平滑转移误差修正模型的转移函数选取问题研究  

Research on the Transition Function Selection of the Nonlinear Smooth Transition Error Correction Model

作　　者：南士敬[1] 赵春艳[1] 

机构地区：[1]西安交通大学经济与金融学院

出　　处：《数量经济技术经济研究》2015年第2期144-160,共17页Journal of Quantitative & Technological Economics

基　　金：国家社科基金项目"非平稳条件下平滑转换自回归模型的线性检验问题研究"(11CTJ002)的资助

摘　　要：针对非线性平滑转移误差修正模型转移函数选取中存在的统计量极限分布非标准、检验统计量功效较低的问题,本文在推导非线性平滑转移协整检验统计量极限分布的基础上构造了如下转移函数选取步骤。首先,计算F_(NST)统计量,进行非线性平滑转移协整检验;其次,计算t_(EST)和t_(LST)统计量及相依概率P_(est)和P_(lst);最后,比较P_(est)和P_(lst)大小并与临界值相比,得出结论。蒙特卡洛仿真模拟结果显示,转移函数选取中各统计量具有良好的功效和势,且转移函数选取中各统计量的功效明显优于其他统计量的功效。实证分析表明我国利率期限结构具有明显的非线性对称调整效应,非线性平滑转移误差修正模型中转移函数应该选取指数函数。Considering the present existing deficiencies in selecting transition function of the nonlinear smooth transition model such as the nonstandard asymptotic statistics and low- er power of the traditional statistics, the paper proposes the following transition function se- lection steps based on deriving the asymptotic distribution of the nonlinear cointegration test statistics. Firstly, calculate FNST statistics to run nonlinear cointegration test. Secondly, de- rive troT and tEST statistics and their dependent probability Pat and p^t. Finally, compare Put with P^t on the condition of both smaller than the critical value 0. 05 and make conclu- sion. The Monte Carlo simulation results show that the statistics of the transition function selection steps have good power and size, and also the power of our proposed statistics is better than that of other traditional statistics. The last empirical results show that Chinese term structure of interest rates has obvious nonlinear symmetric adjustment effects, the transition function of the nonlinear smooth transition error correction model should be the exponential function.

关 键 词：平滑转移误差修正模型 转移函数 功效 检验水平 利率期限结构 

分 类 号：F224.0[经济管理—国民经济]