Prediction of Financial Contagion and Spillover Effects of the US Financial Crisis Using Google Index

This study predicts the contagion and spillover effects of the US financial crisis using the methods of time-varying vector autoregressive with exogenous variable and dynamic copula with ARMAX process, respectively. These two models are modified by incorporating the Google trend data as a new predictor or exogenous variable. We evaluate the contagion and spillover effects between the US stock market and ten other stock markets, including developed and developing markets, from 2007 to 2018. For spillover effects, the Google index could enhance the time-varying vector autoregressive model prediction for all pairs except USA-Thailand. From the aspect of contagion effect, several dynamic copula families are employed, and we find that Gaussian copula seems to perform well for capturing the dependence between the stock markets of interest in many cases. We also find that, for 6 out of 10 pairs, the use of Google index significantly improves prediction accuracy. In addition, the time-varying parameters and correlations reveal a positive and negative impact of the US stock market on both developed and developing markets, especially during the advent of the US-originated global financial crisis.