Jump to comment:

• Do high e-cigarette taxes lead to increases in initiation of cigarette smoking?
  John P Pierce, Karen Messer, David R Strong, Tarik Benmarhnia, Matthew D Stone, Yuyan Shi, Sara B McMenamin and Eric C Leas
  Published on: 2 January 2024
• Scientific concerns
  Michael Pesko
  Published on: 2 January 2024

• Published on: 2 January 2024

  Do high e-cigarette taxes lead to increases in initiation of cigarette smoking?

  • John P Pierce, Professor of Health Policy University of California San Diego
  • Other Contributors:
    • Karen Messer, Professor of Biostatistics
    • David R Strong, Professor of Behavioral Medicine
    • Tarik Benmarhnia, Associate Professor, Scripps Institution of Oceanography
    • Matthew D Stone, Assistant Professor
    • Yuyan Shi, Associate Professor in Health Policy
    • Sara B McMenamin, Associate Professor of Health Policy
    • Eric C Leas, Assistant Professor

  NOT PEER REVIEWED

  In his rapid response to our paper, Dr Pesko expresses concern with the paper’s conclusions. Our study, conducted at the US population level, demonstrated that the prolonged decline in cigarette smoking among young adults was largely unaffected by the introduction of e-cigarettes. Dr Pesko reminds us that this conclusion is contrary to findings from several quasi-experimental designs, which he feels should have primacy over our population-level analysis of time trends. We acknowledge that there are potential threats to the validity and generalizability of the findings of our study, and indeed of every study. Hence, a conclusion of causality requires that an association be robust and replicable across settings and methods. Given the significant public policy implications of any finding that e-cigarette vaping has a role in reducing cigarette smoking prevalence, we feel that presentation of transparent and easily understood population trends have an important role to play in policy discussions. In particular, to support the argument that high e-cigarette taxes cause increased smoking among youth, it would be important to have an exemplar of one or more jurisdictions that imposed a high e-cigarette tax and subsequently experienced an increase in cigarette smoking.

  A substantial portion of Dr Pesko’s critique of our paper is directed towards our commentary on a paper that he co-authored, Abouk et al. (1). This commentary was incorporated at t...

  Show More

  NOT PEER REVIEWED

  In his rapid response to our paper, Dr Pesko expresses concern with the paper’s conclusions. Our study, conducted at the US population level, demonstrated that the prolonged decline in cigarette smoking among young adults was largely unaffected by the introduction of e-cigarettes. Dr Pesko reminds us that this conclusion is contrary to findings from several quasi-experimental designs, which he feels should have primacy over our population-level analysis of time trends. We acknowledge that there are potential threats to the validity and generalizability of the findings of our study, and indeed of every study. Hence, a conclusion of causality requires that an association be robust and replicable across settings and methods. Given the significant public policy implications of any finding that e-cigarette vaping has a role in reducing cigarette smoking prevalence, we feel that presentation of transparent and easily understood population trends have an important role to play in policy discussions. In particular, to support the argument that high e-cigarette taxes cause increased smoking among youth, it would be important to have an exemplar of one or more jurisdictions that imposed a high e-cigarette tax and subsequently experienced an increase in cigarette smoking.

  A substantial portion of Dr Pesko’s critique of our paper is directed towards our commentary on a paper that he co-authored, Abouk et al. (1). This commentary was incorporated at the request of a reviewer. Abouk et al. studied surveillance data for US adolescents and concluded that “there were sizable positive cigarette cross-tax effects, suggesting economic substitution between cigarettes and ENDS for youth,” or in other words, that raising taxes on e-cigarettes would drive youth tobacco consumption away from e-cigarettes and towards traditional cigarettes. Our study considered the same jurisdictions used in Abouk et al., and looked for potential exemplars with substantial populations and with high e-cigarette taxes. We then reported cigarette and e-cigarette prevalence data for these exemplars for the 18-24-year-old population (i.e. for young adults still within the age window for smoking initiation).

  Table 1 of Abouk et al. outlines the 13 jurisdictions that taxed e-cigarettes between 2015 and 2019, providing average tax figures for the first two quarters of the years 2015, 2017 and 2019. Unfortunately, none of the high-quality US tobacco-use surveillance systems (including the two used by Abouk et al.) draws representative samples at the county or city levels. Thus, in three of their jurisdictions, it is not possible to present valid estimates of the change in tobacco use; simply raking the larger sample to the population distribution within the county or city does not make it representative of that jurisdiction. An additional five states did not meet the minimum population criterion of our paper (5 million), needed for reasonably tight confidence limits on the state-level population estimates. Therefore, only the following five states of the original 13 jurisdictions could be considered as possible exemplars in our paper: California, Minnesota, North Carolina, New Jersey, and Pennsylvania.

  Dr Pesko’s suggestion in his critique that we should have used 2019 rather than 2017 e-cigarette tax data is problematic, because the most recent population survey with state-level estimates of tobacco use (TUS-CPS) collected two-thirds of its data in 2018. A fundamental principal of causal inference asserts that to test whether an exposure influenced an outcome, the exposure must have been in place before the outcome was assessed. (2) Accordingly, we used the 2017 e-cigarette tax data, so that the taxes were assessed prior to the prevalence estimates of 2018.

  From Abouk et al. table 1, during the first two quarters of 2017, e-cigarettes taxes (per milliliter) for the five states under consideration were: Minnesota ($2.50), Pennsylvania ($1.05), California ($0.72), North Carolina ($0.05) New Jersey ($0). Pesko argues that we should have focused on the top three states as possible exemplars. However, California introduced a major increase in all tobacco taxes in the latter half of 2017 and, importantly, the tax rate on non-cigarette tobacco products was required to be equivalent to the tax imposed on cigarettes. (3) Thus, the effect of e-cigarette taxes on cigarette smoking in California is confounded by the concurrent increases in cigarette taxes. Accordingly, we concluded that there were only two possible exemplars from the Abouk et al. paper: Minnesota and Pennsylvania.

  In the supplement to our paper (eTable 3) we presented data on exclusive e-cigarette vaping for each of the 24 most populous US states, for both 2014/15 and 2018/19. For Minnesota, exclusive e-cigarette vaping remained constant at 2.8% in both years, in contrast to a 2.7-fold increase in its comparable group of states. For Pennsylvania, exclusive vaping declined from 3.3% to 0.9%, in contrast to an over twofold increase in its comparable group of states. This aligns with economic theory, (4) suggesting that in each of these two states higher taxes are associated with lower demand for e-cigarettes. Thus, both of these states can be used as transparent examples of the Abouk et al. hypothesis. However, in our paper, we noted that smoking prevalence among 18-24-year-olds across all of the 24 most populous US states declined by 43% from 12.9% (95% CI: 12.2%,13.7%) in 2014/15 to 7.4% (95% CI: 6.7% to 8.1%) in 2018/19. Pesko rightly points out the wide confidence limits for individual state smoking prevalence estimates in each of these years. However, there was no discernible evidence, not even suggestive evidence, indicating that, in either of these two exemplar states, the rate of decline in cigarette smoking was slower compared to other US states.

  In summary, it is important to have robust and replicable results before making conclusions that associations are causal. Before making an argument that states should refrain from increasing e-cigarette taxes because such taxes will slow down the decline in smoking initiation, it is particularly important to have transparent examples demonstrating such a cause and effect. None of the 13 jurisdictions used in the Abouk et al. study provides such a demonstration.

  References
  1. Abouk R, Courtemanche C, Dave D, Feng B, Friedman AS, Maclean JC, et al. Intended and unintended effects of e-cigarette taxes on youth tobacco use. J Health Econ. 2023 Jan 1;87:102720
  2. Bradford Hill A. (1965). The Environment and Disease: Association or Causation?. Proceedings of the Royal Society of Medicine. 58 (5): 295–300. doi:10.1177/003591576505800503.
  3. California State Board of Equalization New Tax Rate on Other Tobacco Products Effective July 1 2017 through June 30, 2018. Available at: https://www.cdtfa.ca.gov/formspubs/l502.pdf
  4. International Agency for Research on Cancer (IARC). Handbooks of Cancer Prevention, Tobacco Control, Vol. 14. Effectiveness of Tax and Price Policies for Tobacco Control. Lyon, France., 2011

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 2 January 2024

  Scientific concerns

  • Michael Pesko, J. Rhodes Foster Professor University of Missouri

  NOT PEER REVIEWED

  The current article has a number of problems, including factual errors, unsupported statements, failure to conduct relevant statistical testing, and failure to include relevant studies.

  The authors state: “Abouk et al studied 14 jurisdictions with at least some tax on e-cigarettes,43 but only two of these were large states with taxes large enough to possibly influence behaviour.44 They concluded that taxing e-cigarettes would push young people to cigarettes; however, in the two states with sizeable e-cigarette taxes, we did not find this to be the case.”

  First, the referenced study uses data through mid-2019, at which different jurisdictions have the following standardized tax rates (according to Abouk et al.’s Table 1): District of Columbia ($2.52), Minnesota ($2.49), California ($1.65), Cook County, IL ($1.50), and Pennsylvania ($1.05). The current study treats only Minnesota and Pennsylvania as being large states with large taxes, but the authors’ failure to include these other states that are used in Abouk et al. (2023) makes their comparison incomplete.

  Second, the authors state: “During this period, the prevalence of cigarette smoking in Pennsylvania declined by 6.3% pp and, in Minnesota, it declined by 8.6% pp (online supplemental eTable 4). Thus, the cigarette smoking prevalence decline in both states with meaningful e-cigarette taxes was greater than the 5.6% decline for the USA as a whole (2014–15=13.0%, 95% CI 12...

  Show More

  NOT PEER REVIEWED

  The current article has a number of problems, including factual errors, unsupported statements, failure to conduct relevant statistical testing, and failure to include relevant studies.

  The authors state: “Abouk et al studied 14 jurisdictions with at least some tax on e-cigarettes,43 but only two of these were large states with taxes large enough to possibly influence behaviour.44 They concluded that taxing e-cigarettes would push young people to cigarettes; however, in the two states with sizeable e-cigarette taxes, we did not find this to be the case.”

  First, the referenced study uses data through mid-2019, at which different jurisdictions have the following standardized tax rates (according to Abouk et al.’s Table 1): District of Columbia ($2.52), Minnesota ($2.49), California ($1.65), Cook County, IL ($1.50), and Pennsylvania ($1.05). The current study treats only Minnesota and Pennsylvania as being large states with large taxes, but the authors’ failure to include these other states that are used in Abouk et al. (2023) makes their comparison incomplete.

  Second, the authors state: “During this period, the prevalence of cigarette smoking in Pennsylvania declined by 6.3% pp and, in Minnesota, it declined by 8.6% pp (online supplemental eTable 4). Thus, the cigarette smoking prevalence decline in both states with meaningful e-cigarette taxes was greater than the 5.6% decline for the USA as a whole (2014–15=13.0%, 95% CI 12.2% to 13.7%; 2018–19=7.4%, 95% CI 6.7% to 8.1%).” Examination of the last column of supplemental eTable 4 shows that both Minnesota and Pennsylvania’s smoking rates could have not only decreased less than the national rate, but actually even increased given the confidence intervals. Given these wide confidence intervals for state-specific estimates, the authors cannot argue that, “cigarette smoking prevalence decline in both states with meaningful e-cigarette taxes was greater than the 5.6% decline for the USA as a whole” as this is not supported by statistical testing. In fact, these wide confidence intervals point to the futility of trying to do a state-specific exercise like the authors do, and point to the need to consider all e-cigarette tax states like Abouk at al. (2023) does in order to be powered to say anything meaningful.

  Curiously, eTable 4 also shows California’s smoking decline was -1.2 pp over this time period, which by the authors’ own faulty method would provide evidence opposite of their claim that higher e-cigarette taxes are associated with larger-than-average declines in cigarette smoking prevalence. The omission of California as a large state with a large tax is quite suspicious. California’s e-cigarette tax began 2Q 2017, well before the last period in 2019. There is no justification for Pennsylvania being included but not California is not, since California had higher taxes and a larger population.

  Third, the authors conjecture that there are taxes of a certain threshold that are “large enough to possibly influence behavior.” As far as I can tell, they believe that a $1 per fluid ml e-cigarette tax is the threshold at which behavior is influenced, though they obviously missed California, District of Columbia, and Cook County, IL. The author’s reference (43) was published in 2011 before e-cigarettes were even widely on the marketplace, and as such this doesn’t seem to offer e-cigarette-specific evidence to support their argument. The idea that there is a threshold at all seems arbitrary and poorly supported by decades of research. The literature is full of examples of small cigarette taxes having impacts; for example, one study found a $0.55 per pack cigarette tax increase would reduce maternal smoking by about 22% (2). It’s reasonable to say that taxes of certain sizes may be needed to be powered to detect effects, but the current statement as written suggests small taxes do not influence behavior (in even small ways), which is incorrect.

  Fourth, the current study fails to control for other tobacco policies like cigarette taxes that reduce smoking. For example, Pennsylvania had a sizable cigarette tax in 2016, Minnesota had multiple tax increases since 2013, California had a sizable tax increase in 2017, etc. Of course, other policies are changing as well like indoor air laws. The authors of the current paper essentially compare descriptive evidence of changes in smoking at two points in time, not accounting for cigarette policies that may also be reducing smoking rates. This does nothing to weaken arguments made in Abouk et al. (2023) that conditional on controls, e-cigarette taxes both reduce youth e-cigarette use and raise youth cigarette use. Further, Abouk et al. (2023) use all e-cigarette tax variation and not variation from specific states.

  Fifth, the paper does not include several relevant citations. There are many other quasi-experimental studies that reach a similar conclusion to Abouk et al. (2023) that are not referenced. Other e-cigarette tax studies of youth (4) and young adults (5) find supporting evidence to Abouk et al. (2013). A recent editorial summarized the quasi-experimental literature on youth use as follows: “There are seven natural experiment difference-in-difference-style studies of e-cigarette minimum legal purchase age laws and e-cigarette tax rates on youth, six of which find that these policies increase cigarette use.” (6). The literature further crystallizes when we add older populations and sales data to the mix, with another review finding that 14 of 16 quasi-experimental studies of e-cigarette policies find that e-cigarette availability displaces cigarette use (7). Finally, reference 40 might have been an error, as I don’t see that paper discussing e-cigarette taxes at all, despite the paper claiming it does.

  The preponderance of evidence from quasi-experimental studies supports e-cigarette availability reducing youth cigarette use. The quasi-experimental studies attempt to solve concerns about common liabilities impacting both e-cigarette use and cigarette use by using policy variation to mimic the randomization of a RCT, which allows causal interpretation from observational data (if we’re willing to believe recent Noble Prize award committee statements; and a recent Tobacco Online Policy Seminar panel of world-renowned methodological experts) (8). The current study does not seem to mimic randomization in any way since e-cigarette use is self-selected and influenced by the same factors as cigarette use, and as such the current study suffers from common liability concerns that the quasi-experimental studies have attempted to address.

  It is therefore very difficult for me to understand what the contribution of this paper is, and I recommend against future research using these methods to explore this question when stronger methods are available.

  1. Abouk R, Courtemanche C, Dave D, Feng B, Friedman AS, Maclean JC, et al. Intended and unintended effects of e-cigarette taxes on youth tobacco use. J Health Econ. 2023 Jan 1;87:102720. <p>
  2. Ringel JS, Evans WN. Cigarette taxes and smoking during pregnancy. Am J Public Health. 2001;91(11):1851–6. <p>
  3. Cotti C, Nesson E, Pesko MF, Phillips S, Tefft N. Standardising the measurement of e-cigarette taxes in the USA, 2010–2020. Tob Control. 2021 Dec 15;tobaccocontrol-2021-056865. <p>
  4. Pesko, Warman. Re-exploring the early relationship between teenage cigarette and e-cigarette use using price and tax changes. Health Econ. 2021;
  5. Friedman AS, Pesko MF. Young adult responses to taxes on cigarettes and electronic nicotine delivery systems. Addiction. 2022 Jul 29;add.16002. <p>
  6. Pesko MF, Cummings KM, Douglas C, Foulds J, Miller T, Rigotti NA, et al. United States Public Health Officials Need to Correct E-cigarette Health Misinformation. Addiction. 2022; <p>
  7. Pesko MF. Effects of E-Cigarette Minimum Legal Sales Ages on Youth Tobacco Use in the United States. J Risk Uncertain. 2022; <p>
  8. Pesko MF. How Data Security Concerns Can Hinder Natural Experiment Research: Background and Potential Solutions. JNCI Monogr. 2022 Jul 5;2022(59):89–94. <p>

  Show Less

  Conflict of Interest:
  Dr. Michael F. Pesko has never received commercial funding. His current research is supported by funding from the National Institutes of Health, American Cancer Society, Food and Drug Administration, Health Canada, and the University of Kentucky’s Center for Free Enterprise.

  • Back to top