Welfare to Work: The Need to Take Place Differences into Account


Hal Wolman, Ph.D.

Dr. Wolman is a Professor in the College of Urban, Labor and Metropolitan Affairs and in the Department of Political Science at Wayne State University in Detroit, Michigan. He is serving as a visiting policy analyst in the Office of the Assistant Secretary for Planning and Evaluation through the Intergovernmental Personnel Act.


Technical Analysis Paper No. 45
January, 1996


DHHS, quite understandably, tends to think of the world in terms of individuals, families and households to whom it delivers services. However, those whose concern is with economic development and labor market policy tend to see the world more in terms of places and variations among places. What might be the implications for DHHS concerns, particularly the ability to move from welfare to work, if it attempted to think systematically about the problem of service delivery to individuals, families and households within the context of places or areas which differ in important respects?

Do differences among areas in their characteristics affect the ability of individuals to move from welfare to work? What are these place-related characteristics? If they do affect the ability to move from welfare to work, what are the implications, particularly as we consider substantial changes in the existing welfare system, in terms of policy and legislation, technical assistance, and research and evaluation?

The Effect of Place Differences on Ability to Move from Welfare to Work:

Differences exist by geography in the ability to move from welfare to work. These differences exist among regions and states, between metropolitan and non-metropolitan areas and, within metropolitan areas, between central cities and suburbs. Bane and Ellwood (1983, p. 85) found that, controlling for a variety of other variables, welfare exits due to earnings increase were more common in the South and East than in the West and North Central region. In a model holding other variables constant they estimated (p. 42) that the cumulative percentage of welfare recipients exiting welfare within two years via earnings increase would be 27% in the South, 17% in the North East, 15% in the North Central region and 11% in the West. (See also Hutchens, 1981, p. 232.)

At the state level, data from the Quarterly Public Assistance Statistics indicate substantial variation among the states in AFDC discontinuations as a percentage of average monthly welfare caseload. The mean rate of average monthly exits from AFDC as a percentage of average monthly caseload was 5.53%, but the variation around the mean was substantial: it was, for example, 12.62% in South Dakota, 8.30% in Arkansas, 8.20% in Colorado, and 7.54 in North Carolina, but only 3.45% in the state of Washington, 3.47% in New York, 3.60 in Illinois and Hawaii, and 3.69 in Louisiana. Data for exits due to increased earnings from employment are of very poor quality, but the degree of state by state variation was even greater.

Fitzgerald (1995) and Vartinian (1995a) both examined the likelihood of exiting the welfare rolls by various means (as a result of increased earnings, as a result of marriage, and for other reasons) in panel studies of individuals for which they matched place characteristics, including labor market characteristics, to individuals. Fitzgerald found that, controlling for a variety of personal variables, the probability of exiting from welfare within a two year time period was considerably lower for a welfare family living in a large metropolitan area (population of 250,000 or more) than it would be for the same family living elsewhere. Vartinian found that the likelihood of exiting welfare was lower for residents of metropolitan areas than for rural areas, lower for residents of big cities than for residents of suburban areas and similar for residents of rural areas and suburban areas.

Differences also exist in the ability to move from welfare to work among metropolitan areas. While data for welfare closures are not available nationally for substate levels, an unsystematic sampling of states by telephone provided information on the variation of welfare closure rates among metropolitan areas. In New York, for example, 1994 average monthly exits from welfare due to earnings as a percentage of average monthly caseload was three times greater in the rest of New York state than in New York City. In California, average monthly exits from welfare due to earnings increases as a percentage of average monthly caseload was ten times greater, for example, in Orange County than it was in San Francisco county and twice as great in Santa Clara County as in Sacramento County.

In the remainder of this report I will focus mostly on these differences among metropolitan areas, since they contain most of the U.S. population and since they can be considered as functional labor markets.

There are a variety of reasons why differences among places might affect the ability of welfare recipients to exit welfare and gain employment. Mitchell et. al. (1979) examined the performance of local WIN (Work Incentive Program) programs designed to move welfare recipients into employment and found that 30% of the variation in the performance of these local programs in terms of job entries and 43% in terms of the extent of welfare grant reductions brought about by these entries could be attributed to variation in local labor market and demographic place related factors. We begin the discussion with variation in local labor market conditions.

Labor Markets

Metropolitan areas (MSAs) differ in the condition of their labor markets. These differences exist on both the demand side and the supply side. In particular, there are differences among MSAs in the demand for low-skilled labor and in the employment skills of welfare recipients and their willingness to work.

Demand side

Presumably it should be easier to move from welfare to work in tight labor markets (i.e., areas with low rates of unemployment where employers are actively seeking employees) than in loose labor markets. In 1994, the rate of unemployment in metropolitan areas, at a time when the national unemployment rate was 6.1%, varied from a low of 2.6% in Lincoln, Nebraska to a high of 18.2% in McAllen-Edinburg-Mission, Texas. Among metropolitan areas with unemployment rates above 10% were Bakersfield (14.3%), Fresno (13.7%), Brownsville (12.1%), New Bedford, Mass. (10.0%) and El Paso (10.0%). Those with unemployment rates of 3% or below included Fayeteville (2.7%), Des Moines (2.9%), Omaha (3.0%), and Raleigh-Durham-Chapel Hill (3%).

In addition, since AFDC recipients have below average education levels and cognitive skills they are likely, when they find employment, to do so in the low-skill service sector and in light factory work (see Zill et. al., 1991). It should, therefore, also be easier to move from welfare to work in areas that are experiencing growth in relatively low-skilled jobs and where there are substantial job vacancies for these types of jobs than in areas where such jobs are not readily available. One measure of this is the unemployment rate for persons in low skilled occupations. In 1993 the unemployment rate for handlers, equipment cleaners, helpers and laborers varied among the metropolitan areas for which the Bureau of Labor Statistics collected data from 6.0% in Charlotte to 20.8% in New York. The unemployment rate for service workers varied from 4.0% in the San Jose metropolitan area to 14.5% in the Oklahoma City MSA. (BLS, Geographic Profile of Employment and Unemployment 1993.) Kasarda (1995) points to the divergent experience between Frostbelt and Sunbelt cities from 1970-1990: the former substituted higher skilled information processing jobs for lower skilled jobs in traditional industries, while the latter added jobs in both categories.

While a substantial amount of evidence supports the contention that tight labor markets improve the employment and earnings prospects of low income and low skilled workers and of youth (see Freeman, 1982 and 1991; Ihlanfeldt and Sjoquist, 1991; Cain and Finnie, 1990, and Acs and Wissoker, 1991 - latter two cited in Fitzgerald), there are only a few studies that focus directly on the impact of labor market demand on welfare recipients.

Empirical evidence strongly suggests that tight labor markets improve employment and earnings prospects of low-income and low-skilled individuals in general. Freeman (1991, see also Freeman, 1982), for example, examined the response of young men and young Black men with 12 or fewer years of schooling to variation in labor market conditions. He concluded that, "Local labor market shortages greatly improve the employment opportunities of disadvantaged young men, substantially raising the percentage employed and reducing their unemployment rate. Employment of black youths is particularly sensitive to the state of the local labor markets."

Vartinian (1995a) found that increases in the unemployment rate in the labor market(metropolitan areas) resulted in lower likelihood of welfare exits via either increased earnings or marriage and concluded (p. 15) that, "where employment opportunities are scarce, the likelihood of exiting the AFDC program through either marriage or increased earnings decreases, either because jobs are unavailable or because "marriageable males" are unavailable." Harris (1993, p. 343) found that the probability of exiting welfare immediately after gaining employment is negatively and significantly related to the local (county) unemployment rate. Fitzgerald (1995) found that an increase in the local unemployment rate of 2.72 percentage points would bring about a decline in welfare exit rates for Blacks of 23%.

West et. al (1993) found that a one percentage point increase in the unemployment rate brought about a 0.643 percentage point decline in the rate of welfare recipients who had participated in job training programs and were employed 13 weeks after training terminated. This result is derived from a Department of Labor funded study designed to help the Department devise performance standards for states to apply to local training agencies as part of the Job Training Partnership Act (JTPA). The Department specifies and estimates a regression model explaining post-training employment rates for program participants of local training agencies. A participant is considered employed if he/she is working 20 or more hours per week 13 weeks after termination of training. The model contains a wide range of both personal and environmental variables. The co- efficients of the variables are then utilized as weighting measures to derive an expected performance for each agency, taking into account the difficulty of the task presented by both the mix of clients and local economic and other conditions.

Finally, Bartik (Journal of Urban Economics, forthcoming) estimated that a reduction of one percentage point in the unemployment rate of a metropolitan area would result in a reduction of 0.688 percentage points for men and 1.094 percentage points for women in the poverty rate (not a perfect proxy for the rate of welfare exit by any means, since it does not involve individual behavior, but nonetheless indicative).

Gueron and Pauly (1991, pp. 186-187) note that while it may be easier to move from welfare to work in tight as opposed to loose labor markets, there is some evidence suggesting that the impact of programs designed to assist this movement may actually be greater when unemployment is high. They argue that this results from the fact that under tight labor market conditions the control group also finds it relatively easy to find employment, and thus the program impact, as measured by the difference between the program participants and the control group, is small.

Evidence also exists on the proposition that variation across areas in the rate of total employment growth, particularly in low-skill industries, has an impact on the ability to exit welfare. Vartinian found (pp. 18-19) that change in both the number of manufacturing jobs and retail and wholesale jobs in large cities was positively related to the likelihood of residents of those cities to exit from welfare; a decline in the number of jobs in these sectors was associated with a lower probability of exiting welfare. Albert (1988, p. 128) found that employment change in the apparel industry in California was similarly related to welfare exits. Bartik (Journal of Urban Economics, forthcoming) examined the effect of employment growth in metropolitan areas on the probability of an individual's entry and exit from poverty. He concluded that, " a one percent increase in MSA employment will reduce the poverty rate by 1/3rd of one percent for women and 1/5th of one percent for men."

Supply side

It should be more difficult for a welfare recipient to find work in areas where there are a large number (relative to the population) of other low-skilled individuals seeking work and thus competing with her for a job than in areas where there are a relatively small number seeking employment. This may be of particular importance if a welfare time limit is imposed; those areas which have a relatively high proportion of welfare recipients in the population may suddenly experience a substantial increase in active job seekers without any corresponding short term increase in the number of available low-skill jobs.

Also on the supply side, areas may well differ with respect to the human capital characteristics of their welfare recipients. Welfare recipients may, on average, have more cognitive or job-related skills, job readiness or motivation in one locale than in another. There are many reasons why this may be the case, but the most obvious is differences in the quality of public school systems, particularly central city public school systems, across metropolitan areas.

Variations in local wage rates are likely to affect closures. People are more likely, ceteris paribus, to exit welfare in metropolitan areas where wage rates are high and employment will yield a decent income than where wages are low. Hutchens (1981, p. 233) found that a 10% increase in wage rate, controlling for other variables, was associated with an increase in the probability of an individual's exit from welfare from .359 to .379 (see also Harris, 1993, p. 343, for similar findings.). However, the impact of the local wage rate is relative to the level of reservation wages (the wage at which welfare recipients are willing to enter employment). Average reservation wages may also differ from one metropolitan area to another; welfare recipients with lower reservation wages are more likely to enter employment.

Differences in state policies

Differences in state policies are another place-related characteristic that may differentially affect the ability of recipients to move from welfare to work. Reservation wages (see the above discussion) are related to AFDC benefit levels (including the value of Medicaid and other benefits which are conditioned upon AFDC participation), a policy variable that currently varies from state to state. Reservation wages will certainly be higher in states with high AFDC benefit levels than in states with low AFDC benefit level (see, for example, Bane and Ellwood, 1983, p. 48 and O'Neill et. al, 1984, p. 14); as a consequence exits from welfare to work are likely to be more frequent in states with low benefit levels. Hutchens (1981, p. 233) found that a 10% increase in a state's AFDC payment for a family with no other income, controlling for a variety of other variables, was associated with a decline in the probability of an individual's exit from welfare from .359 to .324 (see also Plotnick, 1983, p. 74, and Gueron and Pauly, 1991, p. 187 for similar findings). Under a system where states have greater discretion - such as a block granted system - the scope for state variation in policy characteristics that affect the ability of recipients to move from welfare to work will be much greater. State policy differences related to eligibility, income disregards, the availability and financing of child care, the existence of public service or community service jobs etc, as well as benefit level, will all affect the ability of welfare recipients to find employment.

Differences in the characteristics of welfare clients

It is much more difficult for certain types of welfare recipients to move from welfare to work than it is for others. Research suggests that long term welfare recipients (those with continuous spells of two or more years) are much less likely to exit welfare to employment than others. West et. al. (1993) find that for each additional percentage point of long term welfare recipients as a percent of all welfare recipients engaged in training programs, the post-training employment rate falls by 0.15 percentage point. They also find that the percentage lacking significant work history, percentage unemployed 15 or more weeks prior to training, and the percentage of high school dropouts are all significantly and negatively related to post-training employment rates. Areas with a higher proportion of such clients in their welfare caseloads will have correspondingly greater difficulty in moving clients to employment.

Similarly, a large number of welfare recipients are characterized by health and/or substance abuse problems that makes their entry into employment problematic. Zill et. al. (1991) present data indicating that one fifth of AFDC recipients has some form of health limitation and one fourth has an alcohol related problem. Data on drug abuse is not easily available, but there is good reason to believe that the incidence of drug abuse among AFDC mothers is also an important barrier to employment. If the proportion of the welfare caseload experiencing these problems differs from area to area there should also be related differences in the movement from welfare to work across areas.

Quality and competence of local institutions

The ability for a welfare recipient to move from welfare to work depends to some extent on the quality and competence of the set of local institutions that are in place to serve and assist her - in particular, local welfare, employment and training and other labor market intermediary institutions (e.g., employment agencies, vocational schools, community colleges, local economic development agencies). We can assume that the quality of these institutions varies across areas as does the degree to which this set of institutions performs in a cooperative, integrated, and coordinative manner. Where local welfare and employment and training agencies co-operate closely - and where these, in turn, are well-integrated with other community institutions - welfare recipients should, ceteris paribus, have an easier path moving from welfare to work.

There is little empirical research available on the quality of local welfare and job training agencies once the characteristics of the labor market, other environmental characteristics and the personal attributes of the clients are taken into account. There is, however, reason to suspect that the impact of local agency action might be quite substantial. Using a very wide range of environmental and personal variables as factors in a model designed to explain the post-training employment rate for local agencies of welfare recipients who had participated in JTPA programs, West et. al (1993, p. 15) could account for only 22% of the variation. That leaves 78% which might be attributable to local agency activities, although some portion of that 78% is undoubtedly due to non- agency related variables that were left out of the original model. (The model was able to explain 36% of the variation in post- training employment rates for all participants - welfare as well as non-welfare recipients - leaving a total of 64% for potential local agency effect.)

Spatial mismatch

It has frequently been argued that inner city residents are isolated from low-skilled jobs for which growth is occurring primarily in suburban areas (see, for example, Wilson, 1987, Kasarda, 1995). Although early research results did not support this contention (see Ellwood, 1986), more recent studies have found a relationship between travel time or transportation access and employment (see, for example, Ihlanfeldt and Sjoquist, 1990, 1991). The degree of this spatial mismatch between the location of potential workers and the location of available jobs is likely to vary from one metropolitan area to another. It is likely to be more severe in areas with poor quality public transportation systems and, ceteris paribus, in larger metropolitan areas.

Ihlanfeldt and Sjoquist (1991) examined 43 MSAs with populations of one million or more and found average travel time for central city youths (aged 16-19) to low income jobs (wage of less than $5.00 per hour) was 26.4 minutes for Blacks with a standard deviation of 5.3 and 20.0 minutes for Whites with a standard deviation of 3.2. They found that travel time was significantly and negatively related to probability of employment for both Black and White central city youth (though the magnitude was greater for the former). They estimated that if commuting time for the MSA where the average commute was the highest were to change to the time found in the MSA with the lowest value the probability of employment for Black youth in that area would increase by 70.9% and of employment for white youth of 30.2% (p. 261).

Discrimination

The degree of employment discrimination by race is likely to vary across MSAs as is the willingness to hire welfare recipients (or to consider welfare receipt as stigmatizing). Fix et. al., (1993, p. 24) used an audit testing approach (sending matched pairs of white and minority housing seekers to respond to advertisements) to estimate the level of housing discrimination in five metropolitan areas. They found that the proportion of Blacks facing discrimination (that is, treated less favorably than their matched White counterpart) varied from 36% in the Los Angeles area to 49% in Atlanta, while the proportion of Hispanics facing discrimination varied from 39% in the Chicago area to 61% in the New York area. In a similar audit with respect to hiring, but conducted in only two metropolitan areas, they found 17% of Blacks were discriminated against in the Chicago area compared to 23% in the Washington D.C. area.

In addition, the proportion of the metropolitan area Black population living in suburbs (at least partially a result of discrimination in suburban housing markets) also varies by metropolitan area (see Massey and Denton, 1993), and access to suburban housing appears for a variety of reasons, including improved access to suburban jobs, to increase the employment prospects of low-income Black women. Popkin et. al. (1993) examined a random sample of participants in the Gautreaux program in Chicago. The Gautreaux program helps Black families who are public housing residents in the city of Chicago to move into private subsidized housing in either the city or its suburbs. Families have very little choice in determining whether their move is within the central city or to a suburb. Popkin et. al found that, controlling for work history, human capital and a variety of other personal factors, suburban movers were significantly more likely to have a job after the move than were city movers.

The "culture" of welfare

The willingness to move from welfare to work depends, to some extent, on the way in which welfare recipients view the "legitimacy" and community acceptance of welfare as a status relative to the legitimacy of work as a status. It is argued that the "acceptability" of welfare is greater where there are large concentrations of welfare recipients and relatively few role models of working adults and that a welfare recipient is less likely to move to work in these areas than she would in other areas where poverty is not so concentrated. (Vartinian, 1995b, for example, finds that exits from welfare via earnings are significantly related to neighborhoods of concentrated poverty and problems - the higher the neighborhood score on an index of economic problems that included the percentage of people on public assistance, the poverty rate and the unemployment rate, the less likely a welfare recipient was to exit welfare within a two year period.)

The extent to which such concentrations of poverty exist varies enormously; a small number of cities contain a substantial number of these areas of heavily concentrated poverty (defined as census tracts where 40% or more of the population is below the poverty line), while most metropolitan areas have few if any such areas. Kasarda (1993) presents data indicating that five cities (New York, Chicago, Detroit, Cleveland, and Milwaukee) contain 36.6% of all such census tracts in the country's 100 largest cities in 1990. The extent to which cities contained areas of heavily concentrated poverty varied substantially by region: in the Midwest nearly 20% of large city census tracts were of this nature compared to 14.7% in the South, 13.7% in the Northeast, and 5.8% in the West. While 11% of the households in these 100 cities received public assistance in 1990, nearly 1/3 of the households in areas of heavily concentrated poverty did so.

Availability and effectiveness of local social services

Welfare recipients suffering from health and substance abuse problems will likely need successful treatment before they are able to move successfully from welfare to work. The availability and effectiveness of such treatment undoubtedly varies from area to area as does the availability of affordable and reliable child care, another necessary prerequisite for moving welfare mothers to employment.

Implications of Differences among Places in Ability of Welfare Recipients to Move from Welfare to Work

The above discussion provides strong support for the contention that the ability to move people from welfare to work will vary by place. A welfare recipient household with a given set of characteristics will find it easier to exit welfare for work in some areas than in other areas.

Performance standards

This means that some states and some local governments will face - because of local conditions - a much more difficult task in meeting performance standards that set a specific and uniform standard of performance (e.g., move 50% of welfare recipients into employment within 5 years) than will others.

Performance standards for welfare to work programs should reflect the fact that it is more difficult, for the various reasons cited above, to move welfare recipients to employment in some areas than in others. Presumably performance standards, if they are applied, should be sensitive to these differences and take them into account. In addition, the ability of a state to meet performance standards (and, thus, receive any rewards or avoid any penalties contingent upon reaching the standards) are not unrelated to many aspects of state program design. In particular, states that set lower benefit levels are much more likely to be able to move recipients to work and thus meet performance standards. Unless carefully thought through, performance standards can stand as a perverse incentive to drive down state benefit levels.

Improving the quality and performance of local welfare and job training agencies

Research suggests that some not insignificant portion of the difference among local agencies in the ability to move people from welfare to work relates to differences in the behavior and internal operations of local welfare and job training agencies. These differences could relate either to differences in program design and/or services provided by local agencies or to differences in agency administrative and management practices. Greater attention needs to be paid to the impact on performance of variations in program design (see below) under different external circumstances and to the internal operations of the local agencies themselves. In particular, broad gauged performance standards (beyond simply those concerned with fraud control) need to be devised for local welfare and related institutions so that those with greater problems can be focussed on through technical assistance, incentives and/or sanctions. Once performance standards and indicators are in place, it should be possible as well to determine the characteristics of effective agencies and to provide some "best practice" suggestions. If, for example, one of the characteristics of those local welfare agencies that perform well is close co-operation with local job training, economic development and other community agencies, attention could be focussed on encouraging greater cooperation at the local level.

Publicizing rankings of performance results, appropriately adjusted for external factors over which the local agency have no control, could be a catalyst for inducing "poor performers" to alter their approach in an effort to improve performance.

Implications for program design

Given differences among areas, it may well be that the appropriateness of services or activities will vary from area to area. A labor force attachment approach emphasizing job search and related services may be more appropriate in tight labor markets and in labor markets with a growing number of low-skilled jobs than in an area where there is high unemployment and few low-skilled jobs available. A human capital development approach may be more appropriate where few low-skilled jobs are available, but there are jobs available requiring modest skill-levels. In areas where a spatial mismatch is particularly severe, transportation related and job network services may be necessary. This suggests an enhanced federal role in sorting out the impact of different varieties of program design under varying circumstances and disseminating these results to state and local governments.

Indeed, the variation in local conditions and the impact of this variation on outcome suggests that substantial scope should be given to locally devised approaches to moving people from welfare to work. This is consistent with the idea of performance partnerships where program objectives are agreed between federal and state and local governments within the context of national policy goals (moving people from welfare to work), benchmarks are established against which performance can be measured, and state and local governments are given substantial discretion in devising approaches to meet the objectives.

Implications for welfare and employment policy

As the above discussion makes clear, it will be much more difficult for welfare recipients to find employment in some areas than in others. There are, as has been noted, a variety of reasons for this. Yet, in some areas, it is difficult to escape the conclusion that the overriding reason is that there is simply insufficient demand in the local labor market for people who have the characteristics of welfare recipients. There simply are not enough low-skilled jobs out there. If this is the case, there may be some areas where any strategy to move large numbers of welfare recipients to employment will require some demand side response such as wage subsidies to private employers or creation of community service jobs. In some areas even this strategy will have to be buttressed by improvement in local support services. And, given the characteristics of welfare recipients and the minimal requirements of any job, it may well be that there is some portion of welfare recipients who will simply be unemployable under any circumstances, even when low skill jobs and local support services are readily available.

Implications for research and evaluation

First, evaluation efforts of welfare to work programs must take into account place differences. Any effort, for example, to compare the results of program design A in place X to program design B in place Y will be confounded if differences in the two places (particularly in local labor market conditions) are not taken into account - even if each of these is, in itself, a well-designed experiment with control groups etc. That is, it will be impossible to determine whether any differences in results are due to differences in the program design or to differences in the two areas in which the program is taking place. Similarly differing results for similar program designs in several areas may reflect differences in the areas themselves. This suggests that one goal of research and evaluation ought to be the identification of area characteristics for which program designs are more or less appropriate.


References

Albert, Vicky N. (1988), Welfare Dependency and Welfare Policy, New York: Greenwood Press.

Bane, Mary Jo and Ellwood, David (1983), The Dynamics of Dependence: The Routes to Self-Sufficiency, Report prepared for DHHS by Urban Systems Research and Engineering, Cambridge, Mass.

Bartik, Tim (forthcoming, 1995), "The Distributional Effects of Local Labor Market Demand and Industrial Mix: Estimates Using Individual Panel Data," Journal of Urban Economics.

Ellwood, David (1986), "The Spatial Mismatch Hypothesis: Are there Teenage Jobs Missing in the Ghetto?" in Richard Freeman and Harry Holzer (eds.), The Black Youth Employment Crisis, Chicago: University of Chicago Press.

Fitzgerald, John (1995), "Local Labor Markets and Local Area Effects on Welfare Duration," Journal of Policy Analysis and Management, vol. 14, no. 1, 43-67.

Fix, Michael; Galster, George; and Struyk, Raymond (1993), "An Overview of Auditing for Discrimination," in Michael Fix and Raymond Struyk (eds.), Clear and Convincing Evidence, Washington D.C.: The Urban Institute Press.

Freeman, Richard (1991), "Employment and Earnings of Disadvantaged Young Men in a Labor Shortage Economy," in Christopher Jencks and Paul Peterson (eds.), The Urban Underclass, Washington D.C.: The Brookings Institution.

Freeman, Richard and Wise, David (1982), "The Youth Labor Market Problem: Its Nature, Causes and Consequences," in Richard Freeman and David Wise (eds.), The Youth Labor Market, Chicago: University of Chicago Press.

Gueron, Judith and Pauly, Edward (1991), From Welfare to Work, New York: Russell Sage.

Harris, Kathleen (1993), "Work and Welfare among Single Mothers in Poverty," American Journal of Sociology, Vol. 99, no. 2, 317-352.

Hutchens, Robert (1981), "Entry and Exit Transitions in a Government Transfer Program: The Case of Aid to Families with Dependent Children," Journal of Human Resources, Vol. 16, no. 2, 217-237.

Ihlandfeldt, Keith and Sjoquist, David (1991), "The Effect of Job Access on Black and White Youth Employment: A Cross-sectional Analysis," Urban Studies, vol. 28, no. 2, 255-266.

Ihlanfeldt, Keith and Sjoquist, David (1990), "Job Accessibility and Racial Differences in Youth Employment Rates," American Economic Review , Vol. 80, no. 1, 267-275.

Kasarda, John (1995), "Industrial Restructuring and the Changing Location of Jobs," in Reynolds Farley (ed.), State of the Union: Vol. 1. Economic Trends, New York: Russell Sage.

Kasarda, John (1993), "Inner City Concentrated Poverty and Neighborhood Distress: 1970 to 1990," Housing Policy Debate, vol. 4, no. 3.

Massey, Douglas and Denton, Nancy (1993), American Apartheid: Segregation and the Making of the Underclass, Cambridge, Mass.: Harvard University Press.

Mitchell, John; Chadwin, Mark; and Nightingale, Demetra (1979), Implementing Welfare-Employment Programs: An Institutional Analysis of the Work Incentive (WIN) Program, Report prepared for the Department of Labor by the Urban Institute.

O'Neill, June; Wolf, Douglas; Bassi, Lauri; and Hannan, Michael, (1984) An Analysis of Time on Welfare, Report prepared for DHHS by the Urban Institute. Washington D.C.

Plotnick, Robert (1983), "Turnover in the AFDC Population: An Event History Analysis," Journal of Human Resources, Vol. 18, no. 1, 65-81.

Popkin, Susan; Rosenbaum, James; and Meaden, Patricia (1993), "Labor Market Experience of Low-Income Black Women in Middle-Class Suburbs: Evidence from a Survey of Gautreaux Program Participants," Journal of Policy Analysis and Management, vol. 12, no. 3.

Vartinian, Thomas (1995a), Location Effects on AFDC Spells: Examining Local Labor Markets and the Spatial Mismatch Hypothesis, unpublished manuscript, Dept. of Economics, Bryn Mawr College.

Vartinian, Thomas (1995b), Neighborhood Effects on AFDC Spells: Examining the Social Isolation, Relative Deprivation, and Epidemic Theories, unpublished manuscript, Department of Economics, Bryn Mawr College.

West, Richard; Williamson, Cynthia; and Kreutzer, Suzanne (1993), Recommended Adjustment Models for PY 94 JTPA Performance Standards for Titles II-A, II-C and III, Report prepared for the Office of Strategic Planning and Policy Development, Employment and Training Administration, U.S. Department of Labor by Social Policy Research Associates, Menlo Park, Cal.

Wilson, William J., The Truly Disadvantaged, Chicago: University of Chicago Press.

Zill, Nicholas; Moore, Kristin; Nord, Christine; and Steif, Thomas (1991), Welfare Mothers as Potential Employees: A Statistical Profile Based on National Survey Data, Washington D.C.: Child Trends, Inc.