Information, risk & cigarette smoking

A major plank in policy efforts to reduce cigarette smoking are public information-based policies. How should these policies be constructed to increase quitting? This is a tough issue.

Telling old people that there are benefits, at any age, to quitting (a truthful statement!) encourages them to delay quitting. It might even encourage younger people to initiate or continue smoking for a time because it might suggest they can they can make an early-enough quit that will limit adverse health effects.

On the other hand telling long-term smokers that the damage they have done is irreversible (generally a false statement) might encourage them to continue smoking on the grounds that stopping now will not advantage them.

This general issue of designing public information messages is taken up in:

Frank A. Sloan, V. Kerry Smith & Donald H. Taylor, The Smoking Puzzle: Information, Risk Perception and Choice, Harvard University Press, Cambridge, 2003.

I learnt much from this. The main message is that general information about the disease and mortality consequences of cigarette smoking is not getting through to many in the population of smokers. While some, such as W. Kip Viscusi (1992), (2003), claim that smokers overestimate risks, Sloan et al. claim that smokers tend to be over-optimistic about their longevity and future health if they quit late in life. In fact, while you can quit early enough to reduce your mortality risks you need to quit very early indeed (more than 16 years before your mid-fifties) to evade the disability consequences of smoking.

Older adult’s decisions to quit smoking seem to require personal experience with a serious health consequence of smoking, such as a heart attack. Less serious symptoms such as shortness of breath will not do it. It is, of course, disadvantageous to wait for serious health events. The trick for Sloan et al is to come up with an information substitute to a serious health event – the one they propose and test is to focus on the disability consequences of smoking among older smokers rather than mortality consequences.

The emphasis is on older smokers who quit in large numbers anyway – the authors do not have a lot to say about dissuading youth from initiating smoking.

The main data base Sloan et al use was a large sample of people born between 1931-1941 who were aged 51-61 in the first year of a survey conducted in 1992. The survey was repeated in 1994, 1996 and 1998. The study has a panel structure with the same individuals asked the same questions about, among other things, their smoking behaviour and their health status and expectations of future health.

The notion that smoking kills is old news to most people - almost no-one disbelieves it. Information may be accurate but not considered relevant to those who use it. But information about the impact of smoking on the quality of life of a smoking-related disease does impact – the claim is that an 11-15% reduction in smoking by older smokers is possible with this approach which is equivalent to a 50% increase in cigarette prices. The prospect of losing the ability to care for oneself is something that grabs people’s attention. Smokers need to be informed of the long-term consequences of disease and be warned of the consequences of quitting too late. In this sense anti-smoking messages substitute for real events such as heart attacks that evoke quitting behaviour.

This can be interpreted as reflecting the concern smokers have both with risk probabilities and the consequences of the risks or the context. Premature death is often viewed as an ‘easy passage’ and taking on a little more risk in this regard is resolved with a decision to continue smoking. The essential point is to emphasise that the transition would be to a lower quality of life because of smoking-induced disability rather than an 'easy passage'.

An interesting feature of the Sloan et al policy prescription is that it partially conflicts with Sloan et al. (2004) where they demonstrate relatively limited disability consequences of smoking – smoking seems to bring disabilities forward but not increase to increase their length.

How do people make smoking decisions? Economic (optimising) and psychological (heuristic) criteria are alternative approaches to determining how people make smoking initiation choices under uncertainty. Smokers start to smoke when young (in Australia, on average, at age 15.9 years), they then go through an ‘discovery phase’ where they learn of some bad effects of smoking, a ‘day of reckoning phase’ where they either quit or face the risk of a serious health problem and a phase where they die.

Rational optimising models that rely on economics suppose predetermined preferences, time constant exponential discount rates and accurate perceptions of future risks and costs.
Perceptions of risk will be subjective probabilities or beliefs and will depend on information the potential smoker is aware of from private and public sources. The objective probabilities of very harmful health consequences are neither minute nor very large. The hazards are however seen as controllable (a smoker can elect to quit) and the smoker may have little or no direct experience of the hazards. These factors can foster optimism biases although, that the hazards are not negligible, should suppress such biases (Weinstein (1989)).

In addition, people may face cognitive limits in their ability to process risk information. Sloan et al in fact provide evidence that heavy smokers do not process risks accurately but that the average smoker does. But smokers do not update their risk perceptions on the basis of general information and need a personalised message such as particularly salient health warnings.

Furthermore, preferences may not be exogenous as economists often suppose but endogenously determined by the process of consumption as in habit-formation and rational addiction models.

They may also be peer-group-determined though this should be reduced by advertising bans and limitations on depictions of smoking in movies. These peer group pressures will be enhanced among low intelligence individuals facing stress and depression.

Discount rates too may be endogenous and time-inconsistent (hyperbolic) rather than exponential.

Government Policy and Advertising. It is difficult to assess the effects of advertising on smoking because it is difficult to measure advertising exposure although individual level studies confirm a link (Lewit et al. (1981)). The Joe Camel ads increased the demand for Camel Cigarettes among youth from 0.5% to 32.8% so it seems difficult to argue with the proposition that marketing works, at least for youth. In that event bans would be effective in reducing smoking. What about the effectiveness of anti-smoking campaigns? The evidence here is mixed with tobacco control enthusiasts being much more confident of campaign successes than economists. Today’s smokers in their fifties started smoking when advertising promoted use and there were few anti-tobacco campaigns. It might take quite a while to see if current campaigns are effective though the rapid decline in adolescent and school-age smoking in Australia provides optimism that there may be significant effects.

Recognising health impacts of smoking. This book argues that people adjust their smoking behaviour on the basis of personalised messages regarding their health. These messages can take the form of health signals particularly in people's late middle age so the emphasis here is on those aged 51-64 years in the final survey 1998. Disability is definitely linked to smoking . Moreover disability effects occur in much the same way for ex smokers as current smokers unless the smokers quit while they are young – here taken to be more than 16 years prior. Ex smokers live longer but still suffer substantially higher disability that 'never' smokers. This is worth pointing out to smokers. One needs to quit early to avoid disability costs.

Risk Perceptions. This investigates how information affects risk perceptions by focusing on responses to the question of what probability individuals attach to living to age 75 and beyond. People generally quite accurately forecast their longevity but smokers and ex smokers are excessively optimistic about their life expectancy while non-smokers are overly pessimistic. Heavy smokers understate risks. The evidence is that health shocks condition forecasts.

People generally pay close attention to their own health and the health experiences of their blood relatives.

Health Shocks and Smoking Behaviour. In the US about 30% of smokers attempt to quit each year but 80% of these attempts fail within a month. Smokers often make 4-5 attempts to quit before they succeed for a sustained period of time and even then cessation is not 100% effective.

Quit rates increase strongly with age presumably because the present value of the costs of smoking increase and information about one’s own health accumulates.

Smoking and quitting decisions are linked to longevity expectations. Serious health complications and messages that mimic effects of such complications induce quitting as do price increases but minor adverse health signals and general health warnings do not. Many smokers quit gradually so that ‘cold turkey’ strategies may be a less effective quit strategy than 'cutting back'.

Personalised Health Messages. Warnings about disability rather than survival are most salient to smokers. Smokers believe you will die anyway but warnings about such things as emphysema are particularly salient among older smokers – these diseases are feared and disliked. This study used a focus group to study how longevity expectations are formed.
Current smokers are more optimistic about their health than former smokers but information warnings were particularly successful among those smokers who had good cognition. Moreover, the messages stayed in their minds for six months after they were administered. Reminders and reinforcing messages would strengthen this effect. People misunderstood the implications of ‘never too late to quit’ messages which they understood as an excuse to delay quitting.

Personalised health messages from physicians based on genetic or other information may, in the future, be a useful way of inducing quits among older smokers.

Longevity Expectations and Cigarette Demands. This tests the information strategy based on emphasising disability and finds it works for older smokers. Providing information about emphysema substitutes for the effect of substantial price hikes which cut into smoking. While prices may be the best triggers for inducing quits in younger smokers prices-cum-information policies might work better for older smokers.

Conclusions. A major issue is why people smoke given its obvious health implications. The key argument in this book is that the quit decision depends on the way information affects risk perceptions and how these jointly impact on smoking behaviour. The main argument is that public messages that relate to effects of smoking on disability among older people is a useful way of inducing quits. Information about risks of premature death seems less effective than information about the disability effects of such things as emphysema. You need to make it clear that smoking-induced death is not an ‘easy exit’. While this is a welcome insight it does not tell you anything about how to induce quits among young smokers - quits that would induce a much greater present value of benefits. Moreover, many of these older smokers will quit anyway.

This paper does however tie information campaigns to an area where marketing will have an impact. It is an innovative and welcome piece of research.

References

E.M. Lewit, D. Coate & M. Grossman, ‘The effects of government regulation on tobacco smoking’, Journal of Law and Economics, 24, 3, 1981, 545-569.

N.D. Weinstein, ‘Optimistic biases about personal risks’, Science, 246, 4935, 1989, 1232-1233.