One of the most difficult choices for democratic societies is whether representatives are to be elected by voters or appointed by other officials. This project asks how the choice affects representation, testing two hypotheses from existing political science research. First, do direct elections lead to more representation of the majority's preferences, relative to appointment? Second, do direct elections cause certain groups - special interests" - to receive better representation than others? I address these questions by leveraging the substantial variation in local institutions, assembling original data sets on three local offices in three states: property tax assessors; school board members; and fire protection district trustees. Upon completion of the project, these unique data sets of political institutions and policy outcomes will be disseminated to the broader scholarly community, constituting a significant public good.
The intellectual merit of this project is related to its contribution to the study of institutions and local governance. The project focuses on these three cases for two key reasons. First, each case includes a significant number of political jurisdictions (towns, school districts, or fire districts) that change from direct elections to appointment (or vice versa) over time. If the analysis were to simply compare jurisdictions that elect or appoint at a single point in time - the approach of most existing studies of this question - it would be difficult to assess whether any difference in policy was due to direct elections, or due to pre-existing differences that are associated with the decision to elect. In contrast, the focus on jurisdictions that change permits comparison of policy outcomes between jurisdictions that elect or appoint, while controlling for any differences that existed prior to the switch. Second, each of the offices under examination is responsible for making decisions that have tangible impacts on citizens' lives. Thus, by examining how direct elections shape property tax assessments, standardized test scores, and fire department response times, the research will facilitate an evaluation of the link between citizens and government more precisely than other studies of representation that analyze indirect measures of policy outcomes.
The broader impacts of the project are associated with its contributions to the study of political institutions and local government. More practically, the findings from this project will help to inform a public policy question - how officials should be selected - with which countless organizations, from state and local governments to private boards, actively wrestle with. Through stronger research designs and original data gathering, this project seeks to offer more credible policy prescriptions than do the handful of conflicting findings from existing studies.
Does it matter whether government officials are directly elected by voters, or appointed by other elected officials? The existing scholarly literature, as well as members of the public, have long assumed that institutions such as direct elections are important for policy. However, existing studies of this question have been severely limited by a lack of good quantitative evidence. The main obstacle is that institutions are not randomly assigned, which means that existing comparisons may be confounded by numerous omitted variables. By approaching this question in three quasi-experimental settings, this project provides the strongest evidence to date on the question of whether direct elections matter for policy. To accomplish this, this project uses an estimation strategy that leverages variation in institutions within units but over time. The estimation strategy is meant to approximate a randomized experiment as much as is feasible in this setting. Units that switch their institutions are designated as "treated", and those that do not are designated "control." The estimation is then a series of simple comparisons. First, we calculate how the treated group's outcomes change before and after the treatment. Second, we calculate how the control group's outcomes changed over the same period. We then take the difference between these two differences, which represents an estimate of the causal effect of institutions. This estimation strategy was employed across three diverse cases to learn as much as possible about the effects of elections versus appointments. The first case examines elected and appointed property tax assessors in 920 New York towns between 1987 and 2012. The tax assessor is responsible for keeping the tax base accurate and up to date, which has implications for inequality. If tax rolls are out of date, some homeowners will pay much more (or less) than other residents with similarly valued homes. Results show a large effect of directly electing tax assessors on the frequency of updates. When assessors are elected, updates are much less common, which increases inequalities in taxation. The second case examines elected and appointed school board members in 132 Virginia school districts between 1987 and 2012. Until 1992, all school boards in Virginia were appointed, which creates a unique opportunity for studying the effects of elected school boards. Although existing literature in political science claims that elected school boards are a fundamental cause of educational policy outcomes, the results show no effect of elections on school district budgets, teacher salaries, or class sizes. The third case examines elected and appointed fire district trustees in over 800 fire protection districts in Illinois between 1994 and 2009. As in the case of school boards, results show that electing or appointing trustees has no impact on fire department budgets. However, another democratic institution does matter. Along with direct elections, some districts are subject to a "direct democracy" law that mandates any tax increases be approved by a majority of voters. The results show that this institution causes a significant decrease in property tax revenue, which in turn increases fire department response times. Overall, these results show that direct elections can make a large impact on policy. However, their potential for influencing policy is limited to cases where policy is highly salient and accessible to voters, such as property taxation. When officials make decisions regarding less salient issues, such as writing budgets, the form of selection is much less important.
