THE CAUSES AND CONSEQUENCES OF GERRYMANDERING.

• Author: Stephanopoulos, Nicholas O.
• Position: Special Issue on Gerrymandering

TABLE OF CONTENTS INTRODUCTION 2117 I. HYPOTHESES 2120 II. DATA AND METHODS 2128 III. RESULTS 2131 A. Main Analysis 2131 B. Nearest-Neighbor Matching 2134 C. Interactive Analysis 2136 D. Representational Analysis 2140 IV. IMPLICATIONS 2143 CONCLUSION 2148 APPENDIX 2150 INTRODUCTION

In recent years, some of the most important questions about partisan gerrymandering have been answered. How should we measure the extent to which a district plan benefits (or disadvantages) a party? Scholars have introduced metrics, like the efficiency gap (1) and partisan bias, (2) that are easy to calculate and so intuitive that courts have begun relying on them. (3) What does the distribution of plans' partisan fairness look like? Based on historical data spanning several decades, the distribution is centered on zero (or no edge for either party) and normal in shape. (4) And how have plans' partisan skews changed over time? In earlier periods, maps tended to assist Democrats, while over the last couple decades, they have tilted ever further in a Republican direction. (5)

Despite this progress, there is still much that we do not know, especially about the causes and consequences of partisan gerrymandering. By causes I mean all of the factors that may affect a district plan's partisan fairness. One set of these factors relates to the institution responsible for redistricting. We might hypothesize (in the absence of reliable evidence) that when a party has full control of the line-drawing process, the resulting map is usually skewed in its favor. Conversely, we might expect that when control of the state government is divided--or when a commission or court crafts the boundaries--the ensuing plan is comparatively neutral.

A second set of factors involves minority representation. A common argument is that Republicans profit as more districts are drawn in which minority voters are able to elect their preferred candidates. The logic is that these districts tend to elect Democratic candidates by overwhelming margins. The districts therefore waste large numbers of Democratic votes, enabling Republican candidates to win more of a plan's remaining seats.

A final concept that is often linked to partisan fairness is political geography. Here the typical claim is that Democratic voters are heavily concentrated in urban areas while Republican voters are more efficiently distributed in suburban, exurban, and rural regions. Accordingly, Democrats are "naturally" packed in a small number of districts, allowing Republicans to win more seats (by slimmer margins) thanks to their superior spatial allocation.

Turning to the consequences of partisan gerrymandering, the most salient is how legislators represent their constituents. The voting records of Democratic and Republican legislators, of course, are almost always different. Most Democrats take more liberal positions while most Republicans adopt more conservative stances. (6) A reasonable hypothesis, then, is that as a district plan skews further in a Democratic (Republican) direction, the ideological midpoint of the legislature becomes more liberal (conservative)--even keeping constant the preferences of the electorate. Electing more of a party's members for the same share of the statewide vote may be expected to yield ideological dividends.

These causes and consequences, it is worth emphasizing, are of more than academic interest. If contemporary maps tend to benefit Republicans, for instance, but this edge is due to compliance with the Voting Rights Act (VRA) or the country's political geography, then there may be little that can or should be done about the imbalance. On the other hand, if Republicans owe much of their advantage to control of the mapmaking process, then the case for intervention, judicial or political, becomes stronger. Similarly, we might not be too concerned about partisan gerrymandering if its damage is limited to bloodless concepts like seat and vote shares. But if gerrymandering distorts legislative representation--the beating heart of a democracy--then there may be more cause for alarm.

In this Article, I tackle these issues using a unique dataset of state house and congressional plans' efficiency gaps from 1972 to 2016. (The efficiency gap is a measure of partisan fairness that captures in a single number how much more "cracked" and "packed" one party's supporters are than the other party's backers.) (7) I pair this dataset with information on the institution responsible for designing each plan, Black and Latino representation, the level of urbanization, and the ideologies of members of Congress. I also employ rigorous techniques, like fixed effects regression and nearestneighbor matching, to come as close as is possible to establishing causation in a nonexperimental setting.

I find, first, that unified control of the redistricting process produces a large and statistically significant shift in the efficiency gap in the direction of the party in charge. This result holds at both the state house and the congressional level. The benefit of unified control has also increased in the last two decades. However, other redistricting institutions have small and inconsistent effects on the efficiency gap. That is, neither party consistently gains from plans designed by courts, commissions, or divided governments.

Second, I show that greater Black representation moves the efficiency gap in a Republican direction while greater Latino representation does not. This result also applies to both state house and congressional plans. The pro-Republican shift in the efficiency gap due to greater Black representation, though, is substantively quite small. And the shift does not occur at all when Democrats are responsible for redistricting. The relationship between Black representation and partisan fairness is therefore contingent, not compulsory.

Third, there is a link between urbanization and the efficiency gap in state house plans but not in congressional plans. At the state house level, Democrats tend to perform somewhat better in less urbanized states, while Republicans usually enjoy a modest advantage in more urbanized states. Interestingly, this link persists no matter which party is in charge of redistricting. It thus seems to be an intrinsic feature of contemporary state house maps (albeit one whose magnitude should not be overstated).

Lastly, the efficiency gap has a statistically and substantively significant impact on the ideological midpoint of a congressional delegation. In fact, a large pro-Democratic (pro-Republican) efficiency gap, on the order of 10 percentage points, results in a liberal (conservative) shift in a delegation's ideological median of about half a standard deviation. Moving from a gerrymander favoring one party to a gerrymander aiding its adversary swings a delegation's ideological median by almost a full standard deviation--all without changing the mind of a single voter.

These findings support two main conclusions. The first is that partisan intent is the most potent driver of district plans' partisan fairness (or lack thereof)- The deliberate manipulation of district lines by a party in unified control of the state government consistently affects the efficiency gap more than any other factor. Compared to full control of the redistricting process, other institutional arrangements, minority representation, and political geography pale in their influence.

The second conclusion is that partisan gerrymandering dramatically distorts congressional representation. Pro-Democratic gerrymanders make House delegations substantially more liberal than their states' electorates. Pro-Republican gerrymanders have an even larger effect in the opposite direction. The harm of gerrymandering is thus more than seat shares that are out of whack with vote shares. It is the ideological skewing of representation--and, with it, the policies that shape people's lives.

I begin this Article by surveying the limited existing literature on the causes and consequences of partisan gerrymandering, using it to generate a series of testable hypotheses. Next, I describe the data and methods that I employ in my analysis. In the core of the Article, I then explore how redistricting institutions, minority representation, and political geography are related to partisan fairness; and how partisan fairness is related to congressional representation. Lastly, I comment briefly on the implications of my results for the legal and political debates over gerrymandering.

1. HYPOTHESES

   As noted at the outset, political scientists have developed several quantitative measures of district plans' partisan fairness, the best known of which are partisan bias and the efficiency gap. The older of these metrics, partisan bias, is the difference between a party's seat share and 50% in a hypothetical, perfectly tied election.

   Suppose, for instance, that a party receives 60% of the vote and 70% of the seats in an election. Suppose also that if the election had ended in a tie, the party would have received 55% of the seats. Then the plan's partisan bias in that election is 5%: 55% minus 50%. (8)

   The newer metric, the efficiency gap, is rooted in the insight that partisan gerrymandering is always carried out in one of two ways: the cracking of a party's supporters among many districts, in which their preferred candidates lose by relatively narrow margins; or the packing of a party's backers in a few districts, in which their preferred candidates win by overwhelming margins. Both cracking and packing produce what are known as wasted votes because they do not contribute to a candidate's victory. In the case of cracking, all votes cast for the losing candidate are wasted; in the case of packing, all votes cast for the winning candidate, above the 50% (plus 1) threshold needed for victory, are wasted. The efficiency gap is simply one party's total wasted votes in an election, minus the other party's total wasted...