Hello everyone!
I would kindly like to ask for some help concerning time dummies (thank you very much).
So here is my problem. I have a panel data of 100 cities belonging to a region of France over a time period from 1980 to 2000. I want to test the effect of saturation (ie. Number of residents+immigrants (in levels) or Number of residents /(Number of residents+immigrants) (as a share)) on the immigration rate for each city. My problem is that I have to consider 2 events. In 1985 some of these cities (I don’t know which ones exactly) were conferred by the state full power in implementing immigration policy regulations. In 1990, all the remaining cities were conferred this power. I want to test the effect of saturation on the migration rate, given the two time events of 1985 and 1990 (I expect that this provision of power from the state to the cities involved more restrictive immigration policies). I was hence thinking about using two year dummies (one for 1985 and one for 1990) even if I do not know very well how to do this in stata..
First of all I think that I cannot insert the two time dummies in a single equation (am I wrong)? So I thought about estimating two different regression equations per each year, in this way:
Immigration rate_it = a0 + a1*saturation_it + a2*saturation_it*DUMMY_1985
(where DUMMY_1985 is a dummy =1 if year >= 1985)
The same logic holds for the other equation:
Immigration rate_it = b0 + b1*saturation_it + b2*saturation_it*DUMMY_1990
(where DUMMY_1990 is a dummy =1 if year >= 1990)
From these, I expect that b2>a2.
My question is: do you think this procedure is correct?
1) Is there a way to estimate one single regression equation?
1) What about removing from the regressions a1 and b1?
2) Is there a way to calculate directly, for instance:
Immigration rate_it = b0 + b1*saturation_it + b2*saturation_it*DUMMY_1990_ONLY ?
(where DUMMY_1990_ONLY = 1 for year 1990 and 0 otherwise). Would this be also feasible? Since I want to derive the impact of saturation if year=1990 compared to the years prior 1990.
I apologize if I was long. Thank you very much whether you kindly decided to help me![]( "Smile")
K
I would kindly like to ask for some help concerning time dummies (thank you very much).
So here is my problem. I have a panel data of 100 cities belonging to a region of France over a time period from 1980 to 2000. I want to test the effect of saturation (ie. Number of residents+immigrants (in levels) or Number of residents /(Number of residents+immigrants) (as a share)) on the immigration rate for each city. My problem is that I have to consider 2 events. In 1985 some of these cities (I don’t know which ones exactly) were conferred by the state full power in implementing immigration policy regulations. In 1990, all the remaining cities were conferred this power. I want to test the effect of saturation on the migration rate, given the two time events of 1985 and 1990 (I expect that this provision of power from the state to the cities involved more restrictive immigration policies). I was hence thinking about using two year dummies (one for 1985 and one for 1990) even if I do not know very well how to do this in stata..
First of all I think that I cannot insert the two time dummies in a single equation (am I wrong)? So I thought about estimating two different regression equations per each year, in this way:
Immigration rate_it = a0 + a1*saturation_it + a2*saturation_it*DUMMY_1985
(where DUMMY_1985 is a dummy =1 if year >= 1985)
The same logic holds for the other equation:
Immigration rate_it = b0 + b1*saturation_it + b2*saturation_it*DUMMY_1990
(where DUMMY_1990 is a dummy =1 if year >= 1990)
From these, I expect that b2>a2.
My question is: do you think this procedure is correct?
1) Is there a way to estimate one single regression equation?
1) What about removing from the regressions a1 and b1?
2) Is there a way to calculate directly, for instance:
Immigration rate_it = b0 + b1*saturation_it + b2*saturation_it*DUMMY_1990_ONLY ?
(where DUMMY_1990_ONLY = 1 for year 1990 and 0 otherwise). Would this be also feasible? Since I want to derive the impact of saturation if year=1990 compared to the years prior 1990.
I apologize if I was long. Thank you very much whether you kindly decided to help me
K