Odds ratio after nested logit

[Charlie Joyez]

Hello everyone,

I am using a regional nested logit to model foreign direct investment location choices.
This way, I am able to remove (partially) the Irrelevant Alternative Assumption (IAA), that implies a more common conditional logit.

In my regression I use interaction terms, and their interpretation in non-linear models is touchy.
Hopefully, thanks to previous topics here, I know that interpreting interaction terms in logit models is possible by displaying odds-ratio instead of marginal effects (see also Maarten Buis paper here).
This is why I used the ",or" option after my conditional logit.

However, the "or" option is not available when running nested logit. (nlogit command).

So two questions :
1) How to compute the odds-ratio after a nested logit, in order to interpret interaction terms in explicative variables?

2)Is there any statistical reason for Stata not to allow the "or" option after nlogit, while it does for most logit related models? ( Because I'd like to understand why it is not allowed, beside to solve my issue.)

Thanks a lot.

[Charlie Joyez]

If anyone interested, I found a solution to my issue :
1) run your nested logit
2) through the outreg command (ssc outreg complete), ask for display of odds ratios (outreg, or)

The worst thing is that this is very simple and I don't know why I haven't though of that earlier.
But I have no answers on why Statacorp doesn't include an "or" option after nested logit, as they do for all kinds of logits models (logit, ologit, clogit,...)

I hope this could help someone else.