Explaining the appearance and success of voter referenda for open-space conservation

https://doi.org/10.1016/j.jeem.2006.02.003Get rights and content

Abstract

This paper provides an empirical investigation of the factors that influence the appearance and success of voter referenda to raise public funds for open-space conservation. We take advantage of a data set that includes detailed information on all such referenda that occurred in the United States between 1998 and 2003. Combining these data with information from the U.S. Census and state-specific variables, we conduct a nationwide analysis along with focused studies of referenda that occurred in New Jersey and Massachusetts. The paper provides the first investigation of how funding mechanisms and funding rates affect voter support for public acquisition of open space. We also provide evidence on the relationship between existing patterns of open space and voter support for open-space referenda. As open-space initiatives continue to gain popularity at the ballot box, the descriptive insights of this paper should prove useful for both policy-makers and advocates working in the area of land use management.

Introduction

The protection of open space from the advance of “urban sprawl” has emerged as one of the more pressing environmental issues in the United States. Open space is generally understood to be a public good that will be under-provided without policy interventions. Policy-makers have begun efforts to protect open space using various instruments—including zoning regulations, development taxes, urban growth boundaries, conservation easements, and public acquisition of undeveloped land. Increasingly, citizens are also becoming directly involved in open-space conservation through ballot initiatives designed to implement mechanisms for public land acquisition. Nearly 1,000 jurisdictions at the state, county, and local levels held open-space referenda between 1998 and 2003, and approximately 80% of these initiatives passed, raising over $21 billion for open-space conservation.1

The proliferation and high success rate of open-space ballot initiatives raise several economic and policy-relevant questions. What factors contribute to the appearance of an open-space referendum in a jurisdiction? How does an initiative's funding mechanism—such as a bond, property tax, sales tax, or income tax—affect the way citizens vote? How responsive are favorable votes to the costs of an open-space initiative? How do socioeconomic characteristics influence voting results? What is the effect of existing patterns of land use? And what other features of a referendum affect voting outcomes?

These questions motivate our analysis in this paper. We construct a data set of open-space referenda that occurred in the United States between 1998 and 2003. Detailed information on each referendum comes from annual reports, titled LandVote, that are published by the Trust for Public Land (TPL) and the Land Trust Alliance (LTA).2 These data include each referendum's political jurisdiction, proportion voting for and against, financing mechanism, financing rate, land characteristics, and other policy-relevant variables. For each jurisdiction we also collect data from the U.S. Census on socioeconomic characteristics. Then, using the combined data, we estimate econometric models to determine the impact of referendum characteristics and socioeconomic variables on voting results.

In addition to the nationwide analysis, we conduct two focused studies of referenda that occurred in New Jersey and Massachusetts. Statewide policies were passed in both states to provide incentives for local jurisdictions to raise taxes for open-space conservation. The result has been numerous referenda in both states: 237 in New Jersey and 137 in Massachusetts between 1998 and 2003. For both states we collect further Census data on all jurisdictions that did not hold a referendum. Additionally, we collect specific data on the amount of open space, recent rates of open-space loss, average property tax burdens, and state-specific features of the referenda. Taking advantage of all these data, we then estimate models for each state in order to determine: (1) what factors influence whether a jurisdiction has held an open-space referendum, and (2) what factors explain the success of a referendum in terms of voting results.

Other researchers have investigated related questions. In a pioneering study of referenda results, Deacon and Shapiro [4] analyze voting outcomes for a law in California to protect coastal zones from development. They find some evidence that the natural coastal environment is a normal good, but the effect is not statistically significant. Kahn and Matsusaka [11] also analyze statewide referenda in California. Three of the referenda they study were to authorize bond issues to purchase park, forest, and wildlife areas. They find evidence that collectively provided open space is a normal good, except when income is very high, in which case it becomes inferior. They also find that people are more likely to vote yes in more urban counties. Another study by Kline and Wichelns [12] uses statewide referenda in Pennsylvania and Rhode Island to investigate demand for the purchase of farmland development rights. They find that the proportion of yes votes increases with a town's population growth, home value appreciation, farmland loss, urbanization, and prevalence of resource sensitive lands.

Because the aforementioned studies use local voting results in statewide referenda, they cannot address the question of what factors contribute to the appearance of an open-space referendum in the first place. Howell–Moroney [9] considers this question in a study of municipalities throughout the Delaware Valley region. He finds that the appearance of a referendum is responsive to patterns of land use, whereby low population density and loss of open space increase the probability of a referendum occurring. He also finds that higher population and median household income increases the probability of a referendum.3

Our paper makes four primary contributions to the literature. First, we construct the most comprehensive data set on open-space referenda to date. Second, we take advantage of variation in the financing mechanism across referenda (e.g., bonds or taxes) in order to investigate whether the type of mechanism proposed affects voter support for open-space acquisition. Third, we exploit variation in the funding rates within the different mechanisms (e.g., bond amounts and tax rates) to determine how responsive voters are to the costs of an open-space initiative. Fourth, we conduct detailed analyses of two states in order to determine the factors that influence the appearance of a referendum, in addition to the factors that influence a referendum's success.4

The results provide new insights into citizen preferences for open space and the relationship between characteristics of an open-space policy and voter support. We find strong evidence that voters are more likely to approve bonds than tax increases. Not surprisingly, funding rates also matter. Higher rates generally decrease the odds of a yes vote, but interestingly, the opposite result emerges at the state and county levels. In general, we find that the factors influencing referenda outcomes differ between the state–county level and the local level.

In New Jersey and Massachusetts, we find evidence that jurisdictions holding open-space referenda differ significantly from those that do not. Referenda tend to occur in wealthier communities that have experienced greater population growth. While jurisdictions with more open space and recent open-space loss are more likely to have held a referendum, these same variables have different effects on referenda success. In particular, we find that, over certain ranges, more open-space loss in the years prior to a referendum actually reduces voter support. We also find further evidence that collectively provided open space is a normal good. Other findings relate to the importance of farmland as a type of open space, to specific features of the proposed open-space policies, and to the effect of existing tax burdens.5

The remainder of the paper is organized as follows. Section 2 describes the data used in the analysis. Section 3 provides details on the econometric specifications. Section 4 reports the results of the nationwide analysis along with the results of the New Jersey and Massachusetts studies. Section 5 concludes with a summary of the main results.

Section snippets

Data

We collected data on open-space referenda from the annual LandVote survey published by the TPL and the LTA. The LandVote survey attempts to provide a comprehensive listing of all open-space referenda that involve the direct acquisition of undeveloped land.6 Using the information contained in the LandVote

Econometric specifications

We estimate regression models to explain the election outcomes of open-space referenda. The dependent variable in our models islogoddsi=lnPi1-Pi,where Pi is the proportion of yes votes out of the total number of votes cast in referendum i. This variable is the log-odds ratio, and it is commonly used in econometric models of aggregate voting results [4], [5], [11], [12], [18], [20].

The equations that we estimate for the nationwide data have the general form:logoddsi=β1Mechi+β2Ratei+β3Extendi+β4

Estimation and results

We report the econometric results in this section. Those for the nationwide analysis are reported first, followed by those for New Jersey and Massachusetts.

Conclusion

The purpose of this paper is to provide an empirical investigation of the factors that influence the appearance and success of voter referenda for open-space conservation. We take advantage of a data set that includes detailed information on all such referenda that occurred in the United States between 1998 and 2003. Combining these data with information from the U.S. Census and state-specific variables, we conduct a nationwide analysis along with focused analyses in New Jersey and

Acknowledgments

This paper was written while Kotchen was an assistant professor at Williams College, and Powers was an undergraduate major in economics; we thank the Department of Economics at Williams for its support. We are grateful for helpful comments from Jon Bakija, Roger Bolton, Bill Fischel, Matt Kahn, Nat Keohane, Erin Mansur, Steve Sheppard, an anonymous referee, and seminar participants at Minnesota, UNH, Yale, and Williams. We are especially grateful to Stacey Schulte for helping to spark the idea

References (22)

  • R. Halvorsen et al.

    The interpretation of dummy variables in semilogarithmic equations

    Amer. Econ. Rev.

    (1980)
  • Cited by (94)

    • Aggregate data yield biased estimates of voter preferences

      2022, Journal of Environmental Economics and Management
      Citation Excerpt :

      From rigis.org, we downloaded a 2016 precinct shapefile, which we then overlaid with the block group shapefile to calculate area weights, which were then used to calculate the approximate socioeconomic mix for each precinct. This construction replicates the method of other research using aggregate data (e.g., Kotchen and Powers 2006; Banzhaf et al., 2010), though many papers use units of analysis larger than precincts. We designed an exit poll survey to elicit votes for GEB and president and several socioeconomic characteristics (age, gender, race/ethnicity, income, education, and homeowner status).9

    • Gaining voter support for watershed protection

      2019, Land Use Policy
      Citation Excerpt :

      Conservation minded individuals in some communities may still want to exchange some of their current environmental quality for other forms of income, such as higher paying jobs (Nelson et al., 2007; Wu and Cutter, 2011; Hochschild, 2018). Voters may also have preferences for which funding rate or mechanism (i.e., a bond or tax) is used, based on perceptions of how costs will likely be allocated to taxpayers (Kotchen and Powers, 2006; Nelson et al., 2007). A large body of literature reveals that voters will consistently rely on cues from trusted sources and heuristics to reach reasoned decisions under stiff information requirements. (

    • The individual determinants of support for open space bond referendums

      2019, Land Use Policy
      Citation Excerpt :

      These studies typically find that people’s perceptions of the consequences of development add explanatory power to their support for open space and growth management policies above and beyond their socioeconomic characteristics. However, these studies rely on surveys of specific locales (e.g. Smutny, 1998; Gainsborough, 2002; Gerber and Philips, 2005; Mohamed, 2008) or states or metropolitan areas that have held referenda or implemented “smart growth” policies (e.g. Kline and Wichelns, 1994; Howell‐Moroney, 2004; Solecki et al., 2004; Kline, 2006; Kotchen and Powers, 2006; Coan and Holman, 2008; Schmidt and Paulsen, 2009), so it is unclear how generalizable their findings are to individuals across states and communities. We add to the literature by examining how individual level characteristics, including perceptions, opinions, and demographics, influence open space bond support on a nationally representative sample of respondents.

    • Voter support for environmental bond referenda

      2018, Land Use Policy
      Citation Excerpt :

      Undeveloped land in an area may also influence voter support for environmental bonds. The unequal distribution of open space areas across a city may make them more or less accessible and thus provide different environmental benefits to some residents (Kotchen and Powers, 2006; Solecki et al., 2004). Similarly, the distribution of recreation opportunities across the city may also influence voter support.

    View all citing articles on Scopus
    View full text