Title: Industry return predictability: Evidence from China
Authors: Yawen Zheng - University of Liverpool (United Kingdom) [presenting]
Michael Ellington - University of Liverpool (United Kingdom)
Michalis Stamatogiannis - University of Liverpool Management School (United Kingdom)
Abstract: The purpose is to demonstrate that Chinese industry portfolio returns are able to forecast the returns of other industry portfolios. The forecastability differs between industry portfolios and some industries are more useful predictors than others. We use a machine learning technique (LASSO) to select the most relevant predictors for each industry and then analyse these relationships at monthly, weekly and daily data frequencies. We find significant cross-industry return forecastability and out-of-sample tests indicate positive forecasting performance. Our results hold to a number of robustness checks.