Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 May 1.
Published in final edited form as: Tob Control. 2020 Apr 8;30(3):264–272. doi: 10.1136/tobaccocontrol-2019-055478

Taxation reduces smoking but may not reduce smoking disparities in youth

Nancy L Fleischer 1, J Travis Donahoe 2, M Chandler McLeod 3, James F Thrasher 4,5, David T Levy 6, Michael R Elliott 7,8, Rafael Meza 9, Megan E Patrick 10
PMCID: PMC7546443  NIHMSID: NIHMS1627309  PMID: 32269173

Abstract

This study examines the extent to which cigarette taxes affect smoking behavior and disparities in smoking among adolescents by gender, socioeconomic status (SES), and race/ethnicity. We used US nationally-representative, repeated cross-sectional data from the 2005-2016 Monitoring the Future study to evaluate the relationship between state cigarette taxes and past 30-day current smoking, smoking intensity, and first cigarette and daily smoking initiation using modified Poisson and linear regression models, stratified by grade. We tested for interactions between tax and gender, SES, and race/ethnicity on the additive scale using average marginal effects. We found that higher taxes were associated with lower smoking outcomes, with variation by grade. Across nearly all of our specifications, there were no statistically significant interactions between tax and gender, SES, or race/ethnicity for any grades/outcomes. One exception is that among 12th graders, there was a statistically significant interaction between tax and college plans, with taxes being associated with a lower probability of 30-day smoking among students who definitely planned to attend college compared to those who did not. We conclude that higher taxes were associated with reduced smoking among adolescents, with little difference by gender, SES, and racial/ethnicity groups. While effective at reducing adolescent smoking, taxes appear unlikely to reduce smoking disparities among youth.

Keywords: disparities, priority/special populations, prices, taxes

Disparities in smoking are pervasive among adolescents in the United States. Adolescents from lower socioeconomic status (SES) households and those who do not plan to attend college have worse smoking outcomes than their more affluent peers [1-3]. Important disparities in smoking also exist among adolescents across gender and race/ethnicity. High school males are more likely to smoke than females [4]. Adolescent non-Hispanic Blacks (hereafter, Blacks) have lower smoking prevalence and probability of initiation than Hispanics and non-Hispanic Whites (hereafter, Whites), although these differences have narrowed over time [4] and non-smoking Hispanic and Black youth have higher susceptibility to future smoking than Whites [2]. Asian Americans have the lowest smoking prevalence [4], probability of initiation [4], and susceptibility to smoking [2].

Despite the well-documented disparities in youth smoking, there is limited evidence on how tobacco policies affect these disparities—especially among adolescents. Prior studies have found that cigarette taxes reduce smoking among adolescents [5-25], although the effectiveness of taxes may be waning in more recent years [26]. However, relatively few studies have examined whether the associations of taxation vary by SES, race/ethnicity, or gender. Despite continued higher smoking rates among low SES youth, studies have found that lower SES youth are generally more price responsive [21, 27], as has been found among low SES adult smokers [28, 29], and therefore should be more likely to be influenced by taxes. Results by race/ethnicity have been mixed: several studies found that racial minority youth were more price responsive than their White counterparts [21, 24, 25, 30], although another study found White adolescents to be more price responsive than Blacks [22]. Results by gender are also mixed, with some studies finding that male adolescents are more price responsive [30, 31], whereas others find the opposite [21, 22, 24].

Prior literature has found mixed results in terms of the effectiveness of taxation on smoking behavior by SES, race/ethnicity, and gender, and few studies have been conducted using data from more recent years. To address these gaps, we used data from the 2005-2016 Monitoring the Future (MTF) surveys to assess the relationship between state cigarette taxes and price, cigarette smoking, and disparities in smoking by SES, race/ethnicity, and gender, stratified by grade.

METHODS

Sample

We obtained restricted access to MTF data from 2005-2016 that included information on each respondent’s smoking behavior, parental education, college plans for 12th graders, mother’s employment status, race/ethnicity, gender, state of residence, and school zip code. Since 1991, MTF has collected nationally representative cross-sections from the 48 contiguous states of 8th, 10th, and 12th graders [3, 32]. While the data are nationally representative, they are not state representative. We used data between 2005 to 2016 to focus on the effects of recent changes in tax on smoking disparities. Analytic sample sizes varied by outcome variable and grade. Due to the use of de-identified data, the University of Michigan Institutional Review Board deemed this research exempt.

Adolescent smoking

The primary dependent variables were past 30-day cigarette smoking (did vs. did not smoke at least one cigarette in the past 30 days); the number of cigarettes smoked per day among respondents who reported any smoking (conditional intensity); initiation for first time smoking in current grade (smoked first cigarette in current grade vs. never smoked a cigarette); and initiation for daily smoking in current grade (started smoking daily in current grade vs. never smoked daily). First cigarette and daily smoking initiation were only assessed among respondents who had not smoked prior to the current grade.

Cigarette taxes and price

The key independent variables were state cigarette taxes and average cigarette sale price (per pack of 20 cigarettes) from the Tax Burden on Tobacco (TBT) [33]. MTF surveys were administered in the spring each year, and students were asked retrospectively about smoking during their current grade. Therefore, we constructed tax and price variables to reflect taxes and prices that students would have been exposed to over the academic year. State and federal cigarette taxes (recorded each year in the TBT for fiscal years ending June 30) were matched to students based during the same year they were surveyed. For prices (recorded on November 1), we used the average of the prices from the November before and after students were surveyed. All dollar amounts were adjusted for inflation to 2016 dollar values using an implicit price deflator [34].

Gender, socioeconomic status, and race/ethnicity

Gender was coded as male or female. We included two markers for SES: parental education and plans to attend college for a four-year degree (12th grade sample only). Parental education was the highest education attained by either of the respondent’s parents coded as: less than high school, high school graduate, some college, or college or higher. For plans to attend college, 12th graders were classified into “definitely will,” “probably will/won’t,” and “definitely won’t” based on their response to whether they expect to attend a four-year college program [3]. Plans to attend college is not a traditional marker of SES and may be correlated with individual time preferences [35]; however, more traditional markers of SES, such as parental education, occupation, and income, affect students’ plans to attend college [36, 37]. Race and ethnicity were self-reported and combined into a single variable defined as non-Hispanic White, non-Hispanic Black, Hispanic, Asian, and other non-Hispanic (including mixed race).

Covariates

To control for individual factors other than gender, SES, and race/ethnicity, we included variables for living arrangements (living alone, with mother or father only, or with both parents), mother’s employment status (grade 8/10: no, part-time, full-time currently, grade 12: no, some, most, all of time growing up), and type of high school (prep, general, or vocational).

To control for time-varying state-level factors that might affect adolescent smoking and exposure to higher state cigarette tax/price, we included variables for state/year unemployment and poverty rates from University of Kentucky’s Center for Poverty Research [38], percent Hispanic and Black population calculated from Survey of Epidemiology and End Results population estimates [39], and percent high school graduates (age 25+) and college graduates (25+) from the American Community Survey [40]. To control for other factors in the tobacco policy environment, we included the percentage of residents covered by workplace and hospitality laws at the county level [41]. To control for anti-smoking sentiment, we used the percentage of adults in each state with home smoking bans, and the percentage by state supporting bans on smoking in bars, calculated from the Tobacco Use Supplement to the Current Population Survey with linear interpolation between survey waves [42, 43]. Finally, we included covariates to control for the four census regions (Northeast, Midwest, South, and West) and year.

Statistical analysis

All analyses were stratified by grade due to variation in price sensitivity by age [22, 44, 45] and interactions between tax and/or price for several outcomes, and accounted for MTF’s complex survey design using strata, school cluster, and individual sample weights. Because MTF does not sample states as a first stage, we did not cluster at the state level. To examine the relationships between cigarette tax and price with smoking participation (past 30-day) and initiation (both first cigarette and daily), we calculated risk ratios using modified Poisson regression with a sandwich-type variance estimator [46]. Linear regression was used to examine conditional smoking intensity. We then separately tested for interactions between either tax or price and gender, SES, and race/ethnicity. We tested interactions on the additive scale by predicting marginal probabilities for our population with covariates at their measured values (known as average marginal effects, AME) [47-49]. We plotted AMEs of tax/price on smoking for each category of gender, SES, or race/ethnicity where the interactions were statistically significant after adjusting for multiple testing using the false discovery rate of Benjamini-Hochberg at 5% [50]. Finally, we calculated elasticities by grade and by sociodemographic subgroups within grades.

Data were missing for 2% to 12% of the variables across samples, particularly for parental education (9% overall, 7%-12% by grade) and gender (4% overall, 3%-7% by grade). To correct for missing values, we performed multiple imputation using IVEware 0.3 [51] using sequential regression imputation under the missing at random assumption [52]. In total, 10 datasets were imputed separately by grade using all covariates (except region) shown in Table 1, as well as indicators for year, school type, ever smoked, smoking participation in last 30 days, first cigarette smoked in current grade, began smoking daily in current grade, ≥5 drinks in a row over last 2 weeks, and marijuana use in last 30 days.

Table 1.

Weighted Descriptive Statistics for all 8th, 10th, and 12th Graders, Monitoring the Future Data, 2005-2016.

Entire Population (30-day Smoking
Participation Analytic Sample)		 Population of Past 30 Day Smokers
(Conditional Intensity Analytic
Sample)
Variables Grade 8
wt. %		 Grade 10
wt. %		 Grade 12
wt. %		 Grade 8
wt. %		 Grade 10
wt. %		 Grade 12
wt. %
Gender
Female 49% 49% 48% 49% 47% 42%
Male 47% 48% 45% 46% 50% 51%
Missing 4% 3% 7% 5% 3% 7%
Race/Ethnicity
Non-Hispanic White 50% 57% 57% 56% 67% 69%
Non-Hispanic Black 13% 11% 12% 8% 6% 6%
Hispanic 18% 14% 14% 16% 11% 11%
Non-Hispanic Asian 4% 4% 4% 1% 2% 2%
Non-Hispanic Other 11% 10% 9% 15% 11% 9%
Missing 4% 3% 5% 4% 2% 4%
Education, Parents' Highest
Less than High School 8% 7% 8% 13% 9% 8%
High School 16% 17% 18% 23% 22% 21%
Some College 14% 17% 19% 17% 19% 21%
College or Greater 50% 53% 47% 35% 43% 44%
Missing 12% 7% 8% 12% 6% 7%
College Plans (Grade 12)
No, Probably/Definitely 16% 26%
Yes, Probably 21% 24%
Yes, Definitely 55% 42%
Missing 8% 8%
Living Arrangement
Neither Mother or Father in Household 4% 4% 6% 8% 7% 9%
Lives with Father 4% 4% 5% 7% 7% 7%
Lives with Mother 18% 18% 21% 23% 21% 22%
Lives with Father and Mother 71% 73% 64% 58% 63% 58%
Missing 3% 2% 5% 3% 2% 4%
Mother's Current Employment (Grade 8/10)
Not Employed 21% 21% 24% 22%
Part Time 19% 16% 17% 14%
Full Time 57% 61% 55% 61%
Missing 4% 3% 4% 2%
Mother's Past Employment (Grade 12)
None 13% 12%
Sometimes 18% 18%
Most of Time 17% 18%
All the Time 47% 48%
Missing 5% 4%
High School Program
College Prep. 33% 46% 49% 20% 31% 38%
General 17% 24% 32% 20% 31% 38%
Vocational/Technical 4% 4% 5% 7% 8% 7%
Other/Don't Know 41% 23% 9% 48% 28% 11%
Missing 5% 3% 6% 5% 2% 6%
Census Region
Northeast 17% 20% 18% 13% 18% 18%
Midwest 22% 24% 23% 24% 27% 27%
South 38% 33% 37% 46% 38% 38%
West 23% 23% 22% 17% 17% 17%
Smoking Participation (past 30 day)
No 91% 86% 79
Yes 6% 11% 17%
Missing 3% 3% 3%
Smoking Intensity (cigarettes/day)
Mean (SE), range 4.0 (8.3), 0.5-40.0 4.1 (7.4), 0.5-40.0 5.0 (7.9), 0.5-40.0
0.5 58% 52% 45%
2.5 27% 30% 30%
10 7% 10% 14%
20 3% 4% 7%
30 2% 1% 2%
40 3% 2% 2%
State tax (mean $ (SE), range) 1.4 (1.0), 0.1-4.7 1.4 (0.9), 0.1-4.7 1.4 (1.0), 0.1-4.7 1.2 (0.8), 0.1-4.7 1.3 (0.9), 0.1-4.7 1.3 (0.9), 0.1-4.7
State price (mean $ (SE), range) 5.8 (1.4), 3.6-10.5 5.7 (1.3), 3.6-10.5 5.8 (1.4), 3.6-10.5 5.4 (1.2), 3.6-10.5 5.5 (1.2), 3.6-10.5 5.6 (1.3), 3.6-10.5
Smoke-free workplace law county coverage (mean %, (SE), range) 52.3 (46.5), 0-100 50.9 (46.9), 0-100 (46.8), 0-100 (46.7), 0-100 43.0 (46.9), 0-100 47.0 (47.2), 0-100
Smoke-free hospitality law county coverage (mean %, (SE), range) 62.7 (46.1), 0-100 65.3 (46.0), 0-100 62.9 (46.6), 0-100 47.9 (47.9), 0-100 55.1 (48.0), 0-100 53.8 (48.0), 0-100
State Unemployment (mean %, (SE), range) 6.6 (2.1), 2.6-13.7 6.7 (2.3), 2.6-13.7 6.7 (2.2), 2.6-13.7 6.6 (2.2), 2.6-13.7 6.7 (2.3), 2.6-13.7 6.7 (2.2), 2.6-13.7
State Poverty (mean %, (SE), range) 13.8 (2.9), 5.6-23.1 13.6 (3.0), 5.4-23.1 13.9 (2.8), 5.4-23.1 14.1 (3.0), 5.6-23.1 13.7 (3.1), 5.4-23.1 13.9 (2.9), 5.4-23.1
State bar smoking opposition (mean %, range) 49.9 (10.2), 26.7-74.5 50.8 (10.2), 27.2-74.0 50.1 (10.3), 26.7-74.5 46.3 (9.6), 26.7-74.5 47.7 (10.4), 27.2-74.0 47.9 (10.3), 27.2-74.5
State % with smoke-free home rules (mean %, (SE), range) 82.8 (6.6), 59.1-93.8 82.7 (7.0), 59.1-96.9 82.5 (6.7), 59.1-93.8 80.5 (6.8), 59.1-93.8 80.7 (7.3), 59.1-96.9 80.8 (6.9), 59.1-93.8
State % Black (mean %, (SE), range) 12.8 (8.1), 0.6-51.9 12.3 (7.7), 0.6-48.6 13.0 (8.0), 0.6-54.4 13.8 (8.8), 0.6-51.0 12.6 (7.8), 0.6-48.6 13.2 (8.0), 0.6-54.4
State % Hispanic (mean %, (SE), range) 16.1 (12.6), 1.0-47.8 16.0 (13.0), 1.0-48.5 16.1 (13.0), 1.1-48.2 13.6 (12.2), 1.0-47.8 14.0 (12.6), 1.0-48.5 14.1 (12.4), 1.1-48.2
State % HS grad (age 25+)(mean %, (SE), range) 85.6 (3.5), 77.8-93.5 85.6 (3.6), 78.4-93.5 85.4 (3.5), 77.8-93.5 85.1 (3.5), 77.8-93.5 85.3 (3.6), 78.4-93.5 85.4 (3.5), 77.8-93.5
State % college grad (age 25+) (mean %, (SE), range) 28.3 (4.7), 16.6-56.7 28.5 (4.5), 16.6-55.0 28.3 (4.6), 17.3-56.7 26.8 (4.5), 16.6-50.1 27.5 (4.5), 16.6-55.0 27.7 (4.7), 17.3-56.7
Unweighted N 193,570 185,943 171,019 11,501 20,019 29,667
Open in a new tab

Abbreviations: Weighted (wt), standard error (SE)

In additional analyses, we examined differences in the source of cigarettes for students by grade using a question in the MTF survey regarding if respondents had bought cigarettes in the past 30 days via friends or relatives, from vending machines, through the mail, in stores where you bring cigarettes to the check-out, in stores where clerks hand you the cigarettes, or in some other way.

In sensitivity analyses we evaluated models that additionally controlled for past 30-day alcohol use, past 30-day marijuana use, and past 2-week binge drinking (i.e., 5 or more drinks in a row) together in one model. We also investigated the potential non-linearity of the associations between tax or price and the four outcome variables using interactions between each year and tax or price.

All analyses were conducted using Stata version 15.0.

RESULTS

Descriptive statistics

Table 1 presents descriptive statistics for 8th, 10th, and 12th graders in the entire population (analytic sample for 30-day smoking participation) and in the analytic sample for smoking intensity for years 2005 to 2016. Smoking participation increased across grades with 6% of 8th graders, 11% of 10th graders, and 17% of 12th graders smoking any cigarettes in the past 30 days. The average inflation-adjusted state and federal cigarette tax in our sample was approximately $1.40 and average sale price was approximately $5.70.

The conditional intensity analytic samples were restricted to those with any smoking participation in the past 30 days, and the initiation samples were restricted to those who had not initiated first cigarette or daily smoking before their current grade (Appendix Table 1). Among past 30-day smokers, average cigarettes smoked per day ranged from approximately 4 to 5. First cigarette smoking initiation ranged from 2% to 4% by grade, and daily smoking initiation ranged from 1% to 2%. Descriptive statistics for the imputed sample are shown for comparison in Appendix Table 2.

Main associations

Table 2 presents the main associations of tax and price on each of the adolescent smoking outcomes for 8th, 10th, and 12th graders after adjusting for individual characteristics, state-level controls, region, and year fixed effects. Findings for tax and price were similar, so we focus on tax below.

Table 2.

Average Marginal Effects of Cigarette Tax (State/Federal) and Average Sale Price on Adolescent Smoking by Grade, Monitoring the Future, 2005-2016. Results Shown are Using Imputed Data (m=10).

8th graders 10th graders 12th graders
aAME (95% CI) P aAME (95% CI) P aAME (95% CI) P
Smoking participation
 Tax −0.007 (−0.010, −0.003) <0.001 0.003 (−0.001, 0.008) 0.156 −0.009 (−0.015, −0.003) 0.002
 Price −0.007 (−0.010, −0.003) <0.001 0.002 (−0.002, 0.007) 0.298 −0.011 (−0.016, −0.005) <0.001
     N 193570 185943 171019
Conditional intensity
 Tax 0.09 (−0.31, 0.50) 0.649 −0.22 (−0.44, 0.00 0.053 −0.30 (−0.49, −0.11) 0.002
 Price −0.02 (−0.38, 0.34) 0.925 −0.23 (−0.43, −0.03) 0.021 −0.27 (−0.44, −0.09) 0.003
     N 11501 20019 29667
First cigarette initiation
 Tax −0.004 (−0.006, −0.002) <0.001 0.000 (−0.002, 0.003) 0.820 −0.002 (−0.006, 0.003) 0.389
 Price −0.004 (−0.006, −0.002) <0.001 0.001 (−0.002, 0.003) 0.701 −0.001 (−0.006, 0.003) 0.474
     N 163427 138073 53933
Daily smoking initiation
 Tax −0.003 (−0.004, −0.001) 0.002 0.002 (−0.006, 0.009) 0.646 −0.003 (−0.006, −0.000) 0.043
 Price −0.003 (−0.004, −0.001) 0.003 0.002 (−0.004, 0.008) 0.541 −0.002 (−0.005, 0.000) 0.104
     N 186398 171541 74231
Open in a new tab

Abbreviations: Average marginal effects (AME), confidence interval (CI)

a

Each average marginal effect is estimated from a single model with either tax or price as the independent variable. Average marginal effects are the average change in predicted probability of the outcome, associated with a one-dollar increase in tax or price. In addition to tax or price, all Poisson models included all variables shown in Table 1 for each grade as well as a year indicator and a state random effect. Conditional intensity was investigated using linear models, adjusted for the same covariates as the Poisson models; interpretation is for a change in the mean cigarettes per day. The initiation questions were only asked for half of the 12th grade forms so sample size is halved.

Among 8th graders, a $1 increase in cigarette tax was associated with a 0.7 percentage point lower probability of smoking in the past 30 days (AME = −0.007; 95% CI: −0.010, −0.003), a 0.4 percentage point lower probability of smoking a cigarette for the first time (AME = −0.004; CI: −0.006, −0.002), and a 0.3 percentage point lower probability of starting daily smoking (AME = −0.003; CI: −0.004, −0.001). Tax was not associated with the number of cigarettes smoked per day among 8th graders.

Among 10th graders, there were no statistically significant associations between cigarette tax or price and smoking participation, first cigarette initiation, or daily smoking initiation. A $1 increase in tax or price was marginally associated with smoking 0.2 fewer cigarettes per day among smokers.

In models of smoking participation for 12th graders, a $1 increase in cigarette tax was associated with a 0.9 percentage point lower probability of smoking in the past 30 days (AME = −0.009; 95% CI: −0.015, −0.003), fewer cigarettes smoked per day (change in mean = 0.30; CI: −0.49, −0.11), and lower daily smoking initiation (AME = −0.003; 95% CI: −0.006, 0.000), but was not associated with first cigarette initiation.

To further understand differences in the taxation/smoking associations, we examined differences in the source of cigarettes for students by grade. Using additional MTF questions, we found 12th graders more often purchased their own cigarettes than 10th and 8th graders (64% from a store clerk vs 24% and 16%, respectively), while 8th and 10th graders more often acquired cigarettes from other sources, such as websites, vending machines, through the mail, or from friends or relatives.

Differential associations of taxation and smoking outcomes by SES, race/ethnicity, and gender

After adjusting for multiple testing, student’s college plans significantly modified the association between taxation and 30-day smoking among 12th graders (Appendix Table 3; P <0.001). Specifically, taxation was associated with lower 30-day smoking probability among students who definitely planned to attend college, while students who did not plan to attend college had no relationship between taxation and 30-day smoking; students who probably planned to attend college fell in between (Figure 1). Although not statistically significant after adjusting for multiple testing, parental education exhibited a similar relationship to that of college bound: students who had at least one parent with a college education had a lower probability of smoking in the past 30 days than students whose highest parental education level was lower (Appendix Figure 1). We found no other interactions by SES, race/ethnicity, or gender for any of the other outcomes in any of the other grades after correcting for multiple testing.

Figure 1.

Figure 1

Open in a new tab

Differential Associations of Tax on Smoking Participation Over Prior 30 Days Among 12th Graders, by Plans to Attend College, Monitoring the Future, 2005-2016. Results Shown are Using Imputed Data (m=10).

We also report tax elasticities overall by grade, as well as stratified by parental education, college bound, race/ethnicity, and gender (Appendix Table 4). Eighth graders were tax responsive for 30-day smoking participation, first cigarette initiation, and daily smoking initiation, while 12th graders were tax responsive for smoking participation, conditional intensity, and daily smoking initiation. Students whose parents had a college education showed greater elasticities than their less educated counterparts for smoking participation (8th and 12th graders), first cigarette initiation (8th graders), and daily smoking initiation (12th graders). Twelfth graders who definitely planned to attend college had stronger tax elasticities than students with other college plans for smoking participation and daily smoking initiation.

In the sensitivity analyses additionally controlling for other substance use, we found similar results. We also investigated the potential non-linearity of the associations between taxation or price and the four outcome variables, and found no evidence of differences in the associations with time (results not shown).

DISCUSSION

Our study found that higher state cigarette taxes and average cigarette price per pack were associated lower 30-day smoking, first cigarette smoking initiation, and daily smoking initiation among 8th graders, lower intensity among 10th grader smokers, and lower 30-day smoking participation and intensity among 12th graders. Our findings are consistent with prior literature finding that taxation reduced smoking participation [5-22, 26, 53], smoking intensity [5, 7, 11, 12, 22-24, 26], and smoking initiation [22, 25] among youth. Interestingly, we found that taxation reduced initiation among 8th graders only, while affecting 30-day smoking among 8th and 12th graders, and intensity among 10th and 12th graders. From our analysis of differences in cigarette sources by grade, 12th graders may more acutely feel the effects of price since they are doing more of their own purchasing, which may help explain why taxation has a stronger association for 30-day smoking among 12th than 10th graders. Regarding stronger associations between taxation and 30-day smoking and initiation for 8th graders compared to 10th graders, it may be that 8th graders are more likely to obtain cigarettes from their parents, whereas 10th graders may be more likely to get cigarettes from their friends. The survey questions could not distinguish between these possibilities, but prior literature suggests that family influence is more important than peer influence on smoking for younger ages [54].

For most taxation/smoking outcome relationships, we saw no heterogeneity of association by SES, race/ethnicity, or gender. Only the relationship between taxation and 30-day smoking participation differed by plans to attend college for 12th graders, and in a direction that indicates an increase in disparities, rather than a decrease. This seems to contradict two studies using older data that found that students whose parents had less than a college degree were more price responsive than students whose parents had a college degree or greater [21, 27], as well as adult studies that have found greater price responsiveness by among low SES adults [28, 29]. However, our results are consistent with the national surveys that show continued disparities in youth smoking by SES. In addition, although we did not find statistically significant differences by parental education in any of the tax/smoking relationships after correction for multiple testing, the pattern for parental education for 12th graders showed a similar pattern to plans to attend college: students whose parents had attended college had lower 30-day smoking participation.

Plans to attend college is not a traditional measure of SES, and another possible interpretation is that students who plan attend college have higher preference for the future. Thus, an increase in cigarette taxes increased the cost of smoking addiction in the future and induced individuals who value the future more to not smoke or smoke less—consistent with forward-looking smoking behavior [55, 56]. Although we do not rule out a time preference interpretation, we think it is more likely that plans to attend college is a SES proxy, given that it is strongly associated with parental education, income, and occupation [36, 37]. Moreover, other studies have shown differences in time preferences do not explain much of differences in health behaviors [57] or the relationship between education and health behaviors [58].

We did not find differences in the association of taxation on smoking outcomes by race/ethnicity or gender, which is in contrast to earlier studies. Several older studies found Black and/or Hispanic youth to be more price responsive than White youth [21, 24, 25, 30], with some variation by gender [30]. However, an analysis of the 2009 federal tax increase found White adolescents were more price responsive than Black adolescents [22]. Previous results were also mixed for gender, with male adolescents more price responsive than female adolescents for smoking prevalence [30] and smoking initiation [31] in some studies, with opposite findings in others for smoking prevalence [21, 22, 24], intensity [21], and initiation [22]. Differences between these studies and our findings could again be due to using newer data, looking at a wider range of outcomes, and examining differences by grade.

Overall, the smoking participation and conditional intensity elasticities that we estimated were smaller than elasticities estimated by prior studies using MTF data from earlier years. A study using MTF data from 1976-1998 had the highest elasticity for smoking participation at −1.41 [24], while studies using MTF data from the 1990s [7, 23, 27, 30] estimated participation elasticities ranging from −0.311 [27] to −0.675 [23] and conditional intensity elasticities ranging from −0.029 [27] to −0.638 [23] when pooling across grades and not differentiating by sociodemographic factors. More recently, a study using MTF data from 1991 to 2010 estimated the participation elasticity at −0.259 and the conditional intensity elasticity at −0.187 [21]. Using data from 2005 to 2016, we found smoking participation elasticities for 8th and 12th graders were −0.28 and −0.13 (respectively), and that the conditional intensity elasticity for 12th graders was −0.12, with other results by grade not statistically significant. The elasticities we estimated may have been smaller than previous in previous papers if youth price-responsiveness has declined over time; decades after the harms of cigarettes were exposed, today’s youth may not be as price-elastic.

This research is subject to several limitations. We examined repeated cross-sectional data, so our findings should not be extrapolated to reflect longitudinal changes within individuals. The substance use data are self-reported, which may lead to reporting bias. However, school-based youth samples typically have less bias than home-based youth samples for substance use reporting [59, 60], and measures of self-reported drug use have been validated within MTF [3]. We examined taxation and smoking behavior over a time period when tobacco control policies were changing and new tobacco products were being introduced. However, we controlled for the changing tobacco control landscape by adjusting for smoke-free policies over time, as well as other state-level characteristics that may be associated with tobacco control policies and smoking behavior. We also tested for changes in the association between taxation and smoking behavior by adding in interaction terms for each year, with the concern that taxation may have become less effective as new products, such as electronic cigarettes, were introduced into the market. We found no evidence of differences in the taxation/smoking relationships over time. However, we acknowledge that the patterns we observed may change due to the increase of e-cigarette use and other products among youth over time, and additional policies implemented to address these changes. Although prior research has also controlled for state fixed effects [18, 26], we did not because MTF data are representative at the national level, but not the state level. Moreover, the number and types of communities sampled within states change over time, and not all states are sampled every year. We did control for several state-level sociodemographic variables to adjust for differences between states, but acknowledge that residual state-level confounding may remain. Local level tax rates were not considered in this analysis, and we did not have data on actual price at point of sale, which may vary within states.

In conclusion, taxation continues to be a useful tobacco control tool for reducing smoking among youth, particularly for 8th and 12th graders in our analysis. However, taxation may not be effective in improving health equity among youth. There is a need for longitudinal analyses to further elucidate these findings as youth transition into adulthood.

Supplementary Material

Supp1
Supp2
Supp3
Supp4
Supp5

What this paper adds.

Cigarette taxes are effective in reducing smoking, and there is evidence that lower socioeconomic status adults are the most price sensitive.

There is little evidence on how taxes impact adolescent smoking disparities, particularly in recent years.

Our findings suggest that cigarette taxes, while effective at reducing smoking among youth, appear unlikely to reduce smoking disparities.

Acknowledgements

Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health [grant number R37CA214787]. Data were collected with support from the National Institute on Drug Abuse of the National Institutes of Health [grant number R01DA001411]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The authors would also like to acknowledge the research assistance of Sarah Skolnick and analytic assistance of Dr. Yanmei Xie.

References

  • [1].Fryar CD, Merino MC, Hirsch R, et al. Smoking, Alcohol Use, and Illicit Drug Use Reported by Adolescents Aged 12–17 years: United States. National Health Statistics Reports 2009;15:1–26. [PubMed] [Google Scholar]
  • [2].US Department of Health and Human Services. Preventing tobacco use among youth and young adults: a report of the Surgeon General. Atlanta, GA: US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2012. [Google Scholar]
  • [3].Miech RA, Johnston LD, O’Malley PM, et al. Monitoring the Future national survey results on drug use, 1975-2017: Volume I, Secondary school students. Ann Arbor, MI: Institute for Social Research, University of Michigan; 2018. [Google Scholar]
  • [4].U.S. National Cancer Institute. A Socioecological Approach to Addressing Tobacco-Related Health Disparities. National Cancer Institute Tobacco Control Monograph 22 NIH Publication No 17-CA-8035A. Bethesda, MD: U.S. Department of Health and Human Services, National Insitutes of Health, National Cancer Institute; 2017. [Google Scholar]
  • [5].Lewit EM, Coate D, Grossman M. The Effects of Government Regulation on Teenage Smoking. The Journal of Law & Economics 1981;24(3):545–569. [Google Scholar]
  • [6].Lewit EM, Hyland A, Kerrebrock N, et al. Price, public policy, and smoking in young people. Tobacco control 1997;6 Suppl 2:S17–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [7].Chaloupka FJ, Pacula RL. Limiting youth access to tobacco: the early impact of the Synar Amendment on youth smoking. 3rd Pacific Rim Allied Economic Organizations Conference 1998. [Google Scholar]
  • [8].Biener L, Aseltine RH, Jr., Cohen B, et al. Reactions of adult and teenaged smokers to the Massachusetts tobacco tax. American journal of public health 1998;88(9):1389–1391. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [9].Harris JE, Chan SW. The continuum-of-addiction: cigarette smoking in relation to price among Americans aged 15-29. Health Econ 1999;8(1):81–86. [DOI] [PubMed] [Google Scholar]
  • [10].Emery S, White MM, Pierce JP. Does cigarette price influence adolescent experimentation? Journal of health economics 2001;20(2):261–270. [DOI] [PubMed] [Google Scholar]
  • [11].Czart C, Pacula RL, Chaloupka R, et al. The impact of prices and control policies on cigarette smoking among college students. Contemporary Economic Policy 2001;19(2):135–149. [Google Scholar]
  • [12].Ross H, Chaloupka FJ. The effect of cigarette prices on youth smoking. Health Econ 2003;12(3):217–230. [DOI] [PubMed] [Google Scholar]
  • [13].Thomson CC, Fisher LB, Winickoff JP, et al. State tobacco excise taxes and adolescent smoking behaviors in the United States. Journal of public health management and practice : JPHMP 2004;10(6):490–496. [DOI] [PubMed] [Google Scholar]
  • [14].Tauras JA. Public policy and smoking cessation among young adults in the United States. Health policy (Amsterdam, Netherlands) 2004;68(3):321–332. [DOI] [PubMed] [Google Scholar]
  • [15].Powell LM, Tauras JA, Ross H. The importance of peer effects, cigarette prices and tobacco control policies for youth smoking behavior. Journal of health economics 2005;24(5):950–968. [DOI] [PubMed] [Google Scholar]
  • [16].Slater SJ, Chaloupka FJ, Wakefield M, et al. The impact of retail cigarette marketing practices on youth smoking uptake. Archives of Pediatrics & Adolescent Medicine 2007;161(5):440–445. [DOI] [PubMed] [Google Scholar]
  • [17].DeCicca P, Kenkel D, Mathios A. Cigarette taxes and the transition from youth to adult smoking: smoking initiation, cessation, and participation. Journal of health economics 2008;27(4):904–917. [DOI] [PubMed] [Google Scholar]
  • [18].Carpenter C, Cook PJ. Cigarette taxes and youth smoking: new evidence from national, state, and local Youth Risk Behavior Surveys. Journal of health economics 2008;27(2):287–299. [DOI] [PubMed] [Google Scholar]
  • [19].Tworek C, Yamaguchi R, Kloska DD, et al. State-level tobacco control policies and youth smoking cessation measures. Health policy (Amsterdam, Netherlands) 2010;97(2-3):136–144. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [20].Choi TC, Toomey TL, Chen V, et al. Awareness and reported consequences of a cigarette tax increase among older adolescents and young adults. American journal of health promotion : AJHP 2011;25(6):379–386. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [21].Tauras JA, Huang J, Chaloupka FJ. Differential impact of tobacco control policies on youth sub-populations. International journal of environmental research and public health 2013;10(9):4306–4322. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [22].van Hasselt M, Kruger J, Han B, et al. The relation between tobacco taxes and youth and young adult smoking: What happened following the 2009 US federal tax increase on cigarettes? Addictive behaviors 2015;45:104–109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [23].Chaloupka FJ, Grossman M. Price, tobacco control policies and youth smoking. National Bureau of Economic Research 1996. [Google Scholar]
  • [24].Ding A Youth are more sensitive to price changes in cigarettes than adults. The Yale journal of biology and medicine 2003;76(3):115. [PMC free article] [PubMed] [Google Scholar]
  • [25].Nonnemaker JM, Farrelly MC. Smoking initiation among youth: The role of cigarette excise taxes and prices by race/ethnicity and gender. Journal of health economics 2011;30(3):560–567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [26].Hansen B, Sabia JJ, Rees DI. Have Cigarette Taxes Lost Their Bite? New Estimates of the Relationship between Cigarette Taxes and Youth Smoking. Am J Health Econ 2017;3(1):60–75. [Google Scholar]
  • [27].Gruber J Youth smoking in the US: prices and policies. National Bureau of Economic Research 2000. [Google Scholar]
  • [28].Brown T, Platt S, Amos A. Equity impact of population-level interventions and policies to reduce smoking in adults: a systematic review. Drug Alcohol Depend 2014;138:7–16. [DOI] [PubMed] [Google Scholar]
  • [29].Hill S, Amos A, Clifford D, et al. Impact of tobacco control interventions on socioeconomic inequalities in smoking: review of the evidence. Tobacco control 2014;23(e2):e89–97. [DOI] [PubMed] [Google Scholar]
  • [30].Chaloupka FJ, Pacula RL. Sex and race differences in young people’s responsiveness to price and tobacco control policies. Tobacco control 1999;8(4):373–377. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [31].Cawley J, Markowitz S, Tauras J. Lighting up and slimming down: the effects of body weight and cigarette prices on adolescent smoking initiation. Journal of health economics 2004;23(2):293–311. [DOI] [PubMed] [Google Scholar]
  • [32].Bachman JG, Johnston LD, O’Malley PM, et al. The Monitoring the Future project after four decades: Design and procedures (Monitoring the Future Occasional Paper No. 82). Ann Arbor, MI: Institute for Social Research, University of Michigan; 2015. [Google Scholar]
  • [33].Orzechowski Walker. The Tax Burden on Tobacco Volume 51, 1970-2016. Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 2018. [Google Scholar]
  • [34].U.S. Bureau of Economic Analysis. Gross Domestic Product: Implicit Price Deflator [GDPDEF]. FRED, Federal Reserve Bank of St. Louis; 2016. [Google Scholar]
  • [35].Fuchs VR. Time Preference and Health: An Exploratory Study. In: Fuchs VR, ed. Economic Aspects of Health. Chicago: University of Chicago Press; 1982:93–120. [Google Scholar]
  • [36].Cabrera AF, La Nasa SM. Understanding the College-Choice Process. New Directions for Institutional Research 2000;107:5–22. [Google Scholar]
  • [37].King JE. The decision to go to college: attitudes and experiences associated with college attendance among low-income students. Washington, DC: College Board; 1996:16. [Google Scholar]
  • [38].University of Kentucky Center for Poverty Research. UKCPR National Welfare Data, 1980-2016. In: Gatton College of Business & Economics UK, ed. Lexington, KY. 2017. [Google Scholar]
  • [39].Surveillance Epidemiology and End Results (SEER) Program Populations. U.S. Population Data 1969-2016 downloaded from Nationa Bureau of Economic Research (NBER). In: National Cancer Institute, DCCPS, Surveillance Research Program, eds. (https://seer.cancer.gov/popdata/download.html) 2017. [Google Scholar]
  • [40].U.S. Census Bureau. American Community Survey, 2005-2016 American Community Survey 1-Year Estimates. American FactFinder 2020. [Google Scholar]
  • [41].Tobacco Control Laws Database. In: American Nonsmokers’ Rights Foundation, ed. 2018. [Google Scholar]
  • [42].DeCicca P, Kenkel D, Mathios A, et al. Youth smoking, cigarette prices, and anti-smoking sentiment. Health economics 2008;17(6):733–749. [DOI] [PubMed] [Google Scholar]
  • [43].Pesko MF, Tauras JA, Huang J, et al. The influence of geography and measurement in estimating cigarette price responsiveness. National Bureau of Economic Research 2016. [Google Scholar]
  • [44].Bader P, Boisclair D, Ferrence R. Effects of tobacco taxation and pricing on smoking behavior in high risk populations: a knowledge synthesis. International journal of environmental research and public health 2011;8(11):4118–4139. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [45].Hawkins SS, Bach N, Baum CF. Impact of Tobacco Control Policies on Adolescent Smoking. J Adolesc Health 2016;58(6):679–685. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [46].Cummings P Methods for estimating adjusted risk ratios. Stata J 2009;9(2):175–196. [Google Scholar]
  • [47].Mustillo S, Landerman LR, Land KC. Modeling longitudinal count data: Testing for group differences in growth trajectories using average marginal effects. Sociological Methods & Research 2012;41(3):467–487. [Google Scholar]
  • [48].Bauer GR. Incorporating intersectionality theory into population health research methodology: Challenges and the potential to advance health equity. Social science & medicine 2014;110:10–17. [DOI] [PubMed] [Google Scholar]
  • [49].Dow WH, Norton EC, Donahoe JT. Multiplicative and marginal effects in nonlinear models. The Stata Journal forthcoming. [Google Scholar]
  • [50].Benjamini Y, Hochberg Y. Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. J R Stat Soc B 1995;57(1):289–300. [Google Scholar]
  • [51].Raghunathan TE, Solenberger PW, Van Hoewyk J. IVEware: imputation and variance estimation software. Ann Arbor, MI: Survey Methodology Program, Survey Research Center, Institute for Social Research, University of Michigan; 2002. [Google Scholar]
  • [52].Raghunathan TE, Lepkowski JM, Van Hoewyk J, et al. A multivariate technique for multiply imputing missing values using a sequence of regression models. Survey Methodology 2001;27(1):85–95. [Google Scholar]
  • [53].Chaloupka FJ. Price, Tobacco Control Policies and Youth Smoking.
  • [54].Liao Y, Huang Z, Huh J, et al. Changes in friends' and parental influences on cigarette smoking from early through late adolescence. J Adolesc Health 2013;53(1):132–138. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [55].Becker GS, Murphy KM. A Theory of Rational Addiction. J Polit Econ 1988;96(4):675–700. [Google Scholar]
  • [56].Gruber J, Koszegi B. Is addiction "rational"? Theory and evidence. Q J Econ 2001;116(4):1261–1303. [Google Scholar]
  • [57].Cutler DM, Glaeser E. What explains differences in smoking, drinking, and other health-related behaviors? Am Econ Rev 2005;95(2):238–242. [DOI] [PubMed] [Google Scholar]
  • [58].Cutler DM, Lleras-Muney A. Understanding differences in health behaviors by education. Journal of health economics 2010;29(1):1–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [59].Gfroerer J, Wright D, Kopstein A. Prevalence of youth substance use: the impact of methodological differences between two national surveys. Drug Alcohol Depend 1997;47(1):19–30. [DOI] [PubMed] [Google Scholar]
  • [60].Kann L, Brener ND, Warren CW, et al. An assessment of the effect of data collection setting on the prevalence of health risk behaviors among adolescents. J Adolesc Health 2002;31(4):327–335. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp1
NIHMS1627309-supplement-Supp1.pdf (60.8KB, pdf)
Supp2
NIHMS1627309-supplement-Supp2.pdf (43.1KB, pdf)
Supp3
NIHMS1627309-supplement-Supp3.pdf (40.5KB, pdf)
Supp4
NIHMS1627309-supplement-Supp4.pdf (79.4KB, pdf)
Supp5
NIHMS1627309-supplement-Supp5.pdf (36.1KB, pdf)

RESOURCES