Demand estimation in industrial organization
For those in search of economics references, I prepared some notes on demand estimation in industrial organization. They survey a selected sample of discrete choice models, from multinomial logit to modern Berry, Levinsohn, and Pakes–type methods.
While originally written as a study guide for a graduate-level industrial organization class, they can also serve as a high-level introduction for anyone who wants to use demand estimation.
The notes discuss conceptual and technical points, such as:
- Differences between product-space and characteristics-space models
- Interpreting the multinomial logit model
- Bridging traditional demand estimation and industrial organization methods, with unobserved quality
- Understanding the random-coefficients demand model behind BLP, and the MPEC and nested fixed-point estimators
- Extensions to standard methods, including adding micro data (Micro BLP) and incorporating a supply-side model
- Econometrics of the BLP estimator
- Remaining difficulties, and some applications (including merger simulation)
Please contact me if you have any comments or suggestions – or just to let me know if you’ve found this summary useful.
(Notes last updated April 2019.)