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

  • Adding new vce options to existing command?

    Dear Statalist,

    I am currently working on a project that incorporates Regression Discontinuity in Time, where time is in addition a discrete variable.

    Because of the high likelihood of auto correlation in the error term, as well as the discrete nature of the running variable, I would need to incorporate HAC standard errors in the model.

    In this, I have two questions:

    1. I have been reviewing the code for the user-written command "rdrobust" and I notice that they denote the vce(types) in the same numbering as, for example, the conventional "regress", i.e., "robust", "hc2", "hc3". Does this mean that there are any built-in functions in STATA that allows one to call upon vce types? And, if so, is there such a function for Newey-West SEs?

    2. Having little to no experience with MATA, could the incorporation of further VCE options to an existing be considered a demanding task? (i.e., is this a reasonable path, or should I try to work around it in some way).

    Johan Karlsson

  • #2
    For starters, you could look at the documentation for the _robust-command
    help _robust
    and then the PDF documentation of it. This might give you some ideas on how to proceed.