMMMFi ,t + β3 AveFamCFCori ,t +

β 4 Famturni ,t + β5CashFami ,t + β 6 log TNAmmfi ,t + β 7 log TNAfami ,t + ε i ,t ,

(3b)

where CFVolMMMFi,t is the volatility of MMMF cash flows in the family i and CFVolFami,t is the volatility of cash flows of the other funds in the family i, calculated as a standard deviation of monthly cash flows in year t, and Maturityi,t is the value-weighted average maturity of MMMFs in the family i in year t. CFVolFam and CFVolMMF can be endogenously determined by right-hand side variables in Equation (2b) and Equation (3b), respectively. In addition, CFVolMMF is the dependent variable in model Equation (2b), but it is the explanatory variable in model Equation (3b). Consequently, CFVolMMF is shown to be defined by some of the right hand variables in the same model. To control for this problem, CFVolFam and CFVolMMF are instrumented on FamFlow, retFam and riskFam, and retMMMF and riskMMMF, respectively, which are believed to be unique instruments. If families use MMMFs as cash centers, as in the second strategy, the relation between cash flow volatility of MMMFs and the rest of the family should be positive. In addition, if maturity is a tool to manage cash and liquidity, it should be negatively related to the MMMFs cash flow volatility. If volatility increase in the rest of the family’s cash flows increases the flow volatility of MMMFs, managers will shorten the maturity of MMMFs’ securities to release more cash for liquidity purposes and to reduce transaction costs.

9

Agapova – The Role of Money Market Mutual Funds in Mutual Fund Families

Table III. Effect of Family Flows on MMMF Flows This table reports results from fixed effect regressions at the family level for the sample of 26,622 family-months from 1992 to 2006. The estimated coefficients are from regressions of the following equation: MMMFflowi ,t = α + β1 Famflowi ,t + β 2 Famturni ,t + β3CashFami ,t + β 4 log TNAmmmf i ,t + β5 log TNAfami ,t + β 6 retMMMFi ,t + β 7 riskMMMFi ,t + ε i ,t ,

where dependent variable is cash flows of MMMFs in the family. The independent variables include family flows (Famflow), family average turnover, ratio of cash holdings in non-MMMFs (CashFam), log of MMMFs and family TNA, and return and risk of MMMFs. T-statistics are reported in parentheses. MMMFflow Fixed Effects

Intercept Famflow Famturn CashFam logTNAmmmf logTNAfam

All

> One MMMF

One MMMF

-302.1***

-496.7***

-15.66*

(-4.54)

(-4.52)

(-1.68)

-0.04***

-0.04***

-0.03***

(-4.09)

(-3.43)

(-3.47)

-4.18

-5.44

-0.30

(-0.45)

(-0.08)

(-0.42)

0.01***

0.01***

(4.99)

(4.18)

20.96**

37.36***

(2.39)

(2.67)

0.002 (0.50) 2.29 (1.48)

4.74

4.59

0.36

(0.51)

(0.32)

(0.25)

40937***

59015***

995.09*

(10.57)

(10.54)

(1.82)

94572***

132023***

(5.22)

(5.05)

(0.81)

Observations

26,622

18,644

7,978

R

0.012

0.015

0.003

retMMMF riskMMMF

2

2038.97

***Significant at the 0.01 level. **Significant at the 0.05 level. *Significant at the 0.10 level.

Definitions for the remaining right hand side variables follow. CFCori,t is the correlation between net cash flows to MMMFs and to the rest of the family in year t. AveFamCFCori,t is the average cash flow correlation between non-MMMFs in the family i in year t.17 The correlation variables control for the possibility of offsetting or matching cash transactions within the family. FamTurni,t is the average portfolio turnover in the family i except MMMFs in year t. CashFami,t is the cash holdings of non-MMMFs in the family i in year t and is used to control for the level of cash AveFamCFCori,t is calculated by finding monthly cash flow correlation between each pair of funds, besides MMMFs, in the family within a year and taking the average of all fund pairs.            17

in the family besides MMMFs. CashFami,t is expected to be negatively related to CFVolMMMFi,t in Equation (2b) and to Maturityi,t in Equation (3b), as more cash availability on nonMMMF level should make it easier for portfolio managers to absorb liquidity shocks without using MMMFs for that purpose. logTNAmmfi,t and logTNAfami,t are logarithms of the total net assets of MMMFs and other funds in family i in year t, respectively. retMMMFi,t and riskMMMFi,t are MMMF value-weighted yearly return and standard deviation of MMMF monthly returns in family i in year t, respectively. Table IV reports the results of our cash management analysis, which show the effects of family non-MMMF cash flow volatility on MMMF flow volatility (Panel A), and

10

Journal of Applied Finance – No. 1, 2011

Table IV. Use of MMMFs by Other Funds in the Family This table reports results from two-stage least squares fixed effects regressions at the family level for the sample of 2,234 family-years from 1992 to 2006. The estimated coefficients are from regressions of the following equation (second step): CFVolMMMFi ,t = α + β1CFVOLFami ,t + β 2CFCori ,t + β3 AveFamCFCori ,t + β 4 Famturni ,t + β5CashFami ,t +

β 6 log TNAmmmfi ,t + β 7 log TNAfami ,t + β8 retMMMFi ,t + β9 riskMMMFi ,t + ε i ,t ,

where dependent variable is the cash flow volatility of MMMFs in the family. The independent variables include volatility of family flows (CFVolFam), correlation of flows between MMMFs and other funds of the family (CFCor), and average cash flow correlation of nonMMMFs in the family (AveFamCFCorr), where CFVolFam is instrumented on family return (Famret), family risk (Famrisk) and family flows (Famflow). Other variables are family average turnover, ratio of cash holdings in non-MMMFs (CashFam), log of MMMFs and family TNA, and MMMF return (retMMMF) and MMMF risk (riskMMMF). T-statistics are reported in parentheses. Panel A. MMMF Cash Flow Volatility CFVolMMMF 2SLS Fixed Effects All Intercept

-338.2 (-1.36)

CFVolFam

0.923** (2.34)

cfcorr AveFamCFcor

logTNAmmmf logTNAfam retMMMF riskMMMF Observations R2

(-1.73)

0.961** (1.98) -26.69 (-0.52)

-0.917 (-0.04)

CashFam

-43.11*

(2.02)

(-0.67) -112.2

-0.018*

One MMMF

-834.4**

-24.44

(-0.97) Famturn

> One MMMF

-260.9 (-1.24) -119.2

0.253 (1.50) -1.99 (-0.47) -4.74 (-0.45) 0.798

(-0.73)

(0.63)

-0.019*

-0.012**

(-1.92)

(-1.70)

(-2.01)

138.5***

248.3***

(5.27)

(6.38)

(5.72)

-45.48*

-83.94**

-1.55

(-1.68)

(-2.00)

(-0.35)

-29,517***

-47,223***

-1,723

(-3.05)

(-3.46)

(-1.48)

97,622**

144,683**

9,379**

(2.32)

(2.41)

(2.07)

2,234

1,570

664

0.23

0.27

0.10

15.87***

(Continued)

MMMF cash flow volatility on MMMF maturity (Panel B). Tests are performed on the whole sample and also separately on sub-samples of families with more than one MMMF and of those with only one MMMF. As Table IV Panel A shows, cash flow volatility of MMMFs and of the rest of the family are positively related at the less than 5% level for the whole sample, as expected,

with a $1 increase in the family flow volatility resulting in a $0.9 increase in MMMF flow volatility. Results support the expectation that families are using MMMFs as cash centers by investing non-MMMF cash directly in the family MMMFs rather than outside of the family. Families with multiple MMMFs have similar results, showing a positive relation between family non-MMMF and MMMF cash flow

11

Agapova – The Role of Money Market Mutual Funds in Mutual Fund Families

Table IV. Use of MMMFs by Other Funds in the Family (Continued) This table reports results from two-stage least squares fixed effects regressions at the family level for the sample of 2,104 family-years from 1992 to 2006. The estimated coefficients are from regressions of the following equation (second step): Maturityi ,t = α + β1CFVolMMFi ,t + β 2CFCori ,t + β 3 AveFamCFCori ,t + β 4 Famturni ,t + β5CashFami ,t +

β 6 log TNAmmfi ,t + β 7 log TNAfami ,t + ε i ,t , where dependent variable is weighted average maturity of MMMF portfolios. The independent variables include volatility of MMMF flows (CFVolMMMF), correlation of flows between MMMFs and other funds of the family (CFCor), and average cash flow correlation of non-MMMFs in the family (AveFamCFCorr), where CFVolMMMF is instrumented on volatility of family flows (CFVolFam), MMMF return (retMMMF) and MMMF risk (riskMMMF). Other variables are family average turnover, ratio of cash holdings in non-MMMFs (CashFam), log of MMMFs and family TNA. T-statistics are reported in parentheses. Panel B. MMMF Maturity Maturity 2SLS Fixed Effect Intercept CFVolMMMF CFcor AveFamCFcor Famturn CashFam

All

> One MMMF

One MMMF

38.19***

26.91***

39.70***

(5.45)

(2.68)

(2.74)

-0.012**

(-1.52)

-0.662

-1.381

-0.678

(-0.67)

(-1.13)

(-0.36)

-1.963

-0.238

-1.832

(-0.70)

(-0.06)

(-0.45)

0.013

1.798

0.039

(0.02)

(0.47)

(0.07)

0.0002 2.289* (1.76)

logTNAfam

-0.202

(-2.35)

(1.48) logTNAmmmf

-0.012**

(-2.02)

-0.996* (-1.68)

0.0001 (1.07) 3.442** (2.01)

0.0027 (0.67) 3.008 (1.03)

-0.671

-1.230

(-0.86)

(-0.96)

Observations

2,104

1,492

612

R

0.002

0.002

0.01

2

***Significant at the 0.01 level. **Significant at the 0.05 level. *Significant at the 0.10 level.

volatility. For all samples, families also have a negative relation between cash in the family non-MMMFs and volatility of MMMF flows. This suggests that as the family funds hold more cash, MMMFs are used as cash centers to a lesser extent, as more cash in non-MMMFs allows the funds to more easily absorb liquidity shocks. Size of MMMFs positively affects cash flow volatility. Fund family size negatively affects MMMF cash flow volatility as larger families probably have more capacity of liquidity shocks absorption across family funds.

Panel B shows that CFVolMMMF is negatively related to MMMF maturity at less than the 5% level, with a $1 million increase in flow volatility resulting in a 0.012 day decrease of average maturity. This suggests that MMMF managers try to reduce overall transaction costs by shortening securities’ maturity in response to increased cash flow volatility in MMMFs. This in turn, is positively related to flow volatility of non-MMMFs in the family. Overall, results show that families can use cash management strategies such as offsetting negatively related

12

Journal of Applied Finance – No. 1, 2011

Table V. Use of MMMFs as Cash Center by Fund Family Investors

The table reports results from the two-stage least squares fixed effect regression analysis of family loads on volatility of MMMFs’ cash flows at the family level for the sample of 2,222 family-years from 1992 to 2006. The estimated coefficients are from regressions of the following equation (second step): CFVolMMMFi ,t = α + β1CFVolFami ,t + β 2 AveTotalLoad i ,t + β3CFCori ,t + β 4 AveFamCFCori ,t + β5 Famturni ,t +

β 6CashFami ,t + β 7 log TNAfami ,t + β8 log TNAMMMFi ,t + β9 retMMMFi ,t + β10 riskMMMFi ,t ε i ,t , where dependent variable is cash flow volatility of MMMFs in the family. Independent variables are volatility of family flows (CFVolFam), average total loads of the family, correlation of flows between MMMFs and other funds of the family (CFCor), average cash flow correlation of non-MMMFs in the family (AveFamCFCorr), family average turnover, ratio of cash holdings in non-MMMFs (CashFam), log of MMMFs and family TNA, MMMFs’ return and risk. CFVolFam is instrumented on family return (Famret), family risk (Famrisk), and family flows (Famflow). T-statistics are reported in parentheses. CFVolMMMF 2SLS Fixed Effect All Intercept

-407.43 (-1.55)

CFVolFam

0.929**

AveTotalLoad CFcor AveFamCFcorr

(-2.17)

(-1.78)

0.980** 14,605*

0.248 (1.46) 132.7

(1.52)

(1.75)

(0.43)

-23.77

-24.63

-1.93

(-0.65)

(-0.47)

(-0.46)

(-0.09) -0.019** (-1.97)

logTNAfam

-46.10*

6,329

-1.92

logTNAmmmf

-944.73**

(2.01)

(-1.05)

CashFam

One MMMF

(2.36)

-121.36

Famturn

> One MMMF

-286.67 (-1.35) -141.66 (-0.84) -0.019* (-1.76)

-4.52 (-0.42) 0.78 (0.62) -0.012** (-2.01)

139.13***

250.79***

(5.22)

(6.29)

(5.75)

-75.72*

-1.23

-39.16

15.96***

(-1.40)

(-1.74)

(-0.27)

-29,820***

-47,267***

-1,723

(-3.06)

(-3.42)

(-1.48)

riskMMMF

97,289**

141,241**

9,399**

(2.30)

(2.33)

(2.08)

Observations

2,222

1,558

664

0.23

0.26

0.10

retMMMF

R

2

***Significant at the 0.01 level. **Significant at the 0.05 level. *Significant at the 0.10 level.

Agapova – The Role of Money Market Mutual Funds in Mutual Fund Families

cash flows between MMMFs and the rest of the family. Results also indicate that families employ cash management strategies by using the MMMFs as family cash centers. B. Fund Family Investors and Cash Center Analysis To examine whether investors use the option of free asset transfers within the same family and how it reflects in cash flow volatility of MMMFs, two-stage least squares regression analysis of panel data is employed on the following model with control for family and year fixed effects:

13

in family MMMFs, thus increasing the cash flow volatility of the MMMFs. This result is not observed in families with one MMMF. Other variables show the same results as in Table IV Panel A. Family cash flow volatility positively affects MMMF cash flow volatility, larger non-MMMF portfolio cash holdings decrease volatility of MMMF cash flows, MMMF size and risk negatively affect MMMF cash flow volatility, and MMMF returns negatively affect MMMF cash flow volatility.

IV. Conclusion

This paper examines the cross-sectional differences among MMMFs and their role in fund families using data β8 LogTNAMMMFi ,t + β9 retMMMFi ,t + β10 riskMMMFi ,t + ε i ,t . (4) from 1992 through 2006. The study finds evidence that fund families and their investors can use MMMFs as cash centers. The first stage is the same as in Equation (3a). The The results show that flows to the family non-MMMFs dependent variable is the volatility of MMMF cash flows are negatively related to family MMMF flows, and family in the family i calculated non-MMMF cash flow as a standard deviation of Contrary to the perception that MMMFs volatility is positively monthly cash flows in year t. are simply homogeneous transitional cash related to family MMMF AveTotalLoadi,t is the valuecash flow volatility. This weighted average of total accounts, this paper finds that MMMFs suggests that families loads of the family i’s non- play a larger role in mutual fund families. can partially offset cash MMMF in year t. The rest of between The substantial cross-sectional differences transactions the variables are as defined in MMMFs and nonin characteristics of MMMFs can be MMMFs and that families Equation (3b). If non-MMMF loads explained by differences in family-specific place non-MMMFs cash increase the value of the in family MMMFs. In option of free asset transfers characteristics, including the use of cash addition, loads positively within the family for management strategies at the family level. affect cash flow volatility investors, the loads should of MMMFs, suggesting be positively related to the volatility of the MMMFs’ cash that family investors also use family MMMFs as cash flows. However, if the loads discourage investors from centers by utilizing the option of free asset transfer within leaving individual funds (Chordia 1996) and the option of the family. free asset transfers is not provided by the family, fund family Contrary to the perception that MMMFs are simply investors will have fewer uses of family MMMFs as cash homogeneous transitional cash accounts, this paper finds centers. that MMMFs play a larger role in mutual fund families. The Table V presents the results from the regression analysis of substantial cross-sectional differences in characteristics of family non-MMMF loads on MMMF cash flow volatility. As MMMFs can be explained by differences in family-specific expected, total loads are positively related to MMMF cash characteristics, including the use of cash management flow volatility at less than the 10% level for the subsample strategies at the family level. The application of these of multi-MMMF families. This suggests that higher non- strategies can reduce operating costs and improve overall MMMF loads make it more attractive for investors to use performance at the family level, which can translate into the option of free transfer within the family to park their cash significant benefits.n CFVolMMMFi ,t = α + β1CFVolFami ,t + β 2 AveTotalLoad i ,t + β3CFCori ,t +

β 4 AveFamCFCori ,t + β5 Famturni ,t + β 6CashFami ,t + β 7 log TNAfami ,t +

14

Journal of Applied Finance – No. 1, 2011

Appendix A. Cases Janus Mutual Fund Family: Example of Cash Management in a Family

Vanguard Mutual Fund Family: Example of Cash Management in a Family

The following example is only meant to illustrate, not prove, the main results of the paper. Janus ranked 14th out of 562 mutual fund families in the US, with $90 billion in total net assets, representing 1.29% market share as of February 2010. As of March 2010, the firm had three bond, 24 equity, and three asset allocation retail funds in addition to seven equity and three fixedincome institutional mutual funds. Janus also offered 15 equity and three fixed-income separately managed portfolios for large institutional investors. The family has a designated “cash center” that performs all cash transactions for all equity, bond, and balanced funds. Most of these investments are overnights repos. All family cash transactions are bundled together and executed by the cash center. Individual fund managers perform no money market transactions. Additionally, the family has a special money market fund, which is available to the family’s funds, but not to outside investors. This fund holds money market instruments with longer maturities than repos for all of the family’s non-MMMFs. The fund had approximately $2 billion in total net assets, and its internal MMMF securities had an average maturity of 20 days. This structure has helped reduce transactions costs and satisfy the cash and liquidity needs of the firm’s non-money market funds. This example illustrates cash management strategies discussed in the paper that are aimed at cost reduction and positive externalities of money market managers’ expertise. Furthermore, Janus creates incentives for its investors to use family MMMFs to temporarily park cash when they redeem assets from other family funds. These incentives include waived transfer fees and/or loads when investors move assets within the family. After discontinuing its institutional MMMFs in 2009, Janus currently has two retail MMMFs. According to firm, investors use its MMMFs when moving money from/to other family funds instead of using money market accounts and/or funds outside of the Janus family. This example illustrates the hypothesis that a mutual fund family can make its MMMFs more attractive to its investors for liquidity transactions, resulting in its investors utilizing the strategy.

The following example is only meant to illustrate, not prove, the main results of the paper. Vanguard is the largest mutual fund family in the US with $1,077 billion in total net assets, representing 15.41% market share as of February 2010. The Vanguard mutual fund family has 117 mutual funds with a total of 219 share classes. As of February 2010, there were three taxable and six tax-exempt money market funds along with 27 bond funds, 25 balanced funds, and 56 equity funds. In line with the family’s overall strategy of cost minimization, cash securities transactions for all Vanguard funds, excluding MMMFs, are done on an aggregate basis through a centralized unit. Individual non-MMMF managers do not participate in the selection and/or trading of money market securities. Prior to 2004, all cash transactions went through an overnight pooled account, which was exempt from Securities & Exchange Commission (SEC) oversight. Since 2004, all non-MMMF funds place their cash holdings in the specially nominated internal MMMF – Market Liquidity Fund. The fund is not available to outside investors and the sole purpose of the fund is to manage cash holdings of other Vanguard funds. The fees of this fund are equal to its operating costs, which are substantially lower than the fees of publicly available Vanguard MMMFs. If the investment objective of a Vanguard non-MMMF does not allow cash to be held in the Market Liquidity Fund, the cash is placed in overnight repos through a centralized family cash account. All fund cash needs in this account are pooled and transactions are aggregated to minimize costs. According to Vanguard, centralized cash management also assists in managing the family’s risk exposure in money market securities. This example demonstrates how a family can utilize cash management strategies, including the use of money market managers’ expertise. In addition, some of Vanguard’s investors use the family’s MMMFs as cash centers for liquidity purposes. According to Vanguard, the family has two types of MMMF investors. One type chooses Vanguard’s MMMFs because of their investment objectives. The other type uses Vanguard’s MMMFs to temporarily park their assets when moving out of other Vanguard funds. It is cheaper for the latter type of investor to keep the cash in the family instead of placing it in outside accounts. As with the Janus example, Vanguard’s practices illustrate that family MMMFs can be more attractive to its investors than outside accounts. In this case, Vanguard’s MMMFs satisfy investors’ liquidity needs at lower costs.

15

Agapova – The Role of Money Market Mutual Funds in Mutual Fund Families

Appendix B. The Choice of Variables for Empirical Model Specification Variable

Definition

Rationale

Expected Sign

MMMFflow

Flows to MMMFs in family i in month t

Dependent variable

Famflow

Flows to other funds in family i in month t

cash flows between the funds can be offset inside the family without transaction costs as described in the first strategy

Famturn

Family portfolio turnover

With higher turnover family need for cash transactions will increase

CashFam

Cash holdings of non-MMMF as percentage of TNA

increase in family portfolios’ cash holding will increase demand for MMMF service as cash center

+

logTNAmmmf

Log of total net assets of MMMF

Size of a fund has an effect on fund flows

+

logTNAfam

Log of total net assets of the family

Size of a family has an effect on fund flows

+

retMMMF

MMMFs’ monthly return

Better performing funds attract more flows

+

riskMMMF

MMMFs’ risk measured as standard deviation of monthly returns in a calendar year

Higher return is associated with higher risk

CFVolMMMF

Volatility of MMMFs cash flows in the family i in year t

Dependent variable

CFVolFam

Volatility of cash flows of other funds in family i in year t

If families use MMMFs as a cash center then family flow volatility affects MMMF flow volatility

CFCor

Correlation between MMMFs and other funds of the family

Controls for possibility of offsetting transaction between MMMFs and the other funds of the family

AveFamCFCor

Average cash flows correlation of non-MMMFs in the family

Controls for the possibility of offsetting transactions across non-MMMFs in the family

Famturn

Family portfolio turnover

Portfolio turnover affects volatility of non-MMMF cash volatility and therefore volatility of MMMF cash flows

+

CashFam

Cash holdings of non-MMMF as percentage of TNA

Controls for level of cash in the family besides MMMFs

-

logTNAmmmf

Log of total net assets of MMMF

Size of a MMMF has an effect on MMMF flow volatility

+

logTNAfam

Log of total net assets of the family

Size of a family has an effect on MMMF flow volatility

-

retMMMF

MMMFs’ average annual return

Higher MMMF returns decrease volatility of MMMF flows

-

riskMMMF

MMMFs’ risk measured as standard deviation of monthly returns in a calendar year

Higher MMMF risk increases volatility of MMMF flows

+

AveTotalLoad

Average total loads of the family

Loads increase the value of the option of free asset transfer within the family

+

Maturity

Value-weighted average maturity of MMMFs portfolio

Dependent variable

CFVolMMMF

Cash flow volatility of MMMFs

If maturity is a tool to manage cash and liquidity, MMMF cash flow volatility negatively affects MMMF maturity

+

+

+

-

-

16

Journal of Applied Finance – No. 1, 2011

References Carhart, M., 1997, “On the Persistence of Mutual Fund Performance,” Journal of Finance Vol. 52 (No. 1), 57-82.

Investment Company Act of 1940, Rule 2a-7, available at: http://taft.law. uc.edu/CCL/InvCoRls/rule2a-7.html.

Chevalier, J. and G. Ellison, 1997, “Risk Taking by Mutual Funds as a Response to Incentives,” Journal of Political Economy Vol. 105 (No. 6), 1167-1200.

Investment Company Institute (ICI) report, December 2006, available at: http://www.ici.org/pdf/2007_factbook.pdf

Chordia, T., 1996, “The Structure of Mutual Funds Charges,” Journal of Financial Economics Vol. 41 (No. 1), 3-39. Christoffersen, S., 2001, “Why Do Money Fund Managers Voluntarily Waive Their Fees?” Journal of Finance Vol. 56 (No. 3), 1117-1140. DeGennaro, R. and D. Domian, 1996, “Market Efficiency and Money Market Fund Portfolio Managers: Beliefs versus Reality,” Financial Review Vol. 31 (No. 2), 453-474. Domian, D., 1992, “Money Market Mutual Fund Maturity and Interest Rates,” Journal of Money, Credit, and Banking Vol. 24 (No. 4), 519527. Domian, D. and W. Reichenstein, 1997, “Performance and Persistence in Money Market Fund Returns,” Financial Services Review Vol. 6 (No. 3), 169-183. Ferri, M. and H. Oberhelman, 1981, “A Study of the Management of Money Market Mutual Funds: 1975-1980,” Financial Management Vol. 10 (No. 4), 24-29. Gaspar, J., M. Massa, and P. Matos, 2006, “Favoritism in Mutual Fund Families? Evidence on Strategic Cross-Fund Subsidization,” Journal of Finance Vol. 61 (No. 1), 73-104.

Guedj, I. and J. Papastaikoudi, 2003, “Can Mutual Fund Families Affect

the Performance of Their Funds?” available at SSRN: http://ssrn.com/ abstract=467282.

Mamaysky, H. and M. Spiegel, 2001, “A Theory of Mutual Funds: Optimal Fund Objectives and Industry Organization,” Working paper, Yale School of Management. Massa, M., 1998, “Why So Many Mutual Funds?” Working paper, INSEAD. Massa, M., 2003, “How Do Family Strategies Affect Fund Performance? When Performance-Maximization is not the Only Game in Town,” Journal of Financial Economics Vol. 67 (No. 2), 249-304 Nanda, V., J. Wang, and L. Zheng, 2004, “Family Values and the Star Phenomenon: Strategies of Mutual Fund Families,” Review of Financial Studies Vol. 17 (No. 3), 667-698 Packer, J. and T. Pencek, 1990, “Taxable Money Market Mutual Funds’ Average Maturity Index and Short-Term Interest Rates,” Papers and Proceedings, Academy of Economics and Finance Fall, 472-480. Seyfried, W. and J. Packer, 2001, “Money Market Mutual Funds and Market Efficiency: Implications for Individual Investors,” Business Quest. Sirri, E. and P. Tufano, 1998, “Costly Search and Mutual Funds Flows,” Journal of Finance Vol. 53 (No. 5), 1589-1622.