Princeton's Event study did not generate the predicted return

[Emma Ien]

Hi everyone,

I've gone through the Princeton event study stata coding, everything run through (I haven't coded the test bit), but I found that my dataset does not contain the predicted return and the accumulative return. What went wrong?

sum predicted_ret cumulative_abnormal_return

Variable | obs Mean Std. Dev. Min Max

-------------+---------------------------------------------------------

predicted_~n | 0

cumulative~n | 8, 214 0 0 0 0




My codes:




gen predicted_return=.

egen id=group(com_id)

forvalues i=1(1)144 { /*note: replace N with the highest value of id */
l id com\_id if id==\`i' & dif==0 reg ret mkt\_ret if id==\`i' & estimation\_window==1 predict p if id==\`i' replace predicted\_return = p if id==\`i' & event\_window3==1 drop p
}




*calculate abnormal return

sort id date

gen abnormal_return=ret-predicted_return if event_window3==1

by id: egen cumulative_abnormal_return = total(abnormal_return)

[Emma Ien]

I'm using stata 16 by the wat, would that affect the results?

[Andrew Musau]

It will help if you posted a link to the code that did not work.

[Emma Ien]

https://dss.princeton.edu/online_hel...study.html#car
This is the manual that i follow

[Leonardo Guizzetti]

Just a hunch, but is this the code you tried (precisely as shown) ?

forvalues i=1(1)144 { /*note: replace N with the highest value of id */
l id com\_id if id==\`i' & dif==0 reg ret mkt\_ret if id==\`i' & estimation\_window==1 predict p if id==\`i' replace predicted\_return = p if id==\`i' & event\_window3==1 drop p
}

If so, it will fail because you have a number of backslashes that simply confuse Stata or prevent macro substitution. And you also have multiple statements on the same line, which will also fail.

If this was not the failure, please post back with the exact code that you ran, and the output that Stata gave, especially where the error occurred. This way you will not waste your time and the time of those who are willing to help you.

[Andrew Musau]

Robert Picard updated the codes from that link. See #4 below and post back if you run into problems.

https://www.statalist.org/forums/for...s-coding-issue

[Emma Ien]

Just a hunch, but is this the code you tried (precisely as shown) ?

forvalues i=1(1)144 { /*note: replace N with the highest value of id */
l id com\_id if id==\`i' & dif==0 reg ret mkt\_ret if id==\`i' & estimation\_window==1 predict p if id==\`i' replace predicted\_return = p if id==\`i' & event\_window3==1 drop p
}

If so, it will fail because you have a number of backslashes that simply confuse Stata or prevent macro substitution. And you also have multiple statements on the same line, which will also fail.

If this was not the failure, please post back with the exact code that you ran, and the output that Stata gave, especially where the error occurred. This way you will not waste your time and the time of those who are willing to help you.

what is a multi-statement? is it the ones in /* ?

[Emma Ien]

Just a hunch, but is this the code you tried (precisely as shown) ?

forvalues i=1(1)144 { /*note: replace N with the highest value of id */
l id com\_id if id==\`i' & dif==0 reg ret mkt\_ret if id==\`i' & estimation\_window==1 predict p if id==\`i' replace predicted\_return = p if id==\`i' & event\_window3==1 drop p
}

If so, it will fail because you have a number of backslashes that simply confuse Stata or prevent macro substitution. And you also have multiple statements on the same line, which will also fail.

If this was not the failure, please post back with the exact code that you ran, and the output that Stata gave, especially where the error occurred. This way you will not waste your time and the time of those who are willing to help you.

I've deleted the /* bit, but the code runs with no error but there is still no return.

[Emma Ien]

the whole code I have:

* setting up company id
encode epic, generate(com_id)
drop epic
drop company

*gen sector var
encode sector, generate(sec_id)
drop sector

*create sector dummy variable?

* event study stata code for trading days
sort com_id date
by com_id: gen datenum=_n
by com_id: gen target=datenum if date==event_date
egen td=min(target), by (com_id)
drop target
gen dif=datenum-td

* codes for pre/post windows, 3 day windows
by com_id: gen event_window3= 1 if dif>=-1 & dif<=1
egen count_event_obs3= count(event_window3), by(com_id)

*estimation window
by com_id: gen estimation_window= 1 if dif<-6 & dif>-26
egen count_est_obs3=count (estimation_window), by(com_id)
replace event_window3=0 if event_window3==. 
replace event_window3=0 if estimation_window==.

* codes for pre/post windows, 5 day windows
by com_id: gen event_window5= 1 if dif>=-2 & dif<=2
egen count_event_obs5= count(event_window5), by(com_id)

replace event_window5=0 if event_window5==. 
replace event_window5=0 if estimation_window==.

* codes for pre/post windows, 11 day windows
by com_id: gen event_window11= 1 if dif>=-5 & dif<=5
egen count_event_obs11= count(event_window11), by(com_id)

replace event_window11=0 if event_window11==. 
replace event_window11=0 if estimation_window==.

*predict return
gen predicted_return=.
egen id=group(com_id) 
 forvalues i=1(1)144 { 
    l id com_id if id==`i' & dif==0
    reg ret mkt_ret if id==`i' & estimation_window==1 
    predict p if id==`i'
    replace predicted_return = p if id==`i' & event_window3==1 
    drop p
}   

*calculate abnormal return
sort id date
gen abnormal_return=ret-predicted_return if event_window3==1
by id: egen cumulative_abnormal_return = total(abnormal_return) 

*Testing for Significance
sort id date
by id: egen ar_sd = sd(abnormal_return) 
gen test =(1/sqrt(3)) * ( cumulative_abnormal_return /ar_sd) 
list com_id cumulative_abnormal_return test if dif==0

export excel com_id event_date cumulative_abnormal_return test using "stats.xls" if dif==0, firstrow(variables) replace

[Emma Ien]

Robert Picard updated the codes from that link. See #4 below and post back if you run into problems.

https://www.statalist.org/forums/for...s-coding-issue

I tried the code, I'm using stata 16 and it seems that

 rangestat (reg) ret2use mkt_ret, interval(com_id 0 0)
command rangestat is unrecognized
r(199);

& if I change to

reg ret2use mkt_ret, interval(com_id 0 0)
option interval() not allowed
r(198);

What went wrong?

[Andrew Musau]

ssc install rangestat, replace

[Emma Ien]

I tried the code, I'm using stata 16 and it seems that

 rangestat (reg) ret2use mkt_ret, interval(com_id 0 0)
command rangestat is unrecognized
r(199);

& if I change to

reg ret2use mkt_ret, interval(com_id 0 0)
option interval() not allowed
r(198);

What went wrong?

I need to instal ssc

use code:

 
 ssc install rangestat

then check:

 
 help rangestat

to see if regenestat is usable

[Nick Cox]

The backslashes in earlier posts arise because of cross-posting on https://www.reddit.com/r/stata/comme..._generate_the/ and of getting confused between two modes of editing allowed there. Please note our policy on cross-posting, which is that you are asked to tell us about it.

In #10 you must install rangestat before you can use it.

ssc install rangestat

As its help explains, regress does not support an option interval(). That's an option of rangestat.

[Emma Ien]

ssc install rangestat, replace

This works, thank you so much!

[Emma Ien]

Also what codes should I add in if I also want to do an event window for

w_event = inrange(days,-1,1)
gen w_event = inrange(days,-5,5)

this is my current code

ssc install rangestat

* setting up company id
encode epic, generate(com_id)
drop epic
drop company

*gen sector var
encode sector, generate(sec_id)
drop sector

*create sector dummy variable?

* event study stata code for trading days
sort com_id date
by com_id: gen tradeday = _n
by com_id: gen event_obs = event_date == date
bysort com_id (event_obs): gen days = tradeday - tradeday[_N] if event_date[_N] == date[_N]

gen w_event = inrange(days,-1,1)
gen w_estim = inrange(days,-26,-6)
by com_id: egen N_w_event = total(w_event)
by com_id: egen N_w_estim = total(w_estim)

drop if N_w_event < 3
drop if N_w_estim < 20

* Estimating Normal Performance
gen ret2use = ret if w_estim
rangestat (reg) ret2use mkt_ret, interval(com_id 0 0)
gen p_return = b_mkt_ret * mkt_ret + b_cons if w_event

* Abnormal and Cumulative Abnormal Returns
gen abnormal_return = ret - p_return
by com_id: egen cumulative_abnormal_return = total(abnormal_return)

* Testing for Significance
by com_id: egen ar_sd = sd(abnormal_return) 
gen test =(1/sqrt(N_w_event)) * ///
    (cumulative_abnormal_return /ar_sd) 

* Testing Across All Events
reg cumulative_abnormal_return if days==0