I want to study the effect of a policy on retail prices in states where a particular policy is imposed and where it isn't, during holidays.
In my data, there are 3 states - CA (4 stores), TX (3 stores), WI (3 stores). The policy is imposed in CA and TX (7 stores then) and not in WI. All stores have the same 40 items in the data and prices are observed weekly for 5 years.
My main variable of interest is the interaction between the policy dummy (=1 if the policy is in place in the state, 0 otherwise - time invariant) and holiday dummy (time-varying).
I want to do a correlated random effects model since I want to estimate the time-invariant policy dummy too.
1. Will the coefficient estimates for the policy dummy, holiday dummy, and their interaction be unreliable/ inflated since there are more stores in states with the policy?
In my data, there are 3 states - CA (4 stores), TX (3 stores), WI (3 stores). The policy is imposed in CA and TX (7 stores then) and not in WI. All stores have the same 40 items in the data and prices are observed weekly for 5 years.
My main variable of interest is the interaction between the policy dummy (=1 if the policy is in place in the state, 0 otherwise - time invariant) and holiday dummy (time-varying).
I want to do a correlated random effects model since I want to estimate the time-invariant policy dummy too.
1. Will the coefficient estimates for the policy dummy, holiday dummy, and their interaction be unreliable/ inflated since there are more stores in states with the policy?
- I don't know if this the right approach to check but I ran the model on i) TX and WI and ii) for all states together - the estimates didn't change except for the holiday dummy but by very little, similarly for p-values.
