How Does Lowering the Vote Threshold for School Bonds Affect Local Education Funding?

This article examines how lowering the vote threshold to approve local school bonds affects education funding. Using a difference-in-differences design and data from over 4,000 local elections in California from 1995 to 2016, the authors study the effects of Proposition 39, which reduced the required vote share from two-thirds to 55 percent. They find that the policy led to a $57 increase in approved bond funding per resident—a more than 100 percent rise. The size of bond proposals increased by $48 per resident, while vote shares fell modestly. Effects were largest in racially diverse, moderately poor, and younger communities.

Full Citation and Link to Article

Grosz, Michel, and Ross T. Milton. 2025. “Relaxing Electoral Constraints in Local Education Funding.” Journal of Policy Analysis and Management 44 (3): 567–592. https://doi.org/10.1002/pam.70023 

Extended Summary

Here is a 1,000-word summary of “Relaxing Electoral Constraints in Local Education Funding” by Michel Grosz and Ross Milton, structured by the requested sections:

Central Research Question

The central question addressed by this study is: How does relaxing voter approval thresholds for local education funding affect the behavior of school boards, voter outcomes, and the eventual level of public investment in school facilities? Specifically, the paper investigates the impact of California’s Proposition 39, passed in 2000, which lowered the vote requirement to approve school and community college bond measures from two-thirds to 55 percent. The authors aim to identify whether this policy change increased the frequency and size of bond proposals, changed voting behavior, and led to higher levels of approved capital investment in education.

Previous Literature

This work draws on and extends several strands of literature in public finance, local political economy, and education policy. One core theoretical foundation is the Romer and Rosenthal (1978, 1979, 1982) framework on agenda-setting in public referenda, where elected officials may prefer higher public spending than voters and are constrained by supermajority rules. Previous empirical research (e.g., Feld and Matsusaka 2003; Funk and Gathmann 2011) has found that voter approval requirements for tax increases reduce public spending.

Within education finance, Balsdon et al. (2003) modeled the interaction between school boards and voters over bond proposals and found that boards propose less than they prefer due to the risk of failure. Cellini et al. (2010) examined the effects of successful school bonds on housing prices, while other studies (e.g., Duflo 2001; Neilson and Zimmerman 2014) investigated the impact of school construction on student outcomes, with mixed results.

This paper adds new evidence by examining how a specific feature of electoral rules—the vote threshold—affects actual public funding decisions, filling a gap left by earlier research which often lacked variation in approval requirements or focused solely on whether a tax limit existed.

Data

The study uses a comprehensive dataset of over 4,500 local tax-related ballot measures in California between 1995 and 2016. The main data source is the California Election Data Archive (CEDA), which provides election results, proposal texts, and vote shares for general obligation (GO) bond measures and other tax proposals. The authors focus primarily on GO bonds for school districts and community colleges, which were directly affected by Proposition 39.

Jurisdiction-level data are drawn from U.S. Census records, the Common Core of Data for school demographics and student achievement, and spatial overlays to estimate populations in community college districts. The final dataset includes a jurisdiction-year panel of 1,589 entities and an election-level dataset of all bond and tax measures.

Methods

The authors implement a difference-in-differences (DiD) research design, comparing outcomes in treated jurisdictions (school and community college districts) to those in control jurisdictions (cities and counties) before and after Proposition 39 took effect in 2001. This approach isolates the causal effect of lowering the vote threshold, under the assumption of parallel trends between treated and control groups.

Two versions of the DiD model are employed: an unconditional model that includes all jurisdictions each year regardless of whether a bond was proposed, and a conditional model that focuses only on those elections in which a bond measure was proposed. The unconditional model estimates changes in proposal frequency and total approved funding, while the conditional model estimates changes in voter support (vote shares) and approval rates among proposed bonds.

The authors also develop a formal theoretical model based on disagreement voting, where a school board proposes a funding level subject to voter approval. The model predicts that reducing the vote share threshold leads to larger bond proposals and more passage, especially when there is disagreement between the board and the median voter. Empirical heterogeneity analyses further explore how effects vary with demographic and political characteristics of districts.

Findings/Size Effects

The study yields several key findings about the effects of relaxing electoral constraints through Proposition 39:

1. Proposal Behavior: There was no significant increase in the likelihood that school or college districts proposed a bond after Proposition 39. However, the average size of proposed bonds increased by $48 per resident (a 59 percent increase), suggesting that boards used their greater flexibility to pursue larger projects rather than more frequent ones.

2. Voting Outcomes: Conditional on a bond being proposed, average voter support (vote share) declined by about 5 percentage points post-reform. This is less than the 11.7-point drop in the required threshold (from 66.7% to 55%), implying that boards strategically increased proposal sizes, which reduced support but not enough to offset the lowered threshold. As a result, the probability of bond approval increased by 21 percentage points.

3. Funding Outcomes: The combination of larger proposals and higher approval rates led to a substantial increase in approved funding. On average, education jurisdictions received $57 more per resident in approved GO bond funding—a more than 100 percent increase compared to pre-reform levels.

4. Timing and Election Type: After the reform, proposals shifted heavily to even-numbered years to meet the new requirement that bond elections be held during regularly scheduled general elections (rather than special elections). This shift coincided with higher voter turnout and potentially more favorable voting environments.

5. Heterogeneity of Effects: The largest funding gains were observed in jurisdictions with:

   
   

   • Higher racial and ethnic diversity, as measured by fractionalization indices.

   • Moderate poverty levels (not the poorest or wealthiest).

   • Moderate shares of non-Hispanic White residents.

   • Lower shares of senior citizens.

   
   

   These patterns are consistent with greater disagreement between voters and elected officials in more diverse or politically heterogeneous communities, leading to larger gains when the supermajority constraint is eased.

6. No Equalization Effect: While Proposition 39 increased overall funding, it did not reduce disparities across jurisdictions. In some cases, funding gaps between low- and high-income areas widened slightly.

Conclusion

This paper offers rigorous evidence that modest changes to electoral rules—such as reducing the vote threshold required to pass local bond measures—can have large effects on public investment outcomes. Proposition 39’s reduction of the vote threshold from two-thirds to 55 percent led to significantly larger bond proposals, higher approval rates, and over 100 percent more capital investment per resident in education.

The findings support the theoretical view that supermajority rules constrain elected officials, especially in contexts with disagreement between voters and officials. When constraints are loosened, elected bodies pursue larger investments even if they receive slightly less voter support. The results also highlight that electoral rule changes can have uneven effects across communities, with greater impacts in more diverse and politically moderate districts.

From a policy perspective, the study raises important considerations about the design of direct democracy tools and their implications for equity in public goods provision. It also suggests that altering vote thresholds may be a powerful lever for increasing investment in infrastructure and education, particularly in contexts where voter-elected body preferences diverge. However, without complementary reforms, such changes may not address—and could even exacerbate—existing inequalities in local public finance.