Jun Zhang, Wei Lan, Nengsheng Fang. A Network-Adjusted Multiple Testing Procedure with Application to Mutual Fund Selection[J]. Quarterly Journal of Economics and Management, 2023, 2(4): 119-142.
Citation: Jun Zhang, Wei Lan, Nengsheng Fang. A Network-Adjusted Multiple Testing Procedure with Application to Mutual Fund Selection[J]. Quarterly Journal of Economics and Management, 2023, 2(4): 119-142.

A Network-Adjusted Multiple Testing Procedure with Application to Mutual Fund Selection

  • With the booming development of China's mutual fund market,mutual funds have been sought after by a large number of investors.At the same time,Fund of Funds (FOF) has also experienced explosive growth.How to select funds that can consistently beat the market from among the many funds has attracted widespread attention from academia to practitioners.On the one hand,facing with many fund options,individual investors hope to choose funds with good performance to achieve higher investment returns in the future; On the other hand,FOF  managers are also eager to select excellent funds to improve their performance to attract new capital inflows,expand the management scale of FOF  and achieve higher returns.However,many studies have shown that there is a clear phenomenon of common stockholding among China's mutual funds.Different funds form a complex relationship network through common stockholding,called a “fund network”.Since the stockholding characteristics between connected funds are relatively similar,their returns tend to be more similar,which increases the difficulty of mutual fund selections.Based on this,this paper focuses on how the fund network affects fund performance and explores how to select funds with real stock-picking ability.

    There are two main issues with existing literature on the performance of funds.Firstly,when studying the impact of fund networks on fund performance,most papers only use the topology of the fund network as an explanatory variable,with fund performance being the explained variable.Few papers directly incorporate the fund network matrix into model estimation,which results in a loss of connection information between funds.Secondly,the current research on choosing funds only takes into account the “luck”,and does not consider the network connections between funds.However,a large amount of existing research shows that fund networks do affect fund performance.Therefore,the current fund selection method is not conducive to investors or FOF managers in selecting funds.

    This paper proposes a mutual fund selection model and its estimation method under network adjustment.The paper innovatively utilizes the fund's heavy stock network matrix to construct a penalty term,which is introduced into the objective function of parameter estimation to restrict the difference in fund alpha(α) between two connected funds,thereby estimating the fund's performance alpha after network adjustment.The paper also theoretically proves the proposed method can effectively control the False Discovery Rate (FDR) under certain conditions.In the empirical study,the paper uses balanced panel data of 596 open-ended funds in China from 2016 to 2020 to explore the impact of fund networks on fund performance and proposes a network-adjusted fund selection strategy based on the FDR method.This paper found that:Firstly,the holdings of China's funds in the past five years had a high degree of overlap,resulting in a high correlation between fund performances.Secondly,the findings of this paper verified the well-known “Matthew effect” phenomenon in the mutual fund market.In terms of the fund market,the fund network would have a positive impact on positive alpha funds,enabling them to achieve better investment performance; on the contrary,it would have a negative impact on negative alpha funds,resulting in lower investment returns.Thirdly,the network-adjusted fund selection strategy proposed in this paper performed well and stably compared with different fund selection portfolios under different holding periods.This indicates that separating the fund network effect from fund performance can help select funds with real stock-picking ability,which can bring sustained investment returns to investors.

    Based on the above conclusions,this article proposes the following suggestions:Firstly,if fund managers want to achieve absolute performance advantages,they should first improve their stock-picking ability,and then rely on the common stockholding phenomenon in the fund market to learn the positions of other funds with stock-picking ability,so as to achieve the “icing on the cake” effect.Secondly,investors or FOF  managers can use network-adjusted fund selection strategies to eliminate the impact of fund networks on fund performance and select excellent funds.

    Finally,some feasible research directions for future work are proposed.Firstly,it is worth expanding our method to encompass situations where funds are linked indirectly,as we seek to examine the effects of indirect common stockholding on fund performance.Secondly,it is of interest to construct a directed dynamic fund network based on the change of fund holdings and explore its dynamic impact on fund performance.Thirdly,the sample period presents unbalanced fund data as a result of liquidation or new establishment.Therefore,it is important to expand the conventional approach to fund selection,which is based on balanced panel data,to include fund selection under unbalanced panel data.Fourthly,since private equity funds can choose to disclose performance autonomously and have strict liquidation mechanisms,their data usually have serious missing features.Extending the method of this paper to private equity fund selection under missing data also has practical significance.
  • loading

Catalog

    Turn off MathJax
    Article Contents

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return