Announcement

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

  • Clustered standard errors in -cmp-

    Hi everyone,

    I want to run an ordered probit model with instrumental variables using the user-written command -cmp-. I found the standard errors remain the same when I included the option vce(cluster idcode). This seems not just happen to my specific model. Here is an example:

    Code:
    webuse laborsup, clear
    gen hhid = ceil(runiform()*50)         
    
    cmp (other_inc = fem_work) (fem_work = kids), indicators($cmp_cont $cmp_probit) quietly
    cmp (other_inc = fem_work) (fem_work = kids), indicators($cmp_cont $cmp_probit) quietly vce(cluster hhid)
    When I tried -ivregress 2sls- in this example, -vce(cluster hhid)- did help me get the clustered standard errors.
    Why does this happen? I think I may miss something important. Is there a way to get the clustered SE with -cmp-?

    Thank you very much!

    Best regards,
    Jie



  • #2
    Oh, that is not good. It looks like I introduced a bug in version 8.2.3, on July 13. I have just fixed it. Please do the following to get the latest version:

    Code:
    net from https://raw.github.com/droodman/cmp/v8.2.6
    net install cmp, replace
    I'll have this posted on SSC.

    Comment


    • #3
      Fixed. Thank you very much, Roodman!

      Best,
      Jie

      Comment


      • #4
        Kit Baum has posted the fix to SSC, so "ssc install cmp, replace" will also fix it now.

        Comment

        Working...
        X