log using state_cig_data_2.log, replace

use state_cig_data


* generate real taxes
gen real_tax=(state_tax+federal_tax)/cpi

* construct new variables

* time trend
gen trend=year-1981
label var trend "=1 in 1st year, 2 in second, etc"


* tobacco producing state
gen tob_state=(state=="NC"|state=="VA"|state=="SC"|state=="KY"|state=="MD"|state=="TN")
label var tob_state "=1 if tobacco producing state"

* run regression with tax and trend
reg packs_pc real_tax trend 

* add tobacco producing state dummy
reg packs_pc real_tax trend tob_state

* correlation between tax and other variables
reg real_tax trend tob_state

* now run a model with seperate dummies
* for each tobacco producing state
* test two hypothesis
* h1:  coef are all zero
* h2:  coef are all equal

gen nc=state=="NC"
gen va=state=="VA"
gen ky=state=="KY"
gen md=state=="MD"
gen tn=state=="TN"
gen sc=state=="SC"

* run unrestricted model for both hypotheses
reg packs_pc real_tax trend nc va ky md tn sc

* do test 1
test nc va ky md tn sc

* do test 2
test nc=va=ky=md=tn=sc

* do test by brute force
* for h1, run restricted model
* (no state effects)
reg packs_pc real_tax trend 


* for h1, run resticted model
* on dummy for all tobacco producing states
reg packs_pc real_tax trend tob_state


log close