Democrats know that they have a problem. As our country has become more politically polarized, we have increasingly sorted ourselves geographically. As Michael Barone has noted, “Democratic voting groups — blacks, Hispanics (in many states), and gentry liberals — tend to be clustered in most central cities, many sympathetic suburbs, and most university towns, while Republican voters are spread more evenly around the rest of the country.” We see this after every election cycle in maps depicting outcomes by county that show America as a sea of red with islands of blue. Although Hillary Clinton won the popular vote for president, she carried only 487 of America’s approximately 3,100 counties.
Given that we elect the U.S. House of Representatives and state legislatures in single-member, winner-take-all districts, the geographic concentration of Democratic votes gives Republicans a natural advantage in legislative races. After the elections in November 2016, the GOP controlled 67 of the nation’s 98 partisan legislative chambers. In the face of this handicap, Democrats have turned to the courts, arguing that Republican success is the result of — or at least enhanced by — partisan gerrymandering.
Historically, suits seeking to overturn legislative maps because they favor Democrats or Republicans have been a nonstarter. From the moment it first adopted a constitutional requirement of “one person, one vote,” in 1964’s Reynolds v. Sims, the Supreme Court has emphasized that “legislative reapportionment is primarily a matter for legislative consideration and determination.” The redistricting process requires weighing a variety of widely accepted objectives — contiguity, compactness, continuity, respect for political boundaries, keeping intact communities of similarly situated voters (such as rural voters), protecting the ability of minority voters to elect candidates of their choice, and even political compromise and the protection of incumbents — that often work at cross-purposes and produce maps that are the product of no identifiable set of principles.
Although constitutional objections to partisan gerrymandering remain theoretically possible, the Supreme Court has been unable to articulate a way to determine how much partisanship is too much. In 2004, in Vieth v. Jubelirer, four justices noted that redistricting litigation had created “one long record of puzzlement and consternation.” Justice Scalia, joined by Chief Justice Rehnquist and Justices Thomas and O’Connor, was prepared to declare that gerrymandering claims were political questions incapable of being adjudicated by the Court. The four rejected the plaintiffs’ appeal and declined to intervene in the redistricting plan (which was in Pennsylvania). In a concurring opinion, Justice Anthony Kennedy agreed that no standard for adjudicating gerrymandering claims had yet been identified, but he was unwilling to foreclose the possibility that someday the courts would find one.
Last year, a group of Democrats in Wisconsin convinced a divided three-judge panel that they had finally located the grail that has eluded detection for the better part of half a century. In Gill v. Whitford, the district-court majority conceded that maps drawn by a GOP majority in the Wisconsin legislature respect traditional redistricting criteria. They are contiguous and compact. They do not resemble a salamander, ink blot, or sacred Mayan bird. But the court nevertheless held that they constituted a partisan gerrymander that unconstitutionally favored Republicans.
The standard that supposedly solves the puzzle that has stumped the courts for so long turns on something called the “efficiency gap.” Calculation of the gap begins with the assumption that all votes cast for a losing candidate, along with any votes cast for a winning candidate that exceed the number of votes necessary for that candidate to win, are “inefficient” or “wasted.” If more “wasted” votes have been cast for the candidates of one party than have been cast for the candidates in another party, there is an “efficiency gap” — i.e., the votes that were cast for candidates of one party were more “efficiently” translated into legislative seats than were the votes for candidates of the other party. Put in more concrete terms, the plaintiffs complain that Republican candidates for the state assembly in Wisconsin won 48.6 percent of the votes cast for assembly candidates in 2012 but 60.6 percent of the seats. In 2014, GOP candidates won 52 percent of all votes but 63.6 percent of the seats.
Gill is now before the Supreme Court. In oral arguments earlier this month, Chief Justice Roberts suggested that this “new” measure might be “sociological gobbledygook.” He could be on to something. It is not that the efficiency gap measures nothing. It does tell us something about differences in the concentration of each party’s voters within districts. But it does not tell us why they are concentrated. There is no reason to presume that voters who prefer Democrats or Republicans will be distributed evenly across a state or that each party’s voters will be geographically concentrated in equal measure. For example, the plaintiffs in Gill argue that any efficiency gap in excess of seven percentage points indicates a problem. Yet under court-drawn maps in place prior to the drawing of the maps being challenged, the pro-Republican efficiency gap averaged 7.6 percentage points.
Of course, the fact that the GOP might have a geographic advantage in the redistricting process doesn’t mean that this tendency of Democratic voters to be more geographically concentrated can explain all the advantages that Republicans have enjoyed. Although the district court allowed for the possibility that the gap might be explained away by the greater concentration of the voters of one party, it rejected that explanation for the Wisconsin maps because it was possible to draw a set of maps with a lower efficiency gap. The plaintiffs also made the argument (although the lower court did not rely on it) that the Wisconsin maps are an “outlier” because computer simulations show that the existing gap in Wisconsin is among the most favorable to Republicans that can be drawn within the boundaries of traditional redistricting principles. These simulations have impressively “scientific” names — e.g., “Markov chain Monte Carlo simulations” — but they cannot be programmed to replicate real-world redistricting. Under court-drawn maps that existed between 1998 and 2010, the “seven point” threshold was exceeded in five of seven elections in Wisconsin.
The efficiency gap (and the alternative measures proposed by redistricting critics) does not reveal how much concentration of voters is “natural.” The measure nonetheless establishes a presumption that the partisan composition of the legislature should track (to some unspecified degree) the statewide vote totals for candidates of each political party. But to presume that deviation from proportionality is constitutionally problematic is to judge the elections we do hold (multiple single-member elections held in geographic districts where voters select an individual) by the degree to which the results resemble those that would follow elections that we do not hold (statewide elections in which voters select a preferred party). To make the latter the measure of the former is to change the nature of legislative elections and place pressure on both legislatures and lower courts to ensure partisan proportionality.
Nor is this presumption about proportionality neutral. To be sure, Wisconsin Republicans wanted to draw the most favorable maps they could. As the Supreme Court has noted, that intention will almost always be present. But if the judiciary countered partisan gamesmanship by mandating that votes be equally efficient and the results proportional, it would be stepping in and helping Wisconsin Democrats solve the natural problem that they face. Ironically, in the interest of rooting out partisan bias, the Gill plaintiffs seek to enlist the courts in their partisan project.
The efficiency gap is nothing new and does not move our redistricting jurisprudence to a law-based, judicially manageable standard. Its failure suggests that the Vieth plurality had it right in 2004. Sometimes the absence of an answer means that we have asked the wrong question.
– Mr. Esenberg is the president and general counsel of the Wisconsin Institute for Law & Liberty, which filed an amicus brief in support of the State of Wisconsin in Gill v. Whitford.