Analyzing electoral timing and outcomes through Georgia's special purpose local option sales taxes.

Author:Collins, Courtney A.

    Georgia was one of the first states to grant local authorities the ability to raise revenue through a local option sales tax (LOST), when in 1975 the state legislature passed enabling legislation. Much of this was the result of greater agitation and sentiments across both the state and country from citizens against what were perceived to be increasingly burdensome property tax rates. This new form of taxation was meant to help alleviate this mood by acting as a substitute for property tax revenue collected by county governments throughout Georgia.

    The success of the LOST in Georgia led the state to pass further legislation allowing for a special purpose local option sales tax (SPLOST) in 1985, followed by an educational special purpose local option sales tax (ESPLOST) in 1996. In fact, Georgia was the first state to adopt this latter form of local finance. Generally, these sales taxes are meant to be used to fund various capital outlay projects within a respective county (in the case of a SPLOST), or for various school district purposes (in the case of the ESPLOST). These methods of public finance have proven highly successful throughout the state, with almost every county having adopted either a SPLOST or ESPLOST at least once since its inception (ACCG 2013; Brunner and Warner 2012).

    Although (E)SPLOSTs (2) have generally proven successful when brought before voters, one of the most recent large-scale SPLOST initiatives, the transportation special purpose local option sales tax (TSPLOST), ended in almost complete defeat. This TSPLOST was meant to raise $18 billion over ten years in order to improve road congestion and other infrastructure problems across Georgia (The Economist Aug. 4th, 2012). If approved, three quarters of the revenue would have been devoted to projects statewide, while the remaining quarter were to be distributed to counties to be used as they so desired on their own infrastructure and road projects. This referendum went to a vote on July 31, 2012 and was placed on the ballot in conjunction with the state's primary election. Interestingly, 106 of Georgia's 159 counties voted down the measure, and did so by a fairly significant margin.

    Overall, this example brings to light an extremely relevant question: to what extent does the timing of a SPLOST or ESPLOST measure placed on a ballot impact the potential for such a measure to pass or fail? In order to address this question, we have compiled data on all (E)SPLOST referenda throughout each of the counties in the state of Georgia between 1985 and 2009 in order to understand what, if any, impact the timing of these elections may have on approval rates, the probability that an (E)SPLOST passes, as well as voter turnout.

    As will be discussed, both SPLOSTs and ESPLOSTs must follow the same procedural formula before appearing on a ballot; however the SPLOST may be brought up at the request of a county commission, while an ESPLOST may be brought up at the request of a county school board (ACCG 2013). Further, both can be placed either on a stand-alone special election (again called for by either the county commission or school board), or during a general primary or other general election. Given this, there exists an interesting opportunity for both county commissions and school boards to attempt to agenda set when an actual vote will be held regarding either a SPLOST or ESPLOST in a way that may lead to more favorable outcomes, and increase the probability that a measure may pass.

    Our current study attempts to explore this possibility. Specifically, we evaluate how the timing of an election impacts the passage rate of both SPLOSTs and ESPLOSTs within the counties throughout the state of Georgia. We do this by analyzing these referenda in order to determine whether those referenda considered during a special election were more likely to pass compared to those considered during a general primary or general election. Overall, we find strong evidence to suggest that (E)SPLOSTs that were placed on a special election ballot were significantly more likely to pass, had much higher voter approval rates, and also saw much lower voter turnout in general.

    Further, we find that these results in general hold for primary elections as well (when compared to general elections), though the impact of a primary election is much smaller than a special election. These results also hold if the data is cut to consider only a SPLOST or ESPLOST election; however the results are somewhat less robust for ESPLOST measures. Finally, and most interesting, it appears that both SPLOST and ESPLOST referenda held in conjunction with either a presidential or gubernatorial primary or general election had an even more profound impact on approval rates (significantly decreasing them), the probability of passage (again significantly decreasing it), and voter turnout (significantly increasing it). In general, we find much larger effects for presidential general or primary elections, compared to gubernatorial elections. This is one of the first studies to systematically evaluate the impact that ballot placement has on overall (E)SPLOST outcomes. (3)

    Overall, this work adds to several important strands of literature. The first is the importance of agenda control on the timing of elections. The seminal work of Niskanen (1971) and Romer and Rosenthal (1978) among others have provided strong theoretical foundations for examining how situations in which certain actors given the ability to bring a measure up for a vote at their discretion may result in an outcome contrary to the wishes of the median voter. More recently, a number of studies have empirically investigated this connection between the ability of certain political actors to determine when a given election may be held, and how that impacts the composition of the electorate and the probability of a certain outcome occurring (Meredith 2009; Berry and Gersen 2010, 2011).

    Further, following the work of Niskanen (1971) Georgia's (E)SPLOST only allows for the tax to increase in increments of 1%. Thus, given this situation it may be possible for elected officials to place the electorate on an all-or-nothing demand curve. In so doing, there is greater potential for the agenda setting public agent to obtain its most desired outcome, and to obtain an outcome beyond what is most preferred and optimally desired by the median voter. This is further compounded by the fact that all (E)SPLOST referenda either pass or fail in an up or down vote, meaning all proposed projects and expenditures either pass or fail as a single package (ACCG 2013).

    Various studies have considered how the outcomes of local school bond measures and other referenda have been impacted based on when the election is actually held and how this impacts the composition of the median voter (Anzia 2011; Dunne, Reed and Wilbanks 1997; Pecquet, Coates and Yen 1996; Rubinfeld and Thomas 1980; Rubinfeld 1977). It is within this context that we attempt to couch our current work as both Georgia county commissions and county school boards have the exclusive authority to determine when a SPLOST or ESPLOST will be placed on a ballot referendum to then be voted on by the general public. Thus, this potential may significantly influence when and how an (E)SPLOST passes, and potentially the extent to which this alters the composition of the median voter.

    Further, this work also adds to a large body of literature which has developed to explore both SPLOSTs and ESPLOSTs generally, and also within Georgia. Several studies have analyzed the various determinants as to why a municipal or county government might adopt a local option sales tax (Burge and Piper 2012; Pajari 1984; Sjoquist, Smith and Walker 2007; Sanders and Lee 2009; Zhao 2005). In relation to the adoption of such a tax in Georgia, Zhao (2005) finds that relatively higher property tax burdens along with the increased ability of a county to tax export led to a greater propensity of a given county to adopt a local option tax. In assessing local option sales taxes in Oklahoma, Burge and Piper (2012) found that both horizontal and vertical spillovers play an important role in their adoption. Finally, numerous studies have examined the fiscal disparities that may or have emerged as a result of the implementation of an (E)SPLOST (Brunner and Warner 2012; Rubinstein and Freeman 2003; Zhao and Han 2008). Broadly, these works do find evidence that (E)SPLOSTs may in fact lead to greater inequalities across a wide spectrum of both fiscal and socioeconomic variables and outcomes.

    The remainder of the paper is structured as follows. Section 2 provides a brief history and description of Georgia's SPLOST and ESPLOST along with the procedural means and methods by which they may be brought up and voted on by the general electorate. Section 3 provides a description of the data as well as the models to be employed in order to test the impact that placement of these measures during different elections may have on the overall probability that the given measure passes. Section 4 presents the results and interpretation of those results, and section 5 concludes.


    Special purpose local option sales taxes have increasingly become a popular means by which state governments and legislatures have granted authority for junior level jurisdictions to raise revenue and taxes through various means other than just a property tax (Burge and Piper 2012). The state of Georgia has been no exception to this fact, and has in many ways been a leader in the development of this method of public finance. Spurred by the "tax revolts" of the 1970s that swept through many states, and also a desire to maintain relatively low property tax rates, the state legislature passed legislation in 1985 that allowed a county government, after majority approval of a county's citizens to implement a 1%...

To continue reading