Difference-in-Differences Panel Data - parallel trend assumption

[Roksana HM]

Hi there,

I am working on a panel dataset with stock prices of multiple companies (control and treatment groups) to see the effect of a policy implementation. Using "xtdidregress" when I try to do the parallel trend assumption (i.e., estat trendplots) I keep getting this error: "treatment assignment times vary; not allowed with estat trendplots".

I have many missing values (stock prices) for this sample.

I very much appreciate it if you could help me.

I tried the following codes:

xtdidregress ( ln_ri ) ( treatment ), group( con_cod ) time( time )

 egen first = min(time) if treatment , by( con_cod )

Regards,
Roksana

[Jeff Wooldridge]

Roksana: It appears you have a staggered intervention , in which case you should use xthdidregress. Does the regulation affect firms initially in different time periods?

[Roksana HM]

Jeff Wooldridge thanks for that. The policy started at the same period for all the companies. I think the problem is that we have a lot of missing values (stock prices) for those companies which I don't know how to deal with.

[Carlo Lazzaro]

Roksana:
see -ipolate- and -mi- and make your choice in the lighr of the missingness mechanism of your unobserved valies.