1: \begin{abstract}

2: Nonparametric regression models with locally stationary covariates have received increasing interest

3: in recent years. As a nice relief of  ``curse of dimensionality'' induced by large dimension of covariates, additive regression model is  commonly used. However, in locally stationary context, to catch the dynamic nature of regression function, we adopt a flexible varying-coefficient additive model where the regression function has the form

4: $\alpha_{0}\left(u\right)+\sum_{k=1}^{p}\alpha_{k}\left(u\right)\beta_{k}\left(x_{k}\right).$

5: For this model, we propose a three-step spline estimation method for each univariate nonparametric function,

6: and show its consistency and $L_{2}$ rate of convergence.

7: Furthermore, based upon the three-step estimators, we develop a two-stage penalty procedure to identify pure additive terms

8: and varying-coefficient terms in varying-coefficient additive model.

9: As expected, we demonstrate that the proposed identification procedure is consistent, and the penalized estimators achieve the same $L_{2}$ rate of convergence as the polynomial spline estimators.

10: Simulation studies are presented to illustrate the finite sample performance of the proposed three-step spline estimation method and two-stage model selection procedure.

11: %In addition to, two real data are investigated

12: %details to demonstrate the usefulness of the proposed model in practice.

13: 

14: \end{abstract}

15: