Announcement

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

  • Psmatch and questions about it

    Hi,

    I would like to investigate the effects of a natural disaster on individuals. Therefore I have been using a difference-in-difference method. My main regression so far is as given below:

    Code:
    xtreg outcome_var FY2007 FY2008 FY2009 FY2011 FY2012 FY2013 didFY2007 didFY2008 didFY2009 didFY2011 didFY2012 didFY2013 if inrange(wave,2008,2014), fe
    This was I was controlling for individual fixed effects (thorugh
    Code:
    ,fe
    ), year fixed effects (through
    Code:
    FY2007 FY2008 FY2009 FY2011 FY2012 FY2013
    , and I was looking at the breakdown of the treatmenteffect (through
    Code:
    didFY2007 didFY2008 didFY2009 didFY2011 didFY2012 didFY2013
    , where each
    Code:
    didFY2007=treatment*financialyear2007
    .

    Now as an alternative way I want to use propensity score matching and want to use the
    Code:
    teffects psmatch
    code. Regarding this I have two questions:

    1. I have read the manual for the function but I am not sure what the function does. I have a treatment group (
    Code:
    intreatmentgroup==1
    ) and a control group (
    Code:
    intreatmentgroup==0
    ). I want the function to match each individual from the treatment group with one individual from the control group. Therefore when I specify the function as below is this the case?

    Code:
     teffects psmatch (outcome_var) (intreatmentgroup cleanedhigh1 cleanesempst hgsex hhtype cleanhstenr,probit)
    2. Does the above code mean that the function does the matching based on the variables
    Code:
    cleanedhigh1 cleanesempst hgsex hhtype cleanhsten
    ?

    3. The shock has been observed in 2010. Therefore a in my difference-in-difference estimation I can look at the effects before and after the shock through
    Code:
    did=treatment*post
    where
    Code:
    post
    is for 2010 and later. How can I do that here? Can I do that here?

    4. Is it possible to control for fixed effects using
    Code:
    teffects psmatch
    , i.e. I want to control for individual as well as year fixed effects? Or is the control for individual fixed effects already done through the matching process? In that case how can I control for year fixed effects?

    I know those are many questions but I have now been reading for 2-3 weeks about this and I still could not solve this issue. Any help or answer to any of the questions would be highly appreciated..

    Kind regards.


  • #2
    You have accidentally posted your topic in Statalist's Mata Forum, which is used for discussions of Stata's Mata language. Your question will see a much larger audience if you post it in Statalist's General Forum, as you have in the past.

    Comment


    • #3
      Thank you for informing me about that. I will immediately do that..
      Best.

      Comment

      Working...
      X