We develop a log-linear structural VAR model to decompose unexpected excess market returns into permanent and temporary cash-flow news, discount-rate news, and non-fundamental news. We make some important findings. First, contrary to recent evidence for the United States, we find that discount-rate news plays a more important role than cash-flow news in global stock markets. Second, non-fundamental factor (i.e., investor sentiment) also plays a role in developed regional stock markets outside the United States. Finally, risk premiums are significantly higher in a down market than in an up market.

Key words: Log-linear SVAR model, Regional stock markets, Decomposed news, Market status