2sls

[Will OLeary]

Hi guys,

I am testing to see if there is a correlation between stricter environmental policy of a country on FDI outflows of USA industries. I want to run a 2SLS where the first regression includes instrumental variables like corruption, government effectiveness, political stability and regulatory quality of the country across 2009-2021 on ESS index (environmental sustainability stringency). However, the F stat I get from this regression is 0, does that mean that I cannot use these instrumental variables in the regression as they are not strongly correlated with the ESS?

In the second stage my aim was to regress the estimate for ESS with trade openness, lagged GDP, GDP per capita, Education and Labour prod to see the effect of ESS on FDI outflow.

Thanks
Will

[George Ford]

just run the thing with ivreg2 and it will give you all the stats about the quality of the instruments. if your F is small (0 would be weird) in the first stage (estimated separately) it will be bad. more IV is not always better. start small.

using lagged GDP and GDPpercap is basically including GDP and the inverse of population as regressors.