The Relationship Between Open Enrollment and School Bond Voting

Abstract

Purpose: The purpose of this study was to identify an association between student mobility through open enrollment and voter support for school bond proposals. Specifically, we hypothesized that higher percentages of nonresident enrollment in a school district and resident exit from a district would be associated with lower levels of voter support for bond proposals. Data Collection and Analysis: We utilized publicly available data on bond proposals placed on ballots between 2009 and 2015, publicly available data from the State of Michigan for information on percentages of nonresident enrollment and resident exit, and publicly available data from the U.S. Census Bureau for data on district resident characteristics. Regression analysis was used to identify associations between nonresident enrollment and resident exit with the percentage of “yes” votes on school bond proposals controlling for community and resident characteristics. Findings: We failed to reject the null hypotheses, finding no statistically significant association between nonresident enrollment and resident exit and average voter support for school bond proposals, ceteris paribus. Implications for Research/Practice: We laid some groundwork for reconceptualizing the relationship between open enrollment policies and communities’ willingness to support local public schools. This has potential implications for both local- and state-level policies regarding enrollment issues and issues of school finance. As local boards continue to struggle with budget shortfalls and mounting capital needs, they may need to further weigh their own communities’ interest in supporting local public schools in the wake of increased student mobility in and out of districts.

Keywords

school choice school bonds open enrollment empirical study

In 1994, Proposal A was passed by Michigan voters thus shifting the funding mechanism for school operating expenditures in Michigan from being primarily based on local property tax revenue to a statewide per-pupil allowance funded largely through an increase in the state sales tax (Mich. Comp. Laws § 388.1620, 1979a; Walcott, 2016). Although this policy shift (voted on by Michigan residents) was designed to significantly reduce property tax burdens (which it did) and help equalize school funding (which it did, to an extent), it substantially increased demand for student enrollment growth as a mechanism to increase school revenues by eliminating local districts’ authority to request tax increases for school operations. Additionally, since 1996, through the addition of Sections 105 and 105c to the Michigan State School Aid Act, Michigan districts have been allowed to enroll nonresident students when open seats are available (Mich. Comp. Laws § 388.1705, 1979b; Mich. Comp. Laws § 388.1705c, 1979c). Although districts are allowed to enroll nonresident students, they are not required to do so when seats are available, with the decision as to whether or not to participate in open enrollment lying with local boards of education. Although the vast majority of Michigan districts enroll at least one nonresident student, not all open their schools to nonresidents even when seats are available.

While open enrollment was purportedly established to allow students in underperforming school districts to exit their residency zoned schools and spur improvements through market competition, it ultimately created a new supply of students for districts to draw from and contributed to drastic changes in many districts’ enrollment patterns (Pogodzinski, Lenhoff, & Addonizio, 2018). Michigan provides a particularly robust market setting for school choice, with well over 500 local traditional public school districts and more than 300 charter schools (enrolling ~145,000 students in 2016-2017; National Alliance for Public Charter Schools, 2018). For example, the three counties constituting Metro Detroit include 83 traditional local districts, or roughly one district per 24 square miles (Holme & Richards, 2009). This market for student enrollment was further fueled by stagnant per-pupil funding allowances and a declining K-12 student population across the state over the past decade, thus leading districts to try to compensate for fewer resources by increasing student enrollment through participation in open enrollment. In 2014-2015, more than 90% of Michigan districts enrolled at least one nonresident student, and approximately 169,000 students (11% of the total student population) attended a traditional public school outside of their resident district. Although the majority of districts had only small percentages of their student enrollment constituted by nonresident students, there were 64 districts with nonresident enrollment of at least 500 students that had 15% of their total enrollment composed of nonresident students. Thus, for at least some districts, a significant percentage of students, and therefore state revenue, was attributed to the enrollment of nonresident students.

Although local operational funding was largely centralized at the state level based primarily on a per-pupil foundation allowance, funding for capital improvements are still generated through local property tax revenues, particularly through the issuance of voter approved bonds. These local investments in capital improvements can affect student outcomes, as researchers have shown that districts’ capital stock is significantly associated with student academic achievement (e.g., Jones & Zimmer, 2001). Yet according to the American Society of Civil Engineers (ASCE, 2017) Schools Infrastructure Report Card, across the United States we should be investing $38 billion more per year than we are to ensure adequate facilities for our nation’s students (ASCE, 2017). Therefore, local property taxes continue to provide an important stream of financing for local school systems in Michigan and most other states.

There has been extensive and valuable research on predictors of the passage of school bonds, most recently by Bowers and Lee (2013) and Ingle, Johnson, Ryan Givens, and Rampelt (2013), but as a whole, this research has failed to theoretically or empirically account for the potential relationship between open enrollment policies and the likelihood that a school bond would be passed by voters. Specifically, it has not adequately addressed questions regarding local residents’ willingness to tax themselves to increasingly fund education for nonresidents, nor the association between resident student exit and support for local education systems. For example, researchers have continually examined the relationship between the percentage of individuals who are 65 years of age or older, or who do not have children, and the likelihood that a bond passes (e.g., Berkman & Plutzer, 2005; Glass, 2008; Zimmer, Buddin, Jones, & Liu, 2011), often reporting conflicting results. These types of variables, though, do not fully capture community commitment to schools when nonresident students are increasingly enrolling in local schools through open enrollment and resident students are exiting via open enrollment, charter schools, and private schools.

This study therefore builds on previous studies through exploratory analysis of the relationship between student mobility and support for local schools as evident in voter support for school bond proposals. More specifically, this work seeks to provide a first step toward identifying the association between the percentage of nonresident student enrollment and resident student exit and the passage of a property tax bond—testing the assumption that residents are less likely to vote to tax themselves to support a local school system that increasingly educates fewer of the district’s resident children.

Open Enrollment Policy

In general, open enrollment policies seek to sever the connection between a student’s residence and school. This has implications for school leadership in a new environment where local district residents and taxpayers correspond less and less closely with district students and their families. Since open enrollment policies either eliminate or expand enrollment boundaries within the traditional public school system, they do not fundamentally alter school finance or governance structures (as compared with expanded choice through the emergence of charter schools or through private school vouchers). At the same time, if state funding of public education is dispersed according to enrollment, as it is in Michigan and almost every other state, such policies can alter the distribution of state funds across districts. Indeed, much of the claimed incentive for school improvement in a choice system arises from the universality of pupil-based funding.

Open enrollment policies were designed in part to allow students to exit “failing” schools and districts, often characterized by high concentrations of low-income and non-White students. Thus, these policies have the potential to mitigate economic and racial segregation prevalent across many regions of the country (Denton, 1995; Fuller & Elmore, 1996). Conversely, interdistrict choice could intensify economic and racial segregation by allowing families to sort themselves across school systems along racial/economic lines without having to physically move their household (Renzulli & Evans, 2005). Researchers have shown that in districts with high concentrations of low-income and non-White students, White and higher income students were more likely to exit through open enrollment policies (Cowen, Creed, & Keesler, 2015).

A further issue that arises from interdistrict open enrollment policies relates directly to how school districts are funded. In school funding systems that rely on local property tax for at least some part of funding, a potential free rider issue arises when nonresident students are enrolled. This problem is exacerbated in Michigan’s school capital funding system, which relies exclusively on local property taxes. In such cases, nonresident parents are not paying the property tax that supports the school where their children attend. Although they pay property tax to support their resident district, there could be extensive variation in the tax rate and overall value generated between districts. Additionally, legislation signed into law in January 2018 allows charter schools (which are not allowed to levy millages) to receive additional funding from countywide enhancement millages, thus spreading generated revenue across a great number of schools (Mich. Comp. Laws § 380.705, 1976).

Given the increased incidence of free ridership and the potential changing demographics of the student population, it is not clear how interdistrict open enrollment policies may alter voter behavior with respect to supporting local school bond referenda. Therefore, we provide an initial analysis of the extent to which nonresident enrollment is associated with the likelihood of school bond passage, thus testing the extent to which voters are sensitive to increasing their tax bills to support local schools that serve nonlocal students. Additionally, we provide an initial analysis of the extent to which resident student exit from the local district is associated with the likelihood of school bond passage, testing the extent to which local voters support local schools when increasing numbers of local children do not attend the community’s schools.

Interdistrict Open Enrollment in Michigan

The Michigan Legislature first approved an interdistrict open enrollment program in 1996 to allow districts to enroll nonresident students from districts within their own intermediate school district (ISD; Mich. Comp. Laws § 388.1705, 1979b). Subsequent changes to the legislation have allowed districts to enroll nonresident students from contiguous ISDs as well (Mich. Comp. Laws § 388.1705c, 1979c). Given that Michigan moved to a system of funding school operations through a state-level per-pupil foundation allowance in 1994 (Mich. Comp. Laws § 388.1620, 1979a), districts with declining or stagnant resident enrollment had an incentive to enroll nonresident students in order to generate operating revenue. The receiving district would get the lesser of the per-pupil foundation allowance of the sending or receiving district (along with all categorical aid that a student was eligible for).

The law requires that districts declare whether or not they accept nonresident students and are required to publicly advertise such information, and they are not allowed to discriminate based on income, race/ethnicity, performance level, or special education status, though they can restrict enrollment of students with records of behavior issues. Districts are, however, given considerable discretion in determining the number of open seats available, the grade levels within which they will accept nonresidents, and the schools that are open to nonresidents. The State does require that if there are more applications for enrollment than open seats, a lottery must be used for determining which students will be enrolled. The State does not collect data on open seats or enrollment capacity, so districts adhere to these policies at their will.

District participation in open enrollment increased steadily in the early years, and by the fifth year, 80% of districts were participating. Although districts were willing to “supply” the open seats, by 2001 only about 2% of students (~33,500) were attending a district outside of their resident boundaries (Cullen & Loeb, 2003). In the early years, participation was most pronounced in rural and urban districts and lowest in high-income districts and districts with a growing resident population (Arsen, Plank, & Sykes, 2000). Arsen and colleagues (2000) also found that urban districts experienced a net loss of students (on average 0.7%) through open enrollment participation. Except for the year 2001-2002, the rate of student participation in open enrollment steadily increased, and in 2014-2015, 11% (169,000) of Michigan students attended a traditional public school district outside of their residency zone. Participation was largely concentrated in urban areas, thus exacerbating the net loss of students in central-city districts due to population declines.

Public School Capital Funding and Bond Referenda

Across the United States, there is great need for investment in school facilities, with a large percentage of schools in need of significant repair (Alexander & Lewis, 2014; ASCE, 2017). Yet concern has been growing in the United States not only over the quality of public school infrastructure but also over the ability or willingness of states and localities to ensure that all children have access to an adequate school building (Plummer, 2006). The commitment of citizens to vote for funding of school facilities may be particularly strained as a growing number of students do not attend their local traditional public schools, opting instead to attend private schools, neighboring traditional public schools, or charter schools.

Passing a School Bond Measure and Student Mobility

Research on school bond passage has focused on several key variables, including resident population characteristics (e.g., age, income status, and race/ethnicity), student characteristics (e.g., race/ethnicity), district characteristics (e.g., long-term indebtedness, employee costs), and bond characteristics (e.g., amount, number of times proposed) (Bowers & Lee, 2013; Bowers, Metzger, & Militello, 2010; Ehrenberg, Ehrenberg, Smith, & Zhang, 2004; Glass, 2008; Ingle et al., 2013; Ladd & Murray, 2001; Sielke, 1998; Zimmer et al., 2011; Zimmer & Jones, 2005). Recent articles by Bowers and Lee (2013) and Ingle et al. (2013) provide extensive reviews of the extant research on predictors of school bond passage. Rather than rehash the quality of work that has already been published, we focus this section specifically on issues related to school choice and school funding that have thus far been underresearched.

There have been relatively few studies focused on issues of school choice and voter preferences for supporting school bond passage, though what work has been done has important implications for practice and future research. For example, using data from California, Colorado, Minnesota, and Wisconsin, Shober (2011) examined the association between enrollment in charter and magnet schools and school bond referenda, positing that school districts with more options through charter schools and/or magnet schools may increase voters’ willingness to support school bond referendums. Subsequently, Shober (2011) reported that while the number of charter schools in a district was associated with an increased likelihood of bond passage, the number of magnet schools was not.

While a district’s willingness to expand choice options may be one way to gain support from voters, an alternative interpretation is that increased competition from charter schools may push voters to pay for upgrades to facilities in order to maintain or increase student enrollment numbers in the face of external competition. In other words, in a market for student enrollment, communities may be more willing to fund capital improvements in order to establish school systems that are more attractive to resident consumers. From Shober’s (2011) study, it is not clear whether nonresident students were populating the charter and magnet schools within the given districts. Additionally, in many places such as Michigan, only a small fraction of charter schools are chartered by traditional school districts, with the vast majority being chartered by institutions of higher education. Thus, charter schools largely operate in direct competition with traditional school districts, not as options within them.

In another study, Ehrenberg and colleagues (2004) utilized panel data from 380 school districts in New York, which included data from 1985-1986 to 1996-1997, to identify predictors of passage of local school budgets (admittedly different from school bond issues). Although a limited measure of school choice, the researchers reported that the rate of private school attendance within a district was negatively associated with voter approval of local school budgets, even after controlling for other resident characteristics such as race, income, and education. This suggests that residents are less likely to financially support local school districts through higher taxes if a significant amount of local resources (as measured by tuition) are being spent in other schools.

Although the study by Ehrenberg and colleagues (2004) has important implications for school choice and school financing, it does not adequately account for the realities of open enrollment, where potentially thousands of students enter or exit the district. More specifically, individuals who send their children to private schools likely spend a significant amount of money on tuition, thus reducing their willingness or ability to pay higher taxes. The researchers did account for a measure of student mobility related to increases/decreases in total student enrollment (which was not statistically significant), but they did not account for the percentage of resident/nonresident students.

In a third study, Ingle et al. (2013) utilized data from Ohio (2007-2010) to estimate logistic regressions to identify the relationship between campaign expenditures and the passage of school levy referenda, while controlling for district and levy characteristics. Included in their controls was a measure of the racial differences between the resident population and the local school population. This measure speaks to an underlying assumption that residents who on average have racial classifications different from the student body would be less willing to support local bond measures. At the same time, this measure does not directly speak to issues of open enrollment, as in many of these districts, there may simply be an aging White population and a younger school-aged non-White population within the district boundaries. The coefficient for this measure of racial difference between resident population and school enrollment was not statistically significant.

Although previous studies have identified several key predictors of school bond passage, collectively they do not fully account for the potentially real and perceived changing role a local community school has in providing education for local residents as well as students from neighboring communities. In other words, in an era of increased parent/student choice related to schooling options (both through open enrollment policies in traditional public schools and the expansion of charter school options), residents of a local community may feel less of an attachment to their local school system. In Michigan, greater school choice has been actively pursued for more than two decades by Betsy DeVos and others. Under the leadership of DeVos as the U.S. Secretary of Education in the Trump administration, the federal government has pursued educational policies geared toward expanding school choice through advocating for more financial resources for charter schools and the expansion of school vouchers and tax credit programs to encourage private school enrollment. Additionally, the rhetoric of calling public schools “government schools” to portray a negative connotation has continued to radiate from the Trump administration as well as from conservative think tanks and media (Stewart, 2017). In other words, we have seen at the attempts of both the federal and state levels to challenge the legitimacy of many of our traditional public schools. This broader policy context regarding school choice and support for traditional public institutions potentially poses additional problems for district leaders to gain voter support for passing school bonds.

District Leadership

Although local school boards have ultimate authority, most often they empower district administrators to take the lead on various policy issues. Therefore, in addition to board members, district administrators play a direct role with regard to both local interdistrict choice policy and maintaining capital funding. In other words, a key role for district administrators (i.e., superintendents and business managers) in an education marketplace is to maintain or increase enrollment, generate revenue for school operations, and maintain, improve, and/or build capital infrastructure.

Interdistrict choice

Although policies regarding interdistrict choice vary across states, district leaders often have considerable discretion in shaping local policies regarding the enrollment of nonresident students. As previously noted, in Michigan districts are not required to participate in open enrollment, and those that do participate decide whether to only enroll nonresidents within their same ISD or also enroll nonresidents from contiguous ISDs. Furthermore, district leaders determine how many open seats are available at each grade level and the schools within their district that are open to nonresidents. For example, for the 2018-2019 school year, the Berkley School District (located in Oakland County in Metro Detroit) indicated that they would enroll up to 120 nonresident Oakland County students in Grades K-5, as well as up to 20 nonresident students in their ninth grade scholars program, which required specific academic credentials such as a minimum 3.50 middle school grade point average, completion of Algebra I, and demonstrated proficiency on standardized test in reading and mathematics (Berkley Schools, 2018). Applications for enrollment for K-5 were only taken from February 19 through March 9 and from February 26 through March 16 for the ninth grade scholars program.

Neighboring Ferndale Schools (also in Oakland County) had a very different policy for the 2018-2019 school year. They accepted applications for open enrollment in Grades K-10 for Oakland County residents across all their schools (application period of March 13 through August 3) but restricted enrollment of students from contiguous counties to Grades 9-12 in two schools, University High school (which primarily serves students from Detroit) and the Tri-County Educational Center (an alternative high school and adult education center). The application window for enrollment for these schools was from March 13 through August 31 for University High School, and from March 13 through September 7 for the Tri-County Educational Center (Ferndale Schools, 2018).

These two examples of neighboring districts not only highlight the varying realities of districts in the education marketplace, but the extent to which district leaders’ decision making can shape the distribution of nonresident enrollment within and across schools and districts. Although the local boards of education ultimately approve nonresident enrollment policies, superintendents and their associates are tasked with maintaining/increasing enrollment while also preserving the academic standards, and being sensitive to the potential racial and class differences between resident and nonresident students, which can not only pose challenges within the schools but also potentially lead to resident enrollment declines (Pogodzinski et al., 2018). This requires the superintendent to not only understand the financial implications of enrollment of both resident and nonresident students but also understand the potential race/class conflict that is often inherent through open enrollment, particularly in metropolitan areas across Michigan.

Ballot proposals

In addition to making decisions regarding enrolling nonresident students, school boards and district administrators make decisions about generating revenue to maintain and improve capital infrastructure, including buildings, technology, and transportation. They are responsible for managing current debt loads, assessing the short- and long-term capital needs of the district, determining the amount and language of any ballot proposal, deciding when to place a proposal on a ballot (e.g., primary election date, November election, during a presidential election cycle), and the political considerations in the district (Ingle, Johnson, & Petroff, 2012). Specifically, district leaders can enact strategies to increase the likelihood that a ballot proposal will be passed. In a summary of research on passing school ballot proposals, Ingle et al. (2012) state that these strategies include “understanding what voters will support, presenting a focused referendum, presenting a unified front from multiple stakeholders, targeting “yes” voters, communicating important information, and developing good relationships with the local media” (p. 820).

What is not yet known is the extent to which student mobility through open enrollment is included in the calculus regarding putting a school ballot proposal forward and the strategies used to gain voter support. For example, district leaders need to understand not only the demographic and population trends within their zoned catchment area but also the patterns of nonresident enrollment within their district and surrounding areas. Furthermore, these calculations are guided by their understanding of voters’ preferences for supporting the local schools, and this calculation possibly varies significantly depending on the extent to which resident students are enrolled in their local school system and the extent to which nonresidents are enrolled. For example, during the 2017-2018 school year in the River Rouge School District (which borders the City of Detroit), approximately 60% of enrolled students were nonresidents (primarily from Detroit), while more than 40% of the resident population enrolled in schools outside of the district (including both charter schools and other traditional public school districts) (MI School data, 2018). Such shifts in enrollment likely factors into district leaders’ vision for the future of their district.

A Framework of Open Enrollment and Bond Passage

The primary focus of the analysis presented in this work relates to identifying the relationship between student mobility across districts and voter support for school bond referenda. Since Michigan school operating funds primarily come from a State allotted per-pupil foundation allowance, it could be argued that a district would seek to meet or build enrollment capacity in order to increase revenue. This could be done through retaining resident students, attracting new residents, and/or enrollment of nonresident students. The demand for students could then motivate school officials to propose capital projects to be funded by local tax payers that would make the district more attractive to families (both resident and nonresident alike; Arsen & Davis, 2006). One could further argue that individuals are more likely to support local schools if they perceive that “good” schools provide a private benefit in the form of increased home value (Hilber & Mayer, 2009). Therefore, to the extent that stable or growing enrollments and capital investments in school infrastructure reflect a “good” school system, voters may be more likely to support school bond referenda.

Our framework challenges these assumptions in fundamental ways. First, as the proportion of nonresident students in a district increases, the problem of free ridership associated with local taxation of the school district becomes greater, which may suppress voter support for bond referenda. In other words, local taxes would be funding school improvements for a larger proportion of students whose parents do not pay the local property tax that supports capital improvements.

Second, the potential for private benefits for voters, in terms of increased home values associated with “good” schools, is not guaranteed in the face of increased nonresident enrollment. Although districts likely try to game the system in some ways, by law they are not allowed to discriminate when enrolling nonresident students. Therefore, districts that enroll large numbers of nonresidents possibly risk adverse peer effects, real or perceived, in terms of enrolling students who are lower achieving and/or economically disadvantaged.

Additionally, issues of racial/ethnic demographics should not be ignored. Recent analysis of open enrollment in Michigan showed that low-income and African American children (who were on average lower performing on State-mandated achievement assessments) were more likely to attend a school outside of their residency zone (Cowen et al., 2015). Additionally, Pogodzinski and colleagues (2018) showed how the enrollment of lower income and African American nonresident students in Metro Detroit played a fundamental role in changing several districts’ student demographics. If the enrollment of nonresident students significantly changes the demographics of the school district in terms of race/ethnicity, economic status, and/or student achievement, it may negatively affect home values.

It is also important to consider resident student exit from a local school system. Open enrollment policies further allow local residents to disengage from the local school by sending their students elsewhere. Specifically, individuals who actively choose to send their children outside of their local district may be even further disinclined to provide support to the local system. This logic follows findings with regard to communities with large private school enrollments and their support for local public schools (Ehrenberg et al., 2004). At the same time, leaders of districts that face large numbers of student exit (or threat of exit) may seek to make capital improvements to not only attract nonresidents but also retain resident students (Arsen & Davis, 2006).

Although one may argue that the average voter is not well-informed about the demographics of the student population, the proportion of nonresident students, or resident student exit in their district, there has been increased media attention in Michigan on open enrollment’s relationship to changing school demographics. For example, an article from Bridge Magazine, which was copublished in the Detroit Free Press, highlighted the potential for racial tension in communities that have school districts that enroll large numbers of lower income African American nonresident students (Wilkinson & Dawsey, 2016). Additionally, given the extent of choice in schooling options for parents, it is more likely that those with school-aged children have a sense (well- or ill-informed) of the student population in the district. Those without school-aged children (particularly those past childbearing age) may be less informed or sensitive to student demographics.

In sum, the conceptual framework that guides our analysis posits that voter support of school bond referenda is influenced by economic concerns such as income and tax share, but support is also influenced by the in-flow of nonresident students through open enrollment and the exit of resident students through open enrollment and other options (e.g., private and charter schools). Although school officials may be motivated to increase spending on capital projects to attract and/or retain students, we suggest that there is likely a point where the proportion of nonresident students reduces voter support due to concerns about free ridership and potential negative impact on school quality, particularly if the nonresident students are on average more likely to be low income, lower achievers, and/or racial/ethnic minorities. Furthermore, we suggest that high rates of resident exit from a local school system signals lack of support for the local school system (voting with their feet), which likely manifests in lack of voter support for school bond referenda.

Although we contend that inflow and outflow of students from a local district likely relates to voting behavior differently, both may be associated with voter preference for supporting school bond measures. Therefore, we tested the following hypotheses:

Hypothesis 1: As the percentage of nonresident students increases in a district, the less likely it is that voters of that district will support a school bond measure.

Hypothesis 2: As the percentage of resident student exits from a district increases, the less likely it is that voters of that district will support a school bond measure.

Methods of Data Collection and Analysis

To test our hypotheses, we drew on publicly available data for the school years 2009-2010 to 2015-2016 (August 2009 through July 2016) from the State of Michigan and the U.S. Census Bureau American Community Survey (ACS). Specifically, data on school bond votes, student enrollment, and property values were collected from the Center for Educational Performance and Information, the Michigan Department of Education, and the Michigan Department of Treasury. Additional data on resident characteristics were accessed through ACS, utilizing the 5-year annual estimates.

Our research focused on the 368 school bond proposals held during this time period across 200 districts (one proposal was not included due to missing data from the ACS, so the starting sample was 367 bond proposals). There were several instances where on a given ballot, there was more than one bond proposal for a given district. Given that the goal of our analysis was to essentially measure community members’ willingness to support public education in the context of nonresident enrollment and resident exit, we only kept the cases for the bond measure that received the highest percentage of yes votes when there were multiple bond proposals on a single ballot. It is our contention that this allows us to more accurately model the relationship between voter support for school bond referenda and nonresident student enrollment and resident exit, as focusing on proposals that received the greatest support could represent a ceiling of voters’ willingness to support school bond referenda in general (while accounting for bond proposal amount). Therefore, the sample size was reduced to 328 bond proposals, which represented the highest supported bond proposals in a given district on a given ballot.

As there were more than 540 traditional public school districts in Michigan during the time period studied, we acknowledge that the sample used in our analysis is not necessarily representative of all districts in Michigan. Not only does the condition of school facilities vary across districts, but school leaders also make calculations as to when to place a bond proposal on the ballot. Therefore, relatively few districts put forward bond proposals in any given year (in the range of years of this study, there was a high of 57 districts with bond referenda on the ballot in a given year).

Analytic Approach

We took a multistep approach to test our hypotheses, beginning with estimating the following general regression:

{PCTYES = β}_{0} {+ β}_{1} {NONRES + β}_{2} {EXIT + β}_{3} P + β_{4} C + β_{5} E + β_{6} {D + β}_{7} Y + e,

(1)

where voter support for a bond measure (measured in percent yes vote; PCTYES) is modeled as a function of the proportion of nonresident students in a district (NONRES), the proportion of resident students who did not attend school in the local district (EXIT), bond proposal characteristics (P), community characteristics (C), and election characteristics (E). We also controlled for district fixed effects (D) and year fixed effects (Y), and the error term was assumed to be normally distributed. For this exploratory study, we prefer this general methodological approach compared with that used in other studies that utilized logistic regression (e.g., Ingle et al., 2013) or discrete time hazard modeling (e.g., Bowers & Lee, 2013), because with a continuous variable, there is additional variation to explore regarding the potential relationship between open enrollment and voter support. In other words, the relationship between open enrollment and voter support may be detectable across communities where bond proposals passed or were defeated. This could be particularly important for district leaders in communities with increasing nonresident enrollment and/or increasing resident student exit.

For the focal predictor variables, the number of nonresident students and resident exit through public school options were accessed through the MI School Data website (www.mischooldata.org). The measure of resident exit through private school attendance was gathered through the ACS data. The measure of EXIT includes the addition of exit through public school choice (traditional public school and charter) as well as estimates of private school enrollment. It is possible that tax preferences vary between those who exit through public school options compared with those who exit through private school, but given the relatively small percentage of private school enrollment in most districts, we grouped them together in this analysis.

Theoretically, we predicted that voters would likely only respond if there were relatively large numbers of nonresident students within a district. We therefore estimated a second model after transforming the continuous variables for NONRES and EXIT to a set of dummy variables representing quartiles of the distribution within the sample (the first quartile was the reference category excluded from the model).

Our selection of control variables was guided by our overall framework and the theory of school bond passage proposed by Bowers and Lee (2013). Table 1 reports descriptive information on the variables included in the analysis. Bowers and Lee’s theoretical framework places the bond as the unit of analysis for research to be more meaningful for district leaders and policymakers alike. It focuses on bond characteristics, community characteristics, and election characteristics. With regard to bond proposal characteristics, we controlled for bond amount, which is related not only to the proposed project but also often to the number of times a particular bond has been proposed. With regard to community characteristics, we controlled for the percentage of resident population aged 65 years and older, percentage of residents with at least a bachelor’s degree, percentage of residents who were White, and urbanicity of the district based on the Common Core of Data designation of city, suburb, town, or rural (rural was the omitted reference category). These types of variables have been found in some research to be associated with bond passage (e.g., Bowers et al., 2010; Bowers & Lee, 2013; Fletcher & Kenny, 2008; Shober, 2011; Tedin, Matland, & Weiher, 2001; Zimmer & Jones, 2005). We further controlled for the log of taxable homestead and nonhomestead property value and log of median income. With regard to election characteristics, we controlled for the percentage of adults 18 years and older who voted in the election. Previous research has shown that higher voter turnout is negatively associated with bond passage (e.g., Bowers & Chen, 2015; Piele & Hall, 1973). Finally, we controlled for year fixed effects to account for any year-specific elements that may be associated with the bond proposal and voter support for bond proposals.

Table 1.

Descriptive Information for Variables.

Variable	Description	M	SD	Min.	Max.
PCTYES	Percentage of voters who voted to pass school bond proposal	52.58	11.24	12.64	85.23
NONRES	Percentage of nonresident students in district	13.72	10.43	0.47	45.53
NONRES Q1	Nonresident 1st quartile (0.000-0.603)	0.25	0.43	0.00	1.00
NONRES Q2	Nonresident 2nd quartile (0.6027-10.746)	0.25	0.43	0.00	1.00
NONRES Q3	Nonresident 3rd quartile (10.747-18.042)	0.25	0.44	0.00	1.00
NONRES Q4	Nonresident 4th quartile (18.043+)	0.25	0.43	0.00	1.00
EXIT	Percentage of resident student exit	23.70	9.66	4.36	57.86
EXIT Q1	Resident exit 1st quartile (0-17.329)	0.25	0.43	0.00	1.00
EXIT Q2	Resident exit 2nd quartile (17.330-22.188)	0.25	0.43	0.00	1.00
EXIT Q3	Resident exit 3rd quartile (22.189-27.991)	0.25	0.43	0.00	1.00
EXIT Q4	Resident exit 4th quartile (27.992+)	0.25	0.43	0.00	1.00
AMOUNT	Bond proposal amount in tens of millions of dollars	2.24	3.69	0.06	50.05
POP 65+	Percentage of population that is 65 years or older	15.38	4.11	6.70	32.10
BA DEGREE+	Percentage of population with at least a bachelor’s degree	21.59	11.49	6.20	78.20
POP WHITE	Percentage of population that is White	91.96	8.22	15.39	99.54
TAX	Log of taxable value	19.72	0.91	17.44	22.75
INCOME	Log of median income	10.82	0.24	10.19	11.54
TURNOUT	Percentage of adult population that voted	23.08	10.49	1.20	66.26
RURAL	CCD designation of rural	0.46	0.50	0.00	1.00
URBAN	CCD designation of urban	0.04	0.20	0.00	1.00
SUBURBAN	CCD designation of suburban	0.29	0.45	0.00	1.00
TOWN	CCD designation of town	0.20	0.40	0.00	1.00

Note. M = mean; SD = standard deviation; Min. = minimum; Max. = maximum; CCD = Common Core of Data.

Results

Table 2 reports the estimates for the two models. Column 1 reports the estimates from modeling PCTYES as a function of the continuous measure of the percentage of nonresident students in a district and resident student exit, controlling for other factors. Column 2 reports the estimates from modeling PCTYES as a function of the quartiles of nonresident enrollment and resident exit (the first quartile was the omitted reference category).

Table 2.

Estimated Effects on Percent Yes Votes.

Variable	(1)	(2)
INTERCEPT	15.119 (45.731)	11.397 (46.601)
NONRES	−0.031 (0.069)
NONRES Q2		0.547 (1.801)
NONRES Q3		0.3914 (1.897)
NONRES Q4		−0.532 (2.080)
EXIT	0.009 (0.072)
EXIT Q2		0.571 (1.760)
EXIT Q3		1.129 (1.882)
EXIT Q4		0.571 (1.881)
AMOUNT	−0.121 (0.221)	−0.128 (0.224)
POP 65+	−0.048 (0.199)	−0.035 (0.200)
BA DEGREE+	0.088 (0.084)	0.089 (0.086)
POP WHITE	0.077 (0.1000)	0.068 (0.103)
TAX	−0.891 (1.200)	−0.894 (1.251)
INCOME	4.994 (4.615)	5.305 (4.764)
TURNOUT	−0.236*** (0.064)	−0.231*** (0.066)
URBAN	3.449 (3.868)	3.477 (3.922)
SUBURBAN	−0.061 (1.856)	0.028 (1.851)
TOWN	0.319 (1.734)	0.264 (1.765)
Year fixed effects	Y	Y
F Value	2.63***	2.16***
R squared	0.1329	0.1346

p < .10. **p < .05. ***p < .01.

In Model 1, neither the coefficient for NONRES nor that for EXIT were statistically significant, and only the coefficient for NONRES was in the expected direction. The only control variables with a statistically significant coefficient was for voter turnout (p < .01), which was negative and consistent with previous research (e.g., Bowers & Chen, 2015). More specifically, higher voter turnout was associated with lower percentages of yes votes.

In Model 2, the coefficients for each of the measures of NONRES failed to reach statistical significance, but only the last coefficient was in the expected negative direction. With regard to the coefficients for resident exit, none of them were statistically significant and similar to the output in Model 1, in an unexpected positive direction. Also, similar to the first model, the only control variable that was statistically significant was for voter turnout (p < .01), again negative. Although our models with a continuous outcome variable was our preferred model, we did estimate some logistic regressions with a binary variable indicating whether or not a bond proposal was passed, but similarly there was no statistically significant association between nonresident enrollment or resident exit and the likelihood that a bond proposal was passed.

Discussion

We hypothesized that districts that are marked by a significant percentage of nonresident students could be penalized by local residents in the extent to which they would approve the issuance of school bonds to fund various school improvements. Similarly, we hypothesized that high levels of resident student exit would be associated with lower support for bond referenda, expecting that residents with children not attending the local schools would be less likely to pay more to support them. The findings we presented ultimately fail to support these hypotheses. In this section, we offer some possible insight into these findings.

Voter Knowledge and Sensitivity to Open Enrollment

One interpretation of the results presented is rather straightforward: Voters may not be influenced by open enrollment because they are on average ignorant of enrollment into and out of their district. Although we expect that parents with school-aged children are likely more aware of student enrollment and demographic patterns, it is less likely that those without school-aged children are as well-informed. Although we indicated that issues of student entrance and exit through open enrollment have garnered some increased media attention in the Metro Detroit area, the overall extent of the inflow and outflow of students, and the demographic shifts associated with such student movement, may not be noticed by the average voter. If a voter is not aware of a particular issue, then the issue cannot directly influence the voter’s voting decision.

Although we expect that many voters are simply unaware of issues around open enrollment, we continue to expect that some are and that this may influence their voting behavior. At the same time, it is not clear as to what direction it may do so. On one hand, following our hypothesis, some voters may be less likely to support school bond referenda when there is a significant percentage of nonresidents in the district due to concerns about free ridership. Relatedly, they may be less likely to support bond referenda when there is a significant percentage of resident exit, as those who voted with their feet to not engage the local district (i.e., they sent their students to a school outside of the district) may also be less likely to financially support the local school system.

On the other hand, following the findings of Arsen and Davis (2006), some voters may be more likely to support bond referenda in response to open enrollment and competition for students. Since the vast majority of districts (particularly in urban areas) are now in competition with each other for a broader pool of students across district lines, some voters may support school capital enhancements in the hopes that such investments will maintain and grow student enrollment to ensure that the district strives and survives market competition. In a sense, it may increase their commitment to supporting their local public school system. Furthermore, some residents may be concerned that degradation of school facilities in their community may ultimately harm their housing value.

Limitations of the Data and Methods of Analysis

Taken together, it is likely that there are multiple dimensions to the association between open enrollment and voter support for school bond referenda, from a null association, to associations working in opposite directions of influence on voter behavior. Given the nature of the data used in this analysis, it is difficult to further test these possible explanations. Without having direct information on voter knowledge of and opinion on open enrollment within their district, we cannot fully identify how it influences overall likelihood of bond passage while considering other covariates.

Although information on individual voter knowledge and voting behavior is not readily available without exit polling or survey research, ideally we would have longitudinal data on bond passage at the district level. However, this is complicated by the unpredictability with which districts place a bond proposal on a ballot that is not a refloat of a failed bond. We were not able to capture how the support of bond proposals in a district may change over time in relation to changes in nonresident enrollment and resident student exit. By focusing only on specific time points, it is difficult to fully account for unobservable variables that influence not only bond passage (e.g., political climate surrounding the community and the school board and/or superintendent) but also the placement of a bond proposal on a ballot in the first place. Therefore, the limits of the data constrain our ability to model the association between open enrollment and voter support for bond referenda.

Given the recent growth of open enrollment programs across the United States and the expansion of other school choice policies, we also contend that such programs increasingly affect local district leadership decision making in ways that are important but not yet well-understood. The research presented here touches on several aspects of local leader decision making, including whether or not to participate in open enrollment, how many nonresidents are allowed to enroll (and whether they come solely from within the ISD or also contiguous ISDs), when to pursue voter approval for school bonds and the amount of such bonds. This ultimately relates to leaders’ ability to understand market forces within their region related to student mobility, the financial needs of the district related to operating revenues, and the financial needs related to infrastructure. Furthermore, it highlights the need for local leaders to understand their constituents and their willingness to support their schools in a changing environment marked by increased free ridership. Although the findings presented here give no definitive guidance, we suggest that district leaders further consider how open enrollment factors into their immediate and long-term plans for sustaining their districts, which includes maintaining/increasing enrollment and generating revenue for capital expenditures. This likely relates to decision making regarding not only how to compete for students in the education market place (both resident and nonresident) but also how to maintain and increase revenue for capital expenses.

As was shown in previous work, in many districts in Michigan, there is a strong relationship between nonresident enrollment and resident student exit, suggesting the possibility that some parents “vote with their feet” by enrolling their children in charter schools or neighboring traditional public schools (Pogodzinski et al., 2018). Although the findings from our current study presented here does not demonstrate that it is necessarily associated with lack of support at the polls, we argue that student mobility through open enrollment (both inflow and outflow of students) should be of concern for district leaders as they seek to maintain their capital infrastructure.

We contend that other researchers should join our efforts to consider additional data collection methods to further investigate the relationship between open enrollment and various school and community outcomes. Specifically, qualitative data collection and analysis could help illustrate how local stakeholders both within the district and broader community grapple with issues related to enrollment and financial support for the district, particularly in districts with changing demographics due to nonresident enrollment and residential patterns. Such data collection and analysis would likely illuminate how district leaders address nonresident enrollment and infrastructure needs and also bring to the forefront community concerns regarding their local schools and concerns of free ridership.

Conclusion

The purpose of this study was not to critique the value of open enrollment policies in promoting educational opportunities for students or measure the impact such policies have on student achievement. Rather, we explored the relationship such policies may have with overall investment in public education and local communities’ continued support for local school systems. At the state level, most of the focus of school financing revolves around issues related to the per-pupil funding allowance and attempts at decreasing the gap between the top and bottom of the variation in fund allocations. Yet issues of local funding streams cannot be ignored as many districts struggle to balance yearly budgets and continually grapple with unmet capital needs for building and technology maintenance and upgrades.

Therefore, if open enrollment policies have the potential to disrupt local financial support for local schools, then the state may need to rethink the ways in which it funds public education, and more specifically modify the rules that govern how state-level funding is used (e.g., operating funds vs. capital expenses). In this sense, issues related to open enrollment and local funding cannot be divorced from larger issues related to intergovernmental financing of public education, particularly with respect to what level of centralization is appropriate for funding public education and what role local property taxes should play in generating revenue for schools. Furthermore, local leaders can utilize the information presented here to focus more attention on the interplay between nonresident enrollment and resident support for their local schools, both in terms of voter support through the passage of bonds as well as by engagement through continued resident student enrollment and community support of district initiatives.

At the heart of the state-level open enrollment policy is the desire to allow parents/students more choice, particularly for those who are designated to attend low-performing schools. Yet such policies fit more broadly within policy issues related to creating a market for education, such as the expansion of charter schools and online education. As such, these policies challenge the traditional form of education delivery—the one provided by local schools to local residents. If such policies fundamentally change people’s conception of education delivery as a local affair, they also challenge the traditional modes of funding education through local property taxes. Additionally, they may change the imperatives of effective leadership, given the continuing role of local millage elections.

Although we ultimately failed to reject the null hypotheses through our analysis, we propose that the framework we presented be further developed to guide research concerning the potentially changing dynamic of local support for school districts, particularly expanding the conceptual understanding of the relationship between the marketization of education and community support for local schools. The dynamics of school bond voting are complex given the general lack of knowledge of local- and state-level issues related to public education and funding, on top of relatively low voter turnout rates. Additionally, there are measures of community engagement with local public schools beyond financial support that should be considered. This speaks to the need to further develop the conceptualization of “local” schools in an era of increased marketization of education and student mobility. An expanded framework could then be used to guide additional data collection and analysis.

Despite the limitations of this study, we laid some groundwork for reconceptualizing the relationship between open enrollment policies and communities’ willingness to support local public schools. As suggested, this has important implications for both local- and state-level policies regarding enrollment and school finance. Public schools in Michigan (similar to many other states) continue to rely on the local tax base for key aspects of school financing. At the same time, the state has created incentives to participate in open enrollment, and for some districts, the need to fill open seats is extensive. As local boards continue to struggle with budget shortfalls and mounting capital needs, they may need to further weigh their own communities’ interest in supporting local public schools in the wake of increased student entrance and exit.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Author Biographies

Ben Pogodzinski is an associate professor in Wayne State University’s College of Education, in the educational leadership and policy studies program. His research focuses on school choice, school organization, and labor relations.

Sarah Winchell Lenhoff is an assistant professor in Wayne State University’s College of Education, in the educational leadership and policy studies program. Her research focuses on educational policy development and implementation, school choice, organizational improvement, and improvement science, with a special interest in Detroit.

Michael Addonizio is a professor in Wayne State University’s College of Education, in the educational leadership and policy studies program. His research focuses on school finance, school governance, and educational policy reform.

References

Alexander

Lewis

(2014). Condition of America’s public school facilities: 2012-13. Washington, DC: National Center for Education Statistics.

American Society of Civil Engineers. (2017). 2017 Infrastructure report card. Reston, VA: Author. Retrieved from https://www.infrastructurereportcard.org/cat-item/schools/

Arsen

Davis

(2006). Taj Mahals of decaying shacks: Patterns in local school capital stock and unmet capital need. Peabody Journal of Education, 81, 1-22.

Arsen

Plank

Sykes

(2000). School choice policies in Michigan: The rules matter. East Lansing: Michigan State University Education Policy Center.

Berkley Schools. (2018). Schools of choice. Retrieved from http://www.berkleyschools.org/about/enrollment/schools-of-choice/

Berkman

M. B.

Plutzer

(2005). Ten thousand democracies: Politics and public opinion in America’s school districts. Washington, DC: Georgetown University Press.

Bowers

A. J.

Chen

(2015). Ask and ye shall receive? Automated text mining of Michigan capital facility finance bond election proposals to identify which topics are associated with bond passage and voter turnout. Journal of Education Finance, 41, 164-196.

Bowers

A. J.

Lee

(2013). Carried or defeated? Examining the factors associated with passing school district bond elections in Texas, 1997-2009. Educational Administration Quarterly, 49, 732-767.

Bowers

A. J.

Metzger

S. A.

Militello

(2010). Knowing what matters: An expanded study of school bond elections in Michigan, 1998-2006. Journal of Education Finance, 35, 374-396.

10.

Cowen

Creed

Keesler

(2015). Dynamic participation in inter-district open enrollment: Evidence from Michigan 2005-2013 (Working Paper No. 49). Retrieved from http://education.msu.edu/epc/publications/documents/WP49-Dynamic-Participation-In-Interdistrict-Open-Enrollment-Cowen.pdf

11.

Cullen

J. B.

Loeb

(2003). K-12 education in Michigan. In Ballard

Courant

P. N.

Drake

D. C

Fischer

Gerber

E. R.

(Eds.), Michigan at the millennium (pp. 299-321). East Lansing: Michigan State University Press.

12.

Denton

N. A.

(1995). The persistence of segregation: Links between residential segregation and school segregation. Minnesota Law Review, 80, 795-824.

13.

Ehrenberg

R. G.

Ehrenberg

R. A.

Smith

C. L.

Zhang

(2004). Why do school district budget referenda fail? Educational Evaluation and Policy Analysis, 26, 111-125.

14.

Ferndale Schools. (2018). Enrollment. Retrieved from http://www.ferndaleschools.org/enrollment/

15.

Fletcher

Kenny

L. W.

(2008). The influence of the elderly on school spending in a median voter framework. Education Finance and Policy, 3, 283-315.

16.

Fuller

B. F.

Elmore

(1996). Who chooses? Who loses? Culture, institutions and the unequal effects of school choice. New York, NY: Teachers College Press.

17.

Glass

G.V.

(2008). Fertilizers, pills and magnetic strips: The fate of public education in America. Charlotte, NC: Information Age.

18.

Hilber

C. A.

Mayer

(2009). Why do households without children support local public schools? Linking house price capitalization to school spending. Journal of Urban Economics, 65, 74-90.

19.

Holme

J. J.

Richards

M. P.

(2009). School choice and stratification in regional context: Examining the role of inter-district choice. Peabody Journal of Education, 84, 150-171.

20.

Ingle

W. K.

Johnson

P. A.

Petroff

R. A.

(2012). “Hired guns” and “legitimate voices”: The politics and participants of levy campaigns in five Ohio school districts. Educational Administration Quarterly, 48, 814-858.

21.

Ingle

W. K.

Johnson

P. A.

Ryan Givens

Rampelt

(2013). Campaign expenditures in school levy referenda and their relationship to voter approval: Evidence from Ohio, 2007-2010. Leadership and Policy in Schools, 12, 41-76.

22.

Jones

J. T.

Zimmer

R. W.

(2001). Examining the impact of capital on academic achievement. Economics of Education Review, 20, 577-588.

23.

Ladd

H. F.

Murray

S. E.

(2001). Intergenerational conflict reconsidered: County demographic structure and the demand for public education. Economics of Education Review, 20, 343-357.

24.

MI School Data. (2018). Student counts: Non-resident status [Data file]. Retrieved from https://www.mischooldata.org/DistrictSchoolProfiles2/StudentInformation/NonResidentStatus.aspx

25.

Mich. Comp. Laws § 380.705, The Revised School Code (Excerpt) Act 451 of 1976. Retrieved from http://www.legislature.mi.gov/(S(gs3auns1zejgabo33u3ss050))/mileg.aspx?page=GetObject&objectname=mcl-380-705

26.

Mich. Comp. Laws § 388.1620, The State School Aid Act of 1979 (Excerpt) Act 94 of 1979a. Retrieved from http://www.legislature.mi.gov/(S(xeiljrj4yerki4abisloil01))/mileg.aspx?page=GetObject&objectname=mcl-388-1620

27.

Mich. Comp. Laws § 388.1705, The State School Aid Act of 1979 (Excerpt) Act 94 of 1979b. Retrieved from http://www.legislature.mi.gov/(S(xrhux3qdex3ofrdrulozdjhd))/mileg.aspx?page=GetObject&objectname=mcl-388-1705

28.

Mich. Comp. Laws § 388.1705c, The State School Aid Act of 1979 (Excerpt) Act 94 of 1979c. Retrieved from http://www.legislature.mi.gov/(S(encfo00a34s33t3excienkha))/mileg.aspx?page=GetObject&objectname=mcl-388-1705c

29.

National Alliance for Public Charter Schools. (2018). National charter school facts. Retrieved from https://data.publiccharters.org/

30.

Piele

P. K.

Hall

J. S.

(1973). Budgets, bonds, and ballots: Voting behavior in school financial elections. Lexington, MA: Lexington Books.

31.

Plummer

(2006). The effects of state funding on property tax rates and school construction. Economics of Education Review, 25, 532-542.

32.

Pogodzinski

Lenhoff

S. W.

Addonizio

M. F.

(2018). The push and pull of open enrolment policy in metro Detroit. Educational Review, 70, 622-642.

33.

Renzulli

L. A.

Evans

(2005). School choice, charter schools, and White flight. Social Problems, 52, 398-418.

34.

Shober

A. F.

(2011). Attracting capital: Magnets, charters, and school referendum success. Journal of School Choice, 5, 205-223.

35.

Sielke

C. C.

(1998). Michigan school facilities, equity issues, and voter response to bond issues following finance reform. Journal of Education Finance, 23, 309-322.

36.

Stewart

(2017, July 31). What the “government schools” critics really mean. The New York Times. Retrieved from https://www.nytimes.com/2017/07/31/opinion/donald-trump-school-choice-criticism.html

37.

Tedin

K. L.

Matland

R. E.

Weiher

G. R.

(2001). Age, race, self-interest, and financing public schools through referenda. Journal of Politics, 63, 270-304.

38.

Walcott

(2016). A refresher on Proposal A and local property taxes. Retrieved from http://msue.anr.msu.edu/news/a_refresher_on_proposal_a_and_local_property_taxes

39.

Wilkinson

Dawsey

C. P.

(2016, September 13). School choice, metro Detroit’s new White flight. Bridge Magazine. Retrieved from http://www.bridgemi.com/detroit-journalism-cooperative/school-choice-metro-detroits-new-white-flight

40.

Zimmer

Buddin

Jones

Liu

(2011). What types of school capital projects are voters willing to support? Public Budgeting and Finance, 31, 37-55.

41.

Zimmer

Jones

J. T.

(2005). Unintended consequence of centralized public school funding in Michigan education. Southern Economic Journal, 71, 534-544.