VaR基于分位数的风险管理问题研究

摘要VaR作为一种金融风险度量的重要工具，其计算方法研究异常丰富，可划分为参数法、非参数法以及半参数法三类。参数法是 VaR 计算中最为常用的方法，其中，基于 GARCH 族模型计算 VaR 则是目前的主流，然而参数法需要通过给定分布的假设来预测条件波动率，从而影响了对实际分布的描述，最终可能导致 VaR 计算结果不够准确。因此，本文引入半参数方法下的分位数回归技术来建立 VaR 模型,基于分位数回归方法不需要假设分布等优点进行实证研究,并将其结果与传统参数法进行比较。76136

本文选取 2010 年 1 月 4 日至 2016 年 4 月 29 日的恒生指数日收盘价作为研究对象，分别运用 t 分布下的 GARCH（1,1,）模型以及两个分位数回归模型来计算 VaR ，并对计算结果进行 Kupiec 似然比检验，从而对比传统 GARCH 模型方法和分位数回归方法在计算 VaR 上的优劣势。从两类模型的检验情况来看,在 95%的置信度下，无论是在失败率还是 LR 统计量上,分位数回归模型都表现良好,充分展示了分位数回归模型在 VaR 计算方面的优越性。

毕业论文关键词 VaR 分位数回归 GARCH 模型风险度量

毕业设计说明书外文摘要

Title Risk Measurement Based on Quantile Regression Model

Abstract VaR is an important tool for the measurement of financial risk。 The calculation methods of VaR are perse, and are pided into three groups: parametric methods, non-parametric methods and semi-parametric methods。 Parametric methods are the most commonly used method for calculating VaR。 As a parametric method, using GARCH models to calculate VaR is currently the main trend。 However, for parametric methods, given distribution needs assuming to predict conditional volatility。 Therefore, the description of the actual distribution is affected, leading to the inaccurate calculation of VaR。 In this paper, quantile regression technique of semi-parametric method was introduced to establish VaR models。 Empirical study was conducted and the results of quantile regression technique, which does not require the assumption of distribution, and those of parametric method, were compared。

The daily closing price of Hang Seng Index from January 4, 2010 to April 29, 2016 were taken as the research object, and one GARCH (1,1) model of t- distribution and two quantile regression models were used to calculate VaR。 Kupiec likelihood ratio test was conducted on calculation results in order to compare traditional GARCH method and quantile regression method in terms of calculation of VaR。 As was demonstrated by the test, within the confidence interval of 95%, whether in failure rate or in LR statistic, quantile regression model stood out, thus fully showing superiority of it in terms of calculation of VaR。

Keywords VaR quantile regression GARCH models risk measurement

本科毕业设计第 I 页

1 绪论 1

1。1 研究背景及意义 1

1。2 国内 VaR 及分位数回归研究综述 2

1。3 文的研究内容及框架 3

2 VaR 及分位数回归理论 5

2。1 VaR 值计算的基本原理 5

2。2 VaR 值计算的主要方法 5

2。3 分位数回归理论 7

3 基于 GARCH 模型的 VaR 实证分析 10

3。1 数据基本分析 10

3。2 建立 GARCH 模型 13

3。3 VaR 计算结果 13

4 基于分位数回归方法的 VaR 实证分析 15 VaR基于分位数的风险管理问题研究:http://www.youerw.com/shuxue/lunwen_87259.html