Event study standard deviation for t-test

[Mike Dooijeweerd]

Hi,

I'm currently doing an event study using a sample of firms and a three days event window [-1,+1]. I'm struggling a bit with which standard deviation I've to use to perform the t-test. Can I simply use a standard deviation over the abnormal returns over all firms and all dates? Or do I have to generate the standard deviation per date and use the one of the event date in my t-test?

I need a standard deviation over all firms in my sample, but wondering of I can perform it over all dates.

Kind regards

[Rich Goldstein]

I find your description completely confusing - in a standard event study, one has censored data; the t-test is not appropriate with censored data; you need to provide more information about your study

[Mike Dooijeweerd]

Hi Rich,

Thanks for your reply. You’re right. I just mean test statistic in stead of t-test. I’m doing a research about certain events during the Covid-19 outbreak on listed equity REITs in the US. After removing firms with incomplete data by sample is left with 299 firms. I’ve calculated the abnormal returns, average abnormal etc.
Now I’m in need of the standard deviation to calculate the test statistic. I was wondering if I just can use: gen sd=sd(abnormal_return) across all companies and dates or I have to do it in another way.

kind regards