Announcement

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

  • Unparallel pre-treatment period in SCM

    Hi there.

    I am currently writing my dissertation on the effect of Populist Radical Right Parties on economic growth in the EU, from 1990-2022. As part of my analysis, I have performed the SCM on the Fidesz party in Hungary from 2010 as the treatment. My RMSE is low (0.024), but my fit could be improved. Here is my current graph (economic growth is measured by GDP/cap growth).
    Click image for larger version

Name:	hungary.png
Views:	1
Size:	82.0 KB
ID:	1775878


    First of all, I would like to know how bad of a fit this is. It is for an undergraduate dissertation, so I'm not sure on the bar here.

    Secondly, I wonder if the poorer fit is due to my relatively small donor pool: I restrict it to EU countries (N=23 once Italy and Poland are excluded) because I wanted to ensure counterfactuals were meaningful, and then included 6 predictors for macroeconomic, structural, and cultural matching.

    Thirdly, I appreciate that yearly variations in GDP/cap growth could be difficult to match on. Could I use GDP/cap instead?

    Thanks so much in advance.
    Last edited by Madison Towndrow; 12 Apr 2025, 10:56.

  • #2
    Madison:
    the fit is actually not exciting.
    What is the Stata module that you used for your SCM exercise?
    Take a look at Comparative Politics and the Synthetic Control Method by Alberto Abadie et al.,. I found footnote 14 enlightening when I drafted our recent article on SCM (shamelessly: https://pubmed.ncbi.nlm.nih.gov/40117192/).
    Kind regards,
    Carlo
    (StataNow 18.5)

    Comment

    Working...
    X