基于投资者情绪和宏观经济情况的股市波动分解

发布时间：2018-01-12 23:03

本文关键词：基于投资者情绪和宏观经济情况的股市波动分解　出处：《山西财经大学》2017年硕士论文　论文类型：学位论文

【摘要】：我国股市一直以来被冠以“政策市”、“散户市”的头衔,这样的说法在一定程度上有其合理性,我国市场投资者以散户居多,广大股民接受的投资教育有限,但也不能就此否认股市波动在很大程度上反映了宏观经济运行情况这一事实。实际上,随着学界对股市研究的不断深入,行为金融学认为投资者的情绪影响其交易行为,进而对市场走势造成影响。而传统金融理论认为股价是围绕其内在价值上下波动这一论断依然有其合理性。与此同时越来越多的人意识到投资者情绪和宏观经济波动可能是推动股市波动背后的主要力量。然而,将长、短期波动结合起来考虑缺乏合适的实证方法。基于以上考虑,本文采用了近年来受到广泛关注的GARCH-MIDAS模型作为本文的实证框架,综合考虑股市波动的长短期两个层面的作用机制。同时在投资者情绪代理变量的确定过程中采用了学界常用到的股市换手率作为比较基准,同时笔者尝试通过结合网络爬虫和文本挖掘提高投资者情绪这一短期因素对模型的贡献度,综合运用Python、R语言编程实现网络文本的的信息采集、文本处理、清洗、分词等研究步骤,构建了直接来源于网络文本的投资者情绪指数。通过一系列的理论和实证研究,本文得到如下结论:第一,在波动分为长期与短期两种成分这一认识上,国内外股市波动的表现是一致的:股市波动包含长期成分和短期成两个层面,且宏观经济信息中包含股市波动长期成分的驱动力量。同时,我国股票市场的短期波动成分能通过投资者情绪得到很好的解释。第二,通过网络爬虫和文本挖掘相结合的方式构建情绪指数不失为良策,为后续研究的指标构建极大的拓宽了信息收集渠道。第三,情感词典在情绪指数的构建当中具有不可替代的重要作用。情感词典选词越全面、筛选越具有针对性,最终构建的投资者情绪指数就越具有代表性。特别是涉及到专业领域的情感分析,需要相关领域的丰富语料以及科学合理的情绪指数构建方法。第四,GARCH模型与MIDAS(混频数据抽样)模型结合而构建的GARCH-MIDAS模型能有效拓展模型研究的纵深,同时也能够最大程度上挖掘混频数据的信息,且在我国股市波动成分模型研究中的应用是有效且适用的。
[Abstract]:The stock market of our country has always been labeled as "policy city" and "retail market". To a certain extent, it is reasonable to say that Chinese market investors are mostly retail investors, and the investment education received by the vast number of investors is limited. But we can not deny the fact that the fluctuation of stock market reflects the macroeconomic situation to a great extent. In fact, with the deepening of the academic research on stock market. Behavioral finance believes that investors' emotions affect their trading behavior. The traditional financial theory that the stock price fluctuates around its intrinsic value is still reasonable. At the same time, more and more people are aware of investor sentiment and macro economy. Volatility may be the main force behind the volatility of the stock market. The combination of long and short term volatility is lack of appropriate empirical methods. Based on the above considerations, this paper uses the GARCH-MIDAS model which has received extensive attention in recent years as the empirical framework. Comprehensive consideration of the stock market volatility of the long-term and short-term mechanisms of the two levels of action. At the same time, in the process of determining investor sentiment proxy variables, the stock market turnover rate commonly used in academia as a comparison benchmark. At the same time, the author tries to improve the contribution of investor sentiment to the model by combining web crawler and text mining, and synthetically uses Python R language to realize the information collection of network text. Text processing, cleaning, word segmentation and other research steps, directly derived from the Internet text investor sentiment index. Through a series of theoretical and empirical research, this paper obtains the following conclusions: first. In the understanding that volatility is divided into long-term and short-term components, the performance of domestic and foreign stock market volatility is consistent: stock market volatility includes long-term components and short-term into two levels. And the macroeconomic information contains the driving force of long-term stock market volatility. At the same time, the short-term volatility of China's stock market can be well explained by investor sentiment. Second. Through the combination of web crawler and text mining, it is a good way to construct emotional index, which greatly broadens the channel of information collection. Emotion dictionary plays an irreplaceable role in the construction of emotion index. The final construction of investor sentiment index is more representative, especially involving the professional domain of emotional analysis, need related to the field of rich corpus and scientific and reasonable emotional index construction method. 4th. The GARCH-MIDAS model constructed by combining GARCH model with midas (mixed frequency data sampling) model can effectively extend the depth of the model research. At the same time, it can mine the information of mixing data to the maximum extent, and the application in the research of volatility component model of stock market in our country is effective and applicable.
【学位授予单位】：山西财经大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F832.51

【参考文献】