Discrete choice models and continuous demand systems in the scanner data age

The growing availability of high-frequency scanner data, collected either at the retail or at the household level, has brought new perspectives and challenges to the estimation of empirical demand models. This chapter reviews the state-of-the-art and perspectives of demand systems and discrete choice models in order to compare them in relation to five key theoretical and technical challenges: (1) dealing with zeroes, that is, non-purchases and corner solutions; (2) product aggregation and category-level policy analysis, quality choices and price data; (3) endogeneity of prices and expenditure; (4) habits, dynamic models and intertemporal optimization (for example, stockpiling); (5) consumer observed and unobserved heterogeneity and tastes.