Announcement

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

  • dichotomous variable in FE regression with AR(1)

    I am using Stata 14.2 and have a data set I want to analyise. My script is as follows: xtregar y x1 x2 x3 x4 x5 x6 , fe

    Now both x5 is a dichotomous variable classifying the sample into two industry types, x 6 has 4 values but again the value is the same for each company in the sample. When I run the script Stata drops x5 and x6 due to collinearity. In order to test assumptions I also run a normal regress and here Stata does not drop those variables. I am sure that makes total sense to you but could you explain why this is happening?

    Thank you!

  • #2
    Patrick:
    -fe- may be a possible explanation.
    More positive replies are conditional on posting what you typed and what Stata gave you back (as per FAQ). Thanks.
    Kind regards,
    Carlo
    (Stata 18.0 SE)

    Comment


    • #3
      Thanks Carlo,

      I had typed xtregar y x1 x2 x3 x4 x5 x6 , fe and from stata I got back the following:

      note: x5 dropped because of collinearity
      note: x6 dropped because of collinearity

      y
      --------------------------
      x1 -0.304
      (1.79)
      x2 0.018
      (0.05)
      x3 -0.353
      (0.37)
      x4 -0.432
      (0.34)
      x5 0.000
      x6 0.000
      _cons 0.241
      (0.83)
      N 1,255
      --------------------------
      * p<0.05; ** p<0.01
      Thanks,

      Patrick

      Comment


      • #4
        You don't show us your -xtset- command, but from context I'm going to guess that your panel variable is the firm. For any given firm, it is always in the same industry, right? Well, in a fixed effects model, any variable that is constant within panels is colinear with the fixed effect for that panel, and it gets dropped. That explains x5.

        You don't tell us what x6 is, but you say it is the same for every company. In that case, it not only is constant within panels, it is simply a constant in the data set. So, of course, it is colinear with the constant term and gets dropped.

        Comment


        • #5
          Thank you Clyde, that makes sense and answers my question.

          Comment

          Working...
          X