Tobacco consumption is a leading cause of preventable death in the world, with 100 million deaths attributed to it during the 20th century and nearly a billion deaths projected for the 21st century [1]. In the United States, despite declines in adult cigarette smoking prevalence during the past 50 years, tobacco use remains the nation’s leading preventable cause of death and disease [2].

Increases in unit price of cigarettes are inversely related to smoking [3–5]. Excise tax is one of the primary public health strategies used to increase price [1,6,7]. However, instead of quitting or reducing consumption, smokers may seek lower-priced cigarettes, thereby potentially undermining the public health benefit of tax increases [8–19].

A common misconception is that excise tax is passed homogeneously to smokers at a rate of 1.0, meaning a $1 tax increase is assumed to raise the retail price by $1 across all smokers. This is also known as the ‘pass-through rate’. The majority of the existing literature, with the exception of two papers [20,21], suggests that the overall pass-through rate of excise taxes in the United States is greater than or equal to 1.0 (also known as over-shifted) [20–27]. In addition, consistent with economic theory, global evidence has also shown that cigarette or alcohol excise taxes are not passed homogeneously to all users [20–26,28–31]. The pass-through rates may vary widely by brand or type of vendors [25,31], manufacturers’ market power [30], the distance to cheaper resources [20,23,24,26] or the type of excise tax [28,29].

By using data from the 2009–10 National Adult Tobacco Survey (NATS), which uniquely collected self-reported cigarette price paid and detailed information on the use of selected price-minimizing strategies, we evaluate average prices paid and pass-through rates of state excise tax by smokers’ strategy.

Table 1 shows demographic and cigarette use characteristics of current smokers by price-minimizing strategy. Among 16 542 smokers, 54.0% (10 393) used one or more strategy. Specifically, 61.3% purchased cigarettes by pack without coupon, 14.9% purchased by pack with coupon, 19.0% purchased by carton without coupon, 4.7% purchased by carton with coupon, 8.8% purchased cigarettes on Indian reservations and 23.6% purchased generic brands.

This study used data from a national representative survey to evaluate smokers’ use of price-minimizing strategies, average price per pack paid and pass-through rates of excise tax. We found that at least 54.0% of current smokers had used one or more strategies in the previous year, and their demographic and smoking characteristics varied by these strategies. Substantial variations also existed in average prices paid with these strategies. Smokers who purchased by carton, who purchased on Indian reservations or who purchased generic brands paid consistently less than those who did not practice these strategies. Coupon use also rendered additional price reductions.

Our study is unique in that we are able to show how the effect of cigarette tax depends upon smokers’ price-minimizing strategies, and to demonstrate that not every smoker bore the same amount of tax. Excise tax was over-shifted to smokers of premium brands who purchased by pack outside Indian reservations: for every $1 tax increase, they paid the entire amount of the tax together with an additional premium of 7–10 cents. The estimated over-shifting is consistent with existing evidence, although our data and methodology differ substantially [22,26,31].

More importantly, carton purchasers (23.7%), generic brand users (23.6%) or those who bought cigarettes on Indian reservations (8.8%) did not realize the same increase in price for the tax increase. These smokers paid only 30–83 cents for the additional tax dollar, representing a saving of 17–70 cents. This might be one of the reasons that some previous studies found that the excise tax was under-shifted to some smokers, while most existing studies found the over-shift of excise taxes [20–27].

Because carton purchasers and generic brand users were more likely to be daily, heavy or more addicted smokers, tax savings rendered among these subpopulations might be subsidized by the premiums paid by pack purchasers, who were more likely to be occasional, light or less addicted smokers. Consequently, price-minimizing strategies may disproportionally mitigate the tax effect among daily, heavy or more addicted smokers.

The federal government recognizes that Native American tribes are sovereign nations. Therefore, tribal members are exempted from state excise tax on cigarettes sold on Indian reservations. However, US Supreme Court decisions have suggested that state taxes can be collected on cigarettes sold to non-members on Indian reservations [37–39]. In practice, however, not all states have established inter-governmental agreements with tribes to collect cigarette tax on sales to non-tribe members on tribal lands, as such agreements can be politically and functionally complicated to negotiate [40]. Therefore, on many reservations, state taxes, which ranged from $0.17 to $4.35 per pack on 1 July 2010 [34], are not applied to these sales. This may partly explain the under-shift of excise tax among those who were likely to purchase cigarettes on Indian reservations.

After adjustment for brands and quantity purchased, the average price paid by smokers who were likely to purchase cigarettes on Indian reservations increased considerably, from $4.74 to 5.67, which is comparable to the price paid for a similar purchase outside Indian reservations. This finding implies that smokers were more likely to buy cigarettes of generic brands, or purchase by carton on Indian reservations. Moreover, considering that the tax pass-through rate was statistically lower for those who were likely to have made their purchases on Indian reservations, and remarkably lower than the rate of carton purchase or that of generic brands, prices in nearby cigarette retailers outside Indian reservations could be much higher than the nation-wide average price [41]. Consequently, smokers who live in close proximity to Indian reservations were likely to spend their time and other resources to purchase ‘state tax-free’ cigarettes there. However, more refined analysis is warranted to further explore the determinants of Indian reservation purchase.

Young adults, African Americans and Hispanics were more likely to purchase by pack without coupons, but less likely to purchase generic brands. Therefore, tax increases were probably passed on to them in the form of higher cigarette prices. This is consistent with the evidence that these subpopulations are more responsive to tax increases [3,42].

We also found that the tax was shifted more heavily to carton purchasers who used coupon or other discounts than their counterparts who did not. More research is needed to understand the combination of carton purchase and coupon use.

This analysis has some limitations. First, not all smokers in the NATS were asked about purchases on Indian reservations during the past 12 months. Because of an approval delay during the first 2 months of data collection, the Indian reservation question was not added to the questionnaire until late November 2009. However, among 16 542 current smokers, 13 008 (78%) provided a response for this question, and data on this question were collected in all states and DC. Secondly, survey questions about cigarette purchases had different time-frames (e.g. carton purchase and coupon use were asked for the latest purchase, whereas brand choices were asked for the purchase during the past 30 days, and Indian reservation purchase was asked for the past 12 months). This time-frame issue is likely to have resulted in an underestimate of the proportion of smokers who used price-minimizing strategies during the previous year. However, the issue should not affect our findings on pass-through rates, besides interpretations of these findings, as these estimates do not necessary reflect the reductions during that time-period. In addition, smokers may endogenously use price-minimizing strategies in response to excise tax increases. To assess potential impacts of the endogeneity on our findings, we performed two sensitivity analyses. We replicated the analysis by excluding smokers from two states where taxes were raised during the NATS sampling period (Pennsylvania and Washington), and by stratifying samples with above and below the average state excise tax. These sensitivity results suggest that potential impacts of endogeneity on our findings are very limited, except for carton purchasers. The pass-through rate was statistically lower for carton purchasers living in states with taxes higher than the average compared to those living in states with taxes lower than the average. Tobacco regulation at local jurisdictions that changed during the sampling period are not captured in the analysis, although we include comprehensive controls for evidence-based state tobacco control policies. However, these local policy changes are less likely to affect the tax pass-through rate at the national level. Finally, like other telephone surveys, the response rate of the NATS is low (37.6% of the National Council of American Survey and Research Organizations response rate) [43], and the information collected is subject to recall bias. However, the sample used for the analysis is weighted to account for the complex survey design and non-response.

Findings show that, during 2009–10, a substantial proportion of smokers used price-minimizing strategies. By reducing pass-through rates of state excise taxes, these strategies are likely to lower the benefit that increased taxes would have on reducing the demand for cigarettes. The proportion of smokers who use price-minimizing strategies may be likely to rise because of increasing excise taxes in the future. Policies that can reduce such cost-reduction opportunities, such as setting a high floor price for cigarettes, prohibiting discounts or promotions or expanding state-level negotiations with tribes to collect taxes from non-member purchases, would probably be effective in reducing cigarette use and promoting cessation.