The Yale Law Journal


A Brief Sur-Reply to Professors Graetz and Warren

16 May 2013


We are grateful for the deep engagement by Michael Graetz and Alvin Warren with our article, What Is Tax Discrimination?, which appeared in the print edition of the Yale Law Journal,1 and for this opportunity to respond to their comments.2 The Court of Justice of the European Union (ECJ) is charged with deciding whether the laws of European Union (EU) member states violate the fundamental freedoms guaranteed by the Treaty on the Functioning of the European Union (TFEU): the free movement of goods, capital, workers, and services, and the right to establish business across borders. The ECJ frequently has held that tax discrimination violates these fundamental freedoms. Although the ECJ has been aggressive in finding that long-standing member-state tax practices violate the EU prohibition on tax discrimination, its failure to clearly articulate a guiding principle in tax cases has attracted extensive critical commentary.3 Accordingly, our goals in What Is Tax Discrimination? were to identify the principle behind the ECJ’s interpretation of tax discrimination, to explain that principle in economic terms, and to describe how to apply that principle.

In that article, we argued that the guiding principle behind the prohibition of tax discrimination is the prevention of protectionism (expressed negatively) or the promotion of competition (expressed affirmatively). In other words, the tax-nondiscrimination principle promotes a level playing field between similarly situated economic actors from different member states. We argued that in the direct tax context, the ECJ interprets the fundamental freedoms to require member states to refrain from using taxes to make it more difficult for cross-border actors than for domestic actors to compete for jobs or investments. Thus, we concluded that the ECJ’s interpretation of the fundamental freedoms amounts to requiring what public finance economists call “capital ownership neutrality” (CON).4 CON is the notion that tax policies should not distort the ownership of assets. (The labor analog of CON is the notion that tax policies should not provide workers from one state with a tax-induced advantage over workers from other member states in securing a job.)

In their fifty-page response to our article, Graetz and Warren raise numerous objections. While we cannot answer all of their objections in these few pages, we will respond briefly to their most serious criticisms.5 Those criticisms can be divided into two broad categories: criticisms of our interpretation of tax nondiscrimination and criticisms of our proposed application of this interpretation.

I. Our Interpretation of Tax Discrimination as Informed by Competitiveness

Graetz and Warren claim to be “mystified” by our theory of constitutional interpretation.6 According to them, we first choose economic efficiency as the paramount norm for evaluating tax discrimination. We then select one efficiency concept, competitive neutrality, over other alternative efficiency concepts without sufficient theoretical or empirical support. Finally, after conceding that the ECJ lacks institutional authority to fully implement competitive neutrality, we urge the ECJ to raise competitive neutrality “to constitutional status.”7

In our view, the above description seriously misconstrues our argument.8 We did not choose competitive neutrality because we concluded, without the benefit of theoretical or empirical support, that competitive neutrality would promote economic welfare better than other norms.9 Rather, we chose competitive neutrality over other norms because competitive neutrality is a superior interpretation of the language of the treaties and because it accords better with the ECJ’s actual decisions in tax cases.10 Thus, whereas Graetz and Warren characterize competitive neutrality as our “assumed constitutional norm,”11 we would characterize it as an “observed constitutional norm.”

Graetz and Warren’s misinterpretation pervades their reply. They criticize us for failing to justify on normative grounds our claim that the doctrine of tax nondiscrimination should be interpreted and applied so as to advance competitiveness.12 For example, Graetz and Warren fault us for not offering any evidence or any reasons for believing that competitive neutrality would do a better job of promoting welfare than competing efficiency norms, such as locational neutrality.13 They argue that “a policy decision based on an economic efficiency standard should be grounded in evidence as to the magnitude of the various distortions.”14 We agree, and acknowledged as much in our original article.15 But our goal was to determine whether any efficiency norm fits the extant tax-discrimination jurisprudence, not to show that the norm that best fit the doctrine was also the norm that would bestpromote efficiency.16

Graetz and Warren’s critique of our article amounts to a lament that we did not take up the question of what interpretation of tax discrimination would best promote economic welfare. But our goal was to answer a more limited set of questions concerning the extant legal definition of tax discrimination17 and the economic implications of that definition.18 Although they do not like our choice of questions, nowhere in their response do Graetz and Warren dispute our conclusion that competitive neutrality is the best fit for the text and doctrine.19

In addition, although we do not claim that competitive neutrality is the welfare-maximizing interpretation of tax nondiscrimination, we do claim that interpreting the tax-nondiscrimination principle to require competitive neutrality is welfare enhancing as compared to a situation in which the EU does not police tax discrimination at all.20 Based on their prior work, it would appear that Graetz and Warren used to agree with that more modest claim and with our broader claim that tax discrimination is informed by competitiveness.

For example, in a 2001 article, Warren conceptualized tax laws that favor domestic producers over foreign producers as a form of discrimination.21 After reviewing various authorities, Warren concluded that “[t]he competitiveness norm would seem more consistent with the current scope of prohibited discrimination than is the [locational] efficiency norm.”22 Warren’s discussion shows that tax discrimination has long been viewed as concerned with promoting competition between in-state and cross-border actors. Thus, it should be no surprise that the ECJ might come to a similar conclusion.

Furthermore, in a 2006 article on tax discrimination that appeared in the print edition of this journal, Graetz and Warren criticized the ECJ’s tax- discrimination jurisprudence as fundamentally inconsistent23 and as creating a “labyrinth of impossibility.”24 As Graetz and Warren describe it in their 2012 article, their 2006 article argued that the ECJ’s “reliance on nondiscrimination as the basis for its decisions did not (and could not) satisfy commonly accepted tax policy norms, such as fairness, administrability, economic efficiency, production of desired levels of revenues, avoidance of double taxation, fiscal policy responsiveness to economic circumstances, inter-nation equity, and so on.”25 And they read the ECJ’s decisions (as of 2006) to “suggest potentially staggering constraints on countries’ freedom to resolve what strike us as quintessentially legislative issues—constraints that are fundamentally inconsistent with the fiscal autonomy retained by the member states in their right to veto EU taxing provisions.”26 Graetz and Warren concluded by arguing that unless the EU moves toward greater harmonization of its tax laws (which is unlikely given the requirement of member state unanimity in fiscal matters), the member states should be given more control over their own tax policies. However, in considering the possibility of the ECJ cutting back on its enforcement of tax discrimination, Graetz and Warren offered a few words of caution:

  [Limiting] the ability of the ECJ to strike down member states’ income tax provisions . . . would permit considerable mischief by the member states. As our review of the ECJ cases has shown, some member state tax provisions are potentially quite protectionist, and some have been adopted to serve precisely that purpose. The dilemma for the nations of Europe is to find a way to retain their autonomy over tax matters without undermining the internal market and, as a practical matter, severely restricting the four freedoms.27  

Thus, in their 2006 article, which is principally a vigorous critique of the ECJ’s tax-discrimination jurisprudence up to that time, Graetz and Warren put their fingers on, but did not develop, many of the points they now dispute. They acknowledged that the four freedoms, which are the source of the principle of tax nondiscrimination, promote competitiveness. They also recognized that promoting competitiveness advances welfare by restraining protectionism. And they warned that without some sort of nondiscrimination principle, the member states would enact protectionist tax policies that would in turn undermine the internal market.

In the same 2006 article, Graetz and Warren proceeded to offer a possible “middle ground between the limited nondiscrimination requirements of international tax and trade treaties and the unduly inhibiting version of nondiscrimination fashioned by the ECJ.”28 The one alternative they suggested was “a slowing of ECJ intervention with more attention to the effect on the member states’ fisc and a greater focus on protectionism as a potential middle ground. The court’s inquiry might, for example, be directed to whether the intent of the provision was protectionist.”29 Presumably, then, because a competitive-neutrality interpretation of nondiscrimination would target protectionist tax provisions for elimination, our proposal at least overlaps with the suggestion that Graetz and Warren made in 2006.30 Expressed somewhat differently, Graetz and Warren in 2006 suggested a compromise between two important values (the internal market and state sovereignty over taxation) that resembles the proposal we developed at length in our 2012 article. Even if Graetz and Warren have changed their position on the proper interpretation of tax discrimination since 2006, they would be hard-pressed to deny that competitive neutrality is at least a reasonable and intuitive interpretation of tax nondiscrimination, or that such an interpretation would increase welfare relative to a regime where protectionist tax policies went unchecked due to a lack of judicial enforcement of tax nondiscrimination.

Graetz and Warren raise several more technical objections to our interpretation of tax discrimination. For example, they argue we did not perform the necessary groundwork for translating CON into the labor context.31 Such an argument again confuses the scope of our project32 with that of the project they would have preferred us to undertake.33 As Graetz and Warren point out, the controversial welfare arguments made advocating CON might not have the same force in the labor context as in the capital context.34 Thus, from a welfare perspective, it may be more important to eliminate protectionist taxes on capital than protectionist taxes on labor.35 Nevertheless, the goal of the tax-nondiscrimination principle may be to root out protectionism and promote competitiveness in both capital and labor. We readily agree that we do not establish competitive neutrality for labor as a crucial normative goal, but that is not necessary for our project. All we need is to translate the idea of competitive neutrality into the labor context, an intuitive and straightforward task.36

Graetz and Warren also claim that our interpretive argument is based on the “unrealistic” assumption that a taxpayer’s residence is fixed.37 They argue that this assumption is too restrictive for a welfare analysis.38 We do not disagree with the latter proposition. But since we did not undertake a welfare economics project, this criticism is beside the point. Again, they confuse our interpretive project with their welfare economics project.

Next, Graetz and Warren claim that our “entire analysis [is based] on an assumption that EU citizens will not take a job in another country if they have to live in the other country for more than six months.”39 Thus, they argue that even if our analysis is right, it is of sharply limited applicability.40 But we never assumed EU nationals would refuse a job in another country if it required them to live there for six months, nor does our analysis require such an assumption.41 Instead, what we say is that when an individual has to change residence in order to take a new job, the legal doctrine of tax discrimination is inapplicable to that person because she resides in the same state where she earns her income.42 EU commentators refer to a case in which an EU taxpayer earns only domestic income as a “purely internal” situation, whereas the right against tax discrimination applies only to cross-border activity.43 Thus, if our analysis is limited, it is limited to the legally relevant set of cases. We did not address purely internal cases simply because purely internal situations are not the focus of the tax-nondiscrimination principle.44

Finally, Graetz and Warren assert that, whatever the relevance of our tax-discrimination analysis to labor taxation, it is inapplicable to capital because in the capital context the assumption of fixed residence “is patently implausible.”45 We agree that investors can change their residence, but we fail to see how that observation undercuts our interpretation of tax discrimination. Although investors may move, at any point in time, they must reside somewhere. Their investments are also located somewhere, perhaps in many states. Any time someone invests in a state other than her own, she could be the target of tax discrimination. Host states might discourage investment from abroad, and residence states might discourage their residents from investing abroad. Thus, cross-border investors can be harmed by tax discrimination. Accordingly, we fail to see how the ability of investors (and service providers) to change residence somehow implies that tax discrimination—which only applies to cross-border investments and provisions of services—cannot or should not be interpreted to promote competitiveness.

II. Applying the Competitive Neutrality Interpretation

After arguing that the best interpretation of the tax nondiscrimination principle is that it seeks to promote competitiveness, our original article described how courts could apply that interpretation on a consistent basis. We showed that competitive neutrality involves two requirements: (1) uniform taxation and (2) universal adoption of one of two possible methods of double tax relief.

First, competitive neutrality requires what we called “uniform” source and residence taxation. (A source tax is uniform if it applies the same way to taxpayers earning income in the jurisdiction without regard to their state of residence. A residence tax is uniform if it applies the same way to all residents of the jurisdiction, no matter where they earn their income.) Second, strict adherence to competitive neutrality requires all states to agree on the same method of double tax relief. Specifically, competitive neutrality requires universal adoption of either worldwide taxation with unlimited foreign tax credits or what we labeled the “ideal deduction method,” one instantiation of which is exemption of foreign-source income.46 Because the ECJ cannot compel states to choose a specific method of double tax relief, we suggested that the ECJ should uphold an allegedly discriminatory tax as long as the law was written and enforced on a uniform source or residence basis.

Thus, we made both a theoretical argument—uniform taxation combined with universal adoption of either worldwide taxation with unlimited foreign tax creditsor the ideal deduction method would achieve competitive neutrality under ideal conditions—and a policy argument—in the absence of authority to impose a universal method of double tax relief, the ECJ should enforce the nondiscrimination principle by striking down non-uniform tax laws. In other words, we offered a compromise of the kind that, in 2006, Graetz and Warren said was needed to balance member states’ tax-sovereignty interests with the welfare interest in preventing protectionism.

Graetz and Warren do not explicitly disagree with our claim that under the usual economic assumptions (such as frictionless markets, no externalities, and a quick convergence to equilibrium) universal adoption of either worldwide taxation with unlimited foreign tax credits or ideal deduction achieves competitive neutrality.47 They do, however, express doubt as to whether we adequately demonstrated that claim.48 Accordingly, in a forthcoming paper we provide a more general derivation along the lines suggested by Graetz and Warren and elaborate upon the example of Françoise and Günther from our original article.49

Graetz and Warren also argue that the ideal deduction method is so far removed from real-world tax systems as to be impractical.50 But this method is closer to actual practice than they acknowledge. As we explain in our original article, under ideal deduction, taxes would proceed in two stages. In the first stage, states would subject all income sourced within their territory to uniform taxation, that is, to taxation without regard to the residence state of the taxpayer. In the second stage, states would subject their residents’ worldwide income to uniform taxation, that is, to taxation without regard to where the income was earned. In calculating how much income will be subject to taxation in the second stage, residence states would allow a deduction for any source taxes (including domestic source taxes) paid by the resident.

This form of two-stage taxation seems unfamiliar, and its adoption by EU member states therefore seems improbable. However, exemption is a form of ideal deduction—one where the residence tax rate is equal to zero. Most of the member states already employ exemption systems. Moreover, there are also real-world examples of two-stage taxation in which the second-stage tax is not zero. For example, many aspects of social welfare and support systems administered through the tax laws have these qualities—source taxation of income and residence-based provision of benefits (i.e., the tax rate in the second stage is negative). However, the clearest example of two-stage taxation with a non-zero second-stage rate is the classical corporate income tax. With such a tax, in the first stage, the corporation is taxed on its income, and in the second stage, the investor is typically taxed on dividends, which are explicitly an after-tax cash flow, and capital gains, which are implicitly so.

Graetz and Warren also assert that ideal deduction does not work with progressive tax rates.51 They offer an example of a taxpayer who earns €10,000 at home, where the source tax rate is 10%, and €10,000 abroad, where the source tax rate is 5%. As Graetz and Warren correctly note, the taxpayer will pay €1,000 in home source taxes and €500 in foreign source taxes. They also correctly conclude that under the ideal deduction method, the taxpayer’s total residence-state-taxable income is €18,500—€20,000 less €1,500 in source taxes—and so her residence tax will be €4,050. But Graetz and Warren then ask, “[H]ow much of that residence tax is attributable to the foreign income? Is it half . . . or 95/185 . . . ? Without an answer to these questions, the taxpayer cannot make the [necessary] calculations . . . .”52 The example stops at that point. But why does the taxpayer need to make such a calculation? Graetz and Warren do not say. Certainly not to calculate her residence taxable income, which is €18,500, or to figure out her residence tax liability, which she can calculate from her taxable income and her residence tax schedule.

In our original article, we argued that one advantage of adopting uniformity as the touchstone for tax discrimination is that the cases are reasonably straightforward to resolve because, among other reasons, courts would only have to look at the law of one state to determine whether that state is discriminating. If a law is uniform either on a source or on a residence basis, the law is not discriminatory; otherwise, it is. Graetz and Warren see our one-state approach as a failing. In their words, “any serious attempt to identify the tax advantages or disadvantages for cross-border income should take account of the tax consequences in both countries.”53 It is difficult to respond to this claim because Graetz and Warren offer no arguments in support of it, though they cite some cases in which the ECJ looked to the law in both states.54 Nonetheless, the U.S. Supreme Court’s tax-discrimination doctrine is instructive on this matter. According to the Court, the constitutionality of a defendant state’s tax law should not “depend on the shifting complexities of the tax codes of 49 other States.”55 We agree: conditioning the legality of one EU member state’s law on the tax provisions of twenty-six other member states would be impractical and impossible to administer.

Finally, Graetz and Warren give three examples they claim our approach does not resolve. First, they give an example of a charity that faces two states’ requirements when it expands into a neighboring state.56 Because the multi-state charity must satisfy the requirements of both its home and its host states, while domestic charities need to satisfy the requirements of only one state, the cross-border charity faces additional burdens. This is an example of double regulatory burdens, not discrimination. Accordingly, the example need not be resolved through any interpretation of tax discrimination.

Their second example involves corporate tax integration. As a method to relieve double economic taxation of corporate income, states may credit the corporate tax paid by the company against the tax due from resident shareholders on dividends. Such relief typically is limited to dividends received from domestic corporations. Graetz and Warren ask whether a state that grants such imputation credits should be required to extend those credits to dividends received by residents from nonresident corporations.57 This is not a case that our theory leaves unresolved, though states may take one of several approaches to resolving corporate double taxation, each of which is consistent with competitive neutrality. First, a state could grant imputation credits to all resident shareholders, regardless of whether the dividend was paid by a foreign or domestic company. In the alternative, a state could grant imputation credits only for dividends paid by domestic companies—but in that case, to maintain competitive neutrality, it would have to grant the imputation credits to both resident and nonresident shareholders. The first option represents a uniform residence credit, while the second option represents a uniform source credit. Finally, a state could simply provide no relief for double taxation. One thing a state may not do consistently with competitive neutrality is to provide relief only to domestic shareholders only for domestic dividends.58 That would be discriminatory, at least on a competitive-neutrality interpretation of tax discrimination.

Their third example is that of a married same-sex couple who, in order to take jobs, move (long enough to establish residency) to a second state that does not recognize same-sex marriages. Graetz and Warren ask whether the second state should be required to recognize their marriage.59 Our answer is that the nondiscrimination principle of the TFEU does not aim to make residence decisions neutral. There is much evidence that one of the overarching goals of the TFEU is to create regulatory competition, at least in those areas (including direct tax) that are not harmonized at the EU level.60 This competition means that different types of taxpayers will be taxed differently in different jurisdictions. However, as long as those differences are not tied to being a national of one or another state, there is no violation. Thus, although differences in marriage law across member states will affect residence decisions, a state does not have to recognize a marriage conducted in another state to avoid a violation of competitive neutrality.


Our project, then, stands in contrast with that of Graetz and Warren. In their 2006 article, Graetz and Warren examined the substantive outcomes in tax-discrimination cases and criticized the ECJ’s resolution of those cases for infringing EU member states’ tax sovereignty.61 They stressed that tax policy ought to be made by the member states, not the ECJ.

Our task was different. We started from the widely accepted premise that the TFEU prohibits states from exercising their tax sovereignty in a manner that discriminates against nonresidents or intra-EU commerce. Thus, we did not reexamine the questions of whether the TFEU ought to constrain member states’ tax sovereignty or whether the ECJ is the right institution to arbitrate disputes about whether particular tax laws violate the nondiscrimination principle. Rather, our goal was to identify the guiding economic principles the ECJ uses in resolving tax-discrimination cases. We argued that competitive neutrality is a better fit for the doctrine and for the language of the TFEU than are competing tax-neutrality norms. Although they say they disagree with that interpretation, Graetz and Warren do not seriously challenge that position; indeed, in earlier writing, they have expressed views consistent with our own interpretation. Graetz and Warren also argue that we have not sufficiently demonstrated that a competitive-neutrality interpretation of tax discrimination would maximize EU welfare. That is true, but normative claims of that kind fall outside the scope of our project.

Having identified competition as the primary efficiency value promoted by the tax-nondiscrimination principle, we then turned to the question of how courts can apply that interpretation in a consistent way. We showed that competitive neutrality requires uniform taxation as well as universal adoption of either worldwide taxation with an unlimited foreign tax credit or ideal deduction. Graetz and Warren do not seriously challenge that conclusion either. Finally, we argued that in light of the ECJ’s inability to require member states to adopt an unlimited foreign tax credit or an ideal deduction system, the ECJ should interpret the principle of tax nondiscrimination to preclude states from enacting and enforcing non-uniform taxes.

Graetz and Warren strongly disagree with our recommendation, but in their response they offer no alternative. Given Graetz and Warren’s sharp criticism of the ECJ’s tax-discrimination jurisprudence in their 2006 article, it is unlikely that they favor sticking with the status quo, in which the ECJ rules without expressly stating its decision-making criteria, at times in the face of fierce external criticism. Nor is it obvious that Graetz and Warren would favor significant limitations on the ECJ’s jurisdiction to decide tax-discrimination cases, given their stated concern that failure to restrict tax discrimination would leave member states largely free to use tax policies to protect domestic interests, thereby “undermining the internal market” and “severely restricting the four freedoms.”62 Although our aim in our original article was not to argue that the ECJ has struck the proper balance between economic integration and member states’ tax sovereignty, we believe that our interpretive recommendation represents an appropriate compromise between the two.

Ruth Mason is Professor of Law and Anthony J. Smits Professor of Global Commerce, University of Connecticut School of Law. Michael S. Knoll is the Theodore K. Warner Professor, University of Pennsylvania Law School; Professor of Real Estate, Wharton School; Co-Director, Center for Tax Law and Policy, University of Pennsylvania. Copyright 2013 by Ruth Mason and Michael S. Knoll. The authors would like to thank Al Dong for his assistance with research.

Preferred citation: Ruth Mason & Michael S. Knoll, A Brief Sur-Reply to Professors Graetz and Warren, 123 Yale L.J. Online 1 (2013),