Double hurdle model: commands, explanatory variables, and policy effects

[Erin Hua]

Dear all,

I am trying to use a double hurdle model for a cross-sectional dataset that involves two decisions:

1. Participation decision (PD) – Binary
2. Willingness to pay (WTP) – Continuous (Note: WTP can be zero even when PD = 1)

I have a few questions:
1. What commands should I use for this case? Is the dhreg command suitable for such a situation where there are still many zeros in the second stage?
2. Can I use the same explanatory variables in both processes (models)?
3. My main concern is the effect of a conservation policy on both PD and WTP. I assume that variables which affect the outcome variables, but that may also be influenced by the conservation policy, should not be included in the models. Am I right?

I would really appreciate any insights or suggestions.

Best regards,
Erin

[George Ford]

dhreg, I suppose.

probably better to have something in select that's different than wtp (an exclusion variable).

if you put stuff in there affected by the "treatment", then they'll absorb part of the effect. a lot of discussion about that in DID models (covariates affected by the treatment).

probably look something like this:

dhreg wtp_amount x1 x2 x3, select(participation = x1 x2 x3 x4) twostep

[Erin Hua]

Thank you for replying George Ford

Although my model is not DID, I assume the same concern applies here. I'll try that code—thanks for the suggestion!

[George Ford]

maybe. I don't see why what affects participation might not also affect wtp, but this is not my area of expertise. I'm sure there's a lot of discussion about that sort of thing.