Abstract
N mutual funds compete for fund flows based on relative performance over their average returns, by choosing between an idiosyncratic and a common risky investment opportunities. The unique constant equilibrium is derived in closed form, which implies that funds generally decrease the investments in their idiosyncratic risky assets under competition, in order to lower the risk of the relative performance. It pushes all funds to herd and hurts their after-fee performance. However, the sufficiently disadvantaged funds with poor idiosyncratic investment opportunities or highly risk averse managers may take excessive risk for a better chance of attracting new investments, and their performance may improve comparing to the case without competition and benefit the investors.
Similar content being viewed by others
Data Availability Statement
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.
Notes
Note that even in a setting of common information in this paper, different fund managers may specialize in different investment opportunities based on their skill and preference, and for simplicity, we summarize this specialization as one idiosyncratic \(S_i\) for each fund (see the same settings in [4, 40]). The analysis in the following can adapt to the case in which each manager has access to the same N risky assets as in [3], with only notational changes.
Similarly in the rest of the paper, with positive integer n, \(v \in \mathbb {R}^n\) and \(D \in \mathbb {R}^{n\times n}\), let \(v_{-i} \in \mathbb {R}^{n-1}\) be the vector after removing v’s ith element, and \(D_{-i} \in \mathbb {R}^{(n-1) \times (n-1)} \) be the matrix after removing D’s ith row and ith column.
According to the proof of Lemma 4.3, \((1-\rho _{1\,m}^2)(1-\rho _{2\,m}^2) \ge (\rho _{12}-\rho _{1\,m}\rho _{2\,m})^2\), and thus \(\kappa _1 >0\).
Though it is a simplified setting in which the first \(N-1\) assets are perfectly correlated and the last one is perfectly negatively correlated with other assets, the manager’s portfolio choice problem is not trivial. For fund \(i \in \{1,\dots ,N-1\}\), the fund flow due to the last fund brings positive exposure to their own idiosyncratic risk. Yet to hedge such a risk manager i may not want to lower the risky investment by too much, because it may hurt the absolute return of the fund. Also, this setting does not create arbitrage opportunities, because each fund only has access to one investment opportunity.
If \({\bar{\lambda }} = N-1\), then the industry average both with and without competition is riskless and Beta coefficients are not well defined.
References
Aivaliotis, G., Palczewski, J.: Investment strategies and compensation of a mean-variance optimizing fund manager. Eur. J. Oper. Res. 234(2), 561–570 (2014)
Anthropelos, Michail, Geng, Tianran, Zariphopoulou, Thaleia: Competition in fund management and forward relative performance criteria. SIAM J. Financial Math. 13(4), 1271–1301 (2022)
Basak, S., Makarov, D.: Strategic asset allocation in money management. J. Finance 69(1), 179–217 (2014)
Basak, S., Makarov, D.: Competition among portfolio managers and asset specialization. Available at SSRN 1563567 (2015)
Basak, S., Pavlova, A., Shapiro, A.: Optimal asset allocation and risk shifting in money management. Rev. Financial Stud. 20(5), 1583–1621 (2007)
Bielagk, J., Lionnet, A., Dos Reis, G.: Equilibrium pricing under relative performance concerns. SIAM J. Financial Math. 8(1), 435–482 (2017)
Bodnaruk, A., Simonov, A.: Loss-averse preferences, performance, and career success of institutional investors. Rev. Financial Stud. 29(11), 3140–3176 (2016)
Boyle, P., Garlappi, L., Uppal, R., Wang, T.: Keynes meets markowitz: the trade-off between familiarity and diversification. Manage. Sci. 58(2), 253–272 (2012)
Brennan, M.J.: The optimal number of securities in a risky asset portfolio when there are fixed costs of transacting: theory and some empirical results. J. Financial Quant. Anal. 10(3), 483–496 (1975)
Brown, K.C., Van Harlow, W., Starks, L.T.: Of tournaments and temptations: an analysis of managerial incentives in the mutual fund industry. J. Finance 51(1), 85–110 (1996)
Brown, N.C., Wei, K.D., Wermers, R.: Analyst recommendations, mutual fund herding, and overreaction in stock prices. Manage. Sci. 60(1), 1–20 (2014)
Brown, S.J., Goetzmann, W.N.: Mutual fund styles. J. Financial Econ. 43(3), 373–399 (1997)
Browne, S.: Stochastic differential portfolio games. J. Appl. Prob. 37(1), 126–147 (2000)
Cai, F., Han, S., Li, D., Li, Y.: Institutional herding and its price impact: evidence from the corporate bond market. J. Financial Econ. 131(1), 139–167 (2019)
Cheng, T., Xing, S., Yao, W.: An examination of herding behaviour of the chinese mutual funds: a time-varying perspective. Pacific Basin Finance J. 74, 101820 (2022)
Chevalier, J., Ellison, G.: Risk taking by mutual funds as a response to incentives. J. Polit. Econ. 105(6), 1167–1200 (1997)
Choi, N., Sias, R.W.: Institutional industry herding. J. Financial Econ. 94(3), 469–491 (2009)
Dal Forno, A., Merlone, U.: Incentives and individual motivation in supervised work groups. Eur. J. Oper. Res. 207(2), 878–885 (2010)
D’Arcangelis, A.M., Rotundo, G.: Herding in mutual funds: a complex network approach. J. Bus. Res. 129, 679–686 (2021)
Deng, C., Zeng, X., Zhu, H.: Non-zero-sum stochastic differential reinsurance and investment games with default risk. Eur. J. Oper. Res. 264(3), 1144–1158 (2018)
Dong, X., Feng, S., Sadka, R.: Liquidity risk and mutual fund performance. Manage. Sci. 65(3), 1020–1041 (2019)
Elliott, R.J., Siu, T.K.: On risk minimizing portfolios under a Markovian regime-switching black-scholes economy. Ann. Oper. Res. 176(1), 271–291 (2010)
Ferreira, M.A., Keswani, A., Miguel, A.F., Ramos, S.B.: The flow-performance relationship around the world. J. Bank. Finance 36(6), 1759–1780 (2012)
Frei, C., Dos Reis, G.: A financial market with interacting investors: Does an equilibrium exist? Math. Financial Econ. 4(3), 161–182 (2011)
Fung, W., Hsieh, D.A.: A primer on hedge funds. J. Emp. Finance 6(3), 309–331 (1999)
Golub, G.H., Van Loan, C.F.: Matrix Computations. Johns Hopkins Studies in the Mathematical Sciences, third edition Johns Hopkins University Press, Baltimore, MD (1996)
Graham, J.R.: Herding among investment newsletters: theory and evidence. J. Finance 54(1), 237–268 (1999)
Grinblatt, M., Titman, S., Wermers, R.: Momentum investment strategies, portfolio performance, and herding: a study of mutual fund behavior. Am. Econ. Rev. 85(5), 1088–1105 (1995)
Gruber, M.: Another puzzle: the growth in actively managed mutual funds. J. Finance 51(3), 783–810 (1996)
Chen, G., Guo, X., Zhang, C.: Analyst target price revisions and institutional herding. Int. Rev. Financial Anal. 82, 102189 (2022)
Han, J., Ma, G., Yam, S.CP.: Relative performance evaluation for dynamic contracts in a large competitive market. Eur. J. Oper. Res. (2022)
Hardy, G.H., Littlewood, J.E., Pólya, G.: Inequalities. Cambridge University Press, Cambridge (1952)
Huang, J., Wei, K.D., Yan, H.: Participation costs and the sensitivity of fund flows to past performance. J. Finance 62(3), 1273–1311 (2007)
Hudson, Y., Yan, M., Zhang, D.: Herd behaviour and investor sentiment: evidence from uk mutual funds. Int. Rev. Financial Anal. 71, 101494 (2020)
Ippolito, R.A.: Consumer reaction to measures of poor quality: evidence from the mutual fund industry. J. Law Econ. 35(1), 45–70 (1992)
Jiang, H., Verardo, M.: Does herding behavior reveal skill? An analysis of mutual fund performance. J. Finance 73(5), 2229–2269 (2018)
Kempf, A., Ruenzi, S.: Tournaments in mutual-fund families. Rev. Financial Stud. 21(2), 1013–1036 (2008)
Kempf, A., Ruenzi, S., Thiele, T.: Employment risk, compensation incentives, and managerial risk taking: Evidence from the mutual fund industry. J. Financial Econ. 92(1), 92–108 (2009)
Lacker, D., Soret, A.: Many-player games of optimal consumption and investment under relative performance criteria. Math. Financial Econ. 14(2), 263–281 (2020)
Lacker, D., Zariphopoulou, T.: Mean field and n-agent games for optimal investment under relative performance criteria. Math. Finance 29(4), 1003–1038 (2019)
Lioui, A., Poncet, P.: Optimal benchmarking for active portfolio managers. Eur. J. Oper. Res. 226(2), 268–276 (2013)
Liu, H.: Solvency constraint, underdiversification, and idiosyncratic risks. J. Financial Quant. Anal. 49(2), 409–430 (2014)
Maug, E., Naik, N.: Herding and delegated portfolio management: the impact of relative performance evaluation on asset allocation. Quarter. J. Finance 1(02), 265–292 (2011)
Ou-Yang, H.: Optimal contracts in a continuous-time delegated portfolio management problem. Rev. Financial Stud. 16(1), 173–208 (2003)
Palomino, F.: Relative performance objectives in financial markets. J. Financial Intermed. 14(3), 351–375 (2005)
Patel, J., Zeckhauser, R., Hendricks, D.: The rationality struggle: illustrations from financial markets. Am. Econ. Rev. 81(2), 232–236 (1991)
Sato, Y.: Fund tournaments and asset bubbles. Rev. Finance 20(4), 1383–1426 (2016)
Scharfstein, D.S., Stein, J.C.: Herd behavior and investment. Am. Econ. Rev. 80(3), 465–479 (1990)
Siemsen, E., Balasubramanian, S., Roth, A.V.: Incentives that induce task-related effort, helping, and knowledge sharing in workgroups. Manage. Sci. 53(10), 1533–1550 (2007)
Sirri, E.R., Tufano, P.: Costly search and mutual fund flows. J. Finance 53(5), 1589–1622 (1998)
Taylor, J.: Risk-taking behavior in mutual fund tournaments. J. Econ. Behav. Org. 50(3), 373–383 (2003)
Uppal, R., Wang, T.: Model misspecification and underdiversification. J. Finance 58(6), 2465–2486 (2003)
Van Nieuwerburgh, S., Veldkamp, L.: Information immobility and the home bias puzzle. J. Finance 64(3), 1187–1215 (2009)
Wei, K.D., Wermers, R., Yao, T.: Uncommon value: the characteristics and investment performance of contrarian funds. Manage. Sci. 61(10), 2394–2414 (2015)
Acknowledgements
G. Wang: Partially supported by National Science Foundation (DMS-2206282)
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Additional information
Communicated by Mihai Sîrbu.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We are very grateful for the insightful comments and suggestions from two anonymous reviewers, which help greatly improve the contribution of our work
Appendix
Appendix
The proof of Theorem 3.1
The first step is to find the optimal portfolio choice of the fund i, given the investment strategies \(\pi _{-i}\) and \(\theta _{-i}\) of other funds. Since we focus on constant equilibria, assume that \(\pi _{-i}\) and \(\theta _{-i}\) are constants. Then, with \({\tilde{X}}_{it} = \exp \left( -\left( r -{\tilde{\psi }}_i\right) t\right) X_{it}\), where \({\tilde{\psi }}_i = \left( 1+\frac{N-1}{N} \alpha _i\right) \psi _i - \frac{\alpha _i}{N} \sum _{j\ne i}^N\psi _j\), Fubini’s theorem implies that
For any \((\pi _i,\theta _i)\in {\mathcal {A}}_i\times \varTheta \) (\(1\le i\le N\)), let \(\tilde{\pi }_{it} = \left( 1+\frac{N-1}{N}\alpha _i \right) \sigma _i \pi _{it}\), \(\tilde{\theta }_{it} = \left( 1+\frac{N-1}{N}\alpha _i \right) b \theta _{it}\), so that \(\pi = (\pi _1,\cdots ,\pi _N) = A_f{\tilde{\pi }}\) and \(\theta = A_m{\tilde{\theta }}\), where \(\tilde{\pi }\) and \(\tilde{\theta }\) are N-dimensional vectors with \((\tilde{\pi })_i = \tilde{\pi }_i\) and \((\tilde{\theta })_i = \tilde{\theta }_i\). Thus,
With \(\phi _{it}= \left[ \begin{array}{c} \tilde{\pi }_{it} \\ \tilde{\theta }_{it} - \sum \limits _{j\ne i}^N c_{ij}\tilde{\theta }_{j} \end{array}\right] , h_i = \left[ \begin{array}{c} \lambda _i \\ \lambda _m \end{array} \right] , F_{it} = \left[ \begin{array}{c} W_{it} \\ B_t \end{array}\right] , \lambda _{-i} = \left[ \begin{array}{cccc} \cdots&\lambda _{i-1}&\lambda _{i+1}&\cdots \end{array}\right] '\), and \(W_{-it} = \left[ \begin{array}{cccc} \cdots&W_{(i-1)t}&W_{(i+1)t}&\cdots \end{array}\right] '\), the dynamics of \({\tilde{X}}\) is
where \(C_i\) is an \((N-1)\)-dimensional matrix with diagonal entries \(c_{ij}\) for \(1\le j \le N\) and \(j\ne i\).
Lemma 4.2 shows that \(\hat{\phi }_i = \frac{1}{\gamma _i} w_i^{-1} h_i + w_i^{-1} w_{-i}C_i\pi _{-i}\) maximizes \(\frac{\mathbb {E}\left[ {\tilde{X}}_{it}^{1-\gamma _i}\right] }{1-\gamma _i}\). Since it is a constant strategy independent of t, it also maximizes the discounted expected utility from management fees for each manager i
\(\hat{\phi }_i = \frac{1}{\gamma _i} w_i^{-1} h_i + w_i^{-1} w_{-i}C_i{\tilde{\pi }}_{-i}\) for \(1\le i \le N\) are 2N equations of constants \({\tilde{\pi }} = ({\tilde{\pi }}_1,\cdots ,{\tilde{\pi }}_N)\) and \({\tilde{\theta }} = ({\tilde{\theta }}_1,\cdots ,{\tilde{\theta }}_N)\):
of which the solution corresponds to the equilibrium strategies of the N funds. Since Lemma 4.3 shows that \(P_f\) and \(P_m\) are invertible, there exists a unique solution \({\tilde{\pi }} = P_f^{-1}\gamma ^{-1}\lambda _f\), \({\tilde{\theta }} = P_m^{-1}\left( \gamma ^{-1} \eta _m + C {\tilde{\pi }} \right) \). Therefore, \(\pi ^* = A_f P_f^{-1} \gamma ^{-1} \lambda _f\), and \(\theta ^* = A_m P_m^{-1} \left( \gamma ^{-1} \eta _m + C A_f^{-1} \pi ^* \right) . \) \(\square \)
Lemma 4.2
Given constant \(\pi _{-i}\) and \(\theta _{-i}\), \(\mathop {\textrm{arg}\,\textrm{max}}\limits _{\phi _i:(\pi _i,\theta _i)\in {\mathcal {A}}_i\times \varTheta } \frac{\mathbb {E}\left[ {\tilde{X}}_{it}^{1-\gamma _i}\right] }{1-\gamma _i} = \hat{\phi }_i = \frac{1}{\gamma _i} w_i^{-1} h_i + w_i^{-1} w_{-i}C_i\pi _{-i}\), for every \(0\le t\le T\), where \( w_i = \left[ \begin{array}{cc} 1 &{} \rho _{im} \\ \rho _{im} &{} 1 \end{array}\right] , w_{-i} = \left[ \begin{array}{c} (\rho _i)_{-i}^\prime \\ (\rho _m)_{-i}^\prime \end{array}\right] \), and \(\rho _i\) is the N-dimensional vector with \((\rho _i)_j = \rho _{ij}\).
Proof
We prove the case of \(0<\gamma _i <1\) and focus on \(\mathbb {E}\left[ {\tilde{X}}_{it}^{1-\gamma _i}\right] \) because \(1-\gamma _i >0\). The case of \(\gamma >1\) follows similarly. Define a stochastic process \(\xi \) such that \(\xi _0 = 1\) and
where \(M_i\) and \(M_{-i}\) are two constant vectors to be determined later, which satisfy \(w_i M_i + w_{-i}M_{-i} = h_i\). Then,
Thus, \(\xi _t \hat{X}_{it}\) is a nonnegative local martingale and hence a supermartingale. Therefore (ignoring the positive \(1-\gamma _i\)), by Hölder’s inequality and noticing that \({\tilde{X}}_{i0} = X_{i0} = 1\),
which is an upper bound for \(\mathbb {E}\left[ \tilde{X}_{it}^{1-\gamma _i}\right] \) corresponding to any \((\pi _i,\theta _i)\in {\mathcal {A}}_i\times \varTheta \).
Next we search for the minimum among all such upper bounds corresponding to different choices of \(M_i\) and \(M_{-i}\), by considering the following constrained minimization problem:
The corresponding Lagrangian function, with Lagrange multiplier l, is
The first-order conditions for \(M_i\), \(M_{-i}\) and l are
Plugging (13) into (14) implies that
Instead of discussing the uniqueness of solutions to the above equation, we pick out one of them \(M_{-i} = -\gamma _i C_i \pi _{-i}\), \(M_i = \gamma _i{\hat{\phi }}_i\), \(l= {\hat{\phi }}_i\), and verify that the candidate strategy \({\hat{\phi }}_i\) can achieve the upper bound corresponding to \(M_{-i}\) and \(M_i\), which verifies that \({\hat{\phi }}_i\) is indeed the maximizer of \(\mathbb {E}\left[ {\tilde{X}}_{it}^{1-\gamma _i}\right] \).
The upper bound corresponding to \(M_{-i} = -\gamma _i C_i \pi _{-i}\) and \(M_i = \gamma _i{\hat{\phi }}_i\) is
On the other hand, for \({\tilde{X}}_i\) corresponding to \(\hat{\phi }_i\),
Thus,
which coincides with the upper bound in (15). \(\square \)
Lemma 4.3
\(P_f\) and \(P_m\) are invertible.
Proof
\(P_f\) and \(P_m\) can be rewritten as \(P_f = A_1P_{diag} P_1 P_{diag}A_2\), and \(P_m = P_{diag}^2A_1P_2A_2\), where \(A_1\), \(A_2\) and \(\textbf{P}_{diag}\) are \(N\times N\) diagonal matrices with \((A_1)_{ii} = \frac{1}{N+(N-1)\alpha _i}\), \((A_2)_{ii} = \alpha _i\) and \((P_{diag})_{ii} = \sqrt{1-\rho _{im}^2}\), and \(P_1\) and \(P_2\) are \(N\times N\) matrices with
On the other hand, for \(i\ne j\), Brownian motions \(W_i\) and \(W_j\) can be written as
where \(Z_{i}, Z_{j}\) are Brownian motions independent of B. Suppose that \(\langle Z_i,Z_j\rangle _t = \rho ^z_{ij} t\), and then \(\rho _{ij} = \rho _{im}\rho _{jm} + \sqrt{(1-\rho _{im}^2)(1-\rho _{jm}^2)} \rho ^z_{ij}\), which implies that \(\frac{(\rho _{ij}-\rho _{im}\rho _{jm})^2}{(1-\rho _{im}^2)(1-\rho _{jm}^2)} = (\rho ^z_{ij})^2 \le 1\). Since \(c_{ii} = \frac{\alpha _i}{N+(N-1)\alpha _i} < \frac{1}{N-1}\), both \(P_1\) and \(P_2\) are strictly diagonally dominated matrices, and hence invertible. Therefore, \(P_f\) and \(P_m\) are invertible, because the diagonal matrices \(A_1\), \(A_2\) and \(P_{diag}\) are also invertible.\(\square \)
The proof of Proposition 3.1
-
(i)
The claim follows from the fact that \(\frac{\partial \pi _i^*}{\partial \lambda _i} = \frac{2(1-\rho _{jm}^2)}{(2+\alpha _i)\sigma _i \gamma _i \kappa _1}\) and \(\kappa _1 >0\).
-
(ii)
The claim follow from \(\frac{\partial \pi _i^*}{\partial \lambda _j} = \frac{2\alpha _i(\rho _{12}-\rho _{1\,m}\rho _{2\,m})}{(2+\alpha _i)(2+\alpha _j)\sigma _i \gamma _j \kappa _1}\).
-
(iii)
The claims are direct results of the following derivatives and the fact that \((1-\rho _{1\,m}^2)(1-\rho _{2\,m}^2) \ge (\rho _{12}-\rho _{1\,m}\rho _{2\,m})^2\) from the proof of Lemma 4.3.
$$\begin{aligned} \frac{\partial \theta _i}{\partial \lambda _i} =\,&-\frac{2(1-\rho _{jm}^2)\rho _{im}}{(2+\alpha _i)b\kappa _2\gamma _i \kappa _1}\left[ \left( 1+\frac{\alpha _1\alpha _2}{(2+\alpha _1)(2+\alpha _2)}\right) (1-\rho _{1m}^2)(1-\rho _{2m}^2)\right. \\&- \left. \frac{2\alpha _1\alpha _2}{(2+\alpha _1)(2+\alpha _2)}(\rho _{12}-\rho _{1m}\rho _{2m})^2\right] , \\ \frac{\partial \theta _i}{\partial \lambda _j} =&- \frac{2\alpha _i(1-\rho _{im}^2)}{(2+\alpha _1)(2+\alpha _2)b \kappa _2 \gamma _j \kappa _1}\left[ \left( 1+\frac{\alpha _1\alpha _2}{(2+\alpha _1)(2+\alpha _2)}\right) \cdot \right. \\&\left. \left( 1-\rho _{1m}^2\right) \left( 1-\rho _{2m}^2\right) \rho _{jm}-\frac{2\alpha _1\alpha _2\rho _{jm}}{(2+\alpha _1)(2+\alpha _2)}\left( \rho _{12}-\rho _{1m}\rho _{2m}\right) ^2 \right. \\&- \left. \left( 1-\frac{\alpha _1\alpha _2}{(2+\alpha _1)(2+\alpha _2)}\right) (1-\rho _{jm}^2) (\rho _{jm}-\rho _{12}\rho _{im})\right] . \end{aligned}$$
\(\square \)
The proof of Proposition 3.2
The claims follow from the derivatives of \(\pi ^*_i\) with respect to \(\alpha _i\) and \(\alpha _j\).
\(\square \)
The proof of Proposition 3.3
Following (7) and (10), with \(\rho _{12}\in (-1,1)\) and \({\bar{\lambda }} \le 1\),
On the other hand,
Thus, \(\pi ^*_2 \le \pi ^M_2\) is equivalent to \(\frac{2\alpha \rho _{12}}{2+\alpha }+ \alpha \left( \frac{\alpha \rho _{12}^2}{2+\alpha }-1\right) {\bar{\lambda }}\le 0\), which, since \(\frac{\alpha \rho _{12}^2}{2+\alpha } <1\), always holds for \(\rho _{12}<0\), and is equivalent to \({\bar{\lambda }} \ge \frac{2\rho _{12}}{2+\alpha (1-\rho _{12}^2)}\) if \(\rho _{12} \ge 0\). Finally,
Thus, if \(\rho _{12} \ge 0\), \({\bar{\lambda }} + \frac{\alpha \rho _{12}}{2+\alpha }>0\), and \(\eta _2^* \le \eta _2^M\) if and only if \(\left( 2+ \alpha (1-\rho _{12}^2)\right) {\bar{\lambda }} - 2\rho _{12}\ge 0\), or equivalently \({\bar{\lambda }} \ge \frac{2\rho _{12}}{2+ \alpha (1-\rho _{12}^2)}\). If \(\rho _{12} < 0\), \( \left( 2+ \alpha (1-\rho _{12}^2)\right) {\bar{\lambda }} - 2\rho _{12} > 0\), and \(\eta _2^* \le \eta _2^M\) if and only if \({\bar{\lambda }} \ge -\frac{\alpha \rho _{12}}{2+\alpha }\). \(\square \)
The proof of Proposition 3.4
Therefore, if \({\bar{\lambda }}=1\), \(\textrm{Beta}^*_1 = \textrm{Beta}^M_1 = 1\). Otherwise, since \(\bar{\lambda }<1\), the sign of \(|\text {Beta}_1^* - 1| -|\text {Beta}_1^M -1| = \textrm{Beta}^*_1 - \textrm{Beta}^M_1\) is the same as that of
If \(\rho _{12} \ge 0\), \(1 + \left( \frac{2+\alpha }{1+\alpha }\frac{1}{\rho _{12}}+\frac{\alpha }{1+\alpha } \rho _{12}\right) \bar{\lambda } + \bar{\lambda }^2 \ge 0\), and hence \(|\textrm{Beta}_1^* - 1| -|\textrm{Beta}_1^M -1| \le 0\). If \(\rho _{12} < 0\), \(|\text {Beta}_1^* -1| - |\textrm{Beta}_1^M - 1| \le 0\) is equivalent to
Since \(\frac{\alpha }{1+\alpha } \rho _{12} + \frac{2+\alpha }{1+\alpha } \frac{1}{\rho _{12}}\) is negative and is decreasing in \(\rho _{12} \in [-1,0)\), the maximum value at \(\rho _{12} = -1\) is \(-2\), and \(\varDelta \ge 0\). Since \(\lambda \le 1\), the two roots of the left hand side of (17) are
Thus, \(|\text {Beta}_1^* - 1|-|\text {Beta}_1^M -1| \le 0\) if \(\bar{\lambda } \le \frac{-\left( \frac{\alpha }{1+\alpha } \rho _{12} + \frac{2+\alpha }{1+\alpha } \frac{1}{\rho _{12}} \right) - \sqrt{\varDelta }}{2}\), and otherwise the inequality is reversed.
On the other hand, algebraic calculations show that \(\textrm{Beta}_2^*-1 = - \left( \textrm{Beta}^*_1 -1\right) \le 0\) and \(\textrm{Beta}_2^M-1 = -\left( \textrm{Beta}^M_1-1 \right) \le 0\). Therefore, the sign of \(|\text {Beta}_2^* - 1| -|\text {Beta}_2^M -1| = -\textrm{Beta}^*_2 + \textrm{Beta}^M_2\) is the same as (16), and equivalent conditions for Fund 1 still hold. \(\square \)
The proof of Proposition 3.5
Let \(\tilde{X}_{it} = \exp \left( -\left( r- {\tilde{\psi }}_i \right) t\right) X_{it}\). Then, with \(\zeta _{it} = \frac{N+(N-1)\alpha _i}{N} \theta _{it} -\frac{\alpha _i}{N} \sum _{j\ne i}^N \theta _{jt}\), \({\tilde{X}}_i\) follows
We first calculate the optimal \(\zeta _i\) (or equivalently the optimal \(\theta _i\)) given \(\theta _j\)’s (\(j\ne i\)) of other funds. With \(d\xi _t /\xi _t = - \lambda _m dB_t\) and \(\xi _0 =1\), \(d(\xi _t\tilde{X}_{it})= \xi _t \tilde{X}_{it} \left( \zeta _{it} b - \lambda _m \right) dB_t\), which is a nonnegative local martingale, and thus a supermartingale. Then, for \(0<\gamma _i <1\) (the case of \(\gamma _i >1\) follows similarly), by Hölder’s inequality, for any \(\theta = (\theta _1,\dots ,\theta _N) \in \varTheta ^N\),
which gives an upper bound of \(\mathbb {E}\left[ \frac{\tilde{X}_{it}^{1-\gamma _i} }{1-\gamma _i}\right] \). On the other hand, with \(\theta _{it} = \frac{N\left( \frac{\lambda _m}{\gamma _i b} + \frac{\alpha _i}{N}\sum \limits _{j\ne 1}^N \theta _{jt}\right) }{N+(N-1)\alpha _i}\), and thus \(\zeta _{it} = \frac{\lambda _m}{\gamma _i b}\), \(\mathbb {E}\left[ \frac{1}{1-\gamma _i} \tilde{X}_{it}^{1-\gamma _i} \right] = \frac{\exp \left( \frac{1-\gamma _i}{2\gamma _i} \lambda _m^2 t \right) }{1-\gamma _i},\) which indicates that \(\zeta _i\) is the maximizer of \(\mathbb {E}\left[ \frac{\tilde{X}_{it}^{1-\gamma _i} }{1-\gamma _i}\right] \). Since \(\zeta _{it} = \frac{\lambda _m}{\gamma _i b}\) is a constant strategy, independent of t, it also maximizes manager i’s expected utility
and the optimal strategy given \(\theta _j\)’s (\(j\ne i\)) is \(\theta _{it} = \frac{N\left( \frac{\lambda _m}{\gamma _ib} + \frac{\alpha _i}{N}\sum \limits _{j\ne i}^N \theta _{jt}\right) }{N+(N-1)\alpha _i}\).
To find the equilibrium, it suffices to solve the system of N equations, each representing the optimal strategy given the portfolio of other funds: \(P_I \theta _t = \frac{\lambda _m}{b}\gamma ^{-1}e\), where \(\theta _t= (\theta _{1t},\cdots ,\theta _{Nt})'\), e is the N-dimensional vector with all entries equal to 1, \(P_I\) the \(N\times N\) matrix with \((P_I)_{i,j} = {\left\{ \begin{array}{ll} \frac{N+(N-1)\alpha _i}{N} &{} \text { if } i =j \\ -\frac{\alpha _i}{N} &{} \text { if } i \ne j, \end{array}\right. }\) and \(\gamma \) is defined in Theorem 3.1. Since \(P_I\) is strictly diagonally dominated and thus invertible, there exists a unique solution \(\theta ^* = \frac{\lambda _m}{b}P_I^{-1} \gamma ^{-1}e\) for every \(0\le t\le T\). Furthermore, since \(P_I = -\frac{1}{N} A_2 (D + ee')\), where \(A_2\) is the diagonal matrix defined in the proof of Lemma 4.3, D is an \(N\times N\) diagonal matrix with \((D)_{ii} = -\frac{1}{c_{ii}}-1\), \(1\le i\le N\), by Sherman–Morrison–Woodbury formula (See equation 2.1.4 in [26]),
and this solution reduces to \(\theta _i^* = \frac{\lambda _m}{b} \left( \frac{1}{1+\alpha _i}\frac{1}{\gamma _i} + \frac{\alpha _i}{1+\alpha _i} \frac{1}{\bar{\gamma }}\right) \) for \(1\le i \le N\). \(\square \)
The proof of Proposition 3.6
-
(i)
Both \(\theta ^*_i\) and \(\theta ^M_i\) are proportion to \(\frac{\lambda _m}{b}\), while the coefficient for \(\theta ^M_i\) is \(\frac{1}{\gamma _i}\) and that for \(\theta ^*_i\) is a convex combination between \(\frac{1}{\gamma _i}\) and \(\frac{1}{{\bar{\gamma }}}\). The claim follows by the comparing the two coefficients.
-
(ii)
Since \(\frac{1}{N}\sum \limits _{i=1}^{N} \left( \frac{1}{1+\alpha _i}\frac{1}{\gamma _i} + \frac{\alpha _i}{1+\alpha _i}\frac{1}{{\bar{\gamma }}}\right) = \frac{1}{{\bar{\gamma }}}\), \({\bar{\theta }}^* = \frac{\lambda _m}{b{\bar{\gamma }}}\), and
$$\begin{aligned} \bar{\theta }^* - \bar{\theta }^M =\,&\frac{\lambda _m}{b} \left( \frac{1}{\bar{\gamma }} - \frac{1}{N} \sum _{i = 1}^N \frac{1}{\gamma _i}\right) = \frac{\lambda _m}{b} \left( \frac{1+{\bar{\alpha }}}{N} \sum _{i = 1}^N \frac{1}{(1+\alpha _i)\gamma _i}- \frac{1}{N} \sum _{i = 1}^N \frac{1}{\gamma _i}\right) \\ =\,&\frac{\lambda _m}{b} \frac{1}{N} \sum _{i = 1}^N \left( \frac{1+\bar{\alpha }}{1+\alpha _i}-1 \right) \frac{1}{\gamma _i}. \end{aligned}$$Since \((\gamma _i-\gamma _j)(\alpha _i-\alpha _j) \ge 0\) for every pair of i and j, \(\left( \frac{1+\bar{\alpha }}{1+\alpha _i}-1 \right) \)’s and \(\frac{1}{\gamma _i}\)’s are similarly ordered, and from Tchebychef’s inequality [32, 2.17.1], the above is greater than or equal to
$$\begin{aligned} \frac{\lambda _m}{b} \frac{1}{N} \sum _{i = 1}^N \left( \frac{1+\bar{\alpha }}{1+\alpha _i}-1 \right) \frac{1}{N} \sum _{i = 1}^N\frac{1}{\gamma _i} =0, \end{aligned}$$and the inequality is reversed if \((\gamma _i-\gamma _j)(\alpha _i-\alpha _j) \le 0\) for every pair of i and j.
-
(iii)
\(\bar{\theta }^* = \bar{\theta }^M\) follows from (ii). Furthermore, since \(\frac{1}{{\bar{\gamma }}} = \frac{1}{N}\sum \limits _{i=1}^N\frac{1}{\gamma _i}\),
$$\begin{aligned}&\theta _i^* - \bar{\theta }^* = \frac{\lambda _m}{b}\left( \frac{1}{1+\alpha } \frac{1}{\gamma _i} +\sum _{j=1}^N\left( \frac{\alpha }{N(1+\alpha )} \frac{1}{\gamma _j} \right) - \frac{1}{N}\sum _{j = 1} \frac{1}{\gamma _j} \right) \\&\quad = \frac{\lambda _m}{b}\left( \frac{1}{1+\alpha } \frac{1}{\gamma _i} - \frac{1}{1+\alpha } \frac{1}{N} \sum _{j = 1}^N \frac{1}{\gamma _j}\right) \\&\quad = \frac{1}{1+\alpha }\frac{\lambda _m}{b} \left( \frac{1}{\gamma _i} -\frac{1}{N} \sum _{j = 1}^N \frac{1}{\gamma _j}\right) = \frac{1}{1+\alpha } \left( \theta ^M_i - \bar{\theta }^M\right) . \end{aligned}$$
\(\square \)
The proof of Proposition 3.7
(i) and (ii) follow the same calculation as for Theorem 3.1. For (iii), notice that \(\pi ^{*\alpha }_N - \pi ^{*0}_N = (k_\rho ({\tilde{\lambda }} +\lambda _{N,\gamma _N}) - \lambda _{N,\gamma _N} )/\sigma _1\). Thus, the comparison between \(\pi ^{*0}\) and \(\pi ^{*\alpha }\) is reduced to
Thus, if \(\lambda _{N,\gamma _N} \ge \varphi \sum _{j =1}^N \lambda _{j,\gamma _j}/N\), then \(\lambda _{N,\gamma _N} \ge k_\rho ({\tilde{\lambda }} + \lambda _{1,\gamma _1})\), and \(\pi ^{*0}_i \ge \pi ^{*\alpha }_i\) for every i. \(\square \)
The proof of Proposition 3.8
(i), (ii) is a direct application of Theorem 3.1 and follows similar arguments to those for Proposition 3.7 (iii).
(iii) Given \(\pi ^*_i\)’s, \( \textrm{Beta}_N^* = \frac{(N+\alpha ){\bar{\lambda }} -(N-1)\alpha }{(1+\alpha )({\bar{\lambda }}-(N-1))}\text {, } \textrm{Beta}_N^M = \frac{N {\bar{\lambda }}}{{\bar{\lambda }}-(N-1)}.\) Thus,
For fund \(i \ne N\), \(\textrm{Beta}_i^* = \frac{1}{1+\alpha }-\frac{N}{(1+\alpha )({\bar{\lambda }}-(N-1))}\frac{{\bar{\gamma }}_{-N}}{\gamma _i}\) and \(\textrm{Beta}_i^M= -\frac{N}{{\bar{\lambda }}-(N-1)}\frac{{\bar{\gamma }}_{-N}}{\gamma _i}\). If \({\bar{\lambda }} > N-1\), both \(\textrm{Beta}\)’s are less than 1, and \(\left|\textrm{Beta}_i^M-1 \right|\ge \left|\textrm{Beta}_i^*-1 \right|\) because
If \({\bar{\lambda }} < N-1\) and \(\frac{{\bar{\gamma }}_{-N}}{\gamma _i} \ge \max (\alpha ,1) \frac{N-1-{\bar{\lambda }}}{N}\), both \(\textrm{Beta}\)s are greater than or equal to 1, and
Other cases of \({\bar{\gamma }}_{-N}/\gamma _i\) follow by similar arguments. \(\square \)
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Wang, G., Ye, J. Fund Managers’ Competition for Investment Flows Based on Relative Performance. J Optim Theory Appl 198, 605–643 (2023). https://doi.org/10.1007/s10957-023-02221-4
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10957-023-02221-4