Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • STATA standardized Cumulative Abnormal Returns

    Hi all,

    My name is Polyxeni and I am a student in Finance and for my thesis I am computing an event study.I have calculated AR and CAR and now I have to make a t test. However in my event study multiple events happening at the same time hence I have to standardized CAR to test my hypothesis H0: CAR=0. I do not have so much experience hence I would like your help if my code is correct since when doing I am loosing 4 values from my event window (total of days: 7) .hence I am not able to use esttab to have my table for significance.

    where dif indicates the days in the event window e.g. dif= -3,-2,-1,0,1,2,3

    **testing for significance

    preserve
    drop if event_window==0
    gen CAR_MMM2 = CAR_MMM^2
    quietly describe
    bys dif:gen N=_N
    collapse ret AR_* CAR_* N, by(dif)
    gen s_MMM_C = sqrt(CAR_MMM2 - CAR_MMM^2)
    gen TS_MMM_C = sqrt(N)*CAR_MMM/s_MMM_C
    egen sbar_MMM = sd(CAR_MMM)
    gen TS_MMM_CR = sqrt(1/dif)*CAR_MMM/sbar_MMM

    restore

    Thank you in advance

    All comments are welcomed

    Best regards,

    Polyxeni

  • #2
    Hi Polyxeni,

    I'm not sure why your Stata code isn't working, but generally it's easier to run those tests directly in EVENTUS (which is part of CRSP / WRDS for Wharton. Eventus is just a program that uses CRSP to calculate abnormal returns for event studes). I don't know how good the documentation on the WRDS site is, but if you search for "Eventus user guide" it will show you how do all of what you ask. (i.e. Eventus can automatically standardize CAR's, can adjust for multiple events happening on the same date, can test whether day(-1) is significantly different from 0, etc).

    Comment

    Working...
    X