The Fiscal and Health Impacts of Professional and Amateur Cigarette Arbitrage

When governments tax goods sold within well-defined geographic boundaries they often create incentives to arbitrage. Local taxation creates such incentives when the taxes raise the local price of the good and when governments in neighboring jurisdictions levy a lower tax. Basic economic theory predicts that arbitrageurs will reduce the amount governments can raise with a local tax because they will trade across borders. We focus on arbitrage induced by local cigarette taxes. We theoretically and empirically investigate how arbitrage affects local cigarette tax revenues, smoking decisions (initiation, consumption, and cessation) and long-term health. We highlight a previously ignored feature of arbitrage in this market—the presence of both professional and amateur arbitrageurs. That feature is particularly important because we show, both theoretically and empirically, that the effects of arbitrage vary systematically with the extent of each type of arbitrage.

We contribute to a rich literature on the relationship between cigarette consumption and taxes and an emerging literature on the distortionary role of jurisdictional taxes. We introduce and describe a novel, hand-collected dataset of city and county-level tax data from 1920 to present. We combine these data with state and federal cigarette taxes to create the most comprehensive historical record of US cigarette taxation compiled to date. We compute the tax burden for residents in every US ZIP code area over time. Using economic theory, we develop new measures of arbitrage incentives for amateur and professional arbitrageurs.

Finally, we estimate intensive and extensive models of cigarette demand that incorporate the more complete cigarette tax measure and that account for systematic variation in the supply of cigarettes available from amateur and professional arbitrageurs. We use aggregate and individual smoking data to document cross-sectional and temporal variation in the systematic arbitrage of each type. We use longitudinal survey data (PSID, NLSY) on retrospectively reported smoking behavior and health to construct individuals’ complete smoking histories and to estimate whether and how arbitrage-induced smoking behavior early in life predicts long-term health.

Our preliminary results indicate that when one fails to account for professional and amateur arbitrage, the average price elasticity of cigarette sales is severely biased upwards. Only accounting for amateur arbitrage, as is standard in the literature, fails to eliminate the bias. We demonstrate the presence and extent of the bias in cigarette demand studies that ignore local cigarette taxes and tax induced arbitrage.

Finally, we expect that other researchers will be able to apply our method of modeling arbitrage to explain behavior in a wide-range of product markets. Our method is most applicable to markets for products that people can easily transport and for which tax differentials are large across jurisdictional boundaries. Two obvious examples are gasoline and alcohol. Sales tax differences across states provide broad opportunities for arbitrage. A new wave of taxes on sugar drinks provides another opportunity to apply this general method.