Public Finance Review - CiteSeerX

1 downloads 0 Views 293KB Size Report
fiscal behavior about how local option sales taxes are adopted remains virtually ...... The special 1% sales tax referendum: Tips and techniques on municipal.
Public Finance Review http://pfr.sagepub.com

Motivations, Obstacles, and Resources: The Adoption Of The General-Purpose Local Option Sales Tax in Georgia Counties Zhirong Zhao Public Finance Review 2005; 33; 721 DOI: 10.1177/1091142105279555

The online version of this article can be found at: http://pfr.sagepub.com/cgi/content/abstract/33/6/721

Published by: http://www.sagepublications.com

Additional services and information for Public Finance Review can be found at: Email Alerts: http://pfr.sagepub.com/cgi/alerts Subscriptions: http://pfr.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

PUBLIC 10.1177/1091142105279555 Zhao FINANCE / MOTIVATIONS, REVIEW OBSTACLES, AND RESOURCES

MOTIVATIONS, OBSTACLES, AND RESOURCES: THE ADOPTION OF THE GENERALPURPOSE LOCAL OPTION SALES TAX IN GEORGIA COUNTIES ZHIRONG ZHAO Eastern Michigan University

Local governments in Georgia have been authorized since the 1970s to levy a 1 percent general-purpose Local Option Sales Tax (LOST), which is earmarked for property tax relief. Using data during 1975 to 2002, this study examines the adoption of the LOST through a discrete-time event history analysis. The dependent variable is the probability that an eligible county will adopt it in a particular year. This probability is negatively related to the obstacles prohibiting the innovation and positively related to (1) the motivations to innovate and (2) the resources for overcoming the obstacles. The findings suggest that the motivations are higher in counties with higher property tax millage rates and the potential of sales tax exportation; the obstacles include high existing sales tax rates and severe tax competition; and the major resource for overcoming these obstacles is the adoption of the LOST in other Georgia counties. Keywords: local option sales tax; property tax relief; policy innovation; policy diffusion; tax mimicking

1. INTRODUCTION

The property tax is still the single most important own-source revenue for local governments in the United States. In recent decades, however, the reliance on the property tax has dramatically declined. This decline has been brought about by several changes over the past AUTHOR’S NOTE: This article is based on my doctoral dissertation, “Property Tax Relief, Additional Revenue, or Tax Mimicking? The Adoption and Budgetary Effects of the GeneralPurpose Local Option Sales Tax in Georgia Counties” (Ph.D. diss., University of Georgia, Athens, 2005). I am indebted especially to my thesis advisor, Laurence J. O’Toole, for his invaluable guidance in my doctoral studies. PUBLIC FINANCE REVIEW, Vol. 33 No. 6, November 2005 721-746 DOI: 10.1177/1091142105279555 © 2005 Sage Publications

721

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

722

PUBLIC FINANCE REVIEW

twenty-five years, most important of which is the “property tax revolt.” Since the late 1970s and early 1980s, there has been a sustained resistance by taxpayers to new taxes or rate increases. For the most part, the “tax revolt” was targeted toward the unpopular property tax (Shuford and Young 2000).1 In 1978, for example, California voters enacted Proposition 13, which abruptly reduced local property tax revenues in the state by half; in 1980, Massachusetts voters approved Proposition 2-1/2, which set an absolute limit on the property tax rate and the annual increase of tax levy. Soon after the adoptions of Proposition 13 and Proposition 2-1/2, many other states also enacted policies designed to reduce the property tax burden (Gold 1979). Since the property tax revolt, local governments have struggled to find alternative sources of revenue to fund their services. With the property tax and expenditure limits, local governments increase their reliance on other taxes, user charges and fees, and state aids (Temple 1996; Shadbegian 1999; Skidmore 1999). In recent years, local option taxes (in particular local option sales taxes) have become the major alternative to the property tax. In 1998, thirty-two states allowed one or more types of local governments to levy local option sales taxes.2 These taxes normally have two features. First, they are optional, that is, local residents have a choice to decide whether to adopt these taxes. Second, they are usually earmarked for some specific purposes, such as education, transportation, capital improvement, or property tax relief.3 The focus of this study is on local option sales taxes that are earmarked for property tax relief. Since the 1970s, local governments in Georgia have been authorized, upon voter approval, to adopt a general-purpose Local Option Sales Tax (LOST), with the requirement that the proceeds to be used for property tax relief. Similar programs have been enacted in other states as well. In Wisconsin, as is the case in Georgia, full amount of LOST proceeds should be used to roll back property tax of the second year (Wisconsin Taxpayers Alliance 2002). In South Carolina, 63 percent of the first-year LOST revenues should be used for property tax rollback, and the portion increases annually up to 71 percent for the fifth and subsequent years (Ulbrich 1996). In Iowa, most counties earmark a specific portion of their LOST proceeds for property tax relief, although that is not required by

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

723

the state.4 Despite the increasing popularity of these programs, local fiscal behavior about how local option sales taxes are adopted remains virtually unexamined. Using data in Georgia counties during 1975 to 2002, this study attempts to identify the factors that determine whether and when a county will adopt the LOST. According to Mohr’s (1969) theory of organizational innovation, these factors are related to a set of motivations, obstacles, and resources. Through a discrete-time event history analysis, the study finds that (1) the motivations to the adoption are related to property tax millage rates and the potential of sales tax exportation, (2) the obstacles to innovate are related to existing sales tax rates and the level of tax competition, and (3) the major resource for overcoming these obstacles is related to tax mimicking behavior at the local level. Understanding these factors provides insight into the nature of this new source of revenue and enhances our knowledge about local government fiscal behavior. The article is organized as follows. The next section briefly describes the LOST in Georgia. Section 3 reviews related literature on state and local tax structure as well as policy innovation and diffusion. Sections 4 and 5 discuss hypotheses and research methods, while section 6 presents research findings. The final section draws conclusions and addresses theoretical and policy implications of this study.

2. THE GENERAL-PURPOSE LOST IN GEORGIA

The general-purpose LOST Act was first passed in 1975 and amended in 1976. It permitted counties, upon public approval, to enact a 1 percent sales and use tax, which was to be shared among the county and cities within the county.5 The LOST Act mandated that the first-year LOST revenues be used to roll back property taxes in the second year. The county government should use its share of LOST revenues to reduce property taxes in unincorporated parts of the county, while city governments should use their share to replace property taxes in the city. In addition, it is required that the property tax rollback financed by LOST revenues be shown on property tax bills. In the second year of the adoption and all the subsequent years, coun-

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

724

PUBLIC FINANCE REVIEW

ties are required to continuously use their LOST revenues to relieve property tax in unincorporated areas. 6 On February 8, 1979, the LOST Act was determined unconstitutional, as the Supreme Court ruled that nothing in the Georgia Constitution authorized counties to share their revenue with cities (Association County Commissioners of Georgia 2002). Subsequent to the Supreme Court’s decision, the General Assembly passed a new legislation, the LOST Act of 1979. The law set up 159 special tax districts that correspond to county boundaries, and it established the LOST as a joint county-city tax. The proceeds should be divided among the county and all “qualified” cities within the county pursuant to an agreement negotiated by the county and those cities. 7 Like the 1975 LOST Act, the 1979 legislation required that all LOST revenues collected in the first year of the tax be used for dollarfor-dollar reduction of property taxes of the second year, and it required the amount of property tax relief to be shown on property tax bills. Somewhat differently, counties should use their LOST revenues to reduce property taxes across the board on all taxable properties in the county, including properties within the incorporated areas (Clements and Weeks 1997). The 1979 legislation, nevertheless, was less explicit about the use of LOST revenues after the second year (Durning 1992). Since the inception of the tax in 1975, the number of counties levying the LOST has gradually increased. As Figure 1 shows, a number of counties adopted the tax soon after 1975, but the rate of adoption gradually leveled off. As of January 2004, all but ten counties in Georgia have adopted the LOST.8 More than half of them adopted the tax within the first five years since the LOST was enacted (Georgia Department of Revenue 2004).

3. LITERATURE REVIEW 3.1. DETERMINANTS OF STATE AND LOCAL TAX STRUCTURE

Hettich and Winer (1988) argued that tax structure is the result of a rational process through which governments balance the political

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

725

Number of counties with the LOST

160 140 120 100 80 60 40 20 0 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002

Year

Figure 1: The Adoption of the Local Option Sales Tax (LOST) in Georgia Counties

costs of generating revenues against the political gains from making increased expenditures. In particular, whether to adopt the LOST in a Georgia county depends on voters’ decision in balancing the costs of increasing sales tax rate against the gains from shifting tax burden or increasing public services. This decision may be influenced by many factors, which have been explored in mainly three approaches: the Dye-Sharkansky-Hofferbert (DSH) comparative studies, the median voter approach, and the multijurisdictional approach. The DSH comparative studies seek to explain variations in policy outcomes across large number of states and localities (Dye 1966; Sharkansky 1970; Hofferbert 1974). These studies have identified many factors that influence revenue policies. Community wealth is found to be a primary determinant of public policy decisions (Dye 1967), in particular taxation and expenditures (Maxwell 1965). Demographic factors, including population density, the percentage of blacks, and the level of education attainment also matter (Sharkansky 1970; Chelf 1984; Nice 1987). In addition, political variables such as party ideology affect revenue policies as well (Elling 1979; Boyne 1998).9 The median voter approach claims that fiscal decisions by local governments reflect median voters’ preferences. In a basic model, each individual taxpayer balances the benefits obtained from a higher level of public services against the costs of higher taxes to choose an optimal “service package.” As a result, public policies in a democratic

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

726

PUBLIC FINANCE REVIEW

system would be determined by the choice of the median voter (Black et al. 1998).10 In a simplified version, the preferences of the public are usually specified in terms of median or average characteristics of the population (Dye 1990).11 A majority of these people prefer, among other things, to shift their tax burdens to others, especially nonresidents or those living outside their jurisdictions (Norstrand 1980; Chicoine and Walzer 1986; Blackley and DeBoer 1987). The multijurisdictional approach holds that the competition among local governments also affect governments’ budgetary decisions (Schneider 1989; Dye 1990). Variance in service provision and tax levy among local governments provides suburban residents with a choice of residency (Tiebout 1956). If residents are not satisfied with the “package” of services and taxes, they can either move to other communities (“vote with their feet”) or voice their dissatisfaction, impose pressure for improvement, and punish elected officials at the polls (Santerre 1986). Consequently, local officials take into account the tax burdens in other jurisdictions when they make tax decisions, in order not to lose tax base or political support (Schneider 1989; Ladd 1992; Case, Rosen, and Hines 1993; Besley and Case 1995; Heyndels and Vuchelen 1998). 3.2. POLICY INNOVATION AND DIFFUSION

Policy innovation and diffusion have been extensively studied for several decades. In general, two principal forms of explanation have been offered: internal determinants models and policy diffusion models (for a survey, see Berry and Berry 1999). Internal determinants models assume that internal characteristics of a state or locality determine whether and when an adoption will occur. Some earlier studies seek to use these models to explain a general innovativeness across multiple issue areas (Walker 1969; Savage 1978). Walker (1969), for instance, created an index that reflects the earliness of adoption for a set of eighty-eight economic and social policies. The index was found to be associated with the presence of slack resources and the level of industrialization and urbanization, and so on. However, Gray (1973) challenged the assumption of the general innovativeness and claimed that the patterns of innovation are issue-

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

727

specific and time-specific. In this study, the adoption of the LOST is assumed to be influenced more by the determinants of tax structure (as discussed above) than by the variables commonly tested for the general innovativeness.12 In contrast to internal determinants models, policy diffusion models focus on “the process by which an innovation is communicated through certain channels over time among the members of a social system” (Rogers 1983, 5). Assuming policy makers in the states emulate each other, two types of diffusion models are most commonly proposed: the “national interaction” model and “neighboring diffusion” models. The national interaction model assumes a national communication network among state officials, in which officials learn about programs from their peers in other states (Gray 1973). Alternatively, neighboring diffusion models posit that states are influenced primarily by their “neighbors,” and the “neighbors” are those states that are in geographical proximity (Berry and Berry 1990), within the same region (Mooney and Lee 1995), or considered economically or demographically similar (Case, Rosen, and Hines 1993). Berry and Berry (1990) argued that pure internal determinants and policy diffusion models provide incomplete or inaccurate explanations for policy innovations. If studies of policy innovations overlook the influence by neighboring states, the models are not properly specified. If studies of policy diffusion fail to control for internal determinants, the validity is undermined by the threat of spuriousness, because states of geographical proximity usually have many other characteristics in common. To solve these problems, the Berrys (1990) proposed an integrated model based on Mohr’s (1969) theory of organizational innovation, and they tested the model through a discrete-time event history analysis in a study of state lottery adoptions. This method is applied here to analyze the determinants of the LOST adoption in Georgia counties.

4. RESEARCH HYPOTHESES

As of 2004, all but ten counties in Georgia have adopted the LOST. A county’s decision to adopt the tax can be attributed to both internal

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

728

PUBLIC FINANCE REVIEW

and external forces. On one hand, the adoption may be determined by the counties’ socioeconomic or political conditions. On the other hand, some counties may become more prone to adopt the LOST because of the influence they receive from other counties. The unit of analysis in this study is each county-year. The dependent variable is conceptualized as the conditional probability that a county without the LOST will adopt it in a particular year. It is measured by a dummy variable that is scored one when a county levies the LOST and zero otherwise. The internal and external determinants are tested simultaneously in a model based on Mohr (1969), who posited that the probability of innovation is negatively related to the strength of the obstacles to innovate and is directly related to (1) the motivations to innovate and (2) the availability of resources for overcoming these obstacles. The motivations for a county to adopt the LOST take several forms. First, as the law requires the LOST to be used for property tax relief, it is hypothesized that counties with higher property tax level are more likely to adopt the LOST. The property tax level is measured by the property tax millage rate and the ratio of property tax to personal income. The ratio of property tax to personal income is the “effective” property tax level relative to the ability to pay. How sensitive local governments and taxpayers are about this measure, nevertheless, remains an empirical question. In contrast, the property tax millage rate, while not an accurate measure of property tax level unless property assessment is controlled, represents the “perceived” level of property tax burden—being highly visible, millage rates are often used for time-series and cross-jurisdictional comparisons in local governments’ budgetary files. Using both “effective” and “perceived” property tax levels in the same model, the study attempts to distinguish whether and how the public responds differently to these two measures.13 Hypothesis 1a: The probability that a county will adopt the LOST is positively related to its property tax millage rate. Hypothesis 1b: The probability that a county will adopt the LOST is higher if the ratio of property tax to personal income is higher.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

729

Second, although the LOST is earmarked for property tax relief, there is an ongoing debate as to whether LOST proceeds have been used in part as additional revenues (Durning 1992; Jung 2001). In this regard, counties facing fiscal pressure may have a higher motivation to adopt the LOST and use the proceeds to supplement their property tax revenues (Alm, McKee, and Skidmore 1993; Furlong 1998). The level of fiscal pressure is measured in two ways. One is the percentage of public roads that are unpaved, which reflects the lack of resources for a county to fulfill one of its primary functions.14 The other is the annual percentage change of real property and utility digest (per capita). This variable reflects the changes in the size of the tax base. A county with a decreased tax base may have difficulties to raise sufficient revenues to fund public services and thus may have a stronger incentive to raise new taxes. Hypothesis 2a: The probability that a county will adopt the LOST is higher if the county has a higher percentage of unpaved public roads. Hypothesis 2b: The probability that a county will adopt the LOST is higher if the county has a lower annual increase of real property and utility digest.

Third, it is expected that counties in a better position to export their sales taxes are more likely to adopt the LOST. The potential of tax exportation is measured in two ways. One is the presence of an interstate highway in a county, because regional shopping centers tend to be located along interstate highways (Pajari 1984). The other is the ratio of taxable sales base to personal income. If a county’s taxable sales base is disproportionally higher than its personal income, it is more likely that the sales base is also contributed by residents from other jurisdictions. Hypothesis 3a: The probability that a county will adopt the LOST is higher if the county contains a portion of interstate highway (see Figure 2). Hypothesis 3b: The probability that a county will adopt the LOST is positively related to the ratio of the taxable sales base to personal income.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

730

PUBLIC FINANCE REVIEW

Figure 2: The Adoption of the Local Option Sales Tax (LOST) and Interstate Highways in Georgia

There are several obstacles for the LOST adoption. First, a county is less likely to adopt the LOST if the existing sales tax rate is already higher in the county than in other places. The rate of state sales tax in Georgia is 4 percent, and local governments can elect to adopt several local option sales taxes.15 Therefore, existing sales tax rates vary across the counties that have not adopted the general-purpose LOST. Second, the “economic cost” (the risk of losing tax base) to levy the LOST in a county may be higher in high-density metropolitan areas where there is a higher degree of spatial competition between market centers (Krmenec 1991). Finally, it may be more difficult to adopt the LOST in a county with a higher percentage of voters who believe in Republican ideology, because these voters may be against any kind of new taxes (Nice 1987; Poterba 1995). Hypothesis 4a: The probability that a county will adopt the LOST is negatively related to the existing rate of sales tax in the county.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

731

Hypothesis 4b: The probability that a county will adopt the LOST is lower if the county is located within the Atlanta metropolitan statistical area (MSA).16 Hypothesis 4c: The probability that a county will adopt the LOST is lower if there is a larger percentage of voters voting for Republican presidential candidates.

The resources for overcoming these obstacles are associated with the tax mimicking behavior in the process of policy diffusion. First, the “political cost” to adopt the LOST—the resistance from taxpayers to a new tax—would be lowered if many other counties in the state have adopted the tax. Second, the “economic cost” to adopt the LOST—losing tax base because of tax competition—would decrease if many neighboring counties have adopted the tax (Ladd 1992). Finally, the “information cost” to adopt the LOST—lacking information and experience with a new source of revenue—would be reduced if many others have adopted the tax (Gray 1973; Berry and Berry 1990). Hypothesis 5a: The probability that a county will adopt the LOST is higher if many neighboring counties have adopted the tax. Hypothesis 5b: The probability that a county will adopt the LOST is higher if many other counties in Georgia have adopted the tax.

In addition, the study controls for two other variables. One is the annual percentage change of real per capita personal income that reflects the fluctuation of economic conditions. Its coefficient is expected to be positive, because voters in better economic conditions may be more tolerant to the increase of tax revenue. The other is the annual percentage change of population. Counties with rapid population growth generally have higher fiscal pressure for expanding public services, and thus the coefficient is expected to be positive (Bergstrom and Goodman 1973).

5. RESEARCH METHOD

The adoption of the LOST in Georgia counties is examined by a discrete-time event history analysis. Event history analysis is a method

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

732

PUBLIC FINANCE REVIEW

to explain a qualitative change (an “event”) that occurs in an individual or unit at a particular point of time. The data for analysis, called an “event history,” is a longitudinal record showing whether and when the event was experienced by a sample of individuals or units during a period of observation (Tuma and Hannan 1984). In this study, the units are Georgia counties, and the event history is whether or when the counties adopted the LOST in a particular year during the period 1975 to 2002. A critical concept in the event history analysis is the “risk set,” which is the set of units in the sample that are “at risk” of the event occurrence (i.e., have a chance of experiencing the event) at a particular time. When the event under analysis is one that a unit cannot repeat (e.g., death), the size of the risk set will decrease over time as units in the sample experience the event (Berry and Berry 1990). In this study, the risk set in each year consists of the counties that are eligible for the LOST but have not yet adopted it. The state law of Georgia limits the total rate of state and local sales taxes at seven percent (Georgia Department of Revenue 1996). Therefore, for counties without the LOST, only those in which the total sales tax rate is lower than 7 percent can elect to adopt the tax. The size of the risk set is generally decreased at the end of each year for two reasons. First, the number of eligible counties changes when the total state and local sales tax rates in some counties reaches or drops from the 7 percent cap. Second, the risk set downsizes when some eligible counties elect to adopt the LOST in a particular year. Accordingly, the data set for analysis consists of the risk sets of different sizes in each year during the study period. There are discrete-time models and duration models of event history analysis. In a discrete-time model, the dependent variable is called the “hazard rate” and defined as the probability Pi,t that a unit i will experience the event at a particular time t, given that the unit is at risk at that time. Alternatively, a duration model can be used to analyze the typical spell of time before an at-risk unit experiences the event (Tuma and Hannan 1984; Temple 1996). This study employs the discrete-time model, and the hazard rate to be analyzed is the conditional probability that an at-risk Georgia county will adopt the LOST in a particular year. It is presumed to be determined by a set of internal

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

733

and external independent variables. Nevertheless, the hazard rate, being a probability, cannot be observed. The observed dependent variable for estimation is a dummy variable that is scored one when an eligible county adopts the LOST and zero otherwise. Since the dependent variable is binary, ordinary least squares (OLS) method is inappropriate because several conventional regression assumptions are violated (Long 1997).17 Parameters will be instead estimated by a logistic model, which has been widely used in the context of binary outcomes (Hosmer and Lemeshow 1989; Menard 1995).

6. RESEARCH FINDINGS

The data for this study are collected from several sources. The Statistical Report, published by the Georgia Department of Revenue, provides information about the taxable sales base and property tax level in Georgia counties from 1970 to 2002. The primary source of socioeconomic data is the Georgia County Guide (Bachtel and Boatright 1982-2003). The guide contains annual demographic and economic data for individual Georgia counties. The election data come from the Office of the Secretary of State of Georgia. Definition, mean, and standard deviation of the independent variables are shown in Table 1. As previously mentioned, the motivations for adopting the LOST include a high property tax level, the existence of fiscal pressure, and the potential of sales tax exportation. The property tax level is measured in two ways, property tax millage rate (MILLAGE) and the ratio of property tax to personal income (PTINCM). The level of fiscal pressure is measured by the percentage of public roads that are unpaved (UNPAVED) and the annual percentage change of real property and utility digest (DGCHANGE). The potential of tax exportation is measured by whether a county contains a portion of interstate highway (HIGHWAY) and the ratio of taxable sales base to personal income (EXPORT). These “motivation” variables, except for DGCHANGE, are expected to have positive relations with the adoption. The coefficient of DGCHANGE is expected to be negative because a decrease of DGCHANGE is an indicator of fiscal pressure.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

734

PUBLIC FINANCE REVIEW

TABLE 1:

Variables Definitions, Means, and Standard Deviations (1975-2002)

Variable MILLAGE PTINCM

UNPAVED

DGCHANGE

HIGHWAY

EXPORT

NEIGHBOR

NLOST REPUBLICAN TAXRATE

ATLMSA

IMCHANGE POPCHANGE

Definition Total property tax millage rate levied in the unincorporated area of a county. The ratio of property tax (dollar) to personal income (thousand dollar), a measure of property tax burden. The percentage of local roads that are unpaved, a measure of fiscal pressure.a The annual percentage change in real property and utility digest, a measure of fiscal pressure. Dummy variable, coded 1 if interstate highway passes through a county, 0 otherwise. The ratio of taxable sales base to personal income, a measure of tax exportation. The percentage of neighboring counties (sharing a common boundary) that have adopted the Local Option Sales Tax (LOST). Number of counties that have adopted the LOST. The percentage of Republican votes in gubernatorial elections. Accumulated state and local sales tax rate that has been collected in a county. Dummy variable, coded 1 if located within the Atlanta Metropolitan Statistical Area, 0 otherwise. The annual percentage change of real per capita personal income. The annual percentage change of population.

Mean

Standard Deviation

23.17

6.80

2.43

0.85

37.30

17.77

0.04

0.70

0.34

.—

0.49

0.21

0.37

0.34

69.57

51.51

0.38

0.16

3.29

0.71

0.13

.—

0.02

0.07

1.62

2.82

a. This variable takes the value in 1985. It is used to measure relative levels of fiscal stress among counties rather than the variation over time.

The obstacles to adoption include (1) Republican ideology, measured as the percentage of Republican votes in gubernatorial elections (REPUBLICAN); (2) the accumulated state and local sales tax rate that has been collected in a county (TAXRATE); and (3) an indicator

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

735

of severe tax competition among shopping centers, measured by whether a county is located within the Atlanta MSA (ATLMSA). These variables are expected to have negative effects on the adoption. The resources for overcoming the obstacles are related to different channels of policy diffusion. One measure of policy diffusion is the percentage of neighboring counties (sharing a common boundary) that have adopted the LOST (NEIGHBOR). The other measure is the number of Georgia counties that have adopted the LOST (NLOST). These two variables are used to test the “neighboring diffusion” model and the “national (statewide) interaction” model, respectively. Their coefficients are expected to be positive. In addition, the study controls for the changes of demographic and fiscal conditions. The model is analyzed by two maximum likelihood estimations. The first estimation includes all the at-risk observations during the period 1975 to 2002, while the second one involves only those during the period of 1975 to 1980.18 As discussed before, half of the adoptions occur within the first five years since the LOST Act was enacted. The two-step analysis attempts to examine whether the early adoptions were triggered by the same factors as the late adoptions. It is hypothesized that counties with higher motivations—higher property tax level, the existence of fiscal pressure, or the potential of tax exportation—might adopt the LOST earlier, while others that adopted the LOST later may have adopted it more because of tax mimicking. In both estimations, year dummies are included to control for omitted variables that change over time but are constant between cases. This method is equivalent to the time fixed-effects model of panel analysis. The results of the estimations are shown in Table 2. Beginning with the analysis of the sample during 1975 to 2002, we see that many hypotheses about the motivations are supported. First, the property tax level is significantly related to the adoption. As expected, MILLAGE is positively related to the adoption (significant at the .05 level). Surprisingly, however, PTINCM significantly lowers the probability of the adoption (at the .1 level). Further reasoning may provide a possible explanation. As MILLAGE is controlled for, PTINCM actually reflects the ratio of property tax digest to personal income, that is, the “effective” size of property tax base. Thus, the result suggests that counties with higher millage rates have higher propensity to adopt the LOST, while counties with broader property tax base tend not to adopt

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

736

PUBLIC FINANCE REVIEW

the LOST. Second, the potential of sales tax exportation increases the probability of the adoption. The coefficient of HIGHWAY is positive and significant at the .05 level, while EXPORT is positive and significant at the .01 level.19 The results suggest that counties passed by an interstate highway or those have disproportionately high taxable sales bases are more likely to adopt the tax. Nevertheless, the effects of fiscal pressure—as measured by UNPAVED and DGCHANGE—are not supported; the two variables do not have significant coefficients. It seems that counties with fiscal difficulties are not more likely to levy the LOST than others. This finding suggests that the public may not have an intention to use the LOST to supplement their property taxes. Most hypotheses about the obstacles to adoption receive strong support. TAXRATE and ATLMSA are both negatively associated with the adoption (significant at the .01 level). It suggests that counties with a higher existing sales tax rate have lower probabilities to adopt the LOST; so do counties located within Atlanta MSA. But the coefficient of REPUBLICAN is not significant.20 The results yield strong support to the effects of policy diffusion as well. NLOST and NEIGHBOR are both significantly related to the LOST adoption.21 It means that counties indeed mimic others in their response to the LOST, and they are influenced not only by their neighbors that share a common boundary with them but also by the others in the same state. The analysis based on the sample in the early period (1975-1980) yields slightly different results. The effects of MILLAGE and HIGHWAY become significant at the .01 level. In contrast, the effects of policy diffusion become insignificant. Evidently, the perceived property tax level and the potential of tax exportation play more important roles in the LOST adoption during the early years, while policy diffusion has significant effects only in the later period. One drawback of the logistic model is the difficulty in interpreting statistical results. Unlike the case in linear regression models, the magnitude of effects in logistic models cannot be viewed directly from the coefficients. In this study, the effects of the independent variables on the LOST adoption are interpreted in two ways: the overall ranges of probabilities and the partial ranges of probabilities.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES TABLE 2:

737

Logit Maximum Likelihood Estimates for Event History Analysis Model of the Local Option Sales Tax (LOST) Adoption

1975-2002 Independent Variable MILLAGE PTINCM UNPAVED DGCHANGE HIGHWAY EXPORT NEIGHBOR NLOST REPUBLICAN a TAXRATE a ATLMSA IMCHANGE POPCHANGE Intercept Number of cases Percentage of adoption LR chi-square Probability > chi-square

Coefficient 0.04 –0.26 0.01 0.00 0.57 1.31 0.87 0.04 1.77 –2.03 –3.02 4.45 0.03 1.41 907 15.38 105.53 .00

1975-1980 Z

2.00** –1.90* 1.08 0.03 2.40** 2.47*** 1.76* 2.46** 1.29 –3.41*** –4.15*** 2.60*** 0.60 0.74

Coefficient

Z

0.07 –0.23 0.01 –0.02 1.06 1.56 0.72 0.01 1.96 — — 5.42 0.00 –5.88

2.53*** –1.08 0.77 –0.12 3.75*** 2.26** 1.02 0.70 1.06 — — 2.36** 0.05 –4.84***

542 16.27 71.25 .00

NOTE: Year dummies have been controlled in the models, but their coefficients are not included in this table. a. The two variables are dropped from the 1975 to 1980 estimation for lacking enough variation during the period. *p < .1. **p < .05. ***p < .01.

The overall ranges of probabilities, given in Table 3, show the biggest impacts the independent variables can have on the LOST adoption. A distinction should be made between the estimated range and the hypothetical range. The estimated range is calculated based on the predicted probabilities of actual observations. For instance, it ranges from .00 to .84 according to the estimation with all the observations during 1975 to 2002. The hypothetical range is calculated with the assumption that the observations can have any combination of independent variables within their ranges in the sample (Long 1997).22 In regards to the 1975 to 2002 estimation, the probability of the LOST adoption can range, hypothetically, from .00 to .94.23 That is, if a county located outside the Atlanta MSA has an interstate highway passing through the county and has the maximum values in MILL-

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

738

PUBLIC FINANCE REVIEW

TABLE 3:

Ranges of Probabilities of the Local Option Sales Tax (LOST) Adoption

1975-2002

a

Estimated probabilities b Hypothetical extremes

1975-1980

Max

Min

Range of Probabil.

.84 .94

.00 .00

.84 .94

Range of Max Min Probabil. .74 .93

.01 .00

.73 .93

a. These are the maximum and minimum predicted probabilities of the observations. b. Hypothetical extremes are estimated based on the range of each individual independent variable that are significant in the estimation.

AGE, EXPORT, and NEIGHBOR, as well as the minimum value in PTINCM, it is almost certain (probability = 94 percent) that the county will adopt the LOST right away. This result shows the model’s explanatory power. The partial ranges of probabilities show the effect of each variable on the predicted probability, that is, the extent to which a change in a variable affects the LOST adoption. The way to do this is to allow one variable to vary from its minimum to its maximum, with all other variables fixed at their means. The results are given in Table 4. MILLAGE, HIGHWAY, and EXPORT show consistent effects in both time periods, but their effects are much stronger in the first five years since the LOST was enacted. For instance, the change of MILLAGE from its minimum to its maximum can contribute to an increase of 18 percent in the probability of the adoption during 1975 to 2002, but the same change in MILLAGE can contribute to an increase of 25 percent of probability during 1975 to 1980. In contrast, while the policy diffusion variables do not have significant effects in the early years, they can make dramatic differences in the later period—the variance of NLOST alone can explain a change of 67 percent in the probability of the adoption! The partial ranges of probabilities can also be presented by figures. Figure 3 illustrated how MILLAGE and EXPORT affect the LOST adoption. As the figure shows, the chance of the adoption rises with counties that have higher property tax millage rates or/and higher potentials of tax exportation, and the combined effects of the two variables are significant. It is almost certain that a county with the minimum values of both MILLAGE and EXPORT would not adopt the

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES TABLE 4:

739

Probabilities of the Local Option Sales Tax (LOST) Adoption over the Range of Several Independent Variables for the Logit Model

1975-2002 Variable

At Max.

At Min.

MILLAGE PTINCM HIGHWAY EXPORT NEIGHBOR NLOST TAXRATE ATLMSA

.24 .05 .17 .26 .20 .69 .00 .01

.06 .20 .10 .07 .09 .02 .17 .16

1975-1980

Range of Probabil. .18 .15 .07 .19 .11 .67 .17 .15

At Max.

At Range of Min. Probabil.

.29 .07 .23 .29 .21 .16 .— .—

.04 .19 .09 .06 .11 .10 .— .—

.25 .12 .14 .23 .10 .06 .— .—

Probability of the adoption (%)

50 45 40 35

EXPORT Max

30

EXPORT 80

25 20

EXPORT 60

15

EXPORT 40

10

EXPORT 20

5

EXPORT Min

0 Min

p20

p40

p60

p80

Max

MILLAGE at different values

Figure 3: Partial Ranges of Probabilities of the Local Option Sales Tax (LOST) Adoption for MILLAGE and EXPORT

LOST (p = 3.1 percent); in contrast, a county with the highest values of MILLAGE and EXPORT is much more likely to adopt the tax (p = 42.6 percent).

7. CONCLUSIONS

This study examines the adoption of the LOST in Georgia counties using socioeconomic and fiscal data during the period 1975 to 2002.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

740

PUBLIC FINANCE REVIEW

The discrete-time event history analysis provides support for most of the hypotheses and yields additional clues for further exploration. To conclude, counties with higher property tax millage rate and higher potential of sales tax exportation show higher propensity to adopt the LOST, and these effects are especially strong in the early years. In contrast, the effects of policy diffusion loom larger later when many counties in Georgia have already adopted the tax. The study does not find that fiscal stress has a significant effect on the LOST adoption. It suggests that the public may not have an intention to use LOST proceeds as additional revenues. However, note that fiscal pressure can be measured in many different ways (Alm, McKee, and Skidmore 1993), and each may have a different effect on the LOST adoption. Other measures, such as the ratio of revenue to expenditure (“fiscal health”) or government deficits/debts, should be incorporated in further studies given data availability. This study adds to the literature on policy innovation and diffusion. Previous studies of policy innovation and diffusion in the United States were almost exclusively focused at the state level. The major reason has been data availability rather than theoretical or epistemological considerations (Blomquist 1999). Almost no study of policy innovation and diffusion has been done except for the introduction of lotteries in the American states (Berry and Berry 1990; Alm, McKee, and Skidmore 1993; Furlong 1998), because few new revenue sources have been created in recent decades. The adoption of local option taxes provides an opportunity to further this line of inquiry. A case can be made, in fact, that the study of policy innovation and diffusion may be better conducted at the local level than at the state level. In Georgia, for instance, the number of counties is several times larger than that of the set of American states (more degrees of freedom); these counties operate within an institutional homogeneous setting (more comparable); and they are close to each other and therefore share geographical proximity (easier for the test of policy diffusion). The study has policy implications as well. First, understanding the factors affecting the LOST adoption enhance our knowledge about the nature of this new revenue source that has become increasingly popular. States can incorporate this knowledge while considering enacting similar programs or revising their current ones; policy entre-

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

741

preneurs at the local level can better assess the “policy windows” (Kingdon 1984) and adjust their strategies to make a policy change (Singer 1987); and the public can make a more informed decision when they cast their votes in LOST referenda. Second, understanding the LOST adoption helps to disentangle related puzzles about the LOST in Georgia. For instance, debates continue on whether LOST proceeds can be or have been used for additional revenues (Durning 1992; Jung 2001). The factors determining the LOST adoption may provide hints for the budgetary effects of LOST proceeds. To the extent that counties adopt the LOST because of fiscal pressure, they are more likely to use the proceeds as additional revenue—but this study does not find the public have an intention to use the LOST for that purpose; to the extent that counties mimic their neighboring counties in the LOST adoption, they might take into account property tax levels in their other counties when they use the LOST for property tax relief. These questions will be explored in my future studies.

NOTES 1. According to public opinion polls, property tax has always been “the least popular tax” (Advisory Commission on Intergovernmental Relations 1994). Property tax is disliked for several major reasons: its visibility or transparency, administrative difficulties in assessments, the imperfect association between homeowner income and tax liabilities over time, and fiscal disparities across jurisdictions (McGuire 2001; Oates 2001). 2. See National Conference of State Legislatures, “Local Option Sales Taxes: A Legisbrief,” http://www.ncsl.org/programs/fiscal/lbloptax.htm (accessed May 22, 2005). 3. In Georgia, for example, the general-purpose Local Option Sales Tax (LOST) is earmarked for property tax relief, the special-purpose Local Option Sales Tax (SPLOST) is earmarked for capital expenditures, the educational-purpose Local Option Sales Tax (ELOST) is earmarked for education, and the Metropolitan Atlanta Rapid Transit Authority tax (MARTA) is earmarked for transportation. 4. See Iowa Department of Revenue and Finance, “2003 County Financial Overview,” http://www.iowacounties.org/fiscalinfo/cfo/cfo2003.htm (accessed April 22, 2004). 5. In counties in which the county government did not enact the LOST, cities could, with voter approval, impose their own sales tax. However, no city has ever adopted the LOST independently. 6. City governments, however, were not required to roll back property taxes after the second year.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

742

PUBLIC FINANCE REVIEW

7. A city was defined as “qualified” if it imposed a tax other than the local option sales tax and offered at least three of the following local government services: (1) water, (2) sewage, (3) garbage collection, (4) police protection, (5) fire protection, and (6) library (Association County Commissioners of Georgia 2002). 8. Five counties, Cobb, Cherokee, DeKalb, Gwinnett, and Rockdale, do not levy LOST. In another five counties, Bulloch, Habersham, Houston, Mitchell, and Rabun, LOST revenues are earmarked for the county school system instead of property tax relief. 9. The Dye-Sharkansky-Hofferbert (DSH) comparative study reveals the complexity of policy decisionmaking in its socioeconomic contexts, but it is critiqued by many for its inductive and data-driven nature (Blomquist 1999, 220) and the “extremely weak theoretical structure” (Tullock 1967, 539) 10. The median voter approach is especially useful for policy decisions that involve direct voting, as is the case with local option sales taxes. In this situation, the median voters actually get to vote! 11. This requires an “inferential leap” that characteristics of the population determined median voter preferences and, hence, community preferences. See Dye (1990). 12. Organizations with greater levels of slack resources seem to be more innovative than others (Walker 1969; Gray 1973; Rogers 1983), but mixed results were also found (Savage 1985). Other commonly utilized socioeconomic factors include population size (Dye and Davidson 1981), industrialization (Walker 1969; Foster 1978), urbanization (Mueller and Comer 1983), education level (Allen and Clark 1981), religion (Wohlenberg 1980), public opinion (Blair and Savage 1981), and socioeconomic change (Clarke 1977), but the empirical results are also mixed. 13. The “perceived” property tax level (MILLAGE) and “effective” property tax level (PTINCM) are barely correlated (R = .055). 14. A caveat with this measure is that the percentage of unpaved public roads may also reflect the level of urbanization. But the correlation between UNPAVED (the percentage of unpaved public roads) and URBAN (the percentage of urban population in a county, not used in this study) is only modest (r = –.46). Because most counties in Georgia are rural, public roads in urban areas may only consist of a small portion of all the public roads in a county—in this regard, the percentage of unpaved public roads may reflect, although to a limited extent, the level of fiscal stress. 15. In addition to the LOST, local governments in Georgia can elect to adopt several other local option sales taxes, such as the SPLOST, the ELOST, or the MARTA tax. 16. Geographic information systems (GIS) analysis (not included in this report) shows the Atlanta metropolitan statistical area has a far higher population density and a far higher accumulation of taxable sales base than other areas in Georgia. 17. The approach is called linear probability model (LPM) when the ordinary least squares (OLS) method is applied to a binary dependent variable. There are many problems with LPM (Long 1997). First, as the errors are heteroskedastic, the OLS estimator is inefficient and the standard errors are biased, resulting in incorrect test statistics. Second, the errors cannot be normally distributed. Without assuming normality, while the OLS estimates can still be unbiased, it raises problem in hypotheses testing. Third, the LPM could easily predict nonsensible values of probability y that are negative or greater than 1. Finally, the model assumes unrealistic linear relation between the independent variables and the probability of an event. 18. A better way to examine the structural change is to estimate the model with two divided samples (1975-1980 and 1981-2002). However, there is not enough variability in the second sample (1981-2002) because very few counties adopted the LOST during that period. As a result, the study examines the models with two overlapping periods. 19. HIGHWAY and EXPORT are only slightly correlated (R = .1615).

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

743

20. REPUBLICAN is measured as the percentage of republican voters in gubernatorial elections. It is possible that this variable does not well reflect the political ideology at the local level. First, local residents may have different political preferences with local issues than at the state level. Second, the variable is measured by vote turnout rates; it may be better measured by voter registration rates, given data availability. 21. NEIGHBOR and NLOST are strongly correlated (R = .616). 22. For instance, the lower extreme of the probabilities is calculated by setting each independent variable associated with a positive β to its minimum and each independent variable associated with a negative β to its maximum. Note that observations with this combination of independent variables do not necessarily approximate any member of the sample (Long 1997). 23. The hypothetical range of probabilities and the partial range of probabilities are predicted by the Clarify software program created by King, Tomz, and Wittenberg (2000).

REFERENCES Advisory Commission on Intergovernmental Relations. 1994. Changing public attitudes on governments and taxes. Reports S-1 to S-23. Washington, DC: Advisory Commission on Intergovernmental Relations. Allen, Russ, and Jill Clark. 1981. State policy adoption and innovation: Lobbying and education. State and Local Government Review 13:18-25. Alm, James, Michael McKee, and Mark Skidmore. 1993. Fiscal pressure, tax competition, and the introduction of state lotteries. National Tax Journal 46 (4): 463-76. Association County Commissioners of Georgia. 2002. 2000-2002: Renegotiation of the local option sales tax revenue distribution certificate. Atlanta, GA: Association County Commissioners of George. Bachtel, Douglas C., and Susan R. Boatright, eds. 1982-2003. The Georgia county guide. Athens: Center for Agribusiness and Economic Development, University of Georgia. Bergstrom, Theodore C., and Robert P. Goodman. 1973. Private demands for public goods. American Economic Review 63 (3): 157-77. Berry, Frances Stokes, and William D. Berry. 1990. State lottery adoption as policy innovation: An event history analysis. American Political Science Review 84 (2): 395-415. Besley, Timothy, and Anne Case. 1995. Incumbent behavior: Vote seeking, tax setting and yardstick competition. American Economic Review 85 (1): 25-45. Black, Duncan, R. A. Newing, Iain McLean, Alistair McMillan, and Burt L. Monroe. 1998. The theory of committees and elections. Rev. 2nd ed. Boston: Kluwer Academic. Blackley, Paul R., and Larry DeBoer. 1987. Tax base choice by local governments. In Land economics, vol. 63, 227. Madison: University of Wisconsin Press. Blair, Diane, and Robert L. Savage. 1981. Abortion policies in the fifty states: A current case of pluralistic responsiveness. Southeastern Political Review 9:34-55. Blomquist, William. 1999. The policy process and large-N comparative studies. In Theories of the policy process, ed. P. A. Sabatier, viii, 289. Boulder, CO: Westview. Boyne, George A. 1998. Party competition and local spending decisions. British Journal of Political Science 28 (1): 210-22. Case, Anne C., Harvey S. Rosen, and James R. Hines Jr. 1993. Budget spillovers and fiscal policy interdependence: Evidence from the states. Journal of Public Economics 52:287-307. Chelf, Carl P. 1984. Public policymaking in America. Glenview, IL: Scott, Foresman.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

744

PUBLIC FINANCE REVIEW

Chicoine, David L., and Norman Walzer. 1986. Factors affecting property tax reliance. Public Choice 49 (1): 17-28. Clarke, Susan E. 1977. Growing and learning: The incidence of state growth management innovations. Paper presented at the annual meeting of the Southern Political Science Association, November, New Orleans, LA. Clements, Betty J., and J. Devereux Weeks. 1997. County and municipal revenue sources in Georgia. Athens, GA: Carl Vinson Institute of Government. Durning, Danny W. 1992. Distributing Georgia’s general-purpose local-option sales tax revenues: An examination of the present policy and some options. Athens, GA: Carl Vinson Institute of Government, University of Georgia. Dye, Thomas R. 1966. Politics, economics, and the public: Policy outcomes in the American states. Chicago: Rand McNally. . 1967. Governmental structures, urban environment and educational policy. Midwest Journal of Political Science 11:353-80. . 1990. The policy consequences of intergovernmental competition. Cato Journal 10 (1): 59-73. Dye, Thomas R., and Dorothy K. Davidson. 1981. State energy policies: Federal funds for paper programs. Policy Studies Review 1 (2): 255-62. Elling, Richard C. 1979. State party platforms and state legislative performance: A comparative analysis. American Journal of Political Science 23 (2): 383-405. Foster, John L. 1978. Regionalism and innovation in the American states. Journal of Politics 40:179-87. Furlong, Edward J. 1998. A logistic regression model explaining recent state casino gaming adoptions. Policy Studies Journal 26 (3): 371-83. Georgia Department of Revenue. 1996. A tax guide for Georgia citizens. Atlanta: Georgia Department of Revenue. . 2004. History rate chart. Atlanta: Georgia Department of Revenue. Gold, Steven David. 1979. Property tax relief. Lexington, MA: Lexington Books. Gray, Virginia. 1973. Innovation in the states: A diffusion study. American Political Science Review 67 (4): 1174-85. Hettich, Walter, and Stanley L. Winer. 1988. Economic and political foundations of tax structure. American Economic Review 78:701-12. Heyndels, Bruno, and Jef Vuchelen. 1998. Tax mimicking among Belgian municipalities. National Tax Journal 51:89-101. Hofferbert, Richard I. 1974. The study of public policy. Indianapolis, IN: Bobbs-Merrill. Hosmer, David W., and Stanley Lemeshow. 1989. Applied logistic regression. New York: John Wiley. Jung, Changhoon. 2001. Does the local-option sales tax provide property tax relief? The Georgia case. Public Budgeting and Finance 21 (1): 73-86. King, Gary, Michael Tomz, and Jason Wittenberg. 2000. Making the most of statistical analyses: Improving interpretation and presentation. American Journal of Political Science 44 (2): 346-61. Kingdon, John W. 1984. Agendas, alternatives, and public policies. New York: HarperCollins. Krmenec, Andrew J. 1991. Sales tax as property tax relief? The shifting onus of local revenue generation. The Professional Geographer 43 (1): 60-67. Ladd, Helen F. 1992. Mimicking of local tax burden among neighboring counties. Public Finance Quarterly 20 (4): 450-67.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

Zhao / MOTIVATIONS, OBSTACLES, AND RESOURCES

745

Long, J. Scott. 1997. Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage. Maxwell, James Ackley. 1965. Financing state and local governments. Washington, DC: Brookings Institution. McGuire, Therese J. 2001. Alternatives to property taxation for local government. In Property tax and local government finance, ed. W. E. Oates. Cambridge, MA: Lincoln Institute of Land Policy. Menard, S. 1995. Applied logistic regression analysis. Thousand Oaks, CA: Sage. Mohr, L. B. 1969. Determinants of innovation in organizations. American Political Science Review 63:111-26. Mooney, Christopher Z., and Mei-Hsien Lee. 1995. Legislating morality in the American states: The case of pre-Roe abortion regulation reform. American Journal of Political Science 39:599. Mueller, Keith, and John Comer. 1983. Disinnovation in the American states: Policy toward health systems agencies. Journal of Sociology and Social Welfare 10:189-202. Nice, David C. 1987. State party ideology and policy making. Policy Studies Journal 13:780-96. Norstrand, Rolf. 1980. The choice between income tax and land tax in Danish municipalities. Public Finance 35 (3): 412-24. Oates, Wallace E. 2001. Property tax and local government finance: An overview and some reflections. In Property tax and local government finance. Cambridge, MA: Cambridge University Press. Pajari, Roger N. 1984. Georgia’s local option sales tax: Who adopts it and why? Southeastern Political Review 12:127-53. Poterba, James M. 1995. Capital budgets, borrowing rules, and state capital spending. Journal of Public Economics 56 (2): 165-87. Rogers, Everett M. 1983. Diffusion of innovations. 3rd ed. New York: Free Press. Santerre, Rexford E. 1986. Representative versus direct democracy: A Tiebout test of relative performance. Public Choice 48 (1): 55-63. Savage, Robert L. 1978. Policy innovativeness as a trait of American states. Journal of Politics 40 (1): 212-24. . 1985. Diffusion research traditions and the spread of policy innovations in a federal system. Publius—The Journal of Federalism 15:1-11. Schneider, Mark. 1989. The competitive city: The political economy of suburbia. Pittsburgh, PA: University of Pittsburgh Press. Shadbegian, Ronald J. 1999. The effect of tax and expenditure limitations on the revenue structure of local government, 1962-87. National Tax Journal 52:221-38. Sharkansky, Ira. 1970. Policy analysis in political science. Chicago: Markham. Shuford, Gordon, and Richard Young. 2000. A report on local government funding: An overview of national issues and trends. Manuscript. Singer, Jerry A. 1987. The special 1% sales tax referendum: Tips and techniques on municipal participation. Urban Georgia 37:9-13. Skidmore, Mark. 1999. Tax and expenditure limitations and the fiscal relationships between state and local governments. Public Choice 99 (1-2): 77-102. Temple, Judy A. 1996. Community composition and voter support for tax limitations: Evidence from home-rule elections. Southern Economic Journal 62:1002-16. Tiebout, Charles M. 1956. A pure theory of local expenditures. Journal of Political Economy 64 (5): 416-24. Tullock, Gordon. 1967. Politics, economics, and the public [Book review]. Journal of Business 40:539.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.

746

PUBLIC FINANCE REVIEW

Tuma, Nancy Brandon, and Michael T. Hannan. 1984. Social dynamics: Models and methods. Orlando, FL: Academic Press. Ulbrich, Holley H. 1996. Local option sales taxes and municipal finance in South Carolina: A look at the first few years. Prepared for the Municipal Association of South Carolina, Columbia. Walker, Jack L. 1969. The diffusion of innovations among the American states. American Political Science Review 63:880-89. Wisconsin Taxpayers Alliance. 2002. Do county sales taxes lower property taxes? Vol. 70. Madison: Wisconsin Taxpayers Alliance. Wohlenberg, Ernest H. 1980. Correlates of Equal Rights Amendment ratification. Social Science Quarterly 60:676-84.

Zhirong Zhao is an assistant professor in the Department of Political Science at Eastern Michigan University. He received his doctorate in public administration from the University of Georgia. His current research interests are in alternative sources of revenue for local governments and related policy issues.

Downloaded from http://pfr.sagepub.com at Serials Records, University of Minnesota Libraries on January 23, 2008 © 2005 SAGE Publications. All rights reserved. Not for commercial use or unauthorized distribution.