考虑GARCH效应的动态无套利Nelson-Siegel模型及国债管理策略分析

发布时间：2018-01-04 11:35

本文关键词：考虑GARCH效应的动态无套利Nelson-Siegel模型及国债管理策略分析　出处：《厦门大学》2014年硕士论文　论文类型：学位论文

【摘要】：利率期限结构不仅是一国宏观经济运行中的一个重要指标,在金融市场中影响各类金融产品的定价,在微观层面调节投融资决策,在世界经济中基于购买力平价对各国之间汇率变化起到基准调节作用。党的十八届三中全会决议更是明确提出“加快推进利率市场化,健全反映市场供求关系的国债收益率曲线”,并将其纳入国家核心发展战略。因此,学术界和实务界一直以来都在寻找能够更好地拟合和预测利率演化过程的期限结构模型。本文在广泛使用的传统Nelson-Siegel模型和具有坚实经济基础的无套利Nelson-Siegel基础上,考虑不同时点、不同期限可能出现的条件异方差因素,得到考虑GARCH的动态无套利Nelson-Siegel模型。 G-AFNS模型仍旧采用NS族模型的水平因子、斜率因子和曲率因子作为三个状态变量,在对2005年1月至2012年11月中国交易所国债市场上的利率期限结构进行研究之后,发现G-AFNS模型样本内拟合程度较高,在进行利率预测时,预测能力明显优于传统的DNS模型和无套利AFNS模型,说明考虑条件异方差之后预测能力显著提高,G-AFNS模型适合应用于我国国债市场上。接下来在国债管理策略分析方面,主要侧重债券免疫组合的构造,与DNS向量久期一致,通过水平因子久期、斜率因子久期和曲率因子久期的完全匹配,来消除利率期限结构变化给债券组合价值带来的利率风险敝口。本文分别考察了每隔1个月、2个月和3个月进行免疫组合调整,结果认为在中国交易所国债市场上,如果利率期限结构不出现突变,每2个月进行一次组合调整来保证免疫效果,应该是较为科学合理的频率。
[Abstract]:Term structure of interest rate is not only an important index in the macroeconomic operation of a country, but also influences the pricing of all kinds of financial products in the financial market and adjusts the investment and financing decisions at the micro level. In the world economy, based on purchasing power parity, the exchange rate changes between countries play a benchmark role. The resolution of the third Plenary session of the 18 CPC Central Committee specifically proposed "accelerating the promotion of interest rate marketization." Improve the bond yield curve that reflects the relationship between supply and demand in the market "and incorporate it into the national core development strategy. The academic and practical circles have been looking for a term structure model which can better fit and predict the evolution of interest rate. In this paper, based on the widely used traditional Nelson-Siegel model and the non-arbitrage Nelson-Siegel with solid economic foundation, different points are considered. Considering the conditional heteroscedasticity factors of different periods, a dynamic arbitrage free Nelson-Siegel model considering GARCH is obtained. G-AFNS model still uses the horizontal factor, slope factor and curvature factor of NS family model as three state variables. After studying the term structure of interest rate in the treasury bond market of China Stock Exchange from January 2005 to November 2012, it is found that the G-AFNS model has a higher fitting degree in the sample. In the forecasting of interest rate, the forecasting ability is obviously superior to the traditional DNS model and the AFNS model without arbitrage, which shows that the forecasting ability is improved significantly after considering conditional heteroscedasticity. G-AFNS model is suitable to be applied to the national debt market of our country. Then, in the analysis of treasury bond management strategy, we focus on the construction of bond immune combination, which is consistent with DNS vector duration, through the complete matching of horizontal factor duration, slope factor duration and curvature factor duration. In order to eliminate the interest rate risk brought by the change of term structure of interest rate on the value of bond portfolio, this paper investigates the immune portfolio adjustment every 1 month, 2 months and 3 months, respectively. The results show that if the term structure of interest rate does not change every two months to ensure the immune effect, it should be a more scientific and reasonable frequency.
【学位授予单位】：厦门大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51;F224

【参考文献】