User Charging in the Federal Government - A Background Document

List of Boxes and Figures

1. Examples of Federal User Charges

2. The "Publicness" of Federal Services

3. Estimating Demand: Contingent Valuation

Figure:

A1. Guidelines for Public Sector Managers on Setting Prices

Acknowledgement

Richard M. Bird and Thomas Tsiopoulos of the International Centre for Tax Studies at the University of Toronto prepared this paper, in consultation with staff of Treasury Board of Canada, Secretariat. This paper sets out the views of prominent practitioners in the fields of economics and public finance on issues such as when user fees are appropriate; what features distinguish public and private goods; and what approaches departments should follow in establishing user fees. In addition, this paper should contribute to a better understanding of user fees and foster better communication between departments and their paying clients.

1. Introduction

This paper has two principal purposes. The first is to present the rationale, from an economic and managerial perspective, for user charges. The second is to set out general guidelines on how to determine where charges should be imposed, and what those charges should be. The primary aim of the paper is to improve understanding of the rationale for this policy, among the public and government officials.

Organization of the Paper

Following the Introduction, Section 2 sets out the economic, policy, and management rationales for introducing user charges to the federal public sector. This section indicates when charges are appropriate and when they are not, and it outlines the principal benefits of imposing correctly designed user charges. A variety of possibilities for financing public sector activity, from general taxation to user charges, are examined, and the circumstances under which each method of financing is most appropriate are described. The characteristics of the goods or services provided are stressed -- for example, a service's inherent "publicness" or its mandatory nature. Finally, the managerial and administrative implications of adopting user charges are discussed.

Section 3 sets out the key principles for determining appropriate charges for different public sector activities, taking into account the factors mentioned in Section 2. It also covers "cost" (in terms of resource use and distribution) of implementing user charges. Attention is paid to the potential benefits of an appropriate charging policy, and to difficulties that may be encountered in determining and implementing appropriate charges for various types of activities. A range of approaches for determining charges, from strict marginal-cost pricing to more market value-oriented strategies, are discussed.

Finally, the importance of establishing an adequate process for determining and reviewing charges is stressed. Charging policies must be appropriate to be accepted by principal stakeholders and the public at large.

Section 4 offers some concluding remarks. Appendix A provides a brief set of "guidelines" for establishing user charges. Appendix B provides additional information on earmarking.

The Changing Environment

Over the last 50 years, the scale and scope of government activity in Canada has increased. Until the mid-1970s, the growth in expenditure at the federal level was financed relatively painlessly by an increasing stream of tax revenues produced by economic expansion. As economic growth slowed, expenditure growth was financed partly by deficits. More recently, federal expenditure growth has not only halted but has been reversed. In the current economic and fiscal environment there is little chance that any significant new expenditures can be financed out of general tax revenues. Indeed, there is, and will likely continue to be, increasing pressure to reduce not only deficits and expenditures but also taxation. In these circumstances, it is important to carefully examine the rationale and structure of all federal expenditures and revenues, including user charge policy.

There is nothing new about user charges at the federal level. They have existed for many years; see Box 1. What is new, however, is the need to ensure--more than ever before--that the right prices are charged for the right activity. This paper is intended to contribute to this goal by making the principles underlying user charges more transparent, more clearly articulated, and better understood. To accomplish this task, three challenges must be met.

The first challenge arises from the perception that, no matter how rationally they are designed, they are "just another tax." Overcoming this can be difficult when: (1) client groups are strongly organized; (2) they can use such general "public policy" arguments as the alleged adverse effects of charges on income distribution; or (3) there appear to be no compensating benefits to offset the new charges. A critical problem in user charge policy is thus how to introduce and communicate such charges to the public at large and especially to those most directly affected. This question is discussed further in sections 3 and 4 of this paper.

The second challenge to user charges is managerial interest and attitude. At first glance, it may not seem to be in themanagerial interest of those responsible for providing a particular public sector service to impose economically efficient charges. If by doing so, they arouse opposition in their clients, this makes their own lives more difficult. If charges also discourage some users of the service, managers will lose clients and have their budgets reduced. At a deeper level, however, proper user charging can and should form part of a new managerial approach, delivering public services efficiently and effectively in today's changed economic and fiscal environment.

Client responses to user charges can provide vital information to public sector managers as to what they should be doing and how. A new dynamic of "client responsive" management may thus be created, particularly where agency budgets are made revenue dependent in some way. This point is discussed further in Section 2.

The third challenge, and the main subject of the paper, is the technical task of determining what user charges should be imposed. Section 2 discusses in broad terms the public sector activities that should be financed in whole or in part by user charges. Section 3 discusses how to determine user charges for different types of services, and Appendix A provides general guidelines for setting the user charges suggested in this paper. It should be noted that each specific public sector activity requires careful and detailed analysis to determine whether charges should be imposed and, if so, what those charges should be.

2. Why Charge for Public Services?

To some Canadians, the idea of charging for public services seems ridiculous. When an activity is carried out by the public sector, presumably Canadians have indicated through the political system that they are not willing to rely on the market for this service. If something could and should be priced as if it were provided by the private sector, presumably it would be so provided. If it is not, there must be a good reason for the public sector to be providing it. In the absence of persuasive argument to the contrary, attempts to impose charges or to raise prices for services provided below cost are likely to be viewed as little more than "revenue grabs."

Such views may be widespread, but they are also misconceived. The federal government is engaged in a vast array of programs. Some of these programs cannot and should not be "priced." But "prices" are already charged for many federal services -- see Box 1 for examples. For many others there are good reasons why they should be priced. Moreover, there are equally good reasons why some charges already in place should be significantly changed.

The basic principle underlying this paper is the precise opposite of the common view cited above. It is that, whenever possible and desirable, public services should be charged for rather than given away. Section 3 sets out how public prices may be determined. The balance of the present section discusses why and when such prices should be charged.

Economic rationale

The primary economic reason for user charges being levied on the direct recipients of public services is to improve the efficiency with which Canadians in general and federal public agencies in particular make use of the resources they have available. Making efficient use of scarce public resources is especially important now, when general fund financing is decreasingly available.

But while user charges may be advocated -- or attacked -- as a potential additional source of revenue, their main economic rationale is not to produce revenue. Rather, it is to promote economic efficiency by providing information to public sector suppliers on how much clients are actually willing to pay for particular services and by ensuring that the public sector supply is valued at least at (marginal) cost by citizens. In this important sense, the objective of effective public pricing policy is not so much recovering costs as it is improving the efficiency with which government uses its resources.

Efficient does not necessarily mean "unfair". User charges may, however, be perceived as having unfair effects on lower income Canadians. The truth of this perception can only be determined by a case-by-case review, but there is some evidence that it is incorrect. The reality and perception of fairness will contribute to the public's appreciation of the equity and efficiency effects of the government's user charges

Political and managerial rationale

Accountability is the key to improved public sector performance, and information is the key to accountability. Efficiency requires that scarce resources be put to their best possible use. One of the most important sources of information for determining efficiency comes from prices. Proper user charges can significantly improve accountability by making clients aware of the costs of the services they receive and managers aware of the benefits and costs of the services they provide.

To gain their support for efficient government, the various user groups need to know what is done, how well it is done, how much it costs, and who pays for it. If users are asked to pay, they should have a say in what programs are delivered, and how. If they are dissatisfied, they will press for greater efficiency and cost effectiveness, explicitly through protests and public actions and implicitly by reducing their use of the services in question. Ideally, imposing user charges will show the real value of public sector services. It will create a new dynamic -- a "market test" -- which forces managers to respond more directly to clients and to allocate their resources as efficiently as possible.

What Services Should be Charged for?

Determining the proper domain and design for user charges is a challenging task. Certain important public sector activities cannot and should not be financed in this manner. Others may be. Still others should be. The characteristics of the activities in question and the nature of the "market" for the service will help determine the appropriate domain. To some extent, the continuum between purely "public" goods and purely "private" goods is matched by a similar one between general fund financing by taxes on one hand and user charges on the other. These two aspects -- the nature of services and the choice of financing method -- are discussed next. In addition, other dimensions of the pricing question that do not fit neatly into this framework are also considered.

- Characteristics of publicly provided goods and services

The federal government provides an enormous variety of goods and services to Canadians. Some of these services should be priced, some should not, and many fall between the two extremes. Where on this continuum particular federal activities fall depends on the characteristics of the services (or goods) provided. At least six distinct characteristics may be relevant in this respect:

One of the main economic rationales for government is to provide citizens with the public goods that they wish, and it is not surprising that many federal activities demonstrate marked "non-rival" characteristics. Box 2 shows federal activities at many points along a continuum that includes goods and services that are purely public at one end and purely private at the other.

A traditional argument for public provision of certain services in Canada has been the sheer size of the initial investment required. With unit costs decreasing as scale increases, efficient private sector provision of such services can be difficult to achieve, because efficient pricing in decreasing-cost industries will not recover full costs.

Technology may alter some of these historical arguments. For example, it is now technically possible to impose efficient road pricing through "smart cards" and similar electronic means.

Education offers an important example of a positive externality, but a better case in the federal sphere concerns telecommunications and transport systems. As each new user is linked to a network, all users presumably benefit as the range of potential interchange is expanded. This is a positive externality. User charges may be appropriate here, but they must be designed with such externalities in mind. How this may be done is set out in Section 3 below.

Negative externalities may also be associated with certain federal programs or private sector activities -- for example, noise, congestion, and pollution in the case of transport. Sometimes public provision of certain activities is justified in part as a controlling or regulating force. In other instances, taxes, prices, or some other form of regulating private activities may be appropriate. Environmental pollution provides a perfect example. Pricing and "user" charges may play an important role in regulating pollution and similar negative externalities.

Most government programs have some redistributive effect, often deliberate. Here again, there seems little reason to impose user charges -- at least with respect to the distributive aspect of a program. For example, a policy's stated objective may be to ensure that every Canadian resident, no matter where he or she lives, can mail a letter of a certain size and weight to any place in the country for the same price. The fact that it costs 10 times more to move the mail to and from one place than to move it to and from another is simply not relevant.

Even with transportation and communication, one of the easiest sectors to price in principle, such objectives have traditionally been important in Canada. The critical question here, however, is whether it is most efficient to achieve such objectives by "cross-subsidization" -- for example, through average-cost pricing, which overcharges some customers to subsidize others -- or to provide incentives to efficient management. Such incentives might involve estimating the extra costs of providing efficient service to high-cost clients and paying a lump sum subsidy to the operating agency to cover this amount. Clearly, the latter approach is better in principle. It is easier to levy user charges for government activities that are not primarily "socio-political" in nature.

- Market structure

An economically efficient price is one that would be charged in a perfectly competitive market. In such a market there are many buyers and many sellers who have full information about the price and cost of an item. They also know about possible substitutes and complementary products. Moreover, the prices of all products fully reflect all external costs and benefits. Finally, any distributional issues are assumed to be dealt with in other ways (e.g., by lump-sum transfers).

Almost by definition, not all these conditions are satisfied when it comes to publicly provided goods and services. Of course, some of them -- such as the competitiveness of markets -- are often not fully satisfied in the private sector either. But the problem is more serious in the public sector. As discussed earlier, the economic rationale for public sector activities is largely based on some or all of these conditions for market efficiency being violated.

The fact that many public sector activitiesare provided by amonopoly supplier is important when it comes to setting user charges. Either the scale and sunk cost factors mentioned above create a so-called "natural" monopoly -- that is, given technology, it is too costly for competitors to emerge -- or public policy itself has created a legal monopoly. This is the case with many regulatory programs. Efficient pricing in a monopoly can be a double-edged sword. From the perspective of the monopoly itself, the most efficient prices are those that extract the maximum return. Such pricing will be discriminatory -- imposing higher prices on those least able to avoid them -- and will result in monopoly output being less than socially desirable. The owners will, however, extract the maximum "rent" from customers.

Preventing such undesirable outcomes is an important rationale for public involvement in providing some services (and in regulating others). In principle, public monopolies should set prices as though they were operating in a fully competitive market. In practice however, the task of determining what those prices should be, let alone implementing them, is a difficult one.

A particular problem in many public services is that they are mandatory. If meat sold in Canada must by law be federally inspected, determining whether fees should be charged and designing such fees is quite different from setting the price of landing rights at a given time of day at a particular airport. One may choose not to land at that airport at that time, but one cannot legally choose to sell (or buy) uninspected meat. Does the degree of choice or compulsion enter into the calculation of what to charge? If so, how? Such questions arise with many potentially "chargeable" services provided by the federal public sector. How best to answer them is a challenging task, discussed further in Section 3.

A related but distinct issue concerns the elasticity of demand for a particular service. If a service is mandatory, all those engaged in a certain activity, or with certain characteristics, must use it. Of course, even where people must use a government service, user charges may still be justified. Why should a service, even a mandatory one, be paid for by someone other than the user (assuming the user benefits from the service)? But such charges have no efficiency rationale because, by definition, they do not affect choices or activities. When demand is inelastic, the task of determining appropriate user charges is thus quite different. Furthermore, in establishing charges, there is no market mechanism to guide decision-making. Users will need to be consulted to ensure that appropriate service is being provided.

There is another important attribute of many government activities which relates to their inherently monopolistic nature. Whenever government has a legal monopoly -- for example, to determine the allocation of the electromagnetic spectrum or the exploitation of a natural resource -- it controls an asset that may be of considerable value to private entrepreneurs who want to land planes, send out television signals, or drill for oil. The value in these situations may exceed the cost of making these resources available. Therefore, as far as government is concerned, any returns from allocating such rights to private parties are pure profit -- or what is often called economic rent.

Appropriate charges in these circumstances extract the maximum rent (profit), while keeping unproductive "rent-seeking" activities (such as lobbying) to a minimum. Given the lack of any good market information on how much people would pay for access to such rights, some form of competitive bidding or auction is probably the best approach to pricing "rentable" resources. This is assuming there are no overriding public policy arguments against selling to the highest bidder.

- Taxonomy

Government services can be financed in a variety of ways. At one extreme, public sector expenditures may be financed through general taxes. At the other, expenditures on a particular service may be financed entirely by user charges. There are three broad classes of major government revenues.

In some instances -- for example, with respect to economic rent -- the appropriate price may be "what the market will bear" and the task is to determine the best way of finding this out, i.e. of "mapping demand." In other instances, public policy considerations (externalities, socio-political objectives, specific characteristics of consumption or production) or informational problems may result in user charges that differ from the competitive level. For example, if a publicly provided service confers external benefits on non-users, as in the case of an immunization program, user charges less than marginal costs are warranted. Any shortfall between total costs and revenues collected from user charges would be covered out of general revenues. It is this form of user charge that is the focus of Section 3 of this paper.

Managerial and Administrative Implications

The taxonomy of public services and financing policies sketched in this section tells only part of the story. To complete this introduction to the appropriate domain for cost recovery and user charges in the federal public sector, brief mention is made of three other dimensions of the problem. The first concerns the relationship between user charges and privatization; the second, the relationship between user charges and earmarking; and the third, the relationship between user charges and fairness or equity. The balance of this section takes up each of these points in turn.

- Moving to a more market-oriented approach

If the public sector is providing a service that could be equally well provided by the private sector, the activity in question should probably be privatized. This conclusion may be valid even when supporting public policies are needed. Examples of such policies include regulation of monopoly providers, taxation to discourage negative externalities, subsidization to encourage positive externalities. Other examples include socio-political objectives such as supporting particular regions or offsetting adjustments in general tax or transfer systems for distributive purposes.

On the other hand, there are many good reasons why some activities that theoretically could be carried out by the private sector are now carried out by the public sector. For instance, the qualifications mentioned in the preceding paragraph may make it unlikely that a private firm could offer the same service as a public provider; the latter may be preferred for this reason. In such instances, there may still be a role for user charges, but such charges would be unlikely to recover full costs or match the prices of an unregulated competitive private provider.

From one perspective, full privatization (or "commercialization") may seem the logical conclusion to applying user charge principles in the federal public sector. From another perspective, however, the possibility of replenishing dwindling budgets with user charges seems to be a desirable alternative to privatization. More broadly, however, user charges and privatization are best seen as just two of many approaches to the same policy objective -- getting the most for Canadians out of the resources at their disposal.

- Should specific revenues and expenditures be linked?

A related issue concerns linkages between revenues and expenditures. In principle, user charges (and similar) revenue should be earmarked. Earmarking may be formal, as in the case of an enterprise self-financed through mechanisms such as vote-netting or a revolving fund. In a less formal way, decisions on appropriations for specific programs will generally take into account the revenue produced by particular activities.

The importance of fairness

Besides helping achieve economic efficiency, appropriately designed user charges may also constitute an important element in achieving "fair taxation." While user charges are often perceived as having adverse distributional effects, on close examination, this perception is too simplistic. What happens in the absence of adequate user charges? Who really benefits from a zero-price policy? These questions need to be answered before the "fairness" of imposing user charges for any particular program can be assessed.

Contrary to common belief, income profiles of consumers of a large range of government services suggest that upper-income households typically benefit disproportionately from consumption of free, or low-cost, public services. For example, low storage and landing fees at local airports and docks disproportionately benefit the few households that own private airplanes and yachts, and subsidized higher education mostly benefits upper-income families. No one should presume that zero-priced services are necessarily the best way to redistribute wealth.

Similarly, if a user charge to cover marginal costs is introduced, fewer general taxes -- a significant share of which are paid by the poor and which carry their own economic costs -- would have to be collected. The result of introducing user charges in this case would not be to raise extra revenue but rather, by rationing the demand for public services, to reduce the size of the public sector. Moreover, such user charges might even (as discussed further in Section 3) extract proportionately more from the rich than from the poor. In such instances -- and there may be many of them -- user charges might improve both efficiency and equity.

3. Designing User Charges

A policy that emphasizes user charges is not designed to obtain revenue, or even to obtain revenue in a way perceived to be fair. Rather, it is designed to redefine government priorities by incorporating more feedback from citizens with respect to the services they want and will pay for. From this perspective, what matters is not just to impose charges but to levy the correct charges.

How can a public sector manager determine the right charge for a particular service? As indicated in Section 2, the answer depends very much on the nature of the service and the structure of the relevant market.

Cost-based Charges

The primary economic rationale for user charges is to encourage efficient use of resources within the public sector. Economic theory indicates clearly that the "best" charges from this perspective are those that are equal to the marginal cost of supplying the good or service in question. If user charges are set below marginal costs, and demand is "elastic" (i.e., sensitive to prices) society will consume more of a service than it otherwise would. But this means society will be worse off as a result of such below-cost pricing. Charging a price lower (or higher) than marginal cost will result in too much (or too little) of a service being produced, from the perspective of what society wants as a whole.

- Efficient pricing

The rule for efficient pricing is simple: prices (or user charges) should equal marginal costs. However, implementing this simple principle can be surprisingly difficult, for two major reasons. First, it may be very difficult to properly define "costs" in many public sector activities. Second, even if such costs can be clearly defined, it may be difficult or impossible to estimate them in the quantitative terms needed to determine appropriate user charges.

"Costs" as viewed by economists advocating marginal-cost pricing are quite different from the costs with which public sector managers, even those operating activities already structured as "cost centres," are likely to be familiar. The notion of cost in financial accounting is simply identifiable monetary outlays incurred in the process of carrying out a particular activity. Such costs include wages, rent, utilities, transportation, supplies, etc. The figures on such costs reflect actual financial outlays made in some particular time period. It can be difficult to allocate such costs to any particular service activity. For example, what proportion of the costs of the director's office should be attributed to delivery of a particular "front-line" service? Cost accountants have developed various methods, albeit sometimes arbitrary, to deal with the problem of allocating such joint or overhead costs.

Economic costs are not simply accounting costs. As already indicated, the fundamental economic concept of cost is opportunity, or the value of the benefits that could have been obtained had the inputs been used for another purpose. From this perspective, the cost of a park does not simply consist of tangible, recorded construction and operating costs. Instead, the relevant cost is the (highest) value that the land could have realized had it been used for some other purpose, such as logging or residential development.

Effective marginal cost pricing thus requires estimating social or opportunity costs. Even in principle this may be a difficult task. Estimating the marginal cost of another unit of a particular service requires identifying all additional costs resulting from this incremental expansion. Adding an extra flight to an already busy airport, for example, gives rise to congestion costs by imposing delays on other flights. Congestion in transport facilities clearly gives rise to real social costs, but it may be hard to convert such costs of waiting into monetary values. This must be done, however, in the process of determining efficient prices.

Even without conceptual problems, there may be problems of measurability. Relevant market prices can often be found -- for example, the value of land used for a park if it were to be logged, or built on. However, such prices can be used to measure socially relevant marginal costs only if it can be assumed that market prices are good approximations to marginal cost prices. For this assumption to be valid, the market from which the information is derived should be close to perfectly competitive -- that is, private sellers set prices close to marginal costs (including a "normal" rate of return on capital).

If marginal cost-based user charges can be calculated, and applied, by definition the amount charged for providing an additional unit of any particular service will just equal the benefit society as a whole derives from providing that unit. In practice, however, it is often necessary to estimate the effects of price changes on demand to determine the desirable level of such charges. While there are various methods of obtaining such demand information -- for example, by using market information on substitute or complementary activities (e.g., travel costs in the case of recreational facilities) or by carefully structured surveys (e.g., contingent valuation analysis) -- it is generally more difficult to get reliable information on demand than on cost (see Box 3).

Several additional issues must be settled. One of the most important is whether "fixed costs" (investment costs) are to be included. In principle, short-run marginal cost (SRMC) prices should be imposed to ensure efficient allocation of resources. Where such charges are used, it must be assumed that the size of a facility is optimal. This condition requires one of two improbabilities: the initial investment decision was based on correct social cost-benefit principles, taking pricing into account or the SRMC prices will be altered as usage changes. That is, when there is excess capacity, no charge will be imposed, but as usage levels rise, user charges will reflect increasing congestion costs. Thus, when it comes time to replace the facility, the funds will be available to do so. The problem with this solution is twofold. First, as noted earlier, congestion costs are difficult to estimate. Second, even if such costs can be estimated, established user charges tend to be difficult to alter without major political and administrative effort. Such efforts are unlikely to be successful in raising prices to pay for a deteriorating level of service. However, if long run marginal costs (LRMC) are used in the initial pricing calculation, overbuilt facilities will be even more under-used and managers will not have demand information with which to work.

Many public activities face decreasing average costs. That is, as output expands, the cost per unit falls. If average costs are decreasing, then marginal costs must, by definition, be below average costs. Facilities with large "sunk" costs and high economies of scale, such as hydro-electric plants and telecommunication networks, are examples. Applying marginal cost pricing in such cases means the user charge will be less than average cost, which in turn means that efficient charges will result in financial deficits.

How can such deficits be financed? One obvious solution is to finance them by general taxes. But why should beneficiaries of a service be subsidized by taxpayers in general? An obvious alternative is to set prices at average rather than marginal costs. All costs would then be recovered, but output would presumably decline. Other solutions to the problem of reconciling efficiency with recovery of financial costs must therefore be sought. None of these solutions are as economically desirable as SRMC pricing, but some may be easier to implement.

- Approximations to efficient pricing

In Canada, at all levels of government, average cost pricing is probably the most common way of setting user charges. Moreover, the "costs" taken into account are usually financial, not social, as discussed above. Presumably, this inefficient practice is followed because such prices are easier to calculate and compelling to client groups, notwithstanding economists' preference for the marginal cost-pricing approach. Such reasoning is understandable, and in many cases average-cost pricing is acceptable. However, a number of approximations to marginal cost pricing are not unduly complex and should also be considered.

The average-cost-pricing approach estimates the total financial cost of providing a particular service and divides this total by the number of units currently provided. This yields the per unit price, and hence user charge. There are two major problems with this approach. First, as already stressed, the result will not be efficient. If costs decline as output increases, the price will be too high and the output less than society really wants. If costs increase as output expands, the price will be too low, too much of the service in question will be demanded and resources will again be misallocated. This is especially true if the "excess" demand at the unduly low price is taken as an indication that output should be increased -- for example, by new investment. Only if unit costs are constant, so that marginal and average costs are equal, is average cost pricing efficient.

As mentioned earlier, the second problem with average-cost pricing (assuming costs are not constant) is that setting such prices effectively requires estimating responsiveness of demand to price changes. Suppose a particular service, such as a national park, has previously been supplied free, or at a charge well below a realistic price. Suppose also that the variable costs of operating that particular park -- that is, the costs to keep it open and operating at some level -- can be estimated without great difficulty, and the number of persons using the park is also known. Clearly, dividing those costs by that number of people is unlikely to result in full "cost-recovery," even of the variable costs, because fewer people will use the park if the price of doing so is substantially increased.

Setting average prices to recover operating costs thus requires not only information on how unit costs change as the number of users change but also on how the number of users will change as prices change. Such information is not easy to obtain, and even if it is available, the resulting prices are not likely to provide an efficient allocation of resources (because of the first problem mentioned above) or any useful information on whether access to the park should be extended or eliminated (because the "fixed cost" aspect has not been factored into the calculation).

As mentioned earlier, the second problem with average-cost pricing (assuming costs are not constant) is that setting such prices effectively requires estimating responsiveness of demand to price changes. Suppose a particular service, such as a national park, has previously been supplied free, or at a charge well below a realistic price. Suppose also that the variable costs of operating that particular park -- that is, the costs to keep it open and operating at some level -- can be estimated without great difficulty, and the number of persons using the park is also known. Clearly, dividing those costs by that number of people is unlikely to result in full "cost-recovery," even of the variable costs, because fewer people will use the park if the price of doing so is substantially increased.

Setting average prices to recover operating costs thus requires not only information on how unit costs change as the number of users change but also on how the number of users will change as prices change. Such information is not easy to obtain, and even if it is available, the resulting prices are not likely to provide an efficient allocation of resources (because of the first problem mentioned above) or any useful information on whether access to the park should be extended or eliminated (because the "fixed cost" aspect has not been factored into the calculation).

Average incremental cost (AIC) pricing may be a useful compromise approach. This approach calculates the costs incurred by an additional user -- like marginal cost pricing -- but it is designed to result in full cost recovery (like some versions of average cost pricing) and to be computationally feasible in the real world faced by public sector managers. This approach simply allocates each element of cost, fixed and variable, financial and (to the extent readily measurable) social, to a particular incremental decision and then assigns to each additional user the incremental cost attributable, on average, to his or her usage.

To offer an example, (based on Haritos and Hildebrand, Canadian Transportation Commission, 1973) when a passenger takes a flight at a particular time from one airport to another, the costs may be broken down into those arising from the addition of one person at this time, at this place (e.g., congestion), those attributable to the time and place (e.g., operating the particular gate and runway), those attributable to the place (e.g., building the gate and runway), and those attributable to the trip (e.g., building and operating the airport). An approximation to efficient pricing in this case might be some combination of a time penalty at peak times, appropriate charges for cost recovery for operating ground facilities, gates and runways, and an access charge to recover the fixed cost of airports and airways. Such charges could be levied in part on passengers (terminal fees, earmarked ticket taxes), in part on airlines (gate and runway charges, varied by time), and in part on both (the costs of fixed installations) through earmarked taxes on aviation fuel, time-sensitive landing fees, etc.

In each of these cases, the appropriate user charge could be calculated on available accounting information, supplemented by information on important social costs (noise, pollution, congestion), and the resulting charges imposed on users (passengers, airplanes, airlines) as a class on an average basis. This system by no means amounts to marginal cost pricing in the strict sense, but it may be as close as one can get in practice.

The multi-part tariff is a somewhat similar approach to pricing. In their simplest form, multi-part tariffs impose a fixed access charge -- for example, for connecting to a network -- and then an additional charge related directly to the amount of use made of the system. Ideally, this last charge should be close to a marginal cost price. Such two-part tariffs are used by some public utilities to recover the deficit arising from the fact that total average incremental costs exceed marginal cost-based user charges.

When it is easy to identify the characteristics of users of a public service, multi-part tariffs -- often called discriminatory prices -- may be useful in achieving efficiency. Suppose, for example, that responsiveness to price change by two distinct groups of users (say, general and commercial aviation or sports and commercial fishing boats) is known, and is different. The less elastic the demand, the higher the price that should be charged on efficiency grounds (such inverse elasticity pricing is called Ramsey pricing). Those with more alternatives, who can switch to other services, will reduce demand in response to the higher price. The way to maximize efficiency (and in this case also revenue) is to charge more to those who have fewer choices.

Charging what the market will bear may not always be considered fair. But it is efficient -- provided, of course, that the market is fully competitive. As mentioned earlier, such pricing is not appropriate when the public sector supplier is a monopoly. As mentioned later, it may also be inappropriate on distributional grounds, since the rich generally have more alternatives than the poor. Nonetheless, it may provide a useful efficiency benchmark for assessing the costs of deviating from efficient pricing for distributive or other reasons.

Another form of price discrimination, which may overlap with Ramsey pricing, is declining block pricing. This is common in some public utilities and is closely related to the practice of imposing lower unit charges on more frequent users (commuters, park visitors, etc.). The premise is "the more you use, the less per unit you pay." Such a pricing strategy may make sense for a facility with substantial excess capacity, but care must be taken to ensure that the extra consumption does not cluster and create a peak-load problem.

In the case of a transit system, if the system is under-used except at peak times, it may make sense to offer discounts to off-peak travellers, perhaps with the unit price declining as such travel increases. But it never makes sense to offer similar reductions to the regular commuters already straining the system 's capacity at peak periods. As this example suggests, the distributional effects of such pricing may be considered good -- the poor, the elderly and students who travel outside peak period. Or it may be considered bad -- commuters who must be at work by 8:30 a.m. are "penalized". But as noted elsewhere, such distributional considerations should not initially be taken into account in deciding the appropriate level and structure of public sector prices.¹

Pricing "Rents" -- A Market-value Approach

An important area for pricing in the federal public sector relates to "rents." As noted earlier, many economic rents are created by public monopolies through allocation of assets highly valued by the private sector, such as natural resources and the electromagnetic spectrum. Often, there are good public policy reasons for either creating such monopolies or, if nature created them, allowing the public sector to determine who uses them. In any case, once such a government monopoly exists, the question arises as to what charge should be imposed for private access to the system and how this charge should be determined. This is quite different from the pricing problem discussed to this point because (with trivial exceptions) there are really no costs to be recovered. Instead, the issue is how to maximize the monopoly rents that result from the issuance of licences while ensuring that the resulting allocation of resources is as efficient as possible.

Auctioning will, in the right circumstances, ensure that licences go to bidders that most value them and will pay the highest possible price for them. Traditionally, the form of auction employed by government has been a single round sealed bid auction, similar to the system used in competing for contracts to supply services to government. More recently, alternative forms of auction have been developed and to some extent implemented, primarily in the United States, to improve the likelihood that the right price is charged and the right person gets the licence. The sequential auction reflects a situation in which, for example, licences for particular frequencies are offered one after the other.

Another, and particularly interesting, form is the simultaneous ascending auction, which has recently been used to allocate part of the radio spectrum in the United States. Under this approach, multiple licences are opened simultaneously for public bidding, which remains open as long as there is any form of bidding on any licence. In some instances, the licences are interdependent in the sense that two licences may cover the same region or part of the spectrum. Licences may also be complementary in the sense that a particular licence may be worth more to an operator who also operates in a contiguous region. Bidding occurs over rounds, with the results of each bid being accounted to bidders before the start of the next round. The auction is run by a computer, allowing for on-line bidding. Bidders are always aware of the value of each licence to competitors as well as the potential aggregation of licences pursued by different participants in the auction. In its only field test to date, this rather complex process yielded considerably more revenue for the government than initially estimated and appears to have produced a logical structure of producers in the market.

The best way to price rent-yielding services efficiently appears to be through some form of auction process. Considerable effort may be needed to design an appropriate process and package the offerings in the optimal way. However, the amounts potentially realisable are concerned are generally sufficient to make such effort worthwhile.

Pricing Externalities

The obvious way to account for positive externalities is to estimate the size of the marginal benefit by providing the service in question to an additional user and then setting the price equal to marginal cost less this external benefit. The resulting financial deficit could then be funded from general revenues. There are two problems with this simple strategy. First, it is virtually impossible to measure external social benefits in any valid way. Discussions on this issue tend to reduce to little more than assertions about individual preferences and perceptions. There are no easy solutions. The few studies of external benefits (e.g., of education) that have been attempted suggest they are usually substantially less (in the range of 10 percent of total marginal cost) than is often asserted (which is in the range of 50 to 90 per cent).

With respect to many public services, the issue is increasing the proportion of costs currently recovered by user charges to a more reasonable level. It seems likely that this can be done without significantly reducing service. The best way to deal with this question may be to put the burden of proof for large subsidization on those advocating such subsidies. That is, the appropriate initial position is that any public service with an easily identifiable beneficiary should be paid for by that beneficiary, unless sound and convincing arguments in favour of an explicit public subsidy can be made.

The second problem is that subsidies must be paid in the most efficient way. One approach much used in the Canadian public sector is to pay a supply subsidy to induce providers to lower their charges to direct users. This approach may be administratively convenient but it has serious problems. For example, if the subsidy depends on the number of users, inefficient over-expansion may be encouraged. Similarly, the subsidy may go to the wrong user from the point of view of public policy -- for instance, the rich (whose demand is more likely to be elastic) compared to the poor (whose demand is inelastic). Supply subsidies may also result in "gold-plated" services that cost more to produce than society is willing to pay.

If such concerns are thought to be significant, shifting from a supply subsidy to a demand subsidy is a possible alternative. For example, a tax credit or transfer payment related to consumption of a service might be directed to (the chosen) consumers. Many such schemes are being implemented in a number of countries, in such public sector programs as health and education, through smart cards, vouchers and so on. Alternatively, if supply subsidies are to be continued, they should, to the extent possible, be pre-determined in the budget process and hence not be subject to manipulation by suppliers.

In addition to external benefits, there are negative externalities that may arise, both from public sector actions and from decisions made by private individuals and firms. How does one price such negative externalities? If an externality can be quantified and attached to a particular action, then a tax or charge can be assessed equal to the marginal costs imposed on those other than direct beneficiaries. If the externality arises out of the non-priced use of a "common property" resource such as air or water, in some instances it can be priced by some variant of the auction technique described above. In other instances it may best be approximated by some form of liability-based tax or charge (e.g., "eco-fees").

The Costs and Process of Pricing

Based on the preceding discussion, properly designed user charges will often require the collection of complex and difficult-to-obtain information. How costly is it to the manager to charge (i.e. to collect the information required to derive an appropriate user charge regime, and the cost of collecting the user charge)?... and does incurring these costs constitute a sound investment? Indeed, thecosts of pricing may be so high as to make it unfeasible to apply user charges. The costs of obtaining the necessary detailed cost and demand information may exceed any conceivable benefits from applying even the best-designed charges. In all cases, such costs, like the actual costs of implementing charges and the (less visible) costs of ensuring compliance should be factored into the final determination.

Costs of many public sector activities -- for example, admitting a person to a facility in the off season -- may readily be calculated. But the cost of collecting the charge may exceed this amount. If so, no charge should be levied. Every road could be a toll road, but the high cost of collecting all those tolls would make it impractical.

An additional factor in determining user charges is the cost of changing public prices once they have been set. So long as an agency is in the public sphere, the prices it sets inevitably reflect the outcome of a political and administrative process and not a pure "market" or economic process. User charges are thus a political institution and hence, inherently "sticky" or hard to change. Once set, they tend to prevail until political circumstances permit, or compel, change. This situation is observed most readily when prices are first set on a service previously provided free, but this applies to some degree to all changes in public prices. For this reason, it is particularly important to set economically efficient prices.

4. Conclusion

Political Realities

Since user charges are somewhat politically determined, it is important at the outset to conduct an adequate process of consultation with affected groups and reviews by the public and central government agencies. This will provide some assurance that the prices set are reasonable and acceptable, and that subsequent adjustments can and will be made as appropriate. The details need to be worked out carefully for each particular area, but a few general principles apply.

First, it is important to set out the parameters within which individual public sector managers can determine prices. What should they take into account? With whom should they consult and how? To what review, if any, are their decisions subject? Political processes are distinguished from market processes primarily by perceived procedural fairness. It is important to set out what this means and how it is to be attained with respect to user charging.

Second, it is important to provide clear and strong incentives for managers to impose efficient user charges. For example, they may obtain a larger budget or one over which they have more control. Unless something is to be gained through the difficult task of designing and implementing such charges, it is not clear what incentive managers have to undertake this task. This problem may be especially important when proper pricing will result in some shrinkage of the agency's budget.

Third, if the prices set by a particular agency are subject to central agency review, the principles that guide such review should be clearly stated and the application of those principles demonstrated to affected managers and the concerned public. Public sector managers cannot and should not be expected to respond solely to the "bottom line" in any financial sense. If they could, the activity in question should not be in the public sector. But they cannot be expected to act efficiently in pursuit of public policy objectives unless the lines and rationale of accountability are clear.

Fourth, it should be apparent to directly affected clients that the real public includes the Canadian people as a whole, represented by their elected officials. This point must be emphasized to ensure that it is not only the direct beneficiaries of the program whose voice is heard.

Serious efforts must also be deployed to communicate to affected groups how the pricing regime was derived and that it is reasonable. To achieve this, the pricing regime must indeed be reasonable. It must also be presented persuasively to a group of people who are accustomed to not having to pay for the service One compelling approach, when circumstances warrant, will be to make the point that if users do not pay a price reasonably related to the benefits they receive, the service may be significantly scaled back or not provided at all. (Users' feedback will then help managers ascertain whether this service should be provided).

Finally, as mentioned earlier, experience suggests that the major area of general public concern is related to real, or perceived, adverse distributional effects. Two alternative strategies may alleviate such concerns. The first is to present a detailed and convincing study of the expected distributional effects of imposing user charges. If such effects are not serious, that should be stated, and the reality of the situation demonstrated. If, however, there are risks of serious adverse distributional effects, the best approach is to recognize these risks and present the proposed strategy that will be pursued to address this situation.

Among possible offsetting devices are "lifeline" pricing schemes (access to an initial basic quantity of the service for everyone at low prices) or compensating changes in general transfer payments. Alternatively, for certain services, a "smart-card" scheme might be appropriate. Under such a scheme, all users would access the service using a card, but low-income users would be given a certain initial credit on their cards. Thus, universality (everyone has the same card) and targeting (those who need it have free or subsidized access) could be achieved.

Similar approaches could be taken to address other areas of public concern over user charges. People must be persuaded that either there will be no real adverse effects or that the goals in question can and will be more efficiently and effectively achieved through explicit budgetary subsidies or in some other way. Close attention to such "political economy" issues may be the key to public acceptance of more rational cost recovery, user charges and pricing policies in the federal public sector.

Concluding Remarks

The preceding discussion summarizes a large, complex, and technical body of literature on appropriate cost recovery and user charging in the federal public sector. Further readings are suggested in the bibliography attached to this paper.

The general principles raised in this discussion are intended to guide public sector managers charged with designing and implementing user charges. Establishing user charges is not a simple task. However, with a clear understanding of the theoretical rationale and broad approaches, the task becomes more manageable and less daunting. It is important to know where it is appropriate to apply charges and where it is not, and to understand why it matters so much that user fees be properly designed and applied. Appendix A provides a "decision-tree" approach to user charging. It offers public sector managers a step-by-step summary of the approach set out in this paper.