中国封闭式基金折价的高频因素分析

发布时间：2018-06-29 00:14

本文选题：封闭式基金 + 折价　；参考：《复旦大学》2013年硕士论文

【摘要】：封闭式基金由于基金净值和二级市场交易价格的不一致,出现溢价或者折价的现象,主要是折价,这种现象被称为“封闭式基金折价之谜”,到目前为止,无论是传统金融视角还是行为金融的视角,都没有得出完善的理论来充分解释封闭式基金折价现象。本文基于偏最小二乘回归的方法,对影响封闭式基金折价的高频数据进行实证分析,考虑到变量的自相关性,加入了所有变量的滞后项,进而对封闭式基金折价现象进行解释。实证结果显示,在所选择的变量中,封闭式基金折价的滞后项具有最强的解释力,显示封闭式基金折价具有很强的自相关性。在所提取的成分中,由封闭式基金折价的滞后项、利率、成交额、市场指数构成的市场景气成分对封闭式基金折价的解释力最强,并且与因变量成正相关；由封闭式基金折价的滞后项、市场指数、指数波动率、成交额构成的市场萧条成分,与折价成负相关。最终,模型提取了自变量71.3%的信息,对因变量具有97.3%的解释力。在此基础上,本文基于拟合效果将数据分为两个阶段,并分别基于偏最小二乘回归进行建模分析,最终的拟合效果都有不同程度的提高。
[Abstract]:Because of the inconsistency between the net value of the fund and the trading price of the secondary market, the closed-end fund has a phenomenon of premium or discount, which is mainly a discount. This phenomenon is called "the mystery of the discount of the closed-end fund". So far, Neither the traditional financial perspective nor the behavioral finance perspective has come to a perfect theory to fully explain the closed-end fund discount phenomenon. Based on the method of partial least square regression, this paper makes an empirical analysis of the high frequency data which affects the discount of closed-end funds. Considering the autocorrelation of variables, the lagging terms of all variables are added, and then the phenomenon of closed-end fund discount is explained. The empirical results show that among the selected variables, the lag term of closed-end fund discount has the strongest explanatory power, indicating that closed-end fund discount has a strong autocorrelation. Among the extracted components, the market boom component, which consists of the backward items of closed-end fund discount, interest rate, turnover and market index, has the strongest explanatory power to the closed-end fund discount, and is positively related to the dependent variable. There is a negative correlation between the market depression component of closed-end fund discount, market index, index volatility and turnover. Finally, the model extracts 71.3% information of independent variables and has 97.3% explanatory power to dependent variables. On this basis, this paper divides the data into two stages based on the fitting effect, and models and analyzes the data based on partial least square regression, and the final fitting effect is improved to some extent.
【学位授予单位】：复旦大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F832.51

【相似文献】