A0793
Title: Inference on three-pass regression filter with high-dimensional target variables
Authors: Shou-Yung Yin - National Taipei University (Taiwan) [presenting]
Abstract: The focus is on the high-dimensional target variables using the three-pass regression filter (3PRF) proposed previously. The 3PRF is designed for extracting useful information from an extra data set, which is supposed to be useful for improving the forecast performance of the target variable. Compared to the traditional principal component approach~(PCA) based on eigendecomposition, the 3PRF delivers a closed-form solution of the estimated factors and the corresponding estimated coefficients. The proposed approach is robust to the different presumed factor numbers, while the performance of the PCA approach is very sensitive to the number of factors. In the empirical study, the proposed method is used to extract the common components which can be used to predict the fundamentals of the dynamics of house prices in the U.S.
