Dear forum,

I follow the paper written by Björnerstedt and Verboven (2013) and Berry (1994) to simulate a merger:

In Berry (1994) they use a nested logit model to estimate demand. Brörnerstedt & Verboven do a merger simulation based on the nested logit model to estimate demand and simulate a merger. They work out an example with commands you need for stata. I want to base the method of my masterthesis on this paper, but I am stuck with a problem of price endogeneity.

On page 10 Börnerstedt and Verboven do a fixed estimates regression instead of an IV regression (to simplify). This is ofcourse not the correct way, because we have to account for price endogeneity. Following the paper from Berry (1994), like they do in the paper, I would like to do an IV regression in stata using product characteristics as instruments. Berry states that you have to invert the function defining the market shares to uncover the mean utility levels of various products. These mean utilities can then be related to product characteristics and prices using instruments.
  1. Can it be explained why this would be a good instrument?
  2. Which commands do I need to make this IV regression with the right instruments work in stata?
Berry 1994: Estimating Discrete-Choice Models of Product Differentiation (jstor.org)
Börnerstedt and Verboven 2013: Bjornerstedt&Verboven_StataJournal.pdf (aguirregabiria.net)

I hope my questions are clear and that someone knows the answer.

Kind regards,
Caro