Are stock market crises contagious? The role of crisis definitions

•Contagion studies analyse excess co-movement between markets during crises.•We examine how results from such studies depend on the chosen 'crisis' sample.•We generalise the main crisis dating approaches used in the literature.•We apply our method to the 1997 Asia crisis and the 2007 financial crisis.•We find no relationship between co-movement and commonly used crisis dates.

Financial contagion studies generally examine whether co-movement between markets increases during a crisis. We use a flexible co-movement measure to examine how conclusions of such analyses depend on the sample chosen as the 'crisis'. To this end, we analyse stock market co-movement during the 1997 Asian crisis and the 2007 global financial crisis for all possible source countries and for all possible time periods or extreme return quantiles. This way we account for the main crisis dating approaches adopted in the literature. Our results suggest there is no clear relationship between excess co-movement and commonly used crisis samples.