New cdist package for counterfactual distribution estimation available from SSC

[Ben Jann]

Thanks to Kit Baum, a new package called cdist is available from SSC. To install cdist, type

. ssc instal cdist

The moremata library is required, so you might also want to type

. ssc install moremata, replace

cdist estimates counterfactual distributions using methods suggested by Chernozhukov et al. (2013). The unconditional (counterfactual) distributions are either obtained by distribution regression using logit models or by a linear quantile regression process. Optionally, counterfactual decompositions of group differences can be obtained.

Also see https://github.com/benjann/cdist.

ben

[Ben Jann]

A new version of cdist is available from SSC. Type

. ssc install cdist, replace

to install the update.

The procedure for estimating the logit models in distribution regression has been revised; perfect predictors are now excluded during model estimation (if possible) to improve numerical stability.
ben