Maximum Likelihood Estimation

[kcchogou]

Diapositive1.JPG (54.42 KiB) Viewed 10838 times

Diapositive2.JPG (73.03 KiB) Viewed 10842 times

[trubador]

Dear kcchogou,

Your problem can be specified in EViews as a LogL object. I've written a few lines of codes for your model. However, as I understand, the parameters you want to estimate are probabilities and variances. Depending on the properties of data, additional assumptions/tricks may be needed. Therefore, I'd like to go over the code and check whether it is properly working before sharing it. If it is not a problem for you, I will kindly ask you to send/share your data (or just a small sample of it). The model might be more difficult than it looks.

[kcchogou]

Dear trubador

Thank you for your feedback. I will send you my database. Send me your email please. My mail is kpenavoun@yahoo.fr. Best regards.

[kcchogou]

Dear trubador
Thank you for your feedback. I do agree to share my database. Send me tour email and I will send it to you. My email is kpenavoun@yahoo.fr
Best regards.

[kcchogou]

Dear all
About my preoccupation on Maximum Likelihood Estimation, I know that my problem can be solved via building a constrained optimization model. The basic problem, for instance, is that the probabilities in my model must add up to 1. But I don't know how to build this king of contraints in Eviews. Mr Trubado helps me to have the following code I would want to share in order to have further help from moderators or other users in the forum. Thank you very much M. Trubado.
@logl agric
f1 = (1/@exp(c(3)/2))*@dnorm((ba-batc)/@exp(c(3)/2))
f2 = (2/@sqrt(exp(c(3))+@exp(c(4))))*@dnorm((ba-batc)/@sqrt(@exp(c(3))+@exp(c(4))))*(1-@cnorm(-(@exp(c(4)/2)/@exp(c(3)/2))*(ba-batc)/@sqrt(@exp(c(3))+@exp(c(4)))))
f3 = (2/@sqrt(@exp(c(3))+@exp(c(5))))*@dnorm((ba-batc)/@sqrt(@exp(c(3))+@exp(c(5))))*(1-@cnorm((@exp(c(5)/2)/@exp(c(3)/2))*(ba-batc)/@sqrt(@exp(c(3))+@exp(c(5)))))

agric = log(@logit(c(1))*f1+@logit(c(2))*f2+(1-@logit(c(1))-@logit(c(2)))*f3)

@param c(1) 0.1 c(2) 0.1 c(3) 10 c(4) 10 c(5) 10
NB. ba is the absolute value of the differnece prices between Bohicon (one region) and Azove (second region) and batc is the transfert costs.
Thank you for your help.
Kcchogou