“Pecking Order of Subscriptions” (Draft coming soon)
Streaming has made media markets look less like a choice of one provider and more like a portfolio: consumers watch several services, subscribe to some, and tolerate ads on others. I develop a framework in which consumers with finite attention jointly decide how to divide their viewing time across competing firms and how much to spend on subscriptions that reduce ad nuisance. Firms earn revenue from both advertising and subscriptions and choose ad rates and subscription prices. I characterize optimal attention allocation and show that subscription spending follows an endogenous pecking order: consumers’ optimal subscription behavior is characterized by a hierarchy that ranks firms by weighingcontent quality against ad burden and price. I explore several variations of this framework that incorporate platforms as a new class of agent, vertical and horizontal differentiation, and ads embedded within content. In a two-firm Hotelling specialization, consumers sort into distinct subscription regimes, with diminishing returns to total content quality generating interior regime boundaries. When a strategic firm competes against an exogenous public broadcaster, its optimal ad rate is interior, because pushing consumers fully into subscriptions sacrifices ad revenue from non-subscribers.
“Contemporary Health Consequences of Atomic Testing in the Semipalatinsk Nuclear Polygon” (with Charles Becker and Sultan Muratov)
Kazakhstan’s Semipalatinsk nuclear polygon witnessed over 450 surface, air, and underground nuclear tests between 1949 and 1962, with underground testing continuing through 1989. Health consequences for the heavily-populated region have been studied extensively. However, the ongoing scope of the effects remainsunexplored: we address this gap. Using rich and detailed district-level population health data from the Kazakhstan Ministry of Health, we find large and persistent effects on a wide range of health conditions many decades after explosions ceased. Matching models link treated and untreated districts with similar conditions using a carefully constructed satellite night-light data series to capture economic characteristics.