Tobacco Documents Online

- 562 The Alleged Costs of Cigarette Smoking S. C. Littlechild, B.Com. Ph.D. Professor of Commerce Head of Department Department of Industrial Economics - and Business Studies " Faculty of Commerce and Social Science .T.he-IIniversity of Birmingham House of Representatives Bill No. HR4957, introduced by Mr. Waxman on November 12th 1981 contains on page 2 the following passage : !'(8) it is estimated that cigarette smoking related deaths and disabilities result in $25,800,000,000 annually in lost productivity to the United States economy and $13,600,000,000 in medical costs." As an applied economist with experience in cost-benefit analysis, I have been interested in this and related issues for several years. As I indicate below, there are numerous problems in making statements of the kind just quoted. In my experience, most of these statements are based upon inade- quate economic analysis and unreliable statistical data. As a result, they tend to be seriously misleading, therefore it seems worthwhile to set out briefly some of these diffi- - culties involved. 5 1. It is by no means estab: is a cause of illness and d< (6) of HR4957. Since this : shall not dwell on it furthf is that the reliability of < any greater than the reliab diagnoses. If the medical fects of smoking, either in it follows that there canno smoking. Indeed, if one ta statistically associated wi the cause of those illaesse tivity and medical costs pr smoking is zero. 2. Even if cigarette, smoki illness and premature death ution remains uncertain. T contributory factors associ and duration of any illness For example, personal habit as personality factors, inf certain illnesses, which it off work or spent under mec

:oking :hild, B.Com. Ph.D. Commerce -tment ' Industrial Economics ess Studies ,mmerce and Social Science :y of Birmingham introduced by i on page 2 the :rette smoking 3 result in st productivity id $13,600,000,000 1n cost-benef it and related issues there are numerous i just quoted. ire based upon In inade- :atistical data. isleading, therefore 3me of these diffi- 0 IF 563 -2- 1. It is by no means established that cigarette smoking is a cause of illness and death as claimed in items (1) to (6) of HR4957. Since this is not my area of expertise I shall not dwell on it further. But the point to establish is that the reliability of any cost calculations cannot be any greater than the reliability of the underlying medical diagnoses. If the medical experts cannot agree on the ef- fects of smoking, either in general or in specific instances, it follows that there cannot be agreement on the costs of smoking. Indeed, if one takes the view that smoking is statistically associated with certain illnesses but is not the cause of those illnesses, then the value of lost produc- tivity and medical costs properly attributable to cigarette smoking is zero. 2. Even if cigarette smoking is claimed to be a cause of - illness and premature death, the extent of any such contrib- ution remains uncertain. There are inevitably a number of contributory factors associated with the incurring, severity and duration of any illness and with.the timing of death. For example, personal habits of diet and exercise, as well as personality factors, influence the onset and severity of certain illnesses, which in turn affect the number of days off work or spent under medical care. Even if smoking were

564 -3- a contributing factor to illness and premature death, it would be quite inaccurate to attribute to smoking the whole of the associated costs. But exactly what proportion ought .to be attributed is unclear, partly because the extent of any contribution of smoking is uncertain, and partly because there is no basis in pure logic for attributing joint effects among the various possibly contributory factors. To illustrate, specific examples of this point are found in items (5) and (6) of HR4957, which claim that "women who take birth control pills and smoke are more likely to suffer a heart attack or stroke" and that "certain occupational haz- ards in conjunction with an individual's smoking increase substantially the risk of disease and death". The first dif- ficulty is an empirical/medical one : how far is it possible to identify and measure accurately the separate effects, if any, of pills and smoking, or occupational hazards and smoking ?~ 'The second difficulty is one of principle : how far should any combined effects of pills and smoking be attributed to smok- ing rather than to taking birth control pills ? If, to take a hypothetical case for illustrative purposes, it were estab- lished that taking a specific pill and smoking had no effect separately, but had a specific effect when combined, it would be quite arbitrary~ t, quent costs) to~smok. ~;. . 3. People engage in of which may result : reducing productivit3 recreational example; holing (cave explorir include travelling bs food, drink, physical on. All these activi ably of substantial m be regretted, it is r the whole of the`~stor be taken into account in these activities m Calculations of the p erally take no accoun In this respect, ciga treated differently, 4. Calculations of tj assume that if people caused by smoking the:

death, it ing the whole portion ought e extent of any ly because g joint effects S, : are found in ; "women'who kely to suffer :cupational haz- ing increase The first dif- is it possible ce effects, if rds and smoking ? as :)w "far 'should any ibuted"to smok- ? If, to take , it were estab- g had no effect mbined, it would a 565 -4- be quite arbitrary to attribute the effect (and its conse- quent costs) to smoking alone. 3. People engage in an enormous variety of activities some of which may result in accidents, illness or death, thereby reducing productivity and generating medical costs. Dramatic recreational examples include skiing, mountain-climbing, pot- holing (cave exploring), and skydiving; more mundane examples include travelling by car or plane, or over-indulgence in food, drink, physical exercise, sunbathing, nightlife and so on. All these activities generate costs in some cases presum- ably ably of substantial magnitude, but while these costs are to be regretted, it is recbgnised that they do not constitute the whole of the story. There are corresponding benefits to be taken into account, which in the eyes of those who engage in these activities more than compensate for the costs incurred. Calculations of the purported costs of cigarette smoking gen- erally take no account of the benefits perceived by smokers. In this respect, cigarette smoking is being singled out and treated differently, for reasons that are not made clear. 4. Calculations of the claimed costs of smoking frequently assume that if people were not affected by illnesses allegedly caused by smoking they would otherwise be perfectly healthy.

~ 566 -5- This, of course, is not the case : They are typically affected by a variety of illnesses. The relevant loss in productivity is thus related not to the total absence from work experienced by smokers, but to the incremental absence, if any, that can properly be attributed to smoking. This would be naturally much lower. 5. Similarly, studies often calculate the total medical costs incurred by smokers. This is incorrect : the appropriate cal- culation is the incremental cost, if any, attributable to smoking. In particular, it needs to be taken into account that nonsmokers who live longer tend to generate very substan- tial medical costs in old age. Indeed, several studies by economists sympathetic to the anti-smoking cause have concluded that the net medical cost of smoking is negative - that is, the lifetime costs of treating smokers are in fact less than the lifetime costs of treating non-smokers. 6. Finally, it might be thought that the costs of smoking are incurred not by the smoker but by the rest of society, so that smokers are effectively subsidised by non-smokers. This is far from the case. Any lost productivity due to absenteeism or ill-health is probably reflected in lower wages, and medical costs are frequently born by the patient (possibly in the form of higher insurance premiums). But any such costs are far outweighed by the heat is no doubt that, on t from smokers to the re For these six main rea of smoking are fraught have studied have gene It is possible that tb upon new and more acce of such evidence I am

ey are typically affected ant loss in productivity nce from work experienced sence, if any, that can his would be naturally e the total medical costs ct : the appropriate cal- any, attributable to be taken into account to generate very substan- d, several studies by :oking cause have concluded is negative - that is, 's are in fact less than iokers. : the costs of smoking are 3 rest of society, so that !ower wages, and medical ient (possibly in the But any such costs are far 567 -6- outweighed by the heavy taxes on tobacco products. There is no doubt that, on balance, there is a transfer of income from smokers to the rest of society. For these six main reasons, calculations of the claimed costs of smoking are fraught with difficulties, and those that I have studied have generally proved inadequate and misleading. It is possible that the figures embodied in HR4957 are based upon new and more acceptable calculations, but in the absence of such evidence I am inclined to view them with suspicion.