Opinion for the Court filed by Circuit Judge SILBERMAN.
Dissenting opinion filed by Circuit Judge RUTH BADER GINSBURG.
SILBERMAN, Circuit Judge:
This case presents a challenge to a decision and order of the Attorney General, pursuant to the Newspaper Preservation Act (“NPA”), 15 U.S.C. §§ 1801-1804 (1982), approving a joint operating arrangement between the Detroit Free Press and Detroit News newspapers. Appellants, which include Michigan Citizens For An [1287]*1287Independent Press,1 seven individuals,2 and the interest group Public Citizen, brought suit against the Attorney General and the two newspapers in the district court alleging that the Attorney General’s decision violates the NPA and the Administrative Procedure Act, 5 U.S.C. § 706 (1982), because it is not based on substantial evidence, is arbitrary and capricious, and is otherwise in violation of law. The district court granted summary judgment in favor of defendants, 695 F.Supp. 1216, and plaintiffs appealed to this court. We conclude that the Attorney General’s decision was based on a permissible construction of the statute, and that his application of the legal standard to the facts of this case was not arbitrary, capricious, or an abuse of discretion. We therefore affirm the judgment of the district court.
I.
A.
Congress passed the Newspaper Preservation Act in 1970 with the stated purpose of “maintaining a newspaper press editorially and reportorially independent and competitive in all parts of the United States.” 15 U.S.C. § 1801. The Act creates an exemption to the antitrust laws that permits a joint newspaper operating arrangement (“JOA”)3 between two newspapers if the Attorney General determines that one of the papers is “a failing newspaper” and that the arrangement will “effectuate the policy and purpose” of the Act. 15 U.S.C. § 1803(b).4 A “failing newspaper” is defined as “a newspaper publication which, regardless of its ownership or affiliations, is in probable danger of financial failure.” 15 U.S.C. § 1802(5).
The first joint newspaper operating arrangement was started by three newspapers in Albuquerque, New Mexico in 1933, and by 1966 there were twenty-two JOAs in effect. In 1964, the Department of Justice initiated an investigation of newspaper JOAs, and in 1965 it sued the publishers of two daily newspapers in Tucson, Arizona, which operated jointly, for violations of sections 1 and 2 of the Sherman Act, 15 U.S.C. §§ 1, 2 and section 7 of the Clayton Act, 15 U.S.C. § 18. The trial court in that suit found violations of all of those provisions, and the Supreme Court upheld that finding in Citizen Publishing Co. v. United States, 394 U.S. 131, 89 S.Ct. 927, 22 L.Ed.2d 148 (1969).
The Court rejected the newspapers’ argument that the so-called “failing company” defense — a judicially created doctrine — absolved them from liability under the antitrust laws. Id. at 137-38, 89 S.Ct. at 930. Under the “failing company” doctrine, conduct which would otherwise violate antitrust laws does not do so if one of the suspect businesses “faced the grave proba[1288]*1288bility of business failure.” International Shoe Co. v. FTC, 280 U.S. 291, 302, 50 S.Ct. 89, 93, 74 L.Ed. 431 (1930). The doctrine is based on the notion that a merger between two competitors, one of which is failing, cannot have an adverse effect on competition, because the failing company would disappear as a competitive factor whether or not the merger occurred.
In Citizen Publishing, the Court narrowly confined the scope of the doctrine. It held that a financially troubled company may not employ the “failing company” defense unless it meets three conditions. The disputed merger may be sought only when the owners of the “failing” company are contemplating liquidation; indeed, the JOA must be the “last straw”. at which the company can grasp. 394 U.S. at 137, 89 S.Ct. at 930. The defendants are required to establish that the company that acquires the failing company is “the only available purchaser,” id. at 138, 89 S.Ct. at 931, and finally, the prospects for successful reorganization under the bankruptcy laws must be “dim or nonexistent.” Id. Because the Tucson papers did not make such a showing, their JOA violated the Sherman Act.
Congress reacted to Citizen Publishing by passing the Newspaper Preservation Act, which established a less stringent test for newspapers seeking a JOA. S. Rep. No. 535, 91st Cong., 1st Sess. 4 (1969). Congress did not question the Court's reasoning in defining the failing company doctrine, but it felt that “the economics of the newspaper industry make it more likely for newspapers to fail when faced with competition than other businesses.” Id. As the Senate Judiciary Committee noted, “when a newspaper is failing it is harder to reverse the process and it is almost impossible to find an outside buyer.” Id. The NPA therefore provides that a newspaper is “failing” and eligible for a JOA when it is “in probable danger of financial failure.” 15 U.S.C. § 1802(5). The statute also in-eludes what Congress meant to be a less strict standard for JO As already existing in 1970. H.R. Rep. No. 1193, 91st Cong., 2d Sess. 10 (1970), U.S.Code Cong. & Admin. News 1970, p. 3547. A pre-statute JOA is not unlawful if at the time at which such arrangement was entered into, not more than one of the newspapers involved was “likely to remain or become a financially sound publication.” 15 U.S.C. § 1803(a).
Since 1970, four new JOAs have been approved and implemented.5 In each of those cases, unlike the Detroit case, the “failing newspaper” was well into what in the newspaper industry is known as the “downward spiral.” The fate of a struggling newspaper is thought to be determined by the close interrelationship between circulation and advertising revenues. Once a paper loses circulation, advertisers are less likely to purchase space in the paper. Readers, in turn, are less likely to buy a paper that is short on advertising, so circulation drops further. The result of this interrelationship is an apparently irreversible downward plunge that ends in business failure. The only court to address the Act, Committee for an Independent P-I v. Hearst Corp., 704 F.2d 467 (9th Cir.), cert. denied, 464 U.S. 892, 104 S.Ct. 236, 78 L.Ed.2d 228 (1983), concluded that a newspaper in the downward spiral satisfies the “probable danger of financial failure” test, as long as it had followed reasonable management practices. Id. at 479.
B.
The Detroit Free Press and the Detroit News are daily newspapers that compete in Detroit, which is the nation’s fifth largest newspaper market. The papers are owned by the two largest news organizations in the United States; Knight-Ridder, Inc. owns the Free Press, and the Gannett Company has controlled the News since February 1986, when it purchased the paper from the Evening News Association. Over the [1289]*1289past fifteen years, the papers have been engaged in fierce competition for absolute dominance of the Detroit market, which was motivated, at least initially, by the knowledge that many junior papers have been unable to survive as the second paper in metropolitan area competition.
This bitter fight has led to large operational losses by both papers. The Free Press has lost money every year since 1979, and it lost over $10 million per year from 1981 to 1986. The News has sustained operational losses since 1980, and it lost over $50 million between 1981 and 1986. A circulation price war has driven the daily prices in Detroit to twenty cents for the News and fifteen cents for the Free Press — probably the lowest daily prices in the United States. In recent years, the News has maintained a consistent circulation lead of approximately 51% to 49%. Perhaps more important, the News has continuously maintained more than a 60% share of total full-run advertising linage.
As a result of their losses, the papers began to consider the alternative of a JOA as early as 1980 when the chief executive officers of Knight-Ridder and the Evening News Association first discussed the possibility. Negotiations continued sporadically from January 1981 to January 1984, but no agreement was reached during that period. In August 1985, Gannett agreed in principle to purchase the News from the Evening News Association, and senior officials of Gannett and Knight-Ridder thereafter met 16 times between August 1985 and April 1986 to shape the final agreement. On April 11,1986 the News and the Free Press executed the JOA.
The agreement, which has an initial term of 100 years, provides — as is typical — that the news and editorial staffs of the two papers are to remain independent and insulated from influence by the other party to the arrangement. The Free Press would publish a morning paper on Monday through Friday, and the News would print a corresponding afternoon edition. On Saturday and Sunday, the parties would publish only one paper, with each paper assuming separate editorial and news responsibilities.
During the first three years of the JOA, the News would receive 55% of the profits of the combined enterprise, while the Free Press would receive 45%. In the fourth and fifth years, the profit split would reduce to 53%/47% and 51%/49%, respectively. Beginning in the sixth year, the profits and losses would be shared equally by the News and the Free Press.
On May 9, 1986, the two papers applied for approval of the JOA by the Attorney General as required by the Act, and the application was referred to the Assistant Attorney General in charge of the Antitrust Division, pursuant to Justice Department regulations. See 28 C.F.R. § 48.7 (1988). The then Assistant Attorney General, Douglas H. Ginsburg (now Judge Ginsburg), issued a report on July 23,1986, concluding that the applicants had “not yet sustained their burden of proof of showing that the Detroit Free Press is a ‘failing newspaper’ within the meaning of the Act and that approval of the application would effectuate the policy and purpose of the Act.” However, he did not advise disapproval of the application; instead, he recommended that the Attorney General order that a hearing be held before an administrative law judge to resolve material issues of fact raised by the application. See 28 C.F.R. § 48.7(b)(2) (1988).
Attorney General Edwin Meese followed his Assistant Attorney General’s advice and pursuant to the Justice Department’s regulations, 28 C.F.R. § 48.10, an administrative law judge was appointed to conduct the hearing. On December 29, 1987, ten months later, the AU issued a decision, recommending that the application be denied. He concluded, inter alia, that the applicants had failed to prove that there exists in Detroit an irreversible market condition that will probably lead to the failure of the Free Press. According to the AU, the Free Press is not dominated by the News and the Free Press is not in a downward spiral toward failure. He instead attributed the losses incurred by the Free Press and the News to “their strategies of [1290]*1290seeking market dominance and future profitability at any cost along with the expectation that failure to achieve these goals would result in favorable consideration of a JOA application.”
The AU did not deny that the fiercely competitive strategies employed by both papers were perfectly rational, given the disastrous history of junior papers in the United States. But he believed that each paper had an eye on a potential JOA application in the event that it turned out to be the loser. They both saw the JOA as a safety net, in other words, and were thereby encouraged to engage in particularly risky competitive acrobatics.
Indisputably, the News was leading the Free Press in most of the circulation, revenue, and advertising linage figures used to measure the relative positions of rival newspapers. The AU maintained, however, that the Free Press was within “striking distance” of the total circulation lead, and that the News’ advertising lead was “vulnerable” to a change in the circulation lead. Neither paper could achieve profitability as long as they both pursued the current price war, but he believed that the record did not support the conclusion that “reader and advertiser demand in Detroit is so inadequate that the market cannot sustain two profitable papers irrespective of changes in pricing policies.” He hypothesized that Detroit could sustain two profitable papers if the Free Press and the News both raised circulation and advertising prices. Still, he recognized that since “neither the Free Press nor the News can raise circulation or advertising prices without regard to what the other paper does, there is no completely unilateral course of action which either paper can pursue which would return it to profitability.”
The AU accepted the testimony of the Antitrust Division’s expert, who calculated that the 50/50 profit split (after five years) represented a perception by Gannett that it could not achieve domination of the market for “at least seven years.” He was, moreover, unpersuaded by Gannett officials’ testimony that if the JOA were denied, they would not raise circulation prices, and he found “contemporaneous evidence” indicating that “absent a JOA Gannett may eventually initiate circulation price increases.” He also assigned “little weight” to testimony from Knight-Ridder’s CEO that he would close down the Free Press if the JOA application were denied. Essentially, the AU predicted that if the JOA were denied, the News would give the Free Press sorely needed relief by raising the News’ circulation and advertising prices, thereby allowing the Free Press to follow suit.
Attorney General Meese, in his opinion of August 8, 1988, disagreed with the conclusions of the AU and granted approval of the JOA. The Attorney General “accepted as accurate the fact findings of the Administrative Law Judge,” but differed with his “ultimate conclusion as to where those facts lead.” Adopting the legal standard enunciated by the Ninth Circuit in Hearst the Attorney General asked: “Is the newspaper suffering losses which more than likely cannot be reversed?” Committee for an Independent P-I v. Hearst, 704 F.2d 467, 478 (9th Cir.), cert. denied, 464 U.S. 892, 104 S.Ct. 236, 78 L.Ed.2d 228 (1983). The Attorney General decided that the answer was yes: the Free Press had met its burden of proof, because it had suffered persistent operating losses over nearly a decade and had no prospect of unilateral action to reverse those losses.
Central to the Attorney General’s opinion was his disagreement with the AU’s prediction of future behavior by the two papers in the event that the JOA is denied. He noted and accepted the AU’s finding that the News was in a stronger competitive position according to all major economic indices. But he determined — or more accurately predicted — that if the News continues its current pricing practices, it “undoubtedly has the ability on such terms to outlast the Free Press.” Given this premise, the Attorney General found persuasive Gannett’s testimony that it would continue its current competitive policies (and not raise prices), which he said “hardly reflects unsound business judgment.” Similarly, the Attorney General felt that the testimony of Knight-Ridder’s CEO concerning a [1291]*1291possible closure of the Free Press “cannot be wholly disregarded,” because it would be “neither counterintuitive nor contradictory” for Knight-Ridder to discontinue the paper if it concluded that it could not outlast the News in a prolonged price war.
In response to the allegation that the papers had pursued their competitive strategies because of the potential of a JOA, the Attorney General read the record as showing that Knight-Ridder was not “principally pursuing any end other than market domination.” (emphasis added). Moreover, he noted that “newspapers cannot be faulted for considering and acting upon an alternative that Congress had created.”
II.
Appellants allege that the Attorney General’s determination is invalid both because it is based on an impermissible interpretation of the statute and is arbitrary or capricious.6 As is not unusual in appeals from agency actions, the claims are interrelated. At the core of appellants’ case is the assertion that the Attorney General could not legally grant approval for a JOA because the Detroit Free Press was not in a tough enough spot to qualify as “in probable danger of financial failure.” Whether the Attorney General legally decided that the Free Press did meet the statutory standard in turn depends to a large extent on whether his prediction of the newspapers’ future course (if he did not approve the JOA) was reasonable. The Attorney General’s interpretation of the probable danger of financial failure test draws content from the factual showing that he requires to meet that test. See INS v. Cardoza-Fonseca, 480 U.S. 421, 107 S.Ct. 1207, 1221, 94 L.Ed.2d 434 (1987) (ambiguous statutory terms “can only be given concrete meaning through a process of case-by-case adjudication”). And there is no question in our mind that if the Attorney General’s statutory interpretation is reasonable, it is entitled to deference under Chevron U.S.A. Inc. v. NRDC, 467 U.S. 837, 842-43, 104 S.Ct. 2778, 2781-82, 81 L.Ed.2d 694 (1984), because we are certainly unable to discern a specific congressional intent governing this case.
Not surprisingly, the exact meaning of the linguistically imprecise phrase “probable danger of financial failure” is not apparent from the statute or the legislative history. The Senate bill, which differed slightly from the House version, “defined” a failing newspaper as one “in danger of probable failure.” The Senate took this phrase — or at least the words “probable failure” — from the Bank Merger Act, 12 U.S.C. § 1828(c)(3) (1982), recognizing that the phrase was informed by the Supreme Court’s decision in United States v. Third National Bank, 390 U.S. 171, 88 S.Ct. 882, 19 L.Ed.2d 1015 (1968). S.Rep. No. 535, 91st Cong., 1st Sess. 2 (1969). The House version, which was eventually adopted, used the clause “probable danger of financial failure” as the standard for new JOAs (emphasis added). Despite its minor difference from the Senate version, the chief House sponsor of the bill explained that “[t]he term ‘probable danger of financial failure’ ... comes out of the Bank Merger Act,” and “is understood by the courts in the field.” 116 Cong.Rec. 23,146 (1970) (statement of Rep. Kastenmeier). The Supreme Court had held in Third National Bank that where managerial deficiencies are responsible for a bank’s financial prob[1292]*1292lems, the banks, in order to prove “probable failure,” must establish that improved management could not achieve profitability. 390 U.S. at 190, 88 S.Ct. at 893. More generally, the parties seeking the merger must “reliably establish the unavailability of alternative solutions” to the financial woes of the failing bank. Id. Thus, strong evidence of probable failure was required.
To be sure, the Attorney General had not previously faced a case such as this. Prior approvals of JOAs had always involved at least one newspaper that had actually entered the downward spiral, whereas the Detroit Free Press could be said to be poised on the brink of the spiral, its future dependent on the competitive behavior of the News.7 Still, the only prior case reviewing an Attorney General’s approval of a JOA — the pr e-Chevron decision of the Ninth Circuit in Hearst — phrased the question before the Attorney General in broader terms than whether one of the newspapers had entered a downward spiral. The court asked: “Is the newspaper suffering losses which more than likely cannot be reversed?” This interpretation of the statutory language, which the court called a “commonsense construction,” id. at 478, was explicitly adopted by the Attorney General in this case, and thus made his own interpretation entitled to Chevron deference. Only for cogent reasons would we reject as unreasonable an interpretation of a statute that a sister circuit had considered a commonsense construction.
The Ninth Circuit thought implicit in its inquiry was an examination of alternative forms of relief for the putatively failing newspaper. Was there, for example, a group of interested buyers or a potential for improved management? Congress’ reference to the Third National Bank case in the legislative history of the statute suggested to the Ninth Circuit that Congress intended the Attorney General to consider alternatives to a JOA before approving an application. We quite agree, but so apparently did the Attorney General. He concluded that if no form of relief was within the control of the sick newspaper — its survival depended only on improbable behavior by its competitor — the statutory test was satisfied. Appellants artificially construe the Attorney General’s decision to permit a JOA without regard to consideration of the competitors’ behavior, but that is not what the Attorney General said.
The dissent suggests that the Attorney General’s statutory construction is impermissible because it did not employ the interpretative canon that exemptions to the antitrust laws — like all exemptions — should be construed narrowly. Group Life & Health Ins. Co. v. Royal Drug Co., 440 U.S. 205, 231, 99 S.Ct. 1067, 1083, 59 L.Ed.2d 261 (1979). Certainly, courts interpret ing the antitrust statutes have often employed that canon. See, e.g., Hearst, 704 F.2d at 473 (“Our review of the Newspaper Preservation Act and its interpretation by the Attorney General is guided by [the] additional rule ... [that] exemptions to the antitrust laws are to be narrowly construed.”).8 But Chevron implicitly precludes courts picking and choosing among various canons of statutory construction to reject reasonable agency interpretations of ambiguous statutes. If a statute is ambiguous, a reviewing court cannot reverse an agency decision merely because it failed to rely on any one of a number of canons of construction that might have shaded the interpretation a few degrees in one direction or another.
We do not mean to say that canons of construction are completely irrelevant in the post-CTiemm era. If employment of an accepted canon of construction illustrates that Congress had a specific intent [1293]*1293on the issue in question, then the case can be disposed of under the first prong of Chevron. See INS v. Cardoza-Fonseca, 480 U.S. 421, 448-49, 107 S.Ct. 1207, 1222, 94 L.Ed.2d 434 (1987) (“ordinary canons of statutory construction” provided “compelling” evidence that Congress intended “well-founded fear or persecution” standard to be different from “clear probability of persecution.”). For example, if Congress banned the importation of apples, oranges, and bananas from a particular country, the canon of expresio unius est exclusio alterius might well indicate that Congress did not intend to ban the importation of grapefruits. In that event, an agency decision to ban grapefruits would be contrary to Congress’ specific intent. As a corollary, we held in American Fed’n of Gov’t Employees v. FLRA, 798 F.2d 1525, 1528 (D.C.Cir.1986), that the FLRA’s interpretation of the Federal Service Labor-Management Relations Statute was impermissible because it “ignore[d] the familiar canon that statutes should be construed ‘to give effect, if possible, to every word Congress used.’ ” The FLRA’s construction had resulted in “an effective repeal” of part of the statute, id. at 1529, and thereby frustrated the intent of Congress.
In this type of case by contrast, the Attorney General is called upon to balance two legislative policies in tension: The pro-consumer direction of the antitrust laws and a congressional desire embodied in the Newspaper Preservation Act that diverse editorial voices be preserved despite the unique economics of the newspaper industry. This is precisely the paradigm situation Chevron addressed. If the agency’s choice “represents a reasonable accommodation of conflicting policies that were committed to the agency’s care by the statute, we should not disturb it unless it appears from the statute or its legislative history that the accommodation is not one that Congress would have sanctioned.” Chevron, 467 U.S. at 845, 104 S.Ct. at 2783 (quoting United States v. Shimer, 367 U.S. 374, 383, 81 S.Ct. 1554, 1560, 6 L.Ed.2d 908 (1961)). To invoke the normal canon of construction is merely to say that the Attorney General put too much weight on the policy of preserving editorial diversity. We are not now after Chevron — if we ever were — permitted to accept such an argument.
Appellants argue that the Attorney General should receive less deference than Chevron requires, because “his interpretation of the statute was different from that of the Antitrust Division, where the Justice Department’s expertise on the Newspaper Preservation Act resides.” We have previously rejected the notion that Chevron deference is based solely on agency expertise. Public Citizen v. Burke, 843 F.2d 1473, 1477 (D.C.Cir.1988); Cablevision Systems Dev. Co. v. Motion Picture Ass’n of America, 836 F.2d 599, 608-09 (D.C.Cir.1988). The rationale of Chevron is also grounded in the principle that the political branches of government, rather than the judiciary, should make policy choices. Chevron, 467 U.S. at 865-66, 104 S.Ct. at 2793.
Nevertheless, our dissenting colleague seems to accept the argument, since she relies heavily on the Antitrust Division’s brief submitted to the AU. Dissent at 1298 n. 3, 1299 n. 5, 1299 n. 6. It is not surprising that the Division — charged with the front line responsibility for enforcing the antitrust laws — sought a narrow interpretation of sections 1802(5) and 1803(b). Congress, however, did not place responsibility for reconciling the conflicting policies and values called for in this type of case upon the Antitrust Division, but rather on the Attorney General, who might be thought to have a broader perspective.
It is also suggested, dissent at 1299, that the Attorney General’s decision is legally defective because he did not explain his interpretation of the conceptual difference between the section of the statute that applied to this transaction and the section that did not apply. The latter, section 1803(a), governs the legality of presta-tute JOAs and is more lenient because a JOA is authorized if, when entered into, not more than one of the newspapers involved was “likely to remain or become a financially sound publication.” We do not understand our dissenting colleague to argue [1294]*1294that the Attorney General’s interpretation of section 1803(b) necessarily overlaps section 1803(a). A financially sound publication might, for instance, be defined as one that is consistently profitable, indeed, profitable enough to recover its cost of capital. That is worlds away from a newspaper that is “suffering losses which more than likely cannot be reversed.” The dissent contends, instead, that the Attorney General was obliged to draw the exact boundaries between the two sections when applying only the one. That seems to us to require an agency to decide cases not before it, to offer dicta that we normally eschew. We know of no authority that would support such a holding.
III.
Even if the Attorney General is statutorily authorized to treat a newspaper as in probable danger of failing before it actually enters the downward spiral, appellants claim that the Attorney General’s decision was arbitrary and capricious because not rationally connected to the facts before him. Bowman Transp., Inc. v. Arkansas-Best Freight System, Inc., 419 U.S. 281, 285, 95 S.Ct. 438, 442, 42 L.Ed.2d 447 (1974) (quoting Burlington Truck Lines v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 246, 9 L.Ed.2d 207 (1962)) (agency must articulate a “rational connection between the facts found and the choice made”). This, as we have noted, is in essence a challenge to the reasonableness of the Attorney General’s determination that the News had the economic power to outlast the Free Press and was not likely to reduce competitive pressure by raising prices.
The only specific challenge, as far as we can determine, to the Attorney General's appraisal of the respective competitive strengths of the two newspapers is based on the different opinion of che AU (and the Antitrust Division’s brief to the AU). It is true that the Attorney General’s crucial conclusions that the Free Press “has no realistic prospect of outlasting the News given the latter’s substantial advertising and persistent circulation lead” and that the News “undoubtedly has the ability ... to outlast the Free Press” was predicated on the AU’s findings recounting the News’ lead in all major indices. It is also true that the AU went on to offer a somewhat different conclusion: that the Free Press was still within “striking distance” of the News and the latter’s lead was “vulnerable.” The Attorney General would not, however, be legally obliged to conform his judgment to that of a statutorily-required AU,9 much less this one, who was employed as a matter of discretion rather than law. See 28 C.F.R. § 48.8 (1988). Both men relied on the very same facts to make different evaluations of the competitive strength of the Free Press. But, it is only the Attorney General’s conclusions that have legal significance, and we cannot say that his determination is unreasonable. It is undisputed, after all, that the News has maintained the lead for a long time and that the Free Press had suffered extensive losses. Debatable, the Attorney General’s appraisal may well be, but hardly unreasonable.
Similarly, appellants rely on the AU’s contrary prediction to dispute the Attorney General’s conclusion that the News would not release the pressure on the Free Press by raising prices if the JOA were disapproved. Gannett officials testified that they had no intention of raising prices regardless of the Attorney General’s decision. The AU refused to credit this testimony, not on account of the witnesses’ demeanor, but because he, the AU, thought that course would only cause more losses for the News and was therefore irrational.10 Cf. Universal Camera Corp. [1295]*1295v. NLRB, 340 U.S. 474, 494-95, 71 S.Ct. 456, 468, 95 L.Ed. 456 (1951) (“The significance of [the AU’s] report ... depends largely on the importance of credibility in the particular case.”). The Attorney General’s judgment of the News’ likely future behavior was premised on his determination, which we have already found reasonable, that the News had the competitive strength to outlast the Free Press.11 The ALJ never squarely found otherwise,12 and if the News had such strength, we do not see how the Attorney General’s projection can be deemed unreasonable. Under those circumstances, Gannett’s refusal to raise prices, as the Attorney General said, “hardly reflects unsound business judgment.”
Knight-Ridder seems to have thought that its competitor’s strategy was rational, since its CEO testified that if the JOA were disapproved, the Free Press would close down. The AU again assigned “little weight” to that testimony, because “if a Free Press closure was imminent, it would have made no economic sense for [the News] to agree to share prospective JOA profits with [the Free Press].” Also, he stressed that an anti-trust division expert had testified that the terms of the profit split (reaching 50/50 after five years) suggested that Gannett thought the Free Press would remain in existence for seven to ten years. The difficulty with the AU’s analysis is that he equated the parties’ bargaining positions prior to submission of a proposed JOA with their strategies after a rejection by the Attorney General. Prior to agreement, the Free Press had every incentive to convince the News that it would compete fiercely, and indefinitely into the future: Damn the losses; full speed ahead. Only by emphasizing its potential longevity could the Free Press extract favorable terms in a JOA. Thus, the Attorney General described the situation as the parties entered into the JOA, as a “competitive stalemate” with market domination “no longer within the grasp of either paper.” But, that the News was willing to negotiate peace terms (the JOA) does not belie the Attorney General’s appraisal of its fundamental superior strength. After this proceeding, in which all cards are placed on the table, it is wholly unrealistic [1296]*1296to assume that the parties will return to the game and play it as if this proceeding had never occurred. If a JOA were denied, the News would have every incentive to force the Free Press to the wall; even if it took seven years to win, at the end of that period the News would have a monopoly. And, if the Free Press were doomed to defeat in the long run, it was not only reasonable but optimal for the paper to close immediately.
The Attorney General had to consider which of the two papers would blink first in the event the JOA were denied. Gannett said that it would not, and Knight-Ridder said that it would. Appellants assert that it was unreasonable for the Attorney General to believe them, because the more likely event was the exact reverse: that if the JOA were denied, Gannett would raise prices and the Free Press would remain in business. The Attorney General, it would seem, did not want to play the high stakes regulatory game that his AU proposed. He obviously was concerned that if he gambled on the AU’s prediction that both newspapers were bluffing, Detroit would lose a newspaper. That is not to say that the Attorney General should put undue stress on self-serving declarations by newspaper executives seeking a JOA. But here, the statements that the Attorney General credited follow a long period of bitter competition. For the News to stay the course, as for the British and French in 1917, promised absolute victory.
It may well be; as appellants argue and the AU found, that under ideal circumstances, Detroit could support two newspapers. The same could also be true of many cities that have lost competing newspapers and are now one newspaper monopoly towns. It is not at all clear whether the newspaper business in some cities is a natural monopoly, and, if so, in cities of what size. This sort of speculation, it seems to us, as it did to the Attorney General, is hardly conclusive. That an omniscient Detroit newspaper czar could set circulation and advertising prices that would permit both papers to return to profitable status is not a useful observation in this context. The Attorney General is required to determine what will actually happen in Detroit if his approval is withheld. It would, moreover, be anomalous for those responsible for enforcing the antitrust laws to try to guide and calibrate the competitive zeal of the two newspapers so as to reach that level of competition at which both newspapers could be profitable.
Appellants might also be understood to complain that the Attorney General did not provide a reasoned explanation for his decision, because his only citations to the record at certain crucial points were to portions of the AU’s opinion that reached different conclusions based on the same facts. Of course, the decision of the AU is part of the record and must be considered by the court when it determines whether the Attorney General’s ruling is supported by substantial evidence or, in this case, arbitrary or capricious. Universal Camera Corp. v. NLRB, 340 U.S. 474, 493, 71 S.Ct. 456, 467, 95 L.Ed. 456 (1951). We have said that an agency must both express an awareness that it is disagreeing with an AU and set forth the basis of the disagreement. Local 441, IBEW v. NLRB, 510 F.2d 1274, 1276 (D.C.Cir.1975); Greater Boston Television Corp. v. FCC, 444 F.2d 841, 853 (D.C.Cir.1970), cert. denied, 403 U.S. 923, 91 S.Ct. 2233, 29 L.Ed.2d 701 (1971). To reverse the Attorney General, however, for failure to state at the exact point of the citations the obvious nature of his disagreement with the AU would be excessive judicial nitpicking. His difference with the AU is clear throughout the opinion, and although “[t]he explanation may have been curt, ... it surely indicated the determinative reason for the final action taken.” Camp v. Pitts, 411 U.S. 138, 143, 93 S.Ct. 1241, 1244, 36 L.Ed.2d 106 (1973).
* Of # # * *
The real difficulty with this case — the factor that quite plainly underlies the AU’s discomfort as well as appellants’ quarrel with the Attorney General’s decision — is the effect that the prospect of a JOA has on the behavior of competing newspapers. See also dissent at 1299. It [1297]*1297is feared that the statute authorizing a JOA creates a self-fulfilling prophecy. Newspapers in two newspaper towns will compete recklessly because of a recognition that the loser will be assured a soft landing.
Appellants argue that the Attorney General inadequately considered whether or not “critical aspects of the newspapers’ conduct were influenced by the prospect of obtaining a JOA.” But his opinion addressed this “dual motive” concern at some length; he observed that this was not the classic case that had worried Congress, where a newspaper had “brought itself to the brink of financial failure through improper marketing practices or culpable .management.” Instead, the record of years of fierce competitive and consequent losses to both papers led the Attorney General reasonably to conclude that both papers were principally pursuing market domination and that their strategies had been followed before any mutual discussion of a JOA. Nevertheless, the Attorney General implicitly recognized that it would be impossible completely to preclude competing newspapers from factoring into their business strategy the prospect of a JOA. As he laconically put it, “newspapers cannot be faulted for considering and acting upon an alternative that Congress has created.” 13
We can envision a perfectly rational different policy, one that would require a showing that the weaker paper was more bloodied before approving a JOA and therefore might discourage the sort of competition we saw in Detroit. Congress, however, delegated to the Attorney General, not to us, the delicate and troubling responsibility of putting content into the ambiguous phrase “probable danger of financial failure.” We cannot therefore say that his interpretation of that phrase as applied to this case, with all of its obvious policy implications, was unreasonable. The judgment of the district court therefore is
AFFIRMED.