Appellants, large industrial customers of Portland General Electric (PGE) and Pacific Power & Light (PP&L), brought suit under ORS 756.5801 to vacate and remand the Public Utility Commissioner’s (Commissioner’s) orders granting rate increases to PGE and PP&L, which intervened in the suit. Appellants alleged that the “rate spreads” adopted in those orders unlawfully discriminate against appellants because the spreads shift to them, as industrial customers, costs incurred by residential customers. The circuit court upheld the orders. See ORS 756.598. On appeal, appellants contend that 1) the utilities’ rates must be, but were not, spread on the basis of each customer class’ cost of service; 2) no substantial evidence supported *454the Commissioner’s consideration of relative price elasticities, resulting in departure from cost-of-service allocation; and 3) the Commissioner unlawfully rejected as irrelevant evidence of past rates of return from various customer classes. We affirm.
Appellants participated as parties in separate proceedings before the Commissioner involving both utilities’ requests for increased electric revenues. In each case, the Commissioner’s order increased revenues and designed or “spread” rates to allocate the revenue burden among the utilities’ principal customer classes. No party disputes the total rate increases; the only issue is whether the burden was properly spread, i.e. divided, among the various customer classes.
The Commissioner’s rate-making function involves two steps: he first determines how much revenue each utility is entitled to receive; he then allocates the burden of paying that amount of revenue among the utility’s ratepayers.
To determine authorized revenues, the Commissioner projects what a utility’s actual costs will be for the next year. The utility is entitled to have rates set to recover those costs: actual costs equal authorized revenues. The Commissioner performs this task by comparing actual costs derived from the expenses, capital costs and fair return on equity of a selected “test year” with actual test-year revenues. All test-year amounts are “normalized” for nonrecurring items and for anticipated changes; costs that stockholders alone should bear are disallowed.2 If, after all *455adjustments, actual costs will exceed actual revenues under the existing rate structure, the utility is entitled to increase its revenues, by increasing rates, to recover that excess. Both utilities here were entitled to increased revenues.
The Commissioner’s next task is to determine how to set rates to allocate the burden of producing the increased revenue among the utilities’ principal customer classes: residential, commercial and industrial. Allocation among classes is called “rate spread”; the method of revenue collection within each class is called “rate design.” In the present case, the Commissioner determined that the rate spread for both utilities should reflect increasing costs of providing electricity, i.e. “the age of cheap and abundant energy is over and consumers must conserve.” The first decision here is what “cost standard” to use, i.e. what “costs” are relevant to the customers. There are two possible standards: “Long Run Incremental Costs” (LRIC), and “cost of service.”
A cost-of-service standard would in effect require each class to pay its own way. To use it, the Commissioner would have to determine how much each class contributed to test-year actual costs and set rates accordingly for each class to recover those “contributions.” This standard is almost an historic one, depending on the test year used. Because it does not anticipate cost increases attributable to future demand and consumption increases, it is somewhat inconsistent with and separate from an LRIC standard, although the two are not mutually exclusive. See Publishers Paper Co. v. Davis, 28 Or App 189, 559 P2d 891 (1977).
*456The Commissioner adopted LRIC in these cases as the relevant cost standard, as he had in recent years in other rate cases, including Publishers Paper Co. v. Davis, supra, 28 Or App at 192, where we explained:
“The Commissioner is his order determined that customers of the utility should pay a rate which reflected the cost to the utility of providing energy to that class of customer. It was necessary to select a ‘cost’ standard to apply in setting the ‘rate spread.’ * * * The order adopted long run incremental cost (LRIC) as the relevant cost standard for evaluating the proposed increases. LRIC is an estimate of the costs the utility will incur over a period of time, usually ten years, in constructing additional facilities and in operating these additional facilities to satisfy increased future energy demand. These costs are allocated to the various consumer classes. In other words the utility makes an estimate of the cost of future construction and operation of generating facilities needed in order to meet the future demands for energy from each specified class of consumer * * *. By increasing present rates to reflect these future costs it is hoped the consumers will respond by decreasing their demand thereby reducing the need for increased investment in the future.”
LRIC proponents argue that rates set at these marginal cost levels promote “economic efficiency” (efficient allocation of scarce resources) by signalling to consumers the rising cost of electricity. Appellants disagree with that assessment but do not here contest the Commissioner’s authority to adopt LRIC as the relevant cost standard.
LRIC proponents presume that energy demand is “price elastic” or “price sensitive”: consumers adjust their demand and consumption in response to price; higher prices create pressure to reduce demand and consumption or at least to lessen increased consumption. If consumers respond to that pressure, the need for future construction will be reduced and resources will be conserved.3 This is the “inverse elasticity rule” or theorem.
*457Since the Commissioner first adopted the LRIC standard in electric rate cases, LRIC has exceeded the “actual costs” that equal authorized revenues. In the cases now appealed, appellants presented the following figures:
PP&L PGE
Total LRIC $549,728,000 $684,374,000
Actual Costs (Authorized Revenues) 301,275,000 434,522,000
Excess LRIC over Authorized Revenues $248,275,000 $249,852,000
This excess LRIC over authorized revenues is projected future costs less current costs, and is called a “revenue gap.” Because the utilities may not collect revenues in excess of actual current costs, the gap must be “closed” or reduced to zero.4 The gap is closed by first comparing actual costs with LRIC to determine a variable that may be expressed as a ratio or percentage:
*458actual costs = authorized revenues = X(LRIC)
actual costs = authorized revenues = X = Y/100
LRIC LRIC
That variable or percentage is how much of total LRIC may be recovered as authorized revenues. PP&L’s costs were 54.83 percent of its LRIC; PGE’s were 63.49 percent. In effect, each percentage is a measure of how their current costs compare with their projected future costs.
To spread the burden of projected future costs, the Commissioner must determine how much of it each customer class must pay. This is where LRIC becomes relevant. The Commissioner found it important to signal customers that electricity will be more expensive if demand grows; LRIC is an estimate of that expense. Each class contributes to the LRIC for the total system. Because each class contributes differently, it is important to determine each class’s LRIC and to set rates that will convey to each class its contribution to future high rates, i.e. to set rates to recover the highest percentage of LRIC without exceeding authorized revenues. If those who would contribute most to LRIC instead reduce consumption and demand for new facilities, rates will not increase as much as predicted and all customers benefit.
To determine how much each customer class had contributed to LRIC, the Commissioner in these cases broke down total system LRIC to three basic components: “demand” or “capacity” costs, “energy” costs, and “joint and customer” costs. Both demand and energy LRIC may be generally described as costs of supplying peak capacity and energy in excess of a minimum system; joint and customer costs are costs of connection to a minimum distribution system (with sufficient poles, lines and transformation equipment to deliver one kilowatt hour of energy to each customer) and costs of meter reading, accounting and billing, i.e. customer service costs. Most joint and customer costs are incurred in serving residential and some commercial customers; few, if any, are attributable to large industrial customers, as the table in the margin demonstrates.5
*459Having determined LRIC by class, the Commissioner may or may not set rates to recover from each class the same percentage of its LRIC as is used to close the revenue gap between the utility’s total authorized revenues and its total LRIC. If he does, he is said to apply the “equal percentages rule.” If he does not, he must justify the discrimination.6
Although joint and customer costs are part of total LRIC, in the present case the Commissioner excluded them from the rate spread because he found: that they were “relatively” inelastic or insensitive to price variations; that authorized revenues would not exceed demand and energy LRIC, which are more elastic than customer costs; and that this rate spread would recover the highest possible percentage of elastic or sensitive LRIC components, thus conveying to customers the signal to conserve. This was a broad application of the “inverse elasticity rule.” The Commissioner then set up a ratio to compare authorized revenues with total-system demand and energy LRIC (total LRIC less joint and customer LRIC) and applied that ratio or *460percentage equally (the “equal percentages rule”) to each customer class’ combined demand and energy LRIC (class total LRIC less class joint and customer LRIC) to determine each class’ revenue burden.
This application of the “equal percentages rule” presumed equal elasticity among classes of demand and energy costs, because there was no available empirical evidence of relative elasticities among classes. Evidence of relative elasticities compared system-wide customer costs with demand and energy costs; both a staff witness and a utility witness testified that customer costs were less elastic than were demand and energy costs and that, absent contrary evidence, class elasticities must be presumed equal in order to apply properly the economic theory of LRIC-based rates.
The resulting rate spreads recover authorized revenues as high, equal percentages of class demand and energy LRIC, rather than as lower percentages of class total LRIC. Appellants argue that this method spread customer costs among all customers, even though some costs were incurred almost exclusively by residential customers, so that the large industrial class would subsidize the residential class.7 Appellants do not contest either the utilities’ right to a rate increase or the LRIC foundation in these rate spreads. They do object to the way LRIC exclusions were used to shift costs among classes so that the rates do not mirror cost of service. They also argue that the record lacks substantial evidence to support the Commissioner’s use of the “inverse elasticity rule.” PGE supports appellants. *461PP&L supports the Commissioner, except as to the substantial evidence question.
SCOPE OF REVIEW
Our scope of review is not in issue, but we note:
1) utility regulation, including ratemaking, is a legislative function subject only to constitutional limits and those of the Commissioner’s express, legislatively delegated broad powers;
2) the Commissioner represents utilities’ customers and the public to obtain for them fair and reasonable rates, ORS 756.040, but is not required to use specific formulae other than to exclude certain costs that must be borne by the utilities’ stockholders, see ORS 757.355;
3) the court, in a suit in equity pursuant to ORS 756.580 challenging the Commissioner’s broad authority, may not substitute its judgment for that of the Commissioner on any finding of fact supported by substantial evidence, ORS 756.598(1); and
4) the Commissioner’s order must disclose a rational relationship between the facts found and legal conclusions reached to demonstrate that the action is not arbitrary, but the reviewing court need not agree with the agency’s inferences or reasoning to uphold the order if it contains sufficient fact findings and legal conclusions to determine if the reasoning is rational and to allow the court to test the agency’s actions against its grant of power. See *462Publishers Paper Co. v. Davis, supra, 28 Or App at 193-196. We turn now to the various facets of appellants’ attack on the Commissioner’s order.
COSTS OF SERVICE
“Setting rates to produce sufficient revenue to allow a utility a fair return is only one of the functions assigned to the commissioner in rate making. * * * [H]e is to ensure that no utility collects from any person, more or less compensation for any service than is collected from any other person for like and contemporaneous service.” Publishers Paper Co. v. Davis, supra, 28 Or App at 196.
Appellants first argue that the Commissioner has failed to allocate rates by cost-of-service and therefore has unjustly discriminated against industrial consumers, in violation of certain provisions of ORS ch 757.8 Nothing in ORS chapters 756 or 757 limits rate allocation to cost-of-service. The legislature has granted the Commissioner broad discretion in establishing rates, subject only to prohibitions against unjust, unreasonable or unjustly *463discriminatory rates. Rate-making is a purely legislative function, involving broad discretion in selecting policies and methods of allocating rates among classes of customers. Cost-of-service and LRIC are two permissible choices. See Publishers Paper Co. v. Davis, supra, 28 Or App at 199; American Can Co. v. Davis, supra, n 6, 28 Or App at 228-29.
ORS chapter 757 prohibits discrimination among members of the same consumer class, but does not disallow discrimination among classes, unless that discrimination is unjust. Because cost-of-service is but one possible allocation method, the Commissioner’s choice of an alternative method is not by itself discriminatory, unreasonable or unjust. That he chose to rely on cost-of-service in American Can and Publishers Paper does not preclude a different choice now to meet changed conditions or to overcome possible ineffectiveness of an earlier method. In those cases, too, this court approved as not discriminatory rates that resulted in different percentage increases for each customer class.
In essence, appellants ask us to choose between two economic theories. However, what is fair and reasonable preference or discrimination is within the Commissioner’s authority to determine; we will not disturb the Commissioner’s judgment if it is supported by substantial evidence. ORS 756.598(1). The real issue, therefore, one of substantial evidence.
EVIDENCE AS TO PRICE ELASTICITY
Appellants’ real challenge is to the Commissioner’s premise that the excluded costs are more inelastic than other costs and will not respond to rate increases designed to encourage conservation so quickly as will the included costs. In spreading rates to encourage conservation, the Commissioner believed that recovering a higher percentage of that part of LRIC that was elastic would signal consumers to reduce consumption and demand, but that recovering a high percentage of inelastic costs, such as billing costs, would not encourage consumers to reduce those costs and would not accurately reflect his goal of reducing energy consumption and demand, as opposed to *464consumption of and demand for services and delivery systems.
Appellants and intervenors argue that evidence of relative elasticity of certain costs for certain classes was insubstantial and the premise, therefore, improper. PP&L cites two California rate cases where similarly premised rate decisions were overturned because, “* * * without some expert testimony or empirical data reflecting elasticity of need and demand for the various groups, it cannot be determined which plan will result in least usage.” California Mfrs. Ass’n v. PUC, 24 Cal 3d 251, 259, 155 Cal Rptr 664, 595 P2d 98, 102 (1979) (emphasis added); see California Mfrs. Ass’n v. PUC, 24 Cal 3d 262, 155 Cal Rptr 670, 595 P2d 104 (1979).
We have reviewed the testimony of two expert witnesses. One, Wallace Gibson, was a PUC staff member; the other, John Shue, was in PP&L’s employ. Although Shue disagreed with Gibson on excluding joint and customer costs from LRIC, he did agree that they were relatively inelastic compared with other LRIC costs. He further assumed that if elasticity were considered, relative elasticity of various costs (excluding joint and customer costs) must be presumed similar for each class because there was then available no empirical data to the contrary.
Appellants and intervenors now argue that this evidence was insubstantial because it is not “factual” or, in any event, because it is insufficient to establish that the Commissioner’s rate design would encourage more conservation than would another design. Assuming without deciding that the California test is the proper test, expert testimony or empirical data would be necessary.
There was no empirical data available to prove that the Commissioner’s approach would succeed. However, the experts in these cases testified as to their familiarity with at least one professional article that attempts to “prove” the theory, as a mathematician “proves” a theorem or as Einstein “proved” relativity. Such a theory is a working but not conclusively established factual or conceptual analysis.
*465The Commissioner’s goal in spreading rates was to encourage conservation. All parties agreed that if costs were elastic, i.e. sensitive to price, higher prices would encourage less demand and consumption. The expert witnesses agreed that recovering a high percentage of LRIC, i.e. spreading rates to reflect more closely the increasing future cost of supplying electricity, would convey to consumers the message to conserve.
The Commissioner needed to determine which costs were price sensitive. The experts’ opinions were that joint and customer costs were relatively inelastic and that there was no basis in theory or fact to assume that various classes had substantially different elasticities. The Commissioner therefore deleted inelastic costs from consideration in order to recover a higher percentage of elastic costs. He applied the same percentage to elastic costs of each consumer class. This allocation is supported by the experts’ opinions and is consistent with the rate-spread goal.
Appellants’ complain because their inelastic costs estimated under this plan were nil; therefore, they have more elastic costs subject to recovery, i.e. more of their are assumed to be price sensitive than those of residential consumers. PP&L also argues that the Commissioner could not assume that industrial consumers’ elastic costs were as elastic as those of residential consumers without evidence to that effect. However, its own expert testified that no empirical data existed to show otherwise and that he agreed with the general theory of relative elasticity. He disagreed with excluding inelastic costs from LRIC, because he did not believe it would better convey appropriate price signals to consumers; but he also said that the staffs design would be as economically efficient, i.e. would produce the same level of savings to the utility, as would a design based on total LRIC.
The evidence presented by the experts was that several, reasonable economic theories existed to accomplish the conservation goal. On the basis of those opinions, the Commissioner chose the theory he believed would encourage consumers to reduce elastic costs and would produce savings to the utilities and thus to all their customers. That *466choice was within the Commissioner’s discretion and was supported by substantial evidence in the form of expert testimony. See American Can Co. v. Davis, supra, n 6, 28 Or App at 229.
RATES OF RETURN
Appellants’ final contention is that the Commissioner erred in refusing to consider evidence of rates of return by customer class. This evidence was based on historical costs. Because appellants here expressly acquiesced in the Commissioner’s use of LRIC, which is based on future, not historical, costs, their proposed evidence would have been irrelevant to a rate spread based on LRIC; the Comissioner properly refused to consider it. See OAR 860-14-045(1).
Appellants and intervenors, to the extent they support appellants’ claims, have not shown by clear and satisfactory evidence that the orders are unreasonable or unlawful, as required by ORS 756.594.
Affirmed.
ORS 756.580 provides:
“(1) A party to any proceeding before the commissioner, when aggrieved by any findings of fact, conclusions of law or order, including the dismissal of any complaint or application by the commissioner, may prosecute a suit against the commissioner to modify, vacate or set aside such findings of fact, conclusions of law or order.
“(2) Such suit may be commenced by any party so aggrieved in the Circuit Court for Marion County, in the circuit court for the county in which any hearing has been held in the proceeding in which the order was made, or in the circuit court for the county in which is located the principal office of any defendant in any such proceeding before the commissioner, and jurisdiction of any such suit hereby is conferred upon the circuit court for any of such counties to hear and determine such suit.
“(3) In such suit, a copy of the complaint shall be served with the summons. The commissioner shall serve and file his answer to such complaint within 10 days after the service thereof, whereupon the suit shall be at issue and stand ready for trial upon 10 days’ notice by either party. All suits brought under this section shall have precedence over any civil cause of a different nature pending in the court, and the circuit court always shall be open for the trial thereof.
“(4) Unless application is made for rehearing or reconsideration of the order, any such suit must be commenced within 60 days after the date of service of the order in the proceeding before the commissioner. If application for rehearing or reconsideration is made, such suit must be commenced within 60 days after the date of service of the order denying rehearing or reconsideration or the date of service of the new order if rehearing or reconsideration is granted. If an order granting or denying an application for rehearing or reconsideration has not been made by the 60th day after the filing of the application, an order denying the application shall be considered served on the 61st day after such filing.
“(5) Application for rehearing or reconsideration need not be made as a condition of judicial review.”
The Commissioner adopted 1979 as the test year for PP&L, after denying the motion of an advocate organization for low-income citizens in southwest Oregon to adopt an historic test year. He adopted 1980 as the test “period” for PGE:
“PGE proposed using as a test period the calendar year, 1980. Staff joined PGE in using this test period. Other parties raised no objection. The rates authorized will not go into effect before January, 1981; therefore, 1980 is representative of the period during which the rates will be in effect and is adopted.
“When an historical test period is used, adjustments to the test year data are made to remove abnormal events not expected to recur and adjustments are made to include the effect of known changes in data which are expected to persist into the future. When a future test year is used, the data is drawn *455from budget figures and financial models of the utility. Abnormal events of the past are therefore excluded and all known future changes are included. For most utilities, rates are expected to be in effect for longer than a one year period. Because of this, specific events occurring in the first year that the new rates are expected to be in effect are sometimes excluded from consideration as abnormal.
“Because PGE has been filing for rate increases at least annually for the past six years and is expected to continue its practice, this order will consider and accept for ratemaking purposes the financial effects of certain events expected to occur in 1980, but may not be expected to continue thereafter.” Order No. 80-021 at 24 (Jan 14, 1980).
In the PGE Order No. 80-021, supra n 2, at 17, the Commissioner found:
“From the period December, 1969, to December, 1978, the average sale to residential customers has actually declined slightly, despite an increase in average size of houses and a large increase in the use of electric heat. In 1978, a total of 19,363 living units were connected, and 70.2 percent of them heat *457electrically. The reduction in individual customer consumption is partly attributable to energy conservation measures. However, because of the average annual rate of customer growth, the total annual growth rate for all kilowatt-hours sold to ultimate consumers has been 4.3 percent during this 10-year period — a modest load growth compared to previous 10-year time spans. Owing to inflation, construction delays, safety modifications, and environmental protection additions, the annual growth rate of plant for the same 10-year period has been 16.9 percent, or 12.6 percent above PGE’s annual load growth rate.”
LRIC and the revenue gap are, we gather, not unlike legal fictions: they do not exist other than in theory (LRIC is an educated guess, so the revenue gap can be no more than an educated guess). They approach reality only when the theory is applied to real actions and numbers, as in allocating the actual revenue burden. In the present case, the Commissioner excluded from actual costs a number of items properly borne by shareholders pursuant to ORS 757.355, newly enacted in 1979. Thus, the utilities cannot recover those amounts. However, LRIC is an estimate of future costs and necessarily includes capital costs of facilities not currently serving customers. Its application does not mean that utilities recover those costs or that customers pay those costs presently; rather, it is used in devising a mathmatical formula that efficiently spreads rates to effect the goals of ORS 756.040(1): “to protect [the] customers, and the public generally, from unjust * * * exactions * * * and to obtain for them adequate service at fair and reasonable rates.” Here, the Commissioner believed that such rates in the future depend on present conservation to minimize the need for more construction and consumption of scarce resources.
Appellants presented the following table compiled from the order and record ((000) omitted).
*459
See Publishers Paper Co. v. Davis, 28 Or App 189, 559 P2d 891 (1977), and American Can Co. v. Davis, 28 Or App 207, 559 P2d 898, rev den (1977), where the commissioner designed rates to recover different percentages of class LRIC, with highest for industrial customers to bring them closer to their full cost burden, because residential customers had since 1962 borne a greater share of rate increases.
Appellants presented the following comparison of revenue burden to show shift of costs to the large industrial class from the residential class:
*461
ORS 757.020 provides:
“Every public utility is required to furnish adequate and safe service, equipment and facilities, and the charges made by any public utility for any service rendered or to be rendered in connection therewith shall be reasonable and just, and every unjust or unreasonable charge for such service is prohibited.”
ORS 757.310 provides:
“(1) Except as provided in ORS 757.315, no public utility or any agent or officer thereof shall, directly or indirectly, by any device, charge, demand, collect or receive from any person a greater or less compensation for any service rendered or to be rendered by it than:
“(a) That prescribed in the public schedules or tariffs then in force or established; or
“(b) It charges, demands, collects or receives from any other person for a like and contemporaneous service under substantially similar circumstances.
“(2) Any public utility violating this section is guilty of unjust discrimination.”
ORS 757.325 provides:
“(1) No public utility shall make or give undue or unreasonable preference or advantage to any particular person or locality, or shall subject any particular person or locality to any undue or unreasonable prejudice or disadvantage in any respect.
“(2) Any public utility violating this section is guilty of unjust discrimination.”