The Micro-Geography of Tax Avoidance: Evidence from Littered Cigarette Packs in Chicago

This paper describes a simple but novel technique for assessing cigarette tax compliance: inspection of littered cigarette packs to determine whether legally required taxes have been paid. In addition to serving as a measure of tax compliance, this approach is also a potential tool for assessing the impact of factors such as cigarette tax increases and tax disparities in adjacent areas on tax avoidance. Between January 2002 and the summer of 2007 at least 36 states and the District of Columbia increased their cigarette tax rates (Goolsbee, Lovenheim and Slemrod 2007) in an effort to discourage smoking and raise revenue. Such tax increases may result in decreased consumption, but may also produce other behaviors such as use of lowerpriced cigarettes, hand-rolling cigarettes from loose tobacco, or smoking a smaller number of cigarettes more intensively. Tax increases have sometimes created large tax disparities in neighboring communities. For example, in July of 2007 the City of Chicago had a combined state and * Institute of Government and Public Affairs, Department of Public Administration and Department of Economics, University of Illinois Chicago, 815 W. Van Buren Street, Suite 525, Chicago, Illinois 60607-3506. dmerrim@uic.edu. Robert Wood Johnson Substance Abuse Policy Research Program (SAPRP) provided financial support. Gilbert Bassett, Fred Bryant, Christina Ciecierski, Micheal Lovenheim, Hana Ross, Joel Slemrod, John Tauras as well as two anonymous reviewers provided useful advice at various stages of the project. Many government officials also cooperated as explained in the text. Seminar participants at the University of Illinois and Northwestern University and National Tax Association annual meetings also provided helpful comments. Numerous assistants worked on the project and cheerfully and patiently engaged in a large variety of unorthodox tasks to collect and code the data reported here. Their names are Eishita Shah, William Barrett, Muhammed Zamar, Istvan Fekete, Lucia Samayoa, Amosh Noviandri, Maura Mahler, Jami Eddington, Robert Madeja, Vinca Merriman, Kyle Peyton, William Pokin, and Jeffrey Wytaniec. local tax of $3.66 per pack, while neighboring Indiana had a state tax of only 55.5 cents per pack and no local cigarette excise taxes. The tiny state of Rhode Island had a tax of $2.46 per pack while its neighboring states of Massachusetts and Connecticut had taxes of only $1.51 per pack. Oklahoma’s tax of $1.03 per pack was 86 cents higher than its border state, and Missouri and Montana’s tax of $1.70 was $1.26 higher than adjacent North Dakota. Non-uniform taxes heighten the possibility of tax non-compliance (failure to pay the applicable tax in the location in which the pack was consumed) and avoidance (behavioral change designed to reduce tax liability). Economic theory predicts that tax avoidance will vary with the absolute level of taxes, tax disparities and the cost of avoidance, such as distance to lower tax sources. High taxes and tax disparities increase the incentive for organized smuggling, tax evasion at the point of sale, and cross border shopping in neighboring lower tax areas. Cigarettes represent a useful laboratory for the study of tax avoidance behavior, given their relatively high taxes, easy access to low-tax or no-tax cigarette sources such as Native American reservations or the Internet, and a relative lack of enforcement efforts. The literature on tax compliance (see Andreoni, Erard and Feinstein, 1998, and Slemrod, 2007 for reviews) often has focused on avoidance or evasion of federal taxes. Slemrod (2007), for example, calculates an aggregate tax gap, defined as the federal income tax that is not collected on personal income, equal to about 14 percent of the revenue collected from federal personal income taxes. In contrast, there have been few studies of the effects of the cost of avoidance on behavior. The geography of cigarette taxes provides an exogenous source of variation in avoidance costs, allowing for a clean estimate of the relationship between tax avoidance and its cost. The approach used in this paper leverages the fact that tax compliance measures can be visually identified on many cigarette packs. In all states except North and South Carolina, wholesalers must affix state tax stamps to cigarette packs before distributing them to retailers to indicate that taxes have been paid. When additional local taxes are imposed the state stamp is altered or an additional local tax stamp is added. As a result, the origin of purchases is observable on littered packs. Working in the “unobtrusive measures” tradition of Eugene T. Webb et. al. (1966), the author organized teams to collect littered cigarette packs in a representative random sample of areas in the city of Chicago and neighboring jurisdictions. A separate survey of appropriately disposed cigarette packs from some of the same areas was also conducted. The tax stamps from the collected packs were examined to determine the extent of non-compliance. In principle this data collection technique provides information about avoidance behavior such as organized smuggling, out-right evasion and cross-border shopping1. Of course, some non-compliance would be expected even if cigarette tax rates were uniform, since many people routinely cross tax borders in the normal course of their activities. Avoidance can be inferred from greater than random non-compliance levels. This is discussed in more detail later in the paper. This study revealed a high degree of noncompliance, with three quarters of Chicago packs not displaying the Chicago tax stamp, and compliance increasing with distance to low-tax borders. This paper examines recent literature on cigarette tax avoidance as well as the rationale for using littered cigarette packs as a possible measure for this behavior. It then describes the sampling, data collection, and data coding procedures employed in this study, followed by summary results and detailed regression data. Finally, it explores the evidence for littered packs as a measure of consumption as well as overall conclusions.