BLP not converging - Hessian is not positive seemidefinite

[Bouba Drame]

I am currently running the Berry Levinsohn Pakes random coefficient logit model and I have been getting the following error message;

. blp s_mkt, stochastic(lp) endog(p= llagp2 llagp3 lagp4 v vv w) markets(mkt2) draws(200) elast(lp,15,pro
> duct)
Iteration 0: f(p) = 1.000e+11
Hessian is not positive semidefinite

Has anyone else run into a similar problem? My dataset is composed on 30 markets with 9 products per markets. The markets shares do not show muh variations from one market to the next, could this be the reason why I am getting the error message? I would hope not since realistically one would expect market shares to be similar (not identical of course) from one market to the next and the blp command ought to be able to solve given these similarities. Any help with this would be much appreciated.

[Stephen Jenkins]

Welcome to Statalist!
What is the "Berry Levinsohn Pakes random coefficient logit model" -- please provide full and complete references, using the detail and format requested in the Forum FAQ.
What is the "blp" command, and where did you get it from? Again, please read the Forum FAQ about how to cite the provenance of code that is not built in to Stata.
Finally, please note that Forum FAQ notes a strong preference for members to post using their real names (firstname familyname). So, I recommend that you send a message to the Forum administrators requesting a change -- it's easy, just hit the "Contact Us" button on the bar at the bottom of the screen and post them your request.
Addressing these remarks, and more generally conforming with Forum etiquette (as summarised in the FAQ), will help you maximise the chances of getting responses that are prompt and helpful to you.

[Yu Chenen]

Hi, I am Yu, I met the same problem as you said, and want to ask you that have you figured it out?

Thanks so much!!

[shuhua He]

I ran into the same problem, too. It seems Bouba Drame's concern is valid-- there is not much variance across markets. At least that's what happened to me ><