ABSTRACT
The psychological changes of investors in crude oil market often cause crude oil price deviating from fundamentals and the Internet search data can effectively reflect the psychological behaviors of investors in crude oil market. The search data aggregated by Google effectively reflects investors’ attention. In this sense, we construct a direct, timely and unambiguous proxy for investor attention in crude oil market by aggregating the Google search volume index (GSVI). Then, we explore the effect and predictive power of the investor attention on crude oil price returns by using the autoregressive integrated moving average (ARMAX) models with exogenous variables and Autoregressive integrated moving average with exogenous variable-Generalized autoregressive conditional heteroscedastic (ARMAX-GARCH). The empirical results indicate that, first, there exists a negative effect of the Google Index on crude oil returns; second, the incorporation of investor attention brings an improvement on the goodness-of-fit; finally, in general, the incorporation of investor attention fails to improve the forecasting performance in crude oil market, specifically, the forecasting performance becomes worse after the incorporation of the investor attention according to almost all the loss functions.
