Housing Policy Debate - Amazon Web Services · letter variables should be italicized. Multiple-letter variables and abbreviations (e.g., AGE) and functions (e.g., exp, min, ln) should

Housing PolicyDebate

Contents Volume 7, Issue 2

Microform, issue, and article copies of Housing Policy Debate are available throughUMI, Ann Arbor, MI. Housing Policy Debate is indexed in ABI/INFORM, Accountingand Tax Database, Current Contents®/Social and Behavioral Sciences, EconomicLiterature Index, International Bibliography of the Social Sciences, Public AffairsInformation Service International (PAIS INT), RealSourceTM, Research Alert, SocialSciences Citation Index® (SSCI®), Social SciSearch®, and World Banking Abstracts.

i

Forum

Networks and Nonprofits: Opportunities andChallenges in an Era of Federal Devolution ............................ 201Langley C. Keyes, Alex Schwartz,Avis C. Vidal, and Rachel G. Bratt

Comment .................................................................................... 231David A. Crowe

Comment .................................................................................... 243James R. Follain

Articles

Local Housing Plans: Learning from Great Britain ................ 253David P. Varady

Economic Shifts and the ChangingHomeownership Trajectory ....................................................... 293James W. Hughes

ii

ISSN # 1051–1482

© Fannie Mae Foundation 1996All Rights Reserved

Housing Policy Debate is published quarterly by the Office of HousingResearch, Fannie Mae Foundation, Washington, DC.

The opinions expressed in this publication are those of the authorsand do not necessarily represent the views of the Editor, Fannie MaeFoundation, or its officers.

Where the Homeless Come From: A Study of thePrior Address Distribution of Families Admittedto Public Shelters in New York City and Philadelphia ........... 327Dennis P. Culhane, Chang-Moo Lee, and Susan M. Wachter

Current Issues

Racial Differences in the Search for Housing:Do Whites and Blacks Use the Same Techniques toFind Housing? ............................................................................ 367Reynolds Farley

Joblessness and Poverty in America’s CentralCities: Causes and Policy Prescriptions ................................... 387John D. Kasarda and Kwok-fai Ting

William C. Apgar, Jr.Harvard University

Robert M. BuckleyThe World Bank

Robert W. BurchellRutgers University

Stephen A. BuserOhio State University

Stuart M. ButlerThe Heritage Foundation

Peter ChinloyThe American University

Phillip L. ClayMassachusetts Instituteof Technology

Denise DiPasqualeUniversity of Chicago

Anthony DownsThe Brookings Institution

James R. FollainSyracuse University

Anthony S. FreedmanPowell, Goldstein, Frazer& Murphy

Stuart A. GabrielUniversity of SouthernCalifornia

Jack M. GuttentagUniversity of Pennsylvania

Chester HartmanPoverty & RaceResearch Action Council

Patric HendershottOhio State University

Franklin JamesUniversity of Coloradoat Denver

Wilhelmina A. LeighJoint Center for Politicaland Economic Studies

David ListokinRutgers University

Kenneth G. LoreSwidler & Berlin

Richard F. MuthEmory University

Mary K. NennoThe Urban Institute

Edgar O. OlsenUniversity of Virginia

John M. QuigleyUniversity of Californiaat Berkeley

Kenneth T. RosenUniversity of Californiaat Berkeley

Anthony SaundersNew York University

Morton J. SchussheimCongressional ResearchService

C. F. SirmansUniversity of Connecticut

Michael A. StegmanU.S. Department ofHousing and UrbanDevelopment

George SternliebRutgers University

Raymond J. StruykThe Urban Institute

Kerry D. VandellUniversity of Wisconsin–Madison

Susan M. WachterUniversity of Pennsylvania

Christine WhiteheadLondon School ofEconomics and PoliticalScience

Susan E. WoodwardConsultant

EditorJames H. Carr

Managing EditorSteven P. Hornburg

Publications EditorKathryn S. McLean

Associate EditorsKaren A. Danielsen Robert E. Lang Isaac F. Megbolugbe

Assistant EditorsCarol A. Bell Amy S. Bogdon Patrick A. Simmons

Fannie Mae FoundationHousing Research Advisory Board

Fannie Mae FoundationOffice of Housing Research

The Office of Housing Research (OHR) of the Fannie Mae Foundationis dedicated to research aimed at providing solutions to the nation’sproblems in housing and community development. Using a multi-disciplinary approach, OHR draws on the skills of staff experts and abroad range of outside scholars and professionals for its research.OHR’s research agenda supports the Fannie Mae Foundation’s mis-sion to expand opportunities for decent and affordable housing andimprove the quality of life in communities throughout America.

Housing Policy Debate

Housing Policy Debate is published quarterly. The goal of this journalis to provide insightful discussion and original research on a broadrange of housing and community development issues. Recent articleshave explored affordable housing shortages, housing policies fordistressed urban neighborhoods, new research on homelessness, andrestructuring the Federal Housing Administration.

Housing Policy Debate regular issues are divided into three sections:Forum, Articles, and Current Issues. The Forum section featuresinformative debate between leading experts on timely topics through alead article and responding comment. The Articles section presentspolicy analysis and research, and in Current Issues authors have theopportunity to present their ideas on issues in housing, communitydevelopment, and finance. Articles in the Forum and Articles sec-tions undergo a double-blind review by members of the Fannie MaeFoundation Housing Research Advisory Board and other respectedscholars.

To obtain more information or submit manuscripts for considerationfor publication, please contact Steven P. Hornburg; Managing Editor,Housing Policy Debate; Office of Housing Research, Fannie MaeFoundation; 4000 Wisconsin Avenue, NW; Washington, DC 20016-2800.

Manuscripts should be original, unpublished works not under consid-eration for publication elsewhere. The author is responsible for obtain-ing any necessary rights or permissions to reproduce quoted materialor illustrations published elsewhere. To be considered for publication,manuscripts need not follow the Housing Policy Debate style guideshown on the following pages but if accepted, the manuscript mustconform to these guidelines. Generally, the editorial style of HousingPolicy Debate follows The Chicago Manual of Style.

iii

Editorial Style Guide for Authors

Please submit one copy of your manuscript, double spaced, alongwith an electronic copy on a 3.5-inch diskette in WordPerfect (version5.1 or higher). Tables and figures may be in any type of software, butthey must be accompanied by camera-ready paper copies.

Because manuscripts will undergo a blind review, submit two titlepages, the first showing title of the manuscript; author name, affilia-tion, and telephone number; date of the manuscript; and a second titlepage containing only the title of the paper.

Third-person style is always preferred. If appropriate, authors maymake limited use of first-person singular, but a single author shouldnot refer to himself or herself as “we.”

Biography. The manuscript should include, on a separate page or the“first” title page described above, a sentence listing each author’sname, title, and affiliation.

Acknowledgments. Generally, we do not include acknowledgmentsto the Office of Housing Research (OHR) or Fannie Mae Foundation asresearch sponsors. Acknowledgments to individual staff members areacceptable. Place any acknowledgments in a separate paragraphfollowing the biographical paragraph. Any disclaimers should also bein this paragraph, following the acknowledgments.

Abstract. Include a two-paragraph abstract not exceeding 150 wordsand place it on the first page of the text. In the first paragraph of theabstract, describe the issue(s) or question(s) the paper addresses.In the second paragraph, state the major findings or conclusions.

Keywords. To help users reference OHR’s published research,keywords are included with journal articles. Please suggest threekeywords for your manuscript.

Abbreviations. Be sure that all abbreviations and acronyms aredefined fully and correctly at first mention in the manuscript.

Text Headings. Headings are not numbered and are placed flush left.First-level headings are bold; second-level headings are italic; andthird-level headings are italic with a period that leads directly intotext.

v

Examples:First-level headingSecond-level headingThird-level heading. The text continues....

Tables and Figures. Use arabic numerals to number tables andfigures consecutively in separate series in order of appearance. In-clude a brief descriptive title at the top of each. Tables and figuresshould be in separate electronic files, not integrated into the text.The text must contain a reference to each table or figure. Any abbre-viations in the tables and figures (including NA) must be defined.

If you draw tabular and other material from other sources, be sure toinclude these sources in the references and obtain copyright permis-sion if necessary. Use a short form of the reference in the Source note:name of the author or agency and date.

Equations. Make sure that all symbols in equations are clear andthat all equations (except those in footnotes) are numbered. Single-letter variables should be italicized. Multiple-letter variables andabbreviations (e.g., AGE) and functions (e.g., exp, min, ln) should notbe italicized; neither should numbers, parentheses, or math opera-tions. Vectors and matrices should be in bold (not italicized).

Footnotes. Footnotes are numbered consecutively within each article,using superscript arabic numerals. Footnotes may be used for ex-planatory information but not strictly for references. We do not useendnotes.

References. The manuscript must include complete and accuratecitations of all materials referenced in the manuscript that are not ofyour original authorship. Please double-check your references toensure that names and dates are accurate and that there are no dis-crepancies between the text and the reference list.

Important Guidelines:The reference section should be alphabetical by author and unnum-bered, and placed at the end of the article. Citations must follow theauthor–date system (see the examples below or Chicago Manual ofStyle [CMS]). All (author–date) references in text must be supportedby full references in the reference list. Give authors’ full firstnames (not just initials). (This is a variation on the CMS author–datestyle.) Include page numbers when citing a journal article or bookchapter. When citing a paper presented at a conference, include theplace, date, and conference title or sponsoring organization.

vi

Give sources for tables in as complete a form as possible in the refer-ence list. If sources are pamphlets or looseleaf updates, include themin the references nevertheless. Use initial-cap style for titles (capital-ize the significant words: nouns, adjectives, verbs, and adverbs). (Thisis a variation on CMS author–date style.)

Examples:Case, Karl E. 1986. The Market for Single-Family Homes in Boston.New England Economic Review (May/June):38–48.

Evans, William N., Wallace Eugene Oates, and Robert M. Schwab.1992. Measuring Peer Group Effects: A Study of Teenage Behavior.Journal of Political Economy 100:967–91.

Hadden, Louise, and Mireille Leger. 1988. Codebook for the AnnualHousing Survey Data Base. Cambridge, MA: Abt Associates.

Knickman, James R., and Beth C. Weitzman. 1989. A Study of Home-less Families in New York City: Risk Assessment Models and Strategiesfor Prevention, Final Report. Prepared for the New York City HumanResources Administration by the Health Research Program of NewYork University.

Myers, Dowell. 1981. A Cohort-Based Indicator of Housing Progress.Population Research and Policy Review 1:109–36.

National Council of State Housing Agencies. 1987–1990. Low-IncomeHousing Tax Credit Activity. Annual Compilations. Washington, DC.

Petty, Phillip N., and Judith C. Chaney. 1991. ImplementingFIREEA’s Affordable Housing Provisions. Paper read at the AREUEAMidyear Meeting, May 28–29, Washington, DC.

Pitkin, John R. Housing Consumption of the Elderly: A Cohort Eco-nomic Economic Model. In Housing Demography: Linking Demo-graphic Structure and Housing Markets, ed. Dowell Myers, 174–99.Madison: University of Wisconsin Press.

Resolution Trust Corporation. 1989. Overview Paper. Washington, DC.

U.S. Department of Commerce, Bureau of the Census. 1991.New York City “Housing and Vacancy Survey Questionnaire 1991.”Washington, DC.

Wall Street Journal. July 30, 1991. Some Large Banks Prepare to SellPiles of Foreclosed Commercial Real Estate.

vii

Networks and Nonprofits: Opportunities and Challenges 201Housing Policy Debate • Volume 7, Issue 2 201© Fannie Mae Foundation 1996. All Rights Reserved.

Networks and Nonprofits: Opportunities andChallenges in an Era of Federal Devolution

Langley C. KeyesMassachusetts Institute of Technology

Alex Schwartz and Avis C. VidalNew School for Social Research

Rachel G. BrattTufts University

Abstract

Community development corporations and other nonprofit organizations areincreasingly responsible for producing and managing low-income housing inurban America. This article examines the network of governmental, philan-thropic, educational, and other institutions that channel financial, technical,and political support to nonprofit housing sponsors. We analyze the relation-ships among these institutions and propose an explanation for their success.We then consider challenges the network must confront if the reinvention offederal housing policy is to succeed.

Block grants and rental vouchers, the dominant emphases of federal policy,present opportunities and constraints for nonprofit housing groups and theirinstitutional networks. While states and municipalities are likely to continueto use block grants for nonprofit housing, the viability of this housing will beseverely tested as project-based operating subsidies are replaced by tenant-based vouchers. We recommend ways that the federal, state, and local govern-ments should help the institutional support network respond to this challenge.

Keywords: Low-income housing; Nonprofit sector; Social capital

Introduction

In the 1980s, the federal government drastically reduced directsubsidies for the development of multifamily housing in urbanareas. The 1990s may well see the government stop subsidizingthe development of special needs housing for the elderly and thedisabled. Moreover, the government now proposes to rid itself ofpast commitments to existing federally subsidized and insureddevelopments. The government will only support the develop-ment of new housing through block grants to localities and statesand, perhaps, the low-income housing tax credit (LIHTC). Even

202 Langley C. Keyes, Alex Schwartz, Avis C. Vidal, and Rachel G. Bratt

more than in the 1980s, federal housing policy will emphasizetenant-based rental vouchers.

Federal housing priorities have shifted from categorical, supply-side programs to block grants and vouchers. At the same time,many cities and states are experiencing a growing shortage oflow-income housing (Joint Center for Housing Studies 1995).Because of these two changes and increased political pressurefrom community organizations, homeless advocates, and othergroups (Goetz 1993), local and state governments are becomingmore involved than ever before in the production, rehabilitation,and preservation of low- and moderate-income housing (Berenyi1989; Goetz 1993; Nenno 1991; Terner and Cook 1990). Agenciesuse block grants and other funding sources to support a broadrange of housing activities. Local and state governments oftensupport the development of low-income housing (especially ininner-city communities) through partnerships with communitydevelopment corporations (CDCs) and other nonprofitorganizations.

This article examines the capacity of nonprofits to produce andmaintain low-income housing in this period of devolution andvouchers.1 This capacity is shaped not just by the competency ofeach individual nonprofit group but by the strength of thenonprofit’s institutional network. Nonprofit housing organiza-tions do not exist in an institutional vacuum. They survive andprosper when they are part of a network of organizations thatsupport and undergird their initiatives. Thus, in consideringtheir present and potential role, we must address the extent towhich nonprofit producers are the point organizations in theirsupporting network of institutions.

In analyzing the institutional support system for nonprofithousing, we draw on the concept of social capital. Sociologists,political scientists, and other social scientists define “socialcapital” as the ability of individuals and organizations to acquireresources through membership in networks and other socialstructures (Coleman 1990; Portes and Sensenbrenner 1993;Powell 1990; Putnam 1993, 1995). The concept points to the waysocial and institutional relationships confer tangible benefits onparticipants, benefits that would not otherwise be readily

1 We are not arguing that nonprofit sponsors are the only or necessarily thebest sponsors of such housing but rather that, at present, they are majorplayers that are viewed by many city governments, foundations, and elementsof the U.S. Department of Housing and Urban Development as being atthe center of any serious effort to revitalize and maintain inner-cityneighborhoods.

Networks and Nonprofits: Opportunities and Challenges 203

available. Most definitions emphasize the role of mutual trust,reciprocity, and a shared sense of identity in forging the socialand institutional relationships that foster social capital. Al-though the notion of social capital is now invoked to explain anever-widening array of phenomena—perhaps to the point ofoveruse—we nevertheless believe that when grounded in specif-ics it offers a useful way of framing institutional support fornonprofit housing.2 This article illustrates the importance ofsocial capital through a discussion of institutional collaborationin two cities to improve the development and management oflow-income housing in the nonprofit sector.

The article is organized into five sections. First, we brieflysketch the changing direction of federal housing policy and theincreasing importance of state and local governments and non-profit organizations in developing low-income housing. Thesecond section examines the elements of the institutional sup-port network for nonprofit housing groups and discusses the roleof social capital in tying these elements together. The thirdsection presents two detailed examples to illustrate the influenceof social capital in shaping some of the nation’s strongest institu-tional support networks for nonprofit community developmentorganizations. The fourth section explores several key challengesfacing the institutional support network; and the concludingsection offers several recommendations for strengthening thenetwork.

From the old paradigm to the new

The production and management of rental housing for low-income families has been fundamentally transformed since 1983when the last federally sponsored multifamily production pro-gram was shut down. The Reagan administration terminatedSection 8 New Construction and Substantial Rehabilitationprograms and slowed to a trickle construction of new publichousing (mostly for replacement units) (Hays 1995; Listokin1991). The administration devoted the bulk of the U.S. Depart-ment of Housing and Urban Development’s (HUD’s) much re-duced budget authority to tenant-based Section 8 certificates andvouchers (Hays 1995; Low Income Housing Information Service1994). The only remaining subsidies designated as supply-side

2 The current debate over Francis Fukuyama’s recent book, Trust: The SocialVirtues and the Creation of Prosperity (Fukuyama 1995; Solow 1995), gives aclear indication that social capital is an “idea in good currency” (Schon 1971,123).


were reserved for the elderly, the disabled, and those living inrural areas.3 The other federal subsidies for production of low-income housing were provided indirectly, through block grantsand tax incentives.

The Community Development Block Grant (CDBG) program and,starting in 1991, the HOME program funnel block grants on aformula basis to states and localities for a range of housing andcommunity development activities (Hays 1995; Rich 1993). TheTax Reform Act of 1986 revamped the structure of tax incentives


existing mortgages and, if necessary, fund critical capital im-provements so that the owners could charge “market rents” andoperate without any project-based subsidies. Qualified tenantswould receive rental vouchers they could use for their currentresidence or for a new one (Bodaken 1995; Dunlap 1995; HUD1994). HUD would also eliminate its remaining categoricalhousing production programs (Section 202, Housing Opportuni-ties for People with AIDS) by consolidating them into a newblock grant program for states and localities. Of course, HUD’sproposed reinvention may never be implemented as planned, butthe basic ideas of vouchers and block grants are likely to shapethe ultimate outcome.

In summary, the production of subsidized housing, particularlyin the inner city, now requires multiple financing sources andtypes of subsidies, entails a decreased role for HUD and anincreased role for state and local governments and other institu-tions, and involves greater participation by nonprofit and lessparticipation by for-profit developers. State and local participa-tion in subsidized housing policy making has increased as HUDhas withdrawn (Goetz 1993; Nenno 1991).

With the demise of deep-subsidy production programs, develop-ing affordable multifamily housing in poor inner cities becamefinancially unattractive to most for-profit developers. Nonprofitorganizations have become an “idea in good currency” for thedevelopment of inner-city affordable housing (Schon 1971, 123).5Embraced both by the political right, which sees in them thehistoric strands of self-help and bootstrapping, and by the

5 Very little data are available on the location of low-income housing producedby for-profit developers. Nationally, the for-profit sector accounts for theoverwhelming share of total housing production, with the nonprofit sectorcontributing a minuscule proportion. However, most for-profit housing devel-opment takes place in suburban and nonmetropolitan areas. Most privatesector housing development in central cities tends to be for relatively affluenthouseholds. Private developers also account for the majority of housing unitsdeveloped through the LIHTC. Unfortunately, no data have been collected onthe intrametropolitan location of this housing. Most of the housing expertsinterviewed for this study believe that most private sector LIHTC housing hasbeen developed in suburban areas, not in inner-city communities. With theprivate sector largely absent from depressed inner-city neighborhoods to date,CDCs and other nonprofit groups often have represented the only source ofnew housing development and housing rehabilitation, especially for low-income multifamily housing. An unpublished study sponsored by the LocalInitiatives Support Corporation provides some support for this contention,finding that CDCs accounted for about 90 percent of affordable housingproduced in the 1980s in the city of Boston, where CDCs are especially active(Vidal 1995). More definitive and current information on the location of LIHTChousing should become available through Jean Cummings and DeniseDiPasquale’s research (in progress) on the LIHTC.


political left, which views them through the community controllens of the 1960s, nonprofits represent a politically viable ap-proach to affordable housing.6

Nonprofit housing producers appeal to state and local govern-ments for several reasons. Perhaps most important in this era ofexpiring use is their goal of permanent affordability (Davis 1994;Goetz 1993; Mayer 1990).7 Community-based housing sponsorsare in business for the long haul and have no intention of reapingcapital gains or charging market-rate rent (Bratt 1989; Davis1994; Vidal 1995). They are committed to housing the poorest,most needy households and often provide more than housing:opportunities for employment, health care, child care, education,and other services (Leiterman and Stillman 1993; Sullivan 1993).

In many cities, nonprofits are the only organizations willing andable to assemble the multiple sources of funding necessary toproduce low-income housing (Committee for Economic Develop-ment 1995; Vidal 1995). Often the relationship between govern-ment agencies and nonprofit organizations is so close that, asGoetz (1993, 130) puts it, “the distinction between the ‘success’ ofthe local public agency and the ‘success’ of the CDCs becomesblurred.” For example, 11 of 18 housing trust funds studied byConnerly (1993) have established priority or set-aside fundingfor community-based organizations. Similarly, in New York Citythe bulk of the 124,000 low- and moderate-income housing unitsbuilt or rehabilitated under the city’s multibillion-dollar capitalbudget housing program have been developed by nonprofit orga-nizations (Bratt et al. 1994).

Nonprofit housing organizations are active throughout the UnitedStates. According to a national survey conducted by the NationalCongress for Community Economic Development (NCCED 1995),there are more than 2,000 nonprofit community-based develop-ment organizations in the United States, 90 percent of which are

6 The following paragraphs summarize the major contributions and limitationsof CDCs and other nonprofit housing groups. For more information andanalysis, see Bratt 1989; Bratt et al. 1994; Clay 1990; Dreier and Hulchanski1993; Keating, Rasey, and Krumholz 1990; Mayer 1990; McNeely 1993;National Congress for Community Economic Development 1995; OMG, Inc.1995; Pickman et al. 1986; Rasey 1993; Schill 1994; Stoeker 1995; Sullivan1993; Vidal 1992, 1995; Walker 1993; and Zdenek 1987.

7 More than 1.9 million units of subsidized housing (Section 221[d]3, Section236, and Section 8) are at risk of loss as federal subsidies expire and ownersbecome eligible to prepay their mortgages and charge market rents (Clay andWallace 1990). HUD estimates that the cost of renewing expiring Section 8contracts alone will total $14 billion through fiscal 1998 (Dunlap 1995).


involved in creating affordable housing. A survey conducted in1989 of the 173 U.S. cities with populations above 100,000 foundone or more CDCs operating in 95 percent of them (Goetz 1993).The same survey also found that 82 percent of the cities withactive nonprofit housing developers provided them with projectfinancing, as did 63 percent of the states. More than half of thesecities and a similar proportion of states also provided nonprofitswith some administrative funding, predevelopment loans, andtechnical assistance (Goetz 1993).

CDCs are by no means the only important nonprofit organiza-tions producing and managing low-income housing. Nonprofitsthat produce housing for low-income people cover a broad rangeof organizational foci—from those primarily concerned withimproving the neighborhood and the quality of life in it to thosetargeting a particular constituency. While CDCs usually focus ona single neighborhood—commonly combining housing develop-ment with other services—other nonprofit housing groups targetspecific client groups (e.g., the elderly, the homeless, the men-tally ill, AIDS patients) and are less concerned with the well-being of particular neighborhoods.8

The role of nonprofit organizations in low-income housing devel-opment is recognized and reinforced in current federal legisla-tion. The National Affordable Housing Act of 1990 earmarks atleast 15 percent of a participating jurisdiction’s HOME funds forqualified nonprofit housing producers.9 Nonprofits are alsodesignated to play a leading role in the housing programs of therestructured HUD.10

8 For example, a recent survey of nonprofit housing organizations in sixdiverse U.S. cities found that 37 percent owned elderly or special needshousing but did not own any multifamily rental housing (Bratt et al. 1994).

9 Some federal legislation and regulations stipulate the involvement of non-profit groups in housing programs. The Financial Institutions Reform, Recov-ery, and Enforcement Act of 1989 gives nonprofits the right of first refusal topurchase from the Resolution Trust Corporation properties that had been inthe portfolios of distressed savings and loans. A similar provision, applicableto properties owned by failed commercial banks, was included in the Compre-hensive Deposit Insurance Act of 1991. Title VI of the Cranston GonzalezNational Affordable Housing Act of 1990 gives nonprofits and other “prioritypurchasers” the right to make the first bona fide offer to purchase a federallysubsidized development whose owner has announced an interest in prepayingthe mortgage. The Internal Revenue Code contains a provision that requires atleast 10 percent of each state’s annual LIHTC allocation to be earmarked forprojects that are at least partially owned by qualified nonprofit organizations.

10 For example, HUD’s original Reinvention Blueprint, issued in December1994, several times emphasizes the importance of nonprofit community-basedorganizations in producing and managing low-income housing.


The growth of state and local housing programs and the increas-ing involvement of nonprofit housing groups are now widelynoted (Bratt 1989, 1992; Davis 1994; Goetz 1993; NCCED 1995;Nenno 1991; Stegman and Holden 1987; Vidal 1992; Walker1993). There is much less research on the relationships betweennonprofit housing groups and state and local governments andother providers of financial, technical, and political support. Theremaining sections of this article examine the nature of theserelationships in the institutional support network for nonprofithousing. The discussion draws from our recent research on themanagement of nonprofit low-income housing in six U.S. cities(Bratt et al. 1994).

The institutional support network: Social capital andthe nonprofit housing system

Nonprofit housing producers, in stitching together the patchworkfinancing of development deals, rely on a host of public, private,and other nonprofit organizations.11 These organizations playmany roles in supporting the production and management ofnonprofit housing. Some provide equity capital, loans, andgrants for housing development; others offer financial assistancefor troubled developments, subsidize the salaries of nonprofitstaff members, and underwrite other operating costs. State and

The focus on the development side of affordable housing is particu-larly important because of the lack of sufficient affordable housingsupply in many jurisdictions and the [lack of] expertise of privateactors (e.g., nonprofit community-based organizations, developers)in this area. . . . To ensure strong participation by community-basedorganizations, set-asides for entities such as the community housingdevelopment organizations (‘CHDOs’) recognized under existing lawwould continue. (HUD 1994, 9, emphasis added)

In addition, the Blueprint’s proposed Community Opportunity Fund (the newname for the expanded Community Development Block Grant) would provideassistance to community-based organizations for neighborhood revitalizationefforts. Finally, in its vision of a radically transformed public housing agenda,the Blueprint notes that in fiscal 1998 “No housing authority would receivefunds directly from HUD. . . . States and localities would have the option ofreplacing non-performing housing authorities with community-based organiza-tions or others” (HUD 1994, 13, emphasis added).

11 We are not arguing that the current financing system is ideal or evenadequate. (See Stegman 1991 and Clancy 1990 for discussion of the negativeexternalities and transaction costs involved in the current approach.) How-ever, given the current system, multiple sources of financing and subsidy areneeded, as well as the capacity to patch those elements together into onedevelopment package.


local governments, as noted above, are critical sources of finan-cial and technical support for nonprofit housing groups in manycities. In fact, nonprofits would probably not be able to producesubsidized housing without the active support of the publicsector. National and local foundations have become key sourcesof funding for predevelopment costs, gap financing, and organi-zational capacity building. CDC coalitions typically act as politi-cal advocates at city halls and state capitols. Educationalinstitutions and consultants are primary sources for training andtechnical assistance (Bratt et al. 1994; Walker 1993). Elementsof the system also serve as monitors or de facto trustees of thenonprofit housing stock, supervising development and manage-ment performance of the nonprofit sponsors on behalf of govern-ment and corporate investors (Goetz 1993; Schwartz et al.forthcoming; Walker 1993).

Many functions of the support system are performed by organi-zations created specifically to respond to the needs of the hous-ing sponsors. For example, the national intermediaries(especially the Local Initiatives Support Corporation [LISC] andthe Enterprise Foundation) raise equity capital through thesyndication of LIHTCs and also provide grants, loans, and tech-nical assistance. Moreover, they often serve as a catalyst informing local housing partnerships. The intermediaries’ connec-tion to major corporations, including philanthropic funders andtax credit investors, serves to legitimize CDCs as viable organi-zations worthy of business and government support. As Walker(1993) argues:

Intermediaries . . . offer implicit guarantees of CDCperformance to private sector financial institutions andpublic sector housing and community developmentagencies. Through organizational needs assessments,monitoring of CDC performance, tying organizationalsupport to performance, and other mechanisms, inter-mediaries will legitimize CDC project efforts. (p. 402)

To understand this institutional network that supports thenonprofit producers of affordable housing, we developed ananalytic framework inspired in part by Robert Putnam’s MakingDemocracy Work (1993). This study introduced the term “socialcapital” to the lexicon of community development in America’sinner cities.12

12 Putnam was not the first to coin the phrase “social capital.” While the termcan be found in the work of James Coleman (1990) and others, its origin seemsto be Jane Jacobs’s The Death and Life of Great American Cities (1963).


Putnam’s detailed research on the network of civic, political,and economic institutions in northern Italy makes a strong casefor the power of social capital in the “sense of mutual reciprocity,the resolution of the dilemmas of collective action and the broad-ening of social identities” (Putnam 1993, 76). At its core, socialcapital is about interdependence and collaboration and is usefulin teasing out the dynamics of the institutional networkundergirding the nonprofit producers of subsidized housing.

Analogous to Putnam’s articulation of social capital is the eco-nomic concept of networks that Walter Powell describes in “Nei-ther Market nor Hierarchy: Network Forms of Organization”:

In network modes of resource allocation, transactionsoccur neither through the discrete exchanges (markets)nor by administrative fiat (firms), but through networksof individuals engaged in reciprocal, preferential, mutu-ally supportive actions. (Powell 1990, 303)

For Powell and other sociologists analyzing business organiza-tions, networks are an operational way of looking at the conceptof social capital among firms and the efficient organization forproduction in an economic setting. Social capital as institutionalnetwork means reciprocity, trust, adaptability, and flexibilityamong individual companies for mutual economic benefit. Whenthe network is working toward and is organized for a sharedvision, Powell suggests that

[T]he basic assumption of network relationships is thatone party is dependent on resources controlled by an-other, and that there are gains to be had by the poolingof resources. In essence the parties to a network agreeto forgo the right to pursue their own interests at theexpense of others. (Powell 1990, 303)

Conceptually, the institutional framework surrounding nonprofithousing sponsors falls between the civic involvement with whichPutnam is concerned and the institutional economic frameworkof Powell. In a sense, the framework is a hybrid—drawing onboth economic interest and civic-mindedness.

The concept of social capital should be more than an engagingimage that has captured the imagination of many urban watch-ers.13 Organizations that develop and manage housing must

13 For example, social capital is a central principle in the Committee forEconomic Development’s recent report, Rebuilding Inner-City Communities(1995).


adopt the terminology of sociologists and economists who focuson firms in the business world and civic associations. This trans-lation requires a closer look at the content of social capital in thenonprofit world of community-based housing.

Four themes constitute the substance of social capital in thoseinstitutional networks that effectively support the nonprofithousing sponsor: (1) long-term relationships of trust and reci-procity, (2) shared vision, (3) mutual interest, and (4) financialnexus.

Long-term relationships of trust and reciprocity

In many cities the institutional network of the nonprofit housingsponsor is sustained by long-standing relationships based ontrust, loyalty, and reciprocity among the individuals within thesupport institutions and the nonprofit housing groups. Keyindividuals are former colleagues, are current board members ofthe same organizations, or have switched positions from one partof the system to another. Staff members at the intermediaries,banks, and foundations previously worked at nonprofit housingorganizations, including those they now assist. Public employeesof community development departments in city hall move to aneighborhood housing sponsor or to a bank’s community develop-ment department.

Tracing the career moves of individuals in the nonprofit institu-tional network in Boston, New York, San Francisco, and Chicago(the cities with the densest networks) would demonstrate thecrossing of career paths and the interlocking memberships onboards, committees, and task forces.

Shared vision

A common ideology is widely shared within the nonprofit institu-tional network, which believes that nonprofit sponsors are theappropriate vehicle for building housing and developing commu-nities in inner cities. The shared vision is perhaps most clearlyheld by those institutions created expressly to support nonprofitsponsors—the Enterprise Foundation, LISC, the NeighborhoodReinvestment Corporation, and the National Community Devel-opment Initiative.


Mutual interest

To the extent that foundations, city hall, and the intermediaries“put their money” on nonprofits, they have an enlightened self-interest in seeing those nonprofits succeed. Simply put, thebetter the nonprofit performs, the wiser and more effective thesupport institution appears. In New York, Oakland, California,and Boston, among others, programs rely on the nonprofit spon-sors. Housing policy has been shaped around them. And largeallocations of city or CDBG funds have been channeled to non-profit-sponsored housing. The community housing sponsors andmembers of the network are mutually dependent on each otherfor program and policy outcomes. The major foundations activein this field are greatly invested in the success of CDCs and intheir efforts to reknit the fabric of the nation’s inner cities.14

Financial nexus

In addition to the first three elements that we identify as mak-ing up social capital in the nonprofit housing world, connectionsare also derived solely from economic self-interest in whichrational choice and financial interest are the bonding element.Here, the language of Powell’s network form of organizationalrelationship comes into play.

Given the multiple funding sources required to develop subsi-dized housing, a range of lending institutions are involved withthe nonprofits and have a financial interest in their well-being.Obviously banks and state housing finance agencies fall into thiscategory, but perhaps the most interested actor in the network isthe equity investor seeking a significant financial return oninvestment. The LIHTC brings large-scale corporate investorsinto the nonprofit orbit. These corporations develop sizablestakes in the housing portfolios of the nonprofit organizationsand face significant financial losses if this housing falters. Thisfact puts pressure on the national intermediaries and theiraffiliates that syndicate the LIHTCs to the corporate community.Not only do the corporations suffer financially if the manage-ment of tax-credit developments violates federal regulations orgets into financial trouble, but the intermediaries would also loseface with their investors and contributors, making it much moredifficult for them to syndicate additional tax credit deals. This

14 These foundations include the Ford Foundation, the John D. and CatherineT. MacArthur Foundation, the Pew Charitable Trusts, The Annie E. CaseyFoundation, the Rockefeller Foundation, and Lilly Endowment, Inc.


pressure to safeguard corporate investment (and their owncorporate credibility) motivates corporations to take a leadershiprole in the institutional support network.15

Grounding theory: Social capital in practice

To put empirical clothes on our social capital framework, weexamine in some detail the institutional network in two of thesix sites we studied in our management research—Minneapolis–St. Paul and Boston (Bratt et al. 1994). Two case studies, ofcourse, do not create a sufficient empirical foundation for thetheory of social capital and networks as the basis for successfulhousing production and management at the neighborhood level.The findings should be viewed as preliminary and exploratory.16

Minneapolis–St. Paul’s Interagency Stabilization Group

The Twin Cities provide a compelling example of social capitalwithin the institutional support network. Spearheaded by theFamily Housing Fund of Minneapolis and St. Paul, the networkhas recently coalesced around the Twin Cities Interagency Stabi-lization Group (ISG). Founded in 1993, ISG includes all of themetropolitan region’s major housing-related institutions.17 Topofficials of each institution meet weekly to develop stabilizationplans for distressed properties selected from a continuously

15 Compare this intense economic interest of the corporations with that ofHUD when it was the financial investor in the old housing developmentparadigm. The magnitude of the distressed HUD portfolio provides testimonialto the lack of market discipline in the relationship between sponsor and HUDin these programs.

16 More extensive and systematic analysis of the role of social capital in theinstitutional support network for nonprofit housing is likely to emerge fromthe recently launched assessment of the National Community DevelopmentInitiative’s second phase. In the first phase, from 1991 to 1994, foundationsand corporations invested $63 million to help CDCs in 21 cities increase theircapacity to produce low-income housing. In the second phase, which began in1995 with an additional $88 million commitment for loans and grants, thefunders (who now include HUD) have broadened the initiative’s focus tostrengthen institutional support for community-based nonprofit groups (OMG,Inc. 1995). Indeed, the request for proposal for the phase II assessment singlesout social capital as a key area for analysis.

17 These institutions include the Family Housing Fund of Minneapolis and St.Paul, the McKnight Foundation, LISC, the Minneapolis Community Develop-ment Agency, the St. Paul Department of Planning and Economic Develop-ment, the Minnesota Housing Finance Agency, and HUD.


revised watch list. The group has decided to give top priority tostabilization of the existing stock; new housing development ison the back burner.

ISG coordinates how each institution invests its resources in thestabilization of distressed publicly subsidized housing. Memberscome to the table and commit some portion of their institution’sbudget to an appropriate element of the stabilization plan (e.g.,debt reduction, deferred maintenance, capital improvements).Most stabilization plans combine financial assistance with rec-ommendations or requirements that the owner modify its assetor property management practices. In some instances, ISG hasconcluded that the property should be sold or, in extreme cases,demolished. By November 1994, ISG had committed more than$5 million to assist 63 developments under nonprofit and for-profit ownership.

The shared commitment and high level of coordination andcooperation among ISG institutions would not be feasible if theseinstitutions had not developed a significant reservoir of socialcapital. Indeed ISG exemplifies all four dimensions of socialcapital outlined above, including a heavy dose of financial self-interest.

ISG draws strength from the personal and institutional relation-ships among its members and between its members and thenonprofit housing sponsors. These trusting relationships make itpossible for ISG members to coordinate the allocation of theirrespective resources to the stabilization plans. These numbersalso designated the Minnesota Housing Finance Agency as thesingle project monitor on behalf of all ISG member institutions,thereby eliminating many duplicate reporting and compliancerequirements for the property owners. These actions show howthe member institutions are bypassing their own internal bu-reaucratic procedures in favor of the collective judgment of ISG.

We cannot overemphasize the significance of a trusting relation-ship that enables organizations with a financial stake to stream-line their paperwork and to give authority to one member. Thereduction in transaction costs has not been calculated in dollaramounts, but it is significant and measurable—and the envy ofother cities that have multiple reporting requirements because ofa lack of reciprocity and trust.

These aspects of ISG’s operation epitomize Powell’s notion ofnetwork relations. Each element of ISG depends on resourcescontrolled by another, and each element agrees to forgo the right


to pursue its own interest at the expense of others. Such collabo-ration is possible because of the participants’ confidence in eachother’s competence and commitment. The sense of confidencearises in part from significant personal relationships and isreinforced by the weekly ISG meetings and the history of suc-cessful collaboration.

Mutual trust is also evident between ISG and the nonprofitowners of troubled properties. To receive stabilization funds,sponsors must be candid with ISG about the state of their devel-opments, producing extensive documentation on their financialstatus, property management procedures, and resident satisfac-tion. In approving a stabilization plan, ISG often requires thesponsor to modify the structure of its property management andasset management operations, sometimes even turning propertymanagement over to another company. Most significant, thesponsor must notify ISG of any new projects it intends to pursue,and ISG reserves the right to terminate stabilization funding ifit determines that the owner’s participation in additional devel-opment will jeopardize its ability to manage its current portfolioeffectively. Thus, the executive directors and boards of housingorganizations are relinquishing some of their decision-makingauthority to the collective judgment of the institutional supportsystem, as embodied by ISG.

ISG explicitly incorporates a sense of shared commitment andvision. As noted above, its membership unanimously emphasizesproject stabilization over new development. Each member recog-nizes the importance of correcting the financial and physicalproblems of the existing stock and of doing so in a systematicway. It is unlikely that the players in the Twin Cities’ institu-tional support network could have collaborated as closely andwith such coordination if they did not have the same vision ofcommunity priorities.

The degree of coordination also reflects mutual interests.Through collaboration each organization in the network canleverage its own resources with those of other institutions,thereby ensuring that its funds contribute to an efficient whole.

While emphasizing the elements of social capital (reciprocity andtrust, shared vision, and mutual interest), one should not mini-mize the significance of financial stake. Most ISG members havecommitted millions of dollars to the development of nonprofit-sponsored housing developments. The failure of these projectswould be a financial and political disaster. Trust, commitment,and financial interest all run in the same direction.


The Boston institutional support network

The institutional support network is well entrenched in Boston,where nonprofit housing sponsors have a history of sustainedoperation since the 1960s when they were an integral part of thecity’s massive urban renewal program. Some sponsors havefallen by the wayside, others have endured, and new ones arosein the 1970s and later in response to local, state, and federalinitiatives. While participation was modified by a recent down-turn in the city’s economy, Massachusetts and the city of Bostontraditionally commit not only federal funds but also their own tothe nonprofit sponsors that have been a vehicle for creating newand rehabilitated units for low- and moderate-income people.

The support network for subsidized housing production andmanagement has many players. The aggressive and innovativeMassachusetts Housing Finance Agency (MHFA) leads a groupof quasi-public statewide entities that finance affordable housingin Boston and provide technical assistance and seed money. TheMetropolitan Boston Housing Partnership (MBHP) is an um-brella organization that provides technical assistance, loans,grants, and asset management support to many community-based sponsors. Its board is made up of leaders in the bankingcommunity and representatives of the public and nonprofitsectors—the classic public-private partnership. HistoricallyMBHP has had tremendous clout with the banks because itsvigorous and deeply committed chairman was one of the mostinfluential bank presidents in the city. The Community Buildersis a highly skilled organization that has been providing technicalassistance to community-based housing organizations for morethan 25 years. It has also been involved in many of the housinginitiatives put forward by nonprofit sponsors in Boston.

The institutional support system for housing in Boston is denseand deeply rooted.18 Players have long-term relationships andoften have worked in more than one organization in the network.Lateral moves are common among city, state, and nonprofitsponsors of housing. For example, the current director of theBoston LISC office was formerly executive director of a success-ful CDC. The executive director of MBHP previously worked forthe state’s Executive Office of Communities and Development(EOCD), which provides supportive funding to the CDCs and isin charge of allocating LIHTCs. Until recently the executive

18 Some might say too dense and deeply rooted. In an era of increasingly scarceresources for housing development, issues of turf and control are ever present.


director of MHFA sat on the board of MBHP. He and two of histop assistants are currently consulting with the city of Bostonabout its future housing agenda.

The power of the shared vision, represented by CDCs in Boston,has been enhanced by foundation support from both local organi-zations and national funders such as the Ford Foundation. Somefor-profit housing developers believe that the system of nonprofitand public support for affordable housing is tilted in favor of thenonprofits. It seems that unless an organization adheres to thecommunity development image of neighborhood empowermentthrough nonprofit efforts, there is little likelihood of funding.Whether or not this is a fair conclusion, no thoughtful observerwould dispute that the city of Boston, the Commonwealth, andthe foundations have placed tremendous reliance and expecta-tion on the vision represented by nonprofit producers of housingoperating from a neighborhood base.

Extensive mutual interest among the various layers in the sup-port network has evolved after many years of incremental link-ing of organizations and agendas. The history of the networkreveals many examples of reciprocal and enlightened self-interest. Perhaps the most vivid example is MBHP’s firstproject, called BHP I. This 700-unit, scattered-site project wasrehabilitated in 1984 by 10 community-based sponsors. BHP Iwas undercapitalized from the start, and the physical conditionof the units and the rent structure established to support thebuildings have always been problems. Since 1990, many of theplayers have undertaken the long and complex process of re-structuring the developments. Despite sustained efforts to refi-nance, to deal with lead paint issues, and to absorb a diminutionof state rental subsidies, 8 of BHP I’s 10 developments were indefault by 1993.

The resolution, which finally came in the fall of 1994, involved acomplex set of transactions requiring cooperation among FannieMae, the city of Boston, MHFA, the individual sponsors, MBHP,EOCD, and a variety of other players. The projects are safe fromforeclosure for at least five years because of the extraordinarywillingness of individual players, through both self- and mutualinterest, to contribute funds, cut back on their return, and sharein the risk and losses associated with the reconfiguration. Thehistory of the negotiations is byzantine, exhausting, and amaz-ing. Many times when the process seemed lost and foreclosureinevitable, one more effort on the part of a combination of play-ers reworked the terms.


The following minutes from the MBHP board meeting of Novem-ber 1994 indicate the degree to which support networks weremobilized to prevent foreclosure:

FNMA [Fannie Mae] is providing $6,619,739 to theworkout in the form of zero interest second mortgageswhich cover the write-down of the present first mort-gages ($3,511,735); funds towards capital improvementand environmental needs ($1,453,955); and funds tocover delinquent accounts payable, legal and closingcosts, the cost of environmental and capital needsassessments and past due bond trustee payments($1,654,049). The balance of the proposed uses are beingfunded by MBHP ($250,000); the City of Boston/PublicFacilities Department ($250,000); the Executive Officeof Communities and Development Lead Paint GrantProgram ($145,000); the Department of Mental Health($83,820); and the Massachusetts Housing PartnershipLead Paint Loan Guarantee Program ($338,640). Theten partnerships are expected to contribute a total of$865,330 from Replacement Reserve accounts, specialescrow and capital accounts and available partnershipcash. Each partnership will retain $5,000 to $10,000 tofund a new replacement reserve account. MHFA willsubordinate its debt ($1,468,150) behind Fannie Mae;contribute approximately $2,000,000 towards securityand . . . [the state rental subsidy program] will befunded under the 1993 declination schedule.

From one point of view, this is a dry and complex list of who ispaying how much for what; but the description is a mosaic ofnetwork players, each contributing to save the buildings andunits that symbolize a shared vision for revitalized neighbor-hoods in Boston.

Summary

The examples of support networks in the Twin Cities and Bostonrepresent the best of social capital in the community develop-ment field. We do not argue that these stories are reproduciblein any city that tries hard enough or that there is widespreadevidence that most cities are moving in this direction. We areconcerned with showing the concept of social capital as it appliesto the most advanced situations and the ways in which theconcept can describe the institutional context that has arisen to


deal with the complexities of financing, building, and maintain-ing subsidized housing in the inner city.

Challenges facing the institutional support network

Institutional support networks—strong or weak—are confrontedby several serious challenges in this era of block grants andvouchers. Four critical issues face these networks: (1) unevendevelopment, (2) the dangers of offering overly aggressive sup-port, (3) competition and conflict among networks supportingdifferent types of nonprofit housing, and (4) the need for stabili-zation of troubled developments.

Uneven development

One shortcoming of the institutional support network is itsuneven development. Because some cities and states host morevibrant nonprofit housing organizations and richer institutionalsupport systems than others, localities with more aggressive andentrepreneurial systems are likely to produce more affordablehousing in the new policy climate than those with fewer, smaller,and institutionally isolated nonprofit housing groups. Thus,cities and regions with equal need for affordable housing willexperience different levels of development, depending on thenumber, size, and strength of the nonprofit housing sponsors andinstitutional support providers. Smaller areas and areas lackingin local foundations and traditions of institutional collaborationare less likely to nurture viable support systems for nonprofithousing sponsors.

The assessment of the first phase of the National CommunityDevelopment Initiative (NCDI) (see footnote 16) underscores theproblems posed by weak or uneven institutional support. Al-though NCDI initially emphasized the goal of increasing thehousing output of individual CDCs, it gradually recognized that“ ‘getting to scale’ entailed a process of improving the environ-ment in which CDCs operate, making it easier for CDCs to gainaccess to new resources and skills for organizational stabilityand project implementation” (OMG, Inc. 1995, 95–96). In study-ing the 21 cities involved in NCDI, the assessors found thatplaces with weaker environments tended to have limited under-standing of community development, limited or underdevelopedstrategies to guide local collaboration, little government partici-pation, and unfavorable economic and local market conditions. Insome of these cities, the support network was relatively new and


undeveloped. In others, it suffered from low levels of coordina-tion and government participation. Perhaps most telling, theNCDI assessment found that CDCs in cities with weaker institu-tional support were less able to meet their housing productiongoals than their counterparts in cities with stronger networks.

The geographic unevenness of nonprofit housing activity under-scores the central role of the national intermediaries. Theseinstitutions are crucial in providing nonprofits located in institu-tionally barren environments with the necessary financial,technical, political, and moral support to foster a viable low-income housing industry. With the national government evermore remote, the intermediaries are increasingly responsible forstandardizing and simplifying the development process andfinding ways to disseminate their assistance as broadly as pos-sible. It is also incumbent on the intermediaries, with theirstrong corporate and foundation backing, to cultivate supportivelocal networks for nonprofit housing groups. LISC, the Enter-prise Foundation, and the Neighborhood Reinvestment Corpora-tion have accepted this challenge, as shown by their growingroster of local offices, their engagement in NCDI, and their jointsponsorship of the Consortium of Housing and Asset Manage-ment (CHAM)19—but the task is enormous.

Overly aggressive support

One challenge for the support network is to become ubiquitous.Another is to refrain from being overly aggressive in areas wherethe network is already strong. There is an inherent tensionbetween the availability of institutional resources for nonprofithousing groups and the increased expectations placed on thesegroups. On one hand, the institutional support network hasbecome something of a safety net for troubled properties andorganizations. On the other hand, the institutional supportnetwork often places increasing pressure on nonprofits to as-sume responsibility for more and more housing. One riskinvolved with this growth is that groups may become overex-tended. Their staff and management systems may not be able toprovide the same quality of service when, for example, theirinventory climbs from 200 to 800 units in two years—a not

19 CHAM is a collaboration among the Enterprise Foundation, LISC, and theNeighborhood Reinvestment Corporation. Launched in 1994, this nationalinitiative is intended to increase the capacity of local nonprofits to perform asresponsible owners and managers of affordable housing by expanding thetraining and technical assistance resources available to them. CHAM hasreceived initial funding from several national foundations (Bratt et al. 1994).


uncommon scenario in New York (Bratt et al. 1994). In Boston,the institutional support network strongly encouraged nonprofitgroups to take on the severely underfinanced BHP I develop-ments—with disastrous results (see above). In Oakland, Califor-nia, the city is pressing a successful CDC active in an Asianneighborhood to work in communities dominated by other ethnicgroups (Bratt et al. 1994).

The support system sees nonprofits as the best—and in somecases the only—resource to assume responsibility for low-incomehousing, especially when the developments are already occupied,poorly constructed, or inadequately financed. But there aredanger signs that without careful organizational development,sufficient resources, and in some cases growth control, commu-nity organizations can become overwhelmed by the magnitude ofthe housing problems.

Competition

The movement toward state and local block grants is likely tobenefit housing groups operating in settings with strong institu-tional networks. State and local governments that already relyon nonprofits to carry out their housing programs, and thatprovide them with financial and technical assistance to do so,will probably treat new housing block grants in the same manneras they did CDBG and HOME block grants. However, the shift toblock grants will mean that sponsors of and advocates for elderlyhousing, homeless housing, multifamily housing, and even publichousing will now compete for the same program dollars. Theinstitutional networks that have developed around particulartypes of housing and worked within the framework of specializedcategorical programs (Section 202, McKinney, Public Housing)must now vie against each other for the same block grants. Thechallenge will be for these distinct networks to find ways ofworking together—in other words, to develop social capital.

Project stabilization

Once open for occupancy, most new nonprofit-owned multifamilyhousing depends on rental income to cover operating costs.Except for Section 202 and other programs for the elderly,disabled, and homeless, the federal government has not offeredoperating subsidies for new developments since the early 1980s.Most housing developments subsidized through state and localprograms (including housing trust funds) and through the


federal CDBG, HOME, and LIHTC programs do not offer subsi-dies for ongoing operating costs (Connerly 1993; Stegman andHolden 1987). Typically, these programs provide upfront subsi-dies in the form of grants and low-interest loans to reduce theamount of debt service expenses that would need to be coveredby rent rolls (DiPasquale and Cummings 1992). While thesesubsidies may make housing initially affordable to low-incomefamilies, they are not structured to cover operating costs such asutilities, maintenance, and taxes over time. Subsidies are notavailable to cover increased water bills, unanticipated repairs, orrevenue shortfalls owing to higher than expected vacancies orarrears. HUD’s proposed elimination of operating subsidies forpublic housing and developments funded under other project-based subsidy programs (e.g., Section 8, Section 236, Section202) means that their owners will join nonprofit sponsors ofnewer multifamily developments in depending on rent rolls forsurvival.

The absence of operating subsidies makes it difficult for manydevelopments to remain financially secure over time. Theseproblems are exacerbated when developments are inadequatelyconstructed or renovated and thus require continual repairs,when they repeatedly suffer from vandalism and other types ofabuse, or when they are located in crime-ridden or drug-infestedareas (Bratt et al. 1994). When, as is often the case, projects arethinly capitalized, they may lack adequate reserves and may notbe fully renovated or may be built with less than optimal materi-als and equipment (Bratt et al. 1994; Walker 1993).

One of the biggest challenges to the institutional support net-work, therefore, is project stabilization. If the nonprofit housingstock is to be self-supporting through rent collection, the institu-tional support network needs to address the inadequate under-writing and the difficult surroundings of many nonprofitdevelopments. This is the intent of the Twin Cities’ ISG, and italso underlies HUD’s mark-to-market proposal. ISG attempts tostabilize distressed properties through debt reduction, capitalimprovements, additional capital and operating reserves, andother measures. HUD’s mark-to-market plan would draw onFederal Housing Administration funds to reduce mortgage prin-cipal and, in some cases, to support capital improvements so thatmarket rents can cover debt services and operating costs(Dunlap 1995). In some instances, rent rolls and reserves are fartoo small to support major capital improvements and thus re-quire outside assistance. New York City’s Department of Hous-ing Preservation and Development, for example, provides


low-interest loans for nonprofit and for-profit owners of low-income housing to replace building systems (e.g., roofs, plumb-ing, boilers, windows) (Bratt et al. 1994; Michetti 1993).

A formidable task for the institutional support network is tostabilize undercapitalized housing developments without simplybailing out irresponsible or incompetent nonprofit sponsors.Perhaps networks in other cities and states should emulateISG’s practice of assessing the management performance ofnonprofit (and for-profit) housing before committing any funds toproject stabilization.

Federal policy and the institutional network:Conclusions

Given the stream of changes flowing from the federal govern-ment, the institutional network that constitutes the nonprofithousing system is facing a difficult challenge. It must forge newrelationships and supportive arrangements to preserve theexisting stock of assisted housing at precisely the time thatuncertainty and fear of declining resources test existing ties andmake the necessary capacity-building investments in institutionslook expensive.

At the federal level, legislation should permit localities to sup-port projects both directly for construction or rehabilitation andindirectly by selectively tying vouchers to developments. Thisflexibility is especially important in localities with soft housingmarkets: Sponsors in neighborhoods with high crime rates orpoor schools will have difficulty competing for tenants withvouchers. CDBG funds, if combined in a new, more inclusiveblock grant, must continue to be usable for housing and will bean essential ingredient in any recipe to make nonprofit ownersmore competitive and better able to oversee their properties.

Perhaps the most immediate federal priority should be the pres-ervation of the LIHTC. The LIHTC has been both the salvationand the bane of low-income multifamily housing development. Itis the only way to provide the equity capital needed to lower therent of new or rehabilitated housing for low- and moderate-income households, but it is complex and usually must be aug-mented by other funding sources (Postyn 1994). But, for all thecontortions and splicing of financing that it has generated, theLIHTC has served a critical financing function. If it is elimi-nated, nonprofit and for-profit sponsors may not be able to con-tinue producing affordable housing.


In addition to reducing urgently needed equity, the demise of theLIHTC would decrease the number of players with a financialinterest in inner-city affordable housing. Large corporations andother investors will no longer have a vested interest in the out-come of inner-city housing development. If the “sunset” of theLIHTC were to be accompanied by the introduction of a newsource of equity for financing inner-city housing, there would belittle protest. But such an alternative appears unlikely.

New responsibilities at the state and local levels will strainmany agencies. The national and local intermediaries, nonprofitcoalitions and advocacy groups, and HUD each have a role toplay in building capacity at the state and local level—particu-larly in cities and states with limited experience with nonprofithousing sponsors.

Competition for funding will shift from the federal budgetingprocess to localities. Cities will be tempted to respond byspreading declining subsidy dollars more thinly, weakeningunderwriting and further reducing fees. Conversely, advocacyorganizations (e.g., nonprofit housing groups, advocates forhomeless persons, public housing residents) may be tempted toundercut one another on behalf of their respective constituen-cies.

HUD is in a position to reward those localities that enhancetheir institutional networks. HUD’s review of Consolidated Plansshould include the extent of nonprofit participation, local invest-ment in capacity-building and institutional infrastructure,collaborative planning and implementation processes, and sensi-tivity to neighborhoods.20 The secretary should use the discre-tionary money in the proposed new block grants (over and abovethe funds distributed on an entitlement basis) to reward citiesthat do well on these issues. HUD is also in a good position toprovide visibility and how-to information about innovative ap-proaches developed by localities.

Whether called “public-private partnerships,” “social capital,” orsimply “mutual support,” the future of the community-basedhousing movement is very much wrapped up in the future ofinstitutional support networks. Over the past 15 years,

20 As of 1995, HUD requires eligible jurisdictions to submit Consolidated Plansdescribing and justifying their intended use of funds provided under theCDBG, HOME, Housing Opportunities for People with AIDS, and McKinneyTitle IV Homeless programs. This replaced the requirement that each jurisdic-tion prepare a Comprehensive Housing Assistance Strategy.


participants in the networks have demonstrated remarkablefacility in devising new program delivery mechanisms and insti-tutional forms that both address problems as they occur andattract new players to the network. This creative and construc-tive infrastructure is one of our most versatile policy tools in thestruggle to create positive living environments in inner citiesand should command our attention accordingly.

Authors

Langley C. Keyes is Ford Professor of City and Regional Planning in theDepartment of Urban Studies and Planning at the Massachusetts Institute ofTechnology. Alex Schwartz is Senior Research Associate at the CommunityDevelopment Research Center and Assistant Professor in the Robert J. MilanoGraduate School of Management and Urban Policy at the New School forSocial Research. Avis C. Vidal is Director of the Community DevelopmentResearch Center and Associate Professor at the Robert J. Milano GraduateSchool of Management and Urban Policy at the New School for Social Re-search. Rachel G. Bratt is Associate Professor in the Department of Urban andEnvironmental Policy at Tufts University.

References

Berenyi, Eileen B. 1989. Locally Funded Housing Programs in the UnitedStates: A Survey of the 51 Most Populated Cities. New York: CommunityDevelopment Research Center.

Bodaken, Michael. 1995. We Must Preserve the Nation’s Supply of AffordableHousing. Shelterforce 17(4):15–17.

Bratt, Rachel G. 1989. Rebuilding a Low-Income Housing Policy. Philadelphia:Temple University Press.

Bratt, Rachel G. 1992. Federal Constraints and Retrenchment in Housing: TheOpportunities and Limits of State and Local Governments. The Journal of Lawand Politics 8(4):651–701.

Bratt, Rachel G., Langley C. Keyes, Alex Schwartz, and Avis C. Vidal. 1994.Confronting the Management Challenge: Housing Management in the Non-profit Sector. New York: Community Development Research Center.

Case, Karl. 1991. Investors, Developers and Supply-Side Subsidies: How MuchIs Enough? Housing Policy Debate 2(2):341–56.

Clancy, Patrick E. 1990. Tax Incentives and Federal Housing Programs:Proposed Principles for the 1990s. In Building Foundations, ed. DeniseDiPasquale and Langley C. Keyes. Philadelphia: University of PennsylvaniaPress.

Clay, Philip L. 1990. Mainstreaming the Community Builders: The Challengeof Expanding the Capacity of Non-Profit Housing Development Organizations.


Cambridge, MA: Massachusetts Institute of Technology, Department of UrbanStudies and Planning.

Clay, Philip L., and James E. Wallace. 1990. Preservation of the ExistingStock of Assisted Private Housing. In Building Foundations, ed. DeniseDiPasquale and Langley C. Keyes. Philadelphia: University of PennsylvaniaPress.

Coleman, James S. 1990. Foundations of Social Theory. Cambridge, MA:Harvard University Press.

Committee for Economic Development. 1995. Rebuilding Inner-City Communi-ties: A New Approach to the Nation’s Urban Crisis. New York.

Community Information Exchange. 1994. Case Studies on Multifamily RentalHousing. Washington, DC.

Connerly, Charles. 1993. A Survey and Assessment of Housing Trust Funds inthe United States. Journal of the American Planning Association 59(3):306–19.

Cooper, Stephen K. 1995. Congress Set to Pass Sunset of Low-Income HousingTax Credit. Housing and Development Reporter 23(28):417, 446.

Davis, John Emmeus, ed. 1994. The Affordable City: Toward a Third SectorHousing Policy. Philadelphia: Temple University Press.

DiPasquale, Denise, and Jean L. Cummings. 1992. Financing MultifamilyRental Housing: The Changing Role of Lenders and Investors. Housing PolicyDebate 3(1):77–116.

Dreier, Peter, and J. David Hulchanski. 1993. The Role of Nonprofit Housingin Canada and the United States: Some Comparisons. Housing Policy Debate4(1):43–80.

Dunlap, Helen M. 1995. Mark to Market: A Strategy to Recapitalize HUD’sAssisted Housing Stock. Shelterforce 17(4):14, 31.

Fukuyama, Francis. 1995. Trust: The Social Virtues and the Creation ofProsperity. New York: The Free Press.

Goetz, Edward G. 1993. Shelter Burden: Local Politics and Progressive Hous-ing Policy. Philadelphia: Temple University Press.

Hays, R. Allen. 1995. The Federal Government and Urban Housing. 2nd ed.Albany: State University of New York Press.

Hebert, Scott, Kathleen Heintz, Chris Baron, Nancy Kay, and James E.Wallace. 1993. Nonprofit Housing: Costs and Funding: Final Report. Vol. I,Findings, and Vol. II, Case Studies. Prepared by Abt Associates, Inc., for U.S.Department of Housing and Urban Development, Office of Policy Developmentand Research.

Hendershott, Patric H. 1990. The Tax Reform Act of 1986 and Real Estate. InBuilding Foundations, ed. Denise DiPasquale and Langley C. Keyes. Philadel-phia: University of Pennsylvania Press.


Jacobs, Jane. 1963. The Death and Life of Great American Cities. New York:Vintage Books.

Joint Center for Housing Studies. 1995. State of the Nation’s Housing 1995.Cambridge, MA: Harvard University.

Keating, Dennis, Keith P. Rasey, and Norman Krumholz. 1990. CommunityDevelopment Corporations in the United States: Their Role in Housing andUrban Redevelopment. In Government and Housing, ed. W. van Vliet and J.van Weesep. Newbury Park, CA: Sage Publications.

Leiterman, Mindy, and Joseph Stillman. 1993. Building Community: A Reporton Social Community Development Initiatives. New York: Local InitiativesSupport Corporation.

Listokin, David. 1991. Federal Housing Policy and Preservation: HistoricalEvolution, Patterns, and Implications. Housing Policy Debate 2(2):157–76.

Low Income Housing Information Service. 1994. At a Snail’s Pace. Washing-ton, DC.

Mayer, Neil S. 1990. The Role of Nonprofits in Renewed Federal HousingEfforts. In Building Foundations, ed. Denise DiPasquale and Langley C.Keyes. Philadelphia: University of Pennsylvania Press.

McNeely, Joseph B. 1993. Building for the Future. Washington, DC: FannieMae Foundation.

Michetti, Felice. 1993. The New York City Capital Program for AffordableHousing. In Housing America, ed. Jess Lederman. Chicago: Probus PublishingCompany.

National Congress for Community Economic Development. 1995. Tying It AllTogether: The Comprehensive Achievements of Community-Based DevelopmentOrganizations. Washington, DC.

Nenno, Mary K. 1991. State and Local Governments: New Initiatives in Low-Income Housing Preservation. Housing Policy Debate 2(2):467–96.

OMG, Inc. 1995. Comprehensive Assessment Report: The National CommunityDevelopment Initiative, Phase I (1991–1994). Philadelphia.

Oser, Alan S. 1994. Unexpected Arrivals Settle a Casualty Condominium. TheNew York Times, January 8.

Pickman, James, Benson F. Roberts, Mindy Leiterman, and Robert N. Mittle.1986. Producing Lower Income Housing: Local Initiatives. Washington, DC:Bureau of National Affairs.

Portes, Alejandro, and Julia Sensenbrenner. 1993. Embeddedness and Immi-gration: Notes on the Determinants of Economic Action. American Journal ofSociology 98(6):1320–50.

Postyn, Sarah H. 1994. The Low Income Housing Tax Credit: A Study of ItsImpact at the Project Level. Master’s thesis, Massachusetts Institute ofTechnology.


Powell, Walter. 1990. Neither Market nor Hierarchy: Network Forms ofOrganization. Research in Organizational Behavior 12:295–336.

Putnam, Robert C. (with Robert Leonardi and Raffaella Y. Nanetti). 1993.Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ:Princeton University Press.

Putnam, Robert C. 1995. Bowling Alone Revisited. The Responsive Community5(2):18–32.

Rasey, Keith P. 1993. The Role of Neighborhood-Based Housing Nonprofits inthe Ownership and Control of Housing in U.S. Cities. In Ownership, Control,and the Future of Housing Policy, ed. R. Allen Hays. Westport, CT: GreenwoodPress.

Rich, Michael J. 1993. Federal Policymaking and the Poor: National Goals,Local Choices, and Distributional Outcomes. Princeton, NJ: Princeton Univer-sity Press.

Schill, Michael H. 1994. The Role of the Nonprofit Sector in Low-IncomeHousing Production: A Comparative Perspective. Urban Affairs Quarterly30(1):74–101.

Schon, Donald A. 1971. Beyond the Stable State. New York: W. W. Norton.

Schwartz, Alex, Rachel G. Bratt, Avis C. Vidal, and Langley C. Keyes. Forth-coming. The Management Challenge: Nonprofit Housing Organizations andtheir Institutional Support System. Journal of Urban Affairs.

Solow, Robert. 1995. But Verify. The New Republic, September 11, pp. 36–39.

Stegman, Michael. 1991. The Excessive Costs of Creative Finance: GrowingInefficiencies in the Production of Low-Income Housing. Housing Policy Debate2(2):357–76.

Stegman, Michael, and J. David Holden. 1987. Nonfederal Housing Programs.Washington, DC: Urban Land Institute.

Stoeker, Randy. 1995. The Political Economy of the Community DevelopmentCorporation Model of Urban Redevelopment. Unpublished manuscript, Univer-sity of Toledo.

Sullivan, Mercer. 1993. More Than Housing. New York: Community Develop-ment Research Center.

Tassos, James. 1996. Telephone interview by Alex Schwartz. Washington, DC:National Council of State Housing Finance Agencies.

Terner, Ian Donald, and Thomas B. Cook. 1990. New Directions for FederalHousing Policy: The Role of the States. In Building Foundations, ed. DeniseDiPasquale and Langley C. Keyes. Philadelphia: University of PennsylvaniaPress.

U.S. Department of Housing and Urban Development. 1994. HUD ReinventionBlueprint. Washington, DC.


Vidal, Avis. 1992. Rebuilding Communities. New York: Community Develop-ment Research Center, New School for Social Research.

Vidal, Avis. 1995. Reintegrating Disadvantaged Communities in the Fabric ofUrban Life: The Role of Community Development. Housing Policy Debate6(1):169–230.

Walker, Christopher. 1993. Nonprofit Housing Development: Status, Trends,and Prospects. Housing Policy Debate 4(3):369–414.

Zdenek, Robert. 1987. Community Development Corporations. In Beyond theMarket and the State, ed. Severyn T. Bruyn and James Mehan. Philadelphia:Temple University Press.


Comment on Langley C. Keyes et al.’s “Networks and Nonprofits” 231Housing Policy Debate • Volume 7, Issue 2 231© Fannie Mae Foundation 1996. All Rights Reserved.

Comment on Langley C. Keyes et al.’s“Networks and Nonprofits: Opportunities andChallenges in an Era of Federal Devolution”

David A. CroweNational Association of Home Builders

Abstract

Housing markets are determined by a complex interplay of consumers andsuppliers. The Keyes et al. article discusses the changing landscape fornonprofit housing providers and what recent developments in federal housingassistance policy will mean to them. But this perspective is too narrow topredict the effects of changes in federal housing policy because all housingproviders are somewhat interrelated. All housing providers need to be consid-ered, and using the terms “for-profit” and “nonprofit” to distinguish betweenthe two types of providers is unfortunate and misleading.

For-profits and nonprofits are fundamentally different: They place a differentemphasis on community, and nonprofits can often deliver subsidies that for-profits cannot. In addition, the strengths and the skills needed to producehousing under somewhat different objectives have led to some specialization.Ultimately, however, determining the optimal provider or mix of providers isbest left to local and state governments as federal housing assistance devolves.

Keywords: Low-income housing; Nonprofit sector; Social capital

Introduction

The dynamics of a housing market are generated by the indi-vidual decisions of many housing consumers and many housingsuppliers. Within these multiple connections of one householdobtaining housing from one provider, there are some common-alities in housing providers and housing consumers. The articleby Keyes et al. focuses on one group of matches within the hous-ing market dynamic: community housing providers and low-income, inner-city renters.

Keyes et al. discuss the changing landscape for nonprofit housingproviders and what federal housing policy change will mean tothem. This, however, may be too narrow a perspective fromwhich to judge the consequences of change and the appropriateresponse to them.

232 David A. Crowe

The article postulates the future of nonprofit housing providersby describing the networks that contribute to their success. Theauthors’ objective is to prepare nonprofit housing providers forthe future and to identify where problems and conflicts are mostlikely to arise. However, the success and failure of all housingproviders are interrelated, and a look into the future shouldinclude the dynamics between housing providers as well as thedynamics within specific providers’ orbs.

This review begins with a brief critique of “Networks andNonprofits” to note the issues that appear most important. It isfollowed by a discussion of why housing providers are segmentedinto different types and what they have in common.

Critique

Keyes et al. observe that housing programs have become lessoriented toward federal government control and more orientedtoward local operation. Rather than a single source of subsidy,like the Section 8 program, the provision of low-income housingrequires multiple equity and debt sources, layered subsidies, andactive participation from a multitude of players. Although nothighlighted in the article, the absence of significant state andlocal government support has also been a major reason for otherparties to step in and negotiate deals (Belsky 1993).

This devolution of housing policy has left community-basedorganizations to handle the most difficult piece of housing ser-vices: housing low-income renters in the inner city. Keyes et al.describe the role of community housing providers as similar tothat of a local government agency, “housing the poorest, mostneedy households” and acting as a social service agency provid-ing “more than housing.” The influence of the nonprofit has beenenforced through set-asides in federal government grant pro-grams and proposed HUD reinvention strategies, and the au-thors recommend continuing these favorable federal preferences.However, they also caution that well-established communityhousing providers may already have capacity problems thatcould be exacerbated if future federal housing programs continuepreferences while giving more responsibility to state and localgovernments.

Keyes et al. believe the success of the nonprofit housing providerlies in nonmarket forces called “social capital.” The elements ofsocial capital that bind the many actors in low-income housingare long-term relationships among the people in community

Comment on Langley C. Keyes et al.’s “Networks and Nonprofits” 233

housing organizations, a common philosophy or belief in thenonprofit approach, and a mutual interest in success. The au-thors present two case studies as examples of how communityhousing organizations, one new and one seasoned, wield socialcapital to achieve success. Although community housing provid-ers are generally referred to as developers and producers of low-income housing in the article, the two case studies do notsupport this view. Instead, they describe organizations focusedon preserving existing housing, often in financial distress frompoor management or initial underfunding. Other more minorillustrations offered in support of the authors’ views in the“Challenges” section of their article also describe managementissues rather than the production of new units.

Given the trend in housing policy and the conditions for success-ful community housing organizations, the authors offer fourareas of concern that should be addressed to ensure the continu-ance of community-based housing providers. First, the presenceof community housing providers is uneven, with some placeshaving rich institutional support while others are “institution-ally barren.” Their recommendation is for the national interme-diaries to step in and facilitate wider representation. However,the authors fail to warn the national intermediaries to heedtheir second piece of advice, which is to limit the breadth of theirresponsibility to what they can accomplish. The national inter-mediaries provide essential links to debt and equity financing,but local solutions call for local partners.

Second, Keyes et al. caution that community housing organiza-tions will be under greater pressure to accept more work thanthey can handle as local governments are given more responsibil-ity and authority to run housing programs. Organizations shouldheed this warning and periodically reconsider their initial char-ter and goals.

Third, community housing providers will find themselves com-peting with other special interest housing providers as federalhousing assistance arrives as a block grant. The concern thatother housing providers (e.g., for elderly, homeless, public hous-ing) will compete for limited dollars reveals some basic differ-ences in the goals that Keyes et al. ascribe to “nonprofits” versusother “advocates.” Oddly, rather than viewing local decisions onthe use of block grant funds as more flexible and more likely tofavor community development, the authors consider a moveaway from federal program prescription as a threat to commu-nity housing groups.

234 David A. Crowe

Fourth, the proposals to use housing subsidies that allow thetenant to choose where to live rather than subsidizing specificprojects will make it difficult for many developments to remainfinancially secure. The authors correctly observe that if propos-als are enacted into law that eliminate housing subsidies tied tobuildings, then buildings in less desirable locations could losetenants.

In summary, the authors recommend that the federal govern-ment permit localities to support project-based housing subsi-dies; that intermediaries, including the U.S. Department ofHousing and Urban Development (HUD), help build the capacityof community housing organizations; and that HUD’s grantapplication criteria should encourage the participation of com-munity housing organizations. They also caution that competi-tion among community housing organizations could lead to lesshousing for the truly needy. Keyes et al. focus on the nonprofithousing providers and the networks that support them. Theirperspective does not include other types of housing providers,notably profit-motivated providers. To understand the strengthsand weaknesses of nonprofits, it is necessary to understand whythere are different types of housing providers.

Differences among housing providers

The provision of housing is accomplished by many differentfirms, organizations, and governments. Each operates in a differ-ent manner with different objectives. The differences amongthem can be collapsed into three broad areas: the distinctions ofprofit motivation, the capture of externalities, and the need forsubsidy. These areas are important in understanding the appro-priate place and function of different actors in response tofederal changes in housing policy. Of these, however, the nomen-clature distinction of the presence or absence of a profit motiva-tion is the least useful to differentiate among housing providers.Certainly, market-oriented firms are established to make a profitwhile community-based organizations do not provide a return toinvestors. But the use of these terms to distinguish between thetwo types of housing providers is unfortunate and misleading.

The profit-motivation distinction is unfortunate because thedifferences between the two types of housing providers aremuch more complex than whether the organization provides areturn to investors and pays taxes. The distinction is misleadingbecause the absence of a profit motivation is not necessarilysynonymous with lower costs.


Profit motivation

Community housing organizations are called “nonprofit housingproviders,” and market-oriented firms are described as “profit-motivated” or “for-profit.” Beyond the simple differences, thepresumption is often made that housing produced by nonprofitsis less expensive than housing produced by for-profits, owing tothe absence of a profit margin in the final price or rent. Further,the demarcation of housing providers implies that the profit-motivated have no social motivation and also that a social moti-vation requires that participants receive no financial reward fortheir good works.

First, home building is a very competitive market, and firmscompete away all but an economic profit. The classic definition ofa competitive market that ensures firms make only economicprofits postulates the following requirements:

a. There are many producers and none is large enough todominate the market. In 1992, over 130,000 firms werecategorized as residential general contractors in the Censusof Construction (U.S. Bureau of the Census 1994). No firmaccounts for more than 1 percent of the national market,and the top 100 account for about 10 percent of single-familyhousing starts.

b. The same or similar product is available from all providers.Building technology is well known, and most firms canproduce comparable homes.

c. The supply of resources is elastic. The components of homebuilding (land, labor, and materials) are widely availablewithout significant supply constraints.

d. All participants have complete knowledge of prices andcosts. Certainly, the price of inputs and the final price ofhomes are well known to buyers and builders (NationalAssociation of Home Builders [NAHB] 1996).

Second, profit is a relatively small portion of the price of a home.The concept is difficult to measure since building firms usedifferent accounting procedures. Most firms are organized asSubchapter S corporations, which can blur the distinction be-tween salary and profit, and much of the hard costs of construc-tion are contracted out. Nevertheless, an NAHB study (1995a)found that home-building firms retain about a 9 percent profitmargin. Hence, at best, if the management and risk premiums

236 David A. Crowe

earned by the entrepreneur were eliminated, prices or rentswould not be reduced by more than 9 percent. This conclusion,however, does not account for the fact that competition andprofit motivation produce efficiencies.

Profits above an “acceptable” or market-driven level can alsotake other forms, whether in a for-profit or nonprofit organiza-tion. High salaries, generous employee benefits, or expensivecompany offices can all be buried in either type of organization’sfinancial statements. The opportunity for incurring unnecessaryor unreasonable expenses is less in a for-profit organization thanin a nonprofit because the owners of the former have a greatereconomic interest in eliminating such expenses.

Third, profit is the return for taking risk. Home builders producehousing with the intention, but not the guarantee, of selling it.Even those firms that build on contract to the owner receivefunds after they are expended and occasionally encounter prob-lems of nonpayment. Larger scale developers who must wadethrough approval and subdivision processes, install infrastruc-ture, and then build homes risk even larger amounts before asale is made and their investment begins to yield a return.Without profit, the return to investors for bearing risk, no in-vestment capital would be available to housing.

Community housing providers also encounter risk. However,community-based organizations typically have little capital oftheir own at risk, and so the risk is usually borne by financialinstitutions and national intermediaries.

Empirical proof of a difference in production costs betweennonprofit and for-profit housing providers is difficult because ofthe wide variation in cost accounting and acquisition procedures.In a small, nonrandom sample of for-profit and nonprofit multi-family construction projects, Abt Associates (Hebert et al. 1993,iii) found that “nonprofit development costs were sometimeshigher and sometimes lower than industry averages.” Thus, thepresence or lack of profit motivation does not determine the realdifferences between housing providers. The more elementarydifferences involve the capture of externalities and the need forsubsidy.

Capture of externalities

A useful distinction between for-profit providers and nonprofitsinvolves the pursuit of different primary objectives: saving or


stabilizing existing neighborhoods and selling or renting housingprofitably. Both types of providers have a stake in thecommunity’s continued well-being, and both expect to achievefinancial success. The difference is that their sensitivity to thesetwo objectives varies. The way in which each firm orders theseobjectives in a given project is perhaps best reflected in thecapture of externalities associated with that project.

Rehabilitating or redeveloping inner-city sites, for example,necessarily entails fitting or refitting homes into existing com-munities. Local residents clearly have a stake in whether andhow housing improvements are done. Improved and fully func-tional housing provides positive externalities to the community,and the community in turn affects the sustainability of housingby providing the right environment. Since nonprofit communityhousing providers often spring from community concerns, it isonly natural that they are primarily seeking the maximumpositive impact of the housing improvement on the community.Conversely, building housing with the expectation of profitablyselling or renting it requires anticipating and meeting prospec-tive buyers’ or renters’ needs. So it is reasonable to expect thatfor-profit, market-driven developers focus primarily on generat-ing the community from the housing to be created and on choos-ing locations and existing surroundings that will appeal toprospective buyers or renters, rather than on fulfilling the needsof the existing neighborhood. To the greatest extent possible, thefor-profit provider is seeking to internalize the positive exter-nalities of the housing improvements for the prospective ownersor renters.

Need for subsidy

Another useful distinction between market-oriented firms andcommunity housing providers is the availability of resourcesand the need for subsidy. New homes are produced by market-oriented firms to fit the incomes of buyers. Producers incorporatesizes, features, and amenities to fit the buyer’s pocketbook. If thebuyer or renter cannot afford the product, the home or apart-ment does not get produced or is offered to someone else.

Providing housing to low-income residents of the central cityoften involves housing that costs more to provide than the resi-dent can afford to spend. As pointed out by Harvard’s JointCenter for Housing Studies (1995), rents of $200 to $300 aregenerally insufficient to cover basic operating costs, much lesscapital costs and improvements. A full-time minimum wage job

238 David A. Crowe

holder can only afford to pay a maximum of $220 per month forgross rent. According to the 1993 American Housing Survey,10 percent of the central-city rental housing stock (1.3 millionunits) rented at or below $220 (excluding units with no cashrent), but 26 percent of the central-city renter households hadincomes below the level considered necessary to afford even$220 per month (U.S. Bureau of the Census 1995).

Thus, some form of assistance is often necessary to make up thedifference between the tenant’s ability to pay and the rentsneeded to maintain decent housing. Federal assistance hasdiminished and will likely continue to do so. The newest forms offederal assistance, such as HOME, require cost sharing. Hence,assistance needs to be assembled from other sources if signifi-cant numbers of families are to be served.

In this scenario, community housing providers act as a surrogatelocal government agency, collecting support from all the likelysources. The two examples discussed by Keyes et al. in Minne-apolis and Boston are very close to this activity of coordinatingand cajoling actors into a unified plan and process.

The authors, however, did not mention the strong magnet tocooperation that the federal government continues to applythrough the Community Reinvestment Act (CRA) and the afford-able housing goals established for the government-sponsoredenterprises of Fannie Mae and Freddie Mac. CRA requirementsbring local lenders to the table and provide some form of financ-ing subsidy to projects benefiting the community. Recentstrengthening of CRA requirements will further encourage theparticipation of financial institutions.

Affordable housing goals for Fannie Mae and Freddie Mac havegenerated new efforts by these financial giants to develop specialmortgage products for households with incomes below theirtraditional customer’s income. In addition, two programs withinthe Federal Home Loan Bank System—the Affordable HousingProgram (AHP) and the Community Investment Program—provide funds for financing very low, low-, and moderate-incomerental and owner-occupied housing for community development.The AHP supplied $100 million to finance affordable housing in1995 (NAHB 1995b). While these incentives are not direct fed-eral funding, they encourage a flow of funds to affordable hous-ing projects typical of community housing providers.

The authors postulate that an increased dependence on localsolutions to housing problems will entail increased demands on


the community housing network. Their advice is to not acceptmore than the system can handle. In many cases, however, theproblem of capacity is not a lack of housing providers in a com-munity but conflicting objectives among the suppliers of housing.The arbitrator of such conflicts should be the local governmentagency with the expertise to efficiently allocate limitedresources.

Local governments have significant experience in administeringhousing block grants. They have also had much greater experi-ence in administering public housing programs. While the publichousing program has received wide publicity for its failures,these failures have been primarily in large cities with severehousing problems inside and outside the public housing system.Many small public housing authorities have been supplyingdecent housing to their clientele for 50 years or more withoutserious problems. Thus, local governments do have the capacityor can develop the capacity to efficiently administer housingprograms.

The authors’ advice to target a portion of federal housing subsi-dies to community housing providers would limit local govern-ments with strong capabilities. A lack of a set-aside does notlimit community provider participation but rather allows localgovernments to choose the correct combination of housing pro-vider, housing consumer, and vehicle of subsidy for that locationand that locality’s voters. Regardless of the network developed tosupport community housing providers, the democratic process ofrequiring elected officials to make allocation decisions that affectthe electorate provides the best guarantee that the wishes oftaxpayers are carried out.

Similarities among housing providers

Notwithstanding the elementary differences described above,some strong similarities are shared among all housing providers.First, all housing providers operate in a market where housingconsumers make choices and react to the different bundles ofhousing offered. Second, all providers are concerned about thefinancial viability of their projects. Third, surrounding neighborshave a stake in the community and the kind of housing they livenear, regardless of who provides it. For these reasons, all provid-ers have a significant stake in the continued integrity of thecommunity, irrespective of their profit motivation.

240 David A. Crowe

An issue raised by Keyes et al. clearly illustrates such similari-ties. They point out a problem of potential concern to all ownersof projects subsidized by the federal government: The HUDreinvention proposal includes the elimination of project-basedsubsidies (i.e., the guarantee of rent payments for all the units ina building). Instead, tenants would be allowed to use their sub-sidy for any unit on the market. If such a proposal is enacted,the projects that were designated by the federal government toserve low-income households would have to compete for tenants.These buildings were constructed or purchased not only to pro-vide housing for a targeted group of people, but also to rejuve-nate, enhance, or stabilize a neighborhood. Those buildings maybe less attractive because of their location, rather than becauseof the housing services they provide. Community housing provid-ers should oppose this proposal because it endangers the neigh-borhoods where these buildings are located. Market-orientedowners should oppose this proposal because the commitment tosubsidize the project was based on a social purpose as well as anindividual purpose. Forcing the project to compete solely on thebasis of individual preferences ignores its original social purposeand provides no incentive for individuals to collectively choose tooccupy the building, thereby providing the positive neighborhoodeffects associated with a viable project.

Conclusion

The HUD reinvention proposal exemplifies potential changesthat would affect all housing providers. Housing suppliers,regardless of their profit motivation, are united in pursuing thebest quality homes and communities for their customers. Theiremphasis on housing and community and the financial means oftheir customers may be different, but all providers are united intheir efforts to make federal assistance flexible and sustainable.

Transferring decision-making authority to state and local gov-ernments will take time and a period of readjustment. Keyes etal. provide issues and cautions for community housing providersto consider as federal policy evolves. However, the authors fail toconsider the niche of community housing providers within thelarger picture of housing provision. For projects that requirenetworks of social services and housing support, nonprofit hous-ing providers may possess a comparative advantage. The theoryof competitive markets suggests that, for projects that requirehousing at the lowest cost, for-profit, market-driven housingproviders possess a comparative advantage. In some cases, acooperative effort is the best approach. In all cases, the most


appropriate provider should be locally determined by the needsof the project, not by an arbitrary federal allocation.

Author

David A. Crowe is Staff Vice President for Housing Policy at the NationalAssociation of Home Builders.

The views expressed in this article are those of the author and should not beconstrued as representing the official views of the National Association ofHome Builders.

References

Belsky, Eric S. 1993. The States and Housing Assistance. Housing Economics41(5):5–8.

Hebert, Scott, Kathleen Heintz, Chris Baron, Nancy Kay, and James E.Wallace. 1993. Nonprofit Housing: Costs and Funding: Final Report. Vol. I,Findings. Prepared by Abt Associates, Inc., for U.S. Department of Housingand Urban Development, Office of Policy Development and Research.

Joint Center for Housing Studies of Harvard University. 1995. The State of theNation’s Housing. Cambridge, MA.

National Association of Home Builders. 1995a. Builders Survey of Construc-tion Costs. Washington, DC.

National Association of Home Builders. 1995b. Financing for Small Buildersand Developers. Washington, DC.

National Association of Home Builders. 1996. The Future of Home Building.Washington, DC.

U.S. Bureau of the Census. 1994. Census of Construction Industries for 1992.CC 92-I-1. Washington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census. 1995. American Housing Survey for the UnitedStates in 1993. Current Housing Reports, H-150-93. Washington, DC: U.S.Government Printing Office.

242 David A. Crowe

Comment on Langley C. Keyes et al.’s “Networks and Nonprofits” 243Housing Policy Debate • Volume 7, Issue 2 243© Fannie Mae Foundation 1996. All Rights Reserved.

Comment on Langley C. Keyes et al.’s“Networks and Nonprofits: Opportunities andChallenges in an Era of Federal Devolution”

James R. FollainSyracuse University

Abstract

The Keyes et al. article is an important contribution to the discussion of thefuture role of nonprofit organizations (NPOs) in the delivery of low-incomehousing. This article does a good job of examining networks and linking theNPO discussion to the concept of social capital. Also, the case studies provideinformative empirical evidence. However, Keyes et al. has several weaknessesin the argument for NPOs. The lack of a careful explanation of the case formarket failure is the primary weakness. The argument can also be strength-ened by expounding on the problems of investing in low-quality housing. Otherpotential weaknesses include the difficulty of identifying stakeholders and thelikely leakage of subsidies to NPO employees.

My comments echo the authors’ call for additional research about NPOs. Moredata and investigation are needed to determine the proper role of NPOs, whetherthey should serve a political purpose, how the services of NPO networks should bepriced, and how efficient NPOs are relative to other private organizations thatprovide similar services.

Keywords: Nonprofit sector; Low-income housing; Social capital

Introduction

The Keyes et al. article advocates the use of nonprofit organiza-tions (NPOs) to provide affordable housing; more specifically, itargues that networks of these NPOs are highly desirable and inneed of further support and development. The article is nicelywritten and organized and offers an important perspective on thefuture of federal housing policy. The major elements of a strongcase for the authors’ position are included and explained. Moregenerally, the article provides further evidence of our country’slong and admirable love affair with community and volunteerorganizations. Unfortunately, from my perspective, the develop-ment of their case suffers from several problems. These problemsare discussed and suggestions are made about the specific infor-mation that is needed to more fully evaluate the importance andpotential of NPOs in the delivery of housing services.


Weaknesses in the case for more NPO involvement

Despite these highly desirable qualities, the case for furtherNPO involvement in housing policy needs further developmentin several areas. These are briefly summarized below.

Case for market failure

In a market-oriented society, the argument for governmentintervention should begin with a discussion of why market forcesare incapable of providing the necessary resources. That is, whatis the evidence for market failure? Such an argument is notoffered in this article. Economists have long argued that low-income households reside in low-quality dwellings primarilybecause of their poverty; the housing market is not necessarilyindicted by such an observation. The challenge to those whosupport the need for supply-side or “place” intervention is todemonstrate something further. Specifically, the argument mustbe made that the price of housing differs from the social cost ofproducing housing. Potential explanations for market failureinclude the existence of racial discrimination, the failure tointernalize important neighborhood externalities, problems inthe financing of low-income housing, and the importance of adecent dwelling to the opportunities for work by low-incomehouseholds.

Although the difficulty of measuring the external costs andbenefits of housing makes this a challenging assignment, I thinkmaking this case is especially important in today’s environmentfor several reasons. First, evidence abounds of “governmentfailure” in the delivery of housing services in the past 20 years orso, including failures in the multifamily mortgage co-insuranceprograms and the dilapidated status of many public housingprograms. Given these failures, policy makers deserve a clearerexplanation of why the government can be expected to do betterthan the market. Second, the growing evidence of a widening gapbetween the wages of unskilled and skilled labor and the stillhigh vacancy rates in multifamily housing suggest that theaffordability problems experienced by many low-income house-holds may be attributable more to labor market failure than tohousing market failure. Third, the optimal housing policy de-pends on the nature of the market failure. If the problem isrelated to the market for equity finance, then policies like thelow-income housing tax credit seem more appropriate. If theproblem is in the market for debt finance, then mortgage insur-ance may be appropriate. Exactly what is the source of market

246 James R. Follain

failure that calls for the public support and reliance on NPOs?The authors should address this directly.

Problems of investing in marginal housing

The authors are concerned that the viability of NPOs is sensitiveto the strength of the housing market. They argue that the“flexibility is especially important to localities with soft housingmarkets.” The vulnerability of any housing investment to a softhousing market is a concern to any organization that invests inhousing, whether run by the private sector, the government, orNPOs. To my way of thinking, a weak or soft housing market canalso be labeled a buyers’ market, which seems to benefit low-income households. The primary goal of policy should not be thesurvival of a developer, whether this developer is an NPO or aprivate organization; improving the welfare of individuals is theprimary focus of policy.

What the authors probably mean is that investment in housingat the low end of the quality distribution (marginal housing) ismore risky than investment in other segments of the market.This logic might follow from a simple filtering model of housing.According to such a model, existing housing in the lowest rangeof housing quality is the most likely to be put out of serviceduring periods of unexpected income growth. This may be true;however, the purchase price of such housing would be expectedto be lower in response to this source of uncertainty. The bestcure for such risk is likely to be accurate appraisals of the prop-erty, not operating subsidies to NPOs. Those NPOs that are bestable to assess this risk will pay lower prices for the property andhave a better chance of surviving unexpected weakness in thelow end of the market. Those who are less able to do this will goout of business. Why should the incentives for accurate appraisalbe any different for nonprofit organizations than for for-profitorganizations?

Lack of stakeholders: Moral hazard problems

Follain and Szymanoski (1995) provide a framework with whichto analyze the case for government intervention in the marketfor debt-financed affordable housing. We specifically address oneof the common complaints of NPOs: Mortgages are often difficultto obtain. Although this is a complex issue, we do argue thatNPOs suffer from a problem less likely to be associated with for-profit organizations in the market for debt finance: Defining the


equity of the NPO in housing projects is more difficult. NPOemployees and board members usually do not invest their ownequity in the projects, and their compensation is not as depend-ent on the build-up of equity. Moreover, many of these organiza-tions consist of volunteers from the community. Salariedemployees of NPOs may have a stake in the success of theproject, but no more than employees of for-profit organizations.These considerations lead us to conclude that NPOs are likely tobe riskier borrowers than for-profit developers, all else equal,because the identification of the stakeholders in NPO projects ismore difficult. As a consequence, we expect NPOs to encountermore difficulty in the mortgage market; for example, lendersmay require tougher underwriting criteria in the form of higherdown payments.

The authors do not address this case of what many refer to as amoral hazard problem. The solutions to the problem in the pri-vate market for housing consist of down payments, regular andaccurate reports on income and balance sheets, and a good trackrecord for investment and management. It would be interestingto know the authors’ assessment of this problem for NPOs.Perhaps they would emphasize the offsetting and implicit assetassociated with the NPO, social capital. If so, and if this is to berelevant in the market for mortgages, ways of measuring itsvalue should be discussed.

Leakage to employees

A standard complaint about all supply-side programs is thatsome of the subsidy is diverted from the tenants to the suppliersin the form of higher salaries, perks, and so on. Why wouldNPOs be immune to this potential for leakage? I think an analy-sis of the salaries, perks, and productivity of NPO employees anddirectors is needed to show otherwise. In brief, why take thechance that some portion of the subsidy goes to anyone otherthan the intended beneficiaries (low-income households)? Theauthors should address this point.

New role for government

An increasing reliance on NPOs to deliver housing servicesalters but does not eliminate the role of government. Under thesupply-side policies of the past 30 years or so, governments haveplayed a major role in the production of housing; this will ceaseunder new policies. Instead, government must become a monitor


of NPO performance. Standards must be established to measureNPO productivity. These standards will, in turn, require moreand better data about NPO operations. Some questions includethe following: Who should bear the cost of monitoring, the gov-ernment or industry? (Under some environmental programs, agovernment monitor resides on-site with the private firm, andthe cost of the monitoring is paid by the firm.) What would bethe specific criteria used to evaluate NPOs? Would evaluationsbe based on the number of people served? Would some placesreceive higher priority than others? These are some of the issuesin the current debate about the establishment of affordablehousing mandates for Fannie Mae and Freddie Mac. I think thesame kind of process is needed if greater reliance is placed onNPOs.

Specific information needed to evaluate theargument

The type of research offered in this article is sorely needed.Indeed, the authors perform valuable research by focusing oncase studies, which is an excellent way to study the generalconcept of social capital and the specific role of NPOs and NPOnetworks. However, the evidence offered in these case studies isalmost exclusively qualitative in nature. Although qualitativeinformation is interesting, quantitative evidence is needed tomake a stronger case for the kinds of policies offered in thisarticle. The following comments suggest areas in which moreevidence is particularly needed and how some of this evidencemight be assembled.

How and where should NPOs intervene?

The authors note the “geographic unevenness” in the distributionof NPOs among markets. Removing this inequality is recom-mended as a high priority for government and NPO networks. Adiscussion of the specific criteria by which this unevenness isassessed and addressed is missing. NPOs may not exist in someareas because the market for affordable housing functions rea-sonably well. In some areas the political process may be workingfine. Of course, the political process may be part of the problemin those areas where strict rent control or excessive building andhousing codes are enforced. For these reasons, it is necessary tohave a better sense of the market and political circumstancesunder which NPOs are most successful. It would also be interest-ing to know about other aspects of organizational design that


lead to successful NPOs. For example, should a franchise ap-proach be developed? How much control should be exerted by thecentral organization? These and many other questions must beaddressed to remedy the geographical unevenness noted by theauthors.

Should NPOs be used to serve a political purpose?

Although I am not a political scientist, it seems to me that thedevelopment of NPOs is a critical political issue. Perhaps theyserve as important representatives of certain groups who areotherwise underrepresented in the political process. In thissense, they serve a potentially valuable role in conveying infor-mation to elected officials and their representatives. This“bottom-up” perspective on NPOs seems closer to the perspectiveheld among many involved in NPOs. Although this role seemslegitimate, the challenge is to be an important element in urbangovernance without being perceived as a partisan force. Provid-ing government funds for partisan initiatives is probably a badidea (witness the ongoing congressional debate regarding thelobbying activity of NPOs).

How should the services provided by NPO networks bevalued?

If NPO networks continue to evolve, a natural question willarise: How will the central NPOs charge individual NPOs fortheir services? The value of these services will depend, in part,on the value of the social capital associated with the national orregional provider. Although some may consider the pricing ofsocial capital unique to NPOs, my own sense is that the pricingissue is faced by many private organizations as well, especiallyorganizations whose value depends on their knowledge of theindustry. At a minimum, the problem may be akin to the pricingissues faced by national associations, such as the National Asso-ciation of Realtors and the National Association of Home Build-ers. However, it may be closer to the relationships among privatefirms, which include franchises (e.g., automobile dealerships). Ineither event, a good portion of the price for services providedmay depend on the value of the social capital.


What is the comparative efficiency of NPOs?

The push to place greater reliance on NPOs raises the need forbetter information about their performance. NPOs are forced toprovide certain types of information for Internal Revenue Servicepurposes and may file U.S. Department of Housing and UrbanDevelopment reports if they receive certain types of financialassistance. Such information should be standardized and madecomparable to the information provided about private develop-ment organizations. Key information should include net operat-ing income, operating expense and its major components,mortgage information, payments and commitments to equityholders, and valuation of the property. In brief, comprehensiveincome and balance sheets should be provided on a regular basis.Information about the tenants of the properties is also needed toassess the social mission of the NPO. Once collected, the infor-mation should be evaluated to rank the NPOs in terms of theirrelative efficiency. Some suggestions for this work come fromrecent attempts to evaluate the efficiency of K–12 public schools(Ruggiero 1994).

Final thought

I was flying from a conference (about the role of nongovernmen-tal government organizations in urban governance) in Berlin,Germany, to New York City when I first read this article. I wasparticularly struck by the authors’ touching description of theexperiences of those involved in NPO activities. These descrip-tions emphasized the cooperative efforts associated with NPOactivity. I began thinking about the enormous amount of coop-eration needed to fly me, my wife, and my luggage from Berlin toSyracuse. Cooperation was needed among pilots, flight attend-ants, baggage movers, customs agents, taxicab drivers, currencyexchangers, and many more. The fact that most of these peoplewere paid for their assistance does not taint their contributions,in my mind. My guess is that many of those involved were alsopleased that they could be of service and that the trip went well.In this sense, some of the differences between the activities ofthese private agents and NPOs seem minor. In sum, NPOs donot have a monopoly on cooperative efforts or on the good feelingthat comes from having done a good job. We ought to keep this inmind as we seek to judge the relative importance of NPOs versusfor-profit firms.


Author

James R. Follain is a Professor of Economics and a Senior Research Associatein the Center for Policy Research at Syracuse University.

References

Follain, James R., and Edward J. Szymanoski. 1995. A Framework for Evalu-ating Government’s Evolving Role in Multifamily Mortgage Markets. Citiscape1(2):151–77.

Ruggiero, John. 1994. Nonparametric Estimation of Cost Efficiency in thePublic Sector with an Application to New York State School Districts. Metro-politan Studies Program Paper No. 165. Syracuse University.


Local Housing Plans: Learning from Great Britain 253Housing Policy Debate • Volume 7, Issue 2 253© Fannie Mae Foundation 1996. All Rights Reserved.

Local Housing Plans: Learning fromGreat Britain

David P. VaradyRutgers University andUniversity of Cincinnati

Abstract

As the U.S. Department of Housing and Urban Development refines the newConsolidated Plans, which replace the Comprehensive Housing AffordabilityStrategies, it should examine Britain’s experience with local housing plans.Case studies of four best-practice cities—Glasgow, Dundee, Birmingham, andYork—highlight the value of these plans in assessing the success of cities intheir new “enabler” role.

Five key lessons for American cities emerge from this article. First, theseplans can serve multiple roles beyond bids to central government. Second,local housing plans should address market-rate as well as below-market-ratehousing issues. Third, American housing plans should use a wider range ofdata sources than census information alone and should incorporate housingmarket analyses dealing with specific areas and population groups. Fourth,the stress on implementation and strategy in British plans should be emu-lated. Finally, aspects of Britain’s competitive bidding system should beconsidered for implementation.

Keywords: Great Britain; Urban planning; Low-income housing

Introduction

America’s record in local housing planning over the past 20 yearshas been mixed at best. Most local officials viewed the HousingAssistance Plans (HAPs), which are required to receive Commu-nity Development Block Grant (CDBG) funds, as a “paperexercise.” As the U.S. Department of Housing and Urban Devel-opment (HUD) refines the new Consolidated Plans, which re-placed the Comprehensive Housing Affordability Strategy(CHAS),1 HUD should draw lessons from other countries where

1 According to the final rule published by HUD, “the planning activitiesembodied in the rule [providing for the Consolidated Plan] are those of theComprehensive Housing Affordability Strategy (CHAS) requirements enactedby the Cranston-Gonzalez National Affordable Housing Act (NAHA) and of theCommunity Development Plan requirements, added to the Community Devel-opment Block Grant Program by NAHA” (Final Rule 1994). According toKathryn Nelson of HUD (1995), the Consolidated Plan does not have as much

254 David P. Varady

housing planning has been more successful. Great Britain’sexperiences are particularly relevant. The U.K.’s system of localhousing planning was instituted at about the same time asAmerica’s HAP. However, Britain’s system has proved to be moresophisticated and has benefited from strong technical guidanceissued by London’s (or Edinburgh’s) central government to itslocalities.

This article describes the emergence of Britain’s system of com-petitive local housing plans and seeks to answer seven sets ofquestions regarding these plans.

1. What are the missions of these plans? Are they simply bidsfor funds from central government? If not, what other func-tions do they serve?

2. What is the scope of the local planning effort? To whatextent do the plans extend beyond low-income housingissues to middle-income housing issues?

3. What approaches are used to measure housing problems?What is the relative importance of housing needs analysis ascompared with housing market analysis (a comparison offuture demand for housing with future supply to identify theneed for new construction)?

4. How are priorities set for different types of policies (e.g.,rehabilitation of inner-city tenements versus the redevelop-ment of peripheral housing estates)?

5. How are citizens and home builders involved in the prepara-tion of plans?

6. To what degree are the plans concerned with implementa-tion issues (e.g., coordination between housing and planningagencies, indicators of management performance)?

7. What is the relative success of neighborhood revitalizationstrategies?

Answers to these questions are sought through case studies offour best-practice British cities: Glasgow (population 672,500),

rigor and oversight regarding priority setting as the CHAS does. The Consoli-dated Plan decreases the number of things that local agencies have to com-plete, including the number of income categories that they have to fill out onforms.

Local Housing Plans: Learning from Great Britain 255

Dundee (172,300), Birmingham (1,014,000), and York (123,100).2The first three are manufacturing cities that experienceddeclines in their industrial base. City leaders attempted torevitalize local economies as centers for the service sector. Incontrast, York is a regional center and much less dependent onmanufacturing, but the successful conservation of its medievalcore has made it an important tourist destination.

Housing and planning officials at both the city and neighborhoodlevels in these four cities were interviewed, as were experts atBritish universities and staff from both the Scottish Office andthe Scottish Building Employers’ Federation.3 In addition, sitevisits to local area housing projects were conducted and relevantplans and reports were studied, particularly the four most recenthousing plans: Birmingham’s Housing Investment Programme,Strategy Statement 1992 (Birmingham City Council 1992),Dundee’s Housing Plan 1990–1995 (City of Dundee n.d.),Glasgow’s Housing Plan for the 1990s (Glasgow City Council1992a), and York’s Housing Strategy Statement 1993/94 (YorkCity Council n.d.).

Recent changes in the housing and planningenvironment4

To understand local housing plans, key characteristics ofBritish housing and planning policy must be recognized. First,

2 The cities were selected based on conversations and correspondence withBritish housing experts including Robina Goodlad and Duncan Maclennan(University of Glasgow); Colin Wood (University of Central England); andBarry Cullingworth (formerly University of Delaware, now Cambridge Univer-sity). These experts defined for themselves what constituted “best practice.”These four local authorities are not necessarily representative of the approxi-mately 450 in Great Britain. In fact, the sample is biased toward Scotland andlarge urban authorities (all controlled by the Labour Party). While importantlessons can be derived from the experiences of these four, the article lacks theexperience of authorities in southern England, who have had to deal with andrecognize the problems of housing affordability and housing shortage. As oneof the anonymous reviewers of this article noted, “southern authorities havebeen much more active and not unsuccessful in an ‘enabling’ approach tosocial/affordable housing provision, e.g., through the use of housing associa-tions, planning powers, low-cost homeownership, etc.”

3 There is a danger in a comparative study such as this of relying too heavilyon statements from local officials and, as a result, uncritically accepting theirsentiments. This is especially true today in Great Britain, because of theincreasing importance given to presentation and “quality of strategy.” I triedto avoid this problem by corroborating conclusions across multiple interviews,including national and community officials, in addition to city staff.

4 This section draws heavily from Best (1994).

256 David P. Varady

housing departments are on the District Council level and areresponsible for providing a wide range of housing services, in-cluding the management of council/public housing.5 Housingdepartments are one part of local authority. The Britishuse the terms “local authorities” and “housing authorities”interchangeably.

Second, beginning in 1979, the direct provision of housing hasbeen deemphasized. Housing authorities are losing quality stockto sitting tenants who are entitled to purchase their homes at adiscounted price. Council housing, which peaked at 30 percent ofthe nation’s homes in 1980, now accounts for 20 percent of thetotal. (The comparable figure for the United States is roughly1.5 percent.) The policy has both positive and negative results.Many new owners take special pride in their homes and havemade substantial improvements. The policy’s downside is that ithas reduced the number of rental properties. Local housingauthorities are increasingly seen as enablers that set the strate-gic context for implementing housing policies but have fewpowers to carry out the policies. The enabling function encom-passes promoting house building,6 fostering house improvement,influencing property management, reducing disadvantage, pro-viding information and advice, and being the nucleus for urbanregeneration efforts such as City Challenge (Bramley 1993;Goodlad 1993).

5 The housing function in British cities is quite different from that in Americancities, where housing authorities are usually separate from city governmentand manage public housing, and where city housing development departmentshandle federal government allocations to nonprofit groups and work withdevelopers to produce affordable housing.

6 Since the 1980s, many local authorities have used legal agreements withdevelopers to obtain a proportion of affordable homes in private housingdevelopments (a process known as achieving planning gain). This arrangementis encouraged in the 1992 Department of the Environment’s Planning PolicyGuidance, Note 3, which makes planning consent conditional on a proportionof affordable homes being provided on each significant site. See “Inquiry intoPlanning for Housing” (1994) for a more detailed discussion of the strengthsand weaknesses of using the planning system to attain affordable housinggoals. Dunmore (1992) provides a detailed discussion of efforts to secureaffordable housing through new private developments. These new guidelinesand targets seem remarkably similar to the various types of inclusionaryhousing policies being implemented in different parts of the United States(e.g., policies stemming from the Mt. Laurel decisions in New Jersey and themixed-income housing programs of the Housing Opportunities Commission inMontgomery County, MD).


Third, to provide low-cost and social housing, more emphasis isbeing given to “housing associations.”7 The pressure for growthamong housing associations has led to intense competition andcost cutting. Consequently, some associations have not only hadto charge higher rents but have also had to cut quality stan-dards. These associations are funded by central government viathe Housing Corporation in England and Scottish Homes inScotland.

Two funding channels for housing in Great Britain—Departmentof the Environment/Scottish Office to local authorities, andHousing Corporation/Scottish Homes to housing associations—are competing with each other for a dominant role. One of thekey housing planning issues is how best to link these two fund-ing channels.

In Scotland, there is a remarkably detailed parallel planningapparatus by Scottish Homes (housing association businessplans) and local authorities (housing plans). Housing associationbusiness plans exist only in Scotland and were introduced in1988. Business plans enable housing associations to explainwhat they do, to say what their objectives are, and to indicatehow they are going to achieve their objectives over a two- orthree-year period. A recent report to Scottish Homes (“StrategicUse of Local Authority Housing Plans” 1995) suggests ways forimproved interaction between strategic planning processes inlocal authorities and Scottish Homes’ district planning process.

Fourth, the Housing Act of 1988 gave tenants of local housingauthorities the power to exercise tenants’ choice in the selectionof a new landlord (typically one or more housing associations) totake ownership of their estate. It was assumed that the competi-tion for ownership between the housing department and thehousing associations would raise levels of service.

Fifth, there has been a switch from bricks-and-mortar subsidiesto personal subsidies. That is, central government has reducedsubsidies to council housing and pressured local authorities toraise rents. In the past, the subsidies have allowed rents to bekept low. Local authorities have been pressured to increaserents. However, the reduced subsidies and higher rents have notled to government savings, because all tenants are entitled toHousing Benefit through the Department of Social Security,

7 “Social housing” is a term used throughout Europe to refer not just to publichousing but also to housing subsidized from public sources. In the U.K., “socialhousing” encompasses both council/public housing and the homes of nonprofithousing associations.

258 David P. Varady

which can reduce the rent to nothing for those who arecompletely dependent on the state. Consequently, the reducedsubsidies to local authorities have been counterbalanced byescalating Housing Benefit costs. Since Housing Benefits is ameans-tested system, the switch may have created a seriousproblem of work disincentives. Further, higher rents may bemaking council housing less attractive for those with higherincomes, thus fostering the “residualization” of this housingstock.

Sixth, major central government initiatives, including “EstatesAction” in England and “Partnership Areas” in Scotland, havebeen established to address poorer public sector rented housing.Cities that use these two programs must provide evidence ofsales to tenants, proposals for alternative landlords, anddevolvement of control to tenant organizations to qualify forfunding. Estates Action involves competitive bidding betweenlocal partnerships of the public, private, and voluntary sectorsfor central government grants. It has been argued that citiesbenefit from this coalition building process even if they do notreceive a central government grant.

Seventh, all housing authorities are required to produce housingplans, which are described in the next section.

Eighth, the regional planning framework varies between Scot-land and England. In Scotland, regional authorities are respon-sible for land supply for a range of uses, including housing. Theirparticular interest today is in providing for the supply of land forthe private sector through the structure planning process. Eachregional authority produces a structure plan every four years(except for the Strathclyde Regional Council, which produces oneevery two years) and includes a section on housing.8 Regionalplanning is done quite differently in England. The metropolitancouncils, which were responsible for strategic planning, wereabolished in the 1980s by the Thatcher administration becauseof ideological differences between central government and thelarge metropolitan authorities. It became necessary to createsome type of strategic planning that could be carried out by localauthorities. Unitary plans were set in place. They have a strate-gic (citywide) as well as a local part to them and are broadlycomparable to comprehensive or master plans in American cities.

8 In 1996 the government plans to scrap Scotland’s 10 Regional Councils and53 District Councils and replace them with about 30 single-tier authorities. Inthis proposal, Glasgow would lose two suburbs within its existing boundariesand would actually be smaller than it is now.


Planning departments in Scotland and England operate quitedifferently. Scottish local authorities have a responsibility towrite local (community) plans that should translate the broaderobjectives of the regional structure plan into more specific localobjectives and targets. Thus, cities like Glasgow and Dundee donot have citywide plans. In the case of housing, the local plansidentify sites for private market housing.

Evolution of local housing plans

When first introduced in 1977, local housing plans replaced asystem in which central government (the Department of theEnvironment in England and the Scottish Office in Scotland)decided on the merits of individual projects.9

Developments in Scotland in the 1970s pushed the field forward.Local Housing Needs and Strategies: A Case Study of the DundeeSubregion (Grant et al. 1976) is a high-quality housing planintended as a model for local governments. A second publication,Assessing Housing Needs: A Manual of Guidance (Working Partyon Assessment of Housing Needs 1977), demonstrates how au-thorities could undertake research work and develop their ownhousing plan.

In England, these plans are called housing investmentprogrammes (HIPs) and are produced annually; in Scotland theyare simply called housing plans and are produced every fouryears (with the exception of Glasgow, which produces plansevery two years). The housing plans have a five-year forwardperspective. Each plan contains two parts: (1) a description ofthe local authority’s strategy for tackling such problems as poorconditions and housing shortages, and (2) a checklist with tablesdealing with housing need, housing stock conditions, and thelocal authority’s capital program. Technically, the housing planis a bid from the local authority to central government forauthority to borrow funds for capital projects.10

Strong government guidelines are an essential feature of theseplans. Both the Department of the Environment and the Scottish

9 See Bramley, Leather, and Murie (1980) for an evaluation of the housinginvestment programmes of the late 1970s.

10 Scotland has a different method for allocating funds than England has. InScotland, the allocation to local authority is either for the public sector stockor for the private sector. In England, central government allocates funds to alocal authority, which then decides how it is divided up.

260 David P. Varady

Office provide advice on timetables, the form of submissions,policy content, and financial arrangements. A number of areas inthese guidelines have been emphasized to reflect key changes inthe housing environment (e.g., the right of tenants to buy theirunits). Therefore, one criterion used in evaluating individualplans is whether they address the priorities of ministers.

When we ask local authorities to prepare housing plans,we tell them [that] ministers are going to pay particularattention to things like homelessness and energy effi-ciency. We wrote to local authorities earlier this yearfor the current round of capital programs telling themministers would give particular attention in makingdecisions to a list of four or five items. They were thingsour ministers told us they wanted to stress. That wouldbe a really good indication of the ministers’ priorities.(Randall and Harrison 1992)

A second criterion is the quality of the strategies and the effec-tiveness and efficiency of the local authority.

Information comes from housing plans and also au-thority’s capital programs, which are submitted everyyear. They actually submit bids each year, which trans-late the housing plan into detailed proposals. We alsotake into account authority effectiveness and efficiency.That looks at things like how well authorities are tar-geting their resources, whether they carried out hous-ing conditions surveys, and whether they use thatinformation to target effectively. Also taken into ac-count are records of overspends and underspends.(Randall and Harrison 1992)

It is important to note, however, that the level of housing needcontinues to be taken into account in making funding allocations.

With new build, we look at demographic trends. We alsolook at the need for special-needs housing. We also lookat information on stock conditions. (Randall andHarrison 1992)

In the United States, if the city is an entitlement communityunder the CDBG program (or a participating community underthe HOME program), and it has a minimally acceptable plan, itis entitled to the same funding as the previous year as long asHUD’s allocation is the same. Britain’s process for evaluatinglocal housing plans is quite different.


We don’t do that in the sense of saying: Well, this planis totally inadequate, send it back, do it again. That isnot the way it works. We do form an opinion on howgood the strategy is and that is taken into account byministers when they form a decision. There is no bench-mark where you say: This plan meets it, this other onedoesn’t. (Randall and Harrison 1992)

There were high expectations for these housing plans when theywere introduced in the 1970s, but in the 1980s a sense of pessi-mism took hold. Most assessments concluded that the strictguidelines had a debilitating effect on the local planning process.When the Conservative Party took power in 1979, housing fundswere slashed11 and plans were used as a tool to control localspending. Cities failing to sell tenants their public stock werepenalized.

Chris Watson (1992), University of Birmingham, who partici-pated in the two Scottish studies mentioned above, is critical ofthe way housing plans have evolved.

The local housing plan has become a bid from the localhousing authority to central government for moreresources (i.e., for permission to borrow more) for thepublic housing program. OK, there may be some addi-tional sums of money earmarked for improvements inprivate sector older housing but the bulk of it has to dowith the building, the maintenance, and the repair ofpublic sector housing. That is one of the sad aspects toit. What we thought we were proposing was a reallycomprehensive approach to all of these matters, but itbecame something which was public finance driven,Treasury driven if you like.

By the 1980s, some housing professionals were disillusionedabout the value of housing plans in general and, consequently,provided the minimum required by law. Today, however, localauthorities are no longer judged by their ability to demonstrateneed but on the quality of their plans. Cities are competing withone another for limited housing allocations. Therefore, to getreasonable capital allocations, they must convince central gov-ernment that their plans are suitably prepared.

Recently, there has been a revival of interest in the housingplan, partly because central government is pushing local

11 Public expenditure cutbacks in housing actually began under the LabourGovernment in 1976.

262 David P. Varady

authorities to assume the enabler role. This role gives legitimacyto housing planning (i.e., how the housing agency analyzes needsand how it evaluates a full range of alternatives, includingprivate sector housing).

No doubt, Britain’s system of local housing plans (includingstrong central government guidance and competitive bids) worksagainst full comprehensive rationality. Local planners do nothave the flexibility and autonomy they expected to have whenthe system was introduced in the 1970s. A Scottish Office officialwho monitors local housing plans observed that the loss of com-prehensive rationality is compensated by the benefits of aquality-driven competitive system.

I am not sure that I agree with the underlying assump-tion that the ideal system is comprehensive planning,that someone sits down, takes a complete overview,assesses all of the different needs, does cost-benefitanalysis and all the rest of it, and sets out priorities. . . . Ican see all of the arguments for that. In practice, thingsdon’t often work out like that and the best way of get-ting results is often by encouraging people to come upwith different projects. . . . Certainly our ministers in aConservative government would not be expected tostand behind 100 percent comprehensive rational plan-ning. They are very much aware that human nature isabout doing things better for yourself and there is a lotof good in that. That would be their view. That is pre-cisely why they had competitions for particular projects.That [competition] brings forth good projects. . . . Ithink that they traded off a bit of enterprise with goodprojects with some loss in strict [rationality in the]addressing of housing need. (Randall and Harrison1992)

Glen Bramley (formerly at the University of Bristol, now atHerriot-Watt University, Edinburgh) credits local housing planswith stimulating thinking about a variety of housing options, notjust one or two as in the past.

In the old days, housing tended to be a mind-set whereparticular problems implied particular solutions. Unfithouses were demolished and replaced. People on wait-ing lists: we would build new houses for them. Owneroccupiers with housing in disrepair would get a repairgrant. There were one-for-one links between these


problems and particular solutions. . . . Now that isn’tquite the case anymore. If you have people who cannotget access to owner occupation, there is a range ofthings that you might do: low cost sales schemes,shared ownership schemes, housing association provi-sion using planning powers, using Housing AssociationGrants. You might try to do something with the existingstock. You might try to persuade some of your existingtenants to move out to create a vacancy and providethem with a form of discount. (Bramley 1992)

However, as Bramley points out, there is a downside to thisemphasis on competition and high-quality plans.

I think that it has encouraged quite a lot of local au-thorities to do a bit more planning, [do] a bit morepolicy analysis, do need surveys, [and] think aboutworking with the private sector. That is the positiveside of it. [On the other hand] I think that some de-prived areas will lose out. It is quite hard to work withthe private sector in a rundown urban area. It is quitehard to use the planning powers to get social housingbecause there is no land value there. It works best in asort of affluent area that does not have too many prob-lems, with [high] land values suitable for using theregulatory powers. Here the private sector would beinterested and the problems would not be overwhelm-ing. (Bramley 1992)

Brown and Carter (1990) note the absence of research on the wayhousing plans are carried out, including the technical issue ofrationality in needs assessments, the political issues of central-local relations, and the quality of interorganizational linkages.Duncan Maclennan (1991, 186) asks, “Does the new competitiveorder for the provision of social housing introduce real conflictsof interest which will make it impossible for a municipality bothto compete effectively and plan honestly?” In other words, somelocal authorities may resist cooperating with alternative land-lords such as Scottish Homes. This sort of competitive stancemay call into question whether the city is the only, or the best,strategic planner for the area. This article examines the extentto which cities have been able to prepare strategic plans in thistype of competitive environment.

264 David P. Varady

Comparative analysis of housing plans in fourbest-practice cities

To assess British local housing plans, a link should be estab-lished between the plans and different positive outcomes (e.g.,the production of affordable housing). Since this was not pos-sible, I will instead evaluate British plans along seven processdimensions: (1) mission, (2) scope, (3) housing analysis,(4) priority setting, (5) consultation, (6) management issues, and(7) neighborhood revitalization. Each of the seven sections beginswith a brief discussion of how that aspect of the planning processshould be carried out based on the housing planning literature(see Varady and Birdsall 1991). This is followed by a discussionof how well that aspect of the planning process is performed inthese four best-practice cities.

Glasgow, Dundee, Birmingham, and York have all producedshort, readable planning documents with lengths ranging from20 pages (York) to 77 pages (Dundee). Statistics are kept to aminimum and included only when necessary. The plans forBirmingham and York contain no tables; all statistics are incor-porated into the text. Glasgow’s plan is distinctive because it isself-promotional. It is the only one to contain color photographsand is obviously aimed at a wide audience.12

Mission

Given the declining financial resources available to meet housingneeds in American cities, the U.S. Conference of Mayors (USCM)recommends that local task forces (including the public, private,and nonprofit sectors) prepare such housing plans. According tothe USCM, this approach would have three advantages over thetraditional approach: increased community awareness of housingneeds, collaboration between public and private agencies, andpolitical support for implementing changes. The key point is thatthe plan should be used not only to provide a means for definingneeded action (the traditional role) but also to provide the basisfor collaboration and to focus community and political actionsthroughout the year. In both the British and American contexts,this means that the plans should be more than bids to centralgovernment. In fact, the four British plans, especially Glasgow’s,were serving far wider purposes.

12 The city has even produced a shrink-wrapped, folded, six-page summarywith photos and charts to ensure that the message is widely disseminated.


Certainly, all four British cities take the bidding function seri-ously. The housing officials interviewed believed that a city witha poor plan would be rewarded with a lower capital allocation.Consequently, all four cities devote a large effort to preparingthe plan while receiving assistance from other agencies, espe-cially their local planning departments. At the beginning of theirHIPs, both Birmingham and York emphasize that they areclosely following guidance from the Department of theEnvironment (i.e., central government) and that their plansbenefited from suggestions in the Audit Commission’s 1992monograph Developing Local Authority Housing Strategies.13

Glasgow’s sophisticated, well-designed plan reflects its broadmission.

[The plan] is us positioning ourself, it’s a learningprocess, it’s an exploring process, it’s negotiating withpartners, it’s a trading device, it’s a marketing ploy, it’ssomething to hit people [central government] over thehead with. This is a document that has been widelycirculated. We give this to builders, influentials inhealth boards, people in economic development net-works. [When we distribute the document we are say-ing:] this is who we are, this is what we are interestedin doing and the types of projects we are involved with.. . . These are the conditions we have, do you want tocome and play? (Brooke 1992)

Glasgow’s flashier plan is particularly important in improvingrelations with a number of agencies that the housing departmenthas not had particularly close ties to in the past: (1) the ScottishEnterprise network, economic development agencies, (2) theStrathclyde Regional Council, primarily the Roads and Educa-tion Departments, and (3) Scottish Homes (Webster 1992). Be-cause these three agencies also produce glitzy plans, the housingdepartment feels the need to produce one of comparable qualityand sophistication.

Robert Towner (1992), York’s Director of Housing Services,highlights another function of the housing plan: educating poli-ticians and citizens about options that deal with housingproblems.

13 The Audit Commission oversees the auditing of local authority finances andpromotes “value for money” from municipal spending. The Commission’s workhas supported Conservative Party claims about waste in local government buthas also criticized the arbitrary and complex nature of central governmentinterventions (Stoker 1988).

266 David P. Varady

However if [the bid] is all [the plan] is seen as, [theplan] would be very limiting. The plan also plays arole in highlighting local choices, which is the caseregardless of the size of the “cake” available to localgovernment.

Dundee’s plan benefits from taking its bid to the Scottish Officeseriously. The city maintains a high-quality forward planningstaff who remain heavily involved with the plan throughoutthe year. Dundee, therefore, stays flexible to respond to newguidelines as they become available (e.g., government programsfor heating or a program for home security). As a result, Dundeehas done unusually well in obtaining funding from sources suchas Scottish Homes. This city’s high-quality housing plan broughtin an estimated additional £25 million beyond what it wouldhave received with an average plan.

Scope

The growth of the local housing function in both Britain andAmerica has increased the need to link housing and communitydevelopment with economic development activities. Cities needto better “coordinate economic development with housing im-provement, for a city’s economic future depends on its livability”(Widner 1980, in Nenno 1989, 5). Sternlieb and Listokin (1985)make the same point using different terms: Housing plansshould have a broad mission, covering both shelter and post-shelter housing needs. “Shelter” refers to the need for adequatephysical housing for low- and moderate-income persons. “Post-shelter” deals with the need of middle- and upper-income peoplefor “housing to serve also as a symbol of prestige and a vehiclefor capital accumulation” (p. 385). Most American housing plansdeal exclusively with “shelter” issues, but there are some excep-tions, such as Hartford’s Housing Component of the Comprehen-sive Plan and Cincinnati’s Housing Blueprint, that show that itis feasible to address both needs (see Varady and Birdsall 1991).

All four best-practice plans do a good job in discussing shelterneeds but do a less than adequate job in addressing post-shelterrequirements. Dundee’s plan, for example, highlights a long listof low-income housing issues: the high rate of deterioration inthe public housing stock, especially in the peripheral housingestates; the continuing need to repair and upgrade low-incomeownership stock in the inner city; the desire of many tenants tobecome owners; and the unmet needs of special groups, such asthe elderly, the disabled, the mentally handicapped and mentally


ill, the homeless, and young single persons. But the plan virtu-ally ignores private market issues.

British planners give so little attention to the middle-incomesection of the market because of priorities set by politicians.Most housing policy makers and practitioners in Britain wouldassert that, given scarce resources, their greatest priority shouldbe the poor since they have less opportunity to find their ownsolutions in the market. Furthermore, the thinking of localauthority housing officials is also influenced by their landlordand management responsibilities. Council housing still consti-tutes a fifth of all stock, and this proportion is much higher incities like Glasgow. A final reason why market-rate issuesreceive short shrift is because of the division of responsibilitybetween the housing and the planning departments. Historically,the former have been responsible for managing social housingwhile the latter have had the responsibility of addressingmarket-rate housing. Birmingham does a good job of coordinat-ing the activities of the housing and planning departments in itsUnitary Development Plan (UDP; Birmingham City Council1991a).14 The housing section of the UDP contains a detaileddiscussion of housing problems, markets, and policies at both thecity and community levels. (The UDP is discussed in more detailin the section on management issues.)

However, Glasgow illustrates a disjuncture between the effortsof the two departments. Specifically, Glasgow’s Housing Planfails to acknowledge the importance of two major planning de-partment reports on private market housing: City Planning Aimsfor the Next Decade (Glasgow City Council 1991), which looks atthe role of housing programs in an effort to promote the eco-nomic regeneration of the city, and A Review of Private Housing

14 UDPs are developed in response to guidance from the secretary of state,who issues a commencement order indicating that the authorities can preparethe plan. The process for creating Birmingham’s UDP was somewhat uniquecompared with other large British cities. The districts and the surroundingauthorities in the Midlands held a conference in 1987, which produced a reportsubmitted to the secretary of the environment. He used the report to draw updraft and later final guidance. A lot of the key land use issues, including thedevelopment of a greenbelt, were agreed to by the local authorities. As aresult, the guidance for the metropolitan area provided by the secretary ofstate was in most respects what the authorities wanted. As of 1992, plannerswere moving toward similar guidance for the region—which consists of theBirmingham metropolitan area and four “shire” counties. UDPs are expectedto comply with planning guidance from the central government. If the recom-mendations in a planning document are different from planning guidance,then a public inquiry process is necessary. Local governments tend to lose oncontroversial issues.

268 David P. Varady

Market Activity in Glasgow 1991 (Glasgow City Council 1992b),which argues that the drop in private housing starts in the citycontributes to housing-led outmigration (i.e., the inability ofmiddle-income families from the city to find attractive suburban-type detached and semidetached housing).15

Glasgow’s plan is, however, noteworthy because of its emphasison population stabilization, a subject that colors the entire plan.Glasgow has experienced an overall decline in population (from774,000 in 1981 to 672,500 in 1991) and increased vacancies inthe peripheral housing estates. The plan proposes a continuationof the city’s Area Renewal policy, which involves thousands ofcouncil flat demolitions, tenure diversification (building private-ownership housing in areas that had been exclusively rental),and improvements in infrastructure, all aimed at making theseareas more attractive. However, the plan sidesteps any seriousdiscussion of how to halt middle-class outmigration. Althoughhousing department staff are willing to discuss the problem ofmiddle-class outmigration, they are unwilling to use the terms“middle income” or “middle class” in the plan, because theyperceive the terminology as a “put-down” of low-income groups.Hence the plan uses such phrases as “catering for those whohave a choice,” “attracting people in employment,” or people on“above average incomes.”16 It is hard to imagine how Glasgowcan begin to develop effective policies to address housing-ledmiddle-class outmigration unless this subject is discussed forth-rightly and until the efforts of the housing and planning depart-ments are better coordinated.

Housing analysis

The literature identifies three approaches to measuring housingneeds: (1) housing market analysis, which compares futuresupply with future demand to indicate the amount of construc-tion required; (2) needs assessment, which relies on professionalstandards typically drawn from the census to measure problemssuch as overcrowding, substandard housing, and rent burden;and (3) preferences or aspirations of householders, which can be

15 I found out relatively recently that A Review of Private Housing MarketActivity was published after the 1992 plan and consequently could not havebeen acknowledged in the plan. This, however, does not alter my main conclu-sion that there was inadequate coordination between the planning and hous-ing departments.

16 Letter from Glasgow Housing Department official choosing to remainanonymous.


obtained directly through surveys or indirectly through clientanalysis (Reiner, Reimer, and Reiner 1963). Grant et al.’s 1976housing plan for Dundee, which was supposed to be a model forBritish local housing plans, is unique in encompassing all threeconceptions of need. Seven different analyses were included inthe plan: (1) the condition of the older housing stock based ona survey of exterior housing conditions and maintenance;(2) estimates of the number and characteristics of housing unitsfor special populations; (3) estimates of the land available fordevelopment; (4) changes in the housing stock, including changesin the incidence of substandard conditions; (5) population andemployment projections; (6) housing preferences and aspirationsbased on a client analysis of waiting lists and transfer recordsfor public housing; and (7) unmet needs for special housing basedon a survey of the elderly.

While none of the four plans achieved the analytic comprehen-siveness of Dundee’s 1976 plan, they did nevertheless displayseveral impressive features. In general, the four plans used awider variety of data sources (e.g., housing conditions surveys,waiting lists for public housing, lists of people applying as home-less under the Homeless Persons Act, the census) to estimatehousing needs than their American counterparts, which relyalmost exclusively on census data.17 All four cities had conductedrecent surveys of conditions of the public housing stock and usedthem in their plans. Both the Scottish Office and the Departmentof the Environment provide detailed guidance on how thesesurveys should be carried out. In addition, Birmingham had usedthe 1988 West Midlands House Conditions Survey, which coversthe private sector.

The housing plans for Glasgow and Dundee show how valuablewaiting lists for housing can be for planning purposes.18 InGlasgow, the existence of a disproportionately large number ofsingle parents on the waiting lists highlights the residualizationof the public housing stock with demand increasingly limited tothose without other choices. Dundee’s analysis of acceptancesand refusals shows how public housing clients are becomingmore discerning. For example, 40 percent of all applicants will

17 Although the Department of the Environment and the Scottish Officeprovide more detailed guidance on policy matters than HUD does, the Britishagencies allow more flexibility on data requirements. Local authorities decidewhat tables to use. As John Randall (1992) from the Scottish Office noted,“most of the authorities have well-qualified staff who can do that type ofwork.”

18 However, central government officials often discount the significance of thisevidence.

270 David P. Varady

consider only “cottage-type” housing, most applicants are veryspecific about the areas in which they are willing to be housed,and few householders want to live in tower blocks.

Glasgow’s evolving approach to measure the need for shelteredhousing is of interest. The 1992 plan used the traditionalapproach recommended by the Scottish Office, with standardsbased on fixed percentages of the population.19 The problem withthe standards is that they do not take into account the fact thatthe frail elderly can be helped in ways other than shelteredhousing. A new approach (see Glasgow City Housing 1993),which will be used in future plans, estimates the requirements ofindividuals for different sheltered forms of care (i.e., servicesdelivered to the home in addition to sheltered housing). The factthat the new technique yields lower estimates of need for shel-tered housing than was the case for the 1992 housing plan isimportant since sheltered housing is quite costly.

All four plans use housing market analysis (HMA) and includeprojections of the city’s housing requirement, along with esti-mates of the amount of public land that should be released forprivate sector development. Only summaries of these computa-tions, prepared by the county or regional planning departments,are included in the housing plan. Little information is providedon the methodology or the type of housing required. In general,there is little connection between HMA and the remainder of theplan. York’s HIP highlights the limited effectiveness of this typeof HMA. According to the plan, an additional 3,000 units wouldbe required between 1991 and 2006. Since insufficient land isavailable within York to accommodate all of these units, someland would have to be found in the suburbs. The plan does not,however, address the problem of how to obtain suburban coop-eration, given the resistance of middle-income suburban resi-dents to living close to lower-income households.

Housing officials in Glasgow, Birmingham, and York emphasizethe need for practical HMAs that will identify demand-supplymismatches for particular population groups and particularareas.

At present, there is no way to tell local authorities whatlevel of owner demand is sustainable on the peripheral

19 The housing department used the following standards based on a 1991Scottish Office circular: (1) very sheltered housing, 20 dwellings per 1,000;(2) sheltered housing, 46 dwellings per 1,000; and (3) medium-dependencyhousing, 80 dwellings per 1,000.


estates. For example, what would happen if the Hous-ing Department demolished part of Easterhouse [one ofGlasgow’s peripheral estates] and gave the land toprivate developers? Would they build there and if sowhat size and types of homes? What other conditionswould be necessary for development? Would the areaneed a private school? At present it is not possible toobtain answers to these questions and the ScottishOffice is not much help. (Milne 1993)

A 1992 best-practice guide prepared for Scottish Homes by theCentre for Housing Research (now the Centre for HousingResearch and Urban Studies) at the University of Glasgowdemonstrates how to carry out this more practical type of HMA.This report evolved into Local Housing Market Analysis andPlanning in Scottish Homes: A Best Practice Guide (ScottishHomes n.d.).

Priority setting

Given limited resources, the housing plan should specify whichpopulation groups will and which will not be served and shouldprovide the reasons for these choices “ranging from perceivedgreatest need to obtaining the most complementary benefits forneighborhood stabilization” (USCM 1985, 51).

The task of setting priorities by program area and allocatingfunds among these categories is a daunting one. For example, ifit is estimated that a city has 1,000 households in substandardunits, what proportions should be helped through public housing,subsidies to owners, vouchers, programs aimed at transformingrenters into owners, and so on? Formulating priorities requirestaking into account social science evidence on the types of pro-grams most appropriate for particular population subgroups,judgments on the importance of different policy goals (e.g.,choice, the opportunity for homeownership), the preferences ofclients, the capacity of the system, and programs that are cur-rently available. The literature offers little guidance as to how todetermine these distinctions. In the 1970s, there was a greatdeal of optimism and enthusiasm for techniques like cost-benefitanalysis and program plan budgeting for comparing differentprogram areas. Grant et al.’s 1976 plan for Dundee, reflectingthis optimism, uses cost-benefit analysis to determine whether toimprove or replace older buildings. Improvement was shown tobe preferable to replacement. However, in recent years,

272 David P. Varady

enthusiasm for techniques like cost-benefit analysis has wanedbecause of the difficulty in measuring noneconomic variables.

To develop realistic priorities, localities need to take into accountfunds that are likely to be available. Varady and Birdsall’s 1991review of American local housing plans showed that they did notdo a very good job of recognizing financial constraints. For ex-ample, Hartford’s planning staff estimated public housing sub-sidy needs at $103 million between 1986 and 1990 but did notindicate how these needs would be met. Hoben and Richardson’s1992 study, based on the first year of CHAS submissions, notedweaknesses in priority setting: “In the majority of the sampleCHASs, priorities, measurable objectives, and implementationplans were too general to permit careful monitoring and perfor-mance assessments” (p. 11). The authors called for more atten-tion to priority setting in future CHAS documents. However,there was no comparable report on 1994 submissions. In itsguidance to localities for preparing the new Consolidated Plans,HUD is stressing the importance of explaining why differentgroups (e.g., low-income renters) are chosen. It is still too earlyto say how well localities will carry out this responsibility(Scalfani 1995).

As a result of central government guidance, British local housingplans are taking a more realistic stance on financial constraints.Until recently, the nature of the bidding system worked againstthe setting of priorities. Cities simply estimated the cost ofmeeting all housing needs (public and private) and translatedthis cost into an annual allocation. Cities were concerned thatthey would be penalized for low bids, which acknowledged lim-ited resources, by reduced allocations. For example, Glasgow’splan estimated that it would cost £1.6 billion to meet councilhousing stock requirements. It, therefore, requested £160 milliona year over a 10-year period. Glasgow actually received a£95 million allocation.

England has moved faster than Scotland toward the implementa-tion of realistic bids. The Department of the Environment askedlocalities to prepare three bids: one based on no change over theprevious year, a second based on a 10 percent increase, and athird based on a 10 percent decrease. These realistic bids willprobably make cities more serious about priorities, but officialsare frustrated by their inability to tackle large unmet needs. Inits 1992 HIP, Birmingham advocates a needs-based programcosting £200 million a year, which would address the bulk ofsocial housing needs. Carrying out the three requested cheaperbids would, according to city officials, mean that “essential worksto [the city’s] own housing stock and in the private sector, many


of which seek to address statutory requirements, would not becarried out” (Birmingham City Council 1992, 39). Therefore,Birmingham presents a “constrained” bid of £84 million based onthe “plus 10 percent” alternative but refuses to bid based on theother two scenarios. Thus, while the realistic bids represent animprovement over the past, the constraints facing housing de-partments will limit the value of the new approach.

British housing planners face a serious dilemma: They areexpected to prepare high-quality plans, but funding is static.Consequently, there is a tendency to make changes along themargins rather than to take a fresh look at priorities each year.

Although you are on an annual allocation system, thereality is that programs that work require a seriouscapital investment. These are the ones that you getheavily involved in and they roll on. So, in any year,most of what you plan to do the next year is largelycommitted by the programs you started implementingin the past. (Bramley 1992)

Comprehensive rationality in priority setting is further limitedby government policies and allocations as well as by statutoryobligations (e.g., meeting the needs of the homeless and provid-ing grants for repairs for private stock housing), which leavesfew resources for other needs.20

Each one of the four plans contains an impressive discussion ofpolicies and plans the local authority intends to implement, aswell as some discussion of the priorities of each plan. Glasgow’s,for example, emphasizes area renewal on peripheral housingestates. Missing from all the plans, however, is an indication ofhow the priorities reflected in the plan were developed.

Cost-benefit analysis (now called “option appraisal” in Britain)has not achieved the expectations promised in the 1976 Dundeestudy. Typically, it is used only for individual improvementgrants on particular sites (e.g., demolition versus rehabilitation).Even at this level, option appraisal is typically only a theoreticalexercise because housing officials have limited choices.

Unless you have a lot of evidence of deterioration in agroup of buildings, you are led in the direction of

20 For example, even if a local authority wanted to expand its supply of councilhousing significantly, it would be deterred by the right-to-buy policy and lowgovernment allocations for the public stock.

274 David P. Varady

improvement regardless of what the statistics show.Now there are major restrictions on demolitions makingoption appraisal analysis of limited value. (Towner1992)

In the past, politics has played a significant role in the establish-ment of priorities among policies and programs. For example, inthe 1970s, Glasgow developed a system for measuring housingneed based on poverty and deprivation, which benefited the city’speripheral housing estates. The system reflected the LabourParty’s desire to win back voters, including those living in coun-cil housing, who had deserted to the Scottish National Party.This politicizing approach can no longer be sustained becausethe demand for these estates is negligible.21 In its new system,the housing department sets priorities based on housingconditions in different parts of the city.22 Housing officials be-lieve this more realistic approach for setting priorities will resultin a shift in funding from the peripheral estates (where centralgovernment agencies such as Scottish Homes and the GlasgowDevelopment Agency are expected to play a key role) to othersections of the city. The new system has been well received bythe Scottish Office, but the question remains whether localpoliticians, some of whom would see their constituencies losefunding, would accept the changes in allocations based on thenew process.

Consultation

With the declining amount of resources available from the fed-eral government to address housing needs, it is increasingly

21 This story implies that there are significant differences between British andAmerican local governments in terms of style of politics and managementregarding housing issues. That is, it is likely that British local authorities aremore dominated by party-based politics and that the parties are themselvesmore ideologically based.

22 In its new system, the housing department has earmarked four blocks offunds for specific things: (1) special needs housing, (2) area renewal for theperipheral estates, (3) a standard funding block (to upgrade housing thehousing department knows it will retain), and (4) a citywide block for miscella-neous things like asbestos removal and energy efficiency programs. Each ofthe housing department’s 16 district offices adds up the need per block, andthe central office adds up the total need for the city broken down by block.From this process, the city might request that 60 percent of the borrowingfinancing go for the standard block. A pro rata share would be allocated todistrict offices.


important that the private and nonprofit sectors be involved inthe preparation of these plans. According to the USCM (1985),the involvement of these two other sectors increases communityawareness of housing needs and stimulates collaboration be-tween public and private agencies and between public and politi-cal support for implementing changes. The importance of citizenparticipation in the planning process, regardless of the func-tional area, is now widely accepted. Call’s 1977 case study ofSpokane, Washington, shows how citizen participation can beused to provide information on housing needs as part of theplanning process. Goetz and Geiger (1995) show that the CHASs’greater requirements for citizen participation compared withthose of the HAPs did lead to heightened awareness of the hous-ing planning process and local and federal housing policy. It alsoled to more assistance to low-income renters.

In general, Britain seems to be behind the United States incitizen participation activities. None of the four cities use townmeetings or other forms of mass consultation with citizens toprepare their housing plan. Officials in Glasgow and Birming-ham cited a lack of time for involving citizens. York’s housingdirector cited the difficulty of obtaining useful input at thecitywide level because tenants had difficulty understandingabstract housing issues.

When you are asking whether there should be onescheme or two schemes or whether 15 percent or25 percent of the housing allocation should go to aparticular program, this is difficult, in terms of gettinguseful participation. The experience of the HousingDepartment is that where residents are directly in-volved in a housing issue in their daily lives their con-tribution is much more obvious and useful. Otherwiseon the more abstract issues, the city should rely on thedemocratic process [i.e., politicians], while welcomingparticipation where it is relevant. (Towner 1992)

However, forms of citizen participation other than mass consul-tation are used. First, Glasgow and Birmingham have preparedpopularized 3- to 6-page summaries of the plan, which are dis-cussed with residents in a variety of settings. Second, Glasgowand Dundee have a system of district housing plans based onclose involvement of residents. Third, York conducts tenantconsultations with each of the city’s 17 tenant associations, andtogether, through the federation of associations. Finally, all fourcities use market research to solicit residents’ views.

276 David P. Varady

Glasgow and Birmingham represent two ends of the spectrum interms of involving home builders. The Scottish Building Employ-ers’ Federation (SBEF, the main organization of builders) hasnot been involved in the preparation of Glasgow’s plan because itdoes not deal enough with the release of land for private hous-ing, which is the key concern of the SBEF (Hopkins 1992). Incontrast, Birmingham, in preparing its HIP, consults with boththe home builders association and the housing association move-ment (Birmingham Housing Association Liaison Committee). Infact, the latter prepares a statement for the city, which is incor-porated into the HIP.

Management issues

Nenno (1989) highlights three key administrative issues thatshould be addressed in local housing plans. First, the housingstrategy should be linked to the city’s long-range comprehensiveplan as well as its short-range development plan. Second, theplan should recognize the expanded role of nonprofit housingorganizations, keeping in mind that these nonprofits vary intheir administrative capacity to implement needed projects.Third, increased housing and community development activitymay indicate the need for a new agency or the reconfiguration ofexisting ones. American local housing plans have devoted limitedattention to such administrative issues. The King County (Wash-ington State) 1987 Affordable Housing Policy Plan highlights thecontinuing importance of the comprehensive plan by recommend-ing affordable housing goals for the different community plans inthe county. Dallas’s Housing Task Force pointed out severalmajor weaknesses in the management of city housing programsand recommended a reorganization of the city’s Department ofHousing and Neighborhood Services. There has been no system-atic research into whether the CHASs examined administrativecapacity.

In general, the four best-practice British plans devote moreattention to administrative issues like coordination and manage-ment performance than the American ones above. Of the fourplans, Glasgow’s devotes the most attention to coordination. Theplan emphasizes a key dilemma: Although the city is expected tosolve low-income housing problems, much of the funding comesthrough other agencies (two-fifths through Scottish Homes). Thisreality led to the development of the Joint Framework for Invest-ment, which lays out principles of partnership between the cityand Scottish Homes (e.g., full and effective consultation betweenthe two agencies on national housing policy issues that affect


Glasgow). However, Glasgow officials recognize that becausethere are key differences in the goals pursued by the two sides,informal methods of consultation can go only so far. Therefore,the plan advocates that central government change some of theobjectives pursued by Scottish Homes, modifying what the citysees as an overemphasis on owner-occupation and on movingpeople out of council housing.

Despite the plan’s stated emphasis on coordination, Glasgow’shousing efforts suffer from the lack of linkage between thehousing and planning departments. That is, the plan does notindicate the extent to which it is compatible with the local plansof the planning department.23

Dundee’s Housing Plan also demonstrates a lack of coordinationbetween the planning and housing departments. There is noindication of how the housing plan is linked to the city’s localplans, nor is there any recognition of work being done by boththe planning department and the economic development depart-ment to spur private market housing as part of an effort topromote economic regeneration for the city as a whole. On theother hand, Dundee’s Housing Plan does indicate a willingnesson the part of the housing department to admit administrativeshortcomings. For example, the Homeless Service came undersome criticism in the customer survey, and in response, theCouncil identified areas in which the service could be improved(e.g., by creating a network of temporary furnished accommoda-tions in a wide range of areas in the city).

Birmingham’s HIP demonstrates strong concern for managementperformance in council housing stock. In an appendix to the HIP,a number of indicators of improved performance are presented:(1) a drop in the vacancy rate and a reduction in the vacancyperiod, and (2) a strong performance in collecting rents owed(94 percent, higher than the target of 90 percent).

As mentioned earlier, Birmingham’s HIP shows a strong workingrelationship between the housing and planning departments. Infact, housing planning in Birmingham can only be understood inthe context of two plans, the HIP and the planning department’sUnitary Development Plan (Birmingham City Council 1991a).Housing is one element of the UDP; the document contains adetailed discussion of housing problems, housing markets, and

23 Unlike cities in England, Scottish cities do not have citywide comprehensiveplans. Instead they have a large number of community plans that locateproposed public improvements and housing projects.

278 David P. Varady

housing policies at both the city and community levels. The UDPis a planning framework for the development of the city over thenext 10 to 15 years. Staff from the planning and housing depart-ments devote considerable time to making sure that one plandovetails with the other.

Recent policy guidance from the Department of the Environmenthighlights the growing role of the UDP and the private sector inaddressing this need. Government guidance recommends thatthe planning department set a target for the provision of afford-able housing throughout the city based on evidence of need andalso set targets for specific sites based on evidence of need andsite suitability. Thus, the planning department is expected tonegotiate with developers for the inclusion of a percentage ofaffordable housing in such schemes, both on sites allocated forhousing and on other sites (Birmingham City Council 1992).These guidelines were issued too late to be part of the 1991 UDP,but it is anticipated that such targets will be included in thefuture review of the UDP.

Of the four plans, York’s HIP best demonstrates the localauthority’s effectiveness as a manager and an enabler. Eachyear, York officials, working with an opinion research firm, holdin-person interviews with a 10 percent sample of council tenants.The results present an impressive record of improvement intenant satisfaction. In 1965, two-thirds of all tenants weresatisfied with the physical condition of homes in the city, but by1991, satisfaction levels had risen to 90 percent. However, cus-tomer surveys are not the only evidence of management quality.Performance reports (mentioned in the HIP; York City Counciln.d.) highlight York’s success in reletting vacant apartments.York performs better than established national targets andachieves average turnaround within two to three weeks.

York highlights its effectiveness as an enabler by giving localauthority land gratis to housing associations, thereby stretchingHousing Association Grants made available from central govern-ment. This magnanimity is not without certain expectations. Asa quid pro quo, York requires housing associations to provide thecity with 100 percent of the nominations (the normal arrange-ment is 50 percent), which allows the city to target those ingreatest need.24 In another section, the HIP endorses centralgovernment’s policy of encouraging the provision of affordablehousing by private developers on land sites in the city and

24 Housing authority–housing association relationships are discussed in detailin Fraser’s 1991 volume.


describes how the effort is being realized in the city’s historicmedieval core, through the use of empty accommodations aboveshops.25

Neighborhood revitalization

For a local housing strategy to be effective, it must be based onan adequate understanding of urban neighborhood dynamics andof when and how to intervene in a housing market to achievestrategic objectives. Unfortunately, the guidebooks available toAmerican planners are of limited utility in developing a neigh-borhood housing strategy. For example, the 1980 guidebook ofthe National Community Development Association understatesthe tensions and contradictions among the goals a local planattempts to promote (e.g., neighborhood stabilization, reductionsin discrimination, effectively mixed low-, middle-, and upper-income housing, and economic development). American localhousing plans have handled neighborhood issues differently. TheUrban Institute’s Detroit study (Rasmussen and Struyk 1981)identified communities that were beginning to experience declineand targeted them for revitalization, which is an example of thetriage approach. In contrast, the Enterprise Foundation’s 1986Chattanooga Plan did not dismiss even the worst neighborhoods.Hartford’s 1986 comprehensive plan focused on low-incomehousing dispersal and fair-housing issues (City of Hartford1986). Cincinnati’s Housing Blueprint largely ignored geographybecause the location of low-income housing is so politicallysensitive (Varady and Birdsall 1991). Those involved in thepreparation of a land use plan for the East End of Cincinnati, alow-income community near the central business district, com-pletely ignored the Housing Blueprint highlighting the absenceof a citywide revitalization strategy (Varady and Raffel 1995).

The four British plans vary in how well they handled geographicissues. Birmingham’s discussion of neighborhood revitalization isthe most thorough.26 The HIP briefly describes the different areainitiatives, many of which are funded by competitive grantprograms run by the central government (e.g., City Challenge,

25 Implementation of this strategy may be difficult, according to planningofficials. Exhortation and negotiation are the only mechanisms that theplanning department has to achieve these goals when it allocates land for newhousing. As a planning official noted, “when you are sitting across fromdevelopers, admonishment does not get you very far” (Cullen 1992).

26 The neighborhood emphasis is much more evident in Birmingham’s 1991HIP (Birmingham City Council 1991b) than it is in the 1992 version.

280 David P. Varady

Estate Action) and also discusses the policies underlying theseinitiatives (public-private partnerships, coordination among cityagencies, consultation with community residents, and “planningin the round,” that is, looking at the relationship between hous-ing, transportation, shopping, etc.).

Birmingham has embraced the role of grant writer in seekingcompetitive grants for a variety of neighborhood initiatives.Robert Blackaby, director of research in the housing department(1992), describes Birmingham’s policy regarding these grants.

[Birmingham’s] HIP sets out a policy of maximizingresources and going after government initiatives. CityChallenge is an example. Some cities reject going afterthe grants. Although we are [of] a different politicalcomplexion than the central government, Birmingham’spolicy is to go after whatever money we can.

One possible danger of participating in such grant competitionsis that cities may lose control of these initiatives to centralgovernment officials. That is, the objectives for a particularrevitalization initiative may reflect the criteria listed in a grantcompetition and may differ from what local politicians want todo. This concern has not been borne out by Birmingham’s Heart-lands project, an effort to bring jobs and better housing to adeteriorating area to the east of the city center. The Heartlandsproject is being carried out by a development corporation like theone in London’s Docklands. A development corporation hasgreater power than a city agency and can take advantage ofspecial central government grants. Birmingham has not experi-enced the same administrative and political problems thatLondon’s Docklands has because the board running the Heart-lands project was established to maintain the city’s initiative.27

The fact that these initiatives are under city control leads to ahigh degree of congruity between neighborhood initiatives andpolicies laid out in the HIP.

Birmingham’s treatment of geography can be contrasted withGlasgow’s, which does not discuss neighborhood initiatives in asmuch detail. Furthermore, many of Glasgow’s key neighborhoodinitiatives are accountable not to the city but rather to centralgovernment agencies in Edinburgh, which leads to incongruence

27 For example, the city’s planning department has the contract to work in theHeartlands. The relations between the city council and the Heartlands Boardare more intertwined than in earlier projects. (For a more detailed discussionof the Heartlands initiatives, see Wood 1993.)


between the projects and citywide housing policy. The 1991Housing Plan does not discuss these weak linkages.

For example, the director of Crown Street Regeneration, a flag-ship project of the Glasgow Development Agency (GDA) that isbeing built on a slum clearance site in the Gorbals area in theinner city, concedes that the project is inconsistent with thecity’s policies and makes no attempt to follow the plan (Galloway1992).28 One example of this dissonance is a dispute between theplanning department and the GDA concerning housing type.The planning department wanted to build low-rise, family-typehousing. The GDA held a competition and decided on atraditional tenement solution, one that echoes the original19th-century row houses (see Glasgow Development Agencyn.d.). While the GDA’s vision was eventually adopted, planningdepartment staff doubt whether such housing will attract fami-lies with children. Furthermore, Crown Street Regeneration mayvery well be undercutting the city’s Area Renewal Policy men-tioned above. Castlemilk,29 one of five areas where the housingdepartment in conjunction with its partner agencies is demolish-ing thousands of units (thereby providing the basis for tenurediversification), is in direct competition with Crown Street forpotential residents. At the very least, these conflicts should havebeen discussed in the housing plan.

In general, the Dundee Housing Plan is not a geographicallyoriented document, though the main exception to this generaliza-tion is the plan’s detailed discussion of the Whitfield Partner-ship, a regeneration effort in one of Dundee’s peripheral housingestates funded by the central government. The District Councilhas had a number of concerns about the initiative. Would it beresident-led or government-led? Would the partnership drawfunds from other areas? Would the Scottish Office require thatthe District Council’s borrowing for projects be concentrated onWhitfield? Would the Partnership acknowledge the DistrictCouncil’s role in the effort? It is still too early to answer thesequestions. There is, however, a growing consensus that the

28 Scottish Enterprise, headquartered in Edinburgh, manages a network oflocal enterprise companies that are involved in local economic development,training, and environmental improvements. Thus, despite its name, GDA is ineffect a central government agency.

29 In 1988, the secretary of state for Scotland released the publication New Lifefor Urban Scotland, which set up four major initiatives for peripheral councilhousing estates, the largest of which is the Castlemilk Partnership (over20,000 residents and approximately 10,000 houses) (see Scottish Office 1988,1990).

282 David P. Varady

initiative has achieved physical improvements in Whitfield’shousing and the environment (Munro 1992). Prospects for eco-nomic and social regeneration are still uncertain.

As is the case with Glasgow and Dundee, York’s HIP does nothave a strong neighborhood focus.30 The success story in York’shousing effort is Aldwark. Although it was implementedmostly through the planning department, it should have beenhighlighted in the HIP, and an update of the project should havebeen provided. The Aldwark project grew out of Lord Esher’sReport, which identified a few cities, including York, for physicalregeneration (Esher’s York 1969). The Esher Report to the cen-tral government focuses on how to bring people back to thesecities. York is a paradigm for establishing conservation areaswithin cities.

Aldwark (an area within sight of York’s historic Minster) hadexperienced decline due to the destruction of 19th-century back-yard industries and planning blight (e.g., uncertainties resultingfrom a proposal to widen medieval streets). The regenerationproject involves family homes as well as apartments, both builtby the private sector in a manner consistent with the historiccharacter of the area. The city particularly wished to encouragemiddle-income people to move back into the city’s historic center.

The HIP should have included an update on the Aldwark, high-lighting the necessity for a close connection between the housingand planning departments to carry out the city’s broader housingstrategy.31 A discussion of this unique housing project wouldmake the plan more readable and widen the audience to includeplanners, designers, and developers in Europe, the UnitedStates, and elsewhere.

30 Two area initiatives in the HIP are not fleshed out sufficiently to explainjust what the city is trying to do in these areas. One involves a proposal totarget renovation grants to areas with high concentrations of elderly persons.No map is included to indicate what this targeting would mean spatially. Thesecond involves the city’s planned investment in an Estate Action project. Thedescription of this initiative is so short and sketchy that the reader has nosense of what problems exist on this estate (presumably they go beyond theneed for physical improvement) and what the council is doing to address theseissues.

31 If a project similar to Aldwark were carried out in the future, city leaderswould probably insist that it have a much larger social housing component. Inother words, the housing department would have a key enabling role in such aproject. Such a discussion—what the city learned from Aldwark—would seemhighly appropriate in a future HIP.


Postscript

Since the completion of my sabbatical research in 1992, housingofficials and planners have initiated a number of changes toaddress weaknesses that I first observed in the system. First,the Scottish Office, following the lead of the Department of theEnvironment, has introduced more realistic bids. Specifically,1993 guidance recommended that “the selection of priority issuesshould reflect . . . a realistic view of resource availability”(Bannister 1993).32

Second, the Scottish Office indicated through a guidance docu-ment (Scottish Office, Environment Department 1993) howoption appraisal could be used at both the strategic and theproject levels. That is, once an authority has determined what itsaims and objectives are, it would then carry out a six-stageoption appraisal on how to meet these goals: (1) definition ofbasic aims, (2) identification of housing objectives, (3) identifica-tion and assessment of alternative strategies, (4) selection ofstrategy and definition of strategic investment priorities,(5) implementation and project identification, and (6) optionappraisal at the project level. Local authorities attempting tofollow the guidance should find useful an extremely well writtenthree-volume report prepared by Pieda Planning (1993), whichprovides examples and case studies that should convince housingdepartment staff in localities throughout Scotland (and Ameri-can officials if given the chance) that strategic priority setting,while difficult, is doable.

Third, the Scottish Office has established a systematic processfor local authorities to obtain feedback on their plans. Specifi-cally, it initiated a system of formal meetings at least once everytwo years with all local authorities.

Under the previous system [the one in operation in1992], we considered both how effective local authori-ties’ strategies appeared on paper and drew on anyanecdotal evidence available as to the actual implemen-tation of the strategies based on our own visits to, andmeetings with, individual authorities together withreports of similar contacts from colleagues in other

32 My interviews suggested that the process of developing realistic bids begana couple of years earlier. This change in mind-set is perhaps most evident inthe case of Glasgow’s bids for Housing Revenue Account funds. In 1991 it bidfor £150 million and got £90 million. In 1992 it bid for £100 million and got thesame £90 million. Thus it appears that even before 1993 local authoritiesaccepted the notion that realistic bids would not hurt them.

284 David P. Varady

relevant branches and divisions. Clearly such informa-tion was available only on an ad-hoc basis because, asyou know, we did not have regular meeting with au-thorities. . . . In future we will hold formal meetingswith all authorities at least once every 2 years. One ofthe main points for discussion at those meetings will bethe strategies which local authorities are now requiredto provide and their progress against the output targetswhich they have set to achieve those strategies.(Bannister 1993)

Fourth, the Scottish Office has proposed ways to achieve closerintegration between housing and planning strategies.

Under the previous system it was clear that a numberof authorities prepared their Housing Plans in isolationfrom the activities of other Departments within theCouncil which impact on housing provision, such asplanning. . . . In the future our examination of the newhousing plans will include a comparison with the localplans for the district to ensure compatibility. We al-ready examine newly submitted local plans to ensurethat they are consistent with, and make reference to,housing plans. Cases of inconsistency will be exploredwith the authority concerned. (Bannister 1993)

Finally, Glasgow’s inattention to geography was rectified in the1994 Housing Plan (Glasgow City Council 1994), which devotesconsiderable time and high priority to the Glasgow RegenerationAlliance, a union of four organizations: Glasgow City Council,Strathclyde Regional Council, Glasgow Development Agency,and Scottish Homes. The alliance seeks to revitalize eight disad-vantaged areas, five of which are peripheral housing estates(Greater Easterhouse, Castlemilk, Greater Pollak, Drumchapel,and Glasgow North) and three of which are inner-city communi-ties (the East End, Gorbals, and Govan). The alliance waslaunched in a set of vision statements for the eight areas con-tained in an attractive document, Glasgow Regeneration Alli-ance, Shaping the Future (Glasgow City Council et al. 1993),issued separately from the plan. The four agencies set a target of1994 for final agreements in such areas as industrial land andproperty and housing redevelopment and tenure change.

Conclusions and policy implications

When Britain initiated its local housing plan system in the late1970s, there was a great deal of optimism among local officials


and housing experts. They felt that the system would promotegreater flexibility and more creative thinking than the previousprojects-based system. Disillusionment set in during the early1980s when housing funds were slashed and strict governmentguidelines introduced. There were fears that the plans wouldsimply become bids to central government and be used to controllocal authorities.

My research based on four best-practice plans indicates thatthese fears were exaggerated. Britain’s local housing planningsystem works well. Britain has made a successful shift from aneeds-driven planning system to one that takes quality intoaccount in the strategy. In an atmosphere of limited publicresources, this means existing funds will be better used.

[Local housing plans were] used to hit the governmentover the head with. [They say:] We need ten times moreresources than are made available. . . . [Now, they aremore likely to] address the nitty-gritty of prioritizingrealistically based on a realistic view of resources.(Randall and Harrison 1992)

The competitive bidding system, which allows the quality of theplan to be included as a factor in making allocations, fostersinnovative thinking about housing options, encourages localauthorities to decide how they will carry out their enabling role(including how they will work with developers to provide afford-able housing), and promotes a concern for management perform-ance. Furthermore, the local housing planning system identifiesand corrects weaknesses. These are no small achievements.

Our analysis suggests five lessons for American housing plan-ning systems. First, in America, as in Britain, these plans couldserve multiple purposes beyond their current role as requests forfunding to the central government. In Britain they are used formarketing, for communicating with partner agencies, for lobby-ing central government, for eliciting citizen participation, and forexploring alternative strategies. There is no reason why theycannot serve the same purposes in the United States.

Second, the British plans demonstrate an overly narrow focus onlow-income housing issues, which are largely irrelevant to someof the most important urban issues such as the decline in theeconomic base. The lesson for American cities is that the plansought to address market-rate housing issues (e.g., the absence ofadequate “move-up” housing), as well as the needs of low-incomegroups such as the homeless.

286 David P. Varady

Third, following these British examples, American housing plansshould use a wider variety of data sources (e.g., waiting lists andphysical condition surveys) as well as the census. Furthermore,improvements in housing market analysis should make thempractical to carry out and useful for dealing with specific areasand population subgroups.

Fourth, the stress given to implementation and strategy inBritish local housing plans from York’s system of tenant surveysto measure customer satisfaction to Glasgow’s plan for coordina-tion with Scottish Homes to Birmingham’s diverse but yet city-controlled revitalization strategies deserves emulation. Therecommendations of the Scottish Office to better link citywidehousing plans with community land use plans should be broughtto the attention of American municipal officials.

Finally, aspects of Britain’s competitive bidding system ought tobe considered for implementation in the United States. Atpresent, HUD regional officers review the individual Consoli-dated Plans to decide if they are minimally acceptable. There islittle incentive built into the American system for cities toproduce better plans (e.g., with realistic performance indicators).Britain’s success with its competitive bidding system impliesthat adding more competition to America’s system would producehigher-quality Consolidated Plans.

HUD officials are aware of the advantages of a competitivesystem but are unlikely to move in the direction of competitiveplans that are similar to local housing plans in Britain.

If HUD got into that [approach], plans would have to berated and ranked. HUD would have to be very specificas to why some plans were rated higher than others.Operationally it is not feasible. This [type of rankingsystem] would avoid lawsuits. We would have to justifywhy say Camden gets it and Kansas City does not. Wewould have to indicate why one does and the other doesnot.33

The above statement raises the question: Why is the system oflocal plan review working in Britain but viewed as impractical inthe United States? The main reason, I suspect, is that Britain

33 Interview with a HUD official who preferred to remain anonymous. Theneed to obtain political support for housing programs by widening benefits tomany districts and states and by ensuring larger allocations to localities withpressing needs by formula measures also drives program design from competi-tive approaches.


has a long tradition of respect for government civil servants(some of the brightest college graduates are attracted to publicservice), and this respect leads citizens and local politicians totrust their judgment (see Foley 1973). In contrast, U.S. federalgovernment workers do not have the status nor do officialsreceive the same esteem that their British counterparts do.34

Thus, there is less need in Britain to develop the type ofrankings and objective standards mentioned in the above quote.

This discussion is relevant to current proposals under consider-ation for reinventing HUD (HUD 1994, 1995). The reinventionplan’s success appears to hinge on the ability to measure andmonitor performance. However, the United States is unlikely tomove in the direction of competitive allocations (Walker 1995).

HUD is being pushed in two directions. One part of HUD (Com-munity Planning and Development) wants more controls andstrategic planning. This division of HUD developed the Consoli-dated Plan. Other divisions of HUD are moving in a differentdirection. FHA insurance and public housing are being priva-tized. Sixty major HUD programs are being merged into threeperformance-based funds—a Community Opportunity Fund, anAffordable Housing Fund, and a Housing Certificate Fund—eachof which will be formula-driven.

These recommendations are severing the nexus between fundingand planning. Since so much of the funding will be formula-driven, it is unlikely that the quality of the plan will be impor-tant in determining allocations. Thus, I am pessimistic about theprospects for harnessing the competitive spirit within the plan-ning process, as is the case in Britain, to improve the delivery ofhousing programs.

34 Most HUD officials are skilled and hardworking. It is a shame that theirwork is undervalued by the general public. My conclusion on the high caliberof British central government housing officials is based primarily on myScottish research—the conclusions may not be generalizable to Department ofthe Environment regional office staff in England. HUD is taking a number ofsteps to improve the quality of local plans, including the development ofguidelines for assessing local housing needs (Bogdon, Silver, and Turner 1993)and the establishment of a new office to guide localities in conducting theirown strategic planning exercises. In addition, HUD is considering an awardsprogram that will recognize and reward examples of excellent strategic localand regional planning (Stegman 1993). Despite these three changes, planreview is likely to continue to be more procedural than substantive.

288 David P. Varady

Author

David P. Varady is a Distinguished Senior Scholar at the Center for UrbanPolicy Research, Rutgers University, and a Professor at the School of Planningat the University of Cincinnati.

The opinions expressed are those of the author and do not necessarily reflectthe positions of Fannie Mae Foundation or its officers. I would also like tothank the housing and planning officials in Glasgow, Dundee, Edinburgh,Birmingham, and York, too numerous to name here, who took time from theirbusy schedules to meet with me. Several persons played a key role in assistingme in this research: Barry Cullingworth, Cambridge University; JudithDugdale, University of Bristol; Robina Goodlad, University of Glasgow; ColinWood, University of Central England; and Kenneth Voytek, formerly withHUD. Finally, the research would not have been possible without an UrbanStudies fellowship from the University of Glasgow and sabbatical support fromthe University of Cincinnati.

References

Audit Commission for Local Authorities and the National Health Service inEngland and Wales. 1992. Developing Local Authority Housing Strategies.London: HMSO.

Bannister, Jan (Scottish Office). 1993. Letter to author, August 20.

Best, Richard. 1994. Recent Successes and Failures in the U.K. Paper pre-sented at the International Forum, Future Visions of Urban Public Housing,November 17–20, Cincinnati, OH.

Birmingham City Council. 1991a. Birmingham Unitary Development Plan.Birmingham, England.

Birmingham City Council. 1991b. Housing Investment Programme, StrategyStatement 1991. Birmingham, England.

Birmingham City Council. 1992. Housing Investment Programme, StrategyStatement 1992. Birmingham, England.

Blackaby, Robert (Director of Research, Birmingham Housing Department).1992. Interview by author. November 9, Birmingham, England.

Bogdon, Amy, Joshua Silver, and Margery Austin Turner. 1993. NationalAnalysis of Housing Affordability, Adequacy, and Availability: A Frameworkfor Local Housing Strategies. Washington, DC: U.S. Department of Housingand Urban Development.

Bramley, Glen. 1992. Interview by author. November 23, Bristol, England.

Bramley, Glen. 1993. The Enabling Role for Local Housing Authorities: APreliminary Evaluation. In Implementing Housing Policy, ed. Peter Malpassand R. Means. Milton Keynes, England: Open University Press.


Bramley, Glen, Philip Leather, and Alan Murie. 1980. Housing Strategies andInvestment Programmes. Bristol, England: University of Bristol, School forAdvanced Urban Studies.

Brooke, Jane (Glasgow Housing Department). 1992. Interview by author.December 4, Glasgow, Scotland.

Brown, Tim, and Norma Carter. 1990. Local Housing Policies and Plans inEngland. Paper presented at the International Sociological Association, July,Paris.

Call, Vaughn P. 1977. Citizen Participation Process in Spokane. Journal ofHousing 34(8):381–86.

Centre for Housing Research, University of Glasgow. 1992. Housing Needs andDemands in Dundee District Council: An Application of the Best PracticeGuide. Glasgow, Scotland.

City of Dundee. n.d. Housing Plan 1990–1995. Dundee, Scotland.

City of Hartford, Department of City Planning. 1986. Hartford Plan forDevelopment: 1985–2000. Hartford, CT.

Cullen, William (York Planning Department). 1992. Interview by author.November 16, York, England.

Dunmore, Kathleen. 1992. Planning for Affordable Housing: A PracticalGuide. Coventry and London: The Institute of Housing and the House BuildersFederation.

Enterprise Foundation. 1986. The Chattanooga 10 Year Program to Make AllHousing Fit and Livable. Columbia, MD.

Esher’s York: A Study in Conservation. 1969. York, England: York and CountyPress.

Final Rule, Consolidated Submission for Community Planning and Develop-ment Programs. 1994. Federal Register 60, no. 3 (January 5): 1878. Microfiche.

Foley, Donald L. 1973. British Town Planning: One Ideology or Three? In AReader in Planning Theory, ed. Andreas Faludi. Elmsford, NY: OxfordUniversity Press.

Fraser, Ross. 1991. Working Together in the 1990s. Coventry, England: Insti-tute of Housing.

Galloway, Michael (Director, Crown Street Regeneration Project). 1992.Interview by author. October 29, Glasgow, Scotland.

Glasgow City Council. 1991. City Planning Aims for the Next Decade. Glasgow,Scotland.

Glasgow City Council. 1992a. Housing Plan for the 1990s. Glasgow, Scotland.

Glasgow City Council. 1992b. A Review of Private Housing Market Activity inGlasgow 1991. Glasgow, Scotland.

290 David P. Varady

Glasgow City Council. 1994. Glasgow’s Housing Plan for the 1990s: A SharperFocus. Glasgow, Scotland.

Glasgow City Council, Strathclyde Regional Council, Glasgow DevelopmentAgency, and Scottish Homes. 1993. Glasgow Regeneration Alliance, Shapingthe Future: A Commitment to Area Regeneration. Glasgow, Scotland: GlasgowCity Council.

Glasgow City Housing. 1993. Special Needs Housing in Glasgow: The WayForward. Glasgow, Scotland.

Glasgow Development Agency. n.d. Crown Street Regeneration Project.Glasgow, Scotland.

Goetz, Edward G., and Shirley Geiger. 1995. Inclusiveness in Local HousingPlanning. Paper presented at Tenth Annual Conference of the Sociology ofHousing, May 5–6, Minneapolis, MN.

Goodlad, Robina. 1993. The Housing Authority as Enabler. Coventry andHarlow, England: Institute of Housing and Longman Group UK.

Grant, R. A., B. W. Thomson, J. K. Bible, and J. N. Randall. 1976. LocalHousing Needs and Strategies: A Case Study of the Dundee Subregion.Edinburgh: Scottish Development Department.

Hoben, James, and Todd Richardson. 1992. The Local CHAS: A PreliminaryAssessment of First Year Submissions. Washington, DC: U.S. Department ofHousing and Urban Development.

Hopkins, Kenneth (Scottish Building Employers’ Federation). 1992. Interviewby author. October 14, Glasgow, Scotland.

Inquiry into Planning for Housing. 1994. Summary (Joseph RowntreeFoundation).

King County (Washington) Housing and Economic Development Division.1987. Affordable Housing Policy Plan. Seattle.

Maclennan, Duncan. 1991. Extending the Strategic Role. In The HousingService of the Future, ed. David Donnison and Duncan Maclennan. Harlow,England: Longman.

Milne, Les (Glasgow Housing Department). 1993. Letter to author, August 10.

Munro, Moira (University of Glasgow, Center for Housing Research). 1992.Interview by author. October 27, Glasgow, Scotland.

National Community Development Association and the U.S. Conference ofMayors. 1980. Developing and Implementing a Local Housing Strategy.Washington, DC: U.S. Department of Housing and Urban Development.

Nelson, Kathryn P. (U.S. Department of Housing and Urban Development).1995. Interview by author. June 24, Washington, DC.

Nenno, Mary K. 1989. Housing and Community Development: MaturingFunctions of State and Local Governments. Washington, DC: NationalAssociation of Housing and Redevelopment Officials.


Pieda Planning. 1993. Strategic Priorities and Option Appraisal in HousingInvestment. Volumes 1–3. Edinburgh, Scotland: Scottish Office, EnvironmentDepartment.

Randall, John, and Fiona Harrison (Scottish Office). 1992. Interview byauthor. November 3, Edinburgh, Scotland.

Rasmussen, David W., and Raymond J. Struyk. 1981. A Housing Strategy forthe City of Detroit. Washington, DC: The Urban Institute.

Reiner, Janet S., Everett Reimer, and Thomas A. Reiner. 1963. Client Analysisand the Planning of Public Programs. Journal of the American Institute ofPlanners 29(4):270–82.

Scalfani, Salvatore (U.S. Department of Housing and Urban Development,Office of Community Development and Planning). 1995. Telephone interviewwith author. June 14.

Scottish Homes. n.d. Local Housing Market Analysis and Planning in ScottishHomes: A Best Practice Guide. Edinburgh, Scotland: Scottish Homes.

Scottish Office. 1988. New Life for Urban Scotland. Edinburgh, Scotland.

Scottish Office. 1990. Urban Scotland into the 90s: New Life—Two Years On.Edinburgh, Scotland: Scottish Office.

Scottish Office, Environment Department. 1993. New Style Housing Plans,Annual Policy Statements and Housing Capital Programmes, Circular No. 12.Edinburgh, Scotland: Scottish Office, April 30.

Stegman, Michael A. 1993. The Role of Planning in the Clinton Administra-tion’s Community Empowerment Agenda. Paper presented at the Associationof Collegiate Schools of Planning meeting, October 29, Philadelphia, PA.

Sternlieb, George, and David Listokin. 1985. Housing. In Setting MunicipalPriorities 1986, ed. Charles Brecher and Raymond D. Horton. New York: NewYork University Press.

Stoker, Gerry. 1988. The Politics of Local Government. London: MacmillanPress.

Strategic Use of Local Authority Housing Plans. 1995. Precis (Scottish Homes)20 (April).

Towner, Robert (York Housing Director). 1992. Interview by author. November16, York, England.

U.S. Conference of Mayors. 1985. Assessing Elderly Housing: A PlanningGuide for Mayors and Local Officials. Washington, DC.

U.S. Department of Housing and Urban Development. 1994. ReinventionBlueprint. Washington, DC.

U.S. Department of Housing and Urban Development. 1995. HUD Reinvention:From Blueprint to Action: Summary. Washington, DC.

292 David P. Varady

Varady, David P., and Charlotte T. Birdsall. 1991. Local Housing Plans.Journal of Planning Literature 6(2):115–35.

Varady, David P., and Jeffrey A. Raffel. 1995. Selling Cities: AttractingHomebuyers through Schools and Housing Programs. Albany, NY: StateUniversity of New York Press.

Walker, Christopher (The Urban Institute). 1995. Telephone interview byauthor. March 9.

Watson, Chris (University of Birmingham). 1992. Interview by author.November 9, Birmingham, England.

Webster, David (Glasgow Housing Department). 1992. Interview by author.December 21, Glasgow, Scotland.

Widner, Ralph R. 1980. Revising Our Development Strategies: A Challenge forthe 1980’s: Commentary. Washington, DC: National Council for UrbanEconomic Development.

Wood, Colin. 1993. Local Urban Regeneration Initiatives: A Study of Birming-ham “Heartlands.” Paper presented at the Seventh Congress of the Associationof European Schools of Planning, June 3–6, Stockholm.

Working Party on Assessment of Housing Needs. 1977. Assessing HousingNeeds: A Manual of Guidance. Edinburgh, Scotland: Scottish DevelopmentDepartment, HMSO.

York City Council. n.d. Housing Strategy Statement 1993/94. York, England:York City Council.

Economic Shifts and the Changing Homeownership Trajectory 293Housing Policy Debate • Volume 7, Issue 2 293© Fannie Mae Foundation 1996. All Rights Reserved.

Economic Shifts and the ChangingHomeownership Trajectory

James W. HughesRutgers University

Abstract

From 1980 to 1988, homeownership rates declined substantially for the firsttime in the postwar era. They stabilized and began to creep upward during the1988–94 period. After presenting a long-term perspective, this article de-scribes and examines two of the underlying forces of this upswing—demo-graphic aging and improved levels of affordability—as well as the impact ofimmigration and minority lags. Fundamental economic factors are thensurveyed: national and regional housing price shifts, housing productioncycles, measures of housing affordability, and employment. Several key econ-omic parameters of the post-recession housing market are presented as a guideto the short-term future.

Post-1988 homeownership rates initially rose because of an aging demography.But gradually, the new affordability became part of the dynamic. The newaffordability was driven by the decade-long slowdown and weakening ofhousing prices, lower post-recession interest rates, and accelerated job cre-ation following the period of “jobless” economic growth.

Keywords: Affordability; Homeownership; Demographics

Introduction

During the 1980s, America experienced a declining rate ofhomeownership for the first time since the Great Depression.1This decline was disturbing for a number of reasons. Home-ownership has long been considered the centerpiece of theAmerican dream, a notion confirmed by Fannie Mae–commissioned surveys conducted by Hart-Teeter Research(Fannie Mae 1992). It has also been regarded as a measure ofthe American standard of living—a barometer of economic andresidential well-being (Myers et al. 1992). Homeownership hasbeen described as binding together a diverse America (Sternlieband Hughes 1982). Thus, declines in homeownership rates

1 This pattern is well documented in the literature on homeownership trends.For example, see Hughes (1991), Myers et al. (1992), Wachter and Megbolugbe(1992), and Joint Center for Housing Studies of Harvard University (1994).

294 James W. Hughes

suggested that housing was becoming less affordable, livingstandards were stagnating or falling, and the American dreamwas fading.2

Homeownership is considered important for other reasons aswell (Eggers and Burke 1995; Myers et al. 1992). For households,it can promote savings and serve as an investment vehicle;provide a psychological sense of stability, well-being, satisfac-tion, and achievement; and yield a higher quality residentialenvironment.3 For municipalities, homeownership is seen as animportant building block for neighborhood stability and commu-nity commitment. For the broader economy, it is viewed as agrowth locomotive, stimulating construction and the productionof housing-related goods and services. Given all of these per-ceived benefits, the decreases in homeownership rates broughtabout increased attention on homeownership as a policy goal andrenewed interest in increasing the national homeownership rate(Eggers and Burke 1995).

The pattern of declining homeownership rates slowed by the endof the 1980s. The homeownership trajectory began to shift up-ward slightly and then plateau, powered by a conjunction ofeconomic shifts and demographic dynamics. The homeownershipsag of the 1980s was most visible in the young adult popula-tion—particularly the last cohorts of the baby boom generation—with affordability contributing to the economic problems. Theresulting ownership “shortfall” became a unique reservoir ofhousing demand, which asserted itself in the post-recessionperiod of the 1990s.4

A “new affordability” helped unleash this demand backlog as thepost-recession era unfolded in 1991. The first dimension of thisaffordability was a decade-long weakening of home prices. Pricesactually retreated in the once price-frenzied markets of the1980s (e.g., New England, the Middle Atlantic states, and South-ern California). The second dimension was the sustained declineof interest rates to single digits from 1982 to 1987 and the un-precedented decline from 1991 to 1993 that brought them to

2 For an analysis of the shifts in affordability from 1974 to 1989, see Gyourkoand Linneman (1993).

3 But as Eggers and Burke (1995) point out, there are challenges to the as-sumption that the financial benefits of homeownership accrue to all house-holds in all situations.

4 Also problematic in the 1980s was the economic weakness of nontraditionalhouseholds, which were the fastest growing demographic segment in theUnited States.

Economic Shifts and the Changing Homeownership Trajectory 295

levels not seen since the early 1970s. The merging of morerealistic pricing and low mortgage rates yielded the highestaffordability levels in a generation. The third dimension, jobgrowth, finally gained momentum following the first full year ofnational economic recovery (post–March 1992), enhancing con-sumer confidence and ability to move in the housing market.Some of the key 1980s’ impediments to homeownership disap-peared, although a fourth dimension—a lack of accumulatedresources to cover down payment requirements—was only par-tially mitigated by improved affordability and job growth. Conse-quently, many new homeowners in the 1990s were drawn fromthe shortfall ranks of the 1980s.

This article describes each of the above shifts that have contrib-uted to the halt in the decline in homeownership rates inAmerica. The purpose is to detail the economic and demographicchanges corresponding to the homeownership rate shift and tosuggest some of the possible linkages. The historical sweep ofhomeownership rates is presented first, followed by a moredetailed evaluation of rates by demographic subsectors (house-hold age, household configuration, and race and Hispanic origin).The impact of demographic aging on braking the homeownershiprate decline is then considered. Several economic factors affect-ing homeownership are analyzed: national and regional housingprice shifts, housing production cycles, changes in affordability,and employment growth patterns. The next section reviews theprospects for increased homeownership over the next severalyears. The article concludes with a recap of the findings andfuture prospects.

Changing homeownership trajectory

Homeownership rates: The long sweep

Homeownership in the United States may not have been on awild roller-coaster ride during the 20th century, but neither hasit been on a one-way escalator. After increasing rapidly duringthe 1920s, ownership rates retreated during the Great Depres-sion and World War II. They then soared for three and a halfdecades before suffering a reversal during the 1980s. Rateschanged course again as the 1980s came to a close and the 1990scommenced.

The specific ownership-rate trends of the past half century arepresented in figure 1. The ownership rate in 1940 (43.6 percent)indicates that America was still substantially a nation of renters

296 James W. Hughes

Fig

ure

1. L

ong-

Ter

m T

ren

d i

n H

omeo

wn

ersh

ip R

ates

, 194

0 to

199

4

70 60 50 40

Homeownership Rate (Percent)

1940

1950

1960

1970

1980

43.6

55.0

62.1

64.2

65.6

65.4

64.8

64.6

64.5

63.9

63.8

64.0

63.8

63.9

63.9

64.1

64.1

64.0

64.0

1981

1982

1983

1984

1985

Yea

r

Sou

rce:

U.S

. Bu

reau

of

the

Cen

sus

(197

5, 1

995c

).N

ote:

199

4 h

omeo

wn

ersh

ip r

ates

are

not

com

plet

ely

com

para

ble

wit

h e

arli

er d

ata.

See

foo

tnot

e 6

in t

he

text

.

1986

1987

1988

1989

1990

1991

1992

1993

1994


on the eve of World War II. But in the postwar era, the majorityof Americans were becoming homeowners. Homeownershipquickly accelerated past the halfway mark (55.0 percent) by1950, propelled by the increased sophistication of mortgageinstruments and a protected housing finance industry of special-ized lending institutions for shelter predicated on long-termfixed low-interest rate mortgages (Sternlieb and Hughes 1983).This element, in concert with rising incomes, permitted the vastupgrading of America’s housing inventory. This process wasinitially assisted by the savings accumulated during war-inducedprosperity.5 The following three decades saw a sustained rise inhomeownership—albeit at a steadily decreasing rate—finallypeaking in 1980 (65.6 percent).

After this peak, a major shift occurred, with homeownershiprates yielding some of their long-term gains. With one exception,each year between 1980 and 1988 witnessed a retreat in theownership rate, with the trough eventually reached in 1988(63.8 percent). Thereafter, the trend again reversed, with aplateau evident in 1993 and 1994, when the rate stabilized at64.0 percent.6 Although not a return to the peak level of 1980,the 1993–94 rate represents a modest recapture (0.2 percentagepoints) of the overall losses incurred during the 1980–88 period(1.8 percentage points).7

5 The personal savings rate in 1944 (personal saving as a percent of personaldisposable income) reached 25.1 percent. By 1970, it had fallen to 8 percent(Sternlieb and Hughes 1982).

6 The homeownership rates for 1994 are not completely comparable withearlier data because of two changes in the Current Population Survey (CPS)that took effect in the first quarter of 1994 (U.S. Bureau of the Census 1995c).(The CPS is the source of most of the homeownership rate data in this article.)A new weighting procedure, based on the 1990 decennial census, tends toreduce the reported homeownership rate. The effect of the second change, theimplementation of the Computer Assisted Survey Information Collection, isuncertain.

7 The postwar (1945–80) homeownership surge to some degree reflected—or atleast coincided with—a disciplined high rate of personal saving in the UnitedStates, which provided a pool of low-cost capital for a sheltered housingfinance system. It also provided the individual source for down payments. Andfor most of this period, median family incomes in America soared. Between1950 and 1973, real median family incomes doubled, with little growth be-tween 1973 and 1980 (Hughes 1991).

Personal savings rates began to falter in the 1970s but experienced much moresignificant erosion in the 1980s, when real median family incomes stagnated,with savings reaching a nadir in 1989. (The personal savings rate stood at7.9 percent in 1980; it dropped to 6.4 percent in 1985 and to 4.0 percent in1989. The rate increased to 5.3 percent by 1992 but fell back to 4.0 percent in1993.) A high-savings generation aged (the baby boom procreators); in its

298 James W. Hughes

Impact of age, race, and aging

The new trajectory of homeownership should be interpreted withcaution, and euphoria over its stabilization and improvementmust be restrained. On the surface, the current pattern suggeststhat the economics of homeownership have improved. But theoverall ownership rate, as detailed in table 1, is a simple sum-mary measure that masks underlying trends. The overall ratereflects the specific ownership rates of demographically definedsubsectors whose relative significance (or weight) changes overtime. Homeownership varies widely by age and household type,as well as by other demographic variables such as race.8 Thus,shifts in the age structure of the population, the mix of house-hold types, or racial composition will cause changes in the over-all rate of homeownership even if the economics of ownershipremain constant.

Age. The age dimension for the major household configurations isdetailed in tables 1 and 2. Table 1 provides ownership rates byhousehold type and age for 1982 (base year), 1988 (trough), and1994 (current).9 Table 2 then details the rate changes for the1982–88 (decline) and 1988–94 (recovery) periods. It is not sur-prising that during the decline period, each of the age sectorsunder 65 (for total households) experienced rate decreases, withthe largest losses evident in the brackets under 40 years old.

What is surprising, however, is the preponderance of negatives(decreases) in the demographically defined cells of table 2 duringthe recovery period. For example, the overall homeownershiprate increased by 0.2 percentage points between 1988 and 1994,but increases were registered by only two age sectors—60 to 64(0.3 percentage points) and 65 and over (1.8 percentage points).Younger households, however, still suffered fairly large declinesin ownership rates, with the rate for the 30–34 age sector fallingby 2.6 percentage points. As will be discussed later, this discrep-ancy suggests that the increase in the overall rate of ownership

stead was the emergence of a much larger low-savings generation (the babyboom itself). An emerging consumption ethic may have overwhelmed America’ssaving discipline. While the causal connections are complex, the retreat inhomeownership rates in the 1980s was certainly accompanied by a similarretreat in personal savings rates. A lack of resources for down paymentsemerged as a significant ownership inhibitor (Martinson 1993).

8 See Hughes (1991) for a full presentation of demographic variations inhomeownership rates.

9 The year 1982 is the first year for published data on demographically seg-mented homeownership rates.


Tab

le 1

. U.S

. Ho

meo

wn

ersh

ip R

ate

s b

y H

ou

seh

old

Ty

pe

an

d A

ge,

198

2, 1

988,

an

d 1

994

(Per

cen

t)

Fam

ily

Hou

seh

olds

On

e-P

erso

n H

ouse

hol

ds

Tot

alM

ale

Fem

ale

Mal

eF

emal

e

Age

of

Hou

seh

olds

Mar

ried

Cou

ple

Hou

seh

olde

rH

ouse

hol

der

Hou

seh

olde

rH

ouse

hol

der

Hou

seh

olde

r19

8219

8819

9419

8219

8819

9419

8219

8819

9419

8219

8819

9419

8219

8819

9419

8219

8819

94

Un

der

2519

.315

.814

.932

.629

.127

.121

.617

.621

.68.

98.

48.

613

.913

.612

.77.

59.

56.

225

to

2938

.635

.934

.153

.952

.449

.934

.732

.331

.617

.314

.414

.623

.723

.023

.014

.315

.917

.230

to

3457

.153

.250

.671

.969

.066

.650

.943

.444

.231

.328

.323

.331

.731

.132

.924

.726

.329

.035

to

3967

.663

.661

.280

.477

.876

.561

.255

.646

.943

.539

.636

.137

.637

.437

.035

.636

.240

.640

to

4473

.070

.768

.283

.983

.581

.863

.664

.757

.554

.250

.649

.037

.741

.242

.438

.940

.343

.345

to

4976

.074

.473

.886

.686

.285

.969

.666

.261

.357

.255

.158

.438

.343

.346

.945

.547

.852

.950

to

5478

.877

.176

.888

.287

.587

.674

.967

.670

.366

.760

.861

.940

.745

.650

.451

.756

.257

.255

to

5980

.079

.378

.489

.689

.589

.274

.376

.666

.966

.467

.166

.947

.347

.150

.760

.561

.161

.760

to

6480

.179

.880

.189

.490

.090

.081

.577

.372

.871

.571

.173

.250

.952

.155

.763

.863

.765

.065

& o

ver

74.4

75.6

77.4

86.6

88.8

90.4

75.3

80.0

83.5

75.1

78.3

79.9

58.6

60.1

63.7

62.2

62.0

64.6

Tot

al64

.863

.864

.078

.578

.978

.859

.356

.152

.847

.145

.344

.238

.639

.943

.151

.251

.854

.5

Sou

rce:

U.S

. B

ur e

au o

f th

e C

ensu

s (1

995c

).N

ote:

199

4 h

omeo

wn

ersh

ip r

ates

are

not

com

plet

ely

com

par a

ble

wit

h e

arli

er d

ata.

See

foo

tnot

e 6

in t

he

text

.

300 James W. Hughes

Tab

le 2

. P

erce

nta

ge

Po

int

Ch

an

ge

in H

om

eow

ner

ship

Ra

tes

by

Ho

use

ho

ld T

yp

e a

nd

Ag

e,19

82 t

o 1

988

an

d 1

988

to 1

994

Fam

ily

Hou

seh

olds

On

e-P

erso

n H

ouse

hol

ds

Tot

alM

ale

Fem

ale

Mal

eF

emal

eH

ouse

hol

dsM

arri

ed C

oupl

eH

ouse

hol

der

Hou

seh

olde

rH

ouse

hol

der

Hou

seh

olde

r

Age

of

1982

–19

88–

1982

–19

88–

1982

–19

88–

1982

–19

88–

1982

–19

88–

1982

–19

88–

Hou

seh

olde

r88

9488

9488

9488

9488

9488

94

Un

der

25–3

.5–0

.9–3

.5–2

.0–4

.04.

0–0

.50.

2–0

.3–0

.92.

0–3

.325

to

29–2

.7–1

.8–1

.5–2

.5–2

.4–0

.7–2

.90.

2–0

.70.

01.

61.

330

to

34–3

.9–2

.6–2

.9–2

.4–7

.50.

8–3

.0–5

.0–0

.61.

81.

62.

735

to

39–4

.0–2

.4–2

.6–1

.3–5

.6–8

.7–3

.9–3

.5–0

.2–0

.40.

64.

440

to

44–2

.3–2

.5–0

.4–1

.71.

1–7

.2–3

.6–1

.63.

51.

21.

43.

045

to

49–1

.6–0

.6–0

.4–0

.3–3

.4–4

.9–2

.13.

35.

03.

62.

35.

150

to

54–1

.7–0

.3–0

.70.

1–7

.32.

7–5

.91.

14.

94.

84.

51.

055

to

59–0

.7–0

.9–0

.1–0

.32.

3–9

.70.

7–0

.2–0

.23.

60.

60.

660

to

64–0

.30.

30.

60.

0–4

.2–4

.5–0

.42.

11.

23.

6–0

.11.

365

& o

ver

1.2

1.8

2.2

1.6

4.7

3.5

3.2

1.6

1.5

3.6

–0.2

2.6

Tot

al–1

.00.

20.

4–0

.1–3

.2–3

.3–1

.8–1

.11.

33.

20.

62.

7

Sou

r ce :

U.S

. B

ure

au o

f th

e C

ensu

s (1

995c

).N

ote :

Fig

ur e

s co

mpu

ted

from

tab

le 1

.


was due to demographic shifts: a maturing population increasedthe proportion of households in the older (higher ownership) agebrackets.

Race. The ethnic and racial composition of America’s populationalso affects overall homeownership rates. As shown in table 3,the homeownership rates for black and Hispanic families startedlower, and fell considerably more, than those of their whitecounterparts between 1982 and 1988.10 The pattern is a bit morecomplex during the 1988–93 recovery, but, overall, minoritygains lagged behind those of whites. However, black marriedcouples did make noticeable progress.

Nonetheless, there is still a wide gulf between white and minor-ity homeownership rates. And because minorities are generally

Table 3. Family Homeownership Rates by Race and Hispanic Origin,1982, 1988, and 1993 (Percent)

Male FemaleMarried House- House-

Total Couple holder holder

White 1982 77.4 80.9 67.0 56.41988 75.4 79.8 60.8 51.31993 75.6 80.7 56.6 50.8

Black 1982 52.3 64.4 52.7 35.91988 47.5 61.6 47.0 30.61993 47.0 63.8 48.5 29.7

Hispanic 1982 50.2 56.7 45.1 30.01988 43.7 51.7 32.4 23.31993 43.8 52.5 25.8 23.7

Percentage point change: 1982 to 1988White –2.1 –1.1 –6.2 –5.1Black –4.9 –2.8 –5.7 –5.3Hispanic –6.4 –5.0 –12.7 –6.7

Percentage point change: 1988 to 1993White 0.3 0.9 –4.2 –0.5Black –0.4 2.2 1.4 –0.9Hispanic 0.1 0.8 –6.6 0.4

Source: U.S. Bureau of the Census (1982, 1988, 1993a).Note: Numbers may not add to totals due to rounding.

10 The homeownership rates in table 3 differ somewhat from those of thepreceding and following tables. They are based on the March data of themonthly Current Population Survey (CPS). The March survey is known as theAnnual Demographic File. The homeownership rates in the other tables,drawn from the Housing Vacancy Survey, are also based on the CPS but areaveraged over the full year.

302 James W. Hughes

gaining a larger share of the population in every age sector, this isa force for deflating rates. However, the increasing propor-tion of minorities in America is not yet as influential on home-ownership as an aging population is.

Aging. Demographic aging is confirmed by recent age structureshifts in the U.S. population. As shown in figure 2, between 1988and 1994, the 18–24 and 25–29 age sectors shrank drastically(by 5.5 percent and 9.2 percent, respectively). This baby-bustphenomenon in turn caused a decline in these sectors’ relativeshares of the total population (from 15.1 to 13.4 percent andfrom 12.0 to 10.2 percent, respectively) (U.S. Bureau of theCensus 1993b, 1993c).11 Thus, those age groups with the lowestincidence of homeownership now contribute proportionally lessto the overall ownership rate.

In contrast, the critical mass of the nation’s adult populationbegan to shift from the under-30 age sector to the middle-agedbrackets during the six-year period. This predominantly baby-boom event is revealed by the rate of change data (figure 2). Thehighest population growth rates between 1988 and 1994 werein the 45–49 age sector (24.4 percent), the 40–44 age sector(20.0 percent), and the 50–54 age sector (17.1 percent). The40–44 age sector had the largest absolute population increase(more than 3.2 million persons). The baby-boom trek to fullmiddle age thus continued to dominate the nation’s age structurechanges.

Not surprisingly, shifting the analysis from population to house-holds reveals the same basic pattern. As shown in table 4, thenumber of households with householders under 35 years of agedeclined by 831,000, while those between 35 and 54 years of ageincreased by nearly 6.2 million. Clearly, the “middle-aging” ofhousing demand was the dominant household event of this six-year period.

These population and household transitions correspond to thelargest age-related jump in the ownership rate (table 1). Forexample, in 1994, the rate for the 25–29 age bracket stood at34.1 percent. But the rate was 50.6 percent for the 30–34 agebracket, 61.2 percent for the 35–39 age bracket, 68.2 percent for

11 The baby bust is generally considered to be those born between 1965 and1976. This generation followed the baby boom, those born between 1946 and1964. Births in the United States peaked in 1957 (4.31 million) and wereabove 4.2 million each year between 1956 and 1961. They reached a trough(below 3.2 million) in the 1973–76 period. The latter defines the heart of thefuture demographic shortfall for the housing market (Hughes 1991).


Fig

ure

2. R

esid

ent

Pop

ula

tion

18

Yea

rs a

nd

Ove

r, 1

988

and

199

4

18 t

o 24

25 t

o 29

30 t

o 34

35 t

o 39

40 t

o 44

Age

of

Res

iden

t

Sou

rce:

U.S

. Bu

reau

of

the

Cen

sus

(199

3b, 1

993c

).N

ote:

Nu

mbe

rs s

how

n a

bove

bar

s in

dica

te p

erce

nta

ge c

han

ge in

gro

up

size

.

45 t

o 49

50 t

o 54

55 t

o 59

–5.5

%

–9.2

%5.

1%15

.0%

20.0

%

24.4

%

17.1

%0.

7%–4

.2%

10.1

%

60 t

o 64

1988

1994

(P

roje

cted

)

65 &

ove

r

40 30 20 10 0

Population 18 Years and Over (Millions)

304 James W. Hughes

Table 4. Householders 15 Years Old and Over, 1988 and 1994

Total Households Percent

Age of(Thousands) Change: 1988–94 Distribution

Householder 1988 1994 Number Percent 1988 1994

15 to 24 5,228 5,263 35 0.7 5.7 5.425 to 34 20,583 19,717 –866 –4.2 22.6 20.335 to 44 19,323 22,293 2,970 15.4 21.2 23.045 to 54 13,630 16,837 3,207 23.5 15.0 17.355 to 64 12,846 12,188 –658 –5.1 14.1 12.665 & over 19,456 20,806 1,350 6.9 21.4 21.4

Total 91,066 97,107 6,041 6.6 100.0 100.0

Source: U.S. Bureau of the Census (1989, 1995a).Note: Numbers may not add to totals due to rounding.

the 40–44 age bracket, and 73.8 percent for the 45–49 agebracket. Thus, the maturing of the younger baby boomers hasbeen a potent force in halting the post-1980 homeownershipattrition. Reinforcing this tendency has been additional growthbetween 1988 and 1994 of the more mature middle-aged sectors.In fact, the largest absolute population gains were experiencedby the 40–44 and the 45–49 age brackets that in 1994 shelteredthe oldest baby boomers (figure 2) and have the highest rates ofownership.

Dynamics of change: Age-specific rate shifts

The aging of America’s population has been a major factor in theincrease in the overall rate of ownership, but it has certainly notbeen the only factor. Using the full six-year homeownership“recovery” period obscures some subtle complexities. To uncoverthese, table 5 presents 1988 to 1994 year by year. Although mostownership rates were lower in 1994 than they were in 1988, allbut three age sectors reversed their pattern of sharp declinesometime during the period (they did not necessarily stay on anupward path, however). The shaded cells in table 5 represent theownership rate low points for each age bracket. The trend rever-sal happened earlier for most older age groups (45 years andover) and later for the younger groups.

The preceding analyses show that the overall rate of homeowner-ship in the United States, after eroding through the 1980s,reversed the trend in 1988. The reversal was initially driven bydemographic aging. As households matured from younger agebrackets (lower ownership rates) to less youthful age brackets


Table 5. Homeownership Rates by Age of Householder,1988 to 1994 (Percent)

Age ofHouseholder 1988 1989 1990 1991 1992 1993 1994

Under 25 15.8 16.6 15.7 15.3 14.9 14.8 14.925 to 29 35.9 35.3 35.2 33.8 33.6 33.6 34.130 to 34 53.2 53.2 51.8 51.2 50.5 50.8 50.6*35 to 39 63.6 63.4 63.0 62.2 61.4 61.8 61.2*40 to 44 70.7 70.2 69.8 69.5 69.1 68.6 68.245 to 49 74.4 74.1 73.9 73.7 74.2 73.7* 73.850 to 54 77.1 77.2 76.8 76.1 76.2 77.2 76.8*55 to 59 79.3 79.1 78.8 79.5 79.3 78.9 78.460 to 64 79.8 80.1 79.8 80.5 81.2 80.9* 80.1*65 to 69 80.0 80.0 80.0 81.4 80.8* 80.7* 80.6*70 to 74 77.7 77.8 78.4 78.8 79.0 79.9 80.175 & over 70.8 71.2 72.3 73.1 73.3 73.4 73.5

Total 63.8 63.9 63.9 64.1 64.1 64.0 64.0

Source: U.S. Bureau of the Census (1995c).Note: Shaded figures indicate lowest homeownership rate in the 1988–94 period. 1994homeownership rates are not completely comparable with earlier data. See footnote 6 inthe text.*Indicates all times when the rate declined again after starting upward.

(higher homeownership rates), declines in individual age-specifichomeownership rates slowed. But by 1992, the aging factorwas supplemented by plateaus in a majority of age-specific own-ership rates (table 5), keeping the homeownership rate near64.0 percent.

Economic factors influencing homeownership,1980 to 1994

Housing price shifts

Ownership rate variations are closely linked to housingaffordability, which in turn is tied to housing prices and, obvi-ously, occupancy costs (discussed below). The fluctuations innational housing price in the past 25 years were initiated by adecade of surging rates of growth (the 1970s), followed by aperiod of relatively moderate increases (1980 to 1988), andconcluding with a further price deceleration (1988 to 1994). Thehuge run-up in prices during the 1970s, a period of high inflationand surging household formations, was probably a major con-tributor to the decline in ownership during the 1980s. This run-up forcefully contributed to an increase in entry-level costs foraspiring homeowners, particularly young ones, as the 1970s

306 James W. Hughes

came to a close. In turn, the moderate price increases of the1980s, culminating in a further slowdown after 1988, set thestage for subsequent ownership gains by not increasing entry-level costs.

The changing price of homes can be observed from a number ofperspectives and data series. The following analyses will focuson new single-family homes sold, existing single-family homessold, and regional variations in house prices.

New single-family homes. Table 6 provides baseline median salesprice data for new single-family homes. The median sales priceincreased at an average annual rate of 10.7 percent from 1970 to1980. The average rate of annual increase fell to 7.2 percentbetween 1980 and 1988 and to 2.4 percent between 1988 and1994. Prices escalated vigorously in the 1970s (176.1 percent),nearly tripling over the entire decade. As shown in table 7, thisrate of increase was far in excess of that of the consumer priceindex (CPI) (112.4 percent). The housing price bubble did notburst in the 1980–88 period, but prices increased by only74.1 percent, compared with an increase in consumer prices of43.6 percent. Finally, between 1988 and 1994, the bubble waspunctured as prices grew by only 15.6 percent over seven years,while the CPI grew nearly twice as fast (25.3 percent).12

The scale of deceleration in price appreciation was probablygreater than the data in table 6 suggest, since the quality ofhousing increased steadily after 1980 (as well as in the 1970s).The price index in table 8 measures changes over time from abase year of 1987 (1987 index = 100.0) in the sales price of newsingle-family houses that have important physical characteris-tics in common (U.S. Bureau of the Census 1995b).13 The priceindex for 1987-quality houses grew from 74.6 in 1980 to 103.6 in

12 The price slowdown may be partly attributed to shifting demand-supplyrelationships, with housing production escalating past household growth inthe 1980s. Household growth reached a record pace in the 1970s as most of thebaby boomers moved through the household formation stage of the life cycle,placing furious pressure on the housing market. Between April 1970 andApril 1980, the total number of households in the United States increased by16.9 million while the number of housing starts (1970 through 1979) totaled17.7 million, a difference of approximately 800,000. Household growth thenslowed dramatically. In the 1980s, the number of households increased by11.6 million (April 1980 to April 1990) while the number of housing startstotaled 14.9 million (1980 through 1989, inclusive), a difference of approxi-mately 3.3 million, suggesting a significant housing-development overhang(U.S. Bureau of the Census 1995d).

13 Note that the price index uses average home prices while the immediatelypreceding analyses used median home prices.


Table 6. Median Sales Price of New Single-Family Homes by Region,1970 to 1994

Year Total Northeast Midwest South West

Median sales price ($)1970 23,400 30,300 24,400 20,300 24,0001980 64,600 69,500 63,400 59,600 72,3001988 112,500 149,000 101,600 92,000 126,5001994 130,000 169,000 132,900 116,900 144,000

Percent change1970–80 176.1 129.4 159.8 193.6 201.31980–88 74.1 114.4 60.3 54.4 75.01988–94 15.6 13.4 30.8 27.1 13.8

Average annual percent change1970–80 10.7 8.7 10.0 11.4 11.71980–88 7.2 10.0 6.1 5.6 7.21988–94 2.4 2.1 4.6 4.1 2.2

Source: U.S. Bureau of the Census. Yearly. Characteristics of New Housing. CurrentConstruction Reports, Series C25. Washington, DC: U.S. Department of Commerce.

Table 7. Consumer Price Index, 1970 to 1994

Change from Previous Value (%)

Year Value Total Average Annual

1970 38.8 — —1980 82.4 112.4 7.81988 118.3 43.6 4.61994 148.2 25.3 3.8

Source: U.S. Bureau of Labor Statistics, Monthly Labor Review.Note: The 1982–84 price index equals 100 and reflects buying patterns of all urbanconsumers.

Table 8. Price Index of Houses Sold in the United States, 1980 to 1994

Average Sales Price

Price 1987-Quality Houses AllYear Index* (Estimated from Price Index) Houses

1980 74.6 $ 94,900 $ 76,4001987 100.0 127,200 127,2001988 103.6 131,800 138,3001994 120.2 152,900 154,500

Percent change1980–88 38.9 38.9 81.01988–94 16.0 16.0 11.7

Source: U.S. Bureau of the Census. Yearly. Characteristics of New Housing. CurrentConstruction Reports, Series C25. Washington, DC: U.S. Department of Commerce.*New, single-family houses, including value of lot. The price indexes have been struc-tured so that each index equals 100.0 in 1987.

308 James W. Hughes

1988, a 38.9 percent rate of increase. This is far less than the81.0 percent increase in the prices of all houses sold, suggestingthat a significant proportion of the price increment of homes soldcould be attributed to the shift toward larger houses and houseswith more amenities. Between 1988 and 1994, even thoughprices of housing units overall grew more slowly than pieces ofthe constant 1987-quality house, the absolute price of housingsold in 1994 ($154,500) was still higher than that of theconstant-quality house ($152,900).

Existing single-family homes. Sales of existing homes exhibit thesame broad pattern of price shift as new homes—a massiveratcheting up of prices during the 1970s followed by more mutedincreases after 1980 (table 9). The average annual rate of in-crease in the median sales price of existing single-family homesfor the same three periods analyzed above was 10.5 percent(1970–80), 4.6 percent (1980–88), and 3.5 percent (1988–94). Themajor variation was the more abrupt deceleration in the averageannual price increase during the 1980–88 period for existingsingle-family homes—4.6 percent (table 9) versus 7.2 percent fornew single-family homes (table 6).

Thus, the existing home inventory became even more economi-cally friendly—compared with the inventory of new homes—during the 1980s. In 1980, the median price of existing homes($62,200) was virtually identical to that of new homes ($64,600).But by 1988, the median existing home price ($89,300) was79.4 percent of the new home price ($112,500). As noted above,much of this difference was probably due to an increase in thesize and amenity level of new homes.

Regional variations. The national rates of increase obscure theactual experience in many areas of the country. Certain “hot”markets in the 1980s, largely driven by the bicoastal economicsurge, experienced extraordinary increases in ownership rates.14

As shown in table 6, during the 1980–88 period, the averageannual growth rate in the median sales price of new single-family homes in the Northeast (10.0 percent) was more than one-third larger than that of the United States overall (7.2 percent).For the entire period, prices increased by 114.4 percent in theNortheast compared with 74.1 percent for the nation. This pat-tern of regional differentiation was even more accentuated for

14 The income surges in these areas were key to their housing price frenzies.For example, New England’s per capita income was 103 percent of the nation’sin 1978; by 1988, it was 121 percent. For the same years, Connecticut’s percapita income went from 117 to 136 percent and Massachusetts’s went from104 to 124 percent (Hughes and Seneca 1993).


Table 9. Median Sales Price of Existing Single-Family Homesby Region, 1970 to 1994

Year Total Northeast Midwest South West

Median sales price ($)1970 23,000 25,200 20,100 22,200 24,3001980 62,200 60,800 51,900 58,300 89,3001988 89,300 143,000 68,400 82,200 124,9001994 109,800 139,100 87,900 96,000 146,700

Percent change1970–80 170.4 141.3 158.2 162.6 267.51980–88 43.6 135.2 31.8 41.0 39.91988–94 23.0 –2.7 28.5 16.8 17.5

Average annual percent change1970–80 10.5 9.2 10.0 10.1 13.91980–88 4.6 11.3 3.5 4.4 4.31988–94 3.5 –0.5 4.3 2.6 2.7

Source: National Association of Realtors, Real Estate Outlook, various issues.

the sales prices of existing single-family homes (table 9). TheNortheast’s annual rate of increase (11.3 percent) was more thandouble that of the nation (4.6 percent) between 1980 and 1988.Existing home prices increased by 135.2 percent in the Northeastduring the eight-year period compared with only 43.6 percentnationally.

A similar but more muted pattern was evident in the West forboth new and existing homes. The average annual increases inmedian sales prices of new homes in the 1980–88 period wereslightly higher than those of the nation as a whole and secondonly to those in the Northeast. In 1980, the West had been com-fortably perched at the top of the home-price ladder. Still, theNortheast’s surge was so strong that its home prices had eclipsedthose of the West by 1988.

The reversal of fortune after 1988 for both of these regions pro-vided a correction to the price-increase excesses of earlier years.The Northeast experienced a decline in existing single-familyhome prices between 1988 and 1994 (table 9) while rates ofincrease in new single-family home prices in the West andNortheast were below the national average (table 6).

This correction happened earlier in the Northeast than inthe West, which is dominated by California. Median salesprices for existing single-family homes for selected bicoastalmetropolitan areas are presented in table 10. The shaded cellsindicate the year when prices peaked for each area. Generally,

310 James W. Hughes

Tab

le 1

0. M

edia

n S

ale

s P

rice

of

Ex

isti

ng

Sin

gle

-Fa

mil

y H

om

es b

y S

elec

ted

Met

rop

oli

tan

Are

as,

198

2 to

199

4(i

n T

ho

usa

nd

s o

f D

oll

ars

)

% C

han

ge:

% C

han

ge:

1982

to

Pea

kP

eak

to

1994

Met

roM

etro

Met

ropo

lita

n A

rea

1982

1988

1989

1990

1991

1992

1993

1994

Are

aU

.S.

Are

aU

.S.

Un

ited

Sta

tes

67.8

89.3

93.1

95.5

100.

310

3.7

106.

810

9.8

—61

.9—

—

Nor

thea

stB

osto

n,

MA

80.2

181.

218

1.9

174.

217

0.1

171.

117

3.2

179.

312

6.8

37.3

–1.4

17.9

Har

tfor

d, C

T82

.916

7.6

165.

915

7.3

148.

214

1.1

135.

313

3.4

102.

231

.7–2

0.4

23.0

New

Hav

en/M

erid

en,

CT

NA

169.

416

3.4

153.

315

3.2

145.

814

2.5

139.

6N

A31

.7–1

7.6

23.0

New

Yor

k/N

. N

ew J

erse

y/L

ong

Isla

nd,

NY

/NJ/

CT

77.1

183.

818

3.2

174.

917

3.5

172.

717

3.2

173.

213

8.4

31.7

–5.8

23.0

Ph

ilad

elph

ia,

PA

/NJ

58.1

102.

410

3.9

108.

711

8.4

117.

011

8.0

119.

510

3.8

47.9

0.9

9.5

Pr o

vide

nc e

, R

I49

.713

0.6

130.

212

7.9

124.

311

8.5

116.

311

6.4

162.

831

.7–1

0.9

23.0

Wes

tO

r an

ge C

oun

ty(A

nah

eim

/S

anta

An

a M

SA

), C

A13

1.5

203.

924

1.7

242.

423

9.7

234.

822

0.7

211.

084

.340

.9–1

3.0

15.0

Los

An

gele

s A

r ea,

CA

119.

617

8.9

214.

821

2.8

218.

921

0.8

195.

418

9.1

83.0

47.9

–13.

69.

5R

iver

side

/S

an B

ern

ardi

no,

CA

78.8

106.

712

4.1

132.

113

7.6

136.

213

4.4

129.

174

.647

.9–6

.29.

5S

acra

men

to,

CA

76.4

94.6

111.

713

6.7

137.

713

4.0

129.

212

4.5

80.2

47.9

–9.6

9.5

San

Die

go,

CA

98.9

153.

418

1.9

183.

218

7.5

183.

117

6.9

176.

089

.647

.9–6

.19.

5S

an F

r an

c isc

o B

ayA

rea,

CA

128.

021

2.9

260.

625

9.3

258.

525

9.3

254.

425

5.6

103.

637

.3–1

.917

.9

Sou

rce:

Nat

ion

al A

ssoc

iati

on o

f R

ealt

ors,

Rea

l E

sta

te O

utl

ook

and

Hom

e sa

les

Ye a

r boo

k, v

ario

us

issu

es.

Not

e: S

had

ed a

reas

in

dic a

te p

eak

s in

sal

es p

ric e

s. U

.S.

c han

ges

from

198

2 to

pea

k a

nd

from

pea

k t

o 19

94 a

r e c

ompu

ted

sepa

r ate

ly f

or e

ach

met

ropo

lita

npe

ak.

NA

= n

ot a

vail

able

.


1988 was the most prominent peak year in the Northeast,while 1991 was the most common peak year in the West. Therates of increase in the metropolitan areas of the Northeastexceeded 100 percent between 1982 and the year of their pricepeak. These percentage increases generally tripled those of thenation for the equivalent time periods. For example, the mediansales price in Providence between 1982 and 1988 increased by162.8 percent; for the same period, the U.S. sales price in-creased by 31.7 percent. In contrast, California’s metropolitanareas, starting from a generally higher base in 1982, typicallyhad 1982-to-peak growth rates between only 80 percent and100 percent, compared to national equivalents of under50 percent for the identical period.

By definition, peaks must have a downside. Half of the north-eastern metropolitan areas presented in table 10 had double-digit price declines between their peak price and their 1994price. Most of the other northeastern observations, and two-thirds of those of the West, had single-digit losses, most above5 percent. In contrast, prices for the nation as a whole increasedevery year during the six-year period.

These relative shifts strongly reflect the bicoastal economy of1980s America. The real declines in housing prices experiencedin many noncoastal regions set the stage for subsequent housingaffordability gains as the new economy of the post-recession1990s unfolded. Moreover, those regions whose prices escalateddramatically were not put at a permanent competitive disadvan-tage, since the late 1980s real estate “crash” brought a quickerosion in their lofty value structures. The afterboom phase ofthe recent shelter cycle was clearly severe.

Housing production and boom-bust cycles

The post-1980 housing era—an era of maturing housing demand—had three distinct phases that were particularly pronounced inthe bicoastal arena: boom, afterboom (bust), and post-bust recov-ery. The data on regional metropolitan price surges and retreatsof the 1980s and 1990s serve to define the boom and afterboom(table 10). These phases are also evidenced nationally in housingproduction. The boom was defined by six straight years (1983 to1988) when single-family unit starts exceeded 1 million unitsannually—the longest sustained single-family production run ofthe postwar era (U.S. Bureau of the Census 1995d). The after-boom was then reflected by the production data of 1991, whenthe housing industry tumbled into a virtual economic abyss.

312 James W. Hughes

Total housing starts barely reached 1 million units in therecession-impacted year of 1991, the worst shelter productionrecord since World War II. (Among the major regions, only theMidwest, with a resurging industrial base bulwarked by manu-facturing exports, was spared the full impact.) Finally, the post-bust first saw single-family unit starts surge past 1 million unitsin 1992 and 1993. In 1994, nearly 1.2 million single-family startswere tallied, more than in any single production year in the1980s. Total starts in post-bust remained modest because of adepressed multifamily rental housing sector. Thus, single-familyunit production was the locomotive of the post-recession housingrecovery—multifamily rental units were the caboose.

Housing affordability

The post-recession housing era has been driven by new levels ofaffordability as revealed by the affordability index of the Na-tional Association of Realtors, which brings together housingprices, mortgage interest rates, and median family income. Theindex is the ratio of actual median family income to the incomeneeded to qualify for a mortgage on a median-priced existingsingle-family home at the current effective mortgage rate.Trends in house prices, affordability, and mortgage interestrates for 1972 and 1982 through 1995 are summarized infigure 3.

In 1972, the affordability index stood at 154.8—the medianfamily income in the United States was 154.8 percent of thequalifying income required to purchase the median-pricedexisting single-family home. Ten years later, in 1982, the indexhad plummeted to 59.5, the consequence of a near tripling ofprices ($26,700 to $67,800) and a doubling of mortgage rates(7.50 percent to 15.38 percent). This resulted in monthly princi-pal and interest payments nearly five times as high while me-dian family incomes merely doubled.15 Thus, the aftereffects ofthe high rates of price increase during the 1970s took their tollon the affordability thresholds of the early 1980s, particularlywhen they intersected with the unprecedented interest ratespike.

The affordability index’s low point was 1982, and it did notpierce 100 until 1986. This slow incline was the consequence ofsustained declines in interest rates (from 15.38 percent in 1982

15 National Association of Realtors, Economics and Research Division, ExistingHome Sales, various issues.

Economic Shifts and the Changing Homeownership Trajectory 313F

igu

re 3

. Hou

sin

g A

ffor

dab

ilit

y: S

elec

ted

Yea

rs, 1

972

to 1

995

1972

1983

1985

1987

1989

1991

1993

1982

1984

1986

1988

1990

1992

Oct

199

3

Med

ian

Hou

se P

rice

In

dex

(198

2 =

100)

Aff

orda

bili

ty I

nde

xM

ortg

age

Inte

rest

Rat

e

May

199

519

94

Sou

rce:

Nat

ion

al A

ssoc

iati

on o

f R

ealt

ors,

Eco

nom

ics

and

Res

earc

h D

ivis

ion

, Exi

stin

g H

ome

Sal

es, v

ario

us

issu

es.

Not

e: A

ffor

dabi

lity

inde

x eq

ual

s 10

0 w

hen

med

ian

fam

ily

inco

me

equ

als

qual

ifyi

ng

inco

me.

Mor

tgag

e in

tere

st r

ate

is t

he

effe

ctiv

e ra

te o

n

loan

s cl

osed

on

exi

stin

g h

omes

as

dete

rmin

ed b

y th

e F

eder

al H

ome

Loa

n B

ank

Boa

rd.

200

150

100 50 0

16 14 12 10 8 6

House Price and Affordability Indices

Mortgage Interest Rate (Percent)

314 James W. Hughes

to 10.25 percent in 1986) in the face of only modest priceincreases. The following year, 1987, mortgage rates fell below10 percent and fluctuated around that level through 1991. Withcontinued moderate increases in home prices, and relativelystable interest rates, the affordability index hovered in thevicinity of 110 through 1991, constrained somewhat at the end ofthis period by the income slowdown precipitated by the July1990–March 1991 recession. Affordability took a big jump in1992 and 1993 when, in the context of modest price increases,interest rates fell to levels not seen in more than 20 years. Forsix straight months (September 1993 through February 1994),mortgage rates remained below 7 percent—a potent housingmarket elixir. Peak affordability of the current cycle was at-tained in October 1993 when it reached 142.3, subsequentlyretreating in 1994 and early 1995 because of gently rising inter-est rates. Still, affordability remains much closer to the highs ofthe past two decades than to the lows.

Employment dimension

The shifting course of the national economy—and job creation—has also influenced the homeownership equation. The nationalrecession, which finally put the boom years of the 1980s to rest,began in July 1990 and ended in March 1991, eight months later.In September 1995, the national economy entered the 54thmonth of recovery and expansion, the average length of postwarexpansions. But the post-recession era has not been temporallyuniform nor geographically symmetrical. Geographically, it isquite different from its immediate business-cycle predecessor—the November 1982–July 1990 expansion—the longest peacetimeexpansion in the nation’s history.

Bicoastal economic ebullience marked the boom years of the1980s, with fears about the “hollowing out” of middle America, aphenomenon that was reflected by home-price and house-valuetrends shown earlier (Sternlieb and Hughes 1988). But as theboom tapered off and the nation slipped into recession, bicoastaleconomic malaise prevailed. The downturn largely affected thenation’s seaboards, and the areas hardest hit were those thathad been caught up in the real estate frenzy of the past decadeand the concomitant surge in service-producing employment.

The post-recession economic landscape also has had very differ-ent regional contours that were most apparent during the firstyear of recovery. The harsh aftereffects of the 1980s boom yearsalong the nation’s seaboards seemingly refused to succumb to


national economic expansion, yielding a stubborn and persistentbicoastal lag. Between March 1991 and March 1992, employmentin the United States increased by only 53,000 jobs, which trans-lates into a 0 percent growth rate (table 11). (The phenomenongave rise to the phrase “jobless economic growth.”) But thenationwide average masked significant regional variation. Mod-est but real employment growth in the Midwest (197,800 jobs or0.7 percent) and South (290,000 jobs or 0.8 percent) was ob-scured by post-recession employment declines in the Northeastregion (–441,400 jobs or –1.9 percent) and the Pacific division(–109,000 jobs or –0.7 percent). Thus, the Midwest and Southserved as centers of national economic recovery, while the North-east and Pacific states remained in the recessionary trough.

Jobless economic growth became a historical footnote during thesubsequent 34-month period. As shown in table 11, betweenMarch 1992 and January 1995, the U.S. economy added nearly7.1 million jobs (6.5 percent growth). The Northeast began toparticipate in the employment recovery, adding 651,300 jobs(2.9 percent), but it was the South (9.0 percent) and Midwest(7.4 percent) that still drove the expansion. At the same time,the West showed economic segmentation. While the Mountainstates had the highest growth rate of any division (14.9 percent),the Pacific division was America’s laggard (1.3 percent), withCalifornia’s performance an economic drag.

Thus, after nearly four years of national recovery and expansion,America’s economic lag still had a marked bicoastal cast. TheMiddle Atlantic and New England states are struggling to catchup to the national pacesetters, while the Pacific division, becauseof California’s problems, barely reversed economic course.

Nonetheless, the job growth acceleration after March 1992brought most regional laggards into the economic fold andbrought even more robust job growth to the regional leaders. Theresulting improvement in labor markets meshed with the re-gional price/value changes to reinforce the potential for home-ownership. Vigorous job growth in the South, the Midwest, andthe Mountain division probably provided the economic opportu-nity for households to take advantage of long-term price lags. Inthe Northeast, as job-losing state economies were replaced byjob-creating state economies, households were able to take ad-vantage of afterboom price corrections and interest rates.16

16 The states of the Northeast were buffeted by their job peaks of the 1980sand their recessionary lows. For example, Massachusetts lost 11.4 percent ofits job base (January 1989 to August 1992), Connecticut lost 9.7 percent(February 1989 to December 1992), New Jersey lost 7.1 percent (March 1989

316 James W. Hughes

Tab

le 1

1. T

ota

l E

mp

loy

men

t C

ha

ng

e b

y R

egio

n a

nd

Div

isio

n: M

arc

h 1

991

to J

an

ua

ry 1

995

(in

Th

ou

san

ds)

Ch

ange

: 19

91–9

2C

han

ge:

1992

–95

Reg

ion

Mar

ch 1

991

Mar

ch 1

992

Jan

uar

y 19

95N

um

ber

Per

cen

tN

um

ber

Per

cen

t

Nor

thea

st r

egio

n22

,671

.122

,229

.722

,881

.0–4

41.4

–1.9

651.

32.

9N

ew E

ngl

and

divi

sion

6,09

2.7

5,98

3.9

6,24

4.0

–108

.8–1

.826

0.1

4.3

Mid

dle

Atl

anti

c di

visi

on16

,578

.416

,245

.816

,637

.0–3

32.6

–2.0

391.

22.

4

Mid

wes

t re

gion

26,7

97.7

26,9

95.5

29,0

01.2

197.

80.

72,

005.

77.

4E

ast

nor

th c

entr

al d

ivis

ion

18,7

35.4

18,8

20.4

20,1

63.3

85.0

0.5

1,34

2.9

7.1

Wes

t n

orth

cen

tral

div

isio

n8,

062.

38,

175.

18,

837.

911

2.8

1.4

662.

88.

1

Sou

th r

egio

n36

,536

.536

,826

.540

,125

.629

0.0

0.8

3,29

9.1

9.0

Sou

th A

tlan

tic

divi

sion

19,4

20.0

19,4

68.0

21,2

07.5

48.0

0.2

1,73

9.5

8.9

Eas

t so

uth

cen

tral

div

isio

n6,

209.

06,

354.

06,

899.

114

5.0

2.3

545.

18.

6W

est

sou

th c

entr

al d

ivis

ion

10,9

07.5

11,0

04.5

12,0

19.0

97.0

0.9

1,01

4.5

9.2

Wes

t re

gion

22,4

65.7

22,4

72.3

23,5

86.8

6.6

0.0

1,11

4.5

5.0

Mou

nta

in d

ivis

ion

5,87

6.8

5,99

2.4

6,88

7.8

115.

62.

089

5.4

14.9

Pac

ific

div

isio

n16

,588

.916

,479

.916

,699

.0–1

09.0

–0.7

219.

11.

3

Tot

al10

8,47

1.0

108,

524.

011

5,59

4.6

53.0

0.0

7,07

0.6

6.5

Sou

rces

: D

ata

for

1991

–92:

Hu

ghes

an

d S

enec

a (1

994a

). D

ata

for

1992

–95:

U.S

. B

ure

au o

f L

abor

Sta

tist

ics,

Mon

thly

Lab

or R

e vie

w.


Cyclical recovery and enhanced affordability have significantlyenlarged the pool of market-competitive aspiring homeowners,helping to surmount some of the well-recognized demographicproblems of the current decade’s housing market.

However, broad economic cycles are rarely marked by smooth,even gradients—and this rule applies to the post-recessionupswing as well. The peak year of the cycle was 1994; net jobgrowth took a near vertical leap during the year.17 The resultwas the third highest job growth year in history. Housing mar-kets reflected this labor market strength. In 1994, sales of exist-ing single-family homes nearly tied the historical record, andnew single-family starts surged. But then job growth deceleratedin the first half of 1995, the consequence of seven short-terminterest rate increases, engineered by the Federal Reserve toforestall inflation by slowing the economy without putting it intorecession. One casualty of this action was housing affordability,which fell as interest rates rose throughout 1994 and May 1995(figure 3). Weaker labor markets led to weaker housing marketsin the first half of 1995. However, this does not threaten toderail the homeownership gains of the 1990s; the basic afford-ability gains remain in place. Decreases in interest rates in late1995 brought rates near their 1993–94 lows, improving theaffordability picture.

Post-bust housing market

To summarize, in 1992, a vigorous housing market finallystarted to emerge, the consequence of melding most of the pre-ceding forces and dynamics into four factors. First, mortgageinterest rates below 7.0 percent, which were once unthinkable,became not only thinkable, but available. Their magic hit fullforce during late 1993. Although rates escalated in 1994 and1995, they are again fairly close to their 20-year lows. Otherdimensions of the mortgage market also played a role in makingmortgages more accessible, as well as contributing to the declinein interest rates.

Low interest rates contributed to a second factor—the newaffordability. As declining interest rates intersected with real(and in some cases nominal) home-price deflation, the cost of

to May 1992), and New York lost 7.0 percent (June 1989 to November 1992).See Hughes and Seneca (1994a, 1994b).

17 U.S. Bureau of Labor Statistics, Employment and Earnings, various issues.

318 James W. Hughes

homeownership in many instances became economically com-petitive with renting.

Third, the new affordability brought into the market a largenumber of first-time buyers who had been shut out during thefrenzy years of the 1980s. During the 1980s, many younger babyboomers had not been able to buy homes because of high pricesand mortgage costs. They have since entered the market. Thefiltering-up process is again working—at least in the shortterm—rejuvenating the trade-up market.

Finally, the economic hemorrhage in lagging regions abated,leading to improved labor market conditions. Household financesrebounded as vigorous job growth ensued and low interest ratesmitigated consumer debt burdens.

This new shelter affordability should persist even as the nationalrecovery continues to mature. The following parameters shouldalso serve to define the broader housing market for the balanceof the century.

Suburban trade-up homeownership

An aging demography portends family-raising suburban shelter.A middle-aged, child-rearing baby boom signals trade-up single-family dwellings as the long-term housing market target of thedecade. This aging process by itself should ensure that home-ownership rates will set new records before the end of thedecade. As the 1990s come to a close, however, increasing num-bers of leading-edge baby boomers will become empty nesters,implying post-family-raising suburban shelter. But this does notimply a return to rental status, only new physical configurations.

Filtering-up process

As noted earlier, the backlog of entry-level homeownershipdemand by the youngest baby boomers rejuvenated homeowner-ship and the trade-up market in the early 1990s. In essence, thefiltering-up process was jump-started. Housing market mobilitybenefited from the decline in the homeownership rates duringthe 1980s. So far, this pent-up demand has surmounted baby-bust demographics, but the reservoir is not inexhaustible.

Therefore, the filtering-up process may still slow. The well-learned lesson of the afterboom years was that move-ups can


only occur when buyers are found for the “old” house. Once thebacklog of young baby boomers—who are still aspiring home-owners—is exhausted, baby-bust demographics will impede thetrade-up process later in the decade, producing market softness(Mankiw and Weil 1989).

Immigration

Immigration, however, will buffer some of the negative demo-graphic effects.18 Immigration levels in 1990 enlarged the pool ofpotential homeowners for the balance of the century. The 1965Immigration Act set a quota of 290,000 persons per year, alongwith certain exemptions. By 1985, the worldwide quota wasreduced to 270,000. But the 1990 Immigration Act capped totalimmigration at 700,000 each year for 1992 through 1994 and675,000 afterwards, not counting a variety of exemptions. Projec-tions for the balance of the 1990s, including exemptions, rangebetween 875,000 and 950,000 persons per year (Emrath 1994;Fix and Passel 1991). However, the impacts of this increase varysubstantially by region.

Thus, the 1990 Immigration Act set in motion a considerableincrease in the current and future demand for housing in theUnited States, in turn guaranteeing increased demand forhomeownership in absolute terms. However, the varied economiccapacities of immigrants may work to inhibit the scale of in-crease in the nation’s ownership rate. Although there have beenimpressive improvements of the ownership rates of the immi-grants of the past generation, with rates increasing rapidlyaccording to length of stay, the rates for ownership of the new-comers for the early 1990s will not approach national averagesuntil well into the first decade of the next century (Emrath1994).

Minority homeownership

Lagging minority homeownership will be a persistent problem inthe absence of direct public policy actions. The gulf betweenwhite and minority rates has failed to narrow during the 1980sand 1990s. While this failure may be partially due to immigra-tion bolstering the ranks of minorities, other questions remain.Certainly the overall incidence of homeownership in America

18 Immigration’s impact on homeownership is evaluated in Joint Center forHousing Studies (1994).

320 James W. Hughes

will not reach its full potential without an upgrade of minorityhomeownership. However, research is needed to evaluate thedownside of a public policy commitment to minority homeowner-ship programs (e.g., fostering household investment in geo-graphic areas with little chance of house value appreciation).

Changing housing rationales

American homeowners benefited for four decades as a virtuallyrecession-proof shelter bull market prevailed throughout much ofthe nation. It created a multitude of true believers in residentialreal estate, but shelter confidence was severely shaken duringthe afterboom, particularly in the bicoastal arena. The burstingof the housing-price bubble produced a more sober financialoutlook. Despite a post-bust recovery, housing rationales havebeen altered.

America’s economy will vary in its future growth performance. Itwill be difficult to consistently replicate the exceptional growthexperience of 1994, particularly because corporate rationaliza-tion and downsizing are probably not over. This trend is cer-tainly a concern to areas considered corporate headquarterscountry—the epicenters of the nation’s service economy thatdefined the frenzied housing markets of the 1980s.

Thus, there appears to be a much more conservative outlook onhomeownership’s financial and investment potential. Housing isno longer considered an infallible savings machine (Hughes andZimmerman 1993). It is still seen as a sound long-term invest-ment, but it may not pay for college educations or represent amagic retirement nest egg. Low inflation has not only yieldedlow interest rates, it has also placed limits on home price appre-ciation. Inflation is not available as a “take-out mechanism”—itwill not bail households out of bad shelter decisions, such asbuying into a geographic area with limited real price apprecia-tion potential. Exit strategies are now weighed more heavilywhen “entrance” decisions are made. “Will I be able to sell thehouse?” is more important now than ever. The lack of demo-graphic support for easy mobility is now well recognized; it isdifficult to casually trade up or to trade sideways.

During the boom years of the 1980s, forecasts for 1990s housinggenerally heralded the coming dominance of upscale trade-upmarkets. But casual economic assumptions were rendered impo-tent by the afterboom, corporate restructuring, and the shift inhousing rationales. The result is far less “up” in upscale and far


less “up” in trade-up. Leading-edge “up-market” sectors arebound by significant price constraints, which produce moreconsumer and producer discipline, thereby keeping price struc-tures in line—enhancing affordability, particularly for entry-level homeownership.

Future affordability

The interest rate lows of the fourth quarter of 1993 quicklybecame an artifact, but affordability levels have not collapsednor are they likely to do so in the near future. The higher inter-est rate thresholds of mid-1994 through mid-1995 may haveinhibited the pace of housing price increases, and the decline ininterest rates during 1995 and early 1996 bodes well foraffordability in the near term.19 Moreover, low general inflationshould persist, further minimizing housing price pressures.Similarly, the long-term slowdown in household growth will notexert its full effect on homeownership demand until the trough ofthe baby bust reaches its prime home-buying years. Thus,demand-side pressures will remain close to their post–WorldWar II lows. All of these tendencies imply a continuation of high(but less-than-peak) levels of affordability. If job growth re-bounds from its early 1995 slowdown, the economics will be inplace for homeownership gains to continue beyond those gener-ated simply by demographic aging.20

The gains would be even more substantial if savings rates im-proved, which would lead to a buildup of household net worthand would exert a downward push on interest rates. Despitecompelling reasons for the savings rate to increase during thebalance of the century, it probably will not. The post-recessionreturn to high levels of personal consumption and increasingconsumer debt suggests that habits built over the past threedecades are not easily abrogated. Thus, it is difficult to see first-time aspiring homeowners being any more successful in accumu-lating down payments than their recent predecessors. Moreover,while direct research is lacking, parental assistance in securingdown payments seems to be more tenuous than before. In the

19 At the time of this writing (February 1996), mortgage interest rates areagain approaching the record lows of 1993 and 1994.

20 Preliminary evidence indicates that the homeownership rate rose in 1995.The U.S. Department of Housing and Urban Development released statisticsthat showed homeownership rates increasing from 64.2 percent in the fourthquarter of 1994 to 65.1 percent in the fourth quarter of 1995 (Associated Press1996).

322 James W. Hughes

1970s, generous down payment support from parents wasdubbed “GI” (good in-laws) financing (Sternlieb and Hughes1980). Generally, the first-time home buyers of the 1990s andbeyond are the children of a generation with a lower savingsrate. Their parents’ pocketbooks and home equity positions mayhave already been stressed by the escalating costs of highereducation. Again, these pocketbooks and home equity levelscould reflect—now and for the balance of the century—the de-cline in savings rates and deceleration of housing value increasesover the past two decades. Worried about their own level ofretirement resources and depleted home equity reserves, parentsmay have to be more penurious about down payment assistance.Constraints such as this could continue to hinder the full realiza-tion of the nation’s homeownership potential.

Conclusion

This article has described the basic shifts in homeownershipoverall and by age group from 1982 to 1994. The period startedwith an unprecedented decline in homeownership rates. Thisdecline reached its trough in 1988, followed by slight increasesand then general stabilization after 1990. The basic maturationof the American population into higher ownership age groupsslowed the decline in rates and eventually underpinned theirreversal. Also instrumental in the transition from decline tostabilization were the increasing levels of homeownershipaffordability in the post-recession 1990s, although further em-pirical research is required to determine just how effective thisfactor was relative to demographic aging. This new afford-abilitywas tied to weakening of home prices and substantially lowerinterest rate thresholds, as well as a strengthening nationaleconomic recovery in 1993 and 1994.

The overall homeownership rate plateau in 1993 and 1994 failedto match the peak rate of 1980. Despite the affordability gains ofthe 1990s, the 1994 homeownership rates of all age groups below60 remain far below their peaks of the previous decade, with thewidest gaps among younger age groups. Given the sheer breadthof these gaps, they are not likely to be eliminated in the foresee-able future. The scale of the improvement in general afford-ability to date has been so substantial that future gains willprobably be much smaller. Future research should conduct acomprehensive review and analysis of the reasons for these gapsas a basis for potential policy prescriptions.


Author

James W. Hughes is Dean of the Edward J. Bloustein School of Planning andPublic Policy at Rutgers University.

References

Associated Press. 1996. Home Ownership at 14-Year High. February 9.

Case, Karl E., and Robert Shiller. 1989. The Efficiency of the Market forSingle-Family Homes. American Economic Review 79(1):125–37.

Eggers, Frederick J., and Paul E. Burke. 1995. Stimulating the Impact onHomeownership Rates of Strategies to Increase Ownership by Low-Income andMinority Households. Paper presented at the Fannie Mae Annual HousingConference.

Emrath, Paul. 1994. Immigration and Housing Demand. Housing Economics,March, pp. 5–8.

Fannie Mae. 1992. Fannie Mae National Housing Survey. Washington, DC.

Fix, Michael, and Jeffrey S. Passel. 1991. The Door Remains Open: RecentImmigration to the United States and a Preliminary Analysis of the Immigra-tion Act of 1990. Washington, DC: The Urban Institute Press.

Gyourko, Joseph, and Peter Linneman. 1993. The Affordability of the Ameri-can Dream. Journal of Housing Research 4(1):39–72.

Hughes, James W. 1991. Clashing Demographics: Homeownership andAffordability Dilemmas. Housing Policy Debate 2(4):1217–50.

Hughes, James W., and Joseph J. Seneca. 1993. The Income Roller Coaster.Rutgers Regional Report Issue Paper Number 7. New Brunswick, NJ: RutgersUniversity, Edward J. Bloustein School of Planning and Public Policy.

Hughes, James W., and Joseph J. Seneca. 1994a. Bicoastal Economic Recoveryand Lag: Post-Recession Employment Growth Patterns. Rutgers RegionalReport Issue Paper Number 10. New Brunswick, NJ: Rutgers University,Edward J. Bloustein School of Planning and Public Policy.

Hughes, James W., and Joseph J. Seneca. 1994b. Anatomy of a Recovery: TheGeography of Economic Rebound. Rutgers Regional Report Issue PaperNumber 11. New Brunswick, NJ: Rutgers University, Edward J. BlousteinSchool of Planning and Public Policy.

Hughes, James W., and Todd Zimmerman. 1993. The Dream Is Alive. Ameri-can Demographics 15(8):32–37.

Joint Center for Housing Studies of Harvard University. 1994. The State of theNation’s Housing. Cambridge, MA.

Mankiw, N. Gregory, and David N. Weil. 1989. The Baby Boom, the BabyBust, and the Housing Market. Regional Science and Urban Economics19:235–58.

324 James W. Hughes

Martinson, Henry E. 1993. Home Buyers’ Sources of Down Payment. HousingEconomics, July, pp. 9–12.

Myers, Dowell, Richard Peiser, Gregory Schwann, and John Pitkin. 1992.Retreat from Homeownership: A Comparison of the Generations and theStates. Housing Policy Debate 3(4):945–75.

Sternlieb, George, and James W. Hughes. 1980. America’s Housing: Prospectsand Problems. New Brunswick, NJ: Rutgers University, Center for UrbanPolicy Research.

Sternlieb, George, and James W. Hughes. 1982. The Evolution of Housing andIts Social Compact. Urban Land 41(12):17–20.

Sternlieb, George, and James W. Hughes. 1983. Housing of the Poor in a Post-Shelter Society. Annals of the American Academy of Political and SocialSciences 465(1):109–22.

Sternlieb, George, and James W. Hughes, eds. 1988. America’s New MarketGeography: Nation, Region and Metropolis. New Brunswick, NJ: RutgersUniversity, Center for Urban Policy Research.

U.S. Bureau of the Census. 1975. Historical Statistics of the United States,Colonial Times to 1970. Bicentennial Edition, Part 2. Washington, DC: U.S.Government Printing Office.

U.S. Bureau of the Census. 1982. Household and Family Characteristics:March 1982, 1988, and 1993. Current Population Reports, Series P20. Wash-ington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census. 1988. Household and Family Characteristics:March 1982, 1988, and 1993. Current Population Reports, Series P20. Wash-ington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census. 1989. Money Income of Households, Families, andPersons in the United States: 1987. Current Population Reports, Series P60-162. Washington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census. 1993a. Household and Family Characteristics:March 1982, 1988, and 1993. Current Population Reports, Series P20. Wash-ington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census. 1993b. U.S. Population Estimates, by Age, Sex,Race, and Hispanic Origin: 1980 to 1991. Current Population Reports, SeriesP25-1095. Washington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census. 1993c. Population Projections of the United States,by Age, Sex, Race, and Hispanic Origin: 1993 to 2050. Current PopulationReports, Series P25-1104. Washington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census. 1995a. Income, Poverty, and Valuation of NoncashBenefits: 1993. Current Population Reports, Series P60-188. Washington, DC:U.S. Government Printing Office.

U.S. Bureau of the Census. 1995b. Characteristics of New Housing: 1994.Current Construction Reports, Series C25. Washington, DC: U.S. Departmentof Commerce.


U.S. Bureau of the Census. 1995c. Housing Vacancies and HomeownershipAnnual Statistics: 1994. Current Housing Reports, Series H111/94-A. Wash-ington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census. 1995d. Statistical Abstract of the United States:1995. Washington, DC: U.S. Government Printing Office.

Wachter, Susan M., and Isaac F. Megbolugbe. 1992. Racial and EthnicDisparities in Homeownership. Housing Policy Debate 3(2):333–70.

326 James W. Hughes

Where the Homeless Come From 327Housing Policy Debate • Volume 7, Issue 2 327© Fannie Mae Foundation 1996. All Rights Reserved.

Where the Homeless Come From: A Study ofthe Prior Address Distribution of FamiliesAdmitted to Public Shelters in New York Cityand Philadelphia

Dennis P. Culhane, Chang-Moo Lee, and Susan M. WachterUniversity of Pennsylvania

Abstract

This study investigates hypotheses regarding the association of census tractvariables with the risk for homelessness. We used prior address informationreported by families entering emergency shelters in two large U.S. cities tocharacterize the nature of that distribution.

Three dense clusters of homeless origins were found in Philadelphia and threein New York City, accounting for 67 percent and 61 percent of shelter admis-sions and revealing that homeless families’ prior addresses are more highlyconcentrated than the poverty distribution in both cities. The rate of shelteradmission is strongly and positively related to the concentration of poor,African-American, and female-headed households with young children in aneighborhood. It is also correlated with fewer youth, elderly, and immigrants.Such areas have higher rates of unemployment and labor force nonpartici-pation, more housing crowding, more abandonment, higher rates of vacancy,and higher rent-to-income ratios than other areas.

Keywords: Homeless; Housing; Neighborhood

Introduction

Researchers and policy makers have increasingly emphasizedthe structural and dynamic nature of the homelessness problem(Burt 1992; Interagency Council for the Homeless 1994; Piliavinet al. 1993). Research on the structural factors associated withhomelessness has used primarily intercity homelessness rates(point prevalence) as the dependent measure, attempting toidentify the associated housing, population, income, and policyfactors (Applebaum et al. 1991, 1992; Burt 1992; Elliot and Krivo1991; Quigley 1991; Tucker 1987). This research has yieldedsignificant though inconsistent results, particularly regardingmany predicted housing and income variables. This article ad-dresses the same issue, using intracity data, aggregated bycensus tract, based on the prior addresses of homeless families intwo large U.S. cities.

328 Dennis P. Culhane, Chang-Moo Lee, and Susan M. Wachter

Literature review

Basic research on contemporary homelessness has employedprimarily cross-sectional survey methods designed to enumeratethe population and document its demographic characteristics.While providing a detailed profile of the population and many ofits needs, this method has had limitations. It has produced astatic representation of a dynamic problem; it has identifiedwhere and in what condition people end up as homeless, but notwhere they come from or go to; and while it has identified thecharacteristics of individuals that increase their vulnerability tothe condition, the data have not been well suited to assessing thesocial processes that contribute to that vulnerability. To someextent, public policies and programs designed to address home-lessness have shared these limitations. Most homelessness pro-gram development has focused on expanding the availability ofresidential and supportive services that target currently home-less persons and families. Program development has focused lesson forestalling the housing emergencies of the many more indi-viduals and families who, without intervening assistance, willmove in and out of homelessness over time. Homelessness pro-grams have also targeted individuals for intervention, and notthe communities or institutions from which they come or thesocial and economic forces that have put these individuals at risk.However, evidence has emerged of a shift in both the researchand policy sectors toward a greater understanding of the struc-tural and dynamic nature of the homelessness problem.

In the research sector, several investigators have applied orargued for the use of geographic methods to study structuralaspects of the homelessness problem (Kearns and Smith 1994;Wallace 1989, 1990; Wolch and Dear 1993). Most commonly,researchers have attempted to identify the socioeconomic factorsthat correspond to the spatial distribution of homelessness,using data on intercity homelessness rates as the dependentvariable (Applebaum et al. 1991, 1992; Burt 1992; Elliot andKrivo 1991; Quigley 1991; Ringheim 1990; Tucker 1987). Basedon this research, homelessness appears to vary by socioeconomicconditions, although specific study findings have been inconsis-tent. Tucker (1987), in one of the first applications of thismethod, argued that cities with rent control had higher home-lessness rates, based on data from an early survey of city sheltercapacity by the U.S. Department of Housing and Urban Develop-ment (HUD 1984). Applebaum and colleagues (1991, 1992)identified major flaws in Tucker’s approach and providedcounterevidence that low vacancy rates, as a proxy for tighthousing markets, were more closely related to HUD’s intercity

Where the Homeless Come From 329

homelessness rates. Elliot and Krivo (1991), using the samedata, found that the availability of low-income housing and lowerper capita expenditures on mental health care were significantlyrelated to homelessness rates but that poverty and unemploy-ment rates were not. In a test of several more carefully specifiedmodels of intercity homelessness rates, Burt (1992) found thatper capita income, the poverty rate, and the proportion of single-person households combined to explain more than half the varia-tion in homelessness rates in high-growth cities, interpreted asevidence that more affluent households and a greater number ofhouseholds with single people put pressure on the housingchoices of poorer people.

A limitation of this research, and perhaps an explanation forstudy differences, is the reliability and validity of the dependentvariable. While perhaps the most widely attainable proxy for thesize of the homelessness problem across locales, point prevalencemeasures are difficult to obtain reliably from place to place. TheHUD estimates (1984) used by Tucker (1987), Applebaum et al.(1991), Elliot and Krivo (1991), and Quigley (1991) were basedon a key informant survey in 60 cities. HUD officials asked fieldstaff to report on the capacity of localities’ emergency sheltersand the estimated number of street homeless in their areas;thus, these estimates were not based on a systematic count. Thecomparability of study findings based on the HUD estimates isfurther complicated by the various authors’ use of differentjurisdictional boundaries in calculating rates. The Urban Insti-tute estimates used by Burt (1992) were derived from results of alarger, more systematic survey of shelter providers and based ona hypothetical ratio of street homeless to sheltered homeless; butagain, they were not derived from an actual count.

Even if estimates were reliably obtained across jurisdictions,their validity as comparable measures of the extent of home-lessness across locales would be confounded by the highly vari-ant responses of those locales to the problem of homelessness. Toa significant degree, the daily size of the sheltered population,typically the largest component of the homeless count, is supply-and policy-driven (Burt 1994; Culhane 1992). The elasticity ofthe supply of shelter beds defines access to the shelter system,which in turn is a function of local policies governing admissioncriteria, length-of-stay limits, and the flexibility of resources tomeet demand. Other policies, such as copayment requirements,sobriety checks, and treatment mandates, as well as the overallquality of facilities, are also likely to influence some clients’perceptions of whether accepting accommodations in a shelterhas relative appeal over other options, and for what duration.


Likewise, opportunities for exiting homelessness will affect theduration of episodes; in general, more programs to facilitate exitfrom homelessness should decrease time to exit and correspond-ingly produce a lower daily census.1 Each of these factors islikely to exercise a systematic influence on a city’s averageshelter stay and shelter capacity, which in turn will play a deter-mining role in the point prevalence of homelessness.

Recent longitudinal research has suggested the potential rel-evance of a structural and dynamic model of homelessness andhas raised questions about the adequacy of point prevalence datafor measuring the homelessness problem. Analyses of adminis-trative data (Burt 1994; Culhane et al. 1994), a national tele-phone survey (Link et al. 1994), and a housing survey in NewYork City (Stegman 1993) have all found that as much as3 percent of the population experienced an episode of “literal”homelessness between 1988 and 1992, suggesting a high degreeof turnover in the homeless population. Longitudinal researchbased on tracked samples of homeless persons (Fournier et al.1994; Koegel and Burnam 1994; Piliavin et al. 1993; Robertson,Zlotnick, and Westerfelt 1994; Wright and Devine 1995) has alsodocumented the often transitory, intermittent nature of home-lessness. Most shelter users appear to mobilize resources andcommunity ties to avoid the shelters most of the time. Hopper(1990, 1995) has characterized these informal networks as the“economies of makeshift.” Unfortunately, the nature of thesesupport systems, and the factors that strain or enhance theirsupportive capacity, are not well understood (see related discus-sions in Burt [1994], Piliavin et al. [1993], and Rossi [1994]).

In the policy sector, recent proposals have discussed the dynamicand structural aspects of the homelessness problem. Most re-cently, the Clinton administration’s plan Priority Home: TheFederal Plan to Break the Cycle of Homelessness (InteragencyCouncil for the Homeless 1994) offers a social and economicanalysis of the causes of homelessness, as well as a distinctionbetween chronic and episodic homelessness.2 Based on thisanalysis, the plan argues for making homelessness prevention apriority for future federal policy. The Clinton plan describes

1 Paradoxically, the opposite could also occur, as may occur in some programsthat require a minimum stay to become eligible for exit programs, or as mayoccur as a result of increased demand for emergency shelter to obtain access toexit programs.

2 Kondratas (1994) observed that the Bush administration plan also empha-sized homelessness prevention and the integration of homeless populationsinto mainstream social programs.


broad legislative initiatives intended to approach that goal, suchas the administration’s health care and welfare-reform propos-als, expansion of the earned-income tax credit, and increasedhomeownership and rental-assistance opportunities.3 In addi-tion, the plan’s core policy objective—that localities establish anorganized “continuum of care” for the homeless service system—acknowledges the need for preventive and long-term housingstabilization efforts, as well as traditional remedial strategies, toreduce the prevalence of homelessness.

The plan does not address how localities might plan for preven-tion programs and offers few specifics regarding implementationother than in the broad terms of the major legislative initiativesdescribed above. Given that many of the proposals in the federalplan are placed in the context of the scientific literature, the gapin the plan could well be a reflection of a gap in prior research.Some conceptual elaboration of homelessness prevention pro-gramming has appeared in the literature (Jahiel 1992; Lindblom1991), but the available empirical literature is limited (U.S.Department of Health and Human Services 1991). The literatureon program targeting has been comparably sparse (Knickmanand Weitzman 1989). Researchers have not provided a methodfor helping policy makers to determine where homelessnessprevention resources should be targeted, nor have they clearlydocumented the factors they should focus on.

Our present study is an attempt to contribute to the continuingintegration of a structural and dynamic model of homelessnessin the research and policy sectors, both by beginning to answerthe “where to target” question facing the planners of homeless-ness prevention programs and by adding to researchers’ tools forinvestigating the structural correlates of homelessness (or the“what to target” question facing planners). This study uses theprior-address information reported by persons admitted to thePhiladelphia and New York City shelter systems to construct anintracity index for the rate of homelessness by census tract andidentifies census tract variables that correspond to that distri-bution. An intracity measure has the following methodologicaladvantages over the intercity point prevalence measuresdescribed above: (1) in general, it is concerned not with theexactness of a count for a given day but with identifying arepresentative sample of persons from whom prior-addressinformation can be obtained over a given period of time; and

3 Regardless of the particular merits or shortcomings of many of these propos-als, their future is uncertain in light of recent changes in the composition ofthe U.S. Congress.


(2) it is not confounded by local policies and regulations thataffect shelter supply and stay patterns because those factorswould presumably have a similar impact across a city’s jurisdic-tion, particularly in centrally administered shelter systems suchas those studied here. While intercity analyses permit research-ers to assess the policy and social factors that vary in relation tohomelessness rates among cities, an intracity approach allowsthem to characterize spatial variations within a city. Thus, anintracity approach may contribute to an understanding of the“makeshift economies” that beget homelessness and of the pro-cesses that contribute to the success or failure of the makeshifteconomies in mediating housing instability.

Social selection processes of homelessness

To develop a theory for generating hypotheses, our study buildson previous theoretical work (Blau 1992; Burt 1992; Culhane1990; Hopper and Hamberg 1986; Jahiel 1992; Rossi 1989;among others). Briefly, the model argues that homelessness is aconsequence of a combination of housing, income, population,and policy factors that have significantly increased the probabil-ity that poor persons will live in precarious housing arrange-ments. Among the precariously housed, a shelter admission ismost likely to occur following some household crisis (e.g., jobloss, marital separation, benefit termination, utility disconnec-tion, hospitalization, incarceration, family conflict) and mostfrequently occurs among persons who have the least amount offamilial, social, or public support. These people include unem-ployed single mothers who are caring for young children and donot receive child support payments; adults with disabilities,including people with mental disorders and people addicted todrugs or alcohol; the undereducated and underemployed, par-ticularly those ineligible for unemployment insurance or generalassistance welfare programs; and people with weak familialsupports, such as those fleeing abusive families and individualswho were reared in foster care or otherwise unsupportive familyenvironments. The precariously housed are expected to be con-centrated in certain areas, because of both selective migrationand restrictions on their housing choice.

A family crisis or household disruption does not necessarily leadto shelter use, but such a result is more likely in the context ofshortages of affordable and suitable housing for people with verylow incomes. The risk of homelessness would likely be greaterif the disruption were preceded by residence in poor-qualityhousing or if it resulted in a subsequent move to such housing.


Thus, one would expect to find that public-shelter admissions aremost often generated in the lowest rent neighborhoods wherepoor people exhaust the opportunities most accessible to them.Such areas are more likely to have generally distressed housingconditions, as indicated by more vacancies and abandonment.Moreover, despite having the lowest-cost housing available, suchareas may nevertheless be “unaffordable” to the people who livein them, leading some to live in crowded or doubled-up arrange-ments (in subfamilies).

The relevance of the other major component to the housingaffordability problem—low income—is likely to be evident by thehigher rates of poverty and joblessness in such neighborhoods.Problems with access to the labor market are indicated by higherrates of unemployment, less full-time employment, and lessparticipation in the labor force. Public assistance presumablyreduces the risk of homelessness in an area (compared with poorareas where people receive less public assistance), but it alsomay be associated with an increased risk of homelessness to theextent that receipt of public assistance indicates very low incomeand less participation in the labor market.

It is presumed that the housing and income problems describedabove have differentially affected African Americans because ofhistorical patterns of migration, economic development, residen-tial segregation, and discrimination. Other ethnic minorities,such as Hispanics and immigrant groups, may also face in-creased risk of homelessness due to poverty, restricted labormarket access, and segregation in poorer-quality housing.

Hypotheses and research questions

First, our study explores the spatial distribution of the residen-tial origins of homeless families through spatial statistics andthematic maps, permitting us to compare the degree of cluster-ing and segregation in those distributions between cities andamong boroughs within New York City. The descriptive analysesalso identify the degree to which the homeless and povertydistributions differ in their concentration, unevenness, andclustering, to further qualify the nature of the prior-addressdistribution of homeless families.

To understand the marginal effect of various factors on thespatial distribution of homeless families’ prior addresses, weused cross-sectional data from the 1990 decennial census(measuring demographic composition, economic status, and


housing and neighborhood factors) in a regression analysis totest some of the assumptions of the theoretical model regardingan area’s potential risk. We hypothesize that the variables de-fined in table 1 will be significantly associated with the rate offamily shelter admission by census tract.

Table 1. Variable Definitions and Hypotheses

Variable Definition Expected Sign

DemographicRBLACK Ratio of black persons +RSPAN Ratio of Hispanic persons +RUNDER18 Ratio of persons under 18 +ROVER64 Ratio of persons over 64 +RNOHIGH Ratio of persons without high school diploma +RFHHOLD Ratio of female-headed households +RFYOUCHD Ratio of female-headed households with

children under six years old +ROLDFAM Ratio of families with householder over 64

years old +RSUBFAM Ratio of subfamilies +RGRPQUAT Ratio of noninstitutionalized persons in

group quarters +RFRBRN70 Ratio of the foreign-born who immigrated

after 1970 +

EconomicRUNEMP Ratio of unemployment +MNHHPAI Mean household public assistance income +MEDHHINC Median household income –RNOPOV Ratio of persons below poverty level +RNOWORK Ratio of persons not in labor force +RTMPWORK Ratio of persons working under 18 hours

per week +

Housing and neighborhood qualityMEDVALUE Median property value –MEDCOREN Median contract rent –RRENT Ratio of rental units +RENTHINC Ratio of median contract rent to median

household income +RCROWD Ratio of housing units with more than two

persons per room +RVAC Ratio of vacant units +RBOARDUP Ratio of boarded-up housing units +

Note: Dependent variable is log(ratio of homelessness occurrence +1). All ratios are inpercent.

We expected variations by city to affect our results, given knowndifferences in several housing market factors such as populationloss, a much higher proportion of single-family housing, andoverall lower housing costs in Philadelphia. We also explored


differences between low- and higher-income areas to test forfactors that may differentially expose persons to homelessness inareas disaggregated by median income.

Procedures

Database development

Data sources. New York City and Philadelphia systematicallyregister all users of public shelters through automated clientmanagement information systems (see Culhane et al. 1994). Aspart of the shelter admission process, families in New York Cityand all households in Philadelphia are asked to report their “lastaddress.” This question may be variously interpreted by familiesrequesting shelter. For purposes of the present study, we assumethe addresses, through their aggregation, to be a proxy for theareas in which families entering the shelter have had somerecent residence. For consistency between sites, only data onfamilies were included in the study. To create an admissionrecord in Philadelphia, clients must present two forms of identi-fication that together must include a social security number anda Philadelphia street address.4 The Philadelphia database beginsDecember 21, 1989, and is current to April 1, 1994. It includesrecords for 9,160 families. In New York City, shelter admissioninformation for families may be verified against a family’s infor-mation in the New York State Welfare Management System atthe time of admission, if the family is registered in that system.The data from New York used for this study begin April 1, 1987,and are current to April 1, 1994. They include records for 71,035households.

Geocoding procedures. To construct a database of addressesaggregated by census tract, we overlaid the addresses from thePhiladelphia data set with the census tract coverage from theTIGER/Line file (U.S. Department of Commerce 1993). We

4 Some persons may be admitted to a shelter with a non-Philadelphia streetaddress because they can otherwise prove that they have been in Philadelphiafor a minimum of six weeks (thereby meeting the residency requirement),because they are sheltered as part of the mandatory shelter provision policy ineffect on extremely cold or hot days, or because they have been admitted inviolation of policy. Some persons do not report a prior address because theyenter the shelter system after-hours (after 5 p.m.), thereby avoiding thecomplete intake interview. Families are permitted to avoid the intake inter-view if they stay for only one night; they are required to complete the intakeinterview if they stay for consecutive nights.


processed the address data from New York City through Geosup-port, a program for normalizing street addresses and for produc-ing geocodes for census blocks and tracts maintained by the NewYork City Department of City Planning.

For both cities, we first matched client address data to therespective base map files (see table 2). For New York City,70 percent of the cases had an address that matched the Depart-ment of City Planning’s geographic files. Shelter addresses wereremoved to produce the study population. The unmatched casesconstitute 30 percent of the total and include rejected in-cityaddresses, in-state non–New York City addresses, out-of-stateaddresses, and missing addresses. In Philadelphia, 59 percent ofthe cases had an address that matched the TIGER file. Again,shelter addresses were removed to produce the study population.The unmatched cases (41 percent) include rejected in-city ad-dresses, in-state non–Philadelphia addresses, and out-of-stateaddresses, but are composed largely of missing addresses. Weconducted further analyses to determine the representativenessof the study population, including comparing the race andethnicity of matched versus unmatched cases, comparing thegeographic distribution of in-city addresses (both those that didand those that did not match the respective base maps by zipcode), and comparing the prior addresses of households withsingle and multiple admissions to shelter (see appendix for amore complete discussion).

Table 2. Qualification of Study Populations

New York Philadelphia

Address-matched sample 49,604 5,375Shelter addresses 481 319Family 49,123 5,056

Nonmatched sample 21,431 3,785In-citya 9,990 858In-state (not in city)b 429 24Out-of-statec 2,120 42Missingd 8,892 2,861

Total households 71,035 9,160

a In-city rejected addresses represent 16.8 percent of the total in-city addresses reportedin New York City. The rejected addresses correlate with the matched addresses by zipcode at r = 0.877. For Philadelphia, the rejected addresses represent 13.8 percent of thein-city addresses and correlate at r = 0.972 with the matched addresses by zip code.b In New York, the most frequent counties of origin outside New York City areWestchester (48 cases), Suffolk (46 cases), and Ulster (20 cases).c Outside of New York, the most frequent states/territories of origin are Puerto Rico(422 cases), New Jersey (244 cases), Pennsylvania (137 cases), California (117 cases),South Carolina (93 cases), North Carolina (90 cases), Connecticut (83 cases), andMassachusetts (81 cases).d 12.5 percent missing in New York City, and 31.2 percent missing in Philadelphia.


Descriptive measures of area variations in homelessnessrates

Concentration by census tract. To analyze the two-dimensionalconcentration of the prior addresses of homeless households withthematic maps by census tract, we used the location quotient(LQ). The LQ is frequently used to identify the proportionatedistribution of a given object group among areas (Bendavid-Val1983). The LQ refers to the ratio of the fractional share of thesubject of interest at the local level to the same ratio at theregional level (see appendix). This article uses the census tractas the equivalent of the local unit and the city or borough as theequivalent of the regional unit.5

Although the LQ is used to examine the two-dimensional aspectsof a spatial distribution, other indices are required to quantifythe relational aspects of that spatial distribution within andamong jurisdictions. For this study, we selected three additionalindices to measure these relational aspects: unevenness, contigu-ity, and clustering.

Unevenness. Unevenness refers to how unequally an object orsocial group is distributed among defined areas in a given juris-diction. For example, a minority group is said to be “segregated”if it is unevenly distributed over census tracts in segregationstudies (Massey and Denton 1988; White 1983). The most widelyused measure of unevenness is the index of dissimilarity. Itmeasures departure from evenness by taking the absolute devia-tion of the population-weighted mean of every census tract’sobject-group proportion from the city’s object-group proportionand expressing that quantity as a proportion of its theoreticalmaximum (James and Taeuber 1985) (see appendix).

Contiguity. A second distributional attribute is the degree ofspatial contiguity. While unevenness deals with the distributionof an object group within a set of areal units overall, contiguity isconcerned with the similarity in concentration between adjoiningareal units. In this study, we used an index of spatial autocor-relation, Moran’s I (Odland 1988), to measure the degree ofcontiguity (see appendix).

Clustering. The third dimension to the spatial distribution of anobject group is clustering. The contiguity index captures some

5 Census tracts with populations under 100 were omitted from both the de-scriptive and the regression analyses to avoid the outlier effects produced bysmall denominators.


aspects of clustering because it identifies the extent to which ad-joining areas have similar concentrations of a given phenomenon.However, when the object group forms highly segregated enclavesin space, the contiguity index would fail to distinguish that type ofclustering. Unfortunately, a proper measure of clustering for latticedata is not available in the literature. Therefore, we developed aclustering index based on our own definition of clustering, referringto the close spatial association of areas with a high concentration ofthat object group (see appendix).

Regression analyses

As stated in the conceptual model, we assume the number of theprior addresses of the shelter users in each census tract to be afunction of demographic composition, economic factors, andhousing and neighborhood characteristics in the census tract.The mathematical form of the model can be denoted as follows:

log(HRi) = a + b(X1i) + c(X2i) + d(X3i) + �i, (1)

where HRi is the rate of shelter admission with the number ofhouseholds in tract i; X1i is the set of demographic variables intract i; X2i is the set of economic variables in tract i; X3i is the setof housing and neighborhood variables in tract i; a is intercept;b, c, and d are sets of the coefficients corresponding to the sets ofthe explanatory variables, X1, X2, and X3, respectively; and �i isthe error disturbance in tract i. Sample statistics for the vari-ables are shown in table 3.6

The ordinary least square (OLS) estimation is based on theassumption of constant error variance. However, data based oncensus tract contain sources of unequal error variance. Everycensus tract does not have the same physical size or equal popu-lation. Therefore, the shelter-admission rate in less-populatedcensus tracts tends to fluctuate more than the rate in more-populated census tracts. This situation can worsen when shel-tered households are concentrated in smaller census tracts.

6 In terms of explanatory variables, median property value (MEDVALUE) ismissing in 99 census tracts in New York. The census tracts are mostly low-income neighborhoods that are our main areas of interest (the mean ofMEDHHINC in the 99 tracts is $20,090, while the mean of all the tracts is$31,532). MEDVALUE is presumably missing in these tracts because itmeasures owner-occupied property values, and these areas may have too fewowner-occupied properties. We dropped MEDVALUE in the final modelspecification, since MEDVALUE was not statistically significant in theexploratory model specifications and the loss of the observations is so largethat it may produce a biased result.


Table 3. Sample Statistics

New York Philadelphia

Variable N Mean Corr.* N Mean Corr.*

DemographicRBLACK 2,107 28.675 0.67 342 39.712 0.71RSPAN 2,107 21.985 0.46 342 4.990 0.06RUNDER18 2,107 21.823 0.64 342 22.027 0.46ROVER64 2,107 13.427 –0.50 342 15.523 –0.24RNOHIGH 2,107 21.433 0.46 342 22.225 0.36RFHHOLD 2,107 19.325 0.82 342 31.992 0.79RFYOUCHD 2,107 5.352 0.76 342 9.167 0.64ROLDFAM 2,107 10.451 –0.44 342 18.676 –0.24RSUBFAM 2,107 5.193 0.58 342 8.562 0.69RGRPQUAT 2,107 1.000 0.09 342 2.507 –0.04RFRBRN70 2,107 18.912 –0.05 342 3.810 –0.22

EconomicRUNEMP 2,107 9.632 0.63 342 11.079 0.67MNHHPAI 2,107 1,986 –0.52 342 3,897 –0.21MEDHHINC 2,107 31,532 –0.58 342 25,783 –0.51RNOPOV 2,107 19.268 0.75 342 20.028 0.68RNOWORK 2,107 2.321 0.47 342 2.383 0.54RTMPWORK 2,107 1.773 –0.11 342 2.101 –0.08

Housing and neighborhood qualityMEDVALUE 2,008 203,004 –0.48 337 65,580 –0.45MEDCOREN 2,107 489.000 –0.57 341 364.173 –0.56RRENT 2,107 65.143 0.42 342 39.669 0.24RENTHINC 2,107 1.720 0.54 341 1.542 0.15RCROWD 2,107 1.657 0.34 342 0.383 0.31RVAC 2,107 5.367 0.12 342 10.875 0.54RBOARDUP 2,107 0.336 0.36 342 2.378 0.72

RNOHMLS 2,107 1.530 NA 342 1.239 NALRNOHMLS** 2,107 1.812 1.00 342 0.495 1.00

Note: NA = not applicable.* Correlation coefficient with the dependent variable (LRNOHMLS).** LRNOHMLS is calculated as log(RNOHMLS + 1) to avoid missing values.

To test the existence of heteroskedasticity, we assumed the errorvariance to be a decreasing function (negative exponential) of thenumber of households in each census tract. Technically, the logof squared residuals from the OLS estimation is regressed withthe number of households. The White test for the pooled OLSestimations reveals the existence of heteroskedasticity (NewYork: χ2 = 35.6, p value = 0.00; Philadelphia: χ2 = 2.66,p value = 0.10). To overcome heteroskedasticity, we used thesquare root of the estimated error variance for the weight forthe final weighted least square (WLS) estimations.


Results

Descriptive measures

In both cities in the aggregate, the distribution of homelessorigins is more highly concentrated than the poverty distribu-tion. Both cities have a lower proportion of census tracts withan LQ greater than or equal to 1.01 for homelessness than forpoverty, but a higher proportion of tracts with an LQ greaterthan 2.00 for homelessness than for poverty (see tables 4, 5,and 6 and figures 1, 2, 3, and 4). Thus, while the poverty distri-butions are characterized by areas that are more broadly dis-tributed but have moderately high concentration (LQ > 1.01), thehomeless distributions are characterized by areas that are lessbroadly distributed but have higher concentration (LQ > 2.00).Accordingly, poverty is a modest proxy for homelessness. Thecorrelation coefficient between the two distributions (by LQ bycensus tract) is 0.558 in New York City and 0.640 in Philadel-phia, as the relative shares of poverty are more widely distrib-uted than the relative shares of homeless origins.

Within each city, the concentrations of homeless origins yieldvisually evident clusters as well, as shown in figures 1 and 3.Nearly two-thirds (61 percent) of all homeless families from NewYork City from 1987 to 1994 were from the three major clusters:Harlem (15 percent of total), South Bronx (25 percent), and theBedford-Stuyvesant–East New York neighborhoods (21 percent).Philadelphia also has three major clusters accounting for67 percent of the homeless families’ prior addresses: NorthPhiladelphia (primarily west of Broad Street) (38 percent), WestPhiladelphia (20 percent), and South Philadelphia (primarilywest of Broad Street) (9 percent).

The calculated indices of unevenness, contiguity, and clusteringare given in table 7. For unevenness, Staten Island scores thehighest, and the Bronx scores the lowest among the five bor-oughs in New York. The homeless families’ addresses arehighly segregated in Staten Island, whereas in the Bronx, wherea broad set of areas is affected, homeless origins are not highlysegregated. With the exception of the Bronx, each of theboroughs has much higher unevenness, or more segregation, inthe distribution of the homeless than of the poor. In New Yorkoverall, the unevenness index is 35 percent higher for the home-less distribution than for the poverty distribution, and inPhiladelphia, the index is 57 percent higher for the distributionof homelessness than for poverty.


Tab

le 4

. Sh

are

s o

f th

e H

om

eles

s a

mo

ng

Bo

rou

gh

s in

New

Yo

rk (

1987

–199

4) a

nd

Ph

i la

del

ph

ia (

1990

–199

4)

New

Yor

k

Man

hat

tan

Bro

nx

Bro

okly

nQ

uee

ns

Sta

ten

Isl

and

Tot

alP

hil

adel

phia

Nu

mbe

r of

fam

ilie

s30

5,36

829

1,97

856

3,28

349

5,62

599

,464

1,75

5,71

838

1,33

9N

um

ber

of h

omel

ess

fam

ilie

s11

,207

15,4

7516

,875

4,92

763

949

,123

5,05

6H

omel

ess/

fam

ilie

s (%

)3.

675.

302.

990.

990.

642.

801.

33L

ocat

ion

qu

otie

nt

1.31

1.89

1.07

0.36

0.23

NA

NA

Not

e: N

A =

not

ava

ilab

le.

Tab

le 5

. L

oca

tio

n Q

uo

tien

ts o

f th

e H

om

eles

s (N

um

ber

of

Tra

cts

an

d P

erce

nt

of

To

tal)

New

Yor

k

Loc

atio

n Q

uot

ien

tM

anh

atta

nB

r on

xB

r ook

lyn

Qu

een

sS

tate

n I

slan

dT

otal

Ph

ilad

elph

ia

Zer

o40

5016

219

230

474

229

13.8

4%14

.84%

21.0

7%29

.31%

30.6

1%22

.07%

65.6

2%<

1.0

015

115

438

125

739

1,04

88

52.2

5%45

.70%

49.5

4%39

.24%

39.8

0%48

.79%

2.29

%>

1.0

198

133

226

206

2962

611

233

.91%

39.4

7%29

.39%

31.4

5%29

.59%

29.1

4%32

.09%

Tot

al28

933

776

965

598

2,14

834

9M

issi

ng*

718

1918

365

18

* T

he

nu

mbe

r of

cen

sus

trac

ts w

ith

pop

ula

tion

un

der

100.


Tab

le 6

. L

oca

tio

n Q

uo

tien

ts o

f th

e P

oo

r (b

elo

w P

ov

erty

Lev

el)

(Nu

mb

er o

f T

ract

s a

nd

Per

cen

t o

f T

ota

l)

New

Yor

k

Loc

atio

n Q

uot

ien

tM

anh

atta

nB

ron

xB

rook

lyn

Qu

een

sS

tate

n I

slan

dT

otal

Ph

ilad

elph

ia

Zer

o7

1213

204

5615

2.36

%3.

45%

1.66

%2.

99%

4.00

%2.

55%

4.18

%<

1.0

015

218

548

341

387

1,38

719

851

.18%

53.1

6%61

.84%

61.8

3%87

.00%

63.2

2%55

.15%

> 1

.01

138

151

285

235

975

114

646

.46%

43.3

9%36

.49%

35.1

8%9.

00%

34.2

3%40

.67%

Tot

al29

734

878

166

810

02,

194

359

Mis

sin

g*1

78

51

228

* T

he

nu

mbe

r of

cen

sus

trac

ts w

ith

pop

ula

tion

of

zer o

.

Tab

le 7

. I n

dic

es o

f U

nev

enn

ess,

Co

nti

gu

ity

, a

nd

Clu

ster

ing

of

the

Ho

mel

ess

an

d t

he

Po

or

New

Yor

k

Man

hat

tan

Br o

nx

Br o

okly

nQ

uee

ns

Sta

ten

Isl

and

Tot

alP

hil

adel

phia

Un

even

nes

sH

omel

ess

0.56

0.40

0.49

0.56

0.63

0.54

0.58

Poo

r0.

390.

400.

330.

290.

360.

400.

37

Con

tigu

ity

Hom

eles

s0.

590.

610.

210.

630.

590.

620.

52P

oor

0.50

0.64

0.59

0.31

0.37

0.65

0.54

Clu

ster

ing

Hom

eles

s0.

810.

840.

870.

830.

800.

860.

85P

oor

0.75

0.84

0.79

0.73

0.72

0.80

0.72


Figure 1. Census Tract Map of the Distribution of the Prior Addressesof the Homeless in Philadelphia, 1990–1994

According to the clustering index created for this study, in fourof the boroughs (Manhattan, Brooklyn, Queens, and StatenIsland) and in both cities overall, origins of the homeless are,again, more clustered than those of the poor. The Bronx is theonly jurisdiction with an equal clustering score for poverty andhomelessness, again consistent with the other evidence showinga more widespread area of risk of homelessness that more closelyparallels the poverty distribution.

LQ (percent of the homelessin tract/percent in the city) 0.00 or missing 0.01–0.50 0.51–1.00 1.01–2.00 2.01 or greater

1.4 0 1.4 2.8 Miles

N


Regression results

New York, pooled sample. Among the demographic variables,indeed among all variables in the model, the proportion ofAfrican-American persons in a tract is the most importantpredictor, in terms of the standardized coefficient (table 8). Theratio of female-headed households with children under age six isthe second strongest predictor among demographic variables,even though a variable for the ratio of female-headed householdsis included and is nearly significant in the predicted direction(� = 0.040, p = 0.110). Contrary to our hypothesis, tracts with

Figure 2. Census Tract Map of the Distribution of the Poor inPhiladelphia, 1990


Figure 3. Census Tract Map of the Distribution of the Prior Addressesof the Homeless in New York City Boroughs, 1987–1994

more immigrant households are less likely to have shelter admis-sions. When this variable is removed in New York Model II, thesign for crowding reverses to become negative, suggesting thatthere is a positive relationship between immigrant communitiesand crowding that reduces the likelihood of shelter admissions.Coefficients for other demographic variables—such as the ratioof persons without a high school diploma, the ratio of subfamilies(families with children who are part of a larger household), andthe ratio of Hispanic households—are significant and in thepredicted positive direction, though of relatively lower magni-tude. The ratio of persons under 18 was negatively associatedwith shelter admissions (opposite the predicted direction), aswas the ratio of persons over the age of 64. The coefficient for


the variable for older families with children is in the predicteddirection, and the coefficient for the variable for persons in groupquarters is opposite the predicted direction, but neither is statis-tically significant.

Among economic variables, the ratio of poor households is themost important factor. The coefficient for the rate of labor forcenonparticipation is also significant and in the predicted direc-tion. The effect of the ratio of temporarily employed persons isnot significant but is in the predicted direction. Effects of theratio of unemployed persons and the mean household publicassistance income variables are not significant, although thepublic assistance variable is nearly significant in the positive

Figure 4. Census Tract Map of the Distribution of the Poorin New York City Boroughs, 1990

LQ (percent of the poorin tract/percent in the city) 0.00 or missing 0.01–0.50 0.51–1.00 1.01–2.00 2.01 or greater

2.9 0 2.9 5.8 Miles

N


Tab

le 8

. W

LS

Est

ima

tio

n R

esu

lts

for

the

Po

ole

d S

am

ple

s

New

Yor

k I

New

Yor

k I

IP

hil

adel

phia

Sta

nda

rdS

tan

dard

Sta

nda

rdV

aria

ble

Coe

ffic

ien

tp

Coe

ffic

ien

tp

Coe

ffic

ien

tp

Dem

ogra

phic

R

BL

AC

K0.

363*

**0.

000

0.34

2***

0.00

00.

219*

**0.

001

R

SP

AN

0.09

8***

0.00

00.

081*

**0.

000

0.02

90.

594

R

UN

DE

R18

–0.0

38**

0.03

3–0

.058

***

0.00

1–0

.041

0.52

5

RO

VE

R64

–0.1

21**

*0.

000

–0.1

24**

*0.

000

–0.0

250.

771

R

NO

HIG

H0.

080

0.00

00.

057*

**0.

000

0.01

60.

833

R

FH

HO

LD

0.04

00.

110

0.08

5***

0.00

10.

201*

*0.

042

R

FY

OU

CH

D0.

186*

**0.

000

0.19

6***

0.00

00.

007

0.92

8

RO

LD

FA

M0.

014

0.44

80.

008

0.65

90.

050

0.44

9

RS

UB

FA

M0.

091*

**0.

000

0.08

3***

0.00

00.

089

0.12

1

RG

RP

QU

AT

–0.0

030.

791

–0.0

020.

831

–0.0

470.

234

R

FR

BN

RN

70–0

.148

***

0.00

00.

013

0.69

9

Ec o

nom

ic

RU

NE

MP

–0.0

010.

980

0.01

80.

472

0.20

1**

0.02

2

MN

HH

PA

I0.

040*

0.07

20.

064*

**0.

005

–0.0

240.

702

M

ED

HH

INC

0.06

2**

0.02

30.

095*

**0.

001

–0.0

660.

465

R

NO

PO

V0.

204*

**0.

000

0.24

8***

0.00

00.

264*

**0.

007

R

NO

WO

RK

0.04

2***

0.05

00.

024

0.27

2–0

.181

**0.

013

R

TM

PW

OR

K0.

006

0.56

50.

006

0.55

30.

084*

0.05

1

Hou

sin

g an

d n

eigh

borh

ood

qual

ity

M

ED

CO

RE

N–0

.080

***

0.00

1–0

.128

***

0.00

00.

029

0.72

3

RR

EN

T0.

008

0.66

6–0

.025

0.18

30.

034

0.57

3

RE

NT

HIN

C0.

072*

**0.

000

0.08

2***

0.00

0–0

.150

**0.

027

R

CR

OW

D0.

049*

**0.

001

–0.0

34**

*0.

005

–0.1

35**

0.03

1


Tab

le 8

. W

LS

Est

ima

tio

n R

esu

lts

for

the

Po

ole

d S

am

ple

s (c

onti

nu

ed)

New

Yor

k I

New

Yor

k I

IP

hil

adel

phia

Sta

nda

rdS

tan

dard

Sta

nda

rdV

aria

ble

Coe

ffic

ien

tp

Coe

ffic

ien

tp

Coe

ffic

ien

tp

RV

AC

0.08

0***

0.00

00.

094*

**0.

000

0.01

40.

809

RB

OA

RD

UP

0.05

8***

0.00

00.

058*

**0.

000

0.25

2***

0.00

0

N2,

107

2,10

734

1R

20.

828

0.81

90.

704

* p

< 0

.10.

**

p <

0.0

5. *

** p

< 0

.01.


direction. The effect of median household income, which is oppo-site the predicted direction and statistically significant, mayproxy for housing market tightness.

Among the housing and neighborhood quality factors, the rent-to-income ratio is significant and positively associated with therate of shelter admission. The association of median contractrent is negative and significant, as expected. The effect of theratio of rental units in an area is not significant. All of the otherneighborhood quality variables are significant and positivelyassociated with the rate of shelter admission, including thevacancy rate, the ratio of boarded-up buildings, and the ratio ofhousing crowding.

Philadelphia. In general, the Philadelphia regression resultsproduced findings qualitatively similar to those of New York,though fewer variables achieved a level of statistical signifi-cance. Once again, the proportion of African-American personsproduced the most significant positive coefficient among demo-graphic variables and, in Philadelphia, is the second most impor-tant predictor as measured by the standardized coefficient. Theeffect of the ratio of female-headed households is also significantand positive. Coefficients for the other variables are in the samedirection as in New York (with the exception of percent foreignborn) but do not reach statistical significance.

Among the economic factors, again, the ratio of poor persons isan important predictor (and the largest standardized coefficientin the Philadelphia model). Median household income isnegatively associated but not significant. The impacts of theunemployment rate and the proportion of temporary workers arealso significant (nearly significant in the case of temporaryworkers, p = 0.051) and positively correspond to the rate ofshelter admission, although neither was significant in New York.The coefficient for mean public assistance income is not signifi-cant. The coefficient for persons not in the labor force is nega-tive, opposite that found in New York.

Among the housing and neighborhood variables (including me-dian contract rent as a control variable), the most significantpredictor (and among the most important variables in the Phila-delphia model overall) is the proportion of boarded-up buildings.Coefficients for both the crowding and the rent-to-income ratiovariables are significant, but with negative signs (opposite thatfound for New York), suggesting that homeless families in Phila-delphia come from areas that are less crowded and more “afford-able” than other parts of the city, perhaps because of the low


neighborhood quality and the comparatively lower cost of hous-ing in Philadelphia. Coefficients for the vacancy rate and propor-tion of rental units variables are not significant.

New York, comparison between low-income areas and higher-income areas. We used median household income to define low-and higher-income areas in New York, with the citywide medianvalue of each tract’s median household income as the breakpoint. In New York, census tracts that have a median householdincome lower than $30,609 are categorized as low-income neigh-borhoods and the remainder as higher-income.7

Results for most demographic variables are similar to those ofthe pooled sample (table 9). Coefficients relating to the propor-tion of African-American persons, Hispanics, female-headedhouseholds with young children, subfamilies, immigrants, andpersons lacking a high school education are all significant andhave the same sign in both areas as in the pooled sample.

Among economic factors, effects of the poverty rate and the rateof labor force nonparticipation are also positive and significantin both areas. However, the mean household public assistanceincome is now significant and positive in predicting shelteradmissions in high-income areas, but negative (though notsignificant) in low-income areas. Unemployment and temporarywork remain not significant.

Among the housing and neighborhood variables, the impact ofthe proportion of rental units is now significant in both areas,though positively associated in high-income tracts and nega-tively associated in low-income tracts. The positive association ofhomelessness to an area’s rent-to-income ratio holds only in low-income tracts. The neighborhood quality variables (crowding,vacancy, boarded-up buildings) are all positively associated andsignificant.

Discussion

While homeless households appear to come from areas with highrates of poverty, areas with the greatest risk of homelessness aregenerally more densely clustered than poor areas. In both cities,

7 We did not make a similar comparison for Philadelphia because there weretoo few observations. We used the Chow test to check for structural differenceswith the null hypothesis that the regressions of the low- and high-incomegroups are identical. The results show that there are structural differences ata statistically significant level (F22, 2107 = 6.12, p = 0.00).


Table 9. WLS Estimation Resultsfor Low- and Higher-Income Areas in New York

Higher-Income Tracts Low-Income Tracts

Standard StandardVariable Coefficient p Coefficient p

DemographicRBLACK 0.567*** 0.000 0.362*** 0.000RSPAN 0.150*** 0.000 0.096*** 0.000RUNDER18 –0.015 0.553 –0.038 0.158ROVER64 –0.173*** 0.000 –0.104*** 0.001RNOHIGH 0.068*** 0.007 0.075*** 0.000RFHHOLD –0.015 0.663 0.037 0.325RFYOUCHD 0.079*** 0.001 0.187*** 0.000ROLDFAM –0.019 0.570 0.031 0.213RSUBFAM 0.226*** 0.000 0.082*** 0.000RGRPQUAT –0.050** 0.026 0.030* 0.053RFRBRN70 –0.203*** 0.000 –0.130*** 0.000

EconomicRUNEMP –0.058 0.208 –0.050 0.151MNHHPAI 0.168*** 0.000 –0.041 0.231MEDHHINC –0.028 0.660 0.045 0.429RNOPOV 0.120*** 0.000 0.149*** 0.000RNOWORK 0.100** 0.025 0.077** 0.012RTMPWORK –0.002 0.916 0.002 0.901

Housing and neighborhood qualityMEDCOREN 0.053 0.380 –0.161*** 0.000RRENT 0.076*** 0.007 –0.064*** 0.004RENTHINC –0.028 0.616 0.091** 0.011RCROWD 0.069*** 0.007 0.043** 0.029RVAC 0.051** 0.020 0.093*** 0.000RBOARDUP 0.096*** 0.000 0.041** 0.016

N 1,031 1,030R2 0.704 0.809

* p < 0.10. ** p < 0.05. *** p < 0.01.

the distribution of homeless families’ prior addresses is morehighly segregated than the poverty distributions. An exception tothis pattern is the Bronx, where the rate of shelter admissions ismore evenly distributed among the borough’s poor neighbor-hoods, and where the level of risk appears generally high. But, ingeneral, homeless families come primarily from a subset of poorneighborhoods where some additional set factors contribute totheir increased risk of public shelter admission.

The regression results support several of the hypotheses con-cerning the neighborhood characteristics associated with therate of public-shelter admissions among families. We will focus


primarily on the New York regression results, which benefitedfrom more observations (census tracts) and thus greater statisti-cal power. We will discuss the Philadelphia results in light ofdifferences between the two cities.

The rate of public-shelter admissions from an area increaseswith the proportion of African Americans and female-headedhouseholds (FHH), particularly those with young children, andto a lesser extent with Hispanic households. These results werepredicted, based on previous research, which has shown thathomeless families are disproportionately composed of minoritiesand FHH. Variables for race and FHH with young childrencontinue to be strongly associated with the rate of shelter admis-sion, even controlling for the rate of poverty, welfare receipt,educational attainment, and various housing and labor marketvariables, which indicates that such households face additionalbarriers to residential stability not specified in this model.

Areas with high concentrations of FHH may be at greater riskbecause of a higher level of risk among individuals in thoseareas, such as having more limited social network size, higherrates of substance abuse and mental-health problems, and otherindividual risk factors. However, research comparing housed andhomeless Aid to Families with Dependent Children recipients inNew York City has shown that there are few such differences inindividual risk factors among public assistance recipients andthat such individual risk factors affect a relatively small propor-tion of families entering shelters (Knickman and Weitzman1989). In addition to these individual-level effects, it is likelythat other social and economic barriers, such as restrictedresidential mobility, limited labor-market access, and variousneighborhood effects, have a differential negative impact onFHH with young children and contribute to both their concentra-tion in low-rent areas and their increased risk of public-shelteradmission.

A similar set of dynamics may contribute to race and povertyconcentrations in a neighborhood, which are among the mostsignificant predictors of shelter admissions for both cities. Again,the increased risk of shelter admission may be partially attribut-able to a larger number of individual-level risk factors amongsuch groups. However, research has found that race has anadditional positive effect on public-shelter use that has not beenexplained by individual risk factors. For example, African-American single adults in Philadelphia have been found to havea significantly longer homelessness duration (controlling forhistory of mental-health and substance-abuse treatment), and


African-American homeless families in New York City have beenfound to have a significantly higher probability of readmission toshelters (controlling for reason for homelessness and type ofshelter discharge) (Culhane and Kuhn 1995; Wong, Culhane, andKuhn forthcoming). Thus, apart from individual risk factors,other social and economic factors are likely to contribute to thedifferential exposure of predominantly African-American andpoor neighborhoods to the risk of public-shelter admission. Forexample, the high degree of spatial clustering among homelessfamilies’ prior addresses found in the descriptive results and thesignificance of the effects of race and poverty concentrationsuggest that processes of racial and economic segregation con-tribute to the increased risk of shelter admission. Such an effectwould be consistent with research on the impact of segregationon housing and neighborhood quality. Massey and Denton (1993)have found that increasing racial segregation has interactedwith declining income to produce higher poverty concentrationsamong African Americans over the past two decades, which theauthors argue has promoted disinvestment in these communitiesby concentrating tenants with a decreasing ability to pay marketrents in financially distressed buildings. This concentration canproduce a “hollowing out” effect, in which units and buildings aremore likely to be left vacant or abandoned, a portrait consistentwith the neighborhoods identified in this study as being at great-est risk of generating homelessness. Housing market forces andgovernment policies may contribute to increases in spatial strati-fication by income and race (Schill and Wachter 1995).

The finding that homeless families come from areas with moresubfamilies, together with the significance of the crowdingvariable, provides empirical support for the hypothesis thathomelessness is one consequence of “doubling up” in an area.Families doubling up are presumably doing so because of a lackof income for independent household formation. Aside from beingat greater risk of a housing emergency because of crowding,people in doubled-up arrangements may also expend sources ofsocial support more quickly in the event of a crisis. For example,people in subfamilies are often already living with parents orother family members prior to public-shelter admission; thus,they have exhausted some of the housing alternatives to whichothers might have access in the event of a housing emergency.

The interesting exception to the heightened risk associated withcrowding is found among recent immigrants. The reversal of thesign relating to the crowding variable in the New York Model II,when the ratio for foreign-born persons is excluded from theregression analysis, suggests that immigrant groups mitigate the


risk of homelessness by increased crowding. Such groups mayhave developed adaptations to crowding that prevent or resolvehousing emergencies. Alternatively, such persons may be lesswilling to seek the support of the public-shelter system, eventhough they may need its services. This area deserves furtherstudy, as immigrant communities’ accommodations to crowdingmay help to inform the design of prevention efforts for othercommunities and families confronting crowding or, alternatively,may reveal a greater need for outreach to immigrant families inneed of public shelter. Further research on accommodations tohousing distress may also help to explain differences in shelteradmissions by race and ethnicity.

The results provide support for hypotheses that family home-lessness is related to housing and neighborhood conditions.Homeless families often come from deteriorated and low-rentneighborhoods, as measured by the ratio of boarded-up buildings(among the most important variables in the Philadelphia model)and the median contract rent. Homeless families are also morelikely to come from neighborhoods with higher vacancy rates,suggesting that these areas are viewed as relatively undesirableand that the rental housing in these areas is at risk of under-maintenance and abandonment. Affordability matters as well, asindicated by the positive effect of the rent-to-income ratio, con-firming the hypothesis that shelter admissions are more likely tooccur in areas with a relatively greater rent burden.

The Philadelphia data generally support the findings from NewYork, though with less statistical significance. Some differencesare worth noting, particularly because they might be a functionof differences in housing and labor markets, as well as in publicpolicies between the two cities. Among the demographic vari-ables, the effect of the ratio of foreign-born persons immigratingsince 1970 is not significant in Philadelphia, nor is the effect ofthe Hispanic variable. Hispanics constitute a relatively smallproportion of the population in Philadelphia (5.6 percent versus24.4 percent in New York) and are known to be underrepre-sented among shelter users there. Hispanics and recent immi-grants in Philadelphia may be subject to dynamics similar tothose of the recent immigrants in New York, whose relativelygreater crowding may be an alternate accommodation to housingdistress.

Among economic variables, unemployment is significantly re-lated to shelter admission rates in Philadelphia but not in NewYork. This finding may indicate a relatively greater problem ofunemployment in some of that city’s neighborhoods. (However,


the effect of the rate of labor force nonparticipation is significantand positively related to shelter admissions in New York.) Theimportance of labor market opportunities as a contributing factorin Philadelphia is amplified by the added positive significance ofpersons temporarily employed.

Finally, among the housing and neighborhood variables, thenegative association between the rent-to-income ratio and therate of shelter admission likely reflects the comparatively lowercost of housing in low-rent areas in Philadelphia (compared withNew York), because population loss has resulted in higher va-cancy rates. The relatively greater importance of abandonmentas a predictor in Philadelphia compared with New York couldalso be related to Philadelphia’s continuing population loss, aswell as the higher rate of immigration in New York, where immi-grants fill some of the low-cost housing that might otherwisehave been left vacant. Differences between cities in the disposi-tion of abandoned and tax-foreclosed properties may also help toexplain the more limited effect of abandonment in New York,where local government has assumed more direct responsibilityfor the management and rehabilitation of tax-foreclosedproperties.

Separating tracts by median income in New York also producedsome interesting differences from the pooled sample. First, themodel performed better for low-income than for higher-incometracts. However, among higher-income tracts, the ratio ofAfrican-American persons increases in importance in terms ofthe standardized coefficient, again raising concerns about theincreased risk of homelessness among African-American commu-nities, even those with relatively higher income. The effect of theproportion of rental units also appears more significant in thesemodels, tending to be positive in the case of higher-income areasand significantly negative in low-income areas. This finding maysuggest that in low-income areas, homeowner-related housingproblems, such as the inability of aging parents or their adultchildren to maintain the costs of the home, may play a role inincreasing the risk of homelessness. In higher-income areas,homelessness is more often related to problems with rentalhousing and its unaffordability.

From a policy perspective, this research offers two broad in-sights. First, because the risk of family homelessness is spatiallyand demographically concentrated, homelessness prevention andoutreach efforts would likely benefit from a geographic- andpopulation-targeted strategy. Policies designed to counteractresidential segregation, concentrated poverty, and poor housing


and neighborhood conditions, as well as more narrowly definedhomelessness-prevention programs, could target the neighbor-hoods found to be at greatest risk for generating shelter admis-sions in this study and the population groups they overrepresent.Second, this study has identified some of the associated factorsthat could guide the substance of a prevention-oriented policystrategy. For example, improved household income, throughexpanded rental assistance, improved access to employmentincome, or increased public assistance benefits, would reduce thepoverty and housing unaffordability that this study found to beassociated with a higher rate of shelter admission. An income-support or housing-subsidy program could also reduce crowding,vacancies, and possibly abandonment, as well as the potentialreinforcing effect of these problems on the risk of shelter admis-sions. Further research is needed to model and test the impact ofsuch policy strategies.

Our study was limited in that the dependent variable repre-sented an aggregation of homeless families’ responses to a singlequery regarding their prior address. Although intake forms forboth cities’ systems provide some standardization for collectinginformation, there are no scripts for collecting information frompeople seeking shelter admission. Some unknown rate of falsereporting could also occur because people are responding toquestions that partly determine their access to or eligibility forservices. Moreover, having found significant and theoreticallyconsistent associations among neighborhood-level variables, thestudy’s results do not diminish the importance of other levels ofcausal influence, such as intercity effects, emergency-assistancepolicies, household dynamics, behavioral adaptations, or otherindividual risk variables. Each of these may influence whoamong the persons in these areas is at greatest risk of shelteradmission and how that risk is distributed geographically. Amultilevel or hierarchical analysis would be necessary to exam-ine the differential impact of these factors in a more systematicmanner.

Finally, future research should further develop and refine thisanalytic approach for studying homelessness. This study waslimited in treating shelter-admission data and predictive vari-ables cross-sectionally, whereas a more time-sensitive treatmentof these variables would be better able to capture the dynamicsof change, including population composition, neighborhood qual-ity, and housing conditions. This study was also limited by anaggregation of variables at the level of the census tract, whereasfurther analyses could examine the block-group-level predictorsor even the characteristics of specific properties associated with


the risk of shelter admission. The analysis could also be furtherrefined by including various spatial measures in the specificationof the regression model. Researchers should consider replicatingthis work in other localities. Such research could be undertakenin areas without computerized shelter-tracking systems byselecting a representative sample of shelter admissions over agiven period of time and surveying people about their housinghistory. The prior-address data could be enhanced by includingmore detailed questions regarding the housing arrangementsof clients. Such an approach would bring greater depth to theunderstanding of the interaction of neighborhood- andhousehold-level dynamics of housing instability than we coulddiscern with the data available for this study.

Conclusion

This study has provided empirical support for several hypothesesregarding the influence of housing and income problems ingenerating homelessness in New York City and Philadelphia,particularly as they disproportionately affect women withyoung children and African Americans. The rate of public-shelteradmission was found to be associated with housing crowding,residence in subfamilies, poverty, restricted access to the labormarket, rent burden, and poor neighborhood quality. Futurepublic policies should consider the role of geographic and demo-graphic variations in the risk of homelessness in designinginterventions to reduce that risk.

Appendix

Qualifications of the study population. To assess the degree ofbias in the selection of the matched versus unmatched cases, weconducted t tests comparing the groups in each respective city byrace and gender of household head (see table A-1). In New York,the matched addresses were significantly more likely to be com-posed of African-American households (t = 11.445, p < 0.0001).The matched addresses were also significantly less likely to becomposed of Hispanic households (t = –7.851, p < 0.0001). How-ever, as shown in table A-1, none of the mean differences be-tween groups was large enough to warrant great concern withthe representativeness of the study population (+5.8 percentage-point difference for African American and –3.9 for Hispanic).Nevertheless, study findings will remain qualified by the factthat the study population for New York City is slightly morelikely to represent African-American households and slightly


Table A-1. Demographics of Matched and Nonmatched Samplesin Philadelphia and New York

African American Hispanic

PhiladelphiaMatched sample Proportion 0.917 0.036

N 5,332 5,098Nonmatched sample Proportion 0.875 0.058

N 3,745 3,497t test for H0 t 41.897 23.181

p 0.000 0.000Result Can reject H0 Can reject H0

New YorkMatched sample Proportion 0.651 0.325

N 36,296 36,296Nonmatched sample Proportion 0.593 0.364

N 12,427 12,427t test for H0 t 11.445 –7.851

p 0.000 0.000Result Can reject H0 Can reject H0

Notes: Total number of observations varies due to missing values. H0 = null hypothesis.The means between matched and nonmatched samples are the same.

less likely to represent Hispanics than the overall homelessfamily population.

In Philadelphia, the address-matched group (the study popula-tion) is more likely to include African-American households(t = 41.897, p < 0.0001) and to underrepresent Hispanic house-holds (t = 23.181, p < 0.0001). Again, the mean differences arenot large (+4.2 percentage points for African American and –2.2for Hispanic).

We undertook an additional procedure to assess whether a geo-graphically distributed bias operated in the matching and rejec-tion of reported addresses within each city by the geocodingprocedures. It is possible that inaccurate base maps or system-atically unconventional address reporting resulted in a biaseddistribution of matched versus rejected in-city addresses.Matched and rejected addresses within each city were thusgeocoded by zip code, and the correlation coefficient was com-puted between the distributions (see table A-2). The matchedand rejected addresses are highly similar in distribution in NewYork City (r = 0.877) and nearly identical in Philadelphia(r = 0.972), showing that the geographic distribution of the studypopulation in both cities is highly representative of all householdsreporting in-city addresses, at the zip code level (see table A-2).


Finally, because these systems were designed for managementand not research purposes, both systems are limited in thathouseholds with multiple admissions to shelters have their prioraddress information overwritten at the time of subsequent re-admissions. In other words, an address history is not retained forhouseholds with multiple shelter admissions. It is conceivablethat households readmitted to shelters may have a significantlydifferent locational distribution than households presenting toshelters for the first time. For example, households with mul-tiple admissions may be disproportionately discharged by shelterprograms to housing in more or less stable areas relative to thelocational origins of households with a single admission. Toassess the degree of bias introduced by this possibility, and toassess whether single- and multiple-admission householdsshould be separated for the purposes of the distributional mea-surements for this study, the correlation between the distribu-tion of prior addresses for households with single versus multipleadmissions by zip code for all matched addresses was computedin both cities. Again, however, the distributions are highly simi-lar in New York City (r = 0.992) and Philadelphia (r = 0.999),suggesting that such a locational difference does not occur at thezip code level and would not warrant further adjustment (seetable A-2).

Table A-2. Correlations between the Number of Homeless Familiesin Each Zip Code in New York and Philadelphia

Pair of Comparison New York Philadelphia

Address-matched and nonmatched sample 0.877 0.972

Single and multiple admissions 0.992 0.999

Measures of area variations in homelessness rates include thelocation quotient, unevenness, contiguity, and clustering.

The location quotient (LQ). The LQ cannot have a value less thanzero. When the LQ in a locality is greater than 1.00, the localityhas a higher concentration of the subject of interest relative tothe other localities of the region combined. Thus, the LQ is usedto identify census tracts that contain a higher percentage shareof the prior addresses of the homeless, the poor (people belowpoverty level, as reported in 1990 census), and minority poorthan that of Philadelphia or New York as a whole. Because of itsunitlessness and absolutivity, the LQ also permits intercity andinterborough comparisons of the spatial distribution of thesubject of interest.


Unevenness. The dissimilarity index varies between 0 and 1, andconceptually it represents the proportion of an object group thatwould have to change its location to achieve an even distribution.The dissimilarity index is calculated as follows:

Dissimilarity = ∑i

ti

ri

− R

2TR(1 − R)

, (A.1)

where ti is population of areal unit i; ri is homeless proportion ofareal unit i; and T and R are the total population and the propor-tion of an object group in the whole, respectively.

Contiguity. The difference between contiguity and unevenness iswell illustrated by comparing the case of the “checkerboardproblem” (White 1983); highly concentrated areas are located ina scattered fashion like the dark squares on a checkerboard, witha pattern in which the dark and light areas are each clusteredtogether to form two halves on the board (one light, one dark).Both patterns yield the same unevenness index value, althoughthey clearly have different distributional patterns in terms ofspatial association. A contiguity index is used to capture thisdifference in spatial association. In this study, we used an indexof spatial autocorrelation to measure the degree of contiguity. Ifobjects that are similar in location also tend to be similar inattributes, the pattern as a whole is said to show positiveautocorrelation. Conversely, if objects that are close together inspace tend to be more dissimilar in attributes than objects thatare farther apart, then negative spatial autocorrelation is dis-played (Shen 1994). Moran’s I is used to calculate spatialautocorrelation, and its mathematical notation is as follows(Odland 1988):

I = nw

ijj

∑i∑

wij

pi

− p( )j

∑i∑ p

j− p( )

pi

− p( )2

i∑

, (A.2)

where n is the number of census tracts; the double summationindicates summation over all pairs of tracts; pi is the ratio of anobject group of tract i to the population of tract i; p is the meanof pi; and wij is a proximity weight for the pair of tract i and tractj, which is 0 when i equals j.

In the geographic literature, the quantity wij refers to an ele-ment in “contiguity matrix” that equals 1 when census tracts iand j are contiguous and 0 otherwise. In this article, adjacent


tracts of every census tract were identified by using a geographicinformation system.

Clustering. As a first step in creating a clustering index, wedivided census tracts into two groups, based on the LQ: highlyconcentrated census tracts and census tracts with low concentra-tions. When two highly concentrated tracts are adjacent to eachother, the common boundary lines are deleted and the two poly-gons of the tracts are merged to form one polygon. If this merg-ing process keeps going, a few polygons that represent highlyconcentrated areas are obtained. The more unevenly distributedan object group is, the smaller number of tracts categorized asthe highly concentrated area will be. The more clustered thehighly concentrated tracts are, the more common boundaries areerased, and the smaller the ratio of the sum of the perimeters ofthe merged polygons to the sum of the perimeters of the originalareal units will be. In this concept, the clustering index (CI) canbe denoted as follows:

CI = 1 −b

′i ′j′j

∑′i

∑b

ijj

∑i∑

, (A.3)

where bij is the length of common boundary between censustracts i and j before polygon merging; i´ and j´ are a pair ofcensus tracts that form the boundaries between the highly con-centrated areas and the sparsely concentrated areas.

In the checkerboard example, the sum of the perimeters of highlyconcentrated polygons persists after the merging process, and CIwill be 0. In the opposite extreme, when the object group is con-centrated in a few census tracts adjacent to one another, a singlehighly concentrated polygon will remain after the polygonmerging process, and CI will be close to 1 but will not exceed 1(see Lee and Culhane [1995] for diagrammatic examples).

Recently, Wong (1993) formulated a new segregation index thatuses the length of the common boundary of two areas as anindicator of the degree of social interaction between the resi-dents of the two areas. In a similar context, the total length ofcommon boundaries between the two areal groups (for racialsegregation, minority area, and majority area) may be inter-preted as the total possibility of social interaction between thetwo groups. The total length of the common boundaries betweenthe areas belonging to the same areal group may be interpretedas the total possibility of social interaction within a group.Therefore, the clustering index measures how small total social


interaction between the two groups is compared to the sum oftotal interaction between the two groups and total interactionwithin groups.

Authors

Dennis P. Culhane is Associate Professor in the School of Social Work andResearch Associate Professor in the Center for Mental Health Policy andServices Research at the University of Pennsylvania. Chang-Moo Lee is SeniorFellow at The Wharton School Real Estate Center of the University of Penn-sylvania. Susan M. Wachter is Professor of Real Estate and Finance at TheWharton School of the University of Pennsylvania.

This research was supported by a grant from the Edna McConnell ClarkFoundation, Program for New York Neighborhoods. The authors gratefullyacknowledge the assistance of Mon Louie of the City of New York HumanResources Administration, Heide Lange-Joe of the City of New York Depart-ment of Homeless Services, and Joseph Henry and June Averyt of the Univer-sity of Pennsylvania for their assistance with this project.

References

Applebaum, Richard, M. Dolny, Peter Dreier, and John Gilberbloom. 1991.Scapegoating Rent Control: Masking the Causes of Homelessness. Journal ofthe American Planning Association 57:153–64.

Applebaum, Richard, M. Dolny, Peter Dreier, and John Gilberbloom. 1992.Sham Rent Control Research: A Further Reply. Journal of the AmericanPlanning Association 58:220–24.

Bendavid-Val, Avrom. 1983. Regional and Local Economic Analysis for Practi-tioners. New York: Praeger.

Blau, Joel. 1992. The Visible Poor: Homelessness in the United States. NewYork: Oxford University Press.

Burt, Martha. 1992. Over the Edge: The Growth of Homelessness in the 1980s.New York and Washington: Russell Sage Foundation and The Urban InstitutePress.

Burt, Martha. 1994. Comment. Housing Policy Debate 5(2):141–52.

Culhane, Dennis P. 1990. The Social Selection Processes of Homelessness. InOn Becoming Homeless: The Structural and Experiential Dynamics of Residen-tial Instability, 135–47. Unpublished doctoral dissertation. Boston College.

Culhane, Dennis P. 1992. The Quandaries of Shelter Reform: An Appraisal ofEfforts to “Manage” Homelessness. Social Service Review 66(3):428–40.

Culhane, Dennis P., Edmund F. Dejowski, Julie Ibanez, Elizabeth Needham,and Irene Macchia. 1994. Public Shelter Admission Rates in Philadelphia andNew York City: The Implications of Turnover for Sheltered Population Counts.Housing Policy Debate 5(2):107–40.


Culhane, Dennis P., and Randall Kuhn. 1995. Patterns and Determinants ofShelter Utilization Among Single Adults in New York City and Philadelphia: ALongitudinal Analysis of Homelessness. Paper presented at the AnnualMeeting of the Eastern Sociological Association, March 31, Philadelphia, PA.

Elliot, Marta, and Lauren J. Krivo. 1991. Structural Determinants ofHomelessness in the U.S. Social Problems 38(1):113–31.

Fournier, Louise, Malijai Caulet, Gilles Cote, Jean Toupin, Maurice Ohayon,Micheline Ostoj, and Isabelle Laurin. 1994. Longitudinal Study of the NewHomeless: Preliminary Results. Paper presented at the Annual Meeting of theAmerican Public Health Association, November 1, Washington, DC.

Hopper, Kim. 1990. The New Urban Niche of Homelessness: New York City inthe Late 1980s. Bulletin of the New York Academy of Medicine 66(5):435–50.

Hopper, Kim. 1995. Definitional Quandaries and Other Hazards in Countingthe Homeless: An Invited Commentary. American Journal of Orthopsychiatry65(3):340–46.

Hopper, Kim, and Jill Hamberg. 1986. The Making of America’s Homeless:From Skid Row to New Poor, 1945–1986. In Critical Perspectives on Housing,eds. Rachel Bratt, Chester Hartman, and Ann Meyerson, 12–40. Philadelphia:Temple University Press.

Interagency Council for the Homeless. 1994. Priority Home: The Federal Planto Break the Cycle of Homelessness. Washington: U.S. Department of Housingand Urban Development.

Jahiel, Rene. 1992. Toward the Prevention of Homelessness. In Homelessness:A Prevention Oriented Approach, ed. Rene Jahiel, 315–36. Baltimore: TheJohns Hopkins University Press.

James, David R., and Karl E. Taeuber. 1985. Measures of Segregation. InSociological Methodology 1985, ed. Nancy Tuman, 1–32. San Francisco:Jossey-Bass.

Kearns, Robin A., and Christopher J. Smith. 1994. Housing, Homelessness andMental Health: Mapping an Agenda for Geographical Inquiry. ProfessionalGeographer 46(4):418–24.

Knickman, James, and Beth Weitzman. 1989. Forecasting Models to TargetFamilies at High Risk of Homelessness. New York: New York UniversityHealth Research Program.

Koegel, Paul, and M. Audrey Burnam. 1994. The Course of HomelessnessAmong Homeless Adults in Los Angeles. Paper presented at the AnnualMeeting of the American Public Health Association, November 1, Washington,DC.

Kondratas, Anna. 1994. Comment. Housing Policy Debate 5(2):153–62.

Lee, Chang-Moo, and Dennis P. Culhane. 1995. The Spatial Distribution ofHomelessness Occurrence and the Efficacy of Shelter Programs in Philadel-phia. Wharton Real Estate Center Working Paper Series No. 213. Universityof Pennsylvania.


Lindblom, Eric N. 1991. Toward a Comprehensive Homelessness-PreventionStrategy. Housing Policy Debate 2(3):631–47.

Link, Bruce G., Ezra Susser, Anne Stueve, Jo Phelan, Robert E. Moore, andElmer Struening. 1994. Lifetime and Five-Year Prevalence of Homelessness inthe United States. American Journal of Public Health 84:1907–12.

Massey, Douglas S., and Nancy A. Denton. 1988. The Dimensions of Residen-tial Segregation. Social Forces 67(2):281–315.

Massey, Douglas S., and Nancy A. Denton. 1993. American Apartheid: Segre-gation and the Making of the American Underclass. Cambridge, MA: HarvardUniversity Press.

Odland, John. 1988. Spatial Autocorrelation. New York: Sage Publications.

Piliavin, Irving, Michael Sosin, Alex H. Westerfelt, and Ross L. Matsueda.1993. The Duration of Homeless Careers: An Exploratory Study. Social ServiceReview 67:576–98.

Quigley, J. 1991. Does Rent Control Cause Homelessness? Taking the ClaimSeriously. Journal of Policy Analysis and Management 9:89–93.

Ringheim, Karen. 1990. At Risk of Homelessness: The Roles of Income andRent. New York: Praeger.

Robertson, Marjorie J., Cheryl Zlotnick, and Alex Westerfelt. 1994. The Courseof Homelessness Among Adults and Families in Alameda County. Paperpresented at the Annual Meeting of the American Public Health Association,November 1, Washington, DC.

Rossi, Peter. 1989. Down and Out in America. Chicago: University of ChicagoPress.

Rossi, Peter. 1994. Comment on Dennis P. Culhane et al.’s “Public ShelterAdmission Rates in Philadelphia and New York City: The Implications ofTurnover for Sheltered Population Counts.” Housing Policy Debate 5(2):163–76.

Schill, Michael H., and Susan M. Wachter. 1995. Housing Market Constraintsand Spatial Stratification by Income and Race. Housing Policy Debate 6(1):141–67.

Shen, Qing. 1994. An Application of GIS to the Measurement of SpatialAutocorrelation. Computer, Environment, and Urban Systems 18(3):167–91.

Stegman, Michael A. 1993. Housing and Vacancy Report: New York City, 1991.New York: Department of Housing Preservation and Development.

Tucker, W. 1987. Where Do the Homeless Come From? National Review,September, pp. 32–45.

U.S. Department of Commerce. 1993. TIGER/Line 1992. Washington, DC:Bureau of Data User Services Division.

U.S. Department of Health and Human Services. 1991. Homelessness Preven-tion Programs. Washington, DC: Office of the Inspector General.


U.S. Department of Housing and Urban Development. 1984. A Report to theSecretary on the Homeless and Emergency Shelters. Washington, DC: Office ofPolicy Development and Research.

Wallace, Roderick. 1989. “Homelessness,” Contagious Destruction of Housingand Municipal Service Cuts in New York City: 1. Demographics of a HousingDeficit. Environment and Planning A 21:1585–1603.

Wallace, Roderick. 1990. “Homelessness,” Contagious Destruction of Housingand Municipal Service Cuts in New York City: 2. Dynamics of a HousingFamine. Environment and Planning A 22:5–15.

White, Michael J. 1983. The Measurement of Spatial Segregation. AmericanJournal of Sociology 88(5):1008–18.

Wolch, Jennifer, and Michael Dear. 1993. Malign Neglect. San Francisco:Jossey-Bass.

Wong, David W. S. 1993. Spatial Indices of Segregation. Urban Studies30(3):559–72.

Wong, Irene, Dennis P. Culhane, and Randall Kuhn. Forthcoming. Predictorsof Shelter Exit and Return among Homeless Families in New York City.Available from authors, University of Pennsylvania.

Wright, James D., and Joel A. Devine. 1995. Housing Dynamics of the Home-less: Implications for a Count. American Journal of Orthopsychiatry 65(3):320–29.


Racial Differences in the Search for Housing 367Housing Policy Debate • Volume 7, Issue 2 367© Fannie Mae Foundation 1996. All Rights Reserved.

Racial Differences in the Search for Housing:Do Whites and Blacks Use the Same Techniquesto Find Housing?

Reynolds FarleyUniversity of Michigan

Abstract

Studies report that real estate brokers often provide less information to blackclients than to whites and steer them to different neighborhoods. However, fewstudies have investigated racial differences in housing search procedures. Ifblacks believe they will receive less useful information from brokers, they mayavoid them.

Analyzing 1992 data from Detroit, this study finds that blacks were signifi-cantly less likely than whites to consult brokers. Blacks tended to rely more onmethods such as talking with friends, checking newspaper ads, or drivingthrough neighborhoods. Blacks were also more likely to believe that theymissed housing opportunities because brokers discriminate. Differences in thesocioeconomic characteristics of subjects account for some of the differences inthe use of brokers. Eradicating discrimination by brokers will broaden housingopportunities only for blacks who use brokers. Policy actions that address theperception of discrimination by brokers may be a more powerful tool.

Keywords: Discrimination; Minorities; Markets

Aims of this investigation

For almost 30 years, fair housing audits have assessed possibleracial discrimination in the sale or rental of housing. They fre-quently report that, when matched pairs of black and whitehomeseekers approach the same real estate broker, race makes adifference. Although blacks are generally shown homes andtreated hospitably, they are often provided with less informationabout the housing market and about financing than whites.These audit studies also report racial steering. More blacks thanwhites are shown homes in racially mixed or in all-black neigh-borhoods; white auditors are seldom shown homes in mixedareas.1

1 For comprehensive summaries of audit studies, see Fix, Galster, and Struyk(1993); Galster (1990); Turner (1992); Yinger (1986, 1993, 1995).

368 Reynolds Farley

The Department of Housing and Urban Development systemati-cally measured racial discrimination in the marketing of adver-tised properties in the largest metropolitan areas, once in 1977and again a dozen years later (Turner, Struyk, and Yinger 1991;Wienk et al. 1979). Summarizing these national surveys, Galster(1992, 647) concluded:

Housing discrimination against black and Hispanichomeseekers and apartment seekers occurs in roughlyhalf of the instances when these persons interact withan agent. . . . Typically this discrimination is subtle innature and therefore difficult for the individual todetect. . . . The frequency of this discrimination has notchanged noticeably since 1977.

While there is now a systematic base of information about howbrokers treat black and white homeseekers, we know much lessabout whether blacks and whites use the same search strategieswhen they enter the housing market. The fair housing auditingtechnique assumes a particular search method and thus cannotprovide evidence on how minority housing seekers adjust theirsearches because of discrimination. By looking at actual searchbehavior, we can learn more about differences between black andwhite homeseekers and the costs that discrimination imposes. Ifblacks believe they will be provided with little information bybrokers and will be steered toward a limited array of options,they may avoid using them and rely on informal search methods,such as talking with friends or scanning newspaper ads. Thus,the elimination of discrimination by brokers will only be a firststep toward increasing housing opportunities for minorities. Thisexploratory study seeks to determine whether blacks and whitesin metropolitan Detroit used similar methods when they recentlysought housing.

Whom we asked: Detroit area homeseekers

Data were gathered from April through July 1992 in the DetroitArea Study (DAS), an annual social science survey conducted bythe University of Michigan’s Department of Sociology. Thisinvestigation focused on the causes of continued racial residen-tial segregation, the nature and extent of racial polarization, andthe status of blacks and women in the labor market. The 1992DAS is one component of the Multi-City Study of Urban Inequal-ity. Similar investigations of the causes of residential segrega-tion, the nature of racial polarization, and the status ofminorities in local labor markets were conducted in Atlanta in1993 and in Boston and Los Angeles in 1994.

Racial Differences in the Search for Housing 369

Using a two-stage study design, researchers first selected arepresentative sample of the Detroit area population age 21 andover. Residential blocks (or block groups in the case of blockswith sparse populations) were sampled proportional to theirpopulation in the April 1990 census. All housing units within theselected blocks were identified by DAS staff and, in a secondstage, a sample of dwelling units was randomly chosen. Inter-viewers visited every sampled home and listed all residents. Atthe door stoop, one household member (age 21 or over) wasrandomly selected for interviewing. Because this study focuseson racial differences, we sought a sufficiently large sample ofblacks so that estimated parameters would have small standarderrors. Consequently, blocks or block groups 70 percent or moreblack in 1990 were sampled at a higher ratio. Data presented inthis article are weighted to reflect this over-sampling of denselyblack neighborhoods.2 For the purposes of this study, the Detroitmetropolitan area includes the city of Detroit, the remainder ofWayne County, and two densely populated contiguous counties(Macomb and Oakland).3

Face-to-face interviews were carried out with 1,543 respondents:750 black, 736 white, and 57 who chose other identities.4 Thisanalysis is restricted to those who said they were black or white.The Latino and Asian populations of Detroit are too small todescribe with a sample of this size. The study design sought tomatch race of interviewer and race of respondent to improve thedata quality, especially for responses about sensitive issues suchas stereotypes and reverse discrimination (Bernick, Pratto, andDavis 1994; Schuman and Converse 1971).5 The overall responserate in the 1992 DAS was 78 percent (Steeh 1993).6

2 Reflecting the elevated level of racial residential segregation in metropolitanDetroit, only 6 percent of white respondents in the 1992 DAS lived in blocks orblock groups that were 70 percent or more black at the time of the 1990census. Ninety-four percent of black respondents lived in blocks or blockgroups that were 70 percent or more black.

3 The Detroit primary metropolitan statistical area as defined by the Bureau ofthe Census for 1990 included Lapeer, Livingston, Macomb, Oakland, St. Clair,and Wayne counties. The three counties sampled in the 1992 DAS included90 percent of the white population in the Detroit metropolis as defined by theBureau of the Census and 99.5 percent of the black population (U.S. Bureau ofthe Census 1991).

4 Races reported were as follows: black = 750; white = 736; Asian = 13;Hispanic = 12; American Indian = 5; other = 27.

5 Ninety percent of black respondents in the DAS were interviewed by blackinterviewers, and 92 percent of white respondents were interviewed by non-black interviewers.

6 The response rate was 81 percent for those 765 housing units selected inhigh-density black segments and 75 percent for the 778 housing units selectedin low-density black segments.

370 Reynolds Farley

Characteristics of metropolitan Detroit make it an appropriatelocation for the investigation of racial residential segregation.The 1990 census counted 47 metropolises of one million or morenationwide. The average segregation score comparing the resi-dential distributions of blacks and whites across block groupswas 69. Detroit was the most segregated and was the only loca-tion in which segregation increased in the 1980s. Its score (89)was more than two standard deviations above the mean for largemetropolises (Farley and Frey 1994).

Throughout the nation, blacks fall behind whites on most indica-tors of social and economic status, but racial discrepancies werelarger in Detroit. Six measures pertinent to housing marketissues are summarized in table 1. The black-white difference inDetroit is compared to that found in all U.S. metropolises in1990. The top panel shows that the proportion of adults withcollege degrees was lower in Detroit than across the nation, andthe racial difference was greater. White households in Detroithad mean 1989 incomes above the national average, but blackhousehold incomes were slightly below. The poverty rate forblacks in Detroit was higher than for blacks in all metropolitanareas but was lower for whites. Racial discrepancies in the valueof owned homes and in monthly rental payments were larger inDetroit than elsewhere. The one inversion to this pattern in-volves tenure, because, in Detroit, an unusually large proportionof households owned their homes.

What we asked: Questions to assess racial differencesin housing search

We know little about what people actually do when they are inthe housing market. Economists have developed theories ofhousing search taking race into account (Courant 1978; Courantand Yinger 1977), and a few studies have analyzed whether theracial composition of a neighborhood is linked to the advertisingof homes in major newspapers (Galster, Freiberg, and Houk1987; Turner 1993). A small number of investigations have goneat least one step further. Weisbrod and Vidal (1981) assessed thebarriers constraining the search of low-income renters in Pitts-burgh and Phoenix who were given vouchers allowing them tocompete on the open market. Zonn (1980) studied the actualmethods used by approximately 500 Milwaukee households pur-chasing a home two decades ago; and more recently, Turner andWienk (1993) and Turner (1993) obtained information about thesearch strategies used by 200 home buyers in Washington, DC,and neighboring Prince George’s County in 1990. These last


Table 1. Characteristics by Race for Metropolitan Detroitand Metropolitan United States, 1990

Metropolitan MetropolitanDetroit United States

College graduates(among population age 25 and over)

White (%) 19.5 24.1Black (%) 9.7 12.5Ratio of black to white 0.50 0.52

Mean household income (1989)White ($) 44,850 41,699Black ($) 26,511 26,591Ratio of black to white 0.59 0.64

Poverty rate (all ages)White (%) 6.9 8.4Black (%) 32.5 27.7Ratio of black to white 4.71 3.29

Householders who ownWhite (%) 75.5 66.2Black (%) 48.7 40.5Ratio of black to white 0.65 0.61

Median value of owner-occupied homesWhite ($) 72,050 88,100Black ($) 28,250 54,850Ratio of black to white 0.39 0.62

Median monthly rent for rentersWhite ($) 426 415Black ($) 267 328Ratio of black to white 0.62 0.79

Source: U.S. Bureau of the Census (1991).

studies were restricted to persons whose search led to a homepurchase.

Because of the dearth of previous studies, it was not possible todraw on an array of tested questions about housing search,especially since this was the first investigation of this topic witha representative household sample for a major metropolis. Togain information from people who had knowledge of or experi-ence with the current housing market, only those who hadsearched for a home or apartment within the past five yearswere asked the search questions. Thirty-six percent of black and47 percent of white respondents said they had, giving us asample of 272 black and 344 white searchers.

372 Reynolds Farley

Those who had looked for housing were asked if they had usedany of five techniques: consulting a broker, reading newspaperads, speaking with friends or relatives, looking at For Sale orFor Rent signs, or going to community organizations andchurches that might list openings (figure 1). A respondent couldanswer “yes” or “no” to each of these five methods and had theopportunity to report other search strategies.

Figure 1. Questions Asked about Housing Search,1992 Detroit Area Study

A13a. Talked with friends and relatives

A13b. Newspaper ads

A13c. For Sale or For Rent signs

A13d. Real estate brokers

A13e. Community organizations or churches

A13f. Other (SPECIFY)________________________

________________________

A12. Have you searched for a house or an apartment in the last five years?

A14. In general, which method do you feel is the best way to locate a house orapartment?

_____________________________________________________________________________(WRITE QUESTION NUMBER FROM A13, IF APPROPRIATE)

A13. (RB, P.1) Which of the following methods did you use in your mostrecent search?

1. Yes 5. No

1. Yes 5. No

SKIP TO NEXT PAGE, SECTION B

1. Yes 5. No

1. Yes 5. No

1. Yes 5. No

1. Yes 5. No

1. Yes 5. No


In addition to ascertaining racial differences in search methods,this investigation determined whether there was racial consen-sus about the best way to locate a new home or apartment.Respondents who had searched were asked which one methodwas the best to find a home or apartment. These housing searchqueries were efficient and productive. They were clearly under-stood by respondents, were easy for interviewers to ask, elicitedno objections, and quickly provided relevant information.

What we found: Black-white differences in housingsearch

Race makes a difference in housing search. In their most recentforay into the housing market, Detroit area blacks made signifi-cantly less use of brokers than whites, but much more use ofinformal methods such as newspaper ads and talking withfriends or relatives (figure 2).

Figure 2. Methods Used by Detroit Area Blacks and Whitesin Their Most Recent Housing Search

Source: Detroit Area Study, 1992.Note: Data are limited to persons who searched for a house or apartment in the past fiveyears. Unweighted sample size: blacks = 272; whites = 344.* p < 0.05. ** p < 0.01.

BlacksWhites

BlacksWhites

BlacksWhites

BlacksWhites

BlacksWhites

BlacksWhites

Real estate brokers or agents

Newspaper ads

Talked with friends and relatives

Looked at For Sale or For Rent signs

Community organizations or churches

0% 10% 20% 30% 40% 50% 70%60%

43%**60%

65%**

62%**

57%*42%

12%**3%

20%13%

45%

46%

Other methods

374 Reynolds Farley

Zonn’s 1980 study of Milwaukee home buyers and Turner andWienk’s 1993 study in Washington, DC, imply that consultingwith real estate agents was the most effective way to locatehousing. In metropolitan Detroit, 6 of 10 whites compared withonly 4 of 10 black searchers used this method. This was, how-ever, only one of four methods frequently reported by both races.But there is an important difference: Significantly more whitesthan blacks used the formal method of consulting a broker, whilemore blacks than whites used each of the four less formal strate-gies. Approximately one-sixth mentioned methods other thanthose listed on the interview schedule. Most commonly, this was“driving through a neighborhood,” presumably to find an attrac-tive home with a For Sale sign. The next most common “other”method involved people who made a distinction between brokersand agents and then said they used an agent. Overall, blacksused more methods than whites; an average of 2.5 methods forblacks, which was significantly greater than the two methodsreported by whites.

To describe racial differences in how the housing market is per-ceived, respondents were asked, “In general, which method do youfeel is the best way to locate a house or apartment?” (figure 3).Respondents were constrained to only one best method. Thosewho mentioned using an agent were combined with those whorecommended brokers, and those who thought that drivingthrough a neighborhood was the best strategy were merged withthose who recommended looking at For Sale signs.

Substantial racial differences are evident in judgments about thebest search method, just as they are among methods used in thelast search. Half of whites said that a broker was best, which isdouble the proportion for blacks. Blacks, more so than whites,recommended newspaper ads and talking with friends as thebest methods, while similarly small proportions of both racesranked looking at signs as the best way to find a house orapartment.

This evidence leads to rejection of the hypothesis that Detroitarea blacks and whites use the same methods to seek housing.When they are in the housing market, blacks and whites usedifferent methods, with blacks relying on informal techniquesmore than whites. Additionally, the findings lead to rejection ofthe hypothesis that blacks and whites share similar views aboutthe best way to find housing, since whites recommend usingbrokers about twice as frequently as blacks. Blacks rely to alesser extent on brokers than whites, and thus they use informal


Figure 3. Responses of Detroit Area Blacks and Whites about the BestMethod to Locate a House or Apartment

Source: Detroit Area Study, 1992.Note: Data are limited to persons who searched for a house or apartment in the past fiveyears. Unweighted. Sample size: blacks = 272; whites = 344.* p < 0.05. ** p < 0.01.

strategies, a phenomenon that could potentially increase the costof their housing search.

Black-white differences: The effects of social,economic, and demographic characteristics

The choice of a search method is influenced by a person’s charac-teristics. As table 1 shows, blacks and whites in Detroit differgreatly along many dimensions relevant to the housing search,such as income. These differences may explain why whites useand recommend brokers more frequently than blacks.

Consulting a broker, we hypothesized, would be linked to fivedemographic characteristics. First, age presumably has an im-pact. Young people have less experience in the housing marketand fewer assets and thus may feel less comfortable visiting abroker. Older persons who have purchased several homes alsomay make less use of a broker. Second, extensively educatedpersons probably search most efficiently, so they would consult

Blacks

Whites

Blacks

Whites

Blacks

Whites

Blacks

Whites

Real estate brokers or agents

Newspaper ads

Talked with friends and relatives

Looked at For Sale or For Rent signs

0% 10% 20% 30% 40% 50% 60%

25%

50%**

29%**

27%**

17%

15%

15%

15%

376 Reynolds Farley

brokers more than those with limited educations. Third, familyincome should be positively linked to consulting brokers in thehousing search. Gender is included as the fourth demographicvariable to investigate whether males and females differ in theirsearch methods, although there is no hypothesis regarding howgender may affect housing search. Finally, owners or aspiringowners, presumably, are more likely to consult brokers thanrenters are, because the transaction costs are much higher whenbuying a home.

To summarize the findings, those who used or recommended abroker were distinguished from those who relied exclusively onthe less formal search methods. Several further analyses wereconsidered to distinguish the different informal methods, as wellas to distinguish those who relied on a broker exclusively fromthose who used a broker along with informal methods. Almost allwho consulted a broker also read ads or talked to friends, buteven here there is a racial difference suggesting that blacks relyless on brokers. Three-quarters of whites who consulted brokersalso used other search strategies, but a significantly higherproportion of blacks did so (89 percent). Overall, these analysesprovided little additional information.

In table 2, respondents’ characteristics are related to housingsearch (defined only as use of a broker in the last search andrecommending a broker as the best method).

Concerning the use of a broker, differences by age were small,but for both races, the highest proportion using brokers werethose ages 35 to 44—people who had limited experience in thehousing market. Educational differences were trivial amongwhites, but extensively educated blacks used brokers morefrequently than blacks with a high school education or less.Family income was not significantly linked to consulting brokersamong whites, but among blacks, those with higher incomeswere more likely to use them. Gender did not make a differencein the consulting of brokers during the last search, but amongboth races, homeowners were significantly more likely to usebrokers than renters.

Similar information is reported in table 2 from the questionconcerning the best search strategy. Half of whites and one-quarter of blacks recommended a broker. Among whites therewas little difference by age, educational attainment, or income.Among blacks, socioeconomic status made a difference. Thoseat the top of the educational attainment distribution and


Table 2. Responses about Use of Broker in Last Housing Search andRecommendation of Broker as the Best Method, by Demographic

Characteristics for Whites and Blacks

Used Broker or Agent Broker or Agent Is Bestin Last Search Method

Whites Blacks Whites Blacks

Age (%)Under 35 54 38a,b 47 19a,b

35–44 59 54 56 36a

45–54 58 22a 58 11a

55 & over 54 23a 40 20a

Educational attainment (%)Less than 12 years 57 16a,b 38 7a,b

12 years 52 33 42 19a

13 to 15 years 60 48 54 28a

16 or more years 55 73 55 53

Family income (%)Not reported 60 50b 59 36a,b

Under $20,000 46 22a 34 12a

$20,000 to $39,999 51 45 45 13a

$40,000 to $69,999 65 69 55 51$70,000 or more 55 64 57 55

Gender (%)Male 59 47 53 27a

Female 53 40a 46 25a

Current tenure (%)Owners 71b 69b 61b 51b

Renters 33 28 32 13a

Total (%) 60 43a 50 25a

Unweighted sample size 344 272 344 272

Source: Detroit Area Study, 1992.a The t-test indicates a significant (p < 0.01) black-white difference.b Differences within this racial group for this characteristic are significant (p < 0.01)using the chi-square test.

those at the top of the income distribution recommended brokersmore frequently than other blacks. And among both races thetenure difference was large: Homeowners recommended consult-ing brokers more than renters did.

There is a large black-white difference in consulting or recom-mending brokers. A key hypothesis asks whether this is a realracial difference or the result of those large differences in socialand economic status that distinguish the races in Detroit. Forevery category of each variable, a test was made to determine if

378 Reynolds Farley

there was a racial difference; that is, is there a black-whitedifference once you control for age, education, or income? Signifi-cant racial differences are indicated.

While the overall racial difference in using brokers was large,there were relatively few significant differences once characteris-tics were taken into account. Consider educational attainment:Blacks lacking a complete high school education consulted bro-kers less frequently than whites, but for those with greaterattainments, there was no significant racial difference. Simi-larly, blacks with family incomes under $20,000 were less likelyto consult brokers than similarly low-income whites; but athigher income levels, there were no significant racial differences.Controlling for tenure, there was no significant racial difference.

These findings imply that, although blacks toward the lower endof the socioeconomic distribution may use brokers less thanwhites, racial differences are small once we control for interven-ing variables. In particular, economically successful blacks andtheir white peers did not differ in their use of brokers. Amongrespondents with some college education and with incomes of$40,000 or more, 65 percent of whites and 70 percent of blacksused brokers in their last search.

To more rigorously test the hypothesis that race has no net effecton the use of a broker, regression models were fit, but these datawould not sustain this analysis. The likelihood that educationalattainment and income influence the use of a broker depends onwhether the searcher is seeking to buy or to rent. For example,income presumably affects the probability of using a broker lessfor those seeking an apartment than for those seeking to buy anew home. Unfortunately, information in this study about tenurepertains to the time of the survey, not to the time of the housingsearch. It is not feasible to fit one model for those who soughtrental housing and another model for those seeking to purchase,since there is no reason to assume that current tenure is a satis-factory proxy for tenure at the time of the search. Conclusionsshould be drawn cautiously about the net effect of race on hous-ing search.

Net racial differences in recommending a broker as the bestsearch strategy reveal many more significant differences. Exceptfor current homeowners and those at the top of the income andeducational distributions, blacks were significantly less likely torecommend brokers than whites.


Conclusion

In Detroit—and similar older metropolises—blacks and whitescontinue to live in different neighborhoods (Farley and Frey1994; Massey and Denton 1993). Two contrasting explanationsseek to explain the role of brokers in perpetuating such segrega-tion. One view stresses the key role of brokers in determiningwhere homeseekers will live (Yinger 1993, 1995). Those whodefend this perspective emphasize audit studies that showblacks and whites are provided with different information andare steered to different neighborhoods. Presumably, if brokersand lenders acted in a color-blind manner, black and whitehomeseekers would be shown similar housing opportunities, theywould frequently decide to move into identical neighborhoods,and black-white segregation would decline.

The opposing view emphasizes recent audit investigations thatshow blacks and whites are often treated much the same bybrokers, and then contends that brokers play a passive role inmaintaining segregation (Butters 1993). Presumably, they pro-vide information to clients who already have well-formed ideasabout their housing needs and their preferences. Given racialdifferences in preferences about neighborhood racial composi-tions, segregation will be the outcome even if brokers treat allclients similarly. Spokespersons for this viewpoint report thatwhites overwhelmingly seek neighborhoods with only a few blackresidents, while blacks prefer integrated neighborhoods withsubstantial numbers of black residents. Indeed, 50 percent blackseems to be the preferred racial makeup for black homeseekers,but few whites would consider buying a home or renting anapartment in such a densely black neighborhood (Clark 1986,1988, 1989, 1991; Farley et al. 1993, 1994). Racial segregation,from this perspective, depends not so much on what brokers do,but rather on the preferences of customers.

While these overarching views are frequently discussed, there ismuch we do not know about the operation of local housing mar-kets. This investigation sought to determine whether blacks andwhites differ in their housing search strategies by interviewing alarge representative sample. Among blacks who entered theDetroit housing market in the five years before 1992, 43 percentconsulted a broker; but among whites, it was a significantlyhigher: 60 percent. Blacks, more so than whites, relied on infor-mal methods, such as reading newspaper ads or speaking withfriends. This racial difference was, in large measure, accountedfor by the large social and economic differences that distinguish

380 Reynolds Farley

blacks from whites. If there is a narrowing of black-white differ-ences in education and income, there will likely also be a reduc-tion in the racial differences in housing search.

It is a challenge, perhaps an impossible task, to ascertain whatproportion of racial segregation is caused by illegal discrimina-tion by brokers and lenders and what proportion is caused bypreferences of whites and blacks for neighborhoods where theirown race dominates. In a metropolitan area as riven by race asDetroit, there is a widespread perception that discrimination iscommon throughout the housing market—and this belief mayhave a substantial impact on how homeseekers look for housing.That is, there is a large black-white difference in the perceptionof racial discrimination in the housing market, which may belinked to the patterns described in this article.

The 1992 DAS sought to ascertain whether blacks and whitesshared similar views about the presence of racial discriminationand its impact on blacks. Respondents were asked about threereasons why Detroit-area blacks might miss out on good housing:because white owners will not sell or rent to blacks; because realestate agents will not show, sell, or rent to blacks; or becausebanks and lenders will not loan to blacks. While audit studiesseek to measure the incidence of actual discrimination in thehousing market, there are very few studies of the perception ofdiscrimination in the marketing of housing.

The specific questions are shown in figure 4, which also displaysthe findings based on responses from those who searched in thepast five years. There is widespread agreement that blacks missout on good housing because white owners refuse to sell or rentto them. It may not be surprising that 85 percent of blacks be-lieve that this occurs very often or sometimes. More surprising isthe absence of a racial difference—more than 8 of 10 whites alsoagreed that blacks miss out on good housing for this reason.

Whites differ from blacks in perceiving discrimination by bro-kers. That is, just about the same proportion of blacks believedbrokers discriminate as thought that white owners did, butsignificantly fewer whites believed that real estate agents willnot show, sell, or rent to blacks. Moving to the third questionabout possible racial discrimination by lenders, we discover aneven larger racial difference, with 51 percent of whites and87 percent of blacks reporting that this happens sometimes orvery often.


Figure 4. Perceptions of Discrimination against Blacks in the DetroitArea Housing Market by White Owners, Real Estate Agents, and Banks

and Lenders

DAS QUESTIONS ABOUT HOUSING MARKET DISCRIMINATION

I’m going to mention several reasons why Black people may miss out on goodhousing in the Detroit area. I’d like you to tell me how often you think Blackpeople miss out on good housing for each of the reasons I mention.

The first reason is because White owners will not rent or sell to Blacks. Do youthink that Blacks miss out on good housing because (of this/White ownerswon’t rent or sell to Blacks) very often, sometimes, rarely, or almost never?

The next reason is because real estate agents will not show, sell, or rent toBlacks. Do you think that Blacks miss out on good housing because (of this/real estate agents refuse to show, sell, or rent to Blacks) very often, some-times, rarely, or almost never?

How about because banks and lenders will not loan money to Blacks to pur-chase a home. Do you think that Blacks miss out on good housing because (ofthis/banks and lenders will not loan money to Blacks to purchase a home) veryoften, sometimes, rarely, or almost never?

Source: Detroit Area Study, 1992.Note: Data refer to persons searching for housing within the past five years.a Chi-square test shows racial difference is significant at 0.05 level.b Chi-square test shows racial difference is significant at 0.01 level.

Whites

Blacks

Whites

Blacks

Whites

Blacks

White owners will not rent or sell to blacksa

b

Banks and lenders will not loan money to blacks

Real estate agents will not show, sell, or rent to blacks

b

Very Often Sometimes Rarely Almost Never

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

382 Reynolds Farley

These findings emphasize that both blacks and whites think theFair Housing Act of 1968 failed to accomplish its goal of elimi-nating racial discrimination. Note that the majority of whites—and much more than a majority of blacks—believe that brokersand lenders discriminate, causing blacks to miss out on goodhousing.

Current efforts to eradicate discrimination by brokers, if success-ful, will broaden housing opportunities for those blacks who usebrokers. As this article reports, these are often blacks with thecharacteristics of economic success. Such efforts, in the shortrun, will not eliminate those substantial racial differences in theuse of brokers. Such a change might be encouraged were there awidespread belief that brokers did not discriminate. Real estatebrokers now generally use a logo in their advertisements indicat-ing that they are dedicated to providing equal opportunities;they sometimes advertise on black-oriented radio stations; andmany banks similarly advertise their status as an equal opportu-nity lender. These efforts, however, have not been successful ineliminating the belief—among both blacks and whites—that themarketing of housing in the Detroit area is done in a raciallybiased manner.

This investigation documents racial differences in the search forhousing and finds that blacks were less likely than whites to usebrokers, but it is a first effort and needs to be repeated else-where with similarly large samples. Indeed, one of its majorcontributions is demonstrating the feasibility of easily obtaininginformation about housing search strategies from a multipurposesurvey. Several changes will provide even more explicit informa-tion about the role of race, especially the addition of a questionabout whether the most recent search was to buy or to rent, oneabout the outcome of that search, and another regarding the raceof any broker used. Further research on the rationale behind theuse, and especially the non-use, of real estate brokers by allhousing seekers would be helpful as well. There are likely anumber of alternative explanations why some housing searchersdo not use brokers, and some of these reasons may have impor-tant policy implications.

Author

Reynolds Farley is a Research Scientist at the Population Studies Center andProfessor of Sociology at the University of Michigan.

The author thanks Charlotte Steeh, Director of the Detroit Area Study, for herassistance. Maria Krysan, Keith Reeves, and Tara Jackson aided in the


development of the questionnaire and the collection of interview data. JudyMullin prepared this article, including the figures. Financial support for the1992 Detroit Area Study was obtained from the Ford Foundation and theUniversity of Michigan.

References

Bernick, E. Lee, David J. Pratto, and Robert Davis. 1994. Race and Dissatis-faction with Government Services: A Cautionary Note. Journal of UrbanAffairs 16(4):385–94.

Butters, Robert D. 1993. The Real Estate Industry’s View of Audit Results:Comments. In Clear and Convincing Evidence: Measurement of Discriminationin America, ed. Michael Fix and Raymond J. Struyk. Washington, DC: TheUrban Institute Press.

Clark, William A. V. 1986. Residential Segregation in American Cities.Population Research and Policy Review 5:95–127.

Clark, William A. V. 1988. Understanding Residential Segregation in Ameri-can Cities: Interpreting the Evidence: Reply to Galster. Population Researchand Policy Review 7:113–21.

Clark, William A. V. 1989. Residential Segregation in American Cities:Common Ground and Differences in Interpretation. Population Research andPolicy Review 8:193–97.

Clark, William A. V. 1991. Residential Preferences and Neighborhood RacialSegregation: A Test of the Schelling Segregation Model. Demography28(1):1–19.

Courant, Paul N. 1978. Racial Prejudice in a Search Model of the UrbanHousing Market. Journal of Urban Economics 5:329–45.

Courant, Paul N., and John Yinger. 1977. On Models of Racial Prejudice andUrban Residential Structure. Journal of Urban Economics 4:272–91.

Farley, Reynolds, and William H. Frey. 1994. Changes in the Segregation ofBlacks from Whites in the 1980s: Small Steps toward a More Racially Inte-grated Society. American Sociological Review 59(6):23–49.

Farley, Reynolds, Charlotte Steeh, Maria Krysan, Tara Jackson, and KeithReeves. 1993. Continued Racial Residential Segregation in Detroit: “ChocolateCity, Vanilla Suburbs” Revisited. Journal of Housing Research 4(1):1–37.

Farley, Reynolds, Charlotte Steeh, Maria Krysan, Tara Jackson, and KeithReeves. 1994. Stereotypes and Segregation: Neighborhoods in the DetroitArea. American Journal of Sociology 100(3):750–81.

Fix, Michael, George C. Galster, and Raymond J. Struyk. 1993. An Overviewof Auditing for Discrimination. In Clear and Convincing Evidence: Measure-ment of Discrimination in America, ed. Michael Fix and Raymond J. Struyk.Washington, DC: The Urban Institute Press.

384 Reynolds Farley

Galster, George. 1990. Racial Discrimination in Housing Markets in the 1980s:A Review of the Audit Evidence. Journal of Planning Education and Research9:166–75.

Galster, George. 1992. Research on Discrimination in Housing and MortgageMarkets: Assessment and Future Directions. Housing Policy Debate3(2):639–84.

Galster, George, Fred Freiberg, and Diane L. Houk. 1987. Racial Differentialsin Real Estate Advertising Practices: An Exploratory Case Study. Journal ofUrban Affairs 9(3):199–215.

Massey, Douglas S., and Nancy Denton. 1993. American Apartheid: Segrega-tion and the Making of the Underclass. Cambridge, MA: Harvard UniversityPress.

Schuman, Howard, and Jean Converse. 1971. Effects of Black and WhiteInterviewers on Black Responses in 1968. Public Opinion Quarterly 35:46–68.

Steeh, Charlotte. 1993. Sampling Report for the 1992 Detroit Area Study. AnnArbor, MI: University of Michigan, Detroit Area Study.

Turner, Margery Austin. 1992. Discrimination in Urban Housing Markets:Lessons from Fair Housing Audits. Housing Policy Debate 3(2):185–216.

Turner, Margery Austin. 1993. Limits on Neighborhood Choice: Evidence ofRacial and Ethnic Steering in Urban Housing Markets. In Clear and Convinc-ing Evidence: Measurement of Discrimination in America, ed. Michael Fix andRaymond J. Struyk. Washington, DC: The Urban Institute Press.

Turner, Margery Austin, Raymond J. Struyk, and John Yinger. 1991. HousingDiscrimination Study: Synthesis. Washington, DC: The Urban Institute Press.

Turner, Margery Austin, and Ron Wienk. 1993. The Persistence of Segregationin Urban Areas: Contributing Causes. In Housing Markets and ResidentialMobility, ed. G. Thomas Kingsley and Margery Austin Turner. Washington,DC: The Urban Institute Press.

U.S. Bureau of the Census. 1991. Census of Population and Housing: 1990.CPH-1-24.

Weisbrod, Glen, and Avis Vidal. 1981. Housing Search Barriers for Low-Income Renters. Urban Affairs Quarterly 16(4):465–82.

Wienk, Ronald E., Clifford E. Reid, John C. Simonson, and Frederick J.Eggers. 1979. Measuring Racial Discrimination in American Housing Markets.Washington, DC: U.S. Department of Housing and Urban Development, PolicyDevelopment and Research.

Yinger, John. 1986. Measuring Racial Discrimination with Fair HousingAudits: Caught in the Act. American Economic Review 76(5):881–93.

Yinger, John. 1993. Access Denied, Access Constrained: Results and Implica-tions of the 1989 Housing Discrimination Study. In Clear and ConvincingEvidence: Measurement of Discrimination in America, ed. Michael Fix andRaymond J. Struyk. Washington, DC: The Urban Institute Press.


Yinger, John. 1995. Closed Doors, Opportunities Lost: The Continuing Costs ofHousing Discrimination. New York: Russell Sage Foundation.

Zonn, Leo. 1980. Information Flows in Black Residential Search Behavior.Professional Geographer 32:43–50.

386 Reynolds Farley

Joblessness and Poverty in America’s Central Cities 387Housing Policy Debate • Volume 7, Issue 2 387© Fannie Mae Foundation 1996. All Rights Reserved.

Joblessness and Poverty in America’s CentralCities: Causes and Policy Prescriptions

John D. KasardaUniversity of North Carolina

Kwok-fai TingChinese University of Hong Kong

Abstract

Two paradigms pervade the policy debate on the causes of urban joblessnessand poverty: (1) a structural, nonvoluntaristic perspective emphasizing theroles of urban economic change, residential segregation, and spatial and skillsmismatches, and (2) a rational-choice, voluntaristic perspective contendingthat the generosity and ready availability of welfare programs have removedthe incentive for poor persons to accept low-paying jobs. This article bringstogether propositions of each paradigm into a comprehensive theoreticalmodel. The study measured and tested key causal operators of the model for asample of 67 large U.S. cities, with special attention to race and gender.

Results show that structural and welfare disincentive perspectives are not inconflict but rather operate side by side to reinforce joblessness and poverty.Race and gender, especially the role of urban space for women’s work, areimportant. The article raises pertinent policy issues derived from the twoperspectives and from the analysis.

Keywords: Policy; Welfare; Minorities

Introduction

The rise of joblessness and poverty in America’s major cities,especially among minorities, has generated considerable schol-arly and public debate. Out of that debate, two competing para-digms (or perspectives) have emerged that differ both in causalexplanations and in policy prescriptions. The first emphasizeshousing segregation and structural changes in metropolitaneconomies that have reduced job accessibility and weakenedearnings for a substantial portion of central-city residents (Kain1968; Kasarda 1985; Wilson 1987). The second contends that thesize and ready availability of government welfare programs haveremoved the incentive to work for many city residents by actu-ally making public assistance a rational choice over employment(Anderson 1978; Mead 1989; Murray 1984).

388 John D. Kasarda and Kwok-fai Ting

In this article, we elaborate on these two paradigms and specifytheir fundamental propositions. We then bring together thespecifications of each paradigm in a comprehensive theoreticalmodel. To assess the model, we include key causal operators(spatial and skill mismatches) in the structural approach andpivotal nonrecursive feedback effects (from welfare recipiency tohigher rates of urban joblessness and poverty) in the disincen-tive approach. Our goal is to assess the plausibility and relativeexplanatory power of the two approaches, giving special atten-tion to race and gender.

Structural barriers versus welfare disincentives

Stripped to their basics, both paradigms view post-1970 rises inurban poverty as a direct consequence of less-skilled city resi-dents, particularly African Americans, no longer working asregularly as they once did. The two perspectives differ sharply,however, as to whether such joblessness is voluntary. Thosetaking a nonvoluntary perspective see high inner-city unemploy-ment resulting from limits to residential choice and from struc-tural barriers to less-skilled city residents created by thesuburbanization of blue-collar jobs and their replacement in thecity by white-collar jobs typically requiring education above thatpossessed by most disadvantaged residents. Central to the struc-tural barrier perspective is the mismatch hypothesis, which hastwo versions: spatial mismatch and skills mismatch.

The spatial mismatch hypothesis was introduced by Kain (1968).He argued that the suburbanization of low-skilled jobs, espe-cially in manufacturing, together with housing market discrimi-nation that residentially confined African Americans to thecentral cities, isolated African Americans from decentralizingemployment opportunities. Suburbanization of manufacturingwas seen as directly reducing opportunities for low-skilled jobsin the central cities while increasing the job-search and commut-ing costs of city residents who were able to secure suburbanblue-collar employment. Residential segregation of AfricanAmericans, as Wilson (1987) and Massey (1990) further argued,reinforces their geographic and social isolation from jobs andmagnifies poverty problems.

Kain’s article sparked more than two decades of debate in thescholarly literature regarding the impact of changing job loca-tions and housing proximity on urban unemployment and earn-ings (see, for example, Ellwood 1986; Farley 1987; Harrison1974; Ihlanfeldt and Sjoquist 1989, 1990, 1991; Leonard 1987;

Joblessness and Poverty in America’s Central Cities 389

Mooney 1969; Offner and Saks 1971; Price and Mills 1985;Straszheim 1980; Vrooman and Greenfield 1980; Zax 1990; Zaxand Kain 1991). The wide range of methods and indicators thatthese studies used to measure spatial access and employmentand earnings outcomes produced an equally wide range ofresults. Because of these inconsistent findings, two widely citedreviews of studies addressing the spatial mismatch hypothesisconclude that the status of the hypothesis remains open (Holzer1991; Jencks and Mayer 1990). Both assessments call for bettermeasurement and modeling of spatial mismatch, with controlsfor competing explanations.1

A conceptually related, but analytically distinct, component ofthe structural barrier explanation of central-city joblessness isthe skills mismatch hypothesis. According to this hypothesis,modern advances in transportation, communication, and manu-facturing technologies interacting with the changing structure ofthe national and international economy have transformed majorcities from centers of material goods production to centers ofinformation exchange, finance, and administration (Johnson andOliver 1992; Kasarda 1976, 1985; Noyelle 1987; Stanback 1991).In the transformation process, many manufacturing, warehous-ing, and other goods-processing establishments that once pro-vided ample employment opportunities for less-educated (andpresumably, less-skilled) city residents either completely van-ished or relocated to the suburbs, exurbs, and abroad (Wacquantand Wilson 1989; Wilson 1987). Traditional urban blue-collarindustries were replaced by knowledge-intensive, white-collarservice industries that typically require some education beyondhigh school. Although spatially accessible, these white-collar jobsare not functionally accessible to most poorly educated cityresidents who lack the skills to perform them (Kasarda 1990,1993a, 1995; Lichter 1988).

Structuralists argue that skills mismatches compounded byspatial mismatches create a double barrier to job access for manycity residents (Kain 1992). Lacking sufficient education to par-ticipate in new urban growth industries and the transportationor financial means either to commute to dispersed suburban jobsor to relocate near them, increasing numbers of disadvantagedcity residents find themselves spatially and functionally cut offfrom employment opportunity. Structuralists go on to argue thatresulting high concentrations of inner-city joblessness triggered

1 Kain (1992) provides a comprehensive review of the research on the spatialmismatch hypothesis, including a critique of the Holzer (1991) and Jencks andMayer (1990) reviews.


increasingly high concentrations of poverty that, in turn,necessitated welfare assistance to more households in need.Structuralists contend that, if welfare had not been available,deteriorating economic conditions in the cities since the late1960s would have resulted in far more households in poverty(Wilson 1987).

An opposing view of city joblessness, poverty, and welfare isoffered by conservative scholars who take a voluntaristic per-spective. The problem, according to this group, is not primarily alack of appropriate jobs nearby or skill barriers to employmentbut is the condition that readily available welfare has created inwhich it is rational for many to choose nonwork over work.Extensive welfare programs, introduced in the 1960s and 1970s,effectively changed the economic rewards and penalties of behav-iors such as family dissolution and out-of-wedlock births thatreinforce poverty.

Banfield (1969) was among the first to articulate the positionthat welfare programs discourage employment and hold peoplein poverty. Martin Anderson (1978), the chief architect of RonaldReagan’s economic recovery program, estimated that guaranteedincome programs such as those proposed during the Carteradministration could reduce work effort by as much as50 percent.

In his controversial yet influential book, Losing Ground, Murray(1984) marshals an extensive array of statistics suggesting thata low-income urban family can actually improve its financialsituation by dissolving its marriage, withdrawing its membersfrom the labor force, and subsisting on a variety of welfare pro-grams. According to Murray, the ready availability of incometransfer payments, free medical care, subsidized housing, foodstamps, and other forms of government assistance discouragedwork and made it profitable for the poor to behave in the shortterm in ways that were destructive to them in the long term. Theunanticipated outcome of these programs, Murray contends, wasactually to increase joblessness, out-of-wedlock births, andpoverty populations because many individuals who programdesigners did not intend to be beneficiaries modified their behav-ior to qualify. In Murray’s words, “We tried to provide for thepoor and produced more poor instead. We tried to remove thebarriers to escape from poverty and inadvertently built a trap”(1984, 9).

Murray’s thesis was developed further by advocates of manda-tory work for people receiving welfare. Mead (1988, 1989), for


example, argues that plenty of low-skilled jobs exist in the cen-tral cities but that under the present set of welfare benefits andwage levels there is no incentive for the poor to accept those jobs.Yet, according to Mead, low-wage, entry-level jobs are importantbridges to better-paying jobs that would eventually lift manypeople out of poverty while providing indirect social benefits tofamily and community.

Weidenbaum (1991) echoes Mead’s position that the ready avail-ability and relative generosity of welfare programs (rather thanan insufficient number of low-skilled jobs nearby) made the poorindifferent to low-wage jobs and encouraged unemployment. He,therefore, concludes that it is a mistake to offer welfare benefitswithout imposing work requirements on recipients, even if thepay is low.

As with spatial and skills mismatch arguments, welfare disin-centive explanations of joblessness and poverty did not escapesubstantial scholarly scrutiny and critique (see Danziger andGottschalk 1985; Duncan and Hoffman 1991; Ellwood and Sum-mers 1986; McLanahan, Garfinkel, and Watson 1988). Criticsmarshaled their own evidence, suggesting that the level of wel-fare does not have a significant negative effect on work effort orfamily stability. Among their strongest points was that averagecash assistance to welfare households actually decreased in realterms after 1972 (because states failed to adjust benefit levelsfor inflation) while joblessness and family instability continuedto rise. Murray (1986), Mead (1989), and others, however, coun-tered by pointing out that reduced real cash benefits were morethan offset by the expansion of noncash benefit programs such asfood stamps, Medicaid, and housing assistance. Equally as im-portant, they further argued, was the growing ease since theearly 1970s with which recipients could gain access to a varietyof public assistance programs. Greater program accessibilityincreased both the number and proportion of households receiv-ing multiple forms of welfare, with public assistance being mostpervasive and generous in large cities.

Tanner, Moore, and Hartman (1995) estimated the total incomevalue of six welfare programs (Aid to Families with DependentChildren [AFDC], food stamps, Medicaid, housing assistance,nutrition assistance, and utilities assistance) for a single parentwith two young children in each of the 50 states and 16 selectedcities. They then compared the dollar value of that tax-freebenefits package to the amount of pretax income a person wouldhave to earn in a job to equal the value of the benefits package.They reported that, in virtually all states, the benefits package


pays substantially more than the type of entry-level job that atypical welfare recipient can expect to find, with benefits espe-cially high in large northeastern and midwestern cities (thesame cities that have undergone the greatest economic restruc-turing and traditional blue-collar job loss). For example, Tanner,Moore, and Hartman (1995) calculated that welfare provides theequivalent of an hourly pretax wage of $14.75 in New York City,$12.45 in Philadelphia, $11.35 in Baltimore, and $10.90 inDetroit. Based on their calculations, they concluded that, formost public assistance recipients, the movement from welfare towork would result in a substantial decline in real income, thusposing a strong disincentive to seeking employment.2

Model specification

To assess the relative impact of urban economic restructuringand welfare program disincentives on city joblessness andpoverty, we constructed a theoretical model incorporating andsynthesizing specifications central to each paradigm. This full-form model is presented in figure 1, and its specifications areelaborated below. For all specifications, the city is the unit ofanalysis.

Beginning with background factors, the transformation of citiesfrom centers of goods processing to centers of information pro-cessing contributes to a loss of low-skilled city jobs. Low-skilled-job loss leads to greater spatial and skill mismatches for largeportions of city residents with limited education. The roles of theurban economic base and its transformation are measured by jobcomposition in 1980 and change between 1980 and 1990. Otherthings being equal, a higher proportion of low-skilled jobs in1980 would be expected to lessen the extent of future spatial andskill mismatches, while a loss of low-skilled jobs in the subse-quent 10 years would intensify the degree of mismatches.

Two other pertinent background variables are included in themodel: low-skilled labor supply and residential segregation.From the labor-supply point of view, a larger pool of poorlyeducated city residents (percent of adults with high school educa-tion or less) increases the competition for available low-skilledjobs and creates greater skill and spatial mismatches in the city.Residents of a highly segregated city have fewer choices in locat-ing closer to their job market, thus causing a longer commuting

2 Earlier, Ellwood (1988) provided an excellent analysis of returns to welfareversus work.


Fig

ure

1. T

heo

reti

cal

Mod

el

Ski

llm

ism

atch

Spa

tial

mis

mat

ch

% lo

w-s

kill

edin

dust

ryjo

bs, 1

980

% h

igh

sc

hoo

l or

less

, 199

0

Oth

erva

riab

les

% f

emal

ejo

bles

s

% f

emal

ein

pov

erty

% f

emal

eon

wel

fare

% m

ale

jobl

ess

Oth

erva

riab

les

% lo

ss o

flo

w-s

kill

edin

dust

ryjo

bs, 1

980–

90

Res

iden

tial

segm

enta

tion

inde

x, 1

990

% m

ale

inpo

vert

y

% m

ale

onw

elfa

re

+

–

–

+

+ + +

+

+ –/+

–/+

+ +

+

+

+ +


distance; for low-skilled workers who cannot afford longer com-mutes, unemployment often results.3

In the structuralist view, cities with higher spatial and skillmismatches should exhibit higher jobless rates. Structuralistsfurther hypothesize that higher jobless rates result in greaterpercentages of people in poverty and, in turn, larger proportionsof people receiving welfare.

Those who view welfare as a poverty trap hypothesize a viciouscycle among welfare, joblessness, and poverty. The welfaredisincentive paradigm posits that the relative generosity ofwelfare programs and expanded eligibility of urban householdsdiscourage work while encouraging family dissolution, out-of-wedlock births, and other behaviors that contribute to greaternumbers in poverty. The critical specification in the welfaredisincentive paradigm, however, is that the widespread avail-ability of public assistance seriously reduces the incentive towork among large numbers of low-income persons, therebyincreasing their unemployment rate. If this is the case, a signifi-cant positive feedback effect on city jobless rates should beobservable from the proportion of residents receiving publicassistance, which reinforces the poverty cycle.

Since race and gender may confound aggregated parameterestimates, it is important to evaluate the relationships amongjoblessness, poverty, and welfare receipt separately by race formen and women (Jencks and Mayer 1990; Kain 1992). We, there-fore, include in the model two sets of measures and specify twofeedback loops for men and women, by race. With two sets ofparallel measures, we also specify three correlations acrossequations between the error terms. Such a specification is typi-cally referred to as “the seemingly unrelated regression system”(Hargens 1988, 69). The error terms of the dependent variablesare correlated when there are common omitted variables. Thissituation is highly probable in our model, which includes identi-cal dependent variables for two different populations. Failure toincorporate such specifications will lead to inefficient parameterestimates and faulty significance tests (Hargens 1988).

Most studies have focused their discussions and analyses onAfrican Americans. If propositions from the economic restructur-ing and the welfare disincentive paradigms are valid, they

3 For an elaborated discussion of geographic industrial restructuring andspatial mismatch, including regional differences in urban economic form, theroles of migration and immigration, and whether people follow jobs or jobsfollow people, see Kasarda 1995.


should apply to less-educated urban white residents as well. Wethus estimate the model illustrated in figure 1 separately fornon-Hispanic whites and non-Hispanic blacks (African Ameri-cans) and compare model parameters for interracial consisten-cies or differences.4

Methods, data, and measures

The theoretical model presented in figure 1 is a nonrecursivemodel that involves feedback relationships among welfare, job-


(PUMA) codes and geographic identifiers on the PUMS files, wewere able to identify 67 cities (see table 1) for which we couldobtain both place of residence and place of work of individualssampled as well as other pertinent variables to construct allmeasures matched to city boundaries.5

Table 1. Central Cities Included in the Analysis

Amarillo, TX Memphis, TNAnchorage, AK Minneapolis, MNAustin, TX Mobile, ALBaltimore, MD Montgomery, ALBaton Rouge, LA Nashville-Davidson, TNBoston, MA New Orleans, LABuffalo, NY New York, NYChattanooga, TN Newark, NJChicago, IL Norfolk, VACleveland, OH Philadelphia, PAColorado Springs, CO Portland, ORColumbia, GA Providence, RIDenver, CO Richmond, VADes Moines, IA Rochester, NYDetroit, MI Sacramento, CAFlint, MI San Antonio, TXFort Lauderdale, FL San Diego, CAFort Wayne, IN San Francisco, CAFort Worth, TX San Jose, CAFresno, CA Seattle, WAGary, IN Shreveport, LAGrand Rapids, MI Spokane, WAGreensboro, NC Springfield, MAIndianapolis, IN St. Louis, MOJackson, MI St. Paul, MNJacksonville, FL St. Petersburg, FLJersey City, NJ Stockton, CAKnoxville, TN Syracuse, NYLexington-Fayette, KY Tampa, FLLincoln, NE Virginia Beach, VALong Beach, CA Washington, DCLos Angeles, CA Wichita, KSLubbock, TX Worcester, MAMadison, WI

5 The Urban Underclass Database technical documentation (Kasarda 1993b)describes how the PUMA codes and geographic identifiers are used to aggre-gate PUMS data to city boundaries. For example, if a geographic match ofPUMA codes to city boundaries exists and the PUMS records identify place ofwork for that matched city, the user can obtain actual city job composition(based on the education, industry, etc., of jobholders) regardless of place ofresidence of the jobholders. These place-of-work identifiers enable the user toomit city residents who do not work in the city and to include suburbanresidents and other noncity residents who do work in the matched city, alongwith their full range of individual attributes (e.g., occupation, education, race,gender).


We had to exclude the other 33 cities from the sample becausethe 1990 PUMS files did not provide appropriate geographicidentifiers of the place individuals were working, making itimpossible to construct key measures on the characteristics ofjobs (based on characteristics of jobholders) in these cities. Be-cause the Hispanic and Asian populations are small in many ofthe 67 cities, and the 1 percent and 5 percent sampling schemesin PUMS further reduce the available sample sizes for these twominority groups, we confined our analysis to non-Hispanicwhites and non-Hispanic blacks (African Americans).

Table 2 lists all variables used in the analysis with their meansand standard deviations. Table 3 reports the sample correlationmatrix. The constructed variables are described below.

To measure city industrial mix and its change, we matched morethan 200 census industry codes from the 1980 and 1990 files,focusing on jobs in low-skilled industries. We defined low-skilledindustries as those in which half or more of the employees hadno more than a high school education in 1990 (compiled from the1990 PUMS file). Though an imperfect measure of skill level,education has been found to come closest to tapping this concept(Bound and Freeman 1990; Moss and Tilly 1991). Having deter-mined the educational skill status of industries, we captured thecity industrial base and its transformation by the percentage ofcity jobs in low-skilled industries in 1980 and the percent changein these jobs between 1980 and 1990.

We measured residential segregation between non-Hispanicwhites and African Americans by the index of dissimilarity basedon census tract-level data using the formula

(0.5 �I|NHWi – NHBi|)100,

where NHWi and NHBi denote the proportion of non-Hispanicwhites and the proportion of non-Hispanic blacks residing in thecensus tract. The index can be interpreted as the percentage ofAfrican Americans who would have to switch tracts to attain asettlement pattern proportionate with that of non-Hispanicwhites in all city tracts and equal to the overall city proportion.We measured education (as a proxy for skill level) of city resi-dents by the percent of persons with no more than a high schooleducation among those aged 16 to 64 and not enrolled in school.We measured education separately for non-Hispanic whites andAfrican Americans.


Tab

le 2

. V

ari

ab

le D

efin

itio

ns,

Mea

ns,

an

d S

tan

da

rd D

evia

tio

ns

Var

iabl

eD

efin

itio

nM

ean

Sta

nda

rd D

evia

tion

PC

TL

SJ8

0P

erce

nt

low

-sk

ille

d in

dust

ry j

obs,

198

041

.67

8.12

PC

TL

SJL

Per

cen

t lo

w-s

kil

led

indu

stry

job

los

s, 1

980–

1990

–15.

1125

.09

SE

GIN

DE

XR

esid

enti

al s

egre

gati

on i

nde

x, 1

990

63.1

412

.50

Non

-His

pan

ic w

hit

es (

1990

mea

sure

s)H

SO

RL

S_W

Per

cen

t re

side

nts

wit

h h

igh

sch

ool

edu

cati

on o

r le

ss44

.13

12.2

9E

DM

AT

CH

_WE

duca

tion

(sk

ill)

mis

mat

ch (

perc

ent)

1.71

8.94

SP

MA

TC

H_W

Spa

tial

mis

mat

ch (

perc

ent)

20.4

03.

29JO

BL

ES

S_W

WP

erc e

nt

jobl

ess

(wom

en)

32.5

96.

43JO

BL

ES

S_W

MP

erc e

nt

jobl

ess

(men

)14

.74

4.33

PO

VE

RT

Y_W

WP

erc e

nt

in p

over

ty (

wom

en)

14.0

34.

47P

OV

ER

TY

_WM

Per

c en

t in

pov

erty

(m

en)

10.0

43.

18W

EL

FA

RE

_WW

Per

c en

t r e

c eiv

e w

elfa

r e a

ssis

tan

c e (

wom

en)

7.80

4.69

WE

LF

AR

E_W

MP

erc e

nt

r ec e

ive

wel

far e

ass

ista

nc e

(m

en)

4.13

2.32

Non

-His

pan

ic b

lac k

s (1

990

mea

sur e

s)H

SO

RL

S_B

Per

cen

t re

side

nts

wit

h h

igh

sc h

ool

edu

c ati

on o

r le

ss63

.02

9.23

ED

MA

TC

H_B

Edu

c ati

on (

skil

l) m

ism

atch

(pe

rcen

t)20

.60

7.62

SP

MA

TC

H_B

Spa

tial

mis

mat

c h (

per c

ent)

22.9

75.

08JO

BL

ES

S_B

WP

erc e

nt

jobl

ess

(wom

en)

39.6

79.

24JO

BL

ES

S_B

MP

erc e

nt

jobl

ess

(men

)27

.35

7.67

PO

VE

RT

Y_B

WP

erce

nt

in p

over

ty (

wom

en)

37.4

010

.25

PO

VE

RT

Y_B

MP

erce

nt

in p

over

ty (

men

)24

.24

6.79

WE

LF

AR

E_B

WP

erc e

nt

r ec e

ive

wel

fare

ass

ista

nc e

(w

omen

)24

.12

9.97

WE

LF

AR

E_B

MP

erc e

nt

r ec e

ive

wel

fare

ass

ista

nc e

(m

en)

7.61

4.64


Tab

le 3

. C

orr

ela

tio

n M

atr

ix o

f V

ari

ab

les

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(1)

HS

OR

LS

_W1.

000

(2)

HS

OR

LS

_B0.

479

1.00

0(3

)P

CT

LS

J80

0.49

50.

182

1.00

0(4

)P

CT

LS

JL0.

471

0.47

90.

312

1.00

0(5

)S

EG

IND

EX

0.23

60.

650

0.02

90.

473

1.00

0(6

)E

DM

AT

CH

_W0.

854

0.23

20.

214

0.48

30.

123

1.00

0(7

)S

PM

AT

CH

_W–0

.057

–0.0

32–0

.272

–0.0

260.

266

–0.0

241.

000

(8)

ED

MA

TC

H_B

–0.0

310.

710

–0.3

270.

388

0.55

10.

077

0.02

51.

000

(9)

SP

MA

TC

H_B

–0.0

420.

198

–0.3

260.

097

0.51

4–0

.015

0.87

80.

289

1.00

0(1

0)JO

BL

ES

S_W

M0.

273

0.18

70.

095

0.56

00.

367

0.33

60.

296

0.18

10.

388

1.00

0(1

1)JO

BL

ES

S_W

W0.

376

0.29

60.

251

0.42

60.

302

0.25

90.

326

0.05

50.

348

0.73

3(1

2)JO

BL

ES

S_B

M0.

241

0.13

40.

055

0.42

50.

340

0.32

30.

216

0.15

20.

261

0.60

0(1

3)JO

BL

ES

S_B

W0.

167

0.16

60.

063

0.33

20.

293

0.19

80.

205

0.16

40.

276

0.64

9(1

4)P

OV

ER

TY

_WM

0.14

90.

112

0.24

70.

525

0.27

60.

247

–0.0

080.

185

0.09

00.

577

(15)

PO

VE

RT

Y_W

W0.

349

0.21

40.

192

0.56

30.

267

0.48

7–0

.197

0.26

9–0

.092

0.54

4(1

6)P

OV

ER

TY

_BM

0.00

60.

046

0.13

00.

370

0.09

00.

066

–0.1

960.

123

–0.2

440.

061

(17)

PO

VE

RT

Y_B

W0.

114

0.36

40.

161

0.29

80.

303

–0.0

12–0

.247

0.24

3–0

.177

0.09

1(1

8)W

EL

FA

RE

_WM

0.14

8–0

.073

0.01

50.

327

–0.0

480.

356

–0.0

890.

091

–0.0

250.

538

(19)

WE

LF

AR

E_W

W0.

259

–0.0

51–0

.017

0.33

3–0

.115

0.49

4–0

.176

0.10

0–0

.154

0.47

9(2

0)W

EL

FA

RE

_BM

0.16

0–0

.101

–0.0

670.

207

0.12

20.

393

0.19

10.

082

0.26

40.

454

(21)

WE

LF

AR

E_B

W0.

159

0.04

2–0

.102

0.20

40.

058

0.30

2–0

.085

0.14

8–0

.037

0.32

7


Tab

le 3

. C

orr

ela

tio

n M

atr

ix o

f V

ari

ab

les

(con

tin

ued

)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

(18)

(19)

(20)

(21)

(11)

JOB

LE

SS

_WW

1.00

0(1

2)JO

BL

ES

S_B

M0.

585

1.00

0(1

3)JO

BL

ES

S_B

W0.

564

0.69

11.

000

(14)

PO

VE

RT

Y_W

M0.

453

0.51

90.

438

1.00

0(1

5)P

OV

ER

TY

_WW

0.41

00.

513

0.38

30.

695

1.00

0(1

6)P

OV

ER

TY

_BM

0.09

60.

344

0.09

20.

431

0.36

61.

000

(17)

PO

VE

RT

Y_B

W0.

088

0.18

40.

425

0.20

80.

235

0.49

91.

000

(18)

WE

LF

AR

E_W

M0.

408

0.49

30.

384

0.43

90.

430

0.11

4–0

.083

1.00

0(1

9)W

EL

FA

RE

_WW

0.35

20.

522

0.38

40.

317

0.67

80.

158

0.00

10.

738

1.00

0(2

0)W

EL

FA

RE

_BM

0.34

40.

655

0.42

60.

448

0.46

50.

184

–0.2

240.

623

0.55

21.

000

(21)

WE

LF

AR

E_B

W0.

173

0.52

00.

641

0.17

20.

369

0.18

00.

502

0.43

20.

612

0.32

31.

000


The PUMS information based on place of residence and place ofwork allows us to aggregate and compare the characteristics ofcity residents with those of city jobholders. The characteristics ofcity jobholders (regardless of their place of residence) reflect theattributes of jobs that are available in the city. We compared theeducational distribution of all city jobholders with that of cityresidents aged 16 to 64 and not in school (among whites andamong African Americans) to measure skill mismatch for thesetwo racial groups (percent of city jobs held by persons with morethan a high school education versus percent of out-of-school adultresidents, by race, with more than a high school education). Posi-tive values indicate that available jobs require education greaterthan that possessed by city residents of each race. We measuredspatial mismatch by the average minutes of the one-way workcommute of city residents for each racial group, computed from thePUMS files as well. While this is an imperfect measure of spatialmismatch, it does indirectly tap spatial separation of workers fromjobs (including patterns of urban development) and workers’ reli-ance on typically more time-consuming public transit.

We measured joblessness as the percent of persons not working,poverty as the percent below the poverty threshold (see tableA-1), and welfare receipt as the percent receiving AFDC, supple-mental security income, or general assistance income during1989.6 Since both the economic restructuring and welfare dis-incentive explanations apply primarily to city residents with lesseducation and low potential wages, we constructed these mea-sures based on out-of-school persons aged 16 to 64 with no morethan a high school education. We determined the measuresseparately by race and gender.

Results

Figures 2 and 3 present the results of the fully specified modelfor non-Hispanic whites and for non-Hispanic blacks. The pathcoefficients shown in the figures are unstandardized. Path coeffi-cients can be expressed in either standardized or unstandardizedforms (Bollen 1989); the standardized coefficient can be obtained

6 With few exceptions, the census measure of public-assistance income ex-cludes social security income, disability income, and hospital or other medicalcare vendor payments. General assistance includes cash payments adminis-tered at the state and local levels for low-income persons who do not qualifyfor AFDC or supplemental security income. Persons must show financial needand live in designated areas (or specific states). Terms used by states to referto this program include relief, home relief, poor relief, and direct assistance.


Fig

ure

2. P

ath

Dia

gram

for

Non

-His

pan

ic W

hit

es

Ski

llm

ism

atch

Spa

tial

mis

mat

ch

% lo

w-s

kill

edin

dust

ryjo

bs, 1

980

% h

igh

sc

hoo

l or

less

, 199

0

Oth

erva

riab

les

% f

emal

ejo

bles

s

% f

emal

ein

pov

erty

% f

emal

eon

wel

fare

% m

ale

jobl

ess

Oth

erva

riab

les

% lo

ss o

flo

w-s

kill

edin

dust

ryjo

bs, 1

980–

90

Res

iden

tial

segm

enta

tion

inde

x, 1

990

% m

ale

inpo

vert

y

% m

ale

onw

elfa

re

0.68

0.05

–0.0

1

–0.1

0

–0.3

2

–0.0

2

0.10

0.18

0.65

0.13

0.19

0.03

0.68

0.40

0.25

0.39

0.42

p <

0.05

p >

0.0

5


Fig

ure

3. P

ath

Dia

gram

for

Non

-His

pan

ic B

lack

s

Ski

llm

ism

atch

Spa

tial

mis

mat

ch

% lo

w-s

kill

edin

dust

ryjo

bs, 1

980

% h

igh

sc

hoo

l or

less

, 199

0

Oth

erva

riab

les

% f

emal

ejo

bles

s

% f

emal

ein

pov

erty

% f

emal

eon

wel

fare

% m

ale

jobl

ess

Oth

erva

riab

les

% lo

ss o

flo

w-s

kill

edin

dust

ryjo

bs, 1

980–

90

Res

iden

tial

segm

enta

tion

inde

x, 1

990

% m

ale

inpo

vert

y

% m

ale

onw

elfa

re

0.58

0.07

–0.0

9

–0.1

8

–0.4

9

–0.0

1

0.28

0.04

0.51

0.08

0.45

0.32

0.54

0.08

0.33

0.78

0.17

p <

0.05

p >

0.0

5


by multiplying the unstandardized coefficient by the ratio of thestandard deviation of the independent variable to the standarddeviation of the dependent variable. For group comparisons, suchas ours between subgroups of race and gender, the unstan-dardized form is more desirable because differences in the mag-nitude of the standardized coefficients may be attributed todifferences in variance of the variables between groups. Inaddition, the variance, and hence the standard deviation of avariable, is subject to sampling fluctuation. Thus, it increasesinstability of the standardized coefficients and presents prob-lems, particularly for group comparison.

The t-values (significance tests) and coefficients of multipledetermination (R2) along with the unstandardized path coeffi-cients (tables 4 and 5) provide general support for both theurban economic restructuring and welfare disincentive perspec-tives. However, the relative importance of the two perspectivesvaries among the four race-gender groups, as will be elaboratedlater.

The structure and dynamics of the city job market have signifi-cant effects on skill and spatial mismatches. A larger base(proportion) of low-skilled jobs in 1980 decreases future skillmismatch, while loss of low-skilled jobs in the subsequent period(1980 to 1990) increases skill mismatch. The same patternappears for whites and African Americans. Also, the extent oflow-skilled jobs in the cities in 1980 significantly reduces spatialmismatch. The effect of a low-skilled job base and a low-skilledjob loss on both skill and spatial mismatches is stronger forAfrican Americans than for whites. This finding implies thaturban economic restructuring has indeed had a more detrimentaleffect on job prospects for African Americans than for whites.

Apropos of human capital, low education level of city residents isstrongly related to skill mismatch (as should be expected almostby definition) for whites as well as for African Americans. Thisstrong relationship does not hold for spatial mismatch. Residen-tial segregation also affects spatial mismatch for both racialgroups; and again, as would be expected, it is much stronger forAfrican Americans than for whites.

Continuing forward in the model, both skill and spatial mis-matches have statistically significant positive effects on joblessrates of non-Hispanic white men and women. As for AfricanAmericans, while all four coefficients are in the expected direc-tion, only female joblessness and spatial mismatch are signifi-cantly related, though powerfully so. These somewhat surprising


Tab

le 4

. U

nst

an

da

rdiz

ed P

ath

Co

effi

cien

ts f

or

No

n-H

isp

an

ic W

hit

es

ED

MA

TC

H_W

SP

MA

TC

H_W

JOB

LE

SS

_WW

JOB

LE

SS

_WM

PO

VE

RT

Y_W

WP

OV

ER

TY

_WM

WE

LF

AR

E_W

WW

EL

FA

RE

_WM

HS

OR

LS

_W0.

68–0

.01

(13.

64)

(0.3

6)P

CT

LS

J80

–0.3

2–0

.10

(–4.

56)

(–1.

96)

PC

TL

SJL

0.05

–0.0

2(2

.14)

(–1.

02)

SE

GIN

DE

X0.

10(3

.08)

ED

MA

TC

H_W

0.18

0.13

(2.3

1)(2

.73)

SP

MA

TC

H_W

0.65

0.42

(3.0

2)(3

.23)

JOB

LE

SS

_WW

0.19

(2.3

1)JO

BL

ES

S_W

M0.

25(3

.11)

PO

VE

RT

Y_W

W0.

68(9

.90)

PO

VE

RT

Y_W

M0.

40(6

.63)

WE

LF

AR

E_W

W0.

03(0

.18)

WE

LF

AR

E_W

M0.

39(2

.25)

R2

0.80

0.20

0.19

0.31

0.09

0.18

0.42

0.29

Not

e : T

he

t-va

lue

for

eac h

coe

ffic

ien

t ap

pear

s in

par

enth

eses

ben

eath

th

e c o

effi

cien

t.


Tab

le 5

. U

nst

an

da

rdiz

ed P

ath

Co

effi

cien

ts f

or

No

n-H

isp

an

ic B

lack

s

ED

MA

TC

H_B

SP

MA

TC

H_B

JOB

LE

SS

_BW

JOB

LE

SS

_BM

PO

VE

RT

Y_B

WP

OV

ER

TY

_BM

WE

LF

AR

E_B

WW

EL

FA

RE

_BM

HS

OR

LS

_B0.

58–0

.09

(10.

05)

(–1.

31)

PC

TL

SJ8

0–0

.49

–0.1

8(–

8.08

)(–

2.88

)P

CT

LS

JL0.

07–0

.01

(2.9

7)(–

0.32

)S

EG

IND

EX

0.28

(5.4

6)E

DM

AT

CH

_B0.

040.

08(0

.32)

(0.8

2)S

PM

AT

CH

_B0.

510.

17(2

.78)

(1.1

3)JO

BL

ES

S_B

W0.

45(3

.65)

JOB

LE

SS

_BM

0.33

(3.1

2)P

OV

ER

TY

_BW

0.54

(5.7

6)P

OV

ER

TY

_BM

0.08

(1.0

2)W

EL

FA

RE

_BW

0.32

(4.2

4)W

EL

FA

RE

_BM

0.78

(5.6

6)

R2

0.75

0.44

0.37

0.33

0.26

0.15

0.39

0.06

Not

e: T

he

t-va

lue

for

each

coe

ffic

ien

t ap

pear

s in

par

enth

eses

ben

eath

th

e co

effi

c ien

t.


results suggest that less-educated inner-city whites are at leastas sensitive to structural disarticulations in local job markets astheir African-American counterparts are.

Of perhaps even greater interest and importance is the muchstronger effect of spatial mismatch on joblessness among womenof both races, compared to men. This finding would imply thatincreased distancing of jobs from city residents has had a greaterimpact on job prospects of inner-city women than on those ofinner-city men. Because multiple family obligations (e.g., childcare, housekeeping, grocery shopping, care for elderly or infirmrelatives) fall disproportionately on women, they are likely toneed shorter commutes or to be less willing to accept a longcommute in the first place. Single female householders withyoung children (who make up the majority of urban poor) areespecially likely to have temporal and spatial constraints thateither restrict them to seeking jobs close to home (often part-time or at lower pay) or, for those lacking resources for domesticassistance, encourage them to stay out of the job market entirely(Organization for Economic Cooperation and Development[OECD] 1995).

The strong effect of spatial mismatch on female joblessnessprovides some indirect credence to Hayden’s (1980) contentionthat urban areas, as spatially designed and serviced, can betteraccommodate men’s daily routines than women’s. Researchconsistently demonstrates that urban women depend on publictransportation more than men do. Even when women are em-ployed, their additional domestic responsibilities often requirecomplex journeys to nonwork destinations and off-peak transportservices (OECD 1995). Public transportation does not have thespatial and temporal flexibility of the private car, nor does itoperate at the same frequency all day. Making women’s jobaccessibility even more difficult is their avoidance of publictransportation in dangerous areas of cities (where many reside),particularly for journeys after dark, out of fear of violence orharassment. Such mobility constraints no doubt factor into workdisincentives for inner-city women. Together with the morecomplex travel patterns and domestic responsibilities of women,these constraints serve to limit their job choices, thereby differ-entially contributing to women’s joblessness.

As both the urban economic restructuring and welfare disincen-tive paradigms would predict, there are significant direct effectsof joblessness on poverty and of poverty on welfare. A higherjobless rate results in a higher poverty rate for all four race-gender groups. This linkage is particularly strong among African


Americans. A higher poverty rate, in turn, leads to a higherpercentage of people receiving welfare, with the exception ofAfrican-American men, for whom we found a positive butstatistically nonsignificant relationship.

The significant positive feedback effect running from urbanwelfare assistance rates to jobless rates provides key support forthe welfare disincentive argument. We found statistically signifi-cant positive feedback effects for all race-gender subgroupsexcept non-Hispanic white women. We also found that the feed-back effect from welfare to joblessness is considerably strongeramong African Americans than whites and, among AfricanAmericans, is particularly strong among men. When followedthrough the complete cycle, these results are consistent with thethesis that, by discouraging work, greater welfare programparticipation indirectly contributes to greater urban poverty.7

Summary and policy implications

Our study used data for 67 of the largest U.S. cities to assess thetwo main paradigms seeking to explain urban joblessness andpoverty. To do this, we developed and tested a fully specifiedmodel incorporating primary causal operators from each theo-retical approach. In particular, we introduced specifications andmeasures for skill and spatial mismatches—which are pivotal tostructuralists’ explanations of urban joblessness and poverty—and for nonrecursive (feedback) effects of public assistance.These effects directly influence joblessness, indirectly influencepoverty, and are central to welfare disincentive arguments.

Results from assessing the full-form model revealed that theeconomic restructuring and the welfare disincentive paradigmsare not necessarily conflicting explanations of urban joblessness

7 Some readers may wonder how reciprocal effects can be separated usingcross-sectional data. Our model is frequently used in economics, but its basicassumptions are seldom mentioned in a substantive way (statistical assump-tions are discussed). A system with reciprocal or feedback causation is essen-tially a dynamic system. Given time, the process of mutual interaction willsettle, at least temporarily, in a particular pattern often referred to as anequilibrium. Our model assumed that the system has operated for a periodlong enough to lead to a stable pattern that can be captured from cross-sectional data (Heise 1975). The equilibrium pattern is a consequence of therelative strength among the interacting elements in the system. The purposeof the system parameter, or path coefficient estimation, is to determine themagnitude of causal impact of each element on the other, similar to a Markovprocess in demography.


and poverty. They may, in fact, operate side by side to reinforcejoblessness and poverty. The relative strength of these twoexplanations varies by race and gender.

Resident education levels, urban industrial structure, and racialsegregation all play roles in influencing skill and spatial mis-matches for both African Americans and non-Hispanic whites.Urban industrial structure and racial segregation, however, havea greater impact on skill and spatial mismatches for AfricanAmericans than for whites. In turn, these mismatches, as well aswelfare program participation, contribute to urban joblessnesswith relative effects again differing considerably by race andgender.

Skill and spatial mismatches, in general, have greater effects onjoblessness among white residents than among African Ameri-cans, despite a quite strong association between spatial mis-match and joblessness for African-American women. Conversely,welfare program participation contributes more to joblessnessamong African-American city residents than among white resi-dents, with the complete welfare-joblessness-poverty cycle strongestamong female African Americans.

Although relationships between individual variables in thecomplete poverty cycle all operate in the same direction amongthe four race-gender groups, substantial differences exist in themagnitude of effects within each group. Joblessness contributedto poverty rates least among white women and most amongAfrican-American women. Such differences may be related tofamily structure, since it has been well documented that a lowerpercentage of African-American women have working spouses.8Poor women, regardless of race, typically have greater access towelfare assistance than poor men do, primarily because of thewidely available AFDC program.

As noted, welfare program participation contributes more tojoblessness among African Americans than among whites. Thisfinding may be attributed to fewer job opportunities and lower

8 While working is an attribute of an individual, the Bureau of the Censusdefines poverty in terms of the total income of the family to which the indi-vidual belongs, with the poverty cutoff adjusted for family size. Table A-1presents the matrix for 1989 income poverty thresholds (used for 1990 censusdata analysis). These thresholds range from $5,947 for single persons aged 65or over to $27,595 for nine-or-more-person families with one related childunder age 18. When determining which income threshold to apply, the Bureauof the Census cross-classifies the previous year’s family income with thecurrent year’s family size.


wages that discourage African-American welfare recipients fromseeking employment alternatives. Higher rates of welfare pro-gram participation by men of both races contribute to greaterjoblessness and suggest that the small percentage of men whoactually do receive welfare (see mean rates presented in table 2)may be more likely to be caught in a welfare-driven povertycycle.

Given the considerable support we found for the urban restruc-turing and welfare disincentive paradigms, dismissing eitherwould be premature. Indeed, the results imply that policy effortsto reduce inner-city unemployment and poverty should be simul-taneously targeted (1) to mitigating spatial and skill mismatchesand (2) to reforming welfare. In terms of current policy debates,conservatives must recognize that in America’s cities today,housing segregation and other structural barriers to job accesscontribute significantly to unemployment and poverty. At thesame time, liberals must recognize that welfare does discouragework and, for many, has become a poverty trap.

To reduce structural barriers and to improve mobility and jobaccess by the inner-city disadvantaged, numerous policy pre-scriptions have been offered, including the following:

1. Attack both residential segregation and spatial accessdirectly by dispersing housing assistance programs (such asthe Gautreaux assisted-housing experiments in Chicago).

2. Provide tax incentives for affordable housing construction inthe suburbs by the private sector.

3. Expand housing voucher programs as opposed to providingadditional spatially fixed public housing in the inner city.

4. Develop networks in the inner city to provide informationabout job openings throughout the metropolitan area andbeyond.

5. Partially underwrite more distant job searches by the inner-city unemployed.

6. Assist temporary needs-based relocation once a more distantjob has been secured.

7. Strictly enforce existing fair housing and fair hiring laws.


8. Support public-private efforts to establish vanpools sounemployed inner-city residents can commute to suburbanbusinesses facing worker shortages.

9. Upgrade the quality of public schools in the cities and helpqualified disadvantaged residents obtain higher education.

10. Improve vocational education, apprenticeship, and co-opprograms that smooth the school-to-work transition.

Each of the above prescriptions, of course, has its own complexi-ties and difficulties of implementation. For example, programs toenhance skill levels and employment readiness of inner-cityresidents will do little to reduce unemployment and povertyunless there are real jobs available at the end of the trainingprogram. This means that labor supply and demand approachesmust be better integrated and skills training must be directlyconnected to labor market needs. Moreover, basic skills trainingwill be more meaningful and effective when taught with anactual job as a frame of reference and where both trainees andpotential employers can see the link between skill developmentand job performance (Social Science Research Council 1993).

Much attention was directed in the late 1970s and throughoutthe 1980s to place-based economic development and jobs-to-


Americans was influenced by racial prejudice and stereotypes.For example, Cole and Deskins (1988) present evidence aboutthe discriminatory siting and hiring practices of Japanese auto-parts manufacturers and suppliers who set up in the UnitedStates during the 1980s. These manufacturers and supplierslocated in areas with few African Americans in the commutingrange and hired fewer African Americans than would be ex-pected from their local labor shed racial composition. Likewise,careful examination of the locations of economically boomingouter suburban places indicates that they tend to be within thelabor commuting shed of the metropolis, but at the farthestpoints from concentrations of poor African Americans (Kasarda1995).

These observations reinforce the point that urban African-American joblessness is intertwined with continuing racialprejudice affecting employer site decisions and with otherstructural barriers not addressed in this article. Policies tocombat discriminatory siting of business establishments as aconsequence of racial bias, though not easy to either develop orenforce, should be considered.

The powerful effect of spatial mismatch on female joblessnesssuggests that gender bias in urban design and transportationservices also needs to be considered. The deconcentration ofmetropolitan jobs, together with restricted transport choice,differentially impacts the least mobile—that is, less-educatedinner-city women. These women are most likely to (1) dependentirely on public transportation, (2) travel close to home,(3) seek only jobs with short commute times, (4) avoid work thatrequires traveling through nearby dangerous areas (especiallyafter dark), and (5) need to balance multiple domestic respon-sibilities with work schedules. As a result, job options for thesewomen tend to be much more restricted spatially and temporally,often limiting them to low-paying and part-time work closer tohome. These constraints no doubt pose strong workdisincentives.

Policy makers and urban planners need to be aware that womentypically have more complex travel patterns and accessibilityneeds than men, most of whom travel only to and from work.Public transportation, on which urban women disproportionatelydepend, poorly serves their multiple-destination needs. Forinstance, public transportation is limited (or absent) during off-peak periods and is designed primarily to serve radial routesinto and out of the central business district rather than dis-persed employment sites and nonwork sites such as shopping


and day-care facilities. When municipal budgets are strained,off-peak and secondary, nonradial public transit routes areusually the first to be cut. Improving accessibility of inner-citywomen to alternative work and nonwork sites should be a pri-ority of any employment linkage strategy.

Perhaps the greatest policy challenge to facilitating the transi-tion to work is overcoming spatial and temporal constraints thatprevent women with children from accommodating their domes-tic and work responsibilities. Given the likelihood that womenwill continue to need to work closer to home, programs thatpromote self-employment including home work and nontradi-tional neighborhood employment should be pursued. For ex-ample, restrictions against public housing tenants operatingbusinesses from their units should be relaxed. In addition,public-private partnerships to help women find and keep jobswithin or near their neighborhoods should be encouraged.

One highly successful example of the latter is the MinneapolisNeighborhood Employment Network (Brinda 1994). This pro-gram is targeted to “hard to employ” residents, especiallywomen, and helps them qualify for local jobs, both traditionaland nontraditional, such as fire protection and law enforcement.The network provides neighborhood recruitment and job train-ing, transportation assistance, access to child-care services, andcounseling. In 1993, this community-based partnership helpedmore than 300 Minneapolis women enter their neighborhoodwork force while providing local employers with trainedemployees.

Policy reforms have also been proposed to reduce welfare disin-centives to work (Ellwood 1988). Similar to those suggested toreduce structural barriers to work, they are not as uncompli-cated or easy to implement as may appear on the surface. Mostare based on the well-documented facts that the vast majority ofpeople receiving welfare (particularly AFDC recipients) arecapable of working and that women with infants and childrenare no longer expected to stay home with them. The reformreceiving greatest attention in the mid-1990s is strict time limitson public assistance (two to five years), except for the disabled.Under these proposals, after a set period of time there will be nowelfare checks, just paychecks.

Along with strict time limits, a series of financial rewards andpenalties have been proposed to encourage skills training, jobplacement, and child support as well as to make working moreremunerative than welfare:


1. Boost incomes of all full-time workers above the poverty lineby expanding the earned-income tax credit.

2. Reimburse employers for their share of social security,Medicaid, etc., for welfare recipients they hire.

3. Provide employers with tax credits to pay above-povertywages.

4. Mandate that all welfare recipients participate in skillstraining and job placement programs, including placementin public-sector jobs, if necessary.

5. Require young mothers with children to complete highschool.

6. Hold biological fathers and deadbeat dads accountable formeeting child-support obligations and make strong efforts tocollect payments.

7. Assist in the establishment of private day-care servicesowned and operated by former welfare mothers in low-income neighborhoods.

8. Lower the attractiveness of welfare relative to work byreducing the size of benefits, the range of assistance pro-grams, and the ease with which they can be tapped.

9. Ensure that all government bodies, welfare agencies, andsocial workers continually reinforce the message that wel-fare is not an entitlement (if you are poor, you get money) oran alternative to work but entails a reciprocal obligation tobecome self-sufficient through education, responsible familybehavior, and work.

The strong feedback effects from welfare rates to jobless ratesamong the less-educated in cities raise a major caution flag tothose who believe the best way to help the poor is to increase thesize and scope of public-assistance programs. While certainindividual needs might be temporarily better met, expansion ofthe welfare program could well dissuade many from takingmoderately or poorly paid jobs. Simply put, poor people may bepoor, but they are not stupid. They make rational decisionsabout relative economic benefits, just as the more fortunate do.Thus, more generous transfer payments together with additionalin-kind assistance programs would likely reduce work incentivesand increase inner-city joblessness, while doing nothing to alterrecipient attitudes that welfare as a way of life is acceptable.


Tab

le A

-1.

Po

ver

ty T

hre

sho

lds

in 1

989

by

Siz

e o

f F

am

ily

an

d N

um

ber

of

Rel

ate

d C

hi l

dre

n u

nd

er 1

8 Y

ears

, in

Do

lla

rs

Rel

ated

Ch

ildr

en u

nde

r 18

Yea

rsW

eigh

ted

Ave

rage

Eig

ht

orS

ize

of F

amil

y U

nit

Th

resh

olds

Non

eO

ne

Tw

oT

hre

eF

our

Fiv

eS

ixS

even

Mor

e

On

e pe

rson

(u

nre

late

d in

divi

dual

)6,

310

Un

der

65 y

ears

6,45

16,

451

65 y

ears

or

olde

r5,

947

5,94

7

Tw

o pe

r son

s8,

076

Hou

seh

olde

r u

nde

r 65

yea

r s8,

343

8,30

38,

547

Hou

seh

olde

r 65

yea

r s o

r ol

der

7,50

17,

495

8,51

5

Th

r ee

pers

ons

9,88

59,

699

9,98

19,

990

Fou

r pe

r son

s12

,674

12,7

9012

,999

12,5

7512

,619

Fiv

e pe

r son

s14

,990

15,4

2415

,648

15,1

6914

,798

14,5

72

Six

per

son

s16

,921

17,7

4017

,811

17,4

4417

,092

16,5

6916

,259

Sev

en p

erso

ns

19,1

9220

,412

20,5

4020

,101

19,7

9419

,224

18,5

5817

,828

Eig

ht

per s

ons

21,3

2822

,830

23,0

3122

,617

21,7

3821

,738

21,0

8420

,403

20,2

30

Nin

e pe

rson

s or

mor

e25

,480

27,4

6327

,595

27,2

2926

,921

26,4

1525

,719

25,0

8924

,933

23,9

73

Sou

rce:

U.S

. B

ur e

au o

f th

e C

ensu

s (1

993)

.


Farley, John E. 1987. Disproportionate Black and Hispanic Unemployment inU.S. Metropolitan Areas. American Journal of Economics and Sociology46:129–50.

Hargens, Lowell L. 1988. Estimating Multiequation Models with CorrelatedDisturbances. In Common Problems/Proper Solutions, ed. J. Scott Long,65–83. Newbury Park, CA: Sage Publications.

Harrison, Bennett. 1974. Urban Economic Development. Washington, DC: TheUrban Institute Press.

Hayden, Dolores. 1980. What Would a Non-Sexist City Be Like? Speculationson Housing, Urban Design and Human Work. Signs 5(3) supplement: 170–87.

Heise, David R. 1975. Causal Analysis. New York: Wiley.

Holzer, Harry J. 1991. The Spatial Mismatch Hypothesis. Urban Studies28:105–22.

Ihlanfeldt, Keith R., and David L. Sjoquist. 1989. The Impact of JobDecentralization on the Economic Welfare of Central City Blacks. Journal ofUrban Economics 26:110–30.

Ihlanfeldt, Keith R., and David L. Sjoquist. 1990. Job Accessibility and RacialDifferences in Youth Employment Rates. American Economic Review80:267–76.

Ihlanfeldt, Keith R., and David L. Sjoquist. 1991. The Effect of Job Access onBlack and White Youth Employment: a Cross-Sectional Analysis. UrbanStudies 28:255–65.

Jencks, Christopher, and Susan Mayer. 1990. Residential Segregation, JobProximity and Black Job Opportunity. In Inner City Poverty in the UnitedStates, ed. Lawrence Lynn, Jr., and Michael McGeary, 187–222. Washington,DC: National Academy Press.

Johnson, James H., Jr., and Melvin L. Oliver. 1992. Structural Changes in theU.S. Economy and Black Male Joblessness: A Reassessment. In Urban LaborMarkets and Job Opportunity, ed. George E. Peterson and Wayne Vroman,542–62. Washington, DC: The Urban Institute Press.

Johnston, John. 1984. Econometric Methods. New York: McGraw-Hill.

Judge, George G., W. E. Griffiths, R. Carter Hill, Helmut Lutkepohl, and T. C.Lee. 1985. The Theory and Practice of Econometrics, 2nd edition. New York:Wiley.

Kain, John F. 1968. Housing Segregation, Negro Employment, and Metropoli-tan Decentralization. The Quarterly Journal of Economics 82:175–97.

Kain, John F. 1992. The Spatial Mismatch Hypothesis: Three Decades Later.Housing Policy Debate 3(2):371–460.

Kasarda, John D. 1976. The Changing Occupational Structure of the AmericanMetropolis: Apropos of the Urban Problem. In The Changing Faces of theSuburbs, ed. Barry Schwartz, 113–36. Chicago: University of Chicago Press.


Kasarda, John D. 1985. Urban Change and Minority Opportunities. In TheNew Urban Reality, ed. Paul E. Peterson, 33–67. Washington, DC: BrookingsInstitution.

Kasarda, John D. 1990. Structural Factors Affecting the Location and Timingof Urban Underclass Growth. Urban Geography 11:234–64.

Kasarda, John D. 1993a. Cities as Places Where People Live and Work: UrbanChange and Neighborhood Distress. In Interwoven Destinies: Cities and theNation, ed. Henry G. Cisneros, 81–124. New York: W. W. Norton.

Kasarda, John D. 1993b. Urban Underclass Database: An Overview andMachine-Readable File Documentation. Chapel Hill, NC: Frank HawkinsKenan Institute of Private Enterprise.

Kasarda, John D. 1995. Industrial Restructuring and Changing Job Locations.In State of the Union: America in the 1990s, Vol. 1, ed. Reynolds Farley,215–67. New York: Russell Sage.

Leonard, Jonathan S. 1987. The Interaction of Residential Segregation andEmployment Discrimination. Journal of Urban Economics 21:323–46.

Lichter, Daniel T. 1988. Racial Differences in Underemployment in AmericanCities. American Journal of Sociology 93:771–92.

Maddala, G. S. 1977. Econometrics. New York: McGraw-Hill.

Massey, Douglas S. 1990. American Apartheid: Segregation and the Making ofthe Underclass. American Journal of Sociology 96:329–57.

McLanahan, Sara, Irwin Garfinkel, and Dorothy Watson. 1988. FamilyStructure, Poverty, and the Underclass. In Urban Change and Poverty, ed.Michael G. H. McGeary and Lawrence E. Lynn, Jr., 102–47. Washington, DC:National Academy Press.

Mead, Lawrence M. 1988. The Hidden Jobs Debate. Public Interest 91:40–58.

Mead, Lawrence M. 1989. The Logic of Workfare: The Underclass and WorkPolicy. The Annals of the American Academy of Political and Social Science501:156–69.

Mooney, Joseph D. 1969. Housing Segregation, Negro Employment andMetropolitan Decentralization: An Alternative Prospective. Quarterly Journalof Economics 83:299–311.

Moss, Philip, and Chris Tilly. 1991. Why Black Men Are Doing Worse in theLabor Market. New York: Social Science Research Council.

Murray, Charles A. 1984. Losing Ground: American Social Policy, 1950–1980.New York: Basic Books.

Murray, Charles A. 1986. No, Welfare Really Isn’t the Problem. Public Interest84:3–11.

Noyelle, Thierry J. 1987. Beyond Industrial Dualism: Market and JobSegmentation in the New Economy. Boulder, CO: Westview.


Offner, Paul, and Daniel H. Saks. 1971. A Note on John Kain’s HousingSegregation, Negro Employment, and Metropolitan Decentralization. TheQuarterly Journal of Economics 191:147–60.

Organization for Economic Cooperation and Development. 1995. Women in theCity: Housing, Services, and the Urban Environment. Paris, France.

Price, Richard, and Edwin S. Mills. 1985. Race and Residence in EarningsDetermination. Journal of Urban Economics 17:1–18.

Rothenberg, Thomas J. 1990. Full and Limited Information Methods. InEconometrics, ed. John Eatwell, Murray Milgate, and Peter Newman, 88–94.London: Macmillan.

Social Science Research Council. 1993. Persistent Urban Poverty: IntegratingResearch, Policy and Practice. New York: Social Science Research Council.

Stanback, Thomas M. 1991. The New Suburbanization: Challenge to theCentral City. Boulder, CO: Westview.

Straszheim, Mahlon R. 1980. Discrimination and the Spatial Characteristics ofthe Urban Labor Market for Black Workers. Journal of Urban Economics7:119–40.

Tanner, Michael, Stephen Moore, and David Hartman. 1995. The Work vs.Welfare Trade-Off: An Analysis of the Total Level of Welfare Benefits by State.In the Policy Analysis series, No. 240, pp. 1–53. Washington, DC: CatoInstitute.

U.S. Bureau of the Census. 1993. 1990 Census of Population and Housing.Public Use Microdata Sample: Technical Documentation. Washington, DC.

Vrooman, John, and Stuart Greenfield. 1980. Are Blacks Making It in theSuburbs? Some New Evidence on Intrametropolitan Spatial Segmentation.Journal of Urban Economics 7:155–67.

Wacquant, Loic J. D., and William J. Wilson. 1989. The Cost of Racial andClass Exclusion in the Inner City. The Annals of the American Academy ofPolitical and Social Science 501:8–25.

Weidenbaum, Murray. 1991. Beyond Handouts. Across the Board 28:49–52.

Wilson, William J. 1987. The Truly Disadvantaged: The Inner City, theUnderclass, and Public Policy. Chicago: University of Chicago Press.

Zax, Jeffrey S. 1990. Race and Commutes. Journal of Urban Economics28:336–48.

Zax, Jeffrey S., and John F. Kain. 1991. Commutes, Quits and Moves. Journalof Urban Economics 29:1–13.

Documents

Housing Policy Debate - Amazon Web Services · letter variables should be italicized. Multiple-letter variables and abbreviations (e.g., AGE) and functions (e.g., exp, min, ln) should