The Administrative Geography Of Devolving Social Welfare Programs


On August 22, 1996, President Clinton signed the Personal Responsibility and Work Opportunity Act into law. The act abolished Aid to Families with Dependent Children (AFDC), the federal government’s main cash assistance program for low-income households. Two key features of the new welfare law are time limits on aid (two years consecutive, five years lifetime) and all manner of work requirements. States and cities are responsible for interpreting and administering the new welfare law, subject to certain federal regulations and sanctions for noncompliance.

The demise of AFDC was part of a broader shift toward devolving federal responsibility for welfare, housing, job training, and other anti-poverty programs to the states and cities. In this working paper, Dr. Mark Alan Hughes argues that administrative geography–how jurisdictional and territorial boundaries organize the eligibility, allocation, and delivery of program service–is crucial to understanding and anticipating the implementation challenges posed by the devolution of federal social welfare programs. Hughes is director of the Community Policy Research Center and vice president for policy development at Public/Private Ventures in Philadelphia. He stresses that Public Housing Authorities and the other state and local agencies responsible for implementing new welfare and related programs are up against a huge problem of labor market fragmentation. A map is worth a thousand words. He illustrates the challenges to social welfare devolution with maps of the administrative geography of the nation’s ten largest urbanized areas. He examines the administrative geographies of housing, employment, and welfare policy.

To help avoid turning the federal social safety net into a safety maze in which low-income citizens can find neither public aid nor private-sector jobs, he recommends that “Federal and state policy rnakers should consider metropolitan-wide administration of welfare-to-work programs.”

The Center for Public Management and the Center on Urban and Metropolitan Policy are pleased to circulate this timely and provocative analysis of the often neglected but clearly important spatial dimensions of social welfare policy implementation.

John J. DiIulio, Jr.
Douglas Dillon Non-Resident Senior Fellow
Center for Public Management
Brookings Institution
Bruce J. Katz
Center on Urban and Metropolitan Policy
Brookings Institution

About The Centers

The Brookings Center for Public Management, a part of the Governmental Studies Program, was founded in January 1994. By bringing together scholars, experts in policy and management, and practitioners to discuss policy issues and administration, CPM speaks with a strong voice in the national conversation about the proper form and function of our government.

The Center for Public Management is dedicated to improving the public service and the processes by which democratically-enacted public policies are implemented. The Center supports three main sets of activities: (1) the design, execution, and timely publication of public management research on cutting-edge problems of federal, state, and local governance; (2) civic education and outreach efforts (symposia, reports, op-eds) intended to keep policyrnakers, public administrators, journalists, and others informed about the crucial public management dimensions of public policy and governance; and (3) leadership and management training programs for selected government executives and managers.

In December 1996, with initial support from the Fannie Mae Foundation, the Brookings Institution established the Center on Urban and Metropolitan Policy in the Economic Studies Program. In partnership with academics, private and public sector leaders and locally-elected officials, the new Urban Center strives to change the way decision makers think about urban policy and how disparate domestic policies — together — impact cities and metropolitan regions.

The Urban Center conducts independent and objective research on the experience of places — urban neighborhoods, cities, and metropolitan areas. Specifically, the Urban Center: (1) develops timely, independent and provocative analyses of current government policies; (2) works to shape the direction of future policies; (3) examines economic, political, fiscal, social, and demographic trends that have both immediate and longer term implications for cities and their regions; and (4) places special emphasis on the unique challenges facing the Nation’s capital. In all these approaches, the Urban Center aims to identify and help set in motion integrated approaches and practical solutions that are feasible for national, state, and local implementation. The Center will also try to be an informative source for local, state and federal leaders, the media and other interested parties.

This Joint Occasional Paper of the two Centers, the first in a series of expected collaborations between them, is preliminary material circulated to stimulate discussion and critical comment. Papers in this series have not been exposed to the regular Brookings pre- publication review and editorial process.

The Where of Devolution

Two features of private labor markets are especially relevant to the devolution of social welfare policy. First, labor markets have become metropolitan-wide. Labor markets are organized into the places where people reside, the places where people work, and the commutes that connect these places. (1) Today, more Americans work outside than inside their place of residence, and the fraction doing so increased by 13 percent between 1980 and 1990. Indeed, about one-fourth of all working Americans cross county lines in their journey to work, and this fraction is also increasing. (2) Today’s local labor markets are metropolitan systems in which jobs and people are distributed widely and commutes, especially the now typical suburb-to-suburb commute, are going in all directions.

Second, job opportunities are unevenly distributed within metropolitan areas. “Spatial mismatch” occurs when barriers arise between places of residence (of labor supply) and places of employment (of labor demand) that impede the matching of, in the typical case, suburban employers with entry-level openings and inner-city residents with low skills. (3)

Figure A displays the basic facts of spatial mismatch as it exists between the cities and suburbs of the Nation’s ten largest metropolitan areas. The figure compares the concentration of unemployment in large cites with the deconcentration of job growth toward the surrounding suburbs. The lighter bars (generally, to the right of the vertical axis) measure urban unemployment relative to suburban unemployment, and the darker bars (always to the left of the vertical axis) measure urban job growth relative to suburban job growth. (4) Thus, metropolitan Houston’s urban unemployment rate in 1993 was about one and one-half times (150%) that of its suburban rate; and its rate of 1985-1991 urban job growth was about one- half (50 %) that of its rate of suburban job growth.

The vertical axis (100%) marks the point at which city and suburban rates would be equal. In nine out of ten cases, city unemployment rates were higher than suburban unemployment rates; and, in all ten cases, city job growth rates during a preceding six-year period were lower than suburban job growth rates. Broadly, the combined length of the two bars describes the spatial mismatch in all ten of these metropolitan areas. (5)

Of course, more is going on here than just spatial mismatch and the inability of supply and demand to meet. But most researchers now agree that the suburbanization of employment has negatively impacted the job prospects of disadvantaged job-seekers, who are overwhelmingly concentrated within central cities. (6) As jobs have moved from around the comer to over the horizon, distance has joined the other labor market difficulties facing the disadvantaged: lack of skills and skills-imparting institutions, discrimination based on race and residence, exclusion from the informal networks employers use to hire new workers, and the falling wages and eroding security of non-professional jobs.

Thirty years of social research on this topic have focused on distance and/or commuting time as the barrier created by spatial mismatch: it is difficult or costly for inner- city residents to commute to suburban jobs. Though most researchers have argued that this distance barrier is also a proxy for other, perhaps more important, barriers (especially the information flow about job openings), none have examined in detail the administrative geography of Federal assistance programs. These programs represent the formal networks by which the disadvantaged seek connections to employers. The horizon line over which job opportunities have shifted from cities to suburbs, defines not just a distance barrier standing between the disadvantaged and jobs but also a set of jurisdictional barriers that potentially turn the safety net into a safety maze.

How fragmented are metropolitan labor markets by the administrative geographies of major Federal housing, job training and placement, and welfare programs? If Federal assistance is oriented to employment, then understanding how major assistance programs interact with metropolitan labor markets becomes an important agenda for public management scholars. Understanding administrative geography–the where of social welfare devolution–is a first step.

Since the New Deal and Great Society eras, the Federal government has established approximately 80 assistance programs targeted to low-income individuals and families. (7) Federal expenditures in FY1993 for these programs totaled $223 billion. In this paper I examine the administrative geographies that govern some of the largest of these 80 programs: Low-Rent Public Housing, Section 8 Housing Assistance, Job Training Partnership Act, and Aid to Families with Dependent Children and its component Job Opportunities and Basic Skills program. Together these assistance programs accounted for $35.4 billion in FY1993 Federal expenditures. They represent the Federal assistance programs most directly oriented to connecting the disadvantaged to employment.

Administrative Geography: territorial boundaries that organize the eligibility, allocation, and delivery of program services

By “administrative geographies,” I refer to territorial boundaries that organize the eligibility, allocation, and delivery of program services. In theory, administrative geography could refer to a number of scales. In this paper, I focus on the sub-State scale. This is the scale of organizations and agencies vested with much of the responsibility for actually running the programs: Public Housing Authorities, county welfare agencies, and so on. Therefore, I focus on the number of PHAs in a metropolitan area, and not on the number of offices maintained by each PHA within its territorial jurisdiction. While the latter may relate in interesting ways to the quality and accessibility of service delivery within a PHA, it is the former issue that relates to our interest in fragmentation.

Fragmentation: the division of metropolitan labor markets by administrative geography in ways that may weaken assistance program outcomes.

By “fragmentation,” I refer to the extent to which an administrative geography divides the metropolitan labor market in ways that may weaken assistance programs as a result of spatial mismatch. When metropolitan labor markets are so clearly divided into job- rich and job-poor parts, the question becomes: to what extent might administrative geographies aggravate rather than mitigate that geography of labor market opportunity. Thus fragmentation has two components: (1) regulations and/or laws governing Federal assistance programs that interact with jurisdictional boundaries and that may limit metropolitan-wide access to employment opportunities and (2) the number of jurisdictions administering these programs in a given metropolitan labor market.


I map three administrative geographies in the ten largest metropolitan areas. These ten metropolitan areas are listed in Table 1. For each area, I have identified the relevant administrative geography for three major Federal assistance programs for the disadvantaged: HUD’s Public and Assisted Housing, DOL’s employment and training under JTPA, and HHS’s Job Opportunities and Basic Skills (JOBS) under AFDC. Together, JOBS and JTPA provided 60 percent of the Federal employment and training funds for the disadvantaged. (8) The ten largest Consolidated Metropolitan Statistical Areas (CMSAs) together contained about one-third (32.3 percent) of the total U.S. population in 1990. CMSAs are comprised of counties and are based on social and economic integration–specifically, commuting. They are a good measure of a local labor market. However, being comprised of whole counties, CMSAs tend to include large portions of outlying areas that would commonly be considered rural rather than urban or developed.

To better contain our analysis within a more recognizably urban boundary and to allow us to use density and land area measures in a more defined way, I base. my analysis on the Urbanized Area definition. These are comprised of municipal and township boundaries, allowing for a finer grain in the boundary. Urbanized areas represent the physical settlement–the built-up area that exceeds a certain level of density: people per unit of area. I have constructed the administrative geographies for each urbanized area. This allows me to exclude as less relevant to my analysis a distant county seat that is located in a rural hinterland far from the metropolitan edge, but that has, say, its own Public Housing Authority. The ten largest urbanized areas are still very large and together contained about one-quarter (23.3 percent) of the total U.S. 1990 population.

Table 1 displays the ten areas used in this report. I list the population, land area, and density for the CMSA, the largest urbanized area therein (a CMSA typically has more than one urbanized area), and the largest city in the urbanized area. Figures 1-10 display the ten largest urbanized areas situated within their respective CMSAs. In nearly every case, there is an outer ring of counties in the CMSA that I exclude from my detailed examination of administrative geographies; thus bounding the analysis to program administrative units most relevant to the metropolitan employment prospects of central city residents.



Housing mobility from high-poverty inner cities to low-poverty suburbs is one strategy for enhancing access to suburban jobs: residential proximity to job-rich locations should reduce informational and physical barriers. Research into the impacts of Chicago’s Gautreaux Program (11) provides encouraging results. One of the advantages of the strategy, of course, is that in attempts to address multiple concerns simultaneously: hypothetically providing access to safer streets and better schools as well as better job prospects.

In this section, I consider the level of fragmentation in the administrative geography of Federal housing assistance relative to metropolitan labor markets. Again, the question here is not about housing assistance per se, nor about efficient or equitable delivery of Public and Assisted Housing services per se. Rather, my concern is with fragmentation and the barriers fragmentation may raise for a strategy of using housing assistance to move toward suburban job opportunities.

Administrative Rules

Public Housing and Section 8 Housing Assistance Payments are by far the largest Federal housing programs for the disadvantaged. In FY1993, Low-Rent Public Housing provided 1.4 million units at a Federal expenditure of $6.2 billion. Section 8 Low-Income Housing Assistance supported 2.8 million units with $11.2 billion in Federal funds. (12) Expenditures on these programs are distributed, via HUD Field Offices, to allocation areas which are based on municipalities, counties, or a group of municipalities or counties (24 CFR 791.102). Each allocation area is intended to be the smallest practical area of sufficient size to support at least one feasible program or project (24 CFR 791.404). An entire central city may be set aside as a separate allocation area (24 CFR 941.101). The goal of the allocation process is the equitable distribution of Federal housing assistance in accordance with the relative housing needs (population, poverty, overcrowding, substandard housing, etc.) in each allocation area with the Field Office jurisdiction (24 CFR 791.402-404).

The functioning body of an allocation area is the Public Housing Authority (PHA), defted as “any State, county, municipality, or other government entity … authorized to engage in or assist in the development or operation of housing for low-income families.” (24 CFR 791.102). PHAs develop and manage Public Housing as well as administer Section 8 Housing Assistance Payments.

Public Housing. As HUD and Federal policy generally moves away from the construction of traditional Public Housing developments, the main issue is access to a metropolitan area’s existing stock of Public Housing Units. (13) On this topic, the salient administrative rules are those governing waiting lists.

Waiting lists interact with administrative geography in two ways:
they are unconsolidated
they are preferential

Each PHA is required to maintain its own waiting list for the stock of public housing units located within its geographical jurisdiction (24 CFR 882.513). A family must fill out a separate application for each PHA in which it seeks a unit; this is the meaning of unconsolidated waiting lists. Thus, metropolitan wide opportunities are complicated directly by the simple number of PHAs in a metropolitan area. Currently, it is at the discretion of the local PHA to take applications by mail. Thus a family may have to travel to the PHA office to acquire an application, to get answers to any questions regarding the application (there is no required standard application), and to submit the application for processing. And this must be done for every PHA in which the family seeks a housing opportunity. Thus, both the number and the location of the PHAs and their offices matter.

In addition to managing its “own” waiting list, each PHA is allowed to grant preferences in the order in which families are drawn from the list and placed in housing units (24 CFR 882.209). PHAs are allowed to give priority to residents of their own jurisdiction, although they may not weight this priority by length of residence. In an employment-access- relevant rule, HUD requires that applicants who work in a PHA jurisdiction must be prioritized as a resident of the jurisdiction. Two factors, however, limit the degree to which this rule actually promotes employment access. First, it requires the equality of preference only to applicants who have already found a job in the PHA’s territory. This limits the utility of housing mobility as an instrument for finding a job. Second, it limits the waiting-list equality only to that particular PHA. The more fragmented a metropolitan labor market, the more likely that a suburban housing search by an applicant employed within the jurisdiction of one suburban PHA will land the applicant in another suburban PHA–one in which his or her work status will not be relevant.

Section 8. Housing Assistance Payments provide a far more portable alternative than Public Housing. Participating households receive subsidies they can apply to private market rental housing. The participants are responsible for finding this housing and are free to do so in “any area where the PHA determines that it is not legally barred from entering into contracts” (24 CFR 882.103). PHAs are encouraged to promote choice of housing opportunities by cooperating with other PHAs, and HUD gives preference to PHAs which provide families with the broadest geographical choice of units (24 CFR 882.103).

Under current practice, however, participants have somewhat limited access to housing in PHAs outside their current one. There are two forms of Section 8 Housing Assistance: certificates and vouchers. With the less portable certificates, PHAs may make arrangements between themselves to accept each others housing certificates. (24 CFR 882.209) If an agreement does not exist, however, then a participating family that moves out of its original PHA jurisdiction retains its assistance only if the destination PHA admits the family to the latter’s Section 8 program.

The need for inter-PHA agreements may encumber the portability of Section 8 assistance

Under the voucher program, any PHA must admit a family with a voucher from another PHA as if the family were part of its “own” program. The “sending” PHA must reimburse the “receiving” PHA for the full cost of the voucher and for 80 percent of the administrative costs. The “sending” PHA, however, may in some cases deny the request to move (24 CFR 887.563). Many cities have designed support services to help families move from high-poverty to low-poverty areas, though these moves are typically within a single PHA.

In general, however, Federal public and assisted housing policy is cognizant of the problems created by fragmentation and has moved toward policy instruments that are less constrained by administrative geography (for example, vouchers versus certificates). We note that many of these administrative rules would be even further modified in HUD’s reinvention plan. Under its proposed tenant-based approach, Public Housing and Section 8 Housing Certificates would both operate more like Housing Vouchers. A standardized application would replace the multitude of PHA-specific applications. Once an application has been accepted into the program, a family would be able to use its assistance in any jurisdiction. It appears that a family would still send an application to each PHA to which it hopes to move and that there will be separate waiting lists. Preferences based on residence, however, would be completely eliminated and PHAs would be required to accept applications by mail. (14)

Administrative Geography

Fragmentation is the extent to which an administrative geography divides the metropolitan labor market in ways that may negatively impact program outcomes in light of spatial mismatch. Fragmentation has two components: boundaries and rules that interact with boundaries. I identified two housing assistance rules that may interact with boundaries in negative ways: waiting lists and required inter-PHA agreements in the preceding section, and we now turn to the boundaries.

Table 2 summarizes the administrative geography of Federal housing assistance to the disadvantaged in the ten largest urbanized areas. For context, the table displays the number of housing units and the land area of each urbanized area. The last three columns present three broad measures of fragmentation: (1) the total number of PHAs in the urbanized area, (2) the number of PHAs in the urbanized area per 1,000,000 housing units, and (3) the average land area of PHAs outside the large cities. (15) The urbanized areas are sorted in descending order of fragmentation as measured by the last measure, average size of suburban PHAs.

The strong form of fragmentation takes the urbanized area as a single metropolitan labor market and points to a single PHA as the least fragmented state. The degree to which the total number of PHAs in the urbanized area exceeds one, therefore, is the degree of fragmentation, using this measure. By this standard, all the urbanized areas are clearly fragmented, with the possible exception of Houston. The total number of PHAs represents the number of applications, waiting lists, and parties to inter-PHA agreements needed to maximize a participant’s access to employment throughout the metropolitan labor market.

We can modify the fragmentation standard by controlling for the size of the urbanized area–relaxing the assumption that the entire local labor market, regardless of size, should be accessible via Federal housing assistance. The next two measures provide some such control of size. These relative measures lack the standard of zero fragmentation found in the first measure (how many PHAs per 1,000,000 housing units should there be? how large should a suburban PHA be?), but are useful for identifying more and less fragmentation across urbanized areas. These measures also suggest the relative likelihood that households might live and work in different PHA jurisdictions–thus negating any waiting-list preferences based on working in a PHA territory.


In addition to the arithmetic dimension of fragmentation, I examine the spatial dimension in the following series of maps. This spatial dimension can be important because, for example, suburban PHAs are not each of the average size listed in the table nor are they scattered evenly throughout the metropolitan areas displayed in Figures 1-10. Knowing the fragmentation’s form as well as its extent helps to frame future research questions and policy issues.

Figures 11-20 display the actual boundaries of all PHAs serving the ten urbanized areas in our study. The devil is in the details, and these maps are useful to analysts considering more specific questions than are possible here. In general (always conceptual “thin ice” in geography), the maps reveal a pattern of central-city PHAs with a dense ring of adjacent suburban municipal PHAs and a surrounding layer of suburban county PHAs. Boston is the extreme form of this fragmentation with three times as many PHAs per capita as the next urbanized area, Detroit; and with an average suburban PHA less than half the size of the next urbanized area, New York. In every case outside Texas, the suburban housing market is divided into areas averaging less than 100 square miles–much smaller than the average area of the central cites in these ten areas (the median two central cities in terms of size of Detroit at 139 square miles and Chicago at 227).

There appears to be a significant level of fragmentation in the administrative geography of Federal housing assistance in these ten urbanized areas. Seen in terms of separate applications, preferential waiting lists, required inter-agency agreements, and so on, these maps suggest the extent to which a PHA geography is imposed on metropolitan housing markets and, in particular, on the suburban employment opportunities policy analysts hope to leverage through housing.


Enhanced opportunity for work is the foundation of public assistance to the non-elderly disadvantaged, and appropriate skills are a key component of work readiness. The Job Training and Partnership Act of 1982 (JTPA) was established to improve training and placement for poor and dislocated workers. (16) Title IIA, targeted on disadvantaged adults, is the largest single program under JTPA. It received about $1 billion in FY1993 and served about 350,000 average monthly recipients. (17) In order to qualify, a person’s family income must fall below a specified poverty level or the family must be receiving food stamps or Aid to Families with Dependent Children (AFDC) (20 CFR 628.605).

Administrative Rules

The Department of Labor allocates funding to each State by formula and the States in turn allocate money to local territories known as Service Delivery Areas (SDAs). SDAs must contain all of a State or one or more units of local government. The boundaries of these SDAs are determined by the Governor and must be consistent with labor market areas, with metropolitan statistical areas, or with areas in which related services are provided under state or federal programs. If requested, however, SDA status must be granted to any unit of general local government with a population of at least 200,000 or to any consortium of local governments with a population of at least 200,000.

This provision of the Act, related to the politics of intergovernmental relations and, in particular, to alliances between large cities and the Federal government, has worked to multiply the number of SDAs in metropolitan labor markets. For example, when JTPA was established the Atlanta business community and the Governor of Georgia proposed a seven- county metropolitan SDA to administer funds and contract training programs for Atlanta and its suburban periphery. However, elected officials in the City of Atlanta and in several counties and county consortia exercised their right under the Act to form SDAs. Consequently, metropolitan Atlanta has five SDAs. (18)

There are approximately 650 SDAs throughout the U.S. SDAs are governed by a partnership between the local government(s) and a Private Industry Council (PIC) representing private businesses and other local interests. JTPA is designed to give SDAs and their respective PICs much autonomy in designing local programs and in contracting with local organizations to provide training, basic skills, job search assistance, and so on.

One of the major features of the JTPA system is the set of performance standards applied to SDAs and designed to promote financial accountability and adherence to Federal goals (20 CFR 627.470). Under these standards, SDA performance goals are based on factors such as the percentage of participants placed in a job, their increase in earnings, the program costs per placement. In recognition of variations in local labor market conditions, performance standards are adjusted for certain economic and demographic characteristics of the SDA.

Although there are broad process goals for JTPA providers (e.g., at least 50 percent of expenditures must be used for education and training), these performance standards are outcome-based. These outcomes are measured for participants who reside within the territory of the SDA. This sounds innocuous enough. Indeed, it sounds like nothing more than local administration. Recall, however, spatial mismatch. Suburban SDAs have no program incentives to place participants from outside their administrative area–namely, residents of central cities or counties–into “their” jobs.

The need for inter-PHA agreements may encumber the portability of Section 8 assistance.

Through the Private Industry Council (PIC) mechanism, SDA are presumed to have employer connections regarding training requirements and placement openings. By the same token, since these PICs are organized according to administrative areas smaller than the metropolitan labor market, they under-bound the private-sector connections of JTPA programs and, in the case of central city and country SDAs, limit these connections and program incentives to the very weakest part of the metropolitan labor market. Local administrative units are likely to create localized information flows and localized program- outcome incentives. For the residents of central SDAs this dynamic may isolate them from opportunities shifting to the suburbs.

How many SDAs exist in the ten largest urbanized areas? Table 3 displays the administrative geography of JTPA in these ten areas. The first two columns display the size of the labor force in 1990 and the land area of the urbanized area. The next three columns show three measures of fragmentation: the number of SDAs, the number of SDAs per one million workers, and the average size of SDAs outside those serving the urbanized areas’s largest cities. Here, perhaps more than with regard to PHAs, we might take a single SDA has the standard for reasonable standard for administering JTPA. No urbanized area comes close to that standard–though once again Houston is the least fragmented by any measure.

New York and Los Angeles have the largest number of SDAs. Unlike with PHAs, however, New York appears significantly less fragmented when we control for size–the area has the most SDAs but is in the middle of the distribution of ten urbanized areas in terms of average SDA size per workers and per land area. Figures 21-30 display the boundaries of the SDAs serving the 10 urbanized areas in our study.



Administrative Rules

Welfare reform under the Personal Responsibility and Work Opportunity Act of 1996 PRWOA) seeks to transform the goal of public assistance from income maintenance to sustainable employment placement. While PRWOA shifts nearly all responsibility for program design from the federal government to the states, the implementation of this new world of welfare will be attempted through the existing apparatus of county welfare offices. If we are serious about connecting welfare families with private-sector work, however, then we must recognize that job opportunities vary widely across counties.

Of more than 3,000 counties nationwide, there are about 160 large ones containing 125,000 or more jobs. These contain about two-thirds of the nation’s jobs and a larger fraction of the welfare caseload, and so they’re a natural place to look for welfare-to-work solutions. But upon examination, these counties are by no means uniformly prosperous. Since 1990, about 60 of these large counties lost a net of 1. 6 million jobs and about 90 gained a net of 2.2 million jobs. In general, job growth is occurring in suburban and/or Sunbelt counties while job loss is taking place in central and/or Frostbelt counties.

A recent GAO survey of county welfare administrators examined the nature and extent of their focus on employers, job development, and placement issues [GAO 1995]. Some of the survey results speak directly to issues of spatial mismatch. For example, about 45 percent of administrators reported a lack of transportation as a major reason that job- ready participants do not become employed (another 30 percent cited transportation as a moderately important reason). But more relevant to administrative geography and fragmentation, both the GAO analysts and the program administrators discussed the conditions and prospects of county labor markets. This bounding is not surprising since program jurisdictions are drawn along county lines and since funding levels and streams are organized by those jurisdictions. But, given the unevenness of job growth, this perspective can dramatically under-bound the opportunities available to central county administrators and participants.

Administrative geography can fragment the metropolitan labor markets in more formal ways when it comes to JTPA. JTPA programs are the primary means by which Temporary Assistance for Needy Families (TANF) participants in many counties will likely receive employment-related services [GAO 1994]. Although there are broad process goals for JTPA providers (e.g., at least 50 percent of expenditures must be used for education and training), these performance standards are outcome-based. These outcomes are measured for participants who reside within the territory of the SDA. Therefore, suburban SDAs have no program incentives to place participants from outside their administrative area–namely, residents of central cities or counties–because such placements would not contribute to their performance goals and not add to their funding.

On the other side of the coin, central city and county SDAs have limited means to develop jobs in suburban areas. Through the Private Industry Council mechanism, SDAs are presumed to have employer connections regarding training requirements and placement openings. By the same token, since these PICs are organized according to administrative areas smaller than the metropolitan labor market, they under-bound the private-sector connections of JTPA programs and, in the case of central city and country SDAs, limit these connections and program incentives to the very weakest part of the metropolitan labor market. Figures 31-40 display the county welfare agencies serving the 10 urbanized areas in our study.



The new federal laws governing welfare-to-work retain the administrative geography of county agencies and the performance-standards approach that is driven by what happens within these local jurisdictions. This fragmentation completely neglects the metropolitan reality of today’s larger labor markets. Given the unevenness of caseloads and job growth, new approaches are needed to ensure that city and county boundaries become as irrelevant to the employment prospects of the welfare poor as they are now to the middle class. In sum, Federal and state policymakers should consider metropolitan-wide administration of welfare-to-work programs.

If structural consolidation proves politically difficult or time consuming, then policymakers should consider functional consolidation so that existing county agencies have powerful incentives to work together as parts of a single metropolitan labor market. This might mean that central counties should be oriented to supply-side needs and suburban counties be oriented to serving more demand-side issues. But in principle, the idea would be to allow counties and other localities to play varying rather than fixed functional roles in a labor exchange and employment transition process that now must occur at a metropolitan scale. Without a move toward metropolitan-wide administration, devolution will indeed turn the safety net into a safety maze.

Spatial Mismatch in the 10 Largest Metropolitan Areas

Figures 1-40

Boston Figure 1 Figure 11 Figure 21 Figure 31
Chicago Figure 2 Figure 12 Figure 22 Figure 32
Dallas Figure 3 Figure 13 Figure 23 Figure 33
Detroit Figure 4 Figure 14 Figure 24 Figure 34
Houston Figure 5 Figure 15 Figure 25 Figure 35
Los Angeles Figure 6 Figure 16 Figure 26 Figure 36
New York Figure 7 Figure 17 Figure 27 Figure 37
Philadelphia Figure 8 Figure 18 Figure 28 Figure 38
San Francisco Figure 9 Figure 19 Figure 29 Figure 39
Washington, D.C. Figure 10 Figure 20 Figure 30 Figure 40


1. There is admittedly a growing fraction of the population that telecommutes and so on. But, in 1990, 97 percent of the working population still commuted to somewhere outside the home and they spent an average of 22 minutes getting there.

2 It is probably an even higher percentage in large metropolitan areas. For example, 36 percent of workers in metropolitan Philadelphia cross county lines in their daily commutes. All the data in this discussion is drawn from tables based on 1980 and 1990 Censuses of Population as posted by the Journey-to-Work and Migration Statistics Branch ‘ Population Division, U.S. Bureau of the Census at http//

3 There is literature in land use and transportation planning that debates the existence and implications of “jobs/housing imbalance”–a kind of middle class spatial mismatch. But the classical spatial mismatch literature, and certainly my focus here, centers on disadvantaged, minority, central-city residents and their potential employers.

4 It is worth noting some data constraints, though these constraints all minimize, in fact, the dramatic effect of Figure 1. In the unemployment data, “urban” refers to all cities in the CMSA with populations greater than 300,000 in 1990 (which are listed in the Figure), and “suburban” refers to the remainder of the CMSA. This probably minimizes the difference in rates by including in the suburban rate both smaller cities–such as Newark in the New York CMSA–and rural areas in the CMSAs (both of which probably have higher unemployment than real suburbs. In the job growth data, “urban” refers to the entire central county(ies) containing the city(ies) used in the unemployment rates, and “suburban” refers to the remainder of the CMSA. Again, this probably minimizes the difference in rates since central city job growth is probably lower than central county growth. Indeed, those CMSAs in which central cities and central counties are co-terminus (New York, Philadelphia, Washington/Baltimore) display the greatest difference in urban and suburban rates.

5 For more discussion of employment suburbanization than is possible here, see Mark Alan Hughes, “Over the Horizon: Jobs in the Suburbs of Major Metropolitan Areas,” Research Report, Public/Private Ventures, December 1993.

6 For an excellent review of the scientific literature, see Harry J. Holzer, “Black Employment Problems: New Evidence, Old Questions” in the Journal of Policy Analysis and Management (Fall 1994). For the latest empirical research and its broad policy implications, see Harry J. Holzer and Keith R. lhanfeldt, “Spatial Factors and the Employment of Blacks at the Firm Level,” Working Paper #85, Russell Sage Foundation, December 1995.

7 Welfare Programs: Opportunities to Consolidate and Increase Program Efficiencies (GAO/HEHS-95-139, May 31, 1995), pp. 24.

8 JOBS and JTPA: Tracking Spending, Outcomes, and Program Performance (GAO[HEHS-94-177, July 15, 1994).

9 Does not include the Baltimore Urbanized Area.

10 Does not include the San Jose Urbanized Area.

11James E. Rosenbaum, “Changing the Geography of Opportunity by Expanding Residential Choice: Lessons from the Gautreaux Program” in Housing Policy Debate (1995), pp. 231-269.

12 Welfare Programs: Opportunities to Consolidate and Increase Program Efficiencies (GAO/HEHS-95-139, May 31, 1995), p. 34.

13.U.S. Department of Housing and Urban Development, HUD Reinvention: From Blueprint to Action (March 1995). In an era of substantial Public Housing development, and to the extent that such planning and construction does continue, questions regarding the intra-metropolitan development of new stock are also important, but not our focus in this report.

14 HUD, Reinvention, op. cit., p. 7.

15 As listed in Figure 1, these are all cities with populations exceeding 300,000.

16 Although much of JTPA and its later amendments would be subject to revision under the pending Workforce Development Act, most of the fragmentation issues raised here would remain salient under the new Act.

17 Welfare Programs: Opportunities to Consolidate and Increase Program Efficiencies (GAO/HEHS-95-139, May 31, 1995), p. 36

18 U.S. Department of Labor, Practical Guidance for Strengthening Private Industry Councils (199 1), p. 89.