Announcement

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

  • eventstudy2 updated

    Dear Members of the Statalist,

    A new version of eventstudy2 is available on SSC:

    NEW IN THIS VERSION:

    - GRANK-T test for cumulative average abnormal returns according to Kolari and Pynnönen (2011)
    - Adjusted Patell test (with (Kolari and Pynnönen (2010) adjustment)
    - Wilcoxon (1945) signed ranks test
    - a substantially improved error output when input data does not match eventstudy2's criteria
    - correction of an error in displaying significance levels of buy-and-hold returns and the correction of an error in calculating the bootstrapped skewness-adjusted t-ratio

    Best regards
    Thomas Kaspereit

  • #2
    Originally posted by Thomas Kaspereit View Post
    Dear Members of the Statalist,

    A new version of eventstudy2 is available on SSC:

    NEW IN THIS VERSION:

    - GRANK-T test for cumulative average abnormal returns according to Kolari and Pynnönen (2011)
    - Adjusted Patell test (with (Kolari and Pynnönen (2010) adjustment)
    - Wilcoxon (1945) signed ranks test
    - a substantially improved error output when input data does not match eventstudy2's criteria
    - correction of an error in displaying significance levels of buy-and-hold returns and the correction of an error in calculating the bootstrapped skewness-adjusted t-ratio

    Best regards
    Thomas Kaspereit
    Thanks for your contribution! It is very helpful.

    Comment


    • #3
      Dear Professor Thomas Kaspereit,
      I can not find the AAR test of GRANKt in eventstudy2 as other papaer said that GRANK test can be used on AAR and CAAR. Is there any new version for solving this case?
      Best RGDs!
      James LIN

      Comment


      • #4
        A further question is that if the test results are different from Wilcoxon and Grank, say one is significant but the other is none.

        Comment


        • #5
          A further question is that which one is the better choice if the test results are different from Wilcoxon and Grank, say one is significant but the other is none.

          Comment


          • #6
            Dear Sir,

            I suggest you use the up to 10 CAR-window options to obtain GRANKt stats for single days, e.g. carlb(0) carub(0) will provide you the GRANKt for the event day.

            I cannot answer your second Q, which test is preferable. They are based on different assumptions, and assumptions are more or less reflective of reality but not better, worse, or one preferable over the other a priori.

            Best Thomas

            Comment


            • #7
              Dear Thomas:
              Thanking a lot for your quick response. I have done your suggest but that the GRANKt for AAR still can not be seen as the following images. It does not show the case on my event study2 version or something I don't recognise.
              Click image for larger version

Name:	a aarfile.png
Views:	1
Size:	22.3 KB
ID:	1681143
              Click image for larger version

Name:	a aar test from smcl.png
Views:	1
Size:	21.0 KB
ID:	1681144

              Comment


              • #8
                Dear James,
                Yes, you cannot see it there because you are looking at the AARs. You have to look at the CARs instead, which are now 1-day CARs, which are equivalent to AARs. Thus, you find the GRANKT statistics in the "carfile". But since you defined 1-day CARs, these will essentially be AARs.
                Best
                Thomas

                Comment


                • #9
                  Dear Professor Thomas Kaspereit,

                  I just got a new laptop recently and installed the evenstudy2 package. However, every time I tried to use the eventstudy2 command, I faced an error saying that "invalid syntax" (picture below).
                  Even when I tried to use the Example codes, I still encountered this problem.
                  May I ask you how I can fix it? Is this something wrong with my setting on the laptop since I used to perform the eventstudy2 command previously with no error like this?

                  Best regards,
                  Dan.

                  Click image for larger version

Name:	Stata error.png
Views:	1
Size:	79.8 KB
ID:	1686261


                  Comment


                  • #10
                    Dear Dan,

                    May I ask you to 'set trace on' , run eventstudy2 and post or send me the logfile?

                    Thanks Thomas

                    Comment


                    • #11
                      Dear Thomas,

                      thanks a lot for providing such a useful tool! I've got a question regarding the test statistics in eventstudy2. Are those performed as one-sided test or generally as two-sided test?
                      And would it be possible to switch from one option to another, i.e. from one-sided to two-sided an vice versa?

                      Thank you a lot in advance!

                      Best wishes,
                      Phil

                      Comment


                      • #12
                        Dear Phil,

                        Thanks a lot. The significance levels are based on two-tailed tests and cannot be changed.

                        However, you can access the test statistics themselves and interpret them accordingly.They are provided in one of eventstudy2's output files

                        Best
                        Thomas

                        Comment

                        Working...
                        X