(concurring):
Although I subscribe completely to the majority’s decision to remand to the Environmental Protection Agency this administrative action, two concerns compel additional comment. The more troubling one is that the Court has been obliged to find the power of the Administrator to issue effluent limitations by a series of subtle inferences, rather than by reference to clear statutory language. Also disturbing is the manner in which the Administrator has exercised his statutory duty to consider the costs of pollution control technology in relation to the benefits of abating pollution. This aspect of the Administrator’s decision would seem to constitute another ground for remanding the Phase I iron and steel operation controls to the Administrator for further consideration.
I.
The Water Pollution Control Act of 1972 is an important environmental protection measure, one of several such critical laws enacted in recent years. It holds promise for the American people that within a reasonable time the lakes and rivers of this country will no longer be threatened by unpurified discharges of industrial and residential waste. The undertaking Congress proposed was technologically uncertain and likely to prove costly as new facilities would have to be installed and old practices would have to be abandoned in order to achieve the desirable objectives set forth in the law. Congress, nonetheless, viewed the problem of water pollution with a sense of urgency. It provided that effluent limitations for point sources compatible with the “best practicable control technology currently available” were to be achieved by J uly 1,19771 and that limitations consonant with the “best available technology economically achievable” were to be attained by July 1, 1988.2 The Act expresses the goal that the 1983 limitations provide for water quality suitable for the protection of fish and wildlife and for recreational use, as well, and that all discharge of pollutants into navigable waters be eliminated by 1985.3
The states were chiefly responsible for the administration of the prior Act.4 In contrast, the 1972 Act vested the authority to issue permits for the discharge of pollutants in the federal Environmental Protection Agency (EPA).5 States were empowered to impose more stringent regulations on dischargers,6 and to exercise an initial veto on applications for federal permits,7 but the responsibility for prescribing minimum standards was vested in the EPA.8 The EPA, however, was authorized by Congress to delegate its permit-granting authority to the states subject to conditions and safeguards, including the retention in the EPA of power to establish basic guidelines with which each state’s implementation plan was to comply.9 Each permit issued under a state plan was to be subject to EPA review and veto.10 And should a state plan become inconsistent with the national water pollution regulations, the EPA was given authority to resume control of the issuance of permits in that state.11
*1074A critical issue common to this appeal and the CPC case 12 is the delegation to the states of the authority to establish minimum standards for pollutant discharge. Two issues determine the outcome of this aspect of each decision: First, are the' effluent limitations described in section 301 of the Act to be imposed by a separate set of regulations with which state-issued permits for discharge must comply, or are the effluent limitations to be embodied instance by instance in the terms of each individual permit? Second, if the effluent limitations are separate regulations, is the federal government or the state to issue those regulations?
The Eighth Circuit has interpreted the Act as imposing the effluent limitations as conditions in each permit issued by a state authority, albeit congruent with the federal guidelines. Judge Hunter, in his able analysis, demonstrates that effluent limitations are to be a separate set of regulations, distinct from the permits, and that these regulations are to issue from the EPA, not from the states.
Regardless of the merits of the respective positions of the Eighth Circuit and this Court, both have had to arrive at their divergent interpretations of this fundamental aspect of the Act by inferences from ancillary provisions, and by deciphering the legislative intent from reading scraps and bits of a convoluted legislative history.
The failure to provide a clear procedural structure on so basic a matter in the administration of the Act is disquieting. In one sense, the difficulty of interpretation imposed here is of little consequence; it is the work of the courts to explicate the laws, no matter how complex their structure. In this instance, however, statutory vagueness inflicts harm on the purposes of the Act and impels the courts to determine the allocation of authority between the national and state governments in the administration of this program. The conflict of interpretation between the Circuits does not initiate, but certainly perpetuates confusion in the administration of the legislation. Under these circumstances, neither the states nor the Administrator, nor the industries and municipalities which are to be regulated by the Act, may be confident which agency possesses the legal authority to promulgate effluent limitations. The result of such incertitude is delay in the implementation of the substantive provisions of the Act while the concerned parties engage in costly litigation to determine the legal powers of each enforcement agency. Such controversy postpones the achievement of the Act’s lofty objectives, and imposes a burden on the Courts of Appeals and the Supreme Court which they might have been spared by careful drafting.
In addition, the difficulty in interpretation impinges on the powers of the coordinate branches of government within the federal system. Ours is a cooperative federalism in which the states and the national government share responsibilities for many programs. The definition of the roles of the state and national governments in areas where they share concurrent powers is essentially a matter for Congress, not for the courts.13 The failure to create clear boundaries for the authority of the states and the EPA has thrust upon the courts a responsibility to infer legislative intent from the disparate provisions of this complex legislation. The courts have not evaded their responsibility, but our disagreement with the Eighth Circuit underlines the extent to which the courts can write law, even in areas of Congressional authority, when there has been a failure to manifest legislative intent by clear statutory commands.
*1075II.
The Administrator’s exercise of his discretion in connection with the cost analysis is the nub of my second concern. Congress has delegated to the Environmental Protection Agency a far-reaching power to strike a balance between water quality and the economy. In framing the guidelines for 1977 effluent limitations for existing point sources, the Administrator is directed by law to consider the costs of pollution control technology in relation to the benefits of pollution abatement. A reviewing court, asked by petitioners to declare an administrative action “arbitrary” or “an abuse of discretion,” is not to substitute its judgment for that of the agency. The Court must make a careful inquiry, however, to ascertain whether the decision of the agency was “based on a consideration of the relevant factors and whether there has been a clear error of judgment.”14 By this standard, set forth in Citizens to Preserve Overton Park v. Volpe, the conclusions of the Administrator with respect to the cost of applying the best practicable control technology would appear to be arbitrary. The record does not adequately demonstrate that his decision on the 1977 guidelines for existing point sources was based on a consideration of all the relevant factors.
Section 304(b)(1)(B) instructs the Administrator to consider “the total cost of application of [the best practicable control technology currently available] in relation to the effluent reduction benefits to be achieved from such application . ” when establishing guidelines for the 1977 effluent limitations.15 The elements of the total cost of applying the technology are not identified in the law itself. However, the legislative history of the Act may be used to support the notion that the term “total costs of application” encompasses not only the expense of erecting pollution control facilities but also the “external cost of potential unemployment, dislocation, and rural area development sustained by the community. . . . ”16
Having accounted for the total costs of applying the requisite technology, the Administrator must consider those costs in relation to the benefits to be achieved from the application of the technology. The Act does not mandate an exacting quantitative comparison in the form of a technical benefit/cost analysis. Nevertheless, from the remarks of Senator Muskie in support of the conference bill it appears that, with respect to the 1977 guidelines, the Act does require the Administrator to make an analysis sufficient to indicate that “the additional degree of effluent reduction is [not] wholly out of proportion to the costs of achieving such marginal level of reduction. . . . ”17
There is no doubt that Congress understood the difficulty and perhaps even the impossibility of reducing to monetary terms all, or even many, of the benefits from moderating pollution.18 The Administrator’s duty is to make the difficult comparison of the quantified and the unquantifiable. Nonetheless, by di*1076reefing the Administrator, when developing the guidelines for 1977, to consider the costs in relation to the benefits, even if only to the extent of determining that costs and benefits are not “wholly out of proportion,” Congress may reasonably be understood to have required that the Administrator survey both costs and benefits thoroughly, each in the terms best suited to understanding its significance.
Costs and benefits could not be accurately compared, if the Administrator were to omit significant benefits, or significant costs, in evaluating pollution control technologies for 1977. Only on the basis of a comprehensive survey could the Administrator conscientiously determine that the costs and benefits of reducing pollution were in some way proportionate. Without the aid of such a survey, decisions might be rendered which distort the priorities intended by Congress.19
In this ease, the Administrator has considered the plant costs of applying the control technology; however, his calculations in this respect are incomplete. In particular, he has assumed that all treatment facilities will be constructed on “green field” sites. Consequently, he has neglected to estimate any average costs for site clearance and preparation. Sites are assumed to need no extensive preparation such as rock excavation, pilings, drainage, grading, or tree removal. In addition, “[l]and acquisition costs are not included in cost estimates.” Other costs not considered in the Administrator’s calculations include the costs of expanding utility support systems and interconnecting utility runs. Although the steel companies have presented only conclusory assertions regarding the significance of these excluded costs, the Administrator would seem to be obliged by the Act to consider these items-as part of “the total cost of technology.”20 He cannot leave utterly out of account what might reasonably be expected to be significant cost factors.
If the Administrator is required to consider the external or secondary costs of applying the best practicable pollution control technology, it appears that he has also failed to satisfy the Act in this respect. Although a report commissioned by the Administrator21 noted the possibility of some plant closings, which would create unemployment and dislocate local economies, the Administrator does not set forth the weight he attached to these considerations. Nor does he find the possibility of plant closings too remote to be considered. If some closings are reasonably foreseeable, the report cannot be said to have presented adequate estimates of the total cost of such eventualities. Similarly, the record does not reveal that the Administrator has discounted the necessity for a choice *1077between expanded production capacity and the application of pollution control technology.22 If such a choice must be made, the Administrator should examine the possibility of any substantial economic impact associated with it. To the extent that Congress intended “total costs” to include secondary costs, the Administrator, it seems to me, should give explicit consideration to whether such costs, if they are significant, are likely to be passed on to other industries and ultimately to consumers.
Where costs are to be evaluated “in relation to” the benefits of reducing pollution, as in the case of the 1977 guidelines, the Administrator should survey the extent of the benefits as well as the costs that the pollution control technology may be expected to provide. The Final Development Document, however, describes only the types of harm that pollution control technology will help to abate. It does not account for the extent of the injury to be alleviated by the control technology. Consequently, it cannot be ascertained from the record whether the benefits to be attained are substantial.
The Administrator has concluded that “the benefits of reducing the pollutants discharged justify the associated costs .” of applying pollution control technology to Phase I iron and steel op-©rations.23 However, the failure of the Administrator to account fully for the costs of the control technology and to indicate in any way the extent of the benefits to be derived from reducing pollution, would seem to preclude meaningful comparison of costs and benefits. Any conclusions drawn from inadequate data would be without foundation. In view of these omissions, the Administrator’s determination cannot be said to be based on a consideration of the relevant factors. Accordingly, it would not appear to meet the test of Overton Park.24 It is quite conceivable that if all the facts were set forth, the Administrator’s determination would not be a “clear error of judgment.” But it is not the function of the appellate courts to “supply a reasoned basis for the agency’s action.” 25 Although courts are obliged to uphold an Administrator’s position when the path to his decision is reasonably discernible,26 the proceeding here presents this Court with an unblazed trail.
It is no small matter to which the Administrator is to direct his attention. A decision to protect the environment necessarily imposes heavy costs on the accustomed ways of production, employment, and consumption in this country. In passing the Water Pollution Control Act of 1972, Congress attempted to balance the improvement in the quality of *1078our water with the costs of that worthy endeavor. Despite the ambitious goals expressed in section 101 of the Act, Congress has ordained a measured national effort at the first stage of enforcement, from 1977 to 1983, requiring pollution abatement at existing point sources to be bounded by some proportionality to its cost. The guidelines and effluent limitations to be established by the Administrator have the potential to affect management’s investment decisions and, thereby, the employment prospects of thousands of citizens and the economic future of many communities. When an administrator’s action has such ramifications, he would appear to have an obligation to move cautiously, taking care that his judgments at each important step of the way are founded on sound analysis and detailed study. It does not appear that the Administrator has acted with care sufficient to satisfy his obligations under section 304(b)(1)(B). For that reason I conclude that the inadequate consideration of the element of cost in setting the guidelines for best practicable control technology is an additional ground for remanding this action to the Administrator.
. § 301(b)(1)(A), 33 U.S.C. § 1311(b)(1)(A) (Supp. Ill 1973).
. Id. (b)(2)(A).
. § 101(a), 33 U.S.C. § 1251(a) (Supp. Ill 1973).
. 33 U.S.C. 1151 et seq. (1970), as amended 33 U.S.C. § 1251 et seq. (Supp. Ill 1973).
. § 402(a), 33 U.S.C. § 1342(a) (Supp. Ill 1973).
. § 510, 33 U.S.C. § 1370 (Supp. Ill 1973).
. § 401(a)(1), 33 U.S.C. § 1341(a)(1) (Supp. Ill 1973).
. § 402(c)(2), 33 U.S.C. § 1342(c)(2) (Supp. Ill 1973).
. Id. (b), (c).
. Id. (d).
. Id. (c)(3).
. CPC International, Inc. v. Train, 515 F.2d 1032 (8th Cir. 1975).
. See Franklin Nat’l Bank v. New York, 347 U.S. 373, 378-79, 74 S.Ct. 550, 98 L.Ed. 767 (1954); Palmer v. Massachusetts, 308 U.S. 79, 84, 60 S.Ct. 34, 84 L.Ed. 93 (1939). Cf. Douglas v. City of Jeannette, 319 U.S. 157, 162-63, 63 S.Ct. 877, 87 L.Ed. 1324 (1943).
. Citizens to Preserve Overton Park v. Volpe, 401 U.S. 402, 416, 91 S.Ct. 814, 824, 28 L.Ed.2d 136 (1971); Duquesne Light Co. v. EPA, 522 F.2d 1186, 1192 (3d Cir. 1975).
. 33 U.S.C. § 1314(b)(1)(B) (Supp. Ill 1973). We note that the specific mandate to consider costs in relation to benefits is limited to the 1977 existing point source standards and does not appear to apply to the 1983 standards under subsection (b)(2)(B) or to new sources under section 306(b)(1)(B). In those sections the Administrator is, however, admonished to take costs “into account” or “into consideration.” The admonition may be taken to imply some kind of comparison of costs and benefits, although the proportionality between costs and benefits, indicated by section 304(b)(1)(B), would not seem to be required for these other sections.
. See Library of Congress, Environmental Policy Division, A Legislative History of the Water Pollution Control Act Amendments of 1972, 93d Cong., 1st Sess. 231 (Comm. Print 1973). (Remarks of Rep. Jones).
. See id. 170 (Remarks of Sen. Muskie). Cf. id. 1466 (Senate Report discussing the analysis of costs and benefits required by § 302).
. See id. 1466.
. Senator Muskie’s references to the “additional degree of effluent reduction” and to “such marginal level of reduction” seem to indicate that the comparison required by Congress is not between total benefits and total costs, but between the costs and benefits attributable to each significant step in pollution abatement. Presumably an incremental analysis would avoid the risk of hidden imbalances between cost and benefit. Such imbalances could be of serious magnitude where, as here, major industries are the subjects of regulation. An incremental analysis would, of course, be more complex than a simple comparison of total costs and benefits. To the extent that estimates of total costs and benefits are aggregates of the costs and benefits of each step in the pollution abatement process, however, the basic data for incremental analysis should be readily available.
. The costs in question will vary according to the conditions of each site, and the Administrator cannot be expected to produce exact costs for each site. He may, however, resort to estimated average figures.
. The Administrator appears to adopt the report and its supplement by his statement at 39 Fed.Reg. 24118 (1974) that “[t]he total cost of implementing the effluent limitations guidelines includes . . . the indirect economic costs identified in the supplementary report entitled “Economic Analysis of the Proposed Effluent Guidelines for the Integrated Iron and Steel Industry” (February 1974). The Administrator does not appear to engage in any more detailed discussion of indirect costs in the Final Development Document.
. As justification for a determination that there is no need for an investment choice between expansion of capacity and pollution control, the Administrator submits newspaper articles on the financial condition of the steel industry that have appeared subsequent to the promulgation of the regulations. The Court accepts those materials, relying on the reasoning of Amoco Oil Co. v. EPA, 163 U.S.App.D.C. 162, 501 F.2d 722, 729 n.10 (1974). In the circumstances of this case, where the regulations are to be remanded to the Administrator for reconsideration in many respects, it appears to be unnecessary and perhaps inappropriate for the Court to take notice in this fashion of subsequently developed information.
As a general principle, moreover, it is disturbing that courts should consider subsequently developed information to determine whether an administrative decision is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” To entertain such information may encourage the use of the courts as mere extensions of a continuous administrative process. It may also encourage some administrators to issue decisions that may well be arbitrary at the time of promulgation, hoping like Micawber that “something will turn up.” Neither of these consequences would appear to enhance the efficiency of the courts or the confidence of citizens in the fairness of the administrative process. See SEC v. Chenery Corp., 332 U.S. 194, 196, 67 S.Ct. 1760, 91 L.Ed. 1995 (1947); SEC v. Chenery Corp., 318 U.S. 80, 94-95, 63 S.Ct. 454, 87 L.Ed. 626 (1943).
. 39 Fed.Reg. 24118 (1974).
. 401 U.S. at 416, 91 S.Ct. 814.
. See Bowman Transportation, Inc. v. Arkansas-Best Freight Systems, Inc., 419 U.S. 281, 285-86, 95 S.Ct. 438, 442, 42 L.Ed.2d 447 (1974).
. See id. 286, 95 S.Ct. 438.