Opinion for the Court filed by LEVEN-THAL, Circuit Judge.
LEVENTHAL, Circuit Judge:These cases arise from the final phase, Phase 9, of the Civil Aeronautics Board’s (CAB or Board) comprehensive investigation and re-evaluation of domestic air fares, the Domestic Passenger Fare Investigation (DPFI).1 Phase 9 (Fare Structure) was mainly concerned with establishing a formula for coach fares as a function of distance. There have been no challenges to the formula adopted by the CAB. The present challenges concern the Board’s determination of the proper relation (a) between coach class fares and economy class fares (petitioner, Continental Air Lines; intervening respondent Trans World Airlines), and (b) between coach and first class fares (petitioner Northwest Airlines; intervening petitioner, American Airlines).
I. BACKGROUND
Hearings were held between June 29 and September 30, 1971 before Administrative Law Judge Robert M. Johnson. ALJ Johnson issued his initial decision on April 7, 1972. He found that coach fares should be cost-related, i.e., geared to the cost structure, but also reflecting other considerations (notably value of service), rather than be cost-based, i.e. based strictly on costs. He recommended maintenance of the then existing ratios of the 3 permanent class fares — economy fare at 90% of coach fare, first class fare at 130% of coach fare.
In its opinion and order of March 18, 1974, the CAB found that, to be reasonable, permanent fares, such as first class and economy, had to recover their full share of allocated costs.2 In the case of economy class, the CAB found that the only significant difference from coach class in the service provided, and hence in the cost of providing it, was the absence of the complimentary meal provided on coach flights with meal service.3 Hence, economy fares *338could not be set at more than $4 below coach.
As to first class fares, the CAB mandated that minimum first class fares be increased, over a period of 2V4 years, to 150-163% of coach fare, depending on distance.4 These fares reflect the CAB’s determination of the relative cost of first class service, ascertained primarily by allocating total costs by the cabin space occupied, but also including the cost of a higher level of cabin amenities.5
These fare decisions are based on the CAB’s conclusion that unless fares of the different classes reflect the cost of providing the respective services, there will be a subsidy between classes in the long run. In particular, the CAB found that allowing first class and economy fares below their proportionate costs would burden the predominate coach passenger in the long run, either in the form of higher fares or poorer service.6
II. ECONOMY FARE DETERMINATION
A. Continental’s Petition for Reconsideration and Motion to Reopen the Record
On April 29, 1974, Continental and Northwest, as well as other carriers, petitioned the CAB for reconsideration of its March order. In addition, on October 31, 1974 Continental moved to reopen the record on the economy fare to consider the allegedly increased need for a below-coach fare due to the drastic reduction of discount fares subsequent to the 1971 hearings, and to allow Continental to prove that the correctly allocated cost savings of economy vis á vis coach was greater than $4.7 On December 27, 1974, on reconsideration, the CAB reaffirmed its March decision, making certain adjustments, such as delaying the schedule for increasing first class fares by nine months.8 At the same time, the Board denied Continental’s motion to reopen the record. It ruled that Continental’s proffer of increased use of economy class, with the elimination of most discount fares, was irrelevant to the Board’s determination that in order to be reasonable in the long run fares for a class of service must reflect pertinent costs for the service. It further stated that Continental had had its chance to support its proposed lower economy fare on the basis of lower costs, knowing as it did that costs are, at least, a prime factor in fare setting.9
In support of its challenge of the Board’s refusal to reopen the record, Continental asserts that it was not on notice of the CAB’s intention to require permanent fares to recover their full share of cost until the decision in the U. S. Mainland-Hawaii Fares Investigation,10 in which the Board first required economy fares to cover full costs, and which came six weeks after the ALJ’s decision in the present Phase 9.
The DPFI was instituted in 1970 as a general investigation of all aspects of passenger fares.11 In the decade since the previous comprehensive fare investigation, the 1960 General Passenger Fare Investigation,12 there had been great growth in the airline industry, the introduction of jet aircraft, and the imminence of the introduction of wide-bodied jets. In the DPFI the CAB set out to determine fare policy, balancing the interest of consumers, the air *339carriers, and the national interests in developing air transport and defense. One of the parties to the investigation, the Department of Transportation (DOT), urged the Board to adopt fares based entirely on the cost of the service (including a fair return), based on its economic theory of the airline industry.13 Although Continental presented data on cost savings in economy class to support its 85% of coach fare proposal, it did not claim that the 85% was justified “based strictly on cost.” 14 It advocated a “cost related” rather than a “cost based” approach. Continental justified the 15% lower economy fare on grounds of: the public interest in a low fare service, the effect of economy fare in generating new air traffic, value of service, and increasing management flexibility. The Board rejected this position, adopting the position that permanent fares should be cost based. Continental had full opportunity to make the best case possible for its position and cannot now claim unfair surprise because the Board adopted a competing one.15
In its motion to reopen the record, Continental sought the opportunity to prove that the “fully allocated” cost difference between economy and coach justified a rate difference in excess of the 10% recommended by the ALJ.16 The request to reopen the record to prove a greater economy/coach cost difference was based on Continental’s argument of unfair surprise in the Board’s insistence on cost-based fares, which we have rejected.17 In addition, Continental claims that changed circumstances between 1971 and 1974, primarily the elimination of discount fares, led to a great increase in the use of and “need” for the economy fare at 85-90% of coach, rather than the $4 difference on meal flights established by the Board.18 Of course, it is unfortunate when the Board must rely on a three year old record. Yet that record is supplemented by official notice, as appropriate, of schedules, fares, traffic, and revenue data. In the nature of things, a rulemaking completely reevaluating the domestic fare structure takes time. In the absence of a compelling showing of unfair surprise or changed circumstances, there is no reason to mandate further delay and reopening of the record. The Board is correct in saying that Continental’s contentions of changed circumstances are “of no decisional significance” 19 in light of its policy determination of the need for permanent fares to return their full share of costs in the long run.
*340B. Record Support for the Board’s Theory
We turn now to the record support for the premise underlying the Board’s Phase 9 structure determinations, that all permanent fares should be proportionate to the cost allocated to each type of service. The record material appears in Phase 5, which contains both evidence submitted by the Department of Transportation, and its acceptance in the conclusion of the Board, to the effect that in the long run there are roughly constant returns to scale in the airline industry, i.e., that an increase in size of operations does not produce a reduction in unit costs,20 and that hence there is no cost saving by permitting below-average cost fares to generate additional air traffic, except as a temporary measure to fill excess capacity. Another way of putting it, at least as we understand the matter, is to say that once a carrier reaches the level of air service of these regulated trunk carriers, there are no further economies of scale.
The Board’s conclusion was used in Phase 5 in support of the decision that discount fares would only be permitted (a) for short periods (18 months or less), and (b) on a showing that they would generate sufficient new traffic to offset diversion from regular coach class plus the additional cost of carrying the new traffic (the profit impact test).21
Stated in more detail, the Board’s approach in Phases 5 and 9 runs as follows: All costs are variable in the long run. There are constant returns to scale. Traffic-generating discount fares do not produce cost savings but rather burden the normal fare passenger. There will be more traffic in classes where rates are below cost of service than there would be if those classes were paying their full share of costs. The bargain for economy class will lead to an increase in the average load factor impairing the service received by the coach passenger by reducing flight availability, *341both in flexible scheduling and reasonable assurance of a seat in peak periods when flights run full. In the alternative, or in combination, the airlines could increase capacity,22 which would result either in increased fares for coach passengers to cover increased capital costs, or if such increase of coach fares were denied would result in an adverse effect on the long term financial health of the carriers, a condition that would, in turn, adversely affect service in the long run. In sum, the theory predicts that permanent discounts or below-cost fare classes will burden the coach passenger in the long run in poorer service (higher load factors or financially weak carriers) or higher fares to pay for additional aircraft.
This is a case that turns not on finding of fact, where the legal inquiry on review is whether there is substantial evidence probative of the fact, but rather on the legal issue whether an agency is arbitrary in adopting a policy based on an economic approach or theory. The Board has adopted a policy for the long run, adopting an economic approach that is based on industry data, as analyzed in studies of economists.23 The data-cum-analysis provide support for the Board’s order in the sense that it shows the Board’s approach is reasonable, even assuming that the underlying theory is in the realm of expert debate rather than incontrovertible truism. The test of reasonableness is satisfied unless the court is prepared to intervene on some principled basis, say, to assert that the approach is inconsistent with facts of experience that have been established by proof or judicial notice.
Here the underlying premise, constant returns to scale for trunk carriers, becomes a linchpin of regulatory policy because it is combined with an economic theory that involves a projection of long run carrier response to conditions of below-cost fares for classes of service. This projection is not capable of immediate verification either by way of proof or disproof. The Board faced a similar problem in regard to the long term effect of discount fares in Phase 5 of the DPFI.24 The Board found support for its theory in the Phase 5 record in the repeated testimony of many carriers that in planning their needs for capacity additions, they considered the total projected reserved seat, i.e., not standby, passenger volume.25 That is, the purchase of new aircraft was approached on the assumption that expected passengers had to be flown, regardless of whether at a normal or discount fare.
Although this support appears in the record in a different context, we cannot say that this economic regulatory theory cannot travel well, or that its use in the economy fare context26 reflects such differences of *342condition that the transfer theory must stand condemned as bereft of reasoned judgment. There is no testimony in the record of such differences of condition, and we have no basis to insist that such differences exist. Indeed, assuming merit in the theory of constant returns to scale for trunk carriers, it seems reasonable to project a more confident application to a permanent, regular reduction in fare for economy service than to discount fares. The economy class service at issue is the same as coach except for meal service. Continental offers economy class service in all its markets.27 In contrast, discount fares have restrictions such as “blackout” periods and length of stay restrictions, to divert traffic and smooth out the peak load periods. In general, it would not seem unreasonable to treat economy passengers as essentially equivalent (“additive”) to coach passengers in terms of future capacity needs.
A theory of ratemaking must be reasonable, explained, and supported, but is not subject to the same substantiation principle as the substantial evidence test applicable to fact-finding. “Although this is a finding, it is in the area of prediction and projection, as to which particular latitude is accorded the agency.” Western Air Lines v. CAB, 161 U.S.App.D.C. 319, 327, 495 F.2d 145, 153 (1974). The Board has the function of regulation of rates but not direct regulation of service, and it must discharge its regulation of rates in the light of a projection of the service that will be provided by the carriers in the light of the rate levels and rate profiles set by the Board. Judicial respect for that kind of agency projection is mandated by Mobil Oil Corp. v. FPC, 417 U.S. 283, 318-19, 94 S.Ct. 2328, 41 L.Ed.2d 72 (1974).
The Board’s approach to the economy fare issue is sustainable as (a) consistent with respectable economic theory, (b) applied in a situation of a regular continuing service that is identical to basic coach service except for the meal, (c) supported by general economic data on costs of the airline industry, and the carriers’ testimony in Phase 5. The Board’s general insistence that fare differences for variations in service be restricted in extent to the cost differences resulting from the variations28 is unassailable on review — unless we ascertain that countervailing ratemaking policy factors are mandated by the Federal Aviation Act. We now consider whether there is such a showing in this case.
C. CAB Consideration of Ratemaking Factors
Continental claims that the Board erred in relying solely on costs to the exclusion of the other considerations set forth in the Rule of Ratemaking, § 1002(e) of the Federal Aviation Act of 1958.29 We pointed out, *343in our 1974 American Airlines opinion upholding the Board’s determination of interline joint fares in Phase 4 of the DPFI:
Of the many factors properly considered in setting rates, cost of service is in most instances the most important, J. Bonbright, Principles of Public Utility Rates 67 (1960), and has frequently predominated in many areas of regulation, (citations omitted). The Board is not required to deviate from the cost principle absent countervailing considerations recognized by statute.30
Continental distinguishes American as involving an order “eliminating the above cost fares which unnecessarily burden the public,”31 while here the Board is requiring an average increase of discount fares, allegedly in conflict with the public interest in law fares 32 and having an adverse effect on “the movement of traffic.”33
Section 1002(e) is a reenactment of the provision in the original 1938 statute. Section 1002(e)(1) does identify the public interest in low fares, but it lies in securing “the lowest cost consistent with the furnishing” of “adequate and efficient service.” Subsection (5) expressly identifies the carriers’ need for revenues sufficient to enable efficient management to provide such service. Continental does not seriously challenge the Board’s finding of $4 as the cost savings of not providing a coach meal. Continental’s proposal to offer an economy service at 85-90% of coach is price competition.34 Price competition is one of the chief objectives of the antitrust law, and almost universally in the public interest in the unregulated sectors of the economy not properly subject to restraints. However, the airline industry is regulated, by the CAB. The legislative history of the Civil Aeronautics Act of 1938 makes it clear that the law delegates authority to the agency to cope with and protect against disastrous rate wars and cutthroat competition.35 The Act does embody a general policy that regulation be conducted in such a way as to respect the working of competition among entrepreneurs risking private capital. Continental Air Lines v. CAB, 171 U.S.App.D.C. 295, 519 F.2d 944 (1975), cert. denied, 424 U.S. 958, 96 S.Ct. 1436, 47 L.Ed.2d 365 (1976). Continental is willing to take the risks and burdens of price competition to increase its market share and claims that it will not have to add capacity due to economy service because in the absence of such incentives its route structure leaves it with *344excess capacity.36 However, the Board has authority to state the basic ground rules for rate competition so that one carrier does not imperil general industry performance.
In Phase 9 the Board established the fare structure for the whole airline industry. The Board has made a not unreasonable determination that where a rate structure generally supplies the industry with reasonable returns that match cost of service, the approach of offering a particular below-basic service at a discount that exceeds the cost savings will operate in the long run to burden the basic coach passenger, with poorer service or higher fares. That determination is binding on this court, at least in the present state of the record and experience. It means that we cannot require the Board to permit Continental to impose these risks on the air transport industry.
Similar reasoning disposes of Continental’s contention that the Board failed to consider the advantageous effect on “the movement of traffic”37 of a significantly lower basic fare to allow people who can’t afford coach fares to fly. Clearly, Continental’s proposed economy fares would make airplane travel more available to the less affluent and increase traffic volume. However, the Board has considered this effect and concluded that the result, unjustified by the cost of the difference of service, the meal, would be a burden on the coach passenger in the long run.38 Without considering whether the Act required the Board to adopt that approach, we hold that the Act did not prohibit that approach.
D. Inconsistency With Other Parts of the DPFI
Continental argues that the Board’s rigid, categorical application of its cost-based policy to the economy fare is arbitrary and inconsistent with the Board’s concurrent treatment of other non-cost-based fares. Continental ^ identifies these contrasts: (1) As to economy service, the Board imposed cost-based fares immediately, while it adopted a gradual approach in reaching cost-based fares in the cases of (a) differentials for discount and first class fares and (b) the congruence of costs and fares as a function of distance (the “fare taper”). (2) As to economy service, the Board decided to allow no fare difference between coach and economy fares on flights with no meal service, while it permitted the same fare differential to be maintained for first class fares as against coach fares even on flights where no meal is served. Reasoned decisionmaking requires an agency to explain changes of policy from past decisions and apparent inconsistencies within the same decision.39
1. Distinction between Board’s policies for discount and economy fares
In Phase 5 of the DPFI, the CAB decided that discount fares would only be permitted to alleviate temporary excess capacity conditions to prevent “promotional fares from becoming permanently embedded into the fare structure and thereby burdening the general fare level and the normal-fare passenger.”40 Discount fares were only permitted with a life of 18 months or less and only if they met a “profit impact” test. There is no showing by Continental that its economy service meets the “profit impact” test. The Board distinguishes its Phase 9 decision as involving fare differences between permanent classes of service. Continental replies that although each discount plan is “temporary,” they keep cropping up so as to be a permanent feature of the landscape.
*345We think the Board’s .distinction of its discount fare policy is sufficient to satisfy the requirement of reasoned decisionmaking. It is true that in 1975 the Board allowed new discount fare plans,41 in light of the effect of the recession on the airline industry. But this is no indication that discount fares are a permanent feature of CAB regulation, and will be permitted in a time of recovery of profits. There is a distinction in time limitation; discount fares require reevaluation after 18 months, while Continental’s proposal is for an enduring economy fare. Moreover, the Board imposed other restrictions on discount fares (e. g. blackout periods, length of stay restrictions, and advance purchase requirements) to reduce diversion from coach, which is the objective of the “profit impact” test.
2. Differences in timing of effectiveness of cost based approaches
We turn to the issue that the Board ordered immediate application of the cost-based economy fare at $4 less than coach on meal flights, while it permitted a gradual approach to the “fare taper.” The basic validity of the fare taper is not contested in these cases. In general air fares embodied a “taper” of reduction (on a per mile basis) with increasing distance. This “taper” has reflected the undisputed economic condition that cost per mile decreases with increasing distance. However, prior to Phase 9 coach long hauls did not show fare reduction per mile comparable to the decrease in costs per mile.42 This resulted in coach fares yielding greater revenue in comparison to costs, for long distance flights than for short distances. The Board adjusted the coach fare taper formula in the direction of the cost formula, to be implemented four months after the order on reconsideration,43 and even after four months did not go all the way to a cost-based formula.
As to the Board’s requirement of cost-based first class fares, the Board phased that into effect over a period of 2V4 years.44 Continental protests these differences in timing approach as impermissible inconsistencies.
The Board recognized the possibility of adverse effect on the airlines’ profits in the short run by its new first class fare policy. The first class fare increases were phased in to allow accommodation, e. g., in aircraft configuration, to the new first class rates.45 The Board’s finding never explain in so many words why a similar phase-in was not allowed for economy fares. Its brief before this court states: “Economy service is but a minor part of the carriers’ total traffic (less than 1 percent) and the establishment of a cost-based coach economy fare differential could have no comparable [to a cost-based coach taper formula] impact on the transportation system and traffic patterns.”46 However, since, even at the time of the Phase 9 hearings, Continental carried about 75% of all economy passengers, and it constituted about 10% of its total traffic,47 economy class is significant to Continental. Elimination of the competitive attractive*346ness of economy service, which Continental aggressively promotes, would likely affect Continental’s ability to compete (and fill its currently purchased and scheduled aircraft) in the routes where it lacks the benefits of beyond segment traffic. Although economy traffic and revenue is small compared to first class, the long-run benefits and immediate burdens are proportionate to the traffic volume involved and the difference between present fares and cost-based fares.48
We are not comfortable with the way the Board handled the introduction of cost-based economy fares vis á vis first class ones. However, on March 20, 1974, the Board granted Continental’s motion for a stay of Order 74-12-109’s economy fare provisions.49 Pending a CAB order following our affirmance of Order 74-12-109, Continental has been able to continue economy service at 90% of coach. Thus, Continental has had a two year delay in implementation of the Board’s order on economy without any phase-in. The actual difference in timing impact does not appear to be significant either in absolute terms or impact on Continental. Further consideration of the difference of treatment of economy and first class is unnecessary.
3. Differences in extent of adoption of cost-based approach
The Board’s decision to move only roughly half-way toward a completely cost-based coach fare taper,50 was discussed on reconsideration:
To begin with, our determination to proceed with caution in implementing a cost-based structure of coach fares was not grounded upon, as Continental implies, any acceptance of the desirability of cross-subsidization of short-haul traffic by long-haul traffic, or even any acceptance of value-of-service pricing. Instead, our determination to take only a first step in the direction of a cost-based coach-fare structure, rather than prescribing a definite timetable for implementation of such a structure, was primarily based on the fact that since the airlines’ entire systems are geared to the traffic patterns resulting from the present coach-fare structure, and since the parties’ evidence did not permit us to draw definitive conclusions as to the effect on traffic of a major change in the coach-fare formula, it would be unwise to undertake drastic changes in the structure of coach fares.51
The Board’s core concern is that a fully cost-based fare taper might have a drastically disruptive effect on passenger traffic volumes in many sectors of the domestic air routes of a magnitude unpredictable on the evidence available. The Board acknowledged that its concern was speculative, voiced on a record without pertinent evidence and therefore dependent solely on “informed judgment.”52
*347The Board’s position that a full application of the cost-based theory to the distance taper problem might well have a drastic, disruptive effect on broad segments of the airline industry, provides justification for implementing the approach by first taking a substantial step in the preferred direction, with opportunity thereafter to gauge the effect by experience. The Board’s intermediate premise is that there was good reason to anticipate that the effect of a cost-based fare taper formula would be distinctly more unsettling than cost-based economy and first class fares. We think this is a situation where “the agency’s path may reasonably be discerned.”53 Adoption of a fully cost-based fare formula would have resulted in fare increases in some short-haul markets on the order of 50%.54 The Board feared a severe impact on short haul air traffic volume, which competes with surface transportation. The Board’s judgment of elasticity of the pertinent demand was, in its words, “necessarily speculative.” Although the Board attempted to assure that this would not result in severe losses for the industry,55 there would inevitably be variations among different carriers, depending on the average length of their routes.
Continental does not complain that the Board failed to determine policy exclusively on industry “average” projections. It would rather go further and argue that the Board should and must invariably consider the effects on various individual carriers. An administrative agency should be able to proceed pragmatically, testing theory in the light of experience. The Board was reasonable in acting cautiously so as to ascertain the possible redistributive effects of its cost-taper policy. In contrast, the economy fare had been found to involve a diversionary effect (attracting from coach) outweighing its new market effect. The converse is that the dropping of the economy fare will not lose passengers in marked degree, that the bulk of economy passengers will continue to fly, either at the higher (cost-based) economy fare, or at coach fare. Even assuming, as Northwest and American claim, the required fare increase for first class drastically reduces that demand, most of the affected passengers will still fly, and while their movement in coach reduces airlines’ revenue, there will be some offsetting reduction in costs.56 On the other hand, the cost-based fare taper might have drastic effects on revenue, traffic volume and schedules. Raising short-haül fares might divert significant numbers of passengers to surface transportation.57 Lowering long-haul fares might make these routes less profitable, leading to reductions in scheduling and service. We are not called upon at this juncture to decide whether to what extent or under what conditions an agency could combine a general cost-based rate approach with a different approach, say one that focuses on “value of service”, or the overall effect on revenues of differentials that “charge what the market will bear.” Thus far at least, except for the specialized instance of local service carriers where the Board was trying to reduce Government subsidy, its general cost-based approach is predicated upon the theory expounded by the DOT and respectable economists, that in the long run there will be a constant return to scale. Our acceptance of the Board’s use of a different rate profile, as it were, for the distance taper, as not *348reflecting an impermissible inconsistency, is based on the reasonableness of the Board’s effort to gain experience in light of the unknown and possibly disruptive effects of a cost-based fare taper. This position has a corollary: when there has been reasonable time to conduct and evaluate this experiment, the Board has a duty to reevaluate the matter.58 In reviewing its fare structure policy, the Board must take a consistently explained position concerning its cost-based approach to fares.
4. Local service carriers
We turn to the Board’s decision in Phase 9 to allow local service carriers to charge up to 130% of the prescribed coach fare formula.59 Local service carriers, operating short-haul routes, are not able to operate profitably at the standard formula fare on the often low volume routes they serve. They typically receive subsidies from the government to provide the public service that they do. It is within the CAB’s authority to authorize local service carrier fares to exceed the standard formula prescribed in Phase 9, in order to reduce public subsidy.60
5. The missing meal
We turn now to what Continental has dubbed “the mystery of the missing meal.”61 The issue, it will be recalled, is that the Board permits the $4 economy/coach difference only on flights with meals, reasoning that there is no cost difference between coach and economy services on non-meal flights.62 This is inconsistent, Continental avers, with the permission of the first class/coach differential regardless of whether a meal is provided, since part of the greater cost of first class is the fancier meal. Implicitly, the Board is saying that- the difference in meals is the sole difference between economy and coach, and is only a minor factor in the first class/coach cost différence, which is primarily due to the different space occupied.63 We might not find this convincing if it were solely an appeal to administrative convenience. On the other hand, there is a certain element of substance in the premise that the first class/coach “class-style” difference is a total package, that the meal difference is only a small part of the cost difference, and that in consumer perception it is a relatively minor part of the package. Continental’s counsel has skillfully argued the problem, but we find the Board’s position barely adequate on this score.
III. APPLICATION OF COST-BASED THEORY TO FIRST CLASS FARES
Having upheld the Board’s adoption of the long term burden/cost theory and its application to the economy class, we turn to Northwest’s challenge to its application to first class service. Essentially the Board’s first class fare orders are congruent with the economy fare orders, in approach and application, and we have no basis for a difference in our decision.
The CAB’s determination of cost differences between first class and coach service reflects higher levels of cabin service and food amenities, but the bulk of the difference results from “space allocation” dividing the total costs of the flight among the classes in proportion to the cabin space occupied by each. While the parties challenging the Board’s order concerning first class fares, Northwest and American, object to this method of cost allocation, they do not seriously challenge the ratios (150-163%, depending on the length of the flight) found by the Board using its method of cost allo*349cation.64 Various carriers argued in Phase 9 for different methods of determining the cost of first class service based on their premise that capacity is not scarce now, nor is it likely to be in the future based on the Board’s 55% load factor standard for rate-making,65 scheduling competition, and the Board’s alleged practice of certifying new competition if load factors go significantly above 55%.66 These carriers argued for incremental cost methods which attribute less flight costs to first class than the Board’s “space-allocation.”67
In its Phase 9 decision, the Board considered the rejected proposals for “opportunity” or “displacement” costing.68 While the Board has an obligation to consider proposals by the parties, the Rule of Administrative Law requiring a statement of the reasons for adoption of standards and approaches does not require comparable detail for the reasons for discarding alternative ' approaches preferred by the parties.69 We are satisfied with the Board’s treatment here, beginning with its initial notation that “[w]hatever the relevance of demand considerations may be to the pricing of first class service, they ought not preclude a realistic appraisal of the costs of the service.”70 (Emphasis in original).
The nub of the carriers’ argument is that in an assumed state of a continuing surplus of capacity, it will maximize the carriers’ revenue to offer first class service at below-space allocation-cost to lure more passengers into buying more of the “excess” airplane space. Maximizing revenues, it is argued, benefits the coach passenger by holding down the need for fare increases.
It appears reasonable to us to allocate flight costs on the basis of space occupied, given that cabin space is interchangeable, with relative ease, between first class and coach accommodations.71 Whether this is *350viewed as a problem of cost accounting, economic theory or regulatory judgment, we cannot say that the Board’s conclusion lacks rationality.
Turning to the effect of cost-based fare differentials on revenues, the Board recognized that the required increases in first class fares (to a ratio to coach fares commensurate with relative cost) would operate, by some diversion of first class passengers to coach, to decrease carriers’ revenues, and perhaps profits as well — in the short run. It granted that the carriers’ analysis of the public interest in below-cost first class fares might be correct — in the short run.72 However, the Board determined to establish its cost-based fare policy for the long run. It took account of the tendency of the carriers to indulge in and perpetuate cycles of overcapacity. The Board used long-run standards and approaches in order to create a favorable environment for sound planning about future capacity, while phasing in the implementation of this policy over a period of 2lA years to ameliorate the immediate impact, given the existing excess capacity, and to allow gradual adjustment.73 It viewed the long run consequences of below cost pricing of first class service, by application of its now-familiar long run cost theory, as burdening the coach passenger. The coach passenger will in the long run suffer either poorer service or higher rates. As we have already stated, this regulatory approach is sufficiently grounded in Phase 5 data and in economic theory to permit adoption by the Board as its policy for permanent fare .structure.
In further support of its first class fare determination, the Board adduces two factors. First, if, as the carriers contend, and the Board concedes is likely, the substantial increases ordered in first class fares will result in significant short-run diversion from first class to coach with a corresponding reduction of revenues, the rational reaction to the carriers will be to reduce the amount of first class capacity offered. Since coach seating is denser, this transformation will increase capacity without acquisition of any new aircraft, thereby reducing future capacity increase needs and costs.74 Secondly, the Board bolsters its contention that the carriers have tended to offer too much first class under the existing first class/coach fare ratios.
These long-run burdens are particularly exacerbated if the carriers offer a volume of first-class capacity which is out of proportion to first-class traffic, as has been the case in recent years. Since 1963, first-class load factors, for domestic operations of the domestic trunks, have been below fifty percent in every year but one, and have not exceeded 45 percent since 1967. In 1972, as we noted earlier, while first-class traffic accounted for only 12.2 percent of the trunks’ domestic traffic, the trunks devoted 18 per*351cent of their capacity to first-class service. In that year their first-class load factor was a mere 35.5 percent.75
As we pointed out in our prior opinion, first-class load factors have been considerably below average for the past several years. In the year ended May 1974, the first class load factor for the domestic operations of the domestic trunks was only 38.8%, as compared with a coach load factor of 58.2%. The only reasonable conclusion that can be drawn is that in peak periods, persons who can only afford the coach fare are turned away, while first-class passengers travel in splendid, low load-factor isolation.76
Language like this77 leads Northwest to charge that the Board’s actual goal is the elimination, or drastic reduction, of first class service. Northwest contends that this violates § 401(e)(4) of the Federal Aviation Act,78 which prohibits CAB regulation of the accommodations and facilities provided by carriers.
We considered § 401(e)(4) in Continental Air Lines, Inc. v. C.A.B., 173 U.S.App.D.C. 1, 522 F.2d 107 (1974). There, we struck down the Board’s policy and implementing orders to eliminate the “unnecessary frill” of coach lounges and five abreast seating on narrow bodied jets. This was to be done by encouraging carriers without these frills to apply for a fare reduction, while prohibiting competitors with lower density coach from matching the reduction. Since the Board’s policy in the very Phase 6A (Seating Configurations) under consideration was to establish fares based on costs in the dense configuration (six abreast and no lounge), this rate reduction, not based on costs, was inconsistent with the Board’s cost based fare policy throughout the DPFI. In the present case, however, the Board has based its fare policy on a determination of cost difference.
Our Continental opinion explicitly distinguishes the type of action taken by the Board here;
Consequently, the TWA Siesta Sleeper-Seat case — based on the principle of preservation of the distinction between classes of service on the basis of actual cost — cannot help the Board here. What the Board is doing here is neither preserving the distinction between classes of service, nor attempting to gear revenues to costs. It is simply attempting through artificial ratemaking to eradicate services which it regards as “needless frills.”79
Northwest contends that the Board’s decision on first class fares is inconsistent with its decisions in Phase 5 (Discount Fares) and 6A (Seating Configurations) of the DPFI. In Phase 6A, the Board allowed lower density coach service (five abreast or coach lounges) at the regular coach fare. It dealt with the competitive pressure to match lower density service with the com*352petitive fare differential policy we rejected in our 1974 Continental opinion.80 The Board did not address this apparent inconsistency in its Phase 9 opinions. However, it was not brought to the Board’s attention in any of the motions for reconsideration of the Phase 9 decision.
Before this court, counsel for the Board has offered a two-fold distinction between the Phases 6A and 9 decisions. The first is that Phase 6A only dealt with seating arrangements within the coach class, while Phase 9 dealt with the proper fare structure between classes. This is a formal distinction, but we question its materiality, when the issue is the Board’s consistency in requiring that fares be cost-based, and in adhering to the Siesta Sleeper principle, of maintaining the distinction between types of service based on cost differences. Secondly, the Board says:
Moreover, the Board’s determination not to prescribe a mandatory fare differential in Phase 6A was partially predicated on the finding that carriers offering high-density coach seating were able to compete effectively with carriers operating five-abreast coach seating without need for a fare differential. In contrast, a carrier would not be able to price first-class service at fully allocated cost levels if a competitor offered the service at fares substantially below costs.81
We are frankly troubled by these explanations, yet they relate to issues not raised in petitions for reconsideration before the Board. For all we know, they may be debating points rather than material contentions, and the parties now raising them would swiftly desist if they anticipated that the Board would be led to draw cost-based differentials between 5-abreast and 6-abreast coach service. We are disinclined to probe further into the problem at this juncture, especially since the Board’s Phase 6A position was struck down in part in the 1974 Continental opinion, and the Board must reconsider its seat configuration policy in light of our opinion and its Phase 9 findings and policy.
We come again to the claim of inconsistency between the Phase 9 determination, of the proper relation between different permanent fare classes, and the Phase 5 decision to permit below-cost promotional discount fares under certain circumstances. We have dealt with this in rejecting Continental’s similar" contention with respect to the Phase 9 economy fare determination. In Phase 5, the Board announced its intention to protect the coach passenger from having to carry a higher share of costs due to discounts by setting fare levels as if there were no discount plans.82 Petitioners Northwest and Continental argue that since the basis of the policy of cost based fares is to prevent undue burden on the coach passenger, the Board should have adopted the same approach in Phase 9 as it did in Phase 5. They argue that coach fares should be set as if coach were the sole service. The carriers could then be permitted to offer first class and economy at less than cost, as a business decision and at their own risk, since the coach fare would be insulated by the proposed policy. Continental further argues that this strategy would be in keeping with the Board’s Phase 6B (Load Fac*353tors) decision to encourage carriers to increase average load factors to the 55% the Board found to be the best price/service tradeoff by setting fare levels as if load factors were 55%.83
The carriers take a selective view of Phase 5. The policy of setting the fare level as if there were no discounts is only one of three conditions the Board imposed on discount fares in Phase 5. The others were that the discount fare expire in eighteen months or less and that it meet a short term profit impact test. The Board’s Phase 9 policy set permanent fare structure with no conditions, around which the carriers would plan scheduling and capacity additions in the long run, and required cost-based fares to prevent a long term burden on the coach passenger — in the form of (a) higher fares, or (b) poorer service in higher load factors, or (c) inability of the carriers to provide adequate service because of inadequate long term profitability. These latter two possibilities are obviated by the Phase 5 conditions. Discount fares are allowed only temporarily in periods of excess capacity when load factors are low, and the profit impact test is intended to assure a favorable effect on carriers’ finances. We are satisfied that the basic distinction drawn by the Board in its Phase 9 finding84 is consistent with reasoned decisionmaking. We decline to consider subsidiary and perhaps immaterial distinctions.85
IV. OVERVIEW
We have upheld the Board’s rulings on the core matters involved in this case upon the basis of a regulatory approach and underlying economic theory that cannot be established at this time as true, but can be accepted as neither arbitrary, capricious, nor unreasonable.
In certain respects, the Board’s action depends on cost allocation methodology. Fifty years ago Justice Brandéis pointed out that in such matters
experience teaches us that it is much easier to reject formulas presented as being misleading than to find one apparently adequate.86
As to the various claims that the Board’s opinion reflects internal inconsistencies, most have been dispelled. Some remain as possibilities that are not legally material on the record and objections made in that record. For at least one such claim, the Board’s approach is sustained as provisional, awaiting further experience with the more nearly cost-based coach fare formula. In general, the court sustains the Board’s objective of looking at coach class as the main class of service, and making an effort to assure that it carries a rate based on sound load factor objectives (which in turn imply a corresponding level of service), and that the coach passenger is not burdened with undue “bargains” given to others. Any such modifications or reconsideration in the light of experience may take into account an added burden when “bargains” are available only to a minority, either because they are, for example, restricted to those with personal or business means (first class service), or to the young (youth fares). In such cases there also arise issues of discrimination. The normal coach passenger cannot bring himself within the favored class. However, in the case of a “bargain” fare (below fully allocated costs) for economy class, the normal coach passenger can bring himself within the favored class. The *354task of establishing fare structure is pervaded by zones of discretion, due to uncertainties in projecting short and long term consequences, and due to the need for serving competing policy interests. There is more latitude to conclude that a differential has not been shown to be an “undue discrimination” when it is one that the consumer himself can realistically avoid.
We have sustained the Board’s conclusion that it can maintain the basic objectives announced in Phase 9 with the addition of Phase 5 discount fares as safety valve flexibility. It may be that Phase 5 will be the seed bed for discriminations more noxious than those struck down in Phase 9. So long as the nation retains some structure of government regulation, the carriers and the agency must be accorded elbow room to interact. The regulatory system, combining the spark of private profit, the motive power of private enterprise, and a loose rein of regulatory restraint to frame the appropriate sphere of competition, permits experimentation, followed by evaluation in the light of experience. The basic requirement is reasoned decisionmaking. While the case is not entirely free from difficulty, we do not find that the record in this case warrants a ruling that it has been bypassed.
V. POSTLUDE
Subsequent to the writing of the preceding opinion, petitioner Continental brought to our attention the CAB’s recent decision in the Hawaii Fares Investigation, Opinion and Order 76-10-37 (hereinafter Hawaii), served on October 20, 1976. In determining lawful fares for the U.S. Mainland-Hawaii market, the Board considered, inter alia, issues of discount fares, and the issues of fare structure here under review presented by relationships between coach fares and first class fares, and between coach and economy class fares. We called on the parties for memoranda concerning the implication for the case at bar of the Board’s present thinking as reflected in the Hawaii decision.
In Hawaii, the Board adopted the position of the Administrative Law Judge that the coach fare level should reflect fully allocated costs at a 62% load factor in the standard seating configuration adopted for rate purposes in Phase 6A of the DPFI. It also agreed with the ALJ that first class fares must be no less than 188% of coach fares in order to cover allocated costs and required that this ratio be achieved in a three stage schedule over a period of 2Vi years. The Board rejected the ALJ’s strict application of the Phase 5 discount fares policy. It decided to allow non-discriminatory, nonpredatory discount fares without fixed expiration date and without application of a profit impact test. In keeping with this liberality to below-cost discount fares, the Board allowed economy fares to continue somewhat below the -amount dictated by cost savings. (An $8 coach-economy difference was permitted, while the cost difference was found to be $6.15.)
In its treatment of the first class/coach fare structure, Hawaii takes the same approach as in the Phase 9 order under review. We find nothing to disturb our previous determination of the validity of the Board’s order concerning the first class/coach fare ratio in Phase 9.
As to economy fares, however, Hawaii permits a greater differential below coach fares than is warranted by cost-savings — an approach different from the Phase 9 order- — as an accompaniment of the main part of the Hawaii decision, to allow the carriers flexibility in offering below-cost discount fares while protecting the normal coach passenger from increased fares or poorer service.
There is some strength in Continental’s contention that a similar flexibility is all that it seeks in its offer of economy service at 85-90% of the domestic coach fare. It maintains that the coach passenger would be protected from higher fares by fare level standards calculated as if all passengers in the coach compartment paid the normal coach fare, from lower levels of service by competition, and that Continental itself would take the financial risk of miscalculation, resulting in inadequate finances. This seems to be the approach taken in Hawaii *355to protecting the coach passenger from .a long term burden due to below cost fares. In denying the same flexibility in the economy fare, there is not even the policy consideration of avoiding all possible risk of subsidizing the wealthier passenger or more luxurious service, such as might be involved in lowering first class fares.
Since the allowance of economy fares somewhat below cost in Hawaii was derivative from its primary discount policy, Hawaii does not present a direct conflict with the Phase 9 economy fare order. We are concerned, however, with what appears to be a divergence in principle from the Phase 9 economy fare order, or at least from its theoretical underpinning.
The memoranda filed by counsel for the Board and for TWA distinguish the treatment of lower-than-coach fares in Hawaii, from what was done in the DPFI, on the basis of the unusual characteristics of the Hawaiian market. Hawaii’s economy is heavily dependent on tourists traveling by air. “More than 80% of all scheduled service passengers are discretionary travelers, i.e., vacationers or persons otherwise on personal business.” Order 76-10-37 at 11. The Hawaiian market is heavily competitive among four or more scheduled carriers in the major Hawaii-Mainland routes and between scheduled and charter service.
There are indications that the Hawaii policy, of allowing the carriers flexibility in lower-than-coach-fares, was adopted by the CAB’s although apparently in the teeth of the premises of the DPFI’s economic theory, because of two interrelated factors. In light of the largely tourist, discretionary nature of the bulk of Hawaiian air travel, the Board has concluded that adequate air service in that market depends less on consistent flight availability and more on low fares than in the domestic market. Thus a high 62% load factor was adopted for fare level purposes.87 The Board has opted for lower fares in this largely discretionary market, where the hardship of having a trip rescheduled is less than for business or commuter air travelers. This orientation may bear on the Board’s decision in Hawaii to allow below-cost fares under the coach fare, with the concomitant risk of impairment of service to the coach passenger, but with the coach passenger protected from higher fares by the fare level standards. The other factor, somewhat interrelated, is the tilt toward cheap air fares because of Hawaii’s dependence on tourism and need of a low-cost lifeline to the U.S. Mainland.
Concurring Board members Minetti and West put it that Hawaii represents a “limited departure” from the DPFI’s discount policy. Order 76-10-37, Concurrence of Members Minetti and West at 9. There is certainly logical force in their statement that the discretionary and vacation-oriented nature of the traffic and the dependence of Hawaii on tourism are relevant to ascertaining the proper load factor standard (and level of service implicit), but do not affect the essential Phase 5 analysis. Id. at 5. Messrs. Minetti and West go on to say that nevertheless the Hawaii order is in the public interest.88
*356The outcome seems to be, however, that what is set forth in Phase 5 and repeated in Phase 9 as a basic economic analysis, is treated in Hawaii— more explicitly by the concurring members and implicitly by the majority — as something that is not inevitable, but is subject to reexamination on the issue of probability of occurrence in the particular situation. If the economic analysis that the Board has treated as virtually a theorem in the DPFI is really only a statement of concern or modest probability, the question is what, if any, is the principled basis for determining when it is to be relied on and when it is to be departed from.
We have tried “to discern the agency’s path,’’ and to see the Board’s Hawaii decision as a reasoned application of the DPFI basic approach. We find that we are left, however, with our own speculations and the arguments of counsel. In order to decide the case arising from the Phase 9 economy fare decision, No. 75-1090, we feel we must have the articulated reasoning of the CAB on these matters. The interest of justice will best be served by remanding the record in No. 75-1090 to the Board. 28 U.S.C. § 2106.89
It might be possible, if the timing is right, for the Board to consider the interrelation of the Hawaii decision with its Phase 9 economy policy along with the petition(s) for reconsideration in the Hawaii case. We will then have the benefit, in deciding the pending case, of integration of the Board’s reasoning in these inter-related matters, as well as an assurance that Phase 9 reflects its current views on economy fares, in light of its policy toward granting temporary discount fares since the DPFI. We anticipate that the stay of the Phase 9 economy fare order granted on March 20,1975, Order 75-3-63, will be kept in effect pending the Board’s reconsideration.
The Phase 9 order concerning first class fares, challenged in No. 75-1195, is affirmed. Our ruling on the Phase 9 economy fare order is held in abeyance, pending remand of the record in No. 75-1090 and further proceedings by the CAB.
The Board is free to enter a modification or supplement of its Phase 9 order and decision on this remand of the record. The Board is requested to return the supplemental record on remand, together with any modification or supplement to its order and opinion, as soon as practicable.
So ordered.
. The DPFI was initiated by the Board in January, 1970, CAB Order 70-1-147 (January 29, 1970). The Board divided the DPFI into phases to be carried out concurrently, Order 70-2-121 (February 26, 1970). The phases and the Board decisions therein are:
1. Aircraft Depreciation: PS-45 (April 9, 1971)
2. Leased Aircraft: PS-44 (April 9, 1971)
3. Deferred Federal Income Taxes: PS^46 (April 9, 1971)
4. Joint Fares: 73^4r-42 (April 10, 1971); 72-6-29 (June 6, 1972); 74-3-80 (March 18, 1974)
5. Discount Fares: 72-12-18 (December 5, 1972); 73-5-2 (May 1, 1973); 73-8-55 (August 10, 1973)
6A. Seating Configurations; 72-5-101 (May 26, 1972); 73-6-102 (June 26, 1973)
6B. Load Factors: 71^4-54 (April 9, 1971); 74-3-91 (March 18, 1974)
7. Fare Level: 71^4-59/60 (April 9, 1971); 72-8-50 (August 10, 1972); 72-9-78 (September 21, 1972)
8. Rate of Return: 71 — 4-58 (April 9, 1971); 71-7-43 (July 8, 1971)
9. Fare Structure: 74-3-82 (March 18, 1974); 74-12-109 (December 27, 1974)
We affirmed the Board’s decision in Phase 4 in American Airlines, Inc. v. C.A.B., 161 U.S.App.D.C. 430, 495 F.2d 1010 (1974). We set aside the Board’s orders in Phase 6A in Continental Air Lines, Inc. v. C.A.B., 173 U.S.App.D.C. 1, 522 F.2d 107 (1974).
. CAB Order 74-3-82 at 9-10, J.A. 398-99.
. Id. at 10-11, J.A. 399-400. The ALJ’s determination that economy fares at 90% of coach reflect the difference of cost was partly based on the CAB’s tentative decision in Phase 6A (Seating Configuration) that economy service would have a 34" maximum seat spacing while coach seats would be spaced at 36". Id. at 129, J.A. 518. This decision was not put into effect. CAB Order 72-5-101, J.A. 233. The question was deferred to the present Phase 9, in which the CAB posited economy service equivalent to coach except for the meal. The CAB considered lack of opportunity for advance seat selection for economy passengers a de minimus factor. Order 74 — 3—82 at 132, J.A. 521. Petitioner Continental admits that economy service is the same as coach except for the meal (Testimony of John P. Maher, Phase 9 Hearings at 158-60, J.A. 1165-67) and does not seriously challenge the $4 fare difference based on the cost of the meal. Rather Continental challenges strict adherence to cost in establishing reasonable fares. See Section II C infra.
. Order 74-3-82 at 123-28, J.A. 512-17.
. Initial Decision of the ALJ (April 7, 1972) at 17-18, J.A. 757-58.
. Order 74-3-82 at 125, J.A. 514.
. Continental Motion to Reopen the Record, October 31, 1974, J.A. 975.
. CAB Opinion and Order on Reconsideration 74-12-109 at 41, J.A. 625.
. Id. at 37-40, J.A. 621-24.
. Order 75-5-100 (May 26, 1972).
. See Order 70-1-147 (January 29, 1970), instituting the DPFI.
. 32 C.A.B. 291 (1960), aff'd sub nom. Eastern Air Lines v. CAB, 111 U.S.App.D.C. 39, 294 F.2d 235, cert. denied, 368 U.S. 927, 82 S.Ct. 363, 7 L.Ed.2d 191 (1961).
. See DOT Phase 5 (Discount Fares) Briefs to the Hearing Examiner (Sept. 8, 1970) and to the Board (May 4, 1971). Appendix to the CAB’s April 16, 1976 Memorandum in this case, at 119, 146. The DOT analysis is considered in notes 20 & 22, infra.
While DOT supported setting economy fares at 85% of coach in its Phase 9 brief to the Hearing Examiner (Nov. 12, 1971) at 4, this was justified partly on lower costs and partly to match “unjustly discriminatory” discount fares. Other carriers, including American, Braniff, TWA, and United advocated economy fares equal to, or $3 less than coach fares. Initial Decision of the ALJ, App. B at 1, J.A. 821.
. Continental’s Brief to the Hearing Examiner in Phase 9 (Nov. 12, 1971) at 41, J.A. 849.
. As the CAB said in its opinion on reconsideration:
Continental argues that it failed to produce cost evidence earlier in the proceeding in reliance on the requirement of Section 1002(e)(1) that the Board consider the effect of rates on the movement of traffic and on an alleged Board practice of establishing price differentials based on marketing considerations. Continental argues that the Board reversed its prior practice and ignored Section 1002(e)(1) by limiting its consideration of economy fares to costs, and in light of these changes, Continental should have an opportunity to present evidence on costs. We cannot accept this argument. The very purpose of this proceeding was to decide the basis on which fares for the various classes of service should be set. The fact that Continental laid its eggs in the wrong basket should not be allowed to serve as a reason for delaying the implementation of fares we have found to be lawful.
Order 74-12-109 at 39, n.91, J.A. 623.
. Continental’s Motion to Reopen the Record (Oct. 31, 1974) at 12, J.A. 985.
. Id. at 1, J.A. 974.
. Id. at 3-7, J.A. 376-80.
. Order 74-12-109 at 37, J.A. 621.
. The effect of scale on the economics of airline operations was considered by the Department of Transportation. That party submitted evidence in support of the proposition that the airline industry is one of constant returns to scale, i.e., that an increase in the size of operations does not produce a reduction in unit costs. No party has seriously challenged the DOT evidence, and without necessarily adopting its methodology and conclusions in all of its details, the Board finds that the general conclusions reached are sound.
Order 72-12-18 (Phase 5) at 48, J.A. 300.
The data and analysis supporting this proposition, as presented by DOT to the Board in Phases 5 & 7, are also available in a book by the DOT’S economic consultants on the DPFI, G. Douglas & J. Miller, Economic Regulation of Domestic Air Transport (Brookings Institution 1974) 13-18. Unit cost data of the eleven trunk carriers for 1964 — 70 show a slight tendency to increase with increasing size of the carrier, measured by available ton-miles flown. The same finding of unit costs substantially independent of size over the range of trunk carriers was found in the FY 1974 data. Report of the C.A.B. Special Staff on Regulatory Reform (July 1975) 102-07. Douglas and Miller also cite the fact that the profitability of the small trunk carriers generally has been substantially better than that of the large ones as support for the proposition of no significant economies of scale. Op. cit. at 17-18. (The issue of cost economy from increasing the number of flights, and consequently equipment and personnel, is of course separate and apart from the fact that local service carriers have higher unit costs by reason of their smaller aircraft and shorter average flight distances.)
Constant returns to scale for the domestic trunk carriers has been reported in other studies. R. Caves, Air Transport and Its Regulators (Harvard Univ. Press 1962), ch. 3; Eads, Nerlove, and Raduchel, A Long-Run Cost Function for the Local Service Airline Industry: An Experiment in Non-Linear Estimation, 51 Review of Economics and Statistics 258-70 (1969).
. Order 72-12-18 at 15-17, 54-55, J.A. 267-69, 306-07. The Board indicated reliance on its Phase 5 learning in explaining its Phase 9 determinations in regard to both the first class and economy fare issues, Order 74-3-82 at 125, 132, J.A. 514, 521. The parties were on notice that this might be done. In its order dividing the DPFI into concurrent phases, the Board said:
all evidence presented in any phase of the entire investigation will also be available in the docket for consideration and decision in all other phases.
Order 70-2-121 at 2, J.A. 176. Beyond this general admonition the Board stated its view that discount fares were “inextricably intertwined” with fare level and fare structure questions in its order instituting the DPFI, Order 70-1-147 at 44, J.A. 171.
.Since fare changes by individual carriers are subject to CAB approval, as well as objection by other carriers, there is little flexibility or competition in price among the scheduled carriers. There is intense competition for passengers in the area of service and frills. The main weapon left to the competitors, however, is convenience of scheduling. The Department of Transportation presented a model of the airline industry, in which the CAB sets rates, which establishes a breakeven load factor for the carriers. They then compete in scheduling (and hence capacity) within the constraint of staying at or above that load factor. The CAB’s acceptance of this model is reflected in its Phase 5 opinion, where it said:
As we held in our decision in Phase 6B, over the long term, industry-wide load factors will tend to equal the industry’s breakeven load factor (using “breakeven” in the sense of including a reasonable return on investment).
Order 72-12-18 at 45, J.A. 297.
See Order 74-3-82 at 109-111, J.A. 498-500; Department of Transportation Brief to the CAB in Phase 5 (May 4, 1971) at 146-64; see also Douglas and Miller, supra, note 20 at 40-44.
. See notes 20 and 22, supra.
. Order 72-12-18 at 40-50, J.A. 292-302.
. American, Allegheny, Braniff, Northwest, North Central, Southern, TWA, United, and Ozark all testified to this effect in the Phase 5 hearings. See Phase 5 Hearings at 170, 376, 576-77, 581, 611, 757-58, 1506, 1686-88, 1723, 1834, 2111-12, 2183, 2207, 2293-94. Western, Northeast, and National made similar statements in written responses to requests for information. Appendix to CAB Memorandum of April 16, 1976 at 56-58.
. Continental points to the disparity in impact between discount fares at the time of the Phase 5 hearings, roughly 40 to 50% of total traffic (Continental Reply Memorandum, p. 3), and *342economy class, which was less than 1% of total traffic. This point has some merit, but the Board’s position is that economy, though less extensive than discount, will be proportionate in its long run effects. See note 48, infra. We are in no position to reject that position as impermissible.
. “Continental is the only airline which offers the economy fare in all its markets.” Continental Reply Br. at 3. See note 34, infra.
. The term “general insistence” takes into account that the approach was not followed in all applications in the Board’s order. This problem is dealt with in Section II D, infra.
. Section 1002(e), 49 U.S.C. § 1482(e) states:
In exercising and performing its powers and duties with respect to the determination of rates for the carriage of persons or proporty, the Board shall take into consideration, among other factors—
(1) The effect of such rates upon the movement of traffic;
(2) The need in the public interest of adequate and efficient transportation of persons and property by air carriers at the lowest cost consistent with the furnishing of such service;
(3) Such standards respecting the character and quality of service to be rendered by air carriers as may be prescribed by or pursuant to law;
(4) The inherent advantages of transportation by aircraft; and
(5) The need of each air carrier for revenue sufficient to enable such air carrier, under honest, economical, and efficient management, to provide adequate and efficient air carrier service.
. American Airiines Inc. v. CAB, 161 U.S.App.D.C. 430, 438, 495 F.2d 1010, 1018 (1974).
. Continental Br. at 35.
. Section 1002(e)(2), 49 U.S.C. § 1482(e)(2).
. Section 1002(e)(1), 49 U.S.C. § 1482(e)(1).
. “Today, the economy fare represents the only vestige of price competition in that part of the air transportation industry which is regulated by the Board.” Continental Brief at 5. Continental uses and promotes economy service to compete with the domestic “Big Three” (American, TWA, and United) and make up for its competitive disadvantage of lacking route authority east of Chicago. Continental justifiably feels itself the custodian of the economy fare and the airline most affected by the fare increases mandated in Phase 9.
Continental is the only airline which offers the economy fare in all its markets. Continental is the only airline which promotes the use of economy fare. No other carrier offers the economy fare in any market not served by Continental. No carrier other than Continental carries a significant number of economy passengers. Of the approximately 1,100,-000 interstate economy passengers in 1973 in mainland markets covered by these orders, Continental carried 833,287, or 76 percent of the total. United and Western in 1973 carried only 143,810 and 33,820 economy passengers, respectively.
Continental Reply Br. at 3.
.The House Commerce Committee Report stated as one of the reasons for imposing economic regulation on the airline industry:
Nor, is there any authority in the Federal Government under existing law to prevent competing carriers from engaging in rate wars which would be disastrous to all concerned.
H.R.Rep.2254, 75th Cong. 3rd Sess. 2 (1938). Congressman Lea, the House sponsor of H.R. 9738, stated in debate:
The authority will also exercise rate control, requiring that rates be reasonable and giving power to protect against cutthroat competition. 83 Cong.Rec. 6407 (May 7, 1938).
. See note 34, supra.
. Section 1002(e)(1), 49 U.S.C. § 1482(e)(1).
. We are sensitive to the force of Continental’s appeal on the economy fare, but conclude the Board acted within its discretion in setting the $4 maximum fare difference from coach. See further discussion in Part IV, infra.
. Greater Boston Television Corp. v. F.C.C., 143 U.S.App.D.C. 383, 394, 444 F.2d 841, 852 (1970) cert. denied, 403 U.S. 923, 91 S.Ct. 2229, 29 L.Ed.2d 701 (1971); Columbia Broadcasting System v. F.C.C., 147 U.S.App.D.C. 175, 188, 454 F.2d 1018, 1027 (1971).
. Order 72-12-18 at 53-54, J.A. 305-06.
. See order 75-2-124, J.A. 643, approving extension of Bicentennial excursion fares in long haul flights and Order 75-4-47, J.A. 697, approving “no frill” fares for additional carriers.
. The existing fare formula did not reflect the economies of long haul flights resulting from the introduction of jets, and it was felt that “value of service” considerations of the great saving of time in long flights and the need to keep short haul fares down to compete with surface transportation justified keeping fares less tapered than costs would dictate. See Order 74-3-82 at 59-62, J.A. 448-51.
. Order 74-12-109 at 40, J.A. 624.
. Id. at 41, J.A. 625.
. We recognize that in the short run, drastic changes in first-class fares could have adverse consequences for the carriers. Indeed, this was the basis upon which we decided to phase in the cost-based fares over a reasonable period of time.
Id., at 20-21, J.A. 604-05.
. CAB Br. at 60.
. Order 74-3-82, Appendix 5, J.A. 579. In its Motion to Reopen the Record, supra note 44 at 6, J.A. 979, Continental stated that economy traffic spurted to about one third of its coach/economy traffic with the elimination of discount fares in mid-1974.
. Our admonition to the Board to adopt similar policies in cases where the benefits and burdens are proportionate is in keeping with our acceptance of the Board’s reliance on the Phase 5 (Discount Fares) findings to support the present application of its cost based fares theory to the much smaller economy class. See note 26 supra. Though the effects may be of different magnitudes, if they be proportionate, the policies respecting the respective fares should be similar.
. Order 75-3-63 (March 20, 1975), J.A. 687.
. See graph in Order 74-12-109 at 77, J.A. 466. The magnitude of the discrepancy before Phase 9 at very short and very long distances are stated by the Board:
Under the present structure, passengers traveling 100 miles pay less than 70 percent of the costs of their carriage. At 300 miles ., fares are 13 percent below costs. Fares are below costs up to approximately 595 miles, and long-haul fares are substantially above costs, e. g. $13.04, or 9.1%, at 2,500 miles.
Id. at 72, J.A. 461.
Thus, if the Board had mandated a cost based fare formula, short haul fares would have had to be increased by more than zh or almost 50%. In Phase 9, the Board eliminated about 'h of the fare-cost discrepancy in short hauls and about 2/s of the (smaller as a percentage of the total fare) discrepancy in long hauls.
. Order 74-10-109 at 24, J.A. 608.
. Our discussion of the effects of our fare formula on short-haul traffic was necessarily speculative, because no party had introduced any persuasive analysis of the elasticity of demand for short-haul air transportation. Under *347the circumstances, our own informed judgment was all that was available to us.
Id. at 10, J.A. 594.
. Greater Boston Television Corp. v. FCC, 143 U.S.App.D.C. 383, 393, 444 F.2d 841, 851 (1970), cert. denied, 403 U.S. 923, 91 S.Ct. 2229, 29 L.Ed.2d 701 (1971).
. CAB Br. at 60.
. [Tjhere is no reason to believe that our decision will in fact result in revenue losses for the industry; indeed, we have carefully adjusted the fare formula to assure that no fare level changes will result.
Order 74-12-109 at 36, J.A. 620.
. See id. at 19, J.A. 603.
. Allowing this to happen by eliminating the cross-subsidization of short haul passengers by long haul ones might be a good thing as a matter of consumer preference and long term optimal resource allocation. However, in the short run, it leaves short haul carriers with excess capacity.
. United States v. CAB, 167 U.S.App.D.C. 313, 324, 511 F.2d 1315, 1326 (1975) (capacity reduction agreements).
. Order 74-3-82 at 1, J.A. 384. See also id. at 140-42, J.A. 529-31.
. See Western Air Lines v. CAB, 161 U.S.App.D.C. 319, 328-29, 495 F.2d 145, 152-53 (1974).
. Continental Reply Br. at 13.
. Order 74-3-81 at 1, J.A. 385. See also id. at 129-33, J.A. 518-22.
. CAB Br. at 62.
. The best the carriers can say is that part of the first class section is “common area” used by all passengers — entry ways, aisles, and toilets. (Northwest Br. at 25). We find that the Board was justified in dismissing this usage as de minimis. Order 74-3-82 at 30, J.A. 419. Conceivably, true space allocation, considering the 3:2 ratio of coach/first class seats abreast and the wider pitch (separation) of first class seats, will yield a ratio greater than the 150-163% mandated by the Board, and the carriers chose to leave well enough alone in not challenging the figures arrived at assuming the space allocation method.
. Order 71-4-54 (Phase 6B — Load Factors) at 45, J.A. 199.
. Northwest Br. at 34.
. “Opportunity costing” assigns no extra capacity costs to the first-class compartment, since capacity is not a scarce resource in modem jet aircraft. Under opportunity costing, the costs of first-class service are the added costs of first-class amenities and the revenue loss from the displacement of coach passengers, if any. The “displacement method” allocates capacity costs by use of a ratio which is based on the percentage increase in seats in the first-class compartment which would be obtained if the aircraft were operated in an all-coach configuration.
Northwest Br. at 15, n. 14.
Even, now, there is crowding at peak times as Northwest admits. Northwest Br. at 34, Northwest Reply Br. at 6. See also testimony of Delta in Phase 5:
that at a 58 percent [load factor], that means you have an awful lot of flights that are going out full and an awful lot that are going out at 20 percent. .
Phase 5 Hearing Tr. 2614.
However, it would be inefficient to have so much capacity that there was never any crowding. The Board does not rely on the present burden on the coach passenger due to a capacity shortage, but rather the burden of future capacity additions to keep load factors where they are now, if first class does not pay its fair share of those capacity addition costs.
. Order 74-3-82 at 27-30, J.A. 416-19.
. Delta Air Lines v. CAB, 162 U.S.App.D.C. 21, 28, 497 F.2d 608, 615 (1973), cert. denied, 417 U.S. 930, 94 S.Ct. 2640, 41 L.Ed.2d 233 (1974).
. Order 74-3-82 at 28, J.A. 417.
. See TWA Siesta Sleeper Seat Service, 27 C.A.B. 788 (1958). In that decision, the CAB prohibited providing sleeper seat service at first class rates as it required 30% more space than first class service. However, the Board did not adopt a full space allocated policy since it allowed sleeper seat service at only a 20% surcharge. 27 C.A.B. at 797. See also Continental Air Lines v. CAB, 173 U.S.App.D.C. 1, 13, 522 F.2d 107, 119 (1-974).
.It is true that, given the carriers’ present level of capacity, and the availability — at least prior to the current fuel shortage — of substantial numbers of empty seats, the carriers (and perhaps even coach passengers) benefit from charging some persons a premium fare (albeit below full costs) for a more luxurious service.
Order 74-3-82 at 125, J.A. 514.
The Board essentially accepted, arguendo, the carriers’ general marketing approach to first-class fares as a reasonable analysis of the short-term consequences of the present first-class structure.
Order 74-12-109 at 21, J.A. 605.
. See p. -of 179 U.S.App.D.C., p. 1304 of 551 F.2d supra.
. Cost savings, however, will result in the long run if the carriers prudently adjust the size of first-class compartments for any changes in traffic volumes that may occur. While the increase in seats per aircraft obtained thereby may not permit immediate frequency reductions, it clearly will forestall the need to add frequencies or to upgrade a schedule to a higher capacity aircraft.45
According to American, the total number of seats in most of its aircraft would increase by 4-5% if half of the first class cabin were converted to coach. This would mean that a carrier with 200 aircraft in its fleet would have the equivalent of 8 to 10 additional aircraft without having to buy a single airframe or engine. Considering the high costs of new aircraft, the savings to the carriers and their customers will be substantial.
Order 74-12-109 at 21, J.A. 605.
. Order 74-3-83 at 125-26, J.A. 514-15.
. Order 74-12-109 at 23, J.A. 607.
. “. . . continue to provide too much first-class service . . . ” Order 74-12-109 at 27, J.A. 611; “. . . about the size of their already excessive first-class compartments . . . ” Id., note 60.
. Section 401(e)(4), 49 U.S.C. § 1371(e)(4), dealing with terms, conditions, and limitations on certificates of public convenience and necessity states:
No term, condition, or limitation of a certificate shall restrict the right of an air carrier to add to or change schedules, equipment, accommodations, and facilities for performing the authorized transportation and service as the development of the business and the demands of the public shall require; except that the Board may impose such terms, conditions, or limitations in a certificate for supplemental air transportation when required by subsection (d)(3) of this section.
.Continental Air Lines v. CAB, 173 U.S.App.D.C. 1, 14, 522 F.2d 107, 120 (1974). In its order on reconsideration in Phase 9, the Board disclaims any intention to eliminate first class service. It calls predictions of the demise of first class a “prophecy of doom,” which it counters with statistics of the vitality of first class service on trans-atlantic routes, albeit at a smaller percentage, with similar first class/coach fare ratios. Order 74-12-109 at 25, J.A. 609. We do not explore this matter further, since we hold the Board justified in applying its cost theory. However, it does suggest that the Board considered the carriers’ pleas of injury.
. Order 72-5-101 at 5, J.A. 234, affirmed with some modifications on reconsideration, Order 73-6-102, J.A. 343.
. CAB Br. at 86. The Board’s counsel gives no citation to the Phases 6A and 9 opinions or records for these findings. Of course, on review, we must look at the findings, policies, and explanations made by the agency itself, rather than post hoc rationalization of counsel. SEC v. Chenery Corp., 318 U.S. 80, 95, 63 S.Ct. 454, 87 L.Ed. 626 (1943). Burlington Truck Lines v. U. S., 371 U.S. 156, 168-69, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962).
.Third, in determining revenue requirements for fare-level purposes, the impact of promotional fares will be eliminated. In other words, we intend to fix the basic normal-fare level on the basis of the revenues which would be realized and the expenses which would be incurred in the absence of the promotional fares. This will involve an adjustment of passenger yields to eliminate the dilution caused by the discount fares, together with related adjustments to traffic, expenses, and investments.
Order 72-12-18 at 55, J.A. 307.
. Continental Br. at 44.
. [W]hile treating first-class service under our Phase 5 policy would place the direct burdens resulting from first-class service on the carriers’ shareholders, it would still indirectly burden the coach passengers by creating a long-run drain on the carriers’ financial health which in turn would have an adverse effect on the services they would provide to coach passengers in the future.
Order 74-12-109 at 27, J.A. 611.
. The Board also distinguishes between first class service and discount fares, based on the physical configuration of the aircraft, and the ability to institute and remove “time promotional fares” without having to change that configuration. Opinion 74-12-109 at 26, J.A. 610.
. Groesbeck v. Duluth, S. S. & A. Ry. Co., 250 U.S. 607, 614-15, 40 S.Ct. 38, 41, 63 L.Ed.2d 1167 (1919).
. The Board similarly treated an analogous insular, predominantly tourist-pleasure travel market in the Mainland U.S.-Puerto Rico/Virgin Islands Fares decisions, Supplemental Opinion and Order 76-8-100 (August 18, 1976). In that decision, the Board adopted load factor standards of 68% for New York-San Juan and 60% for the other Mainland-Puerto Rico/Virgin Islands routes, as well as requiring a phased transition to a cost-based first class/coach fare ratio, found to be 200% for those markets.
. They feel the Board should permit broad flexibility for promotional discount fares for Hawaii, and should not insist on restricting such permission to discounts defined by reference to the specific DPFI requirements of short duration and satisfaction of the profit impact tests. They acquiesce in the use for Hawaii of a substantially different policy from that outlined in the DPFI, in view of the near unanimity of the carriers, supported by the State of Hawaii. They pledge to be vigilant in assuring that coach passengers will not be burdened either with higher fares or degradation of service. Finally, they contend that with such vigilance assured, natural economic motivations will lead the carriers themselves to avoid “unproductive discount fares” rather than require the Board to monitor against discount encroachment on yields and profits.
*356If we understand them correctly, they accept the fundamental analysis of the DPFI, concerning the damage wrought by fares below full cost for any level of service. However, they believe there can be departures that are wholesome and that the Hawaii market warrants an exception going beyond that provided in the short-term discount program set out in Phase 5.
. Algonquin Gas Transmission Co. v. FPC, 175 U.S.App.D.C. 215, 217, 221 n.13, 534 F.2d 952, 954, 958 n.13 (1976); Kennecott Copper Corp. v. EPA, 149 U.S.App.D.C. 231, 235, 462 F.2d 846, 850 (1972); Greater Boston TV Corp. v. FCC, 149 U.S.App.D.C. 322, 331, 463 F.2d 268, 277 (1971), cert. denied, 406 U.S. 950, 92 S.Ct. 2042, 32 L.Ed.2d 338 (1972).