PWI P~ Montraal, JanuarY Zf, 1993 he, t, lott~~f, Ir IacratarP oi It~t~ ior flalth, Dlplxtpat of I~al~~ Piohrond laPI~, 19 mitahall, bndo~, aIU P1 Dsar lad~ SacratarY of btate, II: Tha pit Im~ I hava Ild to your Dlpartr~'S B~DQkl Id Certimal BIPoarch Division h letter ol eollrmp~s reqmdiq the cdober, 1992 report it prlparld under ~e spnisiPn of Cl~s Sae and riQaal Parsonage. I anclose a copy of that lottar for your caavaaltnca. II coleats ooncantratt on thc It Roport's h·dl~ag of tho canadian judganant ~dia8 ~a tobacco Products control Ict (I~ql), as oy clint and I feel laa duty to maks Irl that its rasults arL PI~1J understoodl In briai, I have arquad a) that 111 four Canadian jodgan who bavl ason thn avidanec Pnnsantad to Ir~b tha aa ~~w Ulot it doa mt show mi probbiliB that a ban of tobacco ndvortioing Vill bfinq IM any reduction in tha eonsupption oi tobacco produc'cs, b) that ths Loc Roport misunderstands ths trus LpLet cf tha TPCA in its larll ysars, so leh a that its lagrassion analysis (ei yationlbll valua ncn withoot BATCo document for legal Seniees : Health Canada 21 ray 1999 Page 2~ this problem) and the conclusions of that analysis cannot be taken at face value, and c) that there is no teal reason to jug to the conclusion that any goad will come from dictating what people are alloved to see regarding a product Vley are allowed to buy~ Yours truly, Simon V~ Potter BATCo document for Legal Services : Health Canada 21 May 1999 gw) ~·~ bntrall, January tl, 1993 t, Llu~, Pconolios I operational Pesearob PMsica DPP~(Dt 01 1111~ Root 1111 llillbank ralr tl-n~ ~1~J b~gD I1P IpO Dear ~~ Leanay, PI: Tta Slee Ranort I offer the following collents regarding the october, 1PSt report of the Economics ~nd operational Research Division of the Dcpa~ent oi Health, under the sg~isioP oi Cl~e Blee and Michsl parsonage, It is usually referred to as the Slee Report and I will call it tbat in this letter, The Blae Repcrt deals at sets length xith variol evidence as to the likelihood that a ban of tobacco ~~ising night bring about a reduction in the consumption of tobacco products, I have teen involved, fcr 1 client Iaperial Tohleco Pd, ai canada, lengthy trial bavinq to do vi~ just such 1 bastion and believe I can usefully conat on the Slee Report from the vantage paint of the pl8bac Superior Court's finding on the facts, recently confirmed by the $Ibse Court of Appeal, that there is no such likelihood but only a ~palat~l possibility. I hope I' will understand nI concentrating oa the ~a Repert'e handling o~ this canadian jndqenent, tegdlng the Tebacce products Control Act as my client and I feel soae duty to 1~~ sure that its results are properly mder9tod, mat said, p BMCo document lOr Legal Services : Health Canada 21 May 1999 Rage 2, vill nee below that i cannot kaap from commenting on other parts of the Smee Report as wall, rn its Intmduction, on page i, the Smee Report refers to a Canadian "cffJcial raview' of the issue of the influence of tobacco advertising on smoking and explains in a footnote that this is the 'appeal against the Tobacco Products ControlAct" ("nCA")~ It is not ~plite cowect to refer to the judgement as an "official review", or to refer tr,an appeal against the Act · the appeal is against the judgement which round the Act to he unconstit3ti~nal. At paragraph 62, on pale 19, the Smee Report is similarly technically incorrect, refer~ing again to the judgement already rendered as an "official reviev", and as a judgement of t~e pu8bec court ol Appeal tatter than of the puQec superiPr Court. It would be more correct to speak of a constitutional challenge of that Act, which was successful in first instance on both of the grounds which it advanced, resulting in a judge~ent of the quebec Superior Court declaring the Act invalid because (among other reasons) the gover~ent had failed to show that the ad ban contained in the Act would have an~'effect on consumption of tobacco~ That judgement was taken to appBal by the Attorney General of canada, who asked the Court of Appeal to revere the lowe: court on the facts and on the law. Tbs Court of Appeal's judgemen~ was delivered on Januarl 15, 1993. It reverses the lover court's judgement exclusively on points of lau, it does not interfere at all with the finding that the Attorney General was unable, during a trial which lasted many, many weeks, to establish any prcbability that a ban of tobacco advertising would bring about any reduction in tobacco cansumptidn, It simply holds that the attorney General should not be held to such a standard of proof but need only show a pcssibilitY of such a link. I might add that the dissenting judge on the three-judge panel of the Court of $peal found that there had not even been proof of the possibilitY~ That is, all four judges who have looked at the evidence brought by the Attcmey Cenual of Canada in the TPCA Ease have come to the conclusion that it establishes no probability that tobacco ad bans will have the effect held out for them by their proponents ~ ~nd that is the case far any ad ban, let alone the ad ban brought to Canada by the TPCA, At page 19, the Smee Report contends that the TPeA's ad ban MS "ti9ht, wide-ranging and largely im~ediate" (paragraph 63, page 19)1 This is clearly wrong and this error has a clear effect on the conclusions which iollaw, The Act came into force on January I, 1989, Its only immediate effect was on canadian advertising of canadian tobacco products in canadian periodicals, It has had and can have no BATCo document for Legal Services : Health Canada 21 May 1999 page 3, effect on any advertising seen in foreign periodicals, specifically exempted by the Act though that advertising represented a very large percentage of all advertising seen by Canadians in periodicals. It did not de away with billboard advertising until the beginning of 1491. In-store advertising was not set to disappear until January i, 1993, and that disappearance was postponed two weeks by the QuBbec Court of Appeal's decision last December that it would be unjust to require the dismantling of such advertising while that CoILrtls judgement on the validity of the Act is pending, That is, Canadians have seen a good deal of tobacco advertising since the advent of the Act and will, even were the Act valid, continue to see a good deal of overflow advertising, This points to a good deal mere than the "cautious interpretation" which the Smee Report (paragraph 64 at page 20) calls for with regard to its conclusion that the TPCA has had an influence on the downward slide in tobacco consumption in Canada, calculated by the use of a "dummy variable" in a regression analysis. I note that the authors of the study chose not to use a variable for disposable income, even though the price of tobacco products doubled in real terms over the period studiedtt~, The analysis appears also not to have used any variable, subjective or otherwise, for the changing social attitudes toward smoking, legislation restricting the right to smoke in the workplace or in public areas, insurance premium reductions for non- smokers, growing awareness of the health risks associated with smoking, ever more striking health messages on tobacco packaging, greater health-consciousness of the population las seen in the trend away from alcohol and towards "liten foods), government campaigns to get people to stop smoking and to take up exercise, and so on and so on. One is left to wonder whether the "dummy variablen would have picked up anything ii it had been used for any year previous to 1989, The Smee regression used it only for 1989 and 1990, thereby necessarily picking up an effect beginning in the very first year of the Act's life. Even then, the effect which the "d~mnny" picks up is assumed to be the effect of the Act. In all logic, it can only be the effect of those things other than the explanatory variables the Smee regression did use (nothinq more, apparently, than price, a time trend and a lagged dependent variable), That is, the "d~ny" includes all those factors mentioned two paragraphs above and every other conceivable tactor. And still it does not register for more than l,f~, As suggested above, one of these other factors might reasonably be supposed to be the fact that smokers feel more and more alone in their habit. During the first 1? of the 19 years reviewed by the Smee regression, the incidence of smoking in Canada BATCo document for Legal Services : Health Canada 21 May 1999 Page i. declined from 161 of the population in 1973 to just over 30t without any legislated restriction of any kind on tobacco advertising; it stands today at 30,4t, (This decline was faster in the younger age groups, those said by the ad ban proponents to be particularly susceptible to advertising,) Furthermore, the Smee regression used a nlagged dependent variable", to take into account the lagged effects of the variables it was considering, This is reasonable, considering that most ad ban proponents maintain that: an ad ban cannot have an immediate efiect on consumption but only a delayed effect, What is not reasonable is to jump to the conclusion, without any supporting evidence or arg~rment, that the ad ban did have an effect, immediate, in the last two years of the period studied (1989 and 1990) which were the only years in which the Act had any effect at all (the tobacco industry has complied with the Act even while arguing its invalidity), even though the ad ban was partial to say the least In 1989 and only slightly broader in 1990, not even scheduled to apply to all Canadian advertising tlet alone the permanently excluded overflow from the Ilnited States) until 1993, What is perhaps most problematic for the Smee regression, however, is the data base used for the consumption tigures, The data is not disclosed in the Smee Report or any of its annexes, so it is impossible to say with certainty whether the data used, incorporating official statistics and "an estimate" (paraqraph 63 at page 19) to cover smuggling and cross·border shopping, approximate reality, 1989 and 1990, because of the massive price increases to which Canadian taxing authorities subjected Canadian smokers, marked the beginning of a huge influx of smuggled cigarettes into the Canadian market, either American cigarettes or exported Canadian cigarettes which can still be had much cheaper just across the border (a very short drive for the majority of Canadians), By 1991, it has been estimated that a good 25~ of the cigarettes consumed in Canada were cigarettes which had escaped the taxlan; some of this perhaps escaped the statistics used by the Smee Report to measure Canadian consumption, Even if it did not, however, the remarkable increase in price in recent years and the remarkable increase in black market sales of cigarettes in Canada in recent years must throw into doubt the figures plugged into the Smee recession and the conclusions drawn from that regression, it is, in short, at least arguable that, i~ 1989 and 1990, the Hmissing" eons~n~ption is cfuite large enough to throw considerable doubt an the 1l7t yearly decline linked by the Smea Report to the ad ban for 1989 and the 1~11 found for 1990. Furthermore, so dramatic have been the recent Canadian price increases and other changes that a straight-line progression from the seventies through the eighties and into the nineties is simply not a reasonable hypothesis, Indeed, one could very reasonably postulate an inverse relationship between the ad ban, such as it was, and consumption, BATCo document for Legal Services : Health Canada 21 May 1999 Page 5. In this light it f find it difficult to understand that the Smee Report snauld conclude (peaqraph 691ii on page 2t) that "the current evidence available on these four countries (including Canada) indicates a significant effect" for tobacco advertising bans. Indeed, it strikes me as simply wrong for the Smee Report to go on to say that in "each case (that ist including Canada) the banning of advertising was followed by a tall in smoking on a scale which cannot reasonably be attributed to other fadcors"~ The three other countries acntioned in the Lee Report are Horuay, Finland and New Zealand, lorway and Finland were considered in the TPCR ease, All four judges (that is, the Superior Court judge and all three of the Court of Appeal judges) point out their view that the statistics from these countries do not support the idea that there is any probability that a ban of tobacco advertising will bring about a s~aduction in tobacco advertising~ PerhaPs this is because the rate of decline in tobacco consumption in Norvay was just as high before the ad ban there as afterwards . As to Finland, recent statistics there, which I believe you must have but which I am more than glad to send you if that is needed, indicate an increase, since the ad ban, in the incidence of smoking a~ong yaunq people, partieulaly young girls. As to New Zealsnd, that country's ad ban is based on the results of the report of its gover~ent's Toxic Substances Board" The Canadian judgement is quite scathing as to the methodological flaws of this report and labels it as "vithcut probative values. None of the Court of Appeal judges interferes in any way with this finding but, on the contrary, they all explicitly leave it in place, At paragraphs 19 and 40 an page 111 the Smee Report refers to cross-national studies, partieularly those of Lugesen et al. 11991) and of the ~ew Zealand Department of Health (1989], the former having used the latter's ten-point rating of various country's tobacco advertising restrictions· In the Canadian TPCA case, that New Zealand Report was heavily criticised; indeed, it was found to be of no probative value. The ten-point system it proposed was also heavily criticised, indeed even by the canadian goverrnnent's expert witness called upon to present the results of that study, as well as the results of his own study (also declared by the trial judge to be of no probatlve value). The expert did his own study because ha recognised the inescapable weaknesses of the New Zealand study, ~he Smee Report gives the very weak caution that the ten-point system is lacking because, for example, it gives equal weight to a ban of television advartisin4 of tobacco products as to a ban of such advertising in cinemas, Besides the conclusions of the Honourable Mr· Justice Chabot ct pubbec's Superior court, the Smee BATCo document for Legal Senrices : Health Canada 21 May 1999 Page 6~ Report might have pointed ,,t that the New Zealind study also is t~ery selective in the choice of periods trom which to calculate the supposed effects of various tobacco-related provisions, giving so much weight to what has happened in Iceland and Greece that, if they are excluded irom the study, its conclusions are thrown topSY- turvy. Ad ban proponents who are faced with this lack of evidence on which to justify dictating what we shall be allowed to sea often appeal to a visceral mistrust of advertising in general (though, of course, if we mistrust advertising, we are less likely to be led estray by it], arguing that advertising must have some effect on overall demand or the manufacturers would not spend money on it. Quite sensibly, at the end oi paragraph 11 on page i, the Smee Report concedes that 'all ec~apanies could lain from advertising even if there occurred no increase in sales," There is no need to theorise on this point~ This is exactly what has happened. Total sales of tobacco products in Canada have been declining not only on a per capita basis but in ahselute quantities tar several years, but the advertising has continued, This fits with the sworn testimony given in the TPeR ease by tobacco advertisers and by expert witnesses that the goal of the tobacco advertiser is to affect the share in a declining market; the advertiser will not waste scarce advertising budgets seeking to affect the fundamental decision whether or not to smoke, but will concentrate on the smoker's decision what brand to buy: gaining or keeping market share is much mere likely to earn a profit for the manuf acturer ~ Even an the question of market shares, one must be careful in jumping to conclusions. ligrceing that the studies regarding the influence of tobacco advertising on children really only go to young smokers' choice of brands and not to their eventual decision to smoke, the Smee Report (p'~agraph 16 at page 51 points to the survey by Raberts (1990) showing that brand shares among 11 to id year-aids are highest for the products which are the most heavily advertised. It would seem intuitively true that the brand share for any product, in any age group but particularly for 11 to la year-olds, would be highest for the brand getting the mast exposure~ All we can conclude Irom this is that, were that brand not advertised, it would lose share to the brands which are advertised, precisely what the advertisers have been telling us. It certainly does not te11 US that but for the advertising the young smokers would never have smoked. At paragraph 9 on page j, the Smee Report, postulating that it is p04sible that advertising plays a role in the recruiting of young smokers, points out that most smokers adopted the custom before they were twenty, with a minority having started before they were sixteen. Leaving aside the accuracy of these statistics, and leaving aside the conclusion to which they paint (that most smokers did not adopt the habit until they were past the age ct 16, BATCo document for Legal Services : Health Canada 21 May 1999 page 'I, whatever their exposure to advertising) and leaving aside the trite remark that nearly all customs and life-style habits are adopted before one is twenty, it should be pointed out that statistics such as these mask an important truth. The percentage of the total population (cf Canada or of the United Kingdom) who smoke has been declining markedly for several years; most often, the decline is fastest in the lower age·groups; this has been so all the while that these young people have been exposed to tobacco advertising. At paragraph 10 on page 3, we learn that 672 of men who have ever smoked have quit by the age of 60, That Is, even if advertising convinced absolutely all men to take up smoking, there would be twice as many ex·smokers as smokers at age 60, In the face of advertising, why is it that tvice as many leave smoking as keep to it! The statistics regarding quitting in the United States and in Canada are just as thought-provoking. Is it that the third who keep smoking are weaker than the two-thirds who do not! Is their weakness a particular sensitivity to advertising! Only if the answers to both these questions are yes can the ad ban proponents argue that an ad ban will help people to quit and, even then, they must argue that the ad ban is necessary to protect us from ourselves, from our own desires and foibles. I hope that these comments will he round useful and that, if any further information might be required regarding the Canadian proceedings dealing with the Tobacco Products Control Act, you will not hesitate to call upon me, Yours truly, Simon V, Potter cc Mrs. V, Bottomley, Jp, MP, Secretary of State for Health Depart~ent of Health Richmond House 79 Whitehall london SW1A 7NS BATCo document for Legal Services : Health Canada 21 May 1999 MCMASTER MEICHEN I~ltllrlll~ SOLICltOll ~loel.ltvlSGuI tew vm rlllrn MYIIINON)Sn january 251 1993 aa~st ~MHOH1: IlldIFI:l: I~X 1111111141E) mh SEllklt ~LIMU~LL un Mr~ K. Leaney, Economics bt Operational Research Division, Department of Health, Room 2813, MflIbank Tower, 21·24 Millbank, landon SWIP 4QU~ re: SmeeReport _ Our file: 46981~ Dear Mr. laney, I enclose I commentary dealing with certain aspe~ of the Smee report which I was ated to prepare by Rothmanr (UK) Limited I he been a pmctising lawyer since 19M), I am a mwber of be Bar of the provinces of Quebec and Ontario. Whw the Tobacco products Control Act ("TPCA") was enacted in Canada I was retained by the salidtoff of Rjfl·Macdonald ltd. to act as counsel in a court dhanenge to the validity of that statute an the ground, inter alia, that it was an unjustihed ~e~~rh'on of the right of freedom of Ecpression which is guaranteed by the Caradian Charter of Righ~ and Freedoms, legalaction was ta~cen and resulted in a trial of a number of legaland factual issue before Mr. I~ti~ jtadude babot of the ~ebK Superior Court, the equiva- lent of and legally successor to ~e En&Ush High Court. ~ne of the main issues was the efficacy of ad-bans enacted in other eaun~ies. After a trial which ksted for many weelrs Mr, justice C~bot found that the At~omey·Ceneral of Canada had not h~tmdu~ed any credible widwce in support of the proposition that a ban an adv~tising of tobacco would have any impact on over· ah consumption On this basis he found the law to be unmnsbitutiot~l. ~mu~~mmrw~sths~m rorxnno Ncarenar rmdVb v~lfflV~a HONG IZMC BATCo document for legal Services : Health Canada 21 May 1999 MC,1I1ASTER MEICHEN His decision has now been wemoned by the Quetrec Court of Appeal by a majority of Z to i. The Court of Appeal however, accepted entirely Mr. fustice C~aboPs ~nding of fact and his dedsbn was reversed solely on a paint af law. The commentary on the Smee report enclosed is written with particuhr reler· ence to the factual evidence introduced in the TPCA case and deals with those aspects of the report with which my involvement in the TPCA case Ls made me familiar, I be occasionally quoted extracts from documents which were introduced in evidence~ Some of these are voluminous and I have not attached them as annexes. U on renech~on you wish to have copies of any of them I would be pleased to oblige. Yours faithfully, ~l~s~COIIN K. IRW·JG I End. CKl/mk BATCo document for Legal Senrices : Health Canada 21 May 1999 COh~lE~TARY ON THE S~E REPORT Ey: CK, tying McMaster Meighen PREFACE In the comments which follow reference is made from time to time to the jud~ent of MT~ ~ustice jean jude Qlabot of the ~ebec Superior Court ~nded down on july 26, 1991 in which it was found t~t the Tobacco products Con~ol Ad ("TPCA"] which progressively banned all farms of tobacco advertising by Canadian manufacturers in Canada was unconsb:tutional because it violated the guarantee of freedom of expression found in the Canadian C~ter of Rights and Freedoms (the "C~rtw"). On ]anuarl 15, 1993 the Quebee Court of Appeal overturned thdt judgment and found t~t although the TPCA did constitute an infringement of the right of free~ dom of expression it was nevertheless a justifiable limitation under r 1 of the Caadian Charter which permits rights guaranteed by the Charter to be limited by laws which are reasonable and are "demonstrably jllstifiable in a free and dernocratic society: This conclusion was supported by 2 of the 3 judges who hoard the appeal (le8el and Rothman jjA), Mr, lustjce AndrO Brossard dissenting concluded that those parts of the TPCA which banned advertising were unccnstitub~onal Mt. justice C~bot found that no credible evidence had been advanced by the AttomeyCeneral of Canada in support of the proposition that an advertising ban would achieve its stated purpose of redudng tobacco connrmptioh All three judges of the Court of Appeal agreed entirely with this finding and the lower court judg· ment was reversed by the majority solely on a point of law, BATCo document for Legal Senrices : Health Canada 21 May 1999 a CANAtlA 1 ~ an 1 number of an lad d~(OPt omissions in the dcn of the Report daLI with ~ndl which tad to ~~Cg~ the scope and effect of the TP~A~ 2 ne TPCA is desoibed in general hi in rlmgnph 6L That paragraph notes that "lobacco adverbjing carried by ~~ Pg+n mntinuer to he san Ir before'. ~e Report does not note tLt these magatines account for approximately two thirds of YI mblbr drcu]ation in CaMda, it should also be pointed out tlut the TPCA Idlm o~n~ ail foreign rgdm, not just those h01 the Ulh· 1 paragraph al I~ notes ~t vs. broadcast media,'widlly followed in Canada",an suyd to the US. ban of l9n, without noting Lt Canadian broadast rnedia hdYC likCWi5e PDt earried tOLCCO Idt~Pp IhO 1~~ i. Il is Ibtd in pln· 8 ~t 'Coneessions Id aqb· nPWtbha$, be controls brought in b the Ad were 81~1 ~dtnnpg and $rb immediate: 5 1 tY were so it roY Id weight to the conclusion nached that the Id·bn has been a d~ad bbl in reducing tonnunplaP levels. Howwer, the fact is thdt the Klntrols were LklgefY ·t ~m~tL 01 The only immediate effect was to ban tobacco advertisingin news- papers and the periodical pnr n~ re I~e if any, bnnd adverbjing in naL BATCo document for Legal Services : Heaah Canada 21 ray 1999 of ahl advertising SPending in pdor 18P. (8] Point of sale adverdsing was hmd only with id on pnPg 1 1993. in the ~sn Prior to enactment of the TPCA apmte aLlq to Point d sa]e Id~rAhl (khdhg dina plpae to ~len bl lhetf lPace) rePlcrated aPprorimately 9~ olbhldpcnd for I~~Ycd~ld (I Thc"minor maal~ on pe~p' referred to in pl~gp~ 61 are far tim minor, a sPonsorshipl cable a before excePt ~i brand nama mtg no longer be md. SpomnYp using corporate names (e.g. Aym ltd:) eontinuc. 6 The data ad for the tegmdon ~·II is raid to came from 'dbl GnLa a~a' end r hhdc~n Bh b hit ~~ ~aoa aDbl md aoss-border prhs. The lhI data lid LI~~ for smuggling is at given it i I Its of mlan knowledge Lt Imophg of Canadian dylda bm & USA, ~r became endemic A god cammisrioned by Be Canadian Tobacco MlnufamLm Coundt h~ I ~~pm h of forensic accountantl, lid~~ Avsy, Y~orY, bh~L he, n~td dut by 1AZ Yf 9~ of In dg~a ~l~d i Canada ra Jmrpl~ Lb Canada or were purc~d Iply by Canadians ~iLg theUS,A, There dSnan do not aPPear in L m~~on Bps published by Be Clrudian ylaat The dasl of the algLI proYa is illushated by the assignment of substantial Police raonrca, Previously engagd in dealing with narcotics, in an attempt, a Id nnruccessful, to stsn tbe flood Of Canadian dgaretta fram de lS.A. BATCo document for Legal Services : Health Canada 21 May 1999 7. The Smee report does not refer to a study published in Canada in 1991 by Makowsky and W~tehead which measured the effect of the withdrawal in 1981 of a long·standing ban on adverEsing for alcoholic bwvaga in the Province of Saskatchewan. T~t study is attached as Annex 1 to this Paper, 8. As the authors point out, plwious econometric studier, including many teler· red to in the Smee report, suffer from the tandjgp that they were u~ble to measure the effect of advertising on a population which pmiourjlY t~d not been exposed to it, The repeal of the advertising ban in Sdskatchewan thus provided a unique oppor- tunity to test the proposition that over time be presence of advertising would lead to an inwase in total conrumph'oh 9, As indicated in the study, the conclusion reached was t~t advertising had an effect over time on consumer choices within the pmdud liold but did not lead to any in~ease in overall eolmtmption of alcoholic beverages. 10. A final note ronteming the conclusions rerpPctingC~Mda has wider impiica· tiOFS. 11~ The regression equation used to reach the conclusion that the ad·ban has' accelerated the dedine in Gnadian ronsumption fails to take account of two major facfors, In the first place there is no income Mriable although disposable income in Canada was shrinking in the relevant period, 12, Secondly, and of even greater si~ificance, no account is taken of health cdn· cerns and chan~ng lifestyles~ Any regression ejwh'on which fails to take into BATCo document far Legal Services : Health Canada 21 May 1999 account changing sadal attitude with respect to tobacco products runs the risk of attribu~jlg to an advertising ban an effect which is in fact unrelated. 13. Advertising bans, including that imposed in Canada, began to be enacted only after consumption trends had begun to decline. One of the prindpaI witnesses alled by the Government of Canada in the TPCA ase was Dr.~effrry Ha~ who is both an econometridan and a medial doctor and who has testified on numerous ocms~ ions in lawsuits agairst tobacco manufactwrs in the United States. Dr, Har~ prc~ duced consumption figurer for some 22 countries for the period 1960 to 1985. He told the court that the figure mme from Dr. Laugessw, one of the authors of the TSB Report, but that in one or two csss he had substituted otter figures which he considered to be more re~iabfe~ The consumption figures a produced to the court by Dr Harris are attached as Annex 1 They show, as Dr.~Iarris acknowledged, that in general the consumption trend fat the period 1960 to 1975 was upwards and that with rare exception tobacco consumption began to tall in the mid-197~5, It was at that point t~t advertising bans began to appear. In short the polifi91 impetus to ban tobacco advertising gained strength as the popularity of tobacco products dec· lined. Studies attempting to measure the effect of an advertising ban in a period in which consumption is almost universally in decline but without talcing into account the factors which originally triggered that dedine and Dated the political climate which made the ban possible are unlikely to produce meaningfulrerult~, ~Ihe point is underscored by the fact t~t the ~eatest declines in consumption bye been seen in countries such as the United Ki~dom, the Netherlands, Austalia, New lealand, Canada and the United States which at the relevant times did not ban tobacco adver· bjing. t3AfCo document for legal Senrices : Health Canada 21 May 1999 nr NORWAY AND%INLAND 14~ As in the an of Canada, the actual data used in the Smee report on Norway is not provided. It is stated in paragraph n ttult consumption data was provided by Nonva~s National foundl on Tobacco and Health, It is pointed wt in t~t same paragraph that "It is desirable to take a starting date a few years before the chnges under study to give a good run st the problem. In this case 196~ was chosen.' In respect of Finland, however, a noted in paragraph 61, the Pekurinen ca~umption s~dy on wtuch the authors rely uses 1976, the yat before the ban, as its starti~ point, 15. It is noted with respect to Norway that t view of the importance of young people in smoking recn~ihnent it would be desirable to investigate the effect of the Ad on their rmo~i~g prevalence". Ip. II, ioahate I). Although not mentioned in the Smn report this was in fed done in respect of Finland in a report enb~tled "ChanRes in health habits of v6unapeaple in Finiand in 19'77 to 198i" by Messrs, Rimpela et at. A copy of the Engi~h summarl of that report is attached as ARnew 3. its findings are significant in a number of ~spectr. The loUowing at pager 1 and 2 is particularly pertinent: Smoking In 19f3 52$ of the 14 to 18 y~nr·clds smolred daily, ARer this the proporb~on of daily smolrers dropped tr~ plendously and was in February l9fl, just before the implementation of the Tobacco Ad 1696 (table 1; fig 2, p Is), The dwease in smoking coddpued until 1979 after which the trend levelled off, Smoking in~eased again after 1985 and was in 178'7 at the same Iwel as just before the implementafion of the An The inaease was BATCo document for Legal Services : Health Canada 21 Nlay 1999 s~angat among the 16 and 18 Year-old ~ils. However, among the youngest age group (12 ys) the smoking ex- perimentr did not in~ease after 1985, and the proportion of bily smokers (appr, I%) did not in~ase. In all years the boys smoked somewhat more often t~n the girls (t~ble i). 16. ~he mndusion reached is that The ~sult~ on smoldng show that Finland Ls not been suaesstul in redudng young people rmoldng any more in the r98(Ys. How· ever, the results do not implicate that the Tobacco Act has been useless and failed totally'. A number 31 examples of the usefulness elf the Act in the view of the mth os are a out in pp, 9 and 10~ 17. Tnriaairhlthededineinnnaiunginf~.~~ndwaJnPgnatniolloH·~lglr ban than in the years preceding it In addition, it appears dearly tt~t the advertising ban was not iurrasiui in ledudng Emolting inibation among young people. Is. in dea~ng with Nanvly it Should be pointed out that Dr.Kiell B~eit, of the Nab~onal Coundl on Smairing and Health was designated as an apert witness by the AttorneyCenenai of CaMda prior to the TPCA trial As ~equLed by the auehec rules blmud,, Dr. Bjirt~eil prepared and fiid an apart report although in ~e went bis evidence was not Q]led by the Aaomey·Cenpal for unstated reasons. 19. In the report Dr. B~rtvdt included charts showing the pnrmtag of daily smokers in Norway hom 1950 to 1981 and a iurtha chad showing sales al nunufic· tured dgarettes and smoking tobacco per capio~ aged 15 plus for Canal and Norway for the period 1950 to 1981. These hrts, which were sgurer 7 and B in his expert report, are attached as ~ulexes 4 and 5. BATC6 document for Legal Services : Health Canada 21 May 1999 20. It will be noted from fig~ f tf~t with respect to IF~ler the percentage of the Nonvegian population who wee daily smokers declined steadily Iron 1956 to 1984. ~he percentage of female smo~ea increased from 1956 to 1974 and continued to innease thereafter although at a lesser rate, 21, Figure B which, in the Q~n of Norway, is based on data from the MrKtorate of Customs and BdselCentnl Bureau ol5tatistirs shows sb olmanufactured dga· ntter and smoking tobacco dedining rather s~rpiy in Norway in the lew years prior to 1975, rising to a level in ares of 1975 by 1979, and then falling off dnmalcally from 1979 to 1982. levels since have inaeased slightly. k appears fmm Annec 5 tobacco prices in Norway increased 15% in 1981 and a further 11% in 1982, Dr, BjartveiYs statisb:cs provide no support for the mnl~ntion that the Nonve~n ad·ban had any measurable effect. ~e motrast which his chart shows between bends in Canada in the years prior to the TPCA and NoNlay in those following the ad-ban in that rountrj is in~tn~cb~ve. rv moENa FROM SURYFIS OFREASONS FCIR S~AR~INC SMOWJG 22, The comments in the Smb report under (hir,hesding an based on a recent OW survey of secondary schoolchildren and the A]exandR rtudy from Aus~alil in 1983, 23. B should be noted t~t a n~mber of Canadian gPvenannt studies Lye add· ressed t~ds isrue and that it wasl najor subject in the 1979 Surgeon Genenl'l report in the United Sates. BATCo document for Legal Services : Health Canada 21 May 1949 24. The "Summarl report on alcohol, tobacco and marijuana use and forms among young people in ~aMdd, yeat 4" (1986) concluded t~t: *Of those who are daily smokers, 22% state tt~t they tried their first cigarette because their parents smoked, 136 stated that they did because their siblings smoked, and 51$ tried their first dgarette because their friends smoked". 25. The Canadian govemment'Active Healh Report" of 1985 concluded t~t: "U none of a person's friends smokes, the e~snces are ]e~ t~n one in ten tt~t he or she will smoke. if most or all of a person's hiends smoke the dances are siw in ten that he or she will smoke. in total, a person with smoking friends is seven times more likely to smoke than a person with no friends who smoke." 26. This reflects findings reported in the 19?9 Surgeon General's Report. "SmoGing appears to be initiated as a result of social influences or, more particularly, the imitation of models such as peers, media stereotyper and signifiant adults (e.g, parents and teachers)." Chapter I?, p. ii, ElATCo document for Legal Services : Health Canada 21 May 1999 *Parents who smoke dearly influence the smoking beha· viour of their children. In families where both parents smoke, 222% of the boys and 20,71b of the girls ale also smokers, compared to 1TJ% and 7,61b wht:e neither parent smokes." Ibid,, pt 13 "Peer pressure is widely asnuned to be a si~Qnt cau· sal ladot in the initiaiicn of smoking. Ihe strong influ· cnce of pew group pressures is generally evident in young adolescents (38, A, but be precise reiah~owhip oisucs pre~ue to be initiation of smoking is more dif· ficult to establish.,. In a longitudinal study of CaMdian school ehildren, Matthews (51) finds t~t peer influence was a major fac· tar in the inib~tion of smoking in the population sur· veyed. The influence of peers seems to come from "best frjend* Rla~onships, rather than from large or divmi- fled group pressure, In a multiva~te study of rarrela· tive factors in youth dgarette smoking levitt and Edwards (50) report t~t having a best friend or group of friends who smoke appears to be the best predidor of snaking in children from the 5th through the 12tb grade. Bynnet(l3) finds the most important variable in explairdng smoking be~viour in English and Welsh schoolboys is the number of their friends who smoke, Williams (BB) reviews a substanb~l number of studies which also condude t~t pressures from peers and best friends are important influences to smoke." Ibid, pp, 14·15 "In a Task Force Report on Respiratory Diseases, the National Lstitutes of Health ~O) state t~t mass media t~ve been used extensively in antismoking efforts, but exactly how they influence be~vior is unclear, Ward [8n reports that in a sh~dy designed to asce~in attl· tudes toward television eommerdals and to a~lyte the BAT~o document for legal Services : Health Canada 21 May 1999 effects of telwi~ion ~dvert~ng on adol#cents, the tele vision medium appears to influence the formation of ideal and attitudes, yet doer not "~gger" adalesrents to buy a product. Ward'l study indiraler that dgretto ads are perceived by teenagers as hypo~ititai and are listed as 'least·likcd' while antismoking ads Ire pe~ceived as Is~aightfonvard" and are liked. The e~lfeas of mewgs in other media, such as billboards, magadnes and dis- plays need to be mon p~eo~$ studied. Mendelxlhn (54) concludes that, in general, current mass media efforts to educate the public concerning health issues are disappointing. It is possible that because of ragnib~ve and sodal differences in various development stags of children and adolescents, mass communimtionr may not be the most Ippmp~te means to reach children and adolescents with smafing deterrence message~. Moro spedfically, targeted communicadons might be better presented in selected target si~uations~' ~, p· 15 27. Ihis latter candurion is a~oled int publi~Jhrd rpeech given by Mt. lame Mirl2, the aifidai in dLuge of the Cdnad$n govrmnlent anti3mDlung ptognm 'B~k Fne" whiQ utili~ed television to deliver an anii·lmoking me~agt to adols carts. He said: 9t i5 important to W~e that wdat marke~dng pmgtarm, partiru· ]arly in mass media campaigns, QMot tearj0naMy be qected to reduce smoking i"9 group". And in another pasrage front the same speech: 'It is adnbwledged Bdt sodal marketing progran~ alone an ir~suffi· dot by themselves to alter be~viour," BATCo document for Legal Services : Health Canada 21 May 1999 EVIDENCE ON R_EACT[ONS TOADVERTIShlE~ITS is. ~he 5ma report refers in paragraph 1J to an arb~de published in ~AMA by DiFrantd et al 11991) and state in that connection; ·n~e paper quotes extendve evidence pi tobacco Klmpnies'approach to marketink some of it from the TPCA appeal in Ca~b, which sug· gests tt~t campaig~ of his kind are ~ypially aimed at capturing he early teenage market and may also be intended to encourage more people in this age group to take up smoking." 29, Although the Sms rPpod gos on to say Lt the ddente on mnsumptioo is not ndident to 6labiirh that the campaign does artually inwx s;nolung, it should nevertheless be noted that the article in question is iFaecuate in the extreme in dealing with evidence hom the TPCA acb~on, 30. The article refers spedfically to a number ofmarkehng E~alegy document ior Export A, a brand manufactured and sold in Canada by R~R·Matdonaid The authors dbjrrtivity in dealing with there donunenls may be ill~itrated by the bllawing example: 31, The a~b~de stall (p. 3152, eel, 2): ~wo marketing strategy doaunents for Export A also nvPal that it is the youngest children they are Iher, 'Whose behaviour are we tying to affectt Na users.' 'Ihe goal is'optimiring pmdud and user iaug· ery of Export 'A' against young starter smokers.' The average age for starter smokers is 13 years" BAT~o document for Legal Services : Health Canada 21 May 1999 32. This apparently damning dtation is in fact an amalgam of three separate saur· ces, Conhary to what is stated by the authors only one is an ~po~ A *mar~etink strategy document", The actual text of that document is the following: Whose behaviour are we trying to affect! Demo~aphics: adults· males (predominantly) females · (must not be excluded) 18 to 34; emphasis Is to 24 (new usws)'~ 33. nne goal of "opb'sliring product and user imagery of Export A agai~t young s~rtw smokers" is taken from a proposal from an advertising agencl~ 'IS~t proposal was rejected by RJR·MacdoMld as expressly stated in the very document from which the first part of the quotation was taken It states: Since we lannot direct our media or our aeative (sic) to starter smokers the optimal target group is young adult smokers between the ages of 13 to 24: 34. The last sentence "nte average age for starter smokers is 13 years' does not form part of either the company document or the agency proposal. It is taken from a book entitled "Selling Smoke" written by I. Tye, a well·known Amerian mf· tobacco advocate. 35~ In short the authors have altered the text of the one ~rkebhg document cnk tened, attributed a third-party proposal to RFMacdoMld, cmirjng to inform the reader that the proposal was rejected and have ~afted an additional sentence ~om a book written by a third party, leali~g the reader with the impression that it ema· Mtes from the company. This is nott~ng short of deliberate mwrepresentation. A roW of a letter from the Imdersigned to JAMA dealing with the article in qustion and IAMA's reply ate attached as Annex 6. ~o comment has been received from the authars~ BATCo document far Legal Services : Health Canada 21 May 1999 MCMASTER MEICHEIV Annex 6 84111l11111 EOLICltOl~ I#nrat.Uvt~lp~wrm rmRrm uanWOnb0r January 20, 1992 nll~Ht mV113SLlllm·llll rhX l)Lllll(g1 TILUB·IYIII ~c~wla~u~m· The Editor, Journal of the American Medical Association, 515 North State Street, Chicago, Illinois U,S.A. 60610, re: 909 Nabiscols Cartoon Camel promotes Camel cigarettes to children Dear Sits: The article noted above published In your December 11, 1991 edition makes extensive reference to documents pla- ced in evidence in court proceedings In Canada challenging the constitutional validity of a Canadian law banning the advertisement of tobacco products. I acted as counsel to RJR Macdonald ltd,, a subsidiary of R,J, Reynolds International Tobacco Corporation, in that case. As you may be aware, the (luebec Superior Court ruled last summer that the lan in question was Invalid because it infringed the Fight of freedom of expression and could not be justified as a reasonable limitation on that right, This latter point turned to a targe extent on the contention advanced by the government of Canada that adverti- sing has an effect on overall consumption of tobacco pro- ducts, The trial judge after hearing extensive evidence on this Issue concluded that: ".,.thls possibility goes no further than speculation and certainly does not rise to the level of a probability". ~JRUIID~rmldSOLltAm 10Rrm KWMrOnX v~h~eE~n~ Ha~d ~wl BATCo document for Legal Senices : Health Canada 21 May 1999 1_ __ MCMASTER MEICHEN JAMA Page 2 The decision has been appealed and the case will be heard in May 19qt by the Puebec Court of Appeal, I do not suggest that a pending Canadian case is a reason to avoid publishing articles on the important issues which are at stake, Surely however when a journal of your standing and integrity publishes comments on or Quotations from documentary evidence introduced in such 1 case it should ensure that they are fair and accurate, Regrettably the article in ouestion is inaccurate in the extreme in dealing with many of the documents and, what is worse, seems in a number of instances to be delibe- rately misleading, In their effort to demonstrate that the tobacco industry is attempting to induce children to smoke the authors have adopted the simple expedient of altering the context, and at times the actual text, of marketing documents expressly dealing with young adults to create the false Imp- ression that they deal with children, For example; 1, ihe authors state at p. 3152, eol, 1 that; 'The themes used in tobacco advertising that is targetted at children are the result of extensive research an children conducted by the tobacco industry,," and that this research identifies ",.,the major psychological vulnerabili- ties of children which can then be exploited by advertising...'. BATCo document for legal Services : Health Canada 21 May 1999 ·I _ _ C__ MCMASTER M~IGHEN page 3 1 The authors then immediately turn to an 'Ixport In marketing plan which Has introduced In evidence during the care, ~hr inercapab~e inference II that the E13ort I narke- ting plan is part of a schele to sell cigarettes to chil· dren, In fact, as even a casual readinp of that plan ~o~d disclose, the ape gro~P under consideration war 18 to 34 with emphasis on Is to 24, The document states expressly moreover that the target grouD for advertisements under cPn- sideration nere already rnokers, They were described as 'regu]ar length full flavour and full flavout line extension smokers seeking smo~tng sarisfaction and currently smoking bprf )Igen and OuHaurier brandr". rlayerr and auhaurier are bra ndr mar~ctcd In Canada by a c6mpetltor of RJR Haedonald, The aisertlon that the Export d marketing plan is part of lextenrive research on children' and seeks to exglolt thelr'ps~cholo~cal vuinerabilitIer" II pawntly f~ir~ dn examination of the evidence In the tanadian case would Ehan moreover that in every single instance the young· eit group Involved in advertising research carried out by RJR Macdonald Has 18 to 2a, Witnesser appearing for the govern· mcnt, who had In opportunity of examining virtually of such research carried cut over the last 15 years, were un4ble to Indicate a single exception, 2~ The arilcle states, at p, 3152, coi, 2: "Two marketing strategy documents for Export R also reveal that It Is the yPun- gest chi Idren they are af rer, 'Xhose behaviour are we trying to affect7 Nev users,' The goal Is 'aptlmltin9 product and user imagery of Export 'Aagainst BA~Co document for Legal Services : Heath Canada 21 May 1999 _·I_ ·_ ____ MCMAS'IER MEICHE~ JAMA 1 Page 4 I young starter smoker~~' The average age for starter smokers Is 13 years~ 1 This citation Is in fact an amalgam of three seDa- rate sources, Contrary to Hhat II stated by the authors only one is an Export A "aar~eting strategy document", ihe actual 1 text of that document is the following: 'Vhore behaviour are we trying to affect? I O~noprlDhici: 1 adults - males (predominaotly] females · (muSt not be excludedi~ I 18 to 341 emlbasii 18 to 24 Inew usersl", The goal of 'optlnillng product and user imagery of I Export A against young starter smokers' is taken from a pro- posal from an advertising agency, sly rejected by RJR-Maldonald In the very doeulent from uhich I the first part of the guotaticn was taken. It states: "Sinee we cannot direct our media or out r creative to starter smokers the optimal target group is young adult ' ween the ages of 18 to 24," r The final sentence of the citation, 'fhe average age for starter smokers F1 13 years,' is taken from a hook I Written by Or, Joe 8, Tye and has no eonnectlon whatever with RJR-Hacdonald marketing, 1 It is not necessarY to favour tobacco advertising to retognise that the fcregolng is nothin8 rate dirtortion· The authors have altered the text of the I one marketing document concerned, attributed a third party proporal to RJR-Macdonald and omitted to inform the reader LI~ BATCo document f6r Legal Services : Health Canada 21 May 1999 MCMASTER MEIGHE~ JAHA Page 5 that the proposal was rejected, i have great difficulty in believing that tactics of this rind can recommend themselves to JAHA or to your readers, In addition to outright distortions such as those mentioned above the treatment of other evidence in the Cana- dian case is far from objective, To give a single example, and there are many, the authors state, at p, 3152, that "In countries where advertising has been tdtally banned or severely restricted the percentage of young people who smoke has decreased mate rapidly than in countries where tobacco promotion has been less restricted,' The source for this assertion is stated to be the report of the New Zealand Toxic Substances Board ('TSB'), That report was relied upon by the Attorney General of canada in the court case as demonrttat~n~ that the greater the degree of restriction of advertising the greater the dec- line in consumption, This Is Indeed the thesis advanced by that report. Your readers may have been interested to learn, however, that the trial Judge who heard extensive evidence on the matter round that the data relied upon in the TSB report were inaccurate and that: O..,lt contains serious methodological errors and a lack of scientific rigor which renders it for all Intents and pur- poser devoid of any probatlve value", The tSB analysts in support of its conclusion that smoking prevalence had declined more rapidly where there were 9:eater advettislng restrictions has so flawed that Or, Jeffrey Harris, a nell·known scientist who has frequently appeared as a witness against the tobacco industry In the BA~Co document for Legal Senrices : Health Canada 21 May 1999 MCMAS*IER MEIGHE~ JAMA Page 6 U,ited Stater, and who apDeared in the Canadian case on be- half of the F~vernment, tes;ified that if any of hir students had prepared it he or she would have failed, It may be that your authors disagree nith the lon- elusion reached by the judge an the value of the TSa Report, No one would puertion their righl to do so, Would you OOt agree nonovet that a purportedly scholarlY artlcle should a! least have made its readers aware that a study relied upon far an unquallfi ed conllurloo W" iou"d to have been valuo- less on the only occasion on which it had been iuaject to serious scrutiny? The fore~oing are only the most obvious instances of mis-stateientr of source material and material omissions contained in ihe article, if you pllh to exan~lne any of the documents in the record of the Canadian case to verify the accuracy of your authors' treatment of them, nould be pleased to provide you with copies, The debate between there who favour government intervention to protect societY from influencer such as taba- cco advertising which they believe to be harmful and those who resist any such intenention raiser serious iirues and calls for a thoughtful and informed Dublic response. Tbe publitatlon of 50 nirleadln9 a, article does a great dirrer- vice to both sides of that debate, ]ou:"llr with Hell-earned ctedibilltY such as JRHR should not lend their reputations to advocacy disguised as reholarrhiD pd'licu~aily where the advocacY Is rO feeble ,I to retort to the tactics dirplllyed in this article, Yours faithfully, ~11-~9 COLIN K· IRVING CK~lmt BATCo document for Legal Services : Health Canada 21 May 1999 __ ICC~ -------- Ame~ea Medicll AssociatiOn C Fhnirilnl ~Yiald I rh, C~~~d~ 91 Lml d 6( Pi I;onh Llr Stnr( 3t (R~I dmoian~dd~WOCi~ljon CY9~1TIIQW Ut l~·P~b June i, 1992 Colin K, Trying Maastcr Meighen 630 Bene-tovesque Blvd, West 7th Floor kontrcal, Canada H3B lg7 Oear Mr Irvinp: Your letter vas fat too long tour strict limit is 500 uords) and arrived far too late to publish, I have sent it on to the authors to reply personally if they choose, Yours sincerely, Drrond Bcnnie Deputy Editor (West) (312) ~64-1121 DIR:rnvh cc: Joseph R, Dirranza, MD BATCo document for Legal Services : Heath Canada 11 May 1999 ~rRrg " 4EhL EFFECT OF TOBACCO ADVERTISING ON TOBACCO CONSUMPTION: A DISCUSSION DOCUMENT REVIEWING THE EVIDENCE ECONOMICS & OPERATIONAL RESEARCH DIVISION DEPARTMENTOFHEALIH BATCo document for Legal Services : Health Canada 21 May 1999 EFFECT dF TOBACCO ADVERTISING ON TOBACCO CONSUMPTION: A DISCUSSION DOCUMENT REVIEWING THE EVIDENCE This report reviews the evidence on the effects of tobacco advertising an tobacco consumption, including the effect of adve~sing bans, The report draws on the available published material in this field, both from the UK and internationally, together with some original econometric work. The report is being published as a discussion document. The Oeparunent of Health's Economies and Operational Research Division would welcome any comments on the overall analysis, methodology and original pieces of work in the report. Any comments should be received by Friday 29 January 1993, and should be addressed to; Mr K Leaney Economics and Operational Research Division Department of Health Room 2813 Millbank Tower 21·24 Millbank LONDON SWIP 4OU BATCo document for Legal Senrices : Health Canada 21 May 1999 ACKNOWLEDGMENTS This study was carried aut by eeonarnists in the EcPnamics and Operational Research Division of the Department al Health under the supervision of Clive Smee and Michasi Parsonage. Robert Anderson was primarily responsible far drafting the mainbody of the report and Annex A. Simeon Duckworth drafted Annex B as well as undertaking the analysis which it reports. The study benefited from commens made by medical and administrative colleagues within the DepaRment 01 Health and by tconomiJts in the Treasury; Customs and Excise; and the Depanmsnt ot~ade and industry. Thanks are due to the Tobaeed Advisory Council and Anion an Smoking and Health for contributing material and analysis which slrenglhened the body of evidence On which the report is based. Clive Smee Chief Economic Adviser October 1992 ii · o BATCo document for Legal Services : Health Canada 21 May 1999 CONTENTS Page Introduction . 1 Indirect and circumstantial evidence 2 Analysis of incentive structure in tobacco market 2 Mechanisms by which advertising might increase consumption 4 Evidence on reactions to advertisements 5 Evidence from surveys of reasons for starting smoking 6 Other evidence of indirect effects 7 (luantitative direct evidence 7 Cross·section analysis of countries with different levels of controls 10 Time series analysis of fluctuations in advertising expenditure within countries 12 Before·and·after studies of consumption in banning countries 16 Norway 17 Finland 18 Canada 19 New tea[and 20 SvmmaF~ and Conclusions 21 Annex A: Time series analysis of fluctuations in advertising expenditure within countries: further analysis 23 United Kingdom 23 United States 29 Other countries 33 ·iii· CJ BA~Co document for Legal Services : Health Canada 21 May 1999 INTRODUeT1ON 1, There has been much dispute about the influence of tobacco advertising on smoking, The issue has recently been the subject of official reviews in the United States', New Zealand' and Canada", Official reviews are currently under way in Sweden and Germany. Chapman (19861 and Gadfrey (1990) review the literature, This paper examinestheevidence an the effect of tobacco advertising, including the effect of advertising bans. 2, There is a wide range of studies in th$ area following a variety of methods. The studies cover knowledge of advertisements, reasons for sra~ng to smoke, opinions as to the effect of advertising on smoking behaviour and empirical work looking at assecia~ons between smoking behaviour and advertising expenditure over time or across countries. The Report of the VS Surgeon General shows the range of studies and gives references. The farms of evidence clearly differ in nature. We have leaked at the evidence from abroad as well as in this country and we have carn'ed out some analysis of our own where published studies do not exist. We have concentrated attention on studies which follow sound methods with a reasonable prospect of delivering reliable answers. The evidence falls into two broad categories: ial analysis at the incentive struclure in the tobacco market to determine whether it is likely that advertising will be undertaken to increase overall cdnsump~on; mlnuawocrolchnilMbvwhia~amrngn~htena.e~nMpti.: Ic) evidence on reactions to adve~semants, particularly among young people, on the assumption that there has to be knowledge of advertising if it is to have any effect on smoking; Idi surveys of reasons for takiq up smoking; and (e) evidence on the indirect eff acts of advertising, 1 Aepart by the US Surgeon General the Health Consequences of Smoking: US DHHS (1989), i Report by the Toxic Substances Board INt Department of Health teas). 'Appeal against the Tobacco Products Control Act ITPCA), BATCo document far Legal Services : Herilth Canada 21 May 1999 ~ht~ analysis Jm~ig bbi·n~~*a o·l·R on in lha bo ~I rsilonlh the 1III of adVanisiOg; LI varialiens in the level ct c·lle at dn~dlL end tl I ban on adva~lnC #aREC7 AIO ORCUYS~WIAL E#OEICE 3. The first step is to analyse the eh~a~~~~cs ~ the tobacco rnerkat to assess wheiher there exists an incentive a p~wi to undertake advertising with the aim of h~ea~ng overall eanrump~on I h cotlld prove profitable for a monopclistic ~m m edvanise ~ ad~ m expand the market bttlA move be dlmlnd eurve outwards). The amount worth spending depends on the cost and effectiveness of advertising. ~ m In industry with pan than ane supplier ii is much less likely to be :==~:T:~~' 3"611 ·e3 LI I~t ~ol· 131~p & it would pay ma alms b any industry a edbds to IcHavs the enacts of a mondpdh hdudng Idv~l~ng $ expand industry PCr as tobacco industry in this country is ~gMy ·C"'~d a figures smlrbtd by Boot n J ~99O ShOk. Four Rrms aeeauntfor BSY d Qmafe pPdclon 13 16~~0X or ohr The largest ~ng~ ·Rhr lu h, delrfF market has ~~ d o~L While ~ml of the pre·cond~ions for successful cdbion are present in me tobacco industry, is no evidence of cd~~ve arrangements to promote a monopoly. ~ However, firms create a degree af monopoly for their own Product by branding supported by advertising and ofeouae firms advertise brands not genedctobacco oroduots. as Yet 61 brand by~~ merely expresses the monopoly pouff ela brand. h may therefore be Prnfitable for a ~m to advertise to enco~age the uotake or smoking using the sdva~sad ~an~ since in that case the firm captures all the benefit of the increase h industry sales dueto its ~dvlRb~C a ~e industry states that at sds purpose of adve~bhg b (,, each firm to m~ntaio and H posible iocrease nabt Err I·SJI cenied mt ~hh m, industry confirms that advertising La bdad Inrt bind shares. Ho~·a~ ~ is quite posd#l that some people In recruited to srncking by brand advertising even I firms do not 9eso~ set nut a attract bra. Al we shellsea bb· ·2· BATCo document for Legal Services : Health Canada 21 May 1999 most Of these are likely to be teenagers. 9. It is worth boking at the nature of the product and smoking careers for the incentives they create for advertisers, Smoking is not a necessity of life, so that recruitment of new smokers is necessary to maintain the smoking population by replacing those who die orgive up, Advertising may thereforehave a role in promoting recruitment to this non·necessary product. Smokers typically start in their teens. The OFCS survey of smoking among sehoolchildren (OPCS (19891) and the 1988 General Household Survey (OPeS (1990al) show the following pattern of smoking prevalence with age: OCCASIONAL AND REGt]LAR SMOKERS' 1988 Age Males Females 11·12 4 1 13 10 9 14 15 19 15 24 31 16·19 28 28 In 1988 84% of smokers aged 16 or over had become regular smokers before their twentieth birthday; 45% of male and 392 of female smokers in the 16·34 age group started before the age of 16, 10, About two thirds of men who have ever smoked regularly have given up by the age of 60 ICHS 1988 table 5.31. Nevertheless, the typical smoking career lasts far many years. Applying current life·table methods and taking account of - the much higher death rate among smokers from 35 onwards, the typical length of a smoking career for a man of twenty who smokes regularly is just over twenty six years. And of course about half of twenty·year·old smokers already have a five·year smoking career behind them. The average smoking career for a woman of twenty is rather higher at about thirty·three years', In consequence, i age ll·t5: occasional (cunenl smokers usually smoking less than one cigarette a week) or regular tone or more cigarettes a week) smokers, age 16·19: answer of yes to 'Do you smoke cigarettes at all nowadaysl" 'Relative death rates for smokers are taken from the Cancer Prevention Study II, reported in the Surgeon General's Report VS OHHS (19891, pp150·151, The expected smoking career of twenty year old women is longer than men for three reasons: (aJ the relative mortality rates of female smokers to female non·smokers are lower than for males in CPS II; ibl women have lower mortality rates than men; and Ic) the rate of smoking cessation among women by age 60 is lower far men: 5446 against 661. ·3· 1:: BATCo document for Legal Services : Health Canada 21 May 1999 advertising to increasesmoking prevalence, particularly among young people, has the potential to deliversomething like thi~ years of consumption from each new consumer, 11, 70 same, the industry's opposition to a ban suggests that advertising does increase sales. if advertising doer not increase soles, then the indus$ dleedvely loses nothing from a ban and it gains the E100m or so it spends each year an advertising. There are strong counter·arguments. To begin with, there is the obvious point that advertising protects the branded home product against generic imports. Secondly, advertising may ofter the least costly way of increasing or maintaining market share. Thirdly, adve~sing allows me maintenance of brands with (real or imagined) characteristics which command a price premium over a generic product, In this way all companies could gain from advertising even if there occurred no increase in total sales. Ldb~ha~;cm. L,, ...L!.II I .I.... ~10c~1~1~31113 Vrw!llC~A(IVI~SIII M19htJncrease Cansumatinn 12. It may be worth hearing in mind the direct mehalismr by which advertising and promotion might increase tobacco eansumption IUS Surgeon General's Report cus DHHS (1989]): · by inducing children and young people to begin experimenting with tobacco products and in this way initiate regular smoking · by encouraging adults to take up smoking · by encouraging existing smokers to smoke more · by undermining exisu'ng smokers' motivation to give up - by encouraging former smokers to resume the habit, Advertising may else increase smoking indirectly: · the fact that tobacco advertising is allowed at ail may aeate a climate of social support lor smoking by suggesting that smoking eanna~ truly have dangerous effects or the government would ban tobacco advertising; · media reliant on tobacco advertising may restrain their discussion of the harmful effects of tobacco and in this way restrict the now of health education messages, ·4· C BATCo document far Legal Services : Health Canada 21 May 1999 ~anca an Reactions ~dvad8e~em 13, There is great concern about awareness of tobacco advertisements among children and young people, because many (45# of male and 39~ of female smokers in the 16·34 age groupl start smoking before the age of 16, mat is, before the age at which traders are allowed to sail them tobacco, In other words, may start when by implication they are not old enough to take decisions about smoking for themselves. mere is also concern that advertisers may be torgedng mesh age groups to reenk new smokers and doing it successfully, me svidance about awareness is beyond dispute, For example, Aitsn at al ~987) in be UK have shown that most primary school children are aware of tobacco advam~ng and many can supply ma brand names of cigarette advertisements from which this infcrmaticn has bean removed. mere is even evidence of awareness of cigarette brands among pre·sehool children (Fischer at al (199111, 14. mars is some circumstantial evidence mat advertising may increase smoking prevalence pa~~ulaly among children: findings tom a number of countries indicate Sat by and large smokers and Lose who later take up smoking find tobacco advertising more apaedve and have a more positive attitude towards it than non-smokers MeterP (1992)1. t 5, me Camel advertising campaign in me US using be Old Joe character has aroused particular concern on me score of its appeal to young peopie,its influence on their smoking patterns and what it reveals about the tobacco companies' advertising strategy and capabilities, An article by DiFranza at al (1991) shows that a high proportion of teenagers raeognbe the character, higher San among addts~ This campaign provides an example of the success of advertising in caltu~ng brand share, h has taken Camel from nowhere in the under age smoking market to a ons third share, it has unouastionably achieved greater success among children man among adults, me papal guotes extensive evidence of tobacco companies' approach to marketing, some of it from he TPCA appeal in Canada, which suggests that campaigns of this kind are typically aimed at capturing the early teenage market and may also be intended to encourage more people in this age group to take up smoking, However, me evidence on consumption is not sufficient to establish that the campaign does actually increase smoking: it merely relates to market share, 16, it is worth noting that in me UK a campaign of this kind making an appeal to children, whether inten9onlly or not, would not be Jlo~ed under me voluntary agreements between the government and the tobacco industry, more is nevertheless some indication let advertising influences market share among smokers below the age at which may can legally be sold cigarettes In me UK, A survey by Robsrts n 9901 shows mat brand shares among 11 to 14 year olds are highest for the products which are most heavily advertised, 17, A survey tom California reported by pierce at d ~991) confirms DiFranza's findings about the affact of sdven~ng on young pco~e, h shows mat perception of advertising is higher among young smokers (daflned as 12·18) than adults; BATCo document for legal Services : Healfh Canada 21 ray 1999 market share pagsms over a fouryear period followed perceived advertising patterns: and changes in market share'esulling from advertising occur mainly in young smokers. Again, however, the results show that advertising inlluences market share; the evidence provides no information about the inlluenc·5 on the Overall[evelcf tobacco consumption among young people. 18. Aitken st al (19861 argue that the positive attitudes smokers have towards tobacco advertising indicate that they are deriving some benefit Irom advertising, Probably perceiving it as a farm of sclcial approval of smoking. The facl that those who later become smokers share these views suggests that tobacco adve~sing may be playing some part in breaking down the resistance to Smoking based on awareness of the healthrisks. Tobacco sponsorship of sport in panicular may cultivate these positive attitudes by associating tobacco with charaQeristics which Young people admire. Aitken's survey suggests that the sporl·sponsbring brands Of cigarene are indeed associated in YOU"S people's minds with exciting sports (though nolnecessarily the Sport sponsored). However, there was no di~ferenca between smokers and nonsmokers in their awareness of brand sponsorship. fromSllrvaYs of qeasons 19. A recent OPCS 11990bi survey of secondary schoolchildren has uncovered several factors associated pith starting to smoke. These include famny structure, educational aspiration, attitudes ar parents and siblings, and a 'elatively positive attitude towards smoking, The survey interviewed a panel at the beginning of the second, third and fourth years of school. This longitudinalf eature made it possible to determine whether awareness of tobacco advertising (ineluding brand recognition and liking for tobacco advertisingl was associated with subsequent Smoking behaviour, Amang boyr who had been nonrmokers at Ihe outset, there was no evidence of differB"CBs in initial awareness and approval or advedising between those who subsequentiy Smoited and those who did not. Among girls :~~err and approval were greater among those who subsequently went on to 20, Alexander at al (19831 aporl a rather similar eobn study of 10~12 year old children in Aus~alia. Factors lavourabll la the adoelion Olsmoking within a year of an initial survey were uncovered by logistic regression. This technique allows the order of importance in the contributory factors to emerge and quantifies the c""tribution of each factor, Approval of advertising comas fourth out of five factors and accounts for about 7% of the smoking variation explained. The leading tactor, unhelpfully, is age at time of La first survey. Second and third come Smaking prevalence among siblings d"d peeri, Gender and parental smoking do "Ot play a pan, Those stcpeing smoking during the year showed the opposite pattern, with disapproval of advertising coming third in the list. 21. However, these findings and Ihosa reported in the OPCS (1990bl study have their limitations: since there was "O variation in advenising to which smoking L7 behaviour could be linked, La associations uncovered represent eicumstantW C BATCo document for Legal Senrices : Health Canada 21 May 1999 evidence that advertising increases tobacco consumption. It remains possible that the causation runs in the other direction · children disposed to smoke are more likely to react positively to tobacco advertising and show greater awareness of it. Qther Evidence of Idi~tlffects 22, Tobacco advertising may encourage smoking by providing smokers with an excuse to play down the danger to their health. A survey by Marsh at al (1983) shows that despite the Government's own health education messages 44"/a of smokers (but only 26% of ndnsmokers) agreed with the statement that 'Smoking can't be really dangerous dr the Government would ban cigarette advera~sing,' The implication is that a ban would itself send a powerful health message. 23. Evidence from the US suggests that magazines remict their coverage of the dangers of smoking because they fear a lose; of revenue from tobacco advertising, A study by Warner at al 11992) looked at a wide range of magazines over the period 1959-86 and using logistic regression related the probability of publishing an article on the health risks of smoking to acceptance of tobacco advertising in the same year, The effect was strongly established for women's magazines in particular, even after taking account of possible confounding variables (svch as size of readership and coverage of health issues generally). For example, the probability of covering smoking dangers was 38% lower in magalines with average reliance on tobacco advertising compared with magazines not accepting tobacco advertising. There remains the problem of the direction of causation. For example, it may be that a magazine's attitude towards smoking determines whether it will accept tobacco advenising, However, coverage of the healtbrisks of smoking was related to the proportion of advertising revenues derived from tobacco advertising, not merely whether it was accepted or not, and this tends to suggest that some magazines have modified their stance in deference to tobacco advertisers. QUAMITATIVE DIRECT NIDENCE 24. It is not possible to mount a controlled experiment to examine the effect of advertising on eonsumpSon, with one group exposed to advertising and a control group not exposed. However, a number of situations have arisen where variations in advertising have occurred which make it possible to investigate the effects of exposure on consumption. Such "natural experiments" fall into three broad categories: (i) different levels of advertising eonb·ols seen in a cross·secticn of countries; iii) year·to·year fluctuations in advertising expenditure within countries; liii) the period before and after the introduction of advertising bans in ·7· C BATCo document for Legal Services : Health Canada 21 Yay 1999 certain countries. 25. Advertising is only one of many potential inflvences on smoking in these contexts, and it is important to control as far as possible for the influence ot other variables to avoid "confolJnding" - attributing the effects of other variables to advertising. Most studies in these contexts we multivariate regression for this purpose; those which do not are lass reliable, We do not report studies whose methods have obvious defects, such as those which note that smoking declined following a ban and conclude that the ban caused the decline without adeg~ate examination of either trends before me ban or of other influences such as price rises which accompanied a ban. 26, Regression methods nevertheless encounter a range or technical problems, some ct which are more important in one experimental context than another. 2f. Same of these problems relate to the specification of the relationship, others to weaknesses in the data, though Be two sats of problems interact. We first consider specification problems. The specification of a regression ~siatiPnship raters to its mathematical form and the variables it incorporates. 28, The form chosen for the relationship inevitably imposes some restriction on the effects which can be picked up. For example, a specification which expects a ban to usher in an immediate and sustained reduction in smoking will deliver an inaeeurata ertimate if the sue eflea is to impart a downward trend to smoking over a number of years. However, limitations in the data restrict me range of passible effects whichean be allowed for. In practice the form of the relationship represents a compromise based on informed judgment, 29, An issue of specification in the context of year·lo·year fluctuations in advenising expenditure is the direction ofeausa~~n, since the level of consumption can innuence companies' decisions as to their level of expenditure on advertising. If this happens, failure to take it into account can lead to bias in the estimate of the effect of advertising an smoking, 30. As we sew above the purpose of using a multivariate technique is to avoid confounding, It is therefore important that the list of explanatory variables should include all variables likely to affect consumption, otherwise same of their effects may bias the estimate of the effect of advertising. Obviously, the regression should include variables such as price and income. But it is also important to bear in mind the possible effects of variables which cannot be included because they are difficuit to measve and operationalise, such as attitudes towards smoking and restrictions on the availability and use of tobacco, 31. Two forms of time lag may be imporlant in tobacco adveiliring. Ta begin with, the effect 01 advertising may lake some Bme to die away, Secondly, the addictive nature of smoking and the life·cycle or smoking behaviour suggests that the effect of advertising on consumption may last for many years, by inducing a ·8· L·~ BATCo document for Legal Senrices : Health Canada 21 May 1999 teenager to startsmoking, for example· This feature indicates a need to include lagged consumption among Ihs erplanatari variab er la Gable ~e full lone term effect of the other explanatory variables to be uncovered, 32, The data base far the regression may not be ideal, potential explanatory variables often move together over time and this makes it difficult to distinguish their separate effects Imuiticollineariryl· The data may contain measurement anors. B is sometimes incomplete .for example, the published data on advertising expenditure in the UK omits billboard adve~sing 33~ These problems affect the various expedmental contexts differently. International crass·sec~onal studies tend to raise the suspicion that omitted variables such as social anitudes to~vardr smokinp play a pan 1 levels and in creation of a social environment favouring a ban. 34, The bulk of the published matsriai s~Jdies the affect of yea'-tn·yaar fluctuations in advertising expenditures within counuies, This experimental context has its drawbacks as a guide to the effect of a ban: .the effect of advertising is likely to be small in comparison with price and income and is therefcre mere likely to fail statistical tests due to the imprecision of estimates; ,the annual nucluations presumably relate to be lean producdve dice of expenditure, on the reasonable assumption that advertising is subject to diminishing retums~ AecordinSly, the effect of an outfight ban which cuts away Itie most eHective care of advertising is likely to be greater than extrapolation would suggest. The US Surgeon General's Report (US DHHS 1198911 presents a strong if somewhat speculative version of this argument: since the total level of advertising expenditure which is worthwhile to contest market share lies well above the level associated with tero marginal impact antotal tobacco consumptian a full ban could have an effect even if the evidencebased on yearlo·yaar fluctuations points to no effect, 35~ A total ban is a much more promising experimentai contsd which gets round many of the difficulties inherent in Ihe study of year·to-year nuetuatians, particularlY causation running in both direcriono Mo~wver, ii there is an effect it wl be on a larger seals and should show up more clearly. It is worth bearing in mind that the effect may also be enhanced by the aecdmpanymg publicity and by the implicit message that government now believes that the dangers of smoking justify a ban, 36, Predicting the effect of a ban has to rely to same degree on evidence from other countries, A number of factcrs need to be b6me in mind in transferring results between countries: ·9· ii; BATCo document for Legal Services : Health Canada 21 May 1999 · in predicting the effects of a ban it is important to take account of the degree of restriction already in place. Finland, Norway and Canada imposed bans from a much lower base of restriction an advertising than New feaiand. The UK's eunent level of restriction lies closer to the preban level in New Zealand than to pre·ban levels in the other three countries (Laugesen (1991)); · in some countries various other measur~s accompanied the introduction of a ban and these are likely to have had some effect on smokingl · because of cultural factors, tobacco consumption may not react to changes in price and advsrfising variablesin quits the same way in different countries; - the dbgree of market power in Vie tobacco market may vary from country to country and with ii the proportion of advertising undertaken with a market share objective, C!orr:section A_?alvsiss~!~Co4ntries ~j~h Dit~enl lwe~s! Contob 37, There are two international eroJs·seetional studies. 38, A study by Cox at al (1984 makes an attempt to assess the effectiveness of the different approaches to tobaccoadvenising control policy seen in different countries, The study identifies twobasie philosophies: a legislativtl approach and a piecemeal approach basedon voluntary agreements. In the absence of data on advertising expenditures inthe different Countries, the study resorts to an indirect method of assessing the effect of control policy on smoking, ihe method is to carry out a timeseries regression over a similar time pen'od for each country using a common specification, explaining tobacco eonsumpdon by price, income and a time trend, To the extent that tobacco control policy is influencing consumption its effect should show up in the regression equation in two ways. Firstly, the proponion of the variation in tobacco consumption explained by the regression should be lower in the countries where the control policy is more effect've in influencing consumption. Secondly, all the estimators correlated with the omitted variable will be biased and inefficient, but the authors focus on the effect on the time trend, arguing that it should plot a steeper downward course in the countries which have achieved more efsctive control as it picks up the long term effects on tobacco consumption. These effects are indeed found and the authors conclude that legislative regimes work better than voluntary arrangements. However, this evidence is suggestive rather than conclusive, Other factors could account for be regression results. Social trends away hem smoking may be stronger in the "legislative" eauntries, Indeed there may be reverse causation · a stronger social trend away from smoking which shows up as a negative time trend in the regression equation may have created a climate of opinion favourable to legislative control of tobacco advertising, ·io· VI BATCo document for Legal Services : Health Canada 21 May 1999 39. Laugesen at at (1991) have carried out a multivariate pooled cross- Sectionltime series international Study using a sample of annual data covering 22 OECD countries over the period 1960·1986'. The explanatory variables include price, GDP per head, a score on a scale of 0·10 to capture the level of advemsing restrictions and certain other variables such as the proportion of tobacco accounted for by manufactured cigarettes, The advertising coefficient suggests that for each point on the advertising scare consumption falls by 0.62. In a second speifieadon the coefficients are allowed to vary over time by combining dummy variables representing succeeding years with the year's values of the explanatory variables, The advertising variable registered a perverse positive sign in the early years but it became negative in the early 19705 and has grown steadily in value since then, there is no obvious explanation for the unexpected posit've sign in the early period, The coefficient for the latest year suggests that each point on the adver6sing score would reduce eonsumpaon by 1K% 10,696 whets the oceHicients are held constant over timel, The UK's advereising score was 6,b (ouT of 101 in 1986, though subse~oent agreements on spom sponsorship and advertising will have increased it marginally~ 40, The system for calculating a country's tobacco control score is set out in Appendix 4 of Health or Tobacco (NZ Department of Health %981. A number of features of the weighting system are open to question. For example, a ban an television advertising gets the same score as a ban on cinema adva~ising, despite the very much higher audiences for television, It would be interesting to see the effect of different scoring systams~ 41. The main difficulty with this experimental context is uncertainty about h~e directon of causation and possible canfaunding factors: negative social attitudes towards smoking are likely to lead to low tobacco consumptidn and to saict controls on tobacco advertising, In these circumstances a eoss·seeticn study would suggest that advertising restrictions are responsible,for law tobacco consumption whereas in reality a third factor causes both, Slewart (1991) has cast doubt on the reliability of some of the data used in this study with all the variables affected in same degree. He argues that errors in the data discredit laugesen'J results, However, the e8ects of data error are not as serious as he suggests. The effect of "white noise' measurement enor on the advertising coetficient is cleat in the case of the dependent variable and the advertising variable, but not in the case of the other explanatory variables IGreene (1991ail, Measurement error in the dependent variable increases the variance of coefficients but does not bias their value~~ Measurement error in the advertising variable leads to "aaenoation" biasing its own coefficient towards Zero. Accardingb, errors in the advertising variable may not be critical, However, Wewarf also makes persuasive recommandau'ons as to better ways of measuring some of the explanatory variables, Errors in thess variables impart bias to the advertising coeReient but the direction of bias is not in general known, Stewart does not offer a fresh analysis on a better data set, In the meantime it would be wrong to rejed Laugesen's results on ~he score of imperfect dala~ 4~ ·It· L~ BATCo document for Legal Services : Health Canada 21 May 1999 Time Series Anacsis of Flvetualions in _Advertisinq Expenditure wilhin I1Puneies 42. The bulk of the published material studies the effect or year-loyear fluctuations in advertising expenditures within countries, mainly this countnl or be United Stares. All the studies reported here take account of other influences on tobacco consumption to avoid confounding, not only price and income variables but also specific health "scares" such as reports of the Royal College of Physicians. Many of these articles are of very high quality using state·al·the·an economeuis technique. Advertising is not always the only or even the main focus of interest. Many have found that advertising has an effect; some have not. There appears to be a tendency for articles which have found an effect to underestimate its scale, The reason is that results are usually expressed in terms of an elasticity figure which implicitly invites comparison between different variables standardised on a 146 change. However, in order to standardise the evidence with that on advertising bans, it is more useful to show the effect using a 1001b change. The results presented in this section follow this method, though as we law in paragraph 34 there are good arguments for thinking the effect of a ban would be greater than simple extrapolatioo would indicate. 43. Special factors are at work in different caun~es. Independent researchers in the UK have to work within the limitations imposed by the ineomplere advertising data released by the tobacco industry, covering broadcast media and the press, but omitting billboard advertising and sponsorship. The effect of this restriction became more serious following the ban on cigarette advertising on television in 1965. Radfar (19851 estimates that in the mid·19805 poster advertising accounted for 30%·POX of expenditure on cigarette advertising. The Melra consulting group commissioned by the tobacco industry had access to the lull advertising data apart from sponsorship which had not then, in 1979, attained the importance it appears to have today. Studies carried out in thi United States are particularly informative, because compnhensive data on advertising expenditure are freely available to independent researchers. 44. The studies differ in specification, eg the presence or absence of a lag on consumption or a carryover effect of advertising from one time period to the next. Most studies have to rely exclusively on time series data. In the US, however, it is possible to use pooled data taking advantage of cross·seetional variation in price across states. 45. The results of the key published studies are summarised in the table overleaf. Since the full effects of a reduction in advertising may take some time to come through, same of the studies have provided short run and long run estimates (al[owing for a habit effectl, A range is shown where studies report short and long term effects or results from differentspecifications. The table also shows the publication dales of the studies and the period cased by the data they have used. ·12· BATCo document for Legal Services : Health Canada 21 May 1999 EFFECT OF ADVERTISING ON TOBACCO CONSUMPTION: MAJOR TIME·SERIES STUDIES Reduction Associated Date of with a Ban 96 Data United KinPdom MeGuinness (1475) 7'/1% 1957·68Q Meua(19791 1 1958·78(1 Wia 119al) 71b 1955·75 Aadfar (19851 3%·516 1965·80(1 Godfray 119861 1956·84 outrv 11991) Is·tl·a·r DH 11992) 1958·87 United States Hamilton it 972) 1953·70 Schmaleosee (1972) t 195667Q Fujii 119801 227% 1929·73 Young (1983) 3% 1929-73 Bahagi(1986) 1963·80 Bishop (19881 79b 1954-80 Seldon 11989) 9~ 1952·84 Tegsne 119911 41b5% 1953-85 Other eounrries Leeflang 11985l1 14%·16% 1960-75 Johnson (1 986)' 1961-83 Chetwynd (1988)5 1%' 1973-85 Harrison (19911" 896·12% 1973·85Q statistically insignificant D quarterly data i Godkey investigates a number of alternative models and data definitions in an attempt to improve the specification of tobacco demand models. A large number, and a considerable range, of results are produced. · See paragraphs Af 8·A22 of annex A. · West Germany. · Australia, I New Zealand. · Result using quarterly data; the result for annual data is not stab~stically significant. ·13· BATCo document for Legal Services : Health Canada 21 May 1999 46, Annex A discusses the studies in detail, bulit will be helplul to touch on their salient features to supplement the bare results shown in the table. In the UK the debate among me earlier studies focused on the issue or the Persistence of an advertising effect and how it should be measured. The results from this period divideinto two camps. Independent researchersusing the incomplete published data on advertising round an effect whereas the Metra Consulting Group which had access to the complete advertising data did not. 47. The later UK studies show greater variety, exploring a number of specifications, bur most of them include lagged consumption, tp represent a habit eff ea, and previous periods' advenising eupendilure to allow for persistence of the effect of advertising, Witt et ai ii9851, however, estimate a very simple specification using oniy current values or the variables. None of the UK studies formally inveslipatss the FossibiliE( ~at consumption could affect the level of advertising in such a i~ay as to bias estimates made of the eff ect of advertbing on eonsumptian. Godfrey's ( 19861 objecdve is to explore diff erent specilicatians and data delilons and she eonrjepventiy Produces a large number of results with a wide range of values. It is interest~ng that results should prove so sensitive to specificadon. Duffy (1991) explores the effect of advertising within a complete system, the "Ratterdam" model, uihich takes account, in particular, of effects between tobacco and the dinerent alcoholic beverages. Our own work reported in Annex B lakes advantage at the data an advertising expenditure supplied to be Departments of Heaith and Education and Science in accordance with agreements between the tabaeco industry and government, ihe statistical analysis represents an example of SeneraCt6specilic modelling, whereby an initial :I~Y:lk"~E~,~':,~d~e"Si'veiv limplsd and Ihe modilicilionr lePed 48. Studies of tobacco advertising in the US present a simpler picture. All investigators have acces to reliable advenjsing data, As successive studies pursue product differena:ation by adding some refinement to the design or specification of previous studies and deleting others. Generany speaking, the more recent studies are of higher technical 9ualinl. Only Bishop et al 119861 fail to include lagged consumption in the list of explanatory variables. Only Bishop et al (19861 and Fujii (198[)1 fail to allow pan advertising to influence current consumption. Hamilbn (1972) and Balragi etal 119961 use pooled data; the others use lime series. Only Sehmalenses (19721 and Seldon at al (1981 entertain the Possibility that a link between advsrtising and consumption could renect the industry basing advertising decisions on cbnaumption, Only Seldpn at al 11989) adapt a specification which seeks to explain La level of advertising expenditure as WBII as the effect ofadvenising on consumption. 49. Studies from theother countries are more r~xed, though they generally follow Ihe methods established in the inlornational literature; Harnsen at al 119911 in particular apply a range cl the latest diagnostic tests in their analysis of New Zeaiand data, The drawback to the New Zsaland work is the unavailability of comprehensive data on advenjsing. The Australian study 1Johnson (19861) and .14' i BATCo document for Legal Services : Health Canada 21 May 1999 the West German study (leefiang at el 11985)1 are included here to extend the coverage of countries, 50. We now came to take stock of these studies and to consider what overall conclusion they indicate. Some studies have found a sta~stieally significant effect whilst others, including our own study, have failed to do so, One approach to a conclusion would be through a judgment as to the relative merits of the different studies. The studies are already selected for their use of multivariate statistical analysis to avoid confounding, There exist a number of aLer teas of Jtatisdcai performance. Harrison at al (1991) illustrate good practice, applying tests of the direction of causation, Specification, parameter stability and characteristics of anors. One possibility would be to rank the different studies according tO their performance an these criteria~ However, few studies have carried out these tests and their performance in these respects is unknown, 51, There exists a more straightforward way of drawing a conclusion from a wide range olstudies. If advertising had no effect on consumption we would not expect ail the results to be exacdy zero. Because of sampling we would expect a range of results but with the tally of pasilivs and negative results broadly in balance. MA studies understandably highlight only their positive results and dismiss negative results on the reasonable grounds that they do not make sense. Accordingly, we have gone through the published studies cited above collecting all t~e results to find the balance of positive and negative findings. The list excludes results from equations which the authors report as misspecified. The results are as follows: EFFECT OF ADVERTISING ON CONSUMPTION: SUMMARY DF TIME·SERIES RESULTS' Positive Nega~ve SigniScant 68 2 insignificant 103 39 total 171 41 Most ct the studies report only posilve eff Bets, with only Johnson (1 9861 failing to report any positive effects: 'including the results of the analysis of UK data presented in Annex B, ·15· C BA"TCo document for Legal Senriees : Health Canada 21 May 1999 NUMBER OF STUDIES BY SIGN OF RESULTS Positive only 13 Flegativs only soth The preponderance d positive results Paints to the conclusion that advenising does have a positive consumption, ~su~smst~in 52. ~he evidence Com yea~·lo.year fluctuations in advertising expenditures has certain drawbacks, p, ban Is a much m018 Promising experimental context which Sets round many of these ditficulties, Advertising bans in broadcast media which many countries imposed in the sixties have some of the features ola total ban and we discuss them in the following Paragraph before going on to the more informative context of a full ban. 53. Hamilton (19751 examines the impact on c0"sumptibn in the individual countries which introduced bans in broadcast media in the rixler. HiS Study USeS data covering a period from the fihies U"til 1973. The explanatory variables include lagged consumption, price, pe' capita GDP and dummy variables to represent the adrenising ban and Ihe standard health scares. The result are very mixed as the following table shows: EFFECT ON CONSVMPTION OF AN ADVERTISING BAR IN BAOAOCIST MEDIA RESVCTS FROM SEVEN COUNTAIES Reduction Increase Significant Insignificant 3 2 Total 4 3 The results point to a reduction in four countn'es and anincrease in three, a narrow favour or a reduction. It may be worth noting that unlike a complete ban a television ban does not nPcersarily imply a reduction in total expenditure an tobacco advertising, although it will almost Eerrainly entail a reduction in Its productivity. In the linited Starer, where Ihe ban came to late lpr Hamillon'l 1975 study, advertising expenditure lell folloaing lh, 1971 television ban but rose guickiy thereafter, it is also intererling tD Oote that in the study discussed in Annex A, Hamillon 119721 predicted that the US teievision blo,,uld actually Increase consumption because it also ended ~he Obligatery health messages which C ·16· C: BATCo document for Legal Senrices : Health Canada 21 May 1999 accompanied the advertising. However, 9altagi (19861 has shown that the effect was to reduce consumption, 54. We now turn to consider the countries which have introduced bans. 55, Two countries, Norway and Finland, have had bans for a number of years. The recent bans in New Zealand and Canada also repay study. iceland imposed a ban in 1971, but we exclude it on grounds of sample size and data unavailability, Other countries are sometimes included among the countries with bans, but we exclude them because of doubts as to the effectiveness of enforcement, the investigation technique again has to be same form of regression analysis to take account of the many other influences on smoking. There is a published study of this kind only for Finland, Wa have conducted our own study for Norway and, in much less detail, for Canada and New tealand, 56. Norway introduced a Tobacco Act in 1975 banning all tobacco advertising, sponsorship and indirect tobacco advertising. It also introduced health education measures, health warnings on tobacco packaging and restrictions on the sale of tobacco to young people, 57, Biartveit (1990) has commented on the results of the Act but we know of no published study which deploys econometrie techniques. Wa have carried silt our own study using consumption' data supplied by Norway's National Council on Tobacco and Health, A tobacco pries index and an index of real consumers' expenditure were taken from OECD sources, It is desirable to take a starting date a few years before the changes under study to give a good run at the problem. In this ease 1964 was chosen. The last date for which data far all the variables was available was 1989. Pipelrclling tobacco is an important product in Norway, with the dominant share of the market. Our study handled all tobacco products together and looked at the effect from the implementation of the ban in 1975 rather than its enactment in 1973. We considered two specifications, The first takes as the explanatory variable average consumption per adult. The second is a "double hurdle" model recognising that the decision to smoke and the decision on how much to smoke may have different determinants. it involves simultaneous modelling of prevalence and amount smoked par smoker, In view of the imponanee of young people in smoking recruitment it would be desirable to investigate the affect of the Act on their smoking prevalence, Data does exist for smoking prevalence among 13, 14 and 15 year olds in Norway. This data comes to a peak in 1975, but no survey took place between 1963 and 1975, and it is quite possible that prevalence peeked some time before the Act came in, For this reason we are inclined to doubt that this data throws light on the effects of a ban, ·II· BATCo document for Legal Services : Health Canada 21 Yay 1999 56. A reduction in consumption brought about by a ban on advertising could take a number of farms, A ban could trigger a permanent reduction in consumption perhaps taking some time to come through fully; or it could work its effect more gradually by increasing anestablished trend decline or by initialing such a trend. Our study allowed for both these farms of effect. 59. Annex a describes the study. The first model explaining consumption per head suggests that in the long run the Act reduced consumption by 16%. The second model paints to a more modest decrease of about: 9% in the long run. The affect comes through a reduction in smoking prevalence; the amount smoked per smoker does not appear to have been affected. it is not possible to cjuantihl precisely the share of this reductionartributabla to the advertising ban by itself, but, in view of the minor nature of the other provisions of the Act, the ban on advertising is likely to have accounted for the great majority of the effect Einland 60, In 1971 Finland introduced a ban on tobacco advertising in broadcast media. Following extensive public debate a Tobacco Act came into force in 1977. The Act banned all farms of tobacco advertising, but it also included a number of other measures including restrictions on smoking in public places, a prohibition on sales to under 1Byear-olds, and a half percent levy on tobacco tax revenues for anli· smoking policy, including health education. Any test of the effectiveness of the Act inevitably amounts to a test of the whole range of measures and not simply the adve~sing ban. 61, A study by Pekurinen (19891 of Consumption overthe period 1960·87 focuses on the effect of the Act using a specification which follows the VS studies reviewed above, with allowance for an effect of lagged consumption as well as income and price (ineluding asymmetrl in rises and falls in price). A dummy variable divides the periodbefore and after the 1964 health scare (associated with the VS Surgeon General's Report linking smoking and health) and another the period before and after the Tobacco Act. The author chooses 1976, the year before the Act, as the starting point for studying its effect to capture the impact of the publicity created by public debate about the Act. A feature of the Finnish tobacco market is the fairly high share of Pipeirolling tobacco, about a sixth in the mid·seventies. The author chooses to treat these different tobacco products separately. His analysis finds that the Act led to a sustained reduction of 6.79b in cigarette smoking. However, the results also show a 26# rise in pipeiroling tobacco, There is no reason for a positive effect and the author dismisses it as a rogue interaction, particularly since there was very IUe advertising of pipslrolling tobacco. The author highlights (ha 6,7% reduction in cigarette consumption as giving the most reliable estimate of the effect of the Act. ~tB· BATCo document for Legal Services : Health Canada 21 Yay 1999 Canada 62, Canada introduced a ban on 1 January 1989 when tf~e Tobacco Products Control Act came into force, The Act banned all tobacco advertising, sponsorship and indirect advertising such as trademarks an nontobacco goods. It covered media or Canadian origin and advertising by Canadian tobacco companies aimed at Canada, Tobacco advertising carried by US maga2ines continues to be seen as before. The US broadcast media which have a wide following in Canada are subject to the US ban of 1971. The ban an poster and billboard advertising was phased in overtwo years, with a one third reduction in expenditure compared with the 1987 nominal level inIhe first year and two thirds in the second. The ban on point of sale advertising isqualified to allow lour years' glace for advertising in place and contracts entered into before the 25 January 1988, Minor concessions on spans6rships include continued use of the name of the tobacco manufacturer and use of a brand name until pr,existing contracts have run their course. no tobacco company's expenditure on sponsorship may exceed its nominal 1987 level. 'The tobacco industry appealed against the TPCA in the Ouebec Court of Appeal an'd won its ease. The Act was declared an unconstitutional infringement of freedom of expression. However, the Act remains in force while the Government takes the matter to a higher court, 63. Concessions and exceptions notwithstanding, the conuols brought in by me Act were tight, wide-ranging and largely immediate. We have consumption data from 1972 to 1990 including two yeaa since the Act same $a force. This data comes tom official Canadian sources and includes an estimate to take account of smuggling and cross·border purchases. We also have data from official Sources over the same period covering real disposable income per head and the real price of cigarettes. The real price of cigarenas doubled during the eighties as a result of Steady annualrises in tax as the Government pursued a progressiveiy tighter tobacco control policy. 64. The effects of the labaecoadvertiring ban canbe examined using a regression equation seeking to explain annual eigarene consumption per head. ~he explanatory variables providing the best Ct include price, a time trend and the lagged dependent variable;they do not include an income variable. The changes in consumption in 1989 and 1990 and the Ilxpacted changes taking account of the trend followed by the independent variables in the regression ralalionship were as follows: ·19. BATCo document for Legal Services : Health Canada 21 May 1999 CHANGE IN CIGARE~IE CONSUMPTION e4~ 41e~ A~rual_minusE~~CeCted Ann_ual Culn~ari~ 1989 ·6.2 ·4,5 -1,7 ·1,7 1990 ·6.2 ·5.1 ·1.1 ·2.8 The two years since the ban have seen a reduction in consumption of 2.8W more than expected from the price increases and the annual time trend reduerion of H16. A dummy variable representing the two ban years points to a reduction on the slightly greater scale of 4% in the longer term. This estimate requires cautious interpretation because of the possibility that these reductions couid reflect the year·t0·year fluctuations round the fitted trend. Nec~Zealand 65. An outright ban was introduced in New Zealand in December 1990, the culmination of a series of tightening centrals on advertising introduced over a number of years, The 1990 restrictions banned advertising in print media originating in New Zealand, pasters and sPonsorship of sport, The ban does not apply to imported magazines, about half the market, or to Australian television broadcasts of sponsored events. As a aneession p~e·existing advertising contracts were allowed to run their course, Rgures tom the NZ Department of Health show bat cigarette sales fell by about 7 K 1b in be six months following the 1990 ban compared with the six months before. Excise revenue appears to have fallen even further, curiady, probably refiecting the volatile sales to trade rather than retail sales, Real labacco prices rose by 3.70~ and real personal disposable incomes saw a slight fall of 0.3%. Assuming elasticities of about 0,5 for both variables Ihere 8ctofs are likely to have accounted tor about 2% or Ihe lall in consumption, leaving a fall of 5'/,010 as possUy attributable to the advertising ban, :~eo~:8~perable ealion ii re~uired in u Iw ~lthe bnef lime pried covrld by ~20· BATCo document for Legal Services : Health Canada 21 May 1999 of health edueabon messages. 69, fuming now to the evidence based on statistical analysis of tobacco advertising and consumption, the main findings of our review are as tallows: i. iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii~ countries diner in the extent of their controls on tobacco advertising. Allowing also for factors such as price and income, these differences can be related to varia~onsin levels of tobacco consumption per head. These ioter·country studies have found an eHect, but there is a question mark about the direction of causation, Social attitudes towards smoking may differ internationally, leading to lower levels of smoking and stricter conucls on advertising in some eounhies than in others, thus creagng an association between the two without the controlscausing the lower tobacco consumption. Eear~tOlvear variations in there are many studies of this type,hoth for the UK and for other countries, particularly the US. Details of the analysis vary from one study to another, but the basic approach is to model the demand for tobacco products by relating annual changes in consumption to corresponding changes in advertising expenditure and in other variables such as price and income, Some studies have found that advertising has a statisticany significant effect on consumption; others, including our own study, have not. There are several possible reasons for failure to find a statistically significant effect of advenising, including data imperfections and the inherent difficulty of identifying the separate effect of advertising when this is only one of many potential influences on smoking behaviour, Taken together, however, the studies point to a more decisive result. Because the studies differ in specification and data, a range of results is always to be expected. If, however, advertising genuinely has no affect on consumption, it would also be expected that the numbers of studies reporting positive and negative results would be much the same; in other words, same studies would show that advertising increases consumption, but others that advenising reduces consumption. In practice this symmetry is mt observed: the greatmajority of results point in the same direction towards a positive impact. The balance of evidence thus supports the ccnclusian that advertising does have a positive effect on consumption. iii. ~ij~i?g~s~ adverliring bans provide the mast reliable testbed, as they involve large changes in advertising volumes and there are no doubts about the direction of causation, The main camp ication is that bans are often introduced alongside other measures such as stronger health warnings, additional health education and restrictions on sales to young people. Some of the observed effect on smoking may be due to these additionalmeasurei, ihe impact of advertising bans has been assessed in four countries · Norway, Finland, Canada and New Zealand, 'Thoogh there are gualifications (lor example, the bans in Canada and New Zealand are relatively recent and so may not yet have had their full impact), the current evidence available on these four countries indicates a significant eftect. In each ease the banning of advenising was followed by a fail in smoking an a scale which cannot reasonably be attributed to other factors. ·22· BATCo document for Legal Services : Health Canada 21 May 1999 SUMMARY AND CONCLUSIONS 66. This report has reviewed various farms of evidence to assess whether tobacco advertising affects the aggregate demand for tobacco products land not simply the share of the market taken by individual producers), 67, Most or the evidence comes from studies which use statis6cal means to analyse the relationship between tobacco consumption on the one hand and a range of variables including advertising an the other, Many of these studies look at the impact of ysarto·year changes in advertising expenditures within individual countries, but there is also evidence arising from the variation between eounm'es in the level of controls on advertising and perhaps most important of all tom before·andafter studies in those countries which Rave already introduced advertising bans, ihis evidence is summarised and assessed in paragraph 69 below, 66, Other farms of evidence include the following: i, market structure: advertising to increase the overall scale ol demand would almost certainly be proStable if the market for tobacco products were in the hands of a single producer. Particularly profitable would be the recruitment of young smokers, reflecting the addictive nature of tobacco and the subsequent length of smoking career laround thirty years an average), Such advertising is less likely the more firms there are in the market, as any one firm reaps only a fraction of the benefits. However, the existence of brands and brand loyalty among smokers confers same monopoly power on each firm and makes it profitable for the firm to increase the overall sireof the market by recruiting new smokers to a pa~eular brand. Moreover, brandadvenising may lead some people to start smoking even if the firm does not intend this result. ii. rezpsnition of advertisements; there is a great deal of evidence to show that young people recognise tobacco advertisements and that theta who go on to smoke are more likely to recognise them, But awareness of advertising is at most a necessary condition for coming under its influence. It is not reliable evidence that advertising increases consumption.