L.S.E. Essays on Cost
By James M. Buchanan and George F. Thirlby
When I originally suggested the idea for this book, I had hoped to be able to include a considerably wider range of papers with which to underline James M. Buchanan’s challenge on p. 35 of his
Cost and Choice, where he regrets the demise, and calls for a resurrection, of the L.S.E. opportunity-cost tradition (see p. 6 of this book). However the limitations of finance compelled a stricter selection, and, even so, the emergence of the book would not have been possible without institutional as well as personal support and encouragement. The Center for Study of Public Choice, Virginia Polytechnic Institute, has cooperated fully with the L.S.E. Publications Committee throughout the planning and production of the book, which is institutionally a joint product. For this purpose the Center itself was supported by the Earhart Foundation, whose assistance is gratefully acknowledged…. [From the Preface by George F. Thirlby]
First Pub. Date
New York: New York University Press
First published 1973, for the London School of Economics and Political Science, U.K.: Weidenfeld and Nicolson Collected essays, various authors, 1934-1973. First published as a collection 1973 for the London School of Economics. Includes essays by Ronald H. Coase, Friedrich A. Hayek, Lionel Robbins, and more.
The text of this edition is copyright ©1981, The Institute for Humane Studies.
- Buchanan, Introduction, L.S.E. cost theory in retrospect
- Robbins, Remarks on certain aspects
- Hayek, Economics and Knowledge
- Edwards, Rationale of Cost Accounting
- Coase, Business organization and the accountant
- Thirlby, Subjective theory of value and accounting cost
- Thirlby, The Ruler
- Thirlby, The economists description of business behaviour
- Wiseman, Uncertainty, costs, and collectivist economic planning
- Wiseman, The theory of public utility price
- Thirlby, Economists cost rules and equilibrium theory
First published in
Oxford Economic Papers (Oxford University Press, 1957), ix.
Criticism of the analytical validity of public-utility pricing ‘rules’ has resulted over a period of years in the introduction of successive modifications to the original simple (though not unambiguous) marginal-cost ‘rule’, culminating in advocacy of the two-part tariff and of the ‘club’ principle.
*45 While these pricing rules have been regarded with scepticism as practical guides to public-utility pricing policy,
*46 however, there has perhaps been a less general appreciation of the cumulative weight of the theoretical objections to all such rules; there is still interest in the discovery of a ‘right’ rule, and in the estimation of the ‘marginal’ or other costs of particular public-utility enterprises.
It will be argued in this article that no general pricing rule or
rules can be held unambiguously to bring about an ‘optimum’ use of resources by public utilities even in theory. Indeed, failing some universally acceptable theory of the public economy, the economist can offer no
general guidance at all to a government having to decide a price policy for such utilities. To demonstrate this, it will be necessary to begin with a brief survey of the criticisms of the simple marginal-cost rule. This will provide the basis for a demonstration of the possibly less familiar (though no less decisive) analytical shortcomings of the two-part tariff rule both in its simple form and as modified by a club principle. In conclusion the effect of uncertainty on the analysis will be examined, and the broad implications of the whole argument for public policy will be suggested.
Any discussion of a ‘right’ price presupposes criteria of the public interest against which alternative suggested prices can be judged. The criteria from which the marginal-cost rule stems are derived from the analytical model of a perfectly competitive market economy, in which entrepreneurs are assumed to have perfect foresight
*47 and it is a property of the long run equilibrium situation that, given the distribution of income between consumers, no transfer of factors between uses could increase the satisfactions of one consumer without reducing those of another. The optimum conditions of ‘economic welfare’ are consequently said to be fulfilled by the model. For the competitive firm it is an incidental property of the long-run situation that marginal cost (money outlay on factors)=average cost=price of product. Consequently this equality can be regarded as evidence of the existence of an ‘ideal’ situation, and pricing at marginal cost has accordingly been proposed as a general pricing rule (e.g. as the ‘principle of
administration’ of a collectivist economy).
*48 But public utility enterprises are not perfectly competitive firms. By the usual definition an important part of the factors they employ are not perfectly divisible; they can be obtained only in large physical units, or in a durable or specific form, or both. Also the technically efficient production unit is large relative to the possible size of the market, and the utilities are public bodies, often with considerable powers of monopoly protected by law. In such circumstances there may be no possible price equal both to marginal cost (the money outlay required to increase output marginally) and average cost (which includes outlays on the ‘indivisible’ factors excluded from marginal cost). It therefore appears to be necessary to decide whether price should be fixed equal to the one or to the other.
The argument for pricing such public-utility products at average cost is simply that ‘each tub should stand on its own bottom’; all money outlays which would have been avoided if a product had not been produced should be recovered in the price charged by the utility. But the advocates of marginal-cost pricing find this unconvincing. Some of the outlays included in average cost, they argue, are not current opportunity costs but are either payments for technically indivisible factors or past out-payments for durable and specific factors. The inclusion of these outlays in the price charged therefore prevents the achievement of the optimum welfare conditions, which (it is said) require that additional consumption of a good or service should be possible at a price not greater than the additional costs (money outlays) necessarily incurred in providing for that consumption. Accordingly such outlays should be ignored, and the product priced at marginal cost, even though the enterprise runs at a loss as a result.
Clearly the proposal for pricing at marginal cost requires an explanation of how the consequent losses are to be financed. Hotelling, who originated much of the discussion,
*49 suggested the
use of particular types of taxes. The inclusion of charges for ‘overheads’ (past outlays on indivisible factors) was itself, he said, of the nature of a tax. But there were other and preferable taxes (lumpsum taxes on inheritance, income taxes, etc.) which did not offend against the welfare criteria since they affected only the
distribution and not the
size of the national income. If such taxes were used, and public-utility prices were equated with marginal cost, the optimum welfare conditions would be achieved. Later writers have been justifiably sceptical of the possibility of a tax system that would meet Hotelling’s conditions.
*50 In particular income taxes (on which he expected to have to rely) can be shown to affect the marginal welfare conditions directly. In any case the proposal is open to an even more fundamental objection: the welfare ‘ideal’ relates to a
given distribution of income, and that distribution of income must be altered by the proposed taxes (unless these fall on consumers of public-utility products in proportion to their consumption, which is effectively a return to average-cost pricing). Thus to advocate marginal-cost pricing and the meeting of losses out of taxation is to advocate acceptance of income redistribution from non-consumers to consumers of public utility products. The welfare criteria provide no justification for an interpersonal comparison of this kind. In other words any government deciding upon a pricing policy for public utilities has to take simultaneously into account the effects of its decisions upon the fulfilment of the welfare optima (and hence the size of the national product) and upon the distribution of incomes, and there is nothing in the welfare analysis that provides guidance as to the ‘right’ policy about the second of these.
The reason why marginal-cost pricing raises these difficulties is to be found in the fact that the arguments for the marginal-cost rule are logically unsatisfactory in that they attempt to apply welfare criteria derived from an analysis concerned with marginal variations in factor use to a problem whose essence is discrete change; the whole basis of the public-utility discussion is the indivisibility of the factors employed by such utilities. The results of this attempt are not only of dubious relevance to policy; they are also uncertain in themselves.
The type of indivisibility most emphasized in the discussion is that created by the durability and specificity of factors (
*52It is enlightening to examine the nature of such indivisibility more closely. It has been shown that the marginal-cost ‘rule’ distinguishes between ‘current’ and ‘past’ opportunity-cost problems. Once the sacrifices necessary to create a durable and specific asset have been made, it is argued, no further opportunity costs are created by its later use. The opportunity costs having been borne in the past, no account should be taken of them in deciding current prices,
*53 even though, as has been demonstrated,
this results in losses and in income redistribution. Such an argument rests upon a dubious interpretation of the welfare criteria. The long period, from which the welfare postulates derive, is a situation in which all factors of production are considered to be perfectly mobile; this would seem to imply consideration of a time period at least as long as the lowest common multiple of the life span of all the factors of production concerned. If the marginal-cost rule is conceived in terms of a time period shorter than this, then not all the opportunity costs requisite to the manufacture of the product concerned can be imputed to that product, and the time period chosen must itself be arbitrary, so that the marginal-cost rule becomes simply a statement that outlays on factors of some specified durability should be ignored in deciding prices (i.e. should be treated as ‘past’ outlays). The figure treated as marginal cost will thus depend upon the time period selected.
The division of outlays into ‘past’ and ‘current’ is clearly unsatisfactory, and the implications of durability and specificity become less obscure if such a division is abandoned and the problem is presented in the form of a planning process through time. All opportunity-cost decisions, taken at one moment in time, fix the use of factors during some future period of time. All factors embodied in plans implemented by entrepreneurs, that is, become durable and specific to some degree; new opportunity costs arise in respect of them only when their use can be replanned. This being so, it is not possible to separate opportunity costs into two groups, ‘past’ and ‘current’. The most that can be said is that some kinds of factors lend themselves more readily than do others to frequent replanning. There is a difference, for example, between the extent to which factor use will be ‘fixed’ over time by the implementation of a decision to build a railway bridge and by a decision to hire a railway porter. But the difference is one not of kind but of degree; it is possible to conceive of an ‘ordering’ of opportunity-costs decisions in accordance with the length of time for which they commit factors to particular uses (i.e. create
specificity and durability), but it is not possible to divide such decisions into a group that involves a commitment over time and another group that does not.
Marginal (opportunity) cost in these circumstances is represented by a forgone revenue. The use of factors of production in the entrepreneur’s selected plan excludes them from use in some other plan; marginal cost is the forgone marginal revenue from the best plan necessarily excluded because the chosen plan is selected. But the alternative uses to which factors can be put, and hence the opportunity-cost valuation imputed to them in the planning process, will depend on the time period in terms of which the entrepreneur’s plans are themselves conceived; marginal cost will consequently vary according to the time span of the production plans considered. Thus the meaning and results of an instruction to equate marginal cost and price will be determined by the length of the planning period to which the marginal cost is intended to refer. At one extreme the period chosen may be as long as the lowest common multiple of the life periods of the assets required to produce the public-utility product, and the marginal-cost rule would then give a price that took into account the whole of the sacrifice of alternative consumption caused by the implementation of plans to manufacture the utility product. At the other extreme the consideration of ‘current’ opportunity costs only, if interpreted rigorously, would seem to require that products should be given away. Between these two extremes there is a range of possible marginal-cost rules, differing from each other in the planning time period chosen as appropriate and hence in the ‘durable’ assets they ignore and in the opportunity costs they treat as relevant to decisions about price and output.
The only time period in which all factor-use can be clearly attributed is one as long as the lowest common multiple of the life period of the assets concerned; the designation of any other (shorter) time period as the one appropriate to the rule must involve both an arbitrary decision that that period is one relevant to the computation of marginal cost and a value judgement that income should be redistributed over time towards the consumers of goods produced with relatively durable assets. The marginal
cost principle thus becomes, not the assertion of a general welfare ‘ideal’
but the expression of a particular value judgement, that certain long-run opportunity costs for the community as a whole should be ignored in the interests of the greater short-run utilization by consumers of specific factors of some stated degree of durability.*55
The only defence offered against criticism of the marginal-cost rule on such grounds lies in the introduction of a supplementary criterion: the investment principle. This requires that marginal-cost pricing should be used to decide the selling prices of public-utility products once the utilities are in existence, but that the public investment necessary to create a utility initially should be considered justified only if a perfectly discriminating monopolist could (notionally) recover its cost by charging prices that would maximize his returns. The need for such a supplementary principle to a ‘general’ rule is implausible. In any case the investment principle does not answer the criticisms. It still has to be decided which economic decisions are to be treated as ‘investment’ decisions and which as subject to the marginal cost-rule, and no principle has been suggested by reference to which such decisions might be made. Further, since the prices actually charged are to be determined by the marginal-cost rule the discussion of the effects of that rule (e.g. on income distribution) is unaffected by the introduction of the supplementary principle. It is worth pointing out further that the investment criterion itself has redistributive implications: certainly it does not appear to meet the welfare conditions in the same fashion as would a perfectly competitive market.
The foregoing criticisms of marginal-cost pricing have been fairly widely accepted, although the precise nature of the value judgements implied in the treatment of temporal indivisibility is perhaps not generally recognized. Despite this acceptance, there still seems to be considerable support for marginal-cost pricing from those who feel that policies affecting only the
distribution of the national income both are possible and are in some sense superior to alternative policies that would also affect its
size. In the absence of some generally acceptable basis for preference between different income distributions, it is clear that such a position cannot be supported by logic. The two issues cannot be separated, and policies desirable in terms of the welfare criteria may therefore reasonably be rejected because a government chooses to obtain a preferred distribution of income even at the cost of some diminution of its total size. There is no escape from the very special value judgements that marginal-cost pricing implies.
The two- (or more) part tariff is intended to avoid the anomalies of marginal-cost pricing, in that it is designed to meet the marginal ‘welfare’ conditions and also to avoid problems of interpersonal comparison by raising revenues large enough to cover all outlays. The essence of the proposal is that the price to be charged should by the sum of two parts:
*58 a ‘marginal-cost’ element determined by the increase in costs necessarily incurred in providing further consumption for an individual consumer, and a ‘fixed charge’ to
cover costs which do not vary with consumption but which must be incurred if the consumer in question is to be enabled to consume at all. In this way total costs are covered and the payments made for additional consumption are kept equal to the extra costs of provision (marginal cost) alone. The problem appears to be solved, since the ‘welfare’ conditions are satisfied and no income redistribution seems to be implied.
Unfortunately multi-part pricing provides an unambiguous solution only if the two types of costs concerned can be clearly imputed to individual consumers. In fact, when this can be done, the two parts of the price can logically be treated as the prices of separate products, each capable of clear determination by a normal market process. When these conditions do not obtain, however, the situation becomes very different. This can be seen by introducing the possibility of common costs. If problems of time are disregarded, these are simply current costs that do not vary with total output, but are necessarily incurred if
any output is to be produced at all and
are not imputable directly to individual consumers.*60 How should these current ‘fixed and common’ costs be shared between consumers? In principle limits can be set to the charges that individual consumers can and should be asked to pay, by reference to the cost of providing the indivisible service
for them if other consumers ceased to consume, on the one hand, and the minimum possible cost of providing their addition to total consumption on the other.
*61 But there may still remain a variety of possible methods of charging, and some non-arbitrary means of choosing between them is required if the multi-part tariff is to provide an unambiguous solution to the public-utility pricing problem.
There is a suggested means of meeting this common-cost problem that seems to be fairly widely accepted. This is the use of the ‘club’ principle. This principle is not usually stated with precision; its essence appears to be the proposition that the consumers of the utility product can be treated as a club, created by the consumers to arrange both the amount of the good each individual shall consume and the amount that he shall pay for it. Then, if all ‘members’ (potential consumers) are asked what they would voluntarily pay as a fixed charge rather than go without the possibility of consuming a particular product at a price per unit equal to marginal cost (money outlay), and if the sum of the amounts offered would be great enough to cover the total outlays required, the service should be provided and each consumer charged that part of the common cost that he has stated his willingness to bear. It follows that the ‘club’ principle is likely to give rise to price discrimination, in that different individuals need not be required to pay the same amount for a similar volume of consumption. That is, the principle must imply a redistribution of real income, since consumers with given money incomes purchase
a technically homogeneous product at money prices differing from one consumer to another. But, it is argued, the ‘club’ principle allows consumers themselves to make a voluntary decision whether to accept the good and the consequent income redistribution in preference to having neither. If they, as consumers, take the first course, then this must produce a more satisfactory situation from the point of view of consumers’ choice, and the optimum welfare conditions must therefore be better satisfied as a result of the use of the ‘club’ principle despite the consequent redistribution of income.
The ‘club’ principle has deficiencies serious enough to make the extent of its acceptance a matter for some surprise. The deficiencies are of two kinds. First, the value judgements being made in relation to income redistribution are difficult to justify. Second, the ‘club’ proposals require an unusual (and peculiar) interpretation of the concept of voluntary choice.
No one suggests that the ‘club’ principle avoids the need for value judgements about income distribution.
*63 Rather, what is implied is that ‘welfare’ can unambiguously be said to have been improved by a policy which meets the marginal welfare conditions, even though there is a consequent change in the distribution of income,
provided that the changed income distribution is the consequence of the ‘voluntary’ action of consumers. This extension of the welfare criteria is less innocuous than might at first appear. Economic welfare, as normally defined, is concerned solely with the optimum conditions of individual choice, given the distribution of income; the objective of the public-utility discussion might be described as the discovery of a pricing policy to meet those conditions, in the special circumstances of public-utility production.
But if a suggested principle of pricing would affect economic magnitudes other than the conditions of choice, then it becomes necessary to establish further policy criteria concerned with these other magnitudes, by reference to which the proposed principle can be assessed. In the present case, since income distribution is affected by the ‘club’ principle, criteria for choice between income distributions are required. Moreover these criteria must take the form of a statement about the
income-redistributive objectives of a government, since it has to be recognized that the public-utility discussion (although itself conceived in relation to the conditions of individual choice) is concerned to recommend policies to be implemented by a government.
That is, income distribution is a question of public policy, and it is the attitude of the government to it, and not the attitude of particular groups of consumers, that is of significance for policy. The value judgement implied in the ‘club’ principle is that if the members of the public-utility ‘club’ agree to a particular redistribution of income, then the government must necessarily think such a redistribution desirable. This is not plausible; there are likely to be many cases in which the ‘voluntary’ redistribution would be of a kind that the government disapproved.
*64 In short, once a government is committed, by the creation of a utility and the existence of common costs, to a decision about income distribution, there is no reason why it should prefer public-utility pricing policies that cover total costs by use of the ‘club’ principle to other policies which may or may not cover costs, but which accord better with its own attitude to redistribution. A government permitting utilities to use the ‘club’ principle in effect substitutes the authorities of the utility for itself as the final arbiter in matters of income distribution in this particular context.
The argument for the ‘club’ principle depends upon the fact that the charges to which it gives rise are ‘voluntarily’ agreed by consumers. In general this agreement will be ‘voluntary’ only in the special sense that a malefactor voluntarily goes away to prison after a judge has sentenced him; he chooses the best alternative still available. To appreciate this, it is necessary to look more closely at the form these ‘clubs’ can take and at the nature of their ‘regulations’ (i.e. the powers they have to take and enforce decisions about such matters as the payments to be made by members). Three broad types of ‘club’ can be envisaged.
The first type might be called the
direct production club: it is created and administered by the consumers themselves. Thus, if factors services are available for purchase in free competitive markets, groups of consumers may find it convenient to join together to hire certain services whose products will be consumed by all the group, although it would not be worth the while of individual consumers to hire them separately. Effectively the consumers
ask themselves whether it is worth their while to create a ‘club’ to provide the good concerned, and agree together (in
deciding to create it) upon the volume of their individual consumption and upon the payments each shall make. The illustrations given of the ‘club’ principle are usually of this direct-production character.
*66 Provided that there are alternative competing means of satisfying the demand in question without recourse to a ‘club’, then the possibility of forming a ‘club’ simply represents a widening in the range of choices made available by the competitive market, and so increases satisfactions (‘welfare’).
*67 This is true even though members of the ‘club’ pay different amounts for what is technically the same service. However, such cases of direct production would seem unlikely to be of widespread importance, though there may be special instances in which the conditions are quite well satisfied.
The second form that the ‘club’ might take is one in which the organization of production is undertaken, not by the eventual consumers, but by independent producers, who find the use of standing charges advantageous but whose freedom in deciding the charges that consumers (‘members’) shall be asked to pay is restricted by the presence in the market of other, similar clubs competing for the consumer’s membership. An example of this type of ‘club’ (the
competitive-producer club) is provided by the book clubs, offering supplies of books at differential rates related to total guaranteed consumption, but with the discretion of any
one club in deciding its rates circumscribed by the policies adopted (or able to be adopted) by the other book clubs, and by the ability of consumers to transfer their ‘membership’. There is a case for the existence of this type of ‘club’ also, on grounds of economic welfare. But it must be noticed that the consumers are not now taking decisions about how much they are willing to contribute to a venture in joint production; their ‘voluntary’ decisions are concerned solely with the nature and amount of their personal consumption at the prices thrown up by the market. The production decisions are taken by independent producers, and the fulfilment of the welfare conditions depends upon the protection provided for the consumers by competition between these producers.
The third type of ‘club’,
the discriminating-monopoly club, occurs when neither direct production (with factor services provided by a competitive market) nor competition between ‘club’-type producers is present. Only one ‘club’ is in a position to provide the good or service, so that consumers must join this ‘club’ and pay the discriminatory charges asked, or go without the good. In such cases, where the producer has a considerable degree of monopoly power, it is difficult to see how discriminatory charges can be justified by appeal to the ‘club’ principle. The differential charges are fixed by the producer without reference to consumers, who must accept them as a datum when deciding how much to consume – the only ‘voluntary’ decision left with them. Consumers in these circumstances are protected neither by direct association with pricing and production decisions nor by the existence of competition among producers of the good concerned. The distinction between this last formulation of the ‘club’ principle and the earlier ones is clear; it is the difference between my offer (choice) to pay two thirds of the cost of a particular taxi shared with a friend, in preference to travelling by bus, and my choice whether or not to consume electricity at the particular set of discriminatory prices that a monopolistic electricity utility decides
to apply to me. Cases of the latter type are clearly not justifiable on ‘welfare’ grounds; if all that is required to satisfy the ‘club’ principle is that some consumers should pay rather than go without, then any private discriminating monopolist might meet the conditions.
Unfortunately it is only the last and most unsatisfactory type of ‘club’ that is likely to be relevant to the pricing policy of public utilities, whose products often have no close substitutes and whose monopoly power is protected by law, so that the ‘club’ becomes effectively a method of coercion operated by a sole producer.
In summary, there are clear arguments for a multi-part pricing rule only where the services of the indivisible factors (and therefore the ‘standing charges’) can be imputed directly to individual consumers. In such conditions the method avoids the need for interpersonal utility comparisons. This is not so if there are common costs, which is likely to be the general case. In these cases the decision taken about the prices to be charged must involve a value judgement about the distribution of income, and this cannot in general be avoided by an instruction to make use of the ‘club’ principle.
III THE WELFARE MODEL AND UNCERTAINTY
In the simplified conditions of the competitive model so far postulated, the pricing ‘rules’ could be implemented simply on the basis of objective cost computations made by utility managers. The assumptions about knowledge that lie behind the model are such that it does not matter who takes the decisions about the use of factors in production, nor is there any need for economic activity concerned with the discovery of information, framing of expectations, or considering and choosing between alternative and speculative courses of action. Any departure from the ‘ideal’ situation in which the price of any factor (including ‘entrepreneurship’ as rewarded by normal profit) is equal to its value in another
use or to another user must be explained solely by reference to the short-term immobility of factors of production.
A simple model of this kind is inadequate for the derivation of pricing rules intended to have relevance to practical policy. This can be seen by considering the effects of uncertainty.
*71 Once uncertainty is admitted, it becomes necessary to distinguish between the process of decision-taking by which the use of resources is determined (the
ex ante planning process), and the
ex post distribution of factors between uses that is the consequence of that process. The opportunity-cost problems arise at the
ex ante planning stage: costs are incurred when decisions committing factors to particular uses are taken. With uncertainty this
ex ante planning process must involve judgement as well as a capacity for arithmetic; there is no longer any reason why different individuals, working as they must in an atmosphere of doubt and with incomplete information, should make the same assessments or reach the same decisions even in the unlikely event of their acting on the basis of identical data. That is, the
ex post distribution of factors between uses at any time is determined not only by factor mobility but also by the skill of those who plan the use of those factors
In these circumstances the entrepreneurial function cannot be treated simply as a factor of production rewarded in similar fashion to other factors. The decision-taking process is concerned with the selection and implementation of the production plans which, in the view of those taking the decisions, offer combinations of riskiness and expected net revenue superior to those offered by any alternative plans considered. But the implementation of any plan at all involves a risk that the actual revenues and outlays achieved
ex post will differ from the
ex ante forecasts that
provided a basis for action. This risk is borne by those whose resources are utilized in implementing the plan (the ‘owners’). The ‘owners’ and the ‘decision-takers’ need not be identical. The possible combinations of the functions of ownership and planning control (decision-taking) are clearly very numerous, and the returns to risk-bearing and to the planning function are difficult if not impossible to separate in practice, since individuals may share both functions in varying degrees. But the separation is clear in principle.
The reward of the decision-taking function will be that part of the earnings of decision-takers that is not directly dependent upon the
ex post success of their
ex ante planning activities. So regarded, the return to such decision-taking can be treated, like normal profit in the competitive model, as an outlay on a productive factor. But the rewards offered to individual decision-takers will reflect the view taken by owners of their relative abilities; there can be no question of their being treated as homogeneous. The return to risk-bearing, on the other hand, cannot be treated as an outlay at all; its reward is the
ex post (achieved) excess of revenues over outlays (net revenue) in plans actually implemented. It is in no sense a hire payment for a factor, depending as it does upon the ability to obtain a return from the utilization of factors greater than the hire payments that have to be made to those factors. The size of the return obtained is directly determined by the efficiency with which planning decisions are taken
ex ante and by the attitude of risk-bearers to ventures of different degrees of riskiness.
Since net revenue is the return to the essential economic function of risk-bearing, but cannot be treated as an outlay on a factor, it follows that, if factors of production are to be ideally distributed between uses, the total revenues obtained by firms (
ex post) should be greater than their total outlays and not equal to such outlays as in the conditions of the perfectly competitive model. Also the ‘normal profit’ principle cannot be satisfactorily replaced, as a condition of the welfare ‘ideal’ by a requirement that the net revenues obtained by different firms should be equated
ex post. The competitive process does provide a check on the undue divergence of the net revenues actually obtained from different kinds of productive activity, by directing activity towards avenues in which large net revenues seem likely. But even with complete freedom for potential producers to enter any market they wish there is no reason to expect that competition will, or (from the point of view of an ‘ideal’ factor distribution) should, result in a general equality of achieved net revenues. Net revenue depends upon the individual skill of risk-bearers and decision-takers and upon their attitude to risk. If the abilities and risk attitudes of these individuals differ, then net revenues must also be expected to differ. A welfare principle of net-revenue equalization, in accord with the general principle of factor-price equalization, would thus be valid only in a society in which risk-bearers and decision-takers were of precisely equal ability and took the same attitude to risk. Such a situation being unlikely, it seems better to substitute the more realistic, if less precise, formula that some net revenue must be obtained if the employment of factors of production in any use is to be justified, and that some means (such as the competitive process) is necessary to limit the extent of the divergences between the net revenue obtained from different kinds of productive activity.
If this argument is accepted, then a new dilemma arises for public utility pricing policy. The need for skill in making production plans, and the risks involved in implementing those plans, are not peculiar to one form of economic organization. They do not disappear because an industry becomes a public utility; simply the risks are transferred from private owners to the community as a whole. In respect of decision-taking no insuperable difficulties need arise; so long as there is a large private sector, suitable individuals can be hired at prices determined by their earnings in private industry, and their hire prices treated as outlays. The only difficulty in this respect is the discovery of an incentive to efficient
ex ante planning activity that will replace the association of reward with achieved net revenue generally used in private industry. But risk payments cannot be
treated in this way; they are not simply factor outlays. If public utilities are not expected to earn some net revenue, as is the case with the ‘rules’ so far discussed, then factors of production will be utilized in plans that would not be implemented in private industry because the expected returns were too small. On the other hand, public utilities will often have considerable powers of monopoly, so that the competitive process is not available as a check upon the means utilized to obtain revenues. Consequently, if they are required to earn a net revenue, utilities may do so simply by using their monopoly power to raise prices. Some increase in
ex ante planning efficiency may (but need not) also be stimulated by the need to reach a more difficult target. Thus there appears to be some justification for the view that if a public utility, required to achieve a specified net revenue, did so solely by exercise of its monopoly power over prices, yet the need to raise the revenue might serve a useful ‘welfare’ purpose by checking the over-expansion of the public-utility enterprise relative to enterprises in private industry of a similar degree of riskiness. But there seems to be no ‘right’ net revenue that all utilities should be required to earn in all circumstances, since public utilities differ both in riskiness and in the extent of their monopoly power.
The introduction of considerations of uncertainty also draws attention to the problems that would arise for utility managers concerned with interpreting and administering ‘rules’ of the type so far discussed. These problems are particularly important in the case of rules that do not require costs to be covered. For example, the investment principle, interpreted in
ex ante planning terms, requires that managers, when deciding whether or not to create an asset, should base their revenue estimate upon the system of prices (discriminatory or not) that they would expect to maximize such revenues. The asset should be created if any plan shows a potential (
ex ante) excess of revenues over outlays. But if no charge is made for the use of the asset once created, then the plans that prompted its creation will never be implemented. There will therefore be no means of checking upon the efficiency with which the investment decisions are made. This position will be aggravated
by the fact that once charges for the use of durable assets cease to be made, no guidance can be obtained from the success of
ex post (implemented) plans when considering newly current (
ex ante) plans, since the revenues obtained from implemented plans are not an indication of the valuations placed upon the durable factors by consumers. It is difficult to believe that such a situation would be conducive to efficiency in planning the use of factors and hence in the
ex post (achieved) distribution of factors between uses. Similar problems arise with the marginal-cost rule. The marginal-cost-price relationship becomes a manager’s opinion about the results of a marginal increase in factor use in the alternative
ex ante plans considered by him.
*72 There is consequently no possibility of any outside authority checking upon whether a general instruction to implement the marginal-cost rule is being followed, quite apart from the other shortcomings of such a policy. Considered together with the proposition advanced earlier, that what is treated as marginal cost must depend upon the length of the planning period specified, this suggests that the marginal-cost rule could only be made intelligible in an environment of uncertainty if the general rule were replaced by specific individual directions to managers. Such directions would take the form of an instruction to ignore the estimated replacement costs of particular specified durable assets when deciding price policy, which should otherwise aim at the recovery of all outlays. But this would amount to the replacement of the marginal-cost rule by average cost (or multi-part) pricing, associated with a specific subsidy.
It must be concluded that the welfare criteria give rise to no unambiguous general rule for the price and output policy of public utilities, such that for the given distribution of income to which the welfare model refers obedience to that rule must achieve an
ideal use of resources by the utility. An instruction to price at marginal cost, if it was to be intelligible, would need to be supplemented by a specific statement of what costs were to be ignored when fixing prices, in the case of each utility, so that he general ‘rule’ would effectively be replaced by average cost (or multi-part) pricing and specific subsidies decided separately for each utility. Furthermore value judgements about income distribution are unavoidable with marginal-cost pricing. Average cost or multi-part pricing can solve some of the problems, but only if there are no important common costs, or if the ‘club’ principle can be justified in individual cases. In any case any policy ‘rule’ adopted would need adjustment to take account of uncertainty; an optimum use of resources requires that utilities should earn an excess of revenues over outlays, and there is no simple principle by reference to which the appropriate net revenue to be earned on account of the risk factor can be decided. Failure to require an excess of revenues over outlays encourages the use of resources by utilities that could be better employed elsewhere, but a net-revenue requirement may be met by the exploitation of the monopolistic position of the utility concerned. Consequently uncertainty considerations also require the abandonment of general ‘rules’ and the separate determination of pricing policy in respect of each individual utility.
These negative conclusions have an important positive aspect. The failure to establish general pricing rules does not mean that the government need take no pricing decisions. Rather, given the existence of public utilities, it has to consider each utility individually, and decide policy in respect of some or all of the following matters in respect of each one:
1. The net revenue that the utility should be expected to earn.
2. Whether it is considered desirable explicitly to encourage the short-period use of particular durable and specific factors and, if so, what form the requisite subsidy shall take.
3. The nature and extent of the discriminatory pricing to be permitted. That is, if there are common costs, whether these can be satisfactorily allocated by the free use of a ‘club’ principle without this implying a compulsory and undesirable redistribution of income by the utility managers. If the ‘club’ principle is not appropriate, then a decision has to be taken as to what system of charges would best accord with the government’s general policy in regard to income distribution.
4. Whether, quite apart from the considerations at 2) and 3), the industry concerned is thought suitable for use as a means of redistributing income, as a part of the general system of indirect taxes and subsidies. In this regard of course public utilities differ from other industries only in that they are more likely to become the subject of government policy for other reasons,
*74 and in that they provide a convenient method of achieving those ‘indirect’ income redistributions that some economists consider must be one of the purposes of public finance.
Clearly the decisions taken in the case of each utility must be a reflection of the particular attitude of the government concerned. It would therefore appear that, failing some universally acceptable theory of the
public economy by reference to which policy could be decided (and the possibility of such a theory is doubtful), economists would find their efforts better rewarded if they ceased to seek after general pricing rules and devoted attention to the examination of the policies actually adopted by governments, in order to discover their effects and make clear to the government and to the electorate the nature and consequences of the policies actually being pursued. That is, the economists’
general recommendations need to be concerned not with general pricing rules,
but rather with the availability of information about policy and with the methods adopted to keep that policy under review.
A Course in Applied Economics. Cf. also G. F. Thirlby, ‘The Ruler’, reprinted here, pages 163-98; William Vickrey, ‘Some Objections to Marginal Cost Pricing’,
Journal of Political Economy, 56 (1948); Gabriel Dessus, ‘The General Principles of Rate Fixing in Public Utilities’,
International Economic Papers, No. 1 (translation of a report presented to the Congress of the Union Internationale des Producteurs et Distributeurs d’Énergie Électrique, 1949); and T. Wilson, ‘The Inadequacy of the Theory of the Firm as a Branch of Welfare Economics’,
Oxford Economic Papers (February 1952). This list is not comprehensive.
The historical development of the rules and their analytical origins is set out in two articles by Nancy Ruggles: ‘The Welfare Basis of the Marginal Cost Pricing Principle’ and ‘Recent Developments in the Theory of Marginal Cost Pricing’,
Review of Economic Studies, Nos. 42 and 43, (1949-50).
Specific references have been given in the text only where articles are of particular relevance to the issue concerned.
Report of the Committee on National Policy for the use of Fue! and Power Resources (Cmd. 8647), 1952 (Ridley Report), considered the question of whether coal should be priced at marginal cost, and half the members of the Committee in fact favoured the use of some form of marginal-cost pricing.
Econometrica, (1938). Hotelling’s paper was stimulated by the much earlier work of Dupuit, around 1844. The relevant papers have been collected and reprinted with comments by Mario di Bernardi and Luigi Einaudi, ‘De l’utilite et de sa mesure’,
La Riforma sozials (Turin 1932). One of the most interesting papers, ‘On the Measurement of Utility of Public Works’,
Annales des Ponts et Chaussees, (1844), is published in translation in
International Economic papers, No. 2.
inter alia) J. E. Meade, ‘Price and Output Policy of State Enterprise’,
Economic Journal (1944), pp. 321-8, and ‘Rejoinder’ pp. 337-9; P. A. Samuelson,
The Foundations of Economic Analysis, p. 240; R. H. Coase, ‘The Marginal Cost Controversy’,
Economica, N. S. (1946), pp. 169-82; H. P. Wald, ‘The Classical Indictment of Indirect Taxation’,
Quarterly Journal of Economics (1945), pp. 577-97; I. M. D. Little, ‘Direct
v. Indirect Taxes’,
Economic Journal (1951), pp. 577-84.
possibility and not the
fact of compensation is necessary for an economic policy to be accepted as beneficial. However, it has been amply demonstrated that interpersonal comparisons cannot be avoided in this way (cf. M. W. Reder,
Studies in the Theory of Welfare Economics; I. M. D. Little,
Critique of Welfare Economics; W. J. Baumol,
Welfare Economics and the Theory of the State, and the references cited therein). The debate will not be discussed in the text; all that has to be established is that the simultaneous decisions referred to therein are unavoidable, and that the welfare criteria provide guidance about only one of these decisions.
technical indivisibility is not always made clear in the literature (for a clear separation, cf. e.g. Phelps Brown,
A Course in Applied Economics, and Coase, ‘The Marginal Cost Controversy’).
In contrast with the present section, the discussion of the club principle in Section II will be conducted with reference mainly to technical indivisibility. Such indivisibility amounts to no more than the fact that the whole of a productive factor must be employed in order to obtain
any part of the total product of that factor, so that if the factor is an economic good it must have
current alternative uses, and therefore a price (e.g. if a railway carriage can be attached to different trains, opportunity costs are incurred in attaching it to any one train. But no opportunity costs may be incurred in allowing one more passenger to travel once the carriage is attached).
locus classicus of this argument.
technical indivisibility raise similar considerations.
indivisible, in that extra units of product can be carried without cost, there are
no common costs since one van serves only one customer and the transport cost is attributable to that consumer. In these conditions, Coase argued, the price charged should comprise a fixed charge for the transport service and a price per unit for the product. Total costs are then covered, and the additional payment for extra consumption is equal to the price of the product only (i.e. to marginal cost).
The nature of the complications caused by common costs can be illustrated by replacing Coase’s road-system (note 15 above) by a ring road, with the market at the centre and one van serving a number of customers around the circumference, which the van can join at any point. Clearly the pricing problem now becomes much more complex.
temporally indivisible assets. This kind of question cannot suitably be discussed without relaxing the assumptions of the competitive model. The present section therefore ignores these questions of time, which are more fully treated in the following section of the article. It will be appreciated that the criticisms of the two-part tariff and the ‘club’ principle in the present section are in no way invalidated by this simplification.
A Course in Applied Economics, p.260) makes the point very well; poor families in an area may be willing to pay more towards the provision of a playground than richer families in the same area, but there is no presumption that a government will agree that they should. The welfare criteria provide no guidance in such cases since they offer no means of choice between income distributions.
logical reason why only one system of prices should satisfy the ‘club’ principle; what happens, e.g. if the amounts offered to meet standing charges are greater than the total of common costs, but only total cost is to be recovered? In these cases, where more than one set of prices would satisfy the conditions, someone will have to choose between them. Value judgements must be made in the process, and it is difficult to understand why the government should accept those of the utility as superior to its own.
The false plausibility of the argument for voluntary redistribution through the ‘club’ arises from the application of a logical system concerned solely with individual choice and taking no account of the existence of a government with coercive powers, to a situation where governments have to take decisions involving economic matters outside the scope of individual choice. Some attempt has indeed been made to ‘fit’ the behaviour of the public economy into the individual choice (welfare) analysis, by treating the whole of the economy as a ‘club’. This brings out the weakness and unrealism of the ‘club’ argument even more forcefully than the discussion above; it leads to advocacy of an ‘ethically neutral’ system of government income and expenditure, such that the size of the taxes paid and the public services consumed by individuals would be determined by the free agreement of the citizens (taxpayers and consumers) themselves, and to the suggestion that those unwilling to pay such taxes should be treated as ‘pathological’ (see F. Benham, ‘Notes on the Pure Theory of Public Finance’,
Economica, (1934), pp.453-4, and, for a critical discussion, Musgrave, ‘The Voluntary Exchange Theory of the Public Economy’,
Quarterly Journal of Economics (November 1949).
Co-operative Industrial Research. Here the common service is research for a group of firms with a common interest in the results. Firms can, within broad limits, control the direction of research activity, the distribution of benefits between members, and the methods by which common costs are covered. There is also a possibility of using the market as an alternative to the ‘club’. But it is also not without interest, in view of the earlier argument about the role of government (see p. 259), that a decision had to be made as to whether membership should be made compulsory, because the benefits of the cooperative research are not always easily confined to members of the ‘club’.
Critique of Welfare Economics, and W. J. Baumol,
Welfare Economics and the Theory of State). However, such criticism need not concern us here. There is still point in discussing the use of resources in terms of choice, and the logic of the ‘rules’ can be destroyed even accepting the conceptions of the simplest welfare analysis.
Economic Journal, (December 1945), G. F. Thirlby, ‘The Ruler’, and my ‘Uncertainty, Costs and Collectivist Economic Planning’.
An example of a suitable case might be a change of a permanent nature in the geographical environment, as through the diversion of a river.
Review of Economics and Statistics, (November 1955).
I am grateful to Professor H. G. Johnson, to Mr T. Wilson and to colleagues at L.S.E. for reading and criticizing drafts of this article.
Essay 11, Economists’ cost rules and equilibrium theory