Dear Statalists,
I am currently writing my ba thesis on the effectivness of Environmental Policy using a emission dataset including 61 Countries over 17 years, following the paper https://ideas.repec.org/a/eee/jeeman...3p336-354.html.
In oder to obtain a Policy effect and correct for possible self-selection of the treatment I am using an Difference in Difference approach with Instrumental Variables using the Xtivreg2 command.
I collapsed my data into a pre- and post-treatment periods around the policies ratification and assigned a treatment dummy to be 1 in the post-treatment period to the countries that ratified.
Following their approach my model takes the specification:
D. Emissions = a + ß* D.Kyoto + ß2* D.ControlVars + e for t=2
Where D. denotes the first-difference operator, a is a common time trend for the treatment and control group, Kyoto is a dummy coded as 1 for countries that ratified the protocol in period t and 0 otherwise, ControlVars are a multitude of variables incl.: Population, GDP, Polity...

To obtain a first outlook I am using a simple FD-OLS regression:
Code:
reg d.(Log_PROD_CO2 kyoto Log_pop Log_rgdpo), r
However I am not sure if I have to supress the intercept to interpret ß a my policy effect?
I am thinking about this as the intercept would describes the change in emissions (obvious emission growth) when Kyoto=0 and ß therefore overstates the emission reducing effect of ratifying the Kyoto protocol.