Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
- Residential Segregation, Job Proximit:y' and Black Job Opportunities CHRISTOPHER JENCKS AND SUSAN E. MAYER In 1968 John Kain published a seminal paper in which he argued that the high level of joblessness among urban blacks was partly attributable to the fact that a growing fraction of urban jobs-especially blue-collar manufacturing jobs had moved to the suburbs while exclusionary housing practices had kept blacks penned up in central cities. Kain argued that the resulting "spatial mismatch" reduced both employers' willingness to hire black workers and black workers' ability to find jobs that were in principle open to them. In assessing these arguments it is important to draw a sharp distinction between Kain's demand-side and supply-side stories. On the demand side, Kain argued that suburbanization of jobs was likely to reduce employers' willingness to hire black workers because many suburban firms feared that bringing black workers into an all-white suburb would offend white residents. Yet Kain also noted the possibility that residential segregation might benefit blacks, since employers in all-black areas might be more willing to hire blacks than employers in the mixed areas that would come into existence if cities were not segregated. Kain's demand-side analysis of the way residential segregation and the suburbanization of blue-collar employment affected demand for black workers rested on social considerations, many of which appear to have changed since 1968. Perhaps for this reason, subsequent research has seldom tried lo test the validity of Kain's demand-side argument. Yet for those interested in explaining aggregate black employment or earnings, this part of his argument is probably more important than his supply-side argument. . On the supply side, Kain argued that even when suburban employers were willing to hire blacks, the fact that blacks lived farther than whites from 187
188 INNER-CITY POVERTY IN THE UNITED STATES suburban jobs meant that blacks were less likely to hear about suburban job vacancies before they were filled. In addition, he argued that even when blacks could get suburban jobs, the fact that they had to spend more time and money than whites on commuting was likely to result in lower rates of labor force participation among blacks, who would more often conclude that working was not worth the bother. Unlike Kain's demand-side argument, his analysis of the way distance cut the supply of black workers available to suburban firms was based largely on geographic considerations that have not changed in any fundamental way since 1968. Most blacks still live in the central city, and there are probably even fewer good blue-collar jobs available in the central city today than in the late 1960s. Central-city jobs are increasingly likely to require high levels of education (Kasarda, 1989), and while the educational gap between young black and young white workers has narrowed dramatically, there are still substantial disparities in academic skill.t As a result, many scholars, political leaders, and journalists still view the spatial mismatch hypothesis as a plausible explanation of rising black joblessness.2 This chapter reviews the currently available evidence regarding Kain's major hypotheses. The first section looks at Kain's demand-side arguments, reviewing both the theoretical reasons for thinking that residential segrega- tion might reduce demand for black workers and the empirical evidence for this hypothesis. The second section turns to Kain's supply-side arguments, discussing the effects of living far from major centers of employment on workers' job prospects. The third section reviews evidence on the closely related question of whether black workers who live in the suburbs find better jobs than those who live in the central city. The fourth section com- pares the earnings of central-ci~ blacks who commute to the suburbs to the earnings of similar blacks who work in the central-city ghetto. The clos- ing section discusses the implications of the findings and pressing research needs. We do not take up the question of whether a neighborhood's mean socioeconomic status (SES) affects adult black workers' economic prospects. Wilson (1987) and others have suggested that living in an urban ghetto where few adults have steady jobs may sap an individuals motivation to work In addition, Wilson and others have suggested that living in a neighborhood where nobody has a good job makes it harder to find such a ~ lbe proportion of 25- to 29-year-old nonwhites without a high school diploma fell from 41.6 percent in 1970 to 15.7 percent in 1986. Among whites, the proportion declined from 22.2 to 13.5 percent. The National Assessment of Educational Progress shows that the disparity in test performance between black and white 17-year-olds who were enrolled in school also narrowed after 1970, though not as much as the disparity in years of school completed (National Center for Education Statistics, 1988, Tables 8 and 81-87.) 2 Wilson (1987) and Kasarda (1989) both endorse variants of this hypothesis.
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 189 job even if you try. These hypotheses strike us as plausible, but we have not been able to find any empirical studies that assess their validity for adults. The absence of such research reflects the complexity of the problem. An indiv~dual'.s employment status and earnings can correlate with the mean SES of his or her neighborhood for two different reasons. Living in a high-SES neighborhood can increase your chances of finding a good job, but finding a good job can also enable you to move to a high- SES neighborhood. As a result, individuals in low-SES neighborhoods would be less likely to hold steady, well-paid jobs even if their place of residence had no effect whatever on their job prospects. We would need longitudinal data to estimate the contribution of selective migration to the correlation between earnings and community characteristics. No such data were available when we did this review, so there was no way to estimate the effect of a neighborhood's mean SES on adults' job opportunities.3 RESIDENTIAL SEGREGATION AND DEMAND FOR BLACK WORKERS Theoretical Issues Kain argued that a firm's attitude toward hiring black workers was likely to vary with the racial mix of the neighborhood in which it was located. Following economic convention, we refer to such variations in firms' attitudes toward hiring blacks as variations in the demand for black labor. Noneconomists should bear in mind that this usage refers to more than whether firms prefer a black or white worker for a given job. A firm's demand for black labor is highest when it prefers black workers to equally competent white workers, somewhat lower when it is neutral between equally competent blacks and whites, still lower when it prefers whites to equally competent blacks, and lowest of all when it refuses to hire blacks at all. A firm can express its demand for black or white workers in several ways. Firms that prefer white to black workers must set wages high enough to attract white workers, and they must then devise plausible criteria for hiring white rather than black applicants. ~ minimize the cost of preferring whites, these firms are also likely to locate where white workers are relatively cheap and black workers are relatively scarce. Firms that prefer white workers therefore tend to locate either in white rural areas or on the fringes of major metropolitan areas. Firms with a weaker preference 3Because of increased interest in neighborhood effects, the Panel Study of Income Dynamics has been adding data to its files on the characteristics of the census tracts in which respondents lived.
190 INNER-C~ BERM IN THE UNFED STATES for white workers are more likely to locate in the central city, where a large pool of cheap black labor is readily available. Kain did not discuss the effect of firms' racial preferences on their location decisions. Instead, he assumed that location decisions were exoge- nous and discussed their effect on a firm's demand for black workers. He argued that since local whites might resent the firm's bringing black workers into their all-white neighborhood, firms in such neighborhoods would try to avoid hiring blacks. Kain was mainly concerned with black workers' access to manufacturing jobs. The argument that manufacturing firms in white areas would avoid hiring blacks to avoid local resentment does not strike us as very plausible. Blue-collar workers who commute to large suburban manufacturing plants typically drive to the plant's parking lot, go to their job, put in their time, go back to the parking lot, and drive home. They are unlikely to have much contact with the community in which they work especially if they are black and the community is entirely white. Local resentment is therefore likely to be minimal. Furthermore, suburban whites who do resent the fact that a local plant hires blacks have few effective mechanisms for expressing their displeasure unless they resort to violence, which is rare. (Whites who work in a plant have many effective ways of expressing racial resentment, but they can do this no matter where the plant is located.) Kain's argument seems more plausible when applied to the service sector than when applied to manufacturing, and it seems especially plausible when applied to small firms that deal directly with the public. It is easy to imagine suburban restaurants, grocery stores, gas stations, dry cleaners, and repair services refusing to hire blacks because they think their customers prefer dealing with whites and will take their trade elsewhere if the firm does not cater to their prejudices. Kain did not discuss the likely behavior of firms in racially mixed areas. His arguments imply, however, that manufacturing firms in mixed areas would be indifferent between equally competent black and white workers, since they would not be worried about bringing blacks into mixed areas. The picture is more complicated for retail firms. If both blacks and whites preferred doing business with retail firms that hired people of their own race, and if blacks and whites who lived in mixed neighborhoods had roughly equal incomes, retail firms in mixed areas would presumably try to hire a mix of employees similar to the mix of local residents. If whites were quite averse to firms that hired blacks while blacks were not so averse to firms that hired only whites, or if whites had far more money than blacks, most retail firms in mixed areas might end up discriminating against blacks. Finally, Kain noted that firms in all-black areas might hire more blacks than an average firm. This could occur for two reasons. First, ghetto firms
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORlUNITIES 191 would hire disproportionate numbers of blacks if white workers were un- willing to work in the ghetto or demanded "combat pay" for working there.4 Second, retail firms in the ghetto would hire disproportionate numbers of blacks if black customers preferred dealing with blacks. If these general hypotheses are correct, the persistence of all-black neighborhoods can either increase or decrease aggregate demand for black workers. The net effect of a residential ghetto on demand for black workers will depend on whether the increase in demand within the ghetto offsets the decline elsewhere. The balance between the gains in black employment within the ghetto and the losses elsewhere should depend on two factors: the absolute number of jobs inside and outside the ghetto, and the relative strength of ghetto firms' preference for blacks compared to nonghetto firms' preference for whites. All the factors that influence the costs and benefits of residential segregation are likely to have changed over the past 30 years. Thus even if we were to accept Kain's claim that residential segregation reduced aggregate demand for black workers in Chicago and Detroit in the l950s, we could not assume segregation had the same effect today. The effects of residential segregation are also likely to vary from one part of the country to another. Finding that residential segregation lowers demand for black workers in Chicago and Detroit may not, therefore, imply that it has the same effect in California. Finally, residential segregation is likely to have different effects on different ethnic groups. Thus even if we found that Chinese immigrants did better when they moved to highly segregated cities, the same might not hold for Mexican immigrants, much less for blacks. Empirical Evidence Kain's analysis of Chicago and Detroit employment patterns in the l950s convinced him that blacks would have gotten a larger share of all jobs if there were no ghetto, but this was not because his model implied that residential desegregation would increase demand for black workers. Kain's model implied that residential desegregation would have no effect whatever on overall demand for black workers. Residential desegregation increased black employment only because it brought black workers closer to most jobs and thereby increased the supply of black workers available to fill these jobs at any given wage. 4A downward shift in the local supply of white worked does not necessarily lead to an upward shift in local demand for black worked. When the local supply of white worked available at a given price falls, firms may move elsewhere rather than hiring more blacks. We return to this point below.
192 INNER-CITY POVERTY IN THE UNITED STATES Kain used the proportion of nonwhites in a neighborhood (R) as a proxy for demand and used the number of miles from a neighborhood to the edge of a city's major ghetto (dm) as a proxy for supply. He assumed that both these measures had linear effects on blacks' share of neighborhood employment. Using these two measures to predict nonwhites' share of total employment (W1 in a Chicago neighborhood in 1956, Kain's best estimate was that: W = 9.28 + .456R-.409dm (1) Since this model assumes that the effects of R and Dm are linear, it implies that the fraction of all Chicago-area jobs held by blacks (W) depends entirely on the areawide means of R and dm. The dispersion of R (an indicator of the level of segregation) is irrelevant. Segregation will, of course, increase demand for black workers in black neighborhoods and lower demand in white neighborhoods, but the two changes will exactly offset one another, leaving aggregate demand unchanged. Since residential desegregation does not alter the total number of blacks or whites, at least in the short run, it does not change the overall mean of R. Using Kain's assumptions, therefore, residential segregation only affects blacks' share of total employment indirectly, that is, by affecting their average proximity to jobs (d ). While Kain implicitly assumed that desegregation would leave overall demand for black workers unchanged, he did not try to test this assumption. In order to do so we need to ask whether, when all else is equal, demand for black workers in neighborhoods whose racial mix mirrors that of the metropolitan area as a whole is higher or lower than demand in the metropolitan area as a whole. So far as we can discover, only one study has investigated this issue. Offner and Saks (1971) reanalyzed Kain's Chicago data and found that blacks fared quite badly in racially mixed areas. Employers in all-black neighborhoods hired a lot of blacks. Employers in all-white areas hired very few blacks. Employers in racially mixed areas acted like employers in all- white areas, hiring very few blacks. As a result, blacks' share of employment was lower in racially mixed neighborhoods than in the Chicago area as a whole. Offner and Saks's statistical results cannot tell us why employers in mixed neighborhoods hired so few blacks, but several possible explanations come to mind. If white Chicago residents were reluctant to trade with firms that hired blacks in the 1950s, while Chicago blacks were willing to trade with firms that hired only whites, blacks would not have gotten many retail jobs in racially mixed areas. And if white workers were willing to work in racially mixed areas for the same wages as in all-white areas in the 1950s,
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 193 firms in racially mixed areas would have had no more incentive to hire blacks than firms in all-white areas. These conjectures all assume that a firm's employment practices de- pend on the racial mix of its neighborhood. In the racially mixed Chicago neighborhoods with which we are familiar, however, residential mix is a by-product of local employment patterns, as well as the other way round. The areas around the University of Chicago and Michael Reese Hospital, for example, were located in the path of black residential growth. Ordinar- ily, these areas would have become entirely black. They remained racially mixed because firms employing white professionals remained in the area, and many of these white workers wanted to live near their workplace. Whatever the explanation, Offner and Saks's results show that if resi- dential desegregation had led all Chicago firms to act like those in racially mixed neighborhoods, job opportunities for blacks would have contracted, not expanded. Luring blacks and whites to the same neighborhoods would have reduced blacks' chances of getting jobs in what had once been ghetto firms without opening up an equivalent number of new jobs in firms outside the old ghetto. In the long run, of course, residential desegregation might have re- duced white prejudice against blacks, increasing demand for black workers. One way to test the validity of this hypothesis is to ask whether blacks do better economically In relatively desegregated cities. The evidence on this point, while hardly conclusive, suggests that residential segregation has very little effect on blacks' economic success. Masters (1974) studied the ratio of nonwhite to white earnings in 65 large Standard Metropolitan Statistical Areas (SMSAs) in 1959. Contrary to what we would expect if residential segregation reduced demand for nonwhite workers, the ratio of nonwhite to white incomes was slightly higher in highly segregated SMSAs. This relationship was not reliably different from zero, but the fact remains that Masters's findings offer no support for the contention that desegregation increases demand for black workers.5 Masters (1975) replicated this result using 1969 data from 77 SMSAs with large black populations. Unfortunately, Masters's 1969 results changed when he excluded SMSAs with substantial amounts of agricultural employment. Once he dropped these SMSAs, there was some evidence that both black men's chances of having a job and their relative earnings were 5Kain (1974) argues that Masters's results are inconclusive, since they do not take account of city-to-city differences in the degree to which jobs are suburbanized, the relative tightness of ur- ban and suburban labor markets, or the distance from the black ghetto to centers of employment. But Kain offers neither theoretical arguments nor empirical evidence that including these omit- ted variables would make the coefficients of Masters's segregation measures negative instead of positive.
194 INNER-C1IY POVERTY IN THE UNITED STATES more favorable in less segregated SMSAs.6 Thus we cannot say with any confidence whether residential segregation raised or lowered demand for black workers in the 1960s. Nor can we rule out the possibility that the level of residential segregation depended on the relative earnings of blacks and whites rather than the other way round. A]1 we can say with confidence is that the relationship between residential segregation and relative earnings is weak. In any event, the effects of residential segregation on a firm's demand for black workers are likely to change over time. Kain's data described Chicago in the 1950s, when employers were under no legal or social pressure to hire blacks and when hiring blacks for nonmenial jobs often caused firms a lot of trouble. Masters's data describe the world of 1959 and 1969, before affirmative action programs had much impact. Much has changed since then, both outside and inside the ghetto. Outside the ghetto, white customers, appear to have become more confortable seeing blacks in nontraditional roles. Political and social pres- sures have led many managers, especially in large firms, to open up new kinds of jobs to blacks and to make more aggressive recruiting efforts. All these efforts have, however, focused largely on blacks who act more or less like whites. At the same time, some black workers have become more assertive in their dealings with whites, increasing employers' fears that blacks will cause "trouble." The net effect of these trends on demand for black workers Is not obv~ous.7 6 Masters presents bivariate results for 77 SMSAs and multivariate results for 65 of these SMSAs. The 12 excluded SMSAs had high levels of agricultural employment in their more remote sub- urbs and low black-white income ratios. Almost all coefficients change sign in his multivanate model, so trimming the sample makes a big difference. Unfortunately, he presents no bivanate results for the trimmed sample, and his multivariate results include four separate measures of residential segregation, along with two control variables designed to measure job dispersion and accessibility. With these controls, the Taeubem' measure of residential segregation raised blacks' income relative to whites (t = 2.4) but lowered blacks' relative likelihood of having a job (t = .6~. The multivariate model is hard to interpret, however, because Masters's measure of central- city blacks' access to suburban jobs (A) is (JgClJgt)/(JNgc/JNgt)' where JB is the number of jobs held lay blacks in a given area, JNB is the number of jobs held lay nonblacks, the subscript c denotes jobs in the central city, and the subscript t denotes all jobs in the SMSA. Unfortunately, this ratio measures not only the relative accessibility of suburban jobs to blacks and whites but the relative willingness of suburban employers to hire blacks rather than whites ("demand"~. Kain's argument implies that A is at least partly endogenous and should not be controlled when estimat- ing the effects of residential segregation. The use of four different segregation measures, which may well be quite collinear, also makes Mastem's results hard to interpret. Bivariate statistics for the trimmed sample would have made his findings more instinctive. 7Kirschenman and Neckerman (forthcoming) describe a survey of Chicago employers that shows very high levels of prejudice against unskilled black males, but they do not relate these attitudes to the actual numbers of blacks hired in a given firm.
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 195 Predicting changes in demand for black workers inside the ghetto is as hard as predicting changes outside the ghetto. Black ghetto residents may have grown less willing to patronize firms that hire only whites, but we know of no evidence on this point. The wage premium that whites demand for working in the ghetto has probably risen since 1969, because crime has increased even more in the ghetto than elsewhere,8 but again, we know of no hard evidence on this point. If whites do, in fact, demand higher premiums for working in the ghetto, this is likely to have had two contradictory effects on job opportu- nities for blacks. On the one hand, we would expect firms that remained In the ghetto to have substituted cheap black workers for more expensive white workers. On the other hand, we would expect firms that wanted to retain significant numbers of whites to have left the ghetto. Ken together, these trends would lead to an increase In the fraction of ghetto jobs going to blacks, combined with a decrease in the fraction of all jobs located in ghetto areas.9 Again, the net effect on the number of jobs available to blacks in the ghetto is not obvious. The only data we have found that speak to this issue come from Leonard's (1987) study of Chicago and Los Angeles firms' payrolls in 1974 and 1980.~° Leonard's data do not, of course, measure changes In firms' demand for black workers. Rather, they measure changes In the actual level of black employment inside and outside the ghetto. Such changes depend on the supply of black and white workers available to different sorts of firms as well as on firms' demand for such workers. Leonard's data suggest that blacks' share of blue-collar jobs In the ghetto Increased slightly in both Chicago and Los Angeles between 1974 and 1980, which Is what we would expect if whites were less willing to work In the ghetto. Blacks' share of blue-collar jobs outside the ghetto rose In Q Thepercentage increase in crime since the mid-1960s has been greater among whites than among blacks, but the absolute increase has been greater among blacks than among whites. From the viewpoint of the potential victim, it is the absolute increase that matters, so the wage premium required to lure people to all-black areas is likely to have risen. 9 Most ghettos occupy a larger fraction of the metropolitan area in which they are located today than they did in 1960. As a result, the fraction of all jobs located in the ghetto has often grown, even when the number of jobs per square block has fallen. For analytic purposes, however, the crucial issue is not the fraction of all jobs located in the ghetto but the ratio of this fraction to the fraction of all persons living in the ghetto. The argument in the text reflects conventional wisdom in assuming that this ratio has fallen, but we have not seen any hard data supporting that view. i°Leonard's sample covem 1,911 Chicago firms and 2,389 Los Angeles firms surveyed by the Of- fice of Federal Contract Compliance in both 1974 and 1980. Since the sample excludes firms that came into existence or went out of existence between 1974 and 1980, it is not fully representative of all firms surveyed in either year. Nor are firms surveyed by the OFCC fully representative of all firms.
196 INNER-CI-IY POVERTY IN THE UNllED STATES Los Angeles but fell in Chicago. As a result, the jobs held by black male Chicagoans were more likely to be in the ghetto in 1980 than in 1974, even though ghetto firms accounted for a smaller fraction of all Chicago-area jobs. In Los Angeles, in contrast, the jobs held by black men were less likely to be in the ghetto in 1980 than in 1974, even though ghetto firms were growing faster than those in mixed areas. The difference between Chicago and Los Angeles may well reflect differences in the two cities' racial climates, which in turn affect firms' location decisions. The main conclusion we draw from Leonard's findings is that generalizing from a single city to the nation as a whole would be very dangerous. Taking all the evidence together, the effect of residential segregation on firms' overall demand for black workers remains almost as uncertain as it was 20 years ago. If we rely on Masters's findings for 1959 and 1969, the safest conclusion would be that at that time residential segregation had no consistent effect on firms' interest in hiring blacks. It is important to emphasize, however, that Masters' findings reflect the combined effects of supply and demand, and that they describe employment patterns 20 to 30 years ago, not today. It is also important to emphasize that we know nothing whatever about how residential segregation along economic lines affects demand for black workers. EFFECTS OF JOB PROXIMITY ON LABOR SUPPLY Kain showed that as a neighborhood's distance from the central-city ghetto increased the proportion of neighborhood jobs held by nonwhites declined. This was true even with the neighborhood's own racial mix controlled (see equation 1, above), and it was true in Detroit as well as in Chicago. Leonard (1987) found the same pattern for Chicago and Los Angeles in 1974 and 1980. Kain argued that distance from the central-city ghetto affected the racial mix of a neighborhood's labor force by affecting the supply of black workers available to firms in different neighborhoods. Distance affected labor supply in two ways. First, it increased the effort prospective workers had to make in order to learn about jobs before they were filled. Sec- ond, distance raised workers' commuting costs, making remote jobs less attractive. The argument that information and commuting costs encourage pro- spective workers to choose jobs near where they live is not controversial. The controversial part of Kain's supply-side argument was his suggestion that workers who could not find jobs near home were likely to stop looking and withdraw from the labor force. If this were true, the high rate of joblessness among urban blacks could be reduced either by discouraging the
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 197 movement of jobs from the central city to remote suburbs or by encouraging black residential movement from the central city to the suburbs. 1b test this part of Kain's argument, we need to investigate two questions: 1. Do blacks fare better economically when they live in neighbor- hoods that are near a lot of blue-collar jobs? 2. Do blacks fare better when they live in metropolitan areas where blue-collar jobs are still mainly located in the central city rather than in the suburbs? We take up these questions in turn. Neighborhood Comparisons One way to estimate the effect of job proximity on black employment is to measure how far different neighborhoods are from blue-collar jobs and then ask whether blacks who live in neighborhoods with a lot of nearby blue-collar jobs have unusually high rates of employment. In the 1950s, for example, Chicago blacks were concentrated in two areas: the West Side, which was a major center of manufacturing and warehousing, and the South Side, which had relatively few jobs of this kind. One way of testing Kain's arguments about the importance of job proximity would have been to ask whether blacks on the West Side were more likely to have jobs than blacks on the South Side. This strategy runs an obvious risk, since employment influences residential choices as well as the other way round. If Chicago's West Side firms had hired blacks in large numbers, many blacks who worked for these firms would presumably have moved to the West Side, even if they had lived on the South Side when they got their jobs. If West Side firms had paid blacks better or offered them steadier employment than other Chicago firms, as Kain's argument implied they should, West Side blacks would have had higher rates of employment and higher wages than other Chicago blacks. This would not have proved that living on the West Side helped blacks get good jobs. It would only have proved that blacks, like everyone else, prefer a short trip to wore If we set aside this methodological hazard for the moment, it is still important to ask whether a neighborhood's proximity to various kinds of jobs is positively correlated with the fraction of adult residents who hold jobs or negatively correlated with its unemployment rate. Two studies have correlated job proximity with employment among adult males. Hutchinson (1974) used a Pittsburgh traffic survey to construct an index of the number of jobs located at various distances from each respondent's home. He measured the distance between homes and jobs by estimating the time required to reach the job by public transportation. His index of job proximity had a modest positive effect on the probability that white males
198 INNER-CITY POVERTY IN THE UNITED STATES would be employed in 1967, but it had a negative (though statistically unreliable) effect on the probability that black males would be employed. Leonard (1986) also found that the ratio of nearby jobs to neighborhood population had a negative (but statistically unreliable) effect on Los Angeles men's chances of being employed in 1980. The negligible correlation between job proximity and male employ- ment rates in Pittsburgh and Los Angeles is puzzling, since we would expect workers with steady jobs to settle near their workplace. The weak correlation suggests that neighborhoods located near major centers of em- ployment may have undesirable social or aesthetic attributes that discourage workers in nearby firms from living there. The weak correlation may also reflect the fact that most commuters use automobiles in both Los Angeles and Pittsburgh, making distance less important than it is in some other places. We would expect job proximity to have a stronger effect on female employment than on male employment, since women earn less and are more likely than men to work part-time, making long journeys to work less worthwhile. Leonard (1986) did find that the ratio of nearby jobs to neighborhood residents had a positive effect on women's employment rates in Los Angeles in 1980, but the effect was not reliably greater than zero. No one else has investigated this issue. Job proximity does seem to exert a small but statistically reliable influence on teenage employment rates. Ellwood (1986) used both the ratio of neighborhood jobs to neighborhood residents and the mean commuting time of adult blue-collar workers living in a neighborhood to estimate the proximity of different Chicago neighborhoods to jobs. Both measures showed that males between the ages of 16 and 21 who were no longer in school were more likely to be working in 1970 if they lived in neighborhoods with a lot of nearby jobs. Leonard (1986) also found that the ratio of nearby jobs to population had a significant positive effect on Los Angeles teenagers' chances of being employed in 1980. Ihlanfeldt and Sjoquist (1990) found the same pattern in Philadelphia, Chicago, and Los Angeles in 1980. The finding that job prommi~ is more important for teenagers than for adults is consistent with economic theory. Many teenagers work part-time and few earn much money. As a result, teenagers are less likely than ~ ~Hutchinson actually had two separate measures of distance. One (ACT measured travel time by public transportation. The other (ACS) measured travel time lay automobile and had a value of zero for individuals who did not own automobiles. Ibe coefficient of AC? is small but significant for whites. It is even smaller, insignificant, and wrong-signed for blades. Since Hutchinson's equations do not indude a dummy variable for having an automobile, ACS is largely a proxy for automobile ownership. Its coefficient is highly significant. This finding tells us only that having a job increases the likelihood of owning an automobile. Whether automotive commuting time affects an automobile owner's chances of having a job remains unclear.
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 199 adults to own cars. We would expect job proximity to be more important to workers who walk or take public transportation to work than to those who can drive. While all the available evidence suggests that living in a neighborhood with a lot of nearby jobs increases a teenager's chances of working, there is less agreement about the size of these effects. Both Ellwood and Leonard found very small effects in their studies of Chicago and Los Angeles. Using somewhat better measures of job proximity, Ihlanfeldt and Sjoquist found relatively large effects in Philadelphia, Chicago, and Los Angeles. They divided the Philadelphia SMSA into 26 areas, calculated the average travel time of low-wage workers in each area, and used this average as a measure of low-wage jobs' accessibility from that area. A 3.7 minute (1 standard deviation) increase in mean travel time reduced the percentage of 16- to 24-year-old blacks with jobs by 4 to 6 points, depending on their exact age and whether they were in school. The results for young blacks in Chicago and Los Angeles, while less precise, were roughly similar to those in Philadelphia. The effect of job proximity on employment rates among whites was roughly the same as for blacks in Philadelphia, but job proximity did not vary much for whites in Los Angeles, and it did not have much effect in Chicago. Ihlanfeldt and Sjoquist also found that, at least in Philadelphia, blacks lived much further from low-wage jobs than whites did. Young Philadelphia blacks lived in areas where low-wage workers spent an average of 26 minutes getting to work, whereas young whites lived in areas where low- wage workers spent only 18.5 minutes getting to work. Controlling for job proximity reduced the racial gap in employment by roughly a third, both among 16- to 19-year-olds who had left school and among 2~ to 24-year- olds who were still living with their parents. (For young people no longer living with their parents, we cannot tell whether residential choices are a cause or an effect of job location.) Ihlanfeldt and Sjoquist do not tell us whether Chicago or Los Angeles blacks lived in areas with unusually long travel times for low-wage workers, so we cannot tell how much of the racial disparity in youth employment for these cities was due to the fact that young blacks lived in the wrong places. Based on Ihlanfeldt and Sjoquist's work however, we can say that location is a major factor in the high rate of joblessness among young Philadelphia blacks. City-to-City Comparisons A second way of estimating the effect of distance on employment op- portunities is to compare cities with highly dispersed employment patterns to cities with more concentrated patterns. Mooney (1969) used aggregate data on the 25 metropolitan areas with the largest nonwhite populations
where, 200 INNER-CITY POVERTY IN THE UNITED STATES to estimate the impact of firms' moving to the suburbs on nonwhite male employment in 1960. His best estimate was that R = .63 - 2.86U + .19D + .24A (.05) (.56) (.06) (.06) (2) R is the probability that a nonwhite male living in the poorest central-city census tracts held a job in March 1960. U is the unemployment rate for the SMSA as a whole in 1960. D is the fraction of all jobs in manufacturing, wholesale trade, retail trade, and selected services located in the central city. (This is supposed to be a measure of job dispersal.) A is the probability that an employed nonwhite living in the central city will work in the suburbs. (This is supposed to be a measure of the suburban ring's accessibility to central-city nonwhites.) The numbers in parentheses are the standard errors of the coefficients. The overall tightness of the local labor market was clearly the most important determinant of black men's employment rates. A 1-point increase in an SMSA's overall unemployment rate led to a 2.86-point reduction in the employment rate among black men living in central-city poverty areas. But black men also did worse in cities where manufacturing, trade, and services were more suburbanized. A 10-point increase in the proportion of such jobs located in the suburbs led to a 1.9 point decline in employment among black ghetto residents. Parley (1987) conducted an analogous study using 1980 census data on all metropolitan areas of 50,000 or more. After controlling an SMSA's overall racial mix and a measure of black-white educational inequality in the SMSA, the ratio of black to white unemployment was lower in SMSAs where manufacturing, trade, and service jobs were mainly located in the central city. In explaining Mooney and Farley's findings it is important to bear in mind that they did not find that the absolute level of black employment was higher in cities where manufacturing, trade, and services were concentrated in the central city. They only found that the black-white gap was smaller in these cities.~3 The black-white difference could be larger in cities with 12Farley also found that the ratio of black to white unemployment was lower when a high pro- portion of blacks lived in the suburbs rather than in the central city. Whether this indicates that unemployment falls when blacks move to the suburbs or that blacks move to the suburbs when they have steady jobs is unclear. 13Mooney uses the absolute black employment rate as his dependent variable, but since he con- trols the overall unemployment rate, the estimated erect of suburbanized employment is con- ceptually (though not statistically equivalent to its effect on the black-white disparity.
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNIT ES 201 decentralized employment patterns either because whites do better in such cities or because blacks do worse. Ihlanfeldt and Sjoquist (1989) avoid this ambiguity by looking at the effect of job decentralization on the earnings of both blacks and whites who either worked or said they wanted to work in 1978. Their sample came from the Panel Study of Income Dynamics and included only men and women who had not attended college and lived in the central city of an SMS~ Their measure of earnings was total earnings in 1978, less commuting costs. Their primary measure of job decentralization was the percentage of all low-skill jobs in the SMSA located outside the central city of the SMSA in 1980. All their estimates controlled respondents' age, education, job tenure, marital status, and physical health, as well as the SMSA:s overall unemployment rate, racial mix, and rate of population growth between 1970 and 1980. Among black central-city males without higher education, a 16 point (one standard deviation) decline in the percentage of low-skill jobs located in the central city was associated with a $929 (10 percent) decline in 1978 earnings. Among white central-city men without higher education, the decline was $849 (6 percent). Job decentralization had a small (and statistically unreliable) negative effect on white women's earnings, no effect on black women's earnings, and no effect on college-educated workers' earnings. Ihlanfeldt and Sjoquist did not investigate whether job decentralization lowered central-city men's net earnings by lowering their wages, lowering their hours, or raising their commuting costs. This is a serious limitation. When firms move to the suburbs, housing values in the central city often fall. This may lead unskilled workers who find suburban jobs to live in the central city, since they may save enough on housing to offset their extra commuting costs. Because they include commuting costs while excluding housing costs, Ihlanfeldt and Sjoquist probably overestimate the adverse effect of job decentralization on central-city residents' economic welfare. Furthermore, we cannot draw strong conclusions about the impact of job decentralization on either black or white workers' economic welfare unless we know how it affects the welfare of those who live in the suburbs as well as those who live in the central city. Ihlanfeldt and Sjoquist did not investigate this issue. Reconciling Neighborhood Studies With SMSA Studies Comparisons of metropolitan areas imply that job proximity affects black men's employment prospects, while Hutchinson and Leonard's com- parisons of different neighborhoods in the same city suggest that job prox
202 INNER-CITY POVERTY IN THE UNITED STATES imity makes little difference for adults. We can think of at least three explanations for these apparently contradictory findings. First, Mooney and Farley's city-to-city comparisons may be picking up the effect of job proximity on teenagers. This explanation strikes us as unlikely, however, since teenage employment is not a large fraction of total employment. Second, Hutchinson and Leonard's findings may be peculiar to Pitts- burgh and Los Angeles. Job proximity may be more important in cities where public transportation plays a large role in commuting. Third, the correlation between geographically centralized employment and the black white employment ratio in Mooney and Farley's city-to-city comparisons may be attributable to the fact that both firms' location deci- sions and blacks' job prospects depend on other unmeasured characteristics of the SMSA, such as the level of interracial conflict and prejudice. Firms in SMSAs with a lot of racial conflict may be especially likely to move to the suburbs, and firms in these SMSAs may also be especially reluctant to hire blacks. If this explanation is correct, we should dismiss Mooney and Farley's findings as spurious and put our faith in Hutchinson and Leonard's findings. We would need some direct evidence before accepting this explanation, however. CAN BLACKS EARN MORE IN THE SUBURBS? The spatial mismatch hypothesis assumes that blacks are less likely than whites to have steady, well-paid jobs because they live farther from the kinds of firms that might offer them such jobs. This hypothesis depends on two empirical assumptions: that distance affects a worker's chances of finding a job and that there are, in fact, better jobs available to blacks in distant areas (Kasarda, 1989~. One way to test the latter assumption is to ask whether blacks in the central city who commute to the suburbs earn more than those who work in the ghetto (or in the central business district). Danziger and Weinstein (1976) studied Cleveland, Detroit, and St Louis men between the ages of 21 and 64 who lived in central~ibr neighborhoods with high poverty rates in 1970. (Here and throughout we use the term "ghetto" to describe these poor, black central-city neighborhoods.) They found that ghetto blacks in all three cities were about as likely to work in the suburbs as in the ghetto.~4 This finding does not suggest that information 14Since Danziger and Weinstein do not include individuals who worked in the central city but outside the poverty areas where they lived, their findings do not imply that half of all ghetto blacks worked in the suburbs. Hughes and Madden (forthcoming) report that among black male household heads who worked full-time, year-round, in 1980 in the Cleveland SMSA, 49 percent worked in the suburbs. The rates were 42 percent in Detroit and 31 percent in Philadelphia.
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 203 costs posed a very serious problem for blacks seeking jobs in the suburbs, although the percentage of blacks working in the suburbs would presumably have been even higher if information costs had been lower. Blacks who worked in the suburbs earned 11 percent more than those who worked in the ghetto. The main reason blacks earned more in the suburbs was that they worked in better-paid occupations and industries than ghetto workers. Danziger and Weinstein treated a worker's occupation and industry as fixed personal characteristics, so they concluded that commuting to the suburbs was of little economic value to blacks. But a worker's occupation and industry are not, in fact, fixed. About 9 percent of all workers changed occupations every year during the 1970s (Byrne, 1975), and industry changes are likely to be at least as frequent. Furthermore, changing occupation or industry is one common way of getting a better job. It therefore seems reasonable to suppose that blacks commute to the suburbs partly because commuting allows them to get jobs in well- paid occupations and industries. Controlling occupation and industry when estimating the payoff to commuting is therefore likely to underestimate the benefits of commuting. Danziger and Weinstein did not find much educational difference between blacks who worked in the suburbs and blacks who worked in the ghetto. Their findings are therefore consistent with the hypothesis that blacks who worked in the ghetto could have earned almost 11 percent more than they did if they had had jobs in the suburbs. Most researchers agree that distance plays a minor role in determining white-collar workers' job opportunities. The spatial mismatch hypothe- sis should therefore apply primarily to blacks without higher education. Danziger and Weinstein did not investigate whether blacks without higher education gained more than college-educated blacks by commuting to the suburbs, but Straszheim (1980) did. Using data collected in the San Fran- cisco Bay area in 1965, Straszheim found that blacks who had not attended college earned about 10 percent more if they worked in the San Francisco or Oakland suburbs than if they worked in San Francisco or Oakland itself. This pattern was reversed among blacks who had attended college. The Danziger-Weinstein and Straszheim results for 1965-1970 are, in our judgment, consistent with the spatial mismatch hypothesis. But they are also consistent with a more traditional view, namely that suburban jobs pay more than ghetto jobs because they impose higher commuting costs on most workers. These percentages would be lower for blacks who lived in central cities and still lower for blacks living in central-city poverty areas.
204 INNER-CIl.Y POVERTY IN THE UNITED STATES According to Danziger and Weinstein, commuting to a suburb length- ened the average ghetto resident's working day by about 26 minutes. As- suming an 8-hour day, and further assuming that workers are indifferent between spending their time commuting and spending it working, a 26- minute increase in commuting time is equivalent to a 5 percent reduction in hourly wages. Danziger and Weinstein estimated that the extra out- of-pocket expenses associated with suburban jobs reduced black suburban workers' effective wages by an additional 2 to 4 percent. Thus while black suburban workers earned 11 percent more than ghetto workers, their net wages were only 2 to 4 percent higher than those of ghetto workers. Since blacks who worked in the suburbs were a trifle better educated than those who worked in the ghetto, we would have expected the suburban workers to earn a trifle more. Danziger and Weinstein's findings therefore suggest to us that in 1970 the suburban wage premium was just sufficient to cover ghetto residents' commuting costs. This finding is reassuring. If ghetto residents could have realized substantially higher net earnings by commuting to the suburbs, we would have had to ask why so many of them chose to work in the ghetto. At least in 1970, however, the answer is clear: Blacks could not appreciably increase their net earnings by commuting to the suburbs. Ihlanfeldt (1988) used census data to examine wage differentials be- tween various parts of the Atlanta SMSA in 1980. He found significant location differentials for whites but not for blacks. Black service workers made more in one county north of Atlanta (Clayton), but so did white service workers. This seems to resect a general shortage of service workers in Clayton County. Hughes and Madden (forthcoming) also use 1980 census data to inves- tigate wages in different parts of the Cleveland, Detroit, and Philadelphia SMSAs. Unfortunately for our purposes they do not report average wages in central cities and in the suburbs. Instead, they divide SMSAs into 13 to 27 work zones. These work zones include each SMSA:s central business district, the rest of the central city, and 11 to 25 suburban areas. Hughes and Madden identify the work zone in each SMSA where male household heads who worked full-time, year-round earned the most, after adjusting for their age, education, occupation, industry, and commuting costs. They then estimate the gain in net earnings, after adjusting for changes in commuting costs, if all workers were paid at the same rate as those in the best-paid area. These calculations imply that if all black men had worked in the best- paid part of their SMSA in 1980 they would have earned 38 to 48 percent more than they actually did. Hughes and Madden are rightly skeptical about these estimates. If blacks could have increased their net earnings by 38 to 48 percent by, say, working in a remote suburb rather than the central
RESIDENTL4L SEGREGATION AND BLACK JOB OPPORlUN17IES 205 business district, many would surely have done so. Since they didn't, we must aslL why. The most obvious answer is that blacks who worked in poorly paid areas could not get the jobs that were available in the best-paid areas. Hughes and Madden try to deal with this possibility by controlling workers' age, education, occupation, and industry.~5 But because they are estimating wage differentials for small areas using small samples, their results are likely to be quite sensitive to the presence of a few high-wage firms in one of these areas. Firms that pay substantially more than the areawide average are usually quite choosy about whom they hire. Such firms also tend to set high performance standards and fire workers who do not meet them. As a result, there are likely to be large unmeasured differences between the workers in these firms and apparently similar workers in firms with lower pay scales. Our suspicion that intrametropolitan wage gradients reflect unmea- sured differences in worker characteristics is supported by the fact that Hughes and Madden's method also implies that whites could have in- creased their net earnings by 30 to 36 percent if they had worked in different parts of these SMSAs in 1980. Since there were no obvious constraints on where whites worked, this finding suggests that Hughes and Madden's wage equations probably omit important worker characteristics. All else equal, Hughes and Madden found that blacks earned more when they worked near the periphery of the Cleveland and Philadelphia SMSAs rather than near the center, but the difference was substantively trivial and statistically unreliable. Blacks earned significantly less when they worked near the periphery of the Detroit SMSA than when they worked near the center. These results make it hard to argue that central-ci~ blacks work in low-wage areas because they are poorly informed about high-wage jobs in more remote areas. Hughes and Madden's findings are also at odds with those of Danziger and Weinstein, which implied that blacks earned more near the periphery of the Cleveland, Detroit, and St. Louis SMSAs than near the center. The most plausible explanation is that Hughes and Madden studied all blacks 15For reasons already discussed, controlling a worker's occupation and industry could yield downwardly biased estimates of the benefits of changing job location. In practice, however, Hughes and Madden's estimates are almost certainly biased upward. 16Since the estimated gap between observed and optimal wages is larger for blacks than for whites, Hughes and Madden assume that blacks could increase their earnings more than whites by working in different places. The gap between average and optimal wages is, however, also likely to be a function of sample size, since there is more random variation in area-specific wages using small samples than using large ones. The larger gap among blacks could, therefore, be due to the fact that there are fewer of them-or to the fact that unmeasured differences between workers in different areas are larger among blacks than among whites.
206 INNER-CIIYPOVER:TYIN THE UNll~;D STATES in these SMSAs, whereas Danziger and Weinstein studied only ghetto residents. Ghetto blacks will only take suburban jobs if these jobs pay more than they think they can get in the ghetto, since they will want compensation for their extra commuting costs. Conversely, suburban blacks will only take central-city jobs if those jobs pay more than the jobs they think they could get near home, since they too will want to cover their extra commuting costs. Thus even if suburban and ghetto jobs offer the same average wage for a worker with specified characteristics, workers will only commute long distances if they are offered a job that pays more than this average for some reason. In such a world, samples of commuters will always show that commuting "pays." But it will not follow that average wares are higher in one place than the other.~7 Taken together, the available data offer little support for the notion that blacks could get better jobs if they worked in the suburbs rather than in central cities. Neither Danziger and Weinstein's work nor Hughes and Madden's suggests that either improvements in public transportation between central cities and the suburbs or other devices for making suburban jobs more accessible to central-city blacks would have much effect on black workers' potential earnings. The modest role of commuting costs in both studies also suggests that if suburban firms wanted to hire more blacks, they could attract them relatively easily. _~_ --it- ~D- HOW DO BLACKS FARE WHEN THEY LIVE IN THE SUBURBS? At bottom, the spatial mismatch hypothesis is rooted in the observation that exclusionary housing practices have kept blacks from moving to the suburbs, where their job prospects might be better than they are in the central city. If blacks who live in the suburbs do no better than those in the central city, we have to conclude, with David Ellwood (1986), that the problem is "race, not space." Much popular thinldng about these issues is rooted in the easy assump- tion that there are no blacks in the suburbs of major American cities. Yet some blacks have always lived in the suburbs, and their numbers have grown steadily over the past generation. In metropolitan areas of 1 million or more inhabitants, which are the central focus of spatial mismatch theories and which we will hereafter call "large SMSAs," 19 percent of working-age blacks lived outside the central city in 1967. By 1986 the proportion had risen to 27 percent (Bureau of the Census, 1969: 1bble 1; 1988: Able 31~. ~ 7An alternative explanation of the difference between the Hughes-Madden and Danziger-Wein- stein studies is that the suburban wage advantage disappeared in Cleveland and turned negative in Detroit between 1970 and 1980. This explanation strikes us as highly unlikely.
RESIDENTLi4L SEGREGATION AND BLACK JOB OPPORTUNITIES 207 These suburban blacks mostly live in places like Yonkers, Waukegan, and East Palo Alto, not in places like Scarsdale, Winnetka, and Mill Valley. Furthermore, most suburban blacks live next to other blacks, not whites (Massey and Denton, 1988~. When we estimate the economic impact of living in the suburbs, therefore, we are mainly estimating the impact of location, not the impact of racial or economic desegregation. If residential location exerts a significant influence on black workers' job opportunities, and if the best jobs open to blacks are now located in the suburbs, blacks who live in the suburbs should experience less unemployment and should earn more money than blacks who live in central cities. Simply contrasting the unemployment rate or earnings of workers who live in suburbs to analogous statistics for workers who live in central cities cannot, of course, yield definitive estimates of how much residential location influences economic success, because people's earnings affect where they live as well as the other way round. In 1983, for example, the median value of owner-occupied homes was 25 percent higher in suburbs than in central cities, and median monthly rent averaged 19 percent higher in suburbs than in central cities. For blacks, the price differences were even larger: 28 percent for owner-occupied homes and 30 percent for renege Price differences between central-city and suburban housing ensure that migration between central cities and suburbs would be somewhat selective even if residential choices had no effect whatever on job oppor- tunities. Family heads with steady, well-paid jobs can afford to live in the suburbs. Family heads who do not have (or cannot keep) such jobs are likely to have trouble finding housing they can afford in the suburbs. Living in the suburbs may also have other important drawbacks for household heads who do not work regularly. The suburbs are farther from friends and relatives who can help out when you are out of money, farther from the welfare and unemployment offices, and provide fewer opportunities for making money illegally. As a result, those who do not work regularly are likely to gravitate to the central city. Nor are migration patterns between central cities and suburbs likely to depend exclusively on employment status or income. One major reason why parents move to the suburbs is that they think suburban schools are better than inner~ibr schools. Highly educated parents tend to be even more concerned about the quality of their children's schooling than poorly educated parents with similar incomes. As a result, we would expect that even among families with the same incomes, highly educated parents would 18These estimates are from the Annual Housing Survey (Bureau of the Census, 1983), renamed the American Housing Survey thereafter. The 1985 survey used the 1980 census as its sampling frame, es well as a new questionnaire. The pre-1983 and post-1983 surveys evidently do not yield comparable estimates of many housing characteristics, and the Census Bureau had not published data from the 1985 or 1987 surveys when we were writing this paper.
208 INNER-CITY POVERTY IN THE UNITED STATES be more likely to end up in the suburbs while the less educated would be more likely to end up in the central city. Estimating the determinants of selective migration requires longitudi- nal data. Such data exist, but so far as we have been able to discover, no one has used them to analyze black migration between central cities and suburbs.l9 This means that even if we find a relationship between living in the suburbs and economic success we cannot tell how much of the relationship is due to selective migration and how much, if any, is due to the fact that suburbanites have better access to certain kinds of jobs. But if we find no difference between those who live in the suburbs and those who live in the central city, it is hard to argue that living in a suburb improves job opportunities. In 1968, when Kain first proposed the spatial mismatch hypothesis, suburban blacks seem to have been no better off than their central-city counterparts. Harrison (1972, 1974) used data from the Census Bureau's SuIvey of Economic Opportunity (SEO) to compare men who lived in the central cities and the suburbs of the nation's 12 largest metropolitan areas in 1966. When he compared whites with the same number of years of schooling, he found that central-city men were better off than suburban men, which is what we would expect if there were a lot of selective migration. When he compared nonwhites with the same amount of schooling, however, he found that suburban nonwhites were as likely to be unemployed and earned almost exactly the same amount as central-city nonwhites. Bell (1974) conducted a similar study of black wives using the SEQ. When he controlled a wife's personal characteristics, the size of her family, and her family's income from sources other than her earnings, he found that black wives were slightly more likely to have a job if they lived in central cities. Central-city wives who worked also earned slightly more than their suburban counterparts.20 These findings suggest that living in a suburb was of little economic value to blacks in the mid-1960s and that migration between central cities 19The Panel Study of Income Dynamics (PSID) conducted by the Survey Research Center at the University of Michigan and the National Longitudinal Surveys conducted by the Department of Labor provide annual data on both the residential location and employment status of individuals. 20 Both Harrison and Bell divided their central-city samples into two groups based on the av- erage income in the neighborhood. As one would expect, central-city men who lived in poor neighborhoods earned less than those who lived in better neighborhoods. In order to focus on the erects of location, we averaged the estimates for central-city poverty and nonpoverty areas and contrasted the resulting means with the suburban means for individuals at the same educa- tional level. Our urban-suburban comparisons for males are only approximate, because Harrison does not report means for either group. He does report regression coefficients, but he does not report either intercepts or insignificant coefficients. As a result, we had to estimate the mean urban-suburban difference from his graphs.
RESIDENTL4L SEGREGATION AND BLACK JOB OPPORTUNITIES 209 and suburbs was not closely tied to income. The situation may have changed since then, however. Blue-collar employment has continued to leave cen- tral cities (Kasarda, 1989~. Many manufacturing jobs are now located in relatively remote suburbs accessible only by automobile. Central-city blacks may therefore have more trouble getting reliable information about sub- urban job openings today than they did in 1966. Commuting to suburban jobs may also consume a larger fraction of a central-city resident's earnings, making such jobs less economically attractive and making migration to the suburbs more attractive. Every year since 1968 the Census Bureau's Current Population Survey (CPS) has reported the proportion of adults in the central cities and the suburbs who worked full-time throughout the previous year. For our pur- poses the estimates for large SMSAs are especially relevant, since physical access to remote suburbs is more of a problem in these SMSAs. Table 5-1 shows that in 1967-1971 black men who lived in the central cities of large SMSAs were about as likely to work regularly as blacks who lived in the suburbs surrounding these cities.2i This is consistent with Harrison's findings. After 1971, however, the percentage of black men with regular jobs fell sharply in the central city while holding steady in the suburbs. Able 5-1 suggests, in other words, that while Kain may have been wrong about the benefits of suburban residence when he published his paper in 1968, he may be right today. Kasarda (1989) has used CPS data tapes to make analogous calcula- tions for men between the ages of 16 and 64 who were no longer in school, were not in the armed forces, and had not finished high school. In 1969, joblessness among such men averaged 18.8 percent in central cities and 16.3 percent in the suburbs, suggesting that black suburbanites with limited education were not much better off than their central-city counterparts.22 This finding is consistent with both Harrison's work and Able 5-1. Jobless- ness among black dropouts rose steadily after 1969 in both central cities 21The sampling error of the urban-suburban difference for any one year is quite large. In 1967, for example, the CPS estimated that there were Z1 million black men over the age of 14 in the central cities of SMSAs with 1 million or more inhabitants and 530,000 black men in the sub- urbs of these SMSAs. The 1967 CPS sampled roughly 0.73 percent of all U.S. households, so it should have sampled about 1,500 blacks in the central cities and 390 in the suburbs of these SMSAs. Since roughly half of all black men in these SMSAs worked full-time, year-round, in 1967, the sampling error of the difference between central cities and suburbs was roughly [.52/1,500 + .52/390]5 = 0.028. To minimize such random noise, we averaged the results over five-year penods. The sampling error of the mean difference for a five-year interval is roughly 0.013. These sampling errors would be lower if we included blacks in metropolitan areas of less than 1 million, but such areas provide a weaker test of the spatial mismatch model. 22We use the term "jobless" to subsume all males who were not working. The term therefore includes both those who were looking for work (the officially "unemployed") and those who were not.
210 INNER-C1IY POVERTY IN THE UNITED STATES TABLE 5-1 Employment and Earnings of Black Males Aged 14 and Over in SMSAs of 1 Million or More, by Location and Year Percent Working Full-Tune, YearRound Central city Suburbs Suburban Advantage Suburban Eammgs as Percent of Central City 1967-1971 48.8 49.3 .5 1 11.6 1972-1976 41.0 46.3 5.3 1 10.3 1977-1981 38.9 47.4 8.5 109.1 1982-1986b 38.6 47.7 9.1 112.8 aFor full-tune, year-round workers. bData not published for 1984. SOURCE: Bureau of the Census, Current Population Reports, Senes P-60, "Money income of Households, Families, and Persons in the United States," annual. and suburbs, but it rose much faster in central cities than in suburbs. By 1987, 49.5 percent of all black dropouts in central cities were jobless in a typical month, compared to "only" 33.4 percent in the suburbs. Kasarda focused on black men who had not finished high school because the spatial mismatch hypothesis suggests that residential loca- tion should be especially important for such men. Black men who have attended college are likely to work in white-collar jobs, often in the gov- ernment. These jobs are still concentrated in central cities. If college- educated suburban blacks also enjoy an economic advantage over their central-city counterparts, therefore, we would be inclined to attribute the suburban-urban difference to selective migration rather than to differential job opportunities. Table 5-2 shows urban-suburban differences in the mean number of weeks worked in 1959, 1969, and 1979 by black men with different amounts of schooling. There is no clear difference between central cities and suburbs in either 1959 or 1969. This is consistent with Bible 5-1 as well as with both Harrison and Kasarda's findings for the 1960s. By 1979, black men worked more weeks if they lived in the suburbs than if they lived in central cities, although the difference among poorly educated blacks was smaller than the difference Kasarda reported for that period. The big "news" in Table 5-2 is that suburban residence did as much for college-educated blacks as for those who had not finished high school. The apparent benefits of living in a suburb could, of course, reflect selective migration among college-educated blacks and differential job opportunities among those with less education. But the most parsimonious explanation of liable 5-2 is that movement between central cities and suburbs became more selective at all educational levels during the 1970s.
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES TABLE 5-2 Weeks Worked by Black Males Living in Metropolitan Areas Aged 18 to 59 Who Were Not in School or in Institutions, by Location, Education, and Year 211 1959 1969 1979 Outside Outside Outside Years of Central Central Central Central Central Central School City City City City City City 0-8 39.4 40.2 39.1 41.2 30.9 31.0 (N) (903) (228) (612) (174) (322) (104) 9-11 38.9 40.6 39.3 39.1 31.6 36.2** (N) (495) (80) (676) (148) (611) (198) 12 42.1 41.5 41.9 43.1 35.1 39.4** (376) (51) (706) (178) (878) (391) 13+ 45.6 41.0 43.1 45.0 40.8 43.6* (N) (159) (31) (284) (84) (549) (256) All 40.2 40.5 40.5 41.8 34.9 39.0** At) (1,993) (390) (2,278) (584) (2,410) (949) *Urban-suburban differences significant at the 0.05 level. **Urban-suburban differences significant at the 0.001 level. NOTE: Estimates exclude roughly one-fourth of all urban blacks who lived in SMSAs for which the Census Bureau does not provide an urban-suburban identifier because of conf~- dentiality requirements. SOURCE: Census Bureau 111,000 Public Use sample. A second way to test the spatial mismatch hypothesis is to look at black men's earnings when they live in suburbs rather than central cities. If black men have access to better jobs when they live in the suburbs, they should not only work more regularly but earn more when they work. Column 4 of Table 5-1 shows that in large SMSAs black suburban males with steady jobs earned 12 percent more than their central-city counterparts in 1967-1971, 11 percent more in 1972-1976, 9 percent more in 1977-1981, and 13 percent more in 1982-1986.23 These figures are consistent with the claim that job opportunities are better for suburban men, but they could also be due to selective migration. Harrison's findings suggest that, at least in the 1960s, the earnings differential between central-city and suburban blacks should disappear once we control educational attainment. 1b see if this was the case, we calculated weekly earnings by educational level for central-city and suburban blacks 23The sampling error of the suburban/urban wage ratio for blacks in large metropolitan areas in a given year is about seven percentage points. Averaging over five years reduces the sampling error to about two points. The results in Table 5-1 are thus consistent with the null hypothesis that the underlying ratio did not change from 1967 to 1986.
212 INNER-CITY POVERTY IN THE UNITED STATES using 1/1,000 census samples for 1960, 1970, and 1980. Ibble 5-3 summarizes the results. Like Harrison, we find no significant difference in weekly earnings between central-cit~, and suburban blacks in either 1959 or 1969 once we control education.24 By 1979, however, black males who had not attended college were earning appreciably more if they lived in suburbs rather than central cities. Table 5-3 is thus consistent with the view that by 1979 unskilled and semiskilled blacks could find better jobs if they lived in suburbs than if they lived in central cities. But it is also consistent with the view that highly paid blacks became more inclined to move to the suburbs after 1970. Price and Mills (1985) tell a similar story using data from the May 1978 Current Population Survey. They estimated the effect of suburban residence on the earnings of 25- to 59-year-old full-time, year-round workers. With both education and occupation controlled, black males earned 6 percent more if they lived in suburbs rather than central cities, while white males earned 8 percent more.25 Vrooman and Greenfield (1980) also estimated the effect of suburban residence on black men and women's earnings in the early 1970s, but their sample is so small and unrepresentative that we cannot put much weight on their findings.26 24Table 5-3 shows no difference in weekly earnings between central-city and suburban blacks in either 1959 or 1969, even when we do not control education. Table 5-1, in contrast, suggests that regularly employed suburban blacks earned about 10 percent more than central-city blacks in large SMSAs in 1969. We do not know whether the inconsistencies between the two tables are due to differences in geographic coverage, differences in age coverage, or the inclusion of men who did not work full-time, year-round, in Table 5-3. 25The small size of Price and Mills's black sample raises troubling questions about its represen- tativeness. They took their data from the May 1978 CPS. At that time the CPS sampled roughly 0.7 percent of all households. The Bureau of the Census (1980) indicates that Z3 million black men between the ages of 25 and 54 worked full-time, year-round in 1978, so the May CPS should have included at least 1,600 such men. Yet after excluding men who provided less than half their family's total income, Price and Mills ended up with a sample of only 500 such men. Dropping men whose earnings constituted less than half their family's total income could hardly have elim- inated 70 percent of all black men who worked full-time, year-round. Something else must be going on here, but we cannot say what. Because the sample is small, the sampling error of the 6.0 percent difference between urban and suburban blacks is 3.6 percent. The 95 percent confidence interval for the difference therefore ranges from a 13.2 percent advantage for suburbanites to a 1.2 percent advantage for central-city residents. 26Vrooman and Greenfield used a sample surveyed by the Opinion Research Corporation that included 293 blacks. Almost two-fifths of the black sample lived in suburban areas, which is considerably higher than we would expect on the basis of CPS data. Suburban black males earned 74 percent more than inner-city black males, while suburban black females earned 55 percent more than inner-city black females. No census or CPS survey reports urban-suburban differences of this magnitude. We therefore suspect that the Opinion Research Corporation's sampling frame was unrepresentative for some reason. Assessing Vrooman and Greenfield's results is also complicated by their obscure presentation and by arithmetic errors discussed in Reid (1984~.
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 213 Almost all these comparisons between central-city and suburban blacks support two broad empirical conclusions: 1. In the 1960s, central-city blacks did no better economically than comparable suburban blacks. 2. Since 1970 the economic gap between central-city and suburban blacks has grown appreciably.27 These facts are compatible with two very different hypotheses: 1. The spatial mismatch between black job opportunities and black housing options that Kain described in 1968, while exaggerated at that time, has become quite important since then. Because social conditions in black central-city neighborhoods de- teriorated dramatically between 1965 and 1975, black migration between central cities and suburbs became far more selective than it had previously been. We cannot choose between these two explanations without new data on selective migration. Comparing the studies of blacks who live in the suburbs with the studies of blacks who work in the suburbs, discussed in the previous section, also raises an important puzzle. Both Straszheim and Danziger and Weinstein found that black ghetto residents earned slightly more if they worked in the suburbs rather than the central city in 1965-1970. Yet Bell, Harrison, and our Able 5-3 all indicate that blacks who lived in the suburbs in 1965-1970 ., earned no more than blacks with the same amount of schooling who lived in central cities. The difference between the two kinds of studies may be due to the fact that they cover different cities or different kinds of workers. But the difference may also mean that living in the suburbs is not strongly associated with working in the suburbs. This same paradox arises in opposite form in 1980. Most of the evidence we have examined suggests that by 1980 blacks who lived in suburbs earned more per week than blacks with the same amount of education who lived in central cities. Yet Hughes and Madden found that, with all else equal, Cleveland, Detroit, and Philadelphia blacks who worked regularly in 1980 earned no more when they worked far from the center of their SMSA than when they worked near its center. Once again, we need simple descriptive data showing where blacks who live in different parts of an SMSA work, and how much they earn, in order to reconcile these superficially contradictory findings. If we were to find that blacks could earn more by living in the suburbs, we would also have to ask why so many of them continue to live in central 27The main apparent exception to this generalization is the data on earnings among full-time, year-round, nonwhite male workers in large metropolitan areas, shown in Table 5-1.
214 .s - 3 0\ C 'O ._ · _4 of ~ ._ US ~ ~3 ., ~ D ~ _ .S ~ U' - ~ O mod 0 ~ ~ ^ a' 0 , ~ . ~ 0 6,P 04t~- .S P 0 ID - ~ ._ o O U' .= ~ 106 0\ 0\ - 0\ lo 0\ ° 1 ~ ~ .~ ~ _ ~33 ~ .~ at ~ ~ b. o Go~ CalCal Cal .~ o o · . ~ of ~ an ~ ~ ~ en - ~ - - - ~ - ~= ~ s to A oN om ~ ~ to - lo - ~ - - - - ~ ~ ~ ~ A A ~ ~ - - ~4 ~- - ~ - - - o So .= ~ om to Go os ~ oo - oS . . . o - o ~- - ~- `- c~) ~ t~) ~ ~ o - ~oo o oo o ~oo ~o - c~ - - - - aS~ =0g =E
215 o ~o ~Cat of ~Cal Go lo. ~of lo. o _ o o _ _% _% _ _ ~ Ha, ~ ~ ~ ~ ~ ~ ~ ~ o ~ ~o _ _ ~_ C'' _ Cal Cal - ~= ~_ - _ o_= ~no _ ~ ~oo o oo o~ 00 ~O _ In O ~_% ~oo \4, _~ _ ~ oo ~o ~ _ ~OS ~ o o X o o~ .. oo _~o _X .. _o '_ _ _ _% _ ~o ~ ~ ~ ~o, ~ ~ ~ ~ ~ ~ =~; -8- ~a ~ ~- U, C) Ce 00 80~ =~ U) C~ d" ) Q) .S pr: C) ~ U, et U) ._ ~ ._ ,~ 3 CJ ~ 3 6 C) G) U ~ .5 c ~o . ~ . ~ ~ _ ~ E =3 ~ ~ ~ e~ C ~ C _= Cq E a~ ~ ~ O ~ ~ C _- 3.5: ~- 04 C ~ .5 ~ ~ ~ C, ~ ~ ~ ~ ~ ~ C C) ~ ~ V 365 =. ..-_ C.) ~ z~ ° °
216 INNER-CITY POVERTY IN THE UNITED STATES cities. The conventional answer is that blacks cannot buy housing in all- white suburbs, or that they do not feel comfortable in such suburbs. Racial discrimination is certainly a real problem, both in the housing market and in the social lives of black suburbanites. But for our purposes the question is not why so few blacks live in white suburbs but why so few live in black suburbs and black neighborhoods of racially mixed suburbs. If job opportunities are better for blacks who live in black suburbs, blacks should keep moving to these suburbs until local demand for black workers is saturated and the suburban wage advantage disappears. Whites have, of course, resisted the growth of black suburban neigh- borhoods. But they have also resisted the growth of black urban neigh- borhoods. Our question is therefore comparative: Has white resistance to black residential growth been more effective in the suburbs than in the central cities and, if so, why? The easiest way to answer this question is to investigate whether the supply of black suburban housing has kept pace with demand. If it has not, the price differential between housing in black central-city and black suburban neighborhoods should be greater than the price differential between comparable housing units in white central- cib and suburban neighborhoods. The price of housing in black suburbs should also have increased faster over the past 20 years than the price of comparable housing in black central-city neighborhoods. We have not reviewed the literature on housing prices, but our informal inquiries uncovered no work on trends in the relative price of housing in urban versus suburban black neighborhoods.28 Unless the price of black suburban housing has risen faster than the price of comparable housing in other areas, it is hard to argue that discrimination has discouraged blacks from moving to black suburban areas although one can certainly argue that discrimination has discouraged blacks from moving to white areas. If discrimination has not discouraged black movement to the suburbs, the basic premise of the spatial mismatch hypothesis needs to be reexamined. 28Using a 1/1,000 sample of census records, we found that among whites gross rent (i.e., rent plus heat and utilities) in the suburbs exceeded gross rent in the central city by 9 percent in 1960, 10 percent in 1970, and 11 percent in 1980. Among blacks, gross rent averaged 20 percent less in the suburbs than in the central city in 1960, 4 percent more in 1970, and 21 percent more in 1980. If the figures for blacks covered the same housing units in all three Yeats, they would provide strong support for the hypothesis that demand for suburban housing rose faster than supply during this period. In fact, however, the housing stock rented by blacks in 1980 was much larger and of much higher quality than that rented by blacks in 1960. Much of the apparent increase in relative prices may, therefore, reflect improvement in housing quality rather than increased demand for suburban locations. A convincing analysis of this issue would have to compare price trends in specific black, white, and mixed neighborhoods throughout this period. A study of land values in black and white areas would also be instructive.
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES WHERE DO WE GO FROM HERE? We began with four questions: 217 How does residential segregation affect demand for black workers? How does proximity to work affect the supply of black workers available for employment? Can black central-city residents earn more by commuting to the suburbs? Can black central-city residents earn more if they live in the sub- urbs? Our tentative answers to these questions are as follows. Demand for Black Workers Residential segregation is likely to reduce demand for black workers in white areas and increase demand for black workers in black areas. Whether the net result is to increase or decrease aggregate demand for black workers depends on a number of factors: how strongly whites dislike working in black areas, how strongly blacks dislike working in white areas, whether black customers prefer trading with firms that discriminate in favor of black workers, whether white customers prefer trading with firms that discriminate in favor of white workers, whether firms that mainly hire blacks are more or less productive than firms that mainly hire whites, and hence whether firms prefer hiring relatively cheap labor in the ghetto or more expensive labor in white areas. All these factors are likely to change over time. They also vary from place to place. As a result, we cannot predict the economic consequences of residential segregation by using deductive logic. When we turn from theory to evidence, we find that the level of residential segregation in a metropolitan area had very little effect on income differences between black and white men in either 1959 or 1969. No one has investigated changes since 1969. Nor has anyone studied the effect of residential segregation on the earnings of black women or black teenagers. Based on what we now know, it is hard to draw any strong conclusions about how residential segregation currently affects demand for black workers. All we can say is that it did not appear to have much impact in 1959 or 1969. Job Proximity and Labor Supply The supply of black workers available to a firm at any given wage rate clearly diminishes as distance from black residential areas increases. But it does not follow that moving black workers closer to major centers of
218 INNER-CITY POVE~IY IN THE UNITED STATES employment would increase the number of blacks taking jobs at any given wage rate. Comparisons among Pittsburgh and Los Angeles neighborhoods sug- gest that job proximity has little impact on black men's chances of holding a job. Distance does have an effect on black teenagers' chances of working in Philadelphia, Los Angeles, and Chicago, however, and in the best cur- rently available study distance explains 30 to 50 percent of the racial gap in teenage employment rates. While job proximity does not appear very important for adults in Pittsburgh or Los Angeles, blacks do seem to have done better economically in both 1960 and 1980 when they lived in SMSAs where manufacturing, trade, and service jobs were concentrated in the central city rather than in the suburbs. Suburban Versus Central-City Wages In 1965-1970, ghetto blacks who worked in the suburbs of Cleveland, Detroit, St. Louis, and San Francisco earned about 10 percent more than those who worked in the ghetto. This wage premium was roughly sufficient to cover the extra cost of commuting to the suburbs. In 1980, however, Hughes and Madden found that the overall wage level for all black workers was no higher near the periphery of the Cleveland and Philadelphia SMSAs than near the center, and it was slightly lower near the periphery of the Detroit SMSA The most likely explanation of this apparent contradiction is that for blacks suburban wages were roughly equal to central-city wages in both 1970 and 1980, but that inner-city black workers only commute to the suburbs if they happen to find a job that pays more than they think they could earn nearer home. Blacks Who Live in the Suburbs We found no evidence that suburban blacks fared better than compa- rable central-city blacks in 1966-1970, when Kain first advanced the spatial isolation hypothesis. Since 1970, however, joblessness has risen faster among central-city blacks than among suburban blacks. We do not know whether this was because of selective migration or because demand for black workers fell more in central cities than in the suburbs. Policy Implications Taken together, these findings tell a very mixed story. They provide no direct support for the hypothesis that residential segregation affects the aggregate level of demand for black workers. They provide some support
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 219 for the idea that job proximity increases the supply of black workers, but the support is so mixed that no prudent policy analyst should rely on it. Those who argue that moving blacks to the suburbs would improve their job prospects, or that improving public transportation to the suburbs would reduce unemployment in the central-city ghetto, must recognize that there is as much evidence against such claims as for them. Future Research Because of its intuitive appeal, the spatial mismatch hypothesis is unlikely to go away, no matter what the evidence shows. When the evidence is mixed, as it is now, those who find the theory appealing will continue to defend it, and policymakers who find it appealing will continue to listen. Under these circumstances we ought to make a serious effort to improve the quality of the available evidence. Four steps seem especially pressing. First, since theory suggests that the effects of living in the central-city ghetto are likely to have changed over time, since the available descriptive statistics also point to such changes, and since much of the most widely cited evidence in this debate is now more than 20 years old, we need to use the census, the CPS, and the PSID to assemble time series comparing: · blacks who live in suburbs to blacks who live in central cities; blacks who work in suburbs to blacks who work in central cities; blacks who live in very large cities, where the spatial mismatch hypothesis seems likely to have its strongest effect, to blacks who live in smaller cities, where travel time to suburban jobs is likely to be modest; blacks who live in cities where most blue-collar jobs are far from the central-city ghetto to blacks who live in cities where most blue-collar jobs are close to the central-city ghetto; and blacks who live in highly segregated SMSAs to blacks who live in less segregated SMSAs. All these tabulations should be broken down by sex, educational level, and age. Having descriptive material of this kind readily available in a Census Bureau publication would do more than almost any piece of primary research to raise the level of both public understanding and scholarly argument about these issues. Second, we need more fine-grained descriptions of specific cities, in- cluding maps that show where workers with specified characteristics earn the most and the least, which neighborhoods have major centers of blue- collar employment, and which neighborhoods have the highest and lowest rates of labor force participation. This is a field where a picture is worth considerably more than a thousand words.
220 INNER-CITY POVERTY IN THE UNITED STATES Third, we need to look more carefully at the effects of changes over time in cities' residential and employment patterns. Cross-sectional com- parisons among SMSAs can sometimes be instructive, but they have severe limitations. If any one city pursues policies that benefit blacks, its reward will be to attract new black migrants, whose presence will drive up black unemployment and drive down black wages. This means that no city can expect to keep demand for black labor much tighter than the national average for a protracted period (e.g., more than 20 years). It follows that if we want to estimate the erects of naiwnal policies aimed at keeping manufacturing near the central-city ghetto or at helping central-city blacks move to the suburbs, city-to-city comparisons may lead us badly astray, seriously underestimating the benefits of the policies in question. In order to do better, we need dynamic models that take account of intercity as well as intracity migration. Fourth, and perhaps most important, we need to investigate the deter- minants of residential movements between central cities and suburbs. We then need to estimate the contribution of selective migration to employ- ment and earning differences between central~ity and suburban blacks. Social scientists' collective failure to take account of selective migration is probably the single most important reason why we have learned so little about this subject in the two decades since Kain first advanced the spatial mismatch hypothesis. ACKNOWLEDGMENTS We are indebted to Bennett Harrison, Harry Holzer, Jonathan Leonard, and Michael Wiseman for comments on earlier drafts of this chapter. The Center for Urban Affairs and Policy Research at Northwestern University provided financial assistance. REFERENCES Bell, Duran, Jr. 1974 Residential location, economic performance, and public employment. Pp. 55- 75 in Patterns of Racial Discrunination, Vol. 1, George M. Von Furstenberg, Bennett Harrison, and Ann R. Horowitz, eds. Lexington, Mass.: Lexington Books. Bureau of the Census 1%9 Income in 1967 of Persons in the United States. Current Population Reports, Series P 60, No. 60. Washington, D.C.: U.S. Department of Commerce. Money Income of Families and Persons in the United States: 1978. Current Population Reports, Series P 60, No. 123. Washington, D.C.: U.S. Department of Commerce. 1983 Financial Characteristics of the Housing Inventory for the United States and Regions: 1983. Annual Housing Survey: 1983, Part C. Current Housing Reports, Series H-150-83. Washington, D.C: U.S. Department of Commerce. 1gS0
RESIDENTIAL SEGREGATION AND BLACK JOB OPPORTUNITIES 1988 221 Money Income of Families and Persons in the United States: 1986. Current Population Reports, Series P 60, No. 159. Washington, D.C.: U.S. Department of Commerce. Byrne, James J. 1975 Occupational mobility of workers. Special Labor Force Report, No. 176. Washington, D.C.: Bureau of Labor Statistics. Danziger, Sheldon, and Michael Weinstein 1976 Employment location and wage rates of poverty-area residents. Joumal of Urban Economics 3:127-145. Ellwood, David ~ 1986 The spatial mismatch hypothesis. Are there teen-age jobs missing in the ghetto? Pp. 147-190 in We Black Youth Emp~rnent Crisis, Richard B. Freeman and Harry J. Holzer, eds. Chicago: University of Chicago Press. Barley, John E. 1987 Disproportionate black and Hispanic unemployment in US metropolitan areas. American Journal of Economics and Sociology' 46:129-150. Harrison, Bennett 1972 The intrametropolitan distribution of minority economic welfare. Joumal of Regional Science 12:23~3. Discrimination in space: Suburbanization and black unemployment in cities. Pp. 21-53 in Patterns of Racial Discrimination, Vol. 1, George M. Von Gutenberg, Bennett Harrison, and Ann R. Horowitz, eds. ~ "ington, Mass.: Lexington Books. Hughes, Mark Alan 1987 Moving up and moving out: Confusing ends and means about ghetto dispersal. Urban Studies 24:50~517. Hughes, Mark Alan, and Janice Fanning Madden Forth- Residential segregation and the economic status of black workers: New coming evidence for an old debate. Joumal of Urban Economics. Hutchinson, Peter M. 1g74 1be effects of accessibility and segregation on the employment of the urban poor. Pp. 77-96 in Pattems of Racial Discrimination, Vol. 1, George M. Von Gutenberg, Bennett Harrison, and Ann R. Horowitz, eds. Lexington, Mass.: Lexington Books. Ihlanfeldt, Keith R. 1988 Intra-metropolitan variation in earnings and labor market discrimination: An economic analysis of the Atlanta labor market. Southern Economic Joumal 55:123-140. Ihlanfeldt, Keith R., and David L Sjoquist 1989 The impact of pb decentralization on the economic welfare of central city blacks. Journal of Urban Economics 26:110-130. 1990 Job accessibility and racial differences in youth employment rates. American Economic Review 80:267-276. 1974 Kain, John F. 1968 Housing segregation, Negro employment, and metropolitan decentralization. The Quarter) Journal of Economics 82:175-197. 1974 Reply to Stanley Masters' comment on "Housing Segregation, Negro Employ ment, and Metropolitan Decentralization." Quarterly Joumal of Economics 88:511-519.
222 INNER-CITY POVERTY IN THE UNITED STATES Kasarda, John 1989 Urban industrial transition and the underclass. 17'e Annals of the American Academy of Political and Social Science 501:26-47. Kimchenman, Joleen, and Kathryn Neckerman Forth- We'd love to hire them, but . . .: The meaning of race for employers. coming In Christopher Jencks and Paul Peterson, eds. 17'e Urban Ur~erc~*s. Washington, D.C.: Brookings Institution. Leonard, Jonathan S. 1986 Space, Time, and Unemployment: Los Angeles 1980. Unpublished paper, School of Business Administration, University of California, Berkeley. 1987 The interaction of residential segregation and employment discrimination. Journal of Urban Economics 21:323 346. Massey, Douglas S., and Nancy A. Denton 1988 Suburbanization and segregation in U.S. metropolitan statistical areas. AmeT- ican Journal of Sociology 94~3):592-626. Massey, Douglas S., and Mitchell Lo Eggers 1989 The ecology of inequality: Minorities and the concentration of poverty 1970-1980. National Opinion Research Center, University of Chicago. Mastem, Stanley H. 1974 A note on John Kain's "Housing Segregation, Negro Employment, and Metropolitan Decentralization." Quarterly Journal of Economics 88:505-519. Black-White Income Differentials: Empirical Studies and Policy Implications. New York: Academic Press. Mooney, Joseph D. 1969 Housing segregation, Negro employment, and metropolitan decentralization: An alternative perspective. Quarterly Journal of Economics 83:299-311. National Center for Educational Statistics 1988 Digest ofEducational Statistics. Washington, D.C: U.S. Government Printing Office. Odner, Paul, and Daniel H. Saks 1971 A note on John Kain's "Housing Segregation, Negro Employment, and Metropolitan Decentralization." Me Quarterly Joumal of Economies 191:147- 160. Price, R., and E. S. Mills 1985 Race and residence in earnings determination. Joumal of Urban Economics 17:1-18. Reid, Clifford 1984 Are blacks making it in the suburbs? Joumal of Urban Econonucs 16:357-359. Straszheim, Mahlon R. 1980 Discrimination and the spatial characteristics of the urban labor market for black workers. Journal of Urban Economics 7:119-140. Vrooman, John, and Stuart Greenfield 1980 Are blacks making it in the suburbs? Some new evidence on intrametropoli- tan spatial segmentation. Journal of Urban Economics 7:155-167. Wilson, William Julius 1987 The Truly Disadvantaged. Chicago: University of Chicago Press. 1~75