Announcement

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

  • GLM- convergence not achieved

    hello everyone, I am new to statalist. Please pardon if the language is unclear. I am writing me thesis on impact of RCEP RTA and was running the command to estimate the gravity model with PPML estimator using command:
    glm trade EXPORTER_FE* IMPORTER_FE1-IMPORTER_FE98 ln_gdpij contig comlang colony comcol ln_dist rta, family(poisson) link(log) nocons irls robust iter (30) cluster(pair_id).

    At the end I get the result saying, ''convergence not achieved''.

    What does this mean? How to rectify. Any help would be much appreciated.

    Thank you much.

  • #2
    Dear Joao Santos Silva , I would appreciate any insight that I may get on this. Thank you so much for your consideration.

    Comment


    • #3
      Aishwarya:
      the usual recipe to deal with missed convergence is adding one predictor at time and see when Stata starts gasping.
      Kind regards,
      Carlo
      (Stata 18.0 SE)

      Comment


      • #4
        Dear Carlo,

        Thank you very much. I will try that.

        Comment


        • #5
          Dear Aishwarya Nahata,

          You are likely to jave perfect predictors; please use the ppml command rather than glm.

          Best wishes,

          Joao

          Comment


          • #6
            Aishwarya,

            Some of your syntax is contradictory. You invoke both robust and cluster options, but I think that cluster implies robust. Also, the syntax should be vce(cluster pair_id).

            Another issue is that you limited iterations to 30 (with the iter(30) option). Why did you do this? If you did this because you had an infinite iteration log, then you can still inspect the output to see if anything was going wrong with any particular coefficient - look for missing standard errors.

            From what you typed, it looks like you may have up to 98 variables, possibly fixed effects for something. I'm not sure if that is a problem, but is it necessary?
            Be aware that it can be very hard to answer a question without sample data. You can use the dataex command for this. Type help dataex at the command line.

            When presenting code or results, please use the code delimiters format them. Use the # button on the formatting toolbar, between the " (double quote) and <> buttons.

            Comment

            Working...
            X