查看文章

Methods

We use official data from the Israel Ministry of Transport and Road Safety. The data consist of all new cars marketed in Israel between 2007 and 2018. Each observation corresponds to a specific car configuration (year, brand, model name, category, engine size, weight, engine type, safety systems, market price, etc.). Employing configurations with at least 50 sales per year, our dataset contains 4,019 configurations of 1,169,165 cars. We use a differentiated goods econometric model, based on the work of Fershtman and Gandal (1998) and Fershtman et al.(1999). This is a model of an oligopolistic car market containing two equations–demand (1) and pricing (2)–which are estimated simultaneously. To create a realistic substitution pattern between products, we use the nested logit model that assumes that substitution effects between products in this market depend primarily on predetermined car categories. In other words, consumers tend to first choose a car category according to their needs (eg, family cars, mini cars, SUVs) and only then choose a specific car model from that category (eg, Adamou et al., 2014; Greene et al., 2005).