基金投资风险的实证研究——基于GARCH_VaR模型

An Empirical Study on Fund Investment Risk—Based on GARCH_VaR Model

摘要: 随着金融市场的迅速发展，基金行业市场规模不断增长，截至2024年底，公募基金管理总规模已突破32万亿元。准确有效测量基金的投资风险，对于基金市场的稳健发展以及基金投资者进行合理的资产配置具有重要作用。本文选取了安信价值精选股票(000577.OF)、华夏中证500ETF联接A (001052.OF)、国泰安康定期支付混合A (000367.OF)等不同投资类型和规模的9支样本基金自2017年11月至2022年2月的单位净值数据，利用VaR值来衡量基金风险。根据基金收益率序列“尖峰后尾”、“波动聚集”的特性，建立了GARCH_VaR模型，在正态分布和可以调整尾部参数的t分布、GED分布三种假设条件下，分别计算出各只基金的日VaR值，并应用Kupiec失败率检法对计算出的VaR值的进行检验，据此对不同分布假设条件下的模型进行评价，结果显示不同类型的基金风险差异较大，GED分布假设条件下的模型VaR估计更为准确，更能准确反映基金风险。

Abstract: With the rapid development of the financial market, the market size of the fund industry continues to grow. As of the end of 2024, the total managed size of public funds has exceeded 32 trillion yuan. Accurately and effectively measuring the investment risk of funds plays an important role in the stable development of the fund market and the rational asset allocation of fund investors. This article selects the unit net asset value data of 9 sample funds with different investment types and scales, including Anxin Value Selected Stock (000577.OF), Huaxia CSI 500 ETF Connect A (001052.OF), and Guotai Ankang Regular Payment Hybrid A (000367.OF), from November 2017 to February 2022, and uses VaR value to measure fund risk. Based on the characteristics of “peak after tail” and “volatility aggregation” in fund return sequences, a GARCH_VaR model was established. Under three assumptions: normal distribution, t-distribution with adjustable tail parameters, and GED distribution, the daily VaR values of each fund were calculated, and the Kupec failure rate test was applied to test the calculated VaR values. Based on this, the models under different distribution assumptions were evaluated. The results showed that there were significant differences in risk among different types of funds, and the VaR estimation of the model under the GED distribution assumption was more accurate and could better reflect fund risk.