Bandwidth Selection for Estimators of Time-Varying Stochastic Coefficient Models

A recent strand of the literature has proposed stochastic time-varying coefficient models for modelling structural change in the macroeconomy under both exogeneity and endogeneity. Subsequently, a new class of kernel based non-parametric estimators has been introduced for these models. These kernel based approaches allow for flexible modelling of a time-series without relying on a specific parametric class. However, these estimators are to be tuned by a smoothing parameter, the bandwidth. This paper compares different data-driven selectors for the smoothing parameter