Announcement

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

  • how to deal with '' xthenreg'' for dynamic threshold models

    Dear all

    Finally we get the stata' command to deal with the endogeneity issue for threshold model, but it still not clear.

    i mean how can we know if it exists the threshold effect and at which level




    Click image for larger version

Name:	xthenreg.PNG
Views:	1
Size:	30.0 KB
ID:	1495809





    so please how to deal with the outputs using " xthenreg "
    kind regards
    Sedki,

  • #2
    dear all

    please any help ???

    Comment


    • #3
      Dear sedki, I think that "Bootstrap p-value for linearity test" is what you are looking for. Too bad that the authors did not report the corresponding Wald test statistic (and I don't know how to export this critical statistic out of the ado file!
      Ho-Chuan (River) Huang
      Stata 16.1, MP(4)

      Comment


      • #4
        Dear River Huang

        thank you very much for your intervention

        actually yes, i can not make conclusion about it.

        in their article, the authors mention in the options syntax the following command but unfortunately i could not integrate it in the main command
        ( i think it has a relation with the number of thresholds)

        Click image for larger version

Name:	xthenreg. th.png
Views:	1
Size:	13.1 KB
ID:	1496636




        because with the command " xthreg" we find "thnum" in which we can fix the number of the thresholds

        so here, with this command still not clear how to deal with it.
        hope we get more details about that new command

        best regards
        Sedki

        Comment


        • #5
          Dear River Huang and sedki
          How to install the command first? I tried using ssc install xthenreg and findit xthenreg but i didnt get it. please help me.
          Thank you.
          Ermiyas!
          Last edited by Ermiyas Gebrie; 04 May 2019, 11:22.

          Comment


          • #6
            Dear Ermiyas Gebrie
            you need to install the .xthenreg.ado file from here ( https://sites.google.com/site/myunghseo/research )

            kind regards
            SEDKI

            Comment


            • #7
              Thank you a lot SEDKI for your help.
              Ermiyas

              Comment


              • #8
                Dear all, I have some questions on how to interpret results of the xthenreg stata command:
                1. How to interpret the threshold? I got a Boostrap P-value = 0 which confirm the existence of a non lineartity, but the P_value of the threshold is not significant (0.989) . In this case, I reject these results ?
                2. Which variables are below and under the threshold in these results? ( the signifiance of b and d)
                3. Is the model considers by default that the threshold is endogenous?
                4. If I put in the endogenous' option more than one variable, how the program will recognize which instrument belong to each one of them?
                5. Why when I run the kink static model I got only the results of one regime? what is the difference between the static and the default program?
                Any help?
                kinds regard,
                Fairouz

                Comment


                • #9
                  dear fai ot

                  i hope we get answers about all those questions.

                  just one question for you , which variable we take for the P_value of the threshold ?? x_b or x_d ??

                  best regards
                  sedki

                  Comment


                  • #10
                    Dear sedki zn ,

                    When you use the threshold model, your sample will be diveded into two regimes (below and under the threshold level). Which means that you have to take both of variables (x_b and x_d). But i don't know if b means below and d means under or not .

                    Is your threshold variable endogenous? Because you didn't mention your instruments in the regression.

                    best regards,
                    Fairouz

                    Comment


                    • #11
                      dear fai ot

                      about your question, the output is just a try, first i want that be clear about the threshold levels

                      if we follow your suggestion, like this we will have only 1 threshold and no more. in fact with "xthreg" we can have until 3 thresholds
                      i hope you have more details about that

                      best regards
                      Sedki

                      Comment


                      • #12
                        Dear sedki zn ,
                        Actually all I know that the results gives us (mines also) one threshold " r".
                        I mailed the writers and asked them many questions but I got no response.
                        best regards
                        FAirouz

                        Comment


                        • #13
                          dear fai ot

                          thanks

                          i mailed them too but still no answers.

                          One more question. if i got you well, the P Value of "r" gives us if the threshold effect exists or no ?

                          best regards
                          Sedki

                          Comment


                          • #14
                            Dear sedki zn ,
                            • The boostrap p value for linearity test gives us if there is a non linearity relationship or no. And p>|z| indicatee the variable significance.
                            • One question for you related to this, if i got an existence of nonlinearity, but the"r" is not significant, would I reject the results or what?
                            ps: please chek your inbox,

                            best regards,
                            Fairouz

                            Comment


                            • #15
                              I would like to know the interpretation of the results further. I am also using Dynamic panel threshold regression in one of my projects.

                              Comment

                              Working...
                              X