非参数回归N-W核估计法在财政系统中的应用

The Application of N-W Kernel Estimation of Non-parametric Regression in the Finance System

选取1978-2013年的财政收入和经济增长数据为样本,利用参数回归分析和非参数回归N-W核估计法进行实证分析,比较发现GDP增长率对财政收入增长率的影响在经济发展的不同时期有不同的特点,并非简单的线性模型能解释清楚,而且非参数回归模型的拟合效果优于线性回归模型模拟效果;并采用最小平方交叉实证法(CV)和经验选择方法,从拟合效果中综合考虑其优劣,进而给出最佳窗宽。

Taking the fiscal revenue and economic growth data from 1978 to 2013 as the sample, the paper gives an empirical analysis through parameter regression and N-W kernel estimation of nonparametric regression. The result reveals the effects of GDP growth on revenue growth have different characteristics in different periods of economic development, which cannot be explained with a simple linear model. What＇s more, the simulation result of non-parametric regression is better than the linear regression model. Finally, the best window width is given by using the least square cross-validation （CV） and experience selection method.