Tobit specification test

[Ellesar1]

Hello,

I have made a tobit-estimation for a variable ab291, that contains "number of visits to doctor last 12 months" - obviously there is a big mass point a 0.

On page 292 of The Help Guide II it says: "The tobit model imposes the restriction that the coefficients that determine the probability of being censored are the same as those that
determine the conditional mean of the uncensored observations."

This is what I want to test.

I follow the instructions, and when I calculate my last equation- the truncated one from the original tobit, all columns from Std.Error and to the right becomes "NA" and I get the message that "WARNING: Singular covariance - coefficients are not unique"

Let me stress that this problem does not arise if I calculate by OLS.

Then I thought that maybe it was due to my sampling (I'm totally new to Tobit) - because the data is coded in such a way that if a person has not answered the question of doctor consultations, the value is "100". So before all my equations I sampled "ab291<88".

I then tried the whole procedure again, where I did not sample ab291, but instead made "ab291real" which assigned "NA" to all values bigger than 88, i.e. indicators for missing answer (due to the fact only half of my dataset was answered this specific question).

However, I got the same problem in the third step/equation; The "Truncated" equation says "NA" all over the place?


Any idea what I can do? I use Eviews 7.