Bulletin on Narcotics
Volume LII, Nos. 1 and 2, 2000
Economic and social costs of substance abuse
Economic evaluation of policies and programmes: further uses of estimates of the social costs of substance abuseD. COLLINS
Adjunct Professor, Department of Economics, Macquarie University, Sydney, Australia
Senior Lecturer in Health Economics, Faculty of Medicine, University of New South Wales, Sydney, Australia
Social cost estimates and their uses
From aggregate estimates to disaggregated evaluation
General issues in benefit-cost analysis
Economic evaluation of Quit Victoria
Issues in the economic evaluation of a medically supervised injecting centre
The present article identifies the theoretical areas in substance abuse estimation that have not been sufficiently addressed. Those include issues relating to the definition of social costs, a more comprehensive labour market analysis, the social benefits of drug consumption and the distributional impacts of substance abuse. Examples are presented of types of cost estimates, how the results of estimates can be interpreted and the policy use of each type of cost.
Data requirements are identified and the process of proceeding from aggregate estimates to disaggregated evaluation is reviewed. Issues of attribution are considered, and the importance of calculation of avoidable costs of substance abuse is explained.
General issues are reviewed with regard to benefit-cost analysis and evaluation criteria applicable to substance abuse. The article presents as a case study the economic evaluation of Quit Victoria. It uses the calculations of the social costs of tobacco to provide the basis of benefit-cost evaluation of Quit Victoria. The study resulted in a positive benefit-cost ratio under all assumptions.
The article concludes with a review of the issues to be addressed in the economic evaluation of a medically supervised injecting room that is being undertaken in New South Wales, Australia. It emphasizes the importance of estimating social costs in project appraisal and public policy-making.
The objectives of the Third International Symposium on the Economic and Social Costs of Substance Abuse, held in Banff, Canada, from 31 May to 3 June 2000, were set out in the August 1999 planning concept paper of the Canadian Centre on Substance Abuse. The short-term goal was to test and improve the international guidelines formulated by the Centre ; the medium-term goals were to improve the database for cost estimation, to fill existing gaps in methodology and to extend the scope of such studies to include estimates of avoidable costs; and the long-term goal was to expand cost estimation to comprehensive cost-benefit analyses of alternative policy options.
The present paper deals specifically with the long term-goal of the Symposium, the economic evaluation of policy options. Before discussing issues of economic evaluation, however, it should be pointed out that much theoretical and data development remains to be undertaken before estimation of the costs of substance abuse can be considered to have reached a satisfactory level.
Theoretical areas in need of further development include:
(a) Definition of social (external) costs of substance abuse. Consensus has not yet been reached on the definition of social costs, even though it is of crucial importance to the estimation of such costs. For example, the question of addiction and its impact on informed and rational decision-making has yet to be resolved satisfactorily in spite of the popularity among economists of the theory of rational addiction . Another important issue is whether costs imposed by substance abusers on members of their immediate families should be considered to be private or social costs;
(b) Labour market analysis. The issue is whether labour markets fully adjust factor returns so that the costs of substance abuse are internalized upon the abuser and can therefore be treated in some circumstances as private costs. If labour markets do not perfectly adjust in this way (which is more plausible), the question is what proportion of drug-related productivity losses can be considered to be social costs;
(c) Social benefits of drug consumption. Some types of substance use can lead to private and/or social benefits (as is the case with moderate alcohol consumption, which is neither addictive nor abusive). The question is whether such benefits should be set against the costs of substance abuse or whether they should be simply ignored, as they are in most studies;
(d) Distributional impact of substance abuse. The Second International Symposium on the Economic and Social Costs of Substance Abuse, held in Montebello, Canada, from 2 to 5 October 1995, recommended that future studies should indicate the incidence of the social costs of substance abuse among broad community groups (substance abusers, other individuals, business and government). Such a process involves difficult methodological issues akin to those involved in tax incidence analysis. However, from a political point of view, the incidence of the costs of substance abuse is particularly important since such a high proportion of those costs is likely to be borne by individuals and businesses, rather than by government. That area of analysis is in need of substantial further development. There may well be scope for extending it to include an analysis of the incidence of such abuse costs in different household income categories, that is, the distributional consequences of substance abuse.
Areas of primary importance for future data development are:
(a) Development of attribution factors for drug-related crime;
(b) Impact of substance abuse upon productivity (both absenteeism and on-the-job productivity);
(c) Further development of epidemiological information on passive smoking;
(d) Measurement of the costs of the abuse of pharmaceuticals;
(e) Analysis of the impact of tobacco smuggling;
(f) Valuation of intangible costs such as pain, other forms of suffering and loss of life;
(g) Estimation of avoidable costs;
(h) Costs of fires attributable to cigarette smoking;
(i) Costs of drug-related litter and other forms of pollution.
There is no dispute with the primary objective of the Third Symposium, which was to refine the development and use of the social costs of substance abuse. However, the discussion to date has largely ignored the further uses to which the wealth of data generated in the process of producing estimates of social costs can be profitably applied, in particular in the area of project appraisal. Brown and Jackson  used "project appraisal" as a general term referring to attempts by applied welfare economists to evaluate the efficiency of alternative projects or more widely the efficiency of alternative policies. That is the sense in which the term is used in the present article.
Estimation of social costs performs the important function of indicating the size of the substance abuse problem and its component parts. Extending the analysis to estimation of avoidable costs gives valuable information about the potential economic returns available to policies and programmes designed to curb substance abuse. The aggregate estimates have been used in various countries as a powerful political argument for mobilizing public resources to be used against drug abuse. Such aggregate estimates do not, however, indicate to which particular programmes and policies those resources should be directed.
Table 1 presents a summary of various types of exercises involved in the estimation of the social costs of substance abuse, together with the interpretation and significance of each type of exercise.
Table 1. Cost estimates of substance abuse
|Type of estimate||Interpretation of results||Example of policy use|
|Aggregate costs||Total external costs of substance abuse compared with a situation of no substance abuse||Indication of the size of the substance
|Avoidable costs||Potential economic benefits from substance abuse harm minimization strategies||Determination of the appropriate level of resources to be devoted to harm minimization strategies|
|Incidence of costs||Distribution of the external costs of substance abuse among various community groupings||Mobilization of support from various groups (such as the business community) for programmes against substance abuse|
|Disaggregated costs||External costs of substance abuse disaggregated by category||Economic evaluation (cost-benefit or cost-effectiveness analysis) of harm minimization programmes|
|Budgetary impact||Impact of substance abuse on government expenditure and revenue||Assessment of the case for industry making payments to government as compensation for abuse of substances produced by the industry|
|Source: Adapted from J. D. Collins and H. M. Lapsley, "Estimating and disaggregating the social costs of tobacco", The Economics of Tobacco Control: Towards an Optimal Mix, I. Abedian and others, eds. (Cape Town, Applied Fiscal Research Centre, University of Cape Town, 1998).|
Economic evaluation (the topic of the present paper) and, in particular, cost-benefit analysis and cost-effectiveness analysis applied to substance abuse issues are logical developments of the types of analysis referred to in table 1.
Studies estimating the social costs of substance abuse are based on the collection and development of data on all aspects of the economic impact of substance abuse. As an indication, a typical study concerning the aggregate cost of substance abuse would involve the compilation and/or use of data on some or all of the following:
Epidemiological information on substance abuse;
Mortality classified by disease;
Morbidity classified by disease;
Drug use surveys;
Use and abuse of pharmaceuticals and associated costs;
Public and private hospital occupancy;
Public hospital costs of diagnosis-related groups;
Occupancy and costs of public and private nursing homes;
Publicly funded medical services;
Health expenditure and sources of funding;
Labour force absenteeism;
Award rates of pay;
Employee earnings and hours of work;
Household incomes and wages, salaries and supplements;
Company incomes and turnover;
Valuation of pain, suffering and loss of life;
Output of manufacturing industry;
Tobacco and alcohol market shares;
Population size and structure;
Demographic impact of substance abuse;
Income tax and indirect tax revenue;
Subsidies and assistance to relevant industries;
Motor vehicle usage and accidents;
Drug-related crime and law enforcement;
Substance abuse programme costs.
Studies of the aggregate cost of substance abuse provide a high proportion of the data needed for programme evaluation. If the physical output of programmes (for example, reduction in smoking prevalence, liver cirrhosis or drug psychoses) can be identified, aggregate cost studies will normally yield virtually all the data needed for the economic evaluation of the benefits of individual projects.
Such an extension of the use of aggregate data rarely occurs and those project appraisals attempted tend to be undertaken independently of the aggregate cost studies. (A surprising number of evaluations are undertaken by teams that appear to possess little in the way of economic expertise.) A gold mine of economic data is available for project appraisal but appears to be rarely used for such purposes.
A logical development of the Guidelines of the Canadian Centre on Substance Abuse  would be an extension into the area of project appraisal. In the present article, some of the theoretical and practical problems of such evaluation are considered, using examples from Australia with which the authors have been, or are currently, involved.
Aggregate estimates of the social costs of substance abuse give no indication of the benefits potentially available to harm reduction programmes since:
(a) Some of those costs relate to past substance abuse (for example, morbidity attributable to smoking) and are therefore unavoidable costs;
(b) There can be no expectation that it will be possible to completely prevent the abuse of any particular substance. Even for periods well into the future, when the effects of past abuse have "washed out" of the system, it may be possible to reduce the costs of substance abuse but not to eliminate them.
It is necessary therefore to estimate the avoidable costs of substance abuse in order to be able to indicate the extent of potential benefits to harm minimization programmes. Estimates of avoidable costs, however, fail to indicate how those cost reductions might be achieved or whether the social benefits resulting from such programmes would exceed their social costs. Those issues can be settled only by a process of project appraisal.
Project appraisal evaluates the efficiency of alternative projects or alternative policies. Its aims are to determine, by enumerating the benefits and costs of alternative projects or policies, the appropriate level of public resources to be devoted to the problem, and the nature of the solutions to the problem. Its objective is to maximize the social rate of return resulting from the use of public resources so that the resources can be used as efficiently as possible.
Project appraisal is approached from the perspective of the community as a whole, not from the perspective of individuals, firms or the public sector. Such a social perspective complicates the analysis substantially, since private project appraisal avoids many of the difficult theoretical and practical issues that social appraisers must deal with, in particular the valuation of benefits and choice of discount rate. Furthermore, since the viewpoint is that of the community as a whole and not just that of the government, the issues are much more complex than those of public expenditure funding and public revenue benefits.
In principle, the process of project appraisal should lead to the allocation of resources to all public programmes that yield at least a minimum rate of return. That rate of return should take account of the rates of return, calculated on a consistent basis, on investment in the private sector, to ensure an efficient allocation of resources between the private sector and the public sector. In practice, there are many types of public goods and services that the private sector would never supply (unless through private provision facilitated by public funding). Even if it were possible to calculate public and private rates of return on such a consistent basis, political constraints on public expenditure levels mean that the objective of project appraisal is usually to achieve the efficient allocation of previously determined public expenditure levels between competing public sector uses. Governments generally purport to be attempting to reduce the size of the public sector on the grounds that private expenditure is more efficient than public expenditure. The empirical basis for such a generalization is often extremely questionable and it usually appears to be based more on ideology than on analysis. Combined with the equally questionable doctrine that personal and company income tax rates should be reduced as much as possible, it has had a powerful effect on the imposition of constraints on public expenditure. At that level of evaluation, the appropriate evaluation tool is benefit-cost analysis (BCA).
The scope of project appraisal can be extended into programme budgeting, which is a system of managing government expenditure by attempting to compare the programme proposals of all government agencies authorized to achieve similar objectives . It appears that, in the area of health, such lofty objectives are rarely sought, let alone achieved. The paucity of individual project appraisals makes the implementation of programme budgeting seem a distant objective.
In many cases, the objectives of public expenditure analysis may be even more modest. The objective of the evaluation exercise may be predetermined (for example, a reduction of 10 per cent in juvenile smoking prevalence) so that the analysis is reduced to cost comparisons of alternative programmes designed to achieve the same objective. In other situations, it may be considered that it is so difficult to value a programme's output that BCA is impossible. In those circumstances, a cost-effectiveness analysis (CEA) is more appropriate .
CEA can be defined as a detailed comparison of the costs of alternative techniques for achieving the same predetermined objective. In practice, CEA can be used to determine how a given objective can be achieved at the least cost, or how a desired output can be maximized for a given cost. The objectives and outputs of programmes subject to CEA are almost always one-dimensional since, if alternative programmes yield multiple outputs in different ratios, it becomes necessary to assign values to each type of output. That then moves the analysis into the realm of BCA. Murray and others  discuss a broader view of cost-effectiveness in terms of allocating a fixed budget between interventions in such a way as to maximize health in a society. They acknowledge that only a few applications of this broader use-in which a wide range of preventive, curative and rehabilitative interventions that benefit different groups within a population are compared in order to derive implications for the optimal mix of interventions-can be found.
The advantage of CEA in its usual and more limited sense is that there is no need to value output benefits. That makes the analysis much simpler than BCA, since it is necessary to identify only the costs of alternative interventions. That is generally a much more straightforward process than the valuation of programme benefits, even though significant problems may arise in the allocation of overhead costs.
The major disadvantage of CEA is that the policy objective is predetermined rather than arising from the analysis. CEA is of no assistance in determining policy objectives. As Murray and others  pointed out, the implicit assumption of CEA that the required additional resources would need to be transferred from another health intervention or from another sector is rarely discussed.
A further extension of evaluation techniques comes in the form of cost-utility analysis (CUA). While CUA is the least common method of economic evaluation identified in the present article, its use within the health-care sector warrants some discussion. Such a method of analysis calculates the cost per specified health effect (of a programme, a technology or a pharmaceutical intervention) and expresses outcomes as uniform units of health. Those units are presumed to have similar values across all conditions. The health effects are weighted to reflect individual or societal preferences for different health outcomes.
The most common weighting units are quality-adjusted life years (QALYs) and disability-adjusted life years (DALYs). The QALY measurement attempts to compare treatment priorities, identifying and measuring the utility of using resources to treat people of different health status, with different likely outcomes of treatment. QALY is regarded as a unit of health that combines extension of life with a measure of its worth. Its use is focused on societal decisions relating to which good or service to produce relative to one another, that is, allocative efficiency. There have been a number of "league tables" developed comparing QALY measures of quality and quantity of life years gained. It is possible, for example, to compare QALYs from resources spent on smoking cessation programmes with resources spent on organ transplantation.
The DALY measurement combines healthy life years lost because of premature mortality with healthy life years lost because of disability. It is a useful economic tool as the resource implications of each DALY component can be identified and estimated. The total loss of DALYs worldwide reflects the global burden of disease.
CUA can be considered a special form of CEA in which effects are measured in health status; it can contribute to societal decision-making by determining allocative efficiency. While it is still a matter of contention whether QALYs and DALYs can be effective and acceptable public policy tools, they are increasingly being used to contribute to economic evaluation in the health-care sector.
Social benefit-cost analysis attempts to describe and quantify the social benefits and costs of a policy or programme expressed in terms of a common monetary unit. The current value of the flow of social benefits over the life of a project is compared with the current value of the flow of expenditures that have yielded those benefits. Discounting techniques are used to permit comparison of the current value of differential flows of the benefits and expenditures over time. The costs and benefits, once valued in comparable terms, are then compared in terms of a criterion such as a benefit-cost ratio or some measure of the project's rate of return.
The theoretical and practical issues involved in BCA have been discussed at length in the literature and it would be inappropriate to undertake a full review here; however, a brief recapitulation of the major issues is presented below as they form the basis for the discussion by the authors of an example of BCA in the field of drug control.
Valuation of benefits and costs
As indicated earlier, exercises focusing on aggregate costs of substance abuse should identify and place values on all the costs of abuse of the substance under review. Any reduction of those costs resulting from implementation of a particular programme will represent benefits of that programme. The theoretical and practical issues involved in the valuation of programme benefits should already have been addressed in the study on the costs of substance abuse. An approach based on human capital should provide all the necessary information, including discounted future costs. The approach based on demographics will not yield information directly about future substance abuse costs; thus, extra analysis will be necessary.
Estimates of the aggregate costs of substance abuse should already have made the necessary distinctions between private and social costs and should have ensured that double-counting of costs was avoided by including only real (not pecuniary) costs. The estimates should already have taken account of valuation problems, including the impact of private market imperfections such as monopoly power or managed exchange rate regimes, and the difficulties of placing valuations on intangibles such as pain, other forms of suffering and loss of life. (For a review of issues involved in estimating the costs of tobacco use, see J. Lightwood and others ).
It will still be necessary to undertake the extra research involved in identifying programme costs, but that should prove to be a relatively straightforward exercise.
Discounting of benefits and costs
The choice of discount rate, by which the time streams of benefits and costs are to be reduced to values expressed in monetary units in a common year, is one of the major theoretical problems of BCA. It is important for long-lived projects, for which the benefit-cost ratio can be particularly sensitive to the choice of discount rate. For example, it has been concluded that the issue of whether the lifetime health-care costs of smokers are higher than those of non-smokers depends upon the discount rate that is applied .
For a typical public investment project, the higher the discount rate is, the lower the benefit-cost ratio will be. As discussed below, serious theoretical problems remain in the choice of discount rate. That may not be as great a problem, however, for drug programme evaluations as it can be for other forms of public investment analysis.
The most usual application of social BCA is in the evaluation of large public investment projects, such as roads, bridges and power stations, which typically involve a large initial investment, operating costs over the predetermined life of the project and then, possibly, significant expenditure at the end of the project life if, for example, the facility has to be scrapped. The highly uneven time stream of costs is to be compared with a relatively steady flow of benefits. Expenditure patterns over time on drug programmes (such as the Victorian Smoking and Health Program in Australia, discussed below) are quite different, however; there, relatively steady expenditure flows from the beginning lead, after a period of time, to a relatively steady flow of benefits. If the benefit and cost flows of the projects to be compared are relatively steady over time, the ranking of the projects may be relatively insensitive to the choice of discount rate.
Serious issues about the choice of discount rate remain unresolved. Essentially, the choice is between the social opportunity cost rate and the social time preference rate.
The social opportunity cost rate is a measure of the best alternative use to which the community might have put the resources used in the project under review. That reflects the rates of return available to alternative investments, in the private sector or the public sector. The social time preference rate is a measure of the community's valuation of current consumption against future consumption. A society that places the interests of the current generation above those of future generations will have a high social time preference rate. The two rates are, in an imperfectly competitive real world, most unlikely to be the same. One would expect the social time preference rate to be lower than the social opportunity cost rate. The issue of which category of discount rate to choose has not been resolved .
The results of BCA are presented using one or more of the following three measures:
(a) Net present value is the present value of the future time stream of net benefits (that is, benefits less costs ). The future benefits and costs are discounted to present values by use of what is deemed to be an appropriate discount rate. All other things being equal, the higher the discount rate that is chosen, the lower the net present value will be;
(b) Benefit-cost ratio is the ratio of the present value of a programme's benefits to the present value of its costs. Once again, because the process involves discounting the size of the calculated measure, it is crucially dependent on the choice of discount rate;
(c) Internal rate of return is the discount rate that equates the net present value to zero. That measure has the advantage of not being dependent upon a choice of discount rate-the calculation generates the internal rate of return directly. The logic of such a measure is that it effectively represents the rate of return generated by the programme or project.
As a result of different sets of implicit assumptions underlying the calculation of the criteria, the ranking of the projects under review may not be consistent for all three criteria.
In public policy deliberations, most countries take account of distributional considerations, expressed broadly as the extent of poverty and wealth in the community. Different substance abuse programmes may have different distributional impacts. It is possible (although there is no supporting empirical evidence) that the damaging effects of smoking on health are more highly concentrated on the poor than are the effects on health of alcohol abuse. Thus, anti-smoking programmes might have what the community considers to be more desirable distributional consequences than anti-alcohol programmes. BCA can assign distributional weights designed to reflect what are perceived to be society's distributional values, perhaps giving more weight to benefits accruing to the poor than to those accruing to the more affluent.
The authors have recently attempted a benefit-cost analysis of an ongoing programme in Australia called the Victorian Smoking and Health Program, also known as Quit Victoria . The benefit-cost analysis is based on data generated by an aggregate cost study presented in the same paper that also attempts, for the first time in Australia, to estimate the social costs of tobacco use for the state of Victoria. This section of the present article draws heavily upon the methodological discussion in the Quit Victoria report and presents a summary of the results of the analysis. Many of the problems confronted in the Quit Victoria BCA are common to evaluations of programmes in the area of substance abuse.
Quit Victoria was established in 1985 with the objective of reducing the harmful effects of tobacco use in Victoria. Its main source of funding is the Victorian Health Promotion Foundation, which was originally funded through the state tax on tobacco. As a result of the fact that state tobacco taxes have recently been declared to be unconstitutional, the Foundation is now funded through direct grants from the Health Department of Victoria. Other funders of Quit Victoria are the Anti-Cancer Council of Victoria and the National Heart Foundation (Victorian Division). Quit Victoria conducts media campaigns, school and community education and anti-smoking sponsorships and supports legislation designed to reduce tobacco consumption.
Any estimate of the benefits of a public health programme such as Quit Victoria relies upon identification of an implicit mechanism of the link between implementation of the programme and the harm reduction outcome. The mechanism is, in principle, as follows:
Programme -> Rreduction in consumption -> reduction in harm
There are two fundamental links in the above-mentioned mechanism:
(a) Programme leads to a reduction in consumption. This raises two questions:
i(i) What reduction in consumption is attributable to the programme?
(ii) Over what period of time does that reduction occur?
(b) Reduction in consumption leads to reduction in harm. Three questions arise:
ii(i) What reduction in harm is attributable to the reduction in consumption?
i(ii) Over what period of time does the reduction in harm occur?
(iii) Over what period of time does the reduction in costs (whose estimation is the objective of the exercise) occur?
Any such exercise must therefore address the dual issues of quantifying the health effects of anti-smoking programmes and determining the periods of time over which those programmes take effect.
A further significant question is whether programmes such as Quit Victoria produce permanent reductions in smoking prevalence or whether smoking prevalence would return to former levels if such programmes were to be abandoned. Is the effect of programmes such as Quit Victoria like a screw, remaining in its new position when the driving force is removed, or is it like a spring, reverting to its former position? The answer to that question will determine whether there is a need for ongoing anti-smoking programmes or whether such programmes can cease once the desired reduction in smoking prevalence is achieved. The answer is also crucial to the process of estimating the rates of return on anti-smoking programmes.
Expenditures or programmes such as Quit Victoria may lead to positive "externalities", such as reductions in the prevalence of smoking in other states. In estimating the impact of such a programme, account should be taken of such externalities. Other states may benefit from the money spent on the programme by the state of Victoria (for example, when Quit Victoria messages on billboards at the Melbourne Cricket Ground are seen on national cricket telecasts) and Victoria may benefit from expenditure by other states and by the Government of Australia. No attempt was made to estimate those externalities in the Quit Victoria study. In the case of Quit Victoria, all such externalities are positive, that is, the programmes confer benefits (rather than impose costs) on other states. The same is probably true for virtually all harm minimization programmes whose effects spread beyond the jurisdictions funding those programmes.
A related issue is whether the impact of Quit Victoria programmes, in reducing smoking prevalence in one state, could be evaluated using the changes in smoking rates in other states as a comparator. If their prevalence was also reduced, a comparison of Victoria with New South Wales, for example, might well underestimate the effectiveness and positive outcome of expenditure of the Quit Victoria programme.
The calculations of the social costs of tobacco in Victoria in 1988 and 1992  provide the basis for BCA evaluation of Quit Victoria.
Social BCA involves comparison over the life of a programme of its social costs and social benefits. The Quit Victoria study presents estimates of social costs in Victoria for the years 1988 and 1992, on the basis of which an attempt to predict future social costs can be made. The study makes what is deemed to be a conservative assumption that the social costs of tobacco smoking rose linearly between 1988 and 1992 and that, in the absence of anti-smoking programmes, they would remain at 1992 levels, expressed in real terms, in the future.
The problem arose concerning estimation of the reduction in the social costs of tobacco use in Victoria that could be attributed to Quit Victoria expenditure. Generally, the best measure of smoking in any community is taken to be the smoking prevalence rate, that is, the percentage of the relevant population who smoke. Over the life of Quit Victoria, the total smoking prevalence rate in Victoria has declined substantially, mirroring the decline in the national rate. How much of the decline in Victoria is attributable to Quit Victoria?
The smoking prevalence rate in Victoria is subject to a wide range of influences. The major influences can be summarized as follows:
(a) Quit Victoria;
(b) Other anti-smoking policies in Victoria, for example, restrictions on cigarette advertising;
(c) Anti-smoking campaigns of the federal Government and of other state governments;
(d) Federal and state tobacco tax policies;
(e) Other federal and state regulatory policies, for example, sponsorship bans and restrictions on smoking in the workplace;
(f) Broader economic and non-economic factors, such as changes in living standards or in attitudes towards healthy living.
Can such varied influences on the Victorian prevalence rate be quantified separately? Little evaluation work has been undertaken of anti-smoking programmes and policies in Australia. Perhaps the most intensive study in that area is the recently-published evaluation of the National Tobacco Campaign . The results published to date, however, provide no useful quantitative estimates of the Campaign's impact on the national prevalence rate, and little other verifiable information on the topic is available. There is an urgent need for research disaggregating the effects of the various factors influencing prevalence rates; however, it is currently not possible to identify those effects.
The approach adopted in the study was to identify the decline in the prevalence rate in Victoria over the relevant period and to consider the proportion of the decline that could be attributed to the activities of Quit Victoria. The range of attributable proportions adopted in the study was 10-30 per cent, that is, it was assumed that 10-30 per cent of the decline in the prevalence rates in Victoria was attributable to the activities of Quit Victoria. The actual attributable decline is most likely to lie within that range. Chaloupka and Warner  provide international information on the impact of anti-smoking programmes.
Declines in prevalence will lead to reductions in social costs but the lags involved in that process are difficult to identify. A decline in smoking prevalence may lead to a virtually instant decline in some costs, such as those arising from fire-related deaths, injuries and damage. On the other hand, other types of costs may only be responsive to declines in smoking prevalence with a considerable lag. For example, reduced smoking prevalence may lead to a decline in costs related to lung cancer only after a period of many years. It appears impossible, on the basis of currently available research, to estimate the relevant average lag period. For evidence on individual lags, see Armstrong . The Quit Victoria study adopted the approach of making an educated guess of the range of values in which the actual lag might lie and to test the sensitivity of the results to the adoption of different lags. All other things being equal, the longer the assumed lag, the lower the calculated rate of return will be, since the total benefits will be lower and they will accrue later in the life of the programme.
The present study, therefore, assumes that a given proportionate reduction in the prevalence rate will lead to the same proportionate reduction in social costs after a relevant period. Results are calculated on assumed average lags of 6, 8 and 10 years.
There is no problem identifying the annual costs incurred to date in financing Quit Victoria. There is, however, a problem in forecasting expenditure, compounded by the fact that the future prevalence rate will presumably be a function of future expenditure. There are three possible scenarios to be considered:
(a) The "spring" scenario: maintenance of Quit Victoria expenditure at the current real level will keep the smoking prevalence rate at the current levels;
(b) The "screw" scenario: reduction of the Quit Victoria expenditure level to zero will not lead to any change in the current prevalence rate;
(c) The "modified spring" scenario: maintenance of the current expenditure level of Quit Victoria will lead to further declines in prevalence.
There are other possible scenarios but they are likely to be variants of the above-mentioned scenarios.
The scenario assumed to be the least favourable to Quit Victoria is the "spring" scenario. Compared with that scenario, the "screw" scenario would imply lower expenditure for the same benefits while the "modified spring" scenario would imply higher benefits for the same expenditure. The most conservative approach, the "spring" scenario, was used for the present study. Thus, it was assumed that the expenditure level of Quit Victoria would be maintained at the real 1998 level and that the prevalence rate would remain constant at the 1998 level.
From that, an important question arises: over what period should such an analysis for Quit Victoria be undertaken? In investment analysis, it is usually possible to determine the life of the project. In the case of Quit Victoria, the programme has no predetermined life and it was difficult to predict future public policies towards smoking in Victoria. Given that the benefits of Quit Victoria programmes may accrue over many years and that they would certainly continue to accrue for at least a period of time after one programme had ended, the adoption of too short a period of analysis would lead to an underestimation of the programme benefits. On the other hand, it is extremely difficult to predict future developments in medical technology and, thus, future medical costs. Technological improvements may lead to cost reductions, for example, as a result of the development of new vaccines. Alternatively, they may lead to cost increases, for example, as a result of the development of more effective but more expensive medical treatments. The higher the adopted discount rate, the lower the value placed on future benefits will be. For example, one dollar's benefits to be received in 30 years will have a current value of 17 cents at a discount rate of 6 per cent, but only 6 cents at a discount rate of 10 per cent. For the purposes of the Quit Victoria study, a period of analysis of 30 years was adopted.
The absence of adequate research results with regard to the links between programmes like Quit Victoria, smoking prevalence and the social costs of smoking means that a study such as the present one must identify its underlying assumptions and test the sensitivity of the results to changes in those assumptions. In table 2, below, details are presented of the three sets of assumptions underlying the evaluation of Quit Victoria in the present study.
Table 2 shows, for each of the issues identified in the discussion above, the most conservative assumption (in terms of reducing the indicated rate of return to the Quit Victoria programme), the least conservative assumption and the most plausible assumption (in terms of its approximation to reality). Thus, in that context, the term conservative denotes "yielding a comparatively low rate of return".
Table 2. Assumptions underlying the Quit Victoria evaluation
|Set of assumptions||Attributable decline in smoking prevalence (percentage of total decline)||Time lag between reduction in prevalence and reduction in social costs (years)||Characterization of the effect of Quit Victoria expenditure a||Discount rate (percentage)|
|Most conservative||10||10||Acts as a spring||8|
|Least conservative||30||6||Acts as a screw||4|
|Most plausible||20||8||Acts as a spring||6|
|a If the effect of the expenditure on Quit Victoria is assumed to act as a "spring", it is necessary to maintain the Quit Victoria expenditure at the current level in order to maintain the current rate of smoking prevalence. If the expenditure is assumed to act as a "screw", the prevalence rate will remain at its current level even though all Quit Victoria expenditure will cease.|
In table 3, evaluation results are presented on the basis of the three sets of assumptions. The following four evaluation measures are calculated for each set of assumptions:
(a) Net present value in 1987 prices: the value of the time stream of Quit Victoria social benefits less the time stream of its costs with both benefits and costs discounted back to the values in 1987, the year in which Quit Victoria was instituted;
(b) Net present value in 1999 prices: net present value in 1987 prices converted to 1999 prices by the application of the change in the relevant consumer price index between 1987 and 1999;
(c) Benefit-cost ratio: the ratio of the present value of Quit Victoria benefits to the present value of the costs;
(d) Internal rate of return: the discount rate that equates net present value to zero or the benefit-cost ratio to unity; in effect, the social rate of return generated by the programme.
Table 3. Social benefits of Quit Victoria
Net present value
|Set of assumptions||1987 prices||1999 prices||Benefit-cost ratio||Internal rate of return (percentage)|
|Least conservative||1 675||2 416||51.1||55.9|
Two major conclusions can be drawn from table 3.
(a) The evaluation results are extremely sensitive to the choice of assumption sets;
(b) Even on the most conservative set of assumptions, expenditure on Quit Victoria yields extremely high rates of return.
The estimated social benefits of Quit Victoria are high even under the most conservative analysis. The Quit Victoria programme would yield negative net benefits only if the impact of the programme accounts for no more than 1.5 per cent of the overall reduction in smoking prevalence over the relevant period. It appears implausible that the Quit Victoria programme could have such a minimal impact.
The conclusion to be drawn is that it is extremely difficult to conceive of any other public expenditure, outside the area of public health, that would yield social rates of return of the same order of magnitude as those of Quit Victoria.
There seems no doubt that expenditure on Quit Victoria to date has yielded, and will continue to yield, very high social benefits to the residents of Victoria. Presented below are some of the conclusions that can be drawn from the analysis about appropriate future policies towards the programme.
Given the steady decline in smoking prevalence rates in Australia over the last decade, the question that arises is whether there is any reason to believe that it is possible to reduce smoking prevalence further. It would be necessary to determine the minimum smoking, or baseline, rate that could be achieved in a fully informed community at Australia's stage of economic development.
Perhaps the best indication of the baseline rate of Australia is smoking behaviour among general medical practitioners in Australia . Given that zero smoking prevalence is not achievable, the smoking rates of general practitioners could be interpreted as the best achievable in a fully informed population. About 9 per cent of male general practitioners and about 4 per cent of female general practitioners are smokers and only about 4 per cent of doctors under the age of 30 smoke. Surveys of physicians in the United States of America have shown similar results. About 2-3 per cent of medical students in Australia smoke. That may be considered to be the minimum baseline rate, as medical students (because of their age) have all been exposed to tobacco education prior to their decision to smoke. Those figures compare with an actual prevalence rate in Victoria of 25.3 per cent in 1997.
There is substantial scope for continued improvement in the overall prevalence rate, especially if the Australian community were to become fully informed about all relevant aspects of smoking, including not only the health impact of smoking, but also the highly addictive properties of nicotine.
There seems to be no obvious reason to assume that Australia has, in 2000, reached the minimum achievable prevalence rate. With regard to such rates, the "most plausible" evaluation adopted in the present study may well be too conservative. The case for continued Quit Victoria expenditure is extremely strong. Again, if the effect of Quit Victoria expenditure is like a spring, the rate of return to future expenditure would be high, even though it represents a holding operation-preventing the prevalence rate from rising again. Thus, there seems to be every justification for continuing the Quit Victoria programmes in the future.
Given the high calculated rate of return on Quit Victoria expenditure, it would appear that the current level of expenditure on Quit Victoria is far too low. If public resources are being directed to programmes that yield lower marginal rates of return than those attributable to Quit Victoria, then it is clear that resources are being misallocated.
What are the issues to be determined in such a BCA exercise? The following represents the major issues that arose in the Quit Victoria study:
(a) What the aggregate social costs of abuse of the substance under review are;
(b) What programme expenditures, both past and future, amount to;
(c) To what extent the reduction in prevalence could be attributed to the programme;
(d) Whether the effects of anti-abuse programmes are permanent or temporary (can be likened to a screw or to a spring);
(e) Whether the programme confers externalities on other jurisdictions;
(f) What the optimum size of the programme is;
(g) What future medical costs will be;
(h) How long a period the BCA should cover;
(i) What the time lags between reduced prevalence and reduced costs are;
(j) What the discount rate should be;
(k) What the evaluation criterion should be.
The state of New South Wales in Australia has recently passed legislation to enable an 18-month trial of a medically supervised injecting centre, to be situated in an inner-city area in Sydney with high usage of illicit drugs. Heroin will not be supplied, so to that extent the role of the centre can be seen as similar to the provision of clean needles and syringes, as a public health strategy to reduce blood-borne virus transmission. The trial is to be accompanied by a substantial evaluation component involving quantifiable measures, including:
(a) Public health (e.g. drug overdose and blood-borne virus transmission);
(b) Health of clients, (e.g. fewer deaths, utilization of treatment services);
(c) Public amenity (e.g. public injecting and disposal of injecting equipment);
(d) Criminal activity;
(e) Economic analysis, including cost analysis.
The evaluation protocol is still being developed. It is anticipated that, when the centre opens in 2000, it will be accompanied by a comprehensive collection of clinical, epidemiological and economic data. The evaluation is an example of economic data contributing to overall appraisal of a drug policy intervention. The economic analysis to be undertaken for the centre can be directly related to the types of estimates listed in table 1. Firstly, aggregate costs have been used to identify and quantify the extent of the problem, contributing to the political decision to undertake the trial. The second type of estimate, avoidable costs, forms an important component of the economic evaluation, as the anticipated reduction in costs relating to ambulance attendances, treatment of blood-borne viruses and overdose deaths are examples of quantifiable avoidable costs.
The incidence of costs is also relevant to the study. For example, the business community bears some of the costs of substance abuse, while people who live in streets where injecting drug users congregate bear certain costs, including reduced services owing to factors such as public nuisance and lower property values, as the character of their suburb changes.
CEA, requiring disaggregated costs, will identify the extent of harm minimization achieved by the operation of the centre. Some externalities can be anticipated to be identified during the evaluation of the trial phase. After the initial trial, CEA may be useful in making informed decisions relating to economic evaluation of alternative strategies to reduce harm associated with illicit drugs.
The final type of estimate in table 1 is budgetary impact. While expenditure on government-provided programmes is relevant to that category, there is no revenue impact, as heroin is illegal and therefore not taxable.
There are a number of economic issues associated with the trial in addition to those listed in table 1. They include economies of scale (for example, the state of New South Wales is trialling only one centre while the state of Victoria plans to trial five centres), substitution of different levels of health professionals that influence costs (New South Wales, for example, has mandated medical supervision, which is not required in Victoria), and whether exposure to, and uptake of, treatment programmes may result in increased costs to public programmes in the short term.
The present article has addressed what is suggested to be the next stage of the development and use of guidelines concerning the costs of substance abuse. It is envisaged that, as those cost estimates become further refined, their use should also extend into economic evaluation, to enable cost comparisons of policy effectiveness over time and to facilitate economic appraisals of programmes.
The completed case study of Quit Victoria, a cost-benefit evaluation, provides one example of the use of disaggregated cost data, which demonstrates high social benefits from the programme. The second example of economic evaluation, the medically supervised injecting centre, which is yet to be undertaken, involves a range of cost data and a range of specified outcome measures. The economic evaluation of the centre will underline the importance of including cost data and economic criteria in the initial study design.
As economic issues relating to substance abuse and project appraisal of harm reduction strategies become more prominent in public policy-making, the next stage of social cost estimation becomes important. If the extensive data inputs and analytical outputs of studies concerning the costs of substance abuse are not used in project appraisal, substantial (and expensive) data resources will continue to be underutilized.
- E. Single and others, International Guidelines for Estimating the Costs of Substance Abuse (Ottawa , Canadian Centre on Substance Abuse, 1996).
- G. Becker and K. Murphy, "A theory of rational addiction", Journal of Political Economy, 1988, p. 675.
- C. V. Brown and P. M. Jackson, Public Sector Economics (Oxford, Blackwell Publishers, 1990).
- D. N. Hyman, Public Finance: A Contemporary Application of Theory to Policy (London, Dryden Press, 1996).
- C. J. L. Murray and others, "Development of WHO guidelines on generalised cost-effectiveness analysis", GPE discussion paper No. 4, 1999.
- J. Lightwood and others, "Estimating the costs of tobacco use", Tobacco Control in Developing Countries, P. Jha and F. J. Chaloupka (Oxford University Press, 2000).
- J. J. Barendregt, L. Bonneux and P. J. van der Maas, "The health care costs of smoking", New England Journal of Medicine, vol. 337, No. 15 (1997), p. 1052.
- D. J. Collins and H. M. Lapsley, The Social Costs of Tobacco in Victoria and the Social Benefits of Quit Victoria (Melbourne, Australia, Victorian Smoking and Health Program, 1999).
- Australia's National Tobacco Campaign, Evaluation Report, Volume 1 (Commonwealth Department of Health and Aged Care, 1999).
- F. J. Chaloupka and K. E. Warner, The Economics of Smoking, National Bureau of Economic Research Working Paper No. 7047 (Cambridge, Massachusetts, 1999).
- B. K. Armstrong, "Morbidity and mortality in Australia: how much is preventable?", A Handbook of Preventative Medicine, J. McNeill and others, eds. (London, Edward Arnold, 1990).
- M. Winstanley, S. Woodward and N. Walker, Tobacco in Australia: Facts and Issues 1995 (Melbourne, Australia, Victorian Smoking and Health Program, 1995), p. 15.