The Grid

Some monopolies are rooted in the cost of supplying the market. A city needs water carried through its streets, and a second set of mains may cost more than the rivalry between two suppliers could save. Much of the expense lies in making the network available; another unit carried through it adds comparatively little. One supplier can then serve the city more cheaply than several, each building its own works. The difficulty begins where that economy ends: the saving from a single network gives its owner a position from which to charge the people who depend upon it. The Grid concerns the rules that keep the economy of one supplier while constraining the terms he offers. Those rules must finance the service, encourage the work that makes it cheaper, and protect the quality for which the customer is paying. A low number on the bill is only one part of that task.

The natural monopoly is a comparison between the costs of serving the same demand through one firm and through several. Suppose a network costs a hundred and sixty to keep available for a period, including its capital cost, and each unit supplied through it costs another two. One network carrying a hundred units costs three hundred and sixty. Two identical networks dividing the same hundred units cost five hundred and twenty: the running cost is unchanged, while the fixed expense is paid twice. This is the simple case of declining average cost. The general test is whether dividing the required output among firms raises total cost, a condition called subadditivity; declining average cost throughout the relevant range is sufficient in this single-product case. Now suppose the price buyers will pay for the next unit is twelve minus one-tenth of the quantity supplied. With adequate capacity, fixed quality and no outside harm, the efficient quantity is a hundred, where that willingness to pay reaches the marginal cost of two. Yet charging two for every unit brings in only two hundred. The very price that guides usage efficiently leaves the fixed cost unpaid.¹

Recovering the fixed expense requires a decision about who pays it and through which charge. A public subsidy can cover the hundred and sixty while usage remains priced at two, though the funds must be raised and their own costs considered. A uniform price of four offers another arrangement: demand falls to eighty, revenue is three hundred and twenty, and cost is also three hundred and twenty. The network breaks even, including the normal return already counted among its costs. Twenty units that buyers value above their incremental cost are now left unsupplied; their forgone surplus is twenty. A connection fee combined with a usage charge can separate these tasks. The fee helps meet the cost of keeping the network available, while the charge for another unit can remain close to the cost of providing it. That separation works cleanly when every customer can bear the fee without leaving. When customers differ greatly in their needs, a uniform connection charge can exclude people who would have contributed something toward the network and benefited from using it. The tariff therefore distributes access as well as expense. Its level and its structure both matter, as the preceding plate on The Segment prepared us to see.²

Rate-of-return regulation approaches the financing problem through the firm's accounts. The regulator recognizes operating expenses, depreciation and a return on an accepted capital base, then sets allowed revenue to cover them. This gives the utility a way to recover the cost of maintaining a long-lived system, while giving the public an account against which charges can be judged. But the account also becomes an incentive. Harvey Averch and Leland Johnson showed how a profit-seeking firm can respond when the allowed return on capital exceeds its actual cost and the regulatory constraint binds. Adding capital then enlarges the base on which it may earn the favorable return. In their model the firm substitutes capital for the other input beyond the combination that would minimize the economic cost of producing its output. The regulator may have restrained the return and still encouraged an unnecessarily expensive method of service. The result depends on the rule, the allowed return and the firm's ability to substitute inputs; it does not mean that every regulated investment is excessive. It identifies the decision an inquiry must examine: whether an addition to the network earns its place by the service it supplies, or partly by the allowance attached to owning it.³

A price cap changes the incentive by allowing the firm to keep the benefit of reducing costs during a specified interval. The regulator sets the permitted path of prices in advance, perhaps allowing an index of general inflation less an agreed adjustment for efficiency. Stephen Littlechild's 1983 report on British Telecommunications proposed a constraint of this kind for a basket of monopoly services. If actual costs fall more quickly than the prescribed path, the company retains the difference until the next revision. Managers have a reason to find the saving; customers receive the reduction written into the permitted price path. This also places more of the risk of unexpected costs on the firm. A generous cap can leave substantial gains with shareholders, while an unrealistic one can jeopardize the capacity to supply. And a basket is itself a choice: an average limit can permit one charge to rise while another falls. The rule must specify which services enter the index, how they are weighted, and how long its terms will stand. These details determine what the customer's promised protection actually buys.⁴

Both arrangements require someone to distinguish a genuine saving from a deterioration of the service. A company can spend less by repairing a network more intelligently, or by postponing repairs until failures arrive after the period being measured. A connection charge can remain within its limit while applicants wait longer to be connected. Littlechild expressly included quality protection and monitoring in his proposal, because a tariff guarantee alone would leave such responses open. Information remains central under either system: which costs were necessary, which improvements could reasonably have been made, and which deterioration the customer has had to bear. Cost reimbursement can weaken the reward for economizing; a fixed price can sharpen that reward while increasing the attraction of an unobserved cut in quality. Public ownership changes the people to whom the operator answers, but it leaves decisions about cost, quality and investment to be made. Competition in services that use the network may provide additional discipline where sharing it is feasible, provided the terms of access are workable. The fact that one set of pipes is cheapest establishes a claim about that network, and leaves the organization of neighboring activities open to examination.⁵

The grid brings the volume's question of power to a place where preserving the service and disciplining its supplier must be attempted together. A single network may save resources; the tariff decides how that saving is shared and whether the network can be sustained. The regulatory rule then helps determine how diligently the operator seeks further savings and how faithfully he supplies what the customer was promised. None of these judgments is contained in the number of firms alone. They require an account of costs, an account of alternatives and a way for the people dependent on the system to challenge its terms. Where a second network would be wasteful, the customer may have little immediate prospect of walking away. That gives particular force to the next plate, The Exit: the inquiry into what can discipline a price when the person asked to pay it has nowhere else to go.