[*PG35]STRICTNESS AND SUBSIDIARITY: AN INSTITUTIONAL PERSPECTIVE ON AFFIRMATIVE ACTION AT THE EUROPEAN COURT OF JUSTICE

Sean Pager*

Abstract:  The move to strict review of gender equality cases by the European Court of Justice raises questions regarding the institutional role of the Court. Comparisons between the ECJ’s affirmative action case law and US jurisprudence serve to illuminate the very different role played by the ECJ as the central arbiter of a supranational judiciary. In its readiness to decide contextual issues better left to the national courts, the European Court has taken an “American approach” to affirmative action out of keeping with its role. Closer attention to the dynamics of the Court’s partnership with national judiciaries would serve as a step toward a functional conception of judicial subsidiarity.

Introduction

The 1995 judgment of the European Court of Justice (ECJ) in the case of Kalanke v. Freie Hansestadt Bremen1 attracted widespread attention both in the European press and in scholarly commentary.2 [*PG36]Although it was couched in tentative language, many read the opinion as putting the handwriting on the wall for positive action programs in Europe.3 The Kalanke judgment was handed down the same year as the U.S. Supreme Court issued its Adarand4 ruling, which was widely viewed as having an equivalent effect on affirmative action in the United States.5

In hindsight, both predictions have proved premature. Although affirmative action programs in the United States remain on the defensive politically,6 attempts to achieve their total abolition in the courts appear to have stalled.7 Moreover, in Europe, what began as a vigorous legislative counterattack8 ultimately assumed constitutional force [*PG37]as part of the Amsterdam Treaty.9 Meanwhile, the ECJ, no doubt sensing the political tide, has shifted course.10 Its judgment in Marschall v. Land Nordrhein Westfalen11 effectively reversed the result of Kalanke without overruling it.

Beneath their divergent outcomes, however, Kalanke and Marschall raise deeper questions regarding the role played by the ECJ as it moves to “constitutionalize”12 gender equality by applying a “strict interpretation” of derogations.13 Although this jurisprudence has inspired a flood of commentary, the Court’s institutional role in deciding individual rights cases—and in particular its interaction with the national judiciaries and the relevance of the subsidiarity principle in this regard—remain undertheorized.14

[*PG38] The present work locates the values served by subsidiarity in the positive action cases and connects them with the roles played by the ECJ and national judiciaries under the Article 234 (formerly article 177) procedure of preliminary references. Comparisons with American affirmative action jurisprudence help to clarify the meaning of “strict” judicial review and point to the very different institutional role appropriate to the central court of a supranational judiciary. By ruling on affirmative action at the European level, it is argued that the Court of Justice decided a contextually contingent question better left to the national judiciaries.

Part I of this article summarizes the Kalanke and Marschall decisions; part II compares the logic of these decisions with the rationales of American case law; part III investigates the way in which these cases analyze equality as a European right; part IV considers the relevance of the subsidiarity principle to this analysis; and part V situates the positive action cases within ECJ’s fundamental rights jurisprudence.

I.  Facts and Law

The ECJ’s rulings in Kalanke and Marschall provide a striking contrast because factually the cases are almost identical. In both cases, a male state employee challenged a promotion decision in favor of a female rival based on a preference for the “under-represented” sex where the candidates’ qualifications were otherwise equal. The only factual distinction that appears relevant was that in Kalanke the rules governing the preference may have been more inflexible than those at play in Marschall.

A.  Facts of Kalanke

Eckhard Kalanke and Heike Glissmann were candidates shortlisted for a promotion in the Parks Department of the City of Bremen in 1990. Although Kalanke had received the initial recommendation for the post, resistance from the Personnel Committee led to the referral of the matter to a Conciliation Board. The Board ruled that, as both candidates possessed equivalent qualifications, Glissmann should be given priority as a woman. The board relied on a law then in force in the Land (province) of Bremen requiring that female candidates receive preference, all other criteria being equal, in sectors in which women were under-represented. The law defined under-representa-tion as occurring when women constitute less than half of the employees.

[*PG39] Kalanke challenged this outcome under German law. He was unsuccessful in lower courts and then appealed the case to the Bundesarbeitsgericht (the Supreme Labor Court of Germany).15 That court again rejected Kalanke’s German law arguments.16 However, faced with Kalanke’s further claim that the gender preference mandated by Bremen’s statute violated European Community (Community or E.C.) law, the German court requested a preliminary ruling from the ECJ under the Article 234 (formerly article 177) procedure of the E.C. Treaty.

Article 234 provides a mechanism for national courts to refer questions regarding the interpretation of Community law to the ECJ. The judgments which the ECJ issues in response to Article 234 referrals address questions of law in the abstract, leaving it to the referring court to apply the law to the facts of the case.17

B.  The Advocate General’s Opinion in Kalanke

Advocate General Tesauro presented the case to the ECJ in a detailed opinion.18 The opinion began by reciting Article 2(1) of the Equal Treatment Directive’s “peremptory” mandate “that there shall be no discrimination whatsoever on grounds of sex.”19 In the Advo[*PG40]cate General’s mind, the gender-conscious basis on which the Bremen law operated clearly involved sex discrimination.20 The question the Advocate General posed is whether such a law could fall into the exception provided for by Article 2(4) of the Directive, permitting “measures to promote equal opportunity . . . by removing existing inequalities which affect women’s opportunities.”21

Kalanke was not the first case in which the ECJ had interpreted Article 2(4). An earlier case, Commission v. France, examined the provision in connection with a panoply of special rights that French labor law accorded women, including shortened work hours, early retire[*PG41]ment, family leave, and child-care allowances.22 The case primarily analyzed the permissibility of these benefits under a different provision of the Directive, Article 2(3).23 However, when it came to Article 2(4), the Court read that provision as “specifically and exclusively designed to allow measures which, although discriminatory in appearance, actually aim to eliminate or reduce de facto instances of inequality which may exist in actual working life.”24

Advocate General Tesauro interpreted this formulation from Commission v. France as permitting gender-specific measures that are remedial in that they target the obstacles holding women back. By removing inequalities, he argued, positive action under Article 2(4) helps to ensure equality of opportunity, leveling the playing field for men and women. Therefore, despite a gender specific form, such measures do not represent a “genuine derogation[]” from the principle of equal treatment.25 Indeed, Tesauro commented that “it is only in this way that real and effective substantive equality will be achieved.”26

In making these arguments, Tesauro implicitly adopted a process view of equality. Throughout his opinion, he emphasized that Article 2(4) of the Directive addressed equality of opportunities, not outcomes. Equality was to be regulated only as to process, not result.27 This distinction lay at the heart of Tesauro’s analysis: the Advocate General rejected the employment preference applied under the Bre[*PG42]men statute because he viewed it as aimed at outcomes, not opportunities.28

Tesauro argued that employment preferences were ill-suited to the task of equalizing opportunities because he saw the main structural obstacles hindering women’s advancement as lying outside the workplace.29 He believed that what women needed foremost in order to compete in the labor force was vocational counseling and training to upgrade their qualifications as well as progressive social policies, such as subsidized child-care, to reduce the burden of home life.30 By contrast, Tesauro viewed employment preferences as “tak[ing] on a compensatory nature” linked to “historical discrimination.”31 Such ‘paybacks,’ the Advocate General contended, did nothing to address the obstacles which women face today. Instead of tackling the root causes of gender inequality, preferences merely papered over defects in the process by imposing a predetermined result. As such, they offered only “illusory” progress and were “irrelevant” to the goal of enhancing opportunities in the long-run.32

Tesauro’s opinion was roundly criticized in scholarly commentary for relying on faulty premises. By assuming that preferences serve only to compensate for past discrimination, the Advocate General ignored the very real present obstacles which women face in the workplace today.33 Tesauro suggested that because the Bremen preference only applied where “two candidates of different sex have equivalent qualification [this] implies by definition that the two candidates have had and continue to have equal opportunities”34—a conclusion that [*PG43]assumes selections were made solely based on qualifications. In fact, commentators pointed out that the promotion in the initial round of decision-making had gone in favor of the man, Kalanke, based in part on “social criteria” discriminatory to women.35

Furthermore, Tesauro’s assumption that preferences acted only at the level of outcomes, not opportunities, failed to take into account the potential for a preference regime to function dynamically as an instrument of structural change. By breaking through glass ceilings and infiltrating “old boy” networks, an early vanguard of women can enact changes that make it easier for women to compete later on even terms.36 Indeed, the German court which referred the Kalanke case to the ECJ described a very similar phenomenon when it explained that Bremen’s preference scheme may “help to overcome in the future the disadvantages which women currently face” by acclimating people to seeing women in more senior posts.37

[*PG44] Finally, Tesauro was too quick to dismiss the linkage of preferences to under-representation as purely a numbers battle.38 The problem with this reading of the statute, as Tesauro himself acknowledged, was that “under-representation of women in a given sector [may] reflect[] existing inequality.”39 He attempted to deflect such concerns by observing that “under-representation . . . albeit indicative of inequality, is not necessarily attributable to a consummate determination to marginalise women.”40 Intentional discrimination is beside the point.41 Under Community law, gender imbalance constitutes prima facie evidence of indirect discrimination, regardless of intent.42 In any case, the purpose of positive action under Article 2(4) of the Equal Treatment Directive is to address proactively the more subtle barriers that hinder women’s advancement, without the need to prove discrimination of any kind.43

By recognizing that under-representation can be “indicative of inequality,” the Advocate General opened the way to a different reading of the preference.44 Tesauro’s qualifier, “not necessarily,” implicitly acknowledged a crucial ambiguity.45 The fact is that, in some cases, under-representation might indicate gender bias that preferences could therefore serve to counteract. Tesauro dealt with this ambiguity procedurally by invoking the standard of review. He presented Article 2(4) as “a derogation from an individual right [that] must be inter[*PG45]preted strictly.”46 For Tesauro, this strict interpretation could not be reconciled with the “element of arbitrariness in any preferential treatment which is mechanically confined to the under-represented group and based solely on that ground.”47 In other words, the ambiguity would be construed against the statute, and its arbitrariness seemed to belie a remedial intent.

In reaching this conclusion, Tesauro’s assessment of the statute’s function flowed directly from its form. Preferences that are “mechanically” tied to numbers, Tesauro suggested, can only serve “to rebalance the numbers of men and women, but will not remove the obstacles that brought about that situation.”48 In the absence of a more finely tuned instrument, the preference scheme must thus be rejected.

C.  The Kalanke Judgment

The judgment of the ECJ in Kalanke was short and terse.49 However, as one commentator has observed, almost everything that the ECJ did say could be traced to Advocate General Tesauro’s opinion.50 Like Tesauro, the ECJ began with the proposition that the Bremen gender preference would violate Article 2(1)’s antidiscrimination precept unless it fell within the exception carved out by Article 2(4).51 Following Tesauro, the ECJ then characterized Article 2(4) as “a derogation from an individual right laid down in the Directive [which] must be interpreted strictly.”52 The ECJ held that national laws such as Bremen’s failed this “strict” standard for two reasons. [*PG46]First, the ECJ asserted that “[n]ational rules which guarantee women absolute and unconditional priority go beyond promoting equal opportunities and overstep the limits of the exception in Article 2(4).”53 Second, the ECJ held that “in so far as it seeks to achieve equal representation of men and women in all grades and levels within a department, such a system substitutes for equality of opportunity as envisaged in Article 2(4) the result which is only to be arrived at by providing such equality of opportunity.”54

Based on these objections, the ECJ declared that gender preferences of the kind used in Bremen contravened the Equal Treatment Directive.55 The case then returned to the national (German) judiciary for further proceedings in which the Bundesarbeitsgericht annulled the promotion decision in favor of Ms. Glissmann and ordered that the candidates be judged anew.56

D.  Kalanke’s Aftermath and the Run Up to Marschall

In addition to its controversial result, the paucity of analysis in the Kalanke judgment proved unsatisfying to many of its readers. The judgment appeared to state two distinct reasons for invalidating the preference: objecting to both its “absolute and unconditional” form and the numerical equality which it “seeks to achieve.”57 The ECJ did not elaborate further as to the rationale behind either objection, although both appeared traceable to language used in the Advocate General’s opinion.58 Moreover, one is left guessing as to whether the ECJ’s objections represent separate and independent grounds for the decision, or whether they are somehow interrelated.59

This led to much speculation on the part of commentators as to what sorts of positive action schemes would pass muster in the aftermath of the Kalanke ruling.60 The European Commission, in its [*PG47]official pronouncement on the case, chose to emphasize the “absolute and unconditional” language of the judgment. By this reading, the Kalanke judgment did not invalidate gender preferences per se. Rather, the ECJ struck down the Bremen preference only because its terms were too extreme. Yet, in what way is the preference “absolute and unconditional?” Far from offering unconditional priority to women, the statute stipulated that both candidates must have equal qualifications, and it only applied to sectors where one sex is underrepresented.61

One way to make sense of this somewhat cryptic phrase is to read “absolute and unconditional” to mean “automatic,” in so far as the wording of the preference law, on its face, admitted no grounds for deviation once the relevant prerequisites were satisfied. Indeed, when the ECJ summarized its ruling at the end of the judgment, it referred to “rules . . . which automatically give priority to women in sectors where they are under-represented.”62 On this view, the judgment echoed Tesauro’s rejection of the “arbitrariness” of preferences that are “mechanically” applied in the face of ambiguous evidence.63

The ECJ’s second objection, concerning the result which the Bremen statute “seeks to achieve,” also raised questions .64 The Court saw the statute as directly imposing a result instead of intervening at the level of opportunity (i.e., process) as Article 2(4) envisages. The [*PG48]ECJ did not explain why it read the statute as limited to outcomes. However, the contrast it drew between outcomes and process came unmistakably from Tesauro. As such, it would seem that the second objection flowed from the first. Because the statute operated “automatically” to rebalance numbers, it could not be trusted as a remedial tool. Again, form revealed function.

In any event, European jurists did not have to wait long for further clarification from the ECJ as, two years later, a case with almost identical facts appeared before the Court, again as a preliminary reference under Article 234.65 This case also hailed from Germany, this time from the Land of North Rhine-Westphalia. It, too, dealt with a promotion decision between a male and female candidate with equal qualifications. As in Kalanke, the man, Hellmut Marschall, was protesting a preference granted to his female rival based on a Land law virtually identical to that of Bremen’s.66

There was one crucial difference between the two statutes, however. The North Rhine-Westphalian version contained an additional proviso which stated that a preference need not be awarded where “reasons specific to an individual male candidate tilt the balance in his favor.”67 In other words, whereas the Bremen law was phrased in absolute terms, the law from North Rhine-Westphalia had a built-in “savings clause” to permit exceptions for hardship cases.

E.  The Advocate General Opinion in Marschall

Advocate General Jacobs, in presenting the Marschall case to the ECJ, naturally reviewed the logic and structure of the earlier Kalanke opinions. He was of the opinion that the two cases were indistinguishable. Rejecting the argument that the flexibility provided by the “savings clause” sufficed to circumvent Kalanke’s ban on “absolute and unconditional” preferences, Jacobs noted that the Bremen statute itself had a degree of constitutionally-implied flexibility. “Since the Court in Kalanke recognized that the rule in issue in that case was subject to exceptions, the reference to ‘automatic’ priority should be read in that light.”68 Indeed, Jacobs stated flatly that “the national rule at issue in Kalanke was not in fact absolute and unconditional.”69

[*PG49] The Advocate General’s reading of Kalanke looked instead to the “seeks to achieve” language as controlling. Jacobs observed that in both cases, preferences were awarded only where two candidates have equal qualifications. Echoing Tesauro, Jacobs therefore concluded that, by definition, the candidates already had equal opportunity, and that any award of preferences must therefore be viewed as aiming at imposing a result, rather than remedying deficiencies in the process.70 Although the “savings” clause may have still intervened to override the preference, Jacobs pointed out that its application was supposed to be limited to exceptional cases.71 Therefore, although such added flexibility may have mitigated the statute’s flaws, it hardly erased them. The statute remained impermissibly focused on outcomes, instead of opportunities. Accordingly, Advocate General Jacobs recommended ruling against the preference.72

F.  The Marschall Judgment

The Marschall Court started from the same principles and quoted the same law as the earlier opinions. However, it reached very different conclusions. One obvious change between Kalanke and Marschall is that the Marschall Court openly acknowledged the reality of discrimination in the workplace. It observed that:

[E]ven where male and female candidates are equally qualified, male candidates tend to be promoted in preference to female candidates particularly because of prejudice and stereotypes concerning the role and capacities of women in working life and the fear, for example, that women will interrupt their careers more frequently, that owing to household and family duties they will be less flexible in their working hours, or that they will be absent from work more frequently because of pregnancy, childbirth and breastfeeding.73

For these reasons, the ECJ recognized that “the mere fact that a male candidate and a female candidate are equally qualified does not mean that they have the same chances,” thereby rejecting the false equation [*PG50]of equal qualifications with equality of opportunity exhibited in the two Advocate General opinions.74

This belated recognition by the ECJ of the discrimination which female candidates face in the workplace made it possible to view some employment preferences as equalizing opportunities (by remedying an unequal process), instead of merely balancing outcomes. Yet, the question remains: How does one tell the difference? The ECJ was noncommittal. It stated only that preferences to women “may fall within the scope of Article 2(4) if such a rule may counteract [prejudice against] female candidates . . . and thus reduce actual instances of inequality which may exist in the real world.”75 Under this ambiguous formula, preferences could be viewed as going either to outcome or opportunity.76

To resolve this ambiguity, one might expect that the Marschall Court would have examined more closely the context in the which the preference functioned. This would have led the Court to face a further ambiguity which had confronted Tesauro: What is the significance of the mere fact of underrepresentation? Yet, the Marschall judgment did not discuss underrepresentation or any other contextual criteria. Instead, it focused on the savings clause. Somehow, the addition of the savings clause permitted the preference to be deemed a process remedy within the scope of Article 2(4) of the Equal Treatment Directive.

The judgment of the Marschall Court did not explore what additional criteria could warrant invoking the savings clause, but noted only that the Land chose “a legally imprecise expression in order to ensure flexibility.”77 The Court accepted this approach, although it required that “the candidates [be] the subject of an objective assessment [that] take[s] into account all criteria specific to the individual candidates” and also stipulated that “such criteria are not such as to discriminate against female candidates.”78 The Court failed to explain [*PG51]how undertaking such an assessment would resolve the ambiguity that it found fatal in Kalanke.79 Nothing about “criteria specific to the individual candidates” could be indicative of gender bias, which by its very nature is a group phenomenon.80 Although less “automatic,” the preference was no less arbitrary.

Formally, the Marschall opinion continued to label Article 2(4) of the Directive as a derogation, which would imply a strict interpretation.81 However, the judgment appeared anything but strict. Instead, the Court used highly hedged language (“may . . . if . . . may”) to skirt one ambiguity and entirely ignore another. Moreover, in relying on the savings clause in Marschall to distinguish Kalanke, the ECJ let the largely symbolic value of individualized review take the place of a more substantive resolution of the underlying issues.

II.  Rationales for Preferential Treatment

Notwithstanding these theoretical deficiencies, the ECJ has adhered to its formula of accepting employment preferences that are based on an “objective assessment” of individual candidates,82 and rejecting those that are “automatic” (i.e., not subject to individualized review)83 in two subsequent cases. In doing so, the ECJ appears to [*PG52]have opted for a solution much like that introduced by Justice Powell of the U.S. Supreme Court in his celebrated Bakke opinion.84 Although rejecting the use of strict racial quotas in admissions to medical school, Justice Powell sanctioned consideration of race as a factor that could be taken into account among other criteria specific to each individual candidate.85 Although other members of the Supreme Court observed that such a method would in practice yield the same result as a quota,86 the symbolic appeal of Powell’s individualized approach was undeniable, and Powell’s opinion has remained, in practice, the law of the land ever since.87

Justice Powell, however, justified preferences to underrepresented groups based on an underlying theory as to the value of diversity in a university context.88 Positive Action under Article 2(4) of the Equal Treatment Directive, in contrast, is premised on a remedial rationale—promoting equal opportunity by remedying procedural bias in employment decisions—a very different rationale. In this respect, an analogy may be drawn with another line of U.S. affirmative action cases that also considered preferences justified on remedial grounds. In Croson v. City of Richmond,89 the Supreme Court overturned a preference scheme that purported to remedy the effects of discrimination [*PG53]against minority-owed companies in bidding for municipal contracts. The Court held that evidence of societal discrimination against such minorities, by itself, would not suffice to justify such preferential treatment. Instead, the Court required that government actors muster “particularized” evidence of discrimination against specific minority groups in the precise context where the preference would apply.90

In making this distinction between societal and particularized discrimination, the Supreme Court, speaking through Justice O’Connor, emphasized that its rejection of the former should not be read to minimize the widespread social injustice that many minority groups had faced. The problem with societal discrimination was that its contours were too “amorphous” to be measured with the precision required for a judicially-determinable remedy.91 Without more particularized evidence, Justice O’Connor worried that courts could not make “[p]roper findings . . . to define both the scope of the injury and the extent of the remedy necessary to cure its effects.”92 Their inability to do so raised the “danger that a racial [preference could be] merely the product of unthinking stereotypes or a form of racial politics.”93 Therefore, while not denying that minorities had been treated unequally, the Supreme Court refused to accept this as a constitutionally cognizable injury because the Court could not be sure enough of the facts to control the remedy.94

Throughout these affirmative action cases, the Supreme Court has underscored the profound distrust with which it views racial classifications of any kind. While defenders of affirmative action on [*PG54]the Court have argued that a less stringent standard of review is merited where racial classifications are undertaken for “benign” purposes than that applied in earlier racial segregation cases,95 the majority has rejoined that “strict scrutiny” applies to all uses of race precisely because its benign nature cannot be taken for granted.96

It is interesting to note that Advocate General Tesauro made extensive citation to American affirmative action case law in his Kalanke opinion.97 His doing so was quite unusual for an E.C. opinion.98 One may speculate that in citing the U.S. cases, Tesauro was gesturing toward the jurisprudence of distrust that this case law embodies. That both his opinion and the judgment of the ECJ apply a “strict” interpretation of Article 2(4) of the Directive as a derogation from equal treatment suggests, at minimum, a discomfort with ascriptive, status-based norms that parallels their U.S. counterparts.

In ruling out societal justifications for affirmative action, however, the Supreme Court acted based on second order criteria, namely its concern over the limitations of judicial review, rather than any real principle of equality. In doing so, the Court makes a powerful statement of centralized authority from its position at the apex of an integrated federal judiciary. I will argue that this model is inappropriate in Europe, where the ECJ engages in a very different relationship with the national judiciaries of the E.C. Member States. In rejecting the Bremen preference, the Kalanke Court assumed the prerogative to pass judgment on a question better left to the national courts.99

III.  Equality as a European Right

The controversy that accompanied the ECJ’s Kalanke decision has been only partially tempered by its Marschall opinion. Although Mar[*PG55]schall “corrected” the result of Kalanke by imputing a different purpose to the preference at issue, it perpetuated the analytic framework and centralized mode of decision-making which Kalanke established. In this respect, despite their superficial difference in outcome, the two opinions have more in common than may be initially apparent.

A.  Article 2(4) as a Derogation

As a starting matter, the ECJ’s decision in Kalanke to construe Article 2(4) of the Equal Treatment Directive as concerned with opportunities and not outcomes is both reasonable and textually supported.100 The controversy revolves around how the ECJ distinguishes between them. Amendments to employment criteria to remove or offset gender-specific effects seem clearly permissible as means to equalize opportunities.101 Likewise, a decision to strike down a fixed 50–50 gender quota solely concerned with outcomes would probably not be questioned102 Kalanke and Marschall represent more difficult cases because the preferences were less intrusive instruments that arguably could have served a remedial function.

The key move that the ECJ makes to resolve this ambiguity in Kalanke is the decision to treat Article 2(4) of the Directive as a derogation from the Equal Treatment principle of Article 2(1). This approach allows the ECJ to enforce a “strict interpretation” by which ambiguities are construed against the preference and the statute fails unless shown to be remedial. In imposing this “strict” construction, both Tesauro and the Court cite Johnston v. Constable for authority.103 Johnston was a case interpreting Article 2(2), which grants an exception to equal treatment for occupations “in which the sex of the [*PG56]worker constitutes a determining factor.”104 Such a provision operates in the traditional fashion of a derogation running contrary to the very principle of equal treatment.105 Therefore, the Court was correct to adopt a restrictive view in that case. To import such logic to Kalanke, however, is arguably to misunderstand the complementary nature of the right that Article 2(4) promotes. Instead of citing Johnston, the Court might have followed the earlier example of Hofman v. Barmer Ersatzkasse,106 in which Advocate General Darmon pursues this reasoning to its logical conclusion:

The exception set out in Article 2(4) is in a category of its own. . . . It merely appears to make an exception to the principle [of equal treatment]: in aiming to compensate for existing discrimination it seeks to re-establish equality and not to prejudice it. In other words, since it presupposes that there is an inequality which must be removed, the exception must be broadly construed.107

In substance, Darmon’s position mirrors that of the U.S. Supreme Court minority that advocated a more lenient standard of re[*PG57]view in the affirmative action cases. By constructing Mr. Kalanke’s right to equal treatment as the fundamental right recognized by the directive and Ms. Glissmann’s claim to positive action as a derogation from that right, the ECJ rejects this more permissive model and elects to strictly construe Article 2(4) in a manner no different than Article 2(2).

A number of commentators have criticized this approach as reflecting a truncated, and overly procedural conception of equality.108 They see the equality right under European law as encompassing divergent ideals which stand in tension with one another.109 Scholars argue that, by failing to distinguish between Articles 2(2) and 2(4) in Kalanke, the ECJ “does not do justice to preceding Community law.”110 By contrast, the Marschall decision is seen as a shift toward a more progressive reading of equality, whereby the Court acknowledges the “‘lived experience’ of women in the workplace.”111

These analyses overlook the fact that, despite its effective reversal of Kalanke’s outcome, the Marschall Court continues to construe Article 2(4) as a derogation, implying, at least formally, that a “strict interpretation” applies.112 Cases following Marschall have confirmed that [*PG58]derogations from equal treatment require strict review.113 Although the degree of rigor clearly varies according to the case at hand, the appearance of strict review in gender equality cases marks a departure from earlier judgments in which the ECJ stressed the “margin of discretion” that Member States enjoy in implementing directives on social policy.114 A more nuanced understanding of strictness in the European context holds the key to unraveling the conceptual incoherence of the positive action cases.

B.  Strict Review

The usual test applied by the ECJ in reviewing a measure justified under a derogation is that of proportionality, which has three parts: suitability, necessity, and proportionality strictu sensu. Suitability asks whether the measure accords with the purpose of the derogation. Necessity investigates whether the chosen means are necessary to pursue this goal. Proportionality strictu sensu then balances the goal of the derogation against the burdens that it may engender.115

[*PG59] Although the ECJ has never spelled out what it means by “strict” review, this seems to describe the intensity with which the proportionality inquiry is conducted, with the Court applying a more exacting standard to one or more prongs of the test.116 The ECJ most often lays its stress on the second prong, necessity.117 A strict reading of necessity may require showing that no less intrusive measures were available. Thus, Advocate General Jacobs comments in Marschall that “a gender-specific measure will not to my mind be proportionate . . . if the same result could be achieved by a gender-neutral provision.”118 The ECJ’s insistence on flexibility in the positive action cases, however, most likely stems from its assessment of proportionality strictu sensu. In addition to the symbolic value of treating candidates as individuals, flexibility ensures that the preference regime will not penalize “innocent” males unduly.119

An unstated aspect of strict review is skepticism. Justifications for derogations may no longer be taken on faith because “what appears to be a [remedial] measure . . . may easily turn into a measure which is perpetuating and even legitimizing the traditional division of roles between men and women.”120 Instead, “exceptions . . . must be sufficiently transparent so as to permit effective supervision by the Commission [and the courts].”121 This means that potentially dis[*PG60]criminatory measures must be “governed by [] objective criteri[a] defined in a legislative provision.”122

In applying this intense scrutiny of both ends and means, European strict review resembles the “searching judicial inquiry” of American strict scrutiny.123 In both cases, the proceedings are animated by a distrust of political actors operating in the name of protecting individual rights.124 A basic difference between American and European methodology, however, concerns the procedure for preliminary references. In exercising this shared jurisdiction, the ECJ often defers to the national courts to apply proportionality; it does so habitually in its indirect discrimination cases.125 This display of judicial comity comports with Article 234 (formerly article 177), which reserves questions of fact to be determined by the referring court.

The U.S. judicial system, by contrast, permits far greater latitude for centralized adjudication of constitutional rights. The Supreme Court’s direct appellate jurisdiction over both state and federal courts gives it unquestioned control over all facets of federal law, including fact finding and remedies.126 In Kalanke and Marschall, the ECJ appears to arrogate to itself a similar prerogative to rule on proportionality as a matter of Community law. As will be suggested below, in doing so, the ECJ seems to resemble more an American than European court by acting outside of its role in a supranational regime. The key difference between Kalanke and Marschall concerns their reading of [*PG61]purpose: did the preference aim at outcomes or opportunity? Unfortunately, as seen earlier, the ECJ never resolved this fundamental ambiguity. In Kalanke, Tesauro incorrectly assumed that qualifications are synonymous with opportunity. In Marschall, the ECJ recognized that such assumptions could be belied by gender bias, but saw no need to identify the context in which such bias existed. Moreover, in endorsing the deliberately opaque wording of the savings clause, the ECJ violated its own rule that prima facie discrimination can only be justified by objective criteria, not generalizations.127

To some degree, the ECJ appeared to remedy this latter defect in Abrahamsson, where it “emphasise[d] that the application of [selection] criteria must be transparent and amenable to review . . . based on clear and unambiguous criteria.”128 However, the ECJ seemed no clearer in its fourth preliminary ruling on the subject of these criteria than it was in its first.129

In Badeck, the ECJ more explicitly embraced the opportunity/outcome dichotomy, introduced by Advocate General Tesauro, to justify strict gender quotas on allocations of employment training and job interview slots.130 Yet, it has continued to encounter difficulties applying this paradigm to preferential selection schemes. In Abrahamsson, the ECJ was clearly troubled by the fact that, unlike the prior cases, preferences could be granted even though the beneficiary had slightly inferior qualifications. The judgment comes close to suggesting that such preferences must be rejected as solely [*PG62]outcome-oriented.131 Yet, the same arguments about bias used in Marschall to get around Kalanke would seem applicable here: assessment of qualifications can just as easily be infected by bias as the selections based upon them.

It is also telling that the ECJ’s reasoning on this point was intertwined with condemnation of the “automatic” operation of the preference and its lack of individualized recourse to objective criteria. As in Kalanke, the Court thus continued to look at the form of the statute as indicative of its function. A similar approach was taken by the Badeck Court, which endorsed the use of “binding targets” for female advancement and equal representation of men and women on supervisory and administrative bodies. The Court based its decision on the ground that such quotas were flexibly implemented, without bothering to verify that they had a remedial purpose and were necessary in light of that goal.132

In structuring the analysis in its positive action rulings around these twin dichotomies—outcome versus opportunity, automatic versus flexible—the ECJ’s reasoning has assumed an overly formal character that seems inconsistent with the strict proportionality analysis required for derogations from an individual right. The inability of the ECJ to articulate a more convincing framework to assess the proportionality of positive action arguably stems from its failure to consider the context in which the court applied the preference statutes.133 Only [*PG63]by doing so can one speak meaningfully about their purpose and effects.134

C.  Context

There are many contextual criteria which the ECJ could have examined. To begin with, built into the statutes was the precondition of gender imbalance that was itself a prima facie indication of bias.135 The ECJ seemed to ignore this implication of underrepresentation, dismissing gender balancing in Kalanke as purely a matter of outcome, and omitting any discussion of underrepresentation in Marschall, where references to gender bias remain tentative and generalized.136 Tesauro’s was the only opinion to acknowledge, albeit grudgingly, the potential link between gender imbalance and bias.137

Of course, evidence of gender bias need not be confined to a specific workplace. The important factor is the establishment of a context in which proportionality can be assessed on the basis of objective criteria. By ruling against the Bremen preference on principle, Kalanke made the question of context irrelevant.138 Marschall held the opposite, but its ruling was just as peremptory.139 It assumed that gender bias is so pervasive as to legitimize preferences Community-wide. In neither case did the Court permit much, if any, leeway upon re[*PG64]mand to the national court to judge the preferences within the national context.140

Yet, because Article 2(4) of the Directive grounds positive action in an explicitly remedial rationale, attention to context is critical. It may well be that gender bias is pervasive; however, accepting this notion does not end the proportionality analysis. The problem may vary in kind and degree in different contexts, as may the availability of alternative remedies,141 making the proportionality of the remedy dependent on the facts of the specific case.142 The Marschall Court ignored these considerations in favor of a bright-line rule.143

The later judgments followed roughly the same pattern, but in the reverse. Badeck upheld a raft of supporting measures beyond preferential selection, all as a matter of European law.144 At times, the ECJ seemed to gesture toward proportionality assessment,145 but it did not address the issue of necessity, nor did it charge the national court with this task. Abrahamsson rejected the Swedish preference a priori, once again with no consideration of context.146

In issuing these Community-wide judgments, the ECJ followed the example of the U.S. Supreme Court, whether consciously or not.147 In doing so, the ECJ broke with earlier cases in which it de[*PG65]ferred to the national court on such matters. Even in Johnston, the precedent on which Kalanke relied for its “strict” interpretation of the derogation, the judicial review exercised at the supranational level was minimal. The Johnston Court intoned some warnings about the need for periodic reassessments of “social developments,” but left it entirely to the national courts to ensure that the justifications proffered in that case were “well founded.”148 Despite its claim to strictly enforce individual rights, the ECJ thus gave the United Kingdom carte blanche to defend overt discrimination against policewomen in its own home court based on amorphous and unproven assertions of security risks.149

IV.  Judicial Subsidiarity

There are sound reasons for the ECJ to defer to national courts in assessing the justification for positive action. The European Union (E.U.) may aspire to become “an ever closer union,”150 but the current status of Community law is far from the integrated federal system over which the U.S. Supreme Court presides. A U.S. model of centralized adjudication may be therefore inappropriate for positive action cases in Europe.

The direct appellate jurisdiction exercised by the Supreme Court gives it the last word on questions of federal law, from fact-finding to remedies, even for cases arising in state court.151 American affirmative [*PG66]action cases bear the imprints of this centralized authority. The Supreme Court’s decision in Croson, which rules out societal discrimination as a justification for preferential treatment based on second order prudential concerns, provides a strong example of the top-down control that the Supreme Court exercises over constitutional law.152

In comparison, the ECJ stands in a very different relationship with the national judiciaries.153 “The constitutional discourse in Europe must be conceived as a conversation of many actors . . . rather than a hierarchical structure with the ECJ at the top.”154 The foundation of this relationship is the Article 234 procedure for preliminary references, which accounts for well over half of the ECJ’s caseload, and almost all of its gender equality cases.155 Under Article 234, the ECJ is restricted to answering questions on E.U. law certified to it by national courts. Jurisdiction to make factual determinations, to apply the law to the facts, and to determine remedies remains with the referring court.156 These jurisdictional constraints limit the ECJ’s ability to impose its fiat unilaterally. Although the ECJ’s supremacy on issues of E.U. law is no longer challenged overtly, the ECJ depends on national courts both to refer cases and to implement its rulings afterward, which gives national courts subtle powers of resistance.157 Lack[*PG67]ing appellate enforcement power, the ECJ must therefore engage in a discourse of comity and mutual respect in its rulings, if only to encourage national court compliance.158

At some level, the ECJ must also take account of the subsidiarity principle, the E.U.’s basic principle of federalism. Elevated to the status of a fundamental norm of E.U. law at Maastricht in 1992, the subsidiarity principle requires that the Community act only “in so far as the objectives of the proposed action cannot be sufficiently achieved by the Member States and can therefore, by reason of the scale or effects of the proposed action, be better achieved by the Community.”159 Viewed primarily as a rule of comparative efficiency, the rule has a normative political element as well. As expressed in the preamble to the Treaty on European Union, it calls for “decisions [to be] taken as closely as possible to the citizen.”160

The ramifications of subsidiarity on Community law remain unsettled. Although it has confirmed that the principle is justiciable, the ECJ has yet to imbue subsidiarity with much substantive content.161 Most scholarly analysis to date has focused on the way in which subsidiarity might operate as a legislative constraint.162 Yet, subsidiarity considerations would also seem applicable to judicial action taken by [*PG68]the ECJ.163 Even if it is not legally bound by subsidiarity, the Court arguably still has to take account of the constitutional values which the principle embodies.164 Indeed, the subsidiarity principle was introduced into E.C. law in part as a reaction to activist rulings from Luxembourg.165

The division of judicial roles under Article 234 might be said to embody a kind of judicial subsidiarity, broadly (and anachronistically)166 understood, whereby responsibility for decision-making is allocated between national courts and the ECJ according to comparative institutional expertise.167 One example of this logic may be the level at which the proportionality test is applied. Commentators suggest that the ECJ is prone to deferring to the national courts to assess proportionality in cases of political sensitivity or highly complex or localized fact-patterns.168 Such delegation of decision-making authority to national courts permits decisions to be “taken as closely as possible to the citizen.”169

There are practical reasons for ECJ deference when it comes to positive action. The ambiguous nature of preferences, which aim to establish equality through means that on their face violate equal treatment, requires particular attention to context to ensure that they rest on genuine need, not stereotype. Moreover, the proportionality of such remedies hinges upon highly contingent factors: historical patterns of inequality, lingering effects, and availability of alternative forms of relief, among others. National contexts vary,170 and indeed, [*PG69]the relevant context will often be sub-national.171 National courts are better equipped than the ECJ to perform such fact-intensive evaluations, and under the division of labor established by Article 234, the task should be theirs.

To be sure, the ECJ can always interpret the law in a manner that resolves factual issues implicitly by generalizing, as it did in Marschall. Yet, by their very nature, such omnipurpose rulings rest on unsubstantiated assumptions incompatible with the strict review the ECJ has professed.172 No one remedy can anticipate every contingency. Stronger measures may be justified in contexts where discrimination remains entrenched, especially when previous remedies have been ineffective. Such was arguably the case in Abrahamsson, where the ECJ rejected a preference scheme that went beyond those of previous cases without considering the contextual justifications offered by the scheme’s defenders.173 Conversely, preferential treatment may be unwarranted in fields where women have historically predominated, such as nursing.174

A different issue is raised by a judgment such as Kalanke, where the ECJ appears to have acted from reasons of principle, irrespective of the factual context. Clearly, the ECJ has the power to exclude certain justifications as a matter of E.U. law.175 Could the Kalanke opinion have done this regarding societal discrimination, following the example of the U.S. Supreme Court in Croson? An analogy can be made to [*PG70]the ECJ’s indirect discrimination case law, which also deals with structural factors that disadvantage women. Recognizing the heavily factual nature of such investigations, the ECJ has consistently deferred to the national courts in these cases to assess justifications and apply proportionality.176 However, the ECJ has ruled out some justifications on principle as non-transparent or overly generalized.177

Yet, to extrapolate from those rulings a decision as far-reaching as Croson would be rash. The ECJ excluded the justifications in those cases because they masked other, more objective criteria that could have been examined instead.178 The U.S. Supreme Court did not reject societal discrimination in Croson because of a lack of objective criteria supporting it. Rather, it worried that an overabundance of evidence would make it impossible to control the objectivity of the lower courts’ fact-finding used to justify preferential remedies. Yet, unlike its U.S. counterpart, the ECJ has no general jurisdiction over fact-finding.179 Moreover, such a strongly centralized ruling would violate the institutional norms within which the ECJ operates as the central organ of a supranational judiciary. Both the prudential understandings implicit in Article 234 and judicial subsidiarity principles outlined above weigh heavily against it.

This is not to say that the ECJ should have no role in vetting positive action plans. It may rule some things out of bounds entirely. For example, in rejecting outcome-based justifications for preferences, the ECJ acted within its prerogative to interpret an E.C. directive. Deciding which preference regimes operate in the realm of outcomes versus opportunities is a different matter, likely to be settled only through application of the proportionality test. Here, the ECJ should frame the analysis in legal terms and, if possible, identify the factors to [*PG71]balance.180 By supplying extremely detailed guidelines, the ECJ can emphasize that strict review is required.181 The ultimate analysis should rest, however, with the national courts.

Devolving more of such decisions to national courts would help to reconcile the “strict interpretation” demanded by a derogation against an individual right, with the flexibility that Community law has traditionally allowed Member States implementing a directive on social policy.182 The difference would be that, instead of deferring to the discretion of national governments, the ECJ would defer to national judiciaries.

V.  Fundamental Rights: Cooperation vs. Conflict

Along with Johnston, the positive action cases mark the initial stages of the ECJ’s “constitutionalization” of gender equality, extending to gender cases the same “strict interpretation” that the ECJ applies to economic expressions of the equality right. The change can be linked, at least rhetorically, to the ECJ’s fundamental rights jurisprudence. The ECJ has made this linkage explicit in its social security cases.183 In Johnston and Kalanke, the Court merely refers to equality as an individual right contained in a directive, but the fundamental rights dimension emerges in the Advocate Generals’ opinions. Advocate General Darmon, writing in Johnston, states flatly that “a derogation from a human right as fundamental as that of equal treatment [*PG72]must be appraised in a restrictive manner.”184 Tesauro’s opinion in Kalanke terms equality “a fundamental right the observance of which [Community law] ensures” and invokes the catch-words of the ECJ’s fundamental rights jurisprudence, describing equality as both “a fundamental value of every civil society”185 and “a principle which is safeguarded constitutionally in most of the Member States’ legal systems.”186 Previously, the ECJ had enforced such strict interpretations only in its free movement cases, where the equal treatment norm was instantiated in the E.C. Treaty.187 By treating the right to gender equality as “fundamental,” these opinions thus place it on par with the quasi-constitutional force of the “fundamental liberties” of free movement.

Strict review in gender cases, however, cuts against the margin of discretion traditionally granted to Member States implementing directives on social policy.188 Constitutionalizing gender equality has therefore added to the already swelling controversy that surrounds E.U. fundamental rights. In Kalanke, the ECJ construed equal treatment as a Community norm to reverse the outcome reached under an almost identical provision of the German Constitution.189 The provision in question had been amended specifically to permit positive action following a decision by the German Constitutional Court endors[*PG73]ing positive measures in pursuit of gender equality.190 Inevitably, commentators saw the Kalanke decision as “food for a new fundamental rights conflict” between the ECJ and Germany’s Constitutional Court.191

Unsurprisingly, the Kalanke decision prompted muttering in German accents about the need to construe fundamental rights in light of subsidiarity.192 In Marschall, the ECJ defused these tensions by smuggling in Member State discretion through the back door of the savings clause.193 However, the Court’s adherence to a formal structure that treats Article 2(4) of the Equal Treatment Directive as a derogation requiring strict review promises further conflicts to come. Originally conceived as a shield against Community action, E.U. fundamental rights, strictly enforced, will increasingly function as a sword that cuts into the constitutional autonomy of the national judiciaries.194

The inevitable conflicts that will be generated will continue to press the issue of the relevance of subsidiarity in the field of funda[*PG74]mental rights.195 Some commentators have proposed borrowing a “margin of appreciation” approach from the European Court of Human Rights to accommodate differing conceptions of equality among the Member States.196 This article has suggested, however, that at least in the case of positive action and other contextually contingent questions, a potential for diversity already exists within the present structure of Community law. The distinctive roles which the two levels of E.U. judiciaries play under the Article 234 procedure lend themselves naturally to the application of subsidiarity principles. In contrast to the usual portrayal of fundamental rights as an arena in which the ECJ and national courts compete in a zero-sum game,197 this provides a model of judicial subsidiarity in which the two levels can be viewed as allies, not antagonists, in their common role as protectors of individual rights.198

Conclusion

This article has revealed the shortcomings of the European Court of Justice’s positive action jurisprudence. In relying on “flexibility” as its guiding principle to reconcile individual rights with the competing equality claim of working women, the ECJ has substituted a facile formula in place of the substantive analysis that its putative strict review would imply. Moreover, in ruling on these issues at the Community level, the ECJ has precluded the sensitivity to context that would have been possible had it left the assessment of proportionality to the national courts. These defects stem, in part, from a failure of the ECJ [*PG75]to give adequate consideration to its institutional role in interpreting a Community right through its Article 234 interlocution with national courts.

In the years since Kalanke and Marschall, several amendments to the E.C. Treaty have altered the landscape of gender equality law. The Amsterdam Treaty included a specific provision, Article 141(4), designed to reverse the result of Kalanke through a somewhat incongruous appendage to what started as the Equal Pay Article of the E.C. Treaty.199 After the volte face executed by the ECJ in Marschall, this provision has seemed perhaps superfluous.200 In Abrahamsson, the Court declined to give Article 141(4) any independent substantive meaning, although it remains open for it do so in the future.201

The Amsterdam Treaty has also led to greater Community involvement with social equality. Article 13 of the Treaty provides the E.C. with broad enabling powers to enact anti-discrimination measures protecting specified status groups.202 A pair of new directives passed under this Article have added race, ethnicity, religion, belief, disability, age, and sexual orientation to the list of categories protected from discrimination by E.C. employment law.203 The E.U.’s new Charter of Fundamental Rights contains its own expansive non-discrimination and positive action clauses.204

[*PG76] This continued “constitutionalizing” of equality in Community law will put added pressure on the ECJ to balance the vigilant scrutiny needed to protect fundamental rights against a respect for Member State sovereignty in social policy . The recently enacted Race Directive, in particular, will require even greater attentiveness to context than is the case with gender because race as a construct is much more contextually contingent.205 To perform this juggling act, the ECJ will need to pay closer attention to managing its cooperative partnership with the national judiciaries. As the ECJ continues to articulate the institutional logic underpinning this shared jurisdiction, it may be possible to develop a viable conception of judicial subsidiarity.206

?? ??