2SLS with interaction term

[Jord Sluis]

Dear reader,

My study is about testing the relationship between ESG scores and the corporate financial performance of companies as measured by ROA. Where ROA is my dependent variable and ESG independent. For robustness checks I have also run the 2SLS with IV (the ESG_average)

I have run the 2SLS first for the following regression without interaction term as shown in the following command:

ivregress 2sls ROA Size Leverage sales_growth PPE Industry_n year (ESG = ESG_average), robust

Second, I also test whether legal legislation (Law) as a dummy variable also has an impact on ESG score. So, I use the legal legislation with ESG as interaction term in the regression as follows:

reg ROA i.Law##c.ESG Size Leverage sales_growth PPE i.Industry_n i.year, robust

The problem is that I am unable to run the 2SLS including the interaction term...

already tried the following:

ivregress 2sls TQ ESG Law Size Leverage sales_growth c.ESG##i.Law PPE Industry_n year (ESG = ESG_average), robust

but it keeps saying "ESG included in both exogenous and endogenous variable lists"

Anyone who knows how to run the 2SLS with this particular interaction term?

Thanks in advance!

[Jeff Wooldridge]

You need to treat the interaction as endogenous, too. I don't think Stata allows factor notation when it involves an endogenous variable because it's a dependent variable in the first stage. (I think Stata should allow it, but that's for another day).

The following should do it:

gen ESG_Law = ESG*Law
gen ESG_average_Law = ESG_average*Law
ivregress 2sls ROA ESG Law Size Leverage sales_growth PPE Industry_n year (ESG EST_LAW = ESG_average ESG_average_Law), robust

[Jord Sluis]

Dear Mr. Wooldridge,

Thank you for you quick response. However, I still get the following error: "ESG included in both exogenous and endogenous variable lists". Do you know how to solve this error?

Thanks in advance!

[Jeff Wooldridge]

Oops. You should drop the first occurrence of ESG. And correct my other typo:

gen ESG_Law = ESG*Law
gen ESG_average_Law = ESG_average*Law
ivregress 2sls ROA Law Size Leverage sales_growth PPE Industry_n year (ESG ESG_LAW = ESG_average ESG_average_Law), robust

[Jord Sluis]

Dear Mr. Wooldridge,

Thanks for your help! It works now.