XXI. Pricing and Cost Recovery

Ke-young Chu, and Richard Hemming
Published Date:
September 1991
  • ShareShare
Show Summary Details

What is the scope for marginal cost pricing in the public sector?

What deviations from marginal cost pricing are needed to meet revenue and distributional objectives?

Can optimal pricing rules be implemented? What are the alternatives?

Are there advantages to earmarking revenues to particular expenditure programs?

Expenditure issues cannot be discussed independently of the way in which expenditure is paid for. Both taxation and borrowing have possible adverse consequences for efficiency and growth (see the note on Public Expenditure and Sustainable Fiscal Policy). To the extent that the public sector can recover at least part of the costs it incurs in providing goods and services through charges, there is scope to reduce taxation and/or the fiscal deficit. Moreover, if publicly provided goods have private characteristics, and charges can be levied to reflect private benefits, this will improve efficiency.

Principles of Public Sector Pricing

Marginal cost pricing

Efficient resource allocation generally requires that prices be set equal to marginal cost. If consumers are willing to pay a price above (below) marginal cost for a particular good (or service), resources could be transferred to (from) the production of this good and thereby raise consumer welfare (see the note on Public Expenditure and Resource Allocation). In the end, prices will reflect the opportunity cost of the resources used to meet consumer demand.

Public sector production ranges from public goods to private goods. A public good is nonrival in consumption, and the cost of accommodating an additional consumer is zero; the marginal cost pricing principle implies that there should be no charge for such goods, and that the costs of provision should be met from general revenue. Private goods should be priced at their positive marginal cost, and the losses of decreasing-cost industries such as public utilities (for which marginal cost is below average cost) should also be met from general revenue. Strictly speaking, the only goods that fall in between these two extremes are group goods and local public goods, which are nonrival among identifiable beneficiaries; air traffic control and flood prevention have these characteristics. Such goods should be provided to the group or locality free of charge, but the same group or locality should finance their provision. However, goods that are associated with externalities are often thought to fall between public and private goods. For goods characterized by large external benefits, prices can be set below those that would prevail in a free market, with the resulting subsidy—designed to bring marginal social costs and benefits into line—financed from general revenue. Part of the revenue can be derived from taxes on goods characterized by large external costs.

Revenue objectives

If the government could impose nondistortionary lump-sum taxation, it could cover the cost of providing public goods, the losses of decreasing cost industries, and subsidies to internalize externalities without inducing distortions. The case for marginal cost pricing is then a first-best argument. However, lump-sum taxation is not feasible, and, in a second-best world where taxation is distortionary, departures from marginal cost pricing may be desirable. A special case is where the public sector has a revenue target that it cannot meet through taxation, owing to a narrow tax base or weak administrative capability. If the public sector is responsible for a significant share of total output and can control the price of that output—which realistically assumes a significant degree of monopoly power in the public sector—then public sector pricing can be used to generate revenue.

Ramsey pricing

Determining the second-best set of public sector prices that meets a revenue target but minimizes the efficiency loss as a result of deviations from marginal cost pricing is formally equivalent to deriving optimal commodity taxes. In their simplest form, the Ramsey rules imply that prices deviate from marginal cost in inverse proportion to demand elasticity. The intuition is that goods in relatively inelastic demand can bear high (implicit) taxes because taxation does not significantly affect consumption. More complex interpretations of the Ramsey rules have a similar intuition. An implication of Ramsey pricing is that it is generally inappropriate to impose financial targets as a means of setting prices in subsectors of the public sector. In particular, activities characterized by decreasing costs should not be subject to break-even constraints, as is the common practice. Rather, prices should be raised above marginal cost only insofar as demand conditions dictate, and if losses persist, they should be cross-subsidized.

Departures from efficient pricing

While Ramsey pricing assumes a monopolistic public sector, if the private sector exhibits elements of monopoly, marginal cost pricing may not be an appropriate basis for pricing in the public sector and the imposition of Ramsey prices to meet a revenue target may no longer be optimal. Moreover, even with a competitive private sector, the need to generate revenue is not the only justification for departing from marginal cost pricing. If there are difficulties in effectively imposing taxes, then the government’s ability to redistribute through progressive taxation is also constrained. While public sector pricing may generate sufficient revenue to finance redistributive transfers and other social expenditure programs, some prices can also be subsidized for redistributional purposes. Although this is generally an inefficient way of pursuing distributional objectives—it is better to redistribute income—price subsidies may be justified when the scope to use other instruments is limited.


The preceding discussion illustrates some of the limits to the marginal cost pricing principle. While desirable departures from marginal cost and Ramsey pricing may be described in some cases, when one takes into account the full range of market imperfections in the private sector, and the combination of multiple objectives and limited instruments of the public sector, optimal pricing is virtually impossible to characterize. Moreover, even in the simple cases, pricing rules may not be operational. In many cases marginal cost is not observable, although techniques are available to approximate it from other market or cost data. And if marginal cost can be calculated, the information required to compute optimal departures from marginal cost is often unavailable. For example, the demand elasticities on which Ramsey prices should be based are rarely estimated, nor is it straightforward to make such estimates. As a consequence of these limitations, pricing policy in the public sector is by necessity somewhat ad hoc, the objective being to push pricing policy in what is likely to be the right direction on the basis of clearly defined principles.

Public Sector Pricing in Practice

The notion that public goods should be provided free of charge is widely accepted. Similarly, that the public sector should charge market prices for private goods in competitive markets is uncontroversial. In both cases, there is consistency with the marginal cost pricing principle. Where the only source of competition is imports, border pricing is widely practiced. This will approximate marginal cost pricing only with unrestricted international competition in all markets; otherwise, border pricing is adopted to avoid smuggling to which market imperfections may give rise. Beyond these rather straightforward cases, pricing policy has to be considered on a case-by-case basis. In this connection, however, utility pricing and the scope for user charges in the social sectors—especially health and education—are issues that often arise.

Utility pricing

The natural monopoly characteristics of most utilities—which arise from the extensive use of networks such as electricity grids—make them clear candidates as a group for the application of Ramsey pricing to generate the revenue to cover their losses. Moreover, it is in industries such as electricity, water, and telecommunications that marginal cost is easiest to measure. Clearly, the standard problem about whether prices should be based upon short-run marginal cost or long-run marginal cost has to be resolved. Generally speaking, the former is preferable from a resource allocation viewpoint but produces large price variations, which are administratively costly, politically inconvenient, and create uncertainty among consumers and investors. But the main problem is that estimating the demand elasticities to calculate Ramsey prices for a group of industries is difficult. However, this does not preclude the use of a rule of thumb—water is more inelastically demanded than telephone services, for example—to get appropriate qualitative deviations from marginal cost.

If the scope for system-wide application of Ramsey pricing is limited, it may nevertheless be possible to apply it to one industry to meet a revenue target. Thus a breakeven constraint could be placed on the electricity industry, with a higher prices for lifeline services which are inelastically demanded and lower prices for more discretionary uses. The shortcoming of this approach—and the Ramsey rules in general—is immediately apparent. Goods and services in inelastic demand are often necessities, and charging a high price for necessities hurts the poor. This would not be a problem if there are other means of helping the poor, ideally by raising their incomes, but there may be few alternatives. If redistribution justifies a departure from marginal cost pricing, lifeline electricity should perhaps be provided at below marginal cost, and the revenue constraint will imply that some other inelastically demanded service—for example a specified minimum amount of electricity used by middle-class residential and commercial consumers—should be highly taxed. Peak-load pricing, which is intended primarily to match demand to available capacity, may serve the same purpose if the poor have the flexibility to switch their consumption to off-peak periods.

User charges in health and education

Distributional concerns take on particular significance in the case of social goods such as health and education. The principal issue is how to introduce an element of cost recovery through user charges without conceding distributional objectives. At the same time, there is a trade-off between the desirable implications for resource allocation of imposing some charges and the possible failure to internalize the external benefits of health and education provision if charges are pushed to too high a level. These are complex issues. However, a feature of both these services that facilitates charging is that they are nontransferable, and therefore they are amenable to personalized prices.

From the distributional point of view, personalized prices would allow the poor to be charged less than the rich if these groups can be distinguished. Moreover, if the external benefits of health and education are less appreciated by the poor, or if the poor are less likely to make well-informed decisions, this pattern of prices is consistent with efficiency objectives. However, the fact that governments usually provide universal access to health and education services of a certain quality, and in the latter case make consumption compulsory, suggests that pricing policy alone cannot compensate for market failure. These issues are discussed further in the notes on Health and Education.

Other user charges

User charges are not restricted to social goods, and can be applied quite widely in the public sector. However, because marginal costs are difficult to measure and distributional concerns usually are less compelling, different considerations are brought to bear upon their design. They are often applied to postal, civil aviation, coast guard, and port services, for example. However, perhaps their most common application is to road use.

Generally, user charges can be either specific fees or system-wide fees, depending upon the extent to which metering is possible. If consumption can be metered, fees can be charged that reflect the costs of different categories of demand. Thus a toll road can differentiate fees by distance travelled, and charge trucks more than cars because the former do more damage to the road. Fees can also be high for relatively congested roads. If metering is difficult, and in the case of roads this might be a reflection of high administration costs, a less differentiated fee structure may be used instead. For example, a uniform licence fee can be used to recover costs, but it does not distinguish between different users or the intensity of their use. Indirect charges, such as fuel taxes that distinguish between gasoline and diesel fuel, may more successfully internalize road damage externalities, while subsidizing alternative forms of transportation may reduce congestion.


An issue that arises in connection with user charges relates to whether the revenue raised should be used exclusively to pay for the expenditure that gives rise to them. Thus road user charges would be used to pay for road construction and maintenance. Because there is a close link between the payment of the user charge and the associated expenditure, this is an example of strong earmarking. This contrasts with weak earmarking, where the benefit link is weak. The use of lottery proceeds to pay for public education or a specified share of government revenue to pay for investment spending are examples of weak earmarking.

If the benefits derived from an expenditure program are purely private, expenditure should be fully funded by strong earmarked charges. In effect, private activities undertaken by the public sector should be indistinguishable from similar activities in the private sector—the private sector functions entirely on the basis of fully funded earmarking. In practice, earmarked public sector activities, like road construction and maintenance, are only partially funded by earmarked charges. In many cases, however, this extends beyond the level of earmarking that can be justified by reference to the strength of the benefit link, and is usually associated with extensive use of weak earmarking.

The earmarking debate

The most widely held view is that in the absence of a strong benefit link earmarking is not justified. It is argued that the level of expenditure on a particular program becomes significantly dependent upon the revenue generated to pay for it, which limits government discretion over those funds and leads to a misallocation of resources. Expenditure decisions are not dictated by efficiency criteria but more by the ability of politicians and bureaucrats to put in place earmarking arrangements that protect their favored programs. However, proponents of earmarking argue that even without earmarking, expenditure decisions reflect state failure that arises from the detrimental influence of politicians and bureaucrats. Under such circumstances, the fact that earmarking introduces rigidity into the budget reduces the scope for such influence, and this leads to an improved allocation of resources. In other words, while earmarking may be inefficient, it is more efficient than not earmarking.

The scope for earmarking

Resolving the earmarking debate is not easy since it largely depends upon identifying the extent to which public sector activity is impaired by political and bureaucratic interference and the relative effectiveness of earmarking in compensating for this. These judgements are difficult to make. However, even without being able to make such judgements, there may be some limited role for earmarking under certain conditions. For example, during periods of fiscal adjustment, the earmarking of revenue to essential programs that are relatively easy to cut—such as operations and maintenance—may be appropriate despite a weak benefit link. But in general, it is a strong benefit link combined with a desire to impose the discipline of the private sector on public sector activities that provides the main justification for earmarking.

Country Illustrations

Toll roads in Indonesia

The government justifies toll roads as part of its medium-term road development program on the grounds that where roads are congested and road capacity expansion is required, users should finance road construction through tolls that should not be higher than avoided congestion costs. The government finances the initial construction costs, generally through foreign loans, and retains ownership of any new toll road. However, it delegates administrative functions to one or more state corporations. Toll rates are set by presidential decree and in general provide for cost recovery over a defined payback period. Tolls are differentiated by vehicle type and weight and by traffic density on particular routes. Rates are periodically increased more or less in line with inflation.

In designing its strategy, the government integrated equity objectives into the policy. Not all new roads are financed through tolls. While the strategy generally calls for beneficiaries to bear the costs of an expansion in the current road network in order to reduce the burden on public finances, the policy allows for budgetary allocations to be set aside to finance road projects in the less-highly developed areas of the country and so promote more geographically balanced development. As a result, the government has purposely avoided the allocation of budgetary resources for the construction of high-cost roads in already highly developed areas, but continues to finance nontoll roads in rural areas of the country.

Earmarked fees in Ecuador

Revenues generated by charges levied on visitors to the Galapagos Islands are earmarked for national park maintenance across the country. The government introduced the fees in part to reduce damage to these ecologically fragile islands. The political appeal of introducing such fees was enhanced by dedicating their revenues to environmental concerns. The benefit link is, however, weakened to the extent that foreign visitors to the Galapagos Islands provide much of the revenue while local visitors to other national parks benefit from much of the expenditure it pays for.


    Bos D. “Public Sector Pricing,” in Handbook of Public EconomicsVol. 1ed. by A.J. Auerbach and M. Feldstein (Amsterdam: North-Holland1985).

    • Search Google Scholar
    • Export Citation

    Heady Christopher “Public Sector Pricing in a Fiscal Context.” Country Economics Department Working Paper Series 179 (Washington: World Bank1989).

    • Search Google Scholar
    • Export Citation

    Jimenez Emmanuel “Social Sector Pricing Policy Revisited: A Survey of Some Recent Controversies,” Proceedings of the World Bank Annual Conference on Development Economics (Washington: World Bank1989).

    • Search Google Scholar
    • Export Citation

    Newbery David M. “Charging for Roads,” World Bank Research Observer (Washington: World Bank1988).

    Other Resources Citing This Publication