hi Stata people,

I would like to have your expertise on a problem!! I am analyzing the demand estimation ( two-level nested logit model) and merger simulation for the yogurt industry for three years (2007-2009).
I have everything correct when I run the demand estimation, but when it comes to merger simulation, although I run the same codes for three years, for both years 2008 and 2009 I have correct results ( alpha negative, sigma 1, and sigma 2 between zero and one and sigma 1 bigger than sigma 2) . so far so good. But when it comes to 2007, I have positive alpha, but my sigmas are correct.

Why is that? is there any problem in the data structure or should I fix sth?

I would appreciate any answer.