Christopher R. Berry is a Professor at the University of Chicago Harris School of Public Policy and the College. In addition to directing the Mansueto Institute, he is the Director of the Center for Municipal Finance. Professor Berry’s research interests include American politics, metropolitan governance, municipal finance, and intergovernmental fiscal relations. He is the author of Imperfect Union: Representation and Taxation in Multilevel Governments, winner of the Best Book Award in Urban Politics from the American Political Science Association, Theory and Credibility (with Scott Ashworth and Ethan Bueno de Mesquita), and many other scholarly publications. Professor Berry’s research has been featured in the New York Times, the Washington Post, Bloomberg BusinessWeek, the Chicago Tribune, Last Week Tonight with John Oliver, and many other outlets. Professor Berry is a member of the American Academy of Arts and Sciences.
Summary
Promise of the Reform
"Turnout in US local government elections is startlingly low, often less than 20%, and voters are unrepresentative of the electorate overall. Most local elections are held off-cycle. Changing local election times to coincide with higher-profile federal and state elections would generate dramatic increases in turnout and a more representative set of voters."
Key Takeaways from the Research
Turnout in local government elections is extremely low.
Voters in local elections are disproportionately white, affluent, and elderly.
Special interest groups are disproportionately influential in local elections when turnout is low.
Syncing local elections to coincide with higher-level elections roughly doubles the turnout.
Voters in synced local elections are more representative of the electorate overall.
Candidates in synced elections are more likely to hold preferences that are aligned with their constituents.
Important Questions that the Research Does Not Answer
Are voters in synced elections less well-informed about the candidates and issues at stake?
Does having a larger number of races on a synced ballot generate “ballot fatigue,” causing voters to skip voting on lower-level races?
Does syncing elections alter the number or quality of candidates?
Introduction
"US local governments have a multitude of elections but remarkably few voters."
The vast majority of elected officials in the United States serve in local governments, where voter turnout is shockingly low. Most local elections are held off-cycle, on separate days from higher profile state and federal elections. Changing local elections to become concurrent with higher level elections can lead to dramatic increases in turnout almost immediately, while also substantially lowering the costs of election administration. Voters in concurrent elections are likely to be more representative of the electorate overall, but may be less well informed about the candidates and issues.
The Problem of Low Turnout
US local governments have a multitude of elections but remarkably few voters. Typical turnout in mayoral elections is 20%.
While federal and state races garner the most media attention, the vast majority of elections in the United States are for offices in local government. The nation’s roughly 90,000 local governments include cities, counties, townships, school districts, and a plethora of special purpose districts, such as park districts and library districts. Most of these jurisdictions are run by an elected board and many also elect executives and other officers. Collectively, they are governed by nearly half-a-million elected officials.
Due to the decentralized nature of election administration in the US, there are no comprehensive, official data sources of local election outcomes. However, several teams of scholars have independently assembled local election statistics of various kinds. The data are alarming. One group of researchers compiled data for mayoral elections in 46 major cities, including the 30 largest, between 2011 and 2015. The median turnout rate was 20 percent of voting-age citizens.1 Another study found comparable results in their earlier study of mayoral elections in 144 of the largest cities: average turnout of 26 percent for 340 elections held between 1996 and 2011.2 Comparably low figures have been found in other studies of mayoral turnout.3
Aside from mayor, election data for other local offices is even harder to come by, but available evidence suggests that turnout is, if anything, even lower. A study of school board elections in Michigan found an average turnout of 8% across the state’s roughly 500 districts.4 Turnout in many special district elections is likely to be even lower, although systematic data are unavailable. A comprehensive study of local government elections in Nassau County, New York, found that special district (e.g., water, sanitary, library, and fire districts) elections were often less than 5% – and this is in the same precincts that delivered 44% turnout in the gubernatorial election in the same year.5
In short, while complete data on local elections are unavailable, all available evidence suggests that turnout is exceedingly low, with the 20% turnout rate for mayoral elections perhaps being the high point of local participation.
Empirical Research on Local Election Timing
"While there are documented racial, economic, and age gaps in federal elections, the disparities in local elections are significantly larger."
Voters in Local Elections are Unrepresentative of the Electorate
The relatively few voters who do turn out for local elections are unrepresentative of the overall electorate. While biases in turnout are well known in federal elections, the differences between voters and nonvoters in local elections appear to be even greater.
Several studies show that voters in local elections are demographically unrepresentative of the electorate overall.6 Three key differences between voters and nonvoters emerge from the research.
First, voters are less racially diverse than the electorate overall. White voters are significantly more likely to turnout than are nonwhite voters. Latinos and Asian Americans are especially underrepresented among local voters. Secondly, voters are more affluent than nonvoters. Voters have notably higher incomes and higher education levels than nonvoters, while the gap between unemployed and full-time workers is especially large. Finally, voters are older than nonvoters. Voters over the age of 65 are the most overrepresented group in local elections. They cast ballots at 2-5 times the rate of voters aged 18 to 34, resulting in an average gap of 15 years between the median age of voters and the median age of the voting eligible population.7 While there are documented racial, economic, and age gaps in federal elections as well, the disparities in local elections are significantly larger.
Aside from demographics, other differences between voters and nonvoters in local elections are harder to measure, but there are reasons to suspect that they also differ in preferences over issues. First, to the extent that demographics are correlated with issue preferences, the demographic biases alone would yield a pool of voters that may not represent the preferences of the electorate overall. Second, for single-function local elections, in particular, “selective participation” may produce a voter pool that is disproportionately comprised of those who care most about the issue at stake, potentially allowing special interest groups to have more influence over the jurisdiction.8 Research has shown, for example, that teachers are roughly seven times more likely to vote in school district elections than are ordinary voters.9
"Off-cycle local elections experience abysmally low turnout, while on-cycle or 'synced' elections yield turnout levels comparable to higher profile state and federal elections."
Concurrent Elections Generate Massive Increases in Turnout
One of the chief reasons for low turnout in local elections is that most – roughly 4 in 5 – are held on separate days from major state or federal elections. These “local only” elections are typically held in the spring, but can happen almost any time. Elections for different local offices may even take place on separate days. In Nassau County, New York, for example, local elections take place on 24 different dates scattered throughout 11 months of the calendar year; none of them coinciding with major state or federal elections in November. The tradition of holding local elections non-concurrently with higher-level races can be traced to the Progressive Era. Progressives promoted off-cycle municipal elections ostensibly to give voters a chance to focus on local issues and allow elections to be decided by the most interested and knowledgeable voters, though evidence also suggests they saw such elections as likely to weaken political machines and advantage Progressive candidates.10
Whatever the intentions of the Progressives, the end result has been that off-cycle local elections experience abysmally low turnout, while on-cycle or “synced” elections yield turnout levels comparable to higher profile state and federal elections. Indeed, every study of the issue reaches the same conclusion: synced local elections have turnout rates that are dramatically higher than unsynced elections. Indeed, compared to off-cycle elections, local elections held concurrently with national elections have more than double the rate of voter turnout.11
Most of the evidence on election timing and voter turnout comes from comparisons of different jurisdictions with elections held either synced or unsynced with higher level races. While the observed differences in turnout are stark, we might be concerned that there are other factors that influence turnout in jurisdictions that hold elections at different times. What if jurisdictions that would otherwise have lower turnout for unrelated reasons also choose to hold elections off-cycle?
"No other institutional reform we know of–short of making voting compulsory–comes anywhere close to having this kind of impact on turnout."
One way to explore whether the difference is causal is to study jurisdictions that have changed their election timing from synced to unsynced, or vice versa. Finding that turnout changes within the same jurisdiction after the timing of elections changes should bolster confidence that the relationship is causal. In fact, research that examines such withinjurisdiction changes finds that, in the case of California municipalities and school districts, the changes in turnout that occur after a change in election timing are nearly as large as the differences between jurisdictions.12 Importantly, changing local elections to coincide with presidential elections generates the largest increase in turnout, while changing them to coincide with midterm or primary elections generates smaller, but still substantial, increases. Anecdotally, case studies of individual cities that have changed election timing show similarly substantial changes in turnout after the reform, such as was the case in Los Angeles, which experienced record high turnout after changing its mayoral elections to align with races for president and governor.13
The research is unanimous and persuasive: No other institutional reform we know of–short of making voting compulsory–comes anywhere close to having this kind of impact on turnout.
The shift to concurrent elections also leads to a more representative set of voters, according to studies in California.14 The age gap is almost completely erased when local elections are changed to coincide with presidential elections; racial participation gaps are significantly reduced. Differences in participation by income are the least impacted by changes in election timing. On-cycle elections also generate improvements in partisan representation. Democrats are generally underrepresented in offcycle elections, relative to their share of the local electorate, and on-cycle elections effectively restore partisan representativeness.15 Improvements in voter representativeness is greatest when local elections are switched to coincide with presidential elections, while those synced with midterm or primary elections see smaller improvements. Whether these California results extrapolate to other areas is a question calling for additional research.
Theoretical Research on Local Election Timing
"There is empirical evidence that voters are better represented by their elected officials when local elections are held concurrently with higher level races."
Tradeoffs in Switching to Concurrent Elections
If concurrent elections produce significantly higher turnout and voters that are more representative, are there any potential downsides? The most prominent argument in favor of off-cycle elections, and hence against synced elections, dates back to the Progressive Era.16 It is that holding municipal elections on separate, local-only days, allows voters to focus attention on local issues without the distraction of simultaneous state or federal races. Even if fewer voters turn out, the argument goes, those who do turn out will be better informed and more interested. This argument did not die with the Progressives. As one official of the Michigan Association of School Boards stated more recently, in defense of off-cycle elections, "We want to get the most knowledgeable people at the polls, not necessarily the masses.”17
It is possible to critique such an argument on theoretical or normative grounds. First, the idea that poll access should be restricted to the most knowledgeable voters is anti-democratic and runs counter to US election law. Second, the most knowledgeable voters may have interests that diverge from “the masses.” For example, as noted previously, teachers are far more likely to vote in school district elections. Teachers are almost certainly more knowledgeable about district performance than the typical voter is. On the other hand, teachers naturally have a professional interest in obtaining more generous salaries and benefits than the typical voter might support. Consistent with this concern, there is evidence that school districts with off-cycle elections– where teachers presumably constitute a larger proportion of voters–pay higher teacher salaries.18 In other words, even putting aside democratic concerns about ballot access, there may be a tradeoff involved in engaging a more informed electorate, if better informed voters also have material interests that are at odds with those of the larger electorate.
Theoretical considerations notwithstanding, there is empirical evidence that voters are better represented by their elected officials when local elections are held concurrently with higher level races. One study shows that school board members are more likely to hold political preferences that are aligned with their constituents when boards are elected in on- versus off-cycle races.19 Another study shows that voters in school board elections are more likely to hold incumbents accountable for changes in student test scores when those races are synced with presidential elections.20
A related concern is that, even if voters do turn out to the polls in higher numbers when local elections are synced with higher level races, they may not actually mark the ballot for the local contests. In other words, ballot fatigue could lead to roll off, in which voters mark their choice for higher offices, such as president and governor, but fail to mark the ballot for local races. This might be expected if voters were uninformed or simply uninterested in local elections. If true, then roll off might undo the apparent increase in representativeness of the electorate in on-cycle elections. While evidence on this question is limited, an analysis of San Diego mayoral voting in 2012, which was synced with the presidential election, showed relatively low levels of roll off that was not strongly correlated with precinct demographics.21
While this is certainly an area in which more research is called for, the evidence to date challenges the view that on-cycle elections attract uninterested voters who are unable to make informed choices. If anything, the research suggests the opposite.
Conclusion
"Not only is election timing one of the most powerful reforms to improve turnout, it is also relatively simple to achieve."
Changing local election dates to coincide with higher level races is one of the most powerful reforms for increasing voter turnout. Turnout for synced local elections is often more than double the rate of turnout for unsynced elections, and the voters who turn out in synced elections are significantly more representative of the electorate overall. These effects are especially strong when local elections are synced with presidential elections, rather than midterm or primary elections. While some have expressed concerns that voters in synced elections will be less informed about local issues, research suggests that synced elections actually lead to better representation and accountability.
Not only is election timing one of the most powerful reforms to improve turnout, it is also relatively simple to achieve. In many cases, individual local jurisdictions can alter their election timing by revising their charters. In other instances, state law may need to be amended to allow or require concurrent elections.
Endnotes
1 http://whovotesformayor.org (last accessed June 8, 2023)
2 Holbrook, Thomas M., and Weinschenk, Aaron C., “Campaigns, Mobilization, and Turnout in Mayoral Elections,” Political Research Quarterly, vol. 67, no. 1 (2014): 42–55
3 Caren, Neal, “Big City, Big Turnout? Electoral Participation in American Cities,” Journal of Urban Affairs 29 (1) (2007): 31–46
Wood, Curtis, “Voter turnout in city elections,” Urban Affairs Review, 38(2) (2002): 209-231
4 Weimer, Debbi, “School Board Elections,” Washington, DC: The Education Policy Center (2001)
5 Suozzi, Thomas R., “Special District Election Date Study: A Crazy Quilt,” Nassau County Executive news release (2007)
6 Hajnal, Zoltan L. and Trounstine, Jessica, “Race and Class Inequality in Local Politics,” Report of the APSA Task Force on Racial and Class Inequalities in the Americas (2016). https://www.apsanet.org/inequalities (last accessed June 10, 2023)
7 http://whovotesformayor.org (last accessed June 8, 2023)
8 Berry, Christopher, Imperfect Union: Representation and Taxation in Multilevel Governments (Cambridge University Press, 2009)
9 Moe, Terry, “Political Control and the Power of the Agent,” The Journal of Law, Economics, and Organization 22(1) (2006): 1-19
10 Anzia, Sarah, “Partisan Power Play: The Origins of Election Timing as an American Political Institution,” Studies in American Political Development 26(April) (2012): 24-49
11 Hajnal, Zoltan L. and Lewis, Paul G., “Municipal Institutions and Voter Turnout in Local Elections,” Urban Affairs Review 38, no. 5 (2003): 645–68
Hajnal, Zoltan L., America’s Uneven Democracy (New York: Cambridge University Press, 2010)
Hajnal, Zoltan L., Kogan, Vladimir, and Markarian, Z. Agustin, “Who Votes: City Election Timing and Voter Composition,” American Political Science Review, 116(1) (2022): 374-383
12 Berry, Christopher R. and Gersen, Jacob E., “The Timing of Elections,” University of Chicago Law Review: Vol. 77: Iss. 1. (2010): Article 3
Hajnal, Zoltan L., Kogan, Vladimir, and Markarian, Z. Agustin, “Who Votes: City Election Timing and Voter Composition,” American Political Science Review, 116(1) (2022): 374-383
Trounstine, Jessica, “Concurrent Elections and the Policy Priorities of the Electorate.” Working paper, University of California Merced (2020)
13 "Karen Bass Drew More Votes than Any Mayor Candidate in Los Angeles History,” Los Angeles Times, Nov. 25, 2022. https://www.latimes.com/california/story/2022-11-25/los-angeles-mayor-election-voting-record (last accessed June 9, 2023).
14 Hajnal, Zoltan L., Kogan, Vladimir, and Markarian, Z. Agustin, “Who Votes: City Election Timing and Voter Composition,” American Political Science Review, 116(1) (2022): 374-383
15 Fowler, Anthony, “Regular Voters, Marginal Voters, and the Electoral Effects of Turnout,” Political Science Research and Methods, 2015 vol.3 no. 2: 205-219.
16 Anzia, Sarah, “Partisan Power Play: The Origins of Election Timing as an American Political Institution,” Studies in American Political Development 26(April) (2012): 24-49
17 As quoted in: Reed, Lawrence, “Is there a case for election consolidation across the state or should such matters be decided at the local level?” Mackinac Center for Public Policy (2002). https://www.mackinac.org/4409 (last accessed June 9, 2023)
18 Berry, Christopher R. and Gersen, Jacob E., “The Timing of Elections,” University of Chicago Law Review: Vol. 77: Iss. 1. (2010): Article 3
Anzia, Sarah F., “The Election Timing Effect: Evidence from a Policy Intervention in Texas,” Quarterly Journal of Political Science, vol. 7(3) (2012): 209-248
19 Hartney, Michael T. and Hayes, Sam D., “Off-Cycle and Out of Sync: How Election Timing Influences Political Representation,” State Politics & Policy Quarterly, 21: 4 (2021): 335–354
20 Payson, Julia A., “When Are Local Incumbents Held Accountable for Government Performance? Evidence from US School Districts,” Legislative Studies Quarterly, vol. 42, no. 3 (2017): 421–48
21 Hajnal, Zoltan L., Kogan, Vladimir, and Markarian, Z. Agustin, “Who Votes: City Election Timing and Voter Composition,” American Political Science Review, 116(1) (2022): 374-383
About the Democracy Reform Primer Series
Narrowing the gap between research and public dialogue, the University of Chicago Center for Effective Government's Democracy Reform Primers responsibly advance conversations and strategy about proposed changes to our political institutions. Each Primer focuses on a particular reform, clarifies its intended purposes, and critically evaluates what the best available research has to say about it. The Primers do not serve as a platform for either authors or the Center to advance their own independent views about the reform; to the contrary, they serve as an objective and authoritative guide about what we actually know—and what we still don’t know—about the likely effects of adopting prominent reforms to our political institutions.
In some instances, the available evidence may clearly support the claims of a reform’s advocates. In other instances, it may cut against them. And in still others, the scholarly literature may be mixed, indeterminate, or altogether silent. Without partisan judgment or ideological pretense, and grounded in objective scholarship, these Primers set the record straight by clarifying what can be said about democracy reforms with confidence and what requires further study.
About the Series Editor
Anthony Fowler is a Professor at the Harris School of Public Policy at the University of Chicago. His research applies econometric methods for causal inference to questions in political science, with particular emphasis on elections and political representation. Fowler is currently the Co-editor in Chief of the Quarterly Journal of Political Science, and the co-author (with Ethan Bueno de Mesquita) of Thinking Clearly with Data: A Guide to Quantitative Reasoning and Analysis (Princeton University Press, 2021). Fowler earned his Ph.D. in government from Harvard University and completed undergraduate studies at the Massachusetts Institute of Technology.