ABSTRACT
We describe in this chapter a class of global smoothing methods based on basis approximations for estimating the coefficient curves of the time-varying coefficient model (7.1). These basis approximation smoothing methods are natural extensions of the methods in Chapter 4 to models with time-varying covariates. Compared with the local smoothing methods in Chapters 7 and 8, the basis approximation methods can be viewed as a flexible extension of the classical parametric regression models described in Section 2.1 to nonparametric models, which have also been referred to in the literature as extended linear models, for example, Stone et al. (1997).
