Because nine out of ten lung cancer deaths are attributable to smoking, significant reductions in smoking are likely to reduce lung cancer death as well. It is well known that cigarette demand and consumption are negatively related to price, and that cigarette smokers use price minimization strategies to maintain their tobacco use patterns at a reasonably low cost when prices go up. Two consumer strategies that have received significant attention are legal tax avoidance and illegal tax evasion. These strategies are most common when there is a price differential in an area such as an adjacent state that imposes a relatively lower excise tax on cigarettes. Their effect is a reduction in the intended public health effect of excise taxes that is expected to occur via a drop in consumption, and a reduction in state tax revenues, which may be used to fund tobacco control efforts. An increasingly discussed solution is a minimum price law. If the price is set high enough, among-state price disparities of cigarettes and other tobacco products can be eliminated along with a prominent consumer price reduction strategy. Purpose: In this study, cigarette consumption data are used to inform a novel model of consumption that incorporates the effect of adjacent state price differentials. The model is then used to (1) estimate lost (or gained) revenues by state, as well as (2) expected changes in consumption in a scenario involving a minimum price law for cigarettes that sets a nationwide price of $10 per pack (approximately the average price in New York state in 2014), which would eliminate an among-state price differential, and therefore much of the incentive to avoid or evade taxes. This scenario also raises the price of cigarettes substantially in almost all states. Methods: We use yearly state-level cigarette consumption and price data from the Tax Burden on Tobacco from the years 2004–2014. The developed model is a log-linear regression model that uses latent variables (i.e., random effects) to capture basic price effects and adjacent-state price differential effects in a mixed effects model framework. The latent variables offer a simple means of allowing both price effects to vary by state. We analyze the fitted model in two ways. First, we compare model-based consumption predictions under a regime of existing state price and border- state price differentials with predictions from a regime in which the differential is removed. This comparison results in estimates of state-specific consumption lost (or gained) due to border state price differentials; the estimates of lost consumption are multiplied by state-level excise tax and interpreted as lost (or gained) state revenues. A second analysis compares the current regime to one in which cigarette packs are set at$10 each nationwide to determine the expected consumption reduction. Results: Overall, the effect of price on demand is negative, statistically significant, and well within range of the price elasticity estimates available in the literature. The effect of border-state price differential is also negative and statistically significant suggesting that a state's consumption is negatively related to the difference between its cigarette price and the average price of its neighboring states. Both effects are heterogeneous across states. In the first analysis of the fitted model, calculation of lost (or gained) revenue relative to what would be earned if no price differential existed is examined. The analysis reveals that New York and Illinois are, by a large margin, losing the most yearly tax revenue (nearly $140M each) to out-of-state cigarettes. Other top ranking revenue-loss states in order include Florida, Washington, Minnesota, Massachusetts, Arizona, and Ohio. On the other end of the spectrum, states gaining the most revenue under the current price regime are in order, Pennsylvania, New Hampshire, Indiana, West Virginia, Delaware, Missouri, Virginia, and Iowa. When all state gains (or losses) are summed, the net is a loss at$294.6M nationwide. In a second analysis of the fitted model, state-specific consumption estimates are derived under a regime in which a pack of cigarettes always costs the consumer $10 and in which there is no border price differential. The analysis reveals that the 2014 consumption estimate of approximately 13 billion packs of cigarettes drops to just under 8 billion under the nationwide$10 per pack regime. Conclusions: The analysis results suggest that state excise tax revenues are unfairly distributed due to tax avoidance or evasion behavior, and the net effect is a nationwide loss of almost $300 million in state revenues. This is money that could have been spent by high tax states towards their tobacco control goals, but instead went at a discount to states that have a lower excise tax, and likely weaker tobacco control goals. The analysis also revealed that a nationwide minimum price on tobacco could have a very strong effect on cigarette consumption, cutting out over a third of current consumption. These estimates are drawn from a model fitted to real and recent data. Moreover, the nature of the model allows for state specific idiosyncrasies that may affect price and adjacent state price effects to bear on the results, an approach not seen in the literature to date. However, the calculations involve assumptions that may not be realistic. For example, it is not clear that the price effect will remain the same at all price levels (i.e., the price effect may be non-linear). Also, a minimum price on cigarettes would not necessarily remove price differentials as assumed in the 10$ per pack scenario. Thus, the results of this study are best viewed as somewhat stylized views of what we are losing in the current price regime, and what we could achieve under another.

The following are the 17 highest-scoring abstracts of those submitted for presentation at the 40th Annual ASPO meeting held March 13–15, 2016, in Columbus, OH.