中国国债收益率曲线研究

【作者】 陈震；

【导师】 干杏娣；

【作者基本信息】 复旦大学 ， 金融学， 2009， 博士

【摘要】 本文以中国国债收益率曲线为研究对象,主要回答两大问题:一是如何构建国债收益率曲线;二是国债收益率曲线的形态和变化规律如何。国内对利率期限结构的研究还比较薄弱,作者选择利率期限结构作为研究对象,具有一定的理论和现实意义。国外利率期限结构模型大致可以分为理论经济模型和数量模型,前者经济假设较多,要求较为严格,多用于理论研究。后者利用市场债券价格信息,通过贴现函数或者插值的方法获得利率期限结构数据,比较适合构建收益率曲线。通过利率期限结构的数量模型构建国债收益率曲线有几个关键因素需要把握好:一是数据问题,我国债券交易和报价中失真数据较多,需要通过零波动率利差法、主观判断法、相对位置法、最优报价法等方法进行处理,针对数据数量不足的问题,也可以通过增加交易日、人工补点等方法加以改善。二是模型要素设置问题,通过实证分析比较,作者认为样条模型采用3年、7年为节点,分成三段更加合适,采用复合加权可以兼顾长短,效果较好;NSS模型采用复合加权较好;Hermit模型中关键期限点收益率挑选要慎重。本文对Hermite法和样条模型在收益率曲线估计方面进行了综合比较分析,发现Hermite法比较适合银行间市场,多项式样条模型比较适合交易所市场,鉴于两者的估计结果差异不是很大,且银行间市场在我国债券市场中居于主导地位,建议统一采用Hermite插值法作为全市场国债收益率曲线的构建模型。中央国债登记结算公司编制和发布的中债收益率曲线具有很强的权威性、完整性和真实性,可以直接作为国债收益率数据分析使用。对该数据的分析发现,当前我国国债收益率曲线总体呈现完全向上倾斜的形态,历史上(2002年-2008年)也基本如此,虽然这一现象用预期理论、市场分割理论和流动性偏好理论均可以进行解释,但相对而言,流动性偏好理论的解释较为合理。与发达国家相比,当前我国国债收益率曲线无论是绝对水平还是形状都比较正常。分析2002年-2008年我国国债收益率曲线的变动情况发现:国债利差一般与国债收益率水平成正比,但国债收益率下行或上移的原因及非市场因素也是决定国债收益率曲线形变的重要因素。影响国债收益率曲线变动的原因主要有基本面、资金面和政策面三个方面,CPI、银行间7日回购利率以及银行1年定期存款利率是代表性指标,虽然三个指标都与国债收益率正相关,但物价对国债收益率的影响最为显著,定性观察的历史走势也最为吻合。将三方面因素综合来看,国债收益率上升通常只需三方面因素中的一种推动,国债收益率下降却需三方面因素中的所有都不构成阻力。更多还原

【Abstract】 This thesis focus on the T-bond yield and answers two questions: how to establish the T-bond yield curve and what is the shape and the rule of the yield curve. Since the research on the term structure of interest rate is still relatively weak in China, it is significative in theory and practice for this thesis choosing such a topic.The models about term structure of interest rate include theoretic models and quantitative models. The theoretic models have much strict hypothesis which makes them more adapt to theoretic research. The quantitative models get data through discount function or insert-value methods which make them adapt to establish yield curve.During establishing yield cure through quantitative models, we must pay more attention to some key factors. The first one is data. Since there are many abnormal tradings and distortion data in bond markets, it needs to adopt some scientific and rational methods to manage the raw data before using it. The other one is key hypothesis of the model. Through demonstration research the author finds it better to divide the yield curve into three parts with 3 and 7 year as the key nodes in spline model, adopt composite weighting in NSS models, and be cautious to choose key term point in Hermit model.This thesis analyzes the difference betweent Hermite model and spline model in establishing yield curve, which shows that the Hermite model is approrite to the inter-bank market while the spline modet fits the stock change much. Since the difference of results is small and the inter-bank market is dominant in Chinese bond markets, the author propose to adopt Hermite model to estable unitive yield curve.The CDC curve is esblished and issued by China Government Securities Depository Trust & Clearing Co. Ltd.,(CDC).The authority and authenticity of the CDC curve make it ideal data for yield analysis.Throught analyzing the CDC cureve we find the T-bond yiled curve uptilting now and always in history. Although this phenomenon can be explained by Expectation Hypothesis, Market Segmentation Theory and Liquidity Preference Theory, the Liquidity Preference Theory is most reasonable after all. Comparing with the yield cure of the western developed countries, the level and shape of the T-bond yield of China is normal. By analyzing the change of the T-bond yield curve during 2002-2008, the author finds that yield spread of the T-bond is positive correlative to the yield itself. The main factors pushing the yield curve include macro economy, liquidity and policy. CPI, 7-day repo rate in the inter-bank and 1-year deposite rate are three reprentive indexes which have important influences on yield curve, with CPI having the most one. In a general way, the T-bond yield will rise when any of the three indexes rise, while the T-bond yield will drop when all of the three indexes drop.更多还原