Objectives An increasing number of US states have required a tax on electronic nicotine delivery systems (ENDS) in the past few years. This study evaluated the effect of statewide vaping product excise tax policy on ENDS use among young adults.
Methods We used the two recent waves (2014–2019) of the Tobacco Use Supplement to the Current Population Survey. A total of 17 896 US young adults were analysed. Difference-in-differences approach along with weighted multilevel logistic regressions was used to evaluate the association of vaping product excise tax policy adoption with current ENDS use, accounting for the clustering of respondents within the same states.
Results There was an increase in current ENDS use prevalence from 2014–2015 (3.4%) to 2018–2019 (5.4%). Respondents living in states with vaping product excise tax policy showed significantly lower increase in ENDS use prevalence during the study period (interaction between within-state changes and between-state differences: adjusted OR (AOR)=0.57, 95% CI=0.35 to 0.91), controlling for other state-level policies and sociodemographic characteristics. Additional stratified analysis with state-fixed effects by vaping product excise tax policy implementation status showed consistent findings.
Conclusions Our findings suggest that adopting a vaping product excise tax policy may help reduce ENDS use and suppress the increase of ENDS use prevalence among young adults. Considering that there are still a number of US states that have not implemented vaping product excise tax policy, wider adoption of such policy across the nation would likely help mitigate ENDS use prevalence.
- electronic nicotine delivery devices
- public policy
Data availability statement
Data are available in a public, open access repository. Data of the TUS-CPS are publicly available at https://cancercontrol.cancer.gov/brp/tcrb/tus-cps/questionnaires-data.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Although many cigarette smokers use electronic nicotine delivery systems (ENDS) as a cessation or harm reduction strategy,1 ENDS use is not harmless since they typically include nicotine and other toxic chemicals, posing health risks to the users and bystanders.2–4 ENDS use in US young adults (often defined as adults 18–24 years of age4) is increasingly prevalent while holding steady in other adult age groups in recent years.5 The American Lung Association emphasises the fact that young adulthood is considered as a vulnerable developmental period with escalated risk for ENDS use as they are three times more likely to initiate use of ENDS than conventional cigarettes.6 In addition, current evidence suggests that ENDS use in young adults is associated with an increased likelihood of subsequent use of other substances such as cigarette smoking and marijuana.7–9
Levying a tax on tobacco products is a well-documented policy tool to discourage tobacco use.10–12 Given the rapid increase in ENDS use, many US localities have recently levied a vaping product excise tax in order to reduce ENDS use. As of 30 September 2020, 26 US states and the District of Columbia have passed legislation requiring a tax on ENDS products.13 However, extant studies revealed mixed results of the effects of vaping product excise tax policy implementation on ENDS use,14–19 and evidence on how this policy is associated with ENDS use among young adults is lacking. A recent study that evaluated the excise tax policy as well as smoke-free policies and Tobacco 21 has revealed no significant association of excise taxes of electronic cigarettes (e-cigarettes) with declines in adolescent e-cigarette use during 2017–2019.14 However, the results of Choi et al’s study14 may not be generalisable to other age groups or regions given that the authors examined adolescent ENDS use only in 34 US states. In contrast, another recent study by Pesko et al 15 examined the impact of conventional cigarette and e-cigarette tax policies on cigarette and e-cigarette use (ie, cross-tax effects) among general US adult population using more years (2011–2018) of data and found a reduction in vaping among respondents living in localities that adopted an e-cigarette tax policy, compared with those living in localities without the taxes, controlling for a wide range of locality-level variables. Saffer et al’s study16 also examined the cross-tax effects on adult smoking in Minnesota. It is noteworthy that, however, the cross-tax effects that indicate cigarettes and e-cigarettes as substitutes may not be applied to young adults’ tobacco use behaviours given the explosive popularity of ENDS among young adults compared with other age groups.5 Two recent studies have also found that vaping product excise tax adoption was associated with lower ENDS use17 18; however, they did not compare changes in ENDS use over time in states with vaping product excise tax policy with those without the policy. A working paper reported that an increase in e-cigarette taxes leads to an increase in e-cigarette prices and a decrease in e-cigarette sales in eight states and two countries in the US.19 Cotti et al 19 and Pesko et al 15 clustered standard errors at the locality level to account for within-locality clustering. To our best knowledge, none of previous studies examined the association between vaping product excise tax adoption and ENDS use using a multilevel approach, which not only accounts for clustering of individuals (ie, design effects) but also incorporates different error terms for different levels in a data with hierarchy, which further yields more accurate Type-I error rates compared to non-hierarchical methods.20
The present study examined the association between the implementation of statewide vaping product excise tax policy and ENDS use in young adults from a nationally representative US sample between 2014 and 2019, the period when there was a rapid increase in ENDS use prevalence. We used the difference-in-differences approach along with multilevel modelling, which allowed us to compare current ENDS use in the US states before and after the implementation of vaping product excise tax policy, while controlling for observable and unobservable factors between states with and without such policy. We hypothesised that young adults residing in states with the vaping product excise tax policy would be less likely than those living in states without the policy to be a current ENDS user.
Many tobacco use patterns and behaviours are shaped by multiple personal and environmental factors.21–23 Viewed from the perspective of socioecological approach, this study analysed the effects of vaping product excise tax policy on ENDS use behaviours among young adults. The socioecological framework incorporates the multifaceted interplay of policy, community, organisational, interpersonal, and intrapersonal levels of influence of health and related behaviours.24 25 Levels in this framework include both upstream (ie, larger environment) and downstream (ie, individual) factors, which are all closely related to personal health outcomes.22 According to Warnecke et al’s conceptualisation,26 in the socioecological context, policy and social conditions play a role in influencing health outcomes as one of the upstream factors. Thus, the socioecological approach may help understand the link between vaping product excise tax policy and ENDS use behaviours. Given the available information in the secondary data used in this study, we also modified the framework by incorporating multiple risk factors in the context of the socioecological model, which have been considered as predictors for ENDS use and associated behaviours, including the three groups of variables: (1) statewide vaping product excise tax policy, (2) state-level control variables and (3) individual-level control variables.
Study design and participants
We used a quasi-experimental cross-sectional design to examine the association of the implementation of the statewide vaping product excise tax policy with current ENDS use in a sample of US young adults. The treatment group was composed of respondents residing in states that implemented the vaping product excise tax policy, and those living in states without the policy comprised the control group. This study used the Tobacco Use Supplement to the Current Population Survey (TUS-CPS), a nationally representative survey of tobacco use in the USA. The TUS-CPS is a household-based cross-sectional survey of the civilian, non-institutionalised adults aged 18 years or above in the 50 states and District of Columbia in the USA. TUS-CPS collects individuals’ detailed trajectory of use patterns of tobacco products through a multistage and stratified sampling design. The TUS-CPS is a periodical supplement to the CPS which interviews over 50 000 households in the USA and has been conducted every 3–4 years regularly since 1992. Detailed information about the TUS-CPS can be found elsewhere.27 This study used the two most recent waves (July 2014, January and May 2015; and July 2018, January and May 2019) of the TUS-CPS given that data on ENDS use are only available since the 2014–2015 wave. Among 301 093 adult participants, a sample of young adults (18–24 years) was extracted as the target population of the present study (N=18 404). Given that Minnesota implemented vaping product excise tax policy before the study period (1 August 2010) and that New Jersey began to levy the vaping product excise tax during the study period (29 September 2018), we excluded respondents from Minnesota (1.4%, n=260) and New Jersey (1.3%, n=248), reducing the final sample size to 17 896 young adults (2014–2015 n=10 479; 2018–2019 n=7417).
Variables and measures
Outcome variable: ENDS use
The outcome variable was assessed based on a binary measure of current ENDS use. Current ENDS use was measured through the question: ‘Do you now use an e-cigarette every day, some days or not at all?’ (yes/no).
Main predictor: statewide vaping product excise tax policy
The main predictor in the current study was the statewide vaping product excise tax policy status as of 1 July 2018 (the time when the latter wave of data, 2018–2019 TUS-CPS survey, began). As can be seen in table 1, based on each state’s policy status, respondents who lived in the states that implemented the vaping product excise tax policy were coded as 1=policy implemented (vs 0=no policy). The effective date of the policy for each state can be found elsewhere.28 29
State-level control variables
Informed by Pesko et al’s recent study,15 to reduce omitted variable bias, we controlled for five time-varying state-level variables that may be associated with ENDS tax policy and use behaviours: (1) indoor vaping restrictions in bars, restaurant, workplaces (implemented vs not implemented)28; (2) ENDS minimum legal purchase age laws (21 years or older vs less than 21 years)28; (3) cigarette excise tax rates (dollars per 20-pack)30 (4) indoor smoking restrictions in bars, restaurant, workplaces (implemented vs not implemented)31; and (5) marijuana laws (fully illegal, decriminalised or legalised medical marijuana only, or legalised recreational marijuana).32–34
Individual-level control variables
Given that the socioecological approach includes downstream factors such as individual demographics,22 26 we included the following sociodemographic variables based on the literature that examined tobacco use behaviours at individual levels: age, sex (male or female), race (white or non-white), ethnicity (Hispanic or non-Hispanic), educational attainment (high school/general educational diploma or less, or some college or more), family income (less than $25 000, $25 000–$49 999, $50 000–$99 999, or $100 000 and above), and employment status (employed, unemployed or not in labour force).
We used descriptive statistics to examine and compare ENDS use status, state-level policies, and sociodemographic characteristics in 2018–2019 with those in 2014–2015 by computing unweighted frequencies and weighted percentages for categorical variables as well as weighted means and SEs for continous variables. Pearson’s Χ2 statistics for categorical variables and independent samples t-tests for continuous variables were used to test the differences between the two survey waves.
To evaluate the impact of statewide vaping product excise tax policy on young adult ENDS use, we performed a multilevel logistic regression model to examine the association between state vaping product excise tax policy and ENDS use, accounting for clustering of respondents within the same states35 while controlling for state-level policy variables and sociodemographic characteristics. Specifically, we built a random-intercept model that examined whether the statewide ENDS excise tax policy explained between-state variation in respondents’ ENDS use behaviours beyond individual-level explanatory variables. The respondents were positioned at the lowest level (level 1: individual) and were nested within states (level 2: state). Multilevel models incorporate fixed effects and random effects: (1) fixed effects exhibit overall estimates of explanatory variables across all individuals, and (2) random effects reflect a variance estimate of group effects.35
We used the difference-in-differences approach along with multilevel logistic regression model to compare two groups (young adults living in the states where vaping product excise tax policy was implemented vs those living in the states where such a policy was not implemented) by adding a multiplicative interaction term of between-state differences (ie, presence of policy)×within-state changes (ie, pre/post-implementation of policy) to the multilevel logistic regression model. To compare our analysis to a fixed-effect approach, we further fitted a multilevel model with state-fixed effects. We then conducted a stratified analysis by states with and without vaping product excise tax policy.
The difference-in-differences framework assumes the parallel trends (ie, both treatment and control groups have similar trends in terms of outcome variable before the policy implementation). Unfortunately, the parallel trend assumption is untestable in this study given that there are only two time points (ie, 2014–2015 and 2018–2019 waves). To deal with this issue, we alternatively conducted Χ2 and t-tests to compare group differences in outcome variable and covariates between respondents living in states with and without the policy. The results showed that ENDS use prevalence between the two groups (3.4% vs 3.5%) in the pre-period was not statistically different (p=0.482). Similarly, for the covariates, we observed small effect sizes or no significant differences.
For all analyses, associations were presented as adjusted ORs (AORs) and corresponding 95% CIs. We observed no multicollinearity issues among the independent variables and covariates based on the generalised variance inflation factor.36 To account for different selection probabilities of the study participants, we applied person-level survey sampling weights of the TUS-CPS. R V.4.11.1 was employed to perform all statistical analyses.
Table 2 summarises the descriptive statistics of the study sample for both waves. An increase in ENDS use prevalence was observed from 2014–2015 to 2018–2019 (from 3.4% to 5.4%). Current ENDS use prevalence in states with vaping product excise tax policy at 2014–2015 and 2018-2019 waves was 3.5% (n=83) and 4.7% (n=86), respectively (weighted Pearson’s X2 =3.32, p=0.129), and that in states without the policy was 3.4% (n=306) and 5.7% (n=365), respectively (weighted Pearson’s X2 =40.60, p<0.001). Regarding state-level variables, significant differences in vaping product excise tax policy, indoor vaping restriction, ENDS minimum legal purchase age laws, cigarette excise tax rates, and marijuana laws were observed between the 2014–2015 and 2018–2019 waves. Sociodemographic characteristics of young adults in the 2018–2019 wave were generally similar to those in the 2014–2019 wave except household income level and employment status.
Table 3 presents the results of the weighted multilevel logistic regression analysis of current ENDS use using a difference-in-differences approach. A significant interaction occurred between within-state changes and between-state differences (AOR=0.57, 95% CI=0.35 to 0.91). This result revealed that the positive association of time (ie, post-policy period) with current ENDS use was weaker among young adults living in the states that implemented vaping product excise tax policy compared with those living in the states without the policy. That is, respondents living in states with vaping product excise tax policy showed significantly lower increases in ENDS use prevalence during the study period.
Table 4 summarises the results of the weighted multilevel model with state-fixed effects. This analysis provided largely consistent results of the multilevel model without state-fixed effects. A significant interaction between within-state changes and between-state differences (AOR=0.53, 95% CI=0.28 to 0.99) was observed. Table 4 also summarises the results of the stratified analysis by states with and without vaping product excise tax policy, showing that the association between time and current ENDS use is positive and significant among young adults living in states without the policy (AOR=1.77, 95% CI=1.42 to 2.20), but the association was not significant among their counterparts.
The current study evaluated the extent to which statewide vaping product excise tax policy influenced current ENDS use using a controlled design among young adults from a nationally representative US sample. Between 2014 and 2019, a significant increase in current ENDS use among young adults was observed. During the same time period, in the absence of federal regulation, there was an increase in the number of states that have implemented statewide vaping product excise tax policy. Given the rapid increase and popularity of ENDS use in young populations and that young adulthood is marked by enhanced susceptibility to substance use problems, implementation of an effective regulatory policy in more localities to prevent ENDS use in young adults is warranted.
A major finding of the current study is that the implementation of vaping product excise tax policy appears to be effective in suppressing the increase of ENDS use prevalence in the USA, when controlling for other prominent ENDS use regulations, other state-level policy variables and sociodemographic characteristics. Importantly, our results revealed that the association between time (pre-implementation and post-implementation of the policy) and current ENDS use was weaker among young adults living in the states with vaping product excise tax policy compared with those living in the states without the policy. This indicates that respondents living in states with vaping product excise tax policy showed significantly lower increases in ENDS use prevalence during the study period. The finding was further corroborated by the additional stratified analysis showed that the association between time and current ENDS use was positive and significant among young adults living in states without the policy, whereas the association was insignificant among their counterparts. Our findings support findings from recent studies that have found a negative association of e-cigarette tax rates with e-cigarette use among general US adults.15 17 18 Interestingly, our findings were contrary to Choi et al’s recent study that found no significant effects of ENDS excise tax policy on adolescent ENDS use.14 These dissimilar results may be partly due to the difference in the analytical samples. For example, compared with adolescents, ENDS use among young adults may be more frequent and established (ie, fairly regular use rather than experimental use),4 possibly due to relatively longer duration of ENDS use. Thus, it is likely that frequent and established ENDS use among young adults makes them more sensitive to price changes in ENDS products, which may, in turn, make vaping product excise tax policy more effective in suppressing ENDS use among young adults. Additionally, young adults are known to be financially strained and price sensitive because they experience critical situational changes such as reduced financial support from parents. Our findings add another layer of evidence to the literature by demonstrating the significant role of implementation of vaping product excise tax policy in reducing ENDS use, particularly focusing on young adults who are a vulnerable population with a high rate of using ENDS. Therefore, findings of the current study suggest that the implementation of vaping product excise tax policy is clearly essential, especially for states without the policy. For states with the policy already, but only for counties and cities, statewide implementation of the policy is suggested to decrease ENDS use prevalence in young populations.
Another finding of note in the current study is that cross-tax effects (ie, effects of cigarette excise tax on ENDS use) were not observed. Given that ENDS are often used as an alternative to cigarette smoking37 as well as a cessation or harm reduction tool among cigarette smokers1, traditional cigarettes and ENDS could be substitutes. Recent evidence also revealed that that higher conventional cigarette excise tax rate was associated with increased ENDS use as well as reduced cigarette use among US adults.15 However, for young adults, not observing the effect of cigarette excise tax policy on ENDS use is possible because cigarette smoking prevalence is significantly lower in young adults, compared with other age groups in adults.38
Also, financial factors are important components that should be considered when examining the effects of tax policy. While previous literature reported that household income level may not be an important factor in discriminating ENDS use status for general adult population,39 40 it may be a crucial indicator for young adults because they are more financially strained and price sensitive compared with other adult groups. Indeed, in our study, ENDS use was most prevalent among the lowest income level group (<$25 000), followed by $25 000–$49 000, $50 000–$99 000, and ≥ $100 000 groups, and lower income level was associated with an increased likelihood of using ENDS among young adults. Together with the significant effect of vaping product excise tax policy on ENDS use behaviour in young adults, these findings suggest that the policy may be paid disproportionately by people with lower income. Future studies can explore the disparities that may be caused by ENDS tax policy.
This study has limitations. First, due to the small cell size for states with vaping product excise tax policy during the study period, we evaluated the extent to which a general statewide policy influences current ENDS use, while the effects of a specific policy would likely vary. For example, as of 1 July 2018, three states (California, Minnesota and Pennsylvania) and the District of Columbia implemented the policy based on the percentage of the specified cost (eg, 95% of wholesale) and five states (Delaware, Kansas, Louisiana, North Carolina and West Virginia) used volume-based taxation (eg, $0.05/mL). Given that this simplification may hinder sophisticated examination of the effects of vaping product excise tax policy and that more states have recently adopted policies taxing on ENDS products, future research that examines the effects of different taxation designs on ENDS use among young adults is warranted. In addition, while different ENDS products (eg, e-liquid, disposable ENDS or pod/cartridges) are taxed differently, different ways of taxing were not examined in our study due to data limitation. Future replication studies would be desirable that take into account different taxation methods for different device types, which may or may not corroborate the findings of current study. Second, we have limited data on ENDS use behaviours because the questions regarding ENDS use in the TUS-CPS have recently been added to the later wave only (2014–2015 and 2018–2019 waves). Third, because there was only one data collection point at the pre-policy period, it was not possible to test the parallel trend assumption for the difference-in-differences analysis; we did, however, conduct additional tests to make sure that the two groups were comparable in the pre-policy period. Fourth, self-reported responses might have resulted in measurement errors. Fifth, this study did not examine other locality-level (eg, county, city) changes in ENDS use despite variations in vaping product excise tax policy at different locality levels. Thus, our findings are most relevant to state-level policies and may not be generalisable to other locality-level regulations. Lastly, the US Food and Drug Administration has raised the federal minimum age of purchasing ENDS products from 18 to 21 years in December 2019,41 but data used in this study were only up to 2019. Future studies should use more recent data to further evaluate the effectiveness of vaping product excise tax policy among young people along with the federally expanded ENDS regulations such as Tobacco 21.
Despite these limitations, our findings contribute to the literature as the current study adds to the evidence of the effects of vaping product excise tax policy on current ENDS use among US young adults from a nationally representative sample. Considering that there are still a number of US states that have not implemented the vaping product excise tax policy, wider adoption of such policy across the nation would likely help reduce ENDS use rates. Future monitoring for states without such policy is needed since the effects of the policy for these states remain uncovered. Additional studies are also needed to design more strategic and sophisticated policies taxing ENDS products by distinguishing different ways of taxation for different ENDS products.
What this paper adds
Electronic nicotine delivery systems (ENDS) use in US young adults is increasingly prevalent. More US states have levied a vaping product excise tax in order to reduce ENDS use.
Extant studies have revealed mixed results of the effects of vaping product excise tax policy on ENDS use. Little is known about how the implementation of the policy is associated with ENDS use in young populations, especially in young adults.
This multiwave cross-sectional study that used US national data and the difference-in-differences approach along with weighted multilevel logistic regressions revealed that respondents living in states with vaping product excise tax policy showed significantly lower increases in ENDS use prevalence during 2014–2019.
Data availability statement
Data are available in a public, open access repository. Data of the TUS-CPS are publicly available at https://cancercontrol.cancer.gov/brp/tcrb/tus-cps/questionnaires-data.
Patient consent for publication
The Institutional Review Boards of Indiana University considered the study as a non-human subject study because of the public availability of TUS-CPS data.
Correction notice This article has been corrected since it was published online first. Reader feedback post-publication identified areas of inaccuracy and confusion. The editors and the authors have reviewed the article and the associated references. Issues spotted by readers, together with others identified by the editors, have been amended. We thank readers for bringing this to our attention and the authors for their cooperation.
Contributors D-HH conducted literature review, conceptualised the paper, performed statistical analyses and drafted the manuscript. H-CL coordinated the study, acquired the data, conceptualised the paper and provided critical revision of the manuscript. D-CS provided critical revision of the manuscript. All authors read and approved the final manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.