[In my final exams, I often ask students to write “bench memos” to Supreme Court justices in important pending cases. For a change, I thought it might be interesting to put the shoe on the other foot, so to speak, and to require myself to do what I routinely ask of my students.]
The two pending cases to be argued March 26 — Rucho v. Common Cause and Lamone v. Benisek, the first from North Carolina and the second from Maryland — attack congressional gerrymanders on multiple grounds: primarily, the First Amendment; and secondarily, Equal Protection and Article I.
The First Amendment and Equal Protection challenges suffer from what seems to be a conceptual difficulty. But the same problem does not afflict the Article I claim. There are additional reasons, moreover, why the correct understanding of Article I leads to a modest and appropriate role for judicial invalidation of extreme gerrymanders that function as obstacles to the electorate’s ability to remove incumbent representatives whose performance in office the electorate repudiates. The Court, therefore, should embrace Article I as a basis for judicial review of congressional gerrymanders pursuant to the exercise of its interpretative authority under Marbury v. Madison, while simultaneously eschewing the First Amendment and Equal Protection as grounds for this review.
The Problem with the First Amendment and Equal Protection Claims
Both the First Amendment and Equal Protection claims in these cases rest on the same proposition that gerrymandering wrongly discriminates against members of a political party because of their partisan affiliation. Given this commonality, as well as the fact that the plaintiffs themselves prioritize their First Amendment argument, this memo will also focus on the First Amendment. But this memo’s analysis of the First Amendment issue applies as well to plaintiffs’ Equal Protection claims. Both share the same weakness, which is the difficulty of condemning the choice of a legislative district’s boundaries as a form of unconstitutional discrimination based on the party identification of voters who reside in the district.
While some forms of government discrimination on the basis of partisanship violate the First Amendment, not every kind of governmental differentiation based on party affiliation is unconstitutional. The majority party in each chamber of Congress gives itself more seats on legislative committees and subcommittees than it gives to the opposition party. Similarly, administrative agencies, like the FCC, often have more members from one political party than another. And even though the First Amendment has been construed to prevent partisanship as a basis for employment decisions with respect to subordinate federal employees, Rutan v. Republican Party of Illinois, 497 U.S. 62 (1990), a president clearly can make party affiliation a factor in appointing cabinet-level and other policymaking officials.
The question is whether the First Amendment prevents making partisanship a factor in drawing legislative districts. Can the majority party favor itself, at least to some extent, when exercising the discretion of where to draw the district lines, just as it favors itself (at least to some extent) when determining legislative committee assignments, or hiring some policymaking staffers? Or instead is legislative districting the kind of decision, like determining who gets a license for a parade on a public street, or who gets hired as a state university professor, for which the First Amendment requires that partisanship play no role whatsoever?
To ask these questions, drawing this conceptual distinction, would seem to suggest legislative districting belongs on the side of the line that permits consideration of partisanship as part of the government’s discretion to organize its own operations.
Historically, it has long been thought that it is permissible, even if sometimes objectionable, for partisanship to play a role in the drawing of district lines. When Patrick Henry had to decide where to draw the lines for Virginia’s districts in the first congressional election, one way to draw the lines would favor one nascent political faction (to which Henry and James Monroe belonged), while another way to draw the lines would favor a different political faction (to which James Madison belonged). It might have been “pure politics” rather than “good government” to decide to draw the district line to favor the preferred faction, but this political decision was not in and of itself unconstitutional. And although the First Amendment did not exist until after Madison won this first election and drafted the Bill of Rights, it is unreasonable to think that the purposes of the free expression clauses of the First Amendment were even in part to render invalid the kind of political districting decision that Patrick Henry had made.
The plaintiffs in these cases do not agree among themselves on how the First Amendment applies to redistricting. The Common Cause plaintiffs in the North Carolina case take the position that any redistricting motivated by “invidious” partisanship violates the First Amendment regardless of what actual effect that partisan motive might have. That position, however, seems flatly inconsistent with the Court’s decision in LULAC v. Perry, 548 U.S. 399 (2006), a case these plaintiffs do not cite. That position also seems contrary to basic First Amendment principles (the government does not violate a citizen’s rights to free expression simply because government officials have bad thoughts about the citizen’s political views) as well as the Court’s election administration jurisprudence under Anderson-Burdick-Crawford balancing. While the degree of scrutiny under A-B-C balancing varies with degree of burden on voting rights, there presumably needs to be some measurable harm to voting rights in order to trigger even the most minimal level of judicial review. Improper intent, without any showing of actual harm, would not be enough.
To illustrate this point, suppose the government must decide where to locate a new polling place. There are two options, X and Y. The government has no evidence that either X or Y would be better or worse for either Democrats or Republicans. But on a hunch the government officials responsible for this decision think that X might be better for Democrats and for that reason choose X. This invidious intent, without some actual evidence of harm, would not be a sufficient basis under the First Amendment for the judiciary to enjoin the choice of X as the new polling location. All voters, whether Democrats or Republicans, who live in the relevant precinct would be equally entitled to vote at this new location. There would be no formal discrimination among voters on the basis of party affiliation. Without any proof that it actually would be more burdensome for Republicans than Democrats to go to X to cast their ballots, there would be no basis for invalidating the government’s choice of location even with the evidence of improper motive. In the absence of such evidence of discriminatory effect, Y might be just as bad or even worse. If voters did not actually suffer, there is no point undoing the government’s inappropriately motivated decision.
Under the test advocated by the Common Cause plaintiffs, however, a government’s redistricting decision would violate the First Amendment without any showing of actual harm (beyond the bare minimum necessary for standing) simply because the government officials had a hunch that drawing the district one way, rather than another, might have the effect of benefiting one party over another. The North Carolina map at issue in Rucho may be an especially egregious gerrymander, which in fact gives considerable electoral advantage to one party over another. But the test proposed by the Common Cause plaintiffs is hardly limited to such “outlier” maps. Just the opposite. It would invalidate any map upon a showing that any district line was drawn because of favoritism for one party. (See especially pages 44 & 56 of their brief.) That test would rule out, it would seem, any redistricting whatsoever undertaken by a state legislature — a result that might be desirable as a matter of policy, but one that hardly seems compelled as an interpretation of the First Amendment.
The League of Women Voters plaintiffs in the North Carolina case do not advocate such a lenient standard. They say that the discriminatory effect of a gerrymandered map must be “large” and “durable” (at page 55) to be unconstitutional. But they do not offer any principle rooted in the First Amendment for determining when a map has a “large” and “durable” effect of the wrong kind. They offer multiple metrics for showing that a map has an asymmetrical bias in favor of one party. But the First Amendment does not require that legislative districting satisfy a symmetry standard (see LULAC v. Perry); there are entirely valid reasons of geography why a redistricting map would by asymmetrical, and these plaintiffs do not explain the degree of asymmetry that would cause a map to be unconstitutional. After each decennial census, when it is time to draw new legislative districts, there is no federal constitutional obligation on the part of the mapmaker to draw maps that satisfy a partisan symmetry standard at the same time that the mapmaker considers other redistricting criteria.
In a footnote (at 60 n.17), these plaintiffs say that in future litigation an “asymmetry threshold” could be developed that is similar to the “population deviation threshold [of] ten percent” for adjudicating Reynolds v. Sims claims. But this footnote misses a key distinction between the two types of claims. Equality is the constitutional standard under Reynolds, and 10% is considered a tolerable deviation for purpose of practical implementation. By contrast, perfect symmetry (in other words, zero asymmetry in favor of a political party) is not the constitutional standard, and thus the search is not merely for some practical way to implement an identified constitutional standard, but instead the search is to determine the constitutional standard itself. That key component of the First Amendment claim is still missing.
The Benisek plaintiffs in the Maryland case offer yet a third theory for when a gerrymander violates the First Amendment. In their view (see especially pages 34–35 of their brief), the First Amendment constitutionalizes a prohibition against vote dilution on the basis of party affiliation that is equivalent to the statutory prohibition against vote dilution on the basis of race under the Voting Rights Act. But VRA vote dilution claims can be especially difficult to adjudicate; LULAC v. Perry, again, is a good example. If Congress passed a statute that made vote dilution on the basis of party affiliation unlawful, then the judiciary would be required to consider these claims, as it does under section 2 of the VRA. But the difficulty of these dilution claims is reason to pause before concluding that the Constitution itself imposes this standard (and has ever since the adoption of the First Amendment as part of the Bill of Rights).
One way to see why race-based vote dilution claims do not translate easily into similar party-based claims is to consider the issue of at-large elections. If a government employs at-large elections with the intent to dilute African-American voting power, that intentional race discrimination is directly actionable under the Fifteenth Amendment. Rogers v. Lodge, 458 U.S. 613 (1982). But suppose a government decides to use at-large elections, rather than district-based elections, based on a belief that at-large elections will favor one political party in its competition with an opposing political party. (At-large elections are impermissible for elections to the U.S. House of Representatives, but that is only because of a congressional statute. Plaintiffs’ First Amendment theory would apply to all state and local elections.) At-large elections provide all voters equal voting rights in terms of Reynolds v. Sims. But, as is well known, at-large elections can give a hugely disproportionate boost to the majority party in a locality. Imagine a city where recent elections indicate that Democrats have a 60–40 advantage over Republicans citywide. At-large elections for city council seats would make it likely that Democrats could win all the seats, whereas it might be possible with district-based elections for Republicans to have representation on the city council. If Republicans made this proposal, the Democrats in control might reject it solely because they view retaining at-large elections as advantageous to their party. While that kind of partisan motivation for at-large elections is hardly good government, it has never been deemed unconstitutional. It would be a revolution in First Amendment jurisprudence to hold that plaintiffs could invalidate at-large elections, replacing them with district-based elections, based solely on a showing that at-large elections were retained for reasons of partisan advantage.
There is a simple explanation why the three groups of plaintiffs offer three different versions, each conceptually flawed, of how the First Amendment applies to gerrymandering. The explanation is that the First Amendment is ill-suited to imposing constraints on legislative districting, given that all districts already must comply with the basic equality principles of Reynolds v. Sims. How the drawing of district lines unconstitutionally favors one political party in violation of a “neutral redistricting requirement” embedded in the First Amendment is a hard argument to make, as a matter of both history and First Amendment theory. (The strongest version of the First Amendment argument is presented, not by any of the plaintiffs, but instead in the amicus brief submitted by the ACLU. Even in this brief, however, there is a tension between the general “neutrality” idea it advances and the specific “entrenchment” standard it proposes: partisan favoritism in redistricting that falls short of entrenching a structural advantage for one party would contradict nonpartisan neutrality; yet if the First Amendment only bars gerrymandering that amounts to entrenchment, the brief does not explain how this limited anti-entrenchment doctrine is derived from the much broader neutrality idea or from the First Amendment itself.) But the First Amendment is not the only provision at issue, and the difficulties it suffers in this context do not apply to the separate Article I analysis.
Article I’s Aversion to Incumbency Protection
Unlike the First Amendment, Article I does not contain an anti-discrimination principle that is conceptually difficult to apply in the context of drawing legislative districts. Instead, Article I contains the principle that elections to the federal House of Representatives should be responsive to the changing political views of the electorate. This principle is reflected in the basic Article I requirement that all House members must face the voters every two years, whereas Senators have six-year terms.
This responsiveness requirement has implications for judicial review under Marbury v. Madison. If a state legislature decided that it did not like the frequency of biennial elections and wanted to insulate the state’s congressional representatives from the obligation to seek reelection every two years, the state legislature might enact a law saying that the state will not actually hold these elections “every second year” but only every fourth or sixth. Or even if the state undertakes the effort of conducting the polls biennially, it could decide not to use the results as a basis for ousting an incumbent who loses the poll, but instead treats the result as a Gallup-type public opinion survey to learn the mood of the electorate but without any actual electoral consequences for the incumbent. These hypothetical state laws would be blatantly unconstitutional under Article I, and it would be the obligation of the federal judiciary under Marbury to invalidate them. This point would be true whatever Congress had, or had not, said on the matter pursuant to its own supervisory power to regulate congressional elections — a point that makes clear, contrary to the assertion in Paul Clement’s brief for the Rucho Appellants, that state laws regulating congressional elections are not immune from judicial review (under the political question doctrine) simply because Congress has the power to override state laws on this topic.
It is possible to imagine other state laws that would violate this Article I responsiveness principle. Suppose, for example, that a state law required challengers to win two-thirds of the vote in order to unseat House incumbents. This supermajority requirement, although not as extreme as giving challengers zero chance to defeat incumbents, would still be inconsistent with the basic principle that the electorate is entitled to “throw the bums out” every two years, if that’s what the electorate wants. Given the essential nature of the federal House of Representatives created by Article I, as the primary locus of popular sovereignty in the national government, a state’s rules for electing Representatives should not have a built-in structural advantage for incumbents. (If a state legislature endeavored to undermine popular sovereignty in the election of the federal House of Representatives in ways other than the protection of incumbents — for example, by insisting that incumbents win by two-thirds of the vote in order to retain their seats — that kind of impediment to popular sovereignty would also violate Article I. See Cook v. Gralike, 531 U.S. 510 (2001). But to consider the constitutionality of the legislative districting in these cases under Article I, it suffices to focus on the way in which the improper protection of incumbents contravenes the responsiveness principle inherent in Article I.)
Extreme gerrymanders can function as incumbency-protection mechanisms antithetical to this basic responsiveness principle. Although not exactly the same mechanism as a two-thirds voting requirement, a gerrymandered district can operate as the functional equivalent of a supermajority voting rule. Consider how gerrymandering works in comparison to at-large elections. Imagine a state in which at-large elections to the federal House would be intensely competitive (if Congress permitted at-large House elections, as it used to do and constitutionally could do again), with the expectation that the statewide electorate would be divided close to 50–50 between the competing candidates. Now suppose that the state legislature replaced at-large elections with a gerrymandered district-based map so that each gerrymandered district was electorally uncompetitive, and the incumbent had a 65–35 advantage over the challenger. The resulting situation, which is possible for gerrymandering to accomplish even when statewide races would be 50–50 competitive, would be essentially equivalent to imposing a two-thirds supermajority voting requirement on challengers. Just as an explicit two-thirds voting requirement for House elections imposed by state law would violate Article I because of its inconsistency with the basic responsiveness principle, so too be should the gerrymandering of House districts by state law that operates essentially as the same supermajority voting requirement and thus is equally antithetical to the fundamental responsiveness principle.
It is possible to measure the degree to which a gerrymandered map insulates incumbents from electoral competition. The same technique of statistical simulation that plaintiffs use (see pages 12–16 of their brief) in an effort to measure partisan bias could be employed to measure what’s relevant for Article I: improper incumbency protection. Here’s how it would work: the computers would be programmed to generate thousands of random maps using geographically valid districting criteria (equal population, compactness, respect for municipal and other local boundaries, and so forth). For each map, it would be possible to compute an average competitiveness score for all districts; this would be similar to computing a competitiveness score for at-large elections, but instead of a single statewide competitiveness measure, there would be a series of district-specific competitiveness measures, and the average of this series would be calculated. (The Cook Political Report’s PVI is one commonly used measure of a congressional district’s competitiveness.) The average competitiveness scores of all these randomly generated maps would be plotted on a graph, and it presumably would look something like the traditional bell curve. The average competitiveness score of the actual map adopted by the state legislature — the one alleged to be a gerrymander — could be compared to this graph. If the actual map was a statistical outlier, in the extreme tail of the bell curve, then it would mean that the actual map adopted by state law insulated incumbents from electoral competition to an extreme degree and in a way not necessitated by geographically valid districting criteria. To be sure, a state always should be entitled to offer a defense of a map that is a statistical outlier, but if the state can offer no legitimate reason for its incumbency-insulating map, then the map should be invalidated under Article I as inconsistent with its basis responsiveness principle.
This Article I analysis does not suffer from the defect that afflicts the First Amendment inquiry. The Article I analysis identifies the relevant principle (responsiveness) and the problem that contravenes that principle (incumbency insulation). The analysis then employs a measure of the problem itself (degree of incumbency insulation), not any proxy for the problem. It calls for the invalidation of maps if, but only if, they exhibit the problem to a statistically significant degree. Gerrymanders do not violate Article I if they give a slight edge to incumbents, and this is true even if they are motivated by an improper incumbency-protection purpose. (For the same reason, at-large elections to the House, if ever permitted by Congress again, would not violate Article I even if a state adopted at-large elections in the hope that they would give an advantage to incumbents. The capacity to “vote the bums out” always exists in an at-large election, as long as there is no supermajority voting requirement. Only a gerrymander that gives an artificial edge to incumbents beyond what would exist in at-large elections — an artificial edge equivalent to a supermajority voting requirement — would be actionable under Article I.) In sum, a gerrymander must generate a statistically measurable structural advantage for incumbents, amounting to an extreme impediment to electoral competition between incumbents and challengers in comparison to what would occur in the absence of the gerrymandered map, in order to justify judicial invalidation of the map under Article I.
This Article I analysis has additional advantages absent from the First Amendment (or Equal Protection) inquiry. It applies only to congressional elections, not state or local ones. Therefore, with respect to a topic that has proved judicially difficult for decades, it is advantageous that this approach applies only to a relatively small subset of the topic — and the one most appropriate for the federal judiciary’s involvement. The gerrymandering of state and local elections can be handled by state courts under state constitutional law in our federal system. The gerrymandering of congressional elections, by contrast, requires the federal judiciary to remediate (assuming, as analyzed above, that Article I imposes a constraint on gerrymandering).
This Article I analysis also can acknowledge an appropriate role for Congress in policing the gerrymandering of congressional districts by state legislatures, if Congress chooses to exercise this constitutional authority. The fact that Congress has this power does not mean that the political question doctrine bars the exercise of judicial review under Marbury. As indicated above, the courts must block state laws violative of Article I even if Congress has not acted. But insofar as Congress does have legislative authority in this area, the federal judiciary can appropriately take a subsidiary role and defer to Congress if and when Congress chooses to develop its own rules and procedures for policing improper state-law gerrymanders. While not exactly the same as “dormant Commerce Clause” jurisprudence, judicial enforcement of Article I can recognize the primacy of Congress to regulate this political subject matter that minimizes the risk of judicial overreach. In this respect, Article I (like the Commerce Clause) is in sharp contrast to the First Amendment or Equal Protection: with respect to those constitutional clauses, if the Court issues a pronouncement it necessarily occupies the entire field, crowding out any contrary Act of Congress on the matter. By contrast, with respect to Article I, like the Commerce Clause, the judiciary can give Congress the last word, assuming that Congress has acted within the scope of its own authority. The judicial invalidation of a state law on Article I grounds, like the judicial invalidation of a state law on “dormant” Commerce Clause grounds, can yield to supervisory congressional legislation that appropriately enforces Article I. In this way, a judicial invalidation of a state-law gerrymander on Article I grounds is not nearly so momentous a ruling as would be a judicial decision resting on First Amendment or Equal Protection grounds. (It is possible also that extreme partisan gerrymanders might violate the “fair play” principle inherent in due process, but that kind of constraint against outlier maps would not entail strict partisan neutrality in redistricting, and in any event is an analytically distinct claim that the plaintiffs have not argued in these pending cases.)
Based on the foregoing analysis, this memo recommends vacating the decisions below insofar as they rest on the First Amendment or Equal Protection and remanding for consideration of the whether the challenged maps violate Article I’s responsiveness principle because they improperly insulate incumbents from electoral competition. On remand, the courts below can consider whether there is adequate evidence already in the record to conduct this Article I inquiry and, if not, whether it would be appropriate procedurally to reopen the record for this purpose.
This essay originally appeared on the Election Law @ Moritz website.