Heckman method

[Serena Menny]

Hello everyone,


First, I made probit regressions to determine what drives an investor to invest responsibly. Now, I would like to understand what explains the sustainable investment size. For that, I am using the Heckman method. However, I found some difficulties in choosing the variables. Specifically, I am still determining on what basis I should choose the select variables (first step).

• I previously found that many variables explain the decision to invest. Should I include all of them? Should I also include the control variables?
• In the second step, should I include the same control variables?
• To interpret the results, should I re-interpret the first step? (even though I explained the sustainable investing determinants previously)

I really appreciate any help you can provide.

[Maxence Morlet]

I would say yes to the two first questions, and the answer to your third question is quite field-specific, it depends on your paper, its readership, etc.

For the obtention of correct standard errors, I would recommend using the latest community-contributed heckman command (not sure I know what it is, I'm just quite sure they implement necessary standard error adjustments).

[Serena Menny]

I would say yes to the two first questions, and the answer to your third question is quite field-specific, it depends on your paper, its readership, etc.

For the obtention of correct standard errors, I would recommend using the latest community-contributed heckman command (not sure I know what it is, I'm just quite sure they implement necessary standard error adjustments).

Maxence Morlet thank you so much for your answer!

[Serena Menny]

Maxence Morlet After looking for the command that you recommended, I found that standard error correction is internally done with Heckman two-step. So apparently, I don't need to make any more adjustments