with whom Justice Ginsburg joins, dissenting.
The Court today holds a federal statute unconstitutional without giving heed to the simplest reason for sustaining it. We granted certiorari on the question “[w]hether, as applied to casualty insurance for losses incurred during the shipment of goods from locations within the United States to purchasers abroad, the tax imposed by Section 4371 of the Internal Revenue Code violates the Export Clause of the Constitution of the United States (U. S. Const. Art. I, §9, Cl. 5),” Pet. for Cert. I. A straightforward answer to the question presented requires us to address the narrow issue of the continuing validity of our holding in Thames & Mersey Marine Ins. Co. v. United States, 237 U. S. 19 (1915), that a general tax on certain insurance premiums, as applied to exporters, is a prohibited tax on export goods.
Rejecting this course, the Court ventures upon a broad constitutional inquiry not even implicated by the statute. To do so, it rewrites the question presented. In the first sentence of the opinion, the Court says, “We resolve in this case whether the Export Clause of the Constitution permits the imposition of a generally applicable, nondiscriminatory *864federal tax on goods in export transit,” ante, at 845. In so reformulating the question, the Court makes the assumption that § 4371’s insurance tax is a tax on export goods, thereby adopting the premise of Thames & Mersey that I had thought we were to address. In the end the Court assumes the statute to be invalid rather than deciding it to be so. I find no precedent for setting aside an Act of Congress in this peremptory way. Worse yet, the Court’s assumption is wrong; because § 4371 taxes a service distinct from the goods and is not a proxy for taxing the goods, it does not fall within the prohibition of the Export Clause. The Court thus carves out an undeserved exemption from §4371 for exporters, adding significant complexity to its administration. Moreover, in a case in which the Export Clause should not even apply, the Court tackles the great problem of reconciling our Export Clause jurisprudence with modern decisions interpreting the Commerce and Import-Export Clauses, U. S. Const., Art. I, § 8, cl. 2, and Art. I, § 10, cl. 2. This is unwise and unnecessary. I would limit the inquiry to a reconsideration of Thames & Mersey, and uphold the statute as applied to respondent. With respect, I dissent.
HH
We consider a rather simple federal tax. Section 4371 of the Internal Revenue Code imposes a tax of “4 cents on each dollar, or fractional part thereof, of the premium paid on the policy of casualty insurance or the indemnity bond, if issued to or for, or in the name of, an insured . . . 26 U. S. C. §4371(1) (1982 ed.). The term “insured” is defined to include any “domestic corporation or partnership, or an individual resident of the United States, against, or with respect to, hazards, risks, losses, or liabilities wholly or partly within the United States .. .§ 4372(d)(1). The statute does not discriminate against exports. Indeed, it does not even mention them. The tax must be paid not only by domestic tradj-ers but also by any insured, even an individual, who is cov*865ered in whole or in part for domestic casualty risks. The purpose of the tax is to “eliminate an unwarranted competitive advantage now favoring foreign insurers,” H. R. Rep. No. 2333, 77th Cong., 2d Sess., 61 (1942), who do not pay federal income tax. Cf. 26 U. S. C. §4373(1) (1982 ed.) (exempting from §4371 any policy issued by a foreign insurer that is “signed or countersigned by an officer or agent of the insurer in a State, or in the District of Columbia, within which such insurer is authorized to do business” and is therefore subject to the income tax).
Resolution of the case requires us to determine whether the Export Clause has any bearing on taxes on services like insurance provided to exporters, where the service itself is not exported. The plain text of the Clause casts much doubt on the proposition. It states: “No Tax or Duty shall be laid on Articles exported from any State,” U. S. Const., Art. I, §9, el. 5. The majority avoids this necessary question by asserting that the Government failed to argue the point and so abandoned it. Ante, at 855, and n. 3. True, the Government defends § 4371 on the ground that it does not discriminate between exports and other forms of trade, but this is not a concession that there is no distinction between a tax on insurance premiums and a tax on goods. In fact, the Government makes repeated references to the distinction in its briefs, albeit in the context of discussing the nondiscriminatory character of § 4371. See, e. g., Brief for United States 12-13 (The tax “does not apply specifically to export transactions; to the contrary, it applies only to insurance risks that are either ‘wholly’ or ‘partly’ domestic”); id., at 15 (“The tax imposed by Section 4371 of the Internal Revenue Code is not specifically directed to nor directly ‘laid on Articles exported’ (U. S. Const. Art. I, § 9, Cl. 5). Instead, it applies to insurance premiums paid to foreign insurers for many forms of insurance, including any casualty risk that is ‘wholly or partly within the United States’ (26 U. S. C. § 4372(d)(1))”); id., at 34 (“Even as applied to casualty insurance, the tax *866unquestionably has only an incidental and remote relationship to exports and the export process .. .”).
At oral argument, the Assistant to the Solicitor General acknowledged that he had not made a separate argument based on the distinction between export goods and services related to the exporting process. He explained that the nondiscrimination theory had greater utility, sparing courts the nettlesome inquiry into what is an export. Tr. of Oral Arg. 9. When asked why the Government was avoiding the simpler and clearer argument that § 4371 was just a tax on foreign insurers to offset the tax burdens borne by domestic insurers, he responded, “We do not mean to avoid that argument. That’s part of our argument of why this is a tax of general application.” Id., at 12. Later in oral argument, he stated that “it’s problematic to describe a tax on insurance as a tax on the good,” and cited that problem as a reason for calling into question our decision in Thames & Mersey. Tr. of Oral Arg. 40. When asked if his position had foreclosed us from deciding the case on that basis, he responded: “I don’t believe you’re foreclosed ... by our concession from addressing that issue as you see fit.” Ibid. We have relied on statements more equivocal than this to reconsider and overrule a bad precedent even when the parties in their briefs had argued that the precedent should be upheld. See Blonder-Tongue Laboratories, Inc. v. University of Ill. Foundation, 402 U. S. 313, 319-320 (1971).
The Court’s faulty characterization of the Government’s argument leads it down some odd byways. For example, in Part III-B-3, the Court rejects the Government’s attempt to rely upon Department of Revenue of Wash. v. Association of Wash. Stevedoring Cos., 435 U. S. 734 (1978), where we held that a state tax of general applicability imposed upon a stevedoring firm did not violate the Import-Export Clause even though it may have added to the cost of importing and exporting. The Court points out that the tax in Washington Stevedoring did not fall directly on the goods, ante, at 861, *867and that we reserved the question whether States could tax goods in import or export transit, ante, at 862 (citing 435 U. S., at 757, n. 23). So, in the Court’s view, Washington Stevedoring does not support the Government’s argument that “Congress [may] impose generally applicable, nondiscriminatory taxes that fall directly on exports in transit,” ante, at 861. The Government never argues that §4371 imposes a tax on goods in transit, however. See, e. g., Brief for United States 15 (the tax imposed by §4371 “does not fall specifically on articles of export or export transactions”). If the Government can be faulted, it is for urging us to uphold §4371 on a broad theory (a tax that does not discriminate against exports is valid) rather than the narrow theory subsumed within it (this particular tax does not fall on export goods at all). Nothing in the Government’s argument prevents us from deciding the case on the narrower ground.
Even were we to suppose that the Government did not argue the goods and services distinction, the prudential rule against deciding a case on an unargued theory is in any event not absolute. See Arcadia v. Ohio Power Co., 498 U. S. 73, 77 (1990); Erie R. Co. v. Tompkins, 304 U. S. 64, 77-78 (1938) (overturning Swift v. Tyson, 16 Pet. 1 (1842), as unconstitutional); see also 304 U. S., at 82 (Butler, J.) (pointing out that no constitutional question was argued or briefed either in this Court or the court below). Cf. Evans v. United States, 504 U. S. 255, 269 (1992) (addressing a theory not argued by the parties but advanced by Justice Thomas in dissent); United States v. Burke, 504 U. S. 229, 246 (1992) (Scalia, J., concurring in judgment). This rule has less force when the issue before us is whether it is constitutional to apply the statute where Congress intended it to apply. The predicate question of whether the Export Clause prohibits taxes on distinct services like insurance is “essential to the analysis” of the question presented, Procunier v. Navarette, 434 U. S. 555, 559-560, n. 6 (1978), and necessary to “an intelligent resolution of the constitutionality” of the statute, Vance v. *868Terrazas, 444 U. S. 252, 258, n. 5 (1980). It is before us and should be decided. See this Court’s Rule 14.1(a) (“The statement of any question presented "will be deemed to comprise every subsidiary question fairly included therein”).
To give Congress the respect it is owed, we must decide whether the statute is in fact unconstitutional as applied, not make the borderline call that the Government’s litigation position bars us from reaching a question which, as the Court seems to agree, is presented by the case. In interpreting statutes, for example, we have long observed “[t]he elementary rule . . . that every reasonable construction must be resorted to, in order to save a statute from unconstitutionality.” Hooper v. California, 155 U. S. 648, 657 (1895). See also United States ex rel. Attorney General v. Delaware & Hudson Co., 213 U. S. 366, 408 (1909) (“[W]here a statute is susceptible of two constructions, by one of which grave and doubtful constitutional questions arise and by the other of which such questions are avoided, our duty is to adopt the latter”); Murray v. Schooner Charming Betsy, 2 Cranch 64, 118 (1804).
“This approach not only reflects the prudential concern that constitutional issues not be needlessly confronted, but also recognizes that Congress, like this Court, is bound by and swears an oath to uphold the Constitution. The courts will therefore not lightly assume that Congress intended to infringe constitutionally protected liberties or usurp power constitutionally forbidden it.” Edward J. DeBartolo Corp. v. Florida Gulf Coast Building & Constr. Trades Council, 485 U. S. 568, 575 (1988).
We have not considered ourselves foreclosed from adopting saving constructions the parties failed to suggest. See, e. g., Panama R. Co. v. Johnson, 264 U. S. 375, 389-391 (1924) (interpreting Jones Act to allow action to be brought in admiralty); cf. Brief for Plaintiff-in-Error 9-22 and Brief for *869Defendant-in-Error 3-12, in Panama R. Co. v. Johnson, O. T. 1923, No. 369. We cannot here avoid a constitutional question by statutory construction, but we should take all measures to avoid declaring that Congress “usurp[ed] power constitutionally forbidden it,” DeBartolo, swpra, at 575. The majority cites no case in which we have declared a federal statute unconstitutional by disregarding an unargued theory that would save the statute, and I am not aware of any. We should at least consider a construction of the Export Clause that would render it inapplicable to the statute, rather than assuming the issue away and reaching the unnecessary judgment that a coordinate branch violated the Constitution.
There may be instances, even in constitutional cases, when we should eschew alternative theories for sustaining a statute. For example, we might do so if the theories depend upon different provisions of law or require factual development and legal analysis far afield from that done by the parties or the courts below. That is not this case. The question whether the Export Clause applies to taxes on distinct export-related services requires most of the same inquiries the majority undertakes: construing the text of the Export Clause, considering its history and purpose, and reviewing our precedents. It also requires explicit reexamination of the reasoning of Thames & Mersey Marine Ins. Co. v. United States, 237 U. S. 19 (1915), which the Government has asked us to overrule, in particular the idea that a tax on insurance premiums is a tax on the goods. The last is the only step the Court refuses to take.
There is not, as the Court intimates, ante, at 855, a need for statistical development of the relative incidence of this tax on exporters, unless the Court (as appears unlikely) is interested in the statistics from 1942 to determine if the statute was a pretext when it was enacted. The current incidence of the tax on exporters, whatever it is, will reflect market conditions in light of the operation of this tax over more than 50 years, including the strength of foreign insur*870ers in certain lines exporters purchase, cf. R. Holtom, Underwriting Principles & Practices 451 (3d ed. 1987) (ocean marine insurance dominated by foreign companies). There is no law prohibiting persons from being insured under policies of foreign insurers issued abroad, and nothing in the statute exempts nonexporters from its operation. The Court has all the information it needs to decide this case on the proper basis, and it should not rest its decision that § 4371 is unconstitutional upon a dubious assumption that a general tax on insurance premiums is a tax on export goods.
In Massachusetts v. United States, 333 U. S. 611 (1948), the Government had conceded certain matters of statutory construction which, we felt, undermined its entire position. Id., at 624. We refused to accept those concessions, and, giving the statute its proper interpretation, ruled in the Government’s favor. Id., at 625. It mystifies me that in a constitutional case, where our decision is not subject to congressional revision, the Court here accepts the Government’s purported concession of the meaning of the Export Clause without any independent examination of the question, and then invokes the Clause to strike down a statute. See Torres v. Puerto Rico, 442 U. S. 465, 471, n. 3 (1979) (“[E]ven an explicit concession” by the Commonwealth of Puerto Rico that it was subject to the requirements of the Fourth Amendment would not “relieve this Court of the performance of the judicial function of deciding the issue”) (internal quotation marks omitted).
Quite apart from the unnecessary judgment that an Act of Congress is unconstitutional as applied, today’s decision adds significant complexity to the administration of §4371. Under the thumb of the Court’s holding that all premiums paid to insure export goods are exempt from § 4371, but also under the statutory mandate to collect the tax in all other instances, the Internal Revenue Service (IRS) henceforth finds itself faced with an array of new problems unexplained and unmentioned by the Court. Insurance is one of the *871most complex of businesses, with a multitude of coverage and policy options in different product lines, all generated and still evolving in pursuit of the profitable and efficient underwriting of risks. Not every case will fit the simple model here: a policy written for a single shipment; coverage beginning only with a common carrier picking up the goods from the warehouse or manufacturing plant; simple ascertainment of point of entry into the export stream. Stipulation of Facts ¶¶ 13, 16, App. to Pet. for Cert. 37a, 39a; cf. A. G. Spalding & Bros. v. Edwards, 262 U. S. 66, 68-69 (1923) (delivery to common carrier signals commencement of export).
Commercial inland marine transit insurance, the form of casualty insurance which covers domestic transportation of goods, “is usually written on an open basis, under which all shipments of the kind of merchandise described in the policy are covered.” Holtom, supra, at 435. It would appear, from today’s decision, that if a company has an open policy from a foreign insurer covering the domestic leg of the journey for all shipments, the IRS must untangle what portion of the insurance covered goods that had commenced the process of exportation, and then prorate the tax. So too would proration (or some other accommodation) appear necessary if the policy is taken out on a single shipment but part of the shipment is delivered within the country and part abroad.
In addition, the Court’s decision draws the IRS into the factual morass of determining when exportation has begun. That will often be less clear than it is here. For example, a company may have its own trucks carry goods to a freight forwarder or port, or a hiatus in the journey might be extensive enough to remove the goods from the export stream, see Joy Oil Co. v. State Tax Comm’n, 337 U. S. 286, 288-289 (1949); since “not every preliminary movement of goods toward eventual exportation” triggers the constitutional immunity, Kosydar v. National Cash Register Co., 417 U. S. 62, 69, n. 6 (1974), the determination of the commencement of *872exportation is another layer of complexity added to the administration of §4371. Finally, the IRS now must determine which of the many, ever evolving types of insurance fall within the broad prohibition of Thames & Mersey against any tax that burdens the exporting process. See 237 U. S., at 27. Truckers, for example, often take insurance out to cover liability for the loss or damage to merchandise that they are carrying. Holtom, supra, at 435. The cost of that insurance, which may be specific to an export shipment and related to the value of the goods, is likely passed through in some measure to the exporter and therefore “falls upon the exporting process,” Thames & Mersey, 237 U. S., at 27. Questions will also arise whether it violates the Export Clause to tax insurance taken out by an export freight-forwarder to cover a warehouse storing goods in transit, or to tax ocean marine protection and indemnity insurance taken out by a vessel owner to protect against damage to export cargo, cf. Holtom, supra, at 452, if part of the risk covered is domestic.
The severity of these administrative burdens will depend in part upon the penetration of the domestic market by foreign insurers in certain lines. We can anticipate increased burdens with the 4% price cut in foreign insurance for exporters that results from today’s decision. The Court is wrong to frustrate the will of Congress by giving exporters an undeserved exemption from § 4371 and by adding needless complexity to the administration of the statute, all upon the incorrect, unexamined assumption that the tax is on exported goods.
II
Turning to the question that I take to be dispositive, I would hold that the Export Clause does not apply to §4371. The text and history of the Clause, and its interpretation by the Fifth Congress, suggest that taxes on insurance do not fall within its prohibitions. Because §4371 taxes a service distinct from the actual export of the goods, and does not *873function as a proxy for taxing their value, I would uphold its application to International Business Machines Corporation (IBM).
In my view, the Framers understood the Export Clause to prohibit what its text says: any federal tax “laid on Articles exported,” U. S. Const., Art. I, § 9, cl. 5, not taxes on services like insurance that may have indirect effect on the cost of exporting. There was a history of nations’ imposing onerous taxes on exported goods, even in England until the rise of mercantilist trade policy resulted in the repeal of most export taxes by the end of the 17th century, see W. Kennedy, English Taxation 1640-1799, p. 35 (1913). And specific taxes on exported goods were the only taxes mentioned in the debate at the Constitutional Convention over the Export Clause. For example, Gouverneur Morris of Pennsylvania, opposing the Clause, favored taxing exports as an alternative to direct taxes on individuals.
“He considered the taxing of exports to be in many cases highly politic. Virginia has found her account in taxing Tobacco. All Countries having peculiar articles tax the exportation of them; as France her wines and brandies. A tax here on lumber, would fall on the W. Indies & punish their restrictions on our trade. The same is true of live-stock and in some degree of flour. In case of a dearth in the West Indies, we may extort what we please. Taxes on exports are a necessary source of revenue. For a long time the people of America will not have money to pay direct taxes. Seize and sell their effects and you push them into Revolts.” 2 M. Farrand, Records of the Federal Convention of 1787, p. 307 (rev. ed. 1966).
See also id., at 306 (Mr. Madison: taxes on exported goods, like tobacco, in which Americans were unrivalled would shift the tax burden to foreigners); id., at 360 .(Gouverneur Morris: taxes on goods are essential to embargoes, while taxes on *874ginseng and ship masts would shift the tax burden abroad, and taxes on skins, beavers, and other raw materials might encourage American manufactures); id., at 361 (Mr. Dickenson [sic]: suggesting exemption of certain articles from the Export Clause); id., at 362 (Mr. Fitzimmons: discussing duties imposed on wool by Great Britain). Proponents of the Export Clause also focused on taxes on goods. Id., at 307 (Mr. Mercer: a tax on exported goods encourages the raising of articles not meant for exportation); id., at 360 (Mr. Williamson: discussing taxation of North Carolina tobacco by Virginia); id., at 361 (Mr. Sherman: general prohibition on power to tax exports necessary because “[a]n enumeration of particular articles would be difficult invidious and improper”); id., at 363 (Colonel Mason: discussing Virginia tax on tobacco; Mr. Clymer: discussing middle States’ apprehensions of taxes on products like wheat flour and provisions that, unlike tobacco and rice, were sold in competitive markets). Oliver Ellsworth of Connecticut even contended that he opposed export taxes in part because “there are indeed but a few articles that could be taxed at all; as Tobo, rice & indigo, and a tax on these alone would be partial & unjust.” Id., at 360.
In interpreting constitutional restrictions on the taxing power, we must recall that the want of this power in the National Government was one of the great weaknesses of the Articles of Confederation. With its expenses outpacing revenues from requisitions from the States, the central Government had emptied its vaults by 1782 and soon defaulted on its substantial debt. R. Paul, Taxation in the United States 4-5 (1954). As the Convention records indicate, depriving the Federal Government of the power to tax even export goods was a contentious issue, given the concern that it would cut off a needed source of revenue as well as disable Congress from using export taxes as an instrument of policy. Madison’s last-minute proposal that the Export Clause’s total prohibition on taxing exports be replaced with a provision *875requiring a two-thirds vote of each House failed by the vote of only one State. 2 Farrand, supra, at 363. There is no cause for extending the Export Clause beyond the bargain struck at the Convention and embodied in its text.
There is other compelling historical evidence weighing against Thames & Mersey’s view of the Export Clause as a prohibition extending even to taxes on services that have the indirect effect of raising exportation costs. In 1797 the Fifth Congress passed “An Act laying Duties on stamped Vellum, Parchment and Paper.” Among its provisions was a stamp duty upon
“any policy of insurance or instrument in nature thereof, whereby any ships, vessels or goods going from one district to another in the United States, or from the United States to any foreign port or place, shall be insured, to wit, if going from one district to another in the United States, twenty-five cents; if going from the United States to any foreign port or place, when the sum for which insurance is made shall not exceed five hundred dollars, twenty-five cents; and when the sum insured shall exceed five hundred dollars, one dollar . . . .” Act of July 6, 1797, ch. 11, § 1, 1 Stat. 527.
The duties survived until the unpopular Federalist tax system, which was felt to bear too heavily upon those least able to pay, was abolished soon after Jefferson took office. See Paul, supra, at 6.
We have always been reluctant to say a statute of this early origin offends the Constitution, absent clear inconsistency. See Knowlton v. Moore, 178 U. S. 41, 56 (1900) (imposition of legacy taxes in the same 1797 statute casts doubt on claim that Congress lacks such power); see Ludecke v. Watkins, 335 U. S. 160, 171 (1948) (“The [Alien Enemy Act of 1798] is almost as old as the Constitution, and it would savor of doctrinaire audacity now to find the statute offensive to some emanation of the Bill of Rights”). The 1797 *876statute should dispel any doubt on the issue. Taxes on insurance do not offend the Export Clause. It is not likely, moreover, that the Act was passed to circumvent the Export Clause. The early Congresses were scrupulous in honoring the Export Clause by making specific exemptions for exports in laws imposing general taxes on goods. See, e. g., Act of Mar. 3,1791, ch. 15, §51,1 Stat. 199, 210-211 (tax on distilled spirits); Act of June 5, 1794, ch. 51, § 14, 1 Stat. 384, 387 (tax on snuff and refined sugar). Their refusal to grant exporters similar exemptions ..from insurance taxes indicates that those taxes were not viewed as equivalent to taxes on goods.
In Fairbank v. United States, 181 U. S. 283 (1901), the Court struck down an 1898 statute imposing a stamp tax on an export bill of lading despite a similar tax in the 1797 statute. The decision In Fairbank was 5-4, with a strong dissent from the first Justice Harlan urging deference to the implicit exposition of the Export Clause by the Fifth Congress. The Court, though, reserved the contemporaneous-exposition rule for ?‘doubtful cases,’” id., at 311, and had no doubt that the “discriminating and excessive tax” imposed on export bills of' Jading in the 1898 Act (10 times that imposed on internal T>ills of lading, id., at 290) was unconstitutional.
There is no need to reconsider Fairbank, nor to distinguish it by sole reliance upon the interpretation offered in Washington Stevedoring, which observed that the stamp duty at issue in Fairbank “effectively taxed the goods because the bills represented the goods,” 435 U. S., at 756, n. 21. The tax here, unlike the stamp duty in Fairbank, does not discriminate against exports; it taxes a service distinct from the act of exporting; and it has the clear regulatory purpose of eliminating a perceived competitive advantage of foreign insurers. Viewed in this light, the conclusion of the Fifth Congress that the Export Clause did not bar any tax on export insurance should have great weight in assessing the *877constitutionality of §4371, and FairbanJc is not to the contrary.
Turning once more to Thames & Mersey, I note the 1797 statute was neither briefed to the Court there nor discussed in its opinion. The Court, furthermore, did not examine the text or history of the Export Clause, relying instead on the broad theory of the Clause espoused in the companion case, United States v. Hvoslef, 237 U. S. 1 (1915): namely, that it meant the “process of exporting . . . should not be obstructed or hindered by any burden of taxation,” id., at 13 (internal quotation marks and citation omitted). See Thames & Mersey, 237 U. S., at 25. (Hvoslef s holding that a nondiscriminatory tax on charter parties was unconstitutional as applied to export shipments, by the way, is also called into question by the 1797 Act, which imposed a similar tax.)
Besides failing to consider the evidence just cited, the Thames & Mersey Court relied in part on the theory that insurance is not commerce and so, by implication, the regulatory aspect of the tax could not be justified as an exercise of Congress’ Commerce Clause power. See Thames & Mersey, supra, at 25, citing Paul v. Virginia, 8 Wall. 168 (1869). As a result, the Court reasoned, an insurance policy was simply a personal contract and a document which, by custom, was a necessary part of every export transaction. 237 U. S., at 25-26. A tax on the premiums of such a policy, which fell upon the exporting process and increased its costs, was thought to be the equivalent to a tax laid on charter parties, bills of lading, or the goods themselves. Id., at 27. We abandoned long ago the notion that insurance is not commerce and so beyond the power of Congress to regulate. See United States v. South-Eastern Underwriters Assn., 322 U. S. 533, 543-545 (1944). Congress enacted § 4371 to regulate competition within the insurance field, and its authority to do so ought not to be impaired by a strained reading of the Export *878Clause or reliance on the outmoded reasoning of Thames & Mersey.
We have discarded, in Import-Export Clause cases, the idea afoot in Hvoslef and Thames & Mersey that a tax on services necessary to the export process is equivalent to a tax on goods. In Canton R. Co. v. Rogan, 340 U. S. 511 (1951), the Court upheld a state gross-receipts tax on a steam railroad, even as applied to the railroad’s handling of exports and imports from its marine terminal in the port of Baltimore. The tax was “not on the goods but on the handling of them at the port,” we said, and “when the tax is on activities connected with the export or import the range of immunity cannot be so wide.” Id., at 514-515. Following Canton, the Court in Washington Stevedoring decided that taxes on services may be permissible even if levied upon an activity, such as stevedoring, which occurs while imports and exports are in transit. We remarked: “The transportation services in both settings are necessary to the import-export process. Taxation in neither setting relates to the value of the goods, and therefore in neither can it be considered taxation upon the goods themselves.” 435 U. S., at 757. The distinctions drawn between services and goods in those cases did not depend on the differences between the text of the Export and Import-Export Clauses, and should be observed here.
The Court’s effort to justify its decision on the grounds of stare decisis, ante, at 856, is unconvincing. Stare decisis does not protect a constitutional decision where the reasoning is as poor as it is in Thames & Mersey, see Smith v. Allwright, 321 U. S. 649, 665 (1944), nor when the precedent, even if not yet proved unworkable, is at odds with more recent cases, see Fulton Corp. v. Faulkner, 516 U. S. 325, 345-346 (1996). It is, moreover, just a matter of time before Thames & Mersey proves itself unworkable; prior to today, it had not been given the chance to work its mischief on § 4371.
As we move to a more service-intensive and export-oriented economy, and as policymakers and experts debate *879the wisdom of shifting from income to excise taxes, see Lugar, The National Sales Tax: Avoiding the Zero-Sum Scenario, 48 Tax Executive 26 (1996); Bartlett, Replacing Federal Taxes with a Sales Tax, 68 Tax Notes 997 (1995), we should not use shaky precedent to deprive Congress of important regulatory and revenue-raising options. As respondent conceded at oral argument, Tr. of Oral Arg. 31, the reasoning of Thames & Mersey invites claims by export service providers for exemptions from any number of federal excise taxes, for example, a challenge to the diesel-fuel tax, 26 U. S. C. § 4041, by truckers carrying export shipments. The Export Clause cannot bear this reading.
The protections of the Export Clause must extend, perhaps, somewhat beyond specific taxes on goods, for “[i]f it meant no more than that, the obstructions to exportation which it was the purpose to prevent could readily be set up by legislation nominally conforming to the constitutional restriction but in effect overriding it.” Hvoslef, supra, at 13. As a result, the Court has found certain taxes to be proxies for taxes on the goods. See Washington Stevedoring, supra, at 756, n. 21 (discussing sales tax struck down in Richfield Oil Corp. v. State Bd. of Equalization, 329 U. S. 69 (1946), and the tax on a bill of lading struck down in Fairbank v. United States, 181 U. S. 283 (1901)). In Washington Stevedoring, we expressed some doubt that the tax on insurance in Thames & Mersey fell in this forbidden category, but, to avoid overruling the case, “note[d] that the value of goods bears a much closer relation to the value of insurance policies on them than to the value of loading and unloading ships.” 435 U. S., at 756, n. 21.
The insurance premiums taxed here, like those taxed in Thames & Mersey, bear some relation to the value of the goods, but this does not make them a proxy for a tax on the goods. Premiums, i. e., the price of insurance, depend on risk of loss, and value of the goods is only one component factor of risk. So much is made clear by Stipulation 16 in *880this case. Before the premiums for a shipment of IBM goods of a certain value could be fixed, a premium rate had to be determined. The rate was a function of the risk factors specific to a particular shipment: “the place of origin and destination of the goods, the type of goods involved and how they were packaged, the time and distance of the trip, the route and mode(s) of transportation, and the amount of material handling expected during the trip.” Premiums were then determined by multiplying the value of the goods by the shipment-specific premium rate. Stipulation of Facts ¶ 16, App. to Pet. for Cert. 39a. Cf. Holtom, Underwriting Principles & Practices, at 453-457 (discussing various factors taken into account in underwriting ocean marine insurance, such as nationality of the crew, vessel management, seaworthiness of the vessel, suitability of the vessel for specific cargo, packaging, season of travel, perishability, pilferage risks at ports of call, and risks of damage from accompanying cargo). The premium charged to insure a million dollars of goods for the short overland journey from IBM’s computer factory in Richfield, Minnesota, to a customer in Quebec would be trifling in comparison to the premium charged to insure transport of goods of equivalent value from its factory in San Jose, California, across the continent east to New York and then by sea to Russia. Cf. Stipulation of Facts, App. to Pet. for Cert. 36a-37a; Brief for Respondent 3, n. 2. Given the stipulated, undeniable premise that premiums are graded by risk of loss, they are not a predictable proxy for a Congress intent upon taxing export value. Premiums are a rough proxy, however, for the income of foreign insurers, which is why a Congress intent on eliminating the income tax advantages of those insurers would structure §4371 as it did.
Section 4371’s requirement that the insurance cover domestic risks in whole or in part is further evidence that Congress did not intend it to operate as a proxy for taxing exports. A statute that exempts all exporters who use a *881domestic insurer for the inland leg of a shipment is not an effective instrument for taxing export goods.
I would uphold § 4371 as applied to IBM because the statute imposes a tax on a distinct export-related service and is not a proxy for a tax on the exports themselves. The Court, in my view, makes a serious mistake in assuming the opposite and reaching the question whether a nondiscriminatory tax on goods violates the Export Clause. I would reverse the judgment of the Court of Appeals for the Federal Circuit.