Table 6.8. Estimates of the price elasticity of cigarette demand fpr adu!.@ from indjvjduzd-1-ye!d?Aa Study Estimated price elasticities -0.42 Comments Lewit and Coate 1982 Mullahy 1985 -0.47 Chaloupka 1990 -0.60 (men) not statistically different from zero (women) Chaloupka 1991 and 1992 -0.27 to -0.48 Wasserman et al. 1991 0.069 (1970) -0.23 (1988) Hu et al. 1995a -0.46 Ohsfeldt et al. 1997 -0.03 (tax elasticity, males) Centers for Disease -0.25 (full samplei Control and -0.11 (Ivhites) Prevention 1998 -0.32 (blacks) -1.89 (Hispanics) -0.29 (at or beloit, median income) -0.17 (above median income) -0.26 (men) -0.19 (Momen) Evans and Ringe11999 -0.25 lo -0.56 Ohsfeldt et al. 1999 -0.15 (tax elasticity, males) 1976 National Health Interview Survey; ordinary least squares methods; elasticities by age and sex. 1979 National Health Interview Survey; instrumental variables and probit methods; detailed modeling of addiction; elasticities by sex. Second National Health and Nutrition Examination Survey, 1976-1980; instrumental variables methods; detailed modeling of addiction; elasticities by sex. Second National Health and Nutrition Examination Survey, 1976-1980; instrumental variables methods; detailed modeling of addiction; elasticities by age and educational attainment. 1970,1974,1976,1979,1980,1983, and 1985 National Health Interview Surveys; generalized least squares and two-part methods; allow changes in elasticity over time. California Behavioural Risk Factor Surveys, 1985-1991; tlvo-part methods; controls for interdependence of other behavioral risk factors and smoking. 1985 Current Population Survey, males aged 16 years and older; treats taxes and control policies as endogenous; elasticity estimates for prevalence only. 1976-1980, 1982, 1985, 1987-1992 National Health Intervie Surveys; tlvo-part methods. Natality Detail data, 1989-1992, pregnant women; two-part models. 1992/93 Current Population Survey, males aged 16 years and older; treats taxes and control policies as endogenous; elasticity estimates for prevalence only. the full sample but also for subsamples based on age (20-25 years, 26-35 years, and 36-71 years) and sex. Price had a greater impact on Jvhether a respondent smoked at all than on boll- many cigarettes a respon- dent smoked. The estimated elasticitv of demand for smoking prevalence was -0.26 for the full sample, and the total price elasticity of demand \vas -0.42. The ef- fects of price were larger for younger persons: the to- tal estimated price elasticity for persons 20-25 years old \vas approximately double that for persons 26-74 years old. The studv also found that men, particu- larly those aged 20-35 vears, were quite responsive to changes in cigarette prices, lvhereas \vomen \vere al- most unaffected by price. These findings regarding age are substantiated as well by Lewit and colleagues (1981), ~~110 used data from Cycle III of the Health Examination Sur\.ev (1966 1970) to examine the impact that prices and the anti- smoking advertisements broadcast under the Fairness Doctrine had on cigarette smoking among 6,768 ado- lescents (12-17 years old). Using the same basic meth- ods emploved in the study by Le\vit and Coate (1982), this analysis estimated that the impact of price on ado- lescent smoking (measured at a total price elasticitv of -1.44) was about three times that for adult smoking (Lewit and Coate 1982). The study by Lewit and col- leagues (1981) also confirmed that price had a greater impact on the decision to smoke (elasticitv of -1.20) than on the average quantity of cigarettes eon- sumed by smokers (elasticitv of -0.25). These findings lvere generally supported by another analysis of data from the 1974, 1976, 1977, and 1979 National House- hold Surveys on Drug Abuse (Grossman et al. 1983). Mullahy (1985) was the first to estimate cigarette demand on the basis of a theoretical and empirical model treating cigarette smoking as an addictive be- havior. This model implied that a person's smoking decisions at any point in time are dependent on that person's smoking history. However, unlike most of the more recent econometric applications of addictive be- havior, this analysis assumed that individuals behave myopically-that is, they ignore the future conse- quences of their cigarette addiction when making cur- rent smoking decisions. Using data on 13,794 persons who participated in the 1979 National Health Interview Survey, Mullahy (1985) estimated smoking prevalence and average cigarette consumption separately for men and women (aged 17 years and older). In finding that a person's past cigarette smoking had a significant impact on current smoking decisions, the analysis supports the hypothesis that cigarette smoking is an ad- dictive behavior. The study also found that both smok- ing prevalence and average cigarette consumption w-ere in\:erselv related to cigarette prices. Finally, Mullahy estimated that men were somewhat more re- sponsive to price than women (total price elasticities of demand were -0.56 and -0.39, respectively). Wasserman and colleagues (1991) used data from several of the National Health Interview Surveys from the 1970s and 1980s to consider how the price sensi- tivity of cigarette demand changed over time. Using a generalized linear model, the investigators concluded that cigarette demand has become more responsive to price over time. In the earlier years of their sample, they found that increased cigarette prices did not re- duce cigarette smoking. However, they estimated that, beginning in 1985, when the overall price elasticity of cigarette demand was -0.23, increases in cigarette prices Lvould reduce smoking. As part of the same study, these investigators used data on 1,891 youth aged 12-17 years who had participated in the Second National Health and Nutrition Examination Survey (1976-1980). Unlike Lewit and colleagues (1981), Wasserman and colleagues (1991) found that the esti- mated price elasticity for youth was not statistically different from that for adults. Indeed, the estimated effects of price on youth smoking were not statistically different from zero in any of the models. The investi- gators attributed their relatively low estimates of the price elasticity of demand to their including in their demand equations an index that controlled for smok- ing restrictions. This index, which was highly corre- lated with price, had a negative significant effect on smoking (particularly on young people's decision to smoke). Wasserman and colleagues argued that be- cause of the high correlation between the index and cigarette prices, excluding this index would lead to biased estimates of the effect of prices on demand. Indeed, when they excluded the index from their esti- mated equations, their estimated price elasticities were comparable to those from other studies. Chaloupka (1990,1991,1992) used data from the Second National Health and Nutrition Examination Survey (1976-1980) in applying the Becker and Murphy (1988) model of rational addiction to cigarette smoking. The assumption of rational (or nonmyopic) addictive behavior implies that individuals consider, to some degree, the future consequences of their cur- rent smoking decisions (which depend on past choices). Chaloupka's estimates supported the hy- potheses that smoking is an addictive behavior and that the future consequences of this addiction are an important determinant of current cigarette smoking. Moreover, the estimated long-run price elasticity of demand (in the range of -0.27 to -0.48) was well above that obtained when the addictive aspects of cigarette Table 6.9. Estimates of the price elasticity of cigarette demand for youth and young adults from individual-level data Study Lewit et al. 1981 Lewit and Coate 1982 Grossman et al. 1983 Chaloupka 1991 Wasserman et al 1991 Not statistically different from adults (generalized linear modeling); not statistically different from zero (two-part model) Douglas and No significant effect of prices on Hariharan 1994 smoking initiation decisions Chaloupka and Grossman 1996 -0.675 -0.638 -1.313 Chaloupka and Wechsler 1997 -0.53 -0.38 -1.11 Chaloupka et al. 1997 -0.43 -0.16 -0.59 Estimated price elasticities Prevalence Quantity Total -1.20 -0.25 -1.44 -0.74 -0.20 -0.89 0.88 -1.55 -0.67 -0.62 0.11 -0.51 -0.93 0.91 -0.02 -0.89 0.73 -0.16 Not statistically different from zero Comments Health Examination Survey, Cycle III, 1966-1970; ordinary least squares methods for consumption and smoking participation; aged 12-l 7 years. 1976 National Health Interview Survey; ordinary least squares methods; elasticities by age and sex; aged 20-25 years. (1974) (1976) (1977) (1979) National Household Surveys on Drug Abuse, 1974,1976,1977, and 1979; least squares methods; aged 12-17 years. Second National Health and Nutrition Examination Survey, 1976-1980; instrumental variables methods; detailed modeling of addiction; aged 17-24 years. Second National Health and Nutrition Examination Survey, 1976-1980; generalized (iterative weighted) least squares and two-part methods; aged 12-l 7 years. 1988 and 1989 National Health Interview Surveys; hazard models of smoking initiation; detailed modeling of addiction. 1992,1993, and 1994 Monitoring the Future surveys of Bth, IOth, and 12th graders; two-part methods; mostly aged 12-18 years. 1993 Harvard College Alcohol Study; two-part methods; college students mostly aged 18-22 years. 1992,1993, and 1994 Monitoring the Future surveys of Sth, lOth, and 12th graders; smokeless tobacco use by young males; two-part methods; mostly aged 12-18 years. Table 6.9. Continued Estimated price elasticities Study Prevalence Quantity Total Lewit et al. 1997 -0.87 (prevalence) -0.95 (intentions) Centers for Disease -0.37 -0.21 -0.58 Control and Prevention 1998 Douglas 1998 DeCicca et al., unpublished data, April 1998 DeCicca et al., unpublished data, August 1998 Dee and Evans, unpublished data, 1998 Evans and Huang, unpublished data, 1998 Chaloupka and Pacula 1999 No significant effects of prices on 1987 National Health Interview smoking initiation decisions; elasticity Survey; hazard models of smoking of approximately -1 .O for duration initiation and cessation; detailed of smoking modeling of addiction. -1.32 (8th grade) -0.95 (10th grade) -0.71 (12th grade) -0.03 (smoking onset, 8th to 12th grade) 1988 National Education Longitudinal Survey; treats each wave indepen- dently for prevalence; longitudinal data used to estimate effect of price on smoking onset. -1.994 to -0.746 (8th grade) -1.230 to -0.660 (10th grade) -0.982 to -0.274 (12th grade) -0.505 to -0.025 (smoking onset, 8th to 12th grade) -2.19 to -2.Ol(Sth grade) -1.15 to -0.94(12th grade) -0.79 to -0.63 (smoking onset, 8th to 12th grade) -0.20 (1977-1992) -0.50 (1985-l 992) -0.928 (men) -0.595 (women) -0.639 (whites) -1.108 (African Americans) Comments 1990 and 1992 data from COMMIT* sites; 9th graders. 1976-1980, 1982,1985,1987-1992 National Health Interview Surveys; two-part methods; aged 18-24 years. 1998 National Education Longitudinal Survey; treats each wave indepen- dently for prevalence; longitudinal data used to estimate effect of price on smoking onset. Re-analysis of DeCicca et al. April 1998 data with same methods; differences in sample construction and variable definitions. 1977-1992 Monitoring the Future surveys; high school seniors; state- aggregated prevalence rates; allow for state effects and state-specific time trends. 1992,1993, and 1994 Monitoring the Future surveys of Sth, lOth, and 12th graders; prevalence only; mostly aged 12-18 years. *COMMIT = Community Intervention Trial for Smoking Cessation. Table 6.9. Continued Estimated price elasticities Study Prevalence Quantity Total Harris and Chan 1999 -0.831 -0.165 -0.996 (aged 15-l 7 years) -0.524 -0.255 -0.779 (aged 18-20 years) -0.370 -0.274 -0.644 (aged 21-23 years) -0.202 -0.455 -0.657 (aged 24-26 years) -0.095 -0.234 -0.329 (aged 27-29 years) Tauras 1999 Tauras and Chaloupka 1999b Gruber 2000 0.269 to 0.466 price elasticity of cessation -0.121 -0.67 -0.791 -0.666 -0.059 (older teens, Monitoring the Future surveys) -0.210 -0.003 (younger teens, Monitoring the Future sur\,eys) -0.311 -0.029 (all teens, Monitoring the Future surveys) -1.534 -1.576 (older teens, Youth Risk Behavior Sur\,eyJ 0.419 -0.227 (younger teens, Youth Risk Behavior Survey) -0.126 -0.526 (all teens, Youth Risk Behavior Survey) -0.376 -0.145 (older teens, Natality Detail files) -0.240 -0.058 (younger teens, Natality Detail files) -0.353 -0.124 (all teens, Natality Detail files) Comments 1992-1993 Current Population Survey; two-part methods; also considered differential effects of premium and discount brand prices. Monitoring the Future survey longitudinal data; young adults; multiple failure duration analysis; parametric and semi-parametric models. Monitoring the Future longitudinal data formed from high school senior surveys for 1976-1993; mostly aged 18-32 years. 1991-l 997 Monitoring the Future surveys of 8th, IOth, and 12th graders; 1991,1993,1995, and 1997 Youth Risk Behavior Surveys; 1991-l 997 Vital Statistics Natality Detail files for teens giving birth before age 19; two-part models; state and year fixed effects. smoking rvere ignored. Furthermore, these estimates of the price responsiveness of demand rvere not sensi- tive to the inclusion of variables reflecting smoking restrictions. Chaloupka (1990, 1991, 1992) found that young adults \vere not resp0nsiL.e to changes in ciga- rette prices (in contrast to the findings of Lei\-it and Coate [19X2]) and that men and less-educated persons [Vere much more responsive to changes in cigarette prices than \vere \\romen and more-educated persons. Douglas and Hariharan (1994) applied ideas from Becker and Murphy's (1988) economic model of ad- diction to look at smoking initiation decisions. Using data from the 1978 and 1979 smoking supplements to the National Health Inter\%\\- Surly, Dou,qlas and Hariharan estimated a parametric duration model that accounted for obserl-ed patterns of smoking initiation: the "hazard" of smoking initiation rises sharpI>, from ages 12 through 20 and then declines dramatically, \vith initiation being unlikely after age 25. On the basis ot this model, the analysis found that increases in ciga- rctte prices had no impact on teenagers' decision5 tcl begin smoking. Douglas (1998) extended this I\-ark by estimating a model of the hazards of smoking in- tiation and cessation using data from the cancer risk factor supplement to the lY87 National Health Inter- \rie\v Sur\:ey. Douglas also finds little empirical c`\ i- dence that higher cigarette prices ~~~ould reduce smoking initiation. Ho\ve\rer, the ini.estigators noted that their estimated price effects \\ere likelv to be bi- ased do\vn\vard because of problems tvith the mea- surement of the price variables thev emploved. Douglas did find, ho\ve\.er, that increas& in cigarette prices significantly in&ease the likelihood of smoking cessation, concluding that a lo-percent increase in price lvould reduce the duration of smoking by approxi- mately 10 percent. More recent work by Tauras confirms the findings that higher cigarette prices induce smoking cessation (Tauras 1999; Tauras and Chaloupka 1999a). Using the longitudinal data on young adults from the Monitor- ing the Future project, Tauras (1999) estimated paramet- ric and semi-parametric duration models that allow for multiple cessation attempts bv voung adult smokers. His estimates indicate that thi likelihood of an initial cessation attempt and the probabilities of subsequent attempts rise as cigarette prices rise, \vith an average price elasticity of cessation of 0.333. In a somewhat less sophisticated analvsis using the same data that exam- ined the potential-for gender differences in the effects of price on cessation, Tauras and Chaloupka (IYYYb) concluded that the likelihood of smoking cessation among both voung adult men and young adult women rises significantlv as cigarette prices rise. Hu and colleagues (1995a) used data from the 19851991 California Behavior Risk Factor Surveys to estimate smoking prevalence and average cigarette consumption through equations that accounted for the interdependence of smoking and other behavioral risk factors. Using txvo-part methods, Hu and colleagues found that their estimates of the price elasticity of smoking prevalence were significantly lower when allowing for the interdependence of smoking and other behavioral risk factors (such as drinking and obesity), rzrhereas their estimates of the effect of price on aver- age cigarette consumption bv smokers were unaf- fected. The analysis estimated-that the price elasticity of demand \tras -0.46 overall, -0.21 for smoking preva- lence, and -0.22 for cigarette consumption. More recently, data from the 1976-1980, 1983, 1985, and 1987-1992 National Health Interview Sur- \`evs have been used to study the effects of prices on snioking among adults (CDC 199X). Researchers found that both the probability of smoking and the a\`erage cigarette consumption among smokers M'ere inversely related to cigarette prices, \\,ith an overall estimated price rlasticitv of demand of -0.25. In addition, they found significant differences in price responsiveness for \.arious subpopulations, including those defined by ract/ethnicity, age, family income, and gender. They found that blacks are ttl-ice as responsive as ivhites to changes in cigarette prices and that Hispan- ics are e\`en more price sensitive. Similarly, the re- searchers' estimated price elasticity of -0.58 for young adults (aged 18-24 years) is lvell above that estimated for the full sample, lvhereas individuals with family incomes at or belolv the sample median \vere about 70 percent more responsive to price than those with higher family incomes. Finally, they found that men are much more price responsive than \vomen. To determine Irhether smokers engage in any form of compensating behavior in response to higher cigarette taxes, E\rans and Farreliy (1998) focused on the data from the 1979 Smoking and 1987 Cancer Con- trol Supplements to the National Health Interview Sur\,ey. These supplements \l-ere unique in that they collected information on the brand of cigarettes smoked. This information was converted into detailed data on tar and nicotine content, length of cigarette, and type of filter. The investigators found that continuing smok- ers engage in compensating beha\?or in response to higher cigarette taxes. That is, they found that smok- ers in high-tax states were more likelv than smokers in lo\\.-tax states to smoke higher-tar alid higher-nicotine cigarettes as \vell as longer cigarettes. This compensat- ing beha\,ior bv continuing smokers left their average dailv tar and nicotine intake unchanged. Moreover, younger smokers \2-ere much more likely to engage in this compensating behavior, so much so that the higher taxes led to an increase in average daily tar and nico- tine intake among continuing young adult smokers. Recent research by Chaloupka and colleagues fo- cused on the price responsiveness of cigarette smoking among adolescents and young adults. Chaloupka and Wechsler (1997) used 1993 data from 16,277 students in 140 U.S. colleges and universities to estimate the price elasticity of cigarette smoking among young adults. Using two-part methods, the investigators separately estimated the effects of prices on smoking prevalence and on average consumption among smokers after con- trolling for restrictions on cigarette smoking and limits on youth access to tobacco. College students, who were mostly aged 18-22 vears, were very responsive to changes in cigarette pi-ices. The estimated price elastic- ity of smoking prevalence in this population was -0.53, and the elasticity for average cigarette consumption was -0.58, for an overall price elasticity of demand of -1.11. Chaloupka and Grossman (1996) employed simi- lar methods to examine cigarette smoking among more than 110,000 voung people participating in the 1992, 1993, and 199i Monitoring the Future surveys of 8th-, IOth-, and 12th-grade students. Like several other re- searchers, Chaloupka and Grossman found that smok- ing bv vounger persons is vcrv responsive to changes in cigaiette prices. Their estimated elasticitv of smok- ing prevalence for this sample of mostlv 12- through 18-year-olds \~as-0.675, rvith an overall e&mated price elasticity of demand centered on -1.313. Chaloupka and Pacula (1999) used these data to look at the differential response by gender and race, concluding that Young men and young African Americans are more respon- sive to price than young \vomen and voung iyhites. Most recently, Tauras and Chaioupka (John A. Tauras and Frank J. Chaloupka. Price, clean indoor air laws, and cigarette smoking: evidence from longitudi- nal data for young adults, unpublished data, July 1, 1998) used data from the longitudinal component of the Monitoring the Future sur\.eys to estimate the ef- fects of price on young adult smoking. Using 35 pan- els formed from the 1976 through 1993 high school senior surveys, they estimated models controlling for unobserved state and individual factors affecting ciga- rette demand. For their sample ofvoung adults, mostly aged 18-32, Tauras and Chaloupka estimated an over- all price elasticity of demand centered on -0.79. Taken together, these estimates imply that increases in ciga- rette prices Mould lead to relatively large reductions in smoking among adolescents and young adults. This conclusion is supported by recent studies by Lewit and colleagues (1997) and Evans and Huang (William N. Evans and Lynn X. Huang, Cigarette taxes and teen smoking: new evidence from panels of rc. peated cross-sections, unpublished data, April 15,1998; Harris and Chan 1999; Gruber 2000). Lewit and co]- leagues used data for ninth-grade students in 1990 and 1992 collected in the 22 North American communities from the Community Intervention Trial for Smoking Cessation (COMMIT). They found that both youth smoking prevalence and youth intentions to smoke are inversely related to cigarette prices, with estimated price elasticities of -0.87 and -0.95, respectively. Evans and Huang estimated a somewhat smaller effect of -0.20 for high school seniors by using annual, state- level measures of smoking prevalence aggregated from the 1977 through 1992 Monitoring the Future surveys, However, they concluded that this had increased over- time, estimating an elasticity of -0.50 for the period from 1985 through 1992. Harris and Chan (1999), LIS- ing data from the 1992-1993 Tobacco Use Supplement to the Current Population Survey, provide consistent- evidence that price responsiveness falls with age. Their estimated elasticities range from -0.996 for 15- to 17- vear-olds to -0.329 for 27- to 29-year-olds. Gruber (2000) reaches a somewhat different conclusion using data from the 1991 through 1997 Monitoring the Future survevs, the 1991,1993,1995, and 1997Youth Risk Be, ha\,io; Surveys, and the 1991 through 1997 Vital Sta- tistics Natality Detail files for teens giving birth before their 19th birthday. His estimates indicate that older teens are relatively more responsive to price than younger teens (approximately 17 to 18 years of age compared M.ith approximately 13 to 16 years of age) His estimated price elasticity of smoking prevalence for older teens centers on -0.67, ivhile he finds thal younger teens, on average, are not sensitive to price In addition, he concludes that price sensitivity among older teens is greatest for more socioeconomically dis advantaged groups, such as voung blacks or those witk less educated parents. In contrast, DeCicca and colleagues (Phili DeCicca, Donald Kenkel, and Alan Mathios, Puttiq out the fires: \vill higher taxes reduce youth smoking? unpublished data, April 1998) concluded that highe cigarette taxes have a verv small impact on smokin: initiation among youth. -Using data from the 1988 1990, and 1992 waves of the National Education Longi tudinal Study (NELS) of 1988, and treating each wav separately, the investigators estimated price elasticitie for youth smoking prevalence comparable to those di: cussed abo\,e. Horvever, when they used the 1ongitL dinal data to examine the onset of daily smokin Reducirl,o Tobacco U_se betlveen 8th and 12th grade among youth not smok- ing in 8th grade, DeCicca and colleagues found little effect of price. In a separate analvsis of the same data, Dee and E\.ans (Thomas S. Dee and William N. E\rans, A comment on DeCicca, Kenkel, and Mathios, unpb- lished data, May 10, 1998) come to the opposite con- clusion. Dee and El-ans made t\l-o adjustments to the construction of the sample used bv DeCicca and colleagues-including respondents \vith missing data on some co\Tariates (about 20 percent of the sample) and redefining several variables based on the categori- cal data. After making these changes, Dee and E\,ans estimated a price elasticity for the onset of smoking of -0.63, consistent M'ith several of the other recent stud- ies of vouth smoking based on cross-sectional data. in response to Dee and E\.ans, DeCicca and co- leagues (Philip DeCicca, Donald Kenkel, and Alan Mathios, Putting out the fires: \\ill higher taxes reduce vouth smoking?, unpublished data, August 1998) COP ducted a reanalysis of NELS data b!, using an aItern,i- tive approach to dealing Lvith the problem of missing data. Their reanalysis produced some\vhat more sig- nificant estimates for the effect of cigarette taxes on the onset of daily smoking betlveen 8th and 12th grade; the implied price elasticities from alternati\.e specifi- cations ranged from -0.023 to -0.X5. Ho\\-e\.er, smaller, less significant effects are found for models that employ cigarette prices. After obtaining separate estimates based on race and ethnicity, DeCicca and col- leagues concluded that higher cigarette taxes ha\,e little impact on smoking onset bv black and \2-hite vouth but significantly reduce on.& among Hispanic $outh and youth of other races. The use of longitudinal data to research the impact of cigarette tax and price changes on smoking initiation is clearly an important and ap- propriate step. The differing conclusions from earlier studies of the same data suggest, hobyever, that these discordant results should be Meighed cautiously against the prevailing findings of recent studies. Finally, two recent studies by Ohsfeldt and col- leagues (1997, 1999) examined the impact of cigarette and other tobacco taxes on the probabilities of ciga- rette and smokeless tobacco use by males 16 vears of age and older using data from the 1985 and 1942/1993 Current Population Surveys. To account for the po- tential reverse causality between demand and tobacco control policies (including taxes), the researchers esti- mate a simultaneous equations model. They find con- sistent evidence that higher cigarette taxes reduce the probability of smoking. Behavioral Economics Studies of Cigarette Demand Behavioral economics is the relatively new ap- plication of the principles of consumer demand theory to experimental psychology (Hursh and Bauman 1987). In a laboratory setting, behavioral economists study- ing addictio~i-related behaviors focus on the impact of unit price on drug dependence, including nicotine dependence. Price, in this literature, is defined as the response required to receive one dose of the drug (Bickel et al. 1993; Bickel and Madden 1999). As in standard economic theory, a key prediction of this branch of behavioral economics is that drug consump- tion is in\.ersely related to price. One advantage of this experimental approach in the analysis of cigarette demand is that it allows researchers to study the ef- fects of differences in cigarette prices that are many times larger than the price differences observed in cross-sectional data, time series data, or both. One limi- tation, hoive\,er, is that these methods are generally applicable onlv to dependent individuals. Thus, for temple, their do not pertain to initiation. In a seiies of papers, Bickel, DeGrandpre, and their colleagues reported the results of research on ciga- rette smoking in their behavioral economics labora- tar!' (Bickel et al. 1991, 1992; DeGrandpre et al. 1992, 1991; Bickel and DeGrandpre 1996). In the experi- ments, nicotine-dependent smokers were rewarded \vith two puffs on a cigarette after the completion of a specified number of responses. The total number of puffs received is the measure of consumption, and the number of responses required is the measure of price. The number of responses required to receive two puffs varied from 100 to 3,200, thereby allowing the research- ers to study the impact of price on demand over a large range of prices. As in the econometric and other stud- ies described previously, this experimental approach found an inverse relationship between cigarette smok- ing and price. More interesting, however, is the nature of the relationship between price and consumption. The investigators found that the price elasticity of demand rose as price rose. That is, the percentage reduction in consumption for a given percentage rise in price was larger at higher prices. Studies of Smokeless Tobacco Use and Price Although numerous studies have examined the impact of cigarette prices and smoking prevention policies on cigarette smoking, relatively few studies ha\,e examined the corresponding issues for smoke- less tobacco use, and \.irtually none consider such use in di\Terse culture groups. Similarly, few analyses have examined the possible substitution of smokeless to- bacco products or cigarettes in response to changes in their relative prices. Ohsfeldt and colleagues begin to address these gaps in the literature in two studies of smokeless to- bacco use (Ohsfeldt and Boyle 1994; Ohsfeldt et al. 1997, 1999). Using state-level data for males aged 16 years and older who had participated in the Septem- ber 1985 Current Population Survey, Ohsfeldt and Boyle examined the impact of various tobacco taxes on the prevalence of smokeless tobacco use. Their analysis, which controlled for other determinants of demand, found that higher taxes on smokeless tobacco were associated with lower use of smokeless tobacco. The prevalence of smokeless tobacco use, however, was positively related to cigarette excise taxes. The inves- tigators suggested that these findings might partly explain the growth in smokeless tobacco use among young males during the 1980s. During this period, when cigarette excise taxes ivere rising more rapidly than smokeless tobacco taxes, comparatively larger increases occurred in cigarette prices. As the research previously described indicates, increases in cigarette prices significantly reduce cigarette smoking. Ohsfeldt and Boyle's analysis, howe\rer, suggested that tobacco use overall might not be significantly reduced, because some smokers might turn to using the comparatively less expensive smokeless tobacco products. These find- ings were generallv confirmed by the analysis by Ohsfeldt and colleagues (1997) of the individual-level data from the September 1985 Current Population Sur- vey and their subsequent analvsis of data from the September 1992, January 1993, dnd May 1993 surveys (Ohsfeldt et al. 1999). The authors concluded that higher smokeless tobacco taxes reduce the probability of smokeless tobacco use but that higher cigarette taxes, while reducing the probability of smoking, increase the likelihood of smokeless tobacco use. Similarly, using data on young males from the 1992, 1993, and 1994 Monitoring the Future suri'eys of 8th-, lOth-, and 12th~grade students, Chaloupka and colleagues (1997) concluded that both the pre\.alence and the frequencv of smokeless tobacco use are in- xrersely related toits price. Thev estimated an overall price elasticity of smokeless tobacco demand by young males of -0.39, with more than two-thirds of the effect on the prevalence of smokeless tobacco use. Cigarette Prices and Other Substance Use Little is known about the relationships between cigarette prices and other substance use, whereas much is known about the impact of cigarette price on smok-m ing. Economists define two goods as complements if an increase in the price of one good reduces the con- sumption of not only that good but also the consump-- tion of the other. Conversely, substitutes are goods for which an increase in the price of one results in an increase in the consumption of the other. A few very- recent econometric studies have examined the relation- ship between cigarette prices and other substance use (Pacula 1998a,b; Chaloupka et al. 1999; Farrelly et al. 1999; Pacula et al. 2000). Research on patterns of substance use among youth generally concludes that youth begin with to- bacco, or alcohol, or both and that some youth progress to marijuana and other illicit drug use (Kandel 1975; Kandel and Yamaguchi 1993; USDHHS 1994). Other research concludes that cigarette smoking is a signifi- cant predictor of both the probability and the frequency of other drug use (USDHHS 1988; Henningfield et al. 1990). This research suggests that cigarettes and other substances are complements for one another and that higher cigarette prices, by discouraging smoking among youth, could significantly reduce youth and adult drinking and illicit drug use. Pacula (1998a), in the first econometric examinaj tion of this "gateway hypothesis," used data from the National Longitudinal Survey of Youth to examine tht impact of cigarette prices in earlier years on curreni marijuana use by young adults. Her estimates are con- sistent with the gateway hypothesis; that is, higher pas cigarette prices (which are expected to reduce past ciga, rette smoking) reduce the likelihood that a young aduf currently uses marijuana. However, she finds no rela tionship between contemporaneous cigarette price! and marijuana use (Pacula 1998b). Chaloupka ant colleagues (1999) used data from the 1992 through 199~ Monitoring the Future surveys of Bth-, IOth-, and 12th grade students to examine the relationship betweel current cigarette prices and current cigarette smokin; and marijuana use. They found that higher cigarettl prices, in addition to reducing current cigarette smok ing, also reduce current marijuana use. Farrelly an4 colleagues (1999) found similar evidence for adult using several of the recent National Household Sur \`eys on Drug Abuse. In addition, they found tha higher cigarette prices reduced alcohol use. More rE cently, using a longer time series of data from the Monk toring the Future surveys of 12th-grade student! Pacula and colleagues (2000) found little impact c Reducing Tobacco Usr cigarette taxes on youth marijuana use. The growing evidence suggests that cigarettes and marijuana are not substitutes for one another, implying that higher ciga- rette prices Mill not lead to increased marijuana use, with several studies implying the opposite-that higher cigarette prices will reduce both cigarette and marijuana smoking. Much more research is needed, however, to firmly establish these relationships. Discussion A few general conclusions can be dra1z.n from these studies of the effects of cigarette prices on smok- ing. First, increases in cigarette prices lead to signifi- cant reductions in cigarette smoking; most studies, using a wide variety of data and methods \1-ith \,arious strengths and weaknesses, predict that a IO-percent increase in price \2-ill reduce o\,erall cigarette consump tion bv 3-3 percent. Second, the effects of increases in Taxation of Tobacco Products cigarette prices are not limited to reductions in average cigarette consumption among smokers but include sig- nificant reductions in smoking prevalence. These ef- fects on smoking prevalence constitute both an increase in smoking cessation among smokers and a reduction in smoking initiation among potential young smokers. Third, although evidence concerning the effects of prices on adolescent smoking is mixed, the majority of the e\,idence from recent studies indicates that ado- lescents and young adults are significantly more re- sponsive than adults to changes in cigarette prices. Most recent studies found that adolescents and young adults \vere tivo to three times more sensitive than adults to price. Ongoing research, particularly that based on longitudinal data, will help clarify this issue. Finally, the limited number of studies of smokeless tobacco use suggest that increases in smokeless tobacco prices lvould reduce the prevalence of smokeless to- bacco use. As the preceding section indicates, numerous studies of the demand for cigarettes confirm a funda- mental principle of economics: increased tobacco prices will reduce tobacco use. In general, several fac- tors will determine the retail prices of cigarettes and other tobacco products. For example, factors that re- duce the supply of tobacco will raise the prices of to- bacco products. As described previously, these factors include tobacco price support programs, market po\l-er and collusive behavior among firms in the markets for tobacco products, and restrictions on trade in tobacco and tobacco products. The most important policv- related determinants of prices, however, are taxes on tobacco products. In the United States, tobacco is taxed in various ivays by the federal, state, and local governments. The most important of these are the excise, or per unit, taxes imposed on cigarettes and the general sales tax (an ad valorem tax) applied to cigarettes and other tobacco products in most states. Ad valorem taxes are a fixed percentage of the price and thereby increase or de- crease as price changes. Excise taxes, on the other hand, do not change over time ri,ith prices. Tobacco taxes have relatively low administrative costs and can generate substantial revenues. In recent years, increased taxation of tobacco products has been used as a strategy to reduce tobacco consumption and thereby to improve public health. For example, the health benefits of tax-induced reductions in smoking were often cited by supporters of the federal cigarette excise tax proposed as part of the Clinton admini- stration's proposed Health Security Act of 1993, which included an increase of 75 cents per pack. (The act did not pass.) Similarly, anticipated large reductions in youth smoking were, in part, the rationale for tax in- creases of up to $2.00 per pack proposed as part of most proposals for national tobacco legislation and the average 52.00 state and federal tax set as a goal for 2010 by the Healthy People 2010 initiative. The health benefits of higher taxes were also the focus of the large voter-initiated tax increases in Arizona, California, Massachusetts, Michigan, and Oregon, as well as the large legislated tax increases in Alaska, Maine, and elselvhere. Rationales for Tobacco Taxation Alternative approaches have been used to deter- mine the appropriate level of cigarette and other to- bacco taxes. One such approach is the historical or comparative standard, which looks at the relative value of these taxes over time or cross-sectionally. A second approach is to use an efficiency standard based on the external costs of smoking; this approach implies that tobacco taxes can be thought of as "user fees" suffi- cient to cover the external costs of tobacco use. This approach, however, raises questions concerning the fairness of such taxes. A further argument has been made for substantial increases in tobacco taxes, because these tax hikes Mould lead to substantial reductions in the morbidity and mortality associated with ciga- rette smoking. Finally, because taxes on cigarettes and other tobacco products are a relatively simple way to generate revenues, it has been suggested that these taxes can be set at levels that maximize their returns. Each of these alternatives will be discussed. Historical or Comparative Standard Federal Tobacco Taxes Tobacco has been taxed in North America since the British government first imposed taxes during co- lonial times. Beginning in 1794, the U.S. government imposed tobacco taxes that periodically rose \vith rev- enue needs and subsequentlv fell because of consumer opposition. Since 1864, \vhen cigarette and other to- bacco taxes were included in a package to finance the Civil War, taxes on tobacco in one form or another have remained a part of the federal tax system. Taxes con- tinued to rise and fall over the next 87 years, generally increasing with revenue needs during the Spanish- American War, World Wars I and II, and the Korean War (Table 6.10). The final Lear-related increase in the federal excise tax per pack of cigarettes w'as from 7.0 cents to 8.0 cents per pack on November 1,1951, lvhere it remained for the next three decades. The most recent federal tax increases were moti- vated by a need to raise revenues for a different purpose-to reduce the increasing federal budget defi- cit. The first of these hikes in the federal cigarette ex- cise tax came as part of the Tax Equity and Fiscal Responsibility Act of 1982 (Public Lalz- Y7-248), xvhich temporarily doubled the per pack tax to 16.0 cents, effective January 1, 1983. The tax was to revert to 8 cents on October 1, 1985, but after several extensions, the 16-cent tax was made permanent in 1986. As the result of two I-cent increases included in the Omni- bus Budget Reconciliation Act of 1990, the tax per pack was increased to 20.0 cents on January 1, 1991, and- then to 24.0 cents on January 1, 1993. Finally, as a re- sult of the 1998 budget agreement, federal cigarette excise taxes are scheduled to rise by 10 cents per pack in 2000 and by an additional 5 cents per pack in 2002.- Also as part of the Consolidated Omnibus Bud- get Reconciliation Act of 1985, taxes of 8.0, 24.0, and- 45.0 cents per pound were imposed on chewing to- bmacco, snuff, and pipe tobacco, respectively. These were the first new federal taxes on chewing tobacco and snuff since 1965, when the taxation was set at 10 cents per pound. These taxes are currently 12.0,36.0; and 67.5 cents per pound (Table 6.11). This assessment amounts to approximately 2.7 cents per 1.2-ounce can of snuff, 2.3 cents per 3-ounce pouch of chewing to- bacco, and 6.3 cents per 1.5-ounce pouch of pipe to- bacco. Tobacco for roll-your-own cigarettes is not taxed at the federal level. State aud Local Tobacco Taxes All 50 states and the District of Columbia cur- rently impose excise taxes on cigarettes. The first of these was a tax levied by Iowa in 1921. It was fo!- lowed in 1923 by taxes in Georgia, South Carolin% South Dakota, and Utah. On October 1, 1969, NortE Carolina became the last state to impose a tax on ciga- rettes. As of May 1, 2000, these taxes ranged from 2.E cents per pack in Virginia to S1.ll per pack in New York (Table 6.12). Forty-four states currently imposs taxes on tobacco products other than cigarettes (Tabh 6.13); only 17 states imposed such taxes in 1964. Ir general, these other taxes are ad valorem taxes. The general sales tax in most states applies to cigarette: and other tobacco products, with the tax base in mos states including the excise tax. As of November 1,199s these sales taxes added 8-25 cents per pack to the pricl of cigarettes (Table 6.12). In eight states, 450 cities ant counties impose additional taxes on the sale of ciga rettes, and 85 of these also tax other tobacco product: The largest of the local cigarette taxes are those im posed in Chicago (combined county and city taxes c 34 cents per pack) and New York City (8 cents pe pack). At least until the 195Os, state taxes on cigarette were enacted and raised to generate revenues rathe than to discourage consumption. The average ye; such taxes were initiated in the six major tobaccc producing states (1939) slightly predates the averag year for the other states (1940) (Warner 1981). Befor the widespread publicity on the health consequence of smoking, the average tax rate in the six tobacco stat6 was only slightly lower than that in the other statt (2.5 vs. 2.9 cents per pack). Since the release in the mid-1950s of the first reports describing the adverse health effects of cigarette smoking, and even more so since the 1964 release of the initial Surgeon General's report on smoking and health, state governments have actively used cigarette taxes as a principal tool in their Table 6.10. Federal cigarette excise taxes, selected dates, 1864-2002 Effective date June 30, 1864* Aoril 1, 1865+ 1 August 1,1866t March 2,1867 July 20, 1868 March 3,1875 March 3,1883 August 15,1897 June 14,1898 Julv 1, 1901" Julv 1, 1910 October 4, 1917 February 25,1919 July 1, 1940 November 1,1942 November 1,195l January 1,1983 January 1,199l January 1,1993 Tax per pack of 20 cigarettes (cents) 4.0, 8.0, x.0+20'; 10.0 3.0 0.8, 2.-I 3.5 1 .o 2.4,1.0 2.0 3.0 1.08,2.16 2.5 1.1 6.0 6.5 7.0 8.0 16.0 20.0 24.0 campaigns to reduce tobacco use. For example, the grew substantially over this period. By May 1, 2000, number of tax increases has risen from an average of the simple average of cigarette taxes in the six largest less than three per year in the early 1950s to an aver- age of more than eight per year in the late 195Os, and a tobacco-growing states was 7.1 cents compared with record 22 states increased their cigarette taxes in 1965 (Table 6.14). Similar activity occurred during 1967- 46.5 cents in the remaining states and the District of 1970, tlhen antismoking ads were broadcast under the Columbia. Fairness Doctrine and after cigarette advertising on television and radio was banned in 1971. The once- negligible difference in cigarette excise tax rates be- tM-een the tobacco-producing states and other states January I,2000 34.0 January 1,2002' 39.0 The use of increased cigarette and other tobacco taxes to discourage all tobacco use was even more ob- v,ious in the late 1980s and earlv 1990s. In November 1988, California v,oters approved the Tobacco Tax and Health Protection Act (Proposition 99), the then- largest single increase (25 cents per pack) in any state excise tax on cigarettes. New taxes were also imposed on other forms of tobacco. The novel feature of this tax hike M'as that 20 percent of the new revenues gen- erated by the tax increase was earmarked for tobacco- related education activities and 5 percent was allocated to tobacco-related research. The success of Proposition 99 in California led to a similar voter-approved measure in Massachusetts. In November 1992, voters passed Question 1, which raised the state cigarette tax from 26 cents to 51 cents per pack and increased the state tax on chewing Table 6.11. Federal excise tax rates (cents/pound) on chewing tobacco, snuff, and pipe tobacco, selected years, 1986-2002 Year Chewing tobacco Snuff Pipe tobacco *Lower rate applied to cigarettes valued at $6 or less 1986 8.0 24.0 45.0 per 100 packs of 25 each. `Lower rate applied to cigarettes valued at $5 or less per 100 packs of 25 each. $Lower rate applied to cigarettes valued at $8 or less per 1,000. Higher rate applied to cigarettes valued at more than $12 per 1,000. "Lower rate applied to cigarettes valued at $2 or less per 1,000. `Scheduled. Source: Orzechowski and Walker 2000. 1991 10.0 30.0 56.25 1993 12.0 36.0 67.5 2000 17.0 51.0 95.67 2002* 19.5 58.5 109.69 *Scheduled. Sources: Advisory Commission on Intergovernmen- tal Relations 1991; Bureau of Alcohol, Tobacco and Firearms 2000. Table 6.12. State cigarette excise taxes and sales taxes (cents/pack) aDplied to cigarettes State Alabama Alaska Arizona Arkansas California Colorado Connecticut Delaware District of Columbia Florida Georgia Hawaii Idaho Illinois Indiana IoM-a Kansas Kentucky Louisiana Maine Maryland Massachusetts Michigan Minnesota Mississippi Missouri Excise tax rate May 1, 2000 16.5 100.0 58.0 31.5* 87.0 20.0 50.0 24.0 65.0 33.9 12.0 100.0 28.0 58.0 15.5 36.0 24.0 3.0 20.0 74.0 66.0 76.0 75.0 18.0 18.0 17.0 - Sales tax November 1, 1999 11.0 0 16.0 13.0 25.0 0 19.0 0 19.0 17.0 8.0 15.0 11.0 20.0 13.0 11.0 13.0 15.0 11.0 18.0 16.0 18.0 20.0 19.0 19.0 11.0 L State Excise tax rate Sales tax May 1, November 1, 2000 1999 Montana Nebraska Nevada New Hampshire New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Oregon Pennsylvania Rhode Island South Carolina South Dakota Tennessee Texas Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming 18.0 34.0 35.0 52.0 80.0 21.0 111.0 5.0 44.0 24.0 23.0 68.0 31.0 71 .o 7.0 33.0 13.0 41.0 51.5 44.0 75 -. 82.5 17.0 59.0 12.0 0 13.0 20.0 0 17.0 14.0 13.0 10.0 18.0 13.0 12.0 0 17.0 23.0 13.0 11.0 21.0 18.0 15.0 15.0 11.0 23.0 15.0 16.0 11.0 *Arkansas tax can rise to 33 cents if the state does not appropriate adequate funds for breast cancer research and control. Sources: Orzechowski and Walker 2000; Centers for Disease Control and Prevention, Office on Smoking and Health, State Tobacco Activities Tracking and Evaluation System, unpublished data. tobacco by 25 percent. Although Massachusetts law recommended that at least part of the funds be all prevents funds raised by the tax from being earmarked cated to activities related to reducing tobacco use. for tobacco-related education and prevention efforts, More recently, Michigan voters in 1994 enactc the funds are placed into a Health Protection Fund, Proposal A, which changed the financing for MicI and the wording of the approved measure strongly gan public schools. Part of this plan included raisir the general state sales tax (Ivhich is applied to ciga- rettes and other tobacco products) from 4 to 6 percent and tripling the state excise tas on cigarettes to 73 cents per pack, representin g the largest single increase in cigarette taxes ever implemented in the United States. NeM taxes kvere also imposed on \.arious other tobacco products. Six percent of the ne\v revenues \vere ear- marked for health impro\,ement actilrities, including tobacco-related education and pre\.ention efforts. III November 1994, Arizona \.otcrs appro\.ed the Tobacco Tax and Health Care Act, \\-hich included <3 40-cent increase in the state cigarette tax \vith enrmark- ing provisions similar to those in California, Massa- chusetts, and Michigan. At the same time, ho\ve\,er, \-oters in Colorado rejectecl a tax hike of JO cents pei- pack \\ith similar features. In No\,ember 1996, Oregon voters approved Measure 41, \1-hich increased cigarette taxes bv 30 cents per pack, raised the tax on other to- bacco products from 35 to 65 percent of i\.holesale price, and dedicated a portion of the increased re\`- enue to tobacco use pre\,ention and education. Sim- lar large cigarette-tax increases, including some that dedicate significant funds to tobacco control acti\.ities, haire been recently legislated in a number of states, including Alaska, Maine, Nelv Jersey, and I'Ke\\, York. In addition, in 1998, \.oters in California appro\.ed an additional 50-cent per pack increase in the state ciga- rette tax. The relative ease \\ith \2-hich cigarettes and other tobacco products can be transported and the potential profits from illegal acti\,itv of this kind have limited state and local governments' abititv to further raise tobacco taxes. The large disparities-in price resulting from differences in tobacco taxation create incenti\.es to (1) smuggle on a casual level (invol\ring small quan- tities for personal use) or on an organized le\,el (in- \,olving large quantities, generally for resale); (2) purchase cigarettes through tax-free outlets, including military stores and American Indian reservations; and (3) illegally divert cigarettes within the usual distri- bution system by forging tax stamps, \vhich results in underreporting. Altogether, this "butt legging" (ACIR 1977) can result in a net loss of revenues \vhen tobacco taxes are increased. Although casual smuggling has always been a problem, states reported that organized smuggling activities rose significantly after the cigarette tax hikes of the late 1960s. In response to state pressure, the Trafficking in Contraband Cigarettes Act of 1978 (Pub- lic Law 95-575) was enacted. This act, w-hich dealt onlv with the organized smuggling of cigarettes, prohib- ited the single-transaction transport, receipt, shipment, possession, distribution, or purchase of more than 60,000 cigarettes not bearing the tax indicia of the state in lvhich the cigarettes \vere initially sold. The ACIR (1985) suggests that the law was even more effective than its proponents predicted. Casual smuggling, ho\\-e\,er, may become a more significant problem as the differences betlveen cigarette taxes in neighboring states increase as the result of some of the recent large tax hikes in some states. Se\.eral econometric analyses of cigarette demand ha\.e carefullv considered the effects of price differen- tials on organized and casual cigarette smuggling on state cigarette sales (Baltagi and Levin 1986, 1992; Chaloupka and Saffer 1992; Becker et al. 1994; Saba et al. 1995; Jackson and Saba 1997; Yurekli and Zhang X00). In general, these studies concluded that smug- gling has a significant, but small, impact on cigarette demand, implving that a state cigarette tax increase \\ill lead to s&me smuggling. Yurekli and Zhang (3000). for example, estimate that, on average, 6 per- cent of state cigarette tax revenues were lost due to smu~gting activities in 1995. However, given the mag- nitude of these estimates, Merriman (1994) and Baltagi and Le\ in (1992) estimated that state cigarette taxes AI-C belo\\. their rel,enue-maxin~izing levels. Thus, states can raise cigarette taxes and generate increased re\.enues, even as cigarette sales decline and interstate smuggling increases. Cigarette Taxes and Cigarette Prices Increases in cigarette and other tobacco taxes re- sult in higher prices for these products. Most ciga- rette taxes, howe\-er, are excise taxes; unless they are increased regularlv o\`er time, the value of the tax will fall in real terms (ifter analysis accounts for the effects that inflation, as measured by the Consumer Price In- dex, has on the tax). Because taxes are an important component of price, one of the consequences of an ex- cise tax system w.ith relati\rely infrequent increases is that, at least during the period between excise tax in- creases, the real price of cigarettes will fall over time as the prices of other goods and services increase more rapidly. When trends are examined in real cigarette prices o\`er the past four decades, three clear periods are ob- served (Table 6.15). The first is 1955-1971, when states lvere increasing taxes not only to raise revenues but also to discourage smoking. The real value of state taxes during this period approximately doubled from 13.1 cents (1982-1984 dollars) to 26.4 cents per pack. This increase was more than sufficient to offset the re- ductions in the real federal tax (from 29.9 cents to 19.8 Table 6.13. State tax rates on tobacco products other than cigarettes as of January 1,200O State Taxes on other tobacco products Alabama Cigars retailing for: a) ~3.5 cents each or less, 5150 per thousand; b) >3.5 and 55 cents each, $3.00 per thousand; c) >5 and 28 cents each, $4.50 per thousand; d) >8 and 110 cents each, $7.50 per thousand; e) >10 and 520 cents each, $15 per thousand; f) >20 cents each, $20.25 per thousand. Little cigars: 2 cents for each 10 or fraction thereof. Smoking tobacco: a) 11.125 ounces, 2 cents; b) >0.125 ounces and <2 ounces, 5 cents; c) >2 ounces and <3 ounces, 8 cents; d) >3 ounces and 14 ounces, 11 cents; e) 3 cents additional tax for each ounce or fraction part thereof over -l ounces. Chelving tobacco: 0.75 cents of each ounce or fraction thereof. Snuff: a) ~0.625 ounces, 0.5 cents; b) >0.625 ounces, and 11.625 ounces, 1 cent; c) >1.625 ounces and 12.5 ounces, 2 cents; d) >2.5 ounces and 53 ounces, 2.5 cents; e) >3 ounces and 15 ounces (cans, packages, gullets), 3 cents; f) >3 ounces and 15 ounces (glasses, tumblers, bottles), 3.5 cents; g) >5 ounces and 56 ounces, 4 cents; h) 1 cent additional tax for each ounce or fraction thereof over 6 ounces Alaska Arizona 73? of \\-holesale price. Cigars retailing for: a) 15 cents, 6.1 cents for each 3 cigars; b) >3 cents, 6.1 cents each. Little cigars: 12.9 cents for each 20 or fraction thereof. Smoking and che\ving tobacco and snuff: 6.5 cents per ounce or major fraction thereof. Arkansas California* Colorado Connecticut* Delaware District of Columbia Plug tobacco: 1.6 cents per ounce or fraction thereof. 23% of manufacturers' invoice price. 61.56? of \vholesale price.* 20') of manufacturers' price. 20") of manufacturers' price. 15'; of \Vliolesale price. None. *Little cigars taxed at the same rate as cigarettes. `California rate reset at beginning of each fiscal year; Ne\v Hampshire rate reset semiannually. TMaryland tax becomes effective July 1, 2000. Sources: Orzechowski and Walker 2000; Centers for Disease Control and Prevention, Office on Smoking and Health, State Tobacco Activities Tracking and E\raluation System, unpublished data. Table 6.13. Continued State Florida Georgia Taxes on other tobacco products Smoking tobacco, chelving tobacco, and snuff: 25% of wholesale price. Little cigars: 1%.eighing 53 pounds per 1,000, 2 mills each. All other cigars: 13'5 of wholesale price. 40'7 of lvholesale price. 405 of wholesale sales price. 18"; of M-holesale price. 15'; of \vholesale price. 22'; oi ivholesale price. 10"; of original invoice price from the manufacturer to the wholesaler. hone. Cigars: a) a list price of $120 per thousand or less, tax is 8% of net invoice price; b) a list price of o\.er S120 per thousand, tax is 20% of net invoice price. Smoking tobacco: 33 T of net invoice price. Cheitring tobacco and snuff: 62' ; of Mholesale sales price. Cigars and smoking tobacco: 16 (; of Lvholesalc sales price. All other products 15'; of ivholesale price. 75(; of lvholesale price for smokeless tobacco products. 15% of wholesale price for cigars and pipe tobacco. 16? of Mholesale price. 35'; of lzholesale price. 157 of manufacturers' list price. IO? of manufacturers' price. 12.5% of bvholesale price. 15? of wholesale price. 302 of M-holesale price. Chewing tobacco and snuff: 17.9? of wholesale price invoiced to retailer. 482 of wholesale price. 25% of product value. 20% of wholesale price. 2% of wholesale price. 28% of wholesale price. 17% of \2-holesale price. Cigars, cheroots, stogies, etc., weighing >3 pounds per thousand retailing for: a) 14 cents each, $10 per thousand; b) >-1 cents each, $30 per thousand. Little cigars: 9 mills each. Smoking tobacco: 403 of factory list price. Chewing tobacco and snuff: 307 oi factory list price. Hawaii Idaho Illinois Indiana 10M'il* Kansas Kentucky Louisiana hlaine* Maryland' Massachusetts Michigan Minnesota Mississippi Missouri Montana Nebraska Nevada New Hampshire+ New Jersey New Mexico New York North Carolina North Dakota Ohio Oklahoma Table 6.13. Continued State Taxes on other tobacco products Oregon* Pennsylvania Rhode Island South Carolina South Dakota Tennessee* Texas 65':; of ivholesale sales price. None. 20c; oi w,holesale price. Cigars, cheroots, stogies, etc., retailing for: a) 15 cents each, $11 per thousand; b) >5 cents each, S20 per thousand. Little cigars: 2 cents ior each 8 or fraction thereof. Smoking tobacco: 36'4 of manufacturers' price. Chewing tobacco and snuff: 5% of manufacturers' price. 10% of wholesale price. 6% of wholesale price. Cigars: Tax on cigars and tobacco is based on weight per 1,000 and retail selling price. a) 23 pounds per 1,000, 1 cent for each 10 cigars; b) >3 pounds per 1,000 and retailing for 13.3 cents each, $7.50 per 1,000; c) >3 pounds per 1,000, retailing for >3.3 cents each and containing a substantial amount of nontobacco ingredients, $11 per thousand; d) >3 pounds per 1,000, retailing for >3.3 cents each and containing a substantial amount of nontobacco ingredients, $15 per thousand; Utah Vermont Virginia Washington West Virginia Wisconsin Wyoming e) Chelving, pipe, or smoking tobacco, and snuff: 35.213% of the manufacturers' list price exclusi\:e of anv trade discount, special discount, or deal. 35'; of manufacturers' selling price delilwed into state. 41c; of distributors' price. None. 74.9'; of Ivholesale price. None. 2OY of u\-holesale price. All other products 2O'r of rvholesale price. cents per pack); as a result, cigarette taxes continued to account for about 50 percent of cigarette prices. During the 197Os, howe\rer, the real price of ciga- rettes dropped significantly because of the stability of cigarette excise taxes and the relati\,ely rapid increases in the prices of other goods and serv-ices. During this period, the real value of the federal cigarette tax (which was unchanged in nominal terms) fell bv more than 50 percent, and the real value of state taies dropped by nearly as much. The net result \vas a decline of 38.5 percent in the real price of cigarettes. Moreo\,er, during this period, taxes as a share of cigarette prices- fell from 46.8 to 33.1 percent, because the nontax com- ponent of real price was relatively stable. Since 1981, however, the real price of cigarettes has increased sharply, from 69.3 cents to 127.1 cents- per pack in November 1992, and further in early 1993. Important factors behind this increase were the fed- eral tax increases in 1983,1991, and 1993, which tripled- the nominal value of the cigarette excise tax. Also important M'as the steady rise in the real value of av- erage state excise taxes on cigarettes, from a low of Table 6.14. Number of increases and decreases in state excise taxes on cigarettes, July 1,1950-May 1,200O 1950 1951 1952 1953 1954 1955 956 1 957 958 959 960 961 962 963 964 965 966 967 Year 1968 1969 1970 1971 1972 1973 1974 1975 Increases (Decreases) Year 2 7 (1) 0 2 3 11 5 (1) 8 4 15 3 (2) 17 (1) 2 13 5 22 4 (1) 12 8 20 7 16 5 2 2 5 1976 1 1977 4 1978 1 (1) 1979 3 1980 2 1981 6 (1) 1982 10 1983 13 1984 1 1985 11 1 Y86 6 1987 13 1988 3 1989 14 (1) 1990 8 1991 13 (1) 1992 7 1993 15 (2) 1994 8 1995 5 1996 2 1997 9 1998 2 1999 3 2000 1 Increases (Decreases) Sources: Orzechowski and Walker 2000; Centers for Disease Control and Prevention, Office on Smoking and Health, State Tobacco Activities Tracking and Evaluation System, unpublished data. 14.0 cents per pack in 1982 to 19.4 cents per pack in 1993. However, even with the increases in the real values of the federal and state taxes on cigarettes, taxes as a share of price fell substantially from 1981 to 1993 Reducing Tobncco Use (from 33.1 to 24.9 percent). The most important factor behind the rise in real cigarette prices, then, was the sharp rise in nontax (i.e., manufacturer-added) price components. In 1981, the real value of the nontax por- tion of average cigarette prices was 46 cents. By 1993, this amount was 79.5 cents, which is an increase of more than 70 percent. As described earlier in this chap- ter, in "High Tobacco Concentration and the Impact of Prevention Policies," much of this increase was attrib- utable to the less than perfectly competitive supply side of the cigarette market. The result of the increases in both the tax and the nontax components of ciga- rette prices was an increase of almost 85 percent in the real price of cigarettes from 1981 to 1993. Real cigarette prices declined sharply as a result of "Marlboro Friday" in April 1993, when wholesale cigarette prices, first ior Marlboro then soon after for other premium brands, were cut by 25 percent. More recently, however, real cigarette prices have risen sig- nificantly. These increases are partly the result of in- creases in state and federal cigarette excise taxes over the past few years. More important, however, are the significant increases in wholesale cigarette prices be- ginning in 1997. These prices increased by more than 12 percent between March 1997 and April 1998, return- ing to their 1992 nominal level (USDA 1998a1, in part the result of increased costs associated with tobacco industry settlements with Mississippi, Florida, Texas, and Minnesota. Wholesale prices increased an addi- tional 45 cents per pack in November 1998, on the day the Master Settlement Agreement was announced. This increase, the largest in history, was followed nine months later by an additional 1 g-cent per pack increase (USDA 2000). International Tobacco Taxes Among industrialized countries around the world, the United States has one of the lowest average prices and taxes on cigarettes (Table 6.16). As of December 31, 1996, the average tax in the United States was 66.0 cents per pack, well below the taxes imposed in almost every other industrialized country. At that time, taxes in various other countries, in U.S. dollars, ranged from $5.23 per pack in Norway to 47 cents per pack in South Africa. Most developed countries have at least double the average tax in the United States. Some in- teresting features of these taxes include earmarking for tobacco-related education and other health-related activities (in Denmark, Finland, Iceland, Peru, and else- where), the creation of state-based Health Promotion Foundations in Australia and the Health Sponsorship Council in Ne\v Zealand to fund sporting and artistic Table 6.15. Cigarette taxes and cigarette prices, 1955-2000 (cents/pack) Taxes as a percentage of average prices Real average federal tax+' Real Weighted Average Average Real average average federal cigarette average cigarette Year state tax*+ taxt price* state tax+" price' 1955 3.5 8.0 22.7 13.1 84.7 1956 3.8 8.0 23.2 14.0 85.3 1957 3.9 8.0 23.8 13.9 84.7 1958 4.0 8.0 25.0 13.8 86.5 1959 4.2 8.0 25.6 14.4 88.0 1960 4.7 8.0 26.1 15.9 88.2 1961 4.7 8.0 26.1 15.7 87.3 1962 5.1 8.0 26.9 16.9 89.1 1963 5.2 8.0 26.8 17.0 87.6 1964 5.6 8.0 27.9 18.1 90.0 1965 5.9 8.0 28.2 18.7 89.5 1966 6.9 8.0 30.0 21.3 92.6 1967 7.1 8.0 30.5 21.3 91.3 1968 8.4 8.0 32.3 24.1 92.8 1969 9.1 8.0 32.8 24.8 89.4 1970 10.2 8.0 37.1 26.3 95.6 1971 10.7 8.0 38.9 26.4 96.0 1972 11.6 X.0 30.0 27.8 95.7 1973 12.1 8.0 10.3 27.3 90.8 1974 12.1 8.0 41.8 24.5 84.8 - *State taxes are an average of taxes in all taxing states (42 in 1955; 50 in 1970 and thereafter) and the District of Columbia, weighted by tax-paid cigarette sales in those states. `Nominal and real average state and federal tax data are for the fiscal year ending June 30. %ice reflects the median retail price for cigarettes (including generic brands) in all taxing states, generally as of November 1 of the state fiscal year. %`ercentages cannot be calculated directly from the tax and price information, because taxes are weighted average taxes for the entire fiscal year, whereas prices and percentages are generally as of November 1. `Real cigarette taxes and prices are obtained by dividing the nominal taxes and prices by the national Consumer Price Index; the average of 1982-1984 is the benchmark. 7Preliminary estimate. Source: Orzechowski and Walker 2000. 48.7 47.4 48.8 48.0 46.6 48.9 48.6 48.3 49.4 49.3 49.8 51.4 50.8 49.2 48.9 47.7 16.8 47.7 18.4 47.6 29.9 29.4 28.5 27.7 27.5 27.0 26.8 26.5 26.1 25.8 25.4 24.7 24.0 23.0 21.8 20.6 19.8 19.1 18.0 16.2 events previously backed by the tobacco industry, and the differential taxes on cigarettes w+th high-tar and high-nicotine content used in previous years in the United Kingdom (Roemer 1993). One consequence of the differences in cigarette taxes and prices across countries is the potential for casual and organized cigarette smuggling and other forms of tax evasion. The cigarette industry, for ex- ample, frequently argues that cigarette tax increases Table 6.15. Continued Year Weighted average state tax*+ Average federal tax+ Average cigarette price$ Taxes as a percentage of average price5 Real average state tax+' Real average federal taxtA Real average cigarette price' 1975 12.2 8.0 44.5 44.5 22.7 14.9 82.7 1976 12.4 8.0 47.9 41.4 21.8 14.1 84.2 1977 12.5 8.0 49.2 40.5 20.6 13.2 81.2 1978 12.9 8.0 54.3 37.1 19.8 12.3 83.3 1979 12.9 8.0 56.8 35.5 17.8 11.0 78.2 1980 13.1 8.0 60.0 34.3 15.9 9.7 72.8 1981 13.2 8.0 63.0 33.1 14.5 8.8 69.3 1982 13.5 8.0 69.7 29.9 14.0 8.3 72.2 1983 14.7 12.0 81.9 26.8 11.8 12.0 82.2 1984 15.3 16.0 93.7 33.2 14.7 15.4 91.1 1983 15.9 16.0 97.8 32.3 14.8 14.9 90.9 1986 16.2 16.0 104.5 30.8 14.8 14.6 95.3 1987 16.9 16.0 110.0 29.9 14.9 14.1 96.8 1988 18.2 16.0 122.2 28.1 15.4 13.5 103.3 1989 21.8 16.0 127.5 26.5 17.6 12.9 102.8 1990 24.7 16.0 131.1 26.4 18.9 12.2' 110.3 1991 25.9 16.0 153.3 25.6 19.0 11.7 112.6 1992 26.5 20.0 173.5 25.6 18.9 14.3 123.7 1993 28.0 22.0 183.7 23.9 19.4 15.2 127.1 1994 31.5 24.0 169.3 31.4 21.3 16.2 114.2 1995 31.2 24.0 175.8 31 .o 20.5 15.7 115.4 1996 31.7 24.0 179.6 31.6 20.2 15.3 114.5 1997 31.8 24.0 185.4 30.5 19.8 15.0 115.5 1998 34.1 24.0 195.0 31.5 20.9 14.7 119.6 1999 36.4 24.0 217.5 28.2 21.8 14.4 130.6 2000 39.891 29.Oq 292.6 22.1 23.241 16.9_ 170.58 will actually lead to reductions in tax revenues due to nonexistent or relatively weak policies concerning smuggling and other tax evasion (British-American cigarette smuggling and their lack of enforcement Tobacco Company Limited 1994). The smuggling (ACIR 1977,1985; Joossens and Raw 1995; Joossens et problem is exacerbated by the relative ease with which al., in press). Joossens and Raw (1995, 1998) argued tobacco products can be transported, the potential prof- that many of these other factors can be as important its from this illegal activity, the presence of corruption as price differences in spawning cigarette smuggling. and organized crime, the widespread street selling, the For example, they noted that there is little evidence of availabilitv of tax-free and duty-free cigarettes, and the smuggling in some of the highest priced European Table 6.16. Average retail cigarette price and total taxes per pack (U.S. dollars/pack), selected countries, December 31,1996 Country Norlvay United Kingdom Ireland Denmark Finland Australia Sweden Ne\v Zealand Canada (highest provincial taxes) Singapore Hong Kong France Belgium Germany Canada (average provincial taxes) Austria Netherlands United States (highest state taxes) Italy Canada (lowest provincial taxes) United States (average state taxes) Greece Portugal United States (low-est state taxes) Thailand Tailvan Brazil Spain South Africa Average retail price 7.05 3.27 3.94 1.75 4.54 4.50 4.47 4.17 4.09 3.72 3.62 3.47 3.23 3.18 3.00 2.8-l 2.66 2.65 2.17 2.02 I .90 1.82 1.77 1.60 1.58 1.45 1.43 1.08 1.04 Total taxes 5.23 4.30 4.16 4.02 3.48 2.92 3.13 2.79 2.97 1.87 1.76 2.61 2.39 2.28 1.97 2.11 1.94 1.24 1.59 1.12 0.66 1.33 1.43 0.34 0.89 0.62 1.06 0.81 0.47 Tax as a percentage of retail price* 74 82 84 85 77 65 70 66 73 50 49 75 74 72 66 74 73 47 73 55 35 73 81 21 56 43 74 75 45 Notes: (al Figures given are for a package of 20 of the most popular price category; (b) exchange rates are from the Bank of Canada Official Exchange Rates as of December 31, 1996. *The tax as a percentage of retail price refers to the portion of the average retail selling price that composes all applicable taxes and other fees imposed on the product. Source: Smoking and Health Action Foundation (Canada), unpublished data, April 30, 1997. countries, including France, Nor\\-a\; Sl\.eden, and the United Kingdom, ivhereas there is &tensi\.e et-idence of smuggling in countries \vith relati\.ely lo\r prices, such as Spain and Ital\,. Merriman and colleagues (in press) pro\-ide empiri&l evidence that the perceil-ed le\.el of corruption explains more of the 1 ariance in experts estimates of the magnitude of cigarette smug- gling than do cigarette prices. Moreo\w, Joossens and colleagues (Joossens and Rain 1498; Joossens et al., in press) concluded that much of the smuggling that does occur in Europe and elselihere is encouraged by mul- tinational tobacco companies. Thursb\ and Thursb\T (1994) prrn.ided empirical support for this argument, based on their analysis of data from the United States from tvhich thev concluded that increases in federal cigarette excise taxes lead to increased comnicrcial cigarette smuggling. Perhaps the most interestin, 0 international con- parison is betlvcen cigarette tax polic\ in the United States and Canada. In 1970, a\-w-age taxes (includin;: sales taxes) on cigarettes \\.rre 30 cents per pack in Canada and 30 cents per pack in the United States. BL- 1980, the al'erage Canadian tax, 46 cents per pack. ~va7; double the U.S. tax. Real prices in both countries had fallen sharplv throughout the lY7Os, but after lY80, the gap bet\\Teen' the tivo countries \\.idened rapidI\,. One main reason for this change w.as the adoption oi an ad valorem tax bv the federal and pro\,incial jio\wnments in Canada. As a result, cigarette taxes in Canada doubled bet\veen 1980 and 1981, leading to a 25 percent increase in real cigarette prices. In response to pressure from the cigarette industry, hoi\-e\,er, the ad \,alorem tax structure j\.as replaced \vith an excise tax system in 1984. `The growth in Canadian taxes slolved o\w the next fe\v vears. Most taxing took place at the pro\in- cial rather than the federal level. In 1988, ho\ve\w, the Canadian federal government committed to an aggressive campaign to reduce tobacco use; highlight- ing the campaign \vas a ban enacted that vear on to- bacco advertising. In 1989, the federal tax;vas raised b>, 2 cents per cigarette, and another hike of 3 cents per cigarette occurred in 1991. At the same time, pro- vincial taxes \yere increasing rapidlv. By early 1991, the average tax per pack of cigareties lx-as $2.96 (in U.S. dollars), lvhich is more than five times the a\er- age U.S. tax. The large disparities in Canadian and U.S. ciga- rette prices led to substantial smuggling, which \vas enabled by the long stretches of unmonitored border between Canada and the United States, the relativeI\ wak border controls, and the high concentration of the Canadian population near U.S. borders (S\vcanol and Martial 1943). Much of the black market trade that resulted 12-as in Canadian-produced cigarettes that had been exported to the Cnited States (exports were not subject to the Canadian taxes) and then smuggled back into Canada. Relatively little black market trade in\-ol\-ed cigarettes produced in the United States; U.S. cigarettes use a blend of tobacco different from Cana- dian cigarettes and are less desired by Canadian smok- ers (Siveanor and Martial 1993). In a short-lived effort to reduce the smuggling problem, a tax of 80 cents per pack \vas applied to Canadian cigarette exports in mid- February 1992. This tax \vas repealed six weeks later, although preliminary evidence indicated that it had btwl successful in reducing smuggling (Sweanor and Martial iYY4). After the repeal of the export tax, Ca- nadian cigarette exports to the United States rose dra- maticall\: and smuggling increased again. In response to an aggressive industry-sponsored campaign, the federal tax on cigarettes in Canada was reduced bv SJ.00 per carton on Februarv 9, 1993. More- o\-er, the federal go\.ernment agreed to match provin- cial reductions in taxes up to an additional $10.00 per carton. Quebec immediatelv louvered its provincial tax b! Sl 1 .OO per carton for a &tat tax cut of $26.00 per carton, leading to a 50-percent drop in price. By Au- gust 1991, four other provinces had reduced cigarette taxes substantially. These cuts reduced the average Canadian tax per pack from $2.96 before the federal tax cut to 91.97 as of December 31, 1996 (in U.S. dol- lars), lvhich \vas an amount still well above the aver- age U.S. cigarette tax of 66 cents per pack at that time. The Canadian experience \vas cited by the tobacco industrv during the recent debates over the proposed national tobacco settlement as evidence that a black market in cigarettes would develop in the United States in response to large cigarette tax increases. However, there is little evidence to support this contention. Given that Canadian cigarette taxes xvere reduced because of smuggling from the United States, it is likely that these taxes lvould be increased if the United States were to adopt large tax increases, making it unlikely that wide- spread smuggling of cigarettes from Canada into the United States ~vould occur. Cigarette prices in Mexico, hot\-e\.er, are rvell below. those in the United States, and large increases in U.S. prices could make smuggling cigarettes from Mexico a highlv profitable lrenture. TO date, ho\zre\-er, no empirical evidence supports the con- tention of significant smuggling of cigarettes from Mexico into the United States. Furthermore, unlike the U.S.-Canadian border, the border between the United States ancl Mexico is relativ-elv short and heavily guarded, making it much more- difficult to smuggle large quantities of a bulk\, product like cigarettes. Finally, se\-era1 relati\.ely eas\. options exist for limiting cigarette smuggling (Joossens and \ran der MerIve 1997; Joossens et al., in press). These include prominent tax-paid markings on all tobacco products and sizable increases in the penalties for cigarette smuggling. The ACIR (19851, for example, concluded that the Trafficking in Contraband Cigarettes Act (Pub- lic Law 95-575), which prohibited the transportation, receipt, shipment, possession, distribution, or purchase of large quantities of cigarettes that did not bear the tax indicia of the state in which the cigarettes are found, led to a significant reduction in interstate cigarette smuggling resulting from interstate price differentials. Discussion If one applies Cook and Moore's (1993) discus- sion of alcohol taxes to cigarette taxes, a provocati\,e question arises when one compares previous cigarette excise taxes \\ith current ones: whv is the current tax rate deemed appropriate ivhen it ii just over one-half the level that \\`as deemed appropriate in lY51? Un- less it is in the public interest to tax cigarettes at a much lower rate norv than then (an odd notion, given that in 1951 much less evidence ~vas available on the health hazards of smoking), a case can be made for restoring taxes to their earlier levels. Similar arguments can be made at the state level, particularlv in those states Lvhere taxes have not changed or ha;e been increased modestlv and infrequentlv over time. Other, comparative- standards for appropriate taxes could be used. For example, as sl1oiz.n in Table 6.12, state excise taxes on cigarettes differ substantially; these differences reflect several factors, including the importance of tobacco for the local economv. At an- other level of comparison, large differences betiveen cigarette taxes in Canada and the United States ga\-e rise to a significant black market trade, \\,hich in turn resulted in reductions in Canadian taxes. At the glo- bal level, cigarette and other tobacco taxes in the United States are among the lokyest in industrialized coun- tries around the lvorld. Such comparisons suggest that relativelv high taxes mav be appropriate in some ar- eas and lorz- taxes appropriate in others. On the other hand, one could argue that the taxes on all tobacco products should be equivalent. This last issue is dis- cussed in greater detail in the next section, "Fairness Standard and Optimal Cigarette Taxes." Taxes on smokeless tobacco products are much lower than taxes on cigarettes, particularly at the fed- eral level. The limited research suggests that increases in cigarette excise taxes mav have reduced cigarette smoking but also may have contributed to an increased use of smokeless tobacco products (Ohsfeldt and Boyle 1994; Ohsfeldt et al. 1997, 1999). Some public health advocates and others have therefore called for the equalization of taxes on tobacco (CSH 1994; U.S. House of Representatives 1994). Fairness Standard and Optimal Cigarette Taxes Fair tax policy is an issue that is often debated- but difficult to apply when "optimal" taxes of poten- tially hazardous substances are discussed (Cook and Moore 1993). For taxes on cigarettes and other tobacco products, part of the debate revolves around the per- ceived health benefits and reductions in social costs associated with higher taxes. In their analysis of economic interventions to re- duce alcohol abuse, Cook and Moore (1993) noted that several criteria can be included to judge fairness by those on both sides of the debate. These criteria in- clude a horizontal equity criterion, which suggests that equals should be treated equally; a vertical equity cri- terion, lvhich suggests that those with the greatest abil- itv to pav should be taxed more heavily; and a benefit ciiterion-, which suggests that those who receive the greatest benefit from government activities should be tased more heavily. If the basic notion is accepted thai people Mrho are otherwise similar should be taxed dif- ferently because one uses more tobacco products thar the other (a notion that \,iolates the horizontal equit) criterion), then other questions about fairness nrisc These include questions concerning the allegec regressility of the taxes and the external costs of smok ing and other tobacco use (Cook and Moore 1993). Equity, Incidence, and Distribution of the Tobacco Tax Burden As has been discussed previously, increases il cigarette excise taxes are passed on to consumer through higher cigarette prices. Primarily because c the less than perfectly competitive nature of the ciga rette industry, prices have increased by more than rc cent increases in cigarette taxes. Because consumers wi pay at least the full amount of a tax increase in highs cigarette prices, some questions of fairness revolv around the distributional effects of the tax hike. To ur derstand these effects, it is useful to look at the relatior ship betIveen tobacco use and income (or expenditures (As Cook and Moore 119931 note, income or expend tures are not the only scale on which fairness can L- judged, but they are the most commonly used.) A 1990 report by the Congressional Budget Office (CBO), lvhich used data from the 1985-1985 Consumer Expenditure Surve); made se\,eral obser- vations. For example, expenditures on tobacco prod- ucts increased Ivith income except for people in the highest income quintile. As a percentage of posttax income, however, spending on tobacco rras highest in the lowest income quintile (4.0 percent of posttax in- come) and fell almost proportionatelv 11-ith increased income. Also, if expenditures on tobacco are consid- ered as a percentage of expenditures on all goods and services, ho\ve\.er, the share of tobacco expenditures fell graduallv over the first four income quintiles (from 1.6 to 1.1 p¢) and dropped sharplv onlv in the 1 - top quintile (to 0.7 percent). Thus, the CBO notes, if annual familv expenditures are more reflecti1.e of life- time income than annual familv income, then expen- ditures on tobacco are onlv sljghtlv regressi\.e 0l.t'~ income classes. Finally, the CBO noted that vounger tamilies spent a higher percentage of income on to- bacco products and that their share of spending on tobacco products as a percentage of total expenditures \vas higher as ivell. To examine the distributional impact of cigarette excise tax increases on consumers, the CBO simulated iThat the effects on expenditures w,ould be lvere the 1990 federal excise tax on cigarettes ( 16 cents per pack) doubled. At first glance, the simulated increase ap- peared to fall most hea\,ily on the lo\\-est income cate- gories, thereby implying that cigarette taxes are regressive. However, irhen income tax brackets and transfer payments (discussed in the next section, "Es- timates of the Costs of Smoking") Lvere indexed to ac- count for the price increases associated M.ith excise tax hikes, lolvering individual income taxes and raising transfer payments, the apparent regressilrity of the tax \vas reduced. When looking at the tax increase rela- tive to expenditures rather than income, the CBO coil- eluded that cigarette taxes \vere approximatel\ proportional rather than regressive. Finally, the CBO noted that the largest share of the simulated tax increase was paid for by families in the third and fourth income quintiles and that the smallest share was paid by fam- lies in the lowest income (first and second) quintiles. All of the CBO estimates lvere based on measures of current income. Lyon and Schwab (1995) used an alternative approach that used measures of permanent or lifetime income to examine the distributional effects of cigarette and other "sin" taxes. This approach could account for the intertemporal nature of cigarette con- sumption decisions. The investigators concluded that cigarette excise taxes are as regressi\.e as \vas implied by studies based on current income. Although cigarette taxes fall most heavily on louver income groups, two recent studies suggest that increases in cigarette taxes may reduce the perceived regressivity of these taxes. A study using data from the British General Household Survey concluded that people in the lowest income groups were the most re- sponsive to price increases (Townsend et al. 1994). Simi- lar findings have been obtained in the United States using data from 13 of the National Health Interview Surl,eys conducted from 1976 through 1993 (CDC 1998). The price elasticitv of cigarette demand by those at or belol~ the median;ncome M'as estimated to be approxi- matelv 70 percent higher than that for persons above the m;dian. Another study found that less educated persons \1-ere more responsive than more educated persons to cigarette price changes (Chaloupka 1991). Gi\.en the high correlation between income and edu- cation, the three studies implied that increased ciga- rette taxes l\,ould reduce observed differences in smoking among socioeconomic groups (i.e., that smok- ing pre\,alence is higher in the lower socioeconomic groups) and wrould thereby counter the perception that cigarette taxes are regressi\re. Recent research from de\.eloping countries supports the hypothesis that lolzper income populations are relatively more sensi- tive to price (Jha and Chaloupka 1999; see Chaloupka et al., in press, for a thorough review). Indeed, while cisnrette taxes mav fall more heavily on lower income groups, an increase in the cigarette tax, because of the greater price sensitivity of louver income smokers, may actually be progressive. Moreover, given the estimates from these studies, the health benefits resulting from reductions in smoking stimulated by increased ciga- rette taxes Lvould be disproportionate117 larger in the lowest income populations. Finally, as the CBO report pointed out, although the potential regressivity of cigarette taxes is of some concern, the U.S. tax system is a mix of many different taxes. increased progr&sivity of other taxes and trans- fer programs could be used to compensate low income families for the tax increase. The CBO considered three alternative changes-a 5-percent increase in food stamp payments, a lo-percent increase in the earned income tax credit, and a combination of the two-to offset the potential regressivity of an increase in the cigarette excise tax. In each case, the CBO concluded that these changes would spend about 15 percent of the net revenues resulting from the tax increase. A simi- lar idea \vas implicit in the proposed Health Security Act of 1993, which proposed a federal tax increase of 75 cents per pack to partiallv finance the provision of health insurance and the expansion of benefits to the uninsured and underinsured, most of whom are in lo\\ er sniioeic~tiotiiii grc'ups. Like\\ ista, se\.eral rf- cent proposals for national tobacco legislation contain pro\~isions that r~ould offset the potential regressi\Tity of large increases in cigarette taxes. Estimates of the Costs of Smoking An alternati1.e approach to the question of fair- ness deals \vith the notion that smokers and other to- bacco users impose costs on nonusers. One of these costs is the health consequences for nonsmokers of exposure to ETS. A second is the financial external effect caused by collectively financed programs (e.g., Medicaid and Medicare) Mhere payments in and out are not tied to changes in costs and life expectance caused bv smoking. Thus it can be argued that it ~vould be fair fir smokers and other tobacco users to pay for the consequences of their use. Cigarette and other to- bacco taxes are one relatively efficient approach for attaining this result. Ho\ve\,er, to set taxes at a level sufficient to co\.er fhe costs of cigarette smoking and other tobacco ~1st~ requires an estimaCe of these costs. All studies of the economic costs of tobacco LISA' have focused on Ihe costs of cigarette smoking. The Office of Technology Assessment (U.S. House of Rep- resentati\.es 1YY-l) has noted that although measuring these costs is an inexact science, three general compo- nents are included: . The direct costs of pro\idin, 0 health care ser\-ices to those persons L\.ith smokin, ~r-reldW diseases. Such costs include expenditures for pre\.enting, detecting, diagnosing, and treating smoking- related diseases and medical conditions. o The indirect morbidit\; costs associated \vith lost earnings from \\,ork because of smoking;-related illness. o The indirect mortality costs related to the loss of future earnings from premature death from smoking-related causes. Researchers ha1.e tried to estimate the economic costs of cigarette smoking bv using data from the United States (Rice et al. 1986; %anning et al. 1989, lY91; Hodgson 1992; CDC 1991; U.S. House of Representa- tives 1991; Miller et al. 1998, 1999) and elservhere (see Lightlvood et al., in press, for a comprehensi\,e retie\<.). In addition, as part of the research resulting from Propo- sition 99, se\.eral recent studies ha\.e estimated these costs for California (California Department of Health Ser\-ices 1 YY3; Rice and Max lW2; %lav and Rice 1 YY5). Most of the estimates of the economic costs of smoking have been prevalence based. That is, the,] are based on the estimated prevalence of smoking- related illnesses in a given year and on the costs asso. ciated with those illnesses. Because of the long lags between smoking initiation and the onset of mo\t smoking-related illnesses, these estimates reflect his. torical trends in smoking and thus cannot be used to predict the impact of changes in smoking prevention policies except over long periods. However, this ap- preach has been widely used because of its relative]\ simple methodology and the availability of reliab(c data (Rice et al. 1986). Several of the recent estimates of the costs of smoking have been incidence based (Oster et al. 1984; Manning et al. 1989, 1991; Hay 1991; Hodgson 19Y2). That is, these studies attempt to estimate the average additional costs of smoking over the smoker's lifetime. Cost estimates would differ by the person's age, sex, and level of smoking (i.e., a heavy smoker would ha\.e higher lifetime cosfs than a relatively light smoker with the same characteristics). These estimates of the costs of smoking can be useful for policymakers, who can estimate the change in the costs of smoking associated with a change in smoking behavior resulting from a change in policies to reduce smoking. However, these estimates are sensitive to assumptions about future costs and about issues suc11 as technological change and its diffusion (Hodgson 1988). Many of the studies of the economic costs oi smoking ha\,e included notablv different direct costs in their computations. For example, most include the costs of hospital and nursing home care, physicians' fees, and medications used to treat smoking-related illnesses. One such study estimated that these costs in 1993 \vere $50 billion and that 13.3 percent of them \z'ere paid through public sources (CDC 1994). Ho\\.- e\`er, some studies of direct costs have been limited to the costs associated \vith lung cancer only, Lvhereas 0th ers examined a more comprehensive list of smoking- related illnesses, including cardiovascular disease and chronic obstructi1.e pulmonary disease. Other more recent studies have sought a broader measure of the direct costs of smoking by comparing the differences betlveen total health care spending b> smokers and nonsmokers. The most sophisticated of these recent studies control for other risk factors like]\ to be correlated \vith smoking in an effort to isolate the impact of smoking on medical expenditures (Miller et al. 1998, 1999). These recent studies estimateci- smoking-attributable medical care costs of between $53 billion and S73 billion for 1993, or between 6.5 percent and 11.8 percent of all U.S. health care expenditure5.