Strategic Choice of Risk: Evidence from Mutual Fund Families

Literature that examines strategies to increase cash flows to fund families includes Massa (2003), Nanda et al. (2004), Khorana et al. (2005), Gaspar et al. (2006), Guedj and Papastaikoudi (2008), Evans (2010), Chen and Lai (2010), Nohel et al. (2010), Cici et al. (2010), and Bhattacharya et al. (2013).

Here, we add one note related to the importance of Active Shares: Cremers and Petajisto (2009) document that approximately one-third of U.S. equity fund assets are managed by “closet indexers,” i.e., fund managers who should manage their investment portfolios actively but do not deviate enough from their market indexes to justify their fees. This measure has received considerable media coverage, and information services such as the Morningstar Fund Family Report have begun publishing Active Share information as a proxy for fund risk taking tendency.

See Heinkel and Stoughton (1994), Brown et al. (1996), Chevalier and Ellison (1997), Busse (2001), Elton et al. (2003), Qiu (2003), Goriaev et al. (2005), Kempf and Ruenzi (2008), Kempf et al. (2009), Chen and Pennacchi (2009), Basak and Makarov (2012), Schwarz (2012), Cullen et al. (2012), Basak and Makarov (2014).

See Massa (2003); Siggelkow (2003);.Nanda et al. (2004); Gaspar et al. (2006); Kempf and Ruenzi (2008); Pollet and Wilson (2008); Khorana and Servaes (2012).

Specifically, we include funds with the Wiesenberger Objective Codes G, I, GCI, IEQ, LTG, MCG, and SCG; funds with the Strategic Insight Investment Objective Codes AGG, GMC, GRI, GRO, ING, SCG, and SEC; and funds with the Lipper Objective and Classification Codes EI, EIEI, ELCC, G, GI, I, LCCE, LCGE, LCVE, LSE, MC, MCCE, MCGE, MCVE, MLCE, MLGE, MLVE, MR, SCCE, SCGE, SCVE, and SG.

Schwarz (2012) documents that mutual fund tournaments are subject to sorting bias, which is caused by the sorting of first-half risk levels when establishing relative midyear performance. He further demonstrates that the direction of the sorting bias changes depending on the market condition. The author suggests one way to overcome this bias is to use portfolio holdings’ data and calculates the equally-weighted average security standard deviation for each fund based on its midyear holdings. Following his method, to isolate the effect of market condition on fund-level risk taking measures, for the fund level we also calculate the equally-weighted average security standard deviation for each fund based on its prior-quarter holdings. We then calculate the Standard Deviation of Fund Holdings at Family Level as the value-weighted average standard deviation of member funds affiliated with a fund family using each fund’s total net assets as the weight.

Furthermore, for fund level we also calculate the value-weighted average security standard deviation for each fund using the share price provided in the portfolio holding database (i.e., Thomson Reuters Mutual Fund Holding database) and then calculate the Standard Deviation of Fund Holdings at Family Level as the value-weighted average standard deviation of member funds affiliated with a fund family.

We have constructed the empirical analysis based on both equally-weighted and value-weighted methodology, and both methods generate consistent results. We chose to present the equally-weighted results to isolate the effect of market condition on fund-level standard deviation and to include more observations (about 700 more observations are retained based on the equally-weighted method). The results based on value-weighted method are also available upon request.

In this study, we also use two univariate measures to proxy for the flow tournament position of a fund family in period t. In particular, fund families that experience higher relative net flow growth than their own relative return performance are considered interim winners in the fund-flow tournament. Relative Flow Dummy 1 is equal to one when the relative net flow growth of a fund family is higher than its own relative return performance over the previous 1-year period from (t-1) to t year; and 0 otherwise. Relative Flow Dummy 2 is equal to one when the relative net flow growth of a fund family is higher than its own relative return performance over the two-year period from (t-3) to (t-1) year; and 0 otherwise. According to our strategic choice of risk argument, we expect a negative relation between Relative Flow Dummies and family’s risk-taking tendency. Empirical results on Relative Dummies are consistent with our findings using Excess Flows. The detailed procedure used to calculate the variables and the results are reported in Appendix 1 (Tables 8 and 9).

\( {\alpha}_{f,t-1}^{\kern0.2em previous\kern0.2em 1\kern0.2em year\kern0.2em period}=\frac{{\displaystyle {\sum}_{j=1}^J}{\alpha}_{j,t-1}^{\kern0.2em previous\kern0.2em 1\kern0.2em year\kern0.2em period}*TN{A}_{j,t-1}\ }{{\displaystyle {\sum}_{j=1}^J}TN{A}_{j,t-1}\ } \) is defined as the four-factor adjusted return of fund family f over the previous one year period, which is calculated as the value-weighted average of four-factor adjusted returns of all member funds within family f using previous one year return at the end of quarter t-1. TNA _{j,t − 1} is the total net assets of fund j at the end of quarter t-1, and J is the total number of member funds in family f. Similarly, α _{f,t − 1} ^{previous (t − 3, t − 1)year period} is defined as the four-factor adjusted return of fund family f over the two year period from (t-3) to (t-1) year at the end of quarter t-1.

\( Net\ flow\ Growt{h}_{f,t}^{\kern0.2em previous\ 1\ year\ period}=\frac{{\displaystyle {\sum}_{j=1}^J} Cumulative\ Net\ flow\ Growt{h}_{j,t}^{\kern0.2em previous\ 1\ year\ period}*TN{A}_{j,t}}{{\displaystyle {\sum}_{j=1}^J}TN{A}_{j,t}} \) is defined as the net flow growth rate of fund family f over the previous one year period of time t, which is calculated as the value-weighted average of net flow growth rate of all member funds within a fund family f. In particular, TNA _j,t is the total net assets of fund j at the end of quarter t, Cumulative Net flow Growth _j,t ^{previous 1 year period} is the cumulative net flow growth rate of fund j over the previous one year period at the end of quarter t, and J is the total number of member funds in family f at the end of quarter t. Similarly, Net flow Growth _f,t ^{previous (t − 3,t − 1)year period} is defined as the net flow growth rate of fund family f over the two year period from (t-3) to (t-1) year at the end of quarter t.

For brevity, we do not report the results, but the table is available upon request.