Estimating Complementarity With Large Choice Sets: An Application to Mergers

An Empirical Model of Bargaining with Equilibrium of Fear: Application to Retail Mergers in the French Soft Drink Industry

“…For instance, we find an aggregate own-price elasticity of −1.49 for soft drinks (excluding fruit juices) as in Dubois, Griffith and O'Connell (2020). 47 In line with Dubé (2004) and Ershov et al (2021), we also obtain that own-price elasticities for brands in the cola category range from −2.76 to −3.28 on average. Table 4 displays the own and cross-price elasticity 46 Table 11 in Appendix H displays estimates of the first step of the control function approach.…”

An Empirical Model of Quantity Discounts with Large Choice Sets

“…As first argued by Gentzkow (2007), not accounting for correlation in the unobserved preferences of different products may confound the identification of complementarity and substitutability (Allen andRehbeck, 2019, 2020;Ershov et al, 2021;Fox and Lazzati, 2017;Wang, 2019, 2021;Wang, 2019), stressing the importance of allowing for flexible forms of unobserved heterogeneity in the specification of demand for bundles. In applications with thousands of bundles (like the one we study), the estimation of nonparametric models (Compiani, 2019) or even just of mixed logit models (Gentzkow, 2007;Iaria and Wang, 2019;Liu et al, 2010) may however be prohibitive.…”

“…We implement our methods to investigate the welfare consequences of quantity discounts in the market for carbonated soft drinks (CSDs) in the USA. Using household-level purchase data by IRI for the period 2008-2011 (Bronnenberg et al, 2008), we document that households commonly purchase multiple units of CSDs on any shopping trip (6.6L on average) (Chan, 2006;Dubé, 2004;Ershov et al, 2021) and the pervasiveness of quantity discounts for purchases involving larger quantities (e.g., the average unit-price of a Diet Coke is higher for a 12oz can than for a 2L bottle). 5 We observe that, according to intuition, larger households tend to purchase larger quantities of CSDs, both as multiple units of the same product and as combinations of different products.…”

“…Our methods can enable empirical researchers to scale up the numbers of bundles while allowing for realistic forms of unobserved heterogeneity: demand across multiple product categories involving many products, like grocery or online shopping (Reimers and Waldfogel, 2021;Smith, 2006), mergers in markets with both substitutes and complements (Cournot, 1838;Ershov et al, 2021), mixed bundling pricing strategies (Adams and Yellen, 1976;Chu et al, 2011), (unintended) spillovers of taxes from a product category to others (Allcott et al, 2019;Dubois et al, 2020), and many more.…”

“…Two novel approaches to addressing large choice sets in the estimation of demand for bundles were recently proposed by Ershov et al (2021) and Lewbel and Nesheim (2019). Ershov et al (2021) allow for a very large number of products, but restrict the way they can be combined into bundles (at most two different products, one unit each) and the number of parameters capturing unobserved synergies among products (one per market, the same across all bundles).…”