A new nonparametric estimation method of the variance in a heteroscedastic model

Abstract

In most economic phenomena, the assumption of homoscedasticity in the classic linear regression model is not necessarily true, which leads to heteroscedasticity. The heteroscedastic estimate is an important aspect for the problem of heteroscedasticity. For this hot issue, this paper proposes a nonparametric estimation method with simple calculation for the estimation of heteroscedasticity through orthogonal arrays, which does not rely on the distribution of data. The performance of the proposed method is investigated by prediction error in real data sets and simulations. The results suggest that this method offers substantial improvements over the existing tests.