General rank-based estimation for regression single index models (Q1786907)

The authors consider a single index regression model and overcome deficiencies (in efficiency and robustness) of earlier estimators by a general rank-based estimation approach, which is based on minimizing a rank-based objective function of the residuals. They investigate the asymptotic properties of the estimators, especially consistency and asymptotic normality. To demonstrate the appeal of the rank estimators for dealing with heavy-tailed or contaminated error distributions, a simulation study is provided. A real-life example illustrates that the suggested procedure works well even in the presence of outliers.