Dear All,

I am estimating the willingness to pay for insurance over 5 years, whereas the insurance consists of three different products which are increasing in its premium and coverage/payout.
Thus, the outcome variable "purchase" ranges between 0 and 3 and can be looked at as count data. This leads me to the idea of a 2-stage model, such as the Heckman or Hurdle approach.

Some info on my data: Different to many count data, I do not have the problem of excess zeros and there is not any person that never bought an insurance in the 5 observed years:

Array

Code:
tabstat I_options, stats(mean v n)
[CODE]
variable mean variance N
I_options 1.888 1.154103 625
[CODE]
As the variance is smaller than the conditional mean, this may be a sign of underdispersion.

Next, I want to apply a Heckman as well as Hurdle model to compare results. However, I couldnt find any Stata commands how to estimate the Heckman and logit-poisson Hurdle model for panel data. Does anyone of you know about such commands in Stata? If not, I need to do it manually but was struggling to find a proper command for a truncated poisson command.

Any help in this regard is highly appreciated! Thanks!!