The Association Between Excise Taxes and Smoking and Vaping Transitions - Findings from the 2016–2020 ITC United States Surveys

Abstract

Background

While a growing number of studies examined the effect of e-cigarette (EC) excise taxes on tobacco use behaviors using cross-sectional surveys or sales data, there are currently no studies that evaluate the impact of EC taxes on smoking and vaping transitions.

Methods

Using data from the US arm of the 2016–2020 International Tobacco Control Four Country Smoking and Vaping Survey (ITC 4CV), we employed a multinomial logit model with two-way fixed effects to simultaneously estimate the impacts of cigarette/EC taxes on the change in smoking and vaping frequencies.

Results

Our benchmark model suggests that a 10% increase in cigarette taxes led to an 11% reduction in smoking frequencies (p<0.01), while EC taxes did not have a significant effect on smoking frequencies.

Conclusion

Our findings suggest that increasing cigarette taxes may serve as an effective means of encouraging people who smoke to cut back on smoking or quit smoking. The impact of increasing EC taxes on smoking transitions is less certain at this time.

1. Introduction

The use of electronic cigarettes (ECs), also known as e-cigarettes, has significantly increased in the United States over the past two decades, particularly among young people. While ECs may be helpful for people who smoke and are trying to quit cigarette smoking, (Zhu et al., 2017; Zhuang et al., 2016) they may also lead to nicotine addiction among individuals who would not otherwise use any nicotine or tobacco products (Tackett et al., 2021; Vogel et al., 2020). It is important that regulations aim to prevent the use of all forms of nicotine or tobacco products among young people and adults who never use such products. However, ECs are generally considered to be less harmful than cigarettes, (Callahan-Lyon, 2014; Marques et al., 2021; Tan & Bigman, 2014) which has sparked debate about the impact of EC regulations on people who currently smoke and are considering using them to quit smoking, as well as people who formerly smoked and are using ECs to remain abstinent from smoking.

The costs and benefits of regulating ECs for people who smoke and vape depend on how these regulations affect smoking or vaping frequencies, in which reductions in use frequencies refer to smoke or vape less, or quit smoking or vaping, and increases in use frequencies refer to smoke or vape more or start smoking or vaping. Although there’s increasing research on how EC excise taxes, influence behaviors using cross-sectional data, (Abouk et al., 2022; Cantrell et al., 2020; Pesko et al., 2020) there is currently a lack of studies that evaluate the impact of EC taxes on smoking transitions using longitudinal data. As a result, our understanding of the overall impact of EC regulations on nicotine or tobacco use behavior is limited.

The debate over EC taxes illustrates the challenges faced by policymakers when different products pose different risks to human health. Cigarette excise taxes have been successful in reducing cigarette consumption worldwide and are considered one of the most effective regulatory policies for inducing behavioral changes (Callison & Kaestner, 2014; Chaloupka et al., 2011; Chaloupka et al., 2012; Meier & Licari, 1997)). It is anticipated that EC excise taxes will have significant impacts on both EC and cigarette consumption. However, existing research on this topic has largely analysed cross-sectional tobacco surveys and has focused on either the intentional consequences (e.g., reducing EC consumption) or the unintentional consequences (e.g., increased smoking) (Abouk et al., 2022; Pesko et al., 2020). There have been no studies that use longitudinal data to assess how EC taxes simultaneously impact use frequencies and the role that cigarette excise taxes play in relation to EC taxes. As a result, the net effect of a nicotine or tobacco taxation system on behavioral changes remains unclear, for both current smokers who may benefit from quitting smoking with the help of ECs and exclusive EC users who may benefit from quitting all nicotine consumption.

In addition to regulations (e.g., nicotine and tobacco taxes), sociodemographic factors may also influence smoking/vaping transitions. Studies have shown that demographic variables (e.g., ethnicity, gender, and age) were associated with EC and smoking initiation and cessation (Japuntich et al., 2011; Kalan & Brewer, 2023). For example, smoking and EC initiation are more likely to occur at younger ages (Barrington-Trimis et al., 2020; Pérez et al., 2021). However, how these factors influence the transitions among those who already smoke and vape is still not well understood. It is important to identify the individual characteristics that contribute to the change in use frequencies in order to design regulations that take these factors into consideration. E.g., if it is found that younger users are more likely to reduce use frequencies, it would be justifiable to spend more resources on designing campaigns that target young people.

This study utilizes unique longitudinal data collected by the International Tobacco Control (ITC) project to simultaneously estimate the impacts of EC and cigarette taxes on use frequencies among people who smoke(d) and vape(d). This research aims to answer the important question of how to design a taxation system that leads to net benefits for adult people who use nicotine or tobacco. It is the first study of its kind to do so.

2. Data and Methods

2.1. Data

We drew data from the US arm of the longitudinal International Tobacco Control Four Country Smoking and Vaping Survey (ITC 4CV) waves 1–3, conducted in 2016 (N=2,838), 2018 (N=2,937), and 2020 (N=2,528), respectively. The ITC 4CV is a nationally representative longitudinal survey that collects information on smoking and vaping behaviors and related factors from a sample of current and former smokers, as well as current vapers. To be included in the sample, individuals had to meet one of the following criteria: 1) current smokers - those who currently smoke at least occasionally and have smoked at least 100 cigarettes in their lifetime; 2) former smokers – those who quit smoking within the past 24 months ago or more than 24 months ago, and had smoked at least monthly and at least 100 cigarettes in their lifetime; and 3) current vapers - those who currently vape at least weekly, regardless of their smoking status.

2.2. Measures

Outcome – Smoking and vaping transitions in nicotine and tobacco use

The respondents were asked in the survey, “How often, if at all, do you currently smoke ordinary cigarettes?” The variable “smoking frequency” was constructed and grouped into 4 categories based on their answers to this question: (1) current daily smokers, (2) current weekly smokers, (3) current monthly or less-than-monthly smokers, and (4) recent quitters (quit cigarettes within the last two years and has smoked more than 100 cigarettes lifetime) or long-term quitters (quit cigarettes more than 2 years ago and has smoked more than 100 cigarettes lifetime) or non-smokers.

We classified smoking transitions as reduction, increase, or no-change in smoking frequencies by comparing smoking frequencies between two consecutive waves (T and T+1). Specifically, we considered a respondent to have no changes in smoking frequencies in T+1 if his or her self-reported categories of smoking frequencies did not vary between two consecutive waves. We considered a respondent to have a reduction in smoking frequencies in T+1 if the reported categories of smoking frequency move from more frequent to less frequent ones (e.g., daily smoker in T but monthly smoker in T+1) or quit smoking between two consecutive waves. On the other hand, we considered a respondent to have an increase in smoking frequencies in T+1 if the reported categories of smoking frequency moved from less frequent to more frequent ones or started smoking between two consecutive waves.

Similarly, the respondents were asked, “How often, if at all, do you currently use e-cigarettes/vaping devices (i.e., vape)?” The variable “vaping frequency” was constructed and grouped into 4 categories as well: (1) current daily vapers, (2) current weekly vapers, (3) current monthly or less-than-monthly vapers, (4) ever quitters (past at least weekly vaper) or past triers (vaped once or occasionally) or never tried ECs at the baseline wave. Correspondingly, we classified vaping transitions in the same way as smoking transitions.

Explanatory variables

State cigarette taxes and EC taxes

We obtained annual state-level cigarette and EC excise taxes for the years 2016–2020 from the US Centers for Disease Control and Prevention (CDC) State Tobacco Activities Tracking and Evaluation (STATE) System. The variable “cigarette tax” was a continuous variable, measuring the state excise taxes per pack (20 cigarette sticks). A binary variable, “whether EC tax” was created and coded as 1 if the state where the respondent lived had EC taxes in place and 0 if not. Additionally, we used a continuous standardised EC tax variable produced and shared by Cotti et al. (2021) to examine how the magnitude of EC taxes affects smoking and vaping frequencies (Cotti et al., 2021). We used time-invariant population-weighted EC taxes assuming a 35% retailer markup rate) and expressed these taxes in units of 0.7 fluid mL (equivalent to1 pack).

Other control variables from ITC surveys

The ITC surveys also contain socio-demographic variables, which we assessed in the analyses: sex (female vs. male), race/ethnicity (non-Hispanic White vs. others), age groups (18–24, 25–39, 40–54, 55 and up), and annual household income (low: <$30,000; middle: $30,000-$59,999; high: ≥$60,000).

2.3. Analytical Model

$$Smokin\mathit{g\_T}ransitio{n}_{ist+1}={\beta }_0+{\beta }_{1}E\mathit{C\_t}a{x}_{s,t+1}+{\beta }_{2}Ci\mathit{g\_t}a{x}_{s,t+1}+{\beta }_{3}Vapin\mathit{g\_f}r{q}_{it}+Demographic{s}_{it}+{\epsilon }_{ist}$$ (1)

$$Vapin\mathit{g\_T}ransitio{n}_{ist+1}={\beta }_0+{\beta }_{1}E\mathit{C\_t}a{x}_{s,t+1}+{\beta }_{2}Ci\mathit{g\_t}a{x}_{s,t+1}+{\beta }_{3}Smokin\mathit{g\_f}r{q}_{it}+Demographic{s}_{it}+{\epsilon }_{ist}$$ (2)

This study utilizes a multinomial logit model to examine the relationships between state EC/cigarette taxes and smoking/vaping transitions. The estimating equations (1) and (2) show the benchmark model. $Smokin\mathit{g\_T}ransitio{n}_{ist+1}$ and $Vapin\mathit{g\_T}ransitio{n}_{ist+1}$ denote the smoking and vaping transitions (reductions, increases, or no change in use frequencies), respectively, between two consecutive waves. $E\mathit{C\_t}a{x}_{s,t+1}$ and $Ci\mathit{g\_t}a{x}_{s,t+1}$ denote EC and cigarette taxes at $\text{T}+1$, respectively, in states where individual i lived. $Vapin\mathit{g\_f}r{q}_{it}$ and $Smokin\mathit{g\_f}r{q}_{it}$ denote the individual i’s vaping frequency and smoking frequency at the baseline wave, respectively. The models also control for individual i’s sociodemographic variables at the baseline wave, including sex, age group, race/ethnicity, and annual household income. We used two measures of state EC taxes in both equations. In addition to the benchmark model, we also estimated a dynamic model, in which we replaced the EC/cigarette taxes at $\text{T}+1$ with the change in EC/cigarette taxes. This model will investigate the dynamics of EC/cigarette taxes on smoking and vaping transitions. All equations used a difference-in-difference model with repeated times, also known as a two-way fixed effects framework which controls for the time-invariant state-specific unobservable factors and the common trend across states. All regressions were weighted using longitudinal survey weights for each individual who participated in any two consecutive waves. The standard errors were clustered at the state level to adjust for intertemporal correlations over time and correlations among individuals who live in the same state. We used Stata 17 (Stata Corp, Texas US) to conduct the analyses.

3. Results

Table 1 shows the summary statistics of the study sample, which includes 2,534 individuals who participated in any two consecutive waves. Among all respondents, 14.9% reported a decrease, 6.7% reported an increase, and 78.4% indicated no change in smoking frequency. In terms of vaping frequency, 13.9% reported a decrease, 11.5% reported an increase, and 74.6% indicated no change. Figure 1 shows the distribution of outcome variables. Among those (n=378) who had a reduction in smoking frequencies, 31.23% reduced their smoking frequencies but still smoked and 68.77% quit smoking. Among those (n=352) who had a reduction in vaping frequencies, 15.92% reduced their vaping frequencies but still vaped and 84.08% quit vaping. Among those (n=170) who had an increase in smoking frequencies, 52.43% increased their smoking frequencies and 47.57% initiated smoking. Among those (n=291) who had an increase in vaping frequencies, 25.37% increased their vaping frequencies and 74.63% initiated vaping. The average standardised EC tax rate at baseline wave was $0.188 per 0.7 e-liquid mL, lower than $0.288 at T+1, indicating the average EC tax rate or the number of states that taxed ECs has increased from wave 1 to wave 3. Only 19.1% of states have implemented EC taxes at baseline wave, while this number has increased to 35.7% at T+1. State cigarette taxes have steadily increased from $1.71 per pack at baseline to $1.88 per pack at T+1. Figure 2 shows the average state cigarette taxes, average standardised EC taxes, and the number of states that had EC taxes in place at all three waves. From wave 1 to wave 3, both the average state cigarette and EC taxes had a steady increase, from $1.63 to $1.93 per pack, and from $0.11 to $0.30 per 0.7 fluid mL (=1 pack), respectively. Furthermore, there was a notable increase in the number of states implementing EC taxes, from 6 (DC, LA, MN, NC, PA, and WV) in Wave 1 to 9 (CA, DC, DE, KS, LA, MN, NC, PA, and WV) in Wave 2 and to 21 (CA, CT, DC, DE, IL, KS, LA, ME, MN, NC, NH, NJ, NM, NV, NY, OH, PA, VT, WA, WI, and WV) in Wave 3. The proportion of daily, weekly, ≤monthly vapers, and those who did not vape is 8.5%, 3.8%, 14.8%, and 72.9%, respectively. On the other hand, the proportion of daily, weekly, ≤monthly smokers, and those who did not smoke is 59.1%, 6.7%, 3.9%, and 30.4%, respectively.

Table 1:

Summary statistics of the analytical sample (n=2,534)

	# observations	Weighted Mean (SD)/%
Outcome Variables
Smoking Transition
Reduction	378	14.9%
Increase	170	6.7%
No Change	1,987	78.4%
Vaping Transition
Reduction	352	13.9%
Increase	291	11.5%
No Change	1,890	74.6%
Explanatory Variables
State EC Tax (continuous) at T		0.188 (0.530)
State EC Tax (continuous) at T+1		0.288 (0.580)
State EC Tax Implementation (binary) at T	484	19.1%
State EC Tax Implementation (binary) at T+1	905	35.7%
State Cigarette Tax at T		1.706 (1.074)
State Cigarette Tax at T+1		1.883 (1.102)
Vaping Frequency at T
Daily Vaper	215	8.5%
Weekly Vaper	96	3.8%
≤Monthly Vaper	375	14.8%
Don’t Vape	1,847	72.9%
Smoking Frequency at T
Daily Smoker	1,498	59.1%
Weekly Smoker	170	6.7%
≤Monthly Smoker	99	3.9%
Don’t Smoke	770	30.4%
Sex
Male	1,404	55.4%
Female	1,130	44.6%
Race/Ethnicity
White	1,987	78.4%
Non-White	547	21.6%
Age Group at T
18–24	233	9.2%
25–39	844	33.3%
40–54	704	27.8%
55 and Up	753	29.7%
Annual Household Income at T
Low	857	33.8%
Middle	791	31.2%
High	887	35.0%

Figure 1:

The distribution of outcome variables

Figure 2:

The trend for state cigarette and EC taxes

Table 2 shows the marginal effects of the taxes at T+1 on smoking transitions, as estimated using the benchmark multinomial logit model. Columns 2 to 4 show the results using the standardised EC tax as a regressor, whereas columns 5 to 7 present the results using binary EC tax implementation as a regressor. Respondents were more likely to reduce smoking frequencies if cigarette taxes increased. Specifically, a 10% increase in state cigarette taxes at T+1 led to an 11% increase in the likelihood of reducing smoking frequencies. However, the EC taxes at T+1, regardless of whether it is continuous or binary, did not have a significant effect on smoking transitions. Compared to those who vaped daily, those who vaped weekly and did not vape were 49% and 32%, respectively, more likely to reduce smoking frequencies. Race, sex, and income groups were not significantly associated with smoking transitions. Compared to the 18–24 age group, people from all older groups were less likely to reduce smoking frequencies and more likely to maintain unchanged behaviors.

Table 2:

Marginal effects of EC/cigarette taxes at T+1 on smoking transitions, weighted (n=2,534)

Status	Smoking Transition at T+1 (Using continuous EC taxes)			Smoking Transition at T+1 ((Using binary EC taxes))
	Reduction	Increase	No Change	Reduction	Increase	No Change
State EC Tax (continuous)	−.145 (0.118)	0.073 (0.727)	0.022 (0.504)
State EC Tax Implementation (binary)				−.283 (0.175)	−.161 (0.741)	.067 (0.247)
State Cigarette Tax	1.071*** (0.004)	−.622 (0.463)	−0.154 (0.207)	1.078*** (0.002)	−.550 (0.500)	−0.161 (0.154)
Vaping Frequency - vape daily as a comparison
Weekly Vaper	.489** (0.022)	−.056 (0.847)	−.086 (0.114)	.485** (0.023)	−.051 (0.861)	−.086 (0.112)
≤Monthly Vaper	.200 (0.282)	−.644 (0.097)	0.028 (0.588)	.197 (0.288)	−.655 (0.090)	0.030 (0.570)
Don’t Vape	.320 (0.071)	−.437 (0.131)	0.009 (0.847)	.319 (0.071)	−.441 (0.128)	0.008 (0.859)
Male	.075 (0.540)	0.109 (0.665)	−.024 (0.500)	.075 (0.541)	0.104 (0.679)	−.023 (0.507)
White	−.008 (0.960)	−0.260 (0.307)	.024 (0.580)	−.006 (0.970)	−0.259 (0.309)	.023 (0.590)
Age Group - 18–24 as a comparison
25–39	−0.514*** (0.006)	.170 (0.708)	.128 (0.066)	−0.513*** (0.005)	.181 (0.692)	.127 (0.066)
40–54	−0.592*** (<0.001)	−.125 (0.774)	0.171*** (0.002)	−0.592*** (<0.001)	−.112 (0.797)	0.170*** (0.002)
55 and Up	−0.909*** (<0.001)	−.399 (0.379)	0.239*** (<0.001)	−0.912*** (<0.001)	−.387 (0.399)	0.239*** (<0.001)
Income Group - low income as a comparison
Middle	.127 (0.526)	0.266 (0.286)	−0.045 (0.267)	.130 (0.513)	0.254 (0.306)	−0.045 (0.269)
High	.178 (0.305)	0.226 (0.372)	−.052 (0.153)	.177 (0.308)	0.218 (0.391)	−.051 (0.160)

^***Note: p<0.01

^**p<0.05.

Table 3 shows the marginal effects of taxes at T+1 on vaping transitions. We found that a 10% increase in EC taxes at T+1 increased the likelihood of increasing vaping frequencies by 4%. However, the state EC implementation and cigarette taxes did not have significant effects on vaping transitions. Those who smoked weekly and ≤monthly did not differ from those who smoked daily with respect to vaping transitions. Those who did not smoke were 107% less likely to increase vaping frequencies and 15% more likely to maintain unchanged behaviors. Race and sex were not significantly associated with vaping transitions. Compared to the 18–24 age group, people from all older groups were less likely to reduce or increase vaping frequencies and more likely to maintain unchanged behaviors.

Table 3:

Marginal effects of EC/cigarette taxes at T+1 on vaping transitions, weighted (n=2,534)

Status	Vaping Transition at T+1 (Using continuous EC taxes)			Vaping Transition at T+1 (Using binary EC taxes)
	Reduction	Increase	No Change	Reduction	Increase	No Change
State EC Tax (continuous)	−.060 (0.656)	.412** (0.034)	−.053 (0.291)
State EC Tax Implementation (binary)				−.046 (0.871)	.250 (0.634)	−.030 (0.722)
State Cigarette Tax	−.128 (0.818)	−.838 (0.101)	.149 (0.329)	−.139 (0.807)	−.753 (0.168)	.138 (0.361)
Smoking Frequency - smoke daily as a comparison
Weekly Smoker	.218 (0.242)	.072 (0.760)	−.063 (0.294)	.213 (0.251)	.082 (0.730)	−.064 (0.291)
≤Monthly Smoker	.277 (0.292)	.336 (0.229)	−.145 (0.178)	.266 (0.314)	.356 (0.197)	−.148 (0.168)
Don’t Smoke	−.153 (0.266)	−1.074*** (<0.001)	.154*** (<0.001)	−.158 (0.253)	−1.052*** (<0.001)	.152*** (<0.001)
Male	−.194 (0.219)	.070 (0.669)	.025 (0.449)	−.193 (0.218)	.065 (0.693)	.026 (0.440)
White	.123 (0.398)	−.256 (0.161)	.017 (0.668)	.121 (0.401)	−.256 (0.165)	.017 (0.664)
Age Group - 18–24 as a comparison
25–39	−1.078*** (<0.001)	−.717*** (0.005)	.649*** (<0.001)	−1.090*** (<0.001)	−.690*** (0.006)	.641*** (<0.001)
40–54	−.865*** (<0.001)	−1.530*** (<0.001)	.734*** (<0.001)	−.880*** (<0.001)	−1.494*** (<0.001)	.726*** (<0.001)
55 and Up	−.967*** (<0.001)	−1.802*** (<0.001)	.773*** (<0.001)	−.980*** (<0.001)	−1.766*** (<0.001)	.764*** (<0.001)
Income Group - low income as a comparison
Middle	−.080 (0.596)	−.037 (0.864)	.022 (0.617)	−.076 (0.620)	−.055 (0.797)	.024 (0.586)
High	−.266 (0.062)	.023 (0.910)	.045 (0.298)	−.260 (0.067)	.002 (0.991)	.047 (0.278)

Note:

^***p<0.01

^**p<0.05.

4. Sensitivity Analysis

Table 4 shows the marginal effects of the change in taxes between two consecutive waves on smoking transitions, as estimated by the dynamic model. Smoking transitions were not significantly associated with the EC tax magnitudes or implementation. State cigarette taxes were not significantly associated with smoking transitions regardless of the choice of EC taxes. The estimates of other explanatory variables are similar to Table 2. Table 5 shows the marginal effects of the change in taxes between two consecutive waves of vaping transitions. Neither the change in EC taxes nor the change in cigarette taxes had a significant effect on vaping transitions. The estimates of other explanatory variables are similar to Table 3.

Table 4:

Marginal effects on smoking transitions, dynamic model

Status	Smoking Transition at T+1 (Using continuous EC taxes)			Smoking Transition at T+1 (Using binary EC taxes)
	Reduction	Increase	No Change	Reduction	Increase	No Change
The Change in State EC Tax (continuous)	−.005 (0.814)	.065 (0.057)	−.009 (0.346)
The Change in State EC Tax Implementation (binary)				.033 (0.900)	.057 (0.897)	−.011 (0.851)
The Change in the State Cigarette Tax	.027 (0.415)	−.004 (0.921)	−.006 (0.531)	.024 (0.414)	.043 (0.395)	−.011 (0.331)
Vaping Frequency - vape daily as a comparison
Weekly Vaper	.516** (0.014)	−.084 (0.780)	−.087 (0.107)	.514** (0.014)	−052 (0.862)	−.091 (0.086)
≤Monthly Vaper	.216 (0.237)	−.639 (0.106)	.025 (0.630)	.216 (0.240)	−.639 (0.104)	.025 (0.630)
Don’t Vape	.338 (0.055)	−.422 (0.140)	−.013 (0.771)	.337 (0.056)	−.425 (0.141)	−.013 (0.778)
Male	.083 (0.498)	.125 (0.618)	−.026 (0.453)	.084 (0.490)	.113 (0.655)	−.026 (0.469)
White	−.007 (0.967)	−.275 (0.285)	.025 (0.564)	−.007 (0.965)	−.279 (0.277)	.025 (0.561)
Age Group - 18–24 as a comparison
25–39	−.553*** (0.002)	.107 (0.813)	.150** (0.030)	−.556*** (0.002)	.144 (0.749)	.147** (0.032)
40–54	−.632*** (<0.001)	−.166 (0.711)	.191*** (0.001)	−.634*** (<0.001)	−.133 (0.764)	.188*** (0.001)
55 and Up	−.954*** (<0.001)	−.460 (0.298)	.261*** (<0.001)	−.956*** (<0.001)	−.410 (0.349)	.257*** (<0.001)
Income Group - low income as a comparison
Middle	.120 (0.552)	.275 (0.280)	−.044 (0.280)	.123 (0.539)	.244 (0.339)	-.042 (0.298)
High	.181 (0.300)	.257 (0.322)	−.055 (0.139)	.183 (0.298)	.237 (0.357)	−.054 (0.146)

Note:

^***p<0.01

^**p<0.05.

Table 5:

Marginal effects on vaping transitions, dynamic model

Status	Vaping Transition at T+1 (Using continuous EC taxes)			Vaping Transition at T+1 (Using binary EC taxes)
	Reduction	Increase	No Change	Reduction	Increase	No Change
The Change in State EC Tax (continuous)	.023 (0.207)	.041 (0.188)	−.016 (0.124)
The Change in State EC Tax Implementation (binary)				−.040 (0.837)	−.310 (0.383)	.055 (0.397)
The Change in the State Cigarette Tax	−.006 (0.837)	−.038 (0.140)	.007 (0.197)	.014 (0.601)	−.019 (0.552)	.000 (0.962)
Smoking Frequency - smoke daily as a comparison
Weekly Smoker	.201 (0.277)	.083 (0.718)	−.061 (0.302)	.204 (0.268)	.082 (0.725)	−.062 (0.303)
≤Monthly Smoker	.268 (0.318)	.350 (0.205)	−.146 (0.173)	.264 (0.324)	.372 (0.168)	−.152 (0.151)
Don’t Smoke	−.156 (0.230)	−1.040*** (<0.001)	.151*** (<0.001)	−.153 (0.246)	−1.051*** (<0.001)	.151*** (<0.001)
Male	−.187 (0.245)	.061 (0.710)	.025 (0.459)	−.189 (0.236)	.058 (0.721)	.026 (0.437)
White	.119 (0.402)	−.260 (0.158)	.018 (0.639)	.114 (0.426)	−.268 (0.145)	.020 (0.599)
Age Group - 18–24 as a comparison
25–39	−1.115*** (<0.001)	−.687*** (0.006)	.651*** (<0.001)	−1.106*** (<0.001)	−.676*** (0.008)	.639*** (<0.001)
40–54	−.897*** (<0.001)	−1.478*** (<0.001)	.733*** (<0.001)	−.89*** (<0.001)	−1.470*** (<0.001)	.721*** (<0.001)
55 and Up	−1.004*** (<0.001)	−1.760*** (<0.001)	.774*** (<0.001)	−.989*** (<0.001)	−1.743*** (<0.001)	.760*** (<0.001)
Income Group - low income as a comparison
Middle	−.072 (0.647)	−.047 (0.819)	.022 (0.620)	−.078 (0.617)	−.050 (0.812)	.023 (0.594)
High	−.246 (0.080)	−.012 (0.950)	.043 (0.313)	−.252 (0.074)	.008 (0.969)	.045 (0.299)

Note:

^***p<0.01

^**p<0.05.

5. Discussion and Conclusion

The popularity of ECs has sparked a heated public debate about how to regulate these products in the interest of promoting public health. On one hand, ECs may help people who smoke cigarettes quit smoking and reduce their health risks (Beard et al., 2016; Christensen et al., 2014). On the other hand, ECs may pose risks to young people who may become addicted to them (Marques et al., 2021; Tackett et al., 2021; Vogel et al., 2020). Given the differences in the risks that cigarettes and ECs pose to human health, it is important to consider how the tax system can be designed to encourage users to switch to less harmful products or quit smoking, while also discouraging non-users from initiation. This study adds to the existing research by examining the impact of EC/cigarette excise taxes on smoking and vaping transitions using nationally representative longitudinal data of people who smoke or vape in the US.

Our findings show that higher state cigarette excise taxes led to a greater likelihood of reducing smoking frequencies for overall smoker and vaper populations. This suggests that increasing cigarette taxes continues to serve as an effective means of encouraging people who smoke to cut back on smoking or quit smoking in the US. Moreover, this effect was primarily driven by those who quit smoking because 68.77% of those who had a reduction in smoking frequencies actually quit smoking. Therefore, the US federal and state (local) governments that have authority over taxation may consider continuing to raise cigarette taxes to encourage quitting and smoking reduction among people who smoke or vape.

In addition, our results show that cigarette taxes did not have significant effects on transitions in vaping frequencies among adult smokers or vapers, suggesting that smoking frequency reduction due to cigarette taxes is likely achieved without increased vaping frequency. In other words, as cigarette taxes increase, instead of increasing vaping, adult smokers are more likely to reduce cigarette consumption, quit cold turkey, or quit using cessation aids such as nicotine replacement therapies. Nonetheless, there might still be a need to encourage complete transitions from smoking to vaping among smokers who found ECs to be an effective substitute for cigarettes. In fact, our prior study using the ITC data found that cigarette taxes are associated with perceiving cigarettes to be more costly than ECs among this population (He et al., 2024), showing the potential of using cigarette taxes to promote complete transitions from smoking to vaping. Future research is needed to ascertain this relationship.

One alternative possible explanation of the nonsignificant relationship between vaping frequency and cigarette taxes is that cigarette taxes may not be high enough as compared to EC cost to incentivize behavioral changes. Currently, although the tax incidence or burden is higher on cigarettes than on ECs, this difference is primarily driven by cigarette federal taxes. The tax incidence imposed at the state level is similar for ECs and cigarettes (Ma et al., 2022). Therefore, in order to make cigarettes much less affordable than ECs, policymakers will need to continue to raise cigarette excise taxes, particularly in areas where such taxes have not been increased in recent years. Additionally, to maintain the effectiveness of excise taxes and reduce the affordability of cigarettes, policymakers may need to index specific cigarette taxes with inflation. As the cost of living significantly increased in recent years, it is important for policymakers to adjust excise taxes accordingly to decrease the relative affordability of cigarettes compared to ECs and other consumption goods.

Another important consideration of tax design is whether EC taxes may lead to unintended consequences such as increased smoking. We found that EC taxes, regardless of the form of variables (i.e., continuous vs. binary) being used in the regression, were not significantly associated with smoking transitions. In contrast, the existing literature using the experimental marketplace and cross-sectional surveys of general populations both show that ECs and cigarettes are economic substitutes and that EC taxes may have unintended consequences of increasing cigarette use and consumption (Abouk et al., 2022; Bickel et al., 2018; Pesko et al., 2020). Therefore, although we found weak associations that may be consistent with the existing literature, these associations are non-significant for US people who smoke or vape. Further research is needed to determine whether EC taxes lead to net unintended consequences.

We also found that higher EC taxes led to a greater likelihood of increasing vaping frequencies, which is not in line with the law of demand. This effect was primarily driven by initiation among people who currently or formerly smoke. Specifically, 74.63% of the sample who increased their vaping frequencies were people who currently or formerly smoked but did not vape at the baseline. Therefore, it is possible that even with EC taxes, people who smoke remain to have incentives to increase vaping frequencies (Cotti et al., 2021). However, when the binary EC tax variable was used, implementing EC taxes was not significantly associated with vaping transitions. This insignificant association with vaping transitions was further supported by the dynamic model, in which the change in EC taxes, regardless of the form of variables being used in the regression, did not have a significant effect on vaping transitions. Therefore, the best way to describe our findings is that it is uncertain whether EC taxes in fact increase vaping frequencies.

In summary, our research suggests that there are no significant associations between EC taxes and transitions in smoking or vaping behaviors among smoking or vaping individuals in the US. This contrasts with earlier studies indicating that higher EC prices or taxes decrease EC consumption or use but potentially contribute to increased smoking among US general populations (Abouk et al., 2022; Cotti et al., 2022; Pesko et al., 2020). However, those studies used cross-sectional representative samples of general population surveys or experimental marketplace data based on hypothetical scenarios, whereas our studies are based on longitudinal data following US people who smoke or vape. Hence, it is possible that the reduction effects of EC taxes on EC or cigarette use are primarily driven by affecting transitions (e.g., initiation and relapse) among nonusers rather than by transitions among people who smoke or vape. More research is needed to assess the economic relationship between cigarettes and ECs, whether EC taxes lead to unintended consequences (i.e., increased smoking), and how the relationship and impacts differ by tobacco use status and population characteristics (e.g., age) (Bickel et al., 2018; Pope et al., 2019).

Our findings on the associations between socio-demographics and smoking and vaping transitions provide important insights for policymaking. We found that compared to the 18–24 age group, older groups were less likely to reduce use frequencies and more likely to maintain their current use status. This suggests that younger people’s nicotine and tobacco use behaviors may be less established and potentially more responsive to interventions. On the other hand, targeted efforts and innovative policy interventions may be necessary to encourage behavioral changes among older people who smoke or vape.

There are some limitations to our study. First, the ITC 4CV survey has only published three waves of data so far, which means that we were unable to control for other smoking or vaping regulatory policies, such as smoke-free air laws, in our analysis due to a lack of variation in these policies within states during the study period. However, we did control for both state- and year-fixed effects to adjust for time-invariant state-specific unobservable factors and common trends across states. Second, we used the standardised time-invariant EC taxes generated and shared by Cotti et al. (2021). If the time-invariant EC prices used by Cotti et al. are poor predictors of the EC prices, this could cause the standardised EC taxes we used and our estimates to be biased. Lastly, our transition outcome measures are defined as changes in smoking/vaping frequency categories, which do not fully capture the changes in smoking/vaping consumption. Researchers are encouraged to discover the association between excise taxes and cigarette/e-cigarette consumption. Despite these limitations, our study provides novel evidence for policymakers to consider when designing and reforming taxation systems for products that pose different risks to public health, using rigorous approaches and sensitivity analysis.