摘要
利用2001-01 ̄2005-11我国城镇居民人均可支配收入与消费性支出的月度数据,建立了我国城镇居民消费的非参数模型,并用不变窗宽的核估计、不变窗宽的局部线性估计、k-近邻估计、正交序列估计以及线性最小二乘估计分别进行了拟合和预测,结果表明:非线性模型优于线性模型;在4种非线性估计方法中,局部线性估计方法优于其它3种估计方法.
The nonparametric model of consumption functions of China's urban residents was established by using the monthly data of the controllable income and consumption expenditure in 2001-01--2005-11, and the kernel estimation with fixed bandwidth, the local linear estimation with fixed bandwidth, the k-neighbour estimation, the series estimation and OLS estimation methods were used to estimate the regressive function respectively. The result showed that the local linear estimation method is better than any other method and the nonlinear model is better than the linear model.
出处
《海南师范学院学报(自然科学版)》
2006年第3期214-218,238,共6页
Journal of Hainan Normal University:Natural Science
关键词
城镇居民消费
非参数模型
局部线性估计
窗宽
urban residents' consumption
non-parametric model
the local linear estimation
bandwidth