2. Public Health Programs to Reduce Tobacco Consumption

2.1 Introduction

The risk of death from tobacco consumption exceeds that of any other addiction, exposure or injury. In 1996, tobacco smoking caused an estimated 16 875 deaths in Australia (Mathers et al., 1999). This total included an estimated 6262 deaths from lung cancer, 4645 deaths from chronic obstructive pulmonary disease (COPD), 2507 deaths from coronary heart disease, and 3461 deaths from a variety of other conditions (notably several other cancers and strokes).

However, deaths from tobacco smoking would have been much higher if smoking had not declined substantially over the previous twenty-five years. Figure 2.1 shows the decline in the prevalence of smoking among Australian males and females between 1974 and 1995. During this period, on a 1986 age-standardised basis, adult male smokers fell from 45 per cent to 27 per cent of all adult males.¹ Adult female smokers fell from 30 per cent to 23 per cent of adult females.

In this chapter we outline the history of the main public health programs to reduce tobacco consumption and their costs and examine how the programs contributed to the decline in tobacco consumption. We then estimate the premature deaths averted and other health benefits due to lower tobacco consumption and provide an economic evaluation of anti-smoking public health programs. As discussed in Chapter 1, the study evaluates the impacts of the programs overall from 1970 to the present rather than for particular programs or incremental expenditures.

Figure 2.1 Changes in the prevalence of tobacco smoking

Source: AIHW, 1999.

2.2 Public Health Programs to Reduce Tobacco Consumption

Government programs influence tobacco consumption in three main ways:

By providing information on the potential health effects of tobacco consumption;
By regulations that restrict the promotion of cigarette products, restrict access to them, and influence the conditions under which the products are consumed;
By a range of taxes which increase the price of cigarettes and reduce consumption.

Governments provide information in various ways, including education programs, publicly funded anti-smoking campaigns, and packaging and labelling requirements. Commonwealth, State and Territory governments provide teaching materials to schools to make children better informed about the health effects of tobacco smoking before they start smoking. Since the early 1970s, the Commonwealth has funded periodic national mass media campaigns to encourage smokers to quit smoking. In 1972, the Commonwealth required that cigarette packets be labelled to warn of the adverse effects of smoking. The warnings were strengthened in 1987 and 1995.

There are also numerous regulations on the sale and consumption of cigarettes. The Commonwealth has banned the sale of tobacco products to minors for over 80 years. Between 1988 and 1996, the age of minors was lifted from 15 to 17 years of age in various Australian states. The Commonwealth banned advertising of tobacco products on radio and television in 1976, advertising in the print media in 1992, and advertising on billboards in 1995. The state governments also ban advertising of tobacco products in various public places. Bans on smoking in public places have been introduced progressively in places with high fire risks, in public transport, and in government offices, especially since the mid-1980s.

In the period under study, governments levied three main taxes on tobacco products: Commonwealth excise duties on domestically produced products; customs duties on imported tobacco products; and State licence fees and taxes. These taxes have generally been increased at a higher rate than the consumer price index, especially since the mid-1980s. As a component of the retail price of cigarettes, taxes rose from 43.7 per cent in 1983 to 61.1 per cent in 1994 (Industry Commission, 1994). Governments also reduced the effective rate of industry assistance for production of tobacco products from 30 per cent in 1974-75 to zero by the late 1980s. These changes contributed to large increases in the real price of tobacco products: an estimated 40 per cent price increase between 1981 and 1990 and an estimated 75 per cent increase between 1991 and 1996 (see Table 2.4 below).²

Table 2.1 shows the main public health programs for tobacco reduction in Australia since about 1970. It does not include international events or the large numbers of State campaigns. Nevertheless, it can be seen that there has been a rich and complex history of programs to reduce tobacco consumption, especially between 1972 and 1976 and from 1983 onwards. The effects of these programs on behaviour and on health are almost certainly synergistic, cumulative and lagged. In these circumstances, the impacts of individual public health measures on behaviour are not easily identified.

Table 2.1 Public health programs to reduce tobacco consumption

Year	Public Health Program
1964	US Surgeon General’s Report linking smoking and lung cancer.
1960’s	Australian media campaigns start on dangers of smoking to health.
1968	Federal labelling regulations on cigarettes introduced.
1972	Mandatory health warning on cigarette packs.
1972-5	First federal campaign: 'National Warning against Smoking'.
1972-76	Commonwealth ban on advertising of tobacco products on radio and television phased in.
1974	National Survey of Smoking Prevalence (periodical monitoring of smoking rates) began.
1982	Voluntary agreement between tobacco industry and Government to control levels of tar, nicotine and carbon monoxide in Australian made cigarettes.
1983	First mass media QUIT campaigns in NSW and WA.
1984	First mass media QUIT campaign in Victoria.
	National Secondary School Survey to monitor smoking rates every three years.
	State cancer councils, National Heart Foundation and medical colleges establish Action on Smoking and Health (ASH).
Mid-1980's	National Drug Strategy implemented.
1985	Ministerial Council on Drug Strategy (MCDS) formed. First campaign to launch the National Campaign Against Drug Abuse (NCADA) – later renamed National Drug Strategy.
1986	Centre for Behavioural Research in Cancer in Victoria established. Later centres include Health Promotion and Cancer Prevention in Brisbane, and NSW Cancer Council's Cancer Education Research Program in Newcastle.
	National Health and Medical Research Council adopted recommendations on the Effects of Passive Smoking on Health.
1987	Labelling - four warnings introduced on all tobacco products except cigars.
	Victorian Tobacco Act levied a wholesale tax on tobacco products sold in Victoria to fund Victorian Health Promotion Foundation (VicHealth). Quit funding triples and replacement of tobacco sponsorship of sport and art began.
	Government strengthens regulations for package labelling including tar, nicotine and carbon monoxide levels in cigarettes.
	Commonwealth legislates to ban smoking in domestic airline flights.
1988	Major federal court ruling concerning health risks of passive smoking.
	Commonwealth Government bans smoking in Commonwealth government offices
	Foundation SA established and outdoor advertising began to be phased out.
1988-91	Bans on smoking in Victorian government offices introduced over three years.
1989	Legislation banned cigarette advertising in print media.
1989-91	Vic. SA, ACT and then WA state governments phase out outdoor advertising of tobacco.
1990-1	National TV and cinema advertising campaign: 'Smoking – who needs it?'.
1991	National Health Policy on Tobacco endorsed by Ministerial Council on Drug Strategy.
	Healthway established in WA.
1992	NSW then NT ban outdoor advertising and most tobacco sponsorship.
	Commonwealth bans tobacco sponsorship of most sporting and arts events and remaining outdoor advertising, mainly in Queensland.
	First court award of damages for adverse effects of passive smoking.
1993	Nicotine patches approved for use in Australia and available on prescription.
1994	Commonwealth campaign aimed at reducing teenage smoking.
	Smoking banned in public places in the ACT.
1995	Government regulates for stronger health warnings printed on packaging.
	Commonwealth limits average amounts of tar, nicotine and carbon monoxide in cigarettes.
1996	Limitations on point of sale advertising in Tasmania.
1997	National Tobacco Campaign ‘Every cigarette is doing you damage’.
1998-99	Smoking banned in public places in SA and WA.
2000	Smoking bans in restaurants in NSW.

2.3 Expenditure on Anti-Smoking Public Health Programs

The Centre for Behavioural Research in Cancer (CBRC) has assembled detailed data on expenditure on anti-smoking education campaigns since 1983 when the main Australian programs began.³ The data collected included expenditure on educational and counselling activities aimed at adult smokers, quit programs, activities for special groups such as Aborigines and Torres Strait Islanders, prevention education activities, payments to groups sponsored to promote the Quit logo, and administration and overheads. However, as the CBRC notes, compiling the data was not easy as anti-smoking initiatives are organised differently in each state and records of expenditures are made in different ways. The analysis was restricted to expenditure on educational activities and did not cover expenditure on the enforcement of regulation or on general non-specific drug education in schools.

As shown in Table 2.2, annual anti-smoking education expenditures were around $5 million in the mid-80s and were generally between $13 million and $15 million from 1989 onwards in then current dollars. In constant year 2000 dollars, expenditure peaked at $19 million in 1989-90 and has fallen slightly since.

In AIHW’s comprehensive review of health care costs for 1993-94, Mathers et al. (1998b) attempt to estimate total anti-smoking expenditure, including some items that are not included in the CBRC’s estimates. They added $2 million for tobacco legislation enforcement at State and Territory level and $1.1 million of Commonwealth expenditure for components of the National Drug Strategy and drug education programs. Thus Mathers et al. estimate that anti-smoking expenditure totalled about $18.0 million in 1993-94, which was 21 per cent higher than the CBRC’s estimate of anti-smoking education expenditure ($14.0 million). Our estimated total expenditure (the last column in Table 2.2) includes an extra 21 per cent for omitted expenditures in each year.

Table 2.2 Expenditure on anti-smoking public health programs


Year	CBRC estimate ($ million) Current $s Year 2000 $s		Estimated total expenditures (2000 $s)
1983-84	3.0	5.9	7.1
1984-85	3.9	7.4	9.0
1985-86	4.8	8.4	10.2
1986-87	5.8	9.3	11.3
1987-88	8.6	12.8	15.5
1988-89	13.1	18.2	22.0
1989-90	14.6	18.8	22.7
1990-91	13.3	16.2	19.6
1991-92	15.0	18.0	21.8
1992-93	12.2	14.5	17.5
1993-94	14.9	17.4	21.1
1994-95	13.7	15.5	18.8
1995-96	13.6	14.7	17.9
1996-97	14.2	15.1	18.3

Sources: Centre for Behavioural Research in Cancer, Michelle Scollo; Consultant estimates.

In the absence of reliable data on expenditures on anti-smoking programs before 1983-84, we allow that annual anti-smoking health programs totalled $5.0 million per annum in 2000 dollars in these earlier years, which is broadly consistent with the figures shown in Table 2.2.⁴

2.4 The Impact of Public Health Programs on Tobacco Consumption

The prevalence of tobacco consumption has fallen considerably since the early 1970s. The fall was especially marked from the mid-1980s when major anti-smoking campaigns were initiated. It was also stronger among males than females. On a 1986 age-standardised basis, the proportion of adult male smokers fell from 45 per cent in 1974 to 40 per cent in 1983 and down to 27 per cent in 1995. On the other hand, the proportion of adult female smokers did not decline until the mid-1980s, but it fell from 29 per cent in 1986 to 23 per cent in 1995. (See Table 2.3.)

Significantly, reported cigarette consumption fell by more than prevalence, as smokers reduced their tobacco consumption. Average reported cigarette consumption per adult male halved from twelve cigarettes per day in the 1960s to less than six cigarettes a day in the 1990s. Average reported cigarette consumption per adult female fell from six per day to four per day over the same period. (See Table 2.3 and Figure 2.2.)

Table 2.3 also shows estimated real expenditure per adult on tobacco products from 1965, indexed on 1970. Tobacco expenditure per adult peaked at $636 in 1970 (in 2000 dollars). Per capita expenditure fell by 19 per cent in the 1970s, by 22 per cent between 1980 and 1989, and by a further 37 per cent between 1989 and 1996.

Table 2.3 The decline in tobacco consumption


Year	Adult smokers (%)^a Male Female		Real expenditure per capita on tobacco products 1970 = 100	Quinquennium	Cigarettes/day/adult^b Male Female
1965			92.0	1966-70	10.3	5.8
1970			100.0	1971-75	10.7	6.2
1974	45.3	29.6	88.9
1976	43.3	32.5	88.3	1976-80	9.1	5.7
1980	41.1	31.1	80.8	1981-85	7.7	5.0
1983	40.3	30.8	74.9
1986	33.4	28.9	67.1	1986-90	6.1	4.6
1989	30.0	27.0	63.3
1992	28.3	23.6	54.3	1991-95	5.1	3.6
1995	27.1	23.2	42.8
1996			39.8

(a) Rates age-standardised to the 1986 population.
(b) Cosumption age-standardised to the 1996 population.
Sources: AIHW, 1996; Bardsley and Olekalns, 1999; Taylor and Clements, Annex A.

ABS Census figures and population estimates, Customs and Excise Yearbooks and ABS compilations of Australian Customs Service data

Source: Michelle Scollo

Causes of the decline in tobacco consumption

What caused the decline in tobacco consumption? As we have noted, the prevalence of tobacco smoking fell especially after the introduction of the mass media campaigns in the early 1980s and thereafter. However, other factors, notably price increases may also have been influential factors.

Economic analysis generally explains individual consumption in terms of four main variables: individual income, the price of the product, the price of alternative products, and individual taste or preference. Individual preference may be a function of age or sex, or of education, advertising, information, or addiction. There have been numerous studies of the demand for tobacco using some, or most, of these determining variables. The Industry Commission (1994) reviewed the findings of over a hundred studies of tobacco consumption and we start by summarising their main results. We then report on the findings of some specific examinations of the impacts of public health campaigns on tobacco consumption. Thirdly, we examine two recent studies that have attempted to integrate an analysis of economic effects and public health campaigns.

The Industry Commission (1994) reports on seven Australian studies of price elasticities of demand for tobacco products, as well as on some 50 overseas studies of price elasticities.⁵ All four (then) most recent Australian studies reported an estimated price elasticity of about –0.4 (i.e. for each one per cent increase in the real price of tobacco, consumption falls by 0.4 per cent). Overseas studies give similar results. For example, the UK Department of Health (1994) reported that that the estimated price elasticity in the UK was –0.3 to –0.6. Townsend (1988) reported that that the estimated price elasticity across 19 European countries was –0.4. In a meta study of 41 estimates of price elasticity from 12 countries, Andrews and Franke (1991) found the overall price elasticity was –0.36. More recently, Doran and Sanson-Fisher (1999) reported an overall price elasticity for participation in smoking of –0.38 for Australia (with a price elasticity of –0.44 for males and –0.30 for females).⁶

The Industry Commission (1994) found a much greater range of estimated income elasticities from about 0.1 to 0.9 in Australia and overseas, with a median figure of about 0.4. This implies that a 10 per cent increase in income raises tobacco consumption by 4 per cent. The Industry Commission also found that tobacco company advertising has only a small effect on total tobacco consumption. Two cited studies estimated that a 10 per cent increase in advertising increases tobacco consumption by 0.7 per cent. Corporate advertising of tobacco appears to have more effect on adolescent smoking than on adult smoking. It also affects the brands that people smoke.

With respect to public health issues, the Industry Commission found evidence that:

Education programs have a small effect in stopping teenagers from starting to smoke (for example, Armstrong et al., 1990);
Health information programs (major health warnings) had a significant impact on tobacco consumption in the U.K. (Witt and Pass, 1981).
Advertising bans had a small effect on tobacco consumption in Australia (see Clements et al., 1985 and McLeod, 1986) and overseas;
Bans on smoking in certain locations had a small impact on tobacco consumption (Hocking et al., 1991).

It may be inferred that, overall, these specific programs had a smaller impact than the price changes. However, Pierce et al (1990), reached stronger conclusions on the short and long-term effectiveness of mass-media led anti-smoking campaigns in Australia.⁷ Pierce et al. reported that during the years before the major anti-smoking campaigns in Sydney in 1983 and Melbourne in 1984 there was no observable trend in smoking prevalence in either city. At the start of the campaigns there was an immediate drop of more than two percentage points in male and female smoking prevalence in both cities. Thereafter, there was a decline of about 1.5 percentage points per year among males, but no decline among females, in both cities.

More recently, Tan, Wakefield and Freeman (2000) estimate that, following the 1997 National Tobacco Campaign, overall male smoking prevalence fell by 1.8 percentage points (from 26.5 per cent to 24.7 per cent) and female smoking prevalence fell by 1.4 percentage points (from 23.5 per cent to 22.1 per cent). These estimates are based on large pre- and post-campaign household surveys of smoking prevalence and intentions to quit smoking.

However, these estimates of the impact of the National Tobacco Campaign are accurate only if they make sufficient allowance for smokers who would have quit smoking for other reasons (due to other anti-smoking messages or price increases). Also, some people who quit may revert back to smoking. Although nearly half the population of Victoria attempts to quit smoking in any one year, over 90 per cent relapse back into smoking within the year (Silagy et al., 1996). Pierce et al (1998) found that many of the initial effects of programs to reduce smoking in California did not persist (and recommended continuous programs to maintain the original momentum of change).

The Industry Commission (1994) also observed that there is evidence that smoking is addictive with past consumption affecting present consumption. On the other hand, the Industry Commission made no reference to the price or availability of substitute products, which might influence consumption. There appears to have been little study of the possible role or price of alternative products as a determining factor in tobacco consumption. The implicit assumption is that there are no close substitutes for tobacco, including no addictive substitutes. This assumption may be questionable.

More recently, Bardsley and Olekalns (1999) attempted a comprehensive analysis of the determinants of tobacco smoking in Australia, including the impacts of public health campaigns. Bardsley and Olekalns estimate an equation of the form of Equation (2.1)

C_t = β₀ + β₁C_t-1 + β₂C_t+1 + β₃P_t + β₄Y + β₅X_it (2.1)

where C is tobacco consumption (expenditure per capita), P is tobacco price, Y is income, and X_i are other variables including anti-smoking expenditure, smoking bans, and tobacco advertising. The subscript ‘t’ represents the time period. The β's are parameters to be estimated.⁸

Bardsley and Olekalns conclude that ‘virtually all of the reduction in tobacco consumption can be attributed to tobacco taxes’.⁹ They also find that smoking bans in public places in the workplace reduced tobacco consumption by an estimated 5 per cent since their introduction in 1989, which is a significant amount. Health warnings on cigarette packs reduced consumption by a detectable but small amount. On the other hand, Bardsley and Olekalns find that expenditure on anti-smoking advertising and education and the electronic media ban had no detectable impact in aggregate consumption. They note, however, that anti-smoking campaigns may have affected consumption indirectly by creating acceptance of the tax increases and that it may have affected particular groups in the population.

In a wide-ranging multivariate study of smoking in the United States, Chaloupka and Wechsler (1997) concluded similarly that price changes had a much more significant impact on tobacco consumption than smoking restrictions. They estimated that the price elasticity of demand was about -0.4 for adults and -1.1 for adolescents. On the other hand, they found that smoking restrictions had only a marginal impact on tobacco consumption.

The Bardsley and Olekalns paper has been strongly criticised, for example by Borland (2000) and Chapman (2000), on two main grounds. One is that Bardsley and Olekalns use a questionable behavioural model of rational addiction, in which tobacco consumption at any time depends upon past and future consumption as well as on past and future cigarette prices. The second main concern is that an econometric model of this kind cannot measure the many complex causes and effects from the available data.

In our view, the particular econometric model employed by Bardsley and Olekalns is not a critical issue.¹⁰ Table 2.4 shows some summary data on changes in real per capita expenditure on tobacco consumption and the real price of tobacco. Between 1971 and 1996, the real price of tobacco in Australia rose by 154 per cent and per capita tobacco consumption fell by 60 per cent. As we have seen, many studies have estimated that the price elasticity of demand for tobacco is between -0.3 and -0.6. Under most functional forms, and whatever variables are included, the data used by Bardsley and Olekalns would produce the result that changes in price explain most of the drop in tobacco consumption.¹¹ On the other hand, with the limited data on anti-smoking expenditure available to Bardsley and Olekalns, it is unlikely that a multivariate regression equation of any form would show a relationship between tobacco consumption and anti-smoking expenditure.

Table 2.4 Tobacco consumption and prices

Period	Change in real expenditure on tobacco per capita (%)	Change in real price of tobacco (%)
1963-70	+13.7	-9.5
1971-80	-14.9	+1.7
1981-90	-25.2	+39.9
1991-96	-22.1	+74.7
1971-96	-58.1	+154.1

Source: Bardsley and Olekalns (1999), data provided by authors.

However, Applied Economics concurs with the Borland / Chapman view that the anti-smoking expenditure data employed by Bardsley and Olekalns are unlikely to measure the true magnitude and profile of anti-smoking programs.¹² These data include no expenditure on anti-smoking campaigns between 1975 and 1983 and declining expenditure on anti-smoking campaigns after 1990. Dummy variables were used to represent some discrete policy changes. However, warnings on cigarette packs have been strengthened periodically, bans on smoking in public places have been introduced incrementally and variously in difference States, and bans on tobacco advertising have been introduced and strengthened cumulatively. In effect, these anti-smoking policies have a trending cumulative impact that correlates quite closely with the upward trend in cigarette prices. Given the high correlation between the price changes and cumulative anti-smoking policies, it appears that the Bardsley and Olekalns results may overestimate the price effects and underestimate the policy impacts.

In summary, many factors including Australian public health campaigns were responsible for the very large reduction in tobacco consumption in Australia. Despite many studies of the determinants of tobacco consumption, quantifying these impacts is difficult. It is difficult to separate the impacts of price changes on tobacco consumption from the impacts of other public health programs and the impacts of Australian public health campaigns from the impacts of international research and campaigns on tobacco consumption. It is difficult to distinguish short-term from long-term effects. Also, there has been little research into whether the lower prices and greater availability of potential substitutes has resulted in the substitution of other forms of addictive behaviour for tobacco consumption.

Drawing on the research literature as a whole, we will assume conservatively that Australian public health campaigns (excluding tax changes) were responsible for 10 per cent of the reduction in Australian tobacco consumption. As will be seen in Section 2.7, even if these campaigns were responsible for an even smaller part of the decline in tobacco consumption, the benefits of the campaigns would easily outweigh the costs.

2.5 Health Improvements from Reduced Tobacco Consumption

Three-quarters of all premature deaths due to tobacco consumption are associated with lung cancer, chronic obstructive pulmonary disease (COPD), and coronary heart disease (Mathers et al. 1999). Accordingly, to estimate deaths averted from reduced tobacco consumption, Applied Economics commissioned Dr. Richard Taylor to provide estimates of the reductions in premature deaths from lung cancer, COPD and coronary heart disease. Taylor’s reports are attached as Annexes A, B and C. In this chapter we summarise Taylor’s results for lung cancer and COPD¹³ along with the results for coronary heart disease that are described in more detail in Chapter 3.

The decline in lung cancer deaths

To estimate the effect of lower tobacco consumption on lung cancer deaths, Taylor and Clements develop an age-cohort-period model that relates age and sex-specific rates of lung cancer deaths by cohort in each five-year period to cumulative tobacco consumption. After modelling the impact of tobacco consumption on lung cancer mortality, they estimate the lung cancer deaths that would have occurred under alternative tobacco consumption scenarios.

Specifically, Taylor and Clements estimate an age-sex-period-tobacco consumption (Poisson) model for lung cancer mortality λ_ijk for age i, period j and cohort k for each sex,

λ_ijk= μ + α_i + βX_ijk + γ_j (2.2)

where μ is an intercept term, α_Iand γ_j are the corresponding effects due to age and period and β is the estimate the effect due to cumulative consumption of tobacco (X_ijk). The model was estimated for each five-year period (using average data for each period) from 1921-25 through to 1991-95.

Taylor and Clements (Annex A) estimate lung cancer deaths averted to 1998 due to reduced tobacco consumption as the difference between the estimated deaths that would have occurred if tobacco consumption rates had stayed at 1971-75 levels and actual lung cancer deaths. They estimate lung cancer deaths averted to year 2010 as the difference between estimated deaths with continuing 1971-75 smoking levels and the deaths that would occur (i) with a continuation of 1991-95 smoking rates and (ii) with continuing declines in smoking rates. Our evaluation below draws on the lower estimates of averted deaths based on (i).

Table 2.5 shows estimated actual deaths due to lung cancer for persons aged 30 to 74 and estimates of premature deaths to 2010 assuming a continuation of 1991-96 smoking levels. It also shows estimates of lung cancer deaths for persons aged 30 to 74 that would have occurred with a continuation of 1971-75 smoking prevalence levels.

Taylor and Clements estimate that, with constant 1971-75 levels of tobacco consumption, there would have been 6131 male deaths from lung cancer per annum between 1996 and 2000, compared with an average of 2901 deaths at 1991-95 smoking levels (a fall of 53 per cent). By comparison, at 1971-75 smoking levels, there would have been 2093 female deaths from lung cancer annually from 1996-2000 compared with estimated 1325 deaths at 1991-95 smoking levels (a fall of 37 per cent).

These falls in mortality compare with a fall in per capita tobacco consumption of 59 per cent between 1970 and 1995. There is of course a lag in the decline in lung cancer mortality rates. Therefore, lung cancer continued to increase into the 1980s although tobacco consumption had started to decline.

Table 2.5 Annual deaths due to lung cancer: persons aged 30-74


Period	----------- Males ----------- Actual No fall in Deaths Smoking^a averted			------------- Females ------------ Actual No fall in Deaths Smoking^a averted
1961-65	1591			199
1966-70	2067			301
1971-75	2543	2543	0	434	434	0
1976-80	2900	3070	170	649	642	-7
1981-85	3232	3702	470	862	924	62
1986-90	3236	4458	1222	1051	1267	216
1991-95	3235	5304	2069	1214	1668	454
1996-00	2901^b	6131	3230	1325^b	2093	768
2001-05	2523^b	6971	4448	1397^b	2557	1160
2006-10	2271^b	8097	5826	1493^b	3141	1648
Total^c	114205	201380	87175	46625	63630	17005

(a)Assumes that 1971-75 levels of smoking continue.
(b) The projections to 2010 are based on 1991-95 levels of smoking.
(c) Totals for period 1971-2010.
Source: Taylor and Clements (Annex A).

Chronic obstructive pulmonary disease

Taylor and Clements also estimated the impact of lower tobacco consumption on chronic obstructive pulmonary disease mortality (Annex B). Specifically they examine the impact on chronic obstructive pulmonary disease (COPD, ICD9 496), chronic bronchitis (ICD9 491) and bronchitis, unspecified (ICD9 490). These diseases revealed an epidemic pattern that correlated with tobacco consumption over the last 50 years. On the other hand, no such pattern is observable for asthma (ICD9 493) or emphysema (ICD9 492) and these two diseases were not included in the analysis.¹⁴

In order to estimate the effect of lower tobacco consumption on COPD and bronchitis, Taylor and Clements adopt a similar procedure to that used for lung cancer. They estimate an age-cohort-period model that relates age and sex-specific rates of COPD and bronchitis deaths by cohort in each five-year period to cumulative tobacco consumption, using a Poisson model like Equation (2.2). They then vary the levels of tobacco consumption to estimate the premature deaths that would have occurred under alternative scenarios.

In order to estimate deaths averted to-date due to reduced tobacco consumption, Clements and Taylor estimate the early COPD and bronchitis deaths that would have occurred if tobacco consumption rates had stayed at 1971-75 consumption levels and compare these with actual COPD and bronchitis deaths. To project deaths averted to year 2010, they compare the estimated deaths that would have occurred with 1971-75 smoking levels with the deaths that would occur with (i) 1991-96 smoking levels and (ii) continued declines in current smoking levels. Again, our evaluation is draws on the estimates based on (i).

Table 2.6 shows actual premature deaths due to COPD and bronchitis for persons aged 35 to 74 and estimates of premature deaths to 2010 assuming a continuation of 1991-96 smoking levels. It also shows estimates of COPD and bronchitis deaths for persons aged 35 to 74 that would have occurred with a continuation of 1971-75 smoking prevalence levels.¹⁵

Taylor and Clements estimate that, with constant 1971-75 levels of tobacco consumption, premature male deaths from COPD and bronchitis would have doubled in round numbers from 1500 in 1970 to 3000 in year 2000. Female deaths would have risen sevenfold from 300 in 1970 to 2100 in year 2000. They estimate that the decline in smoking averted 1985 premature male deaths and 1131 female deaths annually between 1996 and 2000. Because of the long-term effects of smoking, estimated deaths averted are significantly higher after year 2000.

Table 2.6 Annual deaths due to COPD and bronchitis: persons aged 35-74


Period	----------- Males ----------- Actual No fall in Deaths Smoking^a averted			--------- Females ------------ Actual No fall in Deaths Smoking^a averted
1971-75	1510	1519	9	299	287	-12
1976-80	1468	1505	37	389	390	1
1981-85	1436	1604	168	505	563	58
1986-90	1369	1857	486	599	819	220
1991-95	1283	2425	1142	979	1526	547
1996-00	1069^b	3054	1985	1024^b	2155	1131
2001-05	642^b	3610	2968	1114^b	3948	2834
2006-10	374^b	4286	3912	1061^b	5862	4801
Total^c	45755	99300	53545	28950	77750	47900

(a) Assumes that 1971-75 levels of smoking continue.
(b) Projections based on 1991-95 levels of smoking.
(c) Totals for 1961-2010.
Source: Taylor and Clements (Annex B).

Coronary heart disease

Between 1968 and 1999, reduced tobacco consumption accounted for 24 per cent of the decline in early male deaths due to coronary heart disease and for 12 per cent of decline in early female deaths (see Chapter 3).

As shown in Table 3.6, lower tobacco consumption averted about 1000 early male deaths annually in the late 1970s and 6000 early deaths a year in the 1990s. Because female smoking did not decline until later than male smoking, there was little decline in early female deaths until the mid-1980s. By the late 1990s, lower smoking reduced early female deaths by 900-1000 a year.

Estimated total deaths averted in 1998

Table 2.7 show deaths of persons aged 30-74 due to lung cancer and of persons aged 35-74 due to COPD/bronchitis and coronary heart disease in 1998. It also shows estimated deaths averted due to the decline in tobacco consumption from 1971-75 levels. Approximately, an estimated 14 500 early deaths in these three disease groups were averted in 1998 due to lower tobacco smoking. This includes 7000 deaths from coronary heart diseases averted, 4000 deaths from lung cancer averted and 3500 deaths from COPD and bronchitis averted.

Other cancers and strokes account for 20 per cent of all tobacco-related deaths (Mathers et al., 1999, Table 7.1). If early deaths for these conditions fell in a similar way to deaths for the estimated conditions, a further 2900 early deaths were averted in 1998. This would produce a total of 17 421 deaths averted in 1998 due to the long-run reduction in tobacco consumption since the early 1970s.

Table 2.7 Deaths averted due to lower tobacco consumption: 1998, persons 35-74 years


Disease	Males Actual Averted Deaths deaths		Females Actual Averted Deaths deaths		Total Actual Averted deaths deaths
Lung cancer^a	3126	3230	1275	768	4401	3998
COPD/bronchitis	1075	1992	957	1589	2032	3581
Coronary heart disease	6650	5949	2499	993	9149	6942
Sub-total	10851	11171	4731	3358	15582	14521
Other cancers and strokes						2900
Total						17421

(a) The age group for lung cancer deaths in this study is 30-74 years olds.

Sources: Annexes A, B and C.

2.6 Valuing the Benefits of Health Improvements

In this section we estimate the benefits that accrued in 1998 due to the health improvements from reduced tobacco consumption. This provides an order of magnitude basis for estimating the benefits from 1970 onwards due to the public health programs (see Section 2.7).

There are three main categories of benefits from the estimated health improvements: longevity, improved quality of life for those relieved of disability, and lower health care costs. The dollar values of each category are estimated below.

Value of years of life gained

In order to estimate the value of longevity gained, estimates are needed for the years gained per early death averted and for the value of these years. Mathers et al. (1999, Table 7.1) indicate that an average of 11.1 years are lost for each lung cancer death attributable to tobacco smoking and that an average of 8.7 years are lost for each COPD and coronary heart disease death attributable to tobacco smoking.

However, these averages include deaths from these diseases at all ages, including deaths in old age. Taylor estimates that persons aged 30 to 74 who die from coronary heart disease lose an average of 11.5 years of life (32 per cent more than 8.7 years). Factoring up the average years of life lost for COPD and lung cancer by a similar proportion (32 per cent), an average of 14.6 years are lost for each early lung cancer death attributable to tobacco smoking and 11.5 years are lost for each early COPD death.

Following the discussion in Chapter 1, we allow $60 000 per year of life gained and apply a central discount rate of 5 per cent. This produces a present value of approximately $530 000 for 12 years of life gained and $620 000 for 15 years of life gained. The estimated total value of longevity gained is $9.6 billion, as shown in Table 2.8.

Table 2.8 Value of years of life and health status gained in 1998

Disease	Early deaths averted	Average years gained	Average value of years gained ($)	Total value of years of life gained ($ million)	Value of improved health status ($ million)	Total value of improved health ($ million)
Lung cancer	3998	15	620,000	2478	223	2701
COPD	3581	12	530,000	1898	906	2804
Coronary heart disease	6942	12	530,000	3679	712	4391
Other cancers, strokes	2900	12	530,000	1537	369	1906
Grand total	17421			9592	2210	11802

Value of improved quality of ,life

In order to estimate improvements in the quality of life, we again draw on Mathers et al. (1999, Table 7.1). This table shows that low health status accounted for 8.2 per cent of estimated total DALYs (disability-adjusted life years) associated with lung cancer attributable to tobacco smoking, 32.3 per cent of total DALYs associated with COPD, and 16.2 per cent of total DALYs associated with coronary heart disease.

Adopting similar factors in this report, the value of health status gained would total $2.2 billion (see Table 2.8). The total value of years of life and health status gained in 1998 would be $11.8 billion.

Savings in health care costs

The estimated total health care cost for all lung cancer cases in 1993-94, inclusive of hospital, medical, pharmaceutical and other costs, was $107 million (Mathers et al, 1999).¹⁶ The total health care costs for COPD and coronary heart disease were $300 million and $894 million. In 1998 dollars, the health care costs were 8.2 per cent higher ($116 million, $325 million and $967 million respectively).¹⁷

There are no data on the reduction in health care costs due to reduced lung cancer, COPD and coronary heart disease. However, approximate health care savings in 1998 can be inferred from the ratio of averted deaths to actual deaths as follows:

Health care savings₉₈ = health care costs₉₃ � 1.082 � (averted deaths / actual all age deaths)₉₈

In 1998, estimated averted lung cancer deaths were 56 per cent of actual all-age deaths due to lung cancer. Estimated averted deaths were 58 per cent of actual deaths due to COPD and 25 per cent of deaths due to coronary health disease.

Using round and conservative numbers, savings of 50 per cent of lung cancer and COPD health care costs would represent savings of $58 million and $162 million for lung cancer and COPD costs respectively in 1998 (in 1998 dollars). A saving of 25 per cent of coronary heart disease costs would represent a saving of $242 million in 1998 (in 1998 dollars). Allowing for relatively minor health care savings in relation to other cancers and strokes, estimated total health care savings in 1998 were conservatively in the order of $500 million.

Total benefits from health improvements in 1998

The total estimated benefit of health improvements in 1998 due to lower tobacco consumption from 1970 onwards is $12.3 billion. This is made up of longevity gains valued at $9.6 billion, improved health status valued at $2.2 billion, and lower health care costs valued at $0.5 billion. In other words, longevity gains account for an estimated 78 per cent of the benefits, improved quality of life for 18 per cent of the benefits, and savings in health care treatment costs for 4 per cent of the benefits.

These results compare health outcomes in 1998 with the outcomes that would have occurred with no decline in smoking after 1970. The results are clearly sensitive to the estimates of health improvements made in Section 2.5, which draw in turn on Annexes A, B and C. However, even if the health benefits of lower tobacco smoking were only half of those estimated, the benefits would still be very large.

2.7 Evaluation of Public Health Programs to Reduce Tobacco Consumption

The evaluation of public health programs to reduce tobacco consumption covers the period 1971 to 2010. The estimated expenditures on the programs from 1983 to 1997 are shown in Table 2.2. For the period from 1971 to 1983, we allow conservatively $5 million per annum in 2000 dollars (which may be higher than they actually were) and for 1998 to 2010 we allow $18 million per annum in line with current expenditures.

As we have seen, the estimated benefits of reduced tobacco consumption were $12.2 billion in 1998. We also concluded in Section 2.4 that 10 per cent of these benefits may be attributed to public health campaigns. Therefore, the estimated benefit in 1998 of the public health campaigns from 1970 was $1.22 billion.

In order to estimate the benefits of public health programs from 1971 to 2010, we project this estimated benefit of $1.22 billion back to 1971 and forward to 2010 in proportion to the estimated averted deaths over this period as shown in Table 2.9. Thus the estimated total benefit in any year, say 1990, equals the estimated total benefit in 1998 ($1.22 billion) � (deaths averted in 1990 / deaths averted in 1998).

Table 2.9 Deaths averted due to lower tobacco consumption^a

Year	Lung cancer	COPD	CHD^b	Total
1975	68	45	821	934
1980	315	184	1401	1900
1985	895	272	2682	3849
1990	1872	1163	6481	9516
1995	3013	2419	6798	12230
1998	3988	3581	6942	14521
2000	4644	4414	7146	16204
2005	6354	6797	7515	20666
2010	8220	10186	7904	26310

(a) Excluding other cancers and strokes.
(b) Coronary heart disease.
Source: Annexes A, B and C.

The results of the evaluation based on these ‘Central Case’ assumptions are shown in Table 2.10. Using a 5 per cent discount rate, the present value of the costs of the public health programs in year 2000 dollars discounted back to 1971 is $176 million. The estimated benefits are $8602 million. The net benefits are therefore $8427 million. This represents a benefit –cost ratio of nearly 50:1.

A comprehensive economic evaluation should take into account various other factors. The decline in tobacco smoking would improve air quality and reduce the adverse effects of passive smoking. On the other hand, smoking regulations would reduce the profits of tobacco producers and provide minor inconvenience to smokers. Thirdly, smokers who give up smoking to enjoy better health may nevertheless suffer from a loss of smoking pleasures. This evaluation quantifies the health benefits, but not the loss of smoking pleasure.¹⁸ Such a comprehensive evaluation is outside the scope of this report.

Sensitivity tests

With a net present value of $8427 million and a benefit-cost ratio of nearly 50:1, no realistic alteration of assumptions could change the basic results. For the sensitivity test shown in Table 2.10, we include only 50 per cent of the health benefits of lower tobacco consumption estimated by Taylor and Clements and allow that public health campaigns were responsible for only 5 per cent of these benefits. These assumptions drop the benefits attributable to the public health campaigns by 75 per cent. However, even on these conservative assumptions, the gross benefits of the campaigns would total $2151 million and the net benefits would be $1975 million.

On the other hand, if benefits and costs are not discounted, the total benefits rise to $33.4 billion and the costs to $0.53 billion, so that net benefits would be $32.9 billion.

Table 2.10 Evaluation of public health programs to reduce tobacco consumption ($m 2000 $s)


Year	Costs	Central Case Benefits Net benefits		Sensitivity test Benefits Net benefits
1971	5.0	0.0	-5.0	0.0	-5.0
1972	5.0	20.0	15.0	5.0	0.0
1973	5.0	40.0	35.0	10.0	5.0
1974	5.0	60.0	55.0	15.0	10.0
1975	5.0	77.2	72.2	19.3	14.3
1976	5.0	92.0	87.0	23.0	18.0
1977	5.0	108.0	103.0	27.0	22.0
1978	5.0	124.0	119.0	31.0	26.0
1979	5.0	140.0	135.0	35.0	30.0
1980	5.0	156.0	151.0	39.0	34.0
1981	5.0	188.0	183.0	47.0	42.0
1982	5.0	224.0	219.0	56.0	51.0
1983	5.0	256.0	251.0	64.0	59.0
1984	7.1	288.0	280.9	72.0	64.9
1985	9.0	318.1	309.1	79.5	70.5
1986	10.2	412.0	401.8	103.0	92.8
1987	11.3	504.0	492.7	126.0	114.7
1988	15.5	596.0	580.5	149.0	133.5
1989	22.0	692.0	670.0	173.0	151.0
1990	22.7	786.4	763.7	196.6	173.9
1991	19.6	828.0	808.4	207.0	187.4
1992	21.8	872.0	850.2	218.0	196.2
1993	17.5	916.0	898.5	229.0	211.5
1994	21.1	960.0	938.9	240.0	218.9
1995	18.8	1010.7	991.9	252.7	233.9
1996	17.9	1072.0	1054.1	268.0	250.1
1997	18.3	1136.0	1117.7	284.0	265.7
1998	18.0	1200.0	1182.0	300.0	282.0
1999	18.0	1268.0	1250.0	317.0	299.0
2000	18.0	1339.1	1321.1	334.8	316.8
2001	18.0	1412.0	1394.0	353.0	335.0
2002	18.0	1488.0	1470.0	372.0	354.0
2003	18.0	1564.0	1546.0	391.0	373.0
2004	18.0	1640.0	1622.0	410.0	392.0
2005	18.0	1707.8	1689.8	427.0	409.0
2006	18.0	1800.0	1782.0	450.0	432.0
2007	18.0	1892.0	1874.0	473.0	455.0
2008	18.0	1984.0	1966.0	496.0	478.0
2009	18.0	2080.0	2062.0	520.0	502.0
2010	18.0	2176.8	2158.8	544.2	526.2

NPV @ 5%	175.7	8602.5	8426.8	2150.6	1974.9
NPV @ 3%	263.9	14370.8	14106.9	3592.7	3328.8
NPV @ 7%	122.6	5360.8	5238.2	1340.2	1217.6

Financial benefits to government

In Section 2.6, we estimated that savings in health care expenditures due to lower tobacco consumption accounted for 4 per cent of the benefits ($0.5 billion out of $12.3 billion) in 1998. Allowing that 4 per cent of the benefits of the public health programs accrue in health care savings, the present value of the expenditure savings in the Central Case would be $344 million. This would provide a saving of about $2 for every $1 of government expenditure on public health programs to reduce tobacco consumption.

2.8 Conclusions

Tobacco consumption has fallen substantially over the last 30 years. Among adult males, smokers fell from 45 to 27 per cent of the population; among adult females, smokers fell from 30 to 23 per cent of the population. Moreover, smokers smoked fewer cigarettes per capita.

The health benefits of reduced tobacco consumption were large. In 1998, an estimated 17 421 premature deaths were averted. These included 6942 deaths coronary heart disease averted, 3998 fewer deaths from lung cancer, 3581 fewer deaths from COPD and bronchitis, and 2900 deaths from strokes and other cancers averted.

The estimated total benefit of health improvements in 1998 due to lower tobacco consumption from 1970 onwards is $12.3 billion. This comprised longevity gains of $9.6 billion, improved health status valued at $2.2 billion, and lower health care costs valued at $0.5 billion.

Over the last 30 years, Australian governments have implemented many public health programs aiming to reduce tobacco consumption, including mass media campaigns and other health warnings and regulations that restrict the promotion of cigarette products and influence the conditions under which cigarettes may be consumed. Also, between 1971 and 1996, changes in taxes contributed to a 154 per cent increase in the real price of cigarettes.

Despite numerous studies of the determinants of tobacco consumption, the contribution of Australian public health programs to reduced tobacco consumption is difficult to quantify. This study assumes conservatively that these programs have contributed to 10 per cent of the decline in tobacco consumption.

For our 'Central Case' evaluation of the public health programs, the estimated present value of the costs of the public health programs in year 2000 dollars discounted back to 1971 is $176 million. The estimated benefits are $8602 million. The net benefits are $8427 million.

For a sensitivity test, we include only 50 per cent of the estimated health benefits and allow that public health campaigns were responsible for only 5 per cent of these benefits. These assumptions drop the benefits attributable to the public health campaigns by 75 per cent. Even on these most conservative assumptions, the gross benefits of the campaigns total $2151 million and the net benefits would be $1975 million.

In terms of public finances, the present value of the expenditure savings for government in the Central Case would be $344 million. This would provide a saving of about $2 for every $1 of expenditure on public health programs to reduce tobacco consumption.

¹ Age-standardisation allows for changes in the numbers of people in different age groups, who have different propensities to smoke. Crude per capita data do not do this. As shown in Section 2.4, cigarette consumption per smoker also fell.

² The real price of a packet of Winfield 25s rose by 47 per cent between 1981 and 1990 and by 67 per cent between 1990 and 1996 (Doran and Sanson-Fisher, 1999).

³ The Centre commissioned Michelle Scollo to collect these data, which Michelle kindly provided to us.

⁴ Michelle Scollo estimates that expenditure on educational anti-smoking programs totalled $0.5 million per annum in the three years to 1974-75, which would be nearly $2.0 million per annum in year 2000 prices. However, these are not complete data on anti-smoking expenditure.

⁵ The price elasticity of demand is the percentage change in consumption that occurs with a one per cent change in price.

⁶ Several studies find that the price elasticity is higher (around –1.0) for young persons. If income is properly accounted for, the lower price elasticity for older people could indicate a greater dependency on tobacco. However, it appears that that the higher price elasticity of younger people reflects in part a failure to allow fully for different incomes.

⁷ The Industry Commission (1994) did not cite Pierce et al. (1990).

⁸ A controversial feature of the Bardsley and Olekalns model is the assumption of rational addiction. In the development of the model, Bardsley and Olekalns argue that tobacco consumption is a function of both past and future consumption and past and future tobacco prices. However, in estimating the model, they do not include past and future tobacco prices as determining variables.

⁹ Bardsley and Olekalns estimate a short-run price elasticity of –0.2 and a long run elasticity of about -0.5 from 1965 to 1990, which is consistent with other research findings. Estimated elasticities rise in the 1990s due, according to the authors, to the choice of functional form.

¹⁰ The dependent variable in the Bardsley and Olekalns model may be non-stationary, in which case the t-statistics may not be reliable. This could be dealt with by using first differences and an error-correction mechanism. It may also be argued that advertising is a public good and should not enter the equation in per capita terms.

¹¹ Using the Bardsley and Olekalns data, tests of alternative regression models indicate that the main results do not depend on the inclusion of past or future consumption (communication from Professor Cooper, University of Western Sydney).

¹² Although Bardsley and Olekalns also used figures for anti-smoking expenditures provided by Michelle Scollo, these figures were lower than those shown in Table 2.2 and had a different pattern in the 1990s.

¹³ Mark Clements was co-author of the reports on lung cancer and COPD.

¹⁴ It is of course possible that the prevalance of asthma and emphysema would have been still higher if tobacco consumption had not fallen.

¹⁵ The analysis for COPD and for coronary heart disease is for persons aged 35 to 74, whereas the subject population for lung cancer is persons 30-74 years.

¹⁶ The AIHW has estimated health care costs in detail for a large number of disease conditions in 1993-94. There are no comparable detailed estimates for more recent years.

¹⁷ Based on the total health price index in AIHW Health Expenditure Bulletin, No.16, p.36.

¹⁸ One reader observed that if 'loss of smoking pleasure' is a cost of quitting, dealing with withdrawal symptoms between cigarettes should be accounted a cost of smoking.

Applied Economics