A0983
Title: MOSAICA with application to global factor investment
Authors: Dan Yang - The University of Hong Kong (Hong Kong)
Jianlong Shao - The University of Hong Kong (Hong Kong)
Yan Xu - The University of Hong Kong (Hong Kong)
Haipeng Shen - The University of Hong Kong (Hong Kong) [presenting]
Abstract: International stock returns are known to display non-Gaussian distributional and series dependence patterns, rendering conventional cross-sectional factor models based on Gaussian and temporally independent assumptions, such as PCA, less preferred than independent component analysis (ICA). The aim is to propose a novel ICA model suited for identifying factors from such patterns and to develop a hierarchical ICA model to identify factor structures of stock markets across multiple countries. The proposed model accommodates differences in factor structures between developed countries and emerging markets while maintaining homogeneity within each group. The statistical merits of the method are demonstrated through numerical experiments, and superior forecasts of size and momentum factor returns are leveraged to create a market timing strategy, significantly enhancing the performance of various long-short zero-cost portfolios.
