Race and the Spatial Segregation of Jobless Men
539
RACE AND THE SPATIAL SEGREGATION OF JOBLESS MEN IN URBAN AMERICA* ROBERT L. WAGMILLER, JR. Changes in U.S. metropolitan areas over the past 30 years are thought to have concentrated jobless men in low-income, predominantly minority neighborhoods clustered near the center of the city. Using tract-level data from the Neighborhood Change Database for 1970–2000, I examine how the residential segregation of jobless from employed men has changed over the past three decades. I nd that jobless men in U.S. metropolitan areas have become less uniformly distributed throughout the metropolis and more isolated, concentrated, and clustered since 1970; but they have also become less centralized. Racial and ethnic group differences in the spatial segregation of jobless men are large. Jobless black men occupy a uniquely disadvantaged ecological position in the metropolis: in comparison with other jobless men, they are much less uniformly distributed throughout the metropolis and much more isolated from employed men, they are concentrated in a smaller amount of physical space, and their neighborhoods are more clustered and are located closer to the center of the city. The dimensions of segregation strongly overlap for black jobless men, producing a multidimensional layering of segregation not encountered by other jobless men. Multivariate models reveal that the uniquely disadvantaged ecological position of jobless black men is less a reection of different patterns of regional concentration and metropolitan settlement or of differences in group-status characteristics than it is an inevitable consequence of extreme levels of racial residential segregation in the United States.
any social scientists, most notably William Julius Wilson (1987, 1996), have argued M that structural changes in the metropolis in the 1970s and 1980s concentrated male joblessness in low-income, predominantly African American neighborhoods in the central city. The emergence of these “jobless ghettos” is thought to have fundamentally altered the nature of urban disadvantage. Widespread male joblessness is believed to have (1) reduced the pool of marriageable men in low-income communities, impeding traditional family formation processes (Wilson 1987, 1996); (2) spurred the concentration of poverty (Wilson 1987, 1996) and increased social dislocations, such as crime, juvenile delinquency, teenage pregnancy, and welfare dependency associated with the concentration of disadvantage (Lee 2000; Parker and Pruitt 2000; Sampson 1997; South and Crowder 1999; Wilson 1987, 1996); (3) reduced the number of mainstream role models in low-income neighborhoods, isolating residents from the daily patterns and rhythms of work and making symbols of family stability and achievement less visible (Anderson 1991; Wilson 1987); and (4) drained community institutions (e.g., churches, schools, stores, block clubs, and neighborhood improvement associations) of the social and economic resources they require for sustenance (Massey 1990; Wilson 1987). Although numerous studies have shown that poverty in U.S. metropolitan areas became more concentrated in the 1970s and 1980s (Jargowsky 1997, 2003; Jargowsky and Bane 1991; Kasarda 1993; Kingsley and Petit 2003; Krivo et al. 1998; Massey and Eggers 1990, 1993; Morenoff and Tienda 1997), the extent to which the jobless ghettos described by Wilson (1987, 1996) have materialized is less clear. Case studies of Detroit by Sugrue (1996) and Farley, Danziger, and Holzer (2000) suggest that the rising unemployment rates
*Robert L. Wagmiller, Jr., Department of Sociology, University at Buffalo, SUNY, 430 Park Hall, Box 604140, Buffalo, NY 14260-4140; E-mail:
[email protected]. The author thanks Debra Street, Robert Adelman, Peter K.B. St. Jean, Edward O. Laumann, Susan E. Mayer, and Ross M. Stolzenberg and the anonymous reviewers and editors at Demography for helpful comments on earlier versions of this article. Demography, Volume 44-Number 3, August 2007: 539–562
539
540
Demography, Volume 44-Number 3, August 2007
that Wilson (1987, 1996) observed in the racially segregated communities on the south side of Chicago are also found in Detroit. Larger studies of more-representative samples of metropolitan areas have, however, been less conclusive. Quillian (2003) reported a signicant decline in male employment in low-income neighborhoods—especially in low-income, predominantly black neighborhoods—between 1950 and 1990. Similarly, Wagmiller (2004) found that the number of neighborhoods in which the majority of working-age men do not work increased sharply between 1970 and 2000. By contrast, Krivo et al. (1998) found that the isolation of jobless black and white men in urban America did not increase between 1980 and 1990 and that racial differences in isolation were modest. Moreover, these studies have largely neglected the spatial dimensions of urban male joblessness and eschewed traditional measures of spatial segregation. The case studies of Chicago and Detroit graphically depicted or qualitatively described the changing rates of male joblessness and unemployment in urban neighborhoods but did not attempt to quantify the spatial concentration of jobless men (Farley et al. 2000; Mouw 2000; Sugrue 1996; Wilson 1987, 1996). The studies of larger, more-representative samples of metropolitan areas have employed measures (such as the index of dissimilarity, isolation index, and concentrated male joblessness rate) that fail to capture other important aspects of the spatial segregation of jobless men, such as the clustering, concentration, and centralization of jobless men (Krivo et al. 1998; Quillian 2003; Wagmiller 2004). The spatial location of neighborhoods with high rates of male joblessness is important because it can either amplify or moderate the negative effects of male joblessness. For example, when neighborhoods with high rates of male joblessness are surrounded by similar neighborhoods or are located in aging, deteriorating areas of the central city, the deleterious effect of high rates of male joblessness on neighborhood social and economic organization is magnied. When such neighborhoods are clustered, ties to mainstream norms, values, and social and economic institutions are further weakened, neighborhood resources are further depleted, and jobless men’s ties to informal employment networks are further eroded. When neighborhoods with high rates of male joblessness are located in aging, deteriorating city centers, suburban and exurban areas of new job growth are less accessible to jobless men, potentially creating a spatial mismatch between the location of jobless men and employment opportunities. Following Massey and Denton (1988) and subsequent studies of hypersegregation (Iceland, Weinberg, and Steinmetz 2002; Massey and Denton 1989, 1993; Massey, White, and Phua 1996; Wilkes and Iceland 2004), I examine the residential segregation of jobless men from employed men along ve dimensions: evenness, isolation,1 concentration, clustering, and centralization. Segregation indices for U.S. metropolitan areas in 1970, 1980, 1990, and 2000 are computed by using census-tract data from the Decennial Census of Population and Housing that are normalized to the 2000 census tract boundaries and 1970 metropolitan area denitions to avoid confounding distribution and boundary changes. For years in which labor force participation information is available by race, segregation indices representing the residential segregation of jobless white, black, Hispanic, and Asian men from employed men of any race are also presented. Results reveal the distinctively disadvantaged ecological niche occupied by jobless black men in urban America. Multivariate models are estimated to evaluate whether the uniquely disadvantaged ecological position held by jobless black men is primarily a reection of differences in the types of metropolitan areas in which they live, differences in group socioeconomic status characteristics, or racial residential segregation patterns. These models reveal that the uniquely disadvantaged
1. Although Massey and Denton (1988) referred to this dimension as exposure, I prefer to refer to this dimension as isolation to avoid the potential terminological confusion that might result from using the isolation index to measure exposure.
Race and the Spatial Segregation of Jobless Men
541
ecological position of jobless black men is largely an unavoidable consequence of extreme levels of racial residential segregation. THE STRUCTURE OF SPATIAL SEGREGATION AND THE EMPLOYMENT PROSPECTS OF JOBLESS MEN Although simple measures—such as the number of neighborhoods with high rates of male nonemployment or the residential dissimilarity between employed and jobless men—are intuitively appealing and can be informative of general trends in the concentration of disadvantage, they fail to capture other important, primarily spatial components of contemporary urban male joblessness. Yet, contemporary disadvantage, as Wilson (1987) argued, is distinguished not only by its concentration in certain neighborhoods in the city but also by the clustering of these neighborhoods near the center of the city. Conventional measures of concentrated disadvantage, which implicitly assume that neighborhoods with high rates of disadvantage are unaffected by the rates of disadvantage in surrounding communities or by their location in urban space, overlook important aspects of the ecology of urban inequality. I examine the spatial segregation of jobless men along the ve dimensions identied by Massey and Denton (1988). From this perspective, jobless men may be overrepresented in some neighborhoods and underrepresented in others, varying on the characteristic of evenness. They may be exposed to few or many jobless men in their neighborhoods, varying on the dimension of isolation. They may be concentrated in a small number of neighborhoods, constituting a small proportion of the total physical space of a metropolis, or they may be spread over a large geographic area, varying on the dimension of concentration. Their neighborhoods may be tightly grouped in a large, contiguous area, or they may be scattered throughout the metropolis, varying on the degree of clustering. Finally, their neighborhoods may be concentrated near the center of a city, or they may be distributed uniformly around a metropolis, varying in the extent of their centralization. The structure of spatial segregation is important because it affects both the employment opportunities available to jobless men and the challenges that they are likely to confront when obtaining and sustaining employment. High levels of unevenness, for example, likely have very different implications for jobless men and their neighborhoods than do high levels of isolation, clustering, concentration, or centralization. Similarly, these other dimensions likely have their own distinct implications for the employment prospects of jobless men. Some dimensions, such as evenness and isolation, may adversely affect jobless men’s employment prospects primarily by affecting their access to information about new job opportunities. A high level of unevenness (which implies that jobless men are overrepresented in certain neighborhoods) or a high level of isolation (which implies that jobless men are not exposed to many employed men in their neighborhood) likely fosters social networks among jobless men that are disproportionately composed of other jobless men. Restricted social contact with employed men limits these men’s access to the informal job networks, through which many jobs are found (Granovetter 1974). Although other dimensions, such as centralization, have a lesser effect on jobless men’s capacity to access information about new job opportunities, they inuence the costs and desirability of traveling to these jobs. Changing manufacturing and communications technologies and increased global economic competition have greatly reduced the number of well-paying, low-skill jobs available in central cities because factories have relocated overseas, to the South, or to the suburbs and urban periphery (Berry and Cohen 1973; Berry and Kasarda 1977; Kasarda 1995; Sassen 1990). High levels of centralization, which indicate that jobless men are congregated near a central business district, can create a spatial mismatch between the residential location of jobless men and areas of new employment growth.
542
Demography, Volume 44-Number 3, August 2007
Still other dimensions, such as concentration and clustering, which may limit jobless men’s ability to access jobs or information about new employment opportunities, are likely to more directly affect jobless men’s employment prospects by strengthening the social and cultural forces that limit or impede their capacity to take advantage of the employment opportunities potentially available to them. When jobless men are concentrated in a small amount of physical space or when neighborhoods with high rates of male joblessness are contiguous—forming what Wilson (1996) termed jobless ghettos—a culture of despair and joblessness can emerge that results in leveled aspirations (Auletta 1982; MacLeod 1987), limited exposure to and knowledge of the routines of work life (Auletta 1982; Wilson 1996), and the evolution of a set of attitudes, values, and behaviors that are at variance with those held by employers and the larger society (Auletta 1982; Kirschenman and Neckerman 1991; Massey and Denton 1993). The effect of high levels of segregation on any one of these dimensions on the employment prospects of jobless men, moreover, is likely to be highly dependent on the level of segregation that they experience on other dimensions. The deleterious effects of high levels of concentration and clustering, for example, may be limited when accompanied by low levels of unevenness and isolation because a culture of despair and joblessness is unlikely to develop when overall levels of joblessness in a neighborhood remain relatively low. Similarly, the adverse consequences of high levels of isolation on the employment prospects of jobless men may moderate when levels of clustering, concentration, and centralization are low. When segregation indices are high and the dimensions of segregation for a group of jobless men overlap empirically, employment prospects for these men are likely to be much more restricted, all other things being equal, than for groups of men who experience lower levels of spatial segregation. Examining multiple dimensions of residential segregation is also important because the levels and degree of overlap between dimensions experienced by jobless men of different racial and ethnic backgrounds are likely to vary. Different patterns of residential segregation and rates of male nonemployment across racial and ethnic groups are likely to lead to stark differences in the levels of segregation experienced by jobless men. Based on earlier studies of racial and ethnic differences in residential segregation (Wilkes and Iceland 2004) and male nonemployment (Mincy 2006), I expect that jobless black men will be more highly segregated from employed men on all dimensions and that these dimensions will overlap more strongly than for jobless men of other racial and ethnic backgrounds. Conversely, I expect that jobless white men will experience universally low levels of segregation from employed men. Patterns of residential segregation and male nonemployment for Hispanics and Asians suggest that on some dimensions (namely, concentration and centralization), jobless Hispanic and Asian men are likely to be nearly as segregated from employed men as jobless black men, while on other dimensions (namely, evenness, isolation, and clustering), they are likely to experience relatively low levels of segregation. Trends for the different dimensions also are likely to diverge. Based on rising rates of male nonemployment, especially for black men (Edelman, Holzer, and Offner 2006; Mincy 2006), I expect sharper increases in isolation than in the other dimensions and sharper increases in isolation for black men than for men from other racial and ethnic backgrounds. Developing explicit expectations of change for the other dimensions on the basis of demographic and ecological changes over the past three decades is more difcult. Based on the widening economic inequality among central cities and suburbs (Swanstrom et al. 2004), for example, I expect greater increases in centralization (and possibly concentration) than in the other dimensions. However, the economic deterioration of inner-ring suburban neighborhoods and the progressive suburbanization of racial and ethnic minority groups (Katz and Lang 2003) suggest that increases in these dimensions may have been tempered by the migration of low-income populations into outlying areas of the metropolis. I expect that segregation on most dimensions will increase more sharply for jobless black and Hispanic
Race and the Spatial Segregation of Jobless Men
543
men than for jobless white men as a consequence of more rapidly increasing economic segregation among blacks and Hispanics (Jargowsky 1996). For black men, this increase may be partially offset by a steady, albeit modest, decline in black-white residential segregation in recent decades (Iceland et al. 2002; Logan, Stults, and Farley 2004; Wilkes and Iceland 2004). The well-documented decline in concentrated poverty (Jargowsky 2003; Kingsley and Pettit 2003) and other forms of concentrated disadvantage (Jargowsky and Yang 2006) in the 1990s suggest that jobless men may have become less segregated from employed men in recent years. DATA AND METHODS The data for this study are obtained from the Neighborhood Change Database (NCDB), which integrates population and housing counts from the 1970, 1980, 1990, and 2000 U.S. Census of the Population and Housing Summary Files into a single database (GeoLytics 2003). The NCDB contains information on the race, sex, and age distribution of the population, citizenship status, labor force participation, education and school enrollment, income and poverty status, and housing characteristics of all census tracts in the United States. Data from the NCDB are used to calculate for individual metropolitan areas (MA) residential segregation indexes for jobless men between 1970 and 2000. Metropolitan-level variables were constructed by cumulating the appropriate tract counts. Geographic comparability is an important concern in a longitudinal study that covers as extended a time period as this study does because the Census Bureau responds to changes in the population distribution by periodically redrawing tract and metropolitan area boundaries. Tract boundaries are redrawn to maintain their population size and to reect physical changes in street patterns. As the population of a metropolitan area migrates into untracted areas, new tracts are added. Census tracts that experience substantial population gain are frequently split into multiple tracts, and tracts that experience substantial population loss are typically merged with other tracts. The boundaries of most tracts changed between 1970 and 2000. The boundaries of metropolitan areas also change over time. Much as tract boundaries are redrawn as the population distribution shifts, metropolitan area boundaries shift as the size, distribution, and social and economic integration of core and outlying areas change. Because the focus of this study is changes in the spatial distribution of male joblessness, using temporally consistent, geographic boundaries is important. Otherwise, one cannot determine whether changes in measures of spatial segregation are the result of changes in the distribution of jobless men in urban space or changes in the geographic boundaries of spatial units. To minimize this problem, in this study, I compute measures of spatial segregation from census-tract data normalized to the 2000 census geographic boundaries2 and apply the metropolitan area boundaries as dened by the Ofce of Management and Budget (OMB) on February 23, 1971, to all years. Ideally, more-recent metropolitan area boundaries that account for the spatial expansion or contraction of the metropolis since 1970 would be used. Unfortunately, such boundaries for the crucial pre-1980 period cannot be used because few areas outside of the boundaries of MAs were tracted prior to 1980.3 2. The use of these boundaries is not, however, without its drawbacks. The application of year-2000 tract denitions to earlier census years results in the creation of some tracts with unusually small populations and other tracts with unusually large populations. Segregation indices were estimated both with and without these atypical tracts to assess the robustness of the results to changes induced into the conceptual denition of tracts by the remapping of earlier census data. The results presented here are not unduly inuenced by the inclusion of these atypical tracts. 3. Although the use of comparable boundaries reduces the chance that boundary changes unduly inuence results, ignoring metropolitan area expansion and contraction over the past 30 years is likely to introduce its own measurement bias. Consequently, all descriptive measures and analytic models were also estimated by using contemporaneous metropolitan area denitions. Mean levels of spatial segregation are noticeably higher when contemporaneous boundaries are used, but overall trends are remarkably similar. The use of contemporaneous boundaries substantially increases within-metropolitan area variation in spatial segregation during this
544
Demography, Volume 44-Number 3, August 2007
I examine the spatial segregation of jobless men from employed men in the 227 metropolitan areas dened by the Census Bureau in 1971. Segregation indexes for individual metropolitan areas are computed for all jobless men in 1970, 1980, 1990, and 2000. Racespecic employment information is not available for 1970. Consequently, segregation indexes for racial and ethnic subgroups could be calculated only for 1980, 1990, and 2000. Segregation indexes were calculated for jobless white, black, Hispanic, and Asian men. Prior to the calculation of these segregation indexes, tracts in which more than 40% of the population in any year resided in group quarters were deleted from the le to remove tracts dominated by prisons, military bases, colleges and universities, and other formal institutions (Massey and Denton 1987). Indices were not computed for racial and ethnic groups with fewer than 1,000 jobless men in a metropolitan area because indices for these groups are likely to be unduly inuenced by random factors inuencing settlement patterns and geographic boundary errors, resulting in indices that are less reliable than those for groups with larger populations of jobless men (Frey and Myers 2002; Glaeser and Vigdor 2001; Wilkes and Iceland 2004). I examine residential segregation along the ve dimensions of spatial segregation identied by Massey and Denton (1988). Indexes measure residential segregation of jobless from employed men. For racial and ethnic subgroups, indexes measure residential segregation of jobless men in a racial or ethnic group from employed men of all races and ethnicities. For example, indexes for jobless black men measure the segregation of these men from employed white, black, Hispanic, and Asian men. For each of the ve dimensions of segregation, I use the index preferred by Massey and Denton (1988) . Detailed information on the calculation of each index is available in Massey and Denton (1988, 1989) and Iceland et al. (2002), but I summarize the indexes here. Evenness is measured by using the index of dissimilarity, D (Jakubs 1977, 1979, and 1981). D ranges from 0 to 1, with its value indicating the proportion of jobless men who would have to move to achieve a uniform distribution throughout the metropolis. Isolation is measured by using the isolation index, xP*x (Lieberson 1981a). This index ranges from 0 to 1 and can be interpreted as the likelihood that a randomly selected jobless man shares a tract with another jobless man. As opportunities for contact between jobless and employed men in urban neighborhoods decline, the isolation index increases. Concentration is measured by using the relative concentration index, which is calculated by computing the average amount of physical space occupied by jobless men relative to employed men and comparing this quantity with the ratio that would be achieved if jobless men were maximally concentrated and working men were maximally dispersed. The relative concentration index ranges from –1 to +1. A score of –1 indicates that working men’s concentration exceeds jobless men’s concentration to the maximum extent possible, a score of 0 means that jobless men and employed men are equally concentrated, and a score of +1 indicates that jobless men’s concentration exceeds employed men’s concentration to the maximum extent possible. Clustering is measured by using White’s (1983) index of spatial proximity (SP). If no difference exists between the clustering of jobless and employed men, SP is 1. If jobless men live closer to other jobless men than to employed men, the value of the index is greater than 1. Finally, if jobless men live closer to employed men than to other jobless men, the index is less than 1. Centralization is measured by using the absolute centralization index, which compares the distribution of jobless men by distance from the central business district to the distribution of land area around the central business district (Duncan and Duncan 1957; Duncan, Cuzzort, and Duncan 1961; Glaster 1984). It ranges from –1 to +1, with negative values indicating a tendency for jobless men to reside in more remote areas, a score of 0 period. Close examination of changes in segregation measures in individual metropolitan areas when boundaries do and do not change suggests that most of this variation reects boundary changes rather than distribution changes.
Race and the Spatial Segregation of Jobless Men
545
indicating that jobless men are distributed evenly around the metropolitan area, and positive values indicating a tendency for jobless men to reside near the central business district. I resolve several data questions to facilitate comparisons across years. Employment rates at the tract level for most of these years are published only for all men aged 16 years or older. Published rates of employment, therefore, substantially overestimate rates of nonemployment among working-age men because some of the men classied as jobless in published statistics are undoubtedly retired. Unfortunately, the detailed information on labor force status necessary to exclude retired men (and, similarly, students, disabled men, and so on) from the calculation of the neighborhood male joblessness rate is not available at the tract level in most years. Following Quillian (2003), I adjust employment rates by excluding all men 65 years or older from both the numerator and the denominator of the formula used to estimate the neighborhood male joblessness rate.4 Some of the men 65 years or older will be employed, resulting in an underestimate of the male nonemployment rate in the neighborhood. On the other hand, the inclusion of many young men likely to still be enrolled in school full time probably leads to an overestimation of the neighborhood male nonemployment rate. Because sex-specic school enrollment information for young men is not available at the tract level in the NCDB, I cannot apply the adjustment for school enrollment Quillian (2003) used. Although estimating the precise magnitude of bias in this measure is difcult, neighborhood nonemployment rates are likely slightly overestimated, and this bias probably declines steadily over time.5 Also, this bias is likely slightly larger for blacks and Hispanics and moderately larger for Asians than for whites, although bias for all groups, except Asians, steadily declines over the study period.6 THE SPATIAL SEGREGATION OF JOBLESS MEN The ve indices were computed for jobless men in 227 metropolitan areas. Figure 1 presents the weighted average indices for these metropolitan areas between 1970 and 2000. The weights used in these calculations are proportional to the number of jobless men in the metropolis in that year. Substantively, weighted estimates tell us about the spatial segregation of the typical jobless man, and unweighted estimates tell us about spatial segregation in the typical metropolis. Weighted estimates are presented here because this study is more concerned with the spatial segregation that the average jobless man experiences than with differences across contexts (i.e., metropolitan areas). Weighted indices are generally slightly larger than unweighted indices (with the exception of the centralization dimension), and the gap between weighted and unweighted indices has expanded over the past 30 years, indicating that jobless men tend to live in metropolitan areas with somewhat higher levels 4. The neighborhood joblessness rate in tract i at time t is dened as Jobless Rate it =
(Men it − Elderlyit ) − Employed it Men it − Elderly it
,
where Menit is the number of men 16 years or older in tract i at time t, Elderlyit is the number of men 65 years or older in tract i at time t, and Employedit is number of employed men in tract i at time t. 5. Using data from the 1970, 1980, 1990, and 2000 Public-Use Microdata Samples (PUMS), which include the individual-level information on age, employment status, and school enrollment necessary to calculate the nonemployment rate for the population of urban men aged 16–64 and not enrolled in school, I assessed the direction and magnitude of this bias by comparing for each year this estimate to an estimate based on the formula in footnote 4. For all years, the jobless rate, as calculated with the equation in footnote 4, is biased upward. However, this bias is generally small to moderate and declines over time. Between 1970 and 2000, the difference between these rates declines from 3.9% in 1970 to 1.3% in 2000. 6. Male nonemployment rates using the PUMS data were also calculated for racial and ethnic subgroups. Nonemployment rates for all racial and ethnic groups are overestimated using the formula in footnote 4. The magnitude of this bias is small to moderate, with rates of nonemployment overestimated by 0.7%–2.4% for white men, by 2.1%–4.6% for black men, by 2.1%–4.4% for Hispanic men, and by 6.5%–8.9% for Asian men. For all groups except Asians, bias declines monotonically between 1980 and 2000.
Demography, Volume 44-Number 3, August 2007
546
Figure 1.
Overall Levels of Spatial Segregation of Jobless Men From Employed Men, by Year: 1970–2000 U.S. Census of Population and Housing
1.200 Clustering
Level of Segregation
1.000
0.800
0.600 Centralization 0.400 Evenness 0.200
0.000
Isolation Concentration 1970
1980
1990
2000
Year
of segregation and that segregation increased more in these metropolitan areas. Spatial segregation indices for all 227 metropolitan areas included in this study are available upon request from the author. Overall, jobless men were not very segregated from employed men in 1970, and they have become only moderately more segregated from employed men since 1970. In 1970, jobless men were only slightly more concentrated and clustered than employed men. They were moderately isolated from employed men in their neighborhoods and were relatively evenly distributed throughout the metropolis. They were, however, highly centralized. Between 1970 and 2000, the spatial segregation of jobless men on most dimensions increased moderately, although the magnitude and direction of change varies across dimensions. Jobless men became less evenly distributed, with evenness increasing from 0.214 in 1970 to 0.259 in 1990 before falling slightly to 0.250 in 2000. Isolation grew more sharply, increasing from 0.189 in 1970 to 0.284 in 2000. Clustering increased from 1.024 to 1.052. The concentration of jobless men rose sharply, increasing from 0.057 to 0.152. By contrast, centralization fell from 0.561 to 0.473. A visual examination of patterns of growth in male joblessness in several large metropolitan areas suggests that one reason for this counterintuitive nding is that racial and ethnic succession enabled jobless men to migrate into city neighborhoods further from the central business district (thereby, decreasing centralization), but the concomitant migration of increasing numbers of employed men into larger, less-densely populated suburban neighborhoods in outlying regions of the metropolis led to an increase in concentration. With the exception of the centralization dimension, the average levels of spatial segregation experienced by jobless men remain quite modest. Although overall residential segregation levels for jobless men in U.S. metropolitan areas remained relatively low over the past three decades, the experiences of individual metropolitan areas are more varied. Some metropolitan areas—such as Newark, NJ;
Race and the Spatial Segregation of Jobless Men
547
Detroit, MI, Flint, MI; New Haven, CT; and Milwaukee, WI—experienced noteworthy increases along multiple dimensions, particularly during the 1970s and 1980s. In Detroit, for example, evenness rose by 0.114, isolation by 0.169, concentration by 0.182, and clustering by 0.072. In Newark, evenness rose by 0.105, isolation by 0.126, concentration by 0.232, clustering by 0.068, and centralization by 0.107. By 2000, segregation levels in many of these metropolises were moderately high, with evenness indexes around 0.35, isolation and concentration indexes approaching 0.40, clustering indexes greater than 1.10, and centralization indexes greater than 0.60. However, most metropolitan areas have experienced much smaller increases, relative stability, or even decline over the 1970–2000 period. In Columbus, OH, for example, evenness and clustering fell slightly, centralization fell sharply, and isolation and concentration remained unchanged. In most metropolitan areas, even after steady increases in segregation along most dimensions over the past 30 years, jobless men continued to experience low to moderate levels of spatial segregation. RACIAL DIFFERENCES IN SPATIAL SEGREGATION Although the average jobless man does not appear to be highly segregated from employed men on most dimensions even after several decades of increasing segregation, jobless men from some racial and ethnic groups may nonetheless experience relatively high levels of segregation. Male nonemployment rates vary widely across racial and ethnic groups. Overall rates of male joblessness for white men in this sample of metropolitan areas range from a low of 16.6% in 1990 to a high of 23.4% in 1980. By comparison, nonemployment rates for black men range from a low of 37.2% in 1990 to a high of 41.6% in 2000. Male joblessness rates for Hispanic and Asian men fall between these extremes, with rates for Hispanic men rising from around 26.0% in 1980 and 1990 to 32.8% in 2000 and rates for Asian men increasing from 22.0% in 1980 to 26.8% in 2000. Figure 2 presents weighted average segregation indices on each dimension for white, black, Hispanic, and Asian jobless men for 1980, 1990, and 2000. The weights used in these calculations are proportional to the number of jobless men in the racial or ethnic group in a metropolis in that year. Indices for individual metropolitan areas are available upon request. Mean levels of spatial segregation vary widely among racial and ethnic groups. On all ve dimensions (panels a–e), jobless black men are much more highly segregated from employed men than are other jobless men. Jobless black men are greatly overrepresented in some neighborhoods and grossly underrepresented in others (panel a). More than half of them would need to move to a new neighborhood with fewer jobless black men to achieve an even distribution throughout the metropolis. They tend to reside in neighborhoods with very high levels of male nonemployment (panel b). On average, nearly 4 of every 10 working-age men in their neighborhoods are either unemployed or have dropped out of the labor force. They also tend to be highly concentrated (panel c) and centralized (panel e). Relative to employed men, they occupy a small amount of physical space in the metropolis, and more than 60% of them would need to move farther from the center of a city to achieve a uniform distribution throughout the metropolis. Their neighborhoods do not, however, tend to be surrounded by other neighborhoods with high rates of male nonemployment (panel d). In nearly all metropolitan areas, jobless black men are much more segregated from employed men than are other jobless men. In many major metropolitan areas, particularly in the industrial Northeast and Midwest, jobless black men experience relatively high absolute levels of segregation on multiple dimensions. Evenness values are generally 0.550 or greater, isolation values are 0.400 or greater, concentration values are 0.700 or greater, and centralization values are 0.800 or greater in metropolitan areas such as Chicago, IL; Philadelphia, PA; Detroit, MI; St. Louis, MO/IL; Baltimore, MD; and Cleveland, OH. In a few of these metropolitan areas, including Baltimore and Detroit, jobless black men are also moderately clustered, with index values of 1.20 or greater. In most other metropolitan areas,
Demography, Volume 44-Number 3, August 2007
548
Figure 2.
Average Indices of Spatial Segregation of Jobless Men From Employed Men, by Race and Year: 1980–2000 U.S. Census of Population and Housing a. Evenness
0.600 Black 0.500 Hispanic 0.400 Asian 0.300
0.200
White
0.100
0.000
1980
1990
2000
1990
2000
b. Isolation 0.600
0.500
0.400
0.300
Black Hispanic Asian
0.200
White
0.100
0.000
1980
(continued)
Race and the Spatial Segregation of Jobless Men
549
(Figure 2, continued)
c. Concentration 0.500 Black 0.400 Asian 0.300 Hispanic 0.200
0.100 White 0.000
–0.100
1980
1990
2000
1990
2000
d. Clustering 1.200 1.100 Asian 1.000 0.900
Black White Hispanic
0.800 0.700 0.600 0.500 0.400
1980
(continued)
Demography, Volume 44-Number 3, August 2007
550 (Figure 2, continued)
e. Centralization 0.800
Black
0.700 Hispanic 0.600
Asian
0.500 White 0.400 0.300 0.200 0.100 0.000
1980
1990
2000
jobless black men experience moderate to moderately high levels of spatial segregation on most dimensions; the exception is the clustering dimension, on which values are nearly uniformly low. In a few metropolitan areas with relatively small black populations—including San Jose, CA; Salt Lake City–Ogden, UT; and Riverside–San Bernardino, CA—jobless black men experience low levels of spatial segregation on most dimensions. Jobless white men do not experience the extensive and multidimensional residential segregation from employed men that jobless black men do. Jobless white men tend to be relatively evenly distributed throughout the metropolis (panel a), and they tend to live in neighborhoods in which most men are employed (panel b). Only 20% of them would have to move to neighborhoods with lower rates of white male joblessness to achieve an even distribution throughout the metropolis. Between 75% and 80% of working-age men in their neighborhoods are employed. Their neighborhoods tend to be surrounded by other neighborhoods in which most men are working (panel d); they are, on average, actually slightly less concentrated than employed men (panel c). Although somewhat more congregated around the center of the city than employed men, with mean centralization values ranging from 0.389 to 0.496 over this period, jobless white men are not overwhelmingly concentrated in declining, deteriorating, central-city neighborhoods (panel e). With the exception of a few small metropolitan areas, mostly in the South and the West, segregation levels for jobless white men on all dimensions except centralization are uniformly low to moderate across metropolitan areas. Jobless white men, as result of these factors, are likely to have extensive opportunities to form social ties with employed men in their neighborhoods and surrounding communities, thus experiencing little unmitigated exposure to the deleterious effects of concentrated male joblessness.
Race and the Spatial Segregation of Jobless Men
551
Jobless Hispanic and Asian men neither enjoy the relative ecological advantages that jobless white men do nor suffer the severe ecological disadvantages that jobless black men do. On most dimensions, the segregation of jobless Hispanic and Asian men falls between these extremes. Jobless Hispanic and Asian men are more evenly distributed throughout metropolis (panel a), less concentrated in a small area of physical space (panel c), and less centralized than jobless black men (panel e); but jobless Hispanic and Asian men are more unevenly distributed, concentrated, and centralized than jobless white men. Jobless Hispanic men experience signicantly higher levels of isolation than white men (panel b) but show similarly low levels of clustering. Jobless Asian men experience mean levels of isolation and clustering (panel d) that are comparable to or less than those experienced by jobless white men. Moreover, in no metropolitan area do jobless Hispanic and Asian men experience the high levels of spatial segregation on multiple dimensions that jobless black men experience in metropolitan areas such as Chicago, Detroit, Baltimore, and Philadelphia. In only a few metropolitan areas is evenness greater than 0.500, isolation greater than 0.400, or clustering greater than 1.05 for jobless Hispanic or Asian men. However, jobless Hispanic and Asian men are highly concentrated and centralized in a number of metropolitan areas with large Hispanic and Asian populations. Nonetheless, in only a few of these metropolitan areas are jobless Hispanic and Asian men highly segregated from employed men on more than a single dimension. Individual indices considered in isolation underestimate the true magnitude of spatial segregation jobless black men experience. Not only are jobless black men much more highly segregated from employed men on each of the ve dimensions, but the dimensions of segregation strongly and consistently overlap for them. Table 1 presents the average inter-item correlations between the ve dimensions over this period for each racial and ethnic group. For jobless black men, all ve dimensions are signicantly and positively correlated, with inter-item correlations ranging from 0.234 for evenness and centralization to 0.488 for evenness and clustering. Except for the clustering dimension, which is negatively correlated with the other dimensions for Hispanics and Asians, the dimensions of segregation also are strongly, positively correlated for jobless Hispanic and Asian men. By contrast, the dimensions of segregation for jobless white men only weakly overlap, and in some instances, they are insignicantly or even negatively correlated. Over the past two decades, patterns of spatial segregation have changed only modestly. Jobless men in all racial and ethnic groups have become somewhat less centralized. In most groups, they have also become slightly less isolated from employed men, and their neighborhoods are slightly less clustered. Jobless black and Hispanic men have become slightly more evenly distributed throughout the metropolis, and jobless white and Asian men have become slightly less so. Similarly, jobless white and black men have become slightly less concentrated, and jobless Hispanic and Asian men have become slightly more so. Contrary to my theoretical expectations, segregation along most dimensions did not increase more sharply for jobless black and Hispanic men. In fact, on most dimensions, the gap either did not change or narrowed slightly, especially during the 1990s. Nor are jobless black men more spatially segregated today than they were 20 years ago, as recent studies imply (Farley et al. 2000; Quillian 2003; Wilson 1987, 1996). In fact, the spatial segregation of jobless black men on most dimensions has become more similar to that of other jobless men as important ecological changes, such as the steady decline in black-white segregation (Iceland et al. 2002; Logan et al. 2004; Wilkes and Iceland 2004) and the progressive suburbanization of racial and ethnic minorities (Katz and Lang 2003) have slowly reduced the spatial isolation of blacks in urban America. Nonetheless, jobless black men occupy a uniquely disadvantaged ecological position in comparison with jobless men from other racial and ethnic groups: they are less evenly distributed throughout the metropolis and more isolated from employed men, they are concentrated in a smaller amount of physical space, and their neighborhoods are more proximal
Demography, Volume 44-Number 3, August 2007
552
Table 1.
Average Intercorrelations Among Dimensions of Residential Segregation of Jobless Men From Employed Men, by Race: 1980–2000 U.S. Census of Population and Housing
Dimension White Men Evenness
Evenness
Isolation
Concentration
Clustering
Centralization
1.000
0.237***
0.118
–0.185**
0.110
Isolation
0.237***
1.000
0.133*
–0.143*
0.140*
Concentration
0.118
0.133*
1.000
–0.010
0.123
–0.185**
–0.143*
–0.010
1.000
0.154*
0.110
0.140*
0.123
0.154*
1.000
1.000
0.393***
0.307***
0.488***
0.234**
Isolation
0.393***
1.000
0.343***
0.262***
0.244**
Concentration
0.307***
0.343***
1.000
0.387***
0.466***
Clustering
0.488***
0.262***
0.387***
1.000
0.432***
Centralization
0.234**
0.244**
0.466***
0.432***
1.000
Hispanic Men Evenness
1.000
0.238*
0.570***
–0.139
0.537***
Isolation
0.238*
1.000
0.277***
–0.102
0.225*
Concentration
0.570***
0.277***
1.000
–0.193*
0.451***
–0.193*
1.000
0.038
0.038
1.000
Clustering Centralization Black Men Evenness
Clustering Centralization
–0.139
–0.102
0.537***
0.225*
0.451***
1.000
0.653***
0.508***
–0.513***
0.310*
Isolation
0.653***
1.000
0.359**
–0.530***
0.289*
Concentration
0.508***
0.359**
1.000
–0.264
0.805***
–0.513***
–0.530***
–0.264
0.310*
0.289*
Asian Men Evenness
Clustering Centralization
0.805***
1.000
–0.169
–0.169
1.000
*p < .05; **p < .01; ***p < .001
and are closer to the center of the city. The dimensions of segregation strongly overlap for them, producing a multidimensional layering of segregation. Jobless white men, by contrast, occupy a distinctively advantaged ecological position: they are relatively uniformly distributed throughout the metropolis, live in areas where most men in their neighborhoods and surrounding communities are employed, occupy as much or more physical space than employed men, and are only moderately centralized. The spatial segregation of jobless Hispanic and Asian men typically falls between these two extremes. RACIAL DIFFERENCES IN MULTIVARIATE PERSPECTIVE While jobless black men occupy a uniquely disadvantaged ecological position in the American metropolis, they differ from jobless white, black, Hispanic, and Asian men in many ways that could inuence the spatial segregation they confront. For a number of historical reasons (Lieberson 1981b; Wilson 1987), they differ in their regional concentration and in the types of metropolitan areas in which they tend to reside. They differ in their mean levels of human capital, their relative socioeconomic status, and their relative population
Race and the Spatial Segregation of Jobless Men
553
size (Mincy 2006). Perhaps most importantly, they differ in the levels of residential segregation members of their group experience (Massey and Denton 1993; Wilkes and Iceland 2004). Any of these differences could account for part or all of the observed difference in patterns of spatial segregation. To test whether racial and ethnic group differences in spatial segregation are (1) a consequence of differences in the types of metropolises in which jobless white, black, Hispanic, and Asian men reside, (2) a result of differences in racial and ethnic group characteristics such as their relative size and socioeconomic status, or (3) a reection of distinct patterns of residential segregation, I estimate a series of multivariate models with the levels of the ve dimensions of spatial segregation in 2000 as the dependent variables. Following Massey and Denton (1989) and Wilkes and Iceland (2004), I pool indices for all racial and ethnic groups and regress them on a set of explanatory factors that included dummy variables for each group comparison, as well as variables measuring regional and metropolitan characteristics, racial and ethnic group characteristics, and levels of racial residential segregation. Independent variables in these models have a correlated error structure because the same metropolitan areas are potentially included in the regression model several times. All models are, therefore, estimated by using a generalized linear regression model that accounts for this error structure and produces robust standard errors. For each dimension, four models are estimated. First, models that include only dummy variables representing the groups are estimated. The coefcients for these models serve as a baseline against which to evaluate more complex models. If group differences in spatial segregation reect differences in regional and metropolitan settlement, group size and status characteristics, or racial residential segregation patterns, coefcients for the group indicators will be greatly reduced when these explanatory factors are added to the model. Next, I estimate models that control for differences in regional concentration and metropolitan characteristics that, according to previous studies (Farley and Frey 1994; Jargowsky 1996; Massey and Denton 1993; White, Fong, and Cai 2003; Wilkes and Iceland 2004), inuence residential settlement patterns. Along with dummy variables representing geographic region, (logged) metropolitan area population, the relative size of the minority and foreign-born populations, the newness of the housing stock, per capita family income (in $1,000s), and several measures of type of economy (manufacturing, military, government, and retirement) are included at this stage. Third, measures of a racial or ethnic group’s size and socioeconomic status characteristics (education and income relative to whites) are added to the model. All group characteristic variables refer to the characteristics of the group as a whole. Finally, measures of racial residential segregation are included. For whites, racial residential segregation indices on all dimensions except centralization are set to their theoretical minimum. Because centralization indices compare the distribution of a group with the distribution of space around the central business district, computing centralization indices for whites is possible. Table 2 presents estimated coefcients from these models. Several conclusions can be drawn from the ndings from these models. First, although some metropolitan and group characteristics are strongly associated with the levels of spatial segregation that jobless men experience, these factors account for very little of the racial gap in spatial segregation. Controlling for regional concentration and metropolitan characteristics has a negligible effect on estimated differences between racial and ethnic groups. The distinctively disadvantaged ecological position that jobless black men occupy is, on the whole, not a consequence of the types of metropolitan areas in which these men tend to reside. Nor is this disadvantaged position primarily a reection of differences in the size or socioeconomic status of the group. Although jobless men from racial and ethnic groups that are more afuent and groups with lower rates of high school dropout are less likely to be segregated from employed men, differences in levels of afuence and human capital explain little of the disparity in spatial segregation among jobless white, black, Hispanic, and Asian men.
Demography, Volume 44-Number 3, August 2007
554
Table 2.
Multivariate Regression of Five Segregation Indices on Selected Explanatory Variables: 2000 U.S. Census of Population and Housing
Variable Model 1 Racial/ethnic group White Hispanic Asian Intercept Model 2 Racial/ethnic group White Hispanic Asian
Evenness
Isolation
–0.260*** (0.006) –0.069*** (0.008) –0.131*** (0.008) 0.459*** (0.006)
–0.110*** (0.005) –0.051*** (0.005) –0.106*** (0.007) 0.347*** (0.005)
–0.438*** (0.024) –0.039 (0.024) –0.029 (0.028) 0.409*** (0.024)
–0.016 (0.010) –0.049*** (0.012) –0.117*** (0.010) 0.923*** (0.009)
–0.203*** (0.015) –0.067*** (0.017) –0.113*** (0.026) 0.649*** (0.016)
–0.257*** (0.006) –0.064*** (0.009) –0.128*** (0.009)
–0.109*** (0.004) –0.051*** (0.005) –0.098*** (0.006)
–0.465*** (0.024) –0.039 (0.024) –0.019 (0.026)
–0.014 (0.011) –0.042*** (0.013) –0.109*** (0.010)
–0.225*** (0.014) –0.052** (0.016) –0.075*** (0.022)
0.004 (0.003) 0.165*** (0.030) –0.057 (0.052) –0.003*** (0.000) 0.017 (0.060) 0.379* (0.192) –1.041*** (0.211) 0.238* (0.120) –0.122** (0.045)
–0.029* (0.012) –0.305** (0.108) 0.038 (0.221) 0.001 (0.001) 0.171 (0.235) –0.781 (0.866) –0.857 (0.882) –1.514*** (0.429) –0.380* (0.185)
0.005 (0.003) –0.002 (0.026) –0.130** (0.041) 0.001** (0.000) 0.025 (0.063) –0.516** (0.176) 0.058 (0.207) 0.259* (0.131) 0.097* (0.043)
–0.027* (0.013) –0.016 (0.126) –0.876*** (0.225) –0.002 (0.001) –0.609* (0.243) 0.394 (0.696) –1.714 (0.887) –1.562** (0.558) –0.273 (0.252)
0.013 (0.010) –0.002 (0.008) –0.020* (0.008)
0.069 (0.043) 0.105** (0.038) –0.076* (0.034)
0.000 (0.008) 0.028*** (0.008) 0.041*** (0.006)
–0.079 (0.049) 0.029 (0.042) –0.142*** (0.035)
0.420*** (0.045)
1.059*** (0.197)
0.759*** (0.057)
1.644*** (0.203)
Metropolitan area characteristics Population (logged) 0.010*** (0.003) Percentage minority 0.022 (0.024) Percentage foreign-born –0.173*** (0.046) Per capita income 0.000 (in $1,000s) (0.000) Percentage manufacturing 0.005 (0.050) Percentage government 0.161 (0.172) Percentage military –0.913*** (0.165) Percentage retirement –0.014 (0.095) Percentage new housing –0.159*** (0.045) Region Northeast 0.019* (0.009) Midwest 0.021** (0.007) South 0.024*** (0.006) Intercept
0.332*** (0.040)
Concentration
Clustering
Centralization
(continued)
Race and the Spatial Segregation of Jobless Men
555
(Table 2, continued)
Variable
Evenness
Isolation
–0.211*** (0.011)
–0.100*** (0.008)
–0.302*** (0.037)
–0.132*** (0.020)
–0.182*** (0.029)
Hispanic
–0.115*** (0.008)
–0.074*** (0.008)
–0.130** (0.043)
–0.109*** (0.011)
–0.097*** (0.026)
Asian
–0.137*** (0.011)
–0.108*** (0.010)
–0.030 (0.046)
–0.068*** (0.010)
–0.076 (0.039)
0.005 (0.003)
–0.029* (0.013)
0.015*** (0.003)
–0.026 (0.013)
Model 3 Racial/ethnic group White
Metropolitan area characteristics Population (logged) 0.011*** (0.003)
Concentration
Clustering
Centralization
Percentage minority
–0.007 (0.026)
0.150*** (0.028)
–0.360*** (0.110)
0.008 (0.024)
–0.038 (0.125)
Percentage foreign-born
–0.244*** (0.058)
–0.097* (0.048)
–0.064 (0.227)
–0.207*** (0.053)
–0.932*** (0.230)
Per capita income (in $1,000s)
0.001*** (0.000)
–0.002*** (0.000)
0.002 (0.001)
0.002*** (0.000)
–0.001 (0.002)
Percentage manufacturing –0.098 (0.053)
–0.037 (0.060)
–0.003 (0.243)
–0.058 (0.054)
–0.693** (0.251)
Percentage government
0.481* (0.199)
0.563** (0.202)
–0.302 (0.874)
–0.243 (0.142)
0.642 (0.685)
Percentage military
–0.385* (0.169)
–0.784*** (0.218)
0.021 (0.891)
0.717*** (0.213)
–1.262 (0.919)
Percentage retirement
–0.025 (0.092)
0.225* (0.107)
–1.486*** (0.443)
0.168 (0.125)
–1.568** (0.562)
Percentage new housing
–0.164*** (0.044)
–0.127** (0.041)
–0.385* (0.189)
0.104* (0.041)
–0.275 (0.251)
0.017 (0.009)
0.014 (0.010)
0.059 (0.043)
0.001 (0.008)
–0.082 (0.051)
Midwest
0.027*** (0.008)
0.003 (0.009)
0.114** (0.038)
0.026*** (0.008)
0.033 (0.043)
South
0.022*** (0.006)
–0.019* (0.007)
–0.083* (0.034)
0.030*** (0.007)
–0.145*** (0.036)
–0.011 (0.010)
–0.218*** (0.047)
0.291*** (0.025)
–0.039 (0.031)
0.180*** (0.038)
0.585** (0.187)
0.337*** (0.040)
0.291* (0.122)
0.062** (0.023)
0.065 (0.132)
–0.032 (0.025)
0.035 (0.110)
0.300*** (0.058)
0.861*** (0.238)
0.488*** (0.055)
1.509*** (0.222)
Region Northeast
Group characteristics Percentage of population –0.051** (0.016) Percentage high school dropout
0.352*** (0.037)
Per capita income relative 0.071** to whites’ (0.026) Intercept
0.149 (0.054)
(continued)
Demography, Volume 44-Number 3, August 2007
556 (Table 2, continued)
Variable
Evenness
Isolation
Concentration
Clustering
0.065*** (0.018)
0.062* (0.025)
0.035 (0.052)
–0.010 (0.015)
–0.092*** (0.036)
Hispanic
–0.020* (0.009)
–0.023* (0.010)
0.023 (0.036)
–0.023* (0.009)
–0.055*** (0.022)
Asian
–0.043*** (0.010)
–0.055*** (0.014)
0.072* (0.034)
0.021* (0.009)
–0.028 (0.034)
0.000 (0.003)
–0.040*** (0.010)
0.001 (0.003)
–0.041*** (0.010)
Model 4 Racial/ethnic group White
Metropolitan area characteristics Population (logged) –0.002 (0.002)
Centralization
Percentage minority
–0.036 (0.022)
0.045 (0.033)
–0.138 (0.100)
–0.041 (0.026)
0.133 (0.113)
Percentage foreign-born
–0.127** (0.044)
–0.040 (0.047)
0.127 (0.192)
–0.044 (0.046)
–0.963*** (0.208)
Per capita income (in $1,000s)
0.000*** (0.000)
–0.003*** (0.000)
0.002 (0.001)
0.001* (0.000)
0.002 (0.001)
Percentage manufacturing –0.072 (0.047)
0.000 (0.059)
0.046 (0.209)
–0.056 (0.063)
–0.415* (0.202)
Percentage government
0.502* (0.199)
0.533** (0.206)
–0.622 (0.765)
–0.146 (0.138)
–0.091 (0.532)
Percentage military
–0.165 (0.174)
0.778*** (0.199)
0.231 (0.845)
0.783*** (0.237)
–0.865 (0.577)
Percentage retirement
–0.065 (0.076)
0.252* (0.106)
–0.463 (0.382)
0.146 (0.123)
–1.705*** (0.432)
Percentage new housing
–0.071 (0.038)
–0.076 (0.041)
0.026 (0.199)
0.166*** (0.050)
–0.489* (0.199)
–0.009 (0.008)
0.007 (0.011)
0.056 (0.037)
–0.017 (0.010)
0.054 (0.041)
Midwest
0.001 (0.007)
–0.006 (0.009)
0.078* (0.035)
0.012 (0.009)
0.056 (0.034)
South
0.011* (0.005)
–0.022** (0.008)
–0.068* (0.032)
0.026*** (0.007)
–0.058 (0.031)
Region Northeast
(continued)
Second, these findings highlight the deleterious effects of high levels of racial residential segregation for jobless black men and their communities. The distinctively disadvantaged ecological position that jobless black men occupy relative to other jobless men is largely a reection of the extreme levels of racial residential segregation experienced by blacks in the United States. High levels of racial residential segregation, in combination with disproportionately high rates of male nonemployment for black men, concentrate jobless men in African American communities, as Wilson (1987) and Massey (1990) hypothesized. Racial residential segregation, in turn, isolates jobless black men from employed men, concentrates them in a small amount of physical space, and clusters their neighborhoods near the city center.
Race and the Spatial Segregation of Jobless Men
557
(Table 2, continued)
Variable
Evenness
Model 4 (cont.) Group characteristics Percentage of population –0.082*** (0.013)
Isolation
Concentration
Clustering
Centralization
–0.152*** (0.018)
–0.066 (0.061)
0.227*** (0.018)
–0.015 (0.024)
0.033 (0.042)
0.100 (0.154)
0.115** (0.038)
0.217* (0.096)
Per capita income relative 0.031 to whites’ (0.021)
0.045 (0.026)
–0.099 (0.090)
–0.077*** (0.023)
0.044 (0.090)
Racial residential segregation Evenness 0.482*** (0.028)
––
––
Percentage high school dropout
0.121*** (0.035)
––
––
––
––
0.706*** (0.066)
––
––
Isolation
––
Concentration
––
––
Clustering
––
––
––
Centralization
––
––
––
Intercept
0.183*** (0.045)
0.217*** (0.030)
––
0.380*** (0.057)
0.562** (0.197)
0.352*** (0.024) –– 0.385*** (0.059)
–– 0.485*** (0.115) 1.050*** (0.191)
Notes: Numbers in parentheses are standard errors. The omitted category for the group variable is black, and the omitted category for region is West. *p < .05; **p <.01; ***p < .001
CONCLUSION Many studies over the past two decades have documented worse employment outcomes for black men than for men from other racial and ethnic groups (D’Amico and Maxwell 1995; Edelman et al. 2006; Lichter 1988; Mare and Winship 1984; Mincy 2006; O’Regan and Quigley 1996; Rivkin 1995; Stratton 1993; Vedder and Gallaway 1992). Black men are employed at signicantly lower rates than other men (Edelman et al. 2006; Lichter 1988; Mare and Winship 1984; Vedder and Gallaway 1992), and the gap between the employment rates of black men and white men is expanding (D’Amico and Maxwell 1995; Edelman et al. 2006; Lichter 1988; Mare and Winship 1984; Rivkin 1995; Stratton 1993; Vedder and Gallaway 1992). Black men experience greater employment instability, which leads to substantial accumulated decits in work experience over the life course (Tienda and Stier 1996). The quality of black men’s jobs has declined relative to that of white men’s jobs (Gittleman and Howell 1995), and black men continue to experience relatively high levels of occupational segregation (Spriggs and Williams 1996). Stark differences between black and white male employment patterns persist even after differences in human capital, local labor market conditions, and other factors associated with employment outcomes are controlled (D’Amico and Maxwell 1995; Lichter 1988; Rivkin 1995; Stratton 1993). One possible reason for these persistent differences is that jobless black men occupy a uniquely disadvantaged ecological niche. This research shows that jobless black men are much more segregated from employed men than are men from other racial and ethnic groups: black men are less uniformly
558
Demography, Volume 44-Number 3, August 2007
distributed throughout the metropolis, more isolated from employed men, concentrated in a smaller amount of physical space, and congregated closer to the center of the city. Only for jobless black men do high levels of spatial segregation and overlapping dimensions of segregation create a multidimensional layering of spatial segregation. In many metropolitan areas, including Chicago, Detroit, Philadelphia, Baltimore, and St. Louis, jobless black men experience relatively high absolute levels of segregation from employed men on multiple dimensions. This uniquely disadvantaged ecological position is likely to create distinctive problems for jobless black men and their neighborhoods. Compared with jobless men from other racial and ethnic groups, jobless black men have much more limited opportunities for social contact with employed men in their neighborhoods, making it more difcult for them to access informal job networks and learn about new employment opportunities. The difculty that these men experience in obtaining information about new employment opportunities, whether by formal and informal methods, is further magnied because they tend to be concentrated in a small number of neighborhoods congregated near the center of the city, where employers are unlikely to recruit workers (Neckerman and Kirschenman 1991). Even when employment can be secured, this disadvantageous ecological position substantially increases commuting time and employment costs, which may account for higher reservation wages (Holzer 1986; Petterson 1998) and job quit rates for black men (Zax 1989). Highly concentrated male joblessness, produced by the multidimensional layering of segregation jobless black men in urban America experience, also creates unique challenges for the neighborhoods in which these men reside. Widespread male joblessness creates social and economic conditions that foster illicit economic activity and other criminal activity (Almgren et al. 1998; Anderson 1990, 1999; Sampson 1987; Venkatesh 2006); disrupts family formation processes (Bennett, Bloom, and Craig 1989) and increases nonmarital childbearing (Ku, Sonenstein, and Pleck 1993); drains distressed neighborhoods of muchneeded resources (Wilson 1987, 1996); and fosters a culture of dependency, despair, and joblessness that impedes achievement and helps to transmit disadvantage across generations (MacLeod 1987; Wilson 1987, 1996). Because of the unique problems this multidimensional spatial segregation creates for jobless black men, studies that examine the spatial mismatch between these men’s residential location and the location of jobs may underestimate the effect that spatial factors have on black-white differences in employment outcomes (Cohn and Fossett 1996; Ihlanfeldt and Sjoquist 1990, 1991; Jencks and Mayer 1990; Mouw 2000; Raphael 1998; Stoll 1996). The ecological disadvantage that jobless black men experience is only partially attributable to the distance between their residential neighborhoods and job opportunities. It is also a consequence of their limited opportunities for contact with employed men in their neighborhoods as well as the deleterious effects of concentrated joblessness and other forms of social and economic disadvantage on achievement, motivation, and access to opportunity. One particularly important direction for future research, therefore, would be to examine the effects that these other forms of ecological disadvantage have on racial and ethnic differences in employment outcomes. Both metropolitan and group characteristics contribute to racial and ethnic differences in the spatial segregation of jobless men. However, I nd that the distinctively disadvantaged ecological position of jobless black men is primarily attributable to group differences in patterns of racial residential segregation. Racial residential segregation condemns jobless black men to a residential environment in which widespread joblessness is the norm and opportunities for contact with employed men are limited. As Massey and Denton (1993:2) noted more than a decade ago, “residential segregation is not a neutral fact; it systematically undermines the social and economic well-being of blacks in the United States.”
Race and the Spatial Segregation of Jobless Men
559
REFERENCES Almgren, G., A. Guest, G. Immerwahr, and M. Spittel. 1998. “Joblessness, Family Disruption, and Violent Death in Chicago, 1970–90.” Social Forces 76:1465–93. Anderson, E. 1990. Streetwise: Race, Class, and Change in an Urban Community. Chicago: University of Chicago Press. ———. 1991. “The Story of John Turner.” Pp. 147–79 in Drugs, Crime, and Social, Isolation: Barriers to Urban Opportunity, edited by A.V. Harrell and G.E. Peterson. Washington, DC: The Urban Institute. ———. 1999. Codes of the Street: Decency, Violence, and the Moral Life of the Inner City. New York: W.W. Norton and Company, Inc. Auletta, K. 1982. The Underclass. New York: Random House. Bennett, N.G., D.E. Bloom, and P.H. Craig. 1989. “The Divergence of Black and White Marriage Patterns.” American Journal of Sociology 95:692–722. Berry, B.J.L. and Y.S. Cohen. 1973. “Decentralization of Commerce and Industry: The Restructuring of Metropolitan America.” Urban Affairs and Annual Review 8:431–55. Berry, B.J.L. and J. Kasarda. 1977. Contemporary Urban Ecology. New York: Macmillan. Cohn, S. and M. Fossett. 1996. “What Spatial Mismatch?: The Proximity of Blacks to Employment in Boston and Houston.” Social Forces 75:557–72. D’Amico, R. and N.L. Maxwell. 1995. “The Continuing Signicance of Race in Minority Male Joblessness.” Social Forces 73:969–91. Duncan, O.D. and B. Duncan. 1957. The Negro Population of Chicago: A Study of Residential Succession. Chicago: University of Chicago Press. Duncan, O.D., R.P. Cuzzort, and B. Duncan. 1961. Statistical Geography: Problems in Analyzing Areal Data. Glencoe, IL: Free Press. Edelman, P., H.J. Holzer, and P. Offner 2006. Reconnecting Disadvantaged Young Men. Washington, DC: Urban Institute Press. Farley, R., S. Danziger, and H.J. Holzer. 2000. Detroit Divided. New York: Russell Sage Foundation. Farley, R. and W.H. 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. Frey, W.H. and D. Myers. 2002. “Neighborhood Segregation in Single-Race and Multirace America: A Census 2000 Study in Cities and Metropolitan Areas.” Working paper. Fannie Mae Foundation, Washington, DC. GeoLytics. 2003. Neighborhood Change Database. East Brunswick, NJ: GeoLytics, Inc. Gittleman, M.B. and D.R. Howell. 1995. “Changes in the Structure and Quality of Jobs in the United States: Effects by Race and Gender, 1973–1990.” Industrial and Labor Relations Review 48:420–40. Glaeser, E.L. and J. Vigdor. 2001. “Racial Segregation in the 2000 Census: Promising News.” Center on Urban and Metropolitan Policy, The Brookings Institution, Washington, DC. Glaster, G.C. 1984. “On the Measurement of Metropolitan Decentralization of Blacks and Whites.” Urban Studies 21:465–70. Granovetter, M. 1974. Getting a Job: A Study of Contacts and Careers. Chicago: University of Chicago Press. Holzer, H.J. 1986. “Reservation Wages and Their Labor Market Effects for Black and White Male Youth.” Journal of Human Resources 21(2):157–77. Iceland, J., D.H. Weinberg, and E. Steinmetz. 2002. Racial and Ethnic Residential Segregation in the United States: 1980–2000. U.S. Census Bureau, Census Special Report, CENSR-3. Washington, DC: U.S. Government Printing Ofce. Ihlanfeldt, K. and D.L. Sjoquist. 1990. “Job Accessibility and Racial Differences in Youth Employment Rates.” American Economic Review 80(1):267–76. ———. 1991. “The Effect of Job Access on Black and White Youth Employment: A Cross-Sectional Analysis.” Urban Studies 28:255–65.
560
Demography, Volume 44-Number 3, August 2007
Jakubs, J.F. 1977. “Residential Segregation; The Taeuber Index Reconsidered.” Journal of Regional Science 17:281–303. ———. 1979. “A Consistent Conceptual Denition of the Index of Dissimilarity.” Geographical Analysis 11:315–21. ———. 1981. “A Distance-Based Segregation Index.” Journal of Socio-Economic Planning Sciences 15:129–36. Jargowsky, P.A. 1996. “Take the Money and Run: Economic Segregation in U.S. Metropolitan Areas.” American Sociological Review 61:984–98. ———. 1997. Poverty and Place: Ghettos, Barrios, and the American City. Russell Sage Foundation, New York. ———. 2003. “Stunning Progress, Hidden Problems: The Dramatic Decline of Concentrated Poverty in the 1990s.” Living Cities Census Series. Washington, DC: Brookings Institution. Jargowsky, P.A. and M.J. Bane. 1991. “Ghetto Poverty in the United States, 1970–1980.” Pp. 235–73 in The Urban Underclass, edited by C. Jencks and P.E. Peterson. Washington, DC: The Brookings Institution. Jargowsky, P.A. and R. Yang. 2006. “The ‘Underclass’ Revisited: A Social Problem in Decline.” Journal of Urban Affairs 28(1):55–70. Jencks, C. and S.E. Mayer. 1990. “Residential Segregation, Job Proximity, and Black Job Opportunities.” Pp. 187–222 in Inner-City Poverty in the United States. Washington, DC: National Academy Press. Kasarda, J.D. 1993. “Inner-city Concentrated Poverty and Neighborhood Distress: 1970–1990.” Housing Policy Debate 4:253–302. ———. 1995. “Industrial Restructuring and the Changing Location of Jobs.” Pp. 215–67 in State of the Union: America in the 1990s, Vol. 1, edited by R. Farley. New York: Sage Foundation. Katz, B. and R.E. Lang, eds. 2003. Redening Cities and Suburbs: Evidence From Census 2000. Volume I. Washington, DC: Brookings Institution Press. Kingsley, G.T. and K.L.S. Petit. 2003. “Concentrated Poverty: A Change in Course.” Neighborhood Change in Urban America Series, No. 2. Washington, DC: Urban Institute. Kirschenman, J. and K.M. Neckerman. 1991. “‘We Would Love to Hire Them But…’: The Meaning of Race for Employers.” Pp. 203–34 in The Urban Underclass, edited by C. Jencks and P.E. Peterson. Washington, DC: Brookings Institution Press. Krivo, L.J., R.D. Peterson, H. Rizzo, and J.R. Reynolds. 1998. “Race, Segregation, and the Concentration of Disadvantage: 1980–1990.” Social Problems 45:61–80. Ku, L., F.L. Sonenstein, and J.H. Pleck. 1993. “Neighborhood, Family, and Work: Inuences on the Premarital Behaviors of Adolescent Males.” Social Forces 72:479–503. Lee, M.R. 2000. “Concentrated Poverty, Race, and Homicide.” Sociological Quarterly 41:189–206. Lichter, D.T. 1988. “Racial Differences in Underemployment in American Cities.” American Journal of Sociology 93:771–92. Lieberson, S. 1981a. “An Asymmetrical Approach to Segregation.” Pp. 61–82 in Ethnic Segregation in Cities, edited by C. Peach, V. Robinson, and S. Smith. Athens, GA: University of Chicago Press. ———. 1981b. Piece of the Pie: Black and White Immigrants Since 1880. Berkeley, CA: University of California Press. Logan, J., B. Stults, and R. Farley. 2004. “Segregation of Minorities in the Metropolis; Two Decades of Change.” Demography 41:1–22. MacLeod, J. 1987. Ain’t No Makin’ It: Leveled Aspirations in a Low-Income Neighborhood. Boulder: Westview Press. Mare, R.D. and C. Winship. 1984. “The Paradox of Lessening Racial Inequality and Joblessness Among Black Youth: Enrollment, Enlistment, and Employment, 1964–1981.” American Sociological Review 49:39–55. Massey, D.S. 1990. “American Apartheid: Segregation and the Making of the Underclass.” American Journal of Sociology 96:329–57.
Race and the Spatial Segregation of Jobless Men
561
Massey, D.S. and N.A. Denton. 1987. “Trends in the Residential Segregation of Blacks, Hispanics, and Asians: 1970–1980.” American Sociological Review 52:802–25. ———. 1988. “The Dimensions of Residential Segregation.” Social Forces 67:281–315. ———. 1989. “Hypersegregation in U.S. Metropolitan Areas: Black and Hispanic Segregation Along Five Dimensions.” Demography 26:378–79. ———. 1993. American Apartheid: Segregation and the Making of the Underclass. Cambridge, MA: Harvard University Press. Massey, D.S. and M. Eggers. 1990. “The Ecology of Inequality: Minorities and the Concentration of Poverty, 1970–1980.” American Journal of Sociology 95:1153–88. ———. 1993. “The Spatial Concentration of Afuence and Poverty During the 1970s.” Urban Affairs Quarterly 29:299–315. Massey, D.S., M.J. White, and V. Phua. 1996. “The Dimensions of Segregation Revisited.” Sociological Methods and Research 25:172–206. Mincy, R.B. 2006. Black Males Left Behind. Washington, DC: Urban Institute Press. Morenoff, J.D. and M. Tienda. 1997. “Underclass Neighborhoods in Temporal and Ecological Perspective.” Annals of the American Academy of Political and Social Science 551:59–72. Mouw, T. 2000. “Job Relocation and the Racial Employment Gap in Unemployment in Detroit and Chicago, 1980 to 1990.” American Journal of Sociology 65:730–53. Neckerman, K.M. and J. Kirschenman. 1991. “Hiring Strategies, Racial Bias, and Inner-City Workers.” Social Problems 38:433–47. O’Regan, K.M. and J.M. Quigley. 1996. “Teenage Employment and the Spatial Isolation of Minority and Poverty Households.” Journal of Human Resources 31:692–702. Parker, K.F. and M.V. Pruitt. 2000. “Poverty, Poverty Concentration, and Homicide.” Social Science Quarterly 81:555–70. Petterson, S.M. 1998. “Black-White Differences in Reservation Wages and Joblessness: A Replication.” Journal of Human Resources 33:758–70. Quillian, L. 2003. “The Decline of Male Employment in Low-Income Black Neighborhoods, 1950– 1990.” Social Science Research 32:220–50. Raphael, S. 1998. “The Spatial Mismatch Hypothesis and Black Youth Joblessness: Evidence From the San Francisco Bay Area.” Journal of Urban Economics 43:79–113. Rivkin, S.G. 1995. “Black/White Differences in Schooling and Employment.” Journal of Human Resources 30:826–52. Sampson, R.J. 1987. “Urban Black Violence: The Effect of Male Joblessness and Family Disruption.” American Journal of Sociology 93:348–82. ———. 1997. “Collective Regulation of Adolescent Misbehavior: Validation Results From Eighty Chicago Neighborhoods.” Journal of Adolescent Research 12:227–44. Sassen, S. 1990. “Economic Restructuring and the American City.” Annual Review of Sociology 16:465:90. South, S.J. and K.D. Crowder. 1999. “Neighborhood Effects on Family Formation: Concentrated Poverty and Beyond.” American Sociological Review 64:113–32. Spriggs, W.E. and R.M. Williams. 1996. “A Logit Decomposition Analysis of Occupational Segregation: Results for the 1970s and 1980s.” Review of Economics and Statistics 78:348–55. Stoll, M.A. 1996. “Distance or Discrimination? The Convergence of Space and Race in Understanding Metropolitan Racial Differences in Employment.” Sage Race Relations 21:3–25. Stratton, L.S. 1993. “Racial Differences in Men’s Unemployment.” Industrial and Labor Relations Review 46:451–63. Sugrue, T.J. 1996. The Origins of the Urban Crisis: Race and Inequality in Postwar Detroit. Princeton, NJ: Princeton University Press. Swanstrom, T., C. Casey, R. Flack, and P. Dreier. 2004. Pulling Apart: Economic Segregation Among Suburbs and Central Cities in Major Metropolitan Areas. Washington, DC: Brookings Institution, Center on Urban and Metropolitan Policy.
562
Demography, Volume 44-Number 3, August 2007
Tienda, M. and H. Stier. 1996. “Generating Labor Market Inequality: Employment Opportunities and the Accumulation of Disadvantage.” Social Problems 43:147–65. Vedder, R.K. and L. Gallaway. 1992. “Racial Differences in Unemployment in the United States, 1890–1990.” Journal of Economic History 52:696–702. Venkatesh, S.A. 2006. Off the Books: The Underground Economy of the Urban Poor. Cambridge, MA: Harvard University Press. Wagmiller, R.L. 2004. “Lost Opportunities: Racial Unrest, Violent Crime, and Emergence of the Jobless Ghetto.” Unpublished doctoral dissertation. Department of Sociology, University of Chicago. White, M.J. 1983. “The Measurement of Spatial Segregation.” American Journal of Sociology 88:1008–19. ———. 1987. American Neighborhoods and Residential Differentiation. New York: Russell Sage Foundation. White, M.J., E. Fong, and Q. Cai. 2003. “The Segregation of Asian-Origin Groups in the United States and Canada.” Social Science Research 32:148–67. Wilkes, R. and J. Iceland. 2004. “Hypersegregation in the Twenty-First Century.” Demography 41:23–36. Wilson, W.J. 1987. The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy. Chicago: University of Chicago Press. ———. 1996. When Work Disappears: The World of the New Urban Poor. New York: Random House. Zax, J.S. 1989. “Quits and Race.” Journal of Human Resources 24:469–93.