Hi I have a question about the mlogit command.

As you know, multinomial logit model is estimated by MLE and mlogit command allows the vce(cluster id) option.
Here, I think, when we use the mlogit command with vce(cluster id) option, we cannot use the "usual" form log-likelihood function (the "summation" of log likelihood function of each observation) as the objective function since we are saying that the error terms are correlated within clusters.

So, my question is that when we use the command
Code:
mlogit              , vce(cluster id)
does stata use other kind of likelihood function as the objective function?

Thank you for your time spending to read this question.