基于ARMA-GARCH模型的国债指数实证研究

发布时间：2018-05-05 17:26

本文选题：ARMA-GARCH模型 + 国债指数　；参考：《合肥工业大学》2012年硕士论文

【摘要】：随着欧债危机的爆发，很多经济学家和计量统计学家开始将研究的目光转向国债市场。国债，是一个国家信用的体现，具有彰显国家公信力的价值。那么，使用计量统计学工具对于国债市场的走势进行量化的数据分析，并进行预测，以达到风险控制的目的，将会是非常有意义的课题。本文主要是使用ARCH类模型族研究我国的国债指数收益率序列，通过研究我们发现，我国的国债指数收益率序列由于其数据本身联动的惯性和相对于经济、政策的滞后性，，会呈现出一定的序列相关，所以本文采用ARMA模型来描述国债收益率序列，GARCH模型用来拟合误差。结合考虑金融数据本身所具有的尖峰厚尾及异方差性，以及投资者对于投资风险的承受能力所体现出的有偏性，本文对其收益率序列建立误差分布为正态，GED和t分布的ARMA-GARCH模型，这样更符合实际，研究效果更好。随后，本文尝试使用交叉验证的方法用ARMA-GARCH模型对国债指数的走势进行预测，与已有研究相比，也取得了较好的研究结果。本文的全文内容将分成以下五个部分：第一章对本文的研究背景, ARCH类模型国内外的研究情况进行简要概述，并提出本文的研究目的和分析方法。第二章对涉及到的ARCH类模型的结构进行简要概述，其中包括对基于不同误差分布下ARMA-GARCH模型的介绍。第三章利用EViews5.1软件对上证国债指数日收益率序列的尖峰厚尾、有偏性、自相关性、平稳性、和异方差性等特性进行基本统计分析，以确定合适的分析模型。第四章依据AIC准则进行参数估计以确定ARMA(p,q)-GARCH(r,s)的参数，从而可以选取合理的ARCH类模型对国债收益率序列进行拟合，同时使用交叉验证的方法对国债收益率序列进行预测，最后对基于不同误差分布下模型的拟合与预测效果进行比较，找出相对更适合于进行国债指数研究的模型。第五章研究的结论。
[Abstract]:With the outbreak of the European debt crisis, many economists and statisticians are turning their attention to the bond market. National debt, is the embodiment of a national credit, with the value of highlighting the credibility of the country. Therefore, it will be a very meaningful subject to use econometric statistical tools to analyze and forecast the trend of the national debt market to achieve the purpose of risk control. In this paper, we mainly use the ARCH model family to study the national debt index yield series of our country. Through the study, we find that the series of the national debt index yield rate of our country is lagging behind because of the inertia of the data itself and the lag of the policy relative to the economy. So the ARMA model is used to describe the bond yield series and the GARCH model is used to fit the error. Considering the sharp and thick tail and heteroscedasticity of financial data and the bias of investors' ability to bear investment risk, this paper establishes a ARMA-GARCH model with the error distribution of normal GED and t distribution for the return series. This is more in line with the reality, the research results are better. Subsequently, this paper attempts to use cross-validation method to predict the trend of treasury bond index by using ARMA-GARCH model, and better results are obtained compared with the previous research. The full text of this article will be divided into the following five parts: In the first chapter, the research background of this paper, the research situation of ARCH class model at home and abroad are briefly summarized, and the purpose and analysis method of this paper are put forward. Chapter 2 gives a brief overview of the structure of the ARCH class model, including the introduction of the ARMA-GARCH model based on different error distributions. In chapter 3, we use EViews5.1 software to analyze the characteristics of the sharp, thick tail, bias, autocorrelation, stationarity and heteroscedasticity of the daily yield series of Shanghai Treasury bond index, so as to determine the appropriate analysis model. In chapter 4, we estimate the parameters according to the AIC criterion to determine the parameters of ARMA-PQ / GARCHRN), so that we can select a reasonable ARCH model to fit the yield series of treasury bonds, and use the method of cross-validation to predict the yield series of treasury bonds, at the same time, we can use the method of cross-validation to predict the yield series of treasury bonds. Finally, the fitting and forecasting results of the model based on different error distribution are compared, and the model which is more suitable for the research of national debt index is found out. The fifth chapter is the conclusion of the research.
【学位授予单位】：合肥工业大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F224;F812.5

【参考文献】