A CONSISTENT MODEL SPECIFICATION TEST FOR A REGRESSION FUNCTION BASED ON NONPARAMETRIC WAVELET ESTIMATION
提出一种基于小波估计的回归函数设定检验方法，能有效检验参数回归模型是否正确，蒙特卡洛模拟显示其在小样本下表现优于传统核方法。
In this paper we present a consistent specification test of a parametric regression function against a general nonparametric alternative. The proposed test is based on wavelet estimation and it is shown to have similar rates of convergence to the more commonly used kernel based tests. Monte Carlo simulations show that this test statistic has adequate size and high power and that it compares favorably with its kernel based counterparts in small samples.