City of Milwaukee, Plaintiffs-Appellees/cross-Appellants v. Clayton K. Yeutter, Secretary of Agriculture, Defendants-Appellants/cross-Appellees

*541 EASTERBROOK, Circuit Judge. Several federal programs provide free or subsidized food for nations in need. American vessels get a preference in transporting the food: [T]he appropriate agency or agencies shall take such steps as may be necessary and practicable to assure that at least [75] per centum of the gross tonnage ... which may be transported on ocean vessels shall be transported on privately owned United States-flag commercial vessels, to the extent such vessels are available at fair and reasonable rates for United States-flag commercial vessels, in such manner as will insure a fair and reasonable participation of United States-flag commercial vessels in such cargoes by geographic areas.... 46 U.S.C.App. § 1241(b)(1), as modified by 46 U.S.C.App. § 1241f(a)(l). Our case involves the application of this rule to food bought and shipped by the federal government under Title II of the Agriculture Trade Development and Assistance Act of 1954, 7 U.S.C. §§ 1721-1726, known popularly as the Food for Peace program and within the shipping business as the Title II or P.L. 480 program. Much of the food suitable for this program is grown in or near the midwest and could be exported from ports on the Great Lakes. Modern U.S.-flag ships suitable for this trade can’t squeeze through the Welland Canal, however, and therefore cannot operate above Lake Ontario. Foreign-flag vessels accordingly dominate ocean commerce at upper Great Lakes ports. The cities and labor unions who filed this suit want the government to export more commodities on foreign vessels through the upper Great Lakes. The district court originally dismissed the suit for want of standing, 634 F.Supp. 760 (E.D.Wis.1986), but we directed it to decide the case on the merits, 823 F.2d 1158 (7th Cir.1987). Any rule using a percentage requires definition of the denominator. Seventy-five percent of which shipments, from which ports, over what period of time? The federal agencies responsible for buying and shipping food under the P.L. 480 program — the Agency for International Development (AID) and the Commodity Credit Corporation (CCC), operating under regulations promulgated by the Maritime Administration (MarAd) of the Department of Transportation — chose one year as the accounting period, without protest. They chose the entire nation as the geographic basis. Computing the percentage port-by-port, or by range of ports (Gulf Coast, East Coast, West Coast, and Great Lakes), could lead to substantial deviation from the 75% figure for the nation as a whole. Selecting the nation as the base means that Great Lakes ports cannot load more than 25% of the cargo, and as a practical matter they handle much less. The locks are frozen for four months; more, under a nationwide accounting system every shipment on a foreign-flag vessel from a coastal port reduces the maximum that can be exported via the Great Lakes. The City of Milwaukee and the other 17 plaintiffs (collectively Milwaukee) asked the district court to direct the federal agencies to compute percentages by port. They maintain that if grain appears in Milwaukee for export, it may go on a foreign-flag vessel because no U.S.-flag vessel is available at any price, let alone a “reasonable” one. If grain appears in Baltimore, 75% of it must go on domestic vessels available at prices reasonable “for United States-flag commercial vessels”. That the share of U.S.-flag vessels in total exports under Milwaukee's approach would be less than 75% is, as Milwaukee sees things, simply a consequence of the requirement that U.S.-flag vessels be available at reasonable prices where the cargo is. If U.S.-flag carriers want to carry midwestem grain, they have only to come and get it. ' But of course grain doesn’t just appear at dockside. Somebody sends it there. “Somebody” turns out to be the Kansas City Commodity Office (KCCO) of the Department of Agriculture’s Agricultural Stabilization and Conservation Service, which when informed by AID of the need (and appropriations) buys the food, gets it to a port, and pays for ocean transportation. Until 1987 the KCCO decided what to buy *542 and which port to use by a formula called “lowest landed cost”. 7 C.F.R. § 1496.5 (1985). If AID wanted to give 100 tons of corn to Pakistan, KCCO would take bids on a package of corn, inland transportation, and ocean transportation. Grain merchants such as Cargill would find the combination that produced the lowest cost as delivered (“landed”) in the port of destination. On a given shipment this might mean buying the corn in Illinois, shipping it by rail to Milwaukee, and using a foreign vessel out the Great Lakes, through the Panama Canal, thence to Pakistan. Or it could mean buying corn farther west, rail to the West Coast, and U.S.-flag vessels to Pakistan. This method would have minimized costs if there were no cargo preference. But if the bid assumed shipment on foreign vessels from a coastal port, U.S.-flag vessels might be available there at the same time. Transferring the shipment to satisfy the preference could increase the cost over what it would have been had the food been purchased elsewhere and shipped through a different port. Consistent application of the lowest landed cost method also could send so much food to ports with few U.S.flag vessels that the agencies could not meet the domestic preference — at least not if the preference were administered nationally. In August and September 1985 the KCCO diverted cargo from the Great Lakes (to which it had been assigned under the lowest landed cost formula) to coastal ports to satisfy the U.S.-flag quota. The Food for Peace program directed extra supplies to Africa to cope with a famine in 1985, and because few U.S.-flag vessels serve that continent most of the traffic went on foreign bottoms. Thus the need for extra U.S.-flag shipments. These diversions precipitated this suit. Before the district court could render a decision, both Congress and the federal agencies changed the rules. In December 1985 Congress enacted 46 U.S.C.App. § 1241f which, among other things, raised the domestic preference from 50% to 75%. Because the increase would affect Great Lakes ports adversely, Congress gave them a grandfather clause: [T]he Secretary of Transportation, in administering this subsection and section 1241(b) of this title ... shall take such steps as may be necessary and practicable without detriment to any port range to preserve during calendar years 1986, 1987, 1988, and 1989 the percentage share, or metric tonnage of bagged, processed, or fortified commodities, whichever is lower, experienced in calendar year 1984 as determined by the Secretary of Agriculture, of waterborne cargoes exported from Great Lake ports pursuant to title II of the Agricultural Trade Development and Assistance Act of 1954. 46 U.S.C.App. § 1241f(c)(2)(B). It added for good measure a change in the accounting year. Early in the year is the “best” time for shipments on foreign vessels, because the KCCO does not feel pressure to “make the quota”; in the fall, however, the KCCO may divert shipments so that things come out right. Great Lakes ports are closed at the beginning of the calendar year and open when the KCCO may be diverting traffic. Thus the impetus for a different accounting year: Instead of being coterminous with the calendar year, the accounting year now starts on April 1, when Great Lakes ports open for business. 46 U.S.C.App. § 1241f(c)(2)(A). In February 1987 MarAd changed the rules under which KCCO evaluates bids. 52 Fed.Reg. 5726, 5729 (1987), 7 C.F.R. § 1496.5 (1988). These regulations require KCCO first to “calculate the lowest landed cost using only higher-priced United States-flag ship rates for the portion of its commodities that it believes is necessary to meet its obligations under the act on a nationwide basis”, 823 F.2d at 1162, and then allow full competition between U.S. and foreign-flag vessels for the rest. In other words, the regulations segment the market. KCCO requires its bidders to use only U.S.-flag rates when making bids for 75% of the exports, and it allows bids based on lowest available cost for the remaining 25%. It gave this rationale for the new system: *543 In most cases, [the prior method] resulted in commodities being procured and allocated to ports (or port ranges) on the basis of foreign flag vessel rates since these rates are often lower than U.S. flag rates. Foreign flag and U.S. flag vessels are then contracted to carry portions or all of the cargoes allocated to these ports (or port ranges). Generally, cargo preference requirements have been met by contracting with U.S. flag vessels calling at these ports. On occasion this practice results in higher expenditures for U.S. flag carriage than is necessary to comply with cargo preference because commodities allocated on an LLC basis may not have been allocated to ports serviced by U.S. flag vessels with the least costly U.S. flag rates. The requirement for increased cargo preference will exacerbate the problem. 52 Fed.Reg. 5726 (1987). Milwaukee took up cudgels against this regulation, the source of its current woes. It contends that the KCCO must allocate shipments on the basis of “true” lowest landed cost rather than on the basis of least cost given the segmentation. According to Milwaukee, this would not prevent implementation of the 75% preference, because cargoes allocated to the Great Lakes can be carved out of the denominator. Milwaukee also submits that administering the preference on a national basis violates the Port Preference Clause of the Constitution, Art. I § 9 cl. 6, because it draws cargo away from the Great Lakes. The district court enjoined implementation of the revised 7 C.F.R. § 1496.5. 688 F.Supp. 479 (E.D.Wis.1988). The court did not reach Milwaukee's arguments that only a pure lowest landed cost method complies with the Act and that the preference must be applied port-by-port; it rejected Milwaukee’s contention that the regulation violates the Port Preference Clause. 688 F.Supp. at 488. It held instead that the “plain language” of § 1241(b)(1) forbids use of a nationwide base. 688 F.Supp. at 486-87. Under the statute the government must allocate cargoes “in such manner as will insure a fair and reasonable participation of United States-flag commercial vessels in such cargoes by geographic areas” (emphasis added), which the court thought means by port ranges. Federal officials — joined by the Transportation Institute, two U.S.-flag lines, and two unions of coastal dock workers, which had intervened to oppose Milwaukee’s demands— filed an appeal. Milwaukee has filed a pointless cross-appeal, supporting the judgment but asking us to substitute its rationales for the district court’s, which it does not defend. (Reminders, e.g., Jordan v. Duff & Phelps, Inc., 815 F.2d 429, 439 (7th Cir.1987), that prevailing parties need not file cross-appeals to make arguments in defense of their judgments seem to fall on deaf ears.) The command to allocate cargoes “in such manner as will insure a fair and reasonable participation of United States-flag commercial vessels in such cargoes by geographic areas” speaks of “a fair and reasonable participation of United States-flag commercial vessels in such cargoes”, not of a fair and reasonable participation of ports or port ranges. Section 1241(b)(1) is special-interest legislation, but the interest is that of U.S.-flag lines, not of ports. “By geographic areas” means “by destination”, not “by origin”. This ensures that the government can’t short-haul domestic carriers. It can’t send shipments from Bangor, Maine, to Providence, Newfoundland, on U.S. ships while reserving all the traffic from Philadelphia to Bangkok for foreign bottoms. If the statute were not clear as is, the legislative history fills all need. The language in question comes from other cargo-preference laws in which it is unmistakably a reference to destinations rather than origins, e.g., § 409 of the Mutual Defense Assistance Act of 1949, 63 Stat. 714, 720 (1949). Administrative interpretations of this language before it appeared in § 1241(b)(1) in 1954 uniformly treated it as a reference to destinations. E.g., Department of Defense Directive No. 2110.12 (Nov. 14, 1951) (“The term ‘geographic areas’ is considered to be descriptive of destination areas and not related to origin areas.”). Interpretations along these lines *544 were presented to Congress before it enacted § 1241(b)(1). By choosing tried-and-true language, Congress also chose a tried-and-true interpretation. Ever since, the responsible agencies have interpreted this language as referring to destinations. They could hardly have done otherwise, and the district court erred in holding that this language forbids a nationwide base. Milwaukee defends its victory on the ground that cargo arriving at ports not frequented by U.S.-flag carriers must be carved out of the total to which the 75% preference applies. It reasons that the statute speaks of a domestic preference only when U.S.-flag carriers are available. If U.S.-flag carriers are unavailable — or if they want too much money — then all cargo may move on foreign bottoms. All parties accept these propositions. Vessels’ availability varies by port. Hence, Milwaukee submits, the ratio must be computed port by port. If each port delivers 75% of its cargo to U.S. bottoms available there, the statutory requirement has been satisfied. It is possible to read § 1241(b)(1) this way, but it is not inevitable. As we have pointed out, someone has to choose a denominator in order to produce a fraction. In the ordinary course, the agencies charged with making the program work may make choices of this sort unless the statute and its history show that Congress has made the choices itself. Homemakers North Shore, Inc. v. Bowen, 832 F.2d 408, 411-12 (7th Cir.1987). The role of the Executive Branch in construing laws is hotly disputed; the role of the Executive Branch in filling the interstices left by statutes that commit a subject to executive discretion is beyond question. Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 843-44, 104 S.Ct. 2778, 2781-83, 81 L.Ed.2d 694 (1984). Section 1241(b)(1) does not speak to the question, and neither does the legislative history. The parties pepper their briefs with statements made to Congress in hearings; Milwaukee is especially fond of an exchange between Rep. Allen and Francis T. Greene, testifying on behalf of the Merchant Marine Institute, that occurred during hearings before the House Committee on Merchant Marine and Fisheries. The government replies with legislative history of its own. We are leery of inferences based on exchanges during hearings. Unless repeated in the committee reports (none was), such statements go unnoticed by Congress and cannot inform, or help us to understand, anyone’s vote. Even statements on the floor — unreliable because no one may have been there when the words were delivered — are more useful than colloquies in hearings. Cf. Covalt v. Carey Canada Inc., 860 F.2d 1434 (7th Cir.1988). Statutory structure is much more reliable as a guide to meaning, and the structure of these statutes supports the agencies’ position. Milwaukee uses the “availability” requirement of § 1241(b)(1) as the jumping-off place for its argument: it means, Milwaukee believes, that the preference applies only when U.S.-flag vessels happen to be available in port when the cargo arrives. Yet the year-long accounting period means that the statutory preference is not limited to vessels on hand. U.S.-flag ships are entitled to 75% of a year’s worth of cargo (if available at reasonable prices). If the baseline extends beyond the ships in port each day, why can’t it extend beyond the given port? The contemporary Congress believes that it does. The Great Lakes grandfather clause and the new accounting year in § 1241f make sense only if KCCO is entitled to use a nationwide baseline; if, as Milwaukee submits, the government must proceed port-by-port, it is quite unnecessary to protect the Great Lakes from diversions. The “by geographic areas” clause also suggests the need for a baseline larger than one port. If KCCO is to even out shipments by destination, it has to be able to bring more than one port at a time into the computation. Although the structure of § 1241(b)(1) and § 1241f supports the government’s position, we need not decide whether it compels the government to use a national baseline. Milwaukee’s injury does not flow directly from the choice of a national baseline. It comes from the revised “lowest landed cost” regulations, which allocate to *545 U.S.-flag vessels 75% of the cargo (and minimize the cost to the government, given the decision to do so). Nothing in § 1241(b)(1) or § 1241f says that the government is forbidden to provide U.S.flag carriers with more than the amount of the preference. Indeed, nothing in these statutes says how the KCCO shall decide where to buy the commodities, which port to move them to, and how to export them. If the KCCO were to decide to use U.S.flag vessels for 100% of all exports, this could not be said to violate § 1241(b)(1) or § 1241f — or for that matter any other statute of which we are aware. Such a decision would raise the total costs to the government (more likely, reduce the amount of food delivered, given the appropriation for foreign aid), but not at the expense of any entitlements protected by these statutes. Higher costs associated with using “too many” domestic vessels might be said to be arbitrary and capricious, presenting questions under the Administrative Procedure Act. Yet the APA has never been thought to cover ordinary procurement decisions of federal agencies — whether to use thick Acme rubber bands rather than thin (and cheap) Zippy rubber bands, for example. Agencies as purchasers of goods and services may make the same sorts of judgments private purchasers might, unless a statute covers the subject. Perkins v. Lukens Steel Co., 310 U.S. 113, 127, 60 S.Ct. 869, 876, 84 L.Ed. 1108 (1940); Reeves, Inc. v. Stake, 447 U.S. 429, 439 n. 12, 100 S.Ct. 2271, 2278 n. 12, 65 L.Ed.2d 244 (1980); Peoples Gas Co. v. Postal Service, 658 P.2d 1182, 1200-01 (7th Cir.1981). Some procurement decisions have been governed by special statutes, such as the Service Contract Act, 41 U.S.C. §§ 351-358, or laws requiring the government to use the low bidder. Some have not. If the Army wants gold-plated can openers, and this increases the costs of procurement, this does not violate the rights of a bidder who would like to sell can openers made out of brass. Milwaukee is not the victim of regulation; it is only a disappointed bidder. It believes that it can “sell” the government a port service that is efficient because (a) close to the source of the food, and (b) served by vessels offering low rates. Milwaukee suffers because the government does not want to buy as much of this service as Milwaukee believes would be optimal. Even if no statute compels the government to use a national baseline, no statute protects Milwaukee from this sort of purchasing decision, either. Perhaps no statute, Milwaukee rejoins, but the Port Preference Clause does. It says: No Preference shall be given by any Regulation of Commerce or Revenue to the Ports of one State over those of another: nor shall Vessels bound to, or from, one State, be obliged to enter, clear, or pay Duties in another. Undoubtedly the statutes and the KCCO’s practices mean that less traffic goes through Milwaukee than would do so if the government selected ports on a low-bid basis. Yet failing to use the lowest price is not the same thing as discriminating among ports. The “preference” in the statutes is for U.S.-flag vessels. If domestic vessels called at Milwaukee, the KCCO would be only too happy to use its port. Milwaukee suffers from the disparate effects of statutes and regulations written in a neutral fashion. (Not quite: § 1241f(c)(2)(B), the grandfather clause, expressly uses geography to allocate business to the Great Lakes. Not surprisingly, Milwaukee does not say that this rule violates the Port Preference Clause, and we need not decide whether it does.) Preferences for domestic shipping are not the only federal rules that leave coastal ranges better off. The Navy builds, berths, and repairs its ships on the coasts — at Pascagoula, Mississippi, at Newport News, Virginia, on Long Island, and so on — for much the same reason Milwaukee suffers from § 1241(b)(1): large ships cannot get through the locks to the upper Great Lakes, and even if they could, it would be inconvenient to bring ships far inland. Disparate consequences of neutral rules do not violate the Port Preference Clause. An unbroken line rejects contentions similar to those pressed on us. E.g., Pennsyl *546 vania v. Wheeling & Belmont Bridge Co., 59 U.S. (18 How.) 421, 433-36, 15 L.Ed. 435 (1856) (a bridge that interdicts access to an upriver port does not violate the Clause, remarking, 59 U.S. at 435 that the Clause “seems to import a prohibition against some positive legislation by congress [preferring certain ports], and not against any incidental advantages that might possibly result from the legislation of congress upon other subjects”); South Carolina v. Georgia, 93 U.S. 4, 12-13, 23 L.Ed. 782 (1876) (similar); Armour Packing Co. v. United States, 209 U.S. 56, 80, 28 S.Ct. 428, 435, 52 L.Ed. 681 (1908) (rules changing rail rates and diverting traffic from some ports do not violate the Clause); Louisiana Public Service Comm’n v. Texas & New Orleans R.R., 284 U.S. 125, 131, 52 S.Ct. 74, 76, 76 L.Ed. 201 (1931) (a rail rate allowance for crossing the Mississippi in Louisiana does not violate the Clause: “Congress ... causes many things to be done that greatly benefit particular ports and which incidentally result to the disadvantage of other ports”); Alabama Great Southern R.R. v. United States, 340 U.S. 216, 229, 71 S.Ct. 264, 272, 95 L.Ed. 225 (1951). See also City of Houston v. FAA, 679 F.2d 1184, 1196-98 (5th Cir.1982) (summarizing these cases). Nothing in the background of the Port Preference Clause suggests that the Supreme Court has missed a turn. Maryland’s delegation to the Constitutional Convention demanded the inclusion of this clause in order to avert the possibility that Congress might require vessels to stop and clear customs at Norfolk, Virginia, before entering Chesapeake Bay, to the detriment of Baltimore. Max Farrand, 2 Records of the Federal Convention 417 (Madison: August 25, 1787). Maryland got what it wanted — not only the explicit ban on an obligation to clear customs at the coastal ports but also the more general prohibition of preferences. Yet Luther Martin, the force behind the Clause, was not satisfied and went into opposition after the Convention. Martin complained that the Clause dealt only with express preferences and left states to bear the effects of other rules. He expressed greatest concern about Congress’ power to designate some places as customs ports to the exclusion of others: [A]s the system is now reported, the general government have a power to establish what ports they please in each State, and to ascertain at what ports in every State ships shall clear and enter in such State, a power which may be so used as to destroy the effect of that provision, since by it may be established a port in such a place, as shall be so inconvenient to the State, as to render it more eligible for their shipping to clear and enter in another than in their own State; suppose, for instance, the general government should determine that all ships which cleared or entered in Maryland, should clear and enter at George-Town, on Potowmack [this was before Maryland ceded the swampland at Georgetown to the United States for the national capital!], it would oblige all the ships which sailed from, or were bound to, any other port of Maryland, to clear or enter in some port in Virginia. Luther Martin, Genuine Information (1788) (emphasis in original), reprinted in Philip B. Kurland & Ralph Lemer, 3 The Founders’ Constitution 372 (1987). Hyperbole from opponents must be used with care, but Martin, as the proponent of the Port Preference Clause, was in the ideal position to know what he had and hadn’t obtained from his colleagues at the Convention. He got a ban on express discrimination; he wanted, and couldn’t get, a ban on disparate impact. For two hundred years courts have understood that only explicit discrimination violates the Port Preference Clause, and this dooms Milwaukee’s argument. Milwaukee has emphasized that the upper Great Lakes are good places to load midwestern foodstuffs for export, and that they ought to be allowed to enjoy their natural economic advantages. Milwaukee has a comparative advantage, however, only if we conceive the P.L. 480 program as an unalloyed gift to foreign peoples. A Congress bent on delivering the maximum amount of food to starving persons, given budgetary limitations, would allow the *547 KCCO to pick the “real” lowest landed cost, which would mean sending more food through the upper Great Lakes. Few statutes have only one objective or one set of beneficiaries. Sections 1241(b)(1) and 1241f show that P.L. 480 has three beneficiaries: the starving, American farmers, and the owners of American ships. Shipowners provided political support for the P.L. 480 program at a price: work for domestic vessels. Wise or not, the preference is there. Federal officials are entitled to use the method they have selected to set aside 75% of the cargoes for U.S.-flag vessels. The injunction issued by the district court accordingly is REVERSED.