Externalities in the Fragmented Metropolis

Abstract

A highly fragmented system of local governance represents a market-like arrangement in which competition among service providers benefits citizen-consumers by improving the efficiency of local public service delivery. Yet, the local public market can also fail, producing negative outcomes for some communities and their residents. Because fragmentation can have differential impacts on the welfare of different groups of citizens—benefiting others and harming some—the organization of the local public sector raises important equity questions. This research focuses on the negative outcomes or externalities generated by local government fragmentation, specifically urban sprawl and spatial economic segregation. The results of the empirical analysis suggest that the design of the local public sector structure involves a trade-off between efficiency and equity.

Keywords

government fragmentation externalities urban sprawl economic segregation

Introduction

Local governments are the face of public service. In 2012, there was one local government for every 3,500 Americans. The different types of local governments—counties, municipalities, townships, special authorities, and school districts—collectively spent approximately US$4,650 for each American.¹ Compared with other levels of government, services provided at the local level often have a more direct and palpable impact on the day-to-day lives of citizens.

Not surprisingly, the most effective way of organizing the local public sector has been a subject of intense debate in the field of Public Administration. Regionalists from the institutional reform perspective, which traces its intellectual lineage from traditional public administration, call for a more centralized and hierarchical local governing system (see Stephens and Wikstrom, 2000, for a review of the literature). In contrast, others from the polycentric school,² which is grounded in public choice, advocate for fragmented and overlapping local governments (V. Ostrom, Tiebout, & Warren, 1961; Tiebout, 1956). The early debates between the two camps focused on the service delivery implications of the local government structure, with both arguing that their respective visions of local governance will lead to a more efficient public service (Jimenez & Hendrick, 2010).

An impressive number of studies in Public Administration and its subfields have explored the issue of service efficiency at the local level. The general conclusion from this literature is that local government fragmentation leads to a smaller public sector, which some argue is evidence of greater service efficiency (see Hendrick, Jimenez, & Lal, 2011; Jimenez & Hendrick, 2010, for reviews of the literature). Scholars from other fields, notably Urban Studies, followed a different research track and identified a group of policy issues distinct from service delivery concerns. This body of literature explores how fragmented systems can create incentives for local policymakers and residents alike to pursue policy and lifestyle choices that benefit these actors but may have negative effects on others. These instances are called externalities or spillover effects, which result when groups of citizens are affected, either positively or negatively, by the collective actions or choices of other citizens.

This research explores the possible negative externalities associated with local government fragmentation. Specifically, it examines the role of the design of the local public sector structure in promoting spatial economic segregation and urban sprawl in metropolitan areas in the United States from 1990 to 2000. An important objective of this study is to expand the focus of Public Administration scholarship beyond the concern for the service outcomes of the organization of the local public sector. It argues for a more holistic view that takes into account the different classes of governance problems resulting from local institutional choices. By focusing on externalities, this study hopes to raise the awareness of policymakers and elected officials about the complex set of problems arising from the design of local governing systems.

In addition, because fragmentation can have differential impacts on the welfare of different groups of citizens—benefiting others and harming some—the organization of the local public sector raises important equity questions. The choice of local governing systems therefore involves potential trade-offs between service efficiency and equity (Howell-Moroney, 2008), putting the issue of local externalities at the heart of Public Administration scholarship.³

The next section discusses the theory of government fragmentation and reviews the empirical evidence on its outcomes. It also expounds on the concept of externalities. Succeeding sections develop and test different hypotheses on how fragmentation can influence segregation and urban development patterns in metropolitan areas in the United States, present the empirical model and estimation methodology, and discuss the findings from the econometric analysis.

Theoretical and Empirical Literature

Service Delivery and Local Government Fragmentation

The greatest strength of the competitive private market lies in its capacity to protect and promote consumer sovereignty (Friedman, 1962). In comparison with the state, the private market—which relies on voluntary exchanges among actors—is considered to be more efficient and responsive to the diverse preferences of consumers for goods and services. The choice among numerous sellers protects the consumer from coercion by any seller, in the same manner that the presence of numerous buyers protects the seller from coercion by any buyer (Friedman, 1962). If neither the consumer nor the producer benefits, the economic transaction will not take place. In the language of welfare economics, markets produce Pareto superior outcomes in which scarce economic resources are allocated in such a way that some individuals are made better off without making others worse off.

Scholars from public choice argue that the welfare-enhancing effect of the private market can be replicated in the public sector. It was Tiebout (1956) who argued that a highly fragmented local public sector functions as a market place in which local governments are the producers and citizens are the consumers. This view has come to be known as the local public market model (Schneider, 1989). It is perhaps more accurate to describe the local public market as a collection of numerous markets for different types of public services. Local government-producers offer an assortment of services at different tax prices, and citizen-consumers shop for tax and service packages that they prefer. For the local government, attracting mobile taxpayers is important because much of local public services is funded through locally generated revenues. For taxpayers, the choice afforded by the presence of numerous local governments is key to satisfying their specific preferences for different local public services and the attendant tax prices. In what is perhaps the most quoted phrase in the literature on local public finance, Tiebout (1956) pointed out,

The consumer-voter may be viewed as picking that community which best satisfies his preference pattern for public goods . . . The greater the number of communities and the greater the variance among them, the closer the consumer will come to fully realizing his preference position. (p. 418)

Beyond improving responsiveness to citizens’ diverse preferences for public goods, others argue that competition among local governments in fragmented systems protects voter-consumers in other ways, specifically by preventing unsustainable growth in taxes and government budgets (Brennan & Buchanan, 1980; Schneider, 1989). In the same manner that competition in the private market forces firms to sell products at lower prices by operating efficiently, competition in the local public market forces governments to provide public services at lower tax costs. Specifically, because local governments fear losing mobile taxpaying residents, there is increased pressure to ensure efficient delivery of local public services (Brennan & Buchanan, 1980; Schneider, 1989).

Majority of the empirical research on fragmentation focuses on production or technical efficiency rather than on service responsiveness or preference satisfaction.⁴ In these studies, technical efficiency is equated with lower spending, revenues or number of employees.⁵ In general, studies find that fragmentation of general-purpose governments, specifically municipal governments, leads to smaller local budgets (see Hendrick, Jimenez, & Lal, 2011; Jimenez & Hendrick, 2010, for reviews of the literature).⁶

Very few studies focus on preference satisfaction, and most use surveys as the primary method of gathering information about citizen views on local public services. Some studies conclude that residents of smaller jurisdictions are more satisfied with the delivery of specific types of local services compared with residents of bigger communities (see E. Ostrom, 1976). Others find that there is no systematic difference in satisfaction with local services between residents of consolidated and fragmented regions (see DeHoog, Lowery, & Lyons, 1990).

Externalities and Local Government Fragmentation

Although private markets can improve social welfare, there are instances when they fail, benefiting market participants but causing harm to those not directly participating in the economic transaction.⁷ Market failures represent instances when the operations of private markets can lead to Pareto sub-optimal results. If private markets can fail, so can public ones.

Early empirical research on local government fragmentation largely ignored the issue of externalities in the local public market. Indeed, although many studies focus on the positive service-related outcomes of local government fragmentation, Tiebout’s (1956) numerous caveats about the local public market are often lost in the discussion. Specifically, Tiebout (1956) argued that the local public market can enhance social welfare only under a number of strict assumptions, one of which is that the “public services supplied exhibit no external economies or diseconomies between communities” (p. 419). In other words, the public service in question must not produce any externality.

Externalities are a type of market failure. Negative externalities⁸ represent a form of damage, harm, or costs imposed on others who are not directly participating in a market (see Figure 1 and accompanying discussion about externalities in private markets). In the case of production externalities, the producer who does not pay for the damage inflicted by the good on non-market participants will continue producing more of the good. Such overproduction represents a reduction in social welfare as the market continues to produce the good even though the social marginal cost to society is higher than the social marginal benefit. There are also negative consumption externalities in which consumption of a good reduces the welfare of others who are not party to the economic transaction. Unless the consumer is made to pay for the harm done to others, he or she will continue to consume more of the good to the point that social marginal costs exceed marginal benefits.

Figure 1.

Marginal costs and benefits in the presence of externalities.

Some studies have explored the negative outcomes associated with local government fragmentation, although not within the externality framework. Specifically, research has explored the relationship between government fragmentation and urban sprawl (see, among others, Carruthers, 2003; Carruthers & Ulfarsson, 2002, 2003), as well as the spatial segregation of the population according to race, social class, or other socio-demographic characteristics (see Bischoff, 2008; Burns, 1994; Heikkila, 1996; J. H. Lewis & Hamilton, 2011; Miller, 1981; Morgan & Mareschal, 1999; Stein, 1987; Weiher, 1991). This research argues that urban sprawl and spatial economic segregation can be seen as externalities generated by government fragmentation. Although these developments benefit some communities and their residents, they can impose harm on others.

Urban sprawl is described as,

unplanned, uncontrolled, and uncoordinated single use development that does not provide for an attarctive and functional mix of land uses and/or is not functionally related to surrounding land uses and which variously appears as low density, ribbon or strip, scattered, leapfrog, or isolated development. (Nelson & Duncan, 1995, p. 1)

Rapid expansion in the urban fringes increases the supply of land for development. This lowers the production costs for developers because of the availability of cheap land, and increases the benefits for homebuyers and businesses because of the resulting lower prices for new housing stock and commercial spaces. Local government officials also benefit from sprawled development by expanding their jurisdiction’s property tax base.

Sprawl, however, can have negative consequences for other residents and communities in a region. Specifically, research indicates that sprawled development is associated with losses in economic productivity caused by traffic congestion (Brueckner, 2000), air pollution (Ewing, Pendall, & Chen, 2002), overdevelopment of open space, and public health concerns caused by overreliance on automobiles (see references in Howell-Moroney, 2008). The overconsumption of land in the urban fringes thus, creates costs for others for which they are not compensated for.

Local government fragmentation also benefits residents who want to relocate to communities, not only on the basis of their tastes for services and the corresponding taxes, but also based on their preference to live with people sharing similar socio-economic backgrounds (Lowery, 2000; Weiher, 1991). This can result in spatial segregation of the population by social class. Richer municipalities have more resources to provide services that their well-off residents prefer, whereas the poor are isolated in jurisdictions with limited capacity to finance even basic services (Hill, 1974). Research has shown how spatial segregation affects access to quality education (see Aaronson, 1998), health care (Squires & Kubrin, 2005), and other municipal social services (Jimenez, 2014a). The gap in the capacity of poor and rich communities to supply various services can lead to disparities in opportunities and outcomes among groups of citizens (Altshuler et al., 1999). Inequitable access to basic services not only constrains the opportunities available to the current generation but may also determine the success of future generations through the inter-generational transmission of poverty and limited opportunities (Altshuler et al., 1999; Cutler & Glaeser, 1997). The theory of externalities suggests that the harm created by fragmentation and the resulting segregation of communities should be reflected as a form of MD that must be included in the calculation of social costs and benefits of the operations of the local public market.

Hypotheses

The question of course is, does local government fragmentation really lead to urban sprawl and spatial economic segregation? Before exploring this question, it is important to clarify the meaning of local government fragmentation.

A fragmented local governing system is characterized by the presence of numerous local governments. The literature, however, distinguishes between the vertical and horizontal dimensions of the local public sector structure (see Hamilton, Miller, & Paytas, 2004; Hendrick et al., 2011). The vertical structure refers to overlapping governments. These governments cut across political jurisdictions, which means that they share a tax base and serve a sub-group of residents of jurisdiction-based governments. The existence of numerous special-purpose governments is considered a general indicator of the fragmentation of the vertical structure (Berry, 2008; Jimenez, 2014b). Special- or single-purpose governments deliver a single type of service or a small number of closely related services, and include special districts and school districts. Horizontal fragmentation, refers to the presence of numerous jurisdiction-based general-purpose units such as municipalities and counties. These governments are considered general- or multi-purpose because they deliver numerous types of services. Note that all local governments covered in this research are classified as independent governments which exercise administrative and fiscal autonomy.⁹

In the succeeding sections, theories from public finance, urban planning, and public administration are used to develop testable hypotheses on how the design of the vertical and horizontal dimensions of the local public sector structure influences urban development and spatial economic segregation patterns in metropolitan areas.

Fragmentation and Urban Development Patterns

Both vertical and horizontal fragmentation can lead to urban sprawl through different mechanisms. First, the presence of numerous multi-purpose governments creates competition for the property tax base. Local governments may encourage the development of land in the urban fringes to capture the tax gains. Private developers are likely to support such policies and develop cheap land in outlying areas for residential or commercial purposes (see Dye & McGuire, 2000). Second, with each jurisdiction enjoying exclusive control over land use authority, communities can choose to control growth in their areas and prevent land uses such as multi-family residences and public housing that lower property values, thus pushing demand for commercial spaces and residential housing into outlying areas (Carruthers, 2003; Downs, 1994). Finally, by providing infrastructure subsidized by taxpayers from across a number of jurisdictions, fragmentation, specifically of special districts, also enables development in the urban fringes (Carruthers & Ulfarsson, 2002; Nelson & Duncan, 1995).

Although the theoretical expectations are straightforward, empirical evidence on the relationship between vertical and horizontal fragmentation, on one hand, and urban sprawl on the other, is inconclusive. Focusing on 159 counties in the United States, Pendall (1999) found that fragmentation was not statistically correlated with the ratio of population change to the change in urbanized land area between 1982 and 1992. Razin and Rosentraub’s (2000) empirical analyses involving metropolitan areas in the United States and Canada provided no evidence that fragmentation predicted sprawl. In two separate studies, Carruthers and Ulfarsson (2002, 2006) focused on counties and found that fragmentation of general- and special-purpose governments lowered population density. In a different study involving metropolitan counties, Carruthers (2003) found that fragmentation of municipal and special district governments increased growth outside of incorporated areas. Dye and McGuire (2000) examined urban development patterns in the 100 largest metropolitan areas in the United States and concluded that municipal fragmentation increased urbanized land areas and the share of population in collar counties.

It is likely that fragmentation of general- and special-purpose governments has a two-way or simultaneous relationship with sprawl (see Carruthers & Ulfarsson, 2002). Through Tiebout sorting, new communities are formed which necessitates the creation of new local governments to ensure provision of needed services (P. G. Lewis, 1996). The establishment of new special- and general-purpose governments, in turn, enables further suburbanization (Carruthers, 2003; Dye & McGuire, 2000). The simultaneous relationship means that fragmentation measures are endogenous. Taking the endogeneity issue into account, it is expected that

Hypotheses 1: Horizontal fragmentation or the presence of numerous general-purpose governments facilitates urban sprawl.

Hypotheses 2: Vertical fragmentation or the presence of numerous special-purpose governments facilitates urban sprawl.

Fragmentation and Spatial Economic Segregation Patterns

In the social stratification and government inequality thesis (or SSGI), Hill (1974) argued that “political incorporation by class and status into municipal enclaves is an important institutional mechanism for creating and perpetuating inequalities among residents in metropolitan communities” (p. 1557). Fragmentation of local governments is argued to be associated not only with spatial racial segregation, but also with the spatial separation of the rich from the poor (see also Lowery, 2000; Jimenez, 2014a).

In the Tiebout world, the residential choices of citizens are driven by the search for tax and service bundles. However, fiscal migration also can be fueled by “lifestyle” preferences of residents (Lowery, 2000; Weiher, 1991). To be fair to Tiebout (1956), he recognized this possibility, albeit in a footnote where he wrote, “Not only is the consumer-voter concerned with economic patterns, but he desires, for example, to associate with ‘nice’ people” (p. 418). Residents, in other words, desire to reside in communities to live with people sharing similar socio-economic characteristics. The socio-economic homogeneity of communities can be preserved in two ways. First, Weiher (1991) argued that jurisdictional boundaries function as signaling and recruiting devices, providing information to prospective residents about the social character of the community. Second, through their land use powers, local governments can prohibit multi-family housing and public housing projects so as not to attract poor families to live in their communities (Savitch & Vogel, 2000). These policies are without a doubt discriminatory, but they can also be seen as strategic fiscal behavior on the part of local officials who fear that an increase in the size of the poor population will induce demand for local services without corresponding improvements in revenues, and may even drive away affluent tax payers.

The literature provides evidence that both horizontal and vertical fragmentation lead to racial segregation. In a study of 76 metropolitan areas, Morgan and Mareschal (1999) found that a higher number of cities per 100,000 population is associated with the spatial segregation of Blacks. Another metropolitan level study by J. H. Lewis and Hamilton (2011) found that the diffusion of expenditure responsibilities across numerous municipalities, townships and school districts increased White–Black residential segregation. Bischoff (2008) also found that fragmentation led to greater racial segregation at the school district level.

Whether fragmentation leads to spatial economic segregation is less clear. Hill (1974) conducted the seminal study of the relationship between the local public sector structure and inter-jurisdictional income inequality. He found that the number of cities per capita had a statistically significant direct relationship with inter-municipal income inequality. E. Ostrom (1983) criticized this study because the measure of inequality—the standard deviation of the distribution of the median family income among municipalities—is sensitive to the size of jurisdictions. Focusing on Los Angeles county, Miller (1981) found that separate incorporation facilitated population sorting by both race and income. Weiher (1991) showed that segregation by race, income, and education from 1960 to 1980 occurred increasingly at the city level rather than at the neighborhood level. Heikkila (1996) concluded that municipalities are Tieboutian clubs which are sorted according to variables such as education, home value, professional occupations, income, and ethnicity, among others. Stein (1987) found that fragmentation of municipal governments increased residential sorting by education only in his study of 224 metropolitan areas. Other studies compared the income of residents in the suburbs and central cities but failed to find any significant difference (see Bollens, 1990; Logan & Schneider, 1982; Morgan & Mareschal, 1999; Post & Stein, 2000).

Rather than a unidirectional relationship, it is plausible that the relationship between fragmentation and spatial economic segregation is simultaneous. Multiple jurisdictions facilitate household sorting, creating numerous communities not only with different tax and service preferences but also of different socio-economic status. The desire for social separation, is a central force driving further political fragmentation (Burns, 1994; Weiher, 1991).

The structural dimension blamed for economic segregation is the fragmentation of general-purpose governments (Burns, 1994; Hill, 1974; Morgan & Mareschal, 1999; Weiher, 1991). The presence of numerous jurisdictions matters for local residents as they pursue their lifestyle preferences. However, it can be argued that fragmentation of single-purpose governments is also likely to affect residential choices. This is particularly true in the case of school districts (Orfield, 1997). With more choices of school districts, wealthier households can opt for exit and leave a less desirable district, leading to the concentration of low-income households in some districts.Taking into account the endogeneity of fragmentation caused by simultaneous causation, it is hypothesized that

Hypotheses 3: Horizontal fragmentation or the presence of numerous general-purpose governments facilitates spatial economic segregation.

Hypotheses 4: Vertical fragmentation or the presence of numerous special-purpose governments facilitates spatial economic segregation.

Empirical Methodology

Unit of Analysis and Empirical Models

The hypotheses are tested on data for all 362 metropolitan statistical area (MSA) in the United States. Equations 1 and 2 are the formal representations of the primary models to be estimated in this study. These pooled cross-sectional time series models use data from the 1990 and 2000 Decennial Census (DC) and the 1992 and 2002 Census of Governments (CoG).¹⁰ The CoG is conducted by the Census Bureau every 5 years, and every 10 years for the DC. The years covered in this study ensure that the local government data from the CoG can be matched with socio-economic and urban development data from the DC.

{Sprawl}_{i t} = α + βF {(fragmentation)}_{i t} + λC {(controls)}_{i t} + ξD (dummies) + μ_{i t} .

{Segregation}_{i t} = α + βF {(fragmentation)}_{i t} + λC {(controls)}_{i t} + ξD (dummies) + μ_{i t} .

In Equations 1 and 2, the left-hand-side variables represent measures of urban sprawl and spatial economic segregation in metropolitan area i in t or years 1990 and 2000. The right-hand-side variables include the vector F which includes measures of the fragmentation (or consolidation) of the horizontal and vertical dimensions of the local public sector structure, C is a vector of control variables, and µ is the error term. The vector D includes regional and year dummies.¹¹ The Southeast region is used as the base region, and 2000 as the base year. Table A1 in the appendix provides information about data sources and basic descriptive statistics for all variables.

Estimation Strategy for the Urban Sprawl Model

The selection of control variables is based on the findings from previous studies. First, government expenditures for infrastructures such as roads and sewerage can pave the way for the formation of new communities in the urban fringes (Nelson & Duncan, 1995). Second, racial heterogeneity influences development patterns, given the desire of local residents to establish or migrate to socially and racially homogeneous communities (Weiher, 1991). The “White flight” to outlying suburban areas, for example, is driven by the desire of White households to separate from Blacks, Hispanics, and the poor who are most likely to be concentrated in central cities (Massey & Denton, 1993). Noting how poverty concentration issues and congestion problems present in cities may drive households away from urban centers, the count of central cities in a metropolitan area is included in the models.

The pursuit of lifestyle choices, of course, ultimately depends on financial capacity of households. Thus, the third control variable is per capita income. The effect of income, however, is difficult to specify. It is possible that higher income enables households to stay in central locations in a region contributing to a compact development pattern (Carruthers, 2002). Alternatively, higher income can facilitate the formation of new communities in outlying areas (Margo, 1992). Fourth, taxes have been shown to be an important determinant of sprawl. Business and households relocate to unincorporated areas to avoid high taxes in central cities and older, more industrialized suburbs (Carruthers, 2002). In addition, to the extent that infrastructure is financed through taxes shouldered by all tax payers in the region, an incentive is created for less dense development. This occurs because the cost of infrastructure is not fully borne by taxpayers in the new development (Brueckner, 2000; McGuire & Sjoquist, 2003). Fifth, the urban sprawl model includes percentage employment in manufacturing to capture the general trend of the suburbanization of manufacturing jobs (see Kain, 1992).

As already noted, the simultaneous relationship between sprawl and fragmentation means that the fragmentation measures in Equation 1 are endogenous. Technically, a regressor is said to be endogenous when it is correlated with the error term. Remember that in Equation 1, sprawl is hypothesized to be caused by fragmentation, a group of control variables, and unobserved random process represented by the error terms µ. Given that µ is correlated with sprawl, and sprawl causes fragmentation because of simultaneous causation, then fragmentation logically is also correlated with µ—the technical definition of endogeneity.

Estimates of the effects of fragmentation on sprawl using ordinary least squares (OLS) regression will be biased if fragmentation is correlated with the error term. What happens is that the coefficients for the fragmentation measures will also capture or reflect the correlation between the error term and sprawl, rather than just the effects of fragmentation. To isolate the true effects of fragmentation on sprawl, the correlation between the error term and fragmentation must be first purged. The appropriate statistical tool to use is instrumental variable (IV) regression, which involves a two-step procedure (Wooldridge, 2002, provides a detailed explanation of IV regression).

In the first stage of IV regression (also called the reduced form equation) represented by Equation 1.1, the endogenous variable F is regressed against I which is a set of instruments or control variables that are not in Equation 1, C which are the exogenous control variables in Equation 1, and D which is a set of regional and year dummy variables. All the exogenous controls from Equation 1 must be included in Equation 1.1 to arrive at unbiased parameters in Equation 1.2 (Baltagi, 2002). The predicted values of fragmentation $\hat{F}$ are calculated in the first stage. $\hat{F}$ is already purged of any correlation with the error terms. In the second stage or Equation 1.2, the predicted values $\hat{F}$ are used in place of the actual values F to isolate the true effects of fragmentation on urban sprawl.

{Fragmentation}_{i t} = α + γI {(instruments)}_{i t} + λC {(controls)}_{i t} + ξD (dummies) + ν_{i t} .

{Sprawl}_{i t} = α + β \hat{F} {(fragmentation)}_{i t} + λC {(controls)}_{i t} + ξD (dummies) + μ_{i t} .

The list of instruments I includes fiscal institutions, state laws regulating the creation of new governments, and deeply lagged values of the endogenous variables. Some studies suggest that state-imposed local fiscal institutions, specifically tax and expenditure limits, and debt limits are correlated with the creation of special districts (see, for example, Burns, 1994). Because of the institutional constraints on the capacity of general-purpose governments to raise taxes or issue debt, local officials often resort to the creation of special districts which are not subject to similar restrictions. This is one reason why special districts are the most numerous type of local government in the United States (Jimenez & Hendrick, 2010). Because local governments are a creature of their states,¹² the degree of fragmentation of the local public sector also depends on state laws governing the creation of such governments. It is expected that lax state laws governing municipal incorporation will be associated with greater municipal fragmentation. Alternatively, stricter municipal incorporation laws might force local officials and residents to rely more on special-purpose governments (see Berry, 2008). The final instruments are the 1972 values of fragmentation measures. With a lag of 20 to 30 years, the assumption is that these measures will only be weakly correlated, if at all, to the sprawl and segregation patterns in metropolitan areas from 1992 to 2002, but highly correlated to levels of fragmentation in the same time period.

Estimation Strategy for the Spatial Economic Segregation Model

For the spatial economic segregation models, the control variables are based on the specifications in Massey and Denton (1993) and Jargowski (1997). First, a larger population should facilitate greater differentiation of neighborhoods (Jargowski, 1997). Second, income should also facilitate sorting, with richer households having greater residential mobility. Third, higher percentage of minorities such as Blacks and Latinos, as well as higher incidence of poverty, has been shown to lead to economic segregation (Massey & Denton, 1993). Fourth, the regional economic structure, specifically a smaller share of jobs in manufacturing, and a higher share of employment in professional or managerial occupations may increase spatial income segregation. As manufacturing industries decline and factories relocate from urban centers to the urban fringes, poor workers are concentrated in central city ghettoes, priced out of expensive suburban communities (Kain, 1992). Also, increasing skills requirements as reflected in higher share of jobs in managerial and professional occupations can intensify class differences which may eventually lead to the decision of rich skilled workers to reside apart from typically unskilled poorer workers (Jargowski, 1997).

The spatial economic segregation model is also estimated using IV regression following the expected simultaneous relationship between fragmentation and spatial isolation of social classes. The same sets of instruments I in Equation 1.1 are used in Equation 2.1. The predicted values of fragmentation in the first stage, $\hat{F}$ , and all the exogenous control variables and regional and year dummies in Equation 2 are included in Equation 2.2.

{Fragmentation}_{i t} = α + γI {(instruments)}_{i t} + λC {(controls)}_{i t} + ξD (dummies) + ν_{i t} .

{Segregation}_{i t} = α + β \hat{F} {(fragmentation)}_{i t} + λC {(controls)}_{i t} + ξD (dummies) + μ_{i t} .

Operationalization of Variables

Measuring fragmentation (consolidation)

This research adopts measures of the local public sector structure commonly used in previous studies. Horizontal fragmentation (consolidation) is measured as the total number of general-purpose governments in a metropolitan area per 100,000 people. Vertical fragmentation (consolidation) is measured as the count of special-purpose governments per 100,000 people. Standardizing by population ensures comparability of the measures across metropolitan systems.

Measuring urban sprawl

Unfortunately, there is no generally accepted measure of urban sprawl (McGuire & Sjoquist, 2003; Squires & Kubrin, 2005). The urban development pattern is assessed by looking at the population and residential housing characteristics of metropolitan areas. Population density, or the percentage of the population residing in urbanized areas in metros, is often used as a measure of urban sprawl (Carruthers, 2002). The second measure—percentage housing units that are single-detached—represents the residential dimension of urban growth (Razin & Rosentraub, 2000). Lower population density and higher percentage of single-detached housing units both point to an expansive development pattern. The caveat is that none of these variables measures sprawl directly. Though imperfect, these measures enable comparison of urban development patterns across metropolitan areas.

Measuring spatial affluence and poverty concentration

Concentration of poverty and affluence is measured by the class isolation index or simply the P* index that was originally developed by Lieberson (1980). Constructed using the census tract as the unit of analysis, the index gives the probability of residential contact between members of the same income group. The index ranges between 0 and 1 with higher values representing greater spatial poverty (or affluence) concentration.

The isolation index for the poor is defined as

{{}_{x}P}_{x}^{*} = \sum | (\frac{x_{i}}{X}) \times (\frac{x_{i}}{t_{i}}),

where x_i is the number of poor families in tract i, X is the total number of poor families at the metropolitan level, and t_i is the total population of tract i (Massey & Eggers, 1990). Data for the P* indices are from the DC. The poverty threshold for a family of four was US$12,672 in 1989 and US$17,029 in 1999. These figures have no exact equivalents in the DC income categories. This study used the nearest categories of US$12,499 in the 1990 DC and US$19,999 in the 2000 DC as the family poverty ceilings.

The affluent isolation index is defined as

{{}_{y}P}_{y}^{*} = \sum | (\frac{y_{i}}{Y}) \times (\frac{y_{i}}{t_{i}}),

where y_i is the number of affluent families in tract i, and Y is the metropolitan aggregate. Because there is no official federal threshold for affluence, the analysis here follows Coulton et al. (1990) who used 100% of median income for a family of four as the floor for affluence. The median family income in 1989 was US$34,213 and US$50,046 in 1999. This study used the income categories of greater than US$75,000 in the 1990 DC and US$100,000 in the 2000 DC as the floors for affluence.

Results

Local Government Fragmentation and Urban Sprawl

Tables 1 and 2 present the results of the urban sprawl models. The dependent variable in Models 1 and 2 is population density, whereas Models 3 and 4 focus on percentage housing units that are single-detached. Models 1 and 3 are estimated using OLS, whereas Models 2 and 4 are estimated using IV regression. Standard errors are clustered by state to address possible intra-state error correlation and correct for heteroskedastic error distribution. The natural logs of the fragmentation measures are used because of the non-linear relationships observed in two-way scatterplots. The analysis indicates that the highest correlation among control variables is between income and poverty rate at r = .60. Exclusion of one of these variables does not affect the main results. (Because of space consideration, correlation matrixes are not presented here but are available on request.)

Table 1.

Fragmentation and Population Density.

	Model 1: OLS		Model 2: IV
	DV: Population density		Stage 1 DV: General-purpose governments per 100,000 people (log)		Stage 1 DV: Special-purpose governments per 100,000 people (log)		Stage 2 DV: Population density
Independent variables	Coefficient	SE	Coefficient	SE	Coefficient	SE	Coefficient	SE
Fragmentation measures
General-purpose government per 100,000 population (log)	−5.284***	1.150	—	—	—	—	−7.117***	2.530
Special-purpose government per 100,000 population (log)	−1.314	0.944	—	—	—	—	−1.347	1.743
Control variables
% population Black	0.160**	0.079	−0.004	0.004	−0.009	0.005	0.148*	0.083
% population Latino	0.183***	0.058	−0.012***	0.004	0.001	0.006	0.148**	0.065
% population poor	−0.350	0.215	−0.015	0.017	−0.003	0.033	−0.508	0.310
Number of central cities in MSA	1.917***	0.382	−0.038	0.024	0.001	0.034	1.815***	0.387
Highway direct expenditures per capita	3.129	2.189	0.187	0.118	0.033	0.217	3.279	2.217
Sewerage direct expenditures per capita	−4.706	8.923	−0.154	0.486	0.472	0.743	−4.087	8.756
Total tax revenues per capita	8.164	6.578	0.283	0.350	0.238	0.425	6.991	6.306
Income per capita	0.000	0.000	0.000***	0.000	0.000	0.000	0.000	0.000
% employed in manufacturing	−0.335***	0.099	0.012**	0.005	0.017**	0.007	−0.300***	0.089
New England region	1.811	3.916	0.422**	0.168	0.556**	0.214	3.699	3.998
Mid-Atlantic region	5.760*	3.338	0.448***	0.145	0.424**	0.206	7.550**	3.643
Great Lakes region	13.871***	2.709	0.676***	0.157	0.368	0.241	15.826***	3.382
Plains region	14.266***	3.734	0.455**	0.170	0.393*	0.213	16.827***	3.987
South West region	6.350**	2.767	0.023	0.185	0.380	0.304	6.557**	2.875
Mountain region	16.115***	4.659	0.020	0.213	0.753***	0.260	16.261***	5.167
Far West region	4.706	3.397	−0.520**	0.197	0.369	0.269	3.992	4.099
2000 year dummy	5.343***	0.901	0.077	0.108	0.118	0.182	8.069***	1.794
Excluded Instruments
Binding TELs	—	—	0.042	0.107	0.150*	0.078	—	—
Debt limits	—	—	0.235**	0.101	−0.084	0.193	—	—
1972 General-purpose government per 100,000 population (log)	—	—	0.080***	0.022	0.018	0.012	—	—
1972 Special-purpose government per 100,000 population (log)	—	—	0.008	0.007	0.050***	0.014	—	—
Constant	69.425***	9.339	2.343***	0.374	1.831***	0.521	73.548***	11.372
n	683		683		683		683
OLS adjusted R²/ IV centered R²	.598		.821		.642		.592
F test of excluded instruments F statistic/Prob > F			10.94/0.000		11.93/0.000
Hansen J statistic/χ² p value							4.449/.1081

Note. DV is dependent variable. Standard errors (SE) are heteroskedasticity-robust and have been clustered by state to address possible intra-state error correlation. The Southeast is the base region, and 2000 is the base year. OLS = ordinary least squares regression; IV = instrumental variable regression; MSA = metropolitan statistical area.

Significant at 10%.

Significant at 5%.

***

Significant at 1%.

Table 2.

Fragmentation and Percentage Single-Detached Housing.

	Model 3: OLS		Model 4: IV
	DV: % single-detached housing		Stage 1 DV: General-purpose governments per 100,000 people (log)		Stage 1 DV: Special-purpose governments per 100,000 people (log)		Stage 2 DV: % single-detached housing
Independent variables	Coefficient	SE	Coefficient	SE	Coefficient	SE	Coefficient	SE
Fragmentation measures
General-purpose government per 100,000 population (log)	0.883	0.669	—	—	—	—	0.957	0.888
Special-purpose government per 100,000 population (log)	1.563**	0.738	—	—	—	—	1.363**	0.616
Control variables
% population Black	0.065	0.045	−0.001	0.003	−0.006	0.005	0.073	0.045
% population Latino	0.034	0.033	−0.011**	0.004	−0.002	0.005	0.025	0.035
% population poor	−0.572***	0.147	0.006	0.013	0.009	0.020	−0.560***	0.138
Number of central cities in MSA	−0.210	0.254	−0.052	0.033	−0.006	0.042	−0.215	0.263
Highway direct expenditures per capita	0.547	1.186	0.176	0.121	0.075	0.240	0.649	1.195
Sewerage direct expenditures per capita	0.716	5.235	−0.402	0.436	0.184	0.811	0.994	5.430
Total tax revenues per capita	4.416	5.422	0.517	0.381	0.299	0.729	3.183	4.980
Income per capita	−0.001***	0.000	0.000***	0.000	0.000	0.000	−0.001***	0.000
% employed in manufacturing	0.331***	0.056	0.012**	0.005	0.017**	0.007	0.332***	0.053
New England region	−6.032***	1.754	0.375**	0.164	0.596**	0.225	−5.811***	1.712
Mid-Atlantic region	−6.375***	1.698	0.413***	0.132	0.504**	0.239	−6.137***	1.720
Great Lakes region	1.050	1.310	0.586***	0.122	0.409	0.251	1.307	1.339
Plains region	−0.457	1.696	0.372***	0.142	0.450*	0.242	−0.449	1.665
South West region	0.719	1.615	0.116	0.173	0.435	0.322	1.372	1.563
Mountain region	−0.189	1.917	0.254	0.195	0.907***	0.253	0.338	1.925
Far West region	0.211	1.638	−0.338*	0.182	0.609**	0.291	1.138	2.000
2000 year dummy	4.002***	0.528	−0.040	0.040	0.108	0.066	3.948***	0.509
Excluded instruments
Binding TELs	—	—	0.100	0.100	0.176*	0.090	—	—
Debt limits	—	—	0.166	0.099	0.180*	0.092	—	—
1972 General-purpose government per 100,000 population (log)	—	—	0.118***	0.012	0.031**	0.015	—	—
1972 Special-purpose government per 100,000 population (log)	—	—	0.003	0.005	0.049***	0.014	—	—
Constant	79.623***	3.368	1.679***	0.501	1.467**	0.702	79.184***	3.367
n	652		652		652		652
OLS adjusted R²/ IV centered R²	.529		.840		.652		.524
F test of excluded instruments F statistic/Prob > F			35.54/0.000		17.73/0.000
Hansen J statistic/χ² p value							2.119/.3467

Significant at 10%.

Significant at 5%.

***

Significant at 1%.

The OLS models confirm that local government fragmentation fuels urban sprawl, and the estimates are statistically significant, providing support to the first and second hypotheses. Model 1 shows that greater fragmentation of general-purpose governments leads to lower population density in metropolitan areas. Model 3 shows that an increase in single-purpose governments leads to an increase in the percentage of housing units that are single-detached.

For the IV models, the primary concern is the validity of instruments. The instruments must meet two assumptions to be valid. Specifically, they have to be jointly correlated with the endogenous variable but orthogonal to, or not correlated with, the error process (Wooldridge, 2002). The F test in Stage 1 of the IV regression is used to check if the instruments are jointly and significantly correlated with the measures of fragmentation. Note that the indicator variable for the presence of municipal incorporation limits was not used as an instrument because it failed to satisfy conditions for instrument validity. In Stage 1, the significant F statistic indicates that the remaining four instruments are jointly significant. The partial R² show that these instruments explain between 35% and 43% of the variation in the number of governments in MSAs. In Stage 2, the Hansen J statistic for overidentification is used to test if the instruments are correlated with the error terms. The Hansen J statistic is insignificant and fails to reject the null hypothesis that the instruments are orthogonal to the error process.

As for the main findings, the results of the IV models confirm the OLS results. Measuring the magnitude of effects, Models 1 and 2 show that a 1% increase in general-purpose governments per 100,000 people (or approximately one new government unit) reduces population density between 0.053 (5.28/100), and 0.071 (7.12/100) percentage point. Models 3 and 4 show that a 1% increase in single-purpose governments (which is the same as creating two new government units) increases single-detached units as a percentage of total housing units by 0.014 to 0.016 point.

A number of interesting results can be observed for the control variables. For example, higher percentage of Blacks and Hispanics is associated with an increase in population density, but higher percentage of the population who are poor leads to higher percentage single-detached housing units. It seems that poverty, more than race and ethnicity, was what drove sprawled development in the 1990s. The results also show that metropolitan areas with more central cities tend to have higher population density, whereas higher income leads to lower percentage single-detached housing units. Finally, as expected, higher per capita highway expenditures, and greater share of jobs in manufacturing lead to a less dense and less expansive development pattern.

Local Government Fragmentation and Spatial Economic Segregation

Tables 3 and 4 show the results for models assessing the relationship between measures of local government fragmentation and spatial poverty and affluence concentration. The dependent variable in Models 5 and 6 is the P* index for poverty, whereas the outcome variable in Models 7 and 8 is the P* index for affluence.

Table 3.

Fragmentation and Spatial Poverty Concentration.

	Model 5: OLS		Model 6: IV
	DV: Poverty isolation index		Stage 1 DV: General-purpose governments per 100,000 people log		Stage 1 DV: Special-purpose governments per 100,000 people log		Stage 2 DV: Poverty isolation index
Independent variables	Coefficient	SE	Coefficient	SE	Coefficient	SE	Coefficient	SE
Fragmentation measures
General-purpose government per 100,000 population log	−0.003	0.002	—	—	—	—	−0.006	0.004
Special-purpose government per 100,000 population log	0.000	0.002	—	—	—	—	0.001	0.002
Control variables
% population Black	0.001***	0.000	0.002	0.003	−0.004	0.006	0.001***	0.000
% population Latino	0.000*	0.000	−0.010**	0.004	0.005	0.006	0.000*	0.000
% population poor	0.011***	0.001	−0.011	0.010	0.001	0.023	0.011***	0.001
Income per capita	0.000	0.000	0.000***	0.000	0.000	0.000	0.000	0.000
% employed in manufacturing	0.000	0.000	0.012**	0.005	0.014*	0.008	0.000	0.000
% employed in management position	−0.002***	0.000	0.020***	0.006	0.008	0.009	−0.002***	0.000
Population (log)	0.014***	0.002	−0.080**	0.030	−0.122**	0.047	0.014***	0.002
New England region	0.030***	0.006	0.591	0.361	0.867**	0.350	0.030***	0.006
Mid-Atlantic region	0.019***	0.006	0.559***	0.197	0.672***	0.239	0.019***	0.006
Great Lakes region	0.025***	0.006	0.546***	0.107	0.394*	0.223	0.026***	0.006
Plains region	0.015***	0.006	0.275**	0.115	0.383*	0.202	0.016***	0.006
South West region	0.012*	0.006	0.100	0.170	0.412	0.288	0.011*	0.006
Mountain region	0.017**	0.007	0.087	0.168	0.819***	0.275	0.014**	0.007
Far West region	−0.014**	0.006	−0.365**	0.176	0.473	0.285	−0.017**	0.007
2000 year dummy	0.025***	0.002	−0.108***	0.031	0.003	0.038	0.025***	0.002
Excluded instruments
Binding TELs	—	—	0.091	0.110	0.176*	0.095	—	—
Debt limits	—	—	0.142	0.098	−0.120	0.202	—	—
1972 General-purpose government per 100,000 population log	—	—	0.133***	0.015	0.040*	0.022	—	—
1972 Special-purpose government per 100,000 population log	—	—	0.004	0.006	0.047***	0.015	—	—
Incorporation limits	—	—	−0.289*	0.158	0.342	0.291	—	—
Constant	−0.063**	0.025	2.009***	0.663	2.197***	0.920	−0.061**	0.025
n	668		668		668		668
OLS adjusted R²/ IV centered R²	.847		.846		.666		.846
F test of excluded instruments F statistic/Prob > F			29.55/0.000		15.97/0.000
Hansen J statistic/χ² p value							0.185/.9800

Significant at 10%.

Significant at 5%.

***

Significant at 1%.

Table 4.

Fragmentation and Spatial Affluence Concentration.

	Model 7: OLS		Model 8: IV
	DV: Affluence isolation index		Stage 1 DV: General-purpose governments per 100,000 people (log)		Stage 1 DV: Special-purpose governments per 100,000 people (log)		Stage 2 DV: Affluence isolation index
Independent variables	Coefficient	SE	Coefficient	SE	Coefficient	SE	Coefficient	SE
Fragmentation measures
General-purpose government per 100,000 population (log)	0.003	0.002	—	—	—	—	0.006*	0.003
Special-purpose government per 100,000 population (log)	−0.002	0.001	—	—	—	—	−0.002	0.003
Control variables
% population Black	0.001***	0.000	−0.003	0.004	−0.004	0.005	0.001***	0.000
% population Latino	0.001**	0.000	−0.012***	0.004	−0.002	0.004	0.001***	0.000
% population poor	0.003***	0.001	−0.004	0.011	0.004	0.019	0.003***	0.001
Income per capita	0.000***	0.000	0.000***	0.000	0.000	0.000	0.000***	0.000
% employed in manufacturing	0.001***	0.000	0.012***	0.004	0.008	0.006	0.001***	0.000
% employed in management position	0.004***	0.000	0.019***	0.006	0.004	0.009	0.004***	0.000
Population (log)	0.022***	0.002	−0.083**	0.032	−0.146***	0.039	0.022***	0.002
New England region	−0.003	0.006	0.272	0.200	0.590**	0.234	−0.005	0.006
Mid-Atlantic region	0.002	0.004	0.555***	0.166	0.624***	0.224	0.000	0.004
Great Lakes region	0.007	0.004	0.778***	0.137	0.529**	0.245	0.005	0.004
Plains region	0.014**	0.005	0.518***	0.165	0.546**	0.237	0.011**	0.005
South West region	0.009	0.007	0.058	0.196	0.544**	0.256	0.009	0.007
Mountain region	0.023**	0.010	−0.012	0.196	0.897***	0.260	0.024**	0.011
Far West region	0.006	0.007	−0.349*	0.194	0.823***	0.253	0.008	0.009
2000 year dummy	−0.011***	0.002	−0.050	0.037	0.032	0.039	−0.011***	0.002
Excluded instruments
Binding TELs	—	—	−0.031	0.086	0.174*	0.094	—	—
Debt limits	—	—	0.234**	0.107	−0.076	0.222	—	—
1972 General-purpose government per 100,000 population (log)	—	—	0.078***	0.020	0.018*	0.011	—	—
1972 Special-purpose government per 100,000 population (log)	—	—	0.005	0.006	0.042***	0.013	—	—
Incorporation limits	—	—	−0.131*	0.071	−0.028	0.136	—	—
Constant	−0.573***	0.029	2.957***	0.654	3.159***	0.833	−0.583***	0.029
n	694		694		694		694
OLS adjusted R²/ IV centered R²	.923		.837		.704		.923
F test of excluded instruments F statistic/Prob > F			10.36/0.000		8.74/0.000
Hansen J statistic/χ² p value							4.878/.1810

Significant at 10%.

Significant at 5%.

***

Significant at 1%.

The results from the OLS models—Models 5 and 7—show that fragmentation measures do not have any statistically significant effects on the spatial isolation of both the poor and rich segments of the population in metro areas. The IV models—Models 6 and 8—show that the instruments are valid. In the first stage, the F test indicates that the instruments are jointly significant. The partial R² show that the excluded instruments explain between 15% and 39% of the variation in general- and special-purpose governments per 100,000 people. In Stage 2, the Hansen J statistic confirms that the instruments are not correlated with the error terms. The findings in the IV models confirm the OLS results, which are contrary to the third and fourth hypotheses.¹³

As for the control variables, higher percentages of Blacks, Hispanics, and the poor lead to increased spatial isolation of the poor and affluent in metropolitan areas. Regions with bigger populations show a higher degree of neighborhood differentiation, whereas higher income increases spatial concentration of rich households. Higher percentage of jobs in manufacturing has no effect on poverty concentration, but positively influences affluence concentration. Finally, an increase in employment in professional or managerial occupations reduces spatial isolation of the poor, but increases spatial separation of the affluent.

Discussion and Conclusion

This study makes four primary contributions to the vast literature exploring the outcomes of the organization of the local public sector. First, it carefully distinguished between the horizontal and vertical dimensions of the local public sector structure and examined how fragmentation in each dimension influenced urban development and economic segregation patterns in metropolitan areas in the US.

Second, this study improved on the empirical approach used in extant studies of urban sprawl and spatial economic segregation. By using data from two waves of the DC and the CoG, this research was able to test for the hypothesized causal relationships between fragmentation and urban development and segregation patterns. It also addressed the issue of endogeneity of measures of local government fragmentation, and thus accurately tested for the effects of the design of the local public sector structure.

Third, the results of the empirical analysis strengthen the argument that a more holistic evaluation of the outcomes of the design of the local governing system is needed. Empirical research within Public Administration tended to focus on the service efficiency effects of local government fragmentation. This research explored a different class of wicked policy problems associated with fragmentation, and showed that fragmentation leads to urban sprawl. The results here, evaluated in light of the findings from other studies of the service-related impacts of fragmentation, suggest that the design of the local public sector structure involves important trade-offs. On one hand, as other research has shown (see Hendrick, Jimenez, & Lal, 2011; Jimenez & Hendrick, 2010, for reviews of the literature), the presence of numerous multi-purpose governments provides choices to citizen-consumers, creates competition among governments, and reduces the cost of services. On the other hand, the empirical analysis here shows that fragmentation of both general- and special-purpose governments can generate negative spillovers specifically urban sprawl. This result is consistent with the findings of Carruthers and Ulfarsson (2002, 2006), Carruthers (2003), and Dye and McGuire (2000).

Finally, accounting for endogeneity of the fragmentation measures, the analysis here provides no evidence that fragmentation of either the horizontal or vertical dimensions of the local public sector structure leads to the spatial separation of economic classes in metropolitan areas. Given the strong findings in the literature about the positive relationship between government fragmentation and racial segregation (see Bischoff, 2008; J. H. Lewis & Hamilton, 2011; Morgan & Mareschal, 1999; among others), the results here pointing to the absence of any statistical relationship between fragmentation and income-based segregation suggest that it is race rather than social class that drives local population sorting.¹⁴

However, it is important to point out that spatial economic segregation was measured at the neighborhood level. The segregation measures used here do not provide any information on whether people are sorting themselves into distinct governmental jurisdictions based on social class. Future research will need to examine how fragmentation influences economic segregation occurring at governmental boundaries. In addition, other studies used more complex measures of the public sector structure. Hendrick, Jimenez, & Lal (2011), for example, measured total, horizontal and vertical fragmentation. They also distinguished between fragmentation (the count of governments) and fiscal dispersion (the distribution of expenditure or revenue responsibilities across governments), and developed measures for specific types of local governments. Unfortunately, it was not possible to use all of these measures in this study because of the difficulty of finding suitable instruments to address the issue of simultaneous causation.¹⁵

The existence of trade-offs among the outcomes of local government fragmentation is an important finding that hopefully can inform policymakers and elected officials engaged in the process of reforming their local public sector. This is true not only in the case of the United States but also for other countries. Some nations in Europe, for example, are currently exploring mandatory amalgamation of their local governments to reduce local public spending.¹⁶ Other countries in Asia, in contrast, are pursuing “big-bang” decentralization and are in the process of creating more local governments to improve responsiveness to local service needs.¹⁷ The results of this research indicate that local public sector reforms might lead to policy outcomes other than what they were intended to achieve.

The question that arises, and which is of interest especially to Public Administration scholars, is what institutional design or policy choices are available that maximize consumer choice and sovereignty and promote efficient and responsive delivery of services, but can effectively address the spillover effects of residential sorting? Is wholesale restructuring of the local governing system the only choice? If not, how can local governments be encouraged to address what are clearly regional, as opposed to purely local, problems?

Microeconomics offers two traditional responses to externalities. One is to force the parties involved in the economic transaction to internalize or pay for the cause of the harm they impose on others through taxation. An equivalent policy in terms of controlling urban sprawl is to introduce development impact fees. Nevertheless, some question the efficacy of such fees in controlling sprawl (Downs, 1999). A second option is to impose regulations on the production of the good that caused the externality. Regulatory responses for controlling sprawl include growth management policies in which state governments exercise their preemptive authority over local land use decisions. Studies have shown that well-designed growth management programs have the potential to address urban sprawl (Carruthers, 2002).

Perhaps more difficult to address is the issue of poverty concentration. Some solutions that have been implemented include policies that address access to affordable housing and transportation, as well as programs that promote regional economic development, which are mainly funded by the federal government (Downs, 1999). Some insist on the importance of inter-local cooperation (Feiock, 2007). One example of a regional initiative to address fiscal disparities across jurisdictions is tax-base sharing (Orfield, 1997). Still, others point out that coordination among local governments to address poverty concentration has been sorely lacking, and that most inter-local collaborative arrangements have focused on systems maintenance issues (such as sewerage management or emergency services) rather than redistributive issues (Jimenez, 2014a; Lowery, 2000).

It is clear that addressing wicked governance problems requires complex and multi-level governance arrangements. More research is needed to understand how different institutional and policy solutions perform in comparison with each other in correcting market failures and related multi-jurisdictional issues in the fragmented metropolis.

Footnotes

Appendix

Table A1.

Descriptive Statistics.

Variable	Data source	M	SD	Minimum	Maximum
Urban sprawl measures
Population density	DC	70.06	15.70	0.39	99.41
Single detached units as percentage of all housing units	DC	63.59	7.85	35.82	82.92
Spatial economic segregation measures
Spatial poverty concentration	DC	0.21	0.06	0.08	0.43
Spatial affluence concentration	DC	0.17	0.07	0.04	0.52
Fragmentation/consolidation measures
General-purpose government per 100,000 population	CoG	11.89	11.66	0.11	77.52
Special-purpose government per 100,000 population	CoG	17.65	15.19	0.36	91.76
Control variables
Population (in 1,000)	DC	600	1,407	40	18,300
% population Black	DC	9.88	10.42	0.03	48.52
% population Latino	DC	7.88	13.49	0.21	94.40
% population poor	DC	12.70	4.52	3.97	41.51
Number of central cities in MSA	CoG	1.45	1.31	0.00	10.00
Highway direct expenditures per capita (in US$1,000)	CoG	0.13	0.07	0.01	0.59
Sewerage direct expenditures per capita (in US$1,000)	CoG	0.09	0.06	0.00	0.45
Total tax revenues per capita (in US$1,000)	CoG	0.95	0.35	0.10	2.72
Income per capita	DC	18,794	3,396	8,935	38,350
% employed in manufacturing	DC	15.95	7.48	2.08	44.84
% employed in management position	DC	29.96	5.06	18.70	50.22
Instruments
Binding tax and spending limits (TELs)	ACIR	0.72	0.45	0.00	1.00
Debt limits	ACIR	0.92	0.27	0.00	1.00
Incorporation limits	ACIR	0.90	0.30	0.00	1.00
1972 General-purpose government per 100,000 population (log)	CoG	5.77	4.90	0.22	31.06
1972 Special-purpose government per 100,000 population (log)	CoG	11.92	10.35	0.45	98.24

Note. DC = decennial census (1990, 2000); CoG = census of governments (1992, 2002); MSA = metropolitan statistical areas; ACIR = advisory commission on intergovernmental relations (1993).

Declaration of Conflicting Interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Benedict S. Jimenez is is the recipient of the 2014 Clarence N. Stone Scholar Award and the 2013 Paul A. Volcker Junior Scholar Award from the American Political Science Association. His research focuses on urban public finance and management. His research has been published in in the Journal of Public Administration Research and Theory, Urban Affairs Review, and Public Administration Review, among others.

References

Aaronson

(1998). Using sibling data to estimate the impact of neighborhoods on children’s educational outcomes. Journal of Human Resources, 33, 915-946.

Altshuler

Morrill

Wolman

Mitchell

(Eds.). (1999). Governance and opportunity in metropolitan America. Washington, DC: National Academy Press.

Baltagi

B. H.

(2002). Econometrics. New York, NY: Springer.

Berry

(2008). Piling on: The fiscal effects of jurisdictional overlap. American Journal of Political Science, 52, 802-820.

Bischoff

(2008). School district fragmentation and racial residential segregation: How do boundaries matter? Urban Affairs Review, 44, 182-217.

Bollens

S. A.

(1990). Constituencies for limitation and regionalism: Approaches to growth management. Urban Affairs Review, 26, 46-67.

Brennan

Buchanan

(1980). The power to tax: Analytical foundations of a fiscal constitution. Cambridge, UK: Cambridge University Press.

Brueckner

J. K.

(2000). Urban sprawl: Diagnosis and remedies. International Regional Science Review, 23, 160-171.

Burns

(1994). The formation of American local governments: Private values in public institutions. New York, NY: Oxford University Press.

10.

Carruthers

J. I.

(2002). The impacts of state growth management programmes: A comparative analysis. Urban Studies, 39, 1959-1982.

11.

Carruthers

J. I.

(2003). Growth at the fringe: The impact of political fragmentation in United States metropolitan areas. Papers in Regional Science, 82, 475-499.

12.

Carruthers

J. I.

Ulfarsson

G. F.

(2002). Fragmentation and sprawl: Evidence from interregional analysis. Growth and Change, 33, 312-340.

13.

Coulton

C. J.

Chow

Wang

E. C.

(1996). Geographic concentration of affluence and poverty in 100 metropolitan areas, 1990. Urban Affairs Review 32, 186-216.

14.

Cutler

D. M.

Glaeser

(1997). Are ghettos good or bad. Journal of Economics, 112, 827-872.

15.

DeHoog

R. H.

Lowery

Lyons

W. E.

(1990). Citizen satisfaction with local governance: A test of individual, jurisdictional, and city specific explanations. Journal of Politics, 52, 807-837.

16.

Dowding

John

Biggs

(1994). Tiebout: A survey of the empirical literature. Urban Studies, 31, 767-797.

17.

Downs

(1994). New visions for metropolitan America. Washington, DC: Brookings

18.

Downs

(1999). Some realities about sprawl and urban decline. Housing Policy Debate, 10, 955-974.

19.

Dye

R. F.

McGuire

T. J.

(2000, May 29). Property taxes, schools, and sprawl. State Tax Notes, 18, 1899-1908.

20.

Ewing

Pendall

Chen

(2002). Measuring sprawl and its impact. Washington DC: Smart Growth America.

21.

Feiock

R. C.

(2007). Rational choice and regional governance. Journal of Urban Affairs, 29, 47-62.

22.

Firman

(2013). Territorial splits (Pemekaran Daerah) in decentralising Indonesia, 2000-2012: Local development drivers or hindrance? Space & Polity, 17, 180-196.

23.

Friedman

(1962). Capitalism and freedom, Chicago, IL: University of Chicago Press.

24.

Gruber

(2013). Public finance and public policy (4th ed.). New York, NY: Worth.

25.

Hamilton

Miller

Paytas

(2004). Exploring the horizontal and vertical dimensions of the governing of metropolitan regions. Urban Affairs Review, 40, 147-182.

26.

Heikkila

E. J.

(1996). Are municipalities Tieboutian clubs? Regional Science & Urban Economics, 26, 203-226.

27.

Hendrick

Jimenez

B. S.

Lal

(2011). Does local government fragmentation affect local spending? Urban Affairs Review, 47, 467-510.

28.

Hill

R. C.

(1974). Separate and unequal: Governmental inequality in the metropolis. American Political Science Review, 68, 1557-1568.

29.

Howell-Moroney

(2008). The Tiebout hypothesis 50 years later: Lessons and lingering challenges for metropolitan governance in the 21st century. Public Administration Review, 68, 97-109.

30.

Jargowski

P. A.

(1997). Take the money and run: Economic segregation in US metropolitan areas. American Sociological Review, 61, 984-998.

31.

Jimenez

B. S.

(2014a). Separate, unequal and ignored? Inter-jurisdictional competition and the budgetary choices of poor and affluent municipalities. Public Administration Review, 74, 246-257.

32.

Jimenez

B. S.

(2014b). The fiscal performance of multi-level local governments. Public Finance Review, 43, 606-635.

33.

Jimenez

B. S.

Hendrick

(2010). Is government consolidation the answer? State and Local Government Review, 42, 258-270.

34.

Kain

J. F.

(1992). The spatial mismatch hypothesis: Three decades later. Housing Policy Debate, 3, 375-460.

35.

Krane

Rigos

Hill

Jr. (2001). Home rule in America: A fifty-state handbook. Washington, DC: CQ Press.

36.

Lewis

J. H.

Hamilton

D. K.

(2011). Race and regionalism: The structure of local government and racial disparity. Urban Affairs Review, 47, 349-384.

37.

Lewis

P. G.

(1996). Shaping suburbia: How political institutions organize urban development. Pittsburgh, PA: University of Pittsburgh.

38.

Lieberson

(1980). A piece of the pie: Blacks and White immigrants, 1880-1930. Berkeley: University of California Press.

39.

Logan

J. R.

Schneider

(1982). Governmental organizations and city/suburb income inequality, 1960-1970. Urban Affairs Quarterly, 17, 303-318.

40.

Lowery

(2000). A transactions costs model of metropolitan governance: Allocation versus redistribution. Journal of Public Administration Research and Theory, 10, 49-78.

41.

Margo

(1992). Explaining the post-war suburbanization of the population in the United States: The role of income. Journal of Urban Economics, 31, 301-310.

42.

Massey

D. S.

Denton

N. A.

(1993). American apartheid: Segregation and the making of the underclass. Cambridge, MA: Harvard University Press

43.

Massey

D.S.

Eggers

M.L.

(1990) The ecology of inequality: Minorities and the concentration of poverty, 1970-1980. American Journal of Sociology 95, 1153-88.

44.

McGuire

T. J.

Sjoquist

D. L.

(2003). Urban sprawl and the finances of state and local governments. In Sjoquist

D. L.

(Ed.), State and local finances under pressure (pp. 299-326). Northampton, MA: E. Elgar.

45.

Miller

G. J.

(1981). Cities by contract. Cambridge, MA: MIT Press.

46.

Morgan

D. R.

Mareschal

(1999). Central-city/suburban inequality and metropolitan political fragmentation. Urban Affairs Review, 34, 578-595.

47.

Nelson

A. C.

Duncan

J. B.

(1995). Growth management principles and practices. Chicago, IL: Planners Press.

48.

Orfield

(1997). American metropolitics: The new suburban reality. Washington, DC: Brookings Institution Press.

49.

Ostrom

(1976). The delivery of urban services. Beverly Hills, CA: SAGE.

50.

Ostrom

(1983). The social stratification-government inequality thesis explored. Urban Affairs Quarterly, 19, 91-112.

51.

Ostrom

Tiebout

Warren

(1961). Organization of government in metropolitan areas: A theoretical inquiry. American Political Science Review, 55, 831-842.

52.

Pendall

(1999). Do land use control cause sprawl? Environment and Planning B: Planning and Design 26, 555-71.

53.

Post

S. S.

Stein

R. M.

(2000). State economies, metropolitan governance, and urban-suburban economic dependence. Urban Affairs Review, 36, 46-60.

54.

Razin

Rosentraub

(2000). Are fragmentation and sprawl interlinked? North American evidence. Urban Affairs Review, 35, 821-836.

55.

Savitch

H. V.

Vogel

R. K.

(2000). Paths to new regionalism. State and Local Government Review, 32, 158-168.

56.

Schneider

(1989). The competitive city: The political economy of suburbia. Pittsburgh, PA: University of Pittsburgh Press.

57.

Squires

G. D.

Kubrin

C. E.

(2005). Privileged places: Race, uneven development and the geography of opportunity in urban America. Urban Studies, 42, 47-68.

58.

Stearns

L. B.

Logan

J. R.

(1986). The racial structuring of the housing market and segregation in suburban areas. Social Forces, 65, 28-42.

59.

Stein

R. M.

(1987). Tiebout’s sorting hypothesis. Urban Affairs Quarterly, 23, 140-166.

60.

Stephens

G. R.

Wikstrom

(2000). Metropolitan government and governance: Theoretical perspectives, empirical analysis, and future. New York, NY: Oxford University Press.

61.

Tiebout

(1956). A pure theory of local expenditures. Journal of Political Economy, 44, 416-424.

62.

Weiher

G. R.

(1991). The fractured metropolis: Political fragmentation and metropolitan segregation. Albany: State University of New York Press.

63.

Wooldridge

J. M.

(2002). Econometric analysis of cross-section and panel data. Cambridge, MA: MIT Press.