|
Benefit-cost analysis is a technique to evaluate public policy resource (usually expenditure) decisions. This chapter is a descriptive outline of benefit-cost analysis (also known as cost-benefit analysis; the terms are interchangeable). The next chapter covers discounting, one of the main analytic techniques in benefit-cost analysis, and the following chapter gives a brief discourse on project ranking; that is choosing between projects when budgets are constrained.
The fundamental rule of benefit-cost analysis is:
In any choice situation, we should select the alternative with the greatest nett benefit.
A corollary of this rule is that we should not accept projects with negative net benefits.
This rule is simple, but in application it is difficult.
First, we rarely know all the alternatives. For example, how many different ways are there to fill the transport needs between Sydney and Brisbane? Were there alternatives to Sydney's third runway and Badgerys Creek? We often choose from a limited field, and quite often the decision is simply whether we should proceed with a given project or not; we don't even consider alternatives. In fact benefit-cost analysis is generally the last stage in a process in which we have chosen the ends, and are looking at a limited number of means to serve those ends. But, as those who have studied planning will be aware, means-end confusion is common, and planners are often constrained in their choices.
Second, measurement of costs and benefits is difficult. Stokey and Zeckhauser stress:
All the impacts, both favorable and unfavorable, present and future, on all of society are determined.(1)
That's a big burden, for it encompasses all of time and all of society
Comparing benefits over time is a problem. For example, a construction project which is complete in one year is preferable to one which is complete in two years. But what if that acceleration adds 20 percent to the construction cost? The next chapter covers discounting, which is a way of comparing benefits over time. But though the techniques may be orthodox and rigorous, they still depend on the data fed into the models, which, in turn, depend on forecasting with all its assumptions and pitfalls.
Then there is the issue of 'all of society'? Is a national park worth the same to you as to me? Does redistribution make all of society better off? These enter into issues of utility theory, in itself a major topic. In favor of redistribution from the well-off to the less well-off is the notion of diminishing marginal utility - the more I have the less is each additional unit worth to me - therefore redistribute to those who have less. Utilitarian theory, as espoused by Jeremy Bentham, sees benefits in equality and distribution to achieve greatest satisfaction - 'each to count for one, and none for more than one'. A more restrictive concept is that we can consider that society benefits only if, in making some people better off, no-one is made any worse off. That is Vilfredo Pareto's more cautious approach, which avoids the conceptual difficulty of manipulating utility mathematically across different individuals.
A possible approach to the distribution problem is to see what society reveals in its taxation system about its preferences for distribution. Our highest marginal tax rate is 48.5 percent; low income earners pay no income taz. This could be interpreted as meaning we see a dollar in the hands of a low income earner twice (i.e. 100.0/51.5) as valuable as a dollar in the hands of a high income earner. We might therefore apply weights to benefits using factors derived from marginal tax rates.
But how far should we cast our net when it comes to "all of society" mean. Our own political jurisdiction? Our nation? The whole of humanity? Other species? These are not academic abstractions; with issues like greenhouse gases, acid rain, depletion of fisheries and ozone depletion, they are vital considerations and all cross national boundaries.
An issue in political economy is that for many interventions the parties who benefit are more concentrated and easier to identify than the parties who bear costs. Tariff protection carries benefits for about 100 000 workers in the textiles, clothing, footwear and motor vehicle industries, many of them in Melbourne. Those who pay are 19 million consumers, and an untold number of unemployed who cannot get work because of the costs of tariff protection. (A Pareto approach to tariff reduction may be compensate those who lose their jobs as a result of tariff cuts.) Although benefit-cost analysis will show that the benefits of tariff reduction will outweigh the costs, these benefits are widely dissipated, and the costs are concentrated. Politically, concentrated strong opposition to trade liberalization can overcome a wider community weak preference.(2)
Direct financial outlays and receipts are reasonably easy to trace. Other costs are harder to measure and trace. (Some of these concepts arise again in costing in Chapter 11.) They include:
External Costs and Benefits
Imagine a decision about whether a major road should bypass a city of 20 000 persons. The road carries 1000 trucks and 5000 cars a day.
A bypass would cost around $100 million, and would save ten minutes travelling per vehicle. Is the project worth going ahead with?
On the cost side, the problem is reasonably easy. The cost is the land purchase price plus what is paid for construction. It's on the benefit side that it's difficult. All benefits are external to the road authority - that is, unless they decide to operate the road as a toll road.
The most easily traced benefits are travelling time. If each truck costs $50 an hour to operate, then the saving for each truck will be $8.33. Cars present a harder case. Assuming people are travelling for leisure, then we don't have such a clear set of transactions. But we may be able to approach the problem indirectly. Some people pay others to mow their lawns while they enjoy their own leisure; some are willing to take in ironing for a fee - in other words their time has an opportunity cost. This may be $10 an hour, say. So if each car carries 1.5 people the ten minute saving will be worth $2.50 per car. Adding these benefits ($8.33 * 1000 trucks * 365 days plus $2.50 * 5000 cars * 365) comes to $7.6 million annually. But here we have to be careful; what really is the opportunity cost of people's time? Is it what they pay to have someone work for them, so they can enjoy more leisure (a function replacement cost approach, as outlined above). Or is it what they could earn in the labor market. which may be distorted, for example, by unequal pay for men and women? Is a minute's travelling time for a woman taking a child to school really worth less than a minute's travelling time for a busy executive?
There are other benefits which are real, but harder to measure. If the new bypass is built, pollution will be reduced and safety will be improved. We may get a partial measure from looking at real estate prices in similar situations, and find, say, that they rise by $3 000 per property once a bypass is built. A city of 20 000 will have around 10 000 residences. There's a benefit of $30 million. We still haven't looked at other benefits - reduction in greenhouse gases for example. It may be that we have enough benefits to justify the project without taking these into account.
And here we have to know which community of interests to take into account, which may determine who should ultimately pay for the bypass. In the road example above the city government will have little interest in the road users, for example, but will have an interest in its own people. Should the federal government, which funds the road, take into its account the interests of the city folk? Should it ask the local government to pay a contribution to the project, to be funded from rates on the properties which have risen in value because of the bypass? (If they do, the ratepayers may say, legitimately, that although the value of their properties has risen, they never asked for this to occur, a problem known in economics as free riding.) The city government will argue that local businesses (garages, fast food outlets) will lose form the bypass; the federal government will argue that people still need to buy gasoline and to eat, and all that is happening is a set of transfers from businesses in that city to businesses elsewhere on the road. These transfers cancel themselves out.
If the road saves lives how do we value the savings? One approach is to see what people are willing to pay to save lives in other choice situations. Unfortunately markets give little information; people are willing to go to extraordinary lengths to ensure air safety, for example, but are far less interested in domestic appliance safety to take an opposite extreme. (Rational choice models would suggest we should spend less on air safety and more on appliance safety; more lives could be saved with a given outlay, but communities and their political representatives do not necessarily make rational choices.)
Consumer Surplus
We come across the concept of consumer surplus again when we look at monopoly and pricing in Chapters 14 and 15. It's basically the difference between the price a consumer pays and what he or she would have been willing to pay. In the case of pure public goods there may be no price to the consumer, so the willingness to pay is a complete measure of the benefit.
We might find it through questionnaire: how much are you willing to pay for the ABC? (The answer may depend on how the question is framed; it costs about 8 cents an Australian a day, or $60 per household a year.) In valuing the Gordon below Franklin Wilderness researchers asked Australians how much they would be willing to pay in order to see it preserved. This is an example of contingent valuation. In the road example above we might see how much people are willing to pay for similar benefits; we might, for example, see how many people choose to use Sydney's toll roads to gain similar benefits in comfort and travelling time. This approach, using revealed preferences, is usually less prone to subjective bias than the contingent valuation approach, because we are looking at actual market situations where people make decisions with their own money.
Conservation of Scarce or Under-Valued Goods
In some countries there may be a shortage of foreign exchange, which may be valued in benefit-cost analysis at a higher rate than its official exchange rate. There are cogent arguments that resources like fossil fuel and tropical timber are under-valued. Most commonly these adjustments are handled by factors to scale up market values.
Most probably the Commonwealth's decision in the 1992-93 Budget to stimulate economic activity through construction projects was influenced in part by the consideration that such projects do not make large demands on foreign exchange reserves. By contrast stimulation through encouraging private investment in plant and equipment would place a heavy demand on imports.
Opportunity Costs
These should be taken into account in any investment, commercial or government. There are different options and perspectives in government, however.
It can be hard for a government to value its own assets. The government may be able to create or destroy value through changes in policy - e.g. land re-zoning. For instance, if the Commonwealth decided to go ahead with re-development of the Canberra Hospital site on Acton Peninsula, what is the opportunity cost of the land? If it could be zoned for commercial use, the opportunity cost would be very high.
In the government sector there may be a different decision-maker's perspective when looking at costs. To the government the employment of someone otherwise unemployed may carry a saving on unemployment payments. The Commonwealth's decision to fund a $245 million local capital works program in 1992-93 was probably influenced by such considerations. To consider opportunity costs it is necessary to have a wide perspective. There is a risk that devolution and fragmentation of public budgets is making it difficult for public sector decision makers and their advisers to take opportunity costs properly into account.
Economic and Financial Evaluation
Certain costs and certain benefits are easy to measure. A decision by Elcom whether to buy two 150 MW turbines or three 100 MW turbines is straightforward, and will be carried out in the same way as in a big private corporation. But as we move away from commercial GBEs towards core government activities our problems become more complex. As Apgar and Brown state, public expenditures almost always have effects that extend beyond the budget (but many budgetary theorists overlook this).(3) An all important distinction in this regard is between economic evaluation and financial evaluation. Economic evaluation does attempt to encapsulate all of society; financial evaluation is simply concerned with the budgetary costs in the agency concerned.
For example, a financial analysis may see virtue in reducing budgetary outlays on health care, through encouraging people to take out private insurance. Economic analysis of the same proposition would look at the costs on all of society. How much more expensive is it to collect health insurance premiums than it is to collect taxes? Will cost control be diminished? Will there be more medical over-servicing? (In general, economic analysis of this issue favors funding through the budget.)
At a more local level a hospital may see financial benefits in discharging patients early. But what will be the community costs? What is the time of carers worth, for example? A narrow focus on financial costs runs the risk of cost shifting - from budgetary outlays, on to other parties, being individuals or other tiers of government.
Double-Counting
There are risks of double-counting in any measure of costs and benefits. We may value a pre-school program for poor neighborhoods on the basis of re-distribution objectives. We may be tempted to add the benefit, drawing on published research, that investment in pre-school education reduces the eventual costs of crime, drug abuse and teenage pregnancy. But what if these benefits are already incorporated into society's desire to re-distribute income? In considering shifting a hospital we may do a contingent valuation survey, asking people how much they would be willing to pay to keep the hospital in their area. We may also calculate costs in terms of travel times of patients and visitors. To an extent these two measures will be picking up the same costs.
Cost-Effectiveness Analysis
Quite often it is not possible to put monetary values on the benefits of publicly produced goods and services. The measurement problems may be just too hard.
If we can, somehow, quantify the benefits, it is still possible to use similar decision rules as we use in benefit-cost analysis. If there is one dominant benefit, or if all benefits can be brought to a common measure (not necessarily monetary), then cost-effectiveness analysis is a useful tool.
Problems amenable to cost-effectiveness analysis are those in which:
(1)We wish to get the most benefit for a given level of budgetary outlay, or;
(2)We wish to find the lowest cost means of achieving a given benefit.
As an example of (1) we may have a budgetary appropriation of $1 million to reduce crime. Do we spend it on street lighting, on mobile patrols, or on some combination? (A conceptual methodology for such rationing is outlined in Chapter 10.)
As an example of (2), we may be seeking a budgetary appropriation to bring our crime rates down to a specific level, perhaps that of a comparable city. What funding should we argue for, and how should it be allocated?
Often, of course, it is not possible to quantify benefits. First, the benefits may not be not measurable on any reasonable scale. The community may want to see the Cahill expressway torn down, but what is this ascetic improvement worth? Second, the benefits may be tangible, but not easy to measure before the event. For example, there are real economic benefits in avoiding chemical contamination of beef exports, but in putting a case for quarantine funding, it is hard to make hard ex ante case. Third, there may be multiple benefits, which cannot be brought to a common base.
This situation requires special mention. Imagine, say, there are three proposals to improve access to justice. (Various approaches could be used, such as increased legal aid funding, price control on lawyers, opening new courts, giving more free legal advice etc.) There are two outcomes, reductions in court waiting lists and a reduction in wrongful convictions. The three proposals are evaluated, and, for a given outlay, the benefits are as shown below:
Reduction in court waiting lists (percent). | Reduction in wrongful convictions (percent) | |
Proposal A | 5 | 2 |
Proposal B | 8 | 3 |
Proposal C | 6 | 4 |
Proposal A can be eliminated, as Proposals B and C carry higher benefits on both counts.
That is, they dominate Proposal A, because they are superior in
every dimension. But the choice between B and C is not subject to such mathematical
determinism, because a tradeoff between two incomparable benefits is involved. This
example may seem hypothetical, but the problem has beset community health centers
throughout Australia, which have had both public health and primary care objectives, which
cannot be brought to a single measure. Both aspects are measurable in themselves, but
there is no basis for combining them.
In these cases, the best hope is for informed and enlightened judgement. This is not to suggest an abandonment of rigor; it is often tempting to avoid rigor by citing vague properties such as "intangibles" or "immeasurables". Even if quantification is impossible, the discipline in articulating costs and benefits is well worth the effort.
General References
Jim McMaster and G R Webb Evaluating Government Programs ( CCAE Canberra undated)
Ajit K Dasgupta and D W Pearce Cost-Benefit Analysis: Theory and Practice (Macmillan 1972)
Department of Finance Handbook of Cost-Benefit Analysis (AGPS 1991)
Specific References
1. Edith Stokey & Richard Zeckhauser A Primer for Policy Analysis (W W Norton NY 1978)
2. Gordon Tullock Towards a Mathematics of Finance (University of Michigan Press 1967)
3. William C Apgar and H J Brown Microeconomics and Public Policy (Scott, Foreman and Co 1987)