Disparate Impact, Unified Law.

• Author: Stephanopoulos, Nicholas O.

AUTHOR. Professor of Law and Herbert and Marjorie Fried Research Scholar, University of Chicago Law School. I am grateful to Charles Abernathy, Christopher Elmendorf, Joseph Fishkin, Owen Fiss, Aziz Huq, Vicki Jackson, Janai Nelson, George Rutherglen, Robert Schwemm, Stacy Seicshnaydre, Michael Selmi, and Charles Sullivan for their helpful comments. My thanks also to the workshop participants at Harvard Law School, the University of Chicago Law School, and the Washington and Lee School of Law, where I presented earlier versions of the Article. I am pleased as well to acknowledge the support of the Robert Helman Law and Public Policy Fund.

ARTICLE CONTENTS INTRODUCTION 1569 I. VOTE DENIAL DOCTRINE AND ITS DEFECTS 1574 A. The Emerging Judicial Consensus 1575 B. The Academy's Approval 1580 C. Unanswered Questions 1582 1. Specific Practice or Entire System? 1582 2. Does the Size of the Disparity Matter? 1583 3. Ability to Comply or Effect on Turnout? 1584 4. Is Interaction with Discrimination Necessary? 1585 5. Are Minority Preferences a Defense? 1586 6. How Does Tenuousness Work? 1587 7. What Is the Remedy? 1588 D. Looming Concerns 1589 II. UNIFYING DISPARATE IMPACT LAW 1595 A. The Usual Framework 1596 B. Applicability to Voting 1601 1. Legislative Histories 1601 2. Theoretical Accounts 1604 3. Nature of the Activity 1607 C. Answered Questions 1609 1. Specific Practice or Entire System? 1610 2. Does the Size of the Disparity Matter? 1611 3. Ability to Comply or Effect on Turnout? 1613 4. Is Interaction with Discrimination Necessary? 1614 5. Are Minority Preferences a Defense? 1616 6. How Does Tenuousness Work? 1617 7. What Is the Remedy? 1619 D. Resolved Concerns 1621 III. OBJECTIONS AND RESPONSES 1625 A. Text and Precedent 1626 B. Constitutional Convergence 1630 C. Prior Record 1632 IV. APPLICATIONS 1636 A. Photo ID Requirements 1637 B. Early-Voting Cutbacks 1640 C. All-Mail Voting 1643 CONCLUSION 1646 INTRODUCTION

Say a state passes a law that makes it harder to vote, like a requirement to show photo ID or a cutback to early voting. (This is not a far-fetched scenario; more voting restrictions have been enacted over the last decade than at any point since the end of Jim Crow. (1)) Suppose also that the state's new law has a disparate racial impact: that it affects a higher proportion of minority than nonminority citizens. (This too is a plausible assumption; minority citizens are less affluent, on average, and so more disadvantaged by measures that increase the resources required for voting. (2))

Courts and commentators agree that, standing alone, this racial disparity does not breach section 2 of the Voting Rights Act (VRA), the key statutory provision banning racial discrimination in voting. (3) In the oft-repeated words of one district court, "a plaintiff must demonstrate something more than disproportionate impact to establish a Section 2 violation." (4) Or as a prominent VRA litigator has put it, "Section 2 plaintiffs [must] establish... that the disparate impact of a challenged vote denial practice is not merely a statistical accident." (5)

Now imagine that a litigant does come up with "something more" than a naked racial disparity--specifically, evidence that the disparity is caused by the law's interaction with historical and ongoing patterns of racial discrimination. In a photo ID case, this evidence might show that minority citizens are poorer than nonminority citizens; that their relative poverty is the product of discrimination; that because they are poorer, they are less likely to own cars; and that because they drive less, they are also less likely to have driver's licenses. In an early-voting case, the causal chain might run from discrimination to worse job qualifications to less flexible work conditions to greater difficulty voting on Election Day to heavier reliance on early voting.

Given this land of record, the emerging judicial consensus is that section 2 is violated. In the Fourth, Fifth, Sixth, Seventh, and Ninth Circuits--though not necessarily in the Supreme Court, which has yet to decide a vote denial case under the VRA--liability ensues if an electoral policy (1) has a disparate racial impact that (2) is attributable to the policy's interaction with discriminatory conditions. (6) And properly so, according to many scholars. To cite a high-profile pair, Sam Issacharoff lauds the courts' two-part test as a "breakthrough," (7) while Pam Karlan extols its capacity "to disrupt politics as usual in the service of full civic inclusion for long-excluded minority citizens." (8)

But there is a problem with construing section 2 in this fashion. Section 2 is a disparate impact provision--a law that imposes liability on the basis of discriminatory effect, not invidious intent. Section 2, however, is not the only such provision. Rather, disparate impact theories are also recognized by Title VII of the Civil Rights Act, by the Fair Housing Act (FHA), and by several more statutes. (9) In these other areas, a violation is not established simply because a policy interacts with discriminatory conditions to produce a disparate impact. Instead, courts follow a well-developed framework under which (1) the plaintiff must prove that a particular practice causes a significant discriminatory effect; (2) the defendant next has the opportunity to show that the practice is necessary to achieve a substantial interest; and (3) the plaintiff may then demonstrate that this interest could be attained in a different, less discriminatory way. The emerging consensus in the vote denial context thus threatens to drive a wedge between section 2 and every other disparate impact provision. It risks turning section 2 into a lonely island in the disparate impact sea.

My aim in this Article, then, is to resist this consensus--to urge consistency rather than variety in disparate impact law and to bring vote denial cases into the familiar disparate impact fold. To state my thesis another way: to date, courts have focused on the explanations for racial disparities in voting, especially the extent to which they are intertwined with past and present discrimination. In my view, though, courts should scrutinize the interests that allegedly justify these electoral disparities: how compelling they are, the degree to which they are advanced by challenged practices, and whether they could be furthered through other means. That is what courts do in every other disparate impact domain, and I see no good reason for vote denial cases to break the mold.

Beyond simplicity, what is the case for unifying disparate impact law? One set of answers stresses the similarities between voting and employment, housing, and the other fields where disparate impact claims may be raised. As a historical matter, the VRA, Title VII, and the FHA are kindred spirits: crown jewels of the civil rights era, enacted in a single burst of legislative activity, and sharing the mission of ending racial discrimination. It would be entirely consistent with these statutes' common legacy for disparate impact law to implement them in the same way.

Theoretically, too, the standard accounts of disparate impact law apply as squarely to voting as to employment or housing. One model treats an unjustified discriminatory effect as a proxy for the true concept of interest: a racially discriminatory purpose. Another approach deems a needless disparate impact an evil in and of itself: an unwarranted racial stratification in a society aspiring for racial equality. Both perspectives extend naturally to vote denial. When an electoral policy differentially affects minority and nonminority citizens, and for no good reason, the injury can be understood as an illicit motive that is suspected but not proven. The harm can also be seen as the unnecessary disparate impact, which prevents minority citizens from participating equally in the political process.

Substantively as well, voting resembles employment and housing in that it is a valued good to which access is determined by criteria that not everyone can satisfy. When these criteria disproportionally exclude minority members, they produce racial disparities whether they pertain to the franchise, the workplace, or the roof over one's head. It is true that voting (unlike employment and housing) is exclusively regulated by the state. But this only means that it is public rather than private interests that are the potential justifications for disparate impacts. It is also true that voting (again unlike employment and housing) is a nonmarket, nonrivalrous good: one with no price and no limit to who may enjoy it. This too, though, simply takes off the table one common rationale for racial discrepancies: private actors' pursuit of profit.

The next reason to unify disparate impact law is more doctrinal. Not only does the framework used by Title VII and the FHA apply fully to vote denial cases; it also resolves many of the most contentious issues that have arisen in these disputes. To name a few: Must section 2 litigants establish a large disparate impact, or will any discriminatory effect do ? The former, if Title VII is any guide. Courts enforcing that provision have insisted on a racial disparity that is both statistically significant and substantively meaningful before finding an employer liable. (10) What kind of disparate impact must be shown in section 2 litigation--a difference between minority and nonminority citizens' abilities to comply with an electoral policy or an eventual gap in voter turnout? Again the former, according to Title VII. In a well-known case, the Supreme Court rejected the "suggestion that disparate impact should be measured only at the bottom line." (11)

Must the racial disparity caused by a voting rule also be linked to social and historical discrimination? No, held the Court in another famous Title VII case (thus undermining one of the pillars of the emerging section 2 consensus). (12) It is enough if the...