Estimating random coefficients logit demand models using aggregate data

Discrete choice demand models are popular in applied analysis and can be estimated using market-level data on product shares and characteristics. The random parameters logit model is an extension to the traditional specification and can accommodate heterogeneity in consumer preferences and rich patterns of substitution over a large number of products. The purpose of this presentation is to set out a Stata program that estimates the parameters of this model by using the algorithm proposed by Berry, Levinsohn, and Pakes (Econometrica, 1995) and that can also address the potential issues of price endogeneity. The estimator is coded in Mata and involves an inner-loop contraction mapping to invert the market shares, followed by an outer loop search over the parameters that minimizes a GMM objective function. The estimator allows the user to specify the variables that have random parameters and contains an additional option to generate a matrix of own and cross-price elasticities of demand. The blp routine is available from the SSC Archive.