Panel: Local Politics and Public Good Provision: Evidence on Crime and Education
(Impact of Politics on the Policy Process)

Friday, November 3, 2017: 10:15 AM-11:45 AM
Addams (Hyatt Regency Chicago)

*Names in bold indicate Presenter

Panel Organizers:  Ying Shi, Stanford University
Panel Chairs:  Jason A. Grissom, Vanderbilt University
Discussants:  Katharine O. Strunk, University of Southern California and Patrick Bayer, Duke University


Minority Representation in Local Government and Distributional Outcomes
Daniel B. Jones, University of South Carolina, Brian Beach, College of William and Mary, Randall Walsh, University of Pittsburgh and Tate Twinam, University of Washington - Bothell



School Boards and Student Segregation
John Singleton, University of Rochester and Hugh Macartney, Duke University



Teachers' Unions, School Board Politics, and District Performance
Ying Shi, Stanford University and John Singleton, University of Rochester


Many public goods, such as education and safety, are principally supplied at a local level. Significant literatures in economics, political science, and policy have examined the efficiency, equity, and political economy of local public good provision. Nevertheless, while locally elected officials, such as school board and city council members, administer such goods, there is limited evidence examining causal effects of local government composition on these outcomes.  This session explores the relationship between local representation and public goods, chiefly education and crime outcomes, using quasi-experimental variation to justify causal inference. The inquiries rely on novel datasets that combine newly gathered information about locally elected officials with election records and administrative data from North Carolina and California.

An enduring issue in criminal justice and public policy broadly is the presence of racial disparities in policing. It is often suggested that such disparities are the product of lack of minority representation in local government, though lack of available data and poor identification often stalk such claims. The first paper in our session examines this relationship using a unique California dataset that matches data on local policing with the ethnicity of city council representatives, which were identified using Mechanical Turk and supplemental data resources. Using quasi-random variation in minority representation on city councils from narrowly-decided elections to establish causality, the authors find that an additional ethnic minority on the city council leads to a reduction in non-white arrest rates.

As segregation approaches Brown-era levels in many school districts, racial disparities also come to the fore in how school boards, in drawing attendance zone boundaries, assign students to public schools. A second paper examines the causal effects of school boards on student segregation, focusing on the political composition of the school board. To do so, the authors combine information about North Carolina school board candidates, matched with the voter registration database, with district level racial and economic segregation outcomes obtained from administrative data. As a school board’s composition is potentially correlated with household sorting, the authors similarly implement a regression discontinuity design at the electoral contest level. Their results demonstrate that more Democrat school boards reduce racial segregation, which has important implications for human capital through peer channels, but that these effects also raise residential segregation and increase “white flight” in some cases.

Local public good supply may also be subject to rent-seeking by interest groups. For instance, prior work has presented evidence that teachers’ salaries and per pupil spending are higher in school districts under greater influence of teachers’ unions. The third paper examines these relationships with a focus on local school boards as an underlying mechanism. Drawing upon the significant “ballot order effects” literature, the authors develop a novel identification strategy to generate quasi-random variation in the share of board members aligned with union priorities. The authors examine the extent to which more union-influenced boards shift schooling inputs found in California administrative data, ranging from the hiring of teachers vs. non-teaching staff to salary schedules, planning periods, and the number of school days.