Convinced advocates of long-run marginal cost pricing still exist. A recent example in favour of such a pricing principle i s Electricity Pricing, by M. Munasinghe and J. Warford (1982). The aim of this section is to review their arguments. As they refer to Crew-Kleindorfer, the latter's contribution is also briefly reviewed.
2.4.1 Munasinghe-Warford's arguments.
Munasinghe and Warford claim that a tariff based on "strict LRMC" is consistent with the objective of an efficient allocation of resources in contrast to the traditional "accounting approach concerned with recover-ing historical or sunk costs" (p 11). In their analysis, they emphasize the importance of determining the optimal reliability, not assuming some exogenously given constraints to fulfill. According to Munasinghe and Warford, "the optimal reliability rule ensures that the marginal outage and capacity costs are also equal. Therefore, when the system is optimally planned and operated that is, capacity and reliability are optimal -SRMC and LRMC coincide" (p 23). For a more rigorous proof of this result, they refer to Crew and Kleindorfer (1979), Chapter 7 (we shall later also review this briefly). Munasinghe and Warford stress the intimate link between optimal investment and pricing rules (p 25): "A pricing policy based on LRMC effectively permits the burden for the cost-benefit analysis to be placed on the electricity consumer, because he signals the justification of further investment by his willingness to pay the marginal cost of electricity supply" (p 27). According to Munasin-ghe and Warford the estimation and use of the "strict LRMC" is simplest when the system is near optimum.
They are, however, aware of the fact that there can occur significant deviations between SRMC and LRMC when the system plant is subop-timal. When a significant excess capacity occurs, for instance, demand charges could be reduced below the LRMC level until the demand grows sufficiently. They are also aware that when there are, conversely, sig-nificant shortages, the economically efficient short-run solution would be to raise prices to ration the limited supply.
They find, however, such price increases to be impractical; large price fluctuations are, according to Munasinghe and Warford, unacceptable to customers (p 18). They therefore stress the importance of having prices adapted to LRMC that are quite stable over time. This smooth-ing-out of costs during a long period is especially important given the large size or "lumpiness" of the power system investments (p 12).
On the other hand, they have observed the recent advances in solid state switching, metering and communications technology that have made it possible to vary prices continuously instead of relying on a predeter-mined schedule. Prices can then be set at any moment to reflect marginal supply costs (Cf. Vickery 1971).
2.4.2 Munasinghe-Warfbrd's argument reviewed
Most of Munasinghe and Warford's arguments are not new; they repeat what their predecessors have said before. But to motivate LRMC pricing by referring to the lumpiness of the investments in the electricity power system is a quite new and somewhat curious argument. The problem is that the marginal willingness to pay for a customer's demand of an extra kWh can never motivate a lumpy investment such as a nuclear power plant. Such a plant must be motivated by the total willingness to pay of a collective of customers over its expected lifetime, an investment rule which needs to be repeated. The Ii k between pricing and investment is not as simple as they assume.
Even if they are aware of the existence of over- and underequipment situations during long periods, they simply repeat the Boiteux solution and thus dispel such situations. In their world, the possibilities of rapid adjustments of investments to achieve an optimal situation are assumed to be the normal case. We do not need to stress once again how unrealistic such a picture is of the real world of electricity.
On the other hand, Munasinghe and Warford describe how, by technical progress in electronic metering, it might be possible to use instantaneous pricing, at least for bigger customers, in accordance with instantaneous changes in supply and demand. They seem to raise no criticism of or doubts about such a future possibility. This, however, is pure SRMC pricing aimed at establishing short-run equilibria. Will LRMC pricing then be wrong, or was it wrong all the time? In our opinion it was wrong all the time.
2.4.3 Crew-Kleindorfer'8 arguments
Besides treating the pricing problem under uncertainty referred to above, Crew and Kleindorfer have a dynamic analysis of the pricing problem under deterministic conditions (1979, Chapter 7). Their result is the well known one, p = SRMC = LRMC, and in optimum the capital stock is adjusted to this condition. But, like their predecessors, they have assumed continuously variable capacity, i.e. the existence of in-divisibilities is ignored. That a dynamic analysis is applied does not make their assumptions more realistic.
2.5 Conclusions
Since Boiteux, economists have repeatedly proved the equivalence SRMC s LRMC ex post (Anderson and Turvey, Crew and Kleindorfer, Munasinghe and Warford). But the derivation of this optimum condition is made under restrictive assumptions, namely that the capacity is rapidly and continuously variable back and forth. This assumption ignores the existence of ex ante indivisibilities. The ex post lumpiness of investments in the electricity field is not even fully recognized in the models that have been used. Thus as soon as an investments is made, if not before, it becomes an indivisible, irreversible and durable unit.
Under these conditions there is no clear-cut meaning of LRMC even if demand is growing over time. For an industry that faces a shrinking demand, the concept of LRMC is meaningless.
When such indivisibilities exist, it is necessary to use the concept of willingness to pay as an investment criterion (as has been lucidly demonstrated by O. Williamson and A.A. Walters) in order to reach an efficient allocation of resources. The corresponding pricing rule is based on SRMC. In order to make this rule generally valid, the SRMC must be defined to include the potential scarcity rents needed to secure ex ante equilibrium at the given capacity level. With this interpretation of costs the peak-load pricing scheme is a straightforward application of the optimal pricing rule. The same is true of the price variations between two consecutive years if demand or capacity availability vary in an anticipated way. The problem with the Boiteux-Turvey analysis is that prices have been restricted from varying in such a way from year to year.
This restriction that prices have to be constant over time is however not a result of a thorough analysis but rather an ad hoc constraint. We have challenged the validity of this constraint.
The situation with lumpy investments has made it difficult to define the concept of LRMC operationally. The approximate measures used in the electrical industries are average cost concepts; total cost are divided by an expected number of kilowatt hours produced by a planned plant (or several) over its economic lifetime. Thus LRMC pricing as described in theory boils down to nothing more than average cost pricing in practice.
We believe that it would be wise to dispense with this concept altogether and rely on pricing based on SRMC with due consideration of budget constraints and other second best restrictions.