3 May
2006
3 May
'06
6:49 p.m.
you might want to highlight the unusual fact that you're using A&G's
method and also working on their substantive problem, but in a different
area.
Gary
On Wed, 3 May 2006, Dan Hopkins wrote:
Definitely a cool project. The title asks a question;
you want it to make
an argument, so that someone skimming a table of contents could get a
sense of what conclusions you draw. Something like "Costless Conflicts"
would help communicate your point.
Also, I think the "while at first glance" clause and the information that
follows immediately from it is a bit subtle for an abstract--if there is a
gap between the factual and counter-factual, why isn't that gap the causal
effect? I wasn't quite clear on that--is this a liability in the method
more generally? Also, I'm not sure you get much from the final sentence
about how your result differs from Abadie and Gardazabal's from another
country--although if you highlight a bit more any methodological
improvements you guys offer, you might mention them there.
Best,
Dan
On Wed, 3 May 2006, Daniel Fetter wrote:
The Economic Costs of Conflict? A Case Study of
Punjab
We assess the economic costs of the secessionist conflict in Punjab by
applying the synthetic control group method of Abadie and Gardeazabal
(2003) to data from India, creating a "counterfactual" Punjab without
terrorism as a weighted average of other Indian states. In doing so
we demonstrate the usefulness of placebo analyses ? creating synthetic
controls for states in the original control group ? in inferring
causal effects with only one treated observation. While at first
glance our results show a gap between the actual and counterfactual
GDP growth paths, we find that this gap is comparable to the gap
produced by running placebo analyses on states that did not experience
terrorism. Hence we cannot infer that terrorism had a strong negative
impact on growth in Punjab. This stands in contrast to Abadie's and
Gardazabal's results from Spain, where the GDP gap for the Basque
Country during its period of terrorism exceeds those generated from
placebo analyses on other regions.
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l
_______________________________________________
gov2001-l mailing list
gov2001-l(a)lists.fas.harvard.edu
http://lists.fas.harvard.edu/mailman/listinfo/gov2001-l