中国股票型开放式基金持股二分网络研究

发布时间：2018-09-06 18:15

【摘要】：近年来，我国金融市场的发展和完善极大地促进了基金市场的蓬勃发展，基金数量及规模均获得了大幅提升，基金公司的投资行为和投资策略对证券市场的影响日渐复杂。根据中国银监会统计数据，截止2013年3月份，共有证券投资基金1257只，基金规模为28340.87亿，约占A、B股上市公司股票市价总值的12.13%，基金市场由此成为当前证券市场最受关注的领域之一。基金市场对于证券市场的影响力，随着基金规模的不断壮大而与日剧增，基金持股对股票市场的影响也有了显著提升；同时，研究基金持股有助于从中获取基金持股行业偏好及基金投资组合等有用的信息，从而更加全面地理解基金持股。 20世纪末，自小世界网络和无标度网络开始，复杂网络结构和性质的研究成为一个引人注目的领域，广泛应用于生产流程、社会关系网直至社会各个领域的研究中，复杂网络理论的发展和广泛应用有助于更好的认识和揭示自然和社会中存在的各种实际网络的内在联系和规律。复杂网络理论与方法将是一座拓宽传统金融经济理论的“金矿”。事实上，在复杂网络理论和方法引入金融经济研究领域前，跨学科的研究者已经采用复杂网络的分析方法开展了复杂网络在其它领域的探索工作并取得了突破性进展。本文就我国基金市场中股票型开放式基金持股信息构建“基金-股票”二分网络，以期用复杂网络的研究方法探索我国股票型开放式基金的持股偏好及其基本特征。本文选取2010~2012年股票型开放式基金持股信息，采用复杂网络方法建立“基金-股票”动态二分网络模型并对其进行加权投影得到股票单顶点加权动态网络和基金单顶点加权动态网络，从网络的基本统计特性和社团结构研究其演化规律。通过研究发现，股票单顶点加权动态网络和基金单顶点加权动态网络均不具有无标度特性，均具有小世界效应；网络的聚集程度基本稳定，股票之间的平均路径长度较小并呈现出逐渐增长的趋势。利用加权CNM算法对该动态网络进行社团划分，发现网络最优模块度Q值维持在较低水平，表明股票单顶点加权动态网络和基金单顶点加权动态网络社团化趋势不明显，，体现了基金的多样化投资策略。对股票单顶点网络中的主要社团按行业进行进一步研究发现:随着时间的推移，基金持股行业偏好更加广泛，主要投资行业由2010~2011年的4个扩展到2012年的7个。
[Abstract]:In recent years, the development and perfection of the financial market in China has greatly promoted the vigorous development of the fund market, the number and scale of the fund have been greatly increased, and the influence of the investment behavior and investment strategy of the fund company on the securities market is becoming more and more complex. As of March 2013, there were 1257 securities investment funds with a total size of 2.834087 trillion, according to statistics from the China Banking Regulatory Commission. Accounting for about 12.13% of the total market value of A, B shares, the fund market has become one of the most concerned areas in the current securities market. The influence of the fund market on the securities market has increased dramatically with the increasing size of the fund, and the impact of the fund holding on the stock market has also increased significantly; at the same time, The research on fund holding helps to obtain useful information such as industry preference and portfolio of fund holding, so as to get a more comprehensive understanding of fund shareholding. At the end of the 20th century, since the beginning of the small world network and scale-free network, The study of complex network structure and properties has become a noticeable field, widely used in production processes, social networks and even in the study of various fields of society. The development and wide application of complex network theory is helpful to better understand and reveal the inherent relations and laws of various practical networks existing in nature and society. The theory and method of complex network will be a gold mine to broaden the traditional financial and economic theory. In fact, before the introduction of the theory and method of complex network into the field of financial and economic research, interdisciplinary researchers have used the analytical method of complex network to carry out the exploration of complex network in other fields and made a breakthrough. Based on the information of shareholding in China's fund market, this paper constructs a two-part network of "Fund-Stock", in order to explore the preference and basic characteristics of shareholding of China's stock open-end fund by using the research method of complex network. This paper selects the stock holding information of the stock open-end fund from 2010 to 2012. The dynamic dichotomous network model of "fund stock" is established by using complex network method, and the weighted dynamic network of single vertex of stock and single vertex of fund are obtained by weighted projection. The evolution of the network is studied from its basic statistical characteristics and community structure. It is found that both stock single vertex weighted dynamic network and fund single vertex weighted dynamic network are scale-free and have small world effect, and the aggregation degree of the network is basically stable. The average path length between stocks is small and showing a growing trend. The weighted CNM algorithm is used to divide the community of the dynamic network, and it is found that the optimal modularity Q value of the network is kept at a low level, which indicates that the trend of corporatization of the stock single vertex weighted dynamic network and the fund single vertex weighted dynamic network is not obvious. Reflect the diversification of the Fund's investment strategy. A further study of the major associations in the stock single vertex network by industry shows that with the passage of time, the preference of the fund holding industry is wider, the main investment industry expanded from 4 in 2010 to 2011 to 7 in 2012.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51;F224

【参考文献】