Forecasting the stock returns of Chinese oil companies: Can investor attention help?

Abstract

This paper builds the investor attention index based on the Google Search Volume Index to forecast the stock returns of three major Chinese oil companies during 2004–2019. It takes into account several different model properties, including the time-varying parameters, regime switching and mixed-data frequency, etc. The empirical results suggest that investor attention helps to forecast the stock returns, and its capacity to forecast the US and Hong Kong stock markets is stronger than that of China's A-share market. For the US and Hong Kong markets, forecasting models with investor attention improves the accuracy by up to 35% compared with the benchmark model; but for the A-share market, the new forecasting models are superior only when it is one-month ahead. In addition, this paper confirms that investor attention does have economic value and can help mean-variance investors to get larger certainty equivalent return.