Frank J. ChaloupkaUniversity of Illinois at Chicago

National Bureau of Economic Researchfjc@uic.edu

www.uic.edu/~fjc

Rosalie Liccardo PaculaThe RAND Corporation

National Bureau of Economic Research

Henry SafferKean University of New Jersey

National Bureau of Economic Research

Michael GrossmanCity University of New York

National Bureau of Economic Research

Funded by the Robert Wood Johnson Foundation,the National Cancer Institute, and the Centers for

Disease Control and Prevention

The Impact of Cigarette Advertising on Youth Smoking

Background

Youth smoking increasing throughoutmuch of the 1990s:

� 50% increase in smoking prevalenceamong 8th/10th grade students

� 20% rise among high school seniors

� 6,000 youth try a cigarette for the firsttime each day; half become regularsmokers

� up to half of all regular smokers will dieprematurely from smoking relateddiseases, with half of these deaths inmiddle age

Background

Cigarette advertising and promotion rising over time:

� 1996: $5.1 billion spent on cigaretteadvertising and promotion in US;continuation of upward trend of pastdecade

� shift over time away fromtraditional, image-orientedadvertising to other promotionalactivities

Background

Policy efforts increasingly focusingon youth, advertising and promotion:

� Numerous policies directed atreducing youth smoking; mostnotably the Synar amendmentand FDA rules related to youthaccess to tobacco products

HeadlineBackground

� FDA rules also include significant limitson cigarette advertising and promotion

� State and local governments restrictingadvertising and promotion

� Multi-state tobacco settlement includesban on billboard and transit cigaretteadvertising, elimination of advertising/promotion targeting youth, prohibitionon use of cartoon characters in ads, andother restrictions

Cigarette Advertising and Adult Smoking

Large number of econometric studies onthe impact of aggregate cigaretteadvertising expenditures on aggregatemeasures of smoking produce mixedevidence that suggests advertising haslittle or no impact on cigarette demand

� Saffer (1998) and other critics of econometric studies based on aggregate data argue that econometric methods are ill-suitedto examining the impact of advertising on demand

Cigarette Advertising and Adult Smoking

Smaller number of studies look at theimpact of a ban on cigarette advertisingin one or a few media on demand, againproducing mixed evidence

� partial bans unlikely to be effectivegiven potential for substitution away from banned media towards other marketing activities

A few recent studies, including Safferand Chaloupka (1999), conclude thatcomprehensive bans on advertising andpromotion will lead to significantreductions in smoking

Cigarette Advertising and Youth Smoking

Relatively small econometric literatureon the impact of cigarette advertising onyouth smoking:

• Lewit, Coate, and Grossman (1981):– data from 1966-1970, US Health

Examination Survey– measures of self-reported television

watching combined withinformation on cigarette advertisingand counteradvertising

– find evidence that televised pro-smoking cigarette advertisingsignificantly increased youthsmoking while anti-smoking adssignificantly reduced youth smoking

Cigarette Advertising and Youth Smoking

Beales (1996):

• 1990 California Tobacco survey data

• monthly magazine, newspaper,newspaper supplement, out-of-home,and special market advertising data for5 CA media markets provided by RJReynolds

• concludes advertising has no impact onyouth smoking

• numerous methodological problems

Cigarette Advertising and Youth Smoking

Pollay, et al. (1996):

• data on brand choice from severalsurveys of youth and adult smokerscombined with brand-specificadvertising expenditure data

• looks at impact of "share of advertisingvoice" on brand choice

• concludes that brand choice amongyounger smokers about 3 times moresensitive to advertising than amongolder smokers

Cigarette Advertising and Youth Smoking

Large non-economics literatureconcludes that: cigarette advertising iseffective in getting children's attention;youth recall cigarette ads; and strengthof interest is correlated with current oranticipated smoking behavior orsmoking initiation

– generally do not adequately accountfor the potential endogeneity betweenand interest in smoking and smokingbehavior

Cigarette Smoking/Socioeconomic/Demographic Data

1992, 1993, 1994 Monitoring the FutureSurveys of 8th, 10th, and 12th gradestudents

– 82,995 students, mostly ages 12-18years; restricted to those living inmetropolitan statistical areas giventhe availability of other data

– three indicators of cigarette smoking:

• smoking in past 30 days• average daily cigarette

consumption for smokers• ordered measure of smoking

frequency

Cigarette Smoking/Socioeconomic/Demographic Data

Wide variety of socioeconomic anddemographic information:

– race/ethnicity, gender, age, agesquared, family structure, religiosity,parents' education, average realweekly income, hours workedweekly, mother's work status,marital status, grade, and more

By special arrangement, identifiers foreach youth's county of residence wereadded to the survey data

Alternative measures of magazinecigarette and other tobacco advertisingexpenditures constructed from severalsources:

• Annual cigarette advertisingexpenditures for approximately 190leading magazines from LeadingNational Advertisers

• Annual magazine circulation, bymetropolitan statistical area for severalhundred magazines from the AuditBureau of Circulation

• Fraction of circulation accounted for byhouseholds with youth ages 12 to 17years from Mediamark Research, Inc.

Advertising Data

• Gross advertising expenditures• Advertising expenditures per capita• Advertising expenditures per household

• Total tobacco advertising• Cigarette advertising only

Advertising Data

Cigarette Prices

• Tobacco Institute

– state-level weighted average price perpack of 20 cigarettes

– price differential for youth living incounties within 25 miles of state withlower cigarette prices

– indicators for youth living nearMexico and Canada

Price and Tobacco-Related Policy Data

Tobacco Control Policies• Coalition on Smoking OR Health and

Office on Smoking and Health, Centersfor Disease Control and Prevention

– variety of state tobacco relatedpolicies including: restrictions onsmoking in various public places andprivate worksites, limits on youthaccess to tobacco products, indicatorfor states that earmark tobacco taxrevenues for anti-smoking campaign,smoker protection policies

NCI/ANR• local tobacco control policies related to

restrictions in smoking in publicplaces/private worksites and limits onyouth access to tobacco products

Price and Tobacco-Related Policy Data

• Cragg's Two-Part Model:– Probit estimates of smoking/smokeless

tobacco use prevalence equations– Least squares estimates of conditional

cigarette demand and frequency ofsmokeless tobacco use

• Ordered probit estimates for categoricalmeasure of smoking

• Multiple model specifications– multicollinearity vs. omitted variables

bias

• Subsamples defined by gender and race

• Standard errors unadjusted and adjustedto account for the clustered nature of theMTF samples

Estimation

Cigarette/Tobacco magazine advertising:

• Estimates mixed across measures ofadvertising and measures of cigarettesmoking:

– Gross advertising expenditures(tobacco, cigarette, and youthspecific) generally positive for allmeasures of smoking; rarelysignificant

– Per household or per capita measuresgenerally negative, with exception ofconditional demand equations

Results

• Somewhat more consistent findings forgender-specific population subgroups:

– Generally positive effects ofadvertising on smoking among youngwomen; often significant in equationsfor frequency of smoking andsmoking prevalence

– Mixed effects for advertising andsmoking among young men, withmost consistent finding a positiveeffect on conditional demand

• Mixed evidence for subgroups based onrace, with exception of generallyconsistent positive and significant effectson conditional demand

Results

Cigarette Prices:

• Negative and significant effects in allequations

• Evidence of border crossing in responseto interstate price differentials

• Significant differences by race andgender

Smoking Restrictions:

• Strong limits on smoking in publicplaces/private worksites significantlyreduce the probability of youthsmoking; little impact on conditionaldemand

Results

Limits on Youth Access to TobaccoProducts:

• Estimates suggest that comprehensiveset of policies limiting youth access totobacco can reduce youth smoking

Other Tobacco Related Policies:

• Cigarette Tax Earmarking:– Strong negative and significant

impact on both smoking prevalenceand cigarette consumption bysmokers

• Smoker protection legislation:– Unexpected positive effects on youth

smoking

Results

Estimates provide mixed evidenceconcerning the relationships betweencigarette advertising and youthsmoking; in general, supportive of apositive relationship betweenadvertising and smoking, but rarelysignificant

– Likely due to the less than idealmeasures of exposure to cigaretteadvertising

Summary/Future Plans

– Better measures:• More “youth oriented” measures

based on youth magazinereadership data from SimmonsMarketing Research

• More comprehensive measures ofcigarette and other tobaccoadvertising being collected inImpacTeen

Youth and young adult smoking verysensitive to price

Other tobacco control policiessignificantly reduce youth smoking,although magnitude of effects isrelatively small

Summary/Future Plans