Drawing the Line: Redistricting and Competition

in Congressional Elections

Michael P. McDonald

George Mason University

What if an election were held, but the outcome was already known with a high degree of certainty? Would eligible voters come to the polls to affirm the inevitable? Would representatives be responsive to their constituents? Would we call this “democracy” or would we be talking about a Third-world dictatorship?

Regrettably, I am characterizing the overwhelming majority of congressional and state legislative elections in the United States. Heading into the 2004 general election, handicappers of the 2004 congressional races consider few to be competitive.¹

Congressional Quarterly rates 29 of the 435 House seats as a toss-up or leaning towards one of the two parties’ candidates. The Cook Political Report rates 36 races as toss-up or lean. Campaigns and Elections rates 56 races to have a 50-60% chance of electing a party’s candidate. In the remainder, knowing a district’s 2000 presidential vote and the incumbent’s party is sufficient to accurately predict the winner.² Handicapping and presidential vote totals within state legislative districts are unavailable nationally, but we can examine the 2002 results: only 4.6% of all incumbents lost in the general election in a year when an appreciable number of incumbents were pitted against one another due to redistricting.³ Real democracy would seem to be lacking from a large number of legislative elections in the United States.

Counter-intuitively, where legislative elections are uncompetitive, the national presidential electorate is evenly divided and highly competitive. The national competitiveness masks the ongoing sub-national realignment that is sorting out the populace along party lines.⁴ Rural states, primarily in the South and Midwest, are trending Republican. An intrastate urban-rural divide is deepening in states with large and growing cities that are trending Democratic.

The reduction of electoral competition in legislative elections is a consequence of the United States’ system of geographic representation though single-member districts. As the nation becomes more regionally divided, legislative districts located in polarizing regions naturally become less competitive. But that is not all that is happening; conscious choices are made during redistricting to reduce competition. District boundaries are being manipulated to divide voters into more politically homogenous districts.

Sorting voters into homogenous districts approximates proportional representation, which could be argued is a desirable feature of an electoral system. In 2002, Republicans won 50.5% of the vote in all congressional races and received 52.6% of the seats.⁵ It is well known that the party that receives the highest number of votes tends to receive a seat bonus in electoral systems that use single member districts, so at face value this outcome appears to be fair. Unfortunately for the Democrats, sorting voters into homogenous districts makes the number of seats awarded to a party unresponsive to changes in the vote. A small Republican bias in the electoral system combined with a low degree of responsiveness means that the Democrats need to do extraordinarily well to become the House majority. Given the distribution of voters and incumbents into districts, I estimate the Democrats need to win 57% of the congressional vote to pick up the eleven seats they need to win a majority of the seats. This unlikely, though not impossible, scenario means that the Republicans should enjoy majority status in the House for the remainder of the decade.

The national “fairness” of the 2002 distribution of congressional seats to votes masks severe disproportional results at the state level. In Massachusetts, Democrats won 64.8% of the 2000 presidential vote (a reasonable measure of underlying partisan strength), yet won all of the congressional seats. In Pennsylvania, Democrats received 52.1% of the 2000 presidential vote, yet won 36.8% of the seats. Little relief for partisan gerrymanders appears to be forthcoming from the courts. In a recent and closely followed partisan gerrymandering case before the Supreme Court, Democratic plaintiffs unsuccessfully argued that the Pennsylvania anti-majoritarian outcome is unconstitutional, with four Justices finding that partisan gerrymandering is non-justiciable for lack of a clear standard.⁶ Justice Kennedy’s concurring opinion held the door open to further attempts to fashion an acceptable judicial standard to identify partisan gerrymanders. For the time being, distortions of the seat to vote ratio appear to be legal, as long as they do not violate other traditional redistricting principles, such as equal population, or violate state constitutional requirements.

Despite the urban-rural realignment, it is possible to increase electoral competition during redistricting by drawing districts that tie together politically opposite areas of a state. The Illinois 17th congressional district strings across hundreds of miles of western Illinois, at one point cutting a one-block wide path through the center of the capitol Springfield, incredibly without picking up population by winding its way through shopping centers and golf courses. Many state constitutions forbid the creation of such non-compact districts. However, even compact districts can be made more competitive. In Arizona, adjacent Democratic and Republican Tucson communities could have been combined to create additional competitive state legislative districts, but were rejected by the state’s redistricting commission.⁷

As these examples suggest, conscious choices during redistricting are reducing electoral competition by drawing fewer competitive districts. Politics and structural constraints impede the creation of competitive districts. Reelection seeking incumbents want safe districts and optimal partisan gerrymandering results in uncompetitive districts. Structural constraints such as drawing specially mandated minority-majority districts and adherence to neutral criteria such as compactness and respect for communities of interest can further reduce opportunities to draw competitive districts.

That incumbents seek reelection is a cornerstone of modern inquiry into legislative behavior, and that incumbents would use redistricting to aid in their reelection efforts by infusing their district with more likeminded partisans is not surprising. Although congress members do not play a formal role in the redistricting process, they often play an informal role by proposing their own maps and lobbying those in charge of redistricting. Often incumbency protection maps are adopted as a consequence of a political process that forces bipartisan compromise. Illinois’s 2001 map is an example of such a bipartisan agreement, necessitated by a divided state legislature. Its snakelike lines not only created safe districts for all incumbents, but it also removed the homes of potential challengers in some districts. In all, incumbency protection maps were adopted in twenty states, which affected 231 districts, due to bipartisan plans adopted in larger states such as California and Texas.⁸

While it would be tempting to lay the blame of uncompetitive elections on bipartisan gerrymanders, the political parties have little interest in drawing competitive districts, too. The percentage of congressional districts with a two-party presidential vote in a competitive range is essentially indistinguishable between those states with partisan and incumbent protection gerrymanders.⁹ Burnham argued that parties were most likely to miscalculate when attempting to maximize the number of districts that they could win, and as a byproduct, create too many competitive districts.¹⁰ There are a few examples of this sub-optimal behavior. However, as Owen and Grofman formally show, a partisan gerrymander and an incumbent protection gerrymander both result in uncompetitive districts.¹¹

To informally understand the dynamics of an optimal partisan gerrymander, consider an evenly divided state with three districts. The gerrymandering party could make each district a mirror of the state, and create three districts with a fifty percent chance of electing their candidates, resulting in 1.5 expected seats. But the gerrymandering party can do better by wasting the other party’s votes in districts that they either win overwhelmingly, or in districts that they have no chance of winning; strategies called “stacking” and “cracking” in the redistricting literature. Ideally, they can completely stack the other party’s supporters into a single district, and evenly spread the remaining supporters across the other two districts. (When there are more than three districts, Bruce Cain refers to the ideal percentage of supporters within the gerrymandering party’s districts as the “efficient level of partisan strength.”¹²). True, these districts are ever so slightly more competitive than those drawn for out-of-luck party, but this level is not so weak as to make these two districts competitive. The result is that the gerrymandering party is expected to win two seats, increasing their expected payoff by 0.5 seats while simultaneously reducing their uncertainty, and none of the seats are competitive.

Incumbents and parties are constrained by federal and state districting rules. Among the most important federal rules is the Voting Rights Act (VRA), which calls for the drawing of special minority-majority districts in certain circumstances. These districts tend to be heavily Democratic (except for Cuban-American districts in Florida, which are heavily Republican), so they are not competitive and many incumbents of these districts run for reelection without serious general election opposition. While the districts themselves are not competitive, they may affect competitiveness of other districts. Placing Democrats into heavily Democratic minority-majority districts makes the remainder of a state more Republican. In a Democratic state, this can increase opportunities for competition in the remainder, while in Republican states this can reduce opportunities for competition. This was particularly true in Republican leaning Arizona, where the drawing of two Hispanic, and heavily Democratic, districts reduced the opportunities to draw competitive districts elsewhere.

            In addition to federal laws, redistricting may be constrained by state laws, constitutions, and ad hoc legislative rules. In the recent round of redistricting, nineteen states required districts to be “compact,” though compactness has numerous definitions. Where practicable, twenty-three states required districts to follow political boundaries, such as city or county lines, or to respect boundaries of “communities of interest” (which are usually defined by a redistricting authority).¹³ Insomuch that partisans are geographically segregated, these additional constraints may aid or hinder the creation of competitive districts.

           Public interest groups like Common Cause and the League of Women Voters advocate for redistricting reform, a cry that has been recently taken up in editorials by many political observers. The model of redistricting reform promoted by these reformers is based on the Arizona Independent Redistricting Commission (AIRC), a commission that I served as a consultant. Like many other states where redistricting commissions have been adopted, the AIRC was created as a consequence of the adoption of an initiative, Proposition 106, in the 2000 general election.¹⁴

Some may be surprised that Arizona is the model of reform, not Iowa. Iowa’s system is the darling of those who look at outcome, not process. It is true that four of Iowa’s five congressional districts were labeled as competitive in 2002, while only one of Arizona’s eight was deemed competitive. A close examination reveals that Iowa’s competitive elections were those where Republican incumbents were mismatched in Democratic leaning districts, in a region of the state where creating Democratic leaning districts was inevitable. Unfortunately, in Republican leaning Arizona, there was little room left to draw competitive congressional districts after two Hispanic-majority districts were drawn. Structural constraints, not process, determined the redistricting outcome in Arizona.

The Iowa process emulates bureaucratic boundary commissions used in virtually every other country that does national redistricting. Nonpartisan legislative support staff, known as the Legislative Service Bureau (LSB), draws and proposes maps to the state legislature. But the process is not free from politics; it exists under statute and the state legislature reserves the right to draw districts if they do not adopt by a majority vote one of three maps proposed in sequence by the LSB. A legislative oversight committee directs the LSB, and in 1981, Congressional Quarterly documented a process under a unified Republican government where the LSB was directed to secure the reelection of Republican incumbents to the detriment of a Democrat incumbent. In 2001, a desire to avoid a political battle over redistricting between a Republican legislature and a Democratic governor helped set the stage for adoption of the second map proposed by the LSB. A happenstance of political geography and political circumstance produced the 2002 map that many found virtuous, but there is no guarantee circumstances will repeat themselves in the future.

The AIRC attempts to permanently take the redistricting process out of the hands of the politicians and place it in the hands of a citizen commission through selection of membership and rules guiding the redistricting process. Arizona’s citizen commission is composed of five members selected from a list of twenty-five applicants vetted by the State Supreme Court. The list contains ten Democrats and ten Republicans and five minor party or independent registered voters. For three years prior to the establishment of the commission, applicants cannot have served in office (except school board), as an officer of a political party, or professionally lobbied the government. The majority and minority leaders of each legislative chamber select one member and the four commissioners, and these four members then select a chairman from the five minor party or independents. Members cannot run in the districts that they create.

The commissioners must draw districts to satisfy a rank ordering of redistricting criteria. The first three criteria echo federal requirements: districts must satisfy the VRA, have equal population, and be compact and contiguous. The remainder are requirements found in Proposition 106: districts must respect communities of interest; to the extent practicable must follow visible geographic features and political boundaries and census tracts; and to the extent practicable, competitive districts should be favored where to do so would create no significant detriment to the other goals. The last criterion is striking, and one other state, Washington, also requires competition to be considered by its commission.

The AIRC labored for nearly three years drawing congressional and state legislative districts. Where congressional redistricting went smoothly, the state legislative redistricting failed to satisfy the Department of Justice’s VRA concerns and was the target of multiple lawsuits, which at the time of this writing are still in appeal. The commission was required to redistrict state legislative districts three times. Several issues developed in the course of the commission’s existence that serve to guide future reform efforts. Terms like “compact,” “community of interest,” “competitive,” “extent practicable,” and “no significant detriment” are not defined in Proposition 106, and the commission had to exercise their discretion. The commission was not explicitly granted legislative authority, which opened their discretion making powers to legal review. The subsequent, and prolonged, court action created an uncertain environment for state legislative candidates, and placed a tremendous burden on the citizen commissioners who were not paid for their duty. In the extended legal battles, the commission eventually exhausted their proscribed budget, and needed further authorization, which ran afoul of partisan politics and threatened the “independent” integrity of the commission.

            The genius of the United States federal system of government is that other states have the opportunity to experiment with the Arizona model and improve upon it. A simple solution to the definitional issue would be to grant redistricting commission legislative authority, but that would limit the review of the commission’s work by a court. Related, better definitions would reduce ambiguity; however, once a definition is carved into a constitution, it would be difficult to amend if found problematic to implement.

Another approach would anticipate the political challenges disguised as legal action (since it is often only the political parties that have the resources to mount a legal challenge). The AIRC process allows for legislative leadership input only at the commissioner selection stage. Borrowing from Iowa and Maine’s processes, a legislative super-majority vote on the commission’s map might allow for partisan grievances to be aired and resolved without court action. If the commission’s work were not to be approved by the legislature, a prudent tweak to the process would be to expedite state Supreme Court review of the map, as is done in states like Florida, Idaho, and Kansas.

As Robert Dahl and other democratic theorists have noted, electoral competition is one of the necessary conditions for democracy. That it is lacking from an overwhelming majority of congressional and state legislative elections in the United States is troubling. Reformers are approaching the issue from many angles: reforming the campaign finance system, promoting open or nonpartisan primary systems, and offering other reforms aimed at changing the electoral system of plurality-win single-member districts. Redistricting reform is yet another approach that deserves serious consideration. In the 1990s, Alaska, Arizona, and Idaho adopted redistricting commissions by initiative. More states may follow in the coming decade given the high level of interest following the re-redistricting battles in Texas and Colorado. Scholarly attention needs to be paid to these redistricting institutions and their electoral and political consequences for academics to intelligently guide the reform process to promote democracy and avoid unintended consequences.

---

Notes

1. Counts are as of September 8, 2004 and will undoubtedly change as election events unfold.

            2. For example, see the Center for Voting and Democracy’s Monopoly Politics report and associated dataset online at: http://www.fairvote.org/library/geog/congress/index.html.

            3. See the National Conference of State Legislatures’ tabulation of 2002 election turnover: http://www.ncsl.org/programs/legman/Elect/02Turnover.htm, accessed Sept. 13, 2004.

            4. See: James G. Gimplel and Jason E. Schuknecht, Patchwork Nation: Sectionalism and Political Change in American Politics (Ann Arbor: University of Michigan Press, 2003). For a newspaper analysis of county level partisanship trends, see also: Bill Bishop, “Politics 2004: Preach to the Choir: As Communities Polarize Politically, Presidential Candidates Focus on Stirring up Local Partisans,” Austin American Statesman at http://statesman.com/specialreports/greativide/0502divide.html         

5. See http://www.fec.gov/elections.html.

            6. See Vieth v. Jubelirer 541 U.S. ____ (2004)

            7. I consulted on district competitiveness for the Arizona Independent Redistricting Commission. At the direction of the commission and under court order, I drew three competitive districts (using my definition) in the Tucson area. The commission rejected my recommendation in favor of a plan with two competitive districts, but also respected a commission-defined community of interest in the foothills north of the city

            8. A classification of adopted redistricting plans can be found in Michael McDonald, “2001-02 Comparative State Redistricting Institutions,” State Politics and Policy Quarterly, forthcoming 2004.

9. The number of districts with a two-party presidential vote (Bush/[Bush+Gore]) within a 45-55% range in states with an incumbent protection map is 53 of 231 total districts, or 22.9%; in states with a partisan map, 43 of 152, or 28.2%. Within a 48-52% range, there were 23 of 231 or 10.0% in states with an incumbent protection map and 12 of 152 or 7.9% in states with a partisan map. Partisan gerrymanders are more likely to occur where there is unified government, which would also be in states that aren’t competitive. However, a multivariate analysis shows there is no statistically significant difference between incumbent and partisan gerrymanders and their tendencies to produce uncompetitive districts.

10. Walter Dean Burnham, “Congressional Representation: Theory and Practice in Drawing the Districts,” in Nelson W. Polsby, ed., Reapportionment in the 1970s, (Berkeley, CA: University of California Press, 1971).

11. Guillermo Owen and Bernard N. Grofman, “Optimal Partisan Gerrymandering,” Political Geography Quarterly 7, no. 1 (1988): 5‑22.

12.Bruce E. Cain, The Reapportionment Puzzle (Berkeley, CA: University of California Press., 1984).

13. For a summary of state rules see Barbaras and Jerit, “Redistricting Rules and Race,” State Politics and Policy Quarterly, forthcoming 2004.

14.  See http://azredistricting.org.

 

---