Objectives. We evaluated the combined effects on California cigarette consumption of an additional 50¢ per pack state tax imposed by Proposition 10 of January 1999 and a 45¢ per pack increase in cigarette prices stemming from the Master Settlement Agreement (MSA) of November 1998.
Methods. We used quarterly cigarette sales data for the period 1984–2002 to estimate a time-series intervention model adjusting for seasonal variations and time trend.
Results. Over the period 1999 through 2002, the combined effect was to reduce cigarette consumption by 2.4 packs per capita per quarter (1.3 billion packs total over the 4-year period) and to raise state tax revenues by $2.1 billion. These effects were similar to the effects of a 25¢ per pack tax increase enacted by Proposition 99 a decade earlier, although with decreased relative effectiveness as measured by percentage of reduction in cigarette consumption divided by percentage of increase in taxation (−0.44 vs −0.60).
Conclusions. A major increase in price through taxation and the MSA provided a strong economic disincentive for smokers in a state with a low smoking prevalence. This effect could be reinforced if part of the MSA payments were devoted to tobacco control programs.
In November 1998, California voters passed Proposition 10, the California Children and Families First Act, increasing the state excise tax on cigarettes from 37¢ to 87¢ per pack beginning on January 1, 1999. All revenues raised by the additional tax were earmarked for early childhood development programs. This major increase in the state tax on cigarettes was not the first in California history. In November 1988, California passed Proposition 99, the Tobacco Tax and Health Promotion Act, an unprecedented antismoking ballot initiative that increased the cigarette tax by 25¢ per pack over the rate of 10¢ per pack that had been maintained since 1967. Proposition 99 became effective on January 1, 1989, and 20% of the new tax revenues was earmarked to fund a statewide tobacco control program that was launched in April 1990. In June 1993, the California Breast Cancer Act was passed, raising the cigarette tax by another 2¢ per pack, to 37¢, effective January 1, 1994. This additional revenue was intended to fund breast cancer research and early breast cancer detection services for uninsured and underinsured women.
Several studies have examined the impact on California cigarette consumption of the 25¢ per pack tax increase following Proposition 99. Structural econometric analyses showed that following the implementation of Proposition 99, cigarette consumption was reduced by 8% to 10% in the short run and by 10% to 13% in the long run.1–3 Time-series analysis indicated a 9% to 11% decline in cigarette sales after Proposition 99.4 In addition, the antismoking media campaign funded by the earmarked excise tax revenues further reduced cigarette consumption.5,6 A decade after the implementation of Proposition 99, the annual per capita cigarette consumption in California in 1998 had declined to 52.3 packs, which was 42% lower than its 1988 level of 90.1 packs and 39% lower than the 1998 national average of 85.8 packs7; the adult smoking prevalence for California was 19.2% in 1998, among the 3 lowest US states.8
Coinciding with the passage of Proposition 10, state attorneys general and the tobacco industry signed the landmark Master Settlement Agreement (MSA) on November 23, 1998. The MSA requires 4 major tobacco companies to pay 46 states and the District of Columbia a total of $206 billion over a 25-year period to reimburse the excess costs of treating smoking-related illnesses under Medicaid programs (the other 4 states had settled previously in June 1997). On the same day, the tobacco companies raised the wholesale price of cigarettes by 45¢ per pack, the largest increase in history, to finance their settlement costs. This price increase was expected to reduce cigarette consumption.
Given the significantly lower level of cigarette consumption in California as of 1998, it is important for public health policymakers to investigate whether the additional 50¢ per pack tax increase imposed by Proposition 10 was an effective tobacco control strategy to further reduce cigarette consumption. In this study we intended to address this policy question. However, given that the implementation of Proposition 10 and the MSA happened almost at the same time, we sought to evaluate the combined effects of Proposition 10 and the MSA on California cigarette consumption by using quarterly cigarette consumption data in California. We then compared their effects to the effect of the tax increase raised by Proposition 99.
This study covered the period from January 1984 through December 2002. We obtained monthly cigarette consumption data from the California State Board of Equalization. These data are based on monthly sales of state cigarette tax stamps that cigarette wholesalers, retailers, and manufacturers purchase and affix to each pack of cigarettes before distribution. We summed up monthly cigarette sales (number of packs) by quarter to derive quarterly sales. We then calculated per capita cigarette consumption by dividing quarterly cigarette sales by the state’s civilian population. Military personnel were excluded from the calculation of per capita cigarette consumption because cigarettes sold at federal military installations are exempted from state taxes. The annual California civilian population is based on estimates as of July 1 of each year, obtained from the California Department of Finance.9 We interpolated the quarterly population from the annual numbers assuming a constant exponential growth rate.
In order to estimate the combined effects of Proposition 10 and the MSA on cigarette consumption, we used a time-series intervention analysis developed by Box and Tiao10 and Box and Jenkins.11 In the model, per capita cigarette consumption was specified as a function of seasonal variations, random fluctuations, autocorrelation between quarters, overall declining trend of smoking habit, and exogenous variables measuring the impacts of taxation interventions. The combined effects of Proposition 10 and the MSA were measured by a dummy variable (value of 1 since 1999; 0 otherwise). Given that Proposition 10 was passed almost 2 months before it became effective, some smokers probably stockpiled additional cigarettes during those 2 months in anticipation of the price increase in January 1999. Because cigarettes are perishable goods, smokers would consume their stockpile of cigarettes and purchase fewer cigarettes immediately after the tax increase. To control for this temporary hoarding effect, a pre-tax dummy variable (value of 1 for the fourth quarter of 1998; 0 otherwise) and a post-tax dummy variable (value of 1 for the first quarter of 1999; 0 otherwise) were included in the model. To measure the effects of 2 other California cigarette tax increases resulting from Proposition 99 and the Breast Cancer Act, we included similar dummy variables for state tax increases and hoarding effects.
Note that during the study period 1984–2002, the federal per-pack tax on cigarettes increased 4 times, from 16¢ to 20¢ in January 1991, from 20¢ to 24¢ in January 1993, from 24¢ to 34¢ in January 2000, and from 34¢ to 39¢ in January 2002. We included 4 additional dummy variables to control for the effects of these federal tax increases.
Overall, cigarette consumption showed a declining trend attributable to a variety of factors such as antismoking campaigns, decreased social acceptability, and more concern with personal health. As reported by Pierce et al.,12 the rate of decline slowed after 1993 in California. To capture the decelerated downward trend over the 76 quarters encompassed in the study period, we included a time trend variable expressed by the square root of the time value. In addition, 3 quarterly dummy variables (the fourth quarter being the comparison quarter) were included to account for the seasonal fluctuations in cigarette sales.
The model was estimated by using the PROC ARIMA procedure of the SAS/ETS software (SAS Institute Inc, Cary, NC). As a measure for the goodness of fit, we used the test developed by Ljung and Box13 to determine if the residuals from the estimated time-series model were white noise. We considered statistical significance as 2-tailed P<.05.
The estimated results of the time-series intervention analysis are presented in Table 1. The statistically significant coefficient for the square root of the time trend is −1.36, indicating that there was a continuous decline in per capita cigarette consumption, although at a reduced rate (−0.25 packs per quarter in 1984–1988, −0.13 packs per quarter in 1989–1992, −0.08 packs per quarter in 1999–2002). After controlling for other factors, we found that the combined effects of the 50¢ per pack tax increase resulting from Proposition 10 and the 45¢ per pack price increase stemming from the MSA were very significant: per capita sales dropped by 2.4 packs per quarter. In contrast, the 25¢ per pack tax increase resulting from Proposition 99 reduced per capita sales by 2.7 packs per quarter. The Breast Cancer Act did not show a statistically significant impact on cigarette sales. The coefficients for the 6 pre-tax and post-tax dummy variables had correct signs as expected, and all were statistically significant except during the fourth quarter of 1988. Interestingly, although the 2¢ per pack tax increase attributable to the Breast Cancer Act did not affect cigarette sales, it was associated with the strongest hoarding effect such that per capita sales increased sharply by 3.8 packs in the fourth quarter of 1993 and decreased by 2.7 packs in the first quarter of 1994.
Figure 1 shows the actual values (dots) and predicted values (solid line) of per capita cigarette sales based on the estimated coefficients in Table 1. The extent to which the time-series intervention analysis fit the observed data is demonstrated by the closeness between these 2 sets of values. Figure 1 also shows the simulated values of cigarette sales by assuming that there were no Proposition 10 or MSA to cause a change in cigarette sales, while holding all other factors unchanged. Two other scenarios of simulation are also plotted by assuming that (1) Proposition 10, the MSA, and Proposition 99 all were nonexistent and (2) only Proposition 99 were nonexistent.
When the predicted and simulated values of per capita sales were multiplied by the civilian population, we obtained predicted and simulated total cigarette sales, as shown in Table 2. Subtracting the predicted sales from simulated sales, we estimated that the combined effects of Proposition 10 and the MSA reduced total cigarette sales by 1.314 billion packs during the first 4 years after implementation of Proposition 10. Despite this reduction, California’s excise tax revenues increased by $2.1 billion. In comparison, Proposition 99 had the effect of reducing the total cigarette sales by 1.293 billion packs and raising state excise tax revenues by $2.0 billion during the first 4 years after implementation.
To compare the relative effectiveness of Proposition 10 combined with the MSA versus that of Proposition 99, we calculated “elasticity”—the percentage of reduction in cigarette sales associated with a 1% tax increase relative to price—as shown in Table 3. Based on the retail price data reported by Orzechowski and Walker,7 the 95¢ per pack increase in tax and wholesale price resulting from Proposition 10 and the MSA reflects a 42.2% increase over the 1998 price, whereas the 25¢ per pack tax increase resulting from Proposition 99 reflects a 19.8% increase over the 1988 price. The reduction of 2.4 packs of cigarettes per capita after Proposition 10 reflects an 18.8% reduction from the 1998 level of cigarette sales, and the reduction of 2.7 packs of cigarettes per capita after Proposition 99 reflects an 11.9% reduction from the 1988 cigarette sales. The elasticity for the relative effects on cigarette sales of Proposition 10 and the MSA in combination is −0.44 (−0.188/0.422); for Proposition 99 the elasticity is −0.60 (−0.119/0.198).
Table 1 also shows that the 4¢ per pack increases in federal tax in 1991 and again in 1993 reduced per capita sales in California significantly, by 1.1 packs per quarter and 1.0 packs per quarter, respectively. The elasticities for the relative effects of the 4¢ per pack federal tax increases in 1991 and 1993 on cigarette sales are −2.80 and −3.14, respectively, suggesting a stronger effect than the state tax increases (Table 3). However, the federal tax increases of 10¢ per pack in 2000 and 5¢ per pack in 2002 did not show any noticeable impact on cigarette sales in California, although these tax increases, in terms of the real magnitude (Table 3, Column 3) and the percentage increase to cigarette price (Table 3, Column 4), were not very different from the federal tax increases in 1991 and 1993.
This time-series intervention analysis indicates that the combined effects of Proposition 10 and the MSA were very significant: during the 4 years after 1999, cigarette consumption was reduced by 2.4 packs per capita per quarter—or by 1.3 billion packs in total—and state excise tax revenues were raised by $2.1 billion. Though the separate effects of Proposition 10 and the MSA could not be distinguished in this study, our findings suggest that a major increase in price, through taxation and the tobacco settlement, still provided a strong economic disincentive for smokers in a state with the lowest US smoking prevalence rates.
In terms of the absolute magnitude of reduction in cigarette consumption, Proposition 10 combined with the MSA achieved a very similar effect as the previous major state tax increase enacted by Proposition 99 a decade earlier. However, considering the extent of taxation (plus the MSA-induced price increase), there appear to be differences in their relative effectiveness. Based on an “elasticity” measure defined as the percentage of reduction in cigarette sales associated with a 1% tax increase relative to price, we calculated an elasticity of −0.44 for Proposition 10 and the MSA versus −0.60 for Proposition 99. These 2 point estimates suggest that the combination of Proposition 10 and the MSA appears to be less effective compared with Proposition 99, although whether or not the difference between these point estimates is statistically significant depends on their confidence intervals. Nonetheless, the increasingly aggressive advertising and promotion by the tobacco industry after the tobacco settlement, along with the decreased funding for the California Tobacco Control Program in recent years, may have counteracted the effects of Proposition 10 and the MSA.
One of the important elements in the MSA is the restriction on tobacco marketing aimed at youth. This restriction includes banning billboard and transit advertising, prohibiting cartoon characters in advertising and “branded” tobacco merchandise, and restricting tobacco company sponsorships of sports and entertainment events. However, the MSA failed to limit the tobacco industry’s ability to promote their products. Since the settlement, the tobacco companies have shifted their resources from media advertising (e.g., newspaper, magazine, outdoor, transit) to other forms of marketing, particularly promotions targeting purchases at retail outlets such as “buy one, get one free” deals.
In 2001, the tobacco industry’s total advertising and promotion expenditures skyrocketed to a record of $11.2 billion, a 76% increase over 1998; media expenditures declined to $1.0 billion, and “retail value added” was the single largest marketing category, amounting to $4.8 billion, or 42.5% of total expenditures.14 Focusing on media advertising alone, Keeler et al.15 found that the tobacco industry’s advertising from 1996 to 2000 partially offset the effects of the MSA-induced higher price on cigarette consumption by 33% to 57%. Were all the forms of advertising and promotion strategies used by the tobacco industry after the settlement taken into account, the counteraction of the impact of Proposition 10 and the MSA would be even greater.
Although none of the Proposition 10 tax revenues were earmarked for tobacco control purposes, 20% of the Proposition 99 tax revenues was earmarked to fund the California Tobacco Control Program. This program combines a statewide antismoking media campaign, school-based tobacco prevention education, community-based intervention activities, and smoking cessation programs. Controlling for tax effects and other confounding factors, several multivariate analyses showed that increases in expenditures for tobacco control programs reduced cigarette consumption in California5,6 and in all states.16 Using a trend analysis, Pierce et al.12 found that the rate of decline in cigarette consumption after introduction of the California Tobacco Control Program was greater during the early years (1989–1993) than the later years (1994–1996).
The slowing in the rate of decline coincided with reductions in funding for the California Tobacco Control Program. Indeed, the real total budget for the California Tobacco Control Program (including the media campaign, local health programs, local school programs, and the research program) expressed in 2002 dollars has been constantly decreasing except during 1996–1997, from a high of $250.6 million in 1989 to $143.8 million in 2000. This latter amount was only 12% of what the tobacco industry was spending on marketing tobacco products in California; in comparison, during 1989–1992 the California Tobacco Control Program spent about 25% of what the tobacco industry was spending.17
Our results also suggest that the effects of the federal tax increases in 1991 and 1993 seemed to have a stronger effect than the state tax increases, and that the 2 federal tax increases in 2000 and 2002 did not appear to have a significant effect on reducing cigarette consumption in California. However, several caveats need to be addressed. A few previous studies have compared the relative effects of federal and state tax increases, and their findings are controversial. Pooling state-specific sales data from 1955 to 1994, Meier and Licari18 concluded that federal tax increases were more effective than state tax increases. Showalter19,20 argued that federal and state taxation per se did not differ in their effect on reducing cigarette consumption, but that tobacco companies’ forward-looking and oligopolistic pricing behavior in response to a tax increase was a more important determinant. On the other hand, when the price of cigarettes reaches a higher level, smokers might become less sensitive to frequent small tax increases compared to less frequent but substantial tax increases.18
Time-series analysis is an efficient method for achieving the goal of this case study, which was to evaluate the effects of 2 interventions: the California tax increases of 1989 and 1999. Although our analysis did not explicitly include detailed structural analyses of all other effects (including consumer behavior, governmental tobacco control programs, and industry responses), it nevertheless implicitly incorporates them. It does so because its specification included both a time trend and allowance for intertemporal correlations in the random error terms. Using separate dummy variables to measure the impact of each taxation intervention allowed us to assess their aggregate effects separately. However, to understand why each intervention works or how one is compared with another, further research is needed to examine the interactive effects of the tobacco companies’ advertising and pricing behavior, state taxation, federal taxation, antismoking media campaigns, and other tobacco control programs on the retail price and consumption of cigarettes.
Also, our aggregate analysis cannot determine the effects of these tax and price increases on individual smoking behavior. For example, based on cross-sectional survey data, a recent study found that the increase in cigarette price during 1996–1999 did not affect smoking participation but was very effective in reducing the number of cigarettes smoked by existing smokers.21 Further research is needed to examine the impact of the combination of Proposition 10 and the MSA on smoking initiation, quitting, and intensity among different demographic populations using individual-level data over a longer period beyond 1999.
We assumed that the sales of cigarette tax stamps were equivalent to cigarette consumption. Our dependent variable does not account for cigarettes that were sold at tax-exempted places or sources such as Indian reservations or the Internet, nor any cigarettes bootlegged from neighboring states. Emery et al.22 estimated that more than 7 months after the implementation of Proposition 10, only 5.1% of California smokers (6.3% of total cigarette consumption in California) avoided the state excise tax by purchasing cigarettes from non- or lower-taxed sources. Therefore, our estimates for the impact on cigarette consumption might be slightly overestimated.
Also, one might argue that not including the statewide antismoking measure in the model may have led to a biased estimate for the taxation effect. Lack of access to the actual quarterly antismoking expenditure data precluded us from examining its influence in this analysis. Using time-series analyses to examine the relative effects of taxation and antismoking media campaigns, Hu et al.5 found that the coefficients for the state taxation were robust regardless of whether the media campaign variable was excluded (−0.136) or included (−0.137). With the media campaign variable included in the model, they estimated that the 25¢ per pack state tax increase in 1989 reduced cigarette sales by 1.33 billion packs during 1989–1992,6 which is consistent with our estimate of 1.29 billion packs in this study (Table 2).
It should be noted that an unintended byproduct of the MSA was a reduction in cigarette consumption as a result of the 45¢ per pack price increase raised by tobacco companies to recover their settlement costs. This may be the most important anti-tobacco benefit from the MSA by far.23 The MSA has an unrealized potential to further reduce cigarette consumption and improve health if the state governments allocate more settlement money to fund tobacco control programs.24 In 2001, only a very small proportion of the settlement money was devoted to tobacco control programs–6% for the average state, and zero for California.25,26
In the face of the tobacco industry’s aggressive marketing, multiple tobacco control strategies are needed to meet the Healthy People 2010 adult smoking prevalence goal of 12%.27 This study demonstrates that a major state tax increase combined with a major increase in wholesale price owing to the MSA stimulated a dramatic reduction in cigarette consumption. This effect could be even greater if part of the MSA payments were used to enrich tobacco control programs.
Note. MA(4) = fourth-order moving average. Chi-square statistic from the Ljung–Box test13 up to lag 24 = 28.05 (P = .21); n = 76. *P<.05. Results are 2-tailed. aIncludes the 50¢ per pack tax increase due to Proposition 10 and the 45¢ per pack increase in the wholesale price due to the MSA. b The predicted sales are calculated by assuming that hoarding effects did not exist. Note. NS = not significant. aAverage quarterly per capita sales is calculated for the year before the fourth quarter of the last year to avoid counting hoarding effects. bIn constant January 1999 dollars based on Consumer Price Index for urban consumers for all goods.
Coefficient t Intercept 28.90 83.81* Square root of time trend −1.36 −13.87* Proposition 99 −2.71 −8.14* Proposition 10 and MSA −2.40 −9.42* Breast Cancer Act tax −0.47 −1.84 Federal tax increase 1991 −1.14 −3.53* Federal tax increase 1993 −1.01 −3.51* Federal tax increase 2000 −0.19 −0.64 Federal tax increase 2002 −0.02 −0.06 Quarter 4, 1988 0.58 1.80 Quarter 1, 1989 −1.76 −5.99* Quarter 4, 1993 3.80 11.13* Quarter 1, 1994 −2.68 −7.96* Quarter 4, 1998 2.57 7.65* Quarter 1, 1999 −1.79 −4.92* Quarter 1 dummy −1.58 −5.66* Quarter 2 dummy −0.16 −0.58 Quarter 3 dummy −0.02 −0.07 MA(4) −0.93 −11.73* Proposition 10 and MSA (1999–2002) Proposition 99 (1989–1992) Increase in tax (per pack), ¢ 95a 25 Reduced per capita sales (per quarter), packs 2.40 2.71 Total state cigarette sales, packs Predicted salesb 5.129 billion 8.545 billion Simulated sales 6.443 billion 9.838 billion Reduction in cigarette sales 1.314 billion 1.293 billion Increase in state excise tax revenues, $ 2.078 billion 2.007 billion Year and Intervention (Amount of Tax Increase) Cigarette Sales in the Last Year, Packsa (1) Realb Cigarette Price in the Last Year, $ (2) Realb Tax Increase, $ (3) Tax Increase Relative to Cigarette Price, % (4) = (3)/(2) Cigarette Reduction Relative to Cigarette Sales, % (5)c Relative Effect of Intervention (6) = (5)/(4) 1989, Proposition 99 (25¢) 22.8 1.715 0.339 19.77 −11.89 −0.60 1991, Federal tax (4¢) 18.9 2.280 0.049 2.15 −6.03 −2.80 1993, Federal tax (4¢) 16.5 2.363 0.046 1.95 −6.12 −3.14 1994, Breast Cancer Act tax (2¢) 15.3 2.139 0.022 1.03 NS NS 1999, Proposition 10 and MSA (95¢) 12.8 2.249 0.950 42.24 −18.75 −0.44 2000, Federal tax (10¢) 10.9 3.418 0.097 2.84 NS NS 2002, Federal tax (5¢) 9.3 3.690 0.046 1.25 NS NS
Note. MA(4) = fourth-order moving average. Chi-square statistic from the Ljung–Box test13 up to lag 24 = 28.05 (P = .21); n = 76.
*P<.05. Results are 2-tailed.
aIncludes the 50¢ per pack tax increase due to Proposition 10 and the 45¢ per pack increase in the wholesale price due to the MSA.
b The predicted sales are calculated by assuming that hoarding effects did not exist.
Note. NS = not significant.
aAverage quarterly per capita sales is calculated for the year before the fourth quarter of the last year to avoid counting hoarding effects.
bIn constant January 1999 dollars based on Consumer Price Index for urban consumers for all goods.
This study was supported in part by grant 7ST-0201 from the University of California Tobacco-Related Disease Research Program.
Human Participant Protection No protocol approval was needed for the study.