The Research On Iterated Local Polynomial Yield Curve Model

With the development of modern financial market and the strengthening of macro control, the yield curve of bond receives more and more attention. Using the model to estimate yield curve of bond has important meanings:on one hand, it can price the financial products accurately; On the other hand, it can help the government to grasp the macroeconomic precisely. But compared with the developed countries, Chinese bond market develops later, imperfect; a large amount of bonds are mostly coupon bonds; bonds samples have many abnormal points. All above lead to the effect of foreign advanced yield curve estimation model of application in China is not so good. At present, domestic researches on yield curve estimates are mostly based on mending models abroad and still haven’t put forward estimation model according to China’s market characteristic.According to the problems mentioned above, on the basis of the features of bond market of China and Local Polynomial Model, the dissertation puts forward a series of new yield curve models:Static Estimation Model, Dynamic Estimation Model and Macro Financial Model. These models can not only estimate yield curve of bond accurately, and also have strong prediction ability, more important, they have excellent statistical properties and realistic meaning. It is good for the stability and development of financial market in our country.The research of this dissertation is on the clue of "Static Estimation Model→Dynamic Estimation Model→Macro Financial Model" step by step:firstly, based on the current situation bond market in China and existing static model, the author puts forward the Iterated Local Polynomial Static Estimation Model which is suitable for Chinese national conditions; secondly, under the framework of Dynamic NS Model, the Static Estimation Model will be promoted. By this extension, it can not only increase the depicted effect of the trend of yield curve, more important, also enhance the prediction ability; At last, by combining the Dynamic Model and macro financial variables, the writer puts forward Iterated Local Polynomial Macro Financial Model. By using the information of macro financial variables and yield curve, the prediction ability of model will be advanced. The contents are as follows:From the aspect of Static Estimations Model, the dissertation analyses the actual situation of bond market in our country and existed static estimation models. On the basis of this, through improving Local Polynomial Estimation, which is a new development in Statistics, the dissertation puts forward Iterated Local Polynomial Static Estimation Models. At the same time, the author using theoretical method, Monte Carlo simulation and empirical method to study the estimation effect and statistical property of new method. The result shows that the Iterated Local Polynomial model not only gives better estimation effect, more important, it has good statistical properties, and can be widely used in the practical application.From the aspect of Dynamic Estimation Model, the dissertation analyzes the Dyn’amic Estimation Model, especially Dynamic NS Model which is put forward recently. The author finds that although the traditional dynamic estimation method can depict the regulation of yield curve accurately, it has poor performance on the prediction of future yield curve. Dynamic NS Model improves the dynamic prediction ability effectively by two steps estimation. But in the first step of two steps estimation, Dynamic NS Model uses bootstrap static estimation model. The application of the model goes not well in Chinese bond market. Thus, with the help of framework of Dynamic NS Model and two steps estimation, the dissertation puts forward Iterated Local Polynomial Dynamic NS Model. The empirical analysis shows that Iterated Local Polynomial Dynamic NS Model not only improves the estimation ability of former Dynamic NS Model, but also improves the prediction ability of future yield curve greatly. At the same time, this paper does a research on three key dynamic factors of NS dynamic model. Through the theoretical analysis, it shows that the three factors can not only indicate term features of the yield curve (short-term, medium-term and long-term), but also mean that the shape characteristics of the yield curve (level, the slope and curve). By using empirical analysis, it indicates that correlation between three dynamic factor and deadlines features is weak, but correlation between shape characteristic is strong. It shows that three dynamic factors depict the level, the slope and curve characteristics respectively to yield curve fitting through yield curve.From the aspect of Macro Financial Model, the dissertation combines the Iterated Local Polynomial Dynamic NS Model and macro financial factors, and puts forward raised Iterated Local Polynomial Macro Financial Model. Using this model, the author analyzes the influence on three dynamic factors from macro economic variables attack. Through empirical analysis, it shows that the impacts of macroeconomic variables attack to three dynamic factors are obviously effective; by adding macro economic variables, the Macro Financial Model can make full use of information on macroeconomic and yield curve to improve the prediction ability of the model.Based on theoretical investigation and empirical research, the paper has got some innovation achievements in the following aspects:1. Putting forward the Iterated Local Polynomial Static Estimation Model and studying the statistical property. Based on analyzing the reality of Chinese bond market characteristics and existing yield curve of static estimation method, the Iterated Local Polynomial Static Estimation model which is suitable for Chinese national conditions has been raised. Through the Monte Carlo simulation and empirical analysis, it shows the new model not only has better estimated effect, but also has better statistical property.2. Putting forward Iterated Local Polynomial Dynamic NS Model and analyzing the realistic meaning of three dynamic factors. Under the framework of Dynamic NS Model and two steps estimation, Iterated Local Polynomial Dynamic NS Model is raised. This model not only has better fitting effect, more important, it also has prediction ability. At the same time, the realistic meaning of the three dynamic factors is analyzed. The results show that the three dynamic factors indicate the term characteristics of yield curve and shape features respectively.3. Putting forward Iterated Local Polynomial Macro Financial Model and analyzing the impacts of macroeconomic variables attack to three dynamic factors. Based on combing the Iterated Local Polynomial Dynamic NS Model and macro financial variables, the Iterated Local Polynomial Macro Financial model is raised. The research shows that macro economic variables have obvious effect on dynamic factors. By adding macro factors, the Macro Financial model has better prediction effect. Although the research has made some useful results, due to theoretical limitations and practical economic data, the thesis is lack of research on universality and depth of model application. The writer hopes that with the development of theory and the rich of economic data, the analysis will be improved and perfected in the follow-up study.