There are quite a lot empirical research based on discrete choice models, in which the consumer selects one of J alternative goods to maximize her indirect utility. The key assumption of these models is that there is an idiosyncratic utility component that follows the extreme value distribution.

The problem I am facing is different. I can observe a large number of individuals' consumption of a single good. The values of these choices are continuous quantities, with a long right tail, and roughly 80% of these values are zero. Even if I can discretize the quantity, I don't think I can assume an indepdent utility term for each discretized choice. So a different modelling approach is needed.

I conjecture the choice problem to consist of two steps. In the first step, the consumer decides whether to consume the good. If yes, the consumer then chooses how much to consume. In the empirical IO literature, is there any established model that is used for such "binary-continuous" choices?


It may not be a 100% match (and the details are a bit fuzzy in my memory), but I believe the model would be similar to Ericson and Pakes (1995).

In their model, there are $N$ firms who make sequential choices

  1. Whether to enter the market or not (in your case, to consume or not), which is a discrete choice.
  2. How much to invest if they have entered (in your case, how much to consume), which is continuous.

This is a seminal paper in dynamic discrete choice so I feel there should be some papers along this line.

| improve this answer | |
  • $\begingroup$ Thanks! This paper's seems more complicated than what I need. I only need a static binary-continuous choice framework that can be readily taken to data for parameter estimation. $\endgroup$ – David Xiaoyu Xu Jun 27 at 2:30

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.