by
Xing Li, Guanghua School of Management, Peking University
Wednesday, October 23, 2019 | 2:00pm-3:30pm | Room 335, HSBC Business School Building
Abstract
We document that identification strategies exploiting cross-border differences in treatment are a variant of the preference externality estimator more recently developed in the industrial organization literature. Waldfogel (1999) coined the term preference externality to describe how the aggregate tastes of heterogenous consumers can influence the products made available to one another within a common market. The externality forms the basis for an instrumental variable estimator where, after conditioning on observed preference determinants for a focal consumer type, the aggregate observables within the market, which vary by the preferences of other types, influence the focal type’s “treatment” but are excluded from the focal type’s outcome equation. Variation in treatment across geographic borders similarly arises from an externality where otherwise comparable individuals near a border face different policies because of different externalities from their respective aggregate regions. We use an advertising application to compare the border and IV implementations of preference externality estimators across three dimensions: i) identifying assumptions, ii) sacrifices in statistical power, and iii) local estimates of heterogenous effects.
