Spatial Stratification Within U.S. Metropolitan Areas

Ingrid Gould Ellen

New York University

Nearly 80 percent of the U.S. population lives in metropolitan areas. To say this, however, may be to obscure the enormous range of environments found within these metropolitan areas. Typically, they are divided into political jurisdictions and neighborhoods that are highly segregated by class, race, ethnicity, and income and even marked by vastly different physical conditions. In most metropolitan areas, central cities and older, inner-ring suburbs tend to have lower-skilled and less affluent populations, lower tax bases, as well as more deteriorated housing stocks and infrastructures, than their newer, outer-ring suburban neighbors. And the segregation becomes even more apparent if comparisons are made across individual neighborhoods within these jurisdictions.

What makes this especially troubling is the growing body of evidence that these geographic disparities may have serious consequences. Many recent studies suggest that children growing up in deprived communities face substantially greater obstacles in obtaining a sound education, securing a steady job, and otherwise advancing their status than their counterparts in more prosperous environments (Ellen and Turner, 1997; Ihlanfeldt, this volume; Mayer, 1996). Significantly, a few of these studies also suggest that the costs of poverty concentration reach far beyond those poor neighborhoods themselves—that the deprivations of South Bronx and Anacostia resonate in Scarsdale and Chevy Chase. There is some evidence, that is, that certain neighborhood effects are nonlinear, and that a deconcentration of the poor may thus lead to decreased levels of poverty and its consequent problems not only in distressed neighborhoods but also in society overall (Crane, 1991; Hogan and Kitagawa, 1985).

Before any meaningful assessment of the nature and extent of these conse-



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 192
Spatial Stratification Within U.S. Metropolitan Areas Ingrid Gould Ellen New York University Nearly 80 percent of the U.S. population lives in metropolitan areas. To say this, however, may be to obscure the enormous range of environments found within these metropolitan areas. Typically, they are divided into political jurisdictions and neighborhoods that are highly segregated by class, race, ethnicity, and income and even marked by vastly different physical conditions. In most metropolitan areas, central cities and older, inner-ring suburbs tend to have lower-skilled and less affluent populations, lower tax bases, as well as more deteriorated housing stocks and infrastructures, than their newer, outer-ring suburban neighbors. And the segregation becomes even more apparent if comparisons are made across individual neighborhoods within these jurisdictions. What makes this especially troubling is the growing body of evidence that these geographic disparities may have serious consequences. Many recent studies suggest that children growing up in deprived communities face substantially greater obstacles in obtaining a sound education, securing a steady job, and otherwise advancing their status than their counterparts in more prosperous environments (Ellen and Turner, 1997; Ihlanfeldt, this volume; Mayer, 1996). Significantly, a few of these studies also suggest that the costs of poverty concentration reach far beyond those poor neighborhoods themselves—that the deprivations of South Bronx and Anacostia resonate in Scarsdale and Chevy Chase. There is some evidence, that is, that certain neighborhood effects are nonlinear, and that a deconcentration of the poor may thus lead to decreased levels of poverty and its consequent problems not only in distressed neighborhoods but also in society overall (Crane, 1991; Hogan and Kitagawa, 1985). Before any meaningful assessment of the nature and extent of these conse-

OCR for page 192
quences can take place, however, it is critical to provide a clearer sense of the nature and extent of the spatial disparities themselves. This paper aims to do just this. The first section, in order to set the stage, documents the magnitude of the spatial and jurisdictional disparities within the average metropolitan area and determines how these have changed in recent years. Many researchers go no further and thus overlook the surprising diversity found across different metropolitan areas in the magnitude of disparities. This paper, however, makes this variation its central concern. To this end, the second section classifies metropolitan areas on the basis of the magnitude of their central-city-suburban disparities and identifies certain metropolitan-area characteristics (such as population size, the degree of racial segregation, and the elasticity of the central-city boundaries) that are correlated with greater and lesser disparities. The third section then estimates a simple, cross-sectional regression that tests which, if any, of these correlations persist after controlling for other factors. Although more definitive conclusions regarding the precise causes of the jurisdictional disparities would be desirable, they would require further statistical analysis that lies outside the scope of this particular project. The Magnitude of Spatial Disparities Within Metropolitan Areas City Versus Suburb Most Americans voice a preference for owning a detached, single-family home and view suburban communities as promising better schools, lower crime rates, and less crowding and traffic (Fannie Mae, 1997). Not surprisingly, then, the population of the suburbs continues to swell. In 1940, just over 15 percent of the U.S. population lived in suburban communities. Fifty years later; that proportion had risen to 46 percent (U.S. Bureau of the Census, 1991). As shown in Table 1, however, not everyone is gaining access to the suburban life. Poor and minority households continue to live largely in inner-city communities. In 1990, 57 percent of the black population and 52 percent of the Hispanic population TABLE 1 Breakdown of Residence by Race and Ethnicity, 1990 (in percent) Residence Total Population Non-Hispanic Whites Blacks Hispanics Metropolitan Areas 77.5 74.7 83.8 90.4 Central Cities 31.3 24.5 57.3 51.5 Suburbs 46.2 50.3 26.4 37.3 Nonmetropolitan 22.5 25.3 16.2 9.6   Source: U.S. Bureau of the Census (1990b).

OCR for page 192
lived in central cities.1 By contrast, less than one-fourth of non-Hispanic white households lived in central cities. The remainder of this section highlights some average differences between central cities and suburbs and examines these differences over time. For the purposes of this paper, attributes of a given metropolitan area's suburbs are viewed in the aggregate, for even though there is substantial variation across individual suburbs in a given metropolitan area, the available census data treat suburbs as a single entity. Nonetheless, neighborhood data are briefly used to reveal some interesting variations across suburban jurisdictions. The disparities presented in this section may be grouped into three categories: income disparities (median household income levels, per capita income levels, and poverty rates); labor market disparities (employment and unemployment rates for working-age men); and educational disparities (high school and college completion rates). These simple measures do not, of course, fully capture the differences between central-city and suburban residents, but they may provide some sense of the social and economic contrasts. As shown in Table 2, central-city residents had incomes in 1990 that were considerably lower than those of suburbanites. Compared with their suburban counterparts, the median household income of central-city residents was 74 percent as great, their per capita income was 84 percent as large, and their poverty rate was over twice as high. Disparities in employment and education were somewhat smaller, but central-city residents appear to be more deprived in these realms as well. In particular, compared with suburban communities, 1.7 times as great a proportion of prime-age working men were unemployed in central cities and 94 percent as large a proportion of young men (ages 25-34) had completed high school. Given the differing racial compositions of central cities and suburbs, much TABLE 2 Comparison of Selected Outcomes for Central-City and Suburban Residents, 1990 Outcome Central City Non-Central City Ratio of Central-City to Suburban Outcome Median household income $26,727 $36,314 0.74 Per capita income $13,839 $16,527 0.84 Poverty rate, families 14.1% 6.0% 2.35 Employed,a men 25-54 83.3% 90.0% 0.93 Unemployed, men 25-54 6.8 4.0 1.7 High school graduates, men 25-34 80.7% 85.8% 0.94 a Members of the armed forces are counted as employed. Source: U.S. Bureau of the Census (1990b).

OCR for page 192
of these spatial differences may be largely a reflection of racial differences in income, employment, and education. As shown in Table 3, however, this does not appear to be the case. The group of non-Hispanic whites, blacks, and Hispanics living in the suburbs are all more prosperous than their counterparts in central-city communities. The differences between central-city and suburban minorities—especially blacks—are particularly pronounced. Black households living in central cities, for instance, took home just 68 cents on every dollar brought home by suburban black households. In other words, minorities living in central cities appear to be particularly disadvantaged. This may be because white households of modest means are able to move to suburban areas that are not welcoming to blacks. For blacks and Hispanics to be welcomed to the suburbs, they may need to be more solidly middle-class. As mentioned above, these comparisons lump together all suburbs as a single entity. Yet as Myron Orfield has pointed out in his analysis of the Twin Cities, suburbia is hardly monolithic (Orfield, 1997). The stereotypical vision of suburbia consists of quiet, affluent, residential neighborhoods made up of single-family detached homes with ample and manicured lawns. Such places of course exist, but they are not necessarily synonymous with the official Census Bureau definition of a suburb (a community within a metropolitan area but outside a central city). Many suburbs today are populated by working-class residents and suffer many of the same social and economic ills found in inner-city communities. In fact, such suburbs have existed for decades. As early as 1960, Bennett Berger challenged the conventional understanding of suburbia in his book Working Class Suburb. Unfortunately, despite the widespread acknowledgment that the term ''suburb'' covers a wide range of metropolitan experience, the Census Bureau offers few alternatives to improve upon it. To provide some sense of the diversity in suburban America, we have exam- TABLE 3 Ratio of Selected Outcomes for Central City and Suburban Residents by Race and Ethnicity, 1990   Ratio of Central-City to Suburban Outcome Outcome All Persons Non-Hispanic Whites Blacks Hispanics Median household income 0.74 0.81 0.68 0.73 Per capita income 0.84 0.98 0.8 0.81 Poverty rate, families 2.35 1.66 1.66 1.63 Labor force participation, 0.95 0.96 0.9 0.95 Unemployment rate, 1.56 1.18 1.52 1.27 % high school graduates, 0.92 0.98 0.89 0.85 % college graduates, 0.96 1.14 0.7 0.78   Source: U.S. Bureau of the Census (1990b).

OCR for page 192
ined 1990 neighborhood, or census tract, data within suburban areas. As noted above, the median household income in suburban America was significantly larger than that in central cities ($36,314 compared with $26,727). But not all suburban communities were so fortunate. In approximately one-fifth of suburban neighborhoods, the median household income was actually less than the overall central-city median. (And without the commercial tax base of a central business district, these suburbs were arguably even worse off than their central cities.) The story is similar with the poverty rate. Nationwide in 1990, the poverty rate was 8.1 percent in the suburbs and 18 percent in central cities. But again, high-poverty suburban communities do exist. Approximately 10 percent of suburban census tracts had poverty rates of at least 18 percent. Although we cannot say for sure, it is likely that these are the aging, inner-ring suburbs that Orfield (1997) describes. Perhaps more significant than the magnitude of these disparities are their trends over time. What has happened to the central-city/suburban gap over recent decades? On the one hand, suburbanization has continued at a rapid pace, presumably leaving behind the increasingly disadvantaged. On the other hand, many inner-suburbs have been "urbanized" (Lineberry, 1975; Masotti and Hadden, 1973; Orfield, 1997; Sternlieb and Lake, 1975). Table 4 shows, for selected measures, the ratios of central-city to suburban outcomes for 1960 through 1990. With the exception of education, every measure in the table suggests that the relative status of central-city residents has consistently declined over the last three decades. Relative to their suburban TABLE 4 Ratios of Selected Outcomes for Central City and Suburban Residents, 1960-1990   Ratio of Central-City to Suburban Outcome Outcome 1960 1970 1980 1990 Median family income .89 .85 .81 .77 Per capita income NA .92 .88 .84 Family poverty rate 1.52 1.75 2.08 2.35 % employed, 0.96 0.93 0.92 .91 Unemployment rate, 1.37 1.33 1.34 1.59 % high school graduates, 0.85 0.86 0.9 .94 % college graduates, 0.82 0.84 0.94 .96 Note: The number of metropolitan areas differs somewhat from one decade to another, as more metropolitan areas have been defined. In other words, the table is not limited to just those metropolitan areas that existed in all four decades. NA = not available a For 1960, the employment and unemployment rates correspond to men at least 14 years of age. Sources: U.S. Bureau of the Census (1960, 1970, 1980, 1990b).

OCR for page 192
counterparts, that is, central-city residents now have lower incomes, higher poverty rates, and lower employment rates than they did 10, 20, and 30 years ago.2 The income differences are particularly stark. In 1960, central-city families were taking home 89 cents on every dollar brought home by suburbanites. By 1990, this ratio had fallen to just .77. Per capita income follows a similar pattern, with ratios falling from .92 in 1970 to .84 in 1990. Finally, poverty is becoming increasingly concentrated in central-city areas. In 1960, poverty was 1.5 times as frequent for families living in central cities; by 1990, the ratio had risen to 2.35. The differences in employment and unemployment ratios are not as dramatic, but the direction is the same—that is, the gap between central-city and suburban residents is growing consistently larger. And, as Ihlanfeldt shows in his paper in this volume, these central city-suburban disparities have grown much more rapidly for blacks than they have for the population as a whole (Ihlanfeldt, this volume). As noted, education is a marked exception to the general trend of continued central-city decline. The overall rise in educational attainment appears to have led to some convergence between central-city and suburban residents. Because it is easier to increase the proportion of high school and college graduates in a population that has fewer of them to start with, the relative fortunes of central-city residents improved as a result of the overall societal progress. Between 1960 and 1990, for instance, while the high school graduation rate rose by 66 percent in non-central portions of metropolitan areas, it rose by 79 percent in central cities. Neighborhood Poverty To compare central cities and suburbs is hardly to capture all of the segregation in metropolitan areas. To understand its full extent, it is necessary to examine class and racial segregation at the level of the neighborhood, or even the block, as Paul Jargowsky does in his recent analysis (1997). Looking at neighborhood poverty, he forcefully documents that concentrated poverty is growing more prevalent. According to his analysis, 17.9 percent of all poor persons in 1990 lived in high-poverty areas (census tracts in which the poverty rate is greater than or equal to 40 percent). By comparison, the proportions were just 13.6 percent in 1980 and 12.4 percent in 1970. In other words, the proportion of poor people living in high-poverty areas rose by 98 percent over the 20 years between 1970 and 1990. Another analysis of trends in economic segregation between 1970 and 1990 arrives at a similar, albeit less dramatic, conclusion. In contrast to Jargowsky, Abramson et al. (1995) study the overall segregation of the poor, not simply the prevalence of high-poverty areas. Using the index of dissimilarity, a measure that indicates the proportion of a particular group of households (here, the poor) who would have to move in order to achieve an even distribution of the group throughout the metropolitan area, they avoid the need to set an arbitrary defini-

OCR for page 192
TABLE 5 Segregation of the Poor in the Central Cities and Suburbs of Chicago, Los Angeles, and Washington, D.C.   Poor-Nonpoor Dissimilarity Index City Central City Non-Central City Chicago 40.7 32.1 Los Angeles 34.9 33.8 Washington, D.C. 36.2 31.2   Source: Analysis of Urban Institute Underclass Database. tion of a high-poverty tract, such as 20 or 40 percent poor. Nonetheless, they also find increasing segregation, with the dissimilarity of the poor in the 100 largest metropolitan areas rising by 11 percent between 1970 and 1990, from 32.9 to 36.4. Like poverty itself, high-poverty neighborhoods appear to be primarily a central-city phenomenon. In 1990, 94 percent of all high-poverty neighborhoods (using Jargowsky's definition of at least a 40 percent poverty rate) were located in central-city communities. (Overall, 52 percent of census tracts were located in central-city jurisdictions.) Most of this difference is of course due to the higher overall rates of poverty in central-city areas. Is it also true that poverty households are more segregated in central cities? To test this hypothesis, three metropolitan areas are examined: Chicago, Los Angeles, and Washington, D.C. As shown in Table 5, the poor were indeed somewhat less segregated in the suburbs in 1990 than they were in central cities of all three of these metropolitan areas. In Chicago, the central-city dissimilarity index was 40.7 compared with 32.1 in the suburbs, and in Washington, D.C., the dissimilarity index was 36.2 in the central city, compared with 31.2 in the suburbs. Interestingly, the differential was far smaller in Los Angeles, a metropolitan area in which the central city is nearly as prosperous as its suburbs. But more research into the consistency and causes of these central city-suburban differentials in economic segregation is needed before arriving at any firm conclusions. Neighborhood Racial Segregation The segregation of the poor seems clearly to have increased over the past two decades, but the segregation of whites and blacks appears to have declined. Massey and Denton, for instance, examine the 30 metropolitan areas with the largest black populations, and their figures suggest that the black-white index of dissimilarity declined from 80.8 in 1970 to 73.3 in 1990, for a decline of 9 percent (Massey and Denton, 1993).3 As for the shifts over the 1980s, Jargowsky (1997) reports that the average dissimilarity index between blacks and non-Hispanic whites in the full set of metropolitan areas fell from 70 to 66. Farley and Frey (1994) limit their analysis to the 232 metropolitan areas with substantial black

OCR for page 192
populations and uncover a similar decline. Specifically, they estimate that the average index of dissimilarity in these metropolitan areas fell from 69 in 1980 to 65 in 1990. Nonetheless, the level of racial segregation continues to be quite high, and far higher than the segregation of the poor. Examining the 100 largest metropolitan areas, Abramson et al. (1995) find that the average dissimilarity index for the poor was 36.1 in 1990, and the mean dissimilarity index for blacks was 60.6. Classifying Metropolitan Areas As discussed above, these national averages conceal considerable variation across individual metropolitan areas. Certain central cities, for example, appear to fare far better than others relative to the suburbs surrounding them, and rates of neighborhood poverty vary markedly, too. It is these differences that this section examines. Data and Method The analysis here ranks the full set of metropolitan areas nationwide by the degree of their central-city/suburban disparity. To perform this ranking, an index of disparity is calculated from the central-city to suburban ratio of three outcomes: per capita income, employment rate for men between 16 and 64, and the proportion of persons 25 or older who have at least a high school diploma. In order that each of the three measures be weighted equally, they are each standardized to a scale of 0-1 before being averaged to create an overall index of disparity.4 The higher the number, the better off the central city relative to its suburban counterparts. Significantly and contrary to the conventional wisdom, not all central cities are worse off than their surrounding suburbs. In 1990, central cities were actually more prosperous than their surrounding suburban communities in 101 metropolitan areas (31 percent). With this said, because these metropolitan areas tend to be fairly small, they housed only 14 percent of the total metropolitan population. Thus, the vast majority of metropolitan residents live in areas in which the cities are worse off than their surrounding suburbs. Although the analysis in the rest of the paper focuses on central-city/suburban disparities, these jurisdictional inequalities are in fact quite correlated with neighborhood-level economic segregation. For example, in 95 metropolitan areas for which data are available, the simple correlation between the neighborhood-level dissimilarity index of the poor and the index of central-city/suburban disparity was a highly significant-.69. Thus, many of the findings below may apply more generally to the extent of neighborhood-level inequalities in metropolitan areas. As explained above, the general purpose in classifying metropolitan areas by their degree of central-city disadvantage is to further our understanding about the

OCR for page 192
causes and contributors to spatial and jurisdictional disparities. To be sure, the causes of these disparities are complex, and making any definite conclusions about causation would require certain data and data analysis that lie outside the scope of this paper. For example, even where a strong correlation between a certain variable and city-suburban disparity can be demonstrated using the present data, the direction of causality cannot be. Nonetheless, determining which characteristics of metropolitan areas are correlated with greater and lesser disadvantage should enrich our understanding considerably. Toward this end, the remainder of this section hypothesizes a number of possible correlates of greater city-suburban disparity and then presents a simple comparison of means to test these possible correlates. Hypotheses There are several possible reasons why certain metropolitan areas might have greater central-city/suburban disparities than others. This paper considers the possible significance of four types of metropolitan attributes: regional location, demographic characteristics, racial segregation, and governmental structure. Future work might also examine the importance of density, industrial structure, and labor market conditions.5 Regional Location One might expect the degree of central-city disadvantage to vary across regions because, among other things, regional location serves as a fairly good proxy for the age of a metropolitan area. Areas in the Northeast and the Midwest, for example, are much older than those in the South and the West. And one might further expect that age, in turn, correlates with central-city disadvantage because, as Massey and Denton (1987) and Farley and Frey (1994) have found, older cities tend to be more racially segregated; it is quite possible that they are more economically stratified as well. As Massey and Denton note, "[c]ities built up before the Second World War have ecological structures that are more conducive to segregation, with densely settled cores and thickly packed working-class neighborhoods" (1987:818). In addition, Hill and Wolman (1997) argue that central cities in younger metropolitan areas may have housing stocks that better match the demands of current consumers and thus that are better able to attract more affluent households.6 Different regions also have different local economies that have different consequences for city-suburb disparity. For example, manufacturing jobs have declined disproportionately in the central cities of the Northeast and the Midwest, and thus it may well be that the residents of these central cities have been disproportionately disadvantaged and isolated by such decline. Regions also differ in their overall rates of economic growth, which some have hypothesized should be

OCR for page 192
correlated with lesser disparity as well, the logic being that growth may attract less-skilled, central-city residents into the labor market (Hill and Wolman, 1997). Another significant regional pattern lies in governmental policies and structure. One example of regional policy differences is that metropolitan areas in the Sun Belt tend to rely less on the property tax as a source of revenue, and thus local suburban communities have less motivation to exclude low-income residents (Bollens, 1986). As for regional differences in governmental structure, metropolitan areas in the West and the South tend to have less fragmented governments; that is, they have fewer local governmental entities, with each entity housing a greater number of people. Consider that, in 1997, the mean number of local governments per 100,000 residents was 14.7 and 12.5 in the West and the South, respectively, compared with 27.2 and 19.9 in the Midwest and the Northeast (Census of Governments, preliminary 1997 estimates). Second, metropolitan areas in the West and the South tend to have what David Rusk has called more "elastic" central cities (Rusk, 1993). Their cities, that is, have been able to grow or perhaps to annex more of their surrounding suburbs and thus to encompass a greater share of the overall metropolitan population. The precise manner in which these distinctive characteristics of governmental structure might impact city-suburb disparity is discussed below. Finally, regions may differ as well in their racial attitudes, which are plainly linked to some extent to levels of suburban exclusion. After all, regions do differ in their degree of racial segregation. And it would seem that, as suggested above and discussed in more detail below, the more segregated a metropolitan area, the greater the city-suburb disparity. Demographics The second category of variables one might expect to correlate with central-city disadvantage are demographic, variables such as metropolitan area size, size of the minority population, and per capita income in the metropolitan area as a whole. In terms of metropolitan area size, it may be that central-city/suburban disparities are greater in larger metropolitan areas simply because of the larger geographic distances involved. If, for instance, some of the central-city/suburban difference is due to a spatial mismatch between suburban jobs and central-city residents, longer physical distances should exacerbate the city's relative disadvantage. It is also likely that political units will be more specialized and competitive in larger, more populous metropolitan areas, leading to greater disparities. The size of the minority population is examined as well; to the extent that racial considerations contributed to the exclusionary policies of many suburbs, metropolitan areas with larger minority (especially black) populations are likely to have more disadvantaged central cities. Furthermore, to the extent that minorities are highly segregated from whites and also much worse off, metropolitan areas with larger minority populations are likely to be more economically segre-

OCR for page 192
gated in general. (Abramson et al., 1995, and Jargowsky, 1997, both find some evidence that the poor are more segregated in metropolitan areas with larger black populations.) As for the overall income in the metropolitan area, there are reasons to believe that it may be related to the magnitude of city-suburban disparity. In particular, if metropolitan-area income is driven largely by the prosperity of high-end, and typically suburban, households, then we would expect city-suburban income disparities to be greater in areas with higher overall incomes. It may be, that is, that central-city residents fare roughly the same across most metropolitan areas, whereas suburban prosperity varies markedly. And indeed, there is some evidence to this effect: the variances of the measures of suburban prosperity are substantially larger than those of the measures of central-city prosperity. Moreover, suburban income, education, and employment are more correlated with overall metropolitan-area prosperity than the corresponding central-city measures. Racial Segregation Somewhat surprisingly, researchers appear not yet to have explored the link between racial segregation and city-suburban inequality. Yet there are several reasons to believe that racial segregation, and in particular black-white segregation, should be correlated with central-city/suburban disparities. First, Cutler and Glaeser (1997) have shown that blacks tend to be more disadvantaged relative to whites in more segregated metropolitan areas. And to the extent that blacks tend to live in central cities, greater black-white disparities should lead to greater city-suburban disparities. The hypothesis holds for Hispanics as well, although it is somewhat less persuasive, since there is less evidence that Hispanics are significantly hurt by residential segregation and, moreover, Hispanics are less concentrated in central cities than blacks. A second possible link between racial segregation and city-suburban inequality is suggested by Massey and Denton's argument that racial segregation concentrates poverty (Massey and Denton, 1993). Again, since blacks, and to a lesser extent Hispanics, tend to live in central cities, racial segregation will tend to concentrate poverty and disadvantage in central cities. The possibility that the causality works in the opposite direction should not be ignored, however. Greater central city-suburban disparities, that is, may encourage a greater proportion of affluent whites to live in suburban communities. Although more affluent blacks may share the same inclination to move to the suburbs, housing market discrimination may discourage their departures. Governmental Structure The final and perhaps most critical class of metropolitan-area characteristics considered here relates to the structure of metropolitan government. One of the characteristics is the degree of city elasticity, or the proportion of the metropoli-

OCR for page 192
tan area's population that resides in the central city.7 Consistent with Rusk's theory about city elasticity, the hypothesis is that central-city/suburban disparities should be smaller in more elastic cities that have been able to expand and annex their surrounding suburbs.8 In addition to this measure of elasticity, the fragmentation of local governments in the metropolitan area is also considered. Specifically, we consider the number of general-purpose governments per 100,000 persons. The argument, made by many researchers, is that political fragmentation may heighten spatial stratification. Their theory is that local governments have an incentive to entice homogeneous residents with similar incomes and tastes for services, since they provide services primarily from their own tax base. And they have the means to do so as well, at least partially, through their regulation of land use. The larger the number of governments, the greater the competition, and the greater the pressure on governments to preserve their population. With a greater number of local governments and thus a broad variety of tax-service packages, there is also more of an incentive for households to shop for communities. Results As a preliminary appraisal, Table 6 displays the ratios between central city and suburban outcomes for the 10 metropolitan areas with the greatest as well as the 10 with the least degree of city disadvantage relative to their suburbs. As shown, the overall index of disparity (described at the beginning of this section) ranges from a low of 0 in Benton Harbor, Michigan, to a high of .83 in Laredo, Texas. As for the specific outcomes, the ratio of central city to suburban per capita income ranges from a low of .43 in Benton Harbor to a high of 2.2 in Naples, Florida. (Note that Naples is a significant outlier—the next-highest ratio of per capita income is 1.6 in Laredo, Texas.) The range in employment rate ratios is far smaller, varying from a low of .49 in Benton Harbor to a high of 1.2 in Yuma, Arizona. (Here it should be noted that Benton Harbor is somewhat of an outlier, with the next-lowest employment ratio being .72 in Detroit, Michigan.) As for the ratio of the high school completion rate, it varies from .62, again in Benton Harbor, to 1.46 in Laredo, Texas. The regional pattern is quite pronounced. With the single exception of Fort Pierce, Florida, all 10 of the metropolitan areas with the most relatively disadvantaged central cities are located in the Northeast or the Midwest. The metropolitan areas with the most relatively prosperous cities are all located in the South and the West. Significantly, these 10 central cities are not necessarily the most well-off in absolute terms. Some of them may have fairly low per capita incomes and rates of employment and high school completion but may simply be located in a metropolitan area that is very distressed overall. Indeed, with the exception of Naples, Florida, and Midland, Texas, all of the metropolitan areas with the most relatively prosperous central cities have per capita incomes below the national

OCR for page 192
TABLE 6 U.S. Metropolitan Areas Ranked by Extent of Central City-Suburban Disparity, Top and Bottom 10 Areas, 1990     Ratio of Central-City to Suburban Outcome Metropolitan Area Region Per Capita Income % Employed, Men 16+ % High School Graduates, persons > 25 Overall Index of Disparity Greatest Disparity: Benton Harbor, MI MW .425 .488 .62 0 Newark, NS NE .451 .817 .652 .171 Hartford, CT NE .52 .76 .708 .179 Cleveland, OH MW .535 .763 .719 .188 Detroit, MI MW .562 .717 .795 .202 Middlesex, NJ NE .525 .843 .684 .209 Trenton, NS NE .503 .846 .695 .211 Bergen, NJ NE .466 .884 .689 .22 Ft Pierce, FL S .578 .856 .731 .244 Philadelphia, PA NE .63 .803 .775 .246 Least Disparity: Laredo, TX S 1.63 1.08 1.46 .829 McAllen, TX S 1.52 1.08 1.35 .768 Yuma, AZ W 1.24 1.2 1.32 .762 El Paso, TX S 1.58 1.05 1.3 .746 Naples, FL S 2.24 .917 1.15 .744 Las Cruces, NM W 1.42 1.06 1.27 .709 Visalia-Tulare, CA W 1.24 1.06 1.3 .684 Corpus Christi, TX S 1.29 1.08 1.22 .676 Bakersfield, CA W 1.27 1.07 1.24 .674 Midland, TX S 1.43 1.05 1.14 .654   Source: U.S. Bureau of the Census (1990a). average. The key point is that "central-city disadvantage" here refers only to disadvantage relative to their surrounding suburbs. To further explore these and other patterns, the full set of metropolitan areas are ranked according to the degree of their central-city/suburban disparity and then subdivided into four quartiles of disparity.9 The mean index of disparity ranges from .31 in the most disadvantaged quartile, to .41 in the second, .47 in the third, and .57 in the least disadvantaged quartile. By comparing the mean of metropolitan-area characteristics across quartiles, we can learn which attributes are correlated with greater and lesser central-city disadvantage. Table 7 presents the mean characteristics of selected variables for each of the four quartiles of disadvantage. Once again, the regional pattern is pronounced. In the quartile with the most relatively disadvantaged central cities, 43.9 percent are located in the Northeast, 31.7 percent in the Midwest, 20.7 percent in the South, and 3.7 percent in the West. In the quartile with the least disadvantaged

OCR for page 192
TABLE 7 Mean Characteristics of Four Quartiles of Metropolitan Statistical Areas, Ranked by Degree of Central-City Disadvantage Relative to Suburbs   Quartiles of Central-City Disadvantage Relative to Suburbs   Q1: Greatest Disadvantage 2nd Quartile 3rd Quartile Q4: Least Disadvantage Region Northeast .439 .256 .085 .036 Midwest .317 .366 .22 .108 South .207 .22 .451 .602 West .037 .159 .244 .253 Demographics Population 934,560 537,981 542,056 277,366 % Black 11.7 8.9 9.3 9.3 % Hispanic 5.5 4.2 6.7 11.6 Per capita income $15,352 $14,238 $13,480 $12,321 Racial Segregation Black segregation index 64.2 55.8 51.7 47.0 Hispanic segregation index 43.9 34.2 30.7 29.7 Government Structure Elasticity .35 .445 .494 .473 # of local governments per 100,000 persons 11.3 13.1 11.0 12.0 Population Share % of U.S. metropolitan population living in quartile 40.7 23.6 23.6 12.1 Mean Central-City Suburban Ratios Per capita income .705 .864 .99 1.18 % employed, men 16-64 .855 .921 .95 1.00 % high school grads > 25 .848 .934 1.00 1.09   Sources: Number of governments are preliminary estimates from 1997 Census of Governments provided by the Census Bureau; racial segregation taken from author's estimates and Census Bureau Home Page. All other figures are calculated from U.S. Bureau of Census (1990a). Note that neither the measure of Hispanic segregation nor that of government fragmentation was available for the full set of metropolitan areas. central cites, by contrast, just 3.6 percent are in the Northeast, 10.8 percent are in the Midwest, 60.2 percent are in the South, and 25.3 percent are in the West. In terms of demographics, the metropolitan areas with greater disparities tend to have larger populations, consistent with the notion that geographic distances play a part in determining the degree of central-city disadvantage. There is some evidence, too, that at least the metropolitan areas with the very highest disparities (or the most distressed central cities) tend to have larger proportions of black residents. It also appears true that central cities fare relatively better in metropolitan areas with lower per capita incomes, which is consistent with the

OCR for page 192
hypothesis outlined above that metropolitan areas with higher overall incomes exhibit greater income and skills disparities, and thus greater spatial disparities as well. The final demographic characteristic appears to run counter to the hypothesis outlined above, in that metropolitan areas with the relatively strongest central cities tend to have larger proportions of Hispanic residents. But given the concentration of Hispanics in the Western and Southern regions of the country, where central cities are strongest, this correlation between central-city prosperity and the size of the Hispanic population may be spurious. The figures on racial segregation are quite striking. In the areas with the most disadvantaged central cities, blacks appear to be far more segregated at the neighborhood level. In particular, in these highest-disparity metropolitan areas, the mean black segregation index (measured by the index of dissimilarity) is 64.2 in the metropolitan areas with the most disadvantaged central cities compared with 47.0 in those metropolitan areas with the lowest levels of disparity. Similarly, Hispanic segregation also appears to be greater in the metropolitan areas with the relatively weakest central cities. Finally, one of the two hypotheses regarding governmental structure appears to be supported: metropolitan areas with the greatest city-suburban disparities also tend to be those with the least elastic central cities—with central cities, that is, that contain a smaller proportion of the overall metropolitan-area population. There is little evidence here, however, that the degree of local government fragmentation (measured by the number of county, municipal, and township governments per capita) has any relationship to the level of central-city/suburban disparity. Multivariate Analysis It is not clear of course, whether the simple correlations revealed above between central-city/suburban inequality and metropolitan area characteristics hold up when other variables are held constant. The next step is thus to estimate a regression in which the index of central-city/suburban disparity is regressed on various metropolitan-area characteristics. Four sets of variables are once again considered: regional location, demographics, racial segregation, and governmental structure. Specifically, the regression includes three regional dummy variables (the Northeast is the reference region); four demographic variables (a measure of metropolitan area size,10 the per capita income of the overall metropolitan area, the proportion black, and the proportion Hispanic); two measures of racial and ethnic segregation (the black and Hispanic dissimilarity indices); and two key measures of governmental structure—the number of general purpose local governments (county, municipal, and township) per 100,000 people and the degree of central-city elasticity. In the comparison of the quartiles of central-city disadvantage, central-city

OCR for page 192
TABLE 8 Regression Results of 1990 Central-City Suburban Index of Disparity   Model 3   Coefficient Standard Error Constant .532*** .075 Midwest -.015 .017 South .09*** .018 West .053*** .019 Log of MSA population .007 .007 Per capita income, in MSA -.009*** .0028 % black, in MSA -.272*** .065 % Hispanic, in MSA .158*** .05 Black segregation -.0013*** .0005 Hispanic segregation -.0015*** .0005 Elasticity .197*** .055 Elasticity spline (at 0.45) -.186** .09 # Governments per 100,000 people .0009* .0005 R-squared .543   N 274   MSA = metropolitan statistical area. * = statistically significant at the 10 percent level. ** = statistically significant at the 5 percent level. *** = statistically significant at the 1 percent level. elasticity was clearly lowest in the quartile with the most disadvantaged central cities, but approximately the same across the remaining three quartiles. It seemed, in other words, that the degree of elasticity only mattered when the elasticity was very low—that is, when the central city captured only a very small share of the metropolitan-area population. To test this apparent finding, a spline variable (at 0.45) is added to the regression, which effectively allows for the degree of elasticity to matter at low levels, but not at higher ones. More technically, it allows for a kink in the regression line at a threshold of 0.45. For elasticity values below 0.45, the relationship between elasticity and central-city disadvantage remains measured by the coefficient on the elasticity variable. For metropolitan areas with elasticities above 0.45, however, the correct relationship is estimated by the sum of the coefficient on elasticity and the coefficient on the spline variable.11 Recall that a larger index of disparity implies a more prosperous central city—relative, that is, to its surrounding suburbs. Thus, if a particular variable, say the proportion black, has a negative coefficient, we should interpret this to mean that central cities tend to suffer greater relative disadvantage when the variable in question is larger. If instead a variable has a positive coefficient, then

OCR for page 192
this suggests that central cities tend to experience smaller relative disadvantage when the variable is larger. The results from the first regression are largely consistent with the simple comparison of means above. Central cities, that is, tend to fare the worst relative to their suburbs in metropolitan areas located in the Northeast, and the best in the South and the West. Central cities also tend to be more disadvantaged in metropolitan areas with higher overall incomes, larger black populations, more segregated minority populations, and smaller Hispanic populations. This final result suggests that the correlation between large Hispanic populations and relatively prosperous central cities is not simply a regional phenomenon and may in fact be more meaningful than at first surmised. With this said, it is difficult to understand just why such a relationship should exist. One notable difference here is that the size of a metropolitan area appears insignificant once its other characteristics are taken into account. The coefficients on the elasticity variables are indeed what we expect. The coefficient on elasticity is positive and significant, and the coefficient on the spline variable is negative and significant. And the magnitudes are virtually the same. In other words, when a central city's elasticity is below .45, raising its elasticity should improve its relative prosperity. Once a city's elasticity has reached .45, however, subsequent increases appear to have no effect.12 Curiously, the coefficient on the number of governments per capita is positive and marginally significant (at the 10 percent level of significance), suggesting that central cities do better in metropolitan areas with more fragmented governments. This surprising result is surely the subject for more research, and no definitive conclusions should be made. Indeed, it is quite likely that the simple number of local governments per capita is simply not a good measure of government fragmentation. For although several researchers have found greater inequality in metropolitan areas that rely more heavily on the property tax as a source of revenue, few have found a significant relationship between the number of governments per capita and city-suburban disparities (see Bollens, 1986; Logan and Schneider, 1982; Morgan and Mareschal, 1996). Thus, it may be that government structure can influence residential sorting, but that the count of governments per capita is too simple a measure to capture such structure. Recall, too, that our dependent variable reflects only average differences between cities and suburbs and not differences among what may be highly diverse, individual suburbs. Conclusion This paper attempts to describe in broad terms the nature and extent of spatial inequality found in contemporary metropolitan areas. It concludes that, although racial segregation is declining modestly, both central-city/suburban and neighborhood-level economic inequality are increasing nationwide. There is, however, considerable variation across individual metropolitan areas in the extent

OCR for page 192
of central-city/suburban disparity. With respect to these variations in inequality, this paper further attempts to test certain key correlates. In general, the results suggest that central cities fare worse relative to their suburbs in the Northeast and the Midwest, in larger and wealthier metropolitan areas, and in areas with larger black and smaller Hispanic populations. In addition, unlike past studies, this study examined the role played not only by the size of minority populations but by their degree of segregation and found that city-suburb inequality is significantly greater in metropolitan areas with more segregated black and Hispanic populations. As to government structure, the results here confirm Rusk's basic hypothesis that highly inelastic central cities are more deprived than others relative to their suburbs. Interestingly, however, the degree of elasticity does not appear to make much difference once it rises beyond about .45. Finally, the degree of local government fragmentation was also examined, and, contrary to expectations, city-suburban inequality was found to be, if anything, greater in metropolitan areas with less fragmented local government. These sometimes surprising findings may shed some light on the nature and roots of spatial and jurisdictional disparities in metropolitan areas. But they also raise and leave unanswered many important questions. Thus, this study should also encourage others to further explore the hypotheses generated here as well as to probe other possible links to spatial disparities, such as residential density (known as urban sprawl), industrial structure, and additional measures of local government coordination and competition, including the extent of property tax reliance. Such explorations should not only help us to come to more definitive conclusions regarding spatial inequalities but also guide us ultimately to mitigate them. Acknowledgments The author would like to thank the committee for its direction, as well as Hal Wolman, Alan Altshuler, Tony Downs, Harry Holzer, and the National Research Council's anonymous reviewers for their helpful comments. References Abramson, Alan, Mitchell Tobin, and Matthew VanderGoot 1995 The changing geography of metropolitan opportunity: The segregation of the poor in U.S. metropolitan areas, 1970 to 1990. Housing Policy Debate 6:45-72. Berger, Bennett 1960 Working Class Suburb: A Study of Auto Workers in Suburbia. Berkeley: University of California Press. Bollens, Scott 1986 A political-ecological analysis of inequality in metropolitan areas. Urban Affairs Quarterly 22(2):221-41.

OCR for page 192
Crane, Jonathan 1991 The epidemic theory of ghettos and neighborhood effects on dropping out and teenage childbearing. American Journal of Sociology 96(5):1226-1259. Cutler, David, and Edward Glaeser 1997 Are ghettos good or bad? Quarterly Journal of Economics 112(3):827. Ellen, Ingrid, and Margery Turner 1997 Location, location, location: How does neighborhood environment affect the well-being of families and children? Housing Policy Debate 8(4):833-866. Fannie Mac 1997 The Fannie Mac National Housing Survey: City Life, Homeownership, and the American Dream. Washington, DC: Fannie Mae. Farley, Reynolds, and William Frey 1994 Changes in the segregation of whites from blacks during the 1980s: Small steps toward a more integrated society. American Sociological Review 59:23-45. Greene, William H. 1993 Economic Analysis. New York: Macmillan Publishing Company. Hill, Edward W., and Harold L. Wolman 1997 City-suburban income disparities and metropolitan area employment: Can tightening labor markets reduce the gaps? Urban Affairs Review 32(4):558-582. Hill, Richard Child 1974 Separate and unequal: Government inequality in the metropolis. American Political Science Review 68:1557-1568. Hogan, Dennis, and Evelyn Kitagawa 1985 The impact of social status, family structure, and neighborhood on the fertility of black adolescents. American Journal of Sociology 90(4):825-855. Jargowsky, Paul 1997 Poverty and Place: Ghettos, Barrios, and the American City. New York: Russell Sage Foundation. Lineberry, Robert L. 1975 Suburbia and the metropolitan turf. The Annals of the American Academy of Political and Social Science 422:1-9. Logan, John R., and Mark Schneider 1982 Governmental organization and city/suburb income inequality. Urban Affairs Quarterly 17(3):305-318. Long, Larry, and Donald C. Dahmann 1980 The City-Suburb Income Gap: Is It Being Narrowed by a Back-to-the-City Movement? Bureau of the Census. Washington, DC: U.S. Department of Commerce. Masotti, Louis, and Jeffrey Hadden, eds. 1973 The Urbanization of the Suburbs. Vol. 7 of Urban Affairs Annual Reviews. Beverly Hills, CA: Sage. Massey, Douglas S., and Nancy A. Denton 1993 American Apartheid: Segregation and the Making of the Underclass. Cambridge, MA: Harvard University Press. 1987 Trends in the residential segregation of blacks, Hispanics, and Asians. American Sociological Review 52:802-825. Mayer, Christopher 1996 Does location matter? New England Economic Review May/June:26-40. Morgan, David, and Patrice Mareschal 1996 Central City/Suburban Inequality and Metropolitan Political Fragmentation. Paper presented at the annual meeting of the American Political Science Association, San Francisco.

OCR for page 192
Orfield, Myron 1997 Metropolitics: A Regional Agenda for Community and Stability. Washington, DC: Brookings Institution Press and the Lincoln Land Institute. Rusk, David 1993 Cities Without Suburbs. Washington, DC: The Woodrow Wilson Center Press. Schnore, Leo 1965 The Urban Scene. New York: Free Press. Sternlieb, George, and Robert W. Lake 1975 Aging suburbs and black homeownership. The Annals of the American Academy of Political and Social Science 422:105-117. U.S. Bureau of the Census 1992 Poverty in the United States: 1992. Current Population Reports, Series Pp. 60-185. 1991 Metropolitan Areas and Cities. 1990 Census Profile, Number 3. 1990a Census of Population and Housing Summary Tape File 3C. CD-ROM. 1990b 1990 Census of Population, Social and Economic Characteristics for Metropolitan Areas. 1980 1980 Census of Population, General Social and Economic Characteristics. 1970 1970 Census of Population, General Social and Economic Characteristics. 1960 1960 Census of Population, General Social and Economic Characteristics. Wilson, William Julius 1987 The Truly Disadvantaged. Chicago: University of Chicago Press. Notes 1   In fact, despite their overrepresentation in central cities, the share of blacks who live in suburban areas has been increasing. In 1960, for instance, just 20 percent of blacks lived in suburban communities, compared with 26 percent in 1990. 2   Data from the Current Population Survey presented in Ihlanfeldt (this volume) suggest that the city-suburban disparity in family income continued to rise between 1989 and 1995. 3   Calculated from Massey and Denton (1993:Table 8.1). 4   The minimum value is assigned 0, the maximum value is assigned 1, and the values in between are standardized according to the following formula: Z = (X-Xmin)/(Xmax-Xmin). 5   Hill and Wolman (1997) examine the impact of tight labor market conditions on central-city/suburban differences and find, contrary to conventional wisdom, that economic growth may actually exacerbate disparities. 6   Several researchers have found the age of the metropolitan area to be correlated with greater central-city/suburban inequality (see Bollens, 1986; Hill, 1974; Logan and Schneider, 1982; Schnore, 1965). 7   The official Census Bureau definition of central city is used here to construct a measure of elasticity. As Hill and Wolman (1997) point out, the Census Bureau uses a very broad definition of central city, which encompasses more than merely the largest municipality in the metropolitan area and may include as well cities that might be better described as ''edge cities.'' Thus, using the Census definition of central city may not be ideal for constructing a measure of elasticity. Nonetheless, it is consistent with the other data used here, which also rely on the official Census definition for central city. The results remain the same when using a more restrictive definition of central city (at least for the 152 metropolitan statistical areas for which the more restrictive measure of elasticity is available). 8   Note that this measure of elasticity differs from the more complicated index used by Rusk, but it captures the essence of his concept (see Hill and Wolman, 1997). 9   Significantly, the quartiles are defined so that each includes an equal number of metropolitan areas, rather than residents. It turns out that the metropolitan areas in which central cities fare worse

OCR for page 192
    are far more populous. Thus, for example, the quartile with the most relatively disadvantaged cities houses a full 41 percent of the total metropolitan population. 10   The natural logarithm is used to smooth the distribution of metropolitan population. 11   Specifically, the regression includes a variable that is equal to the difference between the elasticity and .45 when elasticity is above .45 and takes on the value of 0 when the elasticity is below .45. For more information, Greene includes an excellent discussion of spline regressions (see Greene, 1993:234-238.) 12   Hill and Wolman (1997) find no connection between elasticity and the disparity in per capita income between central cities and suburbs. Their analysis differs in three ways, however. First, they include city-suburban human capital differences as an independent variable, which means that their regression model attempts to explain city-suburban differences in returns to education, rather than overall disparities. Second, most of their estimated models also include the level of disparity existing in 1980 as an independent variable, which means that they test whether elasticity has any effect on the change in disparity between 1980 and 1990. Finally, they do not include a spline variable. (When the spline variable is not included here, the elasticity remains significant, but its magnitude is far smaller.)