Effect of e-cigarette taxes on e-cigarette and cigarette retail prices and sales, USA, 2014–2019

Abstract

Objective:

To use a standardized e-cigarette tax measure to examine the impact of e-cigarette taxes on the price and sales of e-cigarettes and cigarettes in the US.

Design:

We used State Line versions of NielsenIQ Retail Scanner data from quarter 4 of 2014 through quarter 4 of 2019 to calculate e-cigarette and cigarette prices and sales in 23 US states. We then estimated how these outcomes are associated with standardized state-level e-cigarette taxes, controlling for state fixed effects, quarter-by-year fixed effects, cigarette taxes, other tobacco control policies, and other state-level time-varying characteristics.

Results:

A real $1 dollar increase in the e-cigarette standardized tax increases the price of one milliliter of e-liquid between $0.43 and $0.59 depending on specification. Controlling for fixed effects and cigarette taxes, a 10% increase in e-cigarette taxes is estimated to reduce e-cigarette sales by 0.5% and increase cigarette sales by 0.1%, though both results are attenuated and statistically insignificant in a model with full controls.

Conclusions:

Our study finds that e-cigarette taxes increase e-cigarette retail prices by approximately half of the tax. Further, e-cigarette taxes are associated with reduced sales of e-cigarettes and increased sales of cigarettes in some specifications. Our estimates are sizably lower than from other studies using sales and survey data.

INTRODUCTION

Tobacco product taxation is a widely adopted tobacco control strategy – particularly to reduce cigarette sales and use.[1–4] Cigarette taxes reduce cigarette use by raising cigarette prices for consumers.[1] Research suggests that the own-price elasticity of demand (the percent change in consumption given a percent change in price) for cigarettes in the short-term (i.e., over 1–2 years) is −0.4 for adults and approximately −0.70 to −0.86 for youth. In other words, for each 10% increase in price, adult use decreases by 4% and youth use decreases by 7 to 8.6%.[5, 6] The elasticity of demand for cigarettes in the long-term (the period after consumers have fully adjusted to the price change) is about twice as high.[6] Relatively little is known about how e-cigarette taxes impact the price and sales of e-cigarettes and other tobacco products. However, increasingly, research demonstrates that e-cigarette sales, like cigarette sales, decline as e-cigarette prices increase.[7–13]

Several studies have examined how the presence of an e-cigarette tax impacts e-cigarette use, finding mixed evidence of its effectiveness in reducing e-cigarette use.[14–18] One reason for the mixed evidence is that e-cigarette taxes vary significantly in magnitude and structure; therefore, measuring only the presence or absence of an e-cigarette tax may result in specification bias. Unlike cigarettes, which are almost exclusively taxed using a specific excise tax per pack (thus providing a consistent standardized measure of the tax amount), e-cigarette taxes are levied in several ways, thus complicating researchers’ ability to study the effect of e-cigarette taxes.

As of December 31, 2019, 17 states plus Washington, DC had some form of e-cigarette taxes in effect. Eleven states – Connecticut, Delaware, Kansas, Louisiana, New Mexico, North Carolina, New Jersey, Ohio, Washington, West Virginia, and Wisconsin – are classified as using a specific excise tax (i.e., e-cigarettes are taxed a certain dollar amount per unit of e-liquid or number of containers). Five states – California, Illinois, Minnesota, Pennsylvania, and Vermont – and Washington, DC are classified as using an ad valorem tax (percentage of the wholesale price). One state – New York – is classified as using a sales tax. Five of these states – Connecticut, New Mexico, New Jersey, Ohio, and Washington – applied different tax structures or rates depending on the product.[19] To aid in understanding the magnitudes of these varied e-cigarette taxation structures, Cotti and colleagues (2021) standardized United States (US) state and local e-cigarette taxes from 2010 to 2020, developing a measure indicating the tax dollar amount per milliliter of e-liquid fluid.[19] This measure unifies all tax rates and can be used to more effectively evaluate how e-cigarette taxes impact tobacco product prices and sales. Some research has begun to study the effect of these standardized e-cigarette tax rates, with six studies finding evidence that higher e-cigarette taxes reduce e-cigarette use and increase cigarette use for adults, [15] pregnant women, [20] young adults, [21] teenagers [22, 23] and overall (using sales data).[24] Two studies find e-cigarette tax increases are associated with reduced smoking cessation and increased teen smoking.[10, 25] Lastly, one study finds limited evidence that e-cigarette taxes reduce cigarette sales.[26]

Beyond studying the relationship between e-cigarette taxes and sales, two studies have explored the pass-through rate, or the rate at which e-cigarette taxes impact after-tax e-cigarette prices. One study found a pass-through rate of 0.90, meaning the full tax was not passed through to the e-cigarette price paid by consumers.[24] The second, which uses Minnesota tax variation only, found a pass-through rate of 1.33, indicating e-cigarette taxes were over-shifted to consumers.[25] Given this somewhat wide discrepancy in e-cigarette tax pass-through estimates, further research on this question is warranted.

In this study, we use the standardized US state and local e-cigarette tax measure developed by Cotti and colleagues (2021) to examine the impact of standardized state-level e-cigarette taxes on the price and sales of e-cigarettes using State Line NielsenIQ Retail Scanner data. The main contribution of this study is to use standardized state-level tax data to provide additional estimates of e-cigarette pass-through and tax responsiveness, and to compare our estimates with those from the Kilts Center Nielsen retail scanner data.[24]

DATA SOURCES AND METHODS

Data

We used State Line versions of Nielsen Retail Scanner data for e-cigarette and cigarette sales licensed from NielsenIQ (Chicago, IL). State Line data includes independent, chain and gas station convenience stores; food, drug, and mass merchandisers, discount, and dollar stores; and military commissaries. The data includes information on total sales dollars and total sales units for each universal product code (UPC) for each 4-week period of time for each of the 23 states included in the State Line file. These 23 states account for 78% of the US population as of 2019 and approximately 80% of all e-cigarette sales dollars tracked by NielsenIQ. In 2018, we observe $2.8B in e-cigarette sales revenue in the State Line file, which is estimated to be 42.4% of the national e-cigarette marketplace according to an industry report.[27] The NielsenIQ State Line data contrasts with the Kilts Center Nielsen retail scanner data by providing extrapolated state-level estimates instead of raw data for each individual retailer.

We used data for the fourth quarter of 2014 through the fourth quarter of 2019 and all price and sales data were aggregated to state-by-quarter measures to match the level of the standardized tax data. The total sample included 483 quarter state periods. During this time period, nine of the 23 states included in the State Line files effected an e-cigarette tax: California, Illinois, and Pennsylvania, which have an ad valorem tax; Louisiana, North Carolina, Ohio, and Washington, which have a specific tax; New York which has a supplemental sales tax; and New Jersey which previously only had a specific tax, but now has a two-tier mixed tax.[19, 28] We used the remaining fourteen states in the State Line file as controls: Alabama, Arizona, Colorado, Florida, Georgia, Indiana, Kentucky, Massachusetts, Michigan, Missouri, Oregon, Tennessee, Texas, and Virginia.

Measures

Dependent Variables

We constructed two dependent variables: e-cigarette sales-weighted average prices per one milliliter of e-liquid, and e-cigarette sales measured as total per capita milliliters of e-liquid. The sales-weighted average price per one milliliter of e-liquid was calculated by summing total sales dollars in each state-by-quarter and then dividing by total milliliters sold. Per capita milliliters of e-liquid sold was calculated using total milliliters of e-liquid sold in each state-by-quarter divided by state population figures from the National Cancer Institute’s Surveillance, Epidemiology and End Results Program (SEER). We focused exclusively on products containing e-liquid and excluded bar codes for cannabidiol (CBD), heated products, hardware, batteries, and starter kits with no e-liquid, which collectively accounted for 4% of sales volume. We identified milliliters of e-liquid for each of the 1,505 unique UPC codes in our data by hand-collecting information on e-liquid volume through online searches. We successfully found information for 97% of observed UPCs and the remaining 3% accounted for 0.01% of total sales dollars.

E-cigarette and Cigarette Taxes

We used publicly available e-cigarette tax data from Cotti and colleagues (2021), standardized as the tax rate per fluid milliliter. Cotti and colleagues (2021) used Kilts Center Nielsen retail scanner data and e-cigarette product characteristics to develop a method to standardize e-cigarette taxes as an equivalent average excise tax rate measured per milliliter of fluid. This variable accounts for both the magnitude and structure of the various e-cigarette taxes. Variation in the standardized e-cigarette tax comes only from effective e-cigarette tax changes. We inflation-adjust the standardized taxes, which can be visualized in Supplemental Figure 1.

Our cigarette tax measure includes the federal cigarette tax ($1.01 per pack), state cigarette taxes (from the Centers for Disease Control and Prevention State Tobacco Activities Tracking and Evaluation [STATE] System), and local cigarette taxes that are population-weighted at the locality level. Local cigarette taxes were population-weighted by multiplying the place-level tax by the place population from the American Non-Smokers’ Rights Foundation (ANRF) and dividing by the county population from the 2010 Census.

Covariates

We attempted to keep our controls as consistent as possible with Cotti et al. (2022), which studied e-cigarette and cigarette tax responsiveness in the Kilts Center Nielsen Retail Scanner data, to enable easier comparison of results. At the state level we controlled for several time-varying population and policy variables and created population-weighted variables for laws restricting indoor cigarette and e-cigarette use (separately) in bars, restaurants, and workplaces using comprehensive state and local policy data from ANRF. We population-weighted the laws applying to bars, restaurants, and workplaces equally, and applied half weights to restrictions in designated areas.[15]^,[29] We used CDC STATE System data to control for the presence of an e-cigarette minimum legal sales age law and for cigarette sales ages of 18, 19, or 21. We also used information from Tobacco21.org to create the percentage of the state population covered by Tobacco 21 laws (including local laws).[30] To control for states that have restricted the sale of flavored tobacco products by the end of 2019 (MA, MI, OR, and WA), we constructed a measure capturing the percent of each quarter in each state with a flavor ban.

To capture each state’s economic environment, we also controlled for the unemployment rate, the percent of residents living below the poverty line, and the effective minimum wage (defined as the larger figure between the federal minimum and state minimum wage).[31] Additionally, we used American Community Survey data to control for the mean total pre-tax personal annual income and the percent of uninsured adults in each state. Lastly, we created a dichotomous variable equal to one if a state expanded Medicaid through the Affordable Care Act.[32]

To control for policies affecting potential substitutes and complements for tobacco, we controlled for state beer excise taxes and created two dichotomous indicators to capture states with medical and recreational marijuana laws in effect.[33, 34]

Empirical methods

We used a two-way fixed effects model and estimated the following regression:

$$Y_{st}={\beta }_0+{\beta }_1{ECigtax}_{st}+{\beta }_2{Cigtax}_{st}+{\beta }_3{X}_{st}+{\beta }_4{W}_{st}+{\delta }_s+{\gamma }_t+{\epsilon }_{st}$$

where ${\mathrm{Y}}_{\mathrm{s}\mathrm{t}}$ is one of two outcomes: average sales-weighted price of one milliliter of e-liquid, or per capita e-liquid milliliters sold, in state $s$ at quarter $t$. For supplemental analyses, we also create cigarette outcomes of the average sales-weighted price of a pack of cigarettes and per capita packs of cigarettes sold. We used data for 23 states and 21 year-quarters, thus providing an analytic sample of 483. We run five models for each outcome. First, we regressed e-cigarette taxes (${ECigtax}_{st}$) alone. Second, we regressed cigarette taxes (${Cigtax}_{st}$) alone. Our third model included both e-cigarette taxes (${Ecigtax}_{st}$) and cigarette taxes (${Cigtax}_{st}$). Our fourth model included $X_{st}$, which is a vector of tobacco control policy variables described earlier, and our last model included a vector of other state-level characteristics, $W_{st}$, which controls for remaining time-varying variables previously described. We included state population in vector $W_{st}$ for price regressions, but did not include it in the per capita sales regressions since population was included as the denominator of the outcome. Lastly, we included state and quarter-by-year fixed effects (i.e., indicators for each state or each of the 21 unique year-quarters) in all models, represented by ${\delta }_s$ and ${\gamma }_t$ respectively. We included state fixed effects to control for any time-invariant state-level factors (e.g., fixed anti-smoking sentiment), and quarter-by-year fixed effects to control for any time-varying national factors (e.g., changes in Food and Drug Administration [FDA] tobacco regulation). In addition to reporting coefficients that represent the effect of a $1 change in tax rate, we also present coefficients as elasticities at the mean using Stata’s -margins- command. For all analyses, we clustered our standard errors at the state level and describe results as statistically significant if they exceed a p-value of 0.05 using a two-tailed test.

RESULTS

Summary Statistics

Table 1 presents summary statistics for all 23 states. States with an e-cigarette tax by the end of the study period have higher average e-cigarette prices than states without an e-cigarette tax. Per capita e-cigarette milliliters sold is about five percent lower in states that have an e-cigarette tax. The unconditional mean (calculated for all states when the tax is in place) for the average e-cigarette tax is 23 cents; however, the conditional mean (calculated only for states with a tax in place at that time) is more than double at 51 cents (data not shown). To further contextualize changes over time in standardized real e-cigarette taxes and real sales-weighted e-cigarette prices we have plotted the first two measures for states that tax e-cigarettes in Supplemental Figures S1 and S2. Large e-cigarette taxes in California, Illinois, and Pennsylvania appear to be reflected in large e-cigarette prices over time in these states.

Table 1.

Summary Statistics

	All states	States that tax e-cigarettes	States that do not tax e-cigarettes	Student T-Test Difference in Means
Real sales-weighted average Prices (2019 Dollars)
E-cigarette price (per one milliliter of e-liquid)	$4.79 (0.92)	$5.03 (1.02)	$4.64 (0.82)	−0.39*** (0.08)
Cigarette price (per one pack)	$6.38 (1.48)	$7.24 (1.55)	$5.83 (1.14)	−1.40*** (0.12)
Per capita sales
E-cigarette milliliters of e-liquid	0.38 (0.30)	0.36 (0.27)	0.38 (0.32)	0.02 (0.28)
Cigarette packs+	9.65 (4.96)	7.64 (3.46)	10.99 (5.34)	3.31*** (0.44)
Taxes (2019 Dollars)
E-cigarette taxes (per one milliliter of e-liquid)	$0.09 (0.31)	$0.23 (0.46)	-	−0.23*** (0.03)
Cigarette taxes (per one pack)	$2.79 (1.29)	$3.58 (1.43)	$2.29 (0.89)	−1.29*** (0.11)
Tobacco Control Policies
Restrictions on indoor smoking (% of population covered)	0.78 (0.22)	0.91 (0.12)	0.70 (0.23)	−0.22*** (0.02)
Restrictions on indoor e-cigarette use (% of population covered)	0.25 (0.33)	0.39 (0.38)	0.16 (0.25)	−0.23*** (0.03)
Law establishing e-cigarette minimum legal sales age (ref: no law)	0.95 (0.23)	0.96 (0.20)	0.94 (0.24)	−0.02 (0.02)
State cigarette minimum legal sales age	18.34 (0.88)	18.50 (1.07)	18.23 (0.71)	−0.27*** (0.08)
Tobacco 21 laws (% of population covered)	0.15 (0.30)	0.23 (0.36)	0.10 (0.24)	−0.14*** (0.03)
Temporary e-cigarette sales bans (% of quarter ban in effect)	0.004 (0.06)	0.005 (0.07)	0.003 (0.05)	−0.001 (0.005)
State-Level Characteristics
Expansion of Medicaid through the ACA (ref: no expansion)	0.64 (0.48)	0.85 (0.36)	0.51 (0.50)	−0.34*** (0.04)
Beer taxes (2019 dollars per gallon)	$0.33 (0.35)	$0.26 (0.16)	$0.38 (0.42)	0.13*** (0.03)
Minimum wage (2019 dollars)	$8.84 (1.46)	$9.13 (1.52)	$8.66 (1.39)	−0.47*** (0.13)
Unemployment rate (%)	4.63 (0.95)	4.99 (0.77)	4.40 (0.98)	−0.58*** (0.08)
Poverty rate (%)	12.50 (2.93)	12.48 (3.25)	12.51 (2.70)	0.03 (0.27)
Pre-tax personal income (2019 dollars)	$40,909 (5,565)	$42,843 (5,487)	$39,667 (5,260)	−3175.94*** (498.81)
Uninsured (%)	0.09 (0.03)	0.08 (0.02)	0.10 (0.04)	0.02*** (0.003)
Recreational marijuana law in effect	0.18 (0.39)	0.18 (0.38)	0.19 (0.39)	0.01 (0.04)
Medical marijuana law in effect	0.47 (0.50)	0.58 (0.49)	0.40 (0.49)	−0.17*** (0.05)
Population	11,194,084 (8,319,679)	14,136,106 (9,734,534)	9,302,785 (6,625,012)	−4,833,321*** (744,542)
Observations	483	189	294	-

The unit of observation is quarter year per state. Standard deviations are presented in parenthesis. For difference in means standard errors are presented in parenthesis.

⁺Variable used in supplemental tables.

^***p<0.01,

^**p<0.05,

^*p<0.1

Supplemental Figure S3 compares average e-cigarette taxes, prices, and per capita milliliters of e-liquid sold for states that tax e-cigarettes versus states that do not tax e-cigarettes. Here, it is difficult to detect significant divergence over time in prices for states with e-cigarette taxes compared to states without e-cigarette taxes, but starting in 2019 states without e-cigarette taxes began to experience faster e-cigarette sales increases relative to states with e-cigarette taxes.

Estimates of the effect of e-cigarette and cigarette taxes on e-cigarette prices

Table 2 presents estimates of the effect of e-cigarette and cigarette taxes on e-cigarette prices from our two-way fixed effects models. Specifically, a real one dollar increase in the e-cigarette standardized tax increases the sales-weighted average price of one milliliter of e-liquid by $1.01 in our first model, where we only regress e-cigarette taxes on e-cigarette prices with state and quarter-by-year fixed effects. However, in column 3, simply controlling for cigarette taxes causes this coefficient to more than halve to $0.43, which is no longer statistically significant. This large change suggests that e-cigarette taxes and cigarette taxes are correlated and so controlling for the other is needed to isolate the effect of the former. Therefore, in focusing on columns 3 to 5, which describe models that include both taxes, we find that a $1 increase in e-cigarette taxes increases e-cigarette prices by between 43 to 59 cents. With regards to cigarette taxes, we find that a real one dollar increase in cigarette taxes increases the sales-weighted average price of one milliliter of e-liquid by 49 to 57 cents.

Table 2.

Estimates of the effect of e-cigarette taxes on e-cigarette prices using a two-way fixed effects model: NielsenIQ Retail Sales Data 2014 Q4 – 2019 Q4

Outcome:	E-Cigarette Real Price (One ML)
Column:	1	2	3	4	5
E-cigarette tax rate (2019 Q4 dollars)	1.01*** (0.104)		0.43 (0.281)	0.49* (0.280)	0.59** (0.220)
Cigarette tax (2019 Q4 dollars)		0.85*** (0.101)	0.55** (0.246)	0.57* (0.284)	0.49* (0.235)
Covariates:
State fixed effects	Y	Y	Y	Y	Y
Quarter-by-year fixed effects	Y	Y	Y	Y	Y
Tobacco Control Characteristics				Y	Y
State-Level Characteristics					Y
Observations:	483	483	483	483	483
Coefficient of Determination	0.786	0.790	0.793	0.807	0.835

^***p<0.01,

^**p<0.05,

^*p<0.1

Robust standard errors in parentheses; standard errors are clustered at the state level.

Columns 4 and 5 include tobacco control time-varying controls: index of indoor smoking restrictions, index of indoor vaping restrictions, temporary vape ban, e-cigarette MLSA, cigarette MLSA and T21 population coverage.

Column 5 includes state-level time-varying controls: state ACA Medicaid expansion coverage, recreational marijuana law, medical marijuana law, beer tax, minimum wage, poverty rate, unemployment rate, mean total pre-tax personal income, uninsured rate, and population.

Results showing the effect of taxes on cigarette prices can be found in Supplemental Table S1, where we find evidence that a one dollar increase in cigarette taxes increases the price of one pack of cigarettes by between $0.90 and $1.03 depending on specification. In models with time-varying controls, we find small statistically insignificant effects of e-cigarette tax rates on cigarette prices.

Estimates of the effect of e-cigarette and cigarette taxes on e-cigarette per capita sales

Table 3 presents estimates of the effect of taxes on per capita e-cigarette sales. We find that e-cigarette taxes reduce per capita e-cigarette sales (columns 1, 3, and 4). Focusing on columns 3 to 5 that include both tax rates at the same time, our results suggest that a one dollar increase in the e-cigarette tax rate decreases total per capita sales by 0.09 to 0.21 milliliters, which translate to own-tax elasticities of demand of −0.02 to −0.05 for e-cigarettes. Results showing the effect of taxes on per capita cigarette sales can be found in Supplemental Table S2, where we find that cigarette taxes reduce per capita cigarette sales in columns 3 and 4. In those same columns, e-cigarette taxes have large positive effects on cigarette sales, and this cross-tax elasticity is statistically significant in column 3. In the last column, adding additional state-specific controls sharply attenuates both tax coefficients and neither are statistically significant.

Table 3.

Estimates of the effect of e-cigarette taxes on total per capita e-cigarette sales using a two-way fixed effects model: NielsenIQ Retail Sales Data 2014 Q4 – 2019 Q4

Outcome:	Total Per Capita E-Cigarettes ML Sold
Column:	1	2	3	4	5
E-cigarette tax rate (2019 Q4 dollars)	−0.14*** (0.040)		−0.21** (0.087)	−0.17** (0.064)	−0.09 (0.073)
Cigarette tax (2019 Q4 dollars)		−0.08** (0.036)	0.06 (0.063)	0.06 (0.047)	0.02 (0.042)
Covariates:
State fixed effects	Y	Y	Y	Y	Y
Quarter-by-year fixed effects	Y	Y	Y	Y	Y
Tobacco Control Characteristics				Y	Y
State-Level Characteristics					Y
Observations:	483	483	483	483	483
Coefficient of Determination	0.823	0.818	0.824	0.852	0.880
E-cigarette own-tax elasticity	−0.03***	n/a	−0.05**	−0.04**	−0.02
E-cigarette own-tax elasticity (Cotti et. al 2022)	n/a	n/a	−0.67	−0.64	−0.63
Cigarette cross-tax elasticity	n/a	−0.59**	0.48	0.45	0.11
Cigarette cross-tax elasticity (Cotti et. al 2022)	n/a	n/a	0.78	0.78	0.83

Notes:

^***p<0.01,

^**p<0.05,

^*p<0.1

Robust standard errors in parentheses; standard errors are clustered at the state level.

Columns 4 and 5 include tobacco control time-varying controls: index of indoor smoking restrictions, index of indoor vaping restrictions, temporary vape ban, e-cigarette MLSA, cigarette MLSA and T21 population coverage. Column 5 included state-level time-varying controls: state ACA Medicaid expansion coverage, recreational marijuana law, medical marijuana law, beer tax, minimum wage, poverty rate, unemployment rate, mean total pre-tax personal income, and uninsured rate.

Tax elasticities are calculated at the means using the margins command from Stata 17.

Table 4 presents estimates of the effect of prices as calculated from the Nielsen data on per capita e-cigarettes sales. These results may be biased by concerns related to endogeneity of prices,[35] but are useful for comparison to other price elasticity estimates from the literature that are potentially biased in the same way. We do not find that prices have a statistically significant (p<0.05) marginal effect on per capita e-cigarette sales. E-cigarette own-price elasticities range between −0.42 to −0.75 depending on model. As shown in Supplemental Table S3, we do not find that prices have a statistically significant effect (p<0.05) on per capita cigarette sales.

Table 4.

Estimates of the effect of e-cigarette prices on total per capita e-cigarette sales using a two-way fixed effects model: NielsenIQ Retail Sales Data 2014 Q4 – 2019 Q4

Outcome:	Total Per Capita E-Cigarettes ML Sold
Column:	1	2	3	4	5
E-cigarette real price (2019 Q4 dollars)	−0.06 (0.036)		−0.05 (0.036)	−0.03 (0.033)	−0.06* (0.028)
Cigarette real price (2019 Q4 dollars)		−0.07 (04.045)	−0.03 (0.046)	−0.03 (0.039)	0.01 (0.034)
Covariates:
State fixed effects	Y	Y	Y	Y	Y
Quarter-by-year fixed effects	Y	Y	Y	Y	Y
Tobacco Control Characteristics				Y	Y
State-Level Characteristics					Y
Observations:	483	483	483	483	483
Coefficient of Determination	0.821	0.817	0.822	0.850	0.884
E-cigarette own-price elasticity	−0.74	n/a	−0.63	−0.42	−0.75**
Cigarette cross-price elasticity	n/a	−1.12	−0.49	−0.50	0.16

Notes:

^***p<0.01,

^**p<0.05,

^*p<0.1

Robust standard errors in parentheses; standard errors are clustered at the state level.

Columns 4 and 5 include tobacco control time-varying controls: index of indoor smoking restrictions, index of indoor vaping restrictions, temporary vape ban, e-cigarette MLSA, cigarette MLSA and T21 population coverage. Column 5 included state-level time-varying controls: state ACA Medicaid expansion coverage, recreational marijuana law, medical marijuana law, beer tax, minimum wage, poverty rate, unemployment rate, mean total pre-tax personal income, and uninsured rate.

Tax elasticities are calculated at the means using the margins command from Stata 17.

DISCUSSION

The recent emergence of products such as e-cigarettes that are potentially less harmful than conventional cigarettes has raised questions about how best to tax these new products. While all 50 states and DC currently impose a specific excise tax on cigarettes, many states do not tax e-cigarettes and those that have imposed taxes are taking diverse approaches with respect to structure and rates. Notably for our study period most variation in e-cigarette tax rates were predominantly driven by California, Pennsylvania, and Illinois.

Our study investigates the effects of e-cigarette and cigarette taxes on the prices and sales of e-cigarettes using a unique standardized e-cigarette tax rate developed by Cotti and colleagues.[19]

We find evidence that e-cigarette taxes are passed on to consumers in the form of higher prices but our estimates suggest that roughly only half of the tax is passed on. Our estimate of the pass-through rate of e-cigarette taxes is smaller than those of two other studies (0.90 and 1.33).[24, 25] Descriptive evidence suggests potential heterogeneity in pass-through rates across states. While outside the scope of the present study to speculate on why pass-through rates may be higher for some states and lower for others, one potential explanation is potential differences in demand and supply conditions and market competitiveness across states.[36] Our estimates of the pass-through rate of cigarette taxes (between 0.90 to 1.03) is consistent with previous literature. Some studies have concluded that cigarette taxes are nearly fully shifted to consumer prices,[37–39] whereas other studies have concluded that cigarette taxes are more than fully shifted to consumer prices.[40–45]

We also find an inverse relationship between e-cigarette taxes and per capita sales of e-cigarettes in all the models we estimated. Our results suggest that the own-tax elasticity of demand for e-cigarettes ranges between −0.02 to −0.05. This is considerably smaller than a tax elasticity estimate ranging from −0.63 to −0.67 from Cotti et al. (2022) using the Kilts Center Nielsen retail scanner data. Additionally, the current study’s own-tax elasticity estimate is considerably smaller than current estimates of e-cigarette tax elasticities for past 30-day use from studies using survey data sources: −0.118 (for all adults),[15] −0.201 (for adults <40 years of age),[15] −0.539 (for adults 18–25 years of age),[21] −0.075 to −0.164 (for teens),[22] and −0.29 for pregnant women.[20]

The paper’s current methods were designed to replicate the Cotti et al. (2022) methods as closely as possible for ease of comparing results, with the only real difference being data source and the latter paper by Cotti and colleagues using data starting seven quarters earlier than ours. This makes the large differences in e-cigarette tax elasticity estimates across studies more surprising and interesting. Since only Minnesota had an e-cigarette tax during these early seven quarters not included in our study, this is unlikely to have a large impact. More significantly, the coverage across data sources differs; the State Line sample contains data on 23 states (nine adopting e-cigarette taxes), whereas the Kilts Center Nielsen retail scanner data employed by Cotti and colleagues (2022) uses data from all states except Alaska and Hawaii (seventeen states + 2 counties + 1 city adopting e-cigarette taxes).[24] Finally, the Kilts Center Nielsen retail scanner data is of raw store-level data, whereas the State Line data uses the same raw data but extrapolates these to provide state estimates. Little is understood about how Nielsen performs their extrapolations, but if the extrapolation method relies heavily on interpolation this could cause delays before e-cigarette tax effects appear in the data, and result in attenuated estimates.[46] One other noteworthy difference is that the Cotti et al. (2022) elasticities as reported in their Table 2 are relatively more stable than the elasticities reported in our study, which may be driven in part by having the current study having fewer states and shorter time horizon.

We identify several additional limitations. First, the data we use captures retail-based e-cigarette sales in 23 states (roughly 42.4% of the e-cigarette market) and do not capture online and/or vape shop sales, which are estimated to have about 20–30% market share.[47] Therefore, our estimates may not be generalizable to all states or types of e-cigarette sales. Second, sales data do not permit studying the effect of taxes on use behaviors or among specific demographic subgroups. Third, we use the standardized taxation rate made publicly available by Cotti et al. (2021) that use time-invariant prices to reduce bias from factors that change prices over time, but if the time-invariant prices used by Cotti et al. (2021) (from the Kilts Center Nielsen retail scanner data) are poor predictors of time-invariant prices from the Nielsen State Line data, this could cause our estimates, particularly pass-through estimates, to be biased.

CONCLUSION

Our study uses standardized e-cigarette tax rate data to provide further insight of the effect of e-cigarette taxes on e-cigarette and cigarette prices and sales. We find evidence that e-cigarette taxes are under-shifted to e-cigarette prices. We also offer surprising evidence that cigarette taxes increase e-cigarette prices, which has not been previously documented in the literature. Our study finds much smaller own- and cross-tax elasticities than a previous study using Kilts Center Nielsen retail scanner data[24], thus highlighting the importance of understanding the nuanced differences across sales data sources when conducting tobacco policy research. Additionally, the current study finds smaller own- and cross-tax elasticities than other studies using survey data sources.[15, 20–22]

The e-cigarette market is rapidly changing. Ongoing research will be necessary to assess the impact of e-cigarette taxes in a dynamic environment and inform policy makers’ decisions about how to effectively tax these products to maximize public health goals.

What is already known on this topic:

Early research provides evidence that e-cigarette taxes are associated with increased e-cigarette prices, decreased e-cigarette sales/use, and increased cigarette sales/use.

What this study adds:

One limitation of the previous sales data source used to study these questions (the Kilts Center version of the Nielsen retail scanner data) is that it covers a small share of the e-cigarette marketplace. We re-explore these questions of e-cigarette tax pass-through and responsiveness using NielsenIQ State Line data that has a different set of benefits and limitations, with the chief benefit being it covers a larger share of the e-cigarette market, though for a smaller number of states.

How this study might affect research, practice, or policy:

Our study provides evidence that e-cigarette taxes increase e-cigarette prices and reduce e-cigarette sales by sizably lower levels than previously estimated. Our study also compares the Kilts Center version of the Nielsen Retail Scanner data with the NielsenIQ State Line data to unpack potential reasons why estimates may vary.