João Granja

I study the effect of universal service regulation on the speed of roll-out of new mobile telecommunications technologies. I develop a dynamic game of entry and technology upgrade under regulation, which I estimate using new data on mobile technology availability in Brazilian municipalities. Counterfactuals show that the regulation accelerated the introduction of 3G technology to lower-income, rural municipalities by just under 1 year, on average, while reducing firms’ aggregate profits by 10%. The regulation can act as a commitment device and marginally deter technology deployment by unregulated firms. A subsidy auction achieves similar acceleration of 3G deployment at 21% of the cost, likely increasing welfare.

We revisit the estimation of dynamic games with continuous control variables, such as investments in R&D, quality, and capacity. We show how to use the recursive characterization of Markov Perfect Equilibria to develop estimators that make full use of the structure of the model. Our estimator resembles an indirect inference estimator, albeit in a two-step procedure that is common in the estimation of dynamic games. We use Monte Carlo experiments based on an empirically-relevant model of investment in R&D to compare the performance of our estimator with alternatives. We find that our estimator outperforms the commonly-used inequality estimator of Bajari, Benkard, and Levin (2007) and a naive implementation of an estimator based on recursive equilibrium conditions.

Product portfolios have a direct effect on prices via optimal pricing decisions and also an indirect effect because they influence retailers’ bargaining positions, and thus the wholesale prices retailers are able to procure. I study the effects of characteristics of retailers’ product portfolios, in particular their offerings of store-brand products, on the retail prices of national brands. I propose a Nash-in-Nash model of wholesale and retail price determination, which I estimate using IRI scanner data. I use the estimated model to simulate a counterfactual in which I eliminate store-brand products and to quantify the welfare effect of double marginalization. I find that the presence of store-brand products decreases the prices of national brands by about 1%, and that the elimination of double marginalization leads to substantial consumer welfare gains.

We develop tools to test between common competing specifications of the Nash-in-Nash bargaining framework, and apply those tools to data from the Washington State legal cannabis market.

Many passenger transportation markets are characterized by low load factors, which suggests large inefficiencies. I quantify these inefficiencies in an environment in which firms compete on their schedules as well as prices. To do so, I use new and detailed data on firms' schedules and sales, collected in partnership with the Brazilian transportation regulator.