Subsidized housing, housing prices, and the living arrangements of unmarried mothers

Abstract

Although many studies estimate the effects of welfare benefits on mothers’ living arrangements, housing subsidies and prices are rarely the focus. This article uses a new longitudinal birth cohort study, the Fragile Families and Child Wellbeing Study, to examine the relationship between subsidized housing, housing prices, and the living arrangements of unmarried mothers three years after a nonmarital birth.Results suggest that the availability of subsidized housing is negatively associated with marriage relative to living alone. Eligibility criteria and means testing in subsidized housing may make marriage a costly choice. Housing prices are positively associated with marriage, cohabitation, and living with family members relative to living alone. Economies of scale may be particularly important for single‐earner households when housing prices increase. Failure to control for housing costs and subsidies leads to underestimates of the effects of welfare and unemployment rates on the living arrangements of unmarried mothers.

Full text

145
Subsidized Housing, Housing
Prices, and the Living Arrangements
of Unmarried Mothers
Marah A. Curtis
Boston University
Abstract
Although many studies estimate the effects of welfare benefits on mothers’
living arrangements, housing subsidies and prices are rarely the focus. This
article uses a new longitudinal birth cohort study, the Fragile Families and
Child Wellbeing Study, to examine the relationship between subsidized hous
-
ing, housing prices, and the living arrangements of unmarried mothers three
years after a nonmarital birth.
Results suggest that the availability of subsidized housing is negatively
associated with marriage relative to living alone. Eligibility criteria and means
testing in subsidized housing may make marriage a costly choice. Housing
prices are positively associated with marriage, cohabitation, and living with
family members relative to living alone. Economies of scale may be particu
-
larly important for single-earner households when housing prices increase.
Failure to control for housing costs and subsidies leads to underestimates of
the effects of welfare and unemployment rates on the living arrangements of
unmarried mothers.
Keywords: Affordability; Demographics; Welfare
Introduction
Households, particularly low-income renters, often spend the greatest
portion of their income on housing (Bureau of Labor Statistics 2002; Hart
-
man 1998; Mulroy and Ewalt 1996). The national shortage of low-income
housing underscores the ongoing difficulties that poor households face in
locating and maintaining an affordable unit (U.S. Department of Housing
and Urban Development [HUD] 2002, 2003). Estimates of housing cost bur
-
den indicate that at least one renter household in three experiences moder
-
ate cost burdens, while about one in five experiences severe cost burdens
(Belsky, Goodman, and Drew 2005). Housing assistance, however, is not
HOUSING POLICY DEBATE VOLUME 18 ISSUE 1
© 2007 FANNIE MAE FOUNDATION. ALL RIGHTS RESERVED.

FANNIE MAE FOUNDATION
146 Marah A. Curtis
an entitlement and has never been available to all eligible households (U.S.
House Committee on Ways and Means 2004). It was estimated that in 1999,
1 million households were on the waiting list for Section 8 housing, with
an average wait of 28 months (HUD 1999), and there is no indication that
the numbers have decreased substantially. The well-documented paucity of
affordable housing, the high levels of housing cost burdens for renter fami
-
lies, and the relative scarcity of subsidized housing suggest the marked impor
-
tance of both housing prices and subsidies in the lives of unmarried mothers
and their children.
Households headed by unmarried mothers are frequently the focus of
public policy research and debate. This attention is largely due to the associa
-
tion between single motherhood and poverty, as well as research suggesting
benefits for children who grow up with both biological parents (McLanahan
and Sandefur 1994; Powell and Parcel 1997; U.S. House Committee on Ways
and Means 2004). Mothers who are unmarried at the birth of a child are of
particular interest to policy makers due to their greater likelihood of pov
-
erty and reliance on public benefits. The Personal Responsibility and Work
Opportunity Reconciliation Act of 1996 (PRWORA, P.L. 104–193) changed
welfare from an entitlement to a block grant while imposing time limits, work
requirements, and family formation goals. Three out of four goals guiding
this legislation refer directly to family formation via promoting marriage,
reducing out-of-wedlock pregnancies, and encouraging two-parent families.
The current policy emphasis on family formation and marriage among poor
unmarried mothers must take into account the myriad factors affecting these
families. Research accounting for the impact of an understudied, but very
valuable, benefit like subsidized housing is needed.
Housing subsidies
Although a large body of literature examines the effects of cash wel-
fare grants on single motherhood, housing subsidies are rarely considered
(Blau, Kahn, and Waldfogel 2004; Danziger et al. 1982; Ellwood and Bane
1985; Hoynes 1997; Moffitt 1994; Moffitt, Reville, and Winkler 1998). The
value of a rental subsidy, however, may be quite substantial and particularly
important considering the size of the benefit and the fact that stable housing
is a prerequisite for establishing a functioning family life. Housing subsidies
help provide shelter and a stable living environment while freeing up income
to be used for food, child care, clothes, medical care, and other essentials.
Waiting lists for subsidized housing are quite long precisely because this is a
scarce and extremely valuable benefit.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 147
Research focusing on subsidized housing and employment offers some
useful insights. PRWORA placed a strong emphasis on work and included
strict time limits on receiving welfare benefits. Leaving the welfare rolls and
becoming self-sufficient are the major goals of this legislation. If housing
subsidies hinder employment by reducing the need to work, this is a con
-
cern for policy makers. On the one hand, the economics literature suggests
that any nonwage income or in-kind benefit is a disincentive to work (Ashen-
felter 1983; Moffitt 1992). On the other hand, it is also possible that housing
assistance, coupled with other benefits like food stamps and Medicaid health
insurance, can provide a level of stability that supports consistent employ
-
ment (Harkness and Newman 2006). Housing subsidies affect income by
freeing up money that would otherwise have been used for rent. Empirical
results have offered a nuanced relationship between housing subsidies and
the work efforts of welfare recipients. For example, Section 8 residents in
California worked considerably more hours than welfare recipients living in
either public housing or private market rentals (Ong 1998). Different effects
on work may be due to the fact that Section 8 recipients can use their vouch
-
ers to relocate to an area with a more robust labor market.
Several studies consider subsidized housing and aspects of household
composition directly (Ellen and O’Flaherty 2006; Freeman 2005; London
2000; Turner 2003). Ellen and O’Flaherty (2006) use data from the Social
Indicators Survey in New York City to determine the factors influencing
household size and find that fewer adults reside in households that receive
housing subsidies. Turner (2003) uses baseline data from the Fragile Families
and Child Wellbeing (FFCW) Study to look at cohabitation among unmar
-
ried parents in government-assisted housing. The FFCW sample is drawn
from 20 U.S. cities selected to provide variation in housing markets, labor
markets, and welfare regimes. This birth cohort study follows sample mem
-
bers longitudinally, with the latest wave of data available from approximately
three years after the birth of the child. Findings suggest that mothers who live
in government-assisted housing are less likely to cohabit than mothers in pri
-
vate rental housing. Using data from the New York City Housing Vacancy
Survey, Freeman (2005) finds that married and cohabiting couples are less
likely to receive housing assistance.
These studies strongly suggest that subsidized housing influences both
the number of adults in the household and the composition of recipient
families. Cross-sectional studies using recipient status to gauge the effect of
subsidized housing cannot identify causal relationships since it is not clear
which way causation may run. Household composition may be determined
by the subsidy, or certain kinds of households may be more likely to receive a

FANNIE MAE FOUNDATION
148 Marah A. Curtis
subsidy. Using household subsidy status as an explanatory variable to study
living arrangements makes it less clear that results are capturing the effects
of subsidies. An exogenous measure that takes into account the availability
of this benefit rather than individual receipt will offer a clearer estimate of
subsidized housing.
London (2000) uses an exogenous control for subsidized housing—the
amount of state spending for Section 8 housing assistance programs—and
finds that spending negatively affects the probability of a mother’s sharing
a unit with others. The present study will use a new measure of subsidized
housing: the number of available public housing units, certificates, Section 8
vouchers, and Low-Income Housing Tax Credit (LIHTC) units under con
-
tract divided by the eligible city population. Results using this measure sug
-
gest that mothers living in metropolitan statistical areas (MSAs) with more
subsidized housing are more likely to live alone or with extended family
relative to marriage (Curtis 2005).
Housing prices
All else being equal, higher housing prices make it more expensive to live
alone. Given their overall lower income, unmarried mothers may be particu
-
larly sensitive to changes in housing prices. The likelihood of budget trade-
offs between increasing housing prices and other consumption may be very
limited. Several studies consider the impact of rent levels on mothers’ living
arrangements (London 2000; Sigle-Rushton and McLanahan 2002; Winkler
1992). These studies find that higher rents increase the probability of living
in shared arrangements relative to living alone. Using 1980, 1990, and 2000
census data, Curtis (2005) tests the impact of both HUD’s fair market rents
(FMRs) and hedonic single-family owner-occupied housing price indexes on
the living arrangements of mothers. Both sets of results are nearly identical;
both measures of housing prices are negatively associated with living alone.
Aggregate and micro data studies focusing on the household formation deci
-
sions of young people and the elderly have found a large and significant
impact of housing costs on household headship rates (Borsch-Supan 1986;
Haurin, Hendershott, and Kim 1993; Hughes 2003; Smith et al. 1984).
Clearly, housing costs matter.
In considering the association between subsidized housing, housing
prices, and the living arrangements of unmarried mothers, it is important to
control for other relevant factors. This ensures that subsidized housing and
prices will be evaluated with markedly less concern about omitted variables.
Failing to include a full set of personal and human capital characteristics may

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 149
misstate the effects of housing subsidies and prices on living arrangements.
The next section will present a brief overview of pertinent economic, socio
-
logical, and demographic theory and findings related to living arrangements,
marriage, and union formation.
Living arrangements
Opportunities, constraints, personal history, tastes, finances, and the
availability of family all have an impact on whether a mother and her chil
-
dren live alone or with others. These decisions depend on a host of factors,
both pecuniary and nonpecuniary, associated with possible arrangements.
Economic theory suggests that mothers will weigh the relative costs and ben
-
efits of each alternative living arrangement and choose the one that has the
most benefits (Becker 1981; Bergstrom 1996). Some benefits and costs will be
obvious; others, less so. Living with a grandmother may involve economies
of scale, available child care, and emotional support but at the same time
result in a lack of privacy and a reduction in autonomy. Cohabiting with a
partner may provide intimacy, shared resources, and child care, while living
alone may involve funds from child support enforcement and less aggrava
-
tion. Certainly, the costs and benefits of living alone, marrying, cohabiting
with a partner, living with a roommate, or living with extended family will
vary. The degree to which household expenses are shared, the level of auton
-
omy, and the desire for intimacy, as well as potential public benefits, may
affect living arrangements.
Sociological theories stress a number of personal characteristics that are
particularly important when considering marriage and cohabitation. Includ
-
ing controls for mothers’ nativity, race/ethnicity, age, and number of children
is meant to adjust for cultural norms and life cycle stages related to family
formation (Bennet, Bloom, and Craig 1989; Duncan and Hoffman 1990).
Studies of marriage patterns and markets show that the overwhelming
majority of unions involve partners of the same racial/ethnic group and level
of education. These studies also point to the importance of controlling for
age by gender since women tend to marry slightly older men (Lichter, Ander
-
son, and Hayward 1995; Lichter, LeClere, and McLaughlin 1991; Lichter et
al. 1992). However, mothers’ characteristics are only part of the story. The
human capital characteristics of potential mates are strongly linked to union
formation (Carlson, McLanahan, and England 2004; Carlson et al. 2005;
Sigle-Rushton and McLanahan 2002). Lower levels of education, children
from a previous union, lack of work, and a history of incarceration all make
a male partner decidedly less attractive (Carlson, McLanahan, and England

FANNIE MAE FOUNDATION
150 Marah A. Curtis
2004; Carlson et al. 2005; Mincy and Dupree 2001; Western, Lopoo, and
McLanahan 2003).
Welfare benefits, child support enforcement,
and unemployment rates
Both economic and sociological theories of marriage note the impor-
tance of external factors that affect decisions on living arrangements. Public
benefits make it easier for a mother to live on her own, thereby reducing the
economic benefits of all shared arrangements. As noted earlier, a large body
of research focuses on the effects of cash welfare grants on single moth
-
erhood (Blau, Kahn, and Waldfogel 2004; Danziger et al. 1982; Ellwood
and Bane 1985; Hoynes 1997; Moffitt 1994). On the whole, prior research
has found a modest positive association between welfare benefits and sin
-
gle motherhood.
1
Moffitt, Reville, and Winkler (1998) examine potential
incentive effects for mothers to cohabit while receiving Aid to Families with
Dependent Children subsidies. Whether the male partner is the child’s father
or is unrelated determines how his income is considered for eligibility. A
father’s income is counted, while the income of an unrelated male is not. This
structure creates an incentive to cohabit with an unrelated male rather than
the father. Mincy and Dupree (2001) use the FFCW baseline survey to look
at the family formation of new, unwed parents and find that welfare benefits
enable low-income cohabiting couples to pool resources and facilitate family
living. Cohabitation can remain unreported to welfare offices to avoid count
-
ing a father’s income and potentially losing all benefits.
Child support enforcement and labor market conditions are important
because they help determine economic possibilities. Higher levels of enforce
-
ment may encourage fathers to seek marriage to avoid paying child support,
while at the same time increasing the potential for mothers to live indepen
-
dently. Using FFCW data, Carlson et al. (2005) find that depending on the
measure and the model specification, child support enforcement and munic
-
ipal unemployment rate have mixed effects on union formation. Winkler
(1992) finds a negative relationship between the local area unemployment
rate and the probability of marriage for single mothers.
Opportunities for employment may allow both partners to increase
their own earnings. Increased male earnings facilitate marriage and cohabi
-
tation, while increased earnings for mothers allow them to opt for more
1
Results in these studies may be sensitive to any unmeasured variables that could influence
both single motherhood and state benefit levels (Blau, Kahn, and Waldfogel 2004; Danziger et
al. 1982; Ellwood and Bane 1985; Hoynes 1997; Moffitt 1994).

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 151
independence. Noneconomic factors may be more important when both
partners are enjoying strong earnings opportunities. Beyond marriage and
cohabitation, mothers may also live with a roommate or extended family.
The roommate option probably entails fewer emotional and practical ben
-
efits than the other choices.
Hypotheses and variable measurement
This section will discuss research hypotheses, expected results, and mea-
surement of variables in light of the literature, theory, and previous empirical
results.
Housing prices
Housing prices are measured using an index.
2
It is expected that housing
prices will be negatively associated with living alone relative to all shared
arrangements. Living alone becomes more costly as housing prices increase.
Shared arrangements offer economies of scale, shared expenses, and reduced
pressure to trade off consumption to pay for housing.
Housing subsidies
Housing subsidies free up income for nonhousing consumption and
may enable a mother to live independently. Targeting rental assistance to
very low income households (U.S. House Committee on Ways and Means
2004) makes it more difficult for an unmarried mother to consider bring
-
ing an income-earning husband into the household because she risks loss of
benefits. The income-conditioned criteria of project-based and tenant-based
assistance impose an implicit tax on earnings by reducing benefits for every
dollar over the set threshold. A spouse may add more earnings that disqualify
2
The housing price index uses Malpezzi, Chun, and Green’s 1990 indexes for single
owner-occupied housing across MSAs (1998). The hedonic index methodology uses “regres
-
sion analysis to isolate the contribution to house price of each individual housing characteris
-
tic” (number of rooms, number of bedrooms, age of the structure and structure type, presence
of a kitchen, sewage, water, and the absence of sewage) (Thibodeau 1992, 1). “House price
indexes are constructed by predicting the market rent or market value for a constant bundle of
housing characteristics…[F]or owner-occupied homes, the hedonic measures the average mar
-
ket price of a constant-quality house” (Thibodeau 1992, 1–2). The indexes are then adjusted
to 2000 figures using the Office of Federal Housing Enterprise Oversight (OFHEO) house
price index to time-adjust the 1990 figures. The OFHEO index is a weighted repeat sales index
designed to measure changes in U.S. single-family home values. The indexes are adjusted by
normalizing the OFHEO housing price index to 100 in the first quarter of 1990 and then mul
-
tiplying the 1990 value by the 2000 index and dividing by 100.
Regression results using HUD’s FMRs are available from the author and are nearly identi
-
cal to those presented.

FANNIE MAE FOUNDATION
152 Marah A. Curtis
the family or reduce the value of the subsidy. Recipients generally pay 30 per-
cent of household income toward rent; more household income increases the
rent burden. Public housing guidelines require a household roster and annual
financial data, including tax returns, at application (HUD 2006).
Housing subsidies may enable a mother to live independently rather than
sharing a unit with others. As noted, income-conditioning may make mar
-
riage expensive. It is expected that the local availability of subsidized housing
will be negatively associated with all shared arrangements relative to living
alone. Subsidies allow for more privacy and independence, while program
requirements may reduce benefits if marriage increases income. Although, a
priori, the expected direction of the results is clear, the reason is not. Reduc
-
tion in nonromantic shared arrangements is straightforward, but reductions
in marriage are not. This analysis will not disentangle the impact of income
effects (purchasing privacy) from price effects (rules that make marriage
costly).
The availability of subsidized housing is constructed from A Picture of
Subsidized Households—1998 (HUD 2005) and the LIHTC database (HUD
2006). This measure includes project-based assistance (public housing) and
tenant-based assistance (certificates and Section 8 vouchers), as well as the
total number of LIHTC units. The final measure, the availability of subsi
-
dized units, is the raw units divided by the eligible population in a given
city. The number of units/vouchers under contract is tabulated by MSA and
matched to each of the FFCW cities.
3
Welfare grants plus food stamps
It is expected that the maximum welfare and food stamp benefit available
in a given state will be positively associated with living alone relative to all
shared arrangements. The combination of welfare and food stamp benefits
is an exogenous measure of the potential income available to a poor unmar
-
ried mother. A mother’s actual receipt of benefits is most likely endogenous
to decisions on living arrangements and so will not be included. Welfare and
3
The eligible population was determined using an extract from the Integrated Public Use
Microdata Series (IPUMS) census data that included household income and number of house
-
hold members (Ruggles et al. 2004). Households were weighted to be nationally representa
-
tive. Households with 50 percent of the area median family income were counted as eligible
for subsidized housing. Income was adjusted for family size. IPUMS data were not available
for Corpus Christi, TX, so the total population is used. City population data are from U.S.
Bureau of the Census (2000).
Using an alternate measure of the availability of subsidized housing calculated for house
-
holds with 30 percent of the area median family income yields results very similar to those
presented. These results are available from the author.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 153
food stamp benefits for a family of four are drawn from the Welfare Benefits
Database (Moffitt 2005).
Unemployment rates
Low unemployment rates suggest a robust labor market and increased
economic opportunities for mothers and their potential mates. The impact of
a strong labor market increases mothers’ ability to live alone while simultane-
ously improving the earnings potential and desirability of male partners for
marriage and cohabitation. Given these two potential effects, it is not clear
how unemployment rates will be associated with living alone relative to mar
-
riage and cohabitation. It is expected, however, that low unemployment rates
will be associated with decreases in other shared arrangements relative to liv
-
ing alone. The unemployment rates are from the Bureau of Labor Statistics
and are measured in the year the baseline interviews were conducted (n.d.).
Child support enforcement
It is not clear how the stringency of child support enforcement in a given
state will affect mothers’ living arrangements. Previous empirical results
and theoretical expectations are mixed. More stringent enforcement may
yield larger and more consistent benefits that allow mothers to live indepen
-
dently while simultaneously increasing noncustodial fathers’ desire to marry
in order to stop paying benefits. All else being equal, stricter enforcement
should be associated with living alone relative to other nonromantic shared
arrangements. Data come from the Office of Child Support Enforcement
(OCSE). This state-level measure uses the number of OCSE cases that have
received Temporary Aid to Needy Families (TANF) benefits, with collections
as the numerator and the total OCSE TANF caseload as the denominator in
the baseline interview year (U.S. Department of Health and Human Services
1998). Given the mixed empirical results with child support enforcement
measures in the FFCW data, two alternate city-level measures are also tested
4
(Carlson et al. 2005).
4
These two alternate measures were constructed using the 5 percent Public Use Microdata
Sample data: the proportion of never-married mothers who receive child support in a given
city and the proportion of never-married mothers who receive any child support in a given
city. Both measures are purged of city differences in enforcement by regressing the probability
of receiving child support on the demographic characteristics of the mother. These regression
estimates are used to acquire a predicted value of child support. The final measure is the ratio
of the actual amount divided by the predicted amount, which is standardized with a mean of
zero and a standard deviation of 1.0. Measures were constructed by Lenna Nepomnyaschy.

FANNIE MAE FOUNDATION
154 Marah A. Curtis
Controls for personal characteristics
Including these controls will sharpen estimates of housing subsidies and
prices while controlling for well-established differences in marriage and
cohabitation by race/ethnicity, age, nativity, education, number of children,
and characteristics of male partners. Mother’s working status and disability
status are also included as controls. In light of the literature, it is expected
that all else being equal, black mothers, less educated mothers, and mothers
with children from another union will be more likely to live alone, while
foreign-born mothers and mothers with educated working partners will be
less likely to live alone.
Dependent variable
Implicit in modeling living arrangements as a mother’s choice is the fact
that marriage, cohabitation, living with a roommate, and living with family
are all joint rather than individual decisions. It is important to consider how
modeling living arrangements as a mother’s decision could be problematic,
particularly when considering marriage and cohabitation versus living alone.
This concern can be modified by two important points. This unique sample
of unmarried mothers, practically and theoretically, enters negotiations over
living arrangements with more to lose in terms of benefits than the similarly
situated men they may be considering marrying or cohabiting with. Low-
income, able-bodied, noncustodial fathers are eligible for very few public
benefits and are more likely to have inconsistent wages. Moving in with a
low-income, unmarried mother does not change a man’s eligibility status but
most definitely will affect a woman’s eligibility for benefits.
Living arrangements are divided into five mutually exhaustive, exclusive
categories by using the household roster and direct questions about marital
and cohabitation status at the three-year survey. These categories are as
follows:
1. Living alone: About 32 percent live alone with children, and no other
adult is present.
5
2. Married: About 16 percent are married to the father of the most recent
child or another partner (some 14 percent of mothers are married to
new partners).
6
5
Children include stepchildren, foster children, unrelated children, adopted children, and
grandchildren, as well as nieces, nephews, or siblings who are 15 or younger.
6
Some 2 percent of married mothers note other adults in the household.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 155
3. Cohabiting: About 32 percent live with the father of the most recent
child or another partner (some 19 percent of mothers are living with
new partners).
7
4. Living with a roommate: About 2 percent live with a roommate, and
no other adult is present.
5. Living with family: About 17 percent live with extended family, and
no spouse or partner is present.
Data
The FFCW Study offers a unique opportunity to examine the impact of
housing subsidies and prices on a sample of unmarried mothers three years
after a nonmarital birth. The impact of public benefits on this group is of
particular policy importance in an ongoing climate of welfare reform and
reauthorization. By combining important personal characteristics about
mothers and their partners, as well as detailed household roster information,
this analysis can focus on housing subsidies and prices in a refined manner.
These longitudinal data allow for a look at changes in living arrangements
over time rather than a cross-sectional snapshot.
The FFCW Study is a national survey that provides longitudinal infor
-
mation about a birth cohort of 3,712 children born to unmarried parents,
as well as a comparison group of 1,186 children born to married parents, in
75 hospitals in 20 U.S. cities with a population of 200,000 or more. When
weighted, this survey is representative of all unwed births in large cities.
8
Parents were interviewed in the hospital shortly after the child’s birth and
approximately one year and three years later. The next follow-up interview
is planned to take place when the child is about five years old. Baseline inter
-
views were done for 2 of the cities in 1998, for 5 of the cities in 1999, and
for 13 of the cities in 2000.
9
The sample consists of all mothers who were unmarried at baseline and
live with the focal child at baseline and three years later. The household
7
Some 4 percent of cohabiting mothers note other adults in the household.
8
The cities were Austin, TX; Baltimore, MD; Boston, MA; Chicago, IL; Corpus Christi,
TX; Indianapolis, IN; Jacksonville, FL; Nashville, TN; New York, NY; Norfolk, VA; Phila
-
delphia, PA; Pittsburgh, PA; Richmond, VA; San Antonio, TX; San Jose, CA; Toledo, OH;
Detroit, MI; Milwaukee, WI; Newark, NJ; and Oakland, CA.
9
Baseline interviews for Boston (MA), Chicago (IL), Corpus Christi (TX), Indianapo-
lis (IN), Jacksonville (FL), Milwaukee (WI), Nashville (TN), New York (NY), Norfolk (VA),
Pittsburgh (PA), San Antonio (TX), San Jose (CA), and Toledo (OH) occurred in 2000; those
for Baltimore (MD), Detroit (MI), Newark (NJ), Philadelphia (PA), and Richmond (VA) took
place in 1999; and those for Austin (TX) and Oakland (CA) were done in 1998.

FANNIE MAE FOUNDATION
156 Marah A. Curtis
roster, marriage, or cohabitation status must be observed at baseline and
three years. Of the 3,712 unmarried births at baseline, 3,067 mothers can
be observed at both points in time. Of the 645 remaining cases, 533 were
not interviewed at year three, 75 did not live with the focal child either at
baseline or at year three, and 37 did not have enough information on their
household roster to determine living arrangements. Cases lost to attrition
were significantly more likely to be Hispanic, foreign born, unemployed, and
less educated. These conclusions hold for the 112 cases dropped because of
lack of household roster information or because the mother did not live with
the child. Information from mothers’ baseline, one-year, and three-year inter
-
views is used to construct covariates. Information from fathers’ one-year and
three-year interviews is used to supplement missing data on age, education,
fertility, work status, and history of incarceration.
10
The FFCW data contain several weaknesses. Since there are only 20 cit-
ies, there is limited variation in housing subsidies and prices. There is also a
relatively short three-year period to observe changes in living arrangements.
It is impossible to control for both city- and state-level policy measures and
fixed effects, so particular norms or city-level characteristics may be uncon
-
trolled for in this analysis and may impact the living arrangements of unmar
-
ried mothers. Future research on the effects of subsidized housing, prices,
and living arrangements should include more cities, as well as fixed effects,
to rule out this possibility.
This analysis focuses on unmarried mothers exclusively and estimates
their living arrangements conditional on having a child. Mothers who are
more inclined to marry may have done so before the birth and are not a part
of this sample. Nearly half of the FFCW mothers have two or more children
with the focal father by the year three survey. However, it is reasonable to
suspect that a number of mothers will not marry these fathers because the
union has soured. These women then become less attractive marriage part
-
ners because of the presence of two or more children from a prior union.
Family formation research shows that children from another union reduce
mothers’ attractiveness to a new marriage partner, presumably because of the
cost of spending time and resources raising another man’s children.
These results can be used to think about the impact of housing subsidies
and prices on the living arrangements of a particularly vulnerable popula
-
tion under consistent scrutiny by researchers and policy makers. Including a
10
Respondents with missing data on covariates are retained in all analyses, and the per-
centage of missing data is shown in table 1. In the regressions, a dummy variable is used to
signify that there is missing information on a given covariate.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 157
rich set of personal characteristics along with controls for important public
benefits can lead to clearer thinking about how to adjust policy. There is
good reason for caution, however, particularly with respect to conclusions
about the impact of housing subsidies. First, the models are not causal, so
inferences should be made cautiously. Second, policy recommendations must
be tempered by the fact that any associations between housing subsidies and
living arrangements cannot distinguish between income and price effects.
Specifically, if housing subsidies are found to reduce the relative risk of mar
-
riage versus living alone, it will not be clear whether this is because mothers
choose to use this benefit to live on their own or whether marrying a partner
jeopardizes it. Future research should seek to disentangle why housing subsi
-
dies may be associated with living arrangements.
Results
Descriptive information
Table 1 presents descriptive information about the characteristics of moth-
ers used in this analysis. More than half of the sample is black, with nearly
a quarter Hispanic and about 15 percent white. On the whole, mothers are
a younger, more highly educated group than their male partners. The over
-
whelming majority of mothers are native born, and more than three-quarters
were working in the previous year. A bit more than half have one child, and
43 percent have children with another partner. Nearly three-quarters of male
partners are working; almost half have children with another partner; and 44
percent have a history of incarceration. This sample represents the complex
-
ity and challenges facing an urban sample of unmarried parents.
Table 2 presents changes in living arrangements between the baseline
and three-year surveys to get a sense of variation in arrangements over time.
Nearly a third of mothers are living alone, compared with about a fifth at
baseline, and 16 percent have moved into marriages, while the percentage of
those cohabiting has been reduced from nearly half to a third. Certainly some
cohabitations are converted to marriages and living alone while living with a
roommate or family decreases over time by about 7 percentage points. Moth
-
ers’ living arrangements prove dynamic and changeable over time.
Table 3 displays the range in the availability of housing subsidies, hous
-
ing prices, and other policy variables in the 20 FFCW cities. Housing prices
vary quite widely across cities; the minimum cost for a comparable single-
family, owner-occupied house is $73,208 in Corpus Christi, TX, while the
maximum is $422,735 in San Jose, CA. The availability of subsidized hous
-
ing varies by city: Austin (TX) has the least availability, while Pittsburgh has

FANNIE MAE FOUNDATION
158 Marah A. Curtis
Race/Ethnicity
a
White, non-Hispanic 15
Black, non-Hispanic 55
Hispanic 27
Other race/ethnicity 3
Parent/Current partner is of a different race/ethnicity
a
13
Mother’s age
Under 20 2
20 to 24 41
25 to 29 31
30 to 34 14
35 or older 12
Mean age 27 (5.6)
Father’s/Current partner’s age
a
Under 23 12
23 to 27 35
28 to 32 26
33 to 37 13
38 or older 13
Mean age 29 (7.0)
Nativity
a
Mother is foreign born 12
Mother’s education
a
Less than high school 32
High school diploma 36
Some college 27
College degree or more 5
Father’s/Current partner’s education
c
Less than high school 28
High school diploma 47
Some college 17
College degree or more 5
Mother worked last year 77
Father/Current partner working 70
Limiting health problem
a
10
Mother has children with another partner 43
Father/Current partner has other children
b
46
Children with the focal father
1 child 54
2 children 30
3 or more children 16
Father/Current partner has incarceration history
b
44
Mother does not know incarceration status 5
Father is in jail 7
Table 1. Sample Description of 3,067 Subjects (in Percents)
Note: Totals may not equal 100 percent because of rounding. The numbers in parentheses are standard deviations.
a
< 1% missing data
b
< 2% missing data
c
< 3% missing data

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 159
eight times more. Maximum welfare grants plus food stamp benefits also
vary widely, with a $520 benefit in Tennessee and a $1,153 benefit in Cali
-
fornia. Unemployment rates also show considerable range across cities, from
a low of 2.0 percent in San Jose to 6.3 percent in Corpus Christi. Child sup
-
port enforcement ranges from 7 percent in Texas to 45 percent in Wisconsin.
All in all, housing subsidies, prices, and policy controls offer considerable
variation across cities.
Multinomial logistic regression (MLR)
MLR is used to examine the impact of housing subsidies, prices, and
other covariates on mothers’ living arrangements.
11
MLR is appropriate to
use for nominal outcome variables with more than two categories—living
arrangements are divided into five categories. MLR simultaneously estimates
a series of binary regressions, comparing each category with the reference
Table 2. Changes in Living Arrangements
FFCW Baseline Wave 3
Number of Subjects 3,067 3,067
Living arrangements (%)
Mother, child, and spouse NA 16
Cohabitation 48 32
Mother and child alone 19 32
Roommate 9 2
Extended family 24 17
Note: Totals may not equal 100 percent because of rounding.
Table 3. Policy Measures in 20 FFCW Cities
Mean Minimum Maximum
Housing prices
a
$163,665 $73,208 $422,735
Average of subsidized housing 6% 2% 16%
Maximum welfare plus food stamp grant
a
$751 $520 $1,153
Unemployment rate 3.70% 2.00% 6.30%
Child support enforcement rate 24% 7% 45%
a
For a family of four, measured in constant 2000 dollars.
11
MLR assumes that the independence of irrelevant alternatives (IIA) would be met. IIA
assumes that deleting or adding alternative outcome categories will not affect the odds among
the remaining outcomes. Hausman tests failed to reject the IIA assumption. The model does
not violate the IIA assumption.

FANNIE MAE FOUNDATION
160 Marah A. Curtis
category. The probability for each category is obtained, and the system of
equations is identified by setting one of the coefficients to 0 and deriving the
log odds of being in each category relative to living alone. The logs odds are
as follows:
P(y = marriage)/P(y = alone) = exp(
b2*x) (1)
P(y = cohab)/P(y = alone) = exp(
b3*x)
P(y = room)/P(y = alone) = exp(
b4*x)
P(y = family)/P(y = alone) = exp(
b5*x)
The coefficients
b2, b3, b4, and b5 represent the log odds of being in
each of those groups (married, cohabiting, roommate, family) relative to the
reference group (living alone). The coefficients can be transformed and inter
-
preted as relative risk ratios (abbreviated as RRR in the equation). The rela
-
tive risk ratio for a multinomial logit is
P(y = 1
|
x + 1)/P(y = base category
|
x + 1) (2)
RRR =
___________________________________
P(y = 1
|
x)/P(y = base category
|
x)
Results in table 4 are relative risk ratios with robust Z statistics in paren
-
theses. The relative risk ratio measures the risk of being in a given living
arrangement relative to living alone for each independent variable. Relative
risk ratios that are less than 1.00 represent a reduced risk of being in a given
arrangement, while those that are greater than 1.00 represent an increased
risk.
12
Multivariate results
Table 4 provides a multivariate analysis of the correlations between the
independent variables and changes in living arrangements after controlling
for all other differences. To control for unobservable factors that may be
associated with living arrangements, all analyses control for mothers’ base
-
line living arrangements.
13
Housing prices and subsidies make a difference
in the direction predicted a priori. Increasing housing prices are associated
with shared arrangements, while increased subsidies decrease the risk of mar
-
riage relative to living alone. There are no significant differences associated
with any other shared arrangements. For a bit of context, the mean housing
price across cities is $163,665, so a $20,000 increase in prices is associated
12
All analyses use robust standard errors to adjust for the clustering of women in cities.
13
It is possible that personal preferences or individual factors that are not controlled for
may affect living arrangements. Controlling for mothers’ baseline living arrangements adjusts
for individual factors that may be related to living arrangements.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 161
Table 4. Multinomial Analysis of Living Arrangements
Living Arrangement
a
Covariate Married Cohabiting Roommate Family
Housing and policy indicators
Housing prices 1.06** 1.08** 0.99 1.11**
(2.79) (4.22) (0.14) (4.93)
Availability of subsidized housing 0.69* 0.87 0.76 1.11
(2.35) (1.06) (0.80) (0.73)
Maximum welfare plus food stamp 0.88* 0.85** 1.14 0.78**
grant (family of 4) (1.89) (2.77) (0.90) (3.75)
Unemployment rate (Reference = 3.01 to 5.00)
Unemployment rate < = 3.00% 0.83 0.66** 1.16 0.52**
(1.01) (2.79) (0.35) (3.82)
Unemployment rate >5.00% 0.61* 0.85 1.17 0.66*
(2.16) (0.86) (0.37) (1.90)
Child support effectiveness 0.83** 0.92 1.05 0.95
(2.78) (1.64) (0.40) (0.79)
Personal characteristics
Race/Ethnicity (Reference = white)
Black, non-Hispanic 0.28** 0.53** 0.42* 0.64*
(6.99) (4.09) (2.12) (2.34)
Hispanic 0.79 0.81 1.33 1.00
(1.10) (1.08) (0.62) (0.01)
Other race/ethnicity 0.52* 0.39** 0.49 0.47*
(1.66) (2.67) (0.82) (1.85)
Parents/Current partner of a 1.19 1.12 0.98 1.05
different race/ethnicity (0.92) (0.72) (0.06) (0.26)
Mother’s age (Reference = 20 to 24)
Mother is 16 to 19 0.93 0.96 2.07 0.81
(0.11) (0.09) (0.97) (0.49)
Mother is 25 to 29 0.80 0.78* 0.88 0.87
(1.37) (1.87) (0.39) (0.90)
Mother is 30 to 34 0.60* 0.54** 0.28* 0.46*
(2.33) (3.57) (2.06) (3.43)
Mother is 35 or older 0.58* 0.55** 0.80 0.47**
(2.04) (2.82) (0.46) (2.89)
Father’s/Current partner’s age (Reference = 23 to 27)
Father/Current partner is 17 to 22 0.82 0.77 0.59 1.08
(0.87) (1.47) (1.18) (0.41)
Father/Current partner is 28 to 32 1.03 0.94 1.22 0.96
(0.20) (0.43) (0.64) (0.28)
Father/Current partner is 33 to 37 1.63* 1.24 1.33 0.98
(2.32) (1.25) (0.65) (0.07)
Father/Current partner is 38 or older 0.86 0.86 1.59 1.19
(0.58) (0.75) (0.92) (0.71)
Mother is foreign born 3.35** 2.91** 2.41* 2.05**
(4.83) (4.84) (2.28) (2.81)

FANNIE MAE FOUNDATION
162 Marah A. Curtis
Table 4. Multinomial Analysis of Living Arrangements Continued
Living Arrangement
a
Covariate Married Cohabiting Roommate Family
Mother’s education (Reference = high school diploma)
Less than 12 years 0.75* 0.94 1.49 1.00
(1.78) (0.53) (1.36) (0.03)
Some college 1.30* 0.90 0.88 1.05
(1.67) (0.84) (0.36) (0.30)
College degree or more 2.08* 0.87 1.14 1.70*
(2.53) (0.50) (0.18) (1.88)
Father’s/Current partner’s education
(Reference = high school diploma)
Less than 12 years 1.24 1.42** 0.91 1.28*
(1.36) (2.82) (0.32) (1.68)
Some college 0.93 1.00 0.53 0.92
(0.44) (0.02) (1.51) (0.50)
College degree or more 1.00 0.46** 1.10 0.84
(0.01) (2.84) (0.16) (0.63)
Mother worked last year 0.77* 0.69** 1.05 0.86
(1.66) (3.01) (0.16) (0.97)
Father/Current partner working 1.57** 1.40** 1.24 0.99
(2.71) (2.70) (0.66) (0.07)
Mother has a limiting disability 0.89 0.79 0.66 0.97
(0.54) (1.41) (0.90) (0.16)
Mother has children with another partner 0.79* 0.66** 0.43** 0.69**
(1.69) (3.67) (2.68) (2.81)
Father/Current partner has other children 1.01 1.53** 1.25 1.25*
(0.04) (3.92) (0.84) (1.78)
Children with the focal father (Reference = 1 child)
2 children 1.78** 1.47** 0.76 0.72*
(3.96) (3.36) (0.89) (2.30)
3 or more children 1.81** 1.42** 0.46 0.59**
(3.24) (2.40) (1.59) (2.60)
Father/Current partner has been incarcerated 0.52** 0.63** 1.15 0.91
(4.63) (4.21) (0.47) (0.68)
Mother does not know incarceration status 0.33** 0.48** 1.02 0.57*
(3.12) (3.08) (0.05) (2.11)
Father is in jail 0.48* 0.48** 1.33 0.89
(2.12) (3.22) (0.61) (0.52)
Baseline control for living arrangements Included Included Included Included
Observations 3,067
Log pseudo likelihood –3694.80
Pseudo R
2
0.15
Note: Standard errors in parentheses are corrected for clustering at the city level.
a
Alone is the reference category.
*p < 0.10. **p < 0.05.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 163
with a 6 percent to 11 percent increase in the odds that a mother will marry,
cohabit, or live with extended family rather than live alone. To give a sense
of the magnitude of the housing subsidy result, an increase of about 14,000
units in a city of nearly 3,000,000 decreases the odds of marrying by 31
percent relative to living alone. The only significant difference in the impact
of housing subsidies on living arrangements is in the comparison between
living alone and marrying. This suggests that income-conditioning may be
an important part of the story.
Welfare grants reduce the risk of living in most shared arrangements
relative to living alone; the impact on marriage, however, is marginally sig
-
nificant. The relative risk of living with family is reduced by 22 percent, sug
-
gesting that benefits allow mothers to seek out arrangements that might best
suit their families. The impact of unemployment rates is instructive, since it
affects shared arrangements and marriage specifically. Residing in a city with
the lowest unemployment rates is associated with a significant decrease in the
risk of cohabiting or living with family. When economic prospects are good
for these mothers, they choose to maintain their own households relative
to certain shared arrangements, and marriage is unaffected. Residing in an
area with high unemployment rates, however, is associated with a 40 percent
decrease in the odds of marrying relative to living alone.
These results confirm previous research noting the negative effect of high
unemployment on marriage. Poor labor market prospects may make mar
-
riage a less viable option. The final policy control, the state child support col
-
lection rate, proves significant. Increased collections among the population
that has received welfare are negatively associated with the probability of
marriage relative to living alone.
14
Overall, these results confirm the impor-
tance of controlling for the array of benefits and economic conditions likely
affecting unmarried mothers as they make decisions about their lives. Focus
-
ing on housing subsidies and prices proves important.
Finally, demographic controls evidence patterns similar to those in previ
-
ous research and, generally, confirm a priori expectations. Black mothers,
older mothers, and working mothers are more likely to live alone than in
shared arrangements, while college-educated and foreign-born mothers are
more likely to marry than to live alone. With current policy aimed at encourag-
ing or facilitating marriage among this population, attention to the impact of
male partners’ characteristics is instructive. Mothers with less educated part
-
ners have increased odds of cohabiting rather than living alone, while having
14
Results using city-level measures are available from the author. Overall, these results
are quite similar.

FANNIE MAE FOUNDATION
164 Marah A. Curtis
a working partner increases the odds of marriage and cohabitation rather
than living alone.
15
Education and working status matter in these decisions,
most likely as a proxy for a partner’s earnings potential. As expected, having
more children with a partner increases both marriage and cohabitation rela
-
tive to living alone. Finally, a partner with a history of incarceration mark
-
edly reduces the odds of marriage and cohabitation relative to living alone.
Does excluding housing prices and subsidies matter?
It does matter when housing prices and subsidies are omitted from
analyses of living arrangements. Table 5 compares the results when hous
-
ing prices and subsidies are included with results when they are omitted.
Again, results are relative risk ratios with robust Z statistics in parenthe
-
ses. When housing measures are omitted, the effects of welfare and food
stamp benefits on the relative risk of cohabitation and living with family
are underestimated. There are no significant associations between welfare
benefits and living arrangements when housing measures are excluded, thus
suggesting that they are not particularly important. When housing measures
are included, however, the relative risk of cohabiting is reduced by 15 percent
and that of living with family drops by 22 percent. This suggests that previ
-
ous research may underestimate the association between welfare benefits and
living arrangements.
Excluding housing measures from the analysis also underestimates the
importance of unemployment rates. Omitting housing prices and subsidies
may lead policy makers to the conclusion that policies aimed at employment
are not particularly important when, in fact, they are. When housing mea
-
sures are included, robust labor markets are associated with sizable reduc
-
tions (from 47 to 33 percent) in the relative risk of all shared arrangements.
When opportunities to earn are strong, these mothers opt to set up their own
households. Conversely, and no less important, high unemployment rates
significantly reduce the risk of marriage by 40 percent relative to living alone.
Marriage is not on the agenda when unemployment rates are highest. These
results confirm the centrality of labor market opportunities and their impact
on the living arrangements of unmarried mothers.
15
In other results that are not included, mothers’ working status is excluded from the
analysis to test for endogeneity between living arrangements and working status. The results
in this analysis are very similar. In results excluding the current partners’ working status, the
magnitude of the relative risk ratios on the incarceration variables is larger, thereby suggesting
that a portion of the incarceration effect is through earnings. I choose to include current part
-
ners’ working status, since its inclusion does not change the magnitude of the housing variables
while it avoids inflating the importance of incarceration on living arrangements.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 165
Table 5. Results of Analysis of Living Arrangements Omitting Housing Prices and Subsidies (3,067 Observations)
Living Arrangement
a
Covariate Married Cohabiting Roommate Family
Housing measures included Yes No Yes No Yes No Yes No
Maximum welfare plus food stamp grant (family of 4) 0.88* 0.99 0.85** 1.01 1.14 1.11 0.78** 0.98
(1.89) (0.11) (2.77) (0.15) (0.90) (1.01) (3.75) (0.44)
Unemployment rate (Reference = 3.01 to 5.00)
Unemployment rate < = 3.00% 0.83 0.91 0.66** 0.79* 1.16 0.89 0.52** 0.74*
(1.01) (0.62) (2.79) (1.87) (0.35) (0.39) (3.82) (2.05)
Unemployment rate >5.00% 0.61* 0.80 0.85 1.09 1.17 1.09 0.66* 0.89
(2.16) (1.07) (0.86) (0.46) (0.37) (0.23) (1.90) (0.55)
Child support enforcement 0.83** 0.81** 0.92 0.92* 1.05 0.98 0.95 0.97
(2.78) (3.17) (1.64) (1.69) (0.40) (0.13) (0.79) (0.57)
Baseline control for living arrangements Included Included Included Included
Log pseudo likelihood
With housing measures –3694.80
Without housing measures –3714.87
Pseudo
R
2
With housing measures 0.15
Without housing measures 0.14
Note: All covariates are included in the above regressions. Numbers represent relative risk ratios with robust Z statistics in parentheses. Standard errors in parentheses are corrected for
clustering at the city level.
a
Alone is the reference category.
*p < 0.10. **p < 0.05.

FANNIE MAE FOUNDATION
166 Marah A. Curtis
Conclusions and policy implications
The FFCW mothers, unmarried at the birth of the focal child, represent
a group that is at high risk for poverty and that has an increased likelihood
of requiring public support. Although the availability of affordable housing
is limited in many cities and housing subsidies are a scarce benefit, very little
research has focused on either. This analysis concentrates on the impact of
housing prices and subsidies while controlling for an array of policy and per
-
sonal characteristics known to be important in living arrangements. Policies
targeting poor unmarried mothers need to take into account the complex
external factors affecting these families. Focusing narrowly on a single ben
-
efit policy or failing to account for housing prices and local area employment
conditions can lead to incorrect conclusions about what is needed.
Findings suggest that means testing in public housing affects marriage
but not other shared arrangements. Further research is required to figure out
precisely how much of this impact is about mothers choosing to set up inde
-
pendently because they can do so financially or because marriage represents
a huge cost in lost benefits. Regardless of the precise answer to this ques
-
tion, it is clear that rules affect living arrangements, although it is not clear
whether this is an intentional policy goal. Housing analysts should consider
the explicit goal—providing a small proportion of the low-income popula
-
tion with stable housing—as well as the unintended effects of policy. Encour
-
aging two-parent families in subsidized housing may mean raising income
guidelines or instituting earnings disregards to minimize the possibility that
stable housing will be lost if another earner enters the household.
Housing prices and subsidies are not only important in their own right,
but their exclusion understates the importance of welfare benefits on living
arrangements. Assuming that unmarried mothers are aware of the safest,
most productive environments for themselves and their children, benefits
allow these women to exercise the best option. If housing prices and sub
-
sidies are not considered, welfare grants seem unimportant. Of particular
concern is the finding that omitting housing prices and subsidies leads to
incorrect assumptions about the association between unemployment rates
and living arrangements. Again, strong labor markets allow mothers to exer
-
cise their judgment and choose the best arrangement for their families. Mar
-
riage is unaffected, while cohabitation and family arrangements are reduced.
Findings also suggest that high unemployment rates certainly do nothing to
promote marriage among this population, although promoting marriages is
an underlying principle of welfare reform. Attention to labor market policies
aimed at reducing unemployment or increasing mothers’ human capital may
encourage family formation in a very direct way.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 167
Author
Marah A. Curtis is an Assistant Professor in the School of Social Work at Boston
University.
The author acknowledges the very helpful comments of Irwin Garfinkel, Jane Wald-
fogel, Brendan O’Flaherty, and Julien Teitler, as well as the Fragile Families Working
Group Seminar participants at Columbia University and Princeton University. The author
also acknowledges the many helpful suggestions by the editor and two anonymous review
-
ers. Robert W. Gray provided invaluable and ongoing assistance with U.S. Department of
Housing and Urban Development data sets.
References
Ashenfelter, Orley. 1983. Determining Participation in Income-Tested Social Programs.
Journal of the American Statistical Association 78:517–25.
Becker, Gary S. 1981. A Treatise on the Family. Cambridge, MA: Harvard University
Press.
Belsky, Eric S., Jack Goodman, and Rachel Drew. 2005. Measuring the Nation’s Rental
Housing Affordability Problems. Cambridge, MA: Harvard University, Joint Center for
Housing Studies.
Bennett, Neil G., David E. Bloom, and Patricia H. Craig. 1989. The Divergence of Black
and White Marriage Patterns. American Journal of Sociology 95:692–722.
Bergstrom, Theodore C. 1996. Economics in a Family Way. Journal of Economic Litera
-
ture 34(4):1903–34.
Blau, Francine D., Lawrence M. Kahn, and Jane Waldfogel. 2004. The Impact of Welfare
Benefits on Single Motherhood and Headship of Young Women: Evidence from the Cen
-
sus. Journal of Human Resources
39(2):382–404.
Borsch-Supan, Axel. 1986. Household Formation, Housing Prices, and Public Policy
Impacts. Journal of Public Economics 30:145–64.
Bureau of Labor Statistics. 2002. Housing Expenditures. Washington, DC: U.S. Depart
-
ment of Labor.
Bureau of Labor Statistics. n.d. Local Area Unemployment Statistics. World Wide Web
page <http://www.bls.gov/lau> (accessed October 2004).
Carlson, Marcia, Irwin Garfinkel, Sara McLanahan, Ronald B. Mincy, and Wendell Pri
-
mus. 2005. The Effects of Welfare and Child Support Policies on Union Formation.
Popu-
lation Research and Policy Review 23(5–6):513–42.
Carlson, Marcia, Sara McLanahan, and Paula England. 2004. Union Formation in Fragile
Families.
Demography 41(2):237–61.
Curtis, Marah A. 2005. Public Policy, Economic Conditions and the Living Arrangements
of Mothers. Ph.D. diss. Columbia University, School of Social Work.

FANNIE MAE FOUNDATION
168 Marah A. Curtis
Danziger, Sheldon, George Jakubson, Saul Schwartz, and Eugene Smolensky. 1982. Work
and Welfare as Determinants of Female Poverty and Female Headship. Quarterly Journal
of Economics 97:519–34.
Duncan, Greg J., and Saul D. Hoffman. 1990. Welfare Benefits, Economic Opportunities,
and Out-of-Wedlock Births among Black Teenage Girls.
Demography 27:519–36.
Ellen, Ingrid Gould, and Brendan O’Flaherty. 2006. Government Policies and House
-
hold Size: Evidence from New York. Population Research and Policy Review 26(4),
forthcoming.
Ellwood, David T., and Mary Jo Bane. 1985. The Impact of AFDC on Family Structure
and Living Arrangements. In Research in Labor Economics, ed. Ronald G. Ehrenberg,
137–207. Greenwich, CT: JAI.
Freeman, Lance. 2005. Household Composition and Housing Assistance: Examining the
Link. Cityscape: A Journal of Policy Development and Research
8(2):49–67.
Harkness, Joseph M., and Sandra J. Newman. 2006. Recipients of Housing Assistance
under Welfare Reform: Trends in Employment and Welfare Participation. Housing Policy
Debate 17(1):81–108.
Hartman, Chester. 1998. The Case for a Right to Housing. Housing Policy Debate 9(2):
223–46.
Haurin, Donald R., Patric H. Hendershott, and Dongwook Kim. 1993. The Impact of
Real Rents and Wages on Household Formation. Review of Economics and Statistics
75(2):284–93.
Hoynes, Hillary Williamson. 1997. Does Welfare Play Any Role in Female Headship?
Journal of Public Economics 65:89–117.
Hughes, Mary Elizabeth. 2003. Home Economics: Metropolitan Labor and Housing Mar
-
kets and Domestic Arrangements in Young Adulthood. Social Forces 81(4):1399–429.
Lichter, Daniel T., Robert N. Anderson, and Mark D. Hayward. 1995. Marriage Markets
and Marital Choice. Journal of Family Issues 6(4):412–31.
Lichter, Daniel T., Felicia B. LeClere, and Diane K. McLaughlin. 1991. Local Marriage
Markets and the Marital Behavior of Black and White Women. American Journal of
Sociology 96:843–67.
Lichter, Daniel T., Diane K. McLaughlin, George Kephart, and David J. Landry. 1992.
Race and the Retreat from Marriage: A Shortage of Marriageable Men? American Socio
-
logical Review 57(6):781–99.
London, Rebecca A. 2000. The Interaction between Single Mothers’ Living Arrangements
and Welfare Participation. Journal of Policy Analysis and Management
19(1):93–117.
Malpezzi, Stephen, Gregory H. Chun, and Richard K. Green. 1998. New Place-to-Place
Housing Price Indexes for U.S. Metropolitan Areas, and Their Determinants. Real Estate
Economics 26(2):235–74.

HOUSING POLICY DEBATE
Subsidized Housing, Housing Prices, and Unmarried Mothers 169
McLanahan, Sara, and Gary Sandefur. 1994. Growing Up with a Single Parent. Cam-
bridge, MA: Harvard University Press.
Mincy, Ronald B., and Allen T. Dupree. 2001. Welfare, Child Support, and Family For
-
mation. Child and Youth Services Review
23(6/7):577–601.
Moffitt, Robert A. 1992. Incentive Effects of the U.S. Welfare System: A Review. Journal
of Economic Literature 30:1–61.
Moffitt, Robert A. 1994. Welfare Effects on Female Headship with Area Effects. Journal
of Human Resources 29(2):621–36.
Moffitt, Robert A. 2005. Welfare Benefits Database. Johns Hopkins University, Depart
-
ment of Economics. World Wide Web page <http://www.econ.jhu.edu/People/Moffitt/
DataSets.html> (last modified February 2, 2002).
Moffitt, Robert A., Robert Reville, and Ann E. Winkler. 1998. Beyond Single Mothers:
Cohabitation and Marriage in the AFDC Program.
Demography 35(3):259–78.
Mulroy, Elizabeth A., and Patricia L. Ewalt. 1996. Affordable Housing: A Basic Need and
a Social Issue. Social Work 41(3):245–49.
Ong, Paul. 1998. Subsidized Housing and Work among Welfare Recipients. Housing
Policy Debate
9(4):775–94.
Powell, Mary Ann, and Toby L. Parcel. 1997. Effects of Family Structure on the Earn
-
ings Attainment Process: Differences by Gender. Journal of Marriage and the Family
59:419–33.
Ruggles, Steven, Matthew Sobek, Trent Alexander, Catherine A. Fitch, Ronald Goeken,
Patricia Kelly Hall, Miriam King, and Chad Ronnander. 2004. Integrated Public Use
Microdata Series: Version 3.0 [machine-readable database]. Minneapolis, MN: Minne
-
sota Population Center. World Wide Web page <http://www.ipums.org> (accessed Sep
-
tember 30, 2006).
Sigle-Rushton, Wendy, and Sara McLanahan. 2002. The Living Arrangements of New
Unmarried Mothers.
Demography 39(3):415–33.
Smith, Lawrence B., Kenneth T. Rosen, Anil Markandya, and Pierre-Antoine Ullmo. 1984.
The Demand for Housing, Household Headship Rates, and Household Formation: An
International Analysis. Urban Studies
21:407–14.
Thibodeau, Thomas G. 1992. Residential Real Estate Prices 1974–1983: From the Stan
-
dard Metropolitan Statistical Area Annual Surveys. Mount Pleasant, MI: Blackstone.
Turner, Mark. 2003. Cohabitation of Unwed Parents in Federally Subsidized Housing:
Effects of Income and Housing Prices. Unpublished paper. Urban Institute.
U.S. Bureau of the Census. 2000. County and City Data Book: 2000. World Wide Web
page <http://www.census.gov/prod/2002pubs/00ccdb/cc00_tabC1.pdf> (last modified
June 25, 2002).

FANNIE MAE FOUNDATION
170 Marah A. Curtis
U.S. Department of Health and Human Services, Administration for Children and Fami-
lies, Office of Child Support Enforcement. 1998. Table 26: Child Support Enforcement,
Twenty-Third Annual Report to Congress: Average CSE Caseload by TANF/FC, Non-
TANF, and TANF/FC Arrears Only, FY 1998. World Wide Web page <http://www.acf.
hhs.gov/programs/cse/pubs/1999/reports/23rd_annual_report_congress/tables/TABLE26.
htm> (accessed October 2004).
U.S. Department of Housing and Urban Development. 1999. Waiting in Vain: An Update
on America’s Rental Housing Crisis. Washington, DC.
U.S. Department of Housing and Urban Development. 2002. 50th Percentile and 40th
Percentile Fair Market Rents for Fiscal Year 2002. Washington, DC.
U.S. Department of Housing and Urban Development. 2003. Affordable Housing Needs:
A Report to Congress on the Significant Need for Housing. World Wide Web page
<http://www.huduser.org/Publications/pdf/AffHsgNeedsRpt2003.pdf> (accessed October
19, 2006).
U.S. Department of Housing and Urban Development. 2005. A Picture of Subsidized
Households—1998. World Wide Web page <http://www.huduser.org/datasets/assthsg/
statedata98/> (last modified March 31).
U.S. Department of Housing and Urban Development. 2006. Low-Income Housing Tax
Credits. World Wide Web page <http://www.huduser.org/datasets/lihtc.html> (last modi-
fied October 18).
U.S. House Committee on Ways and Means. 2004. 2004 Green Book. Washington, DC:
U.S. Government Printing Office.
Western, Bruce, Leonard M. Lopoo, and Sara McLanahan. 2003. Incarceration and the
Bonds among Parents in Fragile Families. In Imprisoning America: The Social Effects of
Mass Incarceration, ed. Mary Patillo, David F. Weiman, and Bruce Western, 21–45. New
York: Russell Sage Foundation.
Winkler, Ann E. 1992. The Impact of Housing Costs on the Living Arrangements of
Single Mothers. Journal of Urban Economics 32:388–40.