Independent U. S. Tanker Owners Committee, (anunincorporated Association), Appellant, v. Drew Lewis, in His Representative Capacity As Secretary Ofthe Department Oftransportation, et al., Appellees, 690 F.2d 908 (D.C. Cir. 1982)

Petitioner, the Independent U. S. Tanker Owners Committee, appeals from an unsuccessful challenge in the district court to both informal rulemaking and adjudicative action by the Secretary of Commerce,¹  the Maritime Administration (MarAd), and the Maritime Subsidy Board. MarAd (to choose a term by which to refer to all government defendants) is charged with responsibility for fostering the development and encouraging the maintenance of a U. S.-flag merchant marine and domestic shipbuilding industry. It proceeds, in part, by subsidizing the domestic construction of U. S.-flag ships intended for foreign trade. Ships receiving such subsidies are then precluded from competing in the unsubsidized domestic trade. MarAd, however, has the authority to accept total repayment of a construction subsidy and, in return, to lift the domestic trading restriction on the vessel.

Pursuant to its authority, MarAd promulgated an interim rule describing the procedures to be followed and factors to be considered in passing upon an application for repayment. Shortly thereafter MarAd purported to apply this rule in granting the application of Richmond Tankers, Inc., a defendant-intervenor in this case, for repayment of the construction subsidy on its tanker, the Bay Ridge.

We find procedural flaws in the promulgation of the interim rule and both procedural and substantive failings in the informal adjudication leading to approval of Richmond's application. Accordingly, we reverse the decision of the district court and remand with instructions to vacate both the rule and the results of the adjudication. The district court should order MarAd to promulgate a permanent rule and to reconsider with all dispatch whether the Bay Ridge should be allowed to continue operating in the domestic trade. For reasons discussed below, however, we do not require that the Bay Ridge cease domestic operations pending this reconsideration.

It has long been recognized that an adequate merchant marine, with U. S.-flag ships and trained American sailors, is vital to both the national defense and the commercial welfare of our country.²  We require a sound merchant marine to protect foreign trade and to provide support for the armed forces in times of war or national emergency. We also require a modern, efficient shipbuilding industry capable of providing military vessels in times of stress.

Unfortunately, the necessary ships, shipyards, and sailors are not readily forthcoming. High wage rates for skilled laborers and stringent safety standards make American production unattractive. High seamen's wages, protective working conditions, and high outfitting, insurance and repair costs make American operating costs uncompetitive. Direct and indirect subsidies given shipyards and operating fleets by other nations further aggravate the disparity in costs. As a result, the American shipping industry, left to its own resources, would have all its ships built abroad, registered under foreign flags, and manned by foreign seamen.³ 

In the domestic trade, this shipping dilemma has been readily resolved. Since the earliest days of the Republic, preferential legislation has mandated that only U. S.-built and U. S.-flag vessels can be operated in commerce between points in the United States.⁴  The Jones Act, § 27 of the Merchant Marine Act of 1920,⁵  provides that only vessels "built in and documented under the laws of the United States and owned by persons who are citizens of the United States" may engage in domestic trade, defined as trade "between points in the United States, including Districts, Territories, and possessions thereof embraced within coastwise laws ....⁶  Thus, high costs impair U.S. domestic shipping only to the extent that alternative modes of transport (railroads, trucking, pipelines, etc.) prove less expensive.

In U.S. foreign commerce, however, such protective legislation is not possible. Every foreign nation with which the United States trades has precisely the same interests and precisely the same right to have cargo passing between the two countries in ships of its flags. If the construction and operating costs of the foreign-flag vessels are lower, then on a purely competitive basis both import and export cargo of the United States will be carried exclusively in foreign-flag vessels. To forestall this highly undesirable situation, Congress for many years has authorized both a construction subsidy for ships to be built in U.S. yards and an operating subsidy for the manning of American-flag vessels by American citizens in accordance with American safety standards. Under the construction-differential subsidy (CDS) program, the Government may pay up to 50% of the construction costs of vessels needed for the U.S. foreign maritime trade.⁷  There is also a loan guarantee program for ship financing.⁸  The guarantee enables a shipbuilder to secure substantial credit to aid in financing the initial construction, reconstruction, or reconditioning of a vessel.

The U.S. merchant fleet is thus divided into two distinct segments. The "Jones Act" fleet operates, by economic necessity, only in the protected U.S. domestic trade. It cannot compete in the foreign trade with either foreign ships or the U.S. subsidized fleet because Jones Act ships are built and operated without subsidy and are thus far more costly to their American owners. The subsidized U.S. merchant fleet, on the other hand, is prevented by law from operating in the domestic trade.⁹  It would be grossly unfair to allow U.S. vessels that have received a subsidy of up to 50% of construction costs to compete with U.S. vessels whose owners paid the full cost of construction in U.S. yards. The Jones Act preference legislation, designed to encourage construction in U.S. shipyards and the employment of U.S.-flag vessels in the domestic trade, all without direct cost to the taxpayers, would be completely undermined if subsidized U.S.-flag competition were allowed to invade this preserve. Consequently, the owner of any ship built with a CDS must agree that the vessel will operate only in foreign trade, with two exceptions. First, CDS vessels may sail in the domestic trade on either the first or last leg of an overseas voyage. Second, MarAd may consent to the operation of a CDS vessel in the domestic trade for up to six months in any twelve-month period whenever it determines that "such transfer is necessary or appropriate to carry out the purposes of this chapter."¹⁰  Proportional repayment of the subsidy is required under either of these exceptions.

U.S.-flag tankers in the foreign market have not fared well despite Congress' program of construction and operating subsidies. In fact, a decline in Middle East oil production (originally precipitated by the Arab oil embargo), coupled with overbuilding, has led to a virtual collapse of the world tanker market. World-scale rates are well below the break-even costs of U.S. tankers, subsidized or not.¹¹ 

The domestic market, on the other hand, especially the West Coast/Panama Canal trade from Alaska, is flourishing. A shortage of very large crude carriers (VLCCs) has led to highly profitable rates. As a result, there has been persistent pressure from subsidized vessels for permission to enter the domestic market on a permanent basis in exchange for total repayment of any construction subsidies received.

MarAd was placed in an acute dilemma by such pressure. First, neither the Merchant Marine Act nor its legislative history offered any express indication that Congress intended to empower MarAd to approve full-repayment/permanent-release transactions. Second, even if MarAd is so empowered, freely granted releases might be contrary to the purposes of the Act since they would tend to undermine the domestic trade, making planning impossible and new construction unwise. The foreign trade would hang over the domestic trade like a sword of Damocles, ready to trim away any and all profit margin.

The first half of this dilemma, at least, was dissolved in 1980. MarAd had granted a series of requests for repayment during the '70s, culminating in the application of Polk Tanker Corp. for repayment of the CDS on the tanker Stuyvesant in 1977.¹²  MarAd dealt with each application on an ad hoc basis, without hearings or notice to interested parties in the domestic trade. When MarAd granted the Stuyvesant application, three competitors in the Alaskan trade brought suit challenging its authority to do so. The district court upheld MarAd's general power to grant permanent release from domestic trading restrictions in exchange for full CDS repayment.¹³  However, the court also held that releasing the Stuyvesant from such restrictions without analyzing the economic effect of that vessel's entry into the Alaskan trade was an abuse of discretion. The opinion deplored MarAd's ad hoc, unpublished determination and strongly urged MarAd to establish guidelines and procedures of general applicability to govern CDS repayment and permanent waivers of domestic trading restrictions so as to ensure equal treatment and a fair opportunity for comments to be heard.

Substantive guidelines and procedural requirements for applications should be established so that all members of the industry stand on equal footing before the Secretary, and these procedures should guarantee interested competitors and members of the public the opportunity to be heard.¹⁴ 

The United States Court of Appeals reversed the district court on the first point, without reaching the second.¹⁵  Citing the total absence of any authorization in either the Act itself or the legislative history, this court held that MarAd could not lift domestic trading restrictions for any period longer than the six months allowed for in the Act. The Supreme Court then reversed, citing MarAd's broad contracting powers and the absence of anything in the Act or the legislative history that forbade MarAd from permanently lifting the restrictions.¹⁶  The Supreme Court specifically did not address the question, discussed only by the district court, of when or how MarAd might exercise this discretion in accordance with the purposes of the Act.¹⁷ 

In 1978 MarAd published a proposed rule prescribing procedural and substantive guidelines for evaluating all applications for the total repayment of CDS in exchange for removal of domestic trading restrictions.¹⁸  No action on the rule was taken, however, pending final resolution of the Stuyvesant litigation. Once MarAd's authority was affirmed by the Supreme Court, the proposed rule was then republished and new comments were solicited.¹⁹ 

The terms of the rule would have allowed entry into the domestic trade only upon a determination of exceptional circumstances. The exceptional circumstances required to be shown were (1) the absence of "favorable opportunities for employment of the vessel during a protracted period of duration in excess of that normally associated with cyclical downturns;" (2) consequent inability of the "owner or other party responsible for payment" to make required principal and interest payments on mortgage debt guaranteed by MarAd under the Act, and imminent bankruptcy of such party; (3) consequent substantial financial loss to MarAd under its guarantee of the mortgage debt; and (4) likely absence of "any significant adverse competitive effect" on other vessels operating in the domestic trade.

The proposed rule also invited comments on a number of related issues: whether an ad hoc approach would be preferable in processing applications; whether the policy should vary for different CDS vessel types; whether additional exceptional circumstances should be included; what consideration should be given to competition; and what type of procedures should be adopted to ensure that all interested parties have an opportunity to present their views.

Written comments by interested parties were invited within sixty days. Comments were filed by twenty-two separate interests, including petitioner.

Contemporaneous with publication of the proposed rule, charterers and owners of six ships built with construction subsidies made application for repayment.²⁰  MarAd employed comment procedures in these dockets as well, and papers were filed by proponents and numerous opponents together with economic studies of the Alaskan coastwise tanker trade, the outlook for domestic shipping, and the impact of admitting the applicants to the trade. The economic aspects of the problem were common to all dockets as well as to the rulemaking.

On 15 October 1980 MarAd adopted, and made immediately effective, an interim rule different from and narrower in scope than the proposed rule.²¹  It applies only to tankers of at least 100,000 dead weight tons (DWT). The interim rule still states that approval will be granted only where exceptional circumstances exist, but "exceptional circumstances" is defined more loosely, with more play for administrative discretion. After determining that "no favorable opportunities exist for viable employment of the vessel in foreign trade during a protracted period," MarAd is to "consider" the following factors "among others" in determining whether exceptional circumstances exist.

(1) The purposes and policy of the Act.

(2) The economic impact on the U.S.-flag foreign and domestic tanker fleets.

(3) The economic impact on shipbuilding in the U.S.

(4) The financial situation of the applicant and its related companies.

(5) The financial interest of the Government including Title XI obligations.

The interim rule also provides for notice and comment procedures and for MarAd to "publish a concise written explanation for its action."

In 1969 Seatrain Shipbuilding Corporation, a subsidiary of Seatrain Lines, Inc., commenced construction of a series of four VLCCs. The third vessel in that series was the Stuyvesant. The fourth was the Bay Ridge. The Bay Ridge was constructed largely through government financing-a $28.8 million CDS and $81 million in debt obligations guaranteed by the United States.²²  The financing was pursuant to a contract between MarAd and Fillmore Tanker Corp., a subsidiary of Seatrain.²³  As required by the Act, that contract contained a provision, Article 9(b), limiting the vessel to the foreign trade.²⁴ 

In May 1979 the Bay Ridge was sold in a leveraged lease transaction to the United States Trust Company, acting not in its individual capacity but solely as owner-trustee on behalf of two owner-participants, Security Pacific Leasing, Inc., and American Road Equity Corp. The vessel was then bareboat chartered to Richmond Tankers, Inc., for 22 years, its full economic life. Richmond was originally organized by Seatrain specifically for this purpose. Substantially all of the obligations of Richmond with regard to the charter-hire payments to the owner-trustee are guaranteed by Seatrain. Seatrain is also required to pay all of the operating costs of the Bay Ridge, and all other costs of Richmond should revenues fail to cover such costs.²⁵ 

The upshot of this somewhat confusing leveraged lease transaction is that the government's only recourse, in the event of default on the loans it has guaranteed, is against Seatrain or Richmond or, ultimately, the vessel itself.²⁶  The owner-trustee and the two owner-participants have managed (such is the nature of the leveraged lease transaction MarAd has allowed) to remain free of any entanglement or any ultimate financial responsibility.

Upon its delivery on 17 May 1979 there was no employment available for the Bay Ridge in the foreign trade. It went immediately into lay-up where it remained until early 1980. On 17 April 1980 it entered the domestic trade pursuant to a six-month waiver of the domestic trade limitations.²⁷  During that time, the Bay Ridge carried Alaskan crude from Valdez to the western terminal of the Panama Canal, completing six voyages. The vessel then went into lay-up again.

It was readily apparent that the vessel had no prospects of supporting itself in the foreign trade, even with a six-month waiver each year allowing it to operate in the domestic trade.²⁸  As a consequence, Richmond was in grave financial straits. Seatrain was all but bankrupt.²⁹ 

By application dated 1 May 1980 Richmond requested that MarAd amend the CDS contract with respect to the Bay Ridge to remove trading restrictions in exchange for CDS repayment. Richmond's application was noticed in the Federal Register on 15 May 1980, and comments were received on the application. By letter dated 13 November 1980, less than a month after publication of the interim rule, MarAd announced its determination that it would grant Richmond's application.³⁰  At the same time, MarAd announced that it had considered and decided to defer other pending CDS-repayment applications. MarAd stated that it acted immediately on Richmond's application because of grave financial difficulties not faced by the other applicants.

On 19 January 1981 the CDS was repaid. The money was raised through the issuance by the owner-trustee of Title XI bonds guaranteed by the government.³¹  Since that time the Bay Ridge has been profitably employed in the Alaskan trade.

On 25 November 1980 the Independent U.S. Tanker Owners Committee (ITOC) filed its complaint in district court for review of the interim rule and of the Bay Ridge decision. Seatrain Lines, Inc., and Richmond Tankers, Inc., filed a Motion to Intervene as defendants, granted on 5 January 1981.

ITOC is composed of owners and operators of nonproprietary, unsubsidized tanker vessels, many of which are operating in the carriage of crude oil from Alaska to points in the U.S. ITOC's members are, thus, directly affected by any new entrants to that trade.

ITOC raised both substantive and procedural challenges to the interim rule. The rule is substantively flawed, ITOC claims, because it does not set meaningful standards to limit MarAd's discretion. The alleged procedural failing is that no statement of basis and purpose was published with the rule as required by the Administrative Procedure Act.³²  ITOC also challenges the Bay Ridge decision, first, as based on an invalid rule, second, as based on a staff study not disclosed to interested parties for comment, and third, as unaccompanied by a statement evidencing reasoned consideration of all material facts and issues. The district court granted summary judgment for the defendants on all counts. This appeal followed.

The proposed rule originally published by MarAd was "outcome determinative." That is, it listed a number of necessary conditions, conditions that had to be satisfied before MarAd would allow CDS repayment. There might be considerable debate over whether those conditions obtained in any particular case, but the inquiry would be an objective and limited one. MarAd would play a fact-finding rather than a discretionary role, and interested observers could try to plan according to their own understanding of the relevant facts.

In its comments on the proposed rule, ITOC stressed the need for just such a rule if the domestic shipping industry was not to be paralyzed by uncertainty.³³  In its briefs in this court, ITOC now claims that MarAd's failure to adopt such a rule was an abuse of discretion, out of harmony with its statutory mandate to foster the domestic trade.³⁴  The interim rule, ITOC claims, affords no guidance to the interests most affected. It speaks merely of "factors" the agency will "consider ... among others."³⁵  Without firm standards or guidelines, it is in effect a nonrule whose vagueness exacerbates the problems it should resolve.

There are major problems with this argument. First, it rests to a considerable extent on the premise that MarAd was required to limit its discretion through the promulgation of rules. But, though sound as a matter of policy, the proposition is false as a matter of law.

In the Stuyvesant litigation the Supreme Court found MarAd had broad discretion to grant applications for CDS repayment. Substantial arguments have been made that MarAd was then bound to promulgate rules to channel that discretion, not only as a matter of proper administrative practice,³⁶  but also as a means of abiding by its statutory mandate. For it is of the essence of MarAd's duty to enhance the possibilities for domestic construction and expansion and to minimize interference with the free market forces normally at work in the domestic trade. And, as already noted, if every ship built with a construction subsidy overhangs the domestic trade, with no recognized principle to restrain its sudden entry, existing and prospective domestic operators will be deterred from building new ships. The trade needs published rules to delimit the range of the otherwise unknowable and make planning possible.³⁷ 

But, despite the obvious wisdom of MarAd's decision to proceed by rulemaking in this instance, we cannot say that it was obliged to do so. SEC v. Chenery³⁸  establishes with unmistakable clarity that "the choice between proceeding by general rule or by individual, ad hoc litigation is one that lies primarily in the informed discretion of the administrative agency." This principle has been reaffirmed by the Supreme Court in subsequent cases,³⁹  and we can but adhere to it. Thus, since MarAd was not legally obliged to limit its discretion through rulemaking at all, it is difficult to find legal fault with its failure to go further than it did.

Of course, the court is still charged with the responsibility of ensuring that any rule MarAd does adopt, whether voluntarily or on compulsion, harmonizes with its statutory mandate. But the primary responsibility for ensuring such harmony lies with the agency, and the court is only to reverse if MarAd's action is "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law."⁴⁰  We cannot say that the interim rule falls outside these necessarily wide bounds.

The rule adopted does serve some useful purposes. First, it sets as a necessary condition of CDS repayment a finding that "no favorable opportunities exist for viable employment of the vessel in the foreign trade during a protracted period of duration in excess of that normally associated with cyclical market downturns ...."⁴¹  Second, it gives a list of relevant factors to which MarAd will primarily direct its attention. And third, it provides for notice to, and comment by, interested parties and requires a concise written explanation for MarAd's action.⁴²  This last provision is especially important to the scheme of the interim rule. Such participation by nonparties who may be affected by the outcome is automatic in rulemaking proceedings, but not in informal adjudication. Given that MarAd decided to retain considerable discretion on individual applications, rather than adopt an outcome determinative rule, it was crucial to provide affected interests with input into the process and a reasoned explanation for any decisions taken.

Thus, the rule adopted at least ensures that interested parties will have their say before MarAd makes its ad hoc determinations and that those parties will receive a statement of reasons responsive to their comments. It may not be an ideal rule, but it is certainly a genuine one, and we cannot say that it is so contrary to MarAd's statutory obligations as to constitute an abuse of discretion.

In sum, it is for MarAd to determine whether an invariable formula for CDS payback is appropriate or whether the complexities of maritime administration are such that some flexibility and discretion must be retained. Petitioner is understandably dissatisfied with MarAd's determination, but that is not a sufficient ground for a court to overturn it.

As noted above, MarAd was not required by law to promulgate any rules limiting its discretion. But it was nonetheless bound by the requirements of the Administrative Procedure Act (APA) when it decided to do so.⁴³  Furthermore, calling the rule adopted an "interim" one does nothing to insulate it from the judicial review authorized in the APA. The rule published 15 October 1980 was made effective immediately, and it is still in effect now, nearly two years later.⁴⁴  It is legally binding on MarAd, in its requirements of notice, opportunity for comment, and a statement of reasons, as well as the required finding concerning opportunities in the foreign trade, and cannot, therefore, be classified as a mere statement of policy.⁴⁵  MarAd was thus bound to incorporate in the rule it adopted "a concise general statement of (its) basis and purpose."⁴⁶ 

This requirement is not to be dealt with lightly. We have cautioned before against an "overtechnical reading" of the words "concise" and "general."⁴⁷  The statement of basis and purpose should "enable us to see what major issues of policy were ventilated by the informal proceeding and why the agency reacted to them as it did."⁴⁸ 

The basis and purpose statement is not intended to be an abstract explanation addressed to imaginary complaints. Rather, its purpose is, at least in part, to respond in a reasoned manner to the comments received, to explain how the agency resolved any significant problems raised by the comments, and to show how that resolution led the agency to the ultimate rule.⁴⁹ 

MarAd has failed completely to fulfill its obligations in this case. It published the interim rule without any attempt to discuss, much less respond to, the comments received. In fact, aside from some statutory boilerplate taken verbatim from the original proposal for rulemaking, the only language even approaching a statement of the rule's basis and purpose is a single sentence.

After careful consideration of all comments received in response to the proposed rulemaking, as well as to the individual notices of application for total CDS repayment, the Board has decided to adopt a rule on an interim basis but to limit the application of its rule to tankers of at least 100,000 DWT.⁵⁰ 

We have recently condemned just such a one-sentence statement as wholly uninformative.⁵¹  As Judge Friendly has noted, "the agencies do not have quite the prerogative of obscurantism reserved to legislatures."⁵² 

Appellees nonetheless claim compliance. They refer us to a nine-page staff memorandum, dated 7 October 1980, recommending adoption of the interim rule. But this memorandum does nothing to alleviate MarAd's failure to provide an adequate statement of basis and purpose. First of all, the memorandum was never published, as the APA specifically requires. It was only tendered to the district court as part of this litigation three months after its formal date. Second, MarAd did not adopt the memorandum. It only adopted the interim rule. Thus, there is no clear indication that the memorandum represents MarAd's considered position. Finally, the memorandum, itself, even if we were to accept it, does not constitute an adequate statement of basis and purpose. It is primarily descriptive, not explanatory. No explanation is given for why the rule originally proposed was abandoned. No explanation is given for why considerable discretion was retained by MarAd in the face of strong industry representations that clear guidelines were needed. No explanation is given of how a vague promise to "consider" the domestic industry's concerns would ameliorate those problems that originally precipitated the rulemaking procedure. These issues are surely "significant enough to step over a threshold requirement of materiality" requiring comment.⁵³  Yet, no comment was made.

Alternatively, MarAd attempts to excuse its failure to produce an adequate explanation for the adopted rule as "merely a technical violation which could be remedied by issuance of a fuller statement of basis and purpose."⁵⁴  Such a post hoc rationalization, however, would be beside the point. The preparation of a statement of basis and purpose should play an integral part in the decisionmaking process. It ensures thoughtful consideration of the various issues raised before a decision is published. Even if the decision is made first and staff members are then instructed to prepare a supporting statement, that statement is likely to be scrutinized by the decisionmakers themselves before it is issued. And as long as the decision has not already been announced it is freely changeable if they are dissatisfied with the statement. Furthermore, in the process of preparing the statement, staff members themselves may become aware of unforeseen difficulties with the decision, with the result that individual decisionmakers change their mind. But if the decision is announced prior to the preparation of the statement the process is no longer one of investigation. There is an overwhelming institutional bias in favor of justifying the result in any way possible.⁵⁵ 

There may well be good reasons for rejecting the domestic industry's contention that an outcome-determinative rule is preferable to MarAd discretion. Undoubtedly, MarAd could produce some of these reasons in a litigation affidavit. But there are none to be found, where they belong, in a contemporaneous statement of basis and purpose. There are none to be found anywhere in the record. Given the substantial economic interests involved, both local and national, MarAd's failure to act responsibly in this regard, in accordance with mandated procedures, is not only surprising, it is certainly to be deplored.

Accordingly, we remand to the district court on this issue with instructions to vacate the "interim" rule and to order new rulemaking procedures. Upon reflection, MarAd may or may not decide to adopt a permanent rule similar to the interim one. All that we require is that MarAd's justification for adopting the version of the rule it now chooses be clearly and thoughtfully presented in a statement published contemporaneously with the rule.

In appraising the Bay Ridge decision, we are faced at the outset with an interesting and unresolved question of administrative law. What is the effect of an invalidly promulgated rule upon an adjudicative decision that relies on that rule when the decision, without any rule, would have been within the discretion of the agency? In other words, we have seen that MarAd was not legally obliged to limit its discretion through rulemaking. It could have chosen to make repayment decisions on a purely ad hoc basis. But instead it chose to establish a rule governing the consideration of each application. MarAd having botched the rulemaking, the question is whether the subsequent adjudication of an application can still be upheld in its own right.

Appellees claim that any invalidity of the interim rule merely wipes the slate clean, as if MarAd had never even attempted to promulgate a rule. The Bay Ridge decision is, thus, "a discrete exercise of (MarAd's) discretionary authority and does not stand or fall on the basis of the rule."⁵⁶  However, no relevant cases are cited to support this view.

Appellant stresses MarAd's express reliance on the interim rule in deciding to allow repayment of the Bay Ridge subsidy. A staff memorandum of 22 October 1980 recommending the decision explains that "the Richmond application ... clearly satisfies all the factors as developed in the Board's interim rule."⁵⁷  And the 13 November letter announcing the decision cites and purports to track the rule item by item.⁵⁸  "Logically," appellant claims, "agency action taken in pursuance of an invalid regulation cannot stand."⁵⁹  Again, however, no supporting case law is cited.

The dearth of precedent on such a fundamental point is surprising. There is a Supreme Court case on point, NLRB v. Wyman-Gordon Co.,⁶⁰  but its result is not dispositive. Six Justices held that a rule promulgated prospectively by the NLRB in an informal adjudication was invalid. An order subsequently based on that rule in another informal adjudication was nonetheless upheld. But only four Justices subscribed to the proposition that the invalidity of the rule did not invalidate the order based upon it. Three Justices thought the original rule was valid, while the other two, in dissent, found both the rule and, therefore, the order invalid.

It would serve little purpose to guess how the current Court would decide the same issue. In the absence of controlling precedent we must rely upon reason, and reason in Wyman-Gordon, despite the 4-2 head count, seems to rest entirely with Justice Harlan's dissent. He points out that SEC v. Chenery⁶¹  established the principle that a court is obligated to remand a case if an agency has relied upon an improper basis to justify its action. Since the major reason the NLRB gave in support of its order was the invalid rule, Justice Harlan felt Chenery required remand.

If, as the plurality opinion suggests, the NLRB may properly enforce an invalid rule in subsequent adjudications, the rule-making provisions of the Administrative Procedure Act are completely trivialized.⁶² 

One purpose of the APA's rulemaking provisions is to ensure that all of the competing considerations are canvassed before a decision is reached. If the NLRB could violate those provisions in adopting a rule and then issue an order based on the rule, without recanvassing the competing considerations, this purpose would be thwarted. Since Justice Harlan, unlike the plurality, was unwilling to speculate as to the result of a full (re)appraisal of the rule by the NLRB, he wanted to remand the order.

To repeat, Justice Harlan's reasoning seems inescapable. But the question remains whether it is applicable to this case. In Wyman-Gordon the challenged order followed automatically from the invalid rule. The former ("Furnish the union with the names and addresses of your employees.") was but a particular instance of the latter ("In an election an employer shall furnish the union ...."). The NLRB did not reconsider the factors pro and con requiring employers to furnish addresses when it issued the order. Thus, the employer could rightfully claim that it was deprived of an opportunity, because of an invalid rule, fully to argue its case.

No such claim can be made in this case. The rule at issue here merely indicates the procedures MarAd must follow and the factors it must consider in passing on repayment applications. The rule did not dictate the Bay Ridge decision in the way the Excelsior rule dictated the order at issue in Wyman-Gordon. Rather, MarAd, in theory, was conducting the informal adjudication in an open and receptive manner within the broad confines of the rule. The key point of distinction is that ITOC has shown no way in which its interests in the adjudication were procedurally impaired by MarAd's adherence to those confines. The rule did not lead MarAd to refuse to entertain any probative evidence or sound arguments against granting Richmond's application. The "factors to be considered" by MarAd did not exclude any factors claimed relevant by ITOC. As we have seen, ITOC's primary complaint against the interim rule is that it left MarAd with too much discretion to reconsider and reweight all the relevant factors in considering each repayment application. Furthermore, ITOC cannot claim to have been harmed either by the threshold requirement that an applicant demonstrate a lack of opportunities in the foreign trade or by the requirement, not ordinarily found in informal adjudications, of notice to, and an opportunity for comment by, interested parties.

In sum, unlike the employer in Wyman-Gordon, ITOC has no claim that its interests in the informal proceeding were impaired by MarAd's adherence to an invalid rule. In fact, its interests were procedurally advanced. Thus, we cannot conclude that the invalidity of the interim rule automatically invalidates the Bay Ridge decision. We must therefore proceed to an independent evaluation of that decision.

Our responsibility in reviewing this informal adjudication⁶³  is essentially two-fold. First, we must review the record to ensure that MarAd's decision is not "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law."⁶⁴  The critical elements of such review are clear. While MarAd's decision is "entitled to a presumption of regularity," the presumption "is not to shield ... action from a thorough, probing, in-depth review."⁶⁵  Thus, though we are not permitted to substitute our judgment for MarAd's, our inquiry must be "searching and careful,"⁶⁶  and we must ensure "both that (MarAd) has adequately considered all relevant factors ... and that it has demonstrated 'a rational connection between the facts found and the choices made.' "⁶⁷ 

Our second obligation is to examine the procedures MarAd employed in reaching its decision to ensure that they comply with the APA and any applicable statutory or constitutional requirements.⁶⁸  The critical elements of this review, however, are far from clear. The APA rulemaking requirements of notice, an opportunity for comment, and a concise general statement of basis and purpose do not extend to informal adjudication. And where, as here, MarAd's authorization to accept repayment applications is inferred from rather than found in the Maritime Act, there are no statutorily mandated procedures to which the court can turn. Yet, some minimum procedures are necessary to provide a record adequate for the court to perform its review.

The distinct and steady trend of the courts has been to demand in informal adjudications procedures similar to those already required in informal rulemaking.⁶⁹  Courts have required some explanation for agency action⁷⁰  and, to ensure the adequacy of that explanation, some opportunity for interested parties to be informed of and comment upon the relevant evidence before the agency.⁷¹  Thus, despite the Supreme Court's dictum in Vermont Yankee Nuclear Power Corp. v. Natural Resources Defense Council, Inc.,⁷²  that courts may not add to the procedural requirements of the APA except in "extremely rare" circumstances, we are justified in demanding some sort of procedures for notice, comment, and a statement of reasons as a necessary means of carrying out our responsibility for a thorough and searching review.

This demand is further bolstered and given a more concrete form by the interim rule adopted by MarAd. It states the procedures MarAd must follow in considering applications for total repayment of unamortized CDS for tankers of at least 100,000 DWT.

With respect to any such request received, the Board will publish a notice in the Federal Register, providing opportunity for comment by interested parties. After the Board has acted upon any such application, the Board will publish a concise written explanation for its action.⁷³ 

We have explained above why MarAd's adherence to an invalid rule should not invalidate the Bay Ridge decision. For some of the same reasons, any failure to adhere to that invalid rule should invalidate the decision. In other words, despite the procedural flaws in the promulgation of the interim rule, we hold that MarAd was bound by the requirements of that rule.

MarAd itself determined that, in the absence of an outcome-determinative rule, procedural protection for all interested parties was desirable. Under ordinary circumstances, that determination, embodied in a rule, would be binding on the agency.⁷⁴  The fact that MarAd failed to provide an adequate explanation for the rule (more particularly, for why it didn't provide substantive protection as well) should not relieve it of the burdens imposed by the rule. MarAd is, effectively, estopped from deriving any subsequent procedural leeway from its past procedural failings.

We must, therefore, ensure not only that MarAd's decision was not arbitrary or capricious, but also, as a related task, that MarAd made available for comment all relevant evidence and produced an adequate explanation for its action, canvassing the competing comments received and explaining why it resolved the differences as it did.

On 20 October 1980 MarAd's Office of Trade Studies and Statistics presented the Office of Subsidy Contracts with a report of twenty-two, single-spaced pages on the Bay Ridge application.⁷⁵  The first three and one-half pages provide background material. There follows nine and one-half pages summarizing the various comments, replies, rejoinders, and motions of interested parties. The remainder of the report discusses the application in terms of the factors listed in the interim rule and ends by recommending its approval.

Two days later the Office of Subsidy Contracts presented a report to the Maritime Subsidy Board, also recommending approval of Richmond's application.⁷⁶  The report is sixty-two pages, double-spaced. After six pages of background material, the report contains forty-nine pages of analysis of the Bay Ridge application in terms of the interim rule. This analysis, though somewhat more detailed, largely tracks that of the 20 October report. In fact, large sections are taken verbatim from the earlier report. The 22 October report concludes with seven, double-spaced pages of recommended findings and determinations to be made, and actions to be taken.

Neither of these reports were published. They only came to light during the course of the litigation following MarAd's decision. Instead of publishing the reports, on 13 November MarAd took the last seven pages of the 22 October memorandum and, after changing the tense of the verbs so that recommendations became actions, sent them to the interested parties in a four-page, single-spaced letter.⁷⁷  Only three paragraphs of that letter, approximately one page, deal with the merits of the Bay Ridge application, and those paragraphs are wholly conclusory, mostly just repeating the language of the interim rule or stating, without any support, ultimate facts relevant to factors listed in the rule.⁷⁸  The remainder of the letter consists of background filler, deferment of other pending applications, details of the repayment transaction and a denial, in advance, of all requests for reconsideration.

MarAd's 13 November letter clearly falls far short of the required explanation for its action. It contains no constitutive facts and no analysis of the major issues so sharply contested by the interested parties to the proceeding. Yet no further explanation was proferred. MarAd could not have more flagrantly violated both judicial precedent and its own binding rule.

We have chided MarAd in the past for having "clipped off the last five pages of (a staff) memorandum and recorded it as its own 'finding and determination' in the matter and sent it along, in haec verba, to appellants by letter ...."⁷⁹  The admonition obviously did not strike home.

Appellees argue that the inadequacy of the 13 November letter as an explanation of MarAd's action is merely a technical failing remedied by subsequent disclosure of the 22 October memorandum, allegedly adopted in toto by MarAd at a 10 November meeting.⁸⁰  But it is, to say the least, very strange procedure for an agency to adopt a document as the foundation for its decision and then reveal only the last seven pages of it until forced to fuller disclosure by litigation. Moreover, the 22 October memorandum gives rise to problems of its own, both procedural and substantive.

Most of the controversy during the Bay Ridge proceeding centered on a single, general issue. It was not seriously disputed that there were no opportunities for the Bay Ridge in the foreign trade, or that both Seatrain and Richmond were in serious financial straits, or that the government stood to lose a great deal of money if the Bay Ridge was unable to generate sufficient income to pay off its guaranteed loans. The crucial issue was the effect that admission of the Bay Ridge would have on the domestic tanker market. Could the Bay Ridge be absorbed without noticeable effect, or would its entrance cause a tanker glut harmful to the stability of the trade? This question resolves into numerous, more specific issues such as anticipated Alaskan crude oil production, probable tanker supply, and demographics. The amount and type of shipping required turns on how much of the oil produced is consumed locally or on the West Coast, which is a short haul from Alaska, and how much proceeds to the East or Gulf Coasts via the Panama Canal, a long haul requiring transshipment at the west side of the Canal. Large vessels, VLCCs such as the Bay Ridge, are most efficiently employed in carriage from Alaska to the Canal.

Prior to its 22 October staff memorandum, MarAd had produced and made available to interested parties two studies of these issues. One, dated 26 August 1980, compared MarAd's own estimates as to production, tanker supply and demand with those submitted by the applicant.⁸¹  The other, dated 24 September 1980, presented and compared two different scenarios, one based on U.S. Department of Energy forecasts and one based on projections of the California Energy Commission.⁸² 

As noted, both these reports were made available. The parties were able to comment upon and respond to them, arguing either that the forecasts should be altered or, agreeing with the forecasts, that certain conclusions should be drawn from them. The facts and figures in the 22 October memorandum, however, were different from those in both the earlier reports. MarAd based its analysis on, and drew its conclusions from, a new set of forecasts without disclosing them to the parties for comment.

We have said before that even in an informal adjudication parties have a right to be informed of and comment on staff positions.⁸³  "Our cases make clear the importance of such comment in allowing a court to review the action taken by the agency, as well as in facilitating informed agency decisionmaking itself."⁸⁴  This importance is augmented in cases such as this one, where MarAd has refused in advance to entertain any petitions for reconsideration.

Richmond claims that appellant was already familiar with most of the basic data in the 22 October report. Indeed, Richmond cites the affidavit of a vice-president of the economic consulting firm employed by ITOC, claiming that "(t)he basic data in the October Memorandum were substantially in agreement with the economic analysis previously submitted by ITOC."⁸⁵  From this Richmond concludes that ITOC's only possible complaint is with MarAd's methodology and analysis. But "an agency has no obligation to give the parties a preview of its analysis or statement of its tentative conclusion."⁸⁶ 

Richmond certainly has a point. An agency is not obliged to publish a tentative opinion for comment. Otherwise, an absurd and endless process could ensure: notice and comment procedures followed by a tentative opinion followed by comment followed by a new tentative opinion (necessary to take account of and respond to the new set of comments) and so on. But where an agency's analytic task begins rather than ends with a set of forecasts, sound practice would seem to dictate disclosure of those forecasts so that interested parties can comment upon the conclusions properly to be drawn from them. We do not wish to lay too much stress upon the particular point. But the failing is part of an overall picture of totally unacceptable agency practice: A staff report was produced. It differed in fundamental respects from prior staff reports to which interested parties directed their comments. (Misdirected their comments, we now should say.) MarAd relied on the report in making a decision with substantial economic consequences. But the decision was published without any explanation. The report remained buried in the bowels of the agency.

All standards of fairness and due process in administrative law preclude such behavior. And the only way to ensure that it will not continue is to refuse to accept the decision resulting from it.

As noted, the 22 October study upon which MarAd's decision was based differs in fundamental respects from the two earlier reports issued by MarAd. The first report, dated 26 August 1980, predicts surplus nonsubsidized tonnage in the Alaskan trade of 984,000 DWT in 1985 and 594,000 DWT in 1990.⁸⁷  In the domestic trade generally it predicts a surplus of 506,000 DWT in 1985 and 657,000 DWT in 1990.⁸⁸  The Report concludes that there will be "sufficient nonsubsidized tonnage to meet not only the tonnage requirements for the Alaskan oil trade, but also the requirements for the other domestic trades, as well."⁸⁹  In other words, entry of the Bay Ridge would exacerbate the problem of anticipated overtonnage in the domestic trade generally and the Alaskan trade in particular.

The 24 September report departs from the 26 August report only in its discussion of West Coast consumption. The report compares consumption forecasts of the Department of Energy (DOE) with those of the California Energy Commission (CEC). It concludes that under the DOE scenario of declining West Coast consumption (and, hence, a greater need for long-haul carriage), the Bay Ridge and other CDS-built VLCCs "would be necessary by 1985 to offset shortfalls in Jones Act tonnage, but the requirement would decline thereafter and by 1990 12-month/year operation by the Bay Ridge would generate surplus capacity."⁹⁰  Under the CEC scenario of modest growth in West Coast consumption (the scenario adopted in the 26 August study), VLCC requirements "can be met throughout the period 1980-1990 by existing Jones Act tonnage ships augmented by existing CDS-built ships operating under six-month waivers."⁹¹ 

The 22 October report, upon which MarAd relies, paints an even bleaker picture of the domestic tanker market. A chart contained in the report⁹²  predicts over-tonnage for the trade generally of 2.76 million DWT in 1985, over five times the excess tonnage predicted in the 26 August report, and 1.89 million DWT for 1990, almost three times the excess tonnage previously predicted. No specific figures are given for the Alaskan trade for which the Bay Ridge was destined.⁹³  But the text is unencouraging, prophesying a continuous decline in demand due to stagnant production levels and a gradual decline in Panama Canal oil transshipments. "By 1986 ... (t)he demand for Alaska/West Coast/Panama tankers is expected to decline by 1.4 million DWT.... From 1986 to 1990 ... (this trade) is expected to decline by another 1.4 million DWT ...."⁹⁴ 

Yet somehow, mirabile dictu, the report finds room for the Bay Ridge which was not there one month before. It does so by providing separate supply and demand figures for vessels of varying sizes.

In the prior two reports, tonnage was apparently viewed as interchangeable. The 22 October study takes a more sophisticated approach. Supply and demand figures are broken down by vessel size and, despite huge overtonnage generally, a deficit is predicted for vessels of over 100,000 DWT through 1985 of at least 244,000 DWT, just enough to accommodate the 225,000 DWT capacity of the Bay Ridge. The demand for such vessels will then rapidly decline so that a surplus of 81,900 DWT is predicted for 1988, increasing each year thereafter.⁹⁵ 

The text of the report explains why tankers of 100,000 DWT and above are treated as an insulated category.

These vessels are engaged exclusively in the Alaskan crude oil trades, primarily between Valdez and the Panama Canal.... Competition with the 40,000-99,999 DWT size vessels could occur in the Alaska/West Coast trades, but it is unlikely to occur in the Alaska/Panama Canal oil trades because of economies of scale inherent in the larger crude carriers.⁹⁶ 

Thus, an anticipated 1985 surplus of 1.841 million DWT in the 40,000-99,999 DWT category⁹⁷  is not thought to alleviate the shortage of the larger vessels.

Unfortunately, this conclusion and the chart generally are directly contradicted two pages earlier in the report, where it is predicted that shipments between Alaska and the Panama Canal will steadily decline. "The partial equalization of domestic and world crude oil prices at the wellhead and the increased refining capacity for sour crude on the West Coast are expected to result in a decline of Panama transshipments to about 200,000 barrels per day."⁹⁸  Panama Canal transshipments should continue to decline "to marginal levels by 1990."⁹⁹ 

Where, then, stems the exclusive need for tankers in excess of 100,000 DWT? Two hundred thousand barrels per day through the canal generates a need for 1 million DWT in tankers of over 100,000 DWT.¹⁰⁰  It follows that in 1986 there will be a surplus of 2.192 million DWT of ships in that category for the Panama trade, not a deficit of 161,000 DWT as the chart predicts. These extra ships will have to ply the Alaska-West Coast trade where, by the report's own admission, they will compete with surplus ships in the 40,000-99,999 DWT category.

Since the statistical chart is presented without any explanation for the derivation of its figures, the apparent contradiction between text and chart is impossible to resolve. In the face of this contradiction, we cannot allow MarAd's conclusion that there is room for the Bay Ridge in the domestic trade to stand.

A further problem is that, even if we were to accept MarAd's chart and ignore the dire predictions in text, entry of the Bay Ridge would begin to generate excess tonnage as early as 1986-87. It would contribute to an increasingly substantial problem during the latter half of the '80s through to 1990 when MarAd's forecast ends. Thus, it seems inevitable that the Bay Ridge will ultimately have a detrimental effect on the domestic trade.

MarAd countered this obvious point in its report by pointing to the "unusually high breakeven level" which makes it the "marginal vessel" in the trade. "Thus, this vessel will be the first to become unemployed as the Alaskan tanker market softens."¹⁰¹ 

The problem with this argument is that it is belied by the discussion, just six pages earlier, of the financial viability of the Bay Ridge in the Alaskan trade. The Bay Ridge already had contracts for one year and, once the repayment application was granted, they were expected to be renewed for five years. "(I)t is considered highly probable that the contracts will be renewed at least until approximately 1986 when the ALPETCO refinery in Alaska is due to go on line."¹⁰²  The ALPETCO refinery was then expected to provide employment for the Bay Ridge for the remainder of its economic life pursuant to an exclusive shipping rights agreement with an operating partner of Seatrain.¹⁰³  As Richmond's repayment application notes "(t)he combination of the above cargo requirements will lead to the employment of the Bay Ridge for the full 20-year period anticipated in this application for subsidy repayment."¹⁰⁴  Again, it seems inevitable that the Bay Ridge will ultimately have a detrimental effect on the domestic trade.

In its brief, Richmond has suggested that MarAd thought "the advantage of redressing an existing deficit may well outweigh the disadvantages of contributing to a potential future surplus."¹⁰⁵  But no such statement by MarAd is to be found anywhere in the record, and we cannot uphold an agency's decision based on speculation as to what the agency would have thought if it had addressed a problem it chose, rather, to ignore.

MarAd's letter announcing the Bay Ridge decision expressly incorporated a finding that "(t)he grant of Richmond's application will have no or, at the most, minimal adverse economic impact on the U.S.-flag ... domestic tanker fleet () ...."¹⁰⁶  We have pointed to two ways in which inconsistencies in the staff study allegedly adopted by MarAd as part of its decision undermine this finding. As a consequence, we can accept neither the finding nor the decision that stems from it.

The inconsistencies in the 22 October report confirm an initial suspicion, raised by MarAd's attempts to keep the report confidential and its refusal to allow any petitions for reconsideration of its decision, that more was involved here than a neutral assessment of how best to serve U.S. interests in a sound merchant marine. MarAd was obviously concerned with its own precarious financial position and the inevitable embarrassment consequent upon a public disclosure of that position.

Total government exposure on the Bay Ridge alone, over three-fourths of which is attributable to MarAd, was $87 million.¹⁰⁷  That sum marks only the beginning, however, of MarAd's interest, at the time, in keeping Seatrain solvent.¹⁰⁸  As noted in the 20 October report,

Seatrain, the parent of Richmond Tankers, Inc. (the owner of the BAY RIDGE), has guarantees covering another $273,776,000 in outstanding loans, all of which are further guaranteed by the U.S. government. Foreclosure on the BAY RIDGE, if it is unable to operate profitably and service its debt, could force Seatrain in default on obligations to its other subsidiaries. The overall effect could be to call all of the U.S. guarantees which, including the BAY RIDGE, total $354,963,000. This "domino" or "cascading" effect would be softened only by the resale of all repossessed vessels, which would probably result in only recovering a small fraction of the guaranteed amounts. Thus, the failure of the BAY RIDGE to operate and repay debt has a much higher potential loss to the Government attached to it than just the vessel's own outstanding debt.¹⁰⁹ 

Thus, MarAd was placed in an extremely embarrassing position. Having continuously financed unnecessary ships for an insolvent company, it was faced by losses that could not but outrage Congress and the public at large. The problem was exacerbated by the leveraged lease transactions allowed by MarAd that enabled financially sound companies like Security Equipment Leasing, Inc. and American Road Equity Corp. to dip their fingers in the subsidized pie without bearing any of the ultimate financial risk and responsibility.¹¹⁰ 

We do not say that MarAd cannot take account of potential government losses in passing upon a repayment application. The Supreme Court has already said that this is a relevant consideration.¹¹¹  But MarAd must be above board in its deliberations. If MarAd decides that its budgetary concerns outweigh any harm the Bay Ridge might cause the domestic trade, it must forthrightly acknowledge that decision instead of offering specious explanations for how the Bay Ridge will be readily absorbed. Moreover, MarAd cannot gloss over its financial predicament in its published decision with a single sentence, noting that "(t) he financial interest of the Government ... is substantial, and warrants approval of Richmond's application."¹¹²  Agency candor is crucial to the accountability that the APA is designed to ensure. Only thus can the court pursue its function of ensuring MarAd stays within its statutory mandate. Only thus can Congress and voters generally decide whether MarAd serves, on the whole, a useful or an inimical function.

The procedural and substantive failings discussed above are such that we must ask MarAd to reconsider its decision that repayment of the CDS on the Bay Ridge should be permitted with consequent lifting of the domestic trading restriction. MarAd should permit a new round of comments upon the application and publish a thorough and reasoned explanation for whatever decision is reached.

We wish to stress that nothing in our opinion should be read to dictate the outcome of this new proceeding. Our holding that there were substantive flaws in MarAd's decision to grant Richmond's application is not a judgment as to the merits of that application. We have merely found that an adequate explanation for granting the application was not provided. We have not found that an adequate explanation could not be provided. In particular, we have not found that inclusion of the Bay Ridge would have an adverse impact on the domestic trade. We have merely found that MarAd's discussion of the issue in its 22 October report was marred by inconsistencies.

The point is worth stressing. There is always the danger that a court, seriously pursuing its responsibility for review but ignorant of the technical side of a regulated trade, will be convinced by zealous litigants that there are substantive flaws in an agency's decision when, in reality, there is merely an unfortunate lack of clarity. Under such circumstances, the agency should stick to its guns, make the same decision again, observing all procedural requirements, and endeavor to substantiate it in the clearest, most coherent manner possible.

Three further points must be made. First, in fairness to Richmond, MarAd must take into account factors prevailing in the industry at the time of the previous decision, that is, as of 13 November 1980. At that time, everyone acknowledges, there was a severe shortage of large tankers in the domestic trade that has apparently persisted to the present day.¹¹³  Given that the Bay Ridge could remain operational into the next century, a tanker deficit from 1980 to 1982 is not a conclusive factor, but it is certainly a factor to be considered.

Second, it remains within MarAd's discretion whether or not the new Bay Ridge decision should await publication of a permanent rule governing repayment applications.

Finally, we think that the status quo should be maintained during the pendency of these proceedings. The Bay Ridge need not cease operations in the domestic trade. Richmond need not breach its shipping contracts made in reliance upon MarAd's decision. The complicated Title XI financing that enabled repayment of the CDS need not be undone. Given a persistent shortage of large tankers in the domestic trade, the Bay Ridge will do no harm during the brief period necessary for MarAd to make its decision.

We thus remand the case to the district court with instructions to enter an order in accordance with this opinion.

So Ordered.