Simple estimators for cross price elasticity parameters with product differentiation: panel data methods and testing
Palabras clave:
Panel Data, Fixed Effects, Chamberlain, Nested Logit, Random UtilityResumen
This article shows how to use simple Panel Data methods to consistently estimate demand parameters that come from the Random Utility model where products are differentiated. We followed Berry (1994) and obtain the Nested Logit specification and show that for a sample of information from the market of beer in Peru, the Random Effects (GLS) assumption in Panel Data is strongly rejected even when using Instrumental Variables. Yet, we test whether the alternative Fixed Effect (FE) model, estimated with the Within Groups estimator with and without Instrumental Variables, provides a correct specification. We show that preliminary evidence favors the FE model using the Angrist and Newey (1991) approach to test Chamberlain (1982) over-identifying restrictions implied in the FE specification