After getting off to a memorable start in M’Culloch v. Maryland, 17 U.S. (4 Wheat.) 316, 4 L.Ed. 579 (1819)1 the doctrine of intergovernmental tax immunity has deteriorated into a morass of “inconsistent decisions and excessively delicate distinctions,” United States v. New Mexico, 455 U.S. 720, 730, 102 S.Ct. 1373, 71 L.Ed.2d 580 (1982).2 We venture into the bog yet again to evaluate Nevada’s latest effort to squeeze some tax revenue from the activities of its largest landowner, the United States Government.
When these parties last came to visit, we held that Nevada’s tax on federal property used by government contractors violated the Supremacy Clause. See United States v. Nye County Nev., 938 F.2d 1040, 1041 (9th Cir.1991). Learning from its mistakes, the Nevada legislature modified its law to tax federal contractors’ beneficial use of federal property, rather than the property itself. See Nev. Rev. Stat § 361.157 (1997) (real property); id. § 361.159 (personal property). The United States and various of its contractors returned to district court, claiming that such linguistic acrobatics cannot save the Nevada statutes from constitutional infirmity.
The appeals before us involve a host of federal contractors: Areata Associates, Loral Aerospace (now Lockheed Martin Aerospace), Raytheon Services Nevada, EG & G Energy Measurements, Wacken-hut Services and Reynolds Electrical & Engineering. Each contractor manages and maintains federal property in Nevada or provides vital services on that property. The United States ultimately bears the full burden of these taxes because it compensates each of its contractors under standard cost-plus-fee agreements. At year’s end, Nevada requires every contractor to list government property it has used in the course of performing its contract(s) and, at the contractor’s option, estimate the percentage of that property it has used.3 The *1084relevant county — in our case, Nye 4 — then assesses a tax based on the value of that federal property.
The way in which Nye County administers the tax has not changed since our 1991 decision. The question before us, then, is whether the mere change in the wording of the Nevada statutes removes their constitutional infirmity.
Statute of limitations
As a preliminary matter, the counties argue that some of the federal government’s claims are time-barred or otherwise procedurally barred under Nevada law. According to the counties, the federal government is acting as a subrogee of the contractors, and hence its rights are limited by the same procedural rules that would limit the contractors were they to seek relief from the tax in state court. See United States v. California, 507 U.S. 746, 758-59, 113 S.Ct. 1784, 123 L.Ed.2d 528 (1993). However, California considered state law substantive claims. Here, the United States asserts a federal constitutional claim on its own behalf, claiming that the counties have taxed federal property. State procedural rules do not govern such a claim. See United States v. Thornburg, 82 F.3d 886, 893 (9th Cir.1996).
A tax on federal property?
Although states may not tax the federal government or its property directly, see M’Culloch, 17 U.S. (4 Wheat.) at 435, they may tax private parties who use federal property, even when these private parties are providing goods and services to the United States, see New Mexico, 455 U.S. at 734, 102 S.Ct. 1373. Which side of this nebulous line do Nevada’s revised statutes fall on?
We start with Nye County. The statute we struck down there provided:
Personal property exempt from taxation which is leased, loaned or otherwise made available to and used by a natural person, association or corporation in connection with a business conducted for profit is subject to 'taxation in the same amount and to the same extent as though the lessee or user were the owner of the property ....
938 F.2d at 1042 (emphasis added) (citing Nev.Rev.Stat. 361.159 (1991)). We held that “the wording of a tax measure is significant.... [W]hen a statute says it taxes property it probably does. And when it says it doesn’t, it probably doesn’t.” Id. Since the Nevada statute said it taxed property, we accepted that characterization. Although we also criticized the county for making no attempt to segregate and tax the contractor’s beneficial use of the property, see id. at 1043, it was the language of the Nevada statute, not the county’s practice, that was critical to the result, see id.
Taking to heart Nye County’s friendly advice that a tax on “a user’s beneficial use of [ ] property owned by the United States” would be constitutional, id., Nevada amended its tax statutes to do just that. The personal property tax statute now reads:
[W]hen personal property ... which for any reason is exempt from taxation is leased, loaned or otherwise made available to and used by a natural person, association or corporation in connection with a business conducted for profit, the leasehold interest, possessory interest, beneficial interest or beneficial use of any such lessee or user of the property is subject to taxation to the extent the:
(a) Portion of the property ... used; and
(b) Percentage of time during the fiscal year that the property is ... used by the user,
*1085can be segregated and identified.
Nev.Rev.Stat. § 361.159. In the same vein, the new real property tax statute taxes “the leasehold interest, possessory interest, beneficial interest or beneficial use of the lessee or user of the [tax-exempt real estate]” subject to the same apportionment and time conditions. Id. § 361.157. Thus, Nevada has shifted the subject of the taxes from the property itself to the beneficial use of that property. Taking the statutes at face value, as Nye County says we should, they appear to be constitutional.
The Supreme Court reached a similar conclusion in United States v. Township of Muskegon, 355 U.S. 484, 78 S.Ct. 483, 2 L.Ed.2d 436 (1958), where it approved a tax on a government-owned manufacturing plant used rent-free by a private motor company to perform supply contracts with the federal government. “[V]ital” to the result was that the contractor “us[ed] the property in connection with its own commercial activities” and retained autonomy “to use the property as it thought advantageous and convenient in performing its contracts and maximizing its profits,” Id. at 486, 78 S.Ct. 483. Because this also describes the situation of the contractors here, Muskegon supports the constitutionality of Nevada’s tax statutes.
We recognize that the Tenth Circuit has read Muskegon narrowly as applying only to contractors who produce goods on government property, as opposed to those who “merely perform! ] [their] contractual obligations on government owned property.” United States v. Colorado, 627 F.2d 217, 220 (10th Cir.1980), affd mem. sub nom. Jefferson County v. United States, 450 U.S. 901, 101 S.Ct. 1335, 67 L.Ed.2d 325 (1981). Intuitively, there’s something to this distinction: A contractor who merely performs maintenance work on a government office building uses that building to a less substantial degree than one who uses the facilities to manufacture goods. But what of a contractor who uses government property to produce an intangible product like scientific analysis? Or a contractor who performs management services while on federal property, and to that end is given office space and the use of government telephones, computers and other equipment? Or a contractor who provides multiple services, including maintenance and management? Attempting to locate government contractors on this goods/services continuum is a slippery task that could well lead to different tax treatment of the different contractors here, or even different treatment for different activities by the same contractor. The United States in its many briefs never tells us just how the Colorado distinction would work in practice, and we doubt that it is possible to draw a satisfactory line of demarcation.
By summarily affirming a lower court opinion, the Supreme Court adopts its “judgment only, [not its] rationale.” Mandel v. Bradley, 432 U.S. 173, 176, 97 S.Ct. 2238, 53 L.Ed.2d 199 (1977) (per curiam). Thus, we are not bound by the Tenth Circuit’s reading of Muskegon. Indeed, the Court has never invoked Colorado’s goods/services distinction in evaluating the constitutionality of a tax. Cf New Mexico, 455 U.S. at 741 n. 14, 102 S.Ct. 1373. Therefore, because we have serious reservations about the practicability of the Colorado distinction, we adhere to our own reading of Muskegon and what, in the Court’s own words, was “vital” to its result. See Muskegon, 355 U.S. at 486, 78 S.Ct. 483. Here, as there, the contractors use government property as they see fit in conducting their commercial activities. Here, as there, it is constitutionally permissible for the state to tax those contractors, employing their use of federal property as the yardstick.
New Mexico reinforces our conclusion. Attempting to clarify “the confusing nature of [its] precedents,” New Mexico, 455 U.S. at 733, 102 S.Ct. 1373, the Court ruled:
tax immunity is appropriate in only one circumstance: when the levy falls on the United States itself, or on an agency or *1086instrumentality so closely connected to the Government that the two cannot realistically be viewed as separate entities, at least insofar as the activity being taxed is concerned.
Id. at 735, 102 S.Ct. 1373. Under this test, private contractors may invoke the federal government’s tax immunity only in very limited circumstances. This is not such an occasion: The federal government has made no convincing argument that any of the contractors here is “so closely connected to the Government that the two cannot realistically be viewed as separate entities.” Id.; see also Arizona Dep’t of Revenue v. Blaze Constr. Co., — U.S. -, -, 119 S.Ct. 957, 960, 143 L.Ed.2d 27 (1999).5
The United States argues that United States v. County of Fresno, 429 U.S. 452, 97 S.Ct. 699, 50 L.Ed.2d 683 (1977), cabs for the opposite result. The Court there suggested that a state might run afoul of the Constitution if it imposed a use tax on a Forest Service employee’s fire ax. See id. at 466 n. 15, 97 S.Ct. 699. This hypothetical, which was not even a holding of Fresno, is readily distinguishable: It posits property used by a federal employee, whereas the tax here falls on private parties who are federal contractors. Contractors may often be taxed when employees may not be. See, e.g., United States v. Boyd, 378 U.S. 39, 46-48, 84 S.Ct. 1518, 12 L.Ed.2d 713 (1964). Indeed, that distinction is at the heart of the Neto Mexico standard. See New Mexico, 455 U.S. at 735, 102 S.Ct. 1373.
Having concluded that the counties may tax the beneficial use of government property, the next question is how they may calculate the tax. Much case law suggests that the tax on the use of federal property may be measured by the value of the property itself. See, e.g., Boyd, 378 U.S. at 44, 84 S.Ct. 1518; United States v. City of Detroit, 355 U.S. 466, 470, 78 S.Ct. 474, 2 L.Ed.2d 424 (1958); United States v. County of San Diego, 53 F.3d 965, 969 (9th Cir.1995). However, the Sixth Circuit has held that the amount of the tax may not exceed the value of the property’s use to the contractor. See United States v. Hawkins County, 859 F.2d 20, 23 (6th Cir. 1988). The Supreme Court has hinted at the same. See New Mexico, 455 U.S. at 741 n. 14,102 S.Ct. 1373.
Whether or not there needs to be a valuation limit, Nevada’s revised statutes already provide one: Federal contractors are taxed based only on the portion of the property they use and the percentage of time they actually use it. See Nev.Rev. Stat. § 361.157(l)(a) & (b); id. § 361.159(l)(a) & (b). As a general matter, this adequately limits contractors’ tax liability. Disputes over the accuracy of specific assessments must be resolved, at least in the first instance, through the state’s administrative and judicial processes.
Discrimination against the federal government?
Nevada’s real property tax exempts, among others, those who use the “[property of any state-supported educational institution.” Id. § 361.157(2)(c).6 This provision, the United States argues, renders the entire tax invalid because it “operates so as to discriminate against the [Federal] Government or those with whom it deals.” City of Detroit, 355 U.S. at 473, 78 S.Ct. 474; see also M’Culloch, 17 U.S. (4 Wheat) at 436-37.
The Supreme Court’s tax discrimination jurisprudence boils down to the requirement that each state “treat those who deal with the [Federal] Government as well as it treats those with whom it deals *1087itself.” Phillips Chem. Co. v. Dumas In-dep. Sch. Dist, 361 U.S. 376, 385, 80 S.Ct. 474, 4 L.Ed.2d 384 (1960). A tax statute that explicitly treats contractors who use state property better than contractors who use federal property passes constitutional muster only if “significant differences” between those taxpayers justify their inconsistent treatment. Davis v. Michigan Dep’t of the Treasury, 489 U.S. 803, 815— 16, 109 S.Ct. 1500, 103 L.Ed.2d 891 (1989) (citing Phillips, 361 U.S. at 383, 80 S.Ct. 474) (internal quotation marks omitted).
The clearest example of a discriminatory tax is where the state completely segregates those who deal with the federal government and taxes them more heavily than similarly situated entities who deal with the state. The Supreme Court addressed this situation in Phillips, where Texas imposed a tax on all federal lessees under one statute, and imposed a less burdensome tax on all state lessees. See Phillips, 361 U.S. at 378-80, 80 S.Ct. 474. Finding the discrimination “apparent,” id. at 382, 80 S.Ct. 474, and the differences between the two classes of taxpayers “impalpable,” id. at 387, 80 S.Ct. 474, the Court held that Texas could not collect revenue from federal lessees under the more burdensome tax.
Since Phillips, some states have attempted to favor themselves in a slightly less transparent manner by exempting a particular class of state employees or contractors from a tax of otherwise general applicability. Still, courts have made it clear that preferential treatment for any state lessees — no matter how small a group — must extend to their federal counterparts. Thus, the Fourth Circuit saw “obvious” discrimination in a Virginia statute that taxed those who used the tangible personal property of any federal, state or local government entity, but exempted those who used the property of the Virginia Port Authority or other state transportation agencies. See United States v. City of Manassas, 830 F.2d 530, 533 (4th Cir. 1987), affd mem., 485 U.S. 1017, 108 S.Ct. 1568, 99 L.Ed.2d 884 (1988). Because the record demonstrated “no distinction ... between lessees from the state and national governments,” the court held that the tax was unconstitutional. Id. at 535. Similarly, the Supreme Court in Davis invalidated a Michigan statute that exempted from its definition of taxable income all retirement benefits paid by the state or its political subdivisions, finding that the discrimination was “undisputed,” and that no significant differences between state and federal retirees justified their differential treatment. See Davis, 489 U.S. at 816-17, 109 S.Ct. 1500. While the Court left it for the state courts to determine how best “to comply with the [Constitution’s] mandate of equal treatment,” it suggested two possible remedies: expanding the exemption or deleting it altogether. Id. at 818, 109 S.Ct. 1500.
The government argues that, like the provisions in Manassas and Davis, subsection 361.157(2)(c) impermissibly preferences certain state contractors over their federal counterparts. However, when we assess this exemption in light of the tax statute as a whole, we find it to be quite different from the exemptions invalidated in those cases.
First, the overall structure of section 361.157 distinguishes it from the statutes in Manassas and Davis. Virginia’s tax contained only a single exemption, and it distinguished between state and federal contractors. See Manassas, 830 F.2d at 533. And while Michigan’s definition of taxable income presumably contained multiple exemptions, only the one for state retirement benefits seems to have drawn any kind of state/federal distinction. See Davis, 489 U.S. at 814, 109 S.Ct. 1500. By contrast, section 361.157 exempts thirteen categories of property from its beneficial use tax. Although the great majority of these are facially neutral with regard to state/federal ownership, subsection (2)(c) is one of at least two provisions that is not: Subsection 361.157(2)(d) operates to favor the federal government, exempting lessees *1088who use certain federal land.7 There is no similar exemption for those who lease state land. Thus, both state and federal contractors benefit under section 361.157, each in a different way, each to the exclusion of the other. Unlike Manassas and Davis, where the targeting of federal interests was “obvious” and “undisputed,” it is less clear whether section 361.157 taken as a whole disadvantages the federal government and those that do business with it.
It is true that certain state contractors — those who deal with state institutions of higher learning — are exempt. And under the broadest reading of Manas-sas, as soon as any state contractor is given an exemption, all federal contractors must be given a similar exemption. We do not read Manassas so broadly. Where, as here, the statute contains a series of exemptions, some of which favor the federal government, others of which favor the state, most of which are unconcerned with the federal/state distinction, we focus on the individual exemption to determine whether each taken on its own terms discriminates between state and federal interests to the detriment of the federal government. Here, the only exemption that favors the state is the one dealing with those who contract with state-supported educational institutions. Does this provision, standing alone, in fact discriminate against the federal government? While it appears to do so by its terms, in fact, it does not because Nevada has no federally-supported educational institutions, while Nevada has a very fine state university system.8 Therefore there are no federal contractors analogous to the state contractors who benefit from subsection (2)(c).9
In light of this well-known fact, it is more than likely that the Nevada legislature did not draft subsection 361.157(2)(c) with the federal government in mind. The term “state-supported educational institutions,” rather, appears to draw a distinction between educational institutions supported by the state and private educational institutions. So read, it reflects an exercise of fiscal discretion' — giving an advantage to public educational institutions visa-vis private ones — rather than an effort to benefit the state at federal expense.10 The Constitution prohibits only the latter. See Davis, 489 U.S. at 815 n. 4, 109 S.Ct. 1500.
In sum, we conclude that Nevada Revised Statutes sections 361.157 and 361.159, as amended in 1993, steer clear of *1089the constitutional prohibitions on taxing federal property directly and discriminating against the federal government and those with whom it deals. Therefore, we uphold the tax as applied by Nye County to the contractors here.
Loose ends
The district court properly ordered Clark County to refund taxes collected under the unconstitutional prior version of the statutes. We have already held that the county’s statute of limitations defense is inapplicable. The reasoning of Nye County — which rests primarily on the wording of the state statutes — applies equally to both counties. Clark County must pay the refund. It also must pay post-judgment interest, see 28 U.S.C. § 1961, and hence the district court erred in failing to award it.11
Finally, the district judges who upheld Nevada’s tax statutes below also held that the federal government was contractually bound to reimburse the contractors for the taxes they paid under the Nevada statutes. As the United States now points out, the district court lacked jurisdiction to decide the federal government’s liability.12 The contractors may appeal only to the agency board of contract appeals, see 41 U.S.C. § 605, or to the Court of Federal Claims, see id. § 609. Therefore, we must vacate the judgment entered in favor of the contractors against the United States. We remand to the district court with instructions to transfer these claims to the Court of Federal Claims pursuant to 28 U.S.C. § 1631. See, e.g., Sessler v. United States, 7 F.3d 1449, 1452 (9th Cir.1993).
The decision in No. 97-15309 is REVERSED. The decisions in the remaining appeals are AFFIRMED, except for those portions dealing with post-judgment interest, which are REVERSED and REMANDED for the district court to calculate the amount of post-judgment interest, and those portions dealing with indemnity claims against the United States, which are VACATED and REMANDED with instructions that they be transferred to the United States Court of Federal Claims. No costs allowed.
. What high school student can't recite Chief Justice Marshall’s aphorism that “the power to tax is the power to destroy”? Alas, however, like some of Marshall's other famous phrases, see, e.g., Alex Kozinski, That Unfortunate Immortal Phrase, 1987 Utah L.Rev. 977, this one, too, has proven much easier to say than to apply.
. Which, of course, makes it very different from other areas of constitutional law. See, e.g., First English Evangelical Lutheran Church of Glendale v. County of Los Angeles, 482 U.S. 304, 340 n. 17, 107 S.Ct. 2378, 96 L.Ed.2d 250 (1987) (Stevens, J., dissenting) (“However confusing some of our criminal procedure cases may be, I do not believe they have been as open-ended and standardless as our regulatory takings cases are.'').
.If the contractor chooses not to make such an allocation, it is taxed on the full value of the property.
. Clark County apparently does not collect taxes under the new versions of the statutes, but doubtless soon will.
. Judge Hunt below did find that Reynolds’s fire protection services should not be taxed because they fit within the New Mexico language. Nye County doesn't contest this ruling on appeal so the matter is not before us.
. This exemption is not contained in the personal property statute. See Nev.Rev.Stal. § 361.159.
. A Subsection (2)(d) exempts lessees of land made available under the Taylor Grazing Act or by the United States Forest Service or the Department of the Interior's Bureau of Reclamation. We also note that subsection (2)(b) exempts lessees of ''[fjederal property for which payments are made in lieu of taxes in amounts equivalent to taxes.” However, this provision seems to be more of a protection against double taxation than an actual exemption.
. The parties agree that the term ‘'stale-supported” for purposes of this exemption refers to educational institutions that are instrumen-talities of the state, not private institutions that obtain state funding through grants or scholarships. By analogy, "federally-supported” institutions are only those (like the military academies) that are directly funded and operated by the federal government, not all those that obtain some indirect federal funding.
. If there were any federally-supported educational institutions in Nevada, we would invalidate subsection 361.157(2)(c) as unconstitutionally discriminatory. Following the Supreme Court's lead in Davis, we would then leave it to the Nevada legislature to fix the problem by eliminating the exemption or expanding it to cover those who contract with federally-supported educational institutions. See Davis, 489 U.S. at 818, 109 S.Ct. 1500.
.Of course, we could not uphold section 361.157 merely because it serves the important purpose of supporting education. Since the Sixth Circuit’s ruling to the contrary in Chrysler Corp. v. Township of Sterling, 410 F.2d 62, 70 (6th Cir.1969), the Supreme Court has made it clear that "[t]he State’s interest in adopting [a] discriminatory tax, no matter how substantial, is simply irrelevant to an inquiry into the nature of the two classes receiving inconsistent treatment.” Davis, 489 U.S. at 816, 109 S.Ct. 1500; see also Manas-sas, 830 F.2d at 533-34.
. Pre-judgment interest is discretionary, and the United States does not appeal the district court’s decision not to award it.
. The district court’s exercise of jurisdiction implicates the government’s sovereign immunity. The United States waived its immunity and consented to the jurisdiction of the Court of Federal Claims in contract disputes. See United States v. Mitchell, 463 U.S. 206, 215, 103 S.Ct. 2961, 77 L.Ed.2d 580 (1983). Because sovereign immunity is a question of subject matter jurisdiction, the United States may raise it at any point in the proceedings. See Fed.R.Civ.P. 12(h)(3).