HAC standard errors and the event study methodology: a cautionary note