Tobit Model

[Prakashani Singh]

Hello all, I want to run Tobit model, in panel data, but I don't understand which type of censoring I should use because there are missing values in my dependent variable but no zero value is there. Then my confusion is whether can I use the lower limit as zero or not also there is no negative value. Also please help me with appropriate commands. Any help will be appreciated thank you in advance.

[Maxence Morlet]

You could recode the missing values as 0s, but the real question is: are you sure they are missing not at random?

Also, the Tobit model, unlike the two-step Heckman, assumes that the selection equation (which determines observability) is identical to the outcome equation; it constrains the two to be the same.

Please give us more information on your data and research question for us to help you further.

[Maxence Morlet]

As an FYI, Johnston and DiNardo (1997) in "Econometric Methods" do a job of explaining the intracacies of the Tobit model, how to run it and when to run Heckman instead.