Possibly a very trivial question, however I am somehow unable to find any mathematical definition of this model. From other posts on this website I perfectly understand the difference between an unordered multinomial probit model and an ordered multinomial probit model, however I cannot seem to find any econometric definition of the first. Clearly, whether the model is alternative-specific or not matters little to me in this case; this specification only means that I will have one less assumption, that is, that the error terms do not have to be uncorrelated, and this I know how to write down.

I have seen definitions of the ordered multinomial probit model, one of those is as follows:

Consider a latent random variable $y_n$ for individual $n = 1$, ..., $N$ that linearly depends on $x_n$.

$y^*_n = x^{'}_n\beta + \epsilon_n$ with $\epsilon_i $ a normally distributed random variable with mean 0 and variance $\sigma^2$.

Is the unordered definition at all similar to this? Is there a similar way of writing it?


Let alternative $j$ have indirect utility $v_j$ for $j=1,...,J$ and let $\epsilon =(\epsilon_1,...,\epsilon_J)$ be multivariate normal with covariance $\Omega$ then choice probability is $$Pr(j=\arg \max_h \{v_h + \epsilon_h\})$$

defines choice probability of agent choosing alternative $j$ in a multinomial probit model. See page 111 of Kenneth Trains book Discrete Choice Methods with Simulation here

  • $\begingroup$ Ah ofcourse, using utility makes loads of sense. $\endgroup$ – Anon Sep 24 '19 at 23:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.