A1371
Title: An alternative bootstrap procedure for factor-augmented regression models
Authors: Takashi Yamagata - University of York (United Kingdom) [presenting]
Peiyun Jiang - Tokyo Metropolitan University (Japan)
Abstract: A novel bootstrap algorithm that is more efficient than existing methods for replicating the asymptotic distribution of the factor-augmented regression estimator for a rotated parameter vector is proposed. The regression is augmented by r factors extracted from a large panel of $N$ variables observed over $T$ time periods. General weak latent factor models with r signal eigenvalues are considered, that may diverge at different rates, $N^a_k$, where $0<a_k\le 1$ for $k=1,2,...,r$. The asymptotic validity of the bootstrap method is established using not only the conventional data-dependent rotation matrix Hhat, but also an alternative data-dependent rotation matrix, $\hat H_q$, which typically exhibits smaller asymptotic bias and achieves a faster convergence rate. Furthermore, the asymptotic validity of the bootstrap is demonstrated under a purely signal-dependent rotation matrix $H$, which is unique and can be regarded as the population analogue of both Hhat and $\hat H_q$. Experimental results provide compelling evidence that the proposed bootstrap method achieves superior performance relative to existing approaches.
