University of Southern California Dissertations and Theses

The impact of mobility and government rental subsidies on the welfare of households and affordability of markets

(USC Thesis Other)

The impact of mobility and government rental subsidies on the welfare of households and affordability of markets

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

The Impact of Mobility and Government Rental Subsidies on the Welfare of Households and
Affordability of Markets

by
Vincent J. Reina

A dissertation submitted to the faculty of The Graduate School of the
University of Southern California
in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

Public Policy and Management
USC Price School of Public Policy
University of Southern California
August 9, 2016

TABLE OF CONTENTS

I. Technical Appendix 93
II. Figures 95
Acknowledgements 1
Chapter 1: Introduction 2
Chapter 2: Preserving Neighborhood Opportunity: Where federal rental subsidies expire 8
I. Introduction 9
II. Prior Research 11
III. Data and Methods 20
IV. Results 24
V. Discussion 29
VI. Figures and Tables 33
Chapter 3: Are they protected? The end of place-based rental subsidies and the welfare of
low-income households
41
I. Introduction 42
II. Program Background 43
III. Theory and Empirical Evidence 44
IV. Data and Methods 47
V. Analysis 50
VI. Discussion 54
VII. Figures and Tables 57
Chapter 4: What happens when a project-based rental subsidy ends? Mobility and
academic outcomes
65
I. Introduction 66
II. Background and Theory 67
III. Data, Sample, and Identification Strategy 69
IV. Methods and Analysis 71
V. Discussion 75
VI. Figures and Tables 77
Chapter 5: Conclusion 82
References 86
Appendix to Chapter 4 92
1
Acknowledgements
There are many people who made completing this dissertation and my doctoral program
possible. First, I would like to acknowledge my dissertation committee. Gary Painter, my Co-
chair, was often the voice of reason. Your balanced perspective and calm demeanor was always
a welcome contrast to mine. You kept me moving along and always provided me with the
support to do so. Raphael Bostic, my Co-chair, gave me just as much crucial feedback with his
facial expressions as his words. Thank you for always pushing me to see the forest from the
trees as I approached this dissertation and research more broadly (despite my normal
inclinations). Marlon Boarnet, thank you for always being accessible, providing me with prompt
and thoughtful feedback on my research and career, and being one of my strongest advocates at
USC and on the job market. And finally Richard Green, thank you for the housing study group,
summer funding, always finding the time to meet with me, and all of the advice and feedback
you gave over the past four years.
Thank you to all of the people not on my committee who took time to help me. Lisa Schweitzer,
your comments on my third chapter and feedback on my research and interests more broadly
have made more of an impact on my research and perspective than you know. Sarah Axeen, you
were my classmate, confidant, proofreader, and the person who answered all of my questions
about coding, models, and even formatting, and for all of that I am forever grateful. Arthur
Acolin, words cannot express how grateful I am for all of your thoughtful and thorough feedback
on my research, presentations, and the many other things you helped with. Danielle Williams,
thank you for being a gracious cube-mate and friend, especially as I came in grumpy and needed
more computer time to complete my dissertation. Thank you to all of the discussants at ACSP,
APPAM, AREUEA, and Urban Affairs. And thank you to Professor Mike Lens, my un-official
committee member and co-author, who is the type of scholar I can only hope to become. Finally,
thank you to Ben Winter, a co-author and friend, without whom I would have never been able to
get some of these data or such an important practitioner perspective.
On a more personal note, thank you to my wife, Jamie, who made many sacrifices so I could
pursue my PhD. You provided limitless support, guidance, and cookies, ensured I celebrated my
successes and made it possible to get over any bumps in the road. Thank you to my parents for
supporting all of my intellectual pursuits and aspirations. A special thank you to Pat and Joe
Drake for all of help in the transition to LA, and your kindness and support throughout my time
here. And finally, thank you to my daughters. Nuala, you gave me much needed perspective and
forced me to put all of my work down and have fun for at least a couple of hours every day. And
Regan, your impending birth made me hunker down and finish all of my revisions, and every coo
and grunt you utter while sleeping in your basinet next to me right now reminds me how lucky I
am.

2

Chapter 1:

Introduction

3
The federal government made a shift in its approach to the provision of affordable
housing in the 1960s from a public development and ownership model, to a private one (Pendall
2000). Since the 1970s over 3.3 million units of privately owned housing were developed
through federal rental subsidy programs (Collinson et al. 2015). Owners of these properties
agreed to develop affordable housing, and maintain it as such for a fixed period of time, in
exchange for the subsidy. There is an abundance of literature that studies the development of
this housing and what it means for cities, neighborhoods, and the welfare of low-income
households. To date, there is no research that focuses on what happens when owners reach the
end of their affordability restriction periods and choose to exit a subsidy program. This
dissertation is the first to study this phenomenon and the implications it has for the households
who live in these properties as well as the broader U.S. housing policy paradigm.

The shift to the private development and ownership of subsidized rental housing
happened at a time when cities were suffering from disinvestment and population loss. Federal
housing programs aim to assist low-income households by increasing the number of affordable
units in rental markets, improving the quality of low-cost housing, and decreasing rent burdens.
Some of these programs also had the goal of promoting investment in housing in cities at a time
when such investments were becoming less common. That historical context is much different
than the current one, where rents are reaching record highs and vacancies are at record lows in
cities across the country. Currently, over 50 percent of households are paying more than 30
percent of their income to rent a unit, thus qualifying them as rent burdened (Joint Center for
Housing Studies 2016). At the same time as the demand for rental units is increasing, private
owners of federally subsidized rental housing are becoming eligible to exit the programs they
entered 20 to 40 years ago.

Going forward, every owner of the 3.3 million privately-owned subsidized units will
reach a point where they can exit a subsidy program, as will all new rental units currently being
developed through federal programs. Current research finds that subsidized housing owners are
reacting to changes in the market, and properties in neighborhoods where values are increasing
are more likely to exit these programs (Reina and Begley 2014). These neighborhoods are also
ones where it is the most difficult for low-income households to afford absent a subsidy. Despite
4
the reality that all current and future subsidize rental properties will be eligible to exit these
programs going forward, relatively little is known about the affects of such exits on tenants,
neighborhoods, and neighborhood accesss.

This dissertation provides a detailed and nuanced understanding of the impact of project-
based subsidy expirations on neighborhood access, the welfare of the low-income households
who live in these properties, and what this means for the private market approach to the
provision of affordable housing more broadly. This research shows that units that left the project-
based Section 8 program were located in neighborhoods that were dramatically improving.
Going forward, units eligible to leave this program are located in the highest opportunity
neighborhoods of all active, new, and expiring subsidized units. Subsidy expirations reduce
neighborhood access for low-income tenants, as they are located in neighborhoods with fewer
private-market affordable units and ones where it is costly to develop new subsidized housing.
Tenants who live in these properties are offered a voucher as protection when an owner exits the
subsidy program, but this voucher does not appear to be an effective safety net. Despite what
market supply and household demand factors should predict, a large share of households never
use the voucher and experience a significant income shock. A small share of households may
benefit from being offered a voucher because they use the voucher and do not receive a financial
loss and move to lower poverty neighborhoods. While this move could be welfare improving for
some, such gains vary based on household characteristics. Finally, households with school-aged
children tend to move after a project-based subsidy ends. This move does not appear to be
associated with worse academic outcomes; and, for some students the move may result in higher
test scores.

The first study in this dissertation analyzes the inflows and outflows of subsidized units.
Specifically, it focuses on the location of units exiting subsidized housing programs and what
this means for access to neighborhood opportunity. One limitation of existing subsidized
housing research is that it often does not acknowledge that units leave the stock. The units that
exit are often those in high opportunity areas, and removing these units from the analysis
confounds the estimated relationship between subsidized housing and neighborhood access.
Existing studies also do not acknowledge that many properties receive multiple forms of federal
5
subsidy. Treating subsidies as mutually exclusive also biases the estimated relationship between
subsidized housing and neighborhood access. This study is the first to combine a national
database of all subsidized housing, including start and end dates and every federal subsidy layer,
with data on units that exited the subsidized housing stock, and detailed neighborhood data.
Using these data we can identify where units are located, entering and exiting the subsidized
housing stock and what this means for access to opportunity neighborhoods.

This analysis presents several key findings. First, units that left the project-based Section
8 program between 2000 and 2010 were located in tracts that offer fewer opportunities, but these
neighborhoods showed the strongest improvements during that period. This reinforces the fact
that private owners are reacting to neighborhood change dynamics when deciding to exit subsidy
programs. Moreover, the project-based Section 8 units that are set to expire between 2011 and
2020 are in particularly high-opportunity neighborhoods, which perform better than all other
subsidized housing programs including vouchers. The only program producing new subsidized
units is the Low Income Housing Tax Credit (LIHTC) program, and these new units are in tracts
similar to those where LIHTC units are currently active, which tend to be lower opportunity
neighborhoods.
An owner exiting an affordable housing program has important welfare implications for
the households who live in these properties. To date, we know nothing about what happens to
low-income households when an owner exits a subsidy program, despite the scale at which it has
occurred and the fact that many more households will be affected by this reality. The second
study in this dissertation begins to answer that question. This chapter is the first to construct a
national census of tenants living in a property when the project-based Section 8 subsidy ended
between 1996 and 2010 and to use that dataset to analyze what happens to these households.
Specifically, these households are offered a voucher as a safety net when the project-based
subsidy ends, and this paper examines whether households use their voucher, and where they use
it to move.
The results show that fewer than 50 percent of households who live in properties where
the project-based Section 8 contract ends actually use their voucher. Those who do not use the
voucher on average forego over $400 per month of rental assistance, or roughly 41 percent of
6
their income. To remain in the same unit, assuming the rent is unchanged, these households
would need to increase their rent payment by over 300 percent, or the equivalent of spending all
of their income on rent, to make up for the loss of the rental subsidy. Over half of those
households who use their voucher move one or more times, and on the whole these moves are
associated with lower poverty neighborhoods. Households with the highest demand for the
subsidy, and those where the head is Black or 62 or older, are particularly negatively affected by
this event. In aggregate this paper highlights a concerning reality, which is that for a large share
of households this event presents a serious income shock because they do not use the voucher
despite having a high level of demand for it. However, for the small fraction of households in
these properties who use their vouchers, this event could be a welfare-improving event as they
are able to use a voucher to move to a lower poverty neighborhood.
The third study in this dissertation further explores the welfare implications of subsidy
expirations by analyzing its effect, and the resulting move, on the academic outcomes of children
who live in these properties. Whereas the other two analyses are national, this study focuses on
Los Angeles County. In Los Angeles County the project-based Section 8 contracts expired for
171 properties containing more than 6,000 units. This paper constructs a database of all 1,500
federally subsidized properties ever built in Los Angeles County, including those where a rental
subsidy ended. Next, we clean and geo-code the 1.7 million unique addresses of every student
who attended a Los Angeles Unified School District (LAUSD) school between 2000 and 2014
and determine whether students lived in any form of subsidized rental housing. This paper is the
first to use the identification strategy of an owner leaving a subsidy program to examine the
plausibly causal relationship between mobility and academic outcomes. This strategy helps
eliminate concerns about selection bias and unobserved variables that could drive or affect a
move decision.

The models employed in this paper show a significant relationship between a subsidy
contract ending and a student moving, but no negative academic outcomes associated with this
move. There is no evidence of a relationship between the move resulting from a Section 8
contract ending and an increase in the share of days absent nor the likelihood of a student being
suspended. In fact, these models show a marginally significant relationship between the move
7
induced by a Section 8 contract ending and improved test scores. Combined, this study presents
a more nuanced picture of the effect of contract expirations on tenants in these properties.
Despite inducing a move, expirations do not result in poorer academic outcomes for students
who live in these properties; and, for some, expirations may actually result in better outcomes.

The subsequent portion of this dissertation is comprised of the three studies discussed,
followed by a conclusion that places the findings within the broader context of housing policy
and future research.

8

Chapter 2:

Preserving Neighborhood Opportunity: Where federal rental
subsidies expire

9
I. Introduction
Housing comprises a growing share of household costs, particularly for low-income
renters. Tepid wage growth and increased demand for rental housing has contributed to this
dynamic, but the expiration of subsidies on privately owned, publicly subsidized rental properties
is also reducing the stock of affordable housing nationwide (Schwartz et al. 2016). Thousands of
additional units of subsidized housing are nearing the end of affordability restriction periods.
Evidence suggests that property owners in neighborhoods with high price appreciation have
higher odds of not renewing their subsidies, otherwise known as opting out (Reina and Begley
2014). This means that some of the few remaining affordable units in many neighborhoods are at
risk of being converted to market rate. As there is a strong connection between neighborhood
attributes and property values (Harris 1999; Jud and Watts 1981; Linden and Rockoff 2008)
these are potentially the same neighborhoods that offer the most opportunity.

Efforts to preserve rental subsidies can be costly because the government is competing
with the private market. However, such efforts may be justified if these properties are located in
areas with greater neighborhood opportunity, as low-income tenants who lose their subsidies
may also be likely to lose access to housing in areas of greater opportunity. This is particularly
true given the federal government’s renewed commitment to fair housing and neighborhood
opportunity, as exemplified by last year’s Affirmatively Furthering Fair Housing rule (U.S.
Department of Housing and Urban Development 2015). In this paper, we look at the
characteristics of neighborhoods where property owners are exiting rental subsidy programs and
those where new project-based subsidized housing is being developed to learn how the inflows
and outflows of units from the subsidized rental stock affect neighborhood access for low-
income households. We use a combination of neighborhood attributes examined in existing
research (McClure 2011; Turner et al. 2011) including: poverty rates, educational attainment,
employment rates, employment accessibility, and transportation costs. We also add variables on
school quality and crime that are often difficult to obtain on a national level. Schools, crime and
employment data are used in recent analyses on the neighborhood attributes of housing subsidy
recipients (Horn et al 2014; Lens et al 2011; Lens 2014) but researchers have yet to examine
subsidy expirations in the context of neighborhood opportunity.

10
The research seeks to identify the extent to which the end of a rental housing subsidy
affects low-income renters’ access to opportunity neighborhoods. Further, we identify the
specific neighborhood attributes—school quality, poverty, crime, low transportation costs, etc.—
to which subsidized renters may lose access when subsidies expire. Our analysis presents several
key findings. First, we find that units that exited the project-based Section 8 program
1
were on
the whole located in low opportunity neighborhoods, but that these neighborhoods showed the
strongest improvements from 2000 to 2010. In addition, the project-based Section 8 units that are
set to expire during the 2010s are in particularly high-opportunity neighborhoods. On the
contrary, new Low-Income Housing Tax Credit (LIHTC) units were developed in tracts similar
to those where LIHTC units are currently active, which tend to be lower opportunity
neighborhoods.

We conclude that a sincere commitment to fair housing and opportunity would require
the decision about where to build or preserve affordable units to account for many neighborhood
characteristics and how these characteristics change over time. Further, it is clear that expirations
on the horizon need to be monitored on an ongoing basis; it is not as simple as a particular
subsidy type being located in higher or lower opportunity neighborhoods, considering that
neighborhoods with weak but improving indicators may be the precise areas where landowners
are looking to convert to market rate. For this reason, we suggest that policymakers target
resources to counter the effects of expiring subsidies on low-income tenants in those areas. Our
analysis shows a higher share of expired project-based Section 8 properties falling into this
category than any other subsidy, including the voucher program.

Additionally, we acknowledge that preserving these existing subsidies is not the only way
to enhance neighborhood opportunity for low-income households. We note that, on average,
voucher households were located in neighborhoods with better opportunity than Active Section 8

1
In this study we focus solely on units built through the Section 8 New Construction and Substantial Rehab
programs, which does not include the whole universe of properties with a Project-based Section 8 subsidy.
Therefore, in this paper project-based Section 8 does not include units financed outside of that program, such as the
Section 8 Moderate Rehab program or Loan Management Set Aside program, but it does include properties
developed through other U.S. Department of Housing and Urban Development program, such as the Section 202
program, that also received a project-based contract through the Section 8 New Construction and Substantial Rehab
programs.

11
and LIHTC properties in 2010, but that story is more nuanced. First, both Active and Expired
Section 8 properties were in neighborhoods that were on a better trajectory than those where
vouchers were located, and units eligible to leave the project-based Section 8 program in the
future are in higher opportunity neighborhoods than the average voucher unit. In addition,
evidence suggests that vouchers do not provide a sufficient safety net for households in
properties where a project-based Section 8 subsidy expires, particularly for those in
neighborhoods improving at a much more rapid rate than that of the average voucher household.
Ultimately, our findings function within a context of many rental markets that are, on the whole,
becoming increasingly unaffordable and the location of these units is only one factor to use to
evaluate the benefits of these programs amidst many others, including the efficiency of the
subsidy and whether these programs actually relieve a household’s rent burden.

II. Prior Research

The goal of this paper is to identify where subsidized units are being developed and
expiring and implications for the distribution of subsidized housing across neighborhoods of
varying levels of opportunity. There is a robust literature that describes the neighborhoods
occupied by subsidized households, but there is no previous research connecting neighborhood
opportunity to the loss of subsidy due to expirations and property owner opt-outs, with the
exception of a recent report (Ellen and Weselcouch 2015). At present, research on housing
subsidies and neighborhoods ignores the contribution of subsidy expirations to increased rent
burdens and potential changes in access to high-opportunity neighborhoods.

Poverty
Neighborhood opportunity is an evolving and enduring concept in research on subsidized
housing. Most commonly, neighborhood opportunity for assisted households has been measured
using poverty rates (McClure 2006; McClure, Schwartz, and Taghavi 2015; Pendall 2000), but
recent research has also examined public safety (Lens, Ellen, and O’Regan 2011), school quality
(Ellen and Horn 2012), and job accessibility (Lens 2014). Most of this research focuses on the
housing voucher and public housing populations, and to a lesser extent, households in LIHTC
properties.

12

Evidence clearly illustrates that public housing residents live in comparatively poor
neighborhoods. Goering, Kamely, and Richardson (1997) found that, in 1990, just under one half
of all public housing tenants lived in high-poverty census tracts (tracts with poverty rates of 40
percent or higher). Similarly, Newman and Schnare (1997) reported that more than 43 percent of
tenants in family public housing lived in high-poverty census tracts in 1990.

Voucher households have also been found to occupy relatively high-poverty
neighborhoods. Pendall (2000), examining census tract-level data from HUD on 1998 voucher
households, found that neighborhoods with voucher holders had a 1990 poverty rate of 20
percent on average, compared to the nationwide average of 15 percent. In addition, tenants
receiving all forms of assistance were more likely than renters as a whole to live in
neighborhoods scoring high on a neighborhood distress index, constructed from poverty rates;
public assistance receipt; and the proportion of female-headed households, high school dropouts,
and labor force participants.

Particularly germane to this paper is the siting of LIHTC units. McClure (2006)
compared locational outcomes for the voucher and LIHTC programs. Using 2002 administrative
data on voucher households and LIHTC units placed in service through that year, he found that
about 30 percent of LIHTC households and 26 percent of voucher households lived in low-
poverty census tracts. And on average, voucher holders lived in very slightly lower poverty
neighborhoods than LIHTC households. Significantly, the proportions of LIHTC and voucher
households in high-poverty tracts were slightly lower than the percentages of poor households
who lived in high-poverty tracts, although higher than the share of all renters who lived in such
tracts. The households assisted through both of these programs, in other words, were somewhat
better off than the average poor households in terms of neighborhood opportunity, but they were
still living in neighborhoods that had significantly higher poverty rates than other renters (at least
in 2002).

Williamson, Smith, and Strambi-Kramer (2009) also examined LIHTC locational
outcomes. They concluded that LIHTC units are infrequently built in low poverty tracts. They

13
assert that, on average, the LIHTC concentrates poverty similarly to how public housing does,
due to the inability of a lot of projects to attract a true income mix and for the preference built
into the credit for qualified census tracts (QCTs), meaning those census tracts with higher
poverty rates. The preference for QCTs leads LIHTCs to be more commonly sited in higher
poverty neighborhoods than they otherwise would.

Ellen, O’Regan, and Voicu (2009) examined national data on LIHTC locations to assess
the siting of these properties in relation to the neighborhood poverty rates in those census tracts,
and painted a more favorable picture. They found that in the 1980s, 1990s, and 2000s, LIHTC
units were 3-4 times more likely to be located in low-poverty (10 percent or less) tracts than
high-poverty (40 percent or more) tracts. They concluded—echoing McClure (2006)- that
LIHTC units are much less likely than public housing to be located in high-poverty census tracts.
The takeaway from research on LIHTC siting and poverty is that LIHTC properties compare
very similarly to voucher locations in terms of neighborhood poverty rates. Each subsidy is
located in areas with much lower poverty rates than where public housing is located but higher
poverty rates than where the general population lives. There is little to no research on location
outcomes of new Section 8 New Construction/Substantial Rehabilitation (Section 8 NC/SR)
properties, likely because those programs were phased out before research on neighborhood
opportunity became common.

Crime
The other neighborhood attribute that is frequently examined with respect to housing
subsidy locations is crime. Much of what we know about this topic comes from voucher
demonstration programs, such as Gautreaux, Moving to Opportunity (MTO), and HOPE VI. The
vast majority of research on these programs concludes that there have been meaningful numbers
of public housing projects located in extraordinarily high-crime areas (Keels et al. 2005;
Kingsley and Pettit 2008; Popkin et al. 2000; Rubinowitz and Rosenbaum 2002). However,
while public housing neighborhoods are typically higher in crime than average, this is not
uniformly the case (Lens, Ellen, and O’Regan 2011).

14
Research on these programs also sheds light on the efficacy of the voucher program at
locating households in lower crime neighborhoods. Among other research, Keels et al. (2005)
find that voucher households participating in in the Gautreaux program are in much lower crime
neighborhoods than they came from in public housing. These findings are echoed in MTO
research (Briggs, Popkin, and Goering 2010; Kingsley and Pettit 2008; Sanbonmatsu et al. 2011)
and studies of HOPE VI (Popkin and Cove 2007).

The one paper that looks at all housing voucher, public housing, and LIHTC households
in large metropolitan areas is Lens, Ellen, and O’Regan (2011). The authors estimated the
neighborhood crime rates faced by the typical voucher household, and compared those rates to
LIHTC and public housing residents, in addition to poor renters and the overall population. They
found that voucher households live in significantly safer neighborhoods than LIHTC units,
which were located in similarly high-crime neighborhoods as public housing, contrary to
research on poverty.

Other measures of neighborhood opportunity
Horn, Ellen, and Schwartz (2014) linked data on housing subsidy recipients to school
location and performance data in order to estimate the extent to which these households live in
areas with high-quality schools. Overall, they found that voucher households with children lived
in areas near to schools with math proficiency rates that were 3 percent higher than public
housing households with children. However, voucher households lived near lower performing
schools than LIHTC, poor renters, all renters, and households in fair market rate (FMR) units.

A recent paper from the Urban Institute (Turner et al. 2011) identifies neighborhood
opportunity across several domains. Three of the indicators are essentially the flip-side of the
underclass and distress measures defined by Ricketts and Sawhill (1988) and Kasarda (1993),
respectively. The resulting measure includes thresholds for work participation, income, college
completion, percent white, and job density. Their goal in this paper was to examine the extent to
which Moving to Opportunity (MTO) program participants were able to access higher
opportunity neighborhoods. Notably, they found that the MTO program did not noticeably

15
increase participants’ occupancy in higher opportunity neighborhoods. Furthermore, many MTO
households that did gain access to these neighborhoods lost access fairly quickly.

McClure (2011) further advocates for a more complex measurement of neighborhood
opportunity: “the development of an opportunity index should examine the potential for
improved educational attainment, greater safety from crime, a higher probability of obtaining
gainful employment, as well as finding a good quality dwelling unit at an affordable rent”
(McClure 2011, p.10). McClure used a factor analysis to produce a neighborhood opportunity
index with the goal of narrowing the list of variables (or factors) that explain the majority of the
variation in the initial variables, possibly due to the high level of correlation between the various
constructs. Using this factor analysis, he recommends that an analysis of neighborhood
opportunity should include: the incidence and level of poverty, educational attainment,
employment rates, employment accessibility, race, and the presence of other assisted households.
He further suggested that the measure should be employed at the block group level where
possible. McClure notes that missing from this measure is school quality and crime rates, due to
a lack of data availability.

In sum, we can conclude that there is little research focusing on neighborhood
opportunity for the Section 8 program, but there is important work that assesses the
neighborhood locations of LIHTC units. Notably, how the LIHTC program compares to other
subsidies on these metrics depends on how neighborhood opportunity is assessed. The LIHTC
program compares favorably to vouchers and public housing neighborhoods in terms of the
performance of zoned schools (Horn, Ellen, and Schwartz 2014). On the other hand, LIHTC
units are in much higher crime areas than housing voucher households. (Lens, Ellen, and
O’Regan 2011). In terms of poverty, voucher and LIHTC households are very comparable.

Rental subsidies and expirations
There is almost no research that focuses on properties leaving the subsidized housing
stock and what this means for neighborhood opportunity. Owners who receive a subsidy through
the LIHTC or Section 8 NC/SR programs agree to abide by affordability restrictions for a fixed
period of time. The length of affordability restrictions on subsidized properties varies by

16
program, state, and when the property was developed. For example, early LIHTC properties only
had a 15-year affordability restriction period, and this was later changed to extend the initial
compliance period by another 15-year extended use restriction period. In some states the
affordability period is even longer—in California, the period extends to 55 years.
2
The Section 8
NC/SR program generally required a 20- to 40-year affordability period. Given that the program
began in the early 1970s, some owners were no longer required to abide by affordability
restrictions starting in the 1990s.

At the end of an initial affordability period, an owner in the Section 8 NC/SR program is
given the option to renew their contract with HUD for another period of time, or leave the
program. Generally HUD can offer contract renewals of varying lengths, i.e. one, five or 20
years, however, obtaining longer-term contracts requires higher levels of approval, which makes
them less common, and all contracts are subject to annual appropriations.
3
If an owner chooses to
renew the contract, they are not offered subsidies beyond the Section 8 program as of right, but
the renewal period provides an opportunity for owners to adjust property rents to better align
with market rents or building operating costs.
4
In the case where an owner chooses to leave the
program, they must go through a formal process of opting out, which, among other things,
requires notifying tenants of their decision. A Section 8 NC/SR subsidy can also end because a
property failed out of the program. HUD inspects properties in this program, and if a property
fails two consecutive inspections, HUD has the right to terminate the contract. A property can
also fail out due to foreclosure. In all three of these cases, tenants are offered a voucher that they
can use to a rent their existing unit or another unit on the private market, as a form of tenant
protection. The analysis of tenants in these properties conducted in chapter 3 finds that only 48
percent of these households use their vouchers, and those households who do not use their
vouchers lose, on average, over $400 in rental support per month. As this study notes, these

2
http://www.treasurer.ca.gov/ctcac/program.pdf
3
“The maximum term of the contract is 20 years. A Contract Administrator can renew a Section 8 Housing
Assistant Payment contract for up to five years. If an owner wishes to renew the contract for more than five years,
the CA must refer the contract to the Account Executive for final approval.”
http://portal.hud.gov/hudportal/documents/huddoc?id=Section8_Renewal_Guide.pdf
4
See HUD’s “Section 8 Renewal Policy Guide” for more detail on the various ways rents can be adjusted:
http://www.hud.gov/offices/hsg/mfh/exp/guide/s8renew.pdf

17
findings are surprising, and there are various mechanisms that could explain a low use rate,
including potential flaws in how these tenant protection vouchers are offered and the constraints
of voucher households more generally (i.e. lack of deposits to secure housing, landlord
resistance, etc.). Based on the analysis in chapter 3, those who do use their vouchers tend to
move from their existing units, and if they change neighborhoods, they are generally moving to
slightly lower poverty neighborhoods, but such gains are marginal and vary widely based on
household characteristics.

Subsidy expirations occur differently in the LIHTC program. All LIHTC properties have
an initial compliance period, where ownership is established as a limited partnership between a
general partner and investors. At the end of this initial compliance period, the limited partnership
has an interest in ending its ownership because the investors no longer receive a benefit from the
partnership and are also no longer subject to penalties for noncompliance with the program
(Khadduri et al. 2012). For those LIHTC properties built before 1990, this is also the point at
which the property is relieved of any affordability restrictions through the LIHTC program. At
the end of the initial subsidy period, these properties owners often take one of four paths: 1)
recapitalize and remain affordable; 2) remain affordable with no additional subsidy; 3) increase
rents; or 4) convert to homeownership (Khadduri et al. 2012). In 1989 the federal government
amended the LIHTC program to extend the affordability restriction to 30 years instead of the
original 15-year restriction period (Begley et al. 2011). As a result, for those properties built
from 1990 on, owners have the options at year 15 to sell the property to the general partner or an
another entity that can take one of two paths with the property: 1) recapitalize and remain
affordable for another 15 years; or 2) remain affordable with no additional subsidy. An owner
cannot formally opt out of the program unless they prove to the administering agency that the
property is no longer financially viable. Also, a LIHTC subsidy is not renewable by design,
meaning owners cannot extend the existing subsidy. Owners who choose to recapitalize must
submit an application for any new public subsidies. Public agencies have shown an interest in
preserving these properties, and often do, but these agencies have fixed resources and must
balance the tension between preserving existing units and developing new ones (Schwartz and
Melendez 2008).

18
Thus far, properties that have exited the LIHTC program, meaning those built before
1990, have largely remained affordable (Schwartz and Melendez 2008; Khadduri et al. 2012). In
this paper we focus specifically on properties that have ended their initial compliance period, did
not recapitalize through the LIHTC program, and did not receive another form of subsidy from
another federal program, despite having to remain affordable for an additional 15 years. We refer
to these properties as Expired LIHTC, and Expiring LIHTC, because the initial compliance
period and benefits from the subsidy ended, or will end, not because the affordability restrictions
ended. This is a significant moment for LIHTC properties because these properties have been
operating for 15 years and likely need some rehabilitation and recapitalization and the ownership
is being restructured because investors are looking to leave the limited partnership. As a result,
while additional subsidy is not as of right, the end of the initial compliance period offers a unique
opportunity for properties to be recapitalized with additional subsidy, and the desire for investors
to exit also offers a valuable leverage point for attracting additional resources.

Consistent with existing research, we find that a large share LIHTC properties that ended
their initial compliance period between 2000-2010 either recapitalized through the LIHTC
program or continued to receive support from another federal program, which make a tract-level
analysis of what happens to properties that expired during this time statistically infeasible.
However, going forward there are far more properties that will approach the end of the initial
compliance period, and administrating agencies will be faced with a new tradeoff of whether to
use scarce resources to: 1) recapitalize existing LIHTC properties that are subject to extended use
restrictions; 2) recapitalization and preserve properties financed through other programs that may
or may not be nearing the end of use restriction periods
5
; or 3) develop new affordable housing
units. While these three options are not mutually exclusive due to subsidy layering on properties,
option one alone could tie up a large share of future LIHTC funds. As a result, one study
suggests that agencies looking to preserve LIHTC subsidies should set clear guiding principles,
carefully examine housing markets, and pay attention to repositioning properties at risk of
increasing rents or converting to homeownership (Khadduri et al. 2012).

5
After 2020, LIHTC properties begin to reach the end of their extended use periods and join this category.

19
Despite the fact that the LIHTC program is the largest financing program for new
subsidized affordable housing, and all owners of LIHTC and Section 8 NC/SR will at some point
be given the option to exit the stock of subsidized rental housing, and government agencies are
increasingly facing a tradeoff between preserving affordability, recapitalizing units, and
developing new affordable housing, there is little research about what this means for
neighborhood opportunity. Generally, theory suggests that the two main reasons owners may
chose to leave rental subsidy programs are: if they believe the immediate or long-term return
then they can obtain from the market minus any costs associated with leaving the program are
higher than the return allowed by the rent structure for these programs, or because the owner no
longer wants to deal with the administrative burdens associated with receiving the subsidy from
the federal government. (Finkel et al. 2007, Ray et al 2015; Reina and Begley 2014; US GAO
2007).

There are four main studies that predict which owners will opt out of rental subsidy
programs. A national study of Section 8 NCR/SR program finds that owners of properties that
target families, properties owned by for profits, those with rents below the local fair market rent,
and those in the worst condition may be more likely to opt out of a subsidy program (Finkel et al.
2007). An update to that study supports those findings but also finds that a strong local rental
market and strong regional home sales markets also increase the likelihood that an owner opts
out (Ray et al 2015). These papers echo findings from another study that focuses on the Mitchell-
Lama program in New York City and finds that owners of properties in areas with high price
appreciation have higher odds of opting out of their rental subsidy. This means that properties in
neighborhoods that are becoming less affordable are the ones where the stock of privately owned
subsidized housing is decreasing (Reina and Begley 2014). That study also found that properties
where all affordability restrictions expired were more likely to opt out, as were those owned by
for profit developers. Similarly, a study of LIHTC expirations found that LIHTC properties are
more likely to be preserved if a non-profit is part of the ownership structure or there are
additional affordability restrictions (Melendez et al. 2008). Interestingly, that study also found
that high rents alone did not contribute to a property converting to market-rate, but that
properties with high rehabilitation costs were more likely to remain affordable because it is
difficult for them to be converted to market.

20

Ellen and Weselcouch (2015) conducted an analysis of subsidized properties in New
York City and found that properties where the owner opted out of a subsidy contract were in
higher cost and higher amenity neighborhoods than those where new and existing subsidized
units were located. This means that there is an increasing concentration of subsidized units in
lower opportunity neighborhoods in New York City.

Existing research has evolved to document neighborhood opportunity for housing subsidy
recipients, focusing on public housing, voucher, and LIHTC programs. This research has been
important in shaping housing policy, specifically in the development of housing mobility
programs. This paper builds on this rich literature by turning the focus on housing subsidies that
either have expired or will soon expire. Given the large numbers of LIHTC and Section 8 NC/SR
subsidy expirations on the horizon, this is a timely inquiry.

III. Data and Methods
Our goal in this paper is to identify the types of neighborhoods in which we see LIHTC
and Project-based Section 8 subsidies expiring and those where new LIHTC units are being built.
Using several neighborhood characteristics, we compare the average neighborhood opportunity
of properties that are exiting the LIHTC and Section 8 programs and those entering the LIHTC
program. We also compare these neighborhood characteristics to outcomes for housing voucher
and public households, in addition to renter households below the poverty line, as credible
comparison groups for households in LIHTC and Section 8 properties.

We link data sets with information on active and expired housing subsidies and append
them to neighborhood-level demographic characteristics (such as poverty, race, and income) and
neighborhood features such as crime rates, school quality, and job accessibility. We proxy for
neighborhoods using the census tract.
6
For public housing and housing voucher subsidies, we use
data from HUD’s Picture of Subsidized Households, which is available publicly for 2000 and

6
The majority of these data are provided using 2010 tract boundaries. For those variables that were in 2000
boundaries, we utilize the Longitudinal Tract Database (LTDB) which is a public-use file for creating 2010
boundary estimates with 2000 boundary data and vice versa (Logan, Xu, and Stults 2014).

21
2004 to 2010. We also have data from 2000 to 2014 from the National Housing Preservation
Database (NHPD), and use these data for the LIHTC
7
and project-based Section 8 unit locations.
The NHPD was built, and is updated, by the Public and Affordable Housing Research
Corporation (PAHRC), and the National Low Income Housing Coalition (NLIHC), and is
publicly available. These organizations also provided us with a non-public database that provides
additional history on the expired properties. One factor that complicates studying subsidized
housing is that many properties receive multiple forms of subsidy, requiring us to identify all
subsidies on a property in order to determine the actual restriction end dates and whether owners
opted out of all subsidy programs (Reina and Williams 2012). The benefit of the NHPD is that it
tracks properties over time and across multiple programs. Thus, if a LIHTC property reaches the
end of its initial compliance period (its effective subsidy expiration date), we can see whether
this property received a new round of LIHTC financing or remained subsidized through another
federal subsidy program.

There are several important limitations of these data. First, we do not have reliable
information about whether owners choose to renew their project-based Section 8 contract. We
can only see whether a property, and the units in those properties, remain in the program or
exited. Another limitation is that the further one goes back in time, the less reliable the data,
which is why we do not focus on subsidy expirations before 2000.

Figure 1 displays the changes in the different subsidy programs since 2000, where we
observe several important trends. First, while the overall stock of subsidized rental housing
increased since 2000, this increase is due to growth in the number of vouchers and additional
units developed through the LIHTC program. During this time the number of public housing and
project-based Section 8 units decreased. In Figure 2, we highlight inflows and outflows of units
in the major privately-owned project-based subsidy programs and again we see that all new units
developed are from the LIHTC program, and almost all units leaving the subsidized stock are
from the project-based Section 8 program. While the number of new LIHTC properties dwarfs
the number of LIHTC properties that expired between 2000 and 2010, the number of LIHTC

7
There are two types of Low Income Housing Tax credits, that provide different levels of subsidy- the 4 percent and
9 percent credit- and we include both in this study.

22
expirations will dramatically increase going forward. For example, fewer than 6,000 LIHTC
passed the year 15 mark before 2010 and were neither recapitalized through the LIHTC program
nor received another subsidy from another federal program. As previously stated, given the small
number of LIHTC properties that expired between 2000 and 2010, we only analyze LIHTC
expirations between 2011 and 2020. For the foreseeable future, growing numbers of LIHTC
properties will continue to reach the end of their initial compliance period, and in 2020 properties
will start to reach the end of extended use restriction periods. In addition, while the number of
project-based Section 8 contracts ending in any given year is a small share of the overall
portfolio, no new units are being developed through this program, meaning an increasing share
of project-based Section 8 units either have expired or will.
8

The raw number of inflows and outflows also do not tell the whole story with respect to
affordability. There are several other factors to consider: the depth or size of the subsidy,
whether the subsidy is renewable, and the location of the unit. The project-based Section 8
program offers a deeper subsidy than the LIHTC program. Households in a property covered by
project-based Section 8 are guaranteed to pay no more than 30 percent of their income in rent,
whereas LIHTC residents may pay a higher share, and the evidence suggests they typically do,
and only some households living in LIHTC properties are low-income (O’Regan and Horn
2013). Second, a project-based Section 8 contract is renewable; whereas once a LIHTC property
reaches the end of its initial compliance period and subsequently its extended use period, there is
no option to renew the tax credit subsidy, which creates a barrier to recapitalization and
continued affordability. Finally, there could be variation in neighborhood opportunity where
units are located, and we want to consider whether the units entering the subsidized portfolio
offer access to similar, higher, or lower opportunity neighborhoods than those leaving the
portfolio. This final point is the focus of this paper.

8
According to HUD, a Public Housing Authority (PHA) can project-base up to 20 percent of its housing choice
vouchers. These contracts represent “new” project-based contracts but “HUD does not reserve additional units for
project-based vouchers and does not provide any additional funding for this purpose” 24 CFR §983.5.

23
To assess neighborhood opportunity for existing and expired subsidized housing units,
we add demographic data from the 2000 U.S. Census (U.S. Census Bureau 2001) and the 5-year
waves of the American Community Survey (U.S. Census Bureau 2015) that begin in 2005 and
end in 2013.
9
We also estimate tract-level employment accessibility using methods replicated
from Lens (2014) and Shen (2001). These papers estimate job openings using multiple years of
data from the U.S. Census Bureau’s Longitudinal Employer Household Dynamics (LEHD) (U.S.
Census Bureau 2013) files and a distance decay function to weigh jobs inversely proportional to
their distance to resident census tracts. We also take advantage of a nationwide database on
reading and math test scores for public schools. These data allow us to estimate the quality of the
public school that is nearest to each census tract.
10
Finally, we use HUD’s Location Affordability
Index (LAI) (U.S. Department of Housing and Urban Development 2015) to compare census
tracts according to the estimated share of income households below 50 percent Area Median
Income spend on transportation costs.
11
Ideally, housing subsidies would not be expiring where
transportation costs are low.

We concentrate on several variables that have been shown to affect household outcomes
and/or are continually cited as the most important neighborhood features to families in
subsidized housing. These are: percent in poverty, percent nonwhite (to identify location in
racially integrated neighborhoods), unemployment rate, percent of female-headed households,
percent with a high school diploma, spatial job accessibility, location affordability, school
performance (2008-09 school year only), and for a subsample of census tracts (91 cities) in 2000,

9
In the panel dataset that we construct, we match the annual data to the midpoint year of the 5-year wave. Thus, the
2005 to 2009 ACS is considered 2007 data.
10
The data were provided to researchers at the NYU Furman Center by the Department of Education. We thank the
Furman Center for supplying these data geocoded to census tracts. There are several limitations to the data, despite
their value as the only national database that can approximate school quality on the census tract level. First, test
scores are by no means the only meaningful indicator of school quality. Second, we have no information on nearby
private or charter schools. Finally, the most geographically proximate school may not be the school that a census
tract’s households are zoned for, although Horn et al (2014) report that the correlation between proximity and formal
zoning is very high. See Horn et al (2014) for more details on these data.
11
It is important to note that transportation costs are estimated based on census tract characteristics, and not actual
household spending on various forms of transportation. These characteristics include: transportation cost estimates
are based on the following census tract-level factors in lieu of actual transportation cost data: density, walkability
and street connectivity, employment access and job density, median commute distance, percent rental units, and
percent single family detached housing units. For more, see:
http://www.locationaffordability.info/lai.aspx?url=user_guide.php.

24
we have data on major violent and property crimes from the National Neighborhood Crime
Study (NNCS) (Peterson and Krivo 2010).
12
Table 1 provides descriptive statistics on the
housing subsidy variables and these neighborhood features. This table and all subsequent
analyses are limited to census tracts with population greater than 200 in counties with population
greater than 100,000 as of the 2000 Census. We then conduct bivariate analyses to provide a
description of the types of neighborhoods where subsidies are entering and exiting.

IV. Results
In What Types of Neighborhoods do Subsidies Expire?
Tables 2 and 3 provide averages for each neighborhood characteristic, weighted by the
prevalence of each housing subgroup in that census tract.
13
These groups are not mutually
exclusive for properties because a property can have more than one form of subsidy on it, and a
property active from 2000-2010 can also be one expiring in 2010-2020. We have 12 distinct
groups in total: LIHTC subsidies expiring between 2011 and 2020, all Active LIHTC properties,
new LIHTC properties that became active between 2000 and 2010, and LIHTC properties not set
to expire until after 2020; all of the same groups for the Section 8 program but substituting new
properties with those that expired between 2000 and 2010; housing voucher, public housing, and
renter households below the poverty line in 2010; and all tracts in cities with population greater
than 100,000. Note that we include Section 8 subsidies that expired in the 2000s, but not LIHTC
properties that expired during the same time, due to very low numbers of LIHTC properties that
expired. We do not include new Section 8 properties because this subsidy program no longer
produces new units.

In Table 2 we provide the results for each group for what we consider structural
neighborhood characteristics—violent and property crime rates, school test scores, transportation
costs as a share of income, median rents, and job accessibility index. These weighted averages

12
NNCS data were obtained from the Inter-university Consortium for Political and Social Research (ICPSR) at the
University of Michigan. For more on the NNCS, see Peterson and Krivo (2010).
13
We conducted tests for whether the weighted averages for each group and neighborhood characteristic are
significantly different from one another, but we do not add these to the tables or discuss differences in statistical
significance terms for the sake of simplicity. Given the large number of observations (over 50,000 census tracts for
most variables), only numbers that are nearly identical are not statistically significantly different. The authors will
provide documentation of statistical significance tests upon request.

25
provide a measure of the neighborhood characteristics available to the typical household within
each housing subgroup. For example, when we look at the Section 8 properties that expired
between 2000 and 2010, the average violent crime rate for that group is 78 violent crimes per
1,000 persons. This is the highest average crime rate faced by any of the groups, and is
significantly higher than the violent crime rate faced by the active Section 8 properties in 2010.
14

Overall, Section 8 units expired in particularly disadvantaged neighborhoods. This is most
evident when looking at violent and property crime rates, and the relatively low median rents and
job accessibility numbers in Table 2. There is little variation in school test scores across all
subsidized groups, with all portfolios being located in tracts with average scores below the
average tract.

For LIHTC properties, the new, active, and expiring properties are generally located in
similarly situated neighborhoods, although median rents are higher in neighborhoods with
LIHTC properties set to expire during the 2010s. When considering these structural
neighborhood characteristics, we can conclude that the Section 8 properties that expired in the
2000s were typically located in more distressed neighborhoods than the entire portfolio of these
properties. Also, the LIHTC properties expiring in this upcoming decade are in slightly more
expensive neighborhoods than the overall portfolio, but all other such characteristics are very
similar across the LIHTC groups.

Table 3 provides the same measures, using neighborhood demographic characteristics.
The differences across all housing groups are more muted for these characteristics, but there are
some notable observations when looking specifically at LIHTC and Section 8 properties. Section
8 properties that expired in the 2000s are in neighborhoods with similar incomes and poverty
rates as the active portfolios, but Section 8 properties expiring in the 2010s are in neighborhoods
with higher incomes and lower poverty, unemployment, and female-headship rates, suggesting
that households may lose access to less distressed neighborhoods if these subsidies are not
preserved.

14
We want to emphasize that the crime data were only available for 91 cities (not entire metropolitan areas), and
only in the year 2000. Thus, while most analyses involve approximately 54,000 census tracts, the violent crime
analyses cover 8,048 tracts and the property crime data cover nearly 11,000.

26

Across the two tables, a consistent pattern emerges. The active portfolio of LIHTC
properties are in slightly higher opportunity neighborhoods than are Section 8 properties. Each of
these portfolios are generally in neighborhoods that look similar to housing voucher and poor
renter households, worse off than the general population, and better off than public housing
properties. Section 8 properties that expired in the 2000s appear to be in particularly low-
opportunity neighborhoods. On the other hand, the properties in the Section 8 program that are
due to expire in the 2010s appear to be in relatively higher opportunity neighborhoods—this is
significant because these are the properties that may result in a loss to program participants in
terms of neighborhood opportunity. Section 8 properties eligible to expire in the 2010s are in
neighborhoods with the lowest violent crime rates, highest math scores, have relatively high
median rents, incomes, and job accessibility, relatively low poverty rates and are more racially
integrated. LIHTC properties expiring in the 2010s have particularly high median rents and
relatively high incomes, but aside from those two measures appear to be in similar
neighborhoods as existing and new LIHTC properties.

To better summarize some of these results, we developed aggregated neighborhood
measures and present them in Table 4. In the first column, we average the rankings of 9 of the
groups
15
across the neighborhood characteristics in Tables 2 and 3. In these columns, low
numbers (i.e. 1-5) reflect higher neighborhood opportunity and high numbers denote
disadvantaged characteristics. Looking at the LIHTC program, we see that this aggregate
measure shows that units set to expire in the 2010s are in relatively desirable neighborhoods. The
average ranking is for those properties is 4.0, compared to 4.3 for the entire portfolio and 5.2 for
those that just entered the portfolio during the 2000s. Given these numbers, new LIHTC
properties are clearly in neighborhoods that perform worse than existing and expiring LIHTC
properties, echoing the small differences we observed in Tables 2 and 3. The story is the same
for Section 8 properties: while the average ranking for those that expired in the 2000s was 6.2
(behind public housing as the second lowest ranked), the average ranking for those that are set to

15
All of the above groups with the exception of the LIHTC and Section 8 groups that do not expire until 2020 and
the full sample of census tracts.

27
expire in the 2010s is 2.6, making these the highest ranked neighborhoods of all the housing
subsidy groups.
Next, we calculated standardized values (Z scores) for each neighborhood characteristic,
and took the average of those Z scores as another way to aggregate these values. These measures
factor in the magnitude of differences between groups, which is not possible by looking at
average rankings. The Z score can be interpreted as the number of standard deviations from the
average U.S. neighborhood across the neighborhood characteristics. For all of the housing
groups, the average Z score is negative, reflecting the fact that they are located in less desirable
neighborhoods than the average U.S. household, and subsidies in higher opportunity
neighborhoods have average Z scores closer to zero. In Table 4, we see that the values range
from 0.30 standard deviations below the national average (expiring Section 8 units), to 0.67
standard deviations (public housing). The results confirm that the neighborhoods where LIHTC
properties are set to expire in the 2010s are slightly worse off than other LIHTC properties (-0.38
for the 2010s versus -0.41 for the active portfolio), and the Section 8 properties eligible to expire
in this decade are in better neighborhoods than the overall portfolio (-0.30 for the 2010s versus -
0.42 for the 2000s). Further, the Section 8 properties soon to expire are in better neighborhoods
than any other portfolio we analyze, including vouchers.

Finally, we aggregated neighborhood characteristics by identifying the percent of each
housing group located in high- and low-opportunity neighborhoods. Neighborhoods qualified if
they were in the top (or bottom) half of the distribution of neighborhoods in all of the following
characteristics: math scores, poverty rates, unemployment rates, high school attainment, percent
nonwhite, and percent female-headed households. Calculations were made theoretically
consistent across all characteristics—we correct for the fact that higher math scores are desirable
and unemployment is undesirable by using the appropriate half of the distribution.

In the final two columns of Table 4, we display the percent of each housing population
that occupies high- and low-opportunity neighborhoods. This measure captures extreme
neighborhoods, as only about 11 percent of neighborhoods qualify as high-opportunity, and
about 14 percent qualify as low-opportunity. We see here that Section 8 expirations from the
2000s occurred much more frequently on these extreme ends of the opportunity spectrum.

28
Expirations in extreme neighborhoods makes some sense from the standpoint of the property
owner. Owners may allow these subsidies to expire in severely distressed neighborhoods because
they find it difficult to rent out even subsidized units there, given these rents do not provide
much of a discount compared to market rents. And in high-opportunity neighborhoods, it makes
economic sense to decline further subsidy and convert their units to market rate.

The properties expiring in the 2010s are similar in terms of location in high- and low-
opportunity neighborhoods as the overall portfolios. Roughly 4 percent of Section 8 and 3
percent of LIHTC properties set to expire in the 2010s are in high opportunity neighborhoods.
Although these numbers are low, these could be the ideal properties for targeted preservation
efforts if the goal is to ensure low-income residents can access particularly strong
neighborhoods.

Neighborhood Trajectories
For a subset of variables—median rent and income, poverty, female headship, and
unemployment rates, percent nonwhite, and high school completion—we look at the change
between 2000 and 2010 in order to get a sense of how the neighborhoods with expired properties
are changing. In Table 5 we display the change in seven indicators from 2000 to 2010, weighted
by the number of each housing subsidy group. All else equal, we would expect expiring Section
8 contracts to not be renewed in improving neighborhoods, and indeed our research shows this to
be the case. The neighborhoods with Section 8 properties that exited the program in the 2000s
were improving relative to the entire portfolio of Section 8 property neighborhoods across every
neighborhood indicator. Of all the housing subsidy groups, Expired Section 8 neighborhoods
were the only ones without an increase in the poverty or female headship rate from 2000 to 2010.
These neighborhoods had the largest increase in median rents, and they had a 4 percent increase
in median incomes while active Section 8 properties Active in 2010 saw a decline in
neighborhood median incomes, on average. These results suggest that Section 8 properties
expired in neighborhoods that were improving, while the active portfolio was in neighborhoods
that were potentially declining.

29
In Table 6, we aggregate these change indicators by housing subsidy group. Here we see
a lot of consistency with our previous conclusions that Section 8 units tended to expire in
improving neighborhoods. Section 8 expirations in the 2000s were in neighborhoods with the
best trajectory according to the average change ranking and the average change Z score. And a
full 20 percent of Section 8 expirations were in neighborhoods with high positive change,
whereas roughly 5.5 percent of all neighborhoods and 8 percent of Active Section 8 properties
were in neighborhoods that met that threshold.

From this array of descriptive measures, a consistent story emerges. Section 8 properties
expired in undesirable neighborhoods in the 2000s. Section 8 units that expired in the 2000s were
in neighborhoods with very high violent and property crime rates, high poverty rates and
minority populations, and relatively low incomes and rents. However, the Section 8 expiration
neighborhoods were improving quite rapidly, suggesting that these units expired in areas where
things were getting better. This is a major concern going forward, because the properties eligible
to expire in the 2010s are in much better neighborhoods than where properties expired in the
2000s. Section 8 properties expiring in the 2010s are in neighborhoods with the lowest violent
crime rates, highest math scores, have relatively high median rents, incomes, and job
accessibility, and relatively low poverty rates. In addition, expiring Section 8 properties tend to
be located in higher opportunity neighborhoods that are also on a better trajectory than those of
the average voucher and new, active, or expiring LIHTC unit. LIHTC properties expiring in the
2010s have particularly high median rents, relatively high incomes, and are in neighborhoods of
similar quality to voucher households, but these properties are generally also in neighborhoods
on the worst trajectory of all portfolios we analyze.

V. Discussion
In this paper, we use several sources of data at the census tract level to identify the
characteristics of neighborhoods where housing subsidies are expiring. We look at the years
2000 to 2010 in order to observe where Section 8 subsidies already expired, and use data on
projected expirations from 2011 to 2020 to examine the location of current and future
expirations. These subsidy programs do not offer the only way for low-income households to
access particular neighborhoods, and in some cases, the loss of subsidy may not immediately

30
result in increased rent. However, all else equal, a number of families equal to the number of
units that expire will suddenly find themselves at a higher risk of having to move. Further, they
are more likely to have rents raised where market rents are higher, which happens in the higher
opportunity neighborhoods to which housing policy makers and advocates hope that low-income
households can access. In addition, as noted earlier, the tenant protection voucher does not
appear to protect households from these realities. This is particularly important as we consider
that those units due to expire from the project-based Section 8 program in the coming years are
in higher opportunity neighborhoods than the average unit leased with a voucher.

For the Section 8 NC/SR program, where no new construction has occurred in decades,
the sole source of change is in these expiring subsidies. What these analyses uncover is that
Section 8 subsidies expired in areas that were relatively disadvantaged, which, on its own is not a
negative outcome from a neighborhood access perfective. However, the neighborhoods that saw
a greater share of lost Section 8 subsidies improved substantially (on average) over the decade.
Also, the Section 8 NC/SR subsidies that are set to expire in the 2010s are in relatively desirable
neighborhoods. Further, we find evidence that these subsidies expired in neighborhoods on each
of the extreme ends of the neighborhood opportunity scale. In all, past Section 8 NC/SR
subsidies expired in very disadvantaged neighborhoods that were improving, but current and
future expirations are frequently in high-opportunity neighborhoods.

For LIHTC properties, the story is more mixed, and our analysis is limited by the fact that
we cannot look at neighborhoods where LIHTC properties expired because the numbers are low.
Although the neighborhood opportunity profile of LIHTC expiration tracts is generally quite
mixed, our analyses suggest that many LIHTC expirations in the current decade will occur in
slightly higher opportunity neighborhoods than new and active LIHTC properties, but that these
neighborhoods are not necessarily on the best trajectory.

All of these trends occurred during a time of increasing renter burdens. Unadjusted for
inflation, median rents rose 57 percent on average compared to an average 25 percent increase in
incomes. Further, in the 2000s, over 100,000 Section 8 units lost their subsidy, which accounts
for over 11 percent of the total number of Section 8 units that were in place at the start of the

31
decade. While the number of units that entered the LIHTC program was much larger than the
number that left the Section 8 program, the LIHTC program is a less generous subsidy and often
does not alleviate rent burdens for low and very low-income renters to the same extent. Further,
the LIHTC program is young enough (implemented in 1986) that the subsidy expiration process
is just beginning.

Given the limited investment in housing subsidies at all levels of government, there are
difficult and constrained decisions about whether to spend these resources developing new units
or recapitalizing and preserving existing units, and which existing units to preserve. The data and
methods used in this paper should be used by policymakers and advocates to prioritize
preservation efforts in neighborhoods with higher opportunity. In particular, when making
decisions about whether to preserve a subsidy, there are a host of characteristics that define the
potential level of opportunity. These characteristics change over time, and since owners are
making opt-out decisions based on both levels and changes to neighborhood opportunity, so
should policymakers. The methods used for evaluating neighborhood opportunity in the context
of new and expiring subsidized rental housing in this paper can also be modified and extended by
communities, which may value some indicators more than others or be concerned about local
subsidy programs not considered here.

Finally, while the current tenant protection voucher does not appear to provide an
adequate safety net, housing vouchers can be an important tool and resource as we think about
access to neighborhood opportunity. One option is to focus on improving the effectiveness of the
tenant protection voucher, but this program only applies to households in project-based Section 8
developments. In theory, this voucher shields households from potential rent increases and
provides them the option to use their subsidies to move to other locations, which could ultimately
be more efficient. However, as noted in 3, the ability for a household to do either likely requires
improved oversight of the housing authorities administering this benefit as well as direct support
to households on how to use their voucher. Another option is project-basing vouchers, which
allows a unique place-based intervention that could prove useful. However, it is estimated that
there are 2.76 million families currently waiting for housing vouchers (PAHRC, 2015) and if we
cannot serve those families, the project-basing of vouchers may not be a scalable solution. These

32
potential solutions reflect the fact that we are in a world of constrained resources, and while the
goal should not be to preserve every place-based subsidy, the decision on when and how to
preserve, recapitalize, or build new affordable units involves tradeoffs. Ultimately, such
decisions require detailed knowledge about where units are located, which tools may work in
that context, and what that means for access to opportunity neighborhoods.

33

VI. Figures and Tables
Figure 1: Housing Subsidies, 2000 to 2010

400,000
900,000
1,400,000
1,900,000
2,400,000
2000 2002 2004 2006 2008 2010
LIHTC Active Project-based Section 8 Active Public Housing Vouchers

34
Figure 2: Inflows and Outflows of LIHTC and Section 8 Program Units

0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
90,000
100,000
110,000
120,000
130,000
2000 2002 2004 2006 2008 2010
New LIHTC Expired LIHTC Expired Section 8 Cumulative Expired

35
Table 1: Descriptive Statistics
Variable Obs Mean SD Min Max
Expired LIHTC, 2000-2010 53,666 0.10 5.40
0.00
1,100.00
Expiring LIHTC, 2011-2020 53,666 6.30 36.90 0.00 1,008.00
Active LIHTC, 2010 53,666 28.90 89.90 0.00 2,123.00
New LIHTC, 2000-2010
53,666 15.60 61.50
0.00
1,460.00
Expired Section 8, 2000-2010 53,666 1.70 42.10
0.00
7,807.00
Expiring Section 8, 2011-2020 53,666 4.30 24.90
0.00
1,154.00
Active Section 8, 2010 53,666 12.40 51.50
0.00
3,690.00
Vouchers, 2010 53,666 33.90 54.10
0.00
1,678.00
Public Housing, 2010 53,666 15.30 83.70
0.00
2,609.00
Violent Crime Rate, 2000 (8,048 census tracts) 7,957 29.60 37.20
0.00
880.00
Prop Crime Rate, 2000 (10,938 census tracts) 10,819 169.20 201.20
0.00
5,878.00
Math Proficiency, 2008 52,032 0.73 0.18 0.05 1.00
Reading Proficiency, 2008 52,037 0.71 0.19 0.09 1.00
Transportation Costs (Low-Income), 2010 53,290 32.30 8.70 7.00 92.00
Median Rent, 2010 53,072 1,044.50 376.90 99.00 2,001
Job Accessibility Index, 2010 50,685 1.30 3.30 -7.00 534
Median Household Income, 2010 53,523 60,074.60 30,014.90 2,499.00 250,001.00
Unemployment Rate, 2010 53,605 0.10 0.06
0.00
1.00
%No High School, 2010 53,654 0.09 0.07
0.00
0.78
%Non-White, 2010 53,654 0.42 0.31
0.00 1.00
%Poverty, 2010 53,523 0.12 0.12
0.00 1.00
%Female-Headed Households, 2010 53,554 0.32 0.12
0.00 1.00
% Change* Poverty Rate, 2000-10 53,501 0.03 0.08 -0.87
1.00
% Change Unemployment Rate, 2000-10 53,583 0.04 0.06 -1.00
1.00
% Change No HS Grad Rate, 2000-10 53,634 -0.03 0.04 -0.72 0.34
% Change Female Headship Rate, 2000-10 53,535 0.02 0.07 -0.61 0.87
% Change Median Rent (2010 $), 2000-10 52,988 0.26 6.42 -0.92 935.00
% Change Median Income (2010 $), 2000-10 53,506 -0.02 0.27 -0.96 31.00
% Change Nonwhite, 2000-10 53,634 0.06 0.09 -0.73 0.65
Note. LIHTC = Low-Income Housing Tax Credit. SD = standard deviation
*Percent change is expressed as percentage point changes from 2000 to 2010. Median rent and median
income are expressed as percent changes from 2000 to 2010.

36
Table 2: Neighborhood Structural Characteristics Weighted by Housing Group Prevalence
Sample: All census tracts in U.S. counties with population 100,000 or greater

Violent
Crime Rate,
2000 (8,048
census tracts)
Prop Crime
Rate, 2000
(10,938
census tracts)
Math and
Reading Test
Score
Proficiency,
2008
Transportation
Costs, 2010
Median
Rent,
2010
Job
Accessibility
Index, 2010
Expiring LIHTC, 2011-2020
54 256 66% 30 $ 906 1.29
Active LIHTC, 2010
54 252 65% 28 $ 862 1.35
New LIHTC, 2000-2010
56 257 65% 29 $ 871 1.38
Expired Section 8, 2000-2010
78 288 64% 27 $ 827 1.66
Expiring Section 8, 2011-2020
40 206 67% 28 $ 853 1.30
Active Section 8, 2010
51 231 63% 27 $ 808 1.47
Vouchers, 2010
40 182 64% 29 $ 941 1.19
Public Housing, 2010
58 227 64% 25 $ 696 1.78
Renters below Poverty Line, 2010
38 177 65% 30 $ 894 1.17
All Tracts
25 151 73% 33 $ 1,069 1.01
Note. LIHTC = Low-Income Housing Tax Credit.

37
Table 3: Neighborhood Demographic Characteristics Weighted by Housing Group Prevalence
Sample: All census tracts in U.S. counties with population 100,000 or greater

Median
Household
Income, 2010
Unemployment
Rate, 2010
%No High
School
Diploma,
2010
%Non-White,
2010
%Poverty,
2010
%Female-
Headed
Households,
2010
Expiring LIHTC, 2011-2020
$ 42,927 13% 12% 62% 21% 41%
Active LIHTC, 2010
$ 40,830 13% 13% 61% 22% 42%
New LIHTC, 2000-2010
$ 40,938 14% 13% 62% 22% 42%
Expired Section 8, 2000-2010
$ 39,507 14% 12% 64% 23% 45%
Expiring Section 8, 2011-2020
$ 44,658 12% 12% 52% 19% 42%
Active Section 8, 2010
$ 39,718 13% 13% 55% 22% 44%
Section 8 Not Expiring, 2011-2020
$ 40,739 13% 12% 54% 22% 44%
Vouchers, 2010
$ 43,256 13% 13% 61% 20% 41%
Public Housing, 2010
$ 31,812 16% 16% 69% 31% 50%
Renters below Poverty Line, 2010
$ 41,439 13% 14% 61% 24% 39%
All Tracts
$ 62,304 10% 9% 41% 12% 31%
Note. LIHTC = Low-Income Housing Tax Credit.

38
Table 4: Aggregated Neighborhood Statistics for Housing Subsidy Groups
Sample: All census tracts in U.S. counties with population 100,000 or greater

2010 Average
Ranking* (From
Tables 2 and 3)
Percent in High
Opportunity
Tracts**
Percent in Low
Opportunity
Tracts**
Tract Average Z
Score*
Expiring LIHTC, 2011-2020
4.0 3.4% 32.4% -0.38
Active LIHTC, 2010
4.3 3.1% 33.5% -0.41
New LIHTC, 2000-2010
5.2 3.1% 35.1% -0.42
Expired Section 8, 2000-2010
6.2 6.3% 44.6% -0.42
Expiring Section 8, 2011-2020
2.6 4.1% 28.6% -0.30
Active Section 8, 2010
4.8 3.0% 33.1% -0.41
Vouchers, 2010
3.3 2.6% 33.2% -0.37
Public Housing, 2010
6.9 1.1% 42.2% -0.67
Renters below Poverty Line, 2010
4.3 2.6% 32.4% -0.43
Note. LIHTC = Low-Income Housing Tax Credit.
*Lower numbers suggest higher opportunity
**Approximately 14 percent of tracts are high opportunity, and 14 percent are low opportunity

39
Table 5: Change Indicators by Housing Subsidy Group (Percent Change from 2000 to 2010)
Sample: All census tracts in U.S. counties with population 100,000 or greater

Median
Household
Income
(2010 $)
Unemployment
Rate
Share No
High
School
Diploma
Share
Non-white
Poverty
Rate
Female
Headship
Rate
Median
Rent
(2010 $)
Expiring LIHTC, 2011-2020
-7% 4% -3% 7% 5% 5% 19%
Active LIHTC, 2010
-5% 4% -3% 6% 4% 3% 21%
New LIHTC, 2000-2010
-5% 4% -4% 6% 4% 4% 22%
Expired Section 8, 2000-2010
4% 3% -4% 2% 0% 0% 30%
Expiring Section 8, 2011-2020
-4% 4% -3% 6% 3% 2% 19%
Active Section 8, 2010
-4% 4% -4% 4% 4% 2% 20%
Section 8 Not Expiring, 2011-2020
-5% 4% -4% 5% 4% 2% 20%
Vouchers, 2010
-4% 4% -3% 6% 4% 3% 22%
Public Housing, 2010
2% 2% -5% 2% 2% 1% 27%
Renters below Poverty Line, 2010
-7% 4% -3% 7% 6% 3% 21%
Note. LIHTC = Low-Income Housing Tax Credit.

40
Table 6: Aggregated Change Indicators by Housing Subsidy Group
Sample: All census tracts in U.S. counties with population 100,000 or greater

Average Change
Ranking* (From A1)
Percent in
Tracts with
High Positive
Change**
Percent in Tracts
with High
Negative
Change**
Tract Average
Change Z Score*
Expiring LIHTC, 2010-2020
7.0 4.6% 11.8% -0.13
Active LIHTC, 2010
4.6 6.8% 8.3% -0.04
New LIHTC, 2000-2010
4.3 6.2% 9.5% -0.05
Expired Section 8, 2000-2010
1.3 20.0% 4.2% 0.25
Expiring Section 8, 2010-2020
4.3 5.9% 4.5% 0.00
Active Section 8, 2010
3.6 7.8% 4.4% 0.04
Vouchers, 2010
3.9 6.2% 6.5% -0.02
Public Housing, 2010
1.6 12.8% 2.2% 0.23
Renters below Poverty Line, 2010
6.1 4.1% 8.9% -0.13
Note. LIHTC = Low-Income Housing Tax Credit.
*Lower numbers suggest higher opportunity
**Approximately 5 percent of tracts had high negative change, and 5 percent had high positive change

41

Chapter 3:

Are they protected? The end of place-based rental subsidies and the
welfare of low-income households

42
I. Introduction
Rental markets are becoming increasingly unaffordable, with research suggesting that
rent increases across the country far outpaced income growth since 2000. During this time the
share of affordable private-market rental units decreased, resulting in federally subsidized units
comprising a larger share of the overall affordable rental stock (Schwartz et al. 2016). There is an
abundance of research on where subsidized multifamily properties are located and whether these
properties enable low-income households to access opportunity neighborhoods. However, less
well understood is what happens to families when they lose subsidized housing, and whether the
protection they receive provides an effective safety net to shield them from welfare losses. This
research examines those questions.
The project-based Section 8 program is the largest rental assistance program used by the
U.S. Department of Housing and Urban Development (HUD) to finance private owners to
provide building-based rental assistance. HUD has financed over 1.2 million units of housing
through the project-based Section 8 program, and no new units have been built through this
program since the 1970s. Between 1996 and 2010, almost 400,000 tenants lived in properties
where either the owner, or HUD, terminated the project-based Section 8 subsidy. Upon
termination, these low-income tenants become eligible to receive a voucher, designed to shield
them from being displaced and/or spending a higher share of their income on rent. This paper is
the first to study this phenomenon and the implications it has for households who have lost their
Section 8 subsidies.
It is important to understand what happens to these households for several reasons. First,
almost all new subsidized properties developed in the United States since the 1970s are privately
owned, which means at some point, owners of these properties will have the choice to renew
their subsidy, apply for a new subsidy, or convert their property to market rate. This means the
number of households affected by subsidy expirations is significant, and will only continue to
increase. This event also allows us to assess the utility of a voucher as a safety net. There is a
range of literature that analyzes the efficiency of vouchers (Olsen 2003) and the merits of a
voucher as an opportunity vehicle (Lens 2014), including studies of the Moving to Opportunity
(MTO) experiment (Kessler et al. 2011), but none have evaluated whether vouchers provide a

43
safety net that can shield households from moving away from opportunity. This is particularly
important because rental vouchers are increasingly being used as a tool for reducing rent burdens
and promoting neighborhood access at a time when rental markets around the world are
becoming less affordable. Finally, there is an active debate in the United States about when to
preserve existing place-based subsidies, but there is little knowledge about what happens to
households when a subsidy ends. The value of preserving the affordability of a property is
comprised of many factors, including the welfare cost to households and society if these place-
based subsidies end.
This paper is the first to use a national census of every tenant who lived in a property
when the project-based Section 8 subsidy ended between 1996-2010 and analyze what happens
to these households, including whether they use their vouchers, and if and where they use their
vouchers to move. Overall, these data show that slightly less than 50 percent of households who
live in properties where the project-based Section 8 contract ends actually use their vouchers.
Absent a benefit from the loss of a lock-in effect, households who do not use their vouchers
experience a financial loss that likely reduces their overall welfare and utility. Those who do not
use the voucher on average forego over $434 per month of rental assistance, which is equivalent
to roughly 41 percent of their income. To remain in the same unit, these households would need
to increase their rent payment by over 300 percent to make up for the loss of the rental subsidy if
their unit remained at the same price, which is the equivalent of spending almost all of their
income on rent. Those who use their vouchers tend to move multiple times and to slightly lower
poverty neighborhoods, which means that the expiration of a project-based section 8 contract
could be welfare-improving for a fraction of the households in these properties. However, the
voucher does not provide a safety net for the majority of households who either do not use their
vouchers or use their vouchers to move to a higher poverty tract. In addition, households with the
highest demand for the subsidy, and those where the head is Black or 62 or older, are particularly
affected by this event.
II. Program Background
The project-based Section 8 program was created under the Housing and Community
Development Act of 1974. Under this program, property owners signed a Housing Assistance

44
Payment (HAP) contract with HUD, which stipulated that low-income tenants paid 25 percent of
their incomes towards rent, and the government paid each owner the difference between that
amount and the HUD-determined contract rent. In 1998, this was modified to require tenants to
spend 30 percent of their incomes on rent, with HUD continuing to pay the balances (Begley et
al. 2012). Over 1.2 million units of affordable housing were developed through the Section 8
program. In 1983, President Reagan ’s Commission on Housing expressed concern that the
project-based Section 8 program allowed owners to inflate their costs. No funds have been
authorized since then for new project-based Section 8 contracts for new construction or
rehabilitation.
HUD requires owners who used its financing or insurance to agree to a prepayment
restriction and affordability period of at least 20 years. Contract renewal under this program is
optional for owners, but if an owner requests a renewal, HUD must comply with the request,
subject to available appropriations (Begley et al. 2012). The Housing Opportunity Program
Extension Act of 1996 permits owners to prepay their mortgages or terminate their mortgage
insurance, and authorizes HUD to provide vouchers to tenants under certain “ e li g ibi li t y e v e nts.”
The goal of this voucher is to act as a protection for tenants and ensure they can remain in place
and spend no more than 30 percent of their income on rent, even if the rent exceeds fair market
rents for the area. Tenants who decide to move could use their vouchers to rent another unit that
is at, or below, fair market rent. This form of support was first referred to as an “ e nha n c e d
vouc he r” when funding was renewed in FY1998. A tenant must be below 95 percent of area
median income in order to qualify for the enhanced voucher. In addition, the EV program
requires that a Public Housing Authority (PHA) approves the tenant, the unit cannot exceed the
family ’s unit size, and the lease must also start within a year of the project-based assistance
contract being terminated.
III. Theory and Empirical Evidence
A broad range of literature analyzes voucher use, whether households who apply for
vouchers move, and where they move. Evidence suggests that the share of households who are
offered a voucher and then lease a unit with the voucher is decreasing over time, with a lease rate
of 81 percent in 1993 (Kennedy and Finkel 1994), 69 percent in 2000 (Finkel and Buron 2001),

45
and 62 percent in 2001 (Shroder 2002). Overall, existing literature suggests that there are three
main factors that could affect whether a household uses a voucher: the demand a household has
for the subsidy, the supply constraints of the market, or search constraints due to household
demographics.
In general we would expect that the lowest-income households have the highest demand
for a voucher subsidy because the subsidy comprises a higher share of their overall income.
Finkel and Burron (2001) find a significant negative relationship between those with income
over 30 percent of the median and voucher use, meaning that those households eligible for a
voucher who have the smallest share of their income comprised by the subsidy are less likely to
use it. For these households, the voucher likely comprises a small share of their incomes and
rents. The administrative burden of using a voucher acts as a cost that outweighs the benefit of
the voucher. Surprisingly, this study does not find any evidence to support the expectation that
those with the highest demand for the subsidy are more likely to use the voucher.
Theoretically, the likelihood that a household will use its voucher is also a function of the
local supply of housing that a household can rent with a voucher. Existing studies test the
relationship between supply and voucher use by using vacancy rates as a proxy for the supply,
under the theory that a lower vacancy rate means fewer units a household can rent with their
voucher, which translates to lower voucher use. Shroder (2002) finds a positive and significant
relationship between the vacancy rate and voucher use. Kennedy and Finkel (1994) find no
significant relationship between vacancy rates and voucher use, and Finkel and Buron (2001)
find a positive and marginally significant relationship between vacancy and voucher use. This
generally suggests that households are more likely to use their vouchers in high vacancy areas
but offer no conclusions about how this subsidy works in areas with high rental demand, which
are also likely higher opportunity neighborhoods.
Finally, household demographics could affect voucher use. For example, older heads of
household may have a more difficult time using their vouchers because they cannot navigate the
private market and HU D ’s approval process as easily as younger heads of household (Finkel and
Buron 2001). In addition, there could be racial discrimination in a market. Some studies suggest
that voucher households experience discrimination based on race (Galvez 2010, Shroder 2002,

46
Popkin and Cunningham 1999). Shroder (2002) found a significant negative relationship
between being Hispanic and voucher use but could not discern the mechanism that drove that
relationship. Overall, variation in voucher use by household characteristics could highlight issues
of discrimination faced by similar unsubsidized households during housing searches.
There are several key studies that look at whether households who receive vouchers use
those subsidies to move from their existing units, and then what types of neighborhoods these
households access. In general we would expect that if the voucher is an opportunity vehicle, then
households will use the subsidy to move to higher opportunity neighborhoods. Jacob and Ludwig
(2012) take advantage of an experiment by the Chicago Housing Authority Corporation and find
that families who expect to receive a voucher often delay moves until they receive the voucher,
but the authors do not explore if and how these moves vary across market factors and household
demographics. Feins and Patterson (2005) analyze a longitudinal dataset of voucher households
from 1995-2002 and find that households with a head of household aged 24 or below are more
likely to use their voucher to move than households where the head is aged 45 or over. They also
find that households where the head is Black are more likely to move when using a voucher and
that only four percent of their sample moves more than once. An analysis of households in the
MTO program finds that during the 10-15-year study period, the average low-income household
moved twice, whereas households who received a traditional housing choice voucher or an MTO
voucher moved three times on average (Kessler et al. 2011). On the whole, households use their
vouchers to access slightly lower poverty neighborhoods (Devine et al. 2003), they often lease a
unit near where they originally lived and then use the subsidy to move to a lower poverty
neighborhood at a later point (Eriksen and Ross 2013), and the small fraction of households who
make multiple moves tend to continuously move to slightly lower poverty neighborhoods
(Eriksen and Ross 2013; Feins and Patterson 2005).
The households in this study are different from the population in the existing literature for
several key reasons. First, as previously stated, households in our study do not apply for
vouchers; they are offered vouchers when either their owner or HUD ends the place-based
subsidy. As a result, these households are not altering their behavior, including altering move
decisions, in anticipation of a voucher. Households subject to owners opting out face similar
choices as those in a traditional housing choice voucher program; they can lease their existing

47
unit or move to another unit or neighborhood. In general, we expect these households to move
only if they can access a higher quality neighborhood or unit. In the case where HUD terminates
the contract, tenants are forced to move in order to use their vouchers. Therefore, their choices
are more constrained than those of typical voucher households because they cannot lease their
existing units. Finally, evidence suggests that owners exit subsidy programs in areas with higher
price appreciation (Reina and Begley 2014) and, as found in chapter 2, in neighborhoods where
conditions overall are improving. Therefore, the calculation about whether to move is different
from the calculation of many of the tenants in previous studies because they live in an improving
neighborhood and accessing an opportunity neighborhood may mean using their vouchers to stay
in place.
IV. Data and Methods
This paper employs a national database of all tenants who lived in properties where a
project-based Section 8 contract ended between 1996 and 2010, including whether these
households subsequently used a voucher. This database was developed by first identifying all
properties where Section 8 contracts ended and the years in which these contracts ended. These
end-dates are affected by multiple factors, including the presence of other subsidies (Reina and
Williams 2012). These data were then merged with tenant-level data provided by HUD on every
tenant that lived in a property at the time that the contract ended. Finally, those data were
combined with tenant-level data provided by HUD on the universe of tenants who received
vouchers between 1996 and 2010. As a result, it identifies who lived in a property when the
contract ended, and then whether and where they used a voucher at any point after the contract
ended. The analysis focuses specifically on the outcomes of tenants who lived in properties
where the contract ended between 2002 and 2010. We focus on these dates because HUD
updated their databases on multifamily properties in 2001, and there is concern that information
on tenants was lost during this conversion.
1
Between 2002 and 2010, the contracts on over
40,000 units ended because of owner opt-outs. The contracts on over 11,000 units ended because

1
Mark D. Shroder, Associate Deputy Assistant Secretary in the Office of Research, Evaluation, and Monitoring at
HUD who read and discussed this paper at a conference, and is a point of contact for these data, informed us of this
reality and firmly believes that using these data before 2002 would bias our results due to missing data.

48
of foreclosures, and the contracts on over 11,000 units ended because building conditions did not
meet HU D’ s housing quality standards.
The database used in this paper includes extensive tenant data including the age, race,
sex, and income level of each head of household as well as the number of dependents and market
data about the neighborhoods and cities in which these properties are located, specifically
vacancy and poverty rates. Table 1 shows the observable characteristics of tenants who lived in
properties where the contracts ended and compares those to the average characteristics of tenants
who lived in all project-based Section 8 properties during the same period according to HUD ’s
Picture of Subsidized Housing (POSH) database. Several key differences stand out. First, the
tenants in properties where the contract ended tend to be poorer than the average tenant who
lived in a unit where a project-based Section 8 subsidy ended. In addition, the share of units
where the head of household is Black is much higher in properties where the contract ended than
the overall universe of Section 8 properties.
This study then tests whether household demand, market supply, and demographic
factors predict who uses a voucher and whether they move. This paper estimates the demand for
a voucher subsidy in several ways. First, the amount of subsidy provided by a voucher is a
function of two factors: household income and the rent charged by a property owner to rent a
unit (which is capped by HUD as a share of the HUD-determined fair market rent). The project-
based Section 8 program requires that a household spend no more than 30 percent of their
income on rent. HUD pays the difference between the tenant payment and the rent for the unit.
This paper assumes that households have a higher demand for this subsidy if they have a larger
share of their rent paid by the subsidy, and vice versa.
Existing studies test the relationship between income and voucher use with household
wage as the variable of interest. In this paper, we calculate a household ’s effective income —
meaning their income plus the monthly rental subsidy —and then determine the share of their
effective income that is comprised by the HUD rental subsidy. Household demand for this
subsidy likely increases when it comprises a larger share of a household ’s effective income, and
vice versa. As a result, we expect those households with a large share of their effective monthly
income derived from the HUD subsidy to be more likely to use the voucher because they value

49
the subsidy more and vice versa. We also expect that households with a higher share of their
effective income comprised by the subsidy and those on fixed forms of income, such as welfare
or social security, will be less likely to use their voucher to move. If these households do move,
we expect they will move to a lower quality unit or neighborhood because of their financial
constraints. We define neighborhood as the census tract where a household lives, and use tract-
level poverty rate as our metric for assessing neighborhood quality.
As previously discussed, there are several supply factors that could affect voucher use.
Households who live in areas with very high vacancy rates may be less likely to use their
voucher for two reasons. First, high vacancy could reflect the poor quality of the units in the
market and therefore fewer units are likely to meet HUD ’s housing quality standards. Areas with
high vacancy also likely have lower rent, which decreases the value of a HUD rental subsidy. If
an owner has the choice to rent their unit to a market-rate tenant or a voucher tenant at the same
rent, the owner will prefer the market-rate tenant because accepting a voucher tenant involves
administrative costs. As a result, households in areas with the most housing demand as reflected
in lower vacancy rates, may have lower odds of using a voucher because of market competition
for units. In general, we expect that households who do use their voucher are more likely to
move if they live in a neighborhood with a low vacancy rate because of their need to find a unit
that accepts the subsidy. Existing studies test the relationship between vacancy rates and voucher
use with the vacancy rate as a continuous variable. This paper tests the relationship at the tale
ends of the vacancy spectrum because that is likely where households face the highest constraint
in their choice set.
This paper also tests whether voucher use and moves are affected by household
characteristics including race, age, and number of dependents. We expect that voucher use could
be lower for those 62 or older, because of difficulty navigating the private rental market. In
addition, households with dependents may be less likely to use a voucher because it is difficult to
find units with a larger bedroom count on the private market. Theoretically, voucher use should
not vary based on race. The number and location of moves may vary for households with
dependents because this event provides an opportunity to find a unit that better suits their
preferences, such as a desire for better schools.

50
Descriptively, we find that fewer than 50 percent of households use their voucher. The
unadjusted share of households using a voucher (in Table 2) appear to be lower for those with
the highest and lowest demand for the subsidy, and those who are elderly, male, or lived in a
property where HUD terminated the project-based Section 8 contract. There appears to be little
variation in voucher use based on the number of dependents and local vacancy rate. But, there
does appears to be some variation in voucher use by race and ethnicity, with a higher share of
heads of households who are Asian and Hispanics using the voucher, a lower share of
households where the head is Native American using it, and those where the head is Black being
just below the mean voucher use rate. We find that 56 percent of voucher households voucher
move to another tract. Of those who move, 72 percent move once to another tract, 21 percent
move twice to two different tracts, and seven percent make three or more moves to different
tracts. We next test how these results vary over the household demand, market supply, and
demographic factors previously discussed by using a series of logistic regressions for whether a
household uses the voucher and moves, and a standard OLS model to determine the poverty rate
of the neighborhoods where households move.
V. Analysis

Voucher Use:
We run several logistic regression models that test the odds of tenant i using a voucher
based on the factor previously discussed. The base model is as follows:

ln(
p
i
1 p
i
) 

0
 
1
SR
i
 
2
A
i
 
3
D
i
 
4
O
i
 
5
Y
i
 
6
LV
i
 
7
HV
i
 
8
B
i
 
9
A
i
 
10
NA
i
 
11
H
i
 
12
M
i
(1)

Where the dependent variable is a dichotomous variable of whether tenant i uses a voucher; SR is
the share of a household i ’s rent that is paid by the HUD subsidy; A is the share of a household
i ’s monthly effective income that comes in the form of a rental subsidy from HUD; D is a count
of the number of dependents in household i; O is a dichotomous variable of whether the head of
household i is age 62 or older; Y is a dichotomous variable of whether the head of household i is
aged 25 or younger; LV is a dichotomous variable of whether the property where household i
lived was in a market with a vacancy rate at the lowest fifth percentile of our sample, which is

51
below a four percent vacancy rate; HV is a dichotomous variable of whether the property where
household i lived was in a market with a vacancy rate at the highest fifth percentile of our
sample, which is above eleven percent; B is a dichotomous variable for whether the head of
household i is Black; A is a dichotomous variable for whether the head of household i is Asian;
NA is a dichotomous variable for whether the head of household i is Native American; H is a
dichotomous variable for whether the head of household i is Hispanic; M is a dichotomous
control variable of whether the head of household i is male. This model is then re-specified to
include time-fixed effects and a control for the PHA that administers the voucher.

The results from this model are in Table 3. We see that, as expected, a higher share of a
house hold’s rent paid by the Section 8 subsidy is associated with higher odds of voucher use.
Interestingly, there is an inverse relationship between the share of a house hold’s income that is
comprised of the rental subsidy and the odds of using a voucher. These relationships could be
affected by those at the tail ends of the spectrum and we explore this dynamic in a model shown
later in this section. Our supply factors show that those households in areas with the lowest
vacancy rates have higher odds of using their voucher and those in areas with the highest
vacancy rates have lower odds of using a voucher, but this relationship does not hold when
controlling for who administers the voucher. The model also shows that a higher number of
dependents decreases the odds of a household using their voucher. Finally, households where the
head is 62 or older have much lower odds of using a voucher when controlling for other factors,
as do households where the head is Black. Households where the head is Asian have higher odds
of voucher use.
We then restrict each regression based on the reason the contract ended. The reason why
the project-based Section 8 contract ends could lead to variation in voucher use. There are three
reasons why a contract ends: an owner opts out, HUD terminates the contract due to foreclosure,
or HUD terminates the contract due to substandard building conditions. We expect that
households who live in a property where the owner opts out may be more likely to use their
voucher because there is a chance these households will be able to continue to lease their existing
unit with their enhanced voucher. Households who live in a property where HUD terminates the
mortgage may be less likely to use their voucher because HUD requires them to move. In
addition, households that remained in distressed properties until the contract was finally

52
terminated may be those with the fewest housing alternatives. The results for these models are in
Table 4. Most of the estimated relationships remain the same as our previous model in direction,
but there are some differences with respect to magnitude and significance. Of particular note is
that the race variables are only significant for properties where the owner opts out, and lower use
rates are associated with both low and high vacancy rates when a HUD contract is terminated
due to foreclosure.
We expect the demand for a housing subsidy may be significantly stronger and weaker at
the tail end of the distribution of our demand variables, and that these tails could be driving the
estimated relationships. For example, a household will have a higher level of demand for a
voucher when the HUD subsidy payment covers all, or almost all, of the rent. In addition, if the
subsidy covers only a small fraction of the rent, then a household will have much less demand
for the subsidy than those in the middle of the distribution. Following the same logic, those
households where the HUD subsidy comprises almost all of their effective income will have
more demand for the subsidy than those in the middle of the distribution, and those where the
subsidy comprises a very small fraction of their income will have much less demand. As a result,
we test the likelihood of household i using a voucher if the HUD subsidy comprises less than 10
percent of their rent payment, and the odds of using the voucher if the subsidy covers over 90
percent of their rent payment. We also test similar variables for effective income in this model.
The results from this test are in Table 5. When we do this, we find that the relationship between
the share of a house ho ld’s rent that is assisted and voucher use becomes more complex.
Households with the highest and lowest share of their rent assisted have lower odds of voucher
use, as is the case with the share of a house hold ’s income that is comprised by the subsidy.
Household moves:
We then use a multinomial logit model to test the odds that a household remains in the
same tract, moves twice, or moves three or more times compared to only moving once, based on
the same characteristics as follows:

53

ln
Pr(Y J)
Pr(Y J
'
)
 
0
 
1
HR
it
 
2
LR
it
 
3
HA
it
 
4
LA
it
 
5
D
it
 
6
O
it
 
7
Y
it
 
8
LV
it


9
HV
it
 
10
B
it
 
11
A
it
 
12
NA
it
 
13
H
it
 
14
M
it
 
15
I
it
 
16
F  
17
V  
18
T(2)

Where HR is a dummy that is 1 if 90 percent or more of a household ’s rent is paid by the HUD
subsidy for household i in year t; LR is a dummy that is 1 if less than 10 percent of a household ’s
rent is paid by the HUD subsidy for household i in year t; HA is a dummy that is 1 if 90 percent
or more of a household ’s income is comprised by the HUD subsidy for household i in year t; LA
is a dummy that is 1 if 10 percent or less of a household ’s income is comprised by the HUD
subsidy for household i in year t; I is a categorical variable for the primary source of a
household ’s income, where 0 is wage, 1 is welfare, 2 is social security or a pension, 3 is child
support, 4 is unemployment and 5 is some other source of income; F is a dummy that is 1 if the
project-based Section 8 contract ended due to foreclosure; and V is a dummy that is 1 if the
contract ended because the building was in violation of HUD ’s housing quality standards. The
results from this test are in Table 6. One limitation of these data is that it is difficult to tell
whether households who stay in the same tract remain in their original unit, used their voucher to
move to another unit in the same building, or moved to another building in the same tract.
In this model we find that, in general, those households with the highest share of their
rent assisted have lower odds of remaining in the same tract, and higher odds of moving multiple
times, as do those households in the highest vacancy areas, those with dependents and those
where the head is Black or below the age of 26, or rely on welfare or child support as their
primary source of income. In contrast, those households where the head is 62 or older have
higher odds of not moving as do households in areas with low vacancy rates. Finally, those
households in properties where HUD terminated the Section 8 contract due to poor building
conditions have higher odds of moving once, whereas those where the contract was terminated
due to foreclosure have higher odds of moving one or more times.
Where a household moves:

Finally, we run a standard OLS model to look at the poverty rate of the neighborhoods
where households who use their voucher move:

54

y  
0
 
1
MO  
2
HR
it
 
3
LR
it
 
4
HA
it
 
5
LA
it
 
6
D
it
 
7
O
it
 
8
Y
it
 
9
LV
it


10
HV
it
 
11
B
it
 
12
A
it
 
13
NA
it
 
14
H
it
 
15
M
it
 
16
I
it
 
17
F  
18
V  
19
T(3)

The results from this test are in Table 7. These results suggest that neighborhood poverty rate
decreases as households make more moves to other tracts. However, households where the head
is Black or Hispanic are associated with higher poverty rate tracts even after we control for the
number of moves and other supply and demand factors. In addition, those households with
welfare or social security as their primary source of income are associated with slightly higher
poverty neighborhoods.
VI. Discussion
The federal government financed over a million units of privately owned affordable
housing through its project-based Section 8 program. Households living in these properties are
offered a voucher as a form of safety net when the subsidy contracts on their property ends. In
this study, we find that less than 50 percent of households lease a unit with this voucher, and that
voucher use, whether a household moves, and where a household moves, varies based on
demand, supply, and demographic factors.
Those households with the highest demand for the subsidy, as measured by the share of
their income comprised by the subsidy and the share of rent paid by the subsidy, have lower odds
of using their voucher. These households could be those with the fewest resources who have the
least time and resources to navigate the private rental market and HU D ’ s approval process. In
addition, households where the heads are 62 or older or Black have lower odds of using their
voucher, which suggests that the loss of a subsidy could impact households differently based on
demographics. For older heads of household, we suspect that lower odds of using the voucher
can be attributed to difficulty navigating the private rental market with it. Descriptively, there
was little evidence that black heads of household had lower voucher use rates. However, when
controlling for market supply and household demand factors that should explain voucher use
rates, we find that black heads of household are associates with lower odds of using a voucher
across all specifications of the model. This finding reinforces concerns about discrimination in
the rental market.

55
Households with dependents also have lower odds of using their voucher, which could
highlight potential search challenges as there are fewer units with larger bedroom counts thus
increasing search costs and competitiveness for those units. Finally, those households who lived
in properties where HUD terminated the contract have lower odds of using the voucher. Again,
this is a concern because these households lived in a distressed property for some time before
HUD took action to end its contract, which means they likely have a high demand for the subsidy
and the fewest housing alternatives and could point to difficulty navigating the private market
with a voucher. Those households with the least demand, as measured by the smallest share of
their income comprised by the subsidy or rent paid by the subsidy have lower odds of using a
voucher, and this could point to a welfare-improving scenario for these households because they
are no longer subject to a lock-in effect.
For those who do use a voucher, this event could be welfare equivalent if they are able to
rent the same unit at the same level of subsidy. We find that 44 percent of households who use a
voucher in our sample remain in the same tract. Households who live in the properties where the
owner opts out have the option to lease their existing unit, and our model finds that this group
has higher odds of remaining in the same tract, which means some of these households likely
remain in the same unit, although we cannot know for sure. We find voucher user households
where the head is 62 or older and those in areas where the low vacancy rates have higher odds of
remaining in the same tract. Again we suspect many of these households are remaining in the
same unit due to search barriers for a different unit, in which case this event at least appears to be
welfare equivalent in the short term. However, we find that those households with the highest
share of their rent assisted have lower odds of remaining in the same tract, and higher odds of
moving multiple times, as do those: with welfare or child support as their primary source of
income; in the highest vacancy areas; with dependents; and where the head is Black or below the
age of 26.
Several of these findings are concerning. For example, those with welfare or child
support as their primary source of income have an income that is largely fixed, which makes
moving costs more burdensome. In addition, those in areas with extremely high vacancy likely
have a difficult time finding a unit nearby that will meet HU D’ s housing quality standards, and

56
those with dependents likely have a difficult time finding a unit that will both accommodate their
family size and accept a voucher.
When we look at where households move, we do find a silver lining, which is that each
move is associated with a lower poverty tract. This suggests that the conversion from a place-
based subsidy to a voucher may allow some households to improve their welfare by moving to a
lower poverty neighborhood. However, such gains are not equalizing across all households; we
find that households where the head is Black are associated with living in higher poverty tracts
than households where the head is of another race but similar income level.
Overall, this study finds that over 50 percent of households in a property where the
project-based Section 8 subsidy ends do not use the voucher they are offered and thus suffer a
substantial income shock. There is also variation in voucher use and moving patterns based on
household demand, market supply, and demographic characteristics, which suggests that
vouchers may not be an effective safety net tool for all households. While this event does result
in a small fraction of households moving to lower poverty neighborhoods, it negatively impacts
many households, particularly those with the highest demand for the subsidy and those where the
head is Black or 62 or older.
These findings are important because more units will exit the place-based subsidized
housing stock going forward and more tenants will be affected by this event. More broadly, this
points to concern about the effectiveness of a voucher as a safety net tool that protects low-
income households from housing instability. This is particularly worrisome because the
increasing lack of affordable units in cities means low-income households are facing more
market pressure that can result in housing instability and moves away from opportunity. As a
result, in many situations, tenants may be looking for vouchers to stay in place. Ultimately, this
analysis improves our understanding of the welfare loses faced by households when existing
place-based subsidies end, how vouchers function across household demand, market supply, and
demographic factors, and potential limitations of this tool as a safety net and opportunity vehicle.

57

VII. Figures and Tables

Table 1: Average household characteristics of tenants in sample versus HUD ’s Picture of
Subsidized Housing (POSH) database

Contract Ended HUD POSH
Tenant payment per month $220 $260
HUD subsidy per month $513 $633
Average household income $8,795 $10,406
Income Level (in percent) Contract Ended HUD POSH
< 5,000 34 12
5,000-9,999 37 40
10,000-14,999 16 28
15,000-20,000 7 13
> 20,000 5 8
Age of head of household (in %) Contract Ended HUD POSH
< 25 14.1 7.7
62 + 26.6 56.8
Race (in %) Contract Ended HUD POSH
Black 49 26
Native American <1 <1
Asian 3 4
Hispanic 11 12

58
Table 2: Voucher usage based on key variables
Share of rent assisted Share using a voucher
Below 10% Rent Assisted 0.31
Over 90% Rent Assisted 0.48
Share of effective income assisted Share using a voucher
Below 10% Income Assisted 0.36
Over 90% Income Assisted 0.45
Number of dependents Share using a voucher
0 0.49
1 0.47
2 0.49
3+ 0.49
Market Share using a voucher
Vacancy at bottom fifth 0.50
Vacancy at top fifth 0.48
Age of head of household Share using a voucher
< 25 0.79
25-61 0.78
62 + 0.25
Race of head of household Share using a voucher
White 0.47
Black 0.48
Asian 0.50
Native American 0.42
Hispanic 0.52
Sex of head of household Share using a voucher
Female 0.52
Male 0.36
Reason contract ended Share using a voucher
Opt-out 0.53
Foreclosure Termination 0.33
Enforcement Termination 0.38

59
Table 3: Odds of a household using a voucher, converted to odds ratios

Base Base + Year Base + Year + PHA
Coefficient Odds ratio

Coefficient Odds ratio

Coefficient Odds ratio
Share of rent assisted
1.638 5.145 *** 1.698 5.463 *** 2.302 9.994 ***
Share of income comprised of
rental subsidy -2.016 0.133 *** -2.049 0.129 *** -2.298 0.100 ***
Number of dependents
-0.404 0.668 *** -0.399 0.671 *** -0.337 0.714 ***
Vacancy rate
Lowest fifth percentile
vacancy (<=4%) 0.188 1.207 *** 0.156 1.169 *** 0.133 1.142
Highest fifth percentile
vacancy (>=11%) -0.294 0.745 *** -0.242 0.785 *** 0.210 1.234
Age of head of household

<26
0.047 1.048 0.062 1.064 0.182 1.200
62+
-3.372 0.034 *** -3.387 0.034 *** -3.846 0.021 ***
Race of head of household

Black
-0.331 0.718 *** -0.343 0.710 *** -0.298 0.742 ***
Asian
0.458 1.581 *** 0.466 1.594 *** 0.295 1.343 ***
Native
-0.437 0.646 *** -0.432 0.649 *** -0.110 0.896
Hispanic
0.282 1.326 *** 0.261 1.298 *** 0.026 1.026
Sex of head of household

Male
-0.394 0.674 *** -0.395 0.674 *** -0.291 0.748 ***
Observations 65,994 65,994 65,994
Pseudo R
2
0.320 0.324 0.414
Significance: *** < 1 percent; ** < 5 percent; * <10 percent

60
Table 4: Odds of a household using a voucher by the reason the subsidy ended, converted to odds ratios

Opt-out Fail- out Foreclosure
Coefficient Odds ratio Coefficient Odds ratio Coefficient Odds ratio
Share of rent assisted
2.803 16.494 *** 1.534 4.637 *** -1.604 0.201 ***
Share of income comprised of
rental subsidy -2.793 0.061 *** -1.709 0.181 *** -0.184 0.832 ***
Number of dependents
-0.399 0.671 *** -0.302 0.739 ***
Vacancy rate
-3.398 0.033 ***
Lowest fifth percentile
vacancy (<=4%) -0.149 0.862 0.297 1.346 -2.809 0.060 ***
Highest fifth percentile
vacancy (>=11%) -1.628 0.196 * -0.065 0.937
Age of head of household
0.064 1.066
<26
0.300 1.350 *** -0.004 0.995 -3.684 0.025 ***
62+
-3.897 0.020 *** -4.272 0.014 ***
Race of head of household
-0.042 0.959
Black
-0.247 0.781 *** -0.014 0.986 -0.102 0.903
Asian
0.290 1.336 *** 0.0697 1.072 -0.074 0.929
Native
-0.103 0.902 -0.095 0.909 -0.221 0.802
Hispanic
0.087 1.091 -0.136 0.873
Sex of head of household
-0.403 0.668 ***
Male
-0.283 0.754 *** -0.110 0.896 1.701 5.479 ***
Observations
40,051 11,207 11,146
Pseudo R
2
0.404 0.475 0.435
Significance: *** < 1 percent; ** < 5 percent; * <10 percent

61

Table 5: Odds of a household using a voucher sensitivity analysis, converted to odds ratios
Coefficient Odds Ratio
Share of rent assisted

Below 10% Rent Assisted
-0.349 0.705 **
Over 90% Rent Assisted
-0.396 0.673 ***
Share of effective income assisted

Below 10% Income Assisted
-0.702 0.496 ***
Over 90% Income Assisted
-0.200 0.819 ***
Number of dependents
-0.342 0.710 ***
Vacancy rate

Lowest fifth percentile vacancy (<=4%)
0.693 2.000 **
Highest fifth percentile vacancy (>=11%)
-0.003 0.997
Age of head of household

<26
0.145 1.156 ***
62+
-4.121 0.016 ***
Race of head of household

Black
-0.250 0.779 ***
Asian
0.324 1.383 ***
Native
-0.172 0.842
Hispanic
-0.047 0.954
Sex of head of household

Male
-0.266 0.766 ***
Reason subsidy contract ended

Foreclosure
-0.106 0.899 *
Enforcement
-1.316 0.268 ***
Observations
51,088
Pseudo R
2
0.449
Significance: *** < 1 percent; ** < 5 percent; * <10 percent

62
Table 6: Odds of a household moving, converted to odds ratios

Same tract 2 moves 3+ moves
Share of rent assisted
Coefficient Odds ratio Coefficient Odds ratio Coefficient Odds ratio
Below 10% Rent Assisted
0.340 1.405 *** 0.050 1.051 -0.270 0.763
Over 90% Rent Assisted
-0.263 0.769 *** 0.169 1.184 *** 0.320 1.377 ***
Share of effective income assisted

Below 10% Income Assisted
0.501 1.650 *** -0.110 0.896 -0.060 0.942
Over 90% Income Assisted
-0.205 0.815 *** 0.051 1.053 -0.032 0.969
Number of dependents
-0.264 0.768 *** 0.058 1.060 *** 0.107 1.113 ***
Vacancy rate

Lowest fifth percentile vacancy
(<=4%) 0.335 1.398 *** -0.099 0.906 *** -0.490 0.613 ***
Highest fifth percentile vacancy
(>=11%) -1.004 0.366 *** 0.250 1.284 *** 0.488 1.629 ***
Age of head of household

<26
-0.455 0.634 *** 0.131 1.140 *** 0.243 1.275 ***
62+
0.550 1.733 *** -0.440 0.644 *** -0.783 0.457 ***
Race of head of household

Black
-0.519 0.595 *** 0.332 1.394 *** 0.425 1.530 ***
Asian
-0.564 0.569 *** 0.065 1.067 -0.147 0.863
Native
-0.336 0.715 *** 0.131 1.140 -0.063 0.939
Hispanic
0.226 1.254 *** 0.190 1.209 *** 0.179 1.196 **
Sex of head of household

Male
0.117 1.124 *** -0.270 0.763 *** -0.520 0.595 ***
Source of Income

Welfare
-0.29 0.748 *** 0.280 1.323 *** 0.478 1.613 ***
Social security or pension
-0.18 0.835 *** 0.010 1.010 0.166 1.181 ***
Child support
-0.178 0.837 *** 0.128 1.137 *** 0.177 1.194 **
Unemployment
-0.156 0.856 ** 0.155 1.168 * 0.288 1.334 **
Other
0.046 1.047 0.148 1.160 *** 0.229 1.257 ***
Reason subsidy contract ended

63
Foreclosure
-1.110 0.330 *** 0.316 1.372 *** 0.412 1.510 ***
Enforcement
-1.527 0.217 *** -0.190 0.827 *** -0.155 0.856 ***
Observations
78,167 78,167
78,167

Pseudo R
2
0.201 0.201
0.201

Significance: *** < 1 percent; ** < 5 percent; * <10 percent

64
Table 7: Poverty rate of tract where household live (in percentage points)
Number of moves

1
-3.23 ***
2
-3.95 ***
3+
-4.84 ***
Share of rent assisted

Below 10% Rent Assisted
0.52
Over 90% Rent Assisted
0.01
Share of effective income assisted

Below 10% Income Assisted
-0.11
Over 90% Income Assisted
0.47 ***
Number of dependents
-0.20 ***
Age of head of household

<26
0.02
62+
-0.16
Race of head of household

Black
1.28 ***
Asian
0.09
Native
0.04
Hispanic
0.87 ***
Sex of head of household

Male
0.43 ***
Source of Income

Welfare
0.76 ***
Social security or pension
0.68 ***
Child support
-0.07
Unemployment
-0.42
Other
0.62 ***
Reason subsidy contract ended

Foreclosure
-2.58 *
Enforcement
-1.16
Observations
75,365
R
2
0.025
Significance: *** < 1 percent; ** < 5 percent; * <10 percent

65

Chapter 4:

What happens when a project-based rental subsidy ends? Mobility
and academic outcomes

66
I. Introduction
The federal government manages multiple affordable housing programs aimed at
promoting neighborhood access and protecting low-income households from high rent burdens.
For example, the U.S. Department of Housing and Urban Development (HUD) financed roughly
1.2 million units of housing through the project-based Section 8 program. However, in Los
Angeles County the project-based Section 8 contract already ended on 171 properties containing
over 6,000 units. Nationally, there are over 3,000 properties, containing more than 150,000 units,
where the project-based Section 8 subsidy ended. This research focuses on one aspect of the
welfare implications of this event, which is the impact that a subsidy contract ending has on the
academic outcomes of youth who live in these properties. When a subsidy contract ends it
induces those who live in the property to move. As a result, this research also provides insight
on the relationship between mobility and academic outcomes more broadly.

The number of tenants affected by a federal rental subsidy ending will only increase
going forward. Almost all federally subsidized rental units developed since the 1970s are owned
by private entities that agreed to maintain their property as affordable for only a fixed period of
time. Owners can choose to renew their subsidy, apply for a new one, or convert their property
to market-rate when the initial affordability restriction ends. In Los Angeles County there are
over 50,000 units in properties where an owner is eligible to leave a rental subsidy program, or
will be eligible to do so within the next ten years. The scale and welfare implications of subsidy
contracts ending are both large and unstudied.

This research is the first to use the identification strategy of an owner leaving a subsidy
program to examine the relationship between mobility and academic outcomes. This strategy
helps eliminate concerns about selection bias and unobserved variables that could drive or affect
a move decision. Another important contribution of this research is that we create a database of
all federally subsidized properties ever built in Los Angeles County, including those where a
rental subsidy ended, and combine those data with a panel of student-level data for all 1.7 million
Los Angeles Unified School District (LAUSD) students between 2000 and 2014. These data, and
the methods employed in this chapter, allow us to better understand the effect of a place-based
subsidy ending on the outcomes of children in these properties. This is particularly important

67
because cities like Los Angeles have witnessed unprecedented rent increases and historically low
vacancy rates over the last ten years (Green et al. 2015). Such rental market pressures create
increased incentive for owners to exit federal rental programs despite high demand for affordable
units from low-income households. Going forward, these data allow for further analysis of the
relationship between subsidized housing and academic outcomes and how this has changed over
time and across subsidy programs.

This research finds a significant relationship between a subsidy contract ending and a
student moving, but no relationship between this move and worse academic outcomes.
Specifically, we do not find any relationship between the move resulting from a Section 8
contract ending and a change in the share of days absent or the likelihood of a student being
suspended. This paper does find a marginally significant relationship between the move induced
by a Section 8 contract ending and improved test scores. Combined, these result show that a
subsidy contract ending does not result in worse academic outcomes, and may actually lead to
higher test scores for some students.

II. Background and Theory
Mobility rates vary based on income, with students in low-income households being
more likely to move and switch schools than those in higher-income households (Reynolds et al.
2009). Current research suggests that higher rates of mobility are associated with poorer
academic performance. Mobility affects outcomes for a number of reasons, including that: it
creates instability during, and after, the move; it may change the school that the student attends
and thus force the student into a new environment and social network; it may cause discontinuity
in the curriculums to which the student is exposed. Despite the abundance of literature that
explores this topic, there remains no conclusive evidence about the relationship between mobility
and academic outcomes.

Studies find that higher rates of mobility are associated with a higher likelihood of
repeating a grade (Alexander et al. 1996; GAO, 1994; Reynolds et al. 1996; Simpson and Fowler
1994; Wood et al. 1993) and dropping out of school (Astone and McLanahan, 1994; Gasper,
Deluca and Estacion, 2012; Ou & Reynolds, 2008; Reynolds et al. 2009; Rumberger and Larson,

68
1998; South, Haynie and Bose 2007; Swanson and Schneider 1999; Teachman, Paasch and
Carver 1996). In addition, those who move within school districts, and those who move
frequently have lower levels of academic achievement. These studies find that the effect of
mobility on academic outcomes is influenced by many factors, not the least of which is family
characteristics.

All of these studies are unable to establish a causal relationship between mobility and
academic outcomes (Hanushek et al. 2004; Mehana and Reynolds 2004; NRCIM 2010). There
are several reasons why establishing a causal relationship is difficult. First, many of the early
mobility studies suffer from selection bias, where they compare students who move to those who
do not move, when the two groups are not equal in expectation (Hanushek et al. 2004; Mehana
and Reynolds 2004; Reynolds et al. 2009). Another challenge to the current literature is that
these studies cannot either identify and/or control for why a household moves. A household may
move for personal reasons, such as a divorce, or financial reasons, such as an increase or
decrease in income. The move could also be school-related, where the family moves to access a
better school or due to a child’s behavioral problems at their current one. As a result, alternative
mechanisms may explain the relationship between moving and academic outcomes in these
studies. This study attempts to overcome the limitations of past research and establish a causal
relationship between mobility and academic outcomes by taking advantage of an exogenous
housing shock, which comes from the loss of a building-level federal rental subsidy.

The federal government developed, or financed the development of, over 100,000
affordable rental units in Los Angeles County. Beginning in the 1990s, the affordability
restrictions on many of the privately owned subsidized properties began to expire. For the first
time, owners could renew their subsidy with HUD, or not renew the subsidy otherwise known as
“opting-out.” Between 1996 and 2010, there were 171 properties, with roughly 6,335 units
where the owner chose to opt out of a HUD subsidy in Los Angeles County. Research suggests
that that owners in areas with higher price-appreciation are more likely to leave subsidy
programs (Reina and Begley 2014). As seen in chapter 2, units that exited the project-based
Section 8 program between 2000 and 2010 were in neighborhoods that were improving. There
are over 50,000 privately owned subsidized rental units in Los Angeles where the owner is

69
currently eligible to opt out or will be eligible to do so within the next ten years. Many of these
properties are in neighborhoods that meet the same criteria as those where subsidies have already
ended and we expect that at least some will exit the subsidized housing stock. This event has
potentially large implications for the welfare of households in these properties, particularly for
the academic outcomes of the kids who live in these properties and must move. An owner’s
decision to opt-out provides a unique opportunity to explore the relationship between mobility
and academic outcomes because this event induces a move. This source of exogenous variation
allows us to overcome the challenge in most existing studies, where mobility may be a product
of some unobserved difference between those who move and those who do not move.

III. Data, Sample, and Identification Strategy
One of the key contributions of this research is that we create a robust database of all
federally subsidized properties in Los Angeles County, and combine it with a detailed panel of
all 1.7 million students in the LAUSD between 2000 and 2014. The subsidized property database
includes detailed information on unit counts as well as when the federal subsidy started, and if
and when that subsidy ended. The LAUSD data includes annual information on student
demographics and academic performance. Combining these data allows us to identify which
students lived in a subsidized property when the subsidy ended, and the impact of this event on
their performance at school.

The identification strategy in this analysis relies on the premise that a HUD project-based
Section 8 contract expiring is an exogenous shock to those households who live in these
properties. The exogenous shock of a contract termination catalyzes a move, which eliminates
concerns about selection bias and unobserved variables that affect existing estimates of moving
and academic outcomes. The decision to end a Section 8 contract is either made by HUD or a
building owner, and not the households who live in these properties. In order for this event to not
be exogenous, tenants must affect the owner’s decision to exit the program. Thus far there is
empirical evident that only market factors predicts the likelihood of owners leaving subsidized
housing programs, and no evidence to suggest that tenants affect this decision (Abt and
Econometrica 2006; Reina and Begley 2014). In order for tenants to collude to get the owner to
leave a subsidy program they would need to know if and when an owner’s contract is going to

70
expire, which is information that government officials and housing advocates often cannot easily
obtain (Reina and Williams 2012). Alternatively, there could be something about particular
tenants in a property that affects an owner’s decision to opt-out of the program. Existing studies
find no relationship between tenant characteristics and opt-outs. However, it is feasible that there
are at least some properties where the tenants influence an owner’s decision to leave the
program. This would be a building specific dynamic, which is something we control for in our
model

A central part of the identification strategy in this paper is determining which students
lived in a project-based Section 8 property, and which lived a in a property when the subsidy
expired. We identify the 593 properties that ever received a project-based Section 8 contract and
the 171 properties where the contract ends, and then match students to properties as highlighted
in the technical appendix. We identify 183 students who live in a property when the subsidy
contract ends and find that 57 of those students moved at least once within the year the subsidy
contract ended. There are two main issues that could affect tis identification strategy. First, if we
do not correctly match students to properties the results will be biased by contamination across
the subsidized treatment, and non-subsidized groups. To address this concern, an overview of our
data cleaning and matching process is provided in the appendix, and we also validate our
matches in a robustness check. Second, some students who experience a sudden move may not
immediately update their address because they are looking for another permanent address or for
other reasons. If this happens, we underestimate the likelihood that a student moves due to a
contract ending and misidentify our key mechanism. We examine trends in moves over time to
test for issues related to the updating of addresses and find that none of the students in our
treatment group moved in the subsequent two years after the subsidy contract ended. This means
that while there could still be concerns about the address data, we are at least capturing that
initial move due to subsidy contract ending and there does not appear to be any lag in a change
of address that would affect our assignment.

One challenge to the mechanism in this paper is that tenants are offered a voucher when a
project-based Section 8 subsidy ends, which could affect a household’s move decision. As show
in 3, only 48 percent of households who live in properties where the subsidy ends use their

71
voucher. We do not know what happens to those who do not use the voucher, but we do know
that 56 percent of households who use their voucher move to another tract. We also know that
households with dependents have lower odds of using the voucher, but if they do use the voucher
they tend to move one or more times. Within this context, moves with a voucher are associated
with lower poverty tracts, and households with dependents are also associated with living in
lower poverty tracts. This means that a voucher could affect a household’s move decision. It
also means that households in our sample may move to lower poverty neighborhoods with better
schools than they would have been able to afford or access without a voucher, and such moves
could lead to improved academic outcomes. In aggregate, this means that households in this
sample could have a different choice set than similar low-income households facing an
exogenously induced move without the offer of a voucher as a safety net.

Table 1 shows how the demographics for the overall LAUSD population compare to our
treatment group (students who lived in a project-based Section 8 property where the owner did
not exist) and our control group (students who lived in a project-based Section 8 development
when the contract ended). Some key differences emerge. First, the share of students who are
Black is larger in our control group than the overall LAUSD population, and the share of
students who are Black increases further for the treatment group. This is consistent with findings
in chapter 3, which finds that the share of households who are Black is higher in properties
where the project-Section 8 contract ends than those where the subsidy is active. The table also
shows that the share of students in the treatment and control group who are foreign born is higher
than the overall LAUSD population, but the share categorized as proficient in English is higher
in both the treatment and control group than it is in the overall LAUSD student population.

IV. Methods and Analysis
This research tests the impact of the end of a project-based Section 8 contract, and
mobility more broadly, on academic outcomes. To do this, we focus on three outcomes: the
share of days that a student is absent; a student’s test scores; and whether a student is suspended.
The share of days absent is determined by dividing the product of the number of official schools
days in year t minus the number of days that student i was present in academic year t by the
number of official schools days in academic year t. We standardize test scores by first

72
calculating the average of the Math and English standardized test scores for all students in a
grade in a given year and then create a z score that established how student i’s test scores
deviates from the mean of all students who were in the same grade and took the same test in that
year.

The initial descriptive statistics of our mechanism and outcomes of interest for the
population in our sample (Table 2) show that students who live in active and expired Section 8
properties have a higher annual probability of moving, and are absent more days on any given
year than the average LAUSD students. Students in expired properties, on average, tend to have
lower standardized test scores than the overall project-based Section 8 population. When
focusing only on the treatment group (Table 3) we find that this population has the same
probability of moving before and after a subsidy contract ends. This number hides the fact that
the odds of a household moving increases to 20 percent in the year the subsidy ends, but then
drops to zero for the subsequent two years. This spike suggests that a subsidy contract ending
does initiate a move. Other important things to note is that the average number of days absent
decreases and test scores decrease quite dramatically for the treatment group post subsidy
expiration.

This paper uses a difference in differences framework that takes advantage of the
exogenous policy shock provided by the expiration of a HUD subsidy contract as shown in
Equation 1.

Y
it
= α + β SIP
it
+ β OM
it
+ SC
it
+ G
it
+ P
it+
ζ
t

+
ω
t
+ ε
ist
[1]

In Equation 1, Y
ist
represents whether student i moved in year t. SIP
it
is a dichotomous
variable indicating whether student i lived in a property where the subsidy contract ended that
becomes one when the contract ends; and OM
it
is a dichotomous variable for whether student i
made a move that was not in the year a subsidy contract ended. The model controls for the
school that student i attended in year t (SC
it
) because school quality can affect outcomes. In
addition, it controls for the grade-level of student i in year t (G
it
) because there could be variation
in outcomes due to the grade, particularly because students naturally transition schools after
certain grades. This model also controls for the subsidized property where student i lived in year

73
t (P
it
). This property control should capture time-invariant building characteristics, which can
include unobservable characteristics of tenants specific to that property that affect an owner’s
decision to opt out. This control helps to address concerns about the exogeneity of the event and
any building-specific dynamics that may affect outcomes. Finally, we include a student fixed
effect (ω
t
) to control for time invariant student characteristics that could affect outcomes, such as
race and country of origin, and a year fixed effect (ζ
t
), and an error term (ε
ist
).

In the next specification we include another level to the model, which comes in the form
of an interaction between our initial treatment and post variable with a variable that signals
whether a household moved in the year the contract ends. This model isolates the effect of the
move as follows:

Y
it
= α + β SIP
it*
M
it
+ β OM
it
+ SC
it
+ G
it
+ P
it+
ζ
t

+
ω
t
+ ε
ist
[2]

In Equation 2, Y
ist
represents the outcome for student i in year t. SIP
it
is a dichotomous
variable indicating whether student i lived in a property where the subsidy contract ended that
becomes one when the contract ends; M
it
is a dichotomous variable that becomes one if student i
moved in the year that the subsidy contract ended; and OM
it
is a dichotomous variable for
whether student i made a move that was not in the year a subsidy contract ended. In addition, the
model controls for the school that student i attended in year t (SC
it
), the grade-level of student i
in year t (G
it
), the subsidized property that the student lived in (P
it
), and includes a student (ω
t
)
and year fixed effect (ζ
t
), and an error term (ε
ist
).

We use Equation 2 in this form to test for the share of days that student i is absent in year
t. We then use the same specification but include a dichotomous variable for whether student i
was suspended in year t to control for the effect of suspensions on days absent. Next we use this
specification to test for the likelihood that student i is suspended in year t, and in this model OM
becomes suspensions that happen in years other than those where a student experiences a
contract ending. Finally, we use a full specification of this model that has the z score for student
i in year t as the outcome variable, and include suspensions, other moves, and the share of days
absent for student i in year t as controls. The results from all of these models are in Table 4.

74
The first model shows that controlling for a host of factors, a student living in a
subsidized property the year the contract expired is associated with a higher likelihood of
moving. Next we test whether a move, conditional on a student living a property when the
subsidy ends, has an effect on several outcomes. We find no relationship between the contract-
expiration move and the share of days absent or whether a student is suspended. However, we
do find that moves generally are associated with poorer outcomes in the form of a higher share of
days absent and likelihood of being suspended. As expected, we find that being suspended is
associated with a higher share of days absent. In our final specification, we find a marginally
significant and positive relationship between a move due to a contract expiration and test scores
as well as a smaller but positive and significant relationship between moving and academic
outcomes. Finally, we find that suspensions and a higher share of days absent are negatively
associated with test scores.

Combined, these models suggest that even when controlling for a myriad of factors that
affect moves we find that a contract expiration does increase the likelihood of moving. Next,
when controlling for factors that affect academic outcomes, we find that moves that did not occur
in the year a subsidy contract ended are associated with a higher share of days absent and
instances of suspension. However, when controlling for both of those factors, these moves are
associated with higher test scores. Of particular note to this research is that we do not find any
relationship between the move induced by a subsidy ending and worse academic outcomes. In
fact, we find that this move is marginally associated with higher test scores.

Robustness
In our data section we mention several threats to the identification strategy used in this
paper. One threat to identification in this paper is that we assign students to the incorrect groups.
To test our assignment we re-run our models using the most conservative matching strategy,
which includes only households assigned to the exact building address HUD provides regardless
of the size of the property. In this version, a contract ending is still associated with a higher
likelihood of moving, but the variable that captures all other moves is no longer significant in
any of the specifications. Interestingly, moves that occur in the year a subsidy contract ends are
still marginally associated with higher test scores, but the magnitude of the effect decreases.

75
V. Discussion
Our results highlight the complex relationship between subsidy expirations, and mobility
more broadly, and academic outcomes. We find evidence that a subsidy contract ending induces
a move but the impact of this move has no negative affect on academic outcomes. When
controlling for a host of factors that normally affect academic outcomes- including suspensions
and share of days absent- we actually find that a move induced by a subsidy contract ending is
marginally associated with higher test scores at a fairly large magnitude. This relationship
maintains the same level of marginal significance but decreases in magnitude in our robustness
check. This finding may seem counterintuitive because moves traditionally are found to be
associated with poorer academic outcomes. However, as previously noted, tenants in these
properties are offered a voucher so improvements in outcomes could be a result of the voucher
offer being a welfare improving one.

Tenants who live in a property when the project-based Section 8 subsidy expires are
offered a voucher that they can use to rent their existing unit or another unit on the private
market. The second chapter in this paper shows that 48 percent of households use this voucher,
and that households with dependents have lower odds of using their voucher and these
households lose all federal rental subsidy support. In addition, it shows that households with
dependents who did use a voucher were more likely to either not move at all, or move multiple
times, if they used that voucher. As a result, we would expect: 1) the majority of households in
our treatment group to move because they did not use their voucher and wanted to avoid a rent
increase; and 2) some households in our treatment group use a voucher to move. Both moves
could lead to better outcomes if a household moves to a higher opportunity neighborhood with
better performing schools. However, there are likely two distinctly different types of moves,
where voucher households make higher opportunity moves and those who not use a voucher
make lower opportunity moves or remain in place. As a result, it is important to explore where
students are moving, and what role the voucher plays in this moving process.

In aggregate many of the null findings in this chapter present a silver lining to the
potential negative affect that a subsidy contract ending has on low-income households. There is
previous evidence that a subsidy contract ending, and the move it induces, could negatively

76
affect the tenants in these properties. The findings in this paper show that such moves are not
associated with worse academic outcomes for children in these properties, and are potentially
associated with higher test scores. This means that the effect of a subsidy contract ending on the
welfare of households in these properties is more nuanced and could result in improved
outcomes for some households.

77
VI. Figures and Tables

Table 1: Characteristics of students in the sample
Race Overall Population Active S8 Expired S8
American Indian 2.40 0.37 0.00
Asian 4.70 5.48 1.32
Black 12.50 18.44 33.55
Hispanic 69.70 70.67 61.84
Pacific Islander 0.40 0.15 0.66
White 10.30 4.88 2.63
Sex

Female 48.90 49.41 51.1
Male 51.10 50.59 48.9
Country of Origin

U.S. 83.30 62.77 68.31
Non U.S. 16.70 37.23 31.69
English Proficiency

Proficient 54.80 58.93 69.74
Limited English 45.20 41.07 30.26
Number of Observations 1,738,788 42,271 183

78
Table 2: Descriptive statistics of outcomes of interest
Overall population Active S8 Expired S8
Annual probability of moving 0.06 0.12 0.11
Average number of days absent 9.13 10.38 11.93
Annual probability of a suspension 5.41 6.50

Average deviation from mean test score 0 -0.05

79
Table 3: Descriptive statistics of outcome of interest for treatment group

Moves Days Absent Suspensions Test Score
Pre 0.11 12.05 4.10 0.24
Post 0.11 11.9 7.13 -0.14
Difference 0 -0.15 3.03 -0.38

80
Table 4: Impact of contract ending on outcomes

Move
Share of days
absent
Share of days
absent
Suspended Test Score
Move

-0.001

-0.001

0.014

-0.114

Post 0.063 *** 0.008

0.008

-0.005

0.010

Move*Post

-0.024

-0.025

-0.016

0.500 *
Other Move 0.000

0.008 *** 0.007 *** 0.010 ** 0.012 **
Absent

-0.750 ***
Suspended

0.017 ***

-0.062 ***
Controls

Building Y Y Y Y Y
Grade Y Y Y Y Y
School Y Y Y Y Y
Student Y Y Y Y Y
Year Y Y Y Y Y
Observations 70,170 60,325 60,325 70,170 44,151
R
2
0.050 0.083 0.086 0.052 0.076
Significance: *** < 1 percent; ** < 5 percent; * <10 percent

81
Table 5: Impact of contract ending on outcomes robustness check
Move
Share of days
absent
Share of days
absent
Suspended Test Score
Move

-0.031

-0.030

-0.112
-0.464
Post 0.059 *** -0.005 -0.004 -0.041 0.090
Move*Post

0.014 0.013 0.037 1.313 *
Other move 0.001 -0.001 -0.001 -0.004 0.010
Absent

-0.740 ***
Suspended 0.014 -0.058 ***
Controls
Building Y Y Y Y Y
Grade Y Y Y Y Y
School Y Y Y Y Y
Student Y Y Y Y Y
Year Y Y Y Y Y
Observations 70,170 60,325 60,325 70,170 44,151
R
2
0.0502 0.083 0.086 0.052 0.076
Significance: *** < 1 percent; ** < 5 percent; * <10 percent

82

Chapter 5:

Conclusion

83
There are several themes that emerge from the findings in these papers that are important
for both research and practice. First, we have moved to a private ownership model of subsidized
housing and with this model comes tradeoffs. One key tradeoff is that we can leverage the
efficiency of the private sector to develop affordable housing but at some point owners can and
do exit subsidy programs. Because these properties are privately owned, we expect profit-
motivated owners to react rationally to market dynamics. As a result, some owners will choose
to exit subsidy programs when property values in the neighborhood are high or increasing.
Because these neighborhoods are more expensive, there are fewer private-market affordable
rental units and it is much more costly to develop new units. This is confirmed by the data, which
shows that while Section 8 properties are leaving improving neighborhoods, active and new
LIHTC units are being developed in lower opportunity ones.

One potential outcome from this trend is that as older subsidized units exit high
opportunity neighborhoods, and new units enter lower opportunity neighborhoods, we could see
clustering of subsidized housing in certain neighborhoods. This could limit the number and
quality of neighborhoods that subsidized households access, crowd out private market
development of units in those areas, and/or result in rent increases for unsubsidized households
in those markets. Alternatively, if these units increase property values, we could see owners
exiting subsidy programs in those markets, largely transforming affordable housing programs
into neighborhood redevelopment tools. Whether there is clustering and the impact that it has on
neighborhoods and low-income households, as well as the patterns and trend of future
expirations, are important future research questions that can be answered using the data in this
dissertation.

Second, this dissertation highlights that vouchers can be an efficient subsidy, but they are
not a panacea. The first study shows that voucher households generally access higher opportunity
neighborhoods than those in other subsidy programs. However, on average, active and expired
project-based Section 8 properties are in neighborhoods that are on a better trajectory than those
accessed by vouchers. Moreover, project-based Section 8 units due to expire between 2010 and
2020 are in higher opportunity neighborhoods that are also on a better trajectory than voucher
neighborhoods. This means that if those units exit the project-based Section 8 program, and
84
tenants are offered a voucher as a safety net, those households will need to use their voucher in
relatively higher opportunity neighborhoods that are on a better trajectory than the average
voucher household in order to not be displaced. This seems problematic considering that the
second study in this dissertation finds mixed evidence about the effectiveness of vouchers as a
safety net. As previously described, despite what market supply and household demand factors
should predict, only 48 percent of households even use the voucher they are offered, and those
who do not use it lose a large share of their effective income. One important thing to note is that
the models employed in that paper control for the variable impact of the capacity of the PHAs
that administers the expiration-related voucher. This variation can be high and is an important
area for future research using these data, because it could highlight differences across PHAs that
can be exploited to better understand how to improve the effectiveness of this subsidy.

Some households may benefit from being offered a voucher when a subsidy contract
ends. For example, households who use the voucher and move are associated with living in
lower poverty tracts, but fewer than 25 percent of households in these properties fall into that
category. In addition, any neighborhood gains must outweigh the cost of moving on a low and/or
fixed income, which is the case for all of these households. As noted, this event has welfare
implications on many levels, for example, it could affect the academic outcomes of students in
these properties. The third study in this dissertation focuses on the academic outcomes of
students in these properties and finds that a contract expiration catalyzes a move but finds no
evidence that this move is associated with worse academic outcomes. In fact, the paper finds
some evidence that students may see improvements in test scores due to such a move. These
findings seem logical in light of the second study, which shows that households with dependents
have higher odds of moving with their voucher and those moves are associated with lower
poverty tracts. The next step is to combine the data from these studies to test whether improved
test scores are being driven by voucher use.

These papers highlight the need to think about affordable housing within the context of
goals and tradeoffs. For example, the findings underline the importance of considering access to
high opportunity and improving neighborhood when evaluating the benefits of preserving
subsidies as opposed to developing new ones. But, when evaluating whether to preserve a
85
property, neighborhood access is just one part of the equation. A subsidy ending presents a shock
to the households in these properties that could result in welfare loses for some and gains for
others. As a result, deciding when to preserve a property is also a function of who lives in these
properties, where they are located, and the negative and positive externalities associated with the
place-based subsidy ending. Finally, cities have changed dramatically since the 1970s and 1980s
when these programs were developed. Traditionally, vouchers are more highly associated with
access to higher opportunity neighborhoods and improved household outcomes than other
subsidy programs; as shown in this dissertation, this is not always the case. There may be times
when accounting for neighborhood access and welfare costs means that preserving a place-based
subsidy is both a more effective and efficient use of resources.

Ultimately, current rental market dynamics, a lack of innovation in federal subsidy
programs, and a challenging funding environment, means we have to build and employ complex
databases like those used in this dissertation to evaluate goals and tradeoffs in a more robust way
when deciding the most effective and equitable way to approach housing policy.

86
References

Alexander, K.L., Entwisle, D.R., & Dauber, S.L. (1996). Children in motion: school transfers
and elementary school performance. The Journal of Educational Research, 90(1), 3–12.

Astone, N.M. & McLanahan, S.S. (1994). Family structure, residential mobility, and
school dropout: A research note. Demography, 31(4), 575-84.

Begley, J., Brazil, C., Reina, V., & Weselcouch, M., (2011). The state of New York City ’s
subsidized housing. NYU Furman Center for Real Estate and Urban Policy.

de Souza Briggs, X.D., (1997). Moving up versus moving out: Neighborhood effects in housing
mobility programs. Housing Policy Debate 8(1): 195-23.

de Souza Briggs, X., Popkin, S.J., & Goering, J. (2010). Moving to opportunity: The story of an
American experiment to fight ghetto poverty. Oxford University Press.

Cannon, J., Jacknowitz, A., & Painter, G. (2011). The impact of attending full day kindergarten
for English language learners,” Journal of Policy Analysis and Management, 30(2), 287-
309.

Collinson, R., Ellen, I. G., & Ludwig, J. (2015). Low Income Housing Policy. InEconomics of
Means-Tested Transfer Programs in the United States, volume 2. University of Chicago
Press.

Devine, D.J., Gray, R.W., Rubin, L., & Tahiti, L.B., (2003). Housing choice voucher location
patterns: Implications for participant and neighborhood welfare. Office of Policy,
Development, and Research, U.S. Department of Housing and Urban Development.

Econometrica, Inc. & Abt Associates, (2006). Opting in, opting out, and remaining affordable.
U.S. Department of Housing and Urban Development, Office of Policy Development
Research.

Ellen, I. G., O’Regan, K. M., & Voicu, I. (2009). Siting, spillovers, and segregation: A
reexamination of the low income housing tax credit program.Housing markets and the
economy: Risk, regulation, and policy, 233-267. Eriksen, M.D., & Ross, A., (2013). The
Impact of Housing Vouchers on Mobility and Neighborhood Attributes. Real Estate
Economics, 41(2): pp. 255–277.

Ellen, I. G., & Horn, K. M. (2012). Do Federally Assisted Households Have Access to High
Performing Public Schools? Civil Rights Research. Poverty & Race Research Action
Council (NJ1).
Ellen, I. G., & Weselcouch, M. (2015). Report: High-Opportunity Neighborhoods in NYC Are
Losing Affordable Housing.
87
Finkel, M., & Buron, L., (2001). Study on Section 8 voucher success rates: Volume I quantitative
study of success rates in metropolitan areas. Abt Associates.

Finkel, M., Hanson, C., Hilton, R., Lam, K., & Vandawalker, M. (2007). Multifamily Properties:
Opting In, Opting Out and Remaining Affordable.

Galvez, M., (2010). What do we know about housing choice voucher program location
outcomes? A recent review of literature. The Urban Institute.

Gasper, J., DeLuca, S., & Estacion, A. (2012). Switching schools: Revisiting the relationship
between school mobility and high school dropout. American Educational Research
Journal, 49(3), 487-519.
Goering, J., Kamely, A., & Richardson, T. (1997). Recent research on racial segregation and
poverty concentration in public housing in the United States.Urban Affairs Review, 32(5),
723-745.
Green, R., Reina, V., and S. Hepp. 2014. Casden real estate economics forecast. Lusk Center for
Real Estate at the University of Southern California.

Hanushek, E. A., Kain, J. F., & Rivkin, S. G. (2004). Disruption versus Tiebout improvement:
The costs and benefits of switching schools. Journal of Public Economics, 88, 1721-
1746.

Harris, D.R. (1999). Property values drop when blacks move in, because...’: Racial and
socioeconomic determinants of neighborhood desirability. American Sociological Review
64 (3): 461–79.
Horn, K.M., Ellen, I.G, & Schwartz, A.E, (2014). Do housing choice voucher holders live near
good schools? Journal of Housing Economics 23 (March): 28–40.
Jacob, B.A., & Ludwig, J., (2012). The effects of housing assistance on labor supply: Evidence
from a voucher lottery. American Economic Review 102(1): 272–304

Joint Center for Housing Studies of Harvard University (2015). The state of the nation's housing
2015. Joint Center for Housing Studies of Harvard University.

Jud, D.G. & Watts, J.M., (1981). Schools and housing values. Land Economics 57 (3): 459–70.
doi:10.2307/3146025.
Kasarda, J.D., (1993). Inner ‐city concentrated poverty and neighborhood distress: 1970 to 1990.
Housing Policy Debate 4 (3): 253–302.
Keels, M., Duncan, G.J., Deluca, S., Mendenhall, R., & Rosenbaum, J., (2005). Fifteen years
later: Can residential mobility programs provide a long-term escape from neighborhood
segregation, crime, and poverty. Demography 42 (1): 51–73.
88
Kennedy, S., & Finkel, M., (1994). Section 8 rental voucher and rental certificate utilization
study: Final report. Abt Associates.

Kessler, R.C., Adam, E., McDade, T.W., & Lindau, S.T., (2011). Moving to opportunity for fair
housing demonstration program final impacts evaluation. U.S. Department of Housing
and Urban Development.

Khadduri, J., Climaco, C., Burnett, K., Gould, L., & Elving, L. (2012). What Happens to Low-
Income Housing Tax Credit Properties at Year 15 and Beyond?. US Department of
Housing and Urban Development, Office of Policy and Development Research.
Kingsley, T.G., & Pettit, K.L.S., (2008). Have MTO families lost access to opportunity
neighborhoods over Time? Urban Institute Press.
Lens, M.C., Ellen, I.G., & O’Regan, K., (2011). Do vouchers help low-income households live
in safer neighborhoods? Evidence on the housing choice voucher Program. Cityscape 13
(3): 135–59.
Lens, M.C., (2014). The impact of housing vouchers on crime in US cities and suburbs. Urban
Studies, 51(6), pp.1274-1289.

Lens, M.C. (2014). Employment accessibility among housing subsidy recipients. Housing Policy
Debate 24 (4): 671–91.
Linden, L., & Rockoff, J.E., (2008). Estimates of the impact of crime risk on property values
from megan’s laws. The American Economic Review 98 (3): 1103–27.
Logan, J.R., Xu, Z. & Stults, B.J., (2014). “Interpolating US decennial census tract data from as
early as 1970 to 2010: A longitudinal tract database.” The Professional Geographer 66
(3): 412–20.
McClure, K., (2006). The low ‐income housing tax credit program goes mainstream and moves to
the suburbs. Housing Policy Debate 17 (3): 419–46.
McClure, K., (2011). Marketing opportunity index: Analysis of data at the tract and block group
level. Washington D.C: U.S. Department of Housing and Urban Development.
http://www.huduser.org/portal/publications/pubasst/housing_choice_voucher.html
McClure, K., Schwartz, A.F., and Taghavi, L.B. (2015). Housing choice voucher location
patterns a decade later. Housing Policy Debate 25 (2): 215–33.
Mehana, M., & Reynolds, A.J., (2004). School mobility and achievement: a meta-analysis.
Children and Youth Services Review 26(1), 93–119.

89
Meléndez, E., Schwartz, A. F., & Montrichard, A. D. (2008). Year 15 and preservation of tax-
credit housing for low-income households: An assessment of risk. Housing Studies,
23(1), 67-87.

National Research Council and Institute of Medicine. (2010). Student Mobility: Exploring the
Impact of Frequent Moves on Achievement: Summary of a Workshop. Committee on the
Impact of Mobility and Change in the Lives of Young Children, Schools, and
Neighborhoods. Board on Children, Youth, and Families, Division of Behavioral and
Social Sciences and Education. Washington, DC: The National Academies Press.

Newman, S. J., & Schnare, A. B. (1997). “… And a suitable living environment”: The failure of
housing programs to deliver on neighborhood quality. Housing Policy Debate, 8(4), 703-
741.

Olsen, E.O., (2003). Should the Housing Voucher Program Be Converted to a Block Grant?
Housing Policy Debate, 14: 283-293.

O’Regan, K.M., & Horn, K.M., (2013). What can we learn about the low-income housing tax
credit program by looking at the tenants? Housing Policy Debate 23 (3): 597–613.
doi:10.1080/10511482.2013.772909.
Ou, S & Reynolds, A. (2008). Predictors of educational attainment in the Chicago Longitudinal
Study. School Psychology Quarterly, 23(2), 199-229.

Pendall, R. (2000). Why voucher and certificate users live in distressed neighborhoods. Housing
Policy Debate, 11(4), 881-910.

Peterson, R.D., & Krivo, L.J., (2010). National neighborhood crime study (NNCS), 2000:
Version 1. http://www.icpsr.umich.edu/RCMD/studies/27501/version/1.
Popkin S.J., & Cunningham. M.K., (1999). CHAC section 8 program: Barriers to successful
leasing up. Washington, DC: The Urban Institute.

Popkin, S.J., Gwiasda, V.E., Olson, L.M., Rosenbaum, D.P., Buron, L., & Blank, R.M., (2000).
The hidden war: Crime and the tragedy of public housing in Chicago. Rutgers University
Press.

Popkin, S.J., & Cove, E., (2007). Safety is the most important thing: How HOPE VI helped
families. Text. June 26. http://www.urban.org/publications/311486.html
Ray, A., Kim, J., Nguyen, D., & Choi, J. (2015). Opting in, Opting
Out a Decade Later. U.S. Department of Housing and Urban Development, Office of
Policy Development and Research

Reina, V., & Belgey, J., (2014). Will they stay or will they go: Predicting subsidized housing
opt-outs. Journal of Housing Economics, 23:1-16.
90

Reina, V., & Williams, M., (2011). The importance of using layered data to analyze housing:
The case of the subsidized housing information project. Cityscape, 14 (1): 215-222.

Reynolds, A.J., Chen, C., & Herbers, J.E. (2009). School mobility and educational
success: A research synthesis and evidence on prevention. Paper prepared for the
Workshop on the Impact of Mobility and Change on the Lives of Young Children,
Schools, and Neighborhoods, June 29-30, The National Academies, Washington, DC.

Ricketts, E. R., & Sawhill, I. V. (1988). Defining and measuring the underclass. Journal of
Policy Analysis and Management, 7(2), 316-325.

Rohe, W. M., & Freeman, L. (2001). Assisted housing and residential segregation: The role of
race and ethnicity in the siting of assisted housing developments. Journal of the American
Planning Association, 67(3), 279-292.

Rubinowitz, L.S., and Rosenbaum, J.E., (2002). Crossing the class and color lines: From public
housing to white suburbia. 1 edition. Chicago: University Of Chicago Press.
Rumberger, R. W., & Larson, K. A. (1998). Student mobility and the increased risk of high
school dropout. American Journal of Education, 107(1), 1-35.

Sanbonmatsu, L., Katz, L.F., Ludwig, J., Gennetian, L.A., Duncan, G.J., Kessler, R.C., Adam,
E., McDade, T.W., and Lindau, S.T., (2011). Moving to opportunity for fair housing
demonstration program final impacts evaluation. Washington D.C: U.S. Department of
Housing and Urban Development.
Schachter, J. P. (2004). Geographic mobility: 2002-2003. Current Population Reports, P20-549.
Washington, DC: U. S. Bureau of the Census.

Schwartz, A., & Mel éndez, E. (2008). After year 15: Challenges to the preservation of housing
financed with low‐income housing tax credits. Housing Policy Debate, 19(2), 261-294.

Schwartz, H., Bostic, R.W., Green, R.K., Reina, V., Davis, L.M., & Augustine. C.H., (2016).
The Preservation of Affordable Rental Housing: An Evaluation of the MacArthur
Foundation ’s Window of Opportunity Initiative. The RAND Corporation.

Shen, Q. (2001). A spatial analysis of job openings and access in a U.S. metropolitan area.
Journal of the American Planning Association, 67, 53 – 68.

Shroder, M., (2002). Locational constraint, housing counseling, and successful lease-up in a
randomized housing voucher experiment. Journal of Urban Economics, 51: 315-338.

Simpson, G. A., & Fowler, M. G. (1994). Geographic mobility and children's
emotional/behavioral adjustment and school functioning. Pediatrics, 93(2), 303-309.
91

South, S., Haynie, D. L., & Bose, S. (2007). Student mobility and school dropout. Social
Science Research, 36 (1)68-94.

Swanson, C. B. and Schneider, B. (1999). Students on the move: Residential and
educational mobility in America’s schools. Sociology of Education, 72(1):54-67.

Teachman, J. D., Paasch, K., & Carver, K. (1996). Social capital and dropping out of school.
Journal of Marriage and the Family,58(3), 773-783.

Turner, M.A., Comey, J., Kuehn, D., and Nichols, A. (2011). Helping poor families gain and
sustain access to high-opportunity neighborhoods. Text. December 5.
http://www.urban.org/publications/412455.html.
U.S. Census Bureau. (2001). Census 2000. Washington, DC.

U.S. Census Bureau. (2013). Longitudinal employer-household dynamics. Washington,DC.

U.S. Census Bureau. (2015). American community survey. Washington, DC.

U.S. Department of Housing and Urban Development. (2003). Physical condition standards;
physical inspection requirements. Retrieved from
https://www.gpo.gov/fdsys/granule/CFR-2003-title24-vol4/CFR-2003-title24-vol4-
sec983-5

U.S. Department of Housing and Urban Development. (2015). Affirmatively furthering fair
housing. 24 CFR Parts 5, 91, 92, 570, 574, 576, and 903.
U.S. Department of Housing and Urban Development. (2015). Section 8 Renewal Policy
Guidebook. Retrieved June 16, 2016, from
http://portal.hud.gov/hudportal/HUD?src=/program_offices/housing/mfh/mfhsec8

U.S. Government Accountability Office. (1994). Elementary School Children: Many Change
Schools Frequently, Harming Their Education. Washington, DC: U.S. General
Accounting Office.

U. S. Government Accountability Office. (2007). Project based rental assistance: HUD should
update its policies and procedures to keep pace with the changing housing market. GAO-
07-290.Retrieved from http://www.gao.gov/new.items/d07290.pdf

Williamson, A.R., Smith, M.T. & Strambi-Kramer, M. (2009). Housing choice vouchers, the
low-income housing tax credit, and the federal poverty deconcentration goal. Urban
Affairs Review 45 (1): 119–32.
Wood, D., Halfon, N., Scarlata, D., Newacheck, P., & Nessim, S. (1993). Impact of family
relocation on children's growth, development, school function, and behavior. Journal of

92

Appendix
to
Chapter 4

93
I. Technical Appendix

In order to determine which students lived in a subsidized property, and who lived in a
property when the subsidy ended, we first identified almost 500 zip codes where students are
eligible to attend a LAUSD school. Next, we developed a database of all 1,500 federally
subsidized rental properties located in an eligible LAUSD zip code. Our exogenous shock
comes from owners leaving the project-based Section 8 program, so we flagged the 593
properties that ever received a project-based Section 8 contract and identified the 171 of those
where the contract ended. We then used our address-level student and property data to determine
if and when a student lived in a property that received a project-based Section 8 subsidy, and if
they lived in a property when that subsidy ended. In order to do this, we first geo-coded the list
of all subsidized properties in LA County (Figure 1) with a 100 percent address-level match rate,
and flagged which of those properties ever had a project based Section 8 contract (Figure 2).
Next, we focused just on those properties that ever had a project-based Section 8 contract and
identified which ones exited the program (Figure 3). Finally, we cleaned over 1.6 million unique
student addresses using Experian’s address cleaning software, Stata programs, GIS, and manual
inspection, and geo-coded each of those addresses with a 96 percent address-level match-rate.
We then layered the student address data onto the property data and determined if there was a
direct match between the two.

It is common in Los Angeles County for privately owned subsidized properties to have
multiple buildings, with none being above three stories tall. As a result, properties with more
units have more individual buildings and cover a larger land area. However, the HUD property
data usually only provides one building address per property, which is often the building in the
center of the development. Using only the HUD-provided address could result in assigning
students to the incorrect buildings and groups. As a result, we created conservative distance
weights based on the unit count in a property to match students to properties when there was not
already an exact match with the single HUD address. For example we considered a student a
match with a subsidized property if they lived within 50 feet of the property, and the property
contained at least 50 units because a 50 units building cover a larger land area than one with
fewer units. Similarly, we applied a 200-foot buffer for a property with more than 200 units.
Based on this approach, we found 42,271 students who ever lived in Project-based Section 8

94
development; 77 students who lived in a property when the project-based Section 8 contract
ended through an exact address match; and an addition 106 students who lived in a property
when the project-based Section 8 contract ended once we included our distance weighting. All
project-based Section 8 properties were developed in the 1970s and have a distinct architectural
design. As a result, we were able to use the google maps street view feature to look at each of the
properties where we assigned a student to the treatment group and confirm we correctly assigned
them to the property.

95
II. Figures
Figure 1: All subsidized properties in Los Angeles County

96

Figure 2: All subsidized and Section 8 properties in Los Angeles County

97
Figure 3: All current and expired Section 8 properties in Los Angeles County

Abstract (if available)

Abstract The federal government made a shift in its approach to the provision of affordable housing in the 1960s from a public development and ownership model, to a private one. Since the 1970s over 3.3 million units of privately owned housing were developed through federal rental subsidy programs. Owners of these properties agreed to develop affordable housing, and maintain it as such for a fixed period of time, in exchange for the subsidy. There is an abundance of literature that studies the development of this housing and what it means for cities, neighborhoods, and the welfare of low-income households. To date, there is no research that focuses on what happens when owners reach the end of their affordability restriction periods and choose to exit a subsidy program. This dissertation is the first to study this phenomenon and the implications it has for the households who live in these properties as well as our broader understanding of mobility, neighborhood choice, and the utility of vouchers.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Household mobility and neighborhood impacts

PDF

The interactions between housing and business

PDF

Affordable south Los Angeles: survival, support, and different futures

PDF

Three essays on housing demographics: depressed housing access amid crisis of housing shortage

PDF

Mitigating the energy efficiency gap through Property Assessed Clean Energy (PACE): an assessment of the HERO Program in Riverside County, CA

PDF

A better method for measuring housing affordability and the role that affordability played in the mobility outcomes of Latino-immigrants following the Great Recession

PDF

Risks, returns, and regulations in real estate markets

PDF

Do low-income households benefit from homeownership?

PDF

The impact of demographic shifts on automobile travel in the United States: three empirical essays

PDF

Property and labor formalization in the age of the sharing economy: Airbnb, housing affordability, and entrepreneurship in Havana

PDF

No place like home: a three paper dissertation on K-12 student homelessness & housing affordability

PDF

The role of public policy in the decisions of parents and caregivers: an examination of work, fertility, and informal caregiving

PDF

The economic and political impacts of U.S. federal carbon emissions trading policy across households, sectors and states

PDF

Crowded with potential: housing and social mobility strategies among China's educated migrants

PDF

County governance reform in California: introduction of the council‐executive model and the elected county‐executive

PDF

Lessons from TAP implementation: obstacles and solutions to improve the transit users experience

PDF

Active travel, outdoor leisure, and neighborhood environment: path analysis, Los Angeles County

PDF

Unraveling decentralization of warehousing and distribution centers: three essays

PDF

Essays on the economics of cities

PDF

Three papers on dynamic externalities and regional growth

Asset Metadata

Creator Reina, Vincent J. (author)

Core Title The impact of mobility and government rental subsidies on the welfare of households and affordability of markets

School School of Policy, Planning and Development

Degree Doctor of Philosophy

Degree Program Public Policy and Management

Publication Date 07/27/2016

Defense Date 04/22/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag affordability,household welfare,housing policy,low-income,mobility,neighborhood change,OAI-PMH Harvest,rental,subsidized housing

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bostic, Raphael (committee chair), Painter, Gary (committee chair), Boarnet, Marlon (committee member), Green, Richard (committee member)

Creator Email vincent.reina@gmail.com,vreina@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-281192

Unique identifier UC11280684

Identifier etd-ReinaVince-4633.pdf (filename),usctheses-c40-281192 (legacy record id)

Legacy Identifier etd-ReinaVince-4633.pdf

Dmrecord 281192

Document Type Dissertation

Format application/pdf (imt)

Rights Reina, Vincent J.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA