Dear members,
I am a student new to STATA and may need your help on the below question.
I have the level regression as below:
π’ππ‘=ππ‘ + πΏπ + Ξ£π‘β₯1400πΌπ‘ WEπ ππ‘ + Ξ£π‘β₯1400π½π‘ ππ΄ππ ππ‘ +πππ‘
(u is log(population); WE is dummy variable if the city is in Western Europe; PAT is dummy variable if the city is involved in Atlantic trade; ππ‘ denotes time effects and πΏπ denotes city effects)
and I am looking to transfer this to first difference regression, I wrote the equation as below (not sure if this is correct)
π’ππ‘ βπ’ππ‘β1 =β³ππ‘ +πΌπ‘ WEπ ππ‘ + π½π‘ ππ΄ππ ππ‘ +πππ‘
And I tried to estimate the difference regression, and I thought that the point estimates of coefficients should be the same as those in the level regression.
I used the code:
gen yearn = 1 if year == 1300
replace yearn = 2 if year == 1400
replace yearn = 3 if year == 1500
replace yearn = 4 if year == 1600
replace yearn = 5 if year == 1700
replace yearn = 6 if year == 1750
replace yearn = 7 if year == 1800
replace yearn = 8 if year == 1850
bysort city: gen du=u-u[yearn-1]
gen du1 = du if year == 1850
reg du1 WE_y8 PAT_y8
However, when I compared the coefficients with those in level regression they are not the same. Please give me some advices on my equation and estimates.
The data sample are below
Thank you very much !!!
I am a student new to STATA and may need your help on the below question.
I have the level regression as below:
π’ππ‘=ππ‘ + πΏπ + Ξ£π‘β₯1400πΌπ‘ WEπ ππ‘ + Ξ£π‘β₯1400π½π‘ ππ΄ππ ππ‘ +πππ‘
(u is log(population); WE is dummy variable if the city is in Western Europe; PAT is dummy variable if the city is involved in Atlantic trade; ππ‘ denotes time effects and πΏπ denotes city effects)
and I am looking to transfer this to first difference regression, I wrote the equation as below (not sure if this is correct)
π’ππ‘ βπ’ππ‘β1 =β³ππ‘ +πΌπ‘ WEπ ππ‘ + π½π‘ ππ΄ππ ππ‘ +πππ‘
And I tried to estimate the difference regression, and I thought that the point estimates of coefficients should be the same as those in the level regression.
I used the code:
gen yearn = 1 if year == 1300
replace yearn = 2 if year == 1400
replace yearn = 3 if year == 1500
replace yearn = 4 if year == 1600
replace yearn = 5 if year == 1700
replace yearn = 6 if year == 1750
replace yearn = 7 if year == 1800
replace yearn = 8 if year == 1850
bysort city: gen du=u-u[yearn-1]
gen du1 = du if year == 1850
reg du1 WE_y8 PAT_y8
However, when I compared the coefficients with those in level regression they are not the same. Please give me some advices on my equation and estimates.
The data sample are below
city | year | PAT | u | WE |
aachen | 1300 | 0 | 3.044523 | 1 |
aachen | 1400 | 0 | 2.302585 | 1 |
aachen | 1500 | 0 | 2.890372 | 1 |
aachen | 1600 | 0 | 3.135494 | 1 |
aachen | 1700 | 0 | 2.70805 | 1 |
aachen | 1750 | 0 | 3.178054 | 1 |
aachen | 1800 | 0 | 3.178054 | 1 |
aachen | 1850 | 0 | 3.951244 | 1 |
aalst | 1300 | 0 | 1.386294 | 1 |
aalst | 1400 | 0 | 1.386294 | 1 |
aalst | 1500 | 0 | 1.609438 | 1 |
aalst | 1600 | 0 | 1.94591 | 1 |
aalst | 1700 | 0 | 2.079442 | 1 |
aalst | 1750 | 0 | 2.079442 | 1 |
aalst | 1800 | 0 | 2.397895 | 1 |
aalst | 1850 | 0 | 2.833213 | 1 |
amsterdam | 1300 | 1 | 0 | 1 |
amsterdam | 1400 | 1 | 1.098612 | 1 |
amsterdam | 1500 | 1 | 2.70805 | 1 |
amsterdam | 1600 | 1 | 3.988984 | 1 |
amsterdam | 1700 | 1 | 5.298317 | 1 |
amsterdam | 1750 | 1 | 5.347107 | 1 |
amsterdam | 1800 | 1 | 5.379897 | 1 |
amsterdam | 1850 | 1 | 5.416101 | 1 |
athens | 1300 | 0 | 3.218876 | 0 |
athens | 1400 | 0 | 3.555348 | 0 |
athens | 1500 | 0 | 2.302585 | 0 |
athens | 1600 | 0 | 3.496508 | 0 |
athens | 1700 | 0 | 2.564949 | 0 |
athens | 1750 | 0 | 2.302585 | 0 |
athens | 1800 | 0 | 2.484907 | 0 |
athens | 1850 | 0 | 3.401197 | 0 |