Discussion: (0 comments)
There are no comments available.
| Journal of Public Policy & Marketing
View related content: Health Care
When R. J. Reynolds (RJR), the second largest U.S. cigarette manufacturer, brought its Joe Camel cartoon character from France in 1988 for use in the United States, it launched a controversy that outlived the character itself (which RJR agreed to cease using in July 1997). [n1] An assessment of the Joe Camel campaigns raises two main issues. One is the advertisements’ effects on smoking, especially by teens under 18 years of age. The second is Joe Camel’s effects on the intellectual, political, and legal environments surrounding the cigarette market. I argue here that the second class of effects vastly exceeds the first, which appears to have been negligible.
Joe Camel and Smoking
Research on whether the Joe Camel campaigns increased smoking has taken two approaches. The first operates under the notion that by targeting teenagers, the Joe Camel advertisements were bound to increase underage smoking (meaning smoking by teens under the age of 18 years, though in the late 1980s some states had lower minimum ages for purchasing or using cigarettes; see Beales 1992). In this approach, which eventually came to rely almost entirely on internal RJR documents, analysts examine the extent to which the Joe Camel advertisements were targeted at teenagers and the success of that targeting as measured by traditional marketing methods, with the assumption that such success was bound to increase youth smoking. The second approach examines market data, including advertising, consumption, and prevalence (e.g., the percentage of teens that smokes daily), in an attempt to separate the effects of the Joe Camel advertisements from other influences on youth smoking.
Scholarly assessment of the Joe Camel advertisements apparently began with three celebrated articles in the December 11, 1991, issue of the Journal of the American Medical Association (JAMA). An article by Fischer and colleagues (1991) presented data that suggested that the Joe Camel character was very familiar to young children. DiFranza and colleagues (1991) conducted surveys that indicated that the Joe Camel advertisements appealed more strongly to adolescents under 18 years of age than to adults and that the Camel brand’s market share among teenagers increased from a negligible level of .5% to about 33% in the three years after the advertisements were launched. Finally, Pierce and colleagues (1991) presented California survey data that indicated that adolescents disproportionately noticed both Marlboro and Camel advertisements and disproportionately purchased them. Two of the three articles argued that advertising causes smoking and that restrictions on advertising would reduce smoking. Therefore, DiFranza and colleagues (1991, p. 3149) stated, “A total ban of tobacco advertising and promotions, as part of an effort to protect children from the dangers of tobacco, can be based on sound scientific reasoning,” and Pierce and colleagues’ (1991, p. 3154) abstract concluded by saying that “Cigarette advertising encourages youth to smoke and should be banned” (though the body of the article states that their results “suggest that tobacco advertising is causally related” to youth addiction to tobacco).
It is apparent, however, that the JAMA studies do not attempt to demonstrate that the Joe Camel advertisements increased smoking, whether by teens or adults. In fact, none of the articles contains any data on total teenage smoking. Any conclusions about the advertisements’ impact on youth or total smoking must rely on the intuitive notion that successful targeting, evidenced by heightened awareness and an increase in market share, necessarily increases total smoking.
Subsequent research and analysis also reveal that the most striking of the JAMA results–those in DiFranza and colleagues’ (1991) article–are unreliable and highly inaccurate. When the complete data became public through court action, it was learned that the peak appeal of the Joe Camel advertisements was not to consumers under 18 years of age (as the article had stated) but to consumers 18-24 years of age. The published article simply compared people under 18 with all adults and moreover dropped the data for consumers aged 18-20 years after the lead author noted in private correspondence (also discovered in litigation) that the results of a pilot sample tended to disprove what the study was designed to prove. In addition, comparisons with other advertising recall data suggested that advertising recognition scores typically tend to decline strongly with age, at roughly the rate found in DiFranza and colleagues’ study (see DuBow 1993, 1995; Martin 1994; the data and other documents were released through the discovery process in the Mangini lawsuit, discussed subsequently). The portion of DiFranza and colleagues’ study devoted to market shares also had serious flaws. It employed old and unreliable pre-Joe Camel market share data (most of it unpublished, and almost all of it dating from 1979 to 1985), which had then been compared with survey data that were unrepresentative (because they were convenience samples from areas where Camel enjoyed unusually high market shares) and biased (because the brand preference question was asked after respondents had been exposed to seven Camel advertisements). More reliable estimates suggested a national youth market share of well under half DiFranza and colleagues’ estimate of 33% (these data are reviewed in the section on trends in market shares; see DuBow 1993, 1995)
This made for an unusual combination of events–massive publicity for the 1991 JAMA studies, bolstered by their apparently unimpeachable source, followed by a little known but quite thorough dismantling of the most important empirical results. These circumstances caused the JAMA studies to play an oddly mixed role in subsequent events. In the public arena, their influence has persisted more or less undiminished, as is described subsequently. In contrast, the JAMA studies played a far smaller role in litigation, in which they would have been subjected to scrutiny by opposing experts if the cases had not been settled.
Therefore, the Mangini litigation, a California lawsuit directed at the Joe Camel advertisements, relied primarily on RJR documents, though the plaintiff’s detailed trial brief cited DiFranza and colleagues’ (1991) and Pierce and colleagues’ (1991) market share studies (see Coughlin and Janecek 1998). That case was settled before the JAMA results could be tested before a jury.
Especially relevant is Federal Trade Commission (FTC) litigation, in which scholarly assessment of empirical scholarship is a mainstay. The 1993 and 1997 FTC complaints did not rely on the JAMA studies, nor did the FTC’s marketing expert (Joel Cohen). [n2] This is a notable omission, because the FTC suit was based not on the usual grounds of deception, but on “unfairness” in advertising (the FTC Act contains a blanket prohibition on “unfair or deceptive acts or practices”; 15 U.S.C. See 45[a]). The FTC’s definition of unfair practices evolved some decades ago into three criteria: whether the practice injures consumers, whether it violates established public policy, and whether it is unethical or unscrupulous. In its 1980 Policy Statement on Unfairness, the FTC clarified its policy by stating that the first and third criteria were largely redundant, meaning that the FTC would regard advertising unfair only if it caused net harm in the sense of imposing costs on consumers greater than its benefits. This policy was recently codified by legislation. [n3] This standard focuses on the effects of advertising on consumers rather than on the intentions of advertisers. The implication for the Joe Camel litigation is that the FTC must address the question whether the Joe Camel advertisements increased total youth smoking, as opposed to shifting market shares, which presumably would involve no net harm to consumers.
The FTC’s adoption of this approach is clear. When the FTC voted to close its investigation of Joe Camel advertising in 1994, it did so explicitly because of a lack of evidence linking the advertisements to increased smoking. Therefore, the majority statement said, “If intuition and concern for children’s health were a sufficient basis under the law for bringing a case, we have no doubt that a unanimous Commission would have taken that action long ago. The dispositive issue here, however, was whether the record showed a link between the Joe Camel advertising campaign and increased smoking among children” (Azcuenaga, Owen, and Starek 1994; also see FTC 1994b). The FTC chairman simply disagreed on whether the advertising caused more underage smoking (Steiger 1994). The 1996 Food and Drug Administration (FDA) final rulemaking, a product of close cooperation between the FTC and FDA staffs, noted that the FTC litigated the Joe Camel advertisements under its unfairness standard, which the FDA summarized by saying that “an action is unfair if it causes substantial consumer injury, without offsetting benefits to consumers or competition, which consumers cannot reasonably avoid” (FDA 1996, p. 44479). Finally, the 1997 FTC complaint expressly stated that the FTC intended to demonstrate an effect on overall youth smoking (FTC 1998b, p. 55). Some may argue (as Cohen does with considerable force in his companion article in this issue) that the FTC’s policy is unfortunate because it can tolerate seller behavior that not only is strongly objectionable but also could cause substantial harm that could not be detected in the time frame encompassed by FTC advertising litigation. That thinking suggests that the FTC should revert to the nebulous policy the 1980 policy statement was designed to end (see Cohen 2000).
Targeting, Incentives, and Intuition
In a very real sense, the approach taken in the JAMA studies was supplanted by one that avoided the use of data on Joe Camel’s actual impact on youth smoking. The alternative approach consists of an intuitive analysis of the effects of marketing, combined with detailed assessment of the contents of internal RJR documents unearthed through litigation. [n4]
An unusually sophisticated example of this non-market data approach is provided by Joel Cohen (2000) in his article in this issue. Based on his expert report as a witness for the FTC in its Joe Camel litigation, Cohen’s analysis moves far beyond the simple argument that cartoon characters appeal to youths. He advances four important points. The first pertains to RJR’s incentive to target youths and adolescents, including smokers under the age of 18 (bearing in mind that in the mid- to late 1980s, the minimum age for purchasing cigarettes was less than 18 years in many states). This incentive did not necessarily arise from an explicit desire on RJR’s part that more teens should smoke, but from a recognition that most smokers do start as adolescents and often become loyal users of the brand they adopt.
Cohen’s second point is that RJR acted on these incentives by explicitly targeting the 18-20-year-old market to increase market share at the expense of Marlboro and other brands. Moreover, many internal documents reveal or at least suggest an intention to target consumers under 18 years, because of a thorough and possibly innovative understanding of the subtle social factors that can encourage teenagers to smoke. Cohen’s third point is that even if RJR did not intend to induce smoking by people under 18 years of age, it is relatively unimportant. What is important is that regardless of the firm’s private beliefs or goals, the sophisticated targeting tools developed by RJR–the Joe Camel character and the campaigns incorporating that character–clearly appealed to adolescents, and that RJR knew that the effect was bound to be an increase in youth smoking and therefore eventually in adult smoking.
Fourth, Cohen emphasizes that RJR’s market research data indicate that its methods were successful as judged by traditional marketing benchmarks. In its plans and its assessment of those plans, RJR applied the standard hierarchy-of-effects model, one that proceeds through a series of stages from attention and recall through attitudes and beliefs, emotional response, and finally purchase–albeit with the restriction (also traditional) that few measures of actual purchase behavior were used by RJR because few such measures were available; moreover, the measures that were available (purchase intentions, for example, as well as total market sales) provided no reliable way to isolate the effects of the Joe Camel advertisements from those of other possible influences on purchase behavior.
From Incentives to Market Effects
The central policy question, which arises directly from the FTC’s legal theory, is whether the Joe Camel campaign increased youth smoking and eventually increased adult smoking. The line of reasoning just outlined necessarily stops short of addressing this question. One difficulty lies in the inherent limitations of the hierarchy-of-effects model itself. Chief among these is that however successful a marketing campaign may be in the early stages of the hierarchy, that success is by no means bound to carry on through actual choices in the marketplace (see, e.g., Vakratsas and Ambler 1999). But many, if not most, applications of this model work exclusively with intervening variables (i.e., after the advertisements but before market behavior) and never actually test advertising’s effects on market choices. Indeed, the chief purpose of the hierarchy-of-effects model is to provide a means for improving on guesswork when empirical tools for measuring actual sales impact are unavailable. The inability of this kind of data, especially at the brand level, to assess the connection between advertising and total smoking has been recognized by several of the most prominent government and closely allied researchers, including the U.K. Department of Health (1992), the 1989 Surgeon General’s Report (U.S. Department of Health and Human Services [USDHHS] 1989), and Chaloupka and Warner (1999).
A second difficulty pertains to the implications of success among youth smokers. As I explain subsequently, the under-18 market by itself seems too small and too evanescent to generate large profits commensurate with substantial and sustained advertising expenditures. Profits from increasing the under-18 market share would accrue primarily as smokers mature into the much heavier smoking of the young adult years. But large youth market shares do not necessarily persist as smokers mature. This is clear from several developments: the dramatic market share shifts in recent decades (which occasioned RJR’s attempt to resuscitate the Camel brand with the Joe Camel character), the rapid growth of discount generic brands that preceded the April 1993 “Marlboro Friday” round of premium brand price cuts, the highly publicized studies in recent years that find market shares among youth to be different from and far more concentrated than those of adults (three brands accounted for 85% of youth smokers in 1993, for example, but only 35% of adults; see USDHHS 1994b), and finally, the simple fact that Camel’s market share among adults has consistently been well below that among teens for the past decade or so.
A third, even more fundamental problem pertains to the implications of an advertising campaign that succeeds in generating a more or less permanent increase in market shares beyond what it would have been without the Joe Camel advertisements. Researchers must then ask what the impact would be on total product consumption. In his focus on certain fascinating details of youth appeals, Cohen has examined what amounts to an elaborate case study of promotional appeals based on image, taste, and so on. If successful, such appeals can improve the fit between differentiated products or images and consumer segments, including segments that are only potential users. There is no guarantee that this would draw more consumers into the market, however, in addition to those who would have entered anyway because they liked the basic characteristics of the product. It is possible for competing firms–each applying a similar hierarchy-of-effects model based on the same learning so assiduously documented and disseminated in the scholarly literature–simply to offset one another’s efforts, leaving competition at a standstill with little brand switching despite large advertising expenditures.
This reasoning applies even if it is assumed that all brand advertising is profitable in the sense of yielding greater sales for each brand than would have occurred without advertising for that particular brand. It cannot be assumed that if each firm does better with advertising than without, the industry as a whole must do better. Such an assumption is an example of a more general error in economic reasoning known as the “fallacy of composition” (Caballero 1992). The fallacy lies in assuming that whatever action helps a member of a group must help the group as a whole. The difficulty becomes apparent when so-called negative advertising in political campaigns is considered, which almost no one assumes increases total voting. It may even reduce voting, even as it increases individual candidates’ shares over what they would otherwise be (see Ansolabehere and Iyengar 1995).
It is therefore notable that in Cohen’s analysis of RJR’s success in applying the hierarchy-of-effects model, the cited RJR documents contain virtually no data linking marketing efforts (e.g., advertising expenditures) all the way to Camel’s actual sales to youths or adults, not to mention a link to total youth smoking.
This absence of data is a consequence of focusing on industry documents. The question of Joe Camel’s impact on total smoking or total youth smoking cannot easily be resolved or even addressed by working with RJR’s proprietary materials. The company had scant financial incentive to discover the extent to which its own advertising increased the sales of competing brands in addition to increasing Camel sales. And the RJR documents emphasized by plaintiffs, particularly in the Mangini case in California (see Coughlin and Janecek 1998) and the FTC case (at least judging by Cohen’s  report), are essentially silent on the advertising’s impact on non-Camel sales. Moreover, as Cohen (2000) notes, RJR’s internal data pertain almost entirely to the intervening variables that mark progress in the hierarchy of effects rather than to market data. Even if RJR documents demonstrated a potent brand effect, that would not imply a total market effect. But as I noted previously, RJR documents reveal a concern about the apparent lack of benefit from Joe Camel advertising in terms of total market share.
Trends in Market Data
I therefore turn to market data to examine the impact of the Joe Camel campaigns on youth smoking and overall smoking. A few simple trends are illuminating. The Joe Camel campaign began in the first quarter of 1988 and within a few years came under intense public attack. The details of these attacks are related subsequently, but I note here that by 1992 or 1993, Joe Camel was the topic of extensive public debate and discussion–including the American Medical Association’s (AMA’s; 1993) school-based anti-Joe Camel education program that included essay and drawing contests–even as the advertising continued. Therefore, only between 1988 and 1991 or 1992 was the marketing with the Joe Camel character relatively unimpeded.
Market Shares Among Teenagers
I am aware of only three publicly available representative national samples of brand choices among under-18 smokers since the late 1980s. The first two are the Teenage Attitudes and Practices Survey (TAPS) and TAPS-II survey conducted in 1989 and 1993, respectively, for the USDHHS (1994b, 1998). The TAPS data indicate that 8% of youths aged 12-18 years smoked Camels in 1989, and 13% in 1993. [n5] In 1998, the University of Michigan’s long-running Monitoring the Future (MTF) survey of high school students added a brand choice question on smoking (Johnston et al. 1999). This yielded a widely publicized estimate that among teens who smoked, approximately 88% smoked one of the three most advertised brands. But the share for Camel, the third-ranking brand, was only 9.6% (see Johnston et al. 1999). Thus, by 1998, Camel’s share of underage smokers was not much larger than in 1988, shortly after the advertisements began to appear.
Overall Market Shares
There are good reasons for looking closely at market shares among smokers over 18 years of age. It is unlikely that the under-18 market alone yields substantial profits. The TAPS data (which are the most reliable source on the quantity of cigarettes smoked by youths) indicate that the average teen smoker consumes only approximately eight cigarettes per day. The MTF data indicate that approximately 12% of twelfth graders smoke a half pack or more a day, and for eighth and tenth graders, the percentages are approximately 4% and 8%, respectively (on the experimental nature of very light teenage smoking, see Beales 1994b; Kovar 1998). Such a small level of consumption by such a narrow demographic group is likely to yield profits from large advertising expenditures only as the smokers move into the much heavier smoking years in their twenties. Close attention therefore should be paid to Camel’s market share for smokers over 18 years of age in the five or ten years after the Joe Camel advertisements began in 1988.
In Table 1, I present the data available to me for Camel media advertising, youth market shares, and overall market shares for the years 1985-98, based partly on proprietary data from RJR. [n6] These numbers indicate a substantial increase in Camel’s market share among all under-24 age groups in the decade after the 1987 inauguration of the Joe Camel advertisements. This surge peaked in 1992-93, after which market shares dropped below 10%. For the 18-24 group, this was what RJR had expected from the Joe Camel advertisements (see “As Intended, Camel’s YAS Share Has Grown,” [n7] one of many RJR documents entered into the Mangini litigation that confirm some of the market share data in Table 1).
Table 1. Camel Brand Market Shares and Advertising for Selected Years, 1985-98
Note: This table may be divided, and additional information on a particular entry may appear on more than one screen.
Camel Brand Share of Market (%)
Years of Age
Years of Age
Years of Age
Years of Age
Advertising Expenditures (Millions of Current Dollars)
Source: RJR proprietary internal tracking data for specific age groups (except under 18). Advertising data include magazines, newspapers, and out-of-home media (e.g., billboards) and exclude other promotional expenditures, such as sponsorship, merchandise, or sales promotion (e.g., coupons). Total market share data are taken from various years of the Maxwell Report. Under-18 data come from TAPS (1989 and 1993) and MTF (Johnston et al. 1999)
The natural question, whether these increases in the youth market would translate into increases in total market share as opposed to offsetting declines in older market segments, was something RJR worried about. At least by 1991, it was apparent that market share increases were not extending beyond a narrow age segment (see “Camel Advertising Overview” and “Camel Ex. Regular 1992 What Happened?” both from “Mangini Case History” 1998, UCSF no. B000635-000710, Bates JM007273-RJM007348). By 1995, it was apparent that Camel was no longer growing in the targeted 18-24 age segment, and total market share continued to languish (“Camel Performance Issues” in “Mangini Case History” 1998, UCSF no. B000607-000630, Bates RJR199434-199457). Camel’s total market share remained remarkably stable during this entire period, increasing from 4.3% in 1986 to a peak of only 4.8% in 1998. Whether the entire Joe Camel advertising enterprise was profitable to RJR would depend on what would have happened to the Camel brand if RJR had never imported the Joe Camel character. These data provide no reason to associate the Joe Camel advertising campaign with increased smoking, which continued to decline during these years (especially between 1987 and 1992; see FTC 1998a).
Trends in Youth Smoking
In Figure 1, I present MTF survey data on daily smoking by high school seniors. It can be seen that teen smoking in 1992 was slightly below that in 1987 (the year preceding Joe Camel) and then increased fairly rapidly (from 28% to 35%) between 1992 and 1997. [n8] Smoking by eighth and tenth graders (which MTF began tracking in 1991) increased somewhat more rapidly after 1991 and peaked in 1996, one year before twelfth-grade smoking peaked. Smoking by young adults (including college students) has followed a similar pattern, increasing since the early 1990s, according to several sources. The MTF survey, which follows a subset of its survey respondents through college, found stable smoking rates during 1986-90 followed by increases to a level of 27% in 1995; young adult smoking thus started upward before high school smoking did. [n9] A more recent survey, devoted entirely to college students, found an increase in smoking rates from 22.3% in 1993 to 28.5% in 1997. The federally run National Health Interview Survey, which is not restricted to college students, reported that smoking among adults aged 18-24 increased from 19.6% for men and 16.0% for women in 1993 to 24.0% and 18.5%, respectively, in 1995. [n10]
These trends provide no compelling reason to associate the Joe Camel advertisements with the virtually simultaneous increases in middle school, high school, and young adult smoking that began at approximately the same time, around 1992 or 1993. It would have to be argued (as some competent analysts no doubt do) that the advertisements had their main impact on all these age groups after they had become the focus of a national debate and had been the target of FTC and California litigation and school-based awareness programs. Unfortunately, I am aware of no attempt to perform a statistical analysis of the relationship between youth smoking, Camel advertising, and other factors. However, J. Howard Beales (2000) has provided me with the results of simple regressions between the MTF data for twelfth graders and the Camel advertising data in Table 1, with advertising lagged up to three years, which yields no advertising coefficients close to statistical significance. Simple regressions between advertising and the Camel market share data in Table 1, in contrast, generally yield significant coefficients on advertising. [n11]
These data suggest that the surge in youth smoking after 1992 arose from the influence of factors other than Camel advertising. These factors do not include total cigarette advertising, however, which declined while smoking was increasing (see the FTC’s annual reports on cigarette advertising), though lower prices may have been a factor for a while.
The importance of factors other than Joe Camel is more apparent even further beyond Joe Camel’s orbit of influence. Canadian and U.K. trends in youth smoking largely parallel those in the United States, despite the absence of Joe Camel from Canada (which has an almost completely different set of brands) and the United Kingdom (where cartoon characters and even human figures were abolished some years ago; see Ibrahim 1997). Youth smoking prevalence in Canada mainly declined through 1991 and then increased (see Pechmann, Dixon, and Layne 1998). The U.K.’s Office for National Statistics (see U.K. Department of Health 1997) reported similar findings: Between 1992 and 1996, weekly smoking by 11-15-year-olds increased from 9% to 11% for boys and from 10% to 15% for girls. Among 16-24-year-olds, smoking increased between 1993 and 1995 from 32% to 36% for male and 33% to 37% for female consumers.
The age at which people begin to smoke can also be examined. In Figure 2, I present the most complete time series I am aware of on this topic. It is apparent that age of initiation began to decline in 1994, six years after Joe Camel arrived and well after Joe Camel had become a national issue. However, this time series could be interpreted differently, perhaps with the idea that the 1992 and 1993 rates were anomalous interruptions in a downward trend. I am unaware of any statistical analysis of these data in conjunction with Joe Camel advertising and other variables.
Statistical Analysis of Market Data
A statistical analysis of cigarette advertising–including advertising that uses the Joe Camel character–could also be performed that would measure the impact of advertising after taking into account prices and other relevant variables. This technique is a fixture in the advertising literature, and cigarettes are the single most studied product. If this research were to demonstrate that aggregate cigarette advertising increases consumption, that result might strengthen the case that the Joe Camel advertisements increase youth smoking. But the bulk of this research finds no effect from advertising. The most recent and comprehensive research on the U.S. (Franke 1994) and U.K. (the econometric analysis in the appendix to the “Smee Report”; U.K. Department of Health 1992) markets both fail to demonstrate a statistically significant impact from advertising on consumption. Several literature reviews have reached the same conclusions. [n12]
Recent U.S. government reports, however, have tended to avoid the econometric literature. The FDA’s review of the literature, contained in its August 1995 Federal Register notice and supplemented in its 1996 notice, relied on other government reviews of the literature, specifically the 1994 reports from the Surgeon General (USDHHS 1994a) and the Institute of Medicine (IOM). The 1994 Surgeon General’s Report (USDHHS 1994a), on which the FDA relied, however, did not review any econometric studies of the U.S. or U.K. market (see pp. 188-95, titled “Research on the Effects of Cigarette Advertising and Promotional Activities on Young People”). The 1994 IOM report reviewed only one study, by Laugesen and Meads (1991). That study simply compared cigarette consumption in nations with various levels of advertising restrictions and has been subjected to severe criticism because it did not control for confounds (see the critiques by Stewart  and the discussion in U.K. Department of Health . The FDA review itself also virtually ignored the many statistical analyses of the U.S. and U.K. cigarette markets. It cited only Roberts and Samuelson (1988), who examined competitive dynamics between the low-tar and high-tar segments during 1971-82 and reached the counterintuitive conclusion that low-tar advertising affected cigarette consumption but did not affect market shares.
It should be borne in mind that Joe Camel advertising was a small proportion of total cigarette advertising, which itself had no effect on sales. This is apparent from the proprietary data in Table 1, which are generally consistent with the publicly available Maxwell data (which estimated that Camel advertising for 1994 was $32 million, or 7.3%, of $442 million total cigarette media advertising and was only 5.6% of total advertising in 1995 and 10.2% in 1996). If fluctuations in total cigarette advertising–fluctuations that often greatly exceed the total volume of Joe Camel advertising–have no effect on consumption, it is highly unlikely that the Joe Camel advertisements alone significantly increased consumption. [n13] I am aware of only one study that examines advertising and youth smoking (as opposed to total consumption). That is Beales’s (1994b, 1996) cross-sectional study, which works with brand advertising for Camel and Marlboro and aggregate advertising, combined with youth smoking data from the 1989 TAPS survey and a 1990 California survey. Employing several specifications and independent variables in a logistic regression model, Beales finds no correlation between advertising and youth smoking. [n14]
Joe Camel’s Influence on the Intellectual and Political Environments
Philip Morris proclaims that it does not want children to smoke, and spends millions each year on education campaigns to prove it. Along comes Joe Camel and, says a tobacco marketer, “makes us all look guilty in the court of public opinion.”
— The Economist
Sometimes advertisers themselves stir up charges of unfair advertising. Is anyone surprised to see the tobacco industry catching hell on every front? . . . RJR’s continued reliance on Joe Camel has placed all cigarette marketers, not only RJR, in a nowin situation.
— Advertising Age
Joe Camel’s Career in the Public Sector
Far more so than for most other products, the intellectual and political environments surrounding tobacco interact with the litigation arena. The direction of causation, however, tends to run from intellectual and political conditions to litigation. Therefore, it was the FDA’s February 1994 announcement that it was officially considering the possibility of regulating nicotine in cigarettes as a drug that launched the sweeping litigation that transformed the cigarette market in the space of five years (Daynard 1997; also see Kessler 1994). Moreover, the most momentous litigation primarily has been brought not by private parties but by state and local governments, with considerable support from public opinion and health organizations (AMA 1997; Daynard, Bates, and Francey 2000).
There are several reasons for believing that the Joe Camel campaigns exercised a substantial effect on the intellectual and political environments–not through their impact on smoking behavior, which was modest at most, but through the lens of pronouncements by antismoking advocates, public health experts, government officials, politicians, editorial writers, and others. Under attack virtually from the time they were introduced, the Joe Camel advertisements had assumed a prominent role in criticisms of the industry by at least the early 1990s. A 1991 National Cancer Institute report, for example, featured a full-color reproduction of Joe Camel (albeit with no explicit analysis of the advertising itself). In December 1991 came the three JAMA studies, released amid tremendous and continuing publicity (see, e.g., Brown 1991). In March 1992, the Coalition on Smoking or Health petitioned the FTC to investigate RJR’s use of the Joe Camel character, an investigation later supported by 27 state attorneys general (Sullum 1998, pp. 97-98). President Bush’s Surgeon General Antonia Novello, former Surgeon General C. Everett Koop, the AMA, and the American Heart Association all called on RJR to halt its campaign (Cohen 1994; Labaton 1998; Sullum 1998, p. 97). By 1993, the AMA was running school-based programs to increase adolescent awareness of and resistance to the Joe Camel advertisements (AMA 1993). Joe Camel played a prominent role in the 1994 Congressional hearings (famous for the sworn testimony of tobacco executives that cigarettes are not addictive) (Segal 1997).
Even after the advertisements themselves departed in July 1997, the Joe Camel character continued to play a role in the political arena, where it was attacked by President Clinton and Vice President Gore (see Labaton 1998, Segal 1997). In a 1998 retrospective look at the politics of tobacco, a Washington Post story (Connolly and Mintz 1998, p. A1) noted that “public anger at the industry had intensified starting in 1991, when studies revealed that RJR’s advertising cartoon figure, Joe Camel, was almost as recognizable to 6-year-olds as Mickey Mouse.” A prominent antismoking activist said in 1999, “Joe Camel helped change a nation that wanted to believe the industry didn’t market to kids to a nation that didn’t doubt that the industry would market to kids” (Tursi, White, and McQuilkin 1999, Ch. 30, Part 1). A recent report from the American Lung Association and Public Citizen (1998) includes a section titled “Threats–Joe Camel Holding American Kids Hostage.”
Government reports have persistently cited the Joe Camel advertisements as a cause of youth smoking, usually on the basis of the original JAMA studies (despite the absence of any analysis in those articles of the connections between the advertisements and total youth smoking). The 1994 Surgeon General’s Report (USDHHS 1994a, pp. 190-91), which was devoted to youth smoking, discussed the Joe Camel advertisements and summarized the December 1991 JAMA studies. Also in 1994, the IOM published an influential report on youth smoking titled Growing Up Tobacco Free. Although the most striking aspect of the IOM report was its lengthy analysis of the potential benefits of FDA regulation, it also contained a long section on the effects of advertising on youth, beginning with a discussion of the Joe Camel advertisements and the market share results from DiFranza and colleagues’ December 1991 study (IOM 1994, p. 117; the report also cited DuBow 1993, but only with respect to the pre-Joe Camel market share).
Federal agencies continue to publish similar views. A striking example is the October 1998 release of a report from the Centers for Disease Control (CDC) on smoking initiation (USDHHS 1998). Although the report simply provided historical survey data without any analysis or discussion of the causes, the press release generated several news stories that emphasized that a decline in the average age of smoking initiation had occurred since the Joe Camel advertisements began to appear (Associated Press 1998; this story was reprinted in the Washington Post, Seattle Times, Raleigh News and Observer, and presumably many other newspapers). As it happened, data from the same study, presented in Figure 2, indicate that the average age of initiation was lower when the Joe Camel advertisements began in 1988 than it was in 1993, and only after 1993 did it begin to fall. [n15]
Public health spokespeople and organizations have continued in the same vein. I have already noted the AMA’s school-based anti-Joe Camel program. More recently, former Surgeon General C. Everett Koop cited one of the December 1991 JAMA studies when he said in a March 1999 newspaper interview that a recent increase in college smoking could be attributed to Joe Camel (Teft 1999). Some of the academic literature also continues to cite the Joe Camel advertisements (e.g., Slade 1999). So do popular accounts of the tobacco industry, an example being the best-selling volume by Hilts (1996, p. 93), who cites Pierce and colleagues’ (1991) article in stating, “Historical data, like that from John Pierce in San Diego, shows that, whatever group [Joe Camel advertisements] claim to approach in documents so far exposed, they have in fact had the greatest effect on those under 18 years old.”
The Dynamics of Politically Provocative Advertising
As early as May 1992, The Economist (in the first epigraph to this section) declared that the Joe Camel advertisements made the entire industry look “guilty [of youth targeting] in the court of public opinion” (The Economist 1992, p. 22). In December 1994, Advertising Age (in the second epigraph) provided a similar assessment: The Joe Camel advertisements were hurting the industry’s reputation.
Why did RJR persist in using the Joe Camel character for years after its symbol had begun to attract unprecedented criticism, political attacks, and FTC litigation? One reason lies in the dynamics of politically provocative advertising. Firms may have an incentive to arouse political forces that are to the firm’s benefit even if they are detrimental to the industry as a whole. These dynamics run parallel to those that apply to “fear advertising,” that is, advertising that arouses fears of smoking to gain market share for cigarettes perceived to be safer. It is now widely accepted that cigarette sellers sometimes have pursued fear advertising to the point of harming overall industry welfare (on fear advertising and its effects, see Calfee 1985, 1997; Franke 1994; IOM 1994, Ch. 4; Ringold and Calfee 1989, 1990; Scheraga and Calfee 1996; USDHHS 1994a, “Changes in the Style of Cigarette Advertising,” pp. 171 ff.). This notion even provided the foundation for an antitrust allegation as part of the state attorney litigation that led to the November 1998 global settlement, in which plaintiffs claimed that an industry conspiracy, rather than the FTC, eliminated the most explicit fear advertising after 1954 (see the complaints in State of Minnesota et al. v. Philip Morris Inc. et al. 1994 and State of Washington et al. v. American Tobacco Co. Inc. et al. 1996; Solow 1999, which is a working paper by an expert witness for the plaintiffs in the Washington trial).
In persisting in its use of the Joe Camel character, RJR was trading off its financial gain against political costs. The firm’s competitors bore most of those costs. Indeed, if the main political threat was an advertising ban, as appeared to be the case until the litigation environment was so radically altered in 1994, it can be imagined that RJR might benefit from a ban. Although RJR was the number-two firm, it had steadily lost market share in the face of decades of Philip Morris advertising for its Marlboro brand. An advertising ban could prevent further losses. Thus, Adweek noted (Stolz 1992, p. 18), “There is also a theory among Philip Morris executives that RJR, by backing Old Joe all the way, is progressing with a secret plan to ensure a ban on cigarette advertising. This would freeze market share and allow RJR–which has had its tobacco troubles and is burdened with a mountain of junk debt–to save hundreds of millions of dollars on advertising each year. Improbable? Yes. But, in a way, strangely Nixonian.” Earlier, Advertising Age (Crain 1990) had also hypothesized that RJR might welcome an advertising ban.
I do not wish to suggest there is any evidence that RJR sought any of the political fallout generated by the Joe Camel character. Rather, the point is that the incentives to rein in a campaign that could harm its industry were relatively weak, certainly weaker than many observers probably realized. This is consistent with the possibility that the Joe Camel advertisements eventually had untoward effects on the political and legal environments that greatly exceeded their modest effects on youth smoking. Also important was the almost totally unanticipated scope and force of the firestorm that engulfed the industry after 1994. For decades, the industry’s political fears had centered on taxes and advertising restrictions. And several states have dramatically raised taxes in recent years. But it turned out that far more important than taxes and advertising law were the prospects of FDA regulation and especially a new and radically different wave of litigation.
The Role of the FDA
When the FDA set out in 1994 to regulate cigarettes as nicotine products, it placed great emphasis on industry documents. That is because to find that cigarettes were drugs (or rather, drug-delivery devices, in the FDA’s final view), the agency had to demonstrate that cigarettes were marketed with the intention that the products would act as a drug. [n16] As a general rule, when the FDA wishes to classify a product as a drug on the basis of its intended use, it relies on an assessment of marketing claims. Tobacco advertising, however, has avoided claims about the effects of nicotine precisely because such claims would invite FDA regulation, which has been an industry concern for decades (on the history of potential FDA regulation of cigarettes, see USDHHS  regarding the dispute over RJR’s smokeless Premier cigarette; Krulwich 1996). This meant that the FDA’s usual approach could not work. The FDA’s solution was to infer marketing intentions from industry actions that did not involve marketing communications (such as controlling nicotine yield in cigarettes) and especially from the contents of documents from the files of tobacco manufacturers. The FDA’s attempt to use industry documents as a route to regulation was ended in March 2000, however, when the Supreme Court ruled that the FDA could not regulate cigarettes under existing law (Mauro 2000).
Thus, industry documents were central to the FDA’s nicotine investigation and were bound to play a role in the FDA’s attempts to generate political support for its actions. In his early pronouncements on the FDA’s intention to regulate tobacco products, then-FDA Commissioner David Kessler strongly attacked the Joe Camel advertisements in the course of pursuing his central point, which was that tobacco marketing is the primary cause of smoking by adolescents. In these attacks, Kessler usually summarized some of the December 1991 JAMA studies. Kessler also emphasized that the FDA was uncovering what it regarded as extremely damaging evidence from industry files (Kessler 1995; Kessler et al. 1996, 1997).
These FDA activities had two effects. The first was to reinforce the popular view that the Joe Camel character caused young people to smoke. The second was to energize the litigation community, primarily through the release and condemnation of industry documents. This was not the first connection between FDA initiatives and the litigation environment. For example, the FDA’s 1991 moratorium on silicon breast implants led directly to the mass litigation that bankrupted that industry (see Angell 1996), as did the FDA’s removal from the market of one of the drugs used in the fen-phen diet drug combinations (see Johannes and Schmitt 1997).
Before moving on to litigation, I should say a few words about the FDA’s use of advertising research, a crucial topic because the FDA’s main tool for reducing youth smoking (other than restrictions on youth access) was to be controls over advertising and marketing. The FDA’s 1995 rulemaking (set forth in its 1995 Federal Register notice) made extensive use of the December 1991 JAMA studies. However, the FDA did not offer any evidence linking the Joe Camel advertisements to overall youth smoking (and the JAMA articles contained no such evidence). Indeed, the FDA’s review of the literature ignored virtually the entire body of work on the market effects of cigarette advertising, relying on a single article that, without using any advertising data at all, sought to demonstrate that advertising had caused an increase in women’s smoking in the 1970s. [n17] The undue prominence of the December 1991 JAMA articles in the FDA’s analysis is another example of how Joe Camel wielded influence not through its effects on smoking itself but through the regulatory and litigation process.
Joe Camel’s Influence on Litigation
We all know about Joe Camel.
— From the plaintiff’s opening statement to the jury in the Minnesota trial, Jan. 26, 1998; State of Minnesota et al. v. Philip Morris et al. (1994).
The Joe Camel advertisements have influenced litigation in two ways. One is by serving as the target of litigation–in the Mangini case, a class action suit brought in California (see “Mangini Case History” 1998), and in FTC litigation. The FTC first sued RJR over the Joe Camel character in 1993. I noted that the FTC charged RJR with unfairness rather than deception in its advertisements, meaning that the staff needed to show that the Joe Camel advertisements imposed harm, presumably by increasing smoking among youth. The FTC could offer little or no empirical evidence beyond what I reviewed previously, however, inasmuch as proprietary information from RJR provided little or no significant additional data on youth smoking (again, see Coughlin and Janecek 1998; FTC 1998b). This litigation ended in June 1994 when the commission voted to close the staff’s investigation of Joe Camel advertising on the grounds that the FTC staff had failed to demonstrate that the Joe Camel advertisements actually increased youth smoking (see Azcuenaga, Owen, and Starek 1994, a statement by the majority that voted to dismiss the case).
The FTC sued again in 1997 after the appointment of two new commissioners (there are five in all). As emphasized by two dissenting commissioners, the FTC cited virtually no new empirical evidence on the impact of Joe Camel advertisements on youth smoking (on cited evidence, see FTC 1997, 1998b; the dissenting statements are Azcuenaga 1997 and Starek 1997). Indeed, the FTC press release announcing the new suit cited no data on youth smoking later than the 1993 TAPS data, which had been available during the earlier litigation (Broder 1997; FTC 1997). Even as the case was about to go to trial late in 1998, the FTC emphasized the same MTF and CDC data reviewed previously (see FTC 1998b; Labaton 1998, which was based on the FTC’s trial brief filed two weeks earlier; USDHHS 1998). In its assertion that Joe Camel advertising increased teen smoking, the FTC’s trial brief (FTC 1998b) relies completely on public data sources from before 1993, plus the 1999 MTF survey showing Camel with an under-18 market share substantially below the TAPS-II results from 1993, though the FTC also suggested that the new lawsuit could be justified on the basis of newly acquired RJR documents from the FDA. [n18] The 1997 litigation ended on January 27, 1999, when the FTC again dismissed the case, not because the commission had changed its mind but because the November 1998 global settlement provided essentially all the remedies the FTC had sought (including an end to the Joe Camel character and funding for antismoking campaigns; see Wall Street Journal 1999).
The Mangini case was filed as a class action in California in the wake of the intense publicity accorded the December 1991 JAMA articles. The plaintiffs (which eventually included 11 California counties and the cities of Los Angeles, San Jose, and San Francisco) sought an end to the Joe Camel advertisements, plus support for antismoking messages (Tursi, White, and McQuilkin 1999, Ch. 30, Part 2). All the relevant Mangini materials, including a detailed complaint memo, an analysis of the documents on which the complaint relied, and the documents themselves, have been made available on the Internet. Other than Pierce and colleagues’ (1991) JAMA data on market shares, the complaint relied almost entirely on industry documents, the main empirical addition being a 1994 California study that described a rough correlation between the arrival of the Joe Camel advertisements and a reversal in a previous decline in youth smoking (Coughlin and Janecek 1998). After years of delay, the Mangini suit was finally settled in 1997, by which time RJR and the rest of the industry had already agreed (in the first abortive “global settlement” of the state attorneys general litigation) to end the use of cartoon characters. [n19] The FTC and Mangini cases thus had little independent effect beyond bringing the Joe Camel character to an end at a time when RJR was prepared to give it up anyway.
The Third Wave of Tobacco Litigation
Far more important than the Joe Camel–specific litigation is the vast enterprise known as the “third wave” of tobacco litigation (for a review from the perspective of a long-time supporter of increased litigation against cigarette manufacturers, see Daynard 1997). Following the unsuccessful individual liability suits of the 1950s and 1960s and again (with greater plaintiff resources) in the 1980s, the third wave of tobacco litigation has three distinguishing characteristics. The first is sheer size and complexity, encompassing individual smoking and from environmental tobacco smoke, antitrust suits, health care cost recovery suits, and more.
The second characteristic is the avoidance of the issue of individual responsibility in smoking. This has been achieved partly by constructing suits in which judges have excluded evidence on the smoker’s role in deciding whether to smoke (as in State of Minnesota et al. v. Philip Morris Inc. et al.  that led directly to the November global settlement; see Rybak, Phelps, and Mason 1998). Equally important has been the identification of tobacco marketing, especially advertising, as the primary cause of smoking, largely on the basis of the argument that advertising and public relations has the power to deceive even the public health community itself about the dangers of smoking. This remarkable notion formed the basis for lawsuits by health care providers and other entities, as well as the Department of Justice’s lawsuit (announced in September 1999 but still in its early stages, and with an uncertain future, at the time of this writing) (Vladek, Heyman, and Zieve 1998).
The third and most important characteristic of the new era of tobacco litigation is the role as litigants of health care insurers and especially governments at all levels, including suits by all 50 state governments and many cities and counties. The importance of governments as plaintiffs is underscored by the fact that the only suits that by the end of 1999 had yielded substantial damages payments were those brought by the state attorneys general and their allies, culminating in the November 1998 nationwide settlement that is estimated to pay more than $250 billion (primarily to state governments and plaintiff attorneys) over 25 years (for a good overview of the already vast literature on the 1998 settlement, see Redhead 1999).
The Joe Camel advertisements, beginning with publication of the 1991 JAMA articles, played several roles in these developments. One, already described, was to provide a focus for rallying political support from the public health community and the public at large. Public support was essential for litigation in which the prime plaintiffs were state governments suing in state courts before judges either appointed by state governors or elected by the population from which juries are drawn.
The Joe Camel saga also provided a template for litigation strategy. Just as the FTC investigation began with the empirically focused JAMA studies and ended by looking almost entirely at RJR documents, the massive state attorneys general suits also came to use the Joe Camel studies primarily for rhetorical purposes while relying in the court-room entirely on industry documents, including RJR documents on Joe Camel and youth targeting. Thus, the leading marketing expert in State of Minnesota et al. v. Philip Morris Inc. et al. (1994) employed no evidence at all beyond industry documents and did not attempt to familiarize himself with econometric or other work using cigarette market data (see, e.g., the expert witness testimony by Robert Dolan, in State of Minnesota et al. v. Philip Morris Inc. et al. ). Throughout, the focus has been on incentives and intentions; virtually no attention has been paid to actual effects in the marketplace. Much the same is true of other litigation, including individual suits that have generated massive judgments (still under appeal). These, too, are relying heavily on many of the same industry documents used in Minnesota, Mangini, and the second FTC case. [n20] Recently, the World Health Organization set out to apply very much the same strategy in developing nations around the world, by means of the Framework Convention on Tobacco Control (Yach 1999).
The overall litigation strategy, then, has been to rely on an assumption–presumably to be adopted by jurors–that marketing must cause smoking if there is evidence that the manufacturers showed a strong interest in youth smoking. Industry documents cannot address this assumption because (as explained previously) manufacturers had no incentive to investigate the main question, which is the effects of their marketing on total smoking. Therefore, it is not surprising that summaries of industry documents, by people involved in or sympathetic to their use in litigation, indicate that the cache of industry documents from RJR and others apparently contains very little evidence (at least after the 1970s) of actual youth targeting or of success in getting teens to smoke (as opposed to evidence of continuing research or interest in the smoking behavior and brand choices of consumers under 18 years of age) (see, e.g., Coughlin and Janecek 1998; JAMA 1995, which is devoted to the Brown and Williamson documents; Mintz and Torry 1998; Perry 1999).
Conclusions: The Legacy of Joe Camel
The Joe Camel advertising campaigns had little or no effect on smoking by youths or adults, though they may have prevented a decline in the Camel brand’s market shares and perhaps modestly increased it. Although Camel’s market share among smokers under 24 years surged from about 4% to perhaps 12% in 1992-93, and then receded to approximately 9%, Camel’s overall market share never increased more than .5% (from 4.3% to 4.8%). There is no evidence that the Joe Camel advertising increased total youth smoking, which declined between 1987 and 1992.
However, the advertisements appear to have wielded substantial influence on the larger political and legal environment. Beginning soon after the publication of three articles in the December 11, 1991, issue of JAMA, Joe Camel became a fixture in attacks on cigarette marketing by public health organizations, advocacy groups, and politicians. The most important of the JAMA results–that the Joe Camel advertisements had their strongest appeal to consumers under 18 years of age and had increased Camel’s share among youth from less than 1% to 33%–turned out to be unsubstantiated. Nonetheless, the idea that Joe Camel advertisements were targeted primarily at underage smokers and dramatically increased smoking persisted in the public health community, the popular press, and government reports. This putative role for Joe Camel appears to have substantially increased the level of public and political support for the most important antismoking activities of the 1990s, including the FDA’s nicotine regulation initiative and a tremendous wave of lawsuits by state governments and others.
The Joe Camel saga, including the original JAMA studies, also bolstered a broad shift in how cigarette advertising and marketing is assessed: away from an examination of market data such as advertising expenditures and smoking prevalence, in favor of an intuitive assessment of the motivations and intentions of cigarette manufacturers, based largely on internal documents revealed through litigation. A crucial role was played by the FDA investigation, in which the specific nature of FDA law dictated a focus on industry documents even as it facilitated the search for those materials. The result was to bring to light many documents for use by the FTC in its own litigation and by the plaintiff bar as a lever in seeking additional documents in other litigation. These documents proved to be immensely powerful in subsequent litigation, including the state attorneys general suits that culminated in the 1998 global settlement that will transfer hundreds of billions of dollars to state government and plaintiff attorneys.
n1 The chronology of the Joe Camel character in the United States is outlined in Sullum (1998) and Cohen (1994).
n2 The FTC’s 1997 brief for the 1997 case (FTC 1998b) did not cite the JAMA studies. Also see Cohen’s (2000) article in this issue.
There are no comments available.
1150 17th Street, N.W. Washington, D.C. 20036
© 2015 American Enterprise Institute for Public Policy Research