version 6
set more off
capture log close
set matsize 500
log using "C:\documents\ellison\event.log",replace
*This do-file implements an event study, to assess whether 
*the retirement (and subsequent unretirement) of M.Jordan 
*from basketball affected the securities of the companies 
*advertised by him and/or the securities of competing companies

*Before running this do-file, download the datset "event.dta"
*from the data web page and change the directories specified below
*(and above) according to those in your own pc. Remember also to make 
*Stata store a sufficient amount of memory.
*The dataset contains a date variable, two measures of market return
*and stock returns for 15 companies: 1) reebok; 2) converse inc del;
*3) coca cola; 4) general motors; 5) pepsi; 6) wringley;
*7) general mills; 8) sara lee corp; 9) quaker oats; 10)kellogg's;
*11) converse inc; 12)mcdonald's; 13)nike; 14) post corp;
*15) tribune co new
use "C:\documents\ellison\event.dta",clear
sum

*Because of missing values, variables r2, r11 and r14 
*are not used for the analysis
drop r2 r11 r14

*Create dummies for the retirement event (October 6 1993)
*and the unretirement event (March 18 1995)
gen ret= (date==12332)|(date==12331)|(date==12333)
gen unret= (date==12860)|(date==12859)|(date==12861)

*Notice that to retreive the Stata values of specific dates
*you can use the following commands interactively
scalar sc_ret=mdy(10,6,1993)
scalar list sc_ret
scalar sc_unret=mdy(3,18,1995)
scalar list sc_unret

*Set of regressions for each company
*DIRECT EFFECT (i.e. effect on advertised companies)
reg r4 ewret ret unret,robust
reg r4 vwret ret unret,robust
reg r6 ewret ret unret,robust
reg r6 vwret ret unret,robust
reg r7 ewret ret unret,robust
reg r7 vwret ret unret,robust
reg r8 ewret ret unret,robust
reg r8 vwret ret unret,robust
reg r9 ewret ret unret,robust
reg r9 vwret ret unret,robust
reg r12 ewret ret unret,robust
reg r12 vwret ret unret,robust
reg r13 ewret ret unret,robust
reg r13 vwret ret unret,robust
reg r15 ewret ret unret,robust
reg r15 vwret ret unret,robust
*INDIRECT EFFECT (i.e. effect on competing companies)
reg r1 ewret ret unret,robust
reg r1 vwret ret unret,robust
reg r3 ewret ret unret,robust
reg r3 vwret ret unret,robust
reg r5 ewret ret unret,robust
reg r5 vwret ret unret,robust
reg r10 ewret ret unret,robust
reg r10 vwret ret unret,robust

*Transforming the dataset into a single-return one
reshape long r,i(date) j(comp)
sum

*Create company dummies
gen comp1= comp==1
gen comp3= comp==3
gen comp4= comp==4
gen comp5= comp==5
gen comp6= comp==6
gen comp7= comp==7
gen comp8= comp==8
gen comp9= comp==9
gen comp10= comp==10
gen comp12= comp==12
gen comp13= comp==13
gen comp15= comp==15

*Create interaction terms
*Between ewret and company dummies
gen mkt1=comp1*ewret
gen mkt3=comp3*ewret
gen mkt4=comp4*ewret
gen mkt5=comp5*ewret
gen mkt6=comp6*ewret
gen mkt7=comp7*ewret
gen mkt8=comp8*ewret
gen mkt9=comp9*ewret
gen mkt10=comp10*ewret
gen mkt12=comp12*ewret
gen mkt13=comp13*ewret
gen mkt15=comp15*ewret
*Between vwret and company dummies
gen mkt1b=comp1*vwret
gen mkt3b=comp3*vwret
gen mkt4b=comp4*vwret
gen mkt5b=comp5*vwret
gen mkt6b=comp6*vwret
gen mkt7b=comp7*vwret
gen mkt8b=comp8*vwret
gen mkt9b=comp9*vwret
gen mkt10b=comp10*vwret
gen mkt12b=comp12*vwret
gen mkt13b=comp13*vwret
gen mkt15b=comp15*vwret
*Between ret and company dummies
gen ret1=comp1*ret
gen ret3=comp3*ret
gen ret4=comp4*ret
gen ret5=comp5*ret
gen ret6=comp6*ret
gen ret7=comp7*ret
gen ret8=comp8*ret
gen ret9=comp9*ret
gen ret10=comp10*ret
gen ret12=comp12*ret
gen ret13=comp13*ret
gen ret15=comp15*ret
*Between unret and company dummies
gen unret1=comp1*unret
gen unret3=comp3*unret
gen unret4=comp4*unret
gen unret5=comp5*unret
gen unret6=comp6*unret
gen unret7=comp7*unret
gen unret8=comp8*unret
gen unret9=comp9*unret
gen unret10=comp10*unret
gen unret12=comp12*unret
gen unret13=comp13*unret
gen unret15=comp15*unret

*Regressions imposing equal direct effects across advertised companies
reg r comp1-comp15 mkt1-mkt15 ret unret,robust noconstant
reg r comp1-comp15 mkt1b-mkt15b ret unret,robust noconstant

*Testing (Wald) whether the direct effects are equal across advertised companies
reg r comp1-comp15 mkt1-mkt15 ret1-ret15 unret1-unret15,robust noconstant
testparm ret1-ret15,equal
testparm unret1-unret15,equal
reg r comp1-comp15 mkt1b-mkt15b ret1-ret15 unret1-unret15,robust noconstant
testparm ret1-ret15,equal
testparm unret1-unret15,equal

clear
log close