Welfare Analysis of Legislative Redistricting

点击翻译按钮获取中文摘要

Recent Supreme Court cases involving both racial and partisan gerrymandering have intensified a national debate over legislative redistricting. There is little agreement, however, in this debate over how alternative redistricting plans should be evaluated from the perspective of promoting the interests of voters. At its most fundamental level, the relevant question is: how should legislatures represent different groups of voters, and how should voters be allocated across districts in order to implement this prescription? This project develops an approach based upon traditional welfare economics in order to address this question and related questions from both a theoretical and empirical perspective. As a normative benchmark, the principal investigators begin by examining how ideological groups should be represented in legislatures. The key findings are that the seats-votes curve, which relates the fraction of seats held by the leftist party to its national support, should be linear, have a positive slope, and be biased in favor of the dominant party. They then examine how voters should be allocated across districts in order to implement this optimal seats-votes curve. The key findings here are that, if the optimal seats-votes curve can be implemented, then districts should be heterogeneous in their allocation of voter ideologies and that some districts should be "safe seats". Finally, the project will develop conditions under which the optimal seats-votes curve can be implemented and then tackle the difficult question of how voters should be allocated across districts when these conditions are not satisfied. The research team will also develop an integrated theoretical and empirical approach to evaluating legislative redistricting from this welfare perspective. It first develops a methodology for estimating seats-votes curves; this methodology requires only data on district-by-district support for the leftist party in legislative elections. The researchers then demonstrate that voter welfare can be directly measured for this observed seats-votes curve as well as for any other possible seats-votes curve. Thus, observed voter welfare can be compared with voter welfare under the optimal seats-votes curve, permitting measurement of any welfare losses from sub-optimal redistricting. In addition, observed voter welfare can be compared with voter welfare under proportional representation, an alternative voting system commonly used in other countries. The methodology also permits a comparison of voter welfare under alternative redistricting regimes, including bipartisan, partisan, and independent commissions. Finally, these ideas are illustrated in an application to voting in U.S. state legislative elections. The third project aims to generalize the ideas developed in the first two parts to a framework in which both voters and candidates are defined by their ideology as well as by their race. This generalized framework will be used address several normative questions related to race in redistricting. For example, how should black voters be represented in legislatures and allocated across districts? While black majority districts tend to elect black candidates, they also concentrate liberal voters, potentially leading to the defeat of leftist candidates in other districts. Thus, we plan to examine this and other potentially interesting interactions between race and ideology.