Earlier this month, the Supreme Court heard oral arguments in Gill v. Whitford, a case in which University of Wisconsin professor William Whitford and a group of plaintiffs (all Democratic voters in the state) contend that the drawing up of Wisconsin’s state legislative districts was an unconstitutional gerrymander.
The Supreme Court has intervened in gerrymandering complaints in the past, but those cases had to do with racial discrimination or malapportionment. That is, the Court has struck down legislative maps that distribute black voters in ways that minimize their electoral power. In the Wisconsin case, race and ethnicity are not at issue, at least not directly. The plaintiffs are asking the Court to invalidate district lines drawn by the state legislature in 2011 because those lines favor Republicans over Democrats.
Based on the oral arguments, the case is another that divides the Court along ideological lines. The four conservative justices—Chief Justice John Roberts, Clarence Thomas, Samuel Alito, and Neil Gorsuch—seem skeptical that the Court should intervene in such matters. It is better, instead, to leave partisan gerrymanders to the political process. The liberals—Ruth Bader Ginsburg, Stephen Breyer, Sonia Sotomayor, and Elena Kagan—appear eager to involve the Court in the process. Once again, that leaves Justice Anthony Kennedy in the middle, the crucial fifth vote who will decide which way the Court swings.
Kennedy has previously expressed openness to the Court’s involving itself in resolving partisan gerrymandering. In Vieth v. Jubelirer (2004), though the Court declined to rule on the constitutionality of Pennsylvania’s congressional redistricting, Kennedy indicated that the justices might have a role to play if they could find a workable standard to apply.
This is what makes the Wisconsin case especially important. The plaintiffs claim to have put forward such a standard—based on a metric known as the “efficiency gap”—that, they argue, is a simple, intuitive, and broadly applicable tool to find the effects of a gerrymander.
However, the plaintiffs are wrong. The efficiency gap is an interesting but deeply problematic metric that should not be imposed by the judiciary. If Kennedy and the liberal justices were to adopt it as the centerpiece of a test to adjudicate partisan complaints about the drawing of political districts, it could have profound, unintended, and controversial consequences on legislatures throughout the country.
At first blush, the efficiency gap seems quite straightforward. The whole approach hinges on the concept of “wasted” votes. For instance, if a party gets 75 percent of the votes in state legislative contests but wins only 50 percent of the seats, then 25 percent of the vote has been “wasted,” or cast in excess of the final seat totals the party won. Generally speaking, the efficiency gap is calculated by comparing votes to seats, and the plaintiffs are asking the Court to use it to decide whether district maps are presumptively valid or invalid. They claim that if the efficiency gap exceeds a certain threshold, then it is likely because the party in charge of drawing the district lines has cheated the opposition.
There are a number of problems with this. The most immediate concern is that the efficiency gap has simply not undergone sufficient peer review. First introduced by Eric McGhee of the Public Policy Institute of California in 2014, in an article for the Legislative Studies Quarterly, it is a relatively new tool for understanding gerrymanders. Given the glacial speed at which the social sciences operate, the efficiency-gap metric is still in its infancy. While McGhee’s article is well argued and theoretically intriguing, the efficiency gap has simply not been put through its paces. Wendy K. Tam Cho, a professor of political science at the University of Illinois, has noted that to date, “its properties have not been rigorously explored.” In a brief essay evaluating it, she concluded that while it is “simple to articulate and may seem intuitive,” it is problematic. “A measure of partisan fairness,” she asserted, “needs to be more nuanced, and cognizant of the partisan context.”
The efficiency gap presents a number of technical problems that have not yet been worked out by scholars. Even its advocates do not think it should apply to smaller states—those with no more than seven or eight congressional districts—which means that should the Court adopt the efficiency gap as its standard, it will then have to find an additional metric for those states that works just as well.
There are computational issues, too. One of the biggest challenges is calculating the efficiency gap in the absence of sufficient data. For instance, when incumbents run unopposed, the efficiency gap will be hopelessly skewed. Strategies have been proposed that “impute” values in those cases—by, for instance, using presidential vote choices in a district where one party fails to field a candidate in a down-ticket race, to guess how a candidate from that party might have done. But these approaches are inevitably full of controversial assumptions. And again, there has been virtually no scholarly debate, let alone a consensus achieved, on how to handle situations like that.
There is even ambiguity on how to calculate the efficiency gap properly. Advocates for the measure have used different though related metrics, which produce different results. Relatedly, there is disagreement over what constitutes a partisan gerrymander even among the advocates of the metric. The final result of the efficiency gap is not a yes/no proposition. Instead, a researcher has to look at the result and interpret whether or not a gerrymander has occurred. There is no scholarly consensus on that, either, and once again even the experts for the plaintiffs disagree.
These technical problems are minor compared with some deeper limitations of the efficiency-gap concept. All else being equal, the efficiency gap as a measure of unfairness may make a lot of sense: A party’s share of legislative seats should correspond to its share of votes. But in the real world of American politics, all else is usually unequal.
Consider the geographical clustering of party voters, a factor that has grown more pronounced in recent decades. Democrats tend to be heavily concentrated in large cities and college towns, while Republicans are broadly distributed around the outlying areas. This means that an effort to draw a neutral map premised on, say, the idea that districts should represent communities of interest, would very likely run afoul of the efficiency gap. Democratic votes would be “wasted” in urban districts while Republican votes would be more efficient.
This is compounded by the dictates of the 1982 amendments to the Voting Rights Act, which have been interpreted to mandate that minority voters be given their own districts wherever it is reasonably possible. This requirement increases the likelihood that Democrats are packed into minority-majority districts, which means that simply following the federal law could increase the efficiency gap and help create the impression of a partisan gerrymander.
Then there is the problem of how to account for so-called campaign effects, which include the advantages of incumbency, the presence or absence of scandal, campaign spending, voter turnout, national waves. All of these and more can have a huge influence on the votes-to-seats ratio in any given election. If that ratio is pushed too far in one direction or another, it will make the efficiency gap signal that a partisan gerrymander has occurred.
The result of these various problems is that judging district lines by the efficiency gap will produce an intolerably high number of false positives and false negatives. That is, the efficiency gap will sometimes conclude that honest maps are partisan gerrymanders and sometimes conclude that partisan gerrymanders are honest maps. In short, this is not a formula that justices should feel empowers them to make nationwide judgments on what is and is not an impermissible partisan gerrymander.
Consider the case of Illinois over the last decade. In 2002, Republicans controlled the state senate and governorship while Democrats controlled the state house. Since neither party had full authority, a bipartisan redistricting plan was hatched that protected incumbents of both parties. And yet, according to the efficiency gap, this redistricting was a Republican gerrymander. Fast forward to 2012. Democrats by that point had taken complete control of the state government and put together one of the most aggressively partisan gerrymanders in the country. And yet, according to the efficiency gap, the map tilted just slightly toward the Democrats.
The reason for these errors has to do with the distribution of partisans in Illinois. Democrats are concentrated in Chicago while Republicans are spread more broadly across the state. So a neutral plan—such as the 2002 map—will inevitably look like a GOP gerrymander under an efficiency-gap analysis. On the other hand, an extreme Democratic gerrymander—such as was adopted in 2012—will look like a neutral plan. Not wanting to squander votes in their extremely liberal Chicago redoubts, Democratic legislators made sure to draw creatively shaped districts in the Chicago area that mixed urban and suburban areas to produce lots of reliably, but not overwhelmingly, Democratic seats.
Next, consider the state of Alabama, where the verdict rendered by the efficiency gap is even more bizarre. This is a state where the Voting Rights Act is enormously consequential. African Americans constitute about 30 percent of the state’s population—that’s substantial but still small enough that legislators given free rein could easily created a white majority in every congressional district. This is where the Voting Rights Act comes into play, to prevent such racially drawn maps from becoming law. But because the efficiency gap does not take this requirement into account, it gives a wrong sense of the partisan dynamics in Alabama.
Democrats held solid majorities in the Alabama legislature through the 1990s, but the lines they drew actually looked like a Republican gerrymander according to the efficiency gap—in every election cycle except 1994. The reason for the sudden shift had to do with campaign-related events on the ground that happened to disfavor Democrats that year. Democrats again controlled redistricting in the state in 2002 and were reasonably successful in drawing lines, but according to the efficiency gap it was once again a Republican gerrymander.
These examples get to the core theoretical problems of the efficiency gap as a standard of impermissible partisanship: It is a one-dimensional metric. Drawing fair district lines requires evaluating a lot of different variables. No doubt it is ideal for a party to get roughly the same share of seats as it does votes, but there are other factors at play as well. How about encouraging competitive races? How about guaranteed representation for historically underrepresented minorities? How about geographical compactness?
All of these other factors are ignored by the efficiency gap, which reduces fairness to a single votes-to-seats ratio. Such a standard would have enormous effects if the Court uses it to decide which maps are presumptively valid or invalid. If a state drew lines that promoted competitive races, and as a consequence its efficiency gap was too high, the burden of proof would be on the state to demonstrate that it had not gerrymandered. If, on the other hand, Democratic and Republican legislators craft a plan to protect incumbents, with a low efficiency gap, the burden of proof would be on advocates of good government to prove that it was a gerrymander. This is simply not the proper way to design a legal test.
And the examples of Illinois and Alabama further demonstrate that, contrary to the claims of the plaintiffs in Gill v. Whitford, identifying partisan gerrymanders remains a difficult and frankly subjective process. Worse than this, the efficiency gap is a kind of Trojan horse that threatens the uniquely American style of representative government. If the Court adopted the test as its standard, it would in effect be mandating a kind of proportional representation—similar to the systems of continental Europe. This has not been the American approach to legislatures, which instead are the meeting places of representatives from different constituencies, anchored by geography and communities of interest. Under an efficiency-gap standard, we would nominally retain such a system, but the Court would be requiring states to draw lines that effectively make us more like European parliaments.
One can, of course, argue that we should alter our system of representation in this way. Our approach to democracy may indeed be outdated. Maybe Europe has a better way. However, such changes are properly left to the people. If some state legislatures want to bind themselves in their redistricting efforts by some version of efficiency-gap analysis, more power to them. The Supreme Court, though, should not impose such a sweeping alteration by judicial fiat. Moreover, it should not pretend to be engaging in some neutral application of bulletproof social science findings.
To be certain, the efficiency gap is an interesting and useful metric. There is no doubt that it is measuring some of the political effects of gerrymandering. Judged alongside other metrics that have been offered to gauge gerrymandering, it is easily the best of the bunch. Its developers—Eric McGhee in particular—deserve credit for advancing the scholarly understanding of gerrymandering.
But the merits of the efficiency gap and its appeal to scholars are not the issues in this case. Instead, the question is whether the efficiency gap should acquire the force of law. The answer is no. The Court should not allow itself to be tempted into a sweeping change in our system of representation.
Justice Kennedy has long been looking for a workable standard to adjudicate claims of partisan gerrymandering. The efficiency gap, as ingenious as it may be, is simply not good enough.
Jay Cost is a contributing editor to THE WEEKLY STANDARD.
