1

A THEORY OF CONSTRAINED REDISTRICTING

The central theoretical foundation of my argument and the subsequent analysis is that congressional redistricting operates under a significant number of constraints. Those responsible for redrawing electoral boundaries must accede to an array of constitutional, institutional, statutory, geographic, and political demands that limit their ability to pursue partisan goals. Because of these constraints, it is very difficult to implement a redistricting plan that institutes severe and long-lasting electoral bias. The nature of the efficient gerrymander, whereby voters of the opposing party are packed into a few safe districts with large majorities of wasted votes, while others are cracked into more competitive adjacent districts that the gerrymandering party hopes to capture by smaller margins, renders it inherently unstable and susceptible to adverse national trends in subsequent elections.
The reason for this is that gerrymandering can only produce distorted electoral outcomes if one party’s support is more efficiently distributed across the electoral landscape than the other’s. As safe seats only serve to reduce overall vote efficiency by creating large numbers of wasted votes, a highly efficient vote distribution necessarily requires a party to win seats by smaller majorities. The more dramatically districts are redrawn in order to benefit a particular party, the greater the erosion of the strong incumbency advantage that exists in congressional elections. The implication of this is that a focus in the redistricting process on partisan goals, rather than incumbency, should have a net positive effect on the competitiveness of U.S. House elections.
These constraints on the redistricting process fall into one of several categories: political constraints, which concern the nature of the redistricting process as it is conducted in the United States and the inherent uncertainty about its effects; legal constraints, which concern the various constitutional and statutory provisions that redistricters must follow when redrawing the district boundaries; structural constraints, which concern the historical and institutional frameworks that redistricting must operate within; and geographic constraints, which concern the underlying populations and voter alignments that redistricting seeks to allocate among the various districts in a given political jurisdiction.

Political Constraints on Redistricting

As with all decisions made by the political branches of government, redistricting must operate within the rules and procedures of the existing democratic process. In the vast majority of U.S. states, the redrawing of congressional district boundaries after each decennial census is the responsibility of the people’s representatives in their state government.

Control of State Government

With limited exceptions, a necessary condition for the implementation of a partisan gerrymander is for one political party to control each of the political branches of state government (both legislative chambers and the governorship) at the time of reapportionment following the decennial census. There are two situations where this may not be the case. First, most states also allow a gubernatorial veto of a redistricting plan to be overridden by a two-thirds majority in both legislative chambers, and so a party with a supermajority in both houses of the state legislature may implement a partisan redistricting scheme even without controlling the governorship. This occurred in 2002, both in Massachusetts, where a Democratic plan was passed over the veto of Republican governor Swift, and in Arkansas, where a Democratic plan was passed without the signature of Republican governor Huckabee, who lacked the votes to sustain a veto (Congressional Quarterly 2008). Second, several states have laws that limit the role of the governor in the redistricting process, thus allowing a party that controls the legislature but lacks a supermajority the opportunity to implement a gerrymander. These are Tennessee, which allows a gubernatorial veto of a redistricting plan to be overridden by a simple majority in both houses, and Connecticut and North Carolina, which do not allow a gubernatorial veto of a congressional redistricting plan. Several other states allow vetoes of congressional plans, but not those for the state house or state senate (Voting and Democracy Research Center 2004). For the two most recent redistricting cycles, partisan control of the redistricting process was dictated by the results of the 1990 and 2000 state legislative elections, and the corresponding gubernatorial elections for terms expiring no earlier than 1991 and 2001.
There are currently eight states that conduct congressional redistricting using an independent nonpartisan commission, most recently California, which established a Citizens Redistricting Commission following the voters’ approval of Proposition 11 in 2008. In addition, six states, consisting of Connecticut, Illinois, Indiana, Mississippi, Oklahoma, and Texas, have constitutional or statutory provisions that require redistricting to be referred to an independent commission if the legislature fails to pass a reapportionment plan by a certain deadline (National Conference of State Legislatures 2008). In the 2000 redistricting cycle, such provisions were triggered by deadlocked legislatures in Connecticut and Indiana (Karch, McConnaughy, and Theriault 2007). Partisan manipulation of the congressional district boundaries in the remaining forty-two states, except in the case of the exceptions noted above, requires unified partisan control of each of the branches of state government. Redistricting plans must be passed by both houses of the state legislature and signed into law by the governor, allowing a minority party in control of just one of these branches to thwart the majority’s wishes and force them to compromise on redistricting. Even with unified state government, a combination of institutional rules, delaying tactics, and legislative logrolling may allow a minority party to exert some influence over the redistricting process, perhaps limiting the ability of the majority to implement an effective unilateral partisan gerrymander. This might not be a significant obstacle if unified state government were the norm in the United States; however, this has not been the case in recent decades.
In the wake of the 2000 election, when the vast majority of congressional redistricting schemes were being drawn up, just twenty-three states had unified partisan control of all three branches of state government. In only thirteen of these, however, was there the potential for partisan gerrymandering of congressional districts. North Dakota, South Dakota, Montana, and Wyoming each had unified partisan state government, but only a single at-large congressional district. Arizona, Hawaii, Idaho, New Jersey, and Washington also had unified state government, but conducted redistricting through independent commissions. Congressional districts in Mississippi, which had unified Democratic control, were redrawn by the courts. Nebraska was the only state with a unicameral nonpartisan legislature, and so cannot be included in the total of states with unified partisan control. For the purposes of this analysis, redistricting in Nebraska was coded as bipartisan. Vermont, in addition to being a state with an at-large congressional district, and Maine had independent governors at the time of redistricting, and so are also coded as bipartisan. In the 1990s, just twenty states had unified partisan control, and in only sixteen of these was there the potential for partisan gerrymandering. The states excluded were South Dakota, an at-large state; Hawaii, which uses an independent redistricting commission; and Florida and Mississippi, whose congressional districts were redrawn by the courts.
The norm in the most recent redistricting cycles has been divided state government, which requires the parties to compromise on the redrawing of electoral boundaries. Partisan gerrymandering is generally not a widespread phenomenon in the United States, and may become even less frequent as state elections become more competitive and divided state government becomes more frequent (Jewell and Morehouse 2001), although the results of the 2010 and 2012 elections indicate that this pattern may be reversing itself.
In states without partisan control of state government in the legislative session immediately following the decennial census, redistricting must involve some degree of compromise between competing political interests. Political parties in some states have attempted to circumvent this constraint by conducting so-called mid-decade redistricting, whereby a party that takes control of state government in an election subsequent to redistricting uses this opportunity to conduct partisan redistricting through an additional redrawing of the electoral boundaries. Though the possibility has been discussed by politicians in a number of states in recent years (Cook 2005), mid-decade partisan redistricting has occurred on only two occasions since the 2000 census: first, in 2003, where it was instigated by Republicans in Texas after they gained control of the state House of Representatives; and again in 2005, at the behest of Republicans in Georgia after they were able to capture control of the governorship and state legislature. A similar effort was undertaken by Republicans in Colorado after they took control of the state senate in 2002, but it was struck down by the Colorado Supreme Court as a violation of Colorado’s state constitution (Salazar v. Davidson, 79 P.3d 1221 (Colo. 2003)).

Incumbency

Gelman and King (1994, 541–42) and Campbell (1996, 126–27) both argue that the redrawing of electoral boundaries is also constrained by the fundamental tension between the competing interests of partisan advantage and incumbent protection. As participants in redistricting, incumbent state legislators have an interest in their own future electoral fortunes, as well as those of their party. Incumbent members of Congress, although not directly responsible for redrawing their own district boundaries, are nonetheless involved in the redistricting process through lobbying or other informal influence. For example, Republican House majority leader Tom DeLay was an instrumental player in Texas’s 2003 redistricting, the purpose of which was by his own admission to increase Republican representation in the House of Representatives. Any redistricting effort, whether partisan or bipartisan, must at least be cognizant of, and in some cases may even be dictated by, the need for incumbent protection arising from mutual self-interest. In bipartisan redistricting, where control of state government is split between the major political parties, there is little potential for either party to gain an electoral advantage through gerrymandering. Bipartisan redistricting therefore allows incumbents greater influence over the drawing of district boundaries, as both parties have incentives to compromise in order to preserve their existing electoral strength. With both parties as veto players in the redistricting process, failure to cooperate most likely leads to redistricting by the judiciary or another independent body, with unpredictable results that both sides have incentive to avoid. Such redistricting tends to result in compromises that benefit incumbents of both parties, protecting their seats and preserving to a significant degree the existing electoral alignment. A good example of this is the state of New York, where divided partisan control of the two houses of the state legislature has been the norm for the last several redistricting cycles, with the Republican Party controlling the state senate from 1965 to 2009, and the Democratic Party having been in the majority in the state assembly since 1975. Successive redistricting has served to preserve and often reinforce this alignment, which was not broken until the Democrats gained a narrow majority in the state senate following the 2008 election (one they quickly lost again in 2010).
When one party controls the redistricting process and intends to implement a partisan gerrymander, the interests of incumbent protection and partisan advantage are in direct competition with one other. A party may gerrymander conservatively and focus on packing its opponents into safe districts while simultaneously protecting its own incumbent candidates, but the efficient gerrymander necessarily involves the creation of competitive seats that the gerrymandering party’s incumbents must defend (Friedman and Holden 2008). The primary interest of partisan redistricting is to target incumbents of the opposing party, either by redrawing them into districts where they must face off against another incumbent, a practice known as pairing, or by redrawing them into marginal districts that the gerrymandering party hopes to capture. It is therefore likely that the net effect of partisan gerrymandering, as compared to bipartisan collusion, will be to create increased incumbent vulnerability, competitiveness, and turnover of legislative seats.
Any attempt to manipulate the House electoral boundaries to the advantage of one party over the other also necessarily involves a party giving up votes in some districts in order to increase its overall number of legislative seats. Even allowing for retirements and changes in apportionment, some incumbent representatives of the party in control of redistricting may have to sacrifice a portion of their own electoral strength and safety in order to further the collective interests of the party. When a party in control of redistricting seeks to target an opposing incumbent by removing blocs of supporters from his or her district, geographic constraints may dictate that those voters be redrawn into the districts of incumbents of the gerrymandering party, therefore diluting those incumbents’ electoral strength. The high degree of uncertainty about future outcomes thus provides incentives for redistricters to be conservative when redrawing the electoral boundaries, in order to protect the electoral interests of incumbent candidates, and the effectiveness of partisan gerrymandering is constrained by this tension between the competing goals of incumbent protection and partisan advantage. Where redistricting does pursue partisan goals, as opposed to incumbent protection, the expectation is that this will have a positive effect on overall electoral competition.

Self-Constraints, Uncertainty, and Risk Aversion

Analyses of the effects of partisan redistricting generally pay little attention to the actual practicalities of the redistricting process itself, and the logistics of redrawing district boundaries in such a way as to benefit one political party over another. Partisan gerrymandering is generally accomplished by two principal strategies that act in tandem with one other to maximize the effective votes of one party while minimizing those of another. The first of these strategies is “cracking,” which refers to the practice of breaking up the targeted party’s geographical bases of support into several different districts, thus diluting their votes and reducing their efficiency. These supporters are thereby unable to vote in sufficient concentrations to win the individual seats into which they are divided, even though they may represent a significant enough voting bloc overall to warrant representation (Butler and Cain 1992).
The second strategy is “packing,” which refers to the practice of combining the targeted party’s geographical bases of support into a few supermajority districts, thus wasting significant numbers of their votes in a small number of overwhelming victories and allowing the party in control of redistricting to capture neighboring seats. Partisan gerrymandering is generally accomplished using a combination of these tactics to pack some of the targeted party’s voters into districts where they constitute a large majority, while cracking the rest of their voters into districts where they are only slightly in the minority. The intended effect is for the targeted party to win a few districts by large majorities with many wasted votes, whereas the gerrymandering party wins a large number of districts by small majorities and a highly efficient vote distribution (Cox and Katz 2002).
The inherent dangers in this practice are immediately evident: the greater reward a party seeks to gain from partisan gerrymandering, the greater risk it must take in implementing it. The capacity for a gerrymander to distort election results depends on the ability of the party that controls redistricting to capture seats by marginal majorities and hold on to them in subsequent elections. With each additional election conducted under a set of gerrymandered boundaries, it becomes increasingly difficult to predict the behavior of voters, the migration patterns of registered partisans, and the entry of new voters into the electorate. The result is that the marginal districts a party was able to capture through redistricting may be lost as a result of relatively small national swings in the popular vote and coattail effects. Artificial majorities created by gerrymanders are often unstable and may be prone to swing back toward parity in subsequent electoral cycles, suggesting that while gerrymandering may be an effective tool for a party to increase its electoral representation in the short term, in the long term, its effects are uncertain.
The strategy of cracking and packing in an attempt to maximize overall vote efficiency in search of the “efficient gerrymander” is thus a dangerous strategy for redistricters to pursue due to the uncertainty about long-term effects. When a future election produces even a relatively minor popular vote swing against the party implementing a gerrymander, the seats they are able to capture by maximizing vote efficiency are extremely susceptible to reverting back to the opposition party’s control. For this very reason, some recent research has cast doubt on the utility of the efficient gerrymander as accomplished through a combination of cracking and packing. It suggests that focusing exclusively on packing, while yielding smaller short-term gains for the party...