Dear Statalist-Users,

I'm investigating the liquidity effects of an increase in reporting frequency. I have data on firms that increased reporting frequency from semiannual to quarterly because of a regulation. This regulation however is implemented in different countries at different points of time, which is why the firms listed in the different countries are affected at different point of time. Moreover I do have data for firms that did not change reporting frequency as they were listed in countries that already mandated quarterly reporting. I thought about using the firms that change reporting frequency as treatment group and thos that did not change as control group. I then would like to do an difference in differences analysis to examine the liquidity effects of that increase in reporting frequency. However, there are 2 things I'm wondering about :

1) Is it possible to use the "normal" diff in diff design : i.e. xtreg liquiditymeasure Post##Treat, fe or do I have to adjust it somehow? For example with time-fixed effect i.year? Im wondernig about that because of the different treatment times. I define Treat as a dummy cariable that equals one for those firms from the treatment group and 0 otherwise. Moreover I define Post as a dummy variable equals 1 for the years after the regulation was implemented. Which leads to my next question:

2) Even though I have data for the implementation of the regulation for all countries, the firms (control group)in the countries that already mandated quarterly reporting are not affected by the regulation. Hence I wonder whether I should neverthelss use the implementation dates of the control firms to define Post for them or whether I have to match them to the treatment firm first and then use the date of the matched treatment firm for defining the Post dummy for the control firm.

Thanks in advance!

king regards, Dominik