I have a panel data where my dependent variable ( Xt) impacts different i's at different times. Say Xt effects Yit from period 9005 and effects Yjt 052015. How exactly can I measure the effect of x on y?
Estimation question


Bump

are you confusing dependent and independent variables?

I am no expert but depending on how large your sample is, I would consider splitting it into the group that is affected in 9095 and the other that is 0515. You should try to understand why you have this discrepancy.

are you confusing dependent and independent variables?
I think that must be the case

No. Let's say Y is industry output where i is the type of industry. X is a determinant of industry output. The issue is that the effect x has on a particular industry is dependent on time. Say x can only effect one industry at a time.
are you confusing dependent and independent variables?

I was thinking that but The issue is that I don't know the exact times of when the change occurs. I guess I can find the structural breaks for each i from the full sample and get the estimates that way
I am no expert but depending on how large your sample is, I would consider splitting it into the group that is affected in 9095 and the other that is 0515. You should try to understand why you have this discrepancy.

I was thinking that but The issue is that I don't know the exact times of when the change occurs. I guess I can find the structural breaks for each i from the full sample and get the estimates that way
I am no expert but depending on how large your sample is, I would consider splitting it into the group that is affected in 9095 and the other that is 0515. You should try to understand why you have this discrepancy.
That sounds reasonable. If you have any other issues it might be worth posting on the Stata forum or maybe even EViews.

use a state space model.

interact x with time dummies

But also industry dummies .. so X * time* industry
interact x with time dummies