基于非参数方法对人民币汇率与利率动态关系的研究

发布时间：2018-01-03 08:33

本文关键词：基于非参数方法对人民币汇率与利率动态关系的研究　出处：《天津财经大学》2016年硕士论文　论文类型：学位论文

【摘要】：我国当前正处于经济结构转变的关键时期,经济数据的结构在不断地发生着变化,分析经济数据的方法和所建的模型也逐步由简单发展到复杂。参数建模的方法都依赖于对模型的基本假设,回归参数的优良性质也是建立在回归残差分布是已知的条件下,因此事先设定模型形式和提供假设条件给参数模型的应用带来了一定的局限性。非参数模型在很大程度上放宽了经典计量经济模型的假设条件,使得模型的表现形式更加灵活,并且更加真实地表现数据的生成过程和更好地体现经济运行规律。因此,对非参数模型及其估计方法进行研究有着重要的理论和现实价值。2015年10月30日人民币加入SDR(特别提款权)货币篮子,这将加强我国金融市场与世界金融市场的接轨程度,同时也为非参数方法的应用提供了更多的机会。文中首先研究了窗宽和核函数,接着对不同的非参数估计方法进行了研究和对比分析,又从非参数时间序列模型形式的确定、估计方法的选择以及模型预测等方面进行了研究。实证分析部分首先运用参数方法对汇率与利率的关系建模,再利非参数方法对汇率和利率的动态关系做进一步分析,分别得出了汇率、利率密度函数的非参数估计和建立了汇率、利率的非参数时间序列模型,并在此基础上建立了汇率与利率动态关系的非参数模型。文章通过对非参数模型及其估计方法的研究得出了以下结论:第一,对常用非参数估计方法的特点进行了研究和对比分析,其中估计方法包括核估计、局部线性估计、k近邻估计、局部多项式估计等。文中探讨了各种估计方法的适用情况,为他人合理地使用提供了便利。第二,建立了人民币汇率与利率动态关系的非参数模型,旨在从变量的数量关系方面进行研究,因而抛开了更多定性因素的影响得出了变量在数据层面的相互影响关系。
[Abstract]:At present, our country is in the key period of economic structure transformation, the structure of economic data is constantly changing. The methods of analyzing economic data and the models built have gradually developed from simple to complex. The methods of parameter modeling all depend on the basic assumptions of the model. The excellent properties of regression parameters are also based on the condition that the regression residual distribution is known. Therefore, the application of the parametric model is limited by pre-setting the model form and providing the hypothetical conditions. The non-parametric model greatly relaxes the assumptions of the classical econometric model. Make the model more flexible, and more true to represent the process of data generation and better reflect the rules of economic operation. It is of great theoretical and practical value to study the non-parametric model and its estimation methods. In October 30th 2015, RMB joined the SDR (Special drawing Rights) basket. This will strengthen the connection between our financial market and the world financial market, and at the same time provide more opportunities for the application of non-parametric methods. In this paper, we first study the window width and kernel function. Then different nonparametric estimation methods are studied and compared, and the non-parametric time series model is determined. In the part of empirical analysis, the relationship between exchange rate and interest rate is modeled by parametric method. Then the non-parametric method is used to analyze the dynamic relationship between exchange rate and interest rate. The non-parametric estimation of exchange rate and interest rate density function and the non-parametric time series model of exchange rate and interest rate are obtained respectively. On this basis, the non-parametric model of the dynamic relationship between exchange rate and interest rate is established. Through the study of the non-parametric model and its estimation methods, the following conclusions are reached: first. The characteristics of common nonparametric estimation methods are studied and compared. The estimation methods include kernel estimation, local linear estimation and k-nearest neighbor estimation. Local polynomial estimation and so on. This paper discusses the application of various estimation methods, which provides convenience for others to use reasonably. Secondly, the non-parametric model of the dynamic relationship between RMB exchange rate and interest rate is established. The purpose of this paper is to study the quantitative relationship of variables, so the interaction of variables at the data level is obtained by setting aside the influence of more qualitative factors.
【学位授予单位】：天津财经大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：F224;F832.6;F822.0

【相似文献】