1: \begin{abstract}

2: The complexity of semiparametric models poses new challenges to

3: statistical inference and model selection that frequently arise

4: from real applications. In this work, we propose new estimation

5: and variable selection procedures for the semiparametric

6: varying-coefficient partially linear model. We first study

7: quantile regression estimates for the nonparametric

8: varying-coefficient functions and the parametric regression

9: coefficients. To achieve nice efficiency properties, we further

10: develop a semiparametric composite quantile regression procedure.

11: We establish the asymptotic normality of proposed estimators for

12: both the parametric and nonparametric parts and show that the

13: estimators achieve the best convergence rate. Moreover, we show

14: that the proposed method is much more efficient than the

15: least-squares-based method for many non-normal errors and that it only

16: loses a small amount of efficiency for normal errors. In addition, it is

17: shown that the loss in efficiency is at most $11.1\%$ for

18: estimating varying coefficient functions and is no greater than

19: $13.6\%$ for estimating parametric components. To achieve sparsity

20: with high-dimensional covariates, we propose adaptive penalization

21: methods for variable selection in the semiparametric

22: varying-coefficient partially linear model and prove that the methods

23: possess the oracle property. Extensive Monte Carlo simulation

24: studies are conducted to examine the finite-sample performance of

25: the proposed procedures. Finally, we apply the new methods to

26: analyze the plasma beta-carotene level data.

27: \end{abstract}