Given a utility function, $U_{nj} = V_{nj} + \varepsilon_{nj}$, it makes sense that we can find the probability the decision maker $n$ chooses alternative $i$ as:

$$Pr(U_{ni} > U_{nj} \forall j \neq i)$$ $$=Pr(V_{ni} + \varepsilon_{ni} > V_{nj} + \varepsilon_{nj} \forall j \neq i)$$ $$=Pr(\varepsilon_{nj} - \varepsilon_{ni} < V_{ni} - V_{nj} \forall j \neq i)$$

I am having an issue understanding the next step though, where:

$$Pr(\varepsilon_{nj} - \varepsilon_{ni} < V_{ni} - V_{nj} \forall j \neq i)$$ $$= \int_{\varepsilon} \text{I}(\varepsilon_{nj} - \varepsilon_{ni} < V_{ni} - V_{nj} \forall j \neq i)f(\varepsilon_n)d\varepsilon_n$$

Specifically, I am having trouble wrapping my head around why the joint density function, $f(\varepsilon_n)$, comes into play. I understand that this is one of the crucial aspects to understanding discrete choice models, since the choice of this joint density function underlies what model is to be used. Hence, I would like to gain a better understanding of why it is in here. If anyone would maybe have an example or a dumbed-down explanation of why this integral is what it is, I would appreciate it.

  • 1
    $\begingroup$ Did you mean $\forall j \ne i$? $\endgroup$
    – Herr K.
    Apr 14 at 17:09
  • $\begingroup$ Yes I did, thank you $\endgroup$
    – jl_1995_95
    Apr 14 at 17:12

This has nothing to do with any specific model. For any event $A$, let $I_A$ be the indicator function such that $I_A(\omega)=1$ if $\omega\in A$ and $I_A(\omega)=0$ if $\omega\notin A.$

Then $\mathbb{P}(A)=\mathbb{E}[I_A],$ and here the expectation is given in terms of a density function.


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.