Trinova Corp. v. Michigan Department of Treasury

Justice Kennedy

delivered the opinion of the Court.

The principal question before us is whether the three-factor apportionment formula of the Michigan single business tax (SBT), Mich. Comp. Laws §208.1 et seq. (1979), violates either the Due Process Clause or the Commerce Clause of the Federal Constitution. The applicability of a three-factor formula to a state income tax is well settled, but we have not considered whether a similar apportionment formula may be applied to a value added tax (VAT). We granted certiorari to consider this question and to determine whether the Michigan SBT discriminates against out-of-state businesses.

*362 r — <

Although in Europe and Latin America VAT s are common, see Lindholm, The Origin of the Value-Added Tax, 6 J. Corp. L. 11 (1980); Due, Economics of the Value Added Tax, 6 J. Corp. L. 61 (1980), in the United States they are much studied but little used. Michigan is the first and, the parties tell us, the only State to have enacted a VAT as a tax on business activity. We begin with a description of value added and VAT’s in general, and then discuss the Michigan SBT.

A

Value added is an economic concept. “Value added is defined as the increase in the value of goods and services brought about by whatever a business does to them between the time of purchase and the time of sale.” Haughey, The Economic Logic of the Single Business Tax, 22 Wayne L. Rev. 1017, 1018 (1976) (hereinafter Haughey). The value a business adds to a single product is “the difference between the value of the product at sale and the cost of goods purchased from other businesses that went into the product.” Taxation and Economic Policy Office, Michigan Department of Treasury, Analysis of the Michigan Single Business Tax 20-21 (1985) (hereinafter SBT Analysis). It follows that the sale price of a product is the total of all value added by each step of the production process to that point. “The value added of a loaf of bread is the sum of the value contributed at each stage of the production and distribution process. Among others, it includes the contribution of the farmer, miller, baker, wholesaler and retailer.” Haughey 1019.

A business “adds value by handling or processing these [goods] with its labor force, machinery, buildings and capital.” R. Kleine, Advisory Commission on Intergovernmental Relations, The Michigan Single Business Tax: A Different Approach to State Business Taxation 1 (1978) (hereinafter Kleine). In this litigation, value added usually refers to the *363 activities of a single business enterprise. The term can, however, be used with regard to a single product, or even an entire economy. “[Value added] is a means of consistently measuring the size of business firms and other economic enterprises comprising the total economy .... Gross National Product is virtually equivalent to national value added.” Haughey 1017.

One of the acknowledged advantages of value added as a measure of taxation is its neutrality. A VAT is neutral in the sense that it taxes all business activity alike: Under a pure VAT, all forms of business organization (corporation, partnership, proprietorship), all types of financing (debt, equity) and all methods of production (capital intensive, labor intensive) bear the same tax burden.

“[T]ax factors are minimized in business decisions; inherent advantages and relative efficiencies are allowed to operate in the market economy with minimum tax distortions.

“This neutrality of a value-added tax is in notable contrast to the effects of both the corporation income tax and the payroll taxes. The former, by definition, is applied only to corporations and varies with their reliance on equity rather than debt capital and the efficiency with which they use equity capital — that is, their net profits.” Smith, Value-added tax: the case for, 48 Harv. Bus. Rev. 77, 79 (Nov.-Dec. 1970).

Though neutral in theory, VAT’s often depart in practice from the pure value added model because of special exemptions, deductions, and other adjustments. These features can eliminate much of the claim to neutrality. See generally The Value-Added Tax: Lessons from Europe (H. Aaron ed. 1981).

A VAT differs in important respects from a corporate income tax. A corporate income tax is based on the philosophy of ability to pay, as it consists of some portion of the profit remaining after a company has provided for its work *364 ers, suppliers, and other creditors. A VAT, on the other hand, is a much broader measure of a firm’s total business activity. Even if a business entity is unprofitable, under normal circumstances it adds value to its products and, as a consequence, will owe some VAT. Because value added is a measure of actual business activity, a VAT correlates more closely to the volume of governmental services received by the taxpayer than does an income tax. Further, because value added does not fluctuate as widely as net income, a VAT provides a more stable source of revenue than the corporate income tax. See generally Kleine 3, figure 1. “‘The logic or rationale of the [VAT] rests squarely on the benefits received principle of taxation — government services are essential to the operation of any business enterprise . . . and a part of these public service costs should properly be included in the cost of doing business.’” Id., at 4 (citation omitted).

The SBT Analysis, at 20-21, provides us with the following simplified example of how value added is determined. Assume a bakery’s sole revenue comes from the sale of bread. The bakery’s costs consist of materials (flour, sugar, spices, utilities), labor (baker, sales clerk), capital (building, mixer, utensils, oven), and credit (interest paid on loans). Any excess of revenues over costs represents profit. Thus:

Revenues = Cost of Labor + Cost of Materials + Depreciation1 + Interest + Profit.

Because value added is defined as the difference between the value of products sold (revenues), and the cost of materials going into the products, we can represent value added (for the entire firm) by a second simple equation:

*365 Value added = Revenues — Cost of Materials.

The same result is reached by another common method. If we subtract Cost of Materials from each side of the first equation above, we have:

Revenues - Cost of Materials = Cost of Labor + Depreciation + Interest + Profit.

So in practice value added can be calculated as either Revenues - Cost of Materials; or Cost of Labor + Depreciation + Interest + Profit. Not surprisingly, these are referred to as the “subtraction” and the “addition” methods. Each provides an identical measurement of a taxpayer’s value added.2 Once value added is determined, the VAT is assessed as a percentage of the value added for the relevant fiscal period.3

*366 B

The Michigan SBT went into effect on January 1, 1976. 1975 Mich. Pub. Acts 228.4 The SBT replaced seven different business taxes. Kleine 22; Brief for Respondent 8. Before 1976, a typical manufacturer with business activity in *367 Michigan would have been subject to a franchise tax, an income tax, an intangible property tax, and an ad valorem property tax upon inventories. Mitchell, Taxes Repealed and Amended, 22 Wayne L. Rev. 1029 (1976); Brief for Respondent 8-9. After enactment of the SBT, the same manufacturer would pay only one tax.

The Michigan SBT is an addition method VAT, although it inevitably permits various exclusions, exemptions, and adjustments that depart from the simple value added examples described above. Subject to exemptions contained at Mich. Comp. Laws §208.35 (1979), the Michigan SBT is levied against any person with “business activity” within the State of Michigan. §208.31(1).5 In order to calculate the amount of a taxpayer’s SBT the taxpayer must, first, determine its total tax base. The total tax base consists of the taxpayer’s value added, calculated by the addition method: Cost of Labor + Depreciation + Interest + Profit. Under §208.9, the taxpayer begins with federal taxable income (representing profit), adds other elements that reflect consumption of labor and capital including compensation, depreciation, dividends, and interest paid by the taxpayer, and makes other detailed adjustments.

Second, if a taxpayer does business both within and without Michigan, it must determine the portion of its total value added attributable to Michigan. That portion, the crux of this case, is the average of three ratios: (1) Michigan payroll *368 to total payroll, (2) Michigan property to total property, and (3) Michigan sales to total sales. §§208.45, 208.46, 208.49, 208.51. The total tax base is multiplied by the portion of business activity attributable to Michigan (under the three-factor formula), and the result, subject to several further adjustments, is the taxpayer’s “adjusted tax base.” §208.31(2).

Two further adjustments are relevant here: § 208.23(a), which permits a taxpayer to deduct a portion of its capital acquisitions, and §208.31(5), which permits a labor-intensive taxpayer to reduce its adjusted tax base by a percentage equal to the percentage by which compensation exceeds 63% of the total tax base, but with such reduction not to exceed a maximum of 37%. Actual tax liability equals the adjusted tax base multiplied by a tax rate of 2.35%.6

II

Trinova, an Ohio corporation, manufactures automobile components. Its principal office is located in Maumee, Ohio, a suburb of Toledo located near the Michigan border. During 1980, the tax year in question, Trinova maintained a fixed presence in Michigan: a sales office of 14 employees who solicited orders, maintained contact with Trinova’s Michigan customers, and performed clerical work. Michigan, with its automobile industry, was a major market for Trinova’s products. Indeed, Trinova made $103,981,354 worth of sales to Michigan during 1980, 26.5892% of its total sales of $391,065,866. Trinova calculated its 1980 SBT adjusted tax base as follows:

*369 U. S. taxable income (loss) ($42,466,114)

Add:

Compensation $226,356,271

Depreciation $23,262,909

Dividends, interest, and royalties paid $22,908,950

Other $549,526

Subtotal $230,611,542

Subtract:

Dividends, interest, and royalties received ($9,486,223)

Total Tax Base $221,125,319

Apportionment:

Payroll Factor 0.2328%

Property Factor 0.0930%

Sales Factor 26.5892%

Average Factor 8.9717%

Apportioned Tax Base:

$221,125,319

x 8.9717%

= $19,838,700

See 433 Mich. 141, 150-152, 445 N. W. 2d 428, 431-433 (1989). Trinova further adjusted its tax base by subtracting a capital acquisition deduction ($9,063) and by taking the maximum (37%) reduction for labor-intensive taxpayers. These adjustments resulted in a 1980 adjusted tax base of $12,492,671. When multiplied by the tax rate of 2.35%, Trinova’s tax liability amounted to $293,578 ($12,492,671 x 2.35%).7 Trinova timely filed its return and paid its tax liability.

*370 In 1985, a Michigan intermediate Court of Appeals ruled that taxpayers similarly situated to Trinova were entitled to “relief” under Mich. Comp. Laws §208.69 (1979), a provision of the SBT. Jones & Laughlin Steel Corp. v. Department of Treasury, 145 Mich. App. 405, 377 N. W. 2d 397, leave to appeal and reconsideration denied, 424 Mich. 895 (1986). At the time, §208.69 provided that if the apportionment provisions of the SBT did not “fairly represent the extent of the taxpayer’s business activity” in Michigan, the taxpayer could, among other alternatives, petition for the employment of “any other method to effectuate an equitable allocation and apportionment of the taxpayer’s tax base.”

Soon after the decision in Jones & Laughlin, Trinova filed an amended return and refund claim for the 1980 tax year. Based on the relief granted in Jones & Laughlin, Trinova proposed that despite admitted company-wide value added of $221 million and Michigan sales of over $100 million, for purposes of the Michigan SBT it should be treated as if it had negative total value added. Value added apportioned to Michigan would also have been negative, and Trinova would have been entitled to a refund for its entire 1980 SBT payment.8 Upon denial of relief by the Michigan Department of *371 Treasury, Trinova sued for a refund in the Michigan Court of Claims, which ruled in Trinova’s favor on the authority of Jones & Laughlin. No. 86-10430-CM (May 5, 1987); App. to Pet. for Cert. 42a-51a.

While the Department of Treasury’s appeal was pending in the Michigan Court of Appeals, the legislature amended §208.69. 1987 Mich. Pub. Acts 39. The amended §208.69 creates a presumption that the statutory apportionment formula fairly represents the taxpayer’s business activity in Michigan unless the adjusted tax base meets one of two tests, neither of which Trinova could satisfy, and which do not merit discussion here. See Mich. Comp. Laws Ann. §208.69(3) (West Supp. 1990). The Court of Appeals referred to the legislature’s statement that its Act was intended to be

“curative, expressing the original intent of the legislature that the single business tax ... is an indivisible value added type of tax and not a combination or series of several smaller taxes and that relief from formulary apportionment should be granted only under extraordinary circumstances.” 1987 Mich. Pub. Acts 39, §2.

Relying upon this language, the Court of Appeals determined that the amendment was to be given retroactive effect as a “remedial and procedural” statute and that Trinova was not entitled to statutory relief. 166 Mich. App. 656, 666, 421 N. W. 2d 258, 262 (1988).

The Michigan Supreme Court affirmed the Court of Appeals. 433 Mich. 141, 445 N. W. 2d 428 (1989). Without addressing retroactive application of the amendments to §208.69, it construed §208.69 as a “constitutional ‘circuit *372 breaker’ ” to be applied only if required in order to save the SBT against unconstitutional application. Id., at 156, 445 N. W. 2d, at 434. The court then upheld the SBT against Trinova’s federal constitutional challenges. The Michigan Supreme Court noted that formulary apportionment of income taxes is uncontroversial and that it did “not believe that ‘business activity’ as defined under the [SBT] is susceptible to accurate analysis when only one component of the total business effort is examined.” Id., at 163, 445 N. W. 2d, at 438. The court concluded that Trinova’s averaged ratios of payroll, property, and sales are a fair representation of the extent of its business activity in Michigan, making it ineligible for relief on statutory or constitutional grounds. Id., at 163-166, 445 N. W. 2d, at 438-439. We granted Trinova’s petition for a writ of certiorari. 494 U. S. 1015 (1990).

1 — i I — 1 I — I

The principles which govern the validity of state taxes levied upon multistate businesses seek to accommodate the necessary abstractions of tax theory to the realities of the marketplace. Under the test stated in Complete Auto Transit, Inc. v. Brady, 430 U. S. 274, 279 (1977), we will sustain a tax against Commerce Clause challenge so long as “the tax is applied to an activity with a substantial nexus with the taxing State, is fairly apportioned, does not discriminate against interstate commerce, and is fairly related to the services provided. by the State.” We applied this four-part test in later cases addressing a wide variety of taxes. See Goldberg v. Sweet, 488 U. S. 252, 260, n. 12 (1989) (citing applications in cases involving sales, severance, use, corporate income, and business and occupation taxes).

In Complete Auto, we renounced the formalistic approach of Spector Motor Service, Inc. v. O’Connor, 340 U. S. 602 (1951), which had prohibited a State from taxing the privilege of doing business in the State, treating it as a tax upon interstate commerce and so beyond the authority of the *373 State. We seek to avoid formalism and to rely upon a “consistent and rational method of inquiry [focusing ón] the practical effect of a challenged tax.” Mobil Oil Corp. v. Commissioner of Taxes of Vt., 445 U. S. 425, 443 (1980). The Complete Auto test, while responsive to Commerce Clause dictates, encompasses as well the due process requirement that there be “a ‘minimal connection’ between the interstate activities and the taxing State, and a rational relationship between the income attributed to the State and the intrastate values of the enterprise.” Mobil Oil Corp., supra, at 436- 437; see also Amerada Hess Corp. v. Director, Div. of Taxation, N. J. Dept. of Treasury, 490 U. S. 66, 80 (1989) (Scalia, J., concurring).

In this Court, Trinova does not dispute that its business activities have a substantial nexus with Michigan and subject it to the State’s taxing authority. Nor does Trinova argue that the amount of tax it is required to pay bears no fair relation to the services provided by the State. Complete Auto, supra, at 279. Trinova instead contends that Michigan’s SBT fails the other two prongs of the Complete Auto test: that the SBT is not fairly apportioned as applied to Trinova and that the tax discriminates against interstate commerce. We consider these claims and begin with the matter of apportionment.

A

Trinova’s claim that apportionment of the tax is unconstitutional concentrates on the elements of the apportionment formula. The original rationale for apportionment of income was the difficulty of identifying the geographic source of the income earned by a multistate enterprise. See Underwood Typewriter Co. v. Chamberlain, 254 U. S. 113, 120-121 (1920) (legislature “faced with the impossibility of allocating specifically the profits earned by the [taxpayer’s] processes conducted within its borders”). As we stated the problem in Container Corp. of America v. Franchise Tax Bd., 463 U. S. 159, 192 (1983): “Allocating income among various taxing ju *374 risdictions bears some resemblance ... to slicing a shadow.” Trinova argues that because its SBT tax base is composed in large part of compensation and depreciation, elements which can be assigned to a geographic source, we must reject apportionment altogether.

We can accept the premise that apportionment is permitted only when precise geographic measurement is not feasible, for to allow apportionment where there is no practical or theoretical justification could provide the opportunity for a State to export tax burdens and import tax revenues. The Commerce Clause prohibits this competitive mischief. The issue becomes whether, without an apportionment formula, Michigan can assign the SBT tax base and its principal components to separate geographic locations and to separate accounts in each State. Michigan has decided it cannot do so without serious theoretical and practical difficulty, and upon review of the case we accept that determination.

We reject at the outset, however, arguments by Michigan and some amici curiae that the Michigan SBT can be analyzed as a tax upon “business activity.” Brief for Council of State Governments et al. as Amici Curiae 11. The statute does not say that the SBT is a tax upon business activity, but rather that it is a “tax of 2.35% upon the adjusted tax base of every person with business activity in this state which is allocated or apportioned to this state.” Mich. Comp. Laws §208.31(1) (1979) (emphasis added). While Michigan business activity is a threshold requirement for the tax, and value added is its measure, labeling the SBT a tax on “business activity” does not permit us to forgo examination of the actual tax base and apportionment provisions. “A tax on sleeping measured by the number of pairs of shoes you have in your closet is a tax on shoes.” Jenkins, State Taxation of Interstate Commerce, 27 Tenn. L. Rev. 239, 242 (1960).

Trinova errs in the opposite direction. It would dissect the tax base as if the SBT were three separate and independent taxes: a tax on compensation, a tax on depreciation, and a *375 tax on income, each apportioned. Trinova insists that compensation and depreciation can be located and can be separated from the total value added calculation. As a result, Trinova would be taxed upon its Michigan compensation and Michigan depreciation. It would owe no additional tax upon income apportionable to Michigan, because it had no income during the relevant tax year.

This characterization, and with it Trinova’s constitutional argument, fails. Doubtless Trinova can identify the location of its plant and equipment and much of its compensation. The Michigan SBT, however, is not three separate and independent taxes, and Trinova cannot purport to identify the geographic source of value added by assuming that two elements can be located in a single State while the third cannot. Trinova’s proposed apportionment for the 1980 tax year, n. 8, supra, provides a good example of the problems that accompany its argument.

In 1980, Trinova’s company-wide value added amounted to much less than its compensation plus depreciation. In short, Trinova was unprofitable. Under a VAT, however, tax becomes due in any event. Trinova’s approach would require us to conclude that Trinova added value at the factory through the consumption of capital and labor, but that its products somehow lost value outside of this process, perhaps between the time they left the factory and the time they were delivered to customers in Michigan. This approach is incompatible with the rationale of a VAT and is unsupported in the record.

For all this record shows, Trinova’s production operations might have added little value and its sales offices might have added significant value, through superior marketing skill, liaison between the company and its customers, or mere fortuity. See Moorman Mfg. Co. v. Bair, 437 U. S. 267, 272 (1978) (record lacked analysis of what portion of profits was apportionable to sales, to manufacturing, or to other phase of company’s operations).

*376 But we need not rely upon Trinova’s 14 Michigan sales personnel as the source of all the value added that can be apportioned fairly to Michigan. In a unitary enterprise, compensation, depreciation, and profit are not independent variables to be adjusted without reference to each other. If Trinova had paid an additional $100 million in compensation during 1980, there is no way of knowing whether, or to what extent, value added would have increased. In fact, value added would not have increased so long as revenues did not increase. These elements of value added are inextricable, codependent variables.

Without Trinova’s $100 million in 1980 Michigan sales, the company’s value added would have been lower to a remarkable degree. The market demand that sustained those sales did not arise solely, perhaps not even substantially, from the activities of Trinova’s 14 Michigan sales personnel. But there can be little doubt that requirements of the Michigan market determined the direction of Trinova’s design, production, and distribution process. By serving that market and meeting its demands, Trinova generated value added in the sums that it did. We can and must assume that Michigan sales were a part of the company’s essential economic strategies and were an integral part of company-wide value added. It distorts the tax both in application and theory to confine value added consequences of the Michigan market solely to the labor and capital expended by the resident sales force.

Trinova’s attempted characterization is arguable only because Michigan calculates value added by the addition method. The addition and subtraction methods of calculating value, however, are but two different paths to the same result. See n. 2, supra. Had Michigan calculated the SBT tax base by the subtraction method, reporting total revenues minus total cost of materials, Trinova’s characterization would collapse of its own weight. Trinova could geographically locate its revenues and even determine where it purchased its materials. The Michigan apportionment formula *377 assumes as much. But were Trinova to calculate value added based upon the location of its revenues, it would apportion a much greater share of its value added to Michigan (26.5892%) than was apportioned under Michigan’s three-factor formula (8.9717%). An apportionment of value added based solely on the source of revenues is no less justifiable than an apportionment based solely upon the location of compensation or depreciation.

The difference between the addition and subtraction methods is one of form and lacks constitutional significance. Michigan chose the addition method of calculating value added as a convenience to taxpayers, for whom federal taxable income provided an easy starting point. Kleine 6-7 (discussing advantages of addition method); SBT Analysis 21 (same). The Constitution does not require a formalistic analysis resulting in a penalty for Michigan’s selection of an easier calculation method for its taxpayers.

Both methods of calculation, moreover, illustrate the justification for the State’s adoption of an apportionment formula. Under either method, value added includes a remainder or residual that cannot be located with economic precision. Under the addition method, value added contains the element of income, one calculated by and dependent upon factors (revenues minus total costs) not included in the addition method equation; under the subtraction method, value added is itself a remainder, no more assignable than income. It would be impractical to locate value added by a geographic test. We thus agree with the Michigan Legislature’s statement that the SBT is not, for apportionment purposes, “a combination or series of several smaller taxes,” 1987 Mich. Pub. Acts 39, § 2, but an “indivisible,” ibid., tax upon a different, bona fide measure of business activity, the value added.

This conclusion is no different from the one we have reached in upholding the validity of state apportionment of income taxes. As with a VAT, the discrete components of a state income tax may appear in isolation susceptible of geo *378 graphic designation. Nevertheless, since Underwood Typewriter Co. v. Chamberlain, 254 U. S. 113 (1920), we have recognized the impracticability of assuming that all income can be assigned to a single source. In this respect, Trinova’s argument becomes a familiar and often rejected genre of taxpayer challenge:

“[A]pportionability often has been challenged by the contention that. . . the source of [particular] income may be ascertained by separate geographical accounting. The Court has rejected that contention so long as the intrastate and extrastate activities formed part of a single unitary business. See Butler Bros. v. McColgan, 315 U. S. 501, 506-508 (1942); Ford Motor Co. v. Beauchamp, 308 U. S. 331, 336 (1939); cf. Moorman Mfg. Co. v. Bair, 437 U. S., at 272. In these circumstances, the Court has noted that separate accounting, while it purports to isolate portions of income received in various States, may fail to account for contributions to income resulting from functional integration, centralization of management, and economies of scale. Butler Bros. v. McColgan, 315 U. S., at 508-509. Because these factors of profitability arise from the operation of the business as a whole, it becomes misleading to characterize the income of the business as having a single identifiable ‘source.’ Although separate geographical accounting may be useful for internal auditing, for purposes of state taxation it is not constitutionally required.” Mobil Oil Corp., 445 U. S., at 438.

In a recent challenge to this unitary business principle, we rejected the argument that particular assignable costs of a business should be excluded from a broader tax base. Amerada Hess Corp. v. Director, Div. of Taxation, N. J. Dept. of Treasury, 490 U. S. 66 (1989). We considered the New Jersey corporate income tax, which used federal taxable income as a benchmark and required certain adjustments (as does the Michigan SBT). New Jersey required oil companies to *379 add back into income any deduction taken for taxes paid under the federal windfall profits tax. The taxpayers objected that the windfall profits tax is “an exclusively out-of-state expense because it is associated with the production of oil outside New Jersey.” Id., at 74.

In like manner, Trinova objects to the SBT’s requirement that it add compensation and depreciation to federal taxable income on the grounds that these are, with limited exception, out-of-state expenses. In Amerada Hess Corp. we rejected outright the idea that geographically assignable costs of production must be excluded from an apportionment of income:

“[J]ust as each [taxpayer’s] oil-producing revenue — as part of a unitary business — is not confined to a single State, Exxon Corp., 447 U. S., at 226, ... so too the costs of producing this revenue are unitary in nature. See Container Corp., 463 U. S., at 182 (the costs of a unitary business cannot be deemed confined to the locality in which they are incurred).” Ibid.

The reasoning of Amerada Hess Corp. applies with equal force to the case here. The same factors that prevent determination of the geographic location where income is generated, factors such as functional integration, centralization of management, and economies of scale, make it impossible to determine the location of value added with exact precision. In concluding that Michigan can apportion the SBT, we merely reaffirm what we have written before: “In the case of a more-or-less integrated business enterprise operating in more than one State, . . . arriving at precise territorial allocations of ‘value’ is often an elusive goal, both in theory and in practice.” Container Corp., 463 U. S., at 164.

B

Having determined that Michigan’s SBT attempts to tax a base that cannot be assigned to one location with any precision, and that apportionment is proper, we must next *380 consider whether Michigan’s apportionment formula for Trinova’s value added is fair.

Container Corp. states our test for fair income apportionment:

“The first, and again obvious, component of fairness in an apportionment formula is what might be called internal consistency — that is, the formula must be such that, if applied by every jurisdiction, it would result in no more than all of the unitary business’ income being taxed. The second and more difficult requirement is what might be called external consistency — the factor or factors used in the apportionment formula must actually reflect a reasonable sense of how income is generated.” Id., at 169.

Trinova does not contest the internal consistency of the SBT’s apportionment formula, and we need not consider that question.

Instead, Trinova argues that the SBT apportionment formula fails the external consistency test. In order to prevail on such a challenge, an income taxpayer must prove “by ‘clear and cogent evidence’ that the income attributed to the State is in fact ‘out of all appropriate proportions to the business transacted ... in that State,’ [Hans Rees’ Sons, Inc.,] 283 U. S., at 135, or has ‘led to a grossly distorted result,’ [Norfolk & Western R. Co.,] 390 U. S., at 326.” Moorman Mfg. Co., 437 U. S., at 274. We conclude that the same test applies to apportionment of a VAT. Trinova must demonstrate that, in the context of a VAT, there is no rational relationship between the tax base measure attributed to the State and the contribution of Michigan business activity to the entire value added process. See Container Corp., supra, at 180-181.

The Michigan SBT uses the same three-factor apportionment formula we first approved for apportionment of income in Butler Brothers v. McColgan, 315 U. S. 501 (1942). This standard has become “something of a benchmark against *381 which other apportionment formulas are judged.” Container Corp., supra, at 170; see also Moorman Mfg. Co., supra, at 282 (Blackmun, J., dissenting); id., at 283-284 (Powell, J., dissenting). Although the one-third weight given to each of the three factors — payroll, property, and sales — is not a precise apportionment for every case, the formula “has gained wide approval precisely because payroll, property, and sales appear in combination to reflect a very large share of the activities by which value is generated.” Container Corp., supra, at 183 (emphasis added). The three-factor formula is widely used and is included in the Uniform Division of Income for Tax Purposes Act, 7A U. L. A. 331 (1990 Cum. Supp.) (approved in 1957 by the National Conference of Commissioners on Uniform State Laws and the American Bar Association).

Trinova argues that on the facts of this case, the three-factor formula leads to a distorted result, out of all proportion to the business done by Trinova in Michigan. Trinova’s Michigan payroll constituted 0.2328% of total payroll, its Michigan property constituted 0.0930% of total property, and its Michigan sales constituted 26.5892% of total sales. The three-factor formula averages these ratios, with the result that 8.9717% of Trinova’s value added, or $19,838,700, is assigned to Michigan. Because Trinova is a labor-intensive taxpayer, and can deduct capital acquisitions, the tax base is further reduced to $12,492,671.

In this Court, Trinova proposes an alternative two-factor apportionment, excluding the sales factor. Under the two-factor formula, only 0.1629% of Trinova’s value added, or $360,213, would be assigned to Michigan. Brief for Petitioner 33-34.9

*382 Although the three-factor formula “can be justified as a rough, practical approximation of the distribution of either a corporation’s sources of income or the social costs which it generates,” General Motors Corp. v. District of Columbia, 380 U. S. 553, 561 (1965), Trinova argues that the formula does not reflect how the value added tax base is generated. The principal flaw, it contends, is that the formula includes a sales factor. “Sales have no relationship to, and add nothing to, the value that [compensation and depreciable plant] contribute to the tax base in Michigan.” Brief for Petitioner 31. Trinova’s position finds some support among economists. See Barlow & Connell, The Single Business Tax, in Michigan’s Fiscal and Economic Structure 673, 704 (H. Brazer ed. 1982); Kleine 7, 14, n. 5.

We have, supra, at 376, already concluded that sales (as a measure of market demand) do have a profound impact upon the amount of an enterprise’s value added, and therefore reject the complete exclusion of sales as somehow resulting in more accurate apportionment. We further reject this critique because it cannot distinguish application of the three-factor formula to a VAT from application to an income tax. In fact, nearly identical criticisms were levied against the three-factor formula as a method for apportioning income by economists who theorize that income (like value added) is the product of labor and capital, and that the marketplace contributes nothing to production of income. See Studenski, The Need for Federal Curbs on State Taxes on Interstate Commerce: An Economist’s Viewpoint, 46 Va. L. Rev. 1121, 1131-1132 (1960); Harriss, Economic Aspects of Interstate Apportionment of Business Income, 37 Taxes 327, 362-363 (1959); Harriss, Interstate Apportionment of Business Income, 49 Am. Econ. Rev. 398, 400 (1959). If it were not for their age, these criticisms could have been taken almost verbatim from Trinova’s brief:

*383 “[S]ales-by-destination are not a proper allocation factor ... . Taken by themselves, they do not necessarily represent the location of the company’s productive income-creating effort. Only the location of the company’s capital and labor, which may be wholly different from the destination of the sales, identifies the location of that effort and hence the situs for the imposition of a state income tax upon it.” Studenski, supra, at 1131— 1132.

Despite such criticism, the Uniform Division of Income for Tax Purposes Act decided upon an income apportionment formula that included sales, and the importance of sales in generating value has been acknowledged by this Court. Container Corp., 463 U. S., at 183. Thus, as we responded to a similar argument in Moorman Mfg. Co., whatever the merit of Trinova’s argument that sales do not contribute to value added “from the standpoint of economic theory or legislative policy, it cannot support a claim in this litigation that [the State] in fact taxed profits not attributable to activities within the State during the yea[r 1980].” 437 U. S., at 272. Trinova gives no basis for distinguishing the same arguments that were pressed, and rejected, with regard to the apportionment of income. We could not accept Trinova’s argument that the sales factor distorts Michigan’s apportionment formula without rejecting our precedents which approve the use of the same formula to apportion income.

As we find no distortion caused by the three-factor formula, it follows that the Michigan SBT does not tax “value earned outside [Michigan’s] borders.” ASARCO Inc. v. Idaho Tax Comm’n, 458 U. S. 307, 315 (1982). The argument that the value was added in Ohio, by labor and capital, and that no value has been added in Michigan assumes that value added is subject to geographic ascertainment and assumes further the inappropriateness of a sales factor in apportionment. For the reasons we have given, we reject both arguments.

*384 We need not say for certain which method — unadjusted apportionment by the three-factor formula ($19,838,700), apportionment by Trinova’s alternative two-factor formula ($360,213), Trinova’s Jones & Laughlin apportionment urged in state court (-$2,042,458), or the adjusted tax base as calculated in Trinova’s original 1980 return ($12,492,671)— gives the most accurate calculation of Trinova’s value added in Michigan. Trinova has not convincingly demonstrated which figure is most accurate. Trinova gives no estimate of the value added that would take account of both its Michigan sales activity and Michigan market demand for its products. Michigan, on the other hand, has consistently applied a formula, the elements of which appear to reflect a very large share of the activities by which value is generated, with further relief for labor-intensive taxpayers such as Trinova. Trinova has failed to meet its burden of proving “by ‘clear and cogent evidence,’” Moorman Mfg. Co., supra, at 274, that the State of Michigan’s apportionment provides a distorted result.10

C

Trinova also urges that the Michigan SBT should be struck down because it discriminates against out-of-state businesses in violation of the Commerce Clause. Trinova cannot point to any treatment of in-state and out-of-state firms that is discriminatory on its face, as in the cases it cites. See, e. g., *385 Westinghouse Electric Corp. v. Tully, 466 U. S. 388, 393 (1984) (tax credit was limited to gross receipts from export products shipped from a regular place of business of the taxpayer within New York); Boston Stock Exchange v. State Tax Comm’n, 429 U. S. 318, 324-328 (1977) (tax facially discriminated against transactions on securities exchanges located outside of New York and had been enacted in an effort to discourage growth of such exchanges); Halliburton Oil Well Cementing Co. v. Reily, 373 U. S. 64 (1963) (sales tax exempted isolated sales within State, but use tax lacked a similar exemption for similar isolated sales outside of the State).

In the absence of any facial discrimination, Trinova recalls our statement in American Trucking Assns., Inc. v. Scheiner, 483 U. S. 266, 281 (1987), that “the Commerce Clause has a deeper meaning that may be implicated even though state provisions ... do not allocate tax burdens between insiders and outsiders in a manner that is facially discriminatory.” The Commerce Clause requires more than mere facial neutrality. The content of that requirement is fair apportionment. The “deeper meaning” to which American Trucking refers is embodied in the requirement of fair apportionment, as expressed in the tests of internal and external consistency. Other than the vague accusation of discrimination, Trinova presents no other standard by which we might consider the constitutionality of the Michigan SBT.

In further support of its discrimination argument, Trinova relies upon the 1987 statement of Michigan’s Governor that the SBT was enacted “‘to promote the development and investment of business within Michigan.’” Executive Message of Governor James J. Blanchard to the Michigan Supreme Court, Nov. 6, 1987, App. to Pet. for Cert. 73a. This statement helps Trinova not at all. It is a laudatory goal in the design of a tax system to promote investment that will provide jobs and prosperity to the citizens of the taxing State. States are free to “structur[e] their tax systems to *386 encourage the growth and development of intrastate commerce and industry.” Boston Stock Exchange, swpra, at 336. Although Trinova repeats the Governor’s statement in an attempt to demonstrate an impermissible motive on the part of the State, all the contemporaneous evidence concerning passage of the SBT suggests a benign motivation, combined with a practical need to increase revenues.11 Neither Trinova nor the secondary sources it relies upon present any evidence that the SBT was inspired as a way to export tax burdens or import tax revenues.

M <1

In reviewing state taxation schemes under the Commerce Clause, we attempt “to ensure that each State taxes only its fair share of an interstate transaction.” Goldberg v. Sweet, 488 U. S. 252, 260-261 (1989). We act as a defense against state taxes which, whether by design or inadvertence, either give rise to serious concerns of double taxation, or attempt to capture tax revenues that, under the theory of the tax, belong of right to other jurisdictions. We have always “declined to undertake the essentially legislative task of estab *387 lishing a ‘single constitutionally mandated method of taxation.’” Id., at 261, quoting Container Corp., 463 U. S., at 171. We do not say today whether other States should adopt a VAT, or whether Michigan’s three-factor formula is the only acceptable method of apportionment. We do hold that, as applied to Trinova during the tax year at issue, the Michigan SBT does not violate the Due Process or Commerce Clauses of the Constitution.

The judgment of the Supreme Court of Michigan is

Affirmed.

Justice Souter took no part in the consideration or decision of this case.

In calculating value added, a firm’s use of capital is represented by depreciation. Depreciation is the reduction in value of a firm’s assets, through wear and tear, obsolescence, or other factors, and thus roughly measures consumption of capital. See McGraw-Hill Dictionary of Modern Economics 130 (3d ed. 1983); P. Samuelson & W. Nordhaus, Economics 902 (12th ed. 1985).

See, e. g., Aaron, Introduction and Summary, in The Value-Added Tax: Lessons from Europe 1, 2 (H. Aaron ed. 1981) (hereinafter Aaron); Haughey 1018; Special Committee on the Value-Added Tax of the Section of Taxation, American Bar Association, The Choice Between Value-Added and Sales Taxation at Federal and State Levels in the United States, 29 Tax Lawyer 457, 459 (1976) (addition and subtraction methods “reacfh] the same result by the opposite means”); SBT Analysis 20 (addition and subtraction are “two alternative, but equivalent ways of calculating value added. . . . The important point is that, conceptually, these two calculations are equal”).

The nations of the European Economic Community (EEC) each levy a VAT under yet a third method, as a multistage sales tax. See generally Aaron. Under the EEC system, the bakery in our example would be taxed on each sale of bread and would receive a credit for each purchase of materials going into production of the bread. Similarly, at each other link in the chain of production and distribution, tax is assessed on sales, but credit is provided on purchases. This multistage sales tax system places the burden on the taxpayer to demonstrate that it did, in fact, purchase goods for which it requests a credit. The multistage sales tax version of the VAT has been advocated as promoting tax compliance, though the evidence does not necessarily support this view. See Oldman & Woods, Would a Value-Added Tax System Relieve Tax Compliance Problems, in Income Tax Compliance: A Report of the ABA Section of *366 Taxation Invitational Conference on Income Tax Compliance 317 (1983) (multistage consumption VAT has traditionally been regarded as self-enforcing because the tax credit mechanism is said to induce firms to report transactions accurately).

The requirement of “fiscal frontiers” to record and tax interstate transactions makes the multistage sales tax approach impracticable for an individual State. McLure, State and Federal Relations in the Taxation of Value Added, 6 J. Corp. L. 127, 130-131 (1980); see also Haughey 1025 (“invoice credit method is not workable in a subeconomy without the legal authority and means to control the flow of imports and exports”).

On international transactions, the EEC’s VAT’s are assessed in the jurisdiction of destination. As a result, no tax is applied on exports, while full tax is applied to imports. See id., at 1024-1025; Aaron 4. The destination principle does not, however, purport to determine whether value was added in the jurisdiction of destination, or the jurisdiction of origin.

The SBT was not Michigan’s first experiment with a VAT. Between 1953 and 1967, Michigan had utilized a Business Activities Tax (BAT) similar to the Michigan SBT. Although the BAT was a subtraction method VAT, and permitted different adjustments than the SBT, the BAT tax base included “a company’s payroll, profits, and capital outlay less depreciation allowed,” Lock, Rau, & Hamilton, The Michigan Value-Added Tax, 8 Nat. Tax J. 357, 363 (1955), and was apportioned among States by the same three-factor formula that is challenged here. See Mich. Stat. Ann. §§ 7.557(l)-7.557(24) (Supp. 1955), repealed, 1967 Mich. Pub. Acts 281. The Michigan Supreme Court upheld the BAT against a challenge on facts remarkably similar to those presented here by Trinova. Armco Steel Corp. v. State, 359 Mich. 430, 102 N. W. 2d 552, 555-556 (1960) (Ohio corporation had nominal Michigan property and payroll, but substantial Michigan sales). We dismissed an appeal of the judgment in Armco for want of a substantial federal question. Armco Steel Corp. v. Michigan, 364 U. S. 337 (1960). The arguments in that case focused on whether the BAT was best characterized as a tax on income or a tax on gross receipts, with the concern that under our jurisprudence of the time a “direct” tax on gross receipts from interstate commerce would be unconstitutional.

“Business activity” is broadly defined as “a transfer of legal or equitable title to or rental of property, whether real, personal, or mixed, tangible or intangible, or the performance of services, or a combination thereof, made or engaged in, or caused to be made or engaged in, within this state, whether in intrastate, interstate, or foreign commerce, with the object of gain, benefit, or advantage, whether direct or indirect, to the taxpayer or to others, but shall not include the services rendered by an employee to his employer, services as a director of a corporation, or a casual transaction. . . .” Mich. Comp. Laws § 208.3 (1979).

Any taxpayer can, in the alternative, calculate its adjusted tax base as total gross receipts multiplied by the apportionment figure (derived using the three-factor formula) divided by two. This figure is then multiplied by the 2.35% tax rate to give actual tax liability. § 208.31(2). Under this alternative calculation, no firm’s Michigan SBT liability will ever exceed 1.175% of apportioned gross receipts.

Trinova could have calculated its tax liability under the alternative gross receipts method, Mich. Comp. Laws § 208.31(2) (1979), as follows: *370 Total gross receipts ($391,065,866) multiplied by Michigan apportionment factor (8.9717%) divided by two (equals $17,542,628) multiplied by 2.35% equals tax liability of $412,251. This figure amounts to less than 0.4% of Trinova’s Michigan sales. Of course, Trinova did not use this method, as it was required to pay only $293,578 (or 0.28% of Michigan sales) under the apportionment method challenged here.

The amended return proposed that Trinova’s company-wide compensation and depreciation be excluded from the preapportionment tax base, and actual Michigan compensation and depreciation be added back into the apportioned tax base, as follows:

Total Tax Base — statutory formula: $221,125,319

Deduct Compensation ($226,356,271)

Deduct Depreciation ($23,262,909)

Trinova’s Proposed Total Tax Base: ($28,493,861)

*371 Apportionment (8.9717%) ($2,556,384)

Add Michigan Compensation $511,774

Add Michigan Depreciation $2,152

Apportioned Tax Base: ($2,042,458)

433 Mich. 141, 147, n. 4, 445 N. W. 428, 431, n. 4 (1989).

Trinova’s proposed two-factor apportionment differs drastically from the apportionment Trinova requested in the Michigan state courts. Under the approach Trinova took in state court, following Jones & Laughlin Steel Corp. v. Department of Treasury, 145 Mich. App. 405, 377 N. W. 2d 397 (1985), Trinova's apportioned tax base would be -$2,042,458. See n. 8, *382 supra. Under the two-factor formula that Trinova now urges upon us, it is $360,213.

As an alternative ground for upholding the tax, Michigan reminds us that, instead of taxing value added, it could have taxed gross receipts of sales into Michigan. We have repeatedly upheld such taxes. Standard Pressed Steel Co. v. Washington Revenue Dept., 419 U. S. 560, 564 (1975); General Motors Corp. v. Washington, 377 U. S. 436, 448 (1964); McGoldrick v. Berwind-White Coal Mining Co., 309 U. S. 33, 58 (1940). While we accept the analogy Michigan has drawn between a VAT and an income tax, we recognize that the SBT also bears some similarities to a gross-receipts tax. Further, the SBT’s alternative method of taxation (based upon gross receipts) might provide an additional limit on any distortion or possibility that out-of-state values might be taxed. See n. 6, supra. In light of our disposition, we need not address these arguments.

According to Kleine, proponents of the SBT argued as follows: (1) because of Michigan’s volatile economy, the State’s corporate income tax had proven unpredictably cyclical, and therefore a poor source of revenue. The SBT would provide a much broader tax base, and thus prove a more stable revenue source; (2) the SBT would lessen the tax burden on capital, thereby encouraging new investment; (3) the SBT would replace numerous taxes, resulting in less paperwork for both the taxpayer and the tax collector; (4) the SBT would not discriminate against businesses on the basis of their forms of organization; and (5) the SBT would tax inefficient and efficient firms equally for their use of government services, whereas an income tax would burden more heavily efficient, highly profitable firms. In addition, at the time of the SBT’s enactment, Michigan faced a fiscal crisis. The legislature provided that the new SBT would overlap with the old corporate franchise tax, resulting in additional cash flow of $180 million. Kleine 20-23. The argument that a VAT would permit “exporting” the tax to taxpayers outside the State “was not used to any great extent by the proponents of the Michigan [SBT].” Id., at 23.