University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

A BETTER METHOD FOR MEASURING HOUSING AFFORDABILITY AND THE ROLE THAT
AFFORDABILITY PLAYED IN THE MOBILITY OUTCOMES OF LATINO-IMMIGRANTS
FOLLOWING THE GREAT RECESSION

By

Ray Calnan

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(POLICY, PLANNING, AND DEVELOPMENT)

August 2015

Copyright 2015 Ray Calnan

ii

Acknowledgements
I dedicate this work to my wife, Adina Calnan, without her support and
encouragement it would not have been possible. I would also like to dedicate this to the
memory of my original mentor and the driving force behind me entering academia, Dr.
Donald Bleich. Without him, I would never have been interested in real estate, housing, or
academia.
I would like to thank my committee chair, Dr. Gary Painter. With his guidance I was
able to develop my skills and complete my PhD. I am grateful for all the tough questions
and continual assistance in helping me finish my dissertation.
I would also like to express my thanks to my committee members, Dr. Dowell Myers
and Professor Roberto Suro. Each assisted me with their unique and differing expertise in
their fields. I am also appreciative of Prof. Suro’s practical advice on completing my
dissertation and for Dr. Myers’ push to keep striving to solve a problem.
Additionally, I would like to thank my friend and colleague, Sarah Mawhorter, for
her support and the stimulating conversations about our esoteric research. It is tough to
complete a PhD, but the process is made much better with a great committee, supportive
friends, and the best wife.

iii

Table of Contents
Acknowledgements .............................................................................................................................................. ii
Introduction ............................................................................................................................................................ 1
The Response of Latino Immigrants to the Great Recession: .............................................................. 4
Introduction ................................................................................................................................................... 4
Background and Literature Review ...................................................................................................... 6
Data and Model ............................................................................................................................................. 9
Summary Statistics ................................................................................................................................... 16
Regression results - Mobility and Employment ............................................................................ 23
Conclusion .................................................................................................................................................... 32
Appendices .................................................................................................................................................. 34
References ............................................................................................................................................................ 39
An Alternative Method of Measuring Housing Affordability ............................................................ 42
Introduction ................................................................................................................................................ 42
Background and Literature Review ................................................................................................... 43
Data ................................................................................................................................................................. 49
Summary Statistics ................................................................................................................................... 51
An Alternative Measure .......................................................................................................................... 59
Conclusion .................................................................................................................................................... 71
References ............................................................................................................................................................ 73
Addenda ................................................................................................................................................................. 75
The Response of Latino Immigrants to Housing Affordability During and Following the
Great Recession .................................................................................................................................................. 84
Introduction ................................................................................................................................................ 84
Background and Literature review .................................................................................................... 86
Data and Model .......................................................................................................................................... 88
Summary Statistics ................................................................................................................................... 90
Regression Results.................................................................................................................................. 101
Conclusion .................................................................................................................................................. 105
References .......................................................................................................................................................... 107
Addenda ............................................................................................................................................................... 109
Conclusion........................................................................................................................................................... 115

1

Introduction
Housing affordability has been an ongoing concern for many years. Consideration
for it increases as prices and rents increase, but it is often forgotten about, by policymakers,
as housing prices fall. The method for measuring housing affordability has remained fairly
constant for over 30 years with the use of 30% or less of household income devoted to
housing costs considered affordable (Jewkes and Delgadillo 2010). Modifications of this
measure often come in the form of adjustments for changes in housing quality and external
amenities. However, the basic measure still begins with the household as a single unit.
Other methods of measuring affordability have been suggested, but most become overly
complicated or fail to consider the changing make-up of households (M. Stone 2006).
During the early 2000’s there was a housing boom that boosted the economy as a
whole. The expansion in construction jobs masked the loss of manufacturing jobs during
the boom period. As a result of the economic expansion, housing prices increased
dramatically throughout the US and employment was high. The growth in home prices
sustained until 2007 when the bubble burst and housing prices fell, taking down the overall
economy and jobs. The loss of manufacturing jobs became more evident as there were few
other jobs to mask the long-term loss in manufacturing. The housing bust led to what is
now known as the “Great Recession”.
The Great Recession provides an opportunity for a natural experiment that resulted
in a shock to employment in different sectors at different times. The change in housing
costs and incomes moved at different times and allows researchers the ability to measure
the impact of different changes as the recession began and through to the beginning of the
recovery. Particular groups of people were hit harder by the changing economy due to
2

factors such as employment sector, housing tenure, mobility, and social support (Hurd and
Rohwedder 2010 ).
One of the most vulnerable groups affected by the recession was Latino immigrants
and more specifically, less established Latino immigrants. Research has found that Latino
immigrants are affected to a large degree by changes in the occupations in which they
concentrate and by the support networks in the communities in which they live (C. Z. Liu
2007) (Head and Lloyd-Ellis 2012) (Kochhar, Espinoza and Hinze-Pifer 2010). Given the
influx of Latino immigrants to the United States and the dispersal throughout the country,
not following traditional gateways, the impact of the Great Recession has provided an
opportunity to examine the effects of an economic shock on a vulnerable group.
This dissertation attempts to determine the effects of the overall economic
downturn on the mobility and employment outcomes for Latino Immigrants using key
variables of network effects, changes in employment, and the affordability of housing. The
first essay, co-authored with Gary Painter, PhD, is focused on identifying the occupations
that Latino immigrants tend to concentrate in and the effects of the network for Latino
immigrant communities. One industry that has a high concentration of Latino immigrants is
construction. The expansion of employment in this field, due to the housing boom, didn’t
occur only in the traditional gateways for Latino immigrants. The growth occurred in other
regions of the country as well. The essay measures the likelihood of Latino immigrants to
move given the construction employment changes, and the likelihood of obtaining
employment if a move has occurred.
The second essay considers the changes in the make-up of households in order to
identify an alternative measure for housing affordability. The convenience of the long
3

established measure of affordability is hard to overcome. However, the argument against
its use is that the existing measure does not account for changes in the number of people in
the household that are required to work in order to maintain the same level of affordability.
The alternative measure proposed shows that the change in affordability over the years is
much more dramatic when the addition of earners in a household is considered. Housing
affordability has traditionally been a concern to policy-makers when housing prices are
going up, but less so during economic downturns. The reality is that it should be a
continued concern since housing affordability is not just a result of changes in housing
costs, but is also a result of changes in income (Jewkes and Delgadillo 2010). This essay
helps to show that household composition should be considered in the affordability
equation.
In the third essay, the impacts of housing affordability, using the alternative
measure, on the decision of Latino immigrants to move following the housing bust is
considered. Latino immigrants, being a vulnerable group are more likely to move as a result
of job loss and could potentially move-in together in order to pool funds and reduce overall
individual costs. The ultimate finding in this final analysis is that housing affordability does
alter the likelihood of Latino immigrants to make the decision to move to a different metro
area and that the alternative measure provides a more accurate picture of the movements,
than the use of median housing costs or the traditional measure of housing affordability.

4

The Response of Latino Immigrants to the Great Recession:
Occupational and Residential (Im)mobility (co-authored with Gary Painter, PhD)

Introduction
Despite the broad impact of the Great Recession on households in the US, there were
distinct housing and labor markets that were particularly hard hit. This was primarily due
to the fact that the recession, which was precipitated by a collapse in the housing market,
was most severe in markets where much of the recent economic growth had been fueled by
the housing industry (Charles, 2013). Not surprisingly, these markets had drawn a large
number of recent immigrants because of the number of jobs in the construction industry
and tertiary industries that were available prior to the bursting of the housing bubble.
Despite a sizeable literature on the impact of downturns that has focused on
industry specific shocks and on the general impact of recessions on vulnerable populations,
there has been much less research that focuses on immigrant populations. Historic shocks,
such as the loss of manufacturing jobs in the 1980-90’s, showed that minorities were more
likely to be displaced, in part, due to the occupations in which they were concentrated
(Kletzer, 1991; Hamermesh, 1989). Research also finds that the Black population was
harder hit than the White population during the Great Recession, and in past recessions,
with higher unemployment, higher foreclosure rates, more forced movement within local
areas, and slower reemployment (Stoll, 2013; Fairlie, and Kletzer, 1998). Orrenius and
Zavodny (2010) have found that recent immigrants are especially sensitive to shocks in the
labor market, but this research did not focus on industry specific shocks. The current
recession is particularly interesting to study because the recession started in an industry
5

that employed many recent immigrants and spread to the broader economy. This provides
the opportunity to observe how immigrants respond to severe labor market shocks.
In order to assess how Latino immigrants have responded to the Great Recession,
we will document shifts in the occupational status, residential mobility, and employment
status of the Latino immigrant population after the housing crash using data from the
American Community Survey 2006-2009. We are particularly interested in whether the
loss in construction jobs is a contributor to the aforementioned changes, or whether the
overall job market is the main determinant of any changes. Finally, we test whether the
strength of immigrant networks (Painter and Yu, 2010; 2012; Zhu, Liu, and Painter, 2014)
in these metropolitan areas influence the likelihood that Latino immigrants will move or
remain in the same metropolitan area, and whether this will impact their labor market
outcomes.
The results suggest that there have been large declines in the proportion of the
Latino immigrant population that was working in the construction industry over the course
of the recession, in the metropolitan areas studied. Those who lost jobs during the
recession were most likely to still be unemployed or to have moved from the hardest hit
areas. We observed a much smaller increase in the share of Latino immigrants working in
other industries. We also found that while the decline in construction employment
predicted an increase in moving out of a metropolitan area in 2008, the impact of job losses
in the broader economy in 2008 and 2009 were much larger.

6

Background and Literature Review
There has been a lot of recent research on the topics of mobility and employment in
the Great Recession (Chapple and Lester, 2010; Kochhar, Espinoza, and Hinze-Pifer, 2010;
Kothari, Saporta-Eksten, and Yu, 2013; Painter and Yu, 2013; Singer and Wilson, 2010;
Stoll, 2013), but little research has specifically focused on the impacts that the recession
has had on Latino immigrants. Immigrant employment during and after the recession has a
large impact in metropolitan areas because globalization has encouraged immigration to
many developed countries like the United States, with the 17 largest metropolitan areas
having an average of 27% immigrant population, in 2006, as compared to the national
average of 12% (Card, 2009). The Great Recession has not only reduced the number of
immigrants coming to the United States, it has also altered the movements throughout the
country so that they no longer follow the traditional gateways to upward mobility (Frey,
2009; Singer and Wilson, 2010). Research on where people are moving and what the new
driving forces are is lacking with few studies considering changing employment industry
and location simultaneously (Osberg, Gordon, and Lin, 1994).
The number of unauthorized immigrants in the country changed following 2007
with a reduction from high point of 12 million in March 2007 to 11.1 million in March 2009
(Cohn, 2010). While the number of Mexican immigrants in the US dropped towards the end
of the decade there was little evidence of a large migration back to Mexico, only a reduction
in the number of recent immigrants (Cohn, 2009). This is partly in response to the changing
economic environment and a result of changes in deportation and immigration policy.
Since 2005 the number of total removals recorded by the Department of Homeland
Security increased from under 250,000 to nearly 450,000 in 2013, with Mexican nationals
accounting for nearly 72% of all removals in 2013 (DHS, 2015). Migration choices by
7

immigrants are affected by local immigration enforcement and cooperation, with full
enforcement having the same effect on internal migration as a 15% decline in employment
demand for immigrants (Watson, 2013).
Much of the recent literature discusses aggregate mobility and finds the effect of
employment to be small; however, the literature does not isolate the impacts of the
recession on Latino-immigrants and how mobility played a role in the employment
outcomes for this group (Chapple and Lester, 2010; Kothari, 2013; Notowidigdo, 2011).
Research on the impacts of economic shocks on minority groups generally focus on the
outcomes encountered by the Black population. This group was harder hit than the White
population during the Great Recession and in past recessions with greater unemployment,
higher foreclosure rate and more forced movement within local areas in addition to slower
reemployment (Stoll, 2013; Fairlie, and Kletzer, 1998). Historic shocks, such as the loss of
manufacturing jobs in the 1980-90’s showed that minorities were more likely to be
displaced, in part, due to the occupations they were concentrated in (Kletzer, 1991;
Hamermesh, 1989). However, the research does not delve into the question of outcomes
for Latino-immigrants as most has been focused on the difference between Black and White
populations.
A recent Pew Report (2010) found that the immediate structural change in
employment may not have been as harmful for immigrants as an overall gain in the number
of jobs held by foreign born people, and a loss for native born was observed. Stability in
immigrant employment, after a shock such as the recession, may be the result of strong
ethnic communities and employment demand sensitivity of immigrants (Cadena and
Kovak, 2013). Another reason that immigrant employment outcomes may not have
8

suffered as much is because Latino-immigrants have a higher probability of finding a job, in
part, due to the willingness to have longer commutes, and their willingness to move to
employment in addition to the willingness to change employment type and industry (Zhu,
Liu, and Painter, 2014). Cadena and Kovak (2013) found that the willingness to move was
not just among recent immigrants, but held for longer-term immigrant residents as well.
However, the overall trend prior to and in the beginning of the recession was that fewer
people, from all backgrounds, were willing to make long distance moves to seek out
employment (Molloy, Smith and Wozniak, 2014).
It has also been found that immigrant employment patterns are related to the
strength of the social networks of the areas in which they live (Zhu, Liu, and Painter, 2014).
The size of the network also increases the likelihood of finding employment outside of
agricultural occupations (Munshi, 2003). However, it has been found that smaller networks
are sought as individual English proficiency improves (Bauer, 2005). While these network
effects have been shown to be important to immigrant groups, little research has been
done to gauge the strength of the networks during the recession and in comparison to
other effects such as employment in construction and overall employment in general.
On the other hand, research has noted that employment opportunities for
immigrants are impacted by the economic cycles to a greater degree than the native born
population due in part to the education level of most, and in some cases, the legal working
status (Orrenius and Zavodny, 2009). Pendall et al (2012) find that immigrant status,
among other characteristics, makes one more vulnerable to housing shocks and household
income changes. This can vary based upon the regional resilience provided by the overall
neighborhood. There were many changes circumstances during the recessionary period,
9

including changes to laws in places such as Arizona that made it less hospitable to Latino-
immigrants, especially younger, non-citizens (Bohn, 2014).
The literature shows that lower education levels may increase the job loss rate
amongst this group, but the willingness to move may also lead to finding new employment
faster, thereby reducing the length of unemployment (Orrenius and Zavodny, 2009;
Pissarides, and Wadsworth, 1989). The willingness to move after a shock, such as the Great
Recession, is convoluted due to the possible counter effects encompassed in an immigrant
worker (Borjas, 2003). Thus, it is possible, despite having lower skills, that immigrant
works in these industries may be more likely to move. Both Cadena and Kovak (2013) and
Glewwe and Hall (1998) found that low-skilled immigrant workers are more likely to move
for job opportunities than are higher-skilled native born workers.
In sum, the literature suggests that labor market shocks affect vulnerable
populations the most. Immigrants are likely to be vulnerable, and it is possible that they
may move at higher rates to find employment. Therefore, it is not clear if they have suffered
as much as other vulnerable populations. This study will fill this gap in the literature by
focusing on the responses of Latino immigrants to the recession.
Data and Model
The dataset used for this analysis is drawn from the 2006 - 2009 American
Community Survey (ACS) microdata set. The data were obtained from Integrated Public
Use Microdata Series (IPUMS) provided by the Minnesota Population Center (Steven
Ruggles 2010). The data included household and individual level characteristics that
provide detailed information on demographic, economic, and mobility attributes.
The specific attributes used in this study are the following: Household income, Age
of the head of household and individual age, Mobility within the past year, Employment,
10

Occupation, Education, Marital status, Length of time in the United States (immigrant or
non-immigrant and length of time), and Ethnic background (Latino or non-Latino). Ethnic
background was determined from the “Hispanic Origin” variable provided by the ACS.
People self-declaring that they are “Not Hispanic” were considered to be not Latino; all
others were considered to be of Latino origin. Immigrant status was determined using the
variable YRSUSA2 coded by IPUMS and based upon the ACS variable “year of immigration”.
This is coded with “N/A”- not applicable (native born) and 5-year increments with the final
group being “21+ years”.
Additionally, metropolitan level characteristics are included in the analysis, such as:
Percent of Latino-Immigrants who speak English well, Length of time Latino-Immigrants
have been in the United States, Overall employment in the metro, and Industry specific
employment in the metro. The metro area of origin was used for the regression analysis for
the change in overall employment and the change in construction if the person moved
between metros (refer to table A-2 in the appendix).
Many of the head of household characteristics were assigned to all the members of
the household for classification and analysis purposes. Group home and incarcerated
people were excluded from the dataset as motivations for moving, in these cases, vary from
the motivations of the general population. Individuals were classified as Latino-Immigrant
and Other. The Other category includes all native born people and all people who are
immigrants, but not Latino.
This study aggregates data at the metropolitan level. The metropolitan areas which
are selected do not align directly with the immigrant gateway typology outlined by Audrey
Singer in her 2004 paper. However, there is a lot of similarity in the metropolitan areas
11

selected for this analysis. Instead, we selected the 25 metropolitan areas based on whether
immigrants were highly concentrated in the construction industry prior to the housing bust
as described below.
There are numerous ways to measure concentration in the construction industry.
One method is to simply rank the percentage of immigrant workers in each industry, and
determine a threshold for which particular metropolitan areas have a high percentage of
workers within the construction industry. The problem with this method is that is does not
take into account how these ratios compare to the overall population. Our chosen method
is based on the extent to which the construction industry is an ethnic niche industry.
Following Liu (2011) and Liu and Painter (2012), we will calculate an Industrial
Concentration Quotient (ICQ) as follows:
m
im
j
ij
E
E
E
E
ICQ 
where j = (1,…, n) and refers to industries.
Eij is the number of a certain group employed in an industry, and Ej is the total
employment in that industry. Eim is the employment of a certain group in metropolitan area
and Em is total employment in metro. A ratio of larger than 1 signifies that immigrant
concentration in a certain industry is greater than the metropolitan employment in that
industry. An ICQ of 3 or greater would imply that an immigrant is 3 times as likely to work
in construction as another industry. A score of 3 or greater would indicate that the
particular industry is a strong ethnic niche.
For the purpose of this analysis, the ICQ equation was modified to use groupings of
Census defined occupations as a substitute for the industry, as shown in Table 1. These
consolidated occupation groups provide a better representation of the occupations in
12

which Latino immigrants might work rather than relying upon the industry code which
could include occupations or positions less occupied by Latino-Immigrants. We identified 6
occupations as ethnic niches: Clothing, Construction, Domestic Service, Farm, Food, and
Manufacturing.
Table 1. Occupations used to create consolidated occupations
Consolidated Occupation Census defined occupation
Clothing Dressmakers and seamstresses
Knitters, loopers, and toppers textile
Textile cutting machine operators
Textile sewing machine operators
Pressing machine operators (clothing)
Misc textile machine operators
Construction Masons, tilers, and carpet installers
Carpenters
Drywall installers
Electricians
Electric power installers and repairers
Painters, construction and maintenance
Plasterers
Paperhangers
Plumbers, pipe fitters, and steamfitters
Concrete and cement workers
Glaziers
Insulation workers
Paving, surfacing, and tamping equipment
Roofers and slaters
Sheet metal duct installers
Structural metal workers
Construction trades, n.e.c.

Operating engineers of construction
equipment
Excavating and loading machine operators
Helpers, constructions
Construction laborers

13

Table 1, continued
Domestic Service Housekeepers, maids, butlers, stewards
Supervisors of cleaning and building service
Janitors
Gardeners and groundskeepers
Laundry workers
Vehicle washers and equipment cleaners
Farm Farm workers
Farm managers, except for horticultural
Supervisors of agricultural occupations
Graders and sorters of agricultural products
Food Misc food prep workers
Cooks, variously defined
Bartenders
Kitchen workers
Waiter's assistant
Misc food prep workers
Butchers and meat cutters
Bakers
Manufacturing Patternmakers and model makers
Cabinetmakers and bench carpenters
Furniture and wood finishers
Upholsterers
Sawing machine operators and sawyers
Hand molders and shapers, except jeweler
Batch food makers
Lathe, milling, and turning machine operators
Punching and stamping press operatives
Rollers, roll hands, and finishers of metal
Drilling and boring machine operators
Grinding, abrading, buffing, and polishing
Forge and hammer operators
Molders, and casting machine operators
Metal platers
Heat treating equipment operators
Wood lathe, routing, and planing machine
Nail and tacking machine operators
Other woodworking machine operators
Packers, fillers, and wrappers
Painting machine operators
Packers and packagers by hand

14

Table 2 presents the ethnic niche index in 2006 for each of the industries across the
study. It is first interesting to note that the highest ethnic niche is not construction in all of
the metros although it is highest in many. Further, there are metropolitan areas like
Memphis and Richmond that have extremely high ICQ values for the construction industry.
Construction would be classified as a very strong ethnic niche in all of the metros except
Los Angeles and San Bernardino (Liu, 2011). The immigrant population in the Southern
California metropolitan areas is more settled; therefore, no industry would be classified as
a strong ethnic niche. We chose to keep the greater Los Angeles Metropolitan area in the
sample because the housing crash was so strong there. Across all the study areas, the
industries that we have identified as ethnic niches have much larger concentrations of
Latino Immigrants in them than in all of the other occupations in those metropolitan areas.

Table 2. Ethnic Niche Index - 2006
All Other
Occupations Clothing Construction
Domestic
Service Farm Food
Manu-
facturing
Atlanta,GA 0.46 4.30 5.80 3.80 3.00 2.00 2.90
Austin,TX 0.40 3.50 4.90 4.30 2.40 3.00 2.40
Baltimore,MD 0.58 6.40 5.40 4.20 7.00 3.40 0.41
Charlotte-Gastonia-Rock Hill,NC-SC 0.42 3.00 5.30 3.00 3.60 3.50 2.30
Columbus,OH 0.50 . 5.30 4.90 2.60 3.60 3.30
Dallas-FortWorth,TX 0.55 2.20 4.00 3.20 1.70 2.50 3.50
Detroit,MI 0.65 . 3.90 4.20 3.00 2.00 1.80
Greensboro-WinstonSalem-HighPoint,NC 0.40 3.50 5.00 2.00 0.75 4.00 3.30
Indianapolis,IN 0.48 3.70 4.80 4.40 3.00 3.10 4.30
KansasCity,MO-KS 0.44 5.10 4.60 5.40 0.46 4.80 2.70
LasVegas,NV 0.45 2.70 3.10 3.20 2.40 2.20 2.40
LosAngeles-Long Beach,CA 0.71 2.70 2.30 2.80 2.40 2.00 2.70
Memphis,TN/AR/MS 0.41 . 8.00 2.50 . 1.40 6.50
Nashville,TN 0.44 1.30 6.30 4.10 1.20 2.60 4.00
Phoenix,AZ 0.47 1.90 3.70 3.80 3.40 2.90 2.60
Raleigh-Durham,NC 0.39 0.25 7.10 3.70 3.70 3.30 2.80
Richmond-Petersburg,VA 0.36 . 9.50 3.60 1.60 2.10 3.10
Riverside-SanBernardino,CA 0.72 3.00 2.20 2.40 2.70 1.80 2.60
Sacramento,CA 0.51 2.20 3.70 5.30 6.90 2.10 5.30
SanAntonio,TX 0.59 1.90 3.50 3.40 0.70 2.30 4.70
SanFrancisco-Oakland-Vallejo,CA 0.60 1.60 3.40 4.30 5.80 2.70 3.90
SanJose,CA 0.56 3.50 3.40 5.00 4.40 3.60 3.10
Seattle-Everett,WA 0.47 2.10 3.80 5.70 8.50 4.10 4.10
Washington,DC/MD/VA 0.56 2.80 4.90 4.80 2.40 3.60 2.00
WestPalmBeach-BocaRaton-Delray Beach,FL 0.64 1.60 3.40 2.80 3.60 0.76 1.80
15

Below we display summary statistics on the changes in the employment rate and the
change in industry of employment across the recession. We will then estimate two models.
The first is a multinomial logit model that will examine the correlates of the choices to
move outside the metropolitan area, to move within a metropolitan area, and to stay in
place. Careful attention is placed on how the change in employment in the construction
industry, and in the broader labor market, impacts the decision to move after controlling
for individual and household level characteristics. The model will also control for the
strength of the Latino immigrant network in the metropolitan area as proxied by the
percentage of immigrants in that area who have been in the US more than 10 years (Painter
and Yu, 2010). The impact of job losses in a particular area would be expected to induce
mobility while the predicted impact of the strength of the immigrant network would be to
reduce mobility out of the area. This is because the network may provide better access to
local labor markets (Zhu, Liu, and Painter, 2014) or support that may help an immigrant
withstand the job losses.
The second set of regression models use a multinomial logit framework to test what
are the factors that would influence the probability of working. These models include the
same covariates as the aforementioned mobility model, and also include a variable to see
whether mobility out of particular metropolitan areas predicted lower probabilities of
being employed. In this framework, the expected effect of the job market on employment is
straightforward. However, the expected effects of the immigrant network are ambiguous
because of the reasons discussed above. Better access to jobs may be countered by more
support during times of unemployment.

16

Summary Statistics
The housing market collapse and subsequent recession both lowered employment
for Latino immigrants and generated small changes in the occupation mix of those who
remained working. Table 3 displays the employment rates of Latino Immigrants and the
Other group from 2006-2009 rather than the unemployment rates to avoid issues of
individuals dropping in and out of the labor market. While employment rates fell for nearly
all in the metropolitan areas, the impacts were not even across place and group.
Latino immigrants were hardest hit shortly after the economy began to decline in
contrast to the rest of the population (Labeled Other) which experienced the most dramatic
decline in employment in 2008 and 2009. Latino immigrants were hardest hit in Phoenix (-
9%), Richmond (-7%), and Las Vegas (-6%) over the four year time period. This was in
contrast to much smaller declines in overall employment for the rest of the population.
There were actually small increases in employment of Latino immigrants in San Antonio,
Kansas City, and San Francisco indicating that the recovery had begun in some of the
metros by 2009.
17

It is also important to note that the Latino immigrant population was less settled in
some of the areas that had the greatest change in employment. As evidenced in Table 4,
metropolitan areas differed substantially in the percentage of the population that is Latino
and the percentage of the immigrant population that arrived in the US less than 10 years
ago. Of the 25 areas studied, only four had Latino-immigrant populations that made up
more that 12% of the total population. These metros are in the expected states of
California, Nevada, and Texas.
It is important to highlight that the areas with some of the lowest percentage of
Latino-immigrant populations have the highest percent of recently arrived Latino-
immigrants (within the last 10 years). In 2006, over half of the Latino-immigrants in 15 of
Table 3. Employment rate for Latino Immigrants and Other 2006-2009
Latino
Immigrant Other
Latino
Immigrant Other
Latino
Immigrant Other
Latino
Immigrant Other
Atlanta,GA 78% 74% 73% 70% -5% -3%
Austin,TX 76% 75% 75% 75% 0% -1%
Baltimore,MD 78% 75% 77% 74% -1% -2%
Charlotte-Gastonia-Rock Hill,NC-SC 75% 75% 70% 72% -5% -3%
Columbus,OH 68% 74% 68% 73% 0% -2%
Dallas-FortWorth,TX 75% 75% 73% 73% -2% -1%
Detroit,MI 62% 69% 60% 63% -2% -5%
Greensboro-WinstonSalem-HighPoint,NC 78% 73% 76% 69% -2% -4%
Indianapolis,IN 75% 76% 70% 73% -5% -3%
KansasCity,MO-KS 72% 77% 75% 75% 3% -3%
LasVegas,NV 76% 75% 70% 71% -6% -4%
LosAngeles-Long Beach,CA 72% 71% 71% 69% -1% -2%
Memphis,TN/AR/MS 75% 70% 71% 68% -3% -1%
Nashville,TN 77% 74% 75% 73% -2% -1%
Phoenix,AZ 72% 75% 63% 71% -9% -4%
Raleigh-Durham,NC 77% 75% 77% 73% 0% -2%
Richmond-Petersburg,VA 72% 75% 66% 73% -7% -2%
Riverside-SanBernardino,CA 68% 69% 65% 65% -3% -4%
Sacramento,CA 68% 72% 67% 68% -1% -4%
SanAntonio,TX 67% 71% 72% 71% 5% -1%
SanFrancisco-Oakland-Vallejo,CA 73% 73% 73% 71% 1% -2%
SanJose,CA 73% 72% 71% 71% -1% -1%
Seattle-Everett,WA 76% 76% 70% 74% -5% -1%
Washington,DC/MD/VA 80% 78% 78% 77% -2% -1%
WestPalmBeach-BocaRaton-Delray Beach,FL 75% 74% 74% 69% -1% -4%
2006 2009 2006-09 Difference 2006-09 Employment
18

the 25 metros arrived within the past 10 years to the country, with 10 metros having these
recently arrived Latino-immigrants make up over 60% of the overall Latino-immigrant
population. This stands in contrast to the longer established metros of Riverside and Los
Angeles where less than 30% of the Latino immigrant population has been in the country
less than 10 years, in 2006.

After the recession hit, there was a substantial decrease in the percentage of Latino
immigrants in Phoenix (Table 4). What is particularly interesting is that the decline appears
to be among the recent immigrants. The percentage of recent immigrants in Phoenix fell to
37.8%, a decline of 15 percentage points. The drop in this group of immigrants was only
surpassed by the drop in Memphis where the percent of Latino-immigrants was low to
start with, but the percent of those in the country less than 10 years was very high at
Table 4. Distribution of Latino Immigrant Population
% Latino Of Latino % Latino Of Latino % Latino Of Latino
Immigrant Immig, % New Immigrant Immig, % New Immigrant Immig, % New
Atlanta,GA 5.8 67.4 5.6 54.5 (0.1) (12.9)
Austin,TX 9.9 57.0 9.8 47.1 (0.2) (9.9)
Baltimore,MD 1.4 54.2 1.8 52.7 0.4 (1.6)
Charlotte-Gastonia-Rock Hill,NC-SC 5.0 67.8 5.0 57.3 0.1 (10.5)
Columbus,OH 1.6 68.8 1.7 54.8 0.2 (14.0)
Dallas-FortWorth,TX 12.3 48.5 12.1 41.1 (0.2) (7.5)
Detroit,MI 1.2 48.6 1.1 48.5 (0.1) (0.1)
Greensboro-WinstonSalem-HighPoint,NC 4.5 62.3 4.7 56.0 0.2 (6.3)
Indianapolis,IN 2.3 62.6 2.6 65.2 0.3 2.6
KansasCity,MO-KS 3.2 64.9 3.1 51.8 (0.1) (13.1)
LasVegas,NV 13.3 46.3 12.6 39.6 (0.7) (6.6)
LosAngeles-Long Beach,CA 19.7 28.1 19.6 25.8 (0.1) (2.3)
Memphis,TN/AR/MS 2.3 72.8 2.7 57.3 0.4 (15.5)
Nashville,TN 3.0 67.3 3.3 61.4 0.3 (5.9)
Phoenix,AZ 12.3 52.8 10.8 37.8 (1.6) (15.0)
Raleigh-Durham,NC 5.4 71.3 5.4 58.8 (0.0) (12.6)
Richmond-Petersburg,VA 2.0 66.5 2.2 62.3 0.2 (4.1)
Riverside-SanBernardino,CA 16.1 29.2 15.9 23.3 (0.2) (5.9)
Sacramento,CA 5.9 39.7 5.8 36.6 (0.1) (3.1)
SanAntonio,TX 9.1 34.7 9.5 28.9 0.4 (5.8)
SanFrancisco-Oakland-Vallejo,CA 9.4 39.6 9.5 32.9 0.1 (6.6)
SanJose,CA 10.6 41.0 10.1 35.1 (0.5) (5.9)
Seattle-Everett,WA 3.6 53.9 3.4 51.0 (0.2) (2.8)
Washington,DC/MD/VA 7.4 50.0 7.6 45.8 0.2 (4.2)
WestPalmBeach-BocaRaton-Delray Beach,FL 10.1 49.7 10.7 43.7 0.6 (6.0)
Percentage Point Change in 2009 2006
19

72.8%. The percentage of recent immigrants in all the metropolitan areas fell an average of
7 percentage points, which is partially due to the maturing of the immigrant population and
a reduction in the flow of recent immigrants. While 7 metros had a 10 percentage point or
greater reduction in the percent of recent immigrants, Phoenix clearly experienced the
greatest changes in the Latino immigrant population due to the percent of recent and
established Latino-immigrants who lived there in 2006. This suggests that the recent
immigrants who had lost their jobs in the construction industry or other industries likely
left the metropolitan area altogether.
We next investigated how the change in employment during the recession impacted
the occupation choices of Latino-immigrants. The most dramatic declines were in the
construction occupations, as shown in Table 5 below. Not only did the number of Latino-
immigrants decline dramatically in the construction occupations, but the share of Latino-
immigrants working in construction compared to all occupations also declined. While there
is also a substantial decline in the number in manufacturing occupations, the share of
overall employment changed little over the 2006-09 period, which is similar to what Pew
(2010) had found.

Table 5. Employment by Occupation for Latino Immigrants 2006-2009
People % of Total People % of Total ∆ in People ∆ in % of Total ∆ in % of Total
All other occupations 2,213,062 50.5% 2,242,881 51.2% 1.3% 0.7%
Clothing 67,747 1.5% 67,072 1.5% -1.0% 0.0%
Construction 854,532 19.5% 694,174 15.9% -18.8% -3.7%
Domestic service 632,069 14.4% 700,247 16.0% 10.8% 1.6%
Farm 55,806 1.3% 64,632 1.5% 15.8% 0.2%
Food 385,312 8.8% 461,134 10.5% 19.7% 1.7%
Manufacturing 170,046 3.9% 148,411 3.4% -12.7% -0.5%
Not working 1,636,465 27.2% 1,802,874 29.2% 10.2% 2.0%
2006 2009 2006-09
20

Over the four year period, the working age Latino-immigrant population grew by
approximately 2.8%. While the total number of working people remained relatively
constant overall, Table 5 shows that the percent of people not working grew by 10.2%,
indicating that the other industries did not absorb the losses in construction and
manufacturing jobs and/or the additional workers in the market. The trend lines in the
right column show that the drop in the share of construction and manufacturing jobs
started in 2007, but dropped more rapidly in 2009. However, the other occupation groups
show increases in share by 2008-2009.
Analyzing the data at the metropolitan level provides greater insight into the
changes in construction related occupations in the post-recession period. The top 10
metros,
1
in terms of Latino-immigrants employed in construction in 2006, demonstrated
some of the greatest reductions in the number of people employed in related occupations
(Table 6). While there are other metros that had greater percentage drops, the initial
employment in construction is lower than these top ten.

In Phoenix, the immigrant workforce in construction dropped by over 45% during
the four year period, and the share of Latino-immigrants went from 19.5% to 10.9% of the

1
A full listing of all the metros by occupation group is provided in the Appendix Table A-1.
Table 6. Employment in Construction Occupations 2006-2009
2006 2007 2008 2009 2008-09 2006-09 2006-07 2007-08 2008-09
Atlanta,GA 63,696 57,297 66,450 55,740 (10,710) -12.5% -10.0% 16.0% -16.1%
Austin,TX 28,473 28,445 28,990 29,878 888 4.9% -0.1% 1.9% 3.1%
Dallas-FortWorth,TX 118,502 124,948 125,624 109,958 (15,666) -7.2% 5.4% 0.5% -12.5%
LasVegas,NV 41,382 32,946 30,327 23,335 (6,992) -43.6% -20.4% -7.9% -23.1%
LosAngeles-Long Beach,CA 186,740 185,054 175,672 158,474 (17,198) -15.1% -0.9% -5.1% -9.8%
Phoenix,AZ 75,809 76,607 60,868 41,620 (19,248) -45.1% 1.1% -20.5% -31.6%
Raleigh-Durham,NC 20,911 21,371 21,007 18,141 (2,866) -13.2% 2.2% -1.7% -13.6%
Riverside-SanBernardino,CA 60,280 58,865 53,721 42,756 (10,965) -29.1% -2.3% -8.7% -20.4%
SanFrancisco-Oakland-Vallejo,CA 42,968 46,817 46,579 36,691 (9,888) -14.6% 9.0% -0.5% -21.2%
Washington,DC/MD/VA 57,029 62,311 62,125 56,319 (5,806) -1.2% 9.3% -0.3% -9.3%
21

Latino-immigrant workforce. Similarly, the share in Las Vegas fell from about 20.6% to
11.3% with an overall loss of nearly 44% of the employment in construction. In contrast,
San Francisco and Los Angeles had lower loss of jobs, as a percent, over the full period
while maintaining the relative share of construction employment with San Francisco’s
construction share going from 16.0% to 12.8% and Los Angeles going from 8.5% to 7.2%
from 2006 to 2009. This is likely due to the fact that the immigrant workforce is working in
a broader cross section of industries in these metros.
Also shown in Table 6, the change in employment was not uniform across
metropolitan areas. Double digit losses occurred in nearly all the top ten metros between
2008 and 2009. Only Phoenix, Las Vegas, and Atlanta experienced over 10% losses in the
years prior to 2008-09. In Phoenix, there was a slight increase in the share of food industry
workers, but the largest change was in the number of individuals that were not working.
While there was still a sizeable increase in the number of Latino immigrants that were not
working, Las Vegas had a small increase in manufacturing over the period, and a 3
percentage point increase in workers in the food industry to help cushion the losses in
construction.
The evidence described above suggests that Latino immigrants were more
successful in the more established metropolitan areas than in Las Vegas or Phoenix.
However, immigrants may have also responded to the decline in construction and other
jobs by leaving the metropolitan areas altogether with the hope of finding a job somewhere
else.
22

Despite the recession, the total working population continued to grow in Phoenix
with an increase of 6.2% over the 2006-09 period.
2
This was because more individuals
entered the metropolitan area than left the metro area. However, the Latino-immigrant
working age population dropped by 2% over the same period. Of the top 10 metros listed
in Table 6, the only other metro with a drop in the working age Latino-immigrant
population was Los Angeles, with less than a 1% reduction in the number of people.
The patterns of mobility in Phoenix and Los Angeles were not repeated in many
other study areas with only 5 of the 25 study areas having a net loss in the working age
Latino-immigrant population. The Latino-immigrant working age population grew the
greatest in numbers in Dallas, San Francisco, and San Antonio with the greatest percent
increases in Baltimore, Nashville, and Columbus.
Of the top 10, only Washington, DC and Atlanta had decreases in the Latino-
immigrant working age population between 2006 and 2007. However, six metros had
decreases between 2007 and 2008 with all six having their greatest year to year decrease
during this period. During the final one-year period of the study only four metros had a
decrease with the remainder showing slight improvements over the previous year.
Overall, the data suggest that the impacts of the recession on Latino-immigrants
were much larger in the metropolitan areas of Phoenix and Las Vegas than in other
metropolitan areas. The decline in employment and the shifts away from construction were
both much larger in Phoenix and Las Vegas than most other metros with large Latino-
immigrant populations. Latino-immigrants in Phoenix were particularly hard hit because

2
We are not able to measure mobility out of the United States. However, because the population did not decline
in the metro areas, these missing groups are not likely to be large.
23

there was not an increase in employment in other industries. The largest mobility out of
one of these areas was in Phoenix as well. There are two possible explanations for these
differences. Latino immigrants were much more heavily concentrated in the construction
industry in Las Vegas and Phoenix, which makes them much more vulnerable to the
housing market crash. A second, and although not distinctly different possibility, is that the
Latino immigrant population was less settled in those areas as previously shown in Table 2.
Regression results - Mobility and Employment
Mobility
While the above evidence shows that the construction downturn was an important
determinant of the mobility choices and labor market outcomes of Latino Immigrants, we
now turn to multivariate statistical models to explain the relative impact of household
characteristics and metropolitan level characteristics. Below, we provide evidence on the
predictors of household mobility and on labor market outcomes of Latino immigrants who
left one of our study metropolitan areas. The following models do not show 2006 since it
was used as the base year to determine the change in employment, both construction and
overall, and the base year to determine the mobility/employment outcomes for the groups
observed.
We first test what factors predict mobility using a multinomial logit model.
3
In the
model, we test what household characteristics and metropolitan level characteristics

3
The multinomial logit model is estimated according to the following model. The probability of choosing to not
move, move within the metropolitan area, or to leave the metropolitan area is given by Pj = exp(X'βj )D, j = 1,2,3
and Pm = 1jD where D = 1 + ∑ j = 1 through m-1 exp(X'βj ), (j = 1, 2,3) are the different alternatives, Pj is the
probability of each mobility option j, X is a vector of characteristics, and βj is the vector of coefficients pertaining to
location j. As with the simple bivariate logit model, the coefficients of the MNL are estimated only up to a scale
factor, while the coefficients for the reference choice (βm) are set equal to zero. The MNL model is attractive
because the probability function is of a simple form and is strictly concave; hence, the β vector has a unique
solution which is easily estimable using standard maximum likelihood techniques.
24

increase the likelihood of staying in the same house, making a move within the same
metropolitan, or moving to another metropolitan area. In addition to construction jobs and
the overall job market, we also tested whether the strength of the network ties of the
Latino immigrant community influences their mobility choices. One would expect that
stronger ties would reduce mobility out of the area. We use the percentage of the
population that has been in the area for more than 10 years (Painter and Yu, 2010) as our
proxy of the strength of network ties.
4

Table 7 presents three models that differ only in their inclusion of different
metropolitan level variables. The impacts of household level characteristics on mobility are
as expected. Immigrants who have arrived in more recent year are most likely to move
within or outside of the metropolitan area.
5
Immigrants who have been in the country
more than 10 years are less likely to leave their metropolitan area. Single men are most
likely to move either within or outside of their metropolitan area, followed by single
women who are more likely to move than are married men who are heads of households,
but their probability of moving is substantially less than single men. Interestingly, during
the period between 2007 and 2009, the college educated were most likely to move
between metropolitan areas, but less likely than those with only high school diplomas who
were likely to only move within the metropolitan area. This suggests that the college
educated might be moving across metropolitan areas for job related reasons.

4
We also used a measure of language proficiency to control for how well the immigrant community connects to
the broader market. The percent of Latino Immigrants who speak English only, very well, and well to the overall
Latino Immigrant population was used. The results using this variable did not conform to our expectations that
communities with high levels of this measure would have fewer moves out of the metropolitan area.
5
The omitted category is a Latino immigrant living with a head of household who was born in the US.
25

Table 7. Mobility Regression Model

Model 1
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Moved with Metro Area
Length of Time Head of Household has been in the US
0-10 years 0.616 0.004 0.262 0.004 0.43 0.004
11-20 years 0.068 0.005 -0.301 0.004 0.051 0.004
21+ years -0.512 0.005 -0.739 0.005 -0.559 0.004
Marital Status of the Head of Household
Female HH Married 0.05 0.003 0.125 0.003 0.125 0.003
Male HH Single 0.507 0.003 0.4 0.003 0.509 0.003
Female HH Single 0.372 0.003 0.386 0.003 0.507 0.003
Education level of Head of Household
HS Diploma 0.087 0.003 0.114 0.003 0.163 0.003
Some college 0.14 0.004 0.155 0.004 0.216 0.003
College degree plus 0.038 0.004 0.061 0.004 0.015 0.004
Household Income -1.65E-06 2.68E-08 -2.40E-06 2.80E-08 -2.22E-06 2.80E-08
Change in Construction Employment in Metro 0.052 0.007 0.102 0.003 0.079 0.002
_cons -1.99 0.005 -1.812 0.005 -1.938 0.005
Moved between Metro Areas
Length of Time Head of Household has been in the US
0-10 years 0.267 0.006 0.151 0.006 0.113 0.007
11-20 years -0.239 0.006 -0.455 0.007 -0.522 0.007
21+ years -0.649 0.006 -0.776 0.007 -0.686 0.007
Marital Status of the Head of Household
Female HH Married 0.064 0.005 -0.077 0.005 0.015 0.005
Male HH Single 0.503 0.004 0.37 0.005 0.327 0.005
Female HH Single 0.144 0.005 0.222 0.005 0.243 0.005
Education level of Head of Household
HS Diploma 0.167 0.004 0.046 0.005 0.207 0.005
Some college 0.388 0.006 0.334 0.006 0.534 0.006
College degree plus 0.681 0.005 0.638 0.006 0.658 0.006
Household Income -1.57E-06 3.82E-08 -1.06e 06 3.78E-08 -1.34E-06 4.33E-08
Change in Construction Employment in Metro 0.349 0.01 -0.128 0.006 -0.193 0.004
_cons -2.756 0.007 -2.785 0.007 -2.925 0.008
Observations 7,260,175 7,141,257 7,229,271
Pseudo R2 0.0343 0.0303 0.0323
2007 2008 2009
26

Table 7, continued

Model 2
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Moved with Metro Area
Length of Time Head of Household has been in the US
0-10 years 0.608 0.004 0.25 0.004 0.426 0.004
11-20 years 0.07 0.005 -0.301 0.005 0.055 0.004
21+ years -0.504 0.005 -0.723 0.005 -0.55 0.004
Marital Status of the Head of Household
Female HH Married 0.052 0.003 0.132 0.003 0.13 0.003
Male HH Single 0.509 0.003 0.407 0.003 0.512 0.003
Female HH Single 0.38 0.003 0.395 0.003 0.517 0.003
Education level of Head of Household
HS Diploma 0.087 0.003 0.12 0.003 0.154 0.003
Some college 0.142 0.004 0.164 0.004 0.213 0.003
College degree plus 0.035 0.004 0.06 0.004 0.01 0.004
Household Income -1.6E-06 2.7E-08 -2.37E-06 2.79E-08 -2.18E-06 2.80E-08
Change in Construction Employment in Metro 0.071 0.007 0.042 0.004 -0.116 0.003
Change in Overall Employment in Metro 3.068 0.074 5.655 0.089 -9.06 0.083
_cons -2.022 0.005 -1.967 0.006 -2.205 0.006
Moved between Metro Areas
Length of Time Head of Household has been in the US
0-10 years 0.249 0.006 0.132 0.006 0.108 0.007
11-20 years -0.234 0.006 -0.457 0.007 -0.52 0.007
21+ years -0.629 0.006 -0.751 0.007 -0.679 0.007
Marital Status of the Head of Household
Female HH Married 0.069 0.005 -0.069 0.005 0.019 0.005
Male HH Single 0.509 0.004 0.379 0.005 0.33 0.005
Female HH Single 0.162 0.005 0.236 0.005 0.252 0.005
Education level of Head of Household
HS Diploma 0.17 0.004 0.056 0.005 0.198 0.005
Some college 0.393 0.006 0.349 0.006 0.53 0.006
College degree plus 0.675 0.005 0.636 0.006 0.654 0.006
Household Income -1.5E-06 3.8E-08 -1E-06 3.77E-08 -1.31E-06 4.32E-08
Change in Construction Employment in Metro 0.374 0.01 -0.22 0.006 -0.372 0.005
Change in Overall Employment in Metro 6.853 0.11 8.52 0.137 -8.215 0.141
_cons -2.834 0.007 -3.02 0.008 -3.164 0.009
Observations 7260175 7141257 7229271
Pseudo R2 0.0348 0.0312 0.0340
2007 2008 2009
27

Table 7, continued

The impact of living in an area with positive construction job growth changed over
the period. In 2007, before the general economy fell, but after the housing market had
Model 3
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Moved with Metro Area
Length of Time Head of Household has been in the US
0-10 years 0.594 0.004 0.215 0.004 0.402 0.004
11-20 years 0.069 0.005 -0.307 0.005 0.052 0.004
21+ years -0.494 0.005 -0.704 0.005 -0.526 0.004
Marital Status of the Head of Household
Female HH Married 0.054 0.003 0.142 0.003 0.139 0.003
Male HH Single 0.507 0.003 0.406 0.003 0.515 0.003
Female HH Single 0.382 0.003 0.405 0.003 0.53 0.003
Education level of Head of Household
HS Diploma 0.087 0.003 0.112 0.003 0.154 0.003
Some college 0.138 0.004 0.159 0.004 0.213 0.003
College degree plus 0.026 0.004 0.031 0.004 -0.008 0.004
Household Income -1.6E-06 2.7E-08 -2.35E-06 2.79E-08 -2E-06 2.8E-08
Change in Construction Employment in Metro 0.001 0.007 -0.009 0.004 0.033 0.003
Change in Overall Employment in Metro 1.084 0.09 0.743 0.104 -7.264 0.085
Percent of Latino Immigrants in US over 10yrs -0.404 0.011 -1.059 0.012 -0.932 0.012
_cons -1.77 0.008 -1.183 0.01 -1.616 0.009
Moved between Metro Areas
Length of Time Head of Household has been in the US
0-10 years 0.211 0.006 0.027 0.006 0.041 0.007
11-20 years -0.236 0.006 -0.475 0.007 -0.53 0.007
21+ years -0.603 0.006 -0.692 0.007 -0.617 0.007
Marital Status of the Head of Household
Female HH Married 0.072 0.005 -0.039 0.005 0.037 0.005
Male HH Single 0.504 0.004 0.379 0.005 0.336 0.005
Female HH Single 0.166 0.005 0.266 0.005 0.285 0.005
Education level of Head of Household
HS Diploma 0.169 0.004 0.033 0.005 0.201 0.005
Some college 0.384 0.006 0.333 0.006 0.532 0.006
College degree plus 0.653 0.005 0.553 0.006 0.611 0.006
Household Income -1.5E-06 3.8E-08 -9.4E-07 3.74E-08 -9.9E-07 4.3E-08
Change in Construction Employment in Metro 0.177 0.01 -0.35 0.006 0.019 0.006
Change in Overall Employment in Metro 1.601 0.134 -4.157 0.156 -3.879 0.142
Percent of Latino Immigrants in US over 10yrs -1.074 0.016 -2.907 0.017 -2.406 0.02
_cons -2.171 0.012 -0.922 0.015 -1.663 0.015
Observations 7260175 7141257 7229271
Pseudo R2 0.0354 0.0354 0.0364
2007 2008 2009
28

stagnated, living in a metropolitan area with higher construction job growth predicted
more mobility out of the metropolitan area (Model 1). However, in 2008 and 2009, there
was a clear change in the likelihood of leaving metropolitan areas with higher construction
job growth. We next added a variable which measures overall employment changes in the
metropolitan area (Model 2). Similar to Model 1, both overall employment and
construction jobs predict more mobility in 2007. In 2008, places with more construction
jobs had fewer immigrants leaving them, but this was not true for overall employment. By
2009, immigrants were more likely to leave metropolitan areas that had weaker job
markets. This suggests the evolution of the recession from the housing market to the
overall economy, by 2009, impacted mobility differently.
Next we added the variable that measures the permanence of immigrant population
in the metropolitan area (Model 3). We find very consistent evidence that immigrants
living in metropolitan areas that had a higher percentage of immigrants who had been in
the area more than 10 years, were less likely to move both within the metropolitan area
and to leave the metropolitan area that they were in before. As we would expect, the
reduction in mobility was greatest for moves away from the metropolitan area. Most of the
other variables were similar in the 3 models across the years. The one exception is that the
number of construction jobs in a metro area in 2009 became much less important and even
changed signs.
Our interpretation of this collection of findings is that the changes in construction
jobs were important predictors of mobility in the early parts of the recession as that
industry was experiencing job losses first; then the changes in the overall job market were
the strongest determinant of mobility decisions after the recession was widespread. The
29

permanence of the immigrant population was a consistent predictor of mobility across the
time periods studied, and may have increased in importance as the recession worsened.
Employment
In Table 8, we present evidence concerning the likelihood that someone will be
employed given a set of household characteristics and mobility characteristics. We first
note that immigrants living in households with native-born heads of household are the
least likely to work. Interestingly, there were not large differences in employment status as
it relates to the length of time that an immigrant had been in the country, possibly
suggesting that immigrants are most likely to compete with other immigrants in the labor
market (Orrenius and Zavodny, 2013). The immigrants living with household heads who
had most recently arrived had the highest likelihood of working, and this was true across
the time period 2007-2009. It is important to note that the marital status variables relate to
the head of household. These results suggest that in total, living with a household head that
is a married male reduces the likelihood of working. These results are also consistent
across 2007 - 2009. The importance of being college educated grew over the period, and
the impact of age was relatively stable across the period.
While the literature on mobility suggests that a primary reason that someone makes
an inter urban move is employment related (Gabriel, Mattey, and Washer, 2003), it need
not be the case that the entirety of those moving would be expected to have better labor
market outcomes than those who don’t move. We observe this fact in Table 8. Latino
immigrants who did move across metropolitan areas were less likely to be employed than
those that stayed. The model does not allow for such a distinction, but it very well might be
the case that these individuals had lost their jobs in the previous period and were looking
30

for work. The intra-urban movers were almost as likely to work as those had not moved at
all. The estimate moved from positive to negative over the period, but remained much
smaller than the effect of having moved from another metropolitan area.
Table 8. Probability of being Employed

Model 1
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Employed
Age Group of Head of Household
26-35 years 0.008 0.004 -0.075 0.004 0.002 0.004
36-45 years 0.055 0.004 -0.012 0.004 0.032 0.004
46-55 years 0.057 0.004 0.048 0.004 0.026 0.004
56+ years -0.405 0.004 -0.432 0.005 -0.323 0.004
Length of Time Head of Household has been in the US
0-10 years 0.376 0.004 0.336 0.004 0.327 0.004
11-20 years 0.277 0.004 0.235 0.004 0.261 0.004
21+ years 0.279 0.004 0.222 0.004 0.274 0.003
Marital Status of the Head of Household
Female HH Married 0.132 0.002 0.071 0.002 0.117 0.002
Male HH Single 0.596 0.003 0.443 0.003 0.392 0.003
Female HH Single 0.204 0.003 0.198 0.003 0.182 0.003
Education level of Head of Household
HS Diploma 0.084 0.002 0.134 0.002 0.168 0.002
Some college 0.268 0.003 0.285 0.003 0.338 0.003
College degree plus 0.341 0.003 0.461 0.004 0.428 0.003
Moved
Within Metro 0.089 0.003 0.028 0.003 -0.045 0.003
Between Metros -0.332 0.004 -0.296 0.004 -0.376 0.004
_cons 0.478 0.005 0.658 0.005 0.430 0.005
Observations 6286409 6202074 6291742
Pseudo R2 0.0158 0.0129 0.0108
2007 2008 2009
31

Table 8, continued

Model 2
Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
Employed
Age Group of Head of Household
26-35 years -0.004 0.004 -0.070 0.004 -0.003 0.004
36-45 years 0.040 0.004 -0.003 0.004 0.025 0.004
46-55 years 0.040 0.004 0.055 0.004 0.018 0.004
56+ years -0.423 0.004 -0.427 0.005 -0.345 0.005
Length of Time Head of Household has been in the US
0-10 years 0.342 0.004 0.272 0.004 0.286 0.004
11-20 years 0.267 0.004 0.208 0.004 0.231 0.004
21+ years 0.299 0.004 0.246 0.004 0.276 0.004
Marital Status of the Head of Household
Female HH Married 0.137 0.002 0.096 0.002 0.117 0.002
Male HH Single 0.584 0.003 0.448 0.003 0.381 0.003
Female HH Single 0.215 0.003 0.216 0.003 0.180 0.003
Education level of Head of Household
HS Diploma 0.080 0.002 0.139 0.002 0.176 0.002
Some college 0.268 0.003 0.284 0.003 0.343 0.003
College degree plus 0.320 0.003 0.421 0.004 0.410 0.003
Moved
Within Metro 0.087 0.003 0.017 0.003 -0.019 0.003
Between Metros -0.304 0.005 -0.205 0.006 -0.244 0.006
Change in Construction Employment in Metro -0.618 0.029 0.136 0.018 0.098 0.018
Change in Overall Employment in Metro 1.198 0.356 11.713 0.482 -2.675 0.399
Percent of Latino Immigrants in US over 10yrs -2.258 0.063 -0.793 0.061 0.022 0.066
Las Vegas, NV 0.113 0.010 0.216 0.013 0.175 0.012
Los Angeles-Long Beach, CA 0.432 0.013 0.393 0.014 0.243 0.014
Phoenix, AZ -0.169 0.010 -0.160 0.012 -0.125 0.014
Riverside-San Bernardino,CA 0.325 0.014 0.401 0.017 -0.184 0.021
Dallas-Fort Worth, TX 0.165 0.009 0.141 0.015 0.390 0.011
Washington, DC/MD/VA 0.399 0.010 0.458 0.013 0.661 0.011
Atlanta, GA -0.399 0.016 0.282 0.014 0.312 0.014
San Francisco-Oakland-Vallejo, CA 0.239 0.009 0.245 0.013 0.420 0.013
Detroit, MI -0.545 0.016 0.232 0.016 -0.391 0.024
Baltimore, MD -0.217 0.024 0.367 0.018 0.614 0.016
Seattle-Everett, WA -0.017 0.017 0.259 0.014 0.272 0.013
Sacramento, CA -0.025 0.013 0.442 0.015 -0.136 0.020
Kansas City, MO-KS -0.603 0.024 -0.125 0.025 0.721 0.018
Charlotte-Gastonia-Rock Hill, SC -0.178 0.020 -0.099 0.019 0.155 0.017
San Antonio, TX 0.012 0.014 -0.084 0.017 0.494 0.018
Indianapolis, IN -0.124 0.018 0.265 0.023 0.326 0.020
San Jose, CA 0.340 0.011 0.190 0.012 0.287 0.017
Columbus, OH -0.381 0.023 -0.152 0.027 0.112 0.020
Raleigh-Durham, NC -0.079 0.022 0.104 0.016 0.603 0.018
Austin, TX -0.339 0.023 0.232 0.018 0.736 0.018
Nashville, TN 0.055 0.023 0.141 0.019 0.534 0.019
Greensboro-Winston Salem-High Point, NC 0.125 0.020 0.269 0.016 0.467 0.017
West Palm Beach-Boca Raton-Delray Beach, FL 0.314 0.017 0.488 0.017 0.352 0.011
Richmond-Petersburg, VA -0.053 0.027 -0.001 0.021 -0.017 0.025
Memphis, TN/AR/MS -0.742 0.024 0.627 0.025 0.411 0.023
_cons 1.560 0.036 0.565 0.042 0.079 0.041
Observations 6231200 6147367 6236079
Pseudo R2 0.0194 0.0178 0.0159
2007 2008 2009
32

In Model 2, we added metropolitan level controls and controls for having lived in
one of our study metropolitan areas in the previous year. We find that the changes in the
broader job market were very strongly predictive of working in the periods after the
recession hit (2008-2009). Changes in the construction industry specifically, had a smaller
and mixed effect over the period. We found that living in metropolitan areas with stronger
immigrants networks were less likely to work in the early part of the recession (2007-
2008). One explanation is that individuals can rely on these networks to provide resources
in lieu of employment. Finally, we found that residing in Memphis, Kansas, and Detroit in
the previous period predicted the lowest probability of working in 2007 with improvement
over the following years. Residents in California metros fared better overall.
Conclusion
This research documents the changes in employment for Latino immigrants from
2006-2009 and explores reasons for the mobility patterns and employment outcomes in
the 25 metropolitan areas where Latino immigrants were most concentrated in
construction employment. This provides an interesting example of a natural experiment of
a targeted and a general employment shock, and provides evidence of how Latino
immigrants adjusted over the recession. The data suggest that the loss in construction jobs
was offset only slightly by increases in the share employed in other industries. Instead, the
evidence suggests that the workers who lost jobs in construction were either unemployed,
exited the metropolitan area, or both.
Evidence from the regression models confirms these patterns in the data. We found
that job losses in construction spurred exit between 2007-2008, and that job losses at the
metropolitan level, more generally, dominated the reasons for exit. We also found evidence
that those who moved out of the metropolitan area were less likely to be employed. As
33

noted previously, many of these immigrants would likely not have been employed in the
previous metropolitan area. The permanence of the immigrant population in the
metropolitan area had two effects which might be related. Immigrants were less likely to
move when living in a metropolitan area with more immigrants who had been in the
country for more than 10 years, and less likely to work in those same areas. We posit that
the resources within the immigrant community may have contributed to these two
findings. This may also be connected to the income levels and ability to afford to move as
has been shown in previous research related to the manufacturing job losses in past
decades.
The Great Recession has had dramatic impacts on the labor market in general and
on Latino immigrants specifically. As the recovery continues, additional research will need
to follow those who moved in order to observe whether they eventually improved their
labor market outcomes and whether Latino immigrant employment in construction is able
to return to the pre-recession levels as quickly as other groups.

34

Appendices

Appendix Table A-1. Change in Number of Latino-Immigrants by Occupation Group
* Denotes alternate time period
All other occupations 2006 2007 2008 2009 2006-07 2007-08 2008-09 2006-09
Atlanta,GA 73,764 78,837 75,048 70,769 6.9% -4.8% -5.7% -4.1%
Austin,TX 31,246 41,834 35,933 38,068 33.9% -14.1% 5.9% 21.8%
Baltimore,MD 12,696 11,090 14,362 14,170 -12.6% 29.5% -1.3% 11.6%
Charlotte-Gastonia-Rock Hill,NC-SC 20,122 22,899 25,813 24,207 13.8% 12.7% -6.2% 20.3%
Columbus,OH 5,971 4,987 8,680 8,614 -16.5% 74.1% -0.8% 44.3%
Dallas-FortWorth,TX 215,579 214,219 223,482 214,661 -0.6% 4.3% -3.9% -0.4%
Detroit,MI 16,078 13,902 14,617 12,034 -13.5% 5.1% -17.7% -25.2%
Greensboro-WinstonSalem-HighPoint,NC 13,369 13,536 18,135 18,916 1.2% 34.0% 4.3% 41.5%
Indianapolis,IN 10,052 9,864 13,051 11,451 -1.9% 32.3% -12.3% 13.9%
KansasCity,MO-KS 13,500 15,684 17,069 14,488 16.2% 8.8% -15.1% 7.3%
LasVegas,NV 52,208 61,889 61,241 54,443 18.5% -1.0% -11.1% 4.3%
LosAngeles-Long Beach,CA 926,581 906,387 909,339 890,941 -2.2% 0.3% -2.0% -3.8%
Memphis,TN/AR/MS 5,354 5,420 7,297 6,480 1.2% 34.6% -11.2% 21.0%
Nashville,TN 10,645 11,515 9,661 12,758 8.2% -16.1% 32.1% 19.8%
Phoenix,AZ 109,962 117,614 126,544 112,223 7.0% 7.6% -11.3% 2.1%
Raleigh-Durham,NC 16,462 17,657 24,031 19,899 7.3% 36.1% -17.2% 20.9%
Richmond-Petersburg,VA 4,012 3,842 5,008 6,923 -4.2% 30.3% 38.2% 72.6%
Riverside-SanBernardino,CA 218,813 239,742 227,877 219,720 9.6% -4.9% -3.6% 0.4%
Sacramento,CA 27,087 28,675 25,514 26,506 5.9% -11.0% 3.9% -2.1%
SanAntonio,TX 42,800 52,105 53,912 58,566 21.7% 3.5% 8.6% 36.8%
SanFrancisco-Oakland-Vallejo,CA 137,717 136,620 141,097 146,391 -0.8% 3.3% 3.8% 6.3%
SanJose,CA 54,887 58,180 59,125 54,851 6.0% 1.6% -7.2% -0.1%
Seattle-Everett,WA 22,657 27,040 21,631 22,175 19.3% -20.0% 2.5% -2.1%
Washington,DC/MD/VA 131,469 130,490 136,174 137,434 -0.7% 4.4% 0.9% 4.5%
WestPalmBeach-BocaRaton-Delray Beach,FL 40,031 41,237 40,254 46,193 3.0% -2.4% 14.8% 15.4%
35

Clothing 2006 2007 2008 2009 2006-07 2007-08 2008-09 2006-09
Atlanta,GA 1,882 1,258 1,571 1,610 -33.2% 24.9% 2.5% -14.5%
Austin,TX 709 250 796 -64.7% 218.4% 12.3% *
Baltimore,MD 309 542 84 -84.5% -72.8%
Charlotte-Gastonia-Rock Hill,NC-SC 821 626 712 765 -23.8% 13.7% 7.4% -6.8%
Columbus,OH
Dallas-FortWorth,TX 2,980 4,789 3,430 4,772 60.7% -28.4% 39.1% 60.1%
Detroit,MI 116 104 -10.3% *
Greensboro-WinstonSalem-HighPoint,NC 2,382 2,547 3,124 2,567 6.9% 22.7% -17.8% 7.8%
Indianapolis,IN 242 119 110 841 -50.8% -7.6% 664.5% 247.5%
KansasCity,MO-KS 378 478 120 -74.9% -68.3%
LasVegas,NV 748 777 1,185 1,051 3.9% 52.5% -11.3% 40.5%
LosAngeles-Long Beach,CA 49,906 46,367 46,825 45,662 -7.1% 1.0% -2.5% -8.5%
Memphis,TN/AR/MS 616 238 365 -61.4% 53.4% -40.7% *
Nashville,TN 77 761 888.3%
Phoenix,AZ 753 1,171 2,102 1,085 55.5% 79.5% -48.4% 44.1%
Raleigh-Durham,NC 20 773 994 541 3765.0% 28.6% -45.6% 2605.0%
Richmond-Petersburg,VA 221 35 -84.2% *
Riverside-SanBernardino,CA 2,491 2,826 1,567 3,596 13.4% -44.6% 129.5% 44.4%
Sacramento,CA 192 88 241 450 -54.2% 173.9% 86.7% 134.4%
SanAntonio,TX 447 538 749 442 20.4% 39.2% -41.0% -1.1%
SanFrancisco-Oakland-Vallejo,CA 1,253 1,163 1,479 800 -7.2% 27.2% -45.9% -36.2%
SanJose,CA 617 226 444 441 -63.4% 96.5% -0.7% -28.5%
Seattle-Everett,WA 456 274 231 44 -39.9% -15.7% -81.0% -90.4%
Washington,DC/MD/VA 771 251 958 322 -67.4% 281.7% -66.4% -58.2%
WestPalmBeach-BocaRaton-Delray Beach,FL 313 466 103 614 48.9% -77.9% 496.1% 96.2%
Construction 2006 2007 2008 2009 2006-07 2007-08 2008-09 2006-09
Atlanta,GA 63,696 57,297 66,450 55,740 -10.0% 16.0% -16.1% -12.5%
Austin,TX 28,473 28,445 28,990 29,878 -0.1% 1.9% 3.1% 4.9%
Baltimore,MD 5,441 5,124 5,927 5,915 -5.8% 15.7% -0.2% 8.7%
Charlotte-Gastonia-Rock Hill,NC-SC 17,338 15,383 18,779 16,192 -11.3% 22.1% -13.8% -6.6%
Columbus,OH 2,910 2,359 1,979 3,305 -18.9% -16.1% 67.0% 13.6%
Dallas-FortWorth,TX 118,502 124,948 125,624 109,958 5.4% 0.5% -12.5% -7.2%
Detroit,MI 4,332 6,596 6,312 3,297 52.3% -4.3% -47.8% -23.9%
Greensboro-WinstonSalem-HighPoint,NC 13,103 10,504 10,542 6,730 -19.8% 0.4% -36.2% -48.6%
Indianapolis,IN 6,598 6,266 5,396 5,452 -5.0% -13.9% 1.0% -17.4%
KansasCity,MO-KS 8,623 5,188 8,951 7,740 -39.8% 72.5% -13.5% -10.2%
LasVegas,NV 41,382 32,946 30,327 23,335 -20.4% -7.9% -23.1% -43.6%
LosAngeles-Long Beach,CA 186,740 185,054 175,672 158,474 -0.9% -5.1% -9.8% -15.1%
Memphis,TN/AR/MS 5,384 4,639 3,664 4,145 -13.8% -21.0% 13.1% -23.0%
Nashville,TN 8,113 8,147 11,573 7,647 0.4% 42.1% -33.9% -5.7%
Phoenix,AZ 75,809 76,607 60,868 41,620 1.1% -20.5% -31.6% -45.1%
Raleigh-Durham,NC 20,911 21,371 21,007 18,141 2.2% -1.7% -13.6% -13.2%
Richmond-Petersburg,VA 5,820 4,661 5,104 2,699 -19.9% 9.5% -47.1% -53.6%
Riverside-SanBernardino,CA 60,280 58,865 53,721 42,756 -2.3% -8.7% -20.4% -29.1%
Sacramento,CA 13,941 13,612 9,424 7,314 -2.4% -30.8% -22.4% -47.5%
SanAntonio,TX 20,631 18,355 17,838 20,270 -11.0% -2.8% 13.6% -1.7%
SanFrancisco-Oakland-Vallejo,CA 42,968 46,817 46,579 36,691 9.0% -0.5% -21.2% -14.6%
SanJose,CA 16,557 18,920 17,473 14,933 14.3% -7.6% -14.5% -9.8%
Seattle-Everett,WA 10,552 10,067 14,869 5,657 -4.6% 47.7% -62.0% -46.4%
Washington,DC/MD/VA 57,029 62,311 62,125 56,319 9.3% -0.3% -9.3% -1.2%
WestPalmBeach-BocaRaton-Delray Beach,FL 19,399 12,632 13,354 9,966 -34.9% 5.7% -25.4% -48.6%
36

Domestic service 2006 2007 2008 2009 2006-07 2007-08 2008-09 2006-09
Atlanta,GA 25,851 22,290 26,854 28,268 -13.8% 20.5% 5.3% 9.3%
Austin,TX 16,119 15,154 17,501 16,536 -6.0% 15.5% -5.5% 2.6%
Baltimore,MD 2,887 4,175 5,359 6,212 44.6% 28.4% 15.9% 115.2%
Charlotte-Gastonia-Rock Hill,NC-SC 6,018 6,190 7,507 7,275 2.9% 21.3% -3.1% 20.9%
Columbus,OH 2,228 2,471 232 1,972 10.9% -90.6% 750.0% -11.5%
Dallas-FortWorth,TX 59,068 60,888 70,923 68,701 3.1% 16.5% -3.1% 16.3%
Detroit,MI 4,552 4,007 5,287 3,242 -12.0% 31.9% -38.7% -28.8%
Greensboro-WinstonSalem-HighPoint,NC 2,990 2,982 4,294 3,597 -0.3% 44.0% -16.2% 20.3%
Indianapolis,IN 3,999 5,446 3,043 3,341 36.2% -44.1% 9.8% -16.5%
KansasCity,MO-KS 5,789 8,634 5,820 7,870 49.1% -32.6% 35.2% 35.9%
LasVegas,NV 30,870 34,426 34,676 32,126 11.5% 0.7% -7.4% 4.1%
LosAngeles-Long Beach,CA 214,043 219,185 212,792 231,265 2.4% -2.9% 8.7% 8.0%
Memphis,TN/AR/MS 1,575 1,167 2,655 1,542 -25.9% 127.5% -41.9% -2.1%
Nashville,TN 4,417 4,426 5,057 6,117 0.2% 14.3% 21.0% 38.5%
Phoenix,AZ 48,995 56,790 53,968 44,373 15.9% -5.0% -17.8% -9.4%
Raleigh-Durham,NC 5,969 6,466 4,925 10,911 8.3% -23.8% 121.5% 82.8%
Richmond-Petersburg,VA 2,034 1,529 2,225 1,652 -24.8% 45.5% -25.8% -18.8%
Riverside-SanBernardino,CA 42,300 48,601 40,113 46,714 14.9% -17.5% 16.5% 10.4%
Sacramento,CA 10,889 15,240 11,955 13,190 40.0% -21.6% 10.3% 21.1%
SanAntonio,TX 10,949 12,151 11,697 15,309 11.0% -3.7% 30.9% 39.8%
SanFrancisco-Oakland-Vallejo,CA 42,879 51,748 53,362 54,420 20.7% 3.1% 2.0% 26.9%
SanJose,CA 21,629 23,970 21,909 21,586 10.8% -8.6% -1.5% -0.2%
Seattle-Everett,WA 9,827 6,429 10,720 10,217 -34.6% 66.7% -4.7% 4.0%
Washington,DC/MD/VA 44,940 39,242 45,380 45,623 -12.7% 15.6% 0.5% 1.5%
WestPalmBeach-BocaRaton-Delray Beach,FL 11,252 12,448 15,876 18,188 10.6% 27.5% 14.6% 61.6%
Farm 2006 2007 2008 2009 2006-07 2007-08 2008-09 2006-09
Atlanta,GA 1,530 2,409 618 1,290 57.5% -74.3% 108.7% -15.7%
Austin,TX 480 384 1,267 584 -20.0% 229.9% -53.9% 21.7%
Baltimore,MD 526 479 379 465 -8.9% -20.9% 22.7% -11.6%
Charlotte-Gastonia-Rock Hill,NC-SC 820 1,189 276 763 45.0% -76.8% 176.4% -7.0%
Columbus,OH 126 112 -11.1%
Dallas-FortWorth,TX 2,859 1,994 1,574 3,834 -30.3% -21.1% 143.6% 34.1%
Detroit,MI 229 120 137 368 -47.6% 14.2% 168.6% 60.7%
Greensboro-WinstonSalem-HighPoint,NC 123 765 470 1,045 522.0% -38.6% 122.3% 749.6%
Indianapolis,IN 197 424 374 203 115.2% -11.8% -45.7% 3.0%
KansasCity,MO-KS 34 462 104 296 1258.8% -77.5% 184.6% 770.6%
LasVegas,NV 895 1,510 352 578 68.7% -76.7% 64.2% -35.4%
LosAngeles-Long Beach,CA 13,417 10,615 14,148 19,767 -20.9% 33.3% 39.7% 47.3%
Memphis,TN/AR/MS 56 1,382 350 2367.9% -74.7% 525.0% *
Nashville,TN 148 598 646 253 304.1% 8.0% -60.8% 70.9%
Phoenix,AZ 4,589 3,809 3,855 3,717 -17.0% 1.2% -3.6% -19.0%
Raleigh-Durham,NC 1,045 1,407 2,692 262 34.6% 91.3% -90.3% -74.9%
Richmond-Petersburg,VA 64 131 102 104.7% 59.4%
Riverside-SanBernardino,CA 8,554 10,938 11,126 11,126 27.9% 1.7% 0.0% 30.1%
Sacramento,CA 2,042 1,748 1,263 2,236 -14.4% -27.7% 77.0% 9.5%
SanAntonio,TX 154 490 805 1,187 218.2% 64.3% 47.5% 670.8%
SanFrancisco-Oakland-Vallejo,CA 7,669 8,506 5,117 6,102 10.9% -39.8% 19.2% -20.4%
SanJose,CA 2,821 1,194 2,307 3,478 -57.7% 93.2% 50.8% 23.3%
Seattle-Everett,WA 1,922 1,582 1,107 2,330 -17.7% -30.0% 110.5% 21.2%
Washington,DC/MD/VA 2,363 2,762 3,291 975 16.9% 19.2% -70.4% -58.7%
WestPalmBeach-BocaRaton-Delray Beach,FL 3,199 2,578 3,989 3,209 -19.4% 54.7% -19.6% 0.3%
37

Food 2006 2007 2008 2009 2006-07 2007-08 2008-09 2006-09
Atlanta,GA 12,688 11,754 19,148 16,264 -7.4% 62.9% -15.1% 28.2%
Austin,TX 11,204 10,231 11,343 13,717 -8.7% 10.9% 20.9% 22.4%
Baltimore,MD 2,432 2,823 2,024 3,396 16.1% -28.3% 67.8% 39.6%
Charlotte-Gastonia-Rock Hill,NC-SC 8,944 4,958 4,876 5,304 -44.6% -1.7% 8.8% -40.7%
Columbus,OH 2,086 2,490 3,398 1,851 19.4% 36.5% -45.5% -11.3%
Dallas-FortWorth,TX 39,104 45,561 44,645 60,395 16.5% -2.0% 35.3% 54.4%
Detroit,MI 2,180 2,669 5,199 3,726 22.4% 94.8% -28.3% 70.9%
Greensboro-WinstonSalem-HighPoint,NC 5,751 4,595 2,664 5,020 -20.1% -42.0% 88.4% -12.7%
Indianapolis,IN 2,554 2,315 3,014 3,446 -9.4% 30.2% 14.3% 34.9%
KansasCity,MO-KS 6,293 4,295 5,526 5,515 -31.7% 28.7% -0.2% -12.4%
LasVegas,NV 22,867 23,025 25,659 30,113 0.7% 11.4% 17.4% 31.7%
LosAngeles-Long Beach,CA 110,842 136,914 133,833 132,807 23.5% -2.3% -0.8% 19.8%
Memphis,TN/AR/MS 739 492 1,070 2,693 -33.4% 117.5% 151.7% 264.4%
Nashville,TN 2,587 1,959 3,066 3,444 -24.3% 56.5% 12.3% 33.1%
Phoenix,AZ 30,961 30,346 27,923 30,044 -2.0% -8.0% 7.6% -3.0%
Raleigh-Durham,NC 4,704 5,714 5,556 5,334 21.5% -2.8% -4.0% 13.4%
Richmond-Petersburg,VA 993 1,829 2,312 1,464 84.2% 26.4% -36.7% 47.4%
Riverside-SanBernardino,CA 24,600 25,389 29,496 27,917 3.2% 16.2% -5.4% 13.5%
Sacramento,CA 4,895 6,679 13,218 10,558 36.4% 97.9% -20.1% 115.7%
SanAntonio,TX 8,033 6,654 6,776 8,180 -17.2% 1.8% 20.7% 1.8%
SanFrancisco-Oakland-Vallejo,CA 27,630 32,008 29,802 34,147 15.8% -6.9% 14.6% 23.6%
SanJose,CA 15,040 11,692 11,058 11,247 -22.3% -5.4% 1.7% -25.2%
Seattle-Everett,WA 7,992 12,586 8,777 10,550 57.5% -30.3% 20.2% 32.0%
Washington,DC/MD/VA 27,745 26,382 25,089 29,068 -4.9% -4.9% 15.9% 4.8%
WestPalmBeach-BocaRaton-Delray Beach,FL 2,448 2,964 4,311 4,934 21.1% 45.4% 14.5% 101.6%
Manufacturing 2006 2007 2008 2009 2006-07 2007-08 2008-09 2006-09
Atlanta,GA 5,135 4,334 4,443 5,153 -15.6% 2.5% 16.0% 0.4%
Austin,TX 1,119 1,629 1,855 1,863 45.6% 13.9% 0.4% 66.5%
Baltimore,MD 73 57 96 537 -21.9% 68.4% 459.4% 635.6%
Charlotte-Gastonia-Rock Hill,NC-SC 2,629 3,050 3,103 2,659 16.0% 1.7% -14.3% 1.1%
Columbus,OH 591 732 48 23.9% -91.9%
Dallas-FortWorth,TX 26,174 22,357 23,100 19,947 -14.6% 3.3% -13.6% -23.8%
Detroit,MI 894 1,680 1,681 977 87.9% 0.1% -41.9% 9.3%
Greensboro-WinstonSalem-HighPoint,NC 3,059 6,104 4,030 4,271 99.5% -34.0% 6.0% 39.6%
Indianapolis,IN 1,651 1,500 2,173 2,270 -9.1% 44.9% 4.5% 37.5%
KansasCity,MO-KS 859 1,872 2,123 786 117.9% 13.4% -63.0% -8.5%
LasVegas,NV 2,163 2,091 731 2,293 -3.3% -65.0% 213.7% 6.0%
LosAngeles-Long Beach,CA 75,025 79,699 79,462 63,379 6.2% -0.3% -20.2% -15.5%
Memphis,TN/AR/MS 1,661 875 973 554 -47.3% 11.2% -43.1% -66.6%
Nashville,TN 1,325 887 1,390 990 -33.1% 56.7% -28.8% -25.3%
Phoenix,AZ 5,815 7,519 5,691 4,618 29.3% -24.3% -18.9% -20.6%
Raleigh-Durham,NC 985 970 930 1,489 -1.5% -4.1% 60.1% 51.2%
Richmond-Petersburg,VA 251 262 383 88 4.4% 46.2% -77.0% -64.9%
Riverside-SanBernardino,CA 18,015 16,549 21,431 15,842 -8.1% 29.5% -26.1% -12.1%
Sacramento,CA 2,721 1,048 953 1,596 -61.5% -9.1% 67.5% -41.3%
SanAntonio,TX 3,169 1,525 2,084 2,568 -51.9% 36.7% 23.2% -19.0%
SanFrancisco-Oakland-Vallejo,CA 8,107 7,067 5,711 7,955 -12.8% -19.2% 39.3% -1.9%
SanJose,CA 3,597 3,588 1,992 2,503 -0.3% -44.5% 25.7% -30.4%
Seattle-Everett,WA 2,345 1,440 1,425 2,693 -38.6% -1.0% 89.0% 14.8%
Washington,DC/MD/VA 1,800 1,150 1,621 2,638 -36.1% 41.0% 62.7% 46.6%
WestPalmBeach-BocaRaton-Delray Beach,FL 883 857 1,109 694 -2.9% 29.4% -37.4% -21.4%
38

Appendix Table A-2. Diagram of Employment change Variable Assignment

Did you move?
Yes
Within the SAME metro?
Use the
Construction/Overall
Employment Change from
the EXISTING metro
To a DIFFERENT metro?
Use the
Construction/Overall
Employment Change from
the PREVIOUS metro
No
Use the
Construction/Overall
Employment Change from
the EXISTING metro
39

References
Bauer, Thomas, Gil S. Epstein, and Ira N. Gang. "Enclaves, language, and the location choice
of migrants." Journal of Population Economics, 2005: 649-662.
Bohn, Sarah, Magnus Lofstrom, and Steven Raphael. "Did the 2007 Legal Arizona Workers
Act reduce the state's unauthorized immigrant population?" Review of Economics
and Statistics, 2014: 258-269.
Borjas, George J. The labor demand curve is downward sloping: reexamining the impact of
immigration on the labor market. National Bureau of Economic Research, 2003.
Cadena, Brian C., and Brian K. Kovak. Immigrants equilibrate local labor markets: Evidence
from the great recession. National Bureau of Economic Research, 2013.
Card, David. "How Immigration Affects US Cities." In Making Cities Work, by Robert P.
Inman, 158-200. Princeton, NJ: Princeton University Press, 2009.
Chapple, Karen, and T. William Lester. "The resilient regional labour market? The US case."
Cambridge journal of regions, economy and society, 2010: 85-104.
Charles, Kerwin Kofi, Erik Hurst, and Matthew J. Notowidigdo. Manufacturing decline,
housing booms, and non-employment. National Bureau of Economic Research, 2013.
Cohn, D’Vera, and Jeffrey S. Passel. US unauthorized immigration flows are down sharply
since mid-decade. Washington, DC: Pew Hispanic Center, 2010.
Cohn, D'Vera, and Jeffrey S. Passel. Mexican Immigrants: How Many Come? How Many Leave?
Washington, DC: Pew Hispanic Center, 2009.
DHS. 2013 Yearbook of immigration statistics . Washington DC: Department of Homeland
Security, 2015.
Fairlie, Robert W., and Lori G. Kletzer. "Jobs lost, jobs regained: An analysis of black/white
differences in job displacement in the 1980s." Industrial Relations: A Journal of
Economy and Society, 1998: 460-477.
Frey, William H. The Great American Migration Slowdown: Regional and Metropolitan
Dimensions. Metropolitan Policy Program, Brookings Institution, 2009.
Gabriel, Stuart A., Joe P. Mattey, and William L. Wascher. "Compensating differentials and
evolution in the quality-of-life among US states." Regional Science and Urban
Economics, 2003: 619-649.
Glewwe, Paul, and Gillette Hall. "Are some groups more vulnerable to macroeconomic
shocks than others? Hypothesis tests based on panel data from Peru." Journal of
Development Economics, 1998: 181-206.
Hamermesh, Daniel S. "What do we know about worker displacement in the US?" Industrial
Relations: A Journal of Economy and Society, 1989: 51-59.
40

Head, Allen, and Huw Lloyd-Ellis. "Housing liquidity, mobility and the labour market." The
Review of Economic Studies, 2012.
Hurd, Michael D., and Susann Rohwedder. Effects of the Financial Crisis and Great Recession
on American Households. Working Paper, National Bureau of Economic Research,
2010 .
Jewkes, Melanie, and Lucy Delgadillo. "Weaknesses of housing affordability indices used by
practitioners." Journal of Financial Counseling and Planning, 2010: 43-52.
Kletzer, Lori G. "Job displacement, 1979-86: how blacks fared relative to whites." Monthly
Labor Review , 1991: 17-25.
Kochhar, Rakesh, C. Soledad Espinoza, and Rebecca Hinze-Pifer. After the great recession:
foreign born gain jobs; native born lose jobs. Washington D.C.: Pew Hispanic Center,
2010.
Kochhar, Rakesh, C. Soledad Espinoza, and Rebecca Hinze-Pifer. After the great recession:
foreign born gain jobs; native born lose jobs. Washington D.C.: Pew Hispanic Center,
2010.
Kothari, Siddharth, Itay Saporta-Ecksten, and Edison Yu. "The (un) importance of
geographical mobility in the Great Recession." Review of Economic Dynamics, 2013:
553-563.
Liu, C., Zhuang, D, & Painter, G. "Immigrants and the spatial mismatch hypothesis:
Employment outcomes among immigrant youth in Los Angeles." Urban Studies,
2007: 2627-2649.
Liu, Cathy Yang. "Employment concentration and job quality for low-skilled Latino
immigrants." Journal of Urban Affairs, 2011: 117-142.
Liu, Cathy Yang, and Gary Painter. "Travel behavior among Latino immigrants: the role of
ethnic concentration and ethnic employment." Journal of Planning Education and
Research, 2011: 0739456X11422070.
Molloy, Raven, Christopher L. Smith, and Abigail K. Wozniak. Declining migration within the
US: the role of the labor market. National Bureau of Economic Research, 2014.
Munshi, Kaivan. "Networks in the modern economy: Mexican migrants in the US labor
market." The Quarterly Journal of Economics, 2003: 549-599.
Notowidigdo, Matthew J. The incidence of local labor demand shocks. National Bureau of
Economic Research, 2011.
Orrenius, Pia M., and Madeline Zavodny. "Immigrants in the US labor market." Undecided
Nation, 2014: 189-207.
Orrenius, Pia M., and Madeline Zavodny. "Mexican immigrant employment outcomes over
the business cycle." The American Economic Review, 2010: 316-320.
41

Orrenius, Pia M., and Madeline Zavodny. Tied to the business cycle: How immigrants fare in
good and bad economic times. Washington, DC: Migration Policy Institute, 2009.
Osberg, Lars, Daniel Gordon, and Zhengxi Lin. "Interregional migration and interindustry
labour mobility in Canada: A simultaneous approach." Canadian Journal of
Economics, 1994: 58-80.
Painter, Gary, and Zhou Yu. Caught in the Housing Bubble. Los Angeles: University of
Southern California, 2012.
Painter, Gary, and Zhou Yu. "Caught in the housing bubble: Immigrants’ housing outcomes
in traditional gateways and newly emerging destinations." Urban Studies, 2013:
0042098013494425.
Pendall, Rolf, Brett Theodos, and Kaitlin Franks. "Vulnerable people, precarious housing,
and regional resilience: an exploratory analysis." Housing Policy Debate, 2012: 271-
296.
Pissarides, Christopher A., and Jonathan Wadsworth. "Unemployment and the inter-
regional mobility of labour." The Economic Journal, 1989: 739-755.
Singer, Audrey. The rise of new immigrant gateways. Brookings Institution, 2004.
Singer, Audrey, and Jill H. Wilson. The impact of the great recession on metropolitan
immigration trends. Metropolitan Policy Program at Brookings, 2010.
Steven Ruggles, J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder,
and Matthew Sobek. "Integrated Public Use Microdata Series: Version 5.0 [Machine-
readable database]." Minneapolis: University of Minnesota, 2010.
Stoll, Michael A. Great Recession spurs a shift to local moves. US2010 Project, 2013.
Stone, Michael. "What is housing affordability? The case for the residual income approach."
Housing Policy Debate, 2006: 151-184.
Watson, Tara. Enforcement and immigrant location choice. National Bureau of Economic
Research, 2013.
Zhu, Pengyu, Cathy Yang Liu, and Gary Painter. "Does residence in an ethnic community
help immigrants in a recession?" Regional Science and Urban Economics, no. 47 (2014):
112-127.

42

An Alternative Method of Measuring Housing Affordability

Introduction
During the real estate boom of the early 2000’s a major concern for policy makers
was the affordability of housing. With the national median home price increasing from
$97,300 to $221,900 (128% total increase) from 1990 to 2006, according to the National
Association of Realtors, the concern was that people would not be able to afford to
participate in the “American Dream” of homeownership. After the housing bust, in 2007,
the issue of housing affordability dropped off the radar of many policy makers as a result of
housing prices falling dramatically to $166,200 in 2011 (25% decrease from 2006). While
the housing recovery has been slow, it has outpaced the economic recovery and has again
resulted in unaffordable housing in many areas with the national median price back up to
$197,400(2013).
Housing affordability is not merely a reflection of the changes in the actual prices of
homes; it is a result of the way affordability is measured. The measurement of affordability
is crucial in determining proper interventions and allocating limited funds. However, much
of the literature on affordability agrees that the existing measures are substandard (O’Dell,
Smith and White 2004) (Jewkes and Delgadillo 2010) (Orshansky 1965) (Kutty 2005)
(Stone 2006). These measures have not changed with the changing population, household,
and housing supply and therefore do not adequately capture the true dynamics in housing
affordability and households. Housing policies are made based upon the changes in the
measures of affordability and if the existing measures are unable to provide a true
description of what is happening, then the policies will not be as effective.
43

In order to better assess how affordability has changed over time, an alternative
affordability measure is proposed. This alternative measure considers more than just
changes in home prices and changes in household income. It also provides insight into how
the demographics of households are changing as well.
The results suggest that the existing measure of housing affordability lacks the
detail to properly assess the differences between geographical areas and the changing
make-up of a household. The most common measure fails to properly identify the increase
in housing to individuals as household housing costs change. By considering the number of
individual incomes in a household that are required to achieve 30% or less of income spent
on housing, a clearer picture of the actual change in housing costs overtime emerges.
The alternative measure created in this analysis shows that the housing cost has
become more of a burden since more incomes are required to pay for housing. In 1980 the
percent of households in California paying more than 30% of household income to housing
costs was about 27%, compared to 53% in 2010. If only the primary income is considered
the change is even more dramatic, from 40% to 71% over the same period.
The following section provides the background and a review of recent, relevant
literature relating to housing affordability and the measurement of affordability. The
subsequent sections will discuss the dataset, preliminary findings, an explanation of the
alternative method of measuring affordability, analysis of selected geographies using the
alternative measure, followed by the implications of the findings and finally the conclusion.
Background and Literature Review
Affordability
Over the long-term, the affordability of housing has not improved when compared
to income. Controlling for inflation, the nominal sale price for the nation nearly doubled
44

from 1971 to 2007 (Rappaport 2008). After the housing bubble burst, the unadjusted
median home price in the nation dropped from $217,900 (2007) to $166,200 (2011). This
reflects a home price similar to 2002, still well above the price in 1971 of $24,800. Some
may argue that it hasn’t gotten any worse due to the changes in the costs of other goods,
lower interest rates, and transfers of income that have changed over time. Home-
ownership rates also increased during the boom of the early 2000’s. However, there have
been numerous interventions and policies created to make housing more affordable, but it
hasn’t gotten any better, especially when different groups are considered (Schwartz and
Wilson 2006).
Housing affordability is a problem that has and continues to steadily become worse
even as the rate of home ownership increases. This is possible because a greater portion of
income is being devoted to housing. According to Quigley and Raphael (2004), the overall
home ownership rate in the US has increased from 61% in 1960 to 66% in 2000. The
overall home ownership rate increased to a high of 67% in 2006/07 and has since returned
to 66% as of 2009 (US Census). It is important to note that during this time of increased
homeownership overall, the home ownership rate for poor households decreased from
48% in 1960 to 37% in 2000 (Quigley and Raphael 2004).
Some of the poverty related research argues that a major oversight in the analysis of
affordability is the improving quality of housing and that people at all levels are benefiting
from the increasing quality of housing, even those that the lowest income levels (Stone
2006). While this may be true, since the quality of housing is a result of government
intervention, the cost of the higher quality home is a forced increase in the cost of housing
for people at all levels, especially those at the lower income levels who may have had lower
45

quality/cost options in the past (Rappaport 2008). Since this is not within the control of the
housing user, the justification for including adjustments based on “constant-quality” homes
is weakened.
Part of the seeming lack of improvement in the affordability of housing lies in the
way “affordability” is defined and measured as will be discussed in the following section.
Measuring Affordability
Housing affordability can be measured in numerous different ways using different
indices. While this section is not intended to be a comprehensive study of the various
methods, it will identify some of the more common methods. Some of the methods are
intended to measure the affordability of a geographic area such as “Housing Affordability
Mismatch” or the “National Association of Realtors Housing Affordability Index”. Other
measures are intended to be used at the household level to determine the ability to afford
housing for individuals. Jewkes and Delgadillo(2010) provide a thorough discussion and
summary of common measures of affordability, ultimately concluding that none of the
measures is completely perfect. The table below, Figure 1, replicates a portion of their work
in summarizing the methods (Jewkes and Delgadillo 2010).

46

Figure 1 - Summary of Common Affordability Measures

Additional current measures tend to focus on residual techniques that dedicate
income to other costs with the remainder attributable to housing costs. These are often
complex and difficult to become favored since most policy-makers tend to gravitate
towards easier “rules of thumb” or simple ratios. Historically, ratio methods for calculating
affordability have been used. As previously stated, this tendency goes back to poverty
measures and measures for the affordability of public housing (Feins and Lane 1982). The
use of housing cost to income ratios is identified as a primary method by many researchers
due to its simplicity and the ease with which ratios are understood by practitioners, policy-
makers, and society at large (McKenna and Anderson-Hills 1982) (Feins and Lane 1982)
(Jewkes and Delgadillo 2010).
Even with a multitude of measurement options available, housing affordability is
commonly measured as 30% percent of income based upon the median income for an area
and the median rent or housing costs of the area. This measurement is based upon the
standard set by HUD and is used by lenders and policy-makers, therefore it has become by
default the accepted measure (Schwartz and Wilson 2006). While this measure is
inadequate, due to the lack of consideration to earners in a household, it is widely accepted
Housing Affordability Index Brief Description
HUD Guideline Housing is affordable is 30% or less of gross
monthly household income is spent on
housing costs
National Association of Realtors Housing
Affordability Index
Ability of median-income family to buy
median-priced home
National Association of Home Builders/Wells
Fargo Housing Opportunity Index
Percentage of homes available to median-
income family
47

and entrenched due to its ease of use. Since this measure has become the default measure,
it is important to start with it as the base for any alternative comparison.
The existing measure is based upon the measure for poverty created in the early
20th century that was based upon the food needs of a family of three, using after-tax
income (Orshansky 1965). It has since been used with before-tax income and applied to
housing costs without adjustments to additional necessary costs in a modern society. In the
1930’s a report by the Public Works Administration used the rule of thumb that a family
should pay no more than 25% of income to rental housing costs (Feins and Lane 1982).
These “rules of thumb” have been carried over from poverty and rental housing to
ownership housing. According to O’Dell, et al. (2004), the first use of a ratio for housing
costs was a result of an amendment to the housing act of 1968 in which public rental
housing was based upon 25% of income. This was modified to 30% of income as an overall
standard for mortgages and rental housing in 1980 (O’Dell, Smith and White 2004).
This definition is accepted as the standard because it is the guideline provided by
the Department of Housing and Urban Development (HUD) (Glaeser and Gyourko 2008).
This is a problem at very low income levels, because the cost to eat and provide other
necessities has a threshold for all people of irrespective of income level. Therefore, if low
income households have to pay 30% of income to housing expenses, then the household
will have very little left to pay for other necessities.
There are problems with this method of evaluation in that it does not consider the
other costs of living, nor does it consider the differing distribution of incomes and costs
among different geographies. Major metro areas have a wide distribution of living/housing
costs based upon neighborhoods.
48

The poverty literature often attempts to measure poverty or housing induced
poverty using different measures or considering the changes in the housing supply
(Rappaport 2008) (Kutty 2005). However, most fail to consider the number of incomes in a
household. When considering poverty, the family unit is used instead of the number of
people in the household (Orshansky 1965). In housing affordability, the household is used,
regardless of the relationship of the people in it.
The most concerning problem with this existing definition is that it only considers
the level of a household’s income in relation to the housing costs. It does not take into
account the changes in the number of people in the household or the number of people
working in the household. As the economy has changed and employment opportunities
have fluctuated, households have changed, with more than one family living in the same
dwelling or with multiple incomes needed to afford housing costs. By adding people to a
household the cost of housing for the individual should be reduced, but may also lead to
potential overcrowding.
The definition of overcrowding is arguable; however, the important point is not
necessarily the measure, but the change over time using the same measure (Myers, Baer
and Choi, The Changing Problem of Overcrowded Housing 1996). While cultural differences
have an impact on the changing household make-up, there is increasing density in
households over time that is not accounted for by the change in the ethnic/racial make-up
of metro areas.
The addition of earners in a household is not accounted for in the measure of
affordability therefore a household with 4 people each earning $15,000 would have the
same affordability as a household of 1 person who earns $60,000, as long as the housing
49

costs remain the same for both households. This is true with residual techniques that do
not consider the additional costs for additional household members and the cost of
childcare in households that require care givers to work outside the home in order to
afford housing and other living costs.
Literature pertaining to women participating in the workforce at increasing rates in
past decades considers the impacts of additional workers and benefit/need of the
additional income (Myers, Reliance Upon Wives' Earnings for Homeownership Attainment:
Caught Between the Locomotive and the Caboose 1985). However, the research has mainly
been focused on the additional income from a wife and how that additional income
equalizes households (Treas 1987).
The past research that focuses on the additional income of a wife, assumes that it is
secondary to the primary income of the husband, rather than simply considering additional
incomes in a household without order to position in household or relationship. In recent
decades the primary earner in a household is not necessarily assumed to be male and may
not be identified as the “head of household” in the Census. The structure of a household has
also changed with more households now having unrelated people living together (Tienda
and Glass 1985).
Data
The dataset used for this analysis is drawn from the 2006 - 2011 American
Community Survey (ACS) microdata set and the 1980, 1990, 2000, 2010, Census data. The
data were obtained from Integrated Public Use Microdata Series (IPUMS) provided by the
Minnesota Population Center (Steven Ruggles 2010) and include data at the national, state,
metro summary level, county and at the PUMA level. The data included household and
individual level characteristics that provide detailed information on demographic,
50

economic, and mobility attributes. Additional data from the National Association of
Realtors and the Freddie Mac Mortgage Interest Rate Survey have also been used in the
analysis.
The specific attributes this study is concerned with include: Household income, Age
of the head of household, Cost of housing, Number of families in household, Number of
working age people in household, Number of employed people in household, Education,
Marital status, and Ethnicity/Race. Additionally, metropolitan level characteristics are
included in the analysis, such as: Median cost of housing, Median sale price of home,
Average 30-year conventional mortgage interest rate, and Median household income.
Group home and incarcerated people were excluded from the dataset as income and cost of
living characteristics vary from the general population. The data used in the final
calculation of the alternative measure exclude the people noted in the previous sentence in
addition to households that are owner-occupied and do not have mortgages.
The geographic level of the PUMS data is for statistical areas with at least 100,000
residents called a Public Use Microdata Area (PUMA) (US Census n.d.). While the data is
available for this smaller geographic level, this study is focused on the states in order to
identify differences across the country. Some discussion is provided at the city and county
level to show differences as a result of geographic level, however, the analysis and results
are performed at the state level. While policy interventions are generally concerned with
smaller geographies, such as the metropolitan area, for the purpose of analyzing the
existing measure of affordability and considering the alternative proposed measure, the
state level has been used. Conversion to a metro level measure would not change the
underlying benefit of the alternative measure.
51

Summary Statistics
As stated, arguments have been made that housing has not become less affordable
over time due to factors such as income, interest rates, and costs of living. Depending on
the geography level or area analyzed, this may be true. However, nationally and in many
states housing and related costs have increased over time.
The following maps, Figure 2, show the median owner cost as a percent of income
for each state. The maps have little variation between states in 1990 and little change over
time and between states in 2011. The percent of income dedicated to housing costs has
increased in nearly every state, with the exception of North Dakota and Nebraska. This is
an example of how little use a measure such as the median owner cost as a percent of
income is. The real change is muted due to the measure and aggregation of the individual
incomes into the household.
Figure 2 - Percent of Income Used for Housing Costs 1990 & 2011

The following sections discuss characteristics of households and the real estate
market over time in an effort to portray the vast differences based upon the geography and
the geographic level considered. These show that while there has been little change in the
median cost as a percent of income over time, there have been a lot of changes in
households that are not adequately captured by the existing measure.
52

Real Estate Markets and Median Income
According to the National Association of Realtors and the California Association of
Realtors, the median home prices in the nation and in California have increased/decreased
at different rates since 1990. The differential between CA and the nation median home
price has always been large, but has increased over time, Figure 3. The California median
home prices were 99% higher than the nation in 1990. The smallest difference was in
1996/97 when CA was only 45% higher than the country as a whole and the greatest
difference was in 2007 when CA was 157% higher. After the housing bust, the difference
shrank back down to 80% higher in 2012.
Over the same time, the median household incomes have steadily increased over
time and have remained similar between the US and CA with the national median
household income at $29,943 in 1990 and CA at $33,290 (111% of US). The median
household income of CA has remained roughly 5% to 14% higher than the US household
income, ending at 12% higher in 2012 (US Census, ACS).
Figure 3 - Changing Home Price and Household Income 1990-2012

53

As shown in Figure 3, the median home price in California has changed dramatically
over time with less dramatic changes in the nation. However, the incomes for both
geographies have remained fairly steady over time.
While it is clear that incomes have not kept pace with home prices, the argument
has been made that the effects of changing interest rates reduce this difference, making
housing more affordable than in the past. Figure 4 shows the impact of changes in the
interest rate over time. It should be noted that interest rates at the California state level
and the US level are highly correlated. In contrast the monthly payment, assuming 20%
down payment, is very different between California and the US.
It should be noted that the number of home purchases using 20% decreased over
time and decreased rapidly during the boom period of 2000-2007. This was partially a
result of the increase in home prices and the lack of the ability to make the traditional 20%
down payment. This led to less stringent lending standards and lower down payment
requirements. Many of the purchases in the past decade have been completed with first and
second loans, zero percent down, 3.5% down, and other down payments that were less
than 20%.
In addition to lower down payment loans, other loan types, such as Interest-only
and 40 year mortgages became more popular during the boom period (LaCour‐Little and
Yang 2010). The Interest –only payment for California is shown in the graph below in
addition to the traditional fully amortized loan payment. The chart presented is a
conservative estimate of the monthly payments and the differential between geographies,
because it assumes 20% down. Student loan debt has also increased in the past decade and
54

as a result reduces the amount of income that can be used to repay a home loan making
housing even more unaffordable.
Figure 4 - Changing Interest Rates and The Effect on Monthly Payments 1990-2012

According the National Association of Realtors housing affordability index,
affordability improved in late 2006 following the decline of housing market. However,
affordability in the US has declined since early 2013. The NAR affordability index considers
interest rates, median family income, and median home price. One flaw in the index is that
it does not consider the ability to get a loan and the tightening of loan requirements
following the burst of the housing bubble. It also assumes a 20% down payment.
The percent of income spent on housing costs, as defined by the US Census, has not
improved at the lower income levels. Households making less than $50,000 per year should
be of greater concern than those in the upper income brackets, since the cost of food and
other living expenses is similar regardless of income. Using unadjusted income data from
1990, 2000, and 2010, Figure 5 shows the percent of households spending 35% or more of
55

income on owner-occupied housing. In the United States, the number of households with
incomes of less than $50,000 dropped by just over 700,000 to 29 million. However, the
number of households spending over 35% on housing costs increased from 5.5 million to
12.5 million. Nearly a quarter of households with a mortgage in Los Angeles County used
50% or more of income on housing costs in 2011 with over half of the households above
30% of income for housing costs.
Figure 5 - Percent of Households Spending More than 35% of Income on Housing Costs - Households with $50,000 or less
Income

As evidenced in Figure 5, looking beyond median income and median housing costs
is critical in order to identify the true affordability to the population that is the most
vulnerable to changes in housing costs. But looking beyond the traditional variable of
household income to alternative identifiers of housing affordability changes, it becomes
clear that over time household make-up has changed. When variables such as household
size and number of unrelated people living together are considered, it becomes clearer that
the make-up of a housing unit is different and the number of incomes required to afford a
56

home has changed. There are differences in household composition for different
ethnic/race groups.
Household Characteristics by Group
The characteristics of households become more important when income level is
considered. Since the cost of housing is more of a burden on lower income households, it is
important to consider the changes in the household make-up that have resulted from
stagnant income and increasing costs of living.
One possible measure of changing household composition is the household size.
However, arguments have been made that household size is not a good indicator of
affordability because people of different ethnic/racial backgrounds are more likely to have
extended family living with them. The difference in preference to have extended family,
does not explain differences in the number of occupants per room. It would be expected
that larger families would occupy homes with a greater number of rooms, if such a home
was available in the price range that is affordable. Figure 6 shows the percent of
households for different size homes as compared to the total number of households for
each group for the Nation. The figure shows total rooms, not just bedrooms. For example, a
studio apartment with only one large room and a bathroom would be considered 1 room,
even though it does not have a bedroom.
57

Figure 6 - Percent of Households by Room Number

The above figure shows that 40% of Latino households have 4 or fewer rooms, as
compared to about 26% for non-Latino households, in 2010. This is a decrease in from
1980 which shows that the percentages are 52% for Latino and 32% for non-Latino
households. The decrease in the cumulative percentage for 7 or fewer rooms is less
dramatic for both groups, with a 10% reduction for non-Latino households and only a 5%
reduction for Latino households. It should also be noted that 7 or fewer room households
account for 90% of all Latino households in 2010, but only 78% of non-Latino households.
Equally important to the percent of households in each size home is the number of
people on average in each of those homes. The differences between Latino and non-Latino
households is not just in the fact that Latino households tend to occupy homes with fewer
rooms, but also how many people are in each of those rooms. Figure 7 shows that Latino
Not Latino 1980 1990 2000 2010
1 1% 1% 1% 1%
2 3% 3% 3% 2%
3 10% 9% 9% 8%
4 17% 17% 16% 14%
5 22% 22% 22% 20%
6 21% 20% 20% 19%
7 12% 13% 13% 14%
Cumulative 88% 85% 83% 78%
Latino 1980 1990 2000 2010
1 4% 5% 3% 3%
2 8% 10% 7% 3%
3 17% 17% 16% 12%
4 24% 22% 24% 22%
5 22% 20% 22% 24%
6 14% 13% 15% 17%
7 7% 6% 7% 9%
Cumulative 95% 94% 94% 90%
Percent of Households
58

households have a higher average number of people per room. The average Latino
household has 0.9 people per room in 1980 and 0.7 people per room in 2010. This is higher
than for non-Latino households which have an average of 0.5 in 1980 and 0.4 in 2010.
Figure 7 - Average Number of People per Room

Initially, the impression from the figure appears to be positive since there is less
than one person per room in Latino households with 4 or more rooms, but it is important
to remember that these are not bedrooms, simply rooms. The example of a studio
apartment with one large room and a bathroom would have an average of 2.3 people in it in
2010 for Latino households as compared to 1.3 for non-Latinos. The survey question says
“Count a kitchen as a room but do not count bathrooms. Also exclude kitchenettes, strip or
pullman kitchens, utility rooms, and unfinished attics, basements, and other space used for
storage.” (Steven Ruggles 2010).
Not Latino 1980 1990 2000 2010
1 1.2 1.4 1.3 1.3
2 0.7 0.8 0.8 0.7
3 0.5 0.6 0.6 0.5
4 0.5 0.5 0.5 0.5
5 0.5 0.5 0.5 0.5
6 0.5 0.5 0.4 0.4
7 0.5 0.4 0.4 0.4
All rooms 0.5 0.5 0.5 0.4
Latino 1980 1990 2000 2010
1 2.1 2.7 2.0 2.3
2 1.4 1.7 1.3 1.0
3 0.9 1.1 1.0 0.8
4 0.8 0.9 0.8 0.8
5 0.8 0.8 0.8 0.7
6 0.7 0.7 0.7 0.6
7 0.6 0.6 0.6 0.6
All rooms 0.9 1.0 0.8 0.7
Average Number of People per Room
59

Summary
The existing measure of affordability is based solely on the cost of housing and the
income of the household. It does not adequately represent the increased burden on
individuals, especially those at the lower income levels. The changing make-up of a
household, with different numbers of unrelated people, changes in multi-generational
households, and the addition of earners in a household, are all missed by the existing
measure. These factors show that there are changes happening in households that are not
reflected in the commonly accepted method of measuring housing affordability. Therefore
there must be a better way that helps to illuminate the real change in housing affordability.
Since the real concern of housing affordability is how much income must be
dedicated to housings costs, it would make sense that an alternative measure of
affordability would consider the number of earners that are required to afford the cost of
housing. This is especially important if the cost of childcare is considered. If more earners
are required in order to afford a home, then labor historically utilized in the household for
childcare and other housing labor is not available and has to be contracted out to others, at
an expense.
An Alternative Measure
The above evidence shows that there was a lot of variation in the characteristics of
households over time and between geographic areas. However, the difference in the
percent of household income dedicated to housing costs did not change to a large degree. In
order to better identify how much the cost of housing has affected the income of
individuals, an alternative measure has been constructed. The following sections will show
how the alternative measure of affordability can better reflect the change in income
60

required to afford housing. The series of graphs in the following sections use Census data
for California, obtained from IPUMS (Steven Ruggles 2010).
The process of creating the alternative measure is as follows and will be described
in depth in the following section.
1. Evaluate the existing conventional method of measuring affordability.
2. Invert the existing measure to reflect what is not affordable, rather than what is
affordable.
3. Remove owner-occupied households that do not have mortgages.
4. Disaggregate owners (with mortgages) and renters.
5. Disaggregate households based upon the required number of incomes needed to
maintain housing costs at 30% or less of household income.
As previously stated, the traditional measure of affordability considers the percent
of households that pay 30% or less of income to housing. This measure includes renters,
owners with and without mortgages. Years and geographies with higher percentages of
people meeting this criterion would mean that they are more affordable and therefore
preferred. This is reflected in Figure 8 and shows that the percent of households in
California paying 30% or less of income to housing costs was 69% in 1980, but dropped to
52% in 2010. This indicates that households, renters and owners combined, had a drop in
affordability, with fewer households paying 30% or less to housing costs.

61

Figure 8 - Traditional Measure - Percent of Households Paying 30% or less of Income to Housing Costs

The first step in transitioning to the alternative measure is to invert the traditional
measure. When the traditional measure is translated from “the percent of households that
pay 30% or less of income to housing” to “what percent of people pay MORE than 30% of
HH income for housing costs” the graph shows a positive sloping line. This is achieved by
subtracting the traditional measure from 100%. In 1980 the traditional measure shows
that 69% of households used 30% or less of household income on housing costs. The
conversion translates to 31% of households dedicating more than 30% of household
income towards housing costs as show in Figure 9.

62

Figure 9 - Inverted Traditional - Percent of Households Paying MORE than 30% of Income to Housing Costs

The inverted graph shown in Figure 9 allows for a more intuitive look at how the
cost burden to households is changing. It shows that the index is increasing at the same
time that the housing cost for households is increasing. The traditional index decreases as
costs increase. Therefore by inverting the graph, a user can look at it and see that over time
the number and percent of households, that cannot afford housing with less than 30% of
the household’s income, is increasing as the costs of housing increase as well.
The traditional index is often cited as one number for an area, therefore it is
necessary to disaggregate owners without mortgages from the rest of the households. Since
most housing policy is concerned with people acquiring and maintaining homes, it is most
important to look at those who have high ongoing costs for housing and are at risk of losing
housing if they can no longer afford it. Owner households with no mortgage would be
expected to have lower housing costs, and more options, so these households should be
removed from a measure that is concerned with the affordability of housing.
Over time the number of owner households has changed, but the percent of owner
households with mortgages has also changed, as shown in Figure 10. The traditional
63

measure includes these households and is therefore skewed as the percent of owners
without mortgages fluctuates over time.

Figure 10 - Percent of Home Owners without a Mortgage

By removing the households without mortgages, the index becomes a more accurate
representation of the affordability of most people that policies are concerned with. The
result of this removal of owners without mortgages is shown in Figure 11. This graph
shows the line for the inverted traditional measure, as previously shown in Figure 9,
compared with the line that excludes owners without mortgages. While the two lines seem
to follow a similar trajectory, they do not track exactly since the percent of owners without
mortgages fluctuates over time. The difference between the measures fluctuates from a low
of 6% to a high of 8% over the years.
64

Figure 11 - Comparison of Inverted Traditional Measure with and without Owners without Mortgages

The issue of affordability is important to both renters and owners alike, but policy
interventions differ dramatically depending on which group the policy is intended to
address. In order to make the alternative measure more useful for creating and analyzing
policy, it is important to consider owners and renters separately, so that policies can be
crafted for the changing needs of each group.
Figure 12 - Renters and Owners Paying more than 30% of Household Income to Housing Costs- Excludes Owners without
Mortgages

Figure 12 shows separate lines for renters and owners (with mortgages). In 1980
the affordability gap between owners with mortgages and renters was large, at about a 17
65

percentage point difference. Being either a renter or an owner with a mortgage has become
less affordable. However, the pace is not the same for both groups. As Figure 12 shows,
renting has become less affordable by about 15 points and for owners with a mortgage it
has increased by 26 points. This has the effect of shrinking the margin to only 6 points as
compared to 17 in 1980. This figure also shows that the percent of both groups paying
more than 30% of income towards housing has increased dramatically. In 1980, 44% of
renters paid more than 30% and in 2010 that increased to nearly 60%. The percent of
owners with mortgages paying more than 30% nearly doubled from 27% (1980) to
53%(2010).
Up to this point, the sections have identified basic “housekeeping” items that help to
improve the understanding and use of the traditional measure. For a more complete
understanding of the changes over time it is necessary to account for differences in the
number of earners in a household and the amount of income the earners are making in
relation to housing costs.
This requires a breakdown in the measure created thus far. The following two
graphs show the percent of households that pay more than 30% of income to housing costs
based on a single income, the percent that pay more when a second income is considered,
and the overall affordability when all household incomes are considered. For reference, the
“All Incomes” category reflects the same line as the line presented in Figure 11. While
earners beyond two makes only a small difference, the difference is getting larger over
time.
The data used in the creation of Figure 13 is not simply based on how many earners
are in a household and it also differs from the historic research relating to women
66

participating in the workforce. The earners in each household are sorted by the highest
income to the lowest income, rather than by gender or by who is identified as the head of
household.
The purpose of this is to identify at which earner a household meets the 30% of
income or less spent on housing threshold. All additional incomes in the household are not
deemed to be necessary to cover the cost of housing.
Figure 13 - Percent of Households Paying more than 30% to Housing Costs by Income, by Tenure - Excludes Owners without
Mortgage

Figure 13 provides a graph for owners with a mortgage and a second graph for
renters. The understanding of both graphs is similar; therefore, the following paragraphs
will discuss the graph relating to owners with a mortgage.
Households that paid 30% or more on housing in 1980, with a single income, was
40% percent. This increased to just over 70% in 2010. When two incomes are considered,
the percent of households that paid more than 30% in 1980 was 28%. This means that
12% of households benefited from the second income in keeping housing costs under 30%
of household income. The dashed line in the graphs shows the percent of households
67

paying more than 30% of a single income in 1980. The dashed line extends through to 2010
to show how dual incomes are no longer enough to maintain what a single income could
have afforded in 1980.
An interesting finding is that the percent of households that paid more than 30% of
income with two incomes in 2000 is now approximately equal to the percent of single
incomes in 1980. A similar result is seen with single income in 1990 is compared to dual
income in 2010. This indicates that households may have been able to maintain the same
level of household affordability, but it required the income from an additional earner that
was not required in earlier decades.
The addition of a third income provides slightly better results in achieving the 30%
or less result, but the benefit is very small. The key is to note that the benefit of the third
income has gained in importance over the decades and may become even more important
in future years. For example, in 1980 the benefit of the incomes beyond two was less than
1%, by 2010 it is nearly 2%.
Comparing the outcomes of renters and owners with mortgages shows that both
groups have increased the number of households that can’t afford housing at the 30% or
less level, regardless of the number of incomes considered. However, the situation for
renters has gotten worse at a slower pace than that of the owner group. Single income
affordability is now similar for both groups; however the benefit of a second income is
greater for the owner group.
Ultimately there are still many owner households that pay more than 30% to
housing. In 1980 approximately 27% of households paid more than 30% of income to
housing when all incomes are considered. That grew to over 50% in 2010. This shows that
68

even with all incomes considered in a household, there are still approximately 53% of
owner households that must pay more than 30% of income towards housing.
Figure 14 - Changes in Percent of Households Paying more that 30% of Income to Housing Costs 1980 & 2010

Comparing the outcomes of the alternative measure to the traditional inverted
measure shows that the alternative measure better captures how things are changing. The
inverted traditional measure shows that the percent of households paying more than 30%
grew from 31% to 48% from 1980 to 2010. Figure 14 shows that the changes become more
dramatic when consideration is given to the number of incomes and whether the
household is a renter-occupied household or an owner-occupied household. The traditional
measure understates the changes due to the inclusion of household without mortgages and
the aggregation of all households together regardless of individual incomes.
If a comparison is made between Latino households and non-Latino households, the
power of the alternative measure becomes even more evident. Figure 15 further shows
that the traditional measure hides what is truly happening. There is a 14 percentage point
change in the traditional measure for non-Latino owner households between 1980 and
2010. The same change for Latino households is 23%. When the alternative measure is
considered, the change for households that can’t afford housing costs with only a single
Measure 1980 2010 Change
Traditional Inverted 31 48 17
Owner (Single Income) 40 71 31
Owner (Dual Income) 28 55 27
Renter (Single Income) 55 72 16
Renter (Dual Income) 45 61 16
Percent of Households Paying More than 30% to
Housing Costs
69

income increases by 28% for non-Latino households, however, the change for Latino
households is 35%.
Figure 15 - Percent of Households Paying More than 30% to Housing Costs - by Ethnicity

When comparing Latino and non-Latino households, the interpretation using the
traditional inverted measure would be that both groups were in similar positions in 1980,
31% and 33%, and both groups lost affordability through 2010 with non-Latinos at 45%
and Latinos at 56%. However, when the alternative measure is used and the impact of
additional earners is evident. In 1980 the benefit of incomes beyond two for is 1percentage
point and 2 percentage points for non-Latinos and Latinos, respectively. In 2010, the
impact of the additional incomes for non-Latinos is still only at 1%, however, the impact for
Latinos is 4 percentage points. This indicates that Latino households rely on incomes
beyond two to make housing more affordable. The differences between the percent of
households that cannot afford housing at 30% with a single income versus when all
incomes are considered is even greater, with a 16 point difference for non-Latino
households in 2010 and a 21 point difference for Latino households. These differences are
all muted by the traditional measure of housing affordability.
Non-Latino 1980 2010 Change
Traditional Inverted 31% 45% 14%
Owner (Single Income) 39% 67% 28%
Owner (Dual Income) 28% 52% 25%
Owner (All Income) 27% 51% 24%
Latino 1980 2010 Change
Traditional Inverted 33% 56% 23%
Owner (Single Income) 46% 81% 35%
Owner (Dual Income) 32% 64% 33%
Owner (All Income) 30% 60% 30%
70

Application of the Alternative Measure
As shown in the previous section, the alternative measure provides more detail than
the traditional measure. Analyzing the states and the nation using the alternative measure
provides greater insight into the changing number of earners needed in households. The
inverted traditional measure shows that fewer households are able pay less than 30% of
income towards housing, but it tends to mute the true nature of the change. Graphs for
owner-occupied households using the alternative measure and the inverted, combined,
traditional measure are shown for each state in the addenda.
The graphs show that states vary dramatically in affordability at any level, but vary
even more when incomes are considered. States, such as California, Hawaii, and New York,
show that the inverted traditional measure and the single-income measure had smaller
differences in 1980, then in 2010. Part of this is the result of owners without mortgages
and renters being included in the traditional measure. Without consideration for the
different number of incomes required to afford the housing costs, the policies created will
likely not be as effective.
In 1980 the states with the highest percent of households that were unaffordable
using the single highest income were Nevada (47%) and New Hampshire (46%). In
contrast, California (71%) and Hawaii (73%) had the highest percent of unaffordable
households in 2010.
The impact of adding earners to a household is different for each of these states, as
shown in Figure 16. New Hampshire has the second highest percent of unaffordable
households, based on one income, in 1980, and ranks 6th when two incomes are
considered. In contrast to New Hampshire, California and Hawaii both maintain their
71

positions, as having the highest percent of unaffordable households in 2010, even when the
second income is considered.
Figure 16 - Impact of Additional Earners on Affordability

Conclusion
The existing measures of affordability influence policy creation, but may not be
influencing the correct types of policies. The traditional measure is simple and favored by
government agencies due to its simplicity, therefore is has become the default measure.
The failure in this measure is that it doesn’t adequately capture the changing number of
earners required to maintain adequate income.
The proposed alternative measure helps to identify the changing income/earner
needs to maintain household spending levels for housing. There are advantages to using
the alternative measure over the traditional one.
• The alternative measure considers the different number of earners
required in a household rather than just assuming that all households have the
Single Income 1980 1990 2000 2010
California 0.40 0.57 0.61 0.71
Hawaii 0.43 0.56 0.70 0.73
Nevada 0.47 0.49 0.60 0.66
New Hampshire 0.46 0.59 0.53 0.66
Dual Income 1980 1990 2000 2010
California 0.28 0.40 0.42 0.55
Hawaii 0.26 0.35 0.48 0.58
Nevada 0.32 0.30 0.39 0.50
New Hampshire 0.27 0.35 0.32 0.43
Improvement from 1st with addition of 2nd Earner
1980 1990 2000 2010
California 0.12 0.17 0.19 0.16
Hawaii 0.17 0.21 0.22 0.15
Nevada 0.15 0.19 0.21 0.16
New Hampshire 0.19 0.24 0.21 0.23
72

same number of earners. This helps to disaggregate the household income so
that affordability can be more accurately assessed. This will become even more
useful in the future as incomes beyond two become more important to maintain
affordability.
• Interpreting the alternative measure is intuitive since increases in the
measure reflect a greater number of households that can’t afford housing costs
based upon the standard of 30% or less. The measure shows an increasing index
as the cost burden for households increase. When there is a decrease in the
index, there are fewer households that have housing cost burdens in excess of
the 30% level.
• The major advantage of this alternative over other methods, including
residual methods, is that it is still simple to understand and use. The simplicity is
part of the reason the existing measure has become so entrenched and it is also a
reason why this alternative should be easier to adopt than more complicated
residual techniques.
The alternative measure will be critical in the analysis of future household
formation since it identifies the number of earners required in households in order to make
housing more affordable. If incomes continue to stagnate while housing and other costs
continue to increase, the need for earners, beyond two, will become necessary to maintain
affordability. This could result in fewer people moving out to form new households or
households merging in order to maintain affordability. The traditional measure of
affordability understates the true nature of the change.

73

References
Feins, Judith D., and Terry Saunders Lane. How much for housing?: new perspectives on affordability
and risk. Cambridge: Abt Books, 1982.
Glaeser, Edward L., and Joseph Gyourko. Rethinking Federal Housing Policy. Washington, D.C.:
American Enterprise Institute for Public Policy Research, 2008.
Gyourko, Joseph. "Urban Housing Markets." In Making Cities Work, by Robert Inman, 123-157.
Princeton, NJ: Princeton University Press, 2009.
Jewkes, Melanie, and Lucy Delgadillo. "Weaknesses of housing affordability indices used by
practitioners." Journal of Financial Counseling and Planning, 2010: 43-52.
Kutty, Nandinee K. "A new measure of housing affordability: Estimates and analytical results."
Housing policy debate, 2005: 113-142.
LaCour‐Little, Michael, and Jing Yang. "Pay me now or pay me later: alternative mortgage products
and the mortgage crisis." Real Estate Economics, 2010: 687-732.
McKenna, William F., and Carla Anderson-Hills. The report of the President's Commission on Housing.
Vol. 81, President's Commission on Housing, 1982.
Myers, Dowell. "Reliance Upon Wives' Earnings for Homeownership Attainment: Caught Between
the Locomotive and the Caboose." Journal of Planning Education and Research, 1985: 167-
176.
Myers, Dowell, William C Baer, and Seong-Youn Choi. "The Changing Problem of Overcrowded
Housing ." Journal of the American Planning Association, 1996: 66-84.
O’Dell, William, Marc T. Smith, and Douglas White. "Weaknesses in current measures of housing
needs." Housing and Society, 2004: 29-40.
Orshansky, M. "Counting the poor: Another look at the poverty profile." Social Security Bulletin,
1965: 3-29.
Quigley, John M., and Steven Raphael. "Is Housing Unaffordable? Why Isn't It More Affordable?" The
Journal of Economic Perspectives (American Economic Association) 18, no. 1 (2004): 191-
214.
Rappaport, J. The Affordability of Homeownership to middle-income Americans. Economic Review,
2008.
Schwartz, Mary, and Ellen Wilson. Who Can Afford To Live in a Home?: A look at data from the 2006
American Community Survey. US Census Bureau, 2006.
74

Steven Ruggles, J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder, and
Matthew Sobek. "Integrated Public Use Microdata Series: Version 5.0 [Machine-readable
database]." Minneapolis: University of Minnesota, 2010.
Stone, Michael. "What is housing affordability? The case for the residual income approach." Housing
Policy Debate, 2006: 151-184.
Tienda, Marta, and Jennifer Glass. "Household structure and labor force participation of black,
Hispanic, and white mothers." Demography, 1985: 381-394.
Treas, Judith. "The effect of women's labor force participation on the distribution of income in the
United States." Annual Review of Sociology, 1987: 259-288.

75

Addenda
Graphs for the nation and each state showing the alternative measure form 1980 to
2010.

.8 .7 .6 .5 .4 .3 .2 .1
0
Alabama Alaska Arizona
Year
Graphs by state
1980 1990 2000 2010
Year
Single Income Dual Income
All Incomes Inverted Traditional Measure
76

.8 .7 .6 .5 .4 .3 .2 .1
0
Arkansas California Colorado
Year
Graphs by state
.8 .7 .6 .5 .4 .3 .2 .1
0
Connecticut Delaware District of Columbia
Year
Graphs by state
77

.8 .7 .6 .5 .4 .3 .2 .1
0
Florida Georgia Hawaii
Year
Graphs by state
.8 .7 .6 .5 .4 .3 .2 .1
0
Idaho Illinois Indiana
Year
Graphs by state
78

.8 .7 .6 .5 .4 .3 .2 .1
0
Iowa Kansas Kentucky
Year
Graphs by state
.8 .7 .6 .5 .4 .3 .2 .1
0
Louisiana Maine Maryland
Year
Graphs by state
79

.8 .7 .6 .5 .4 .3 .2 .1
0
Massachusetts Michigan Minnesota
Year
Graphs by state
.8 .7 .6 .5 .4 .3 .2 .1
0
Mississippi Missouri Montana
Year
Graphs by state
80

.8 .7 .6 .5 .4 .3 .2 .1
0
Nebraska Nevada New Hampshire
Year
Graphs by state
.8 .7 .6 .5 .4 .3 .2 .1
0
New Jersey New Mexico New York
Year
Graphs by state
81

.8 .7 .6 .5 .4 .3 .2 .1
0
North Carolina North Dakota Ohio
Year
Graphs by state
.8 .7 .6 .5 .4 .3 .2 .1
0
Oklahoma Oregon Pennsylvania
Year
Graphs by state
82

.8 .7 .6 .5 .4 .3 .2 .1
0
Rhode Island South Carolina South Dakota
Year
Graphs by state
.8 .7 .6 .5 .4 .3 .2 .1
0
Tennessee Texas Utah
Year
Graphs by state
83

.8 .7 .6 .5 .4 .3 .2 .1
0
Vermont Virginia Washington
Year
Graphs by state
.8 .7 .6 .5 .4 .3 .2 .1
0
West Virginia Wisconsin Wyoming
Year
Graphs by state
84

The Response of Latino Immigrants to Housing Affordability
During and Following the Great Recession

Introduction
The Great Recession impacted most participants in the US economy, but affected
Latino immigrants in particular and in slightly different ways than the rest of the
population. This was, in part, a result of the changes in employment in occupations that
Latino immigrants are concentrated (Painter and Calnan). The outcomes for Latino
immigrants are not just a result of moving to job opportunities, the outcomes are also a
result of the cost of housing.
Housing expenses are a large portion of the overall household expense, especially in
the established immigrant gateways and increasingly in the emerging immigrant gateways.
This paper will consider the impact of housing affordability on the prevalence of moving.
Using an alternative measure of housing affordability, this research shows that affordability
appears to have played a part in the decision to move. This may indicate that movers were
willing to forgo additional wages in one location in order to obtain less expensive housing
in another, similarly to the research relating to lower wages and lower unemployment as a
result of migration decisions that are influenced by housing tenure (Head and Lloyd-Ellis)
(Coulson and Fisher).
Housing affordability during and following the recession was often given little
thought since the driver of this recession was the bust in housing prices. However, the
change in employment rates made housing affordability just as important following the
85

bust as it was during the boom period. The loss in jobs and reduction in hours worked
increased the burden on individuals even at a time when housing prices fell.
In order to ascertain the degree to which the affordability of housing played a part
in the moving process, I have analyzed the affordability of select metropolitan areas using
an alternative measure of affordability as described in the previous chapter. Using this
measure helps to provide a better understanding of the true impact of changes in housing
costs and incomes. Data relating to housing, employment, and mobility of Latino
immigrants and all of the population have been obtained from the American Community
Survey for the years of 2006-2011. The last step in the analysis is a regression model that
helps to better communicate the effects of all the variables considered on the propensity to
move.
Much of the research has considered mobility to job opportunities as a reason for
leaving as well as housing ownership as a reason for staying. This research differs in that it
also considers the cost of housing to the individual based upon the alternative measure
created of affordability rather than the historic measure. This study helps to provide
greater understanding of the influences on migration decisions that have not been
thoroughly considered by the existing literature. It also helps to focus the attention on the
specific impacts on Latino immigrants.
The regression models confirm that the effects of increasing housing costs and loss
of affordability tend to decrease the likelihood of people moving to other metro areas.
Further research, likely in the form of interviews, would need to be conducted in order to
better ascertain whether higher paying employment opportunities were forgone in order
to obtain lower costs of housing.
86

The following section provides the background and a review of recent, relevant
literature relating to the impact of housing affordability on migration decisions. The
subsequent sections will discuss the dataset, summary statistics and preliminary findings.
This is followed by the regression analysis, the implications of the findings, and finally the
conclusion.
Background and Literature review
The housing outcomes of the Latino population in general and Latino immigrant
population in particular is of great interest to policy-makers, because these groups
represent a large segment of the US population. There has been a lot of research on the
changes in housing tenure, employment, and mobility following the Great Recession
(Kochhar, Espinoza and Hinze-Pifer) (Chapple and Lester) (Kothari, Saporta-Eksten and
Yu) (Singer and Wilson) (Stoll) (Painter and Yu). Disruption of the normal migration
patterns between metro areas has been documented for the population as a whole (Rhee
and Karahan); however, there has been less focus on the specific outcomes of Latino
immigrants in relation to the affordability of housing and migration.
Research that has considered the impacts on Latinos has found that they have been
disproportionately impacted, along with Blacks, in the loss of home ownership and equity
(Ellen and Dastrup) (Singer and Wilson). The result of greater home loss by these groups
can lead to rental compression is some areas as former owners now become renters and
the pool of vacant housing increases (Ellen and Dastrup). The result has been a slight
uptick in the number of family households that include extended family and households
that have additional, unrelated people living together (Elliott, Young and Dye).
Historically, one of the arguments for housing selection was the concept of
“filtering” of older homes/areas to lower income levels through the market life until
87

revitalization occurs, starting the cycle over. However, during the Great Recession there
was stagnation in incomes and greater unemployment (Hurd and Rohwedder). As a result,
the normal migration of people to employment, and the unemployment rates in particular
areas, did not follow the historic patterns (Rhee and Karahan). This diversion in the normal
life-cycle of households has disrupted the “filtering” process and has led to changes in the
household make-up, including the possibility of “doubling up” in some cases (Elliott, Young
and Dye). This has also potentially led to greater consideration of housing costs in
migration decisions.
During and following the great recession, the formation of households slowed and
the rebound is not expected to be great, but rather slow and steady (Dunne). While
research on household formation has continued following the recession, the research that
considers housing affordability’s impact on the decision to move during and following the
housing bust has been limited to using the traditional measure of affordability or residual
techniques (M. Stone) (M. E. Stone). Some have argued that an approach including the
quality of housing should be considered in the determination of housing
standards/affordability (Bogdon and Can). Others have argued that the attributes of the
neighborhood should be considered in the determination of housing affordability
(Mulliner, Smallbone and Maliene). However, even while giving consideration to housing
standards and local amenities, the consideration for make-up of the household is often
overlooked.
The housing affordability measure that is commonly used, or used as a starting
point for modification in research, is a measurement based upon the standard set by HUD.
This standard states that 30% or less of household income should be dedicated to housing.
88

Since this has been the HUD standard and is used by lenders and policy-makers, it has
become by default the accepted measure (Schwartz and Wilson). The measure hasn’t
changed as the make-up of households has changed, and therefore, does not adequately
reflect what is truly happening to affordability over time as stated in the previous chapter.
Researchers have considered mobility to job opportunities as a reason for leaving as
well as housing ownership as a reason for staying. This research considers the cost of
housing to the individual based upon the alternative measure discussed in the previous
chapter. This study helps to provide greater understanding of the influences on migration
decisions that have not been thoroughly considered by the existing literature. It also helps
to focus the attention on the specific impacts on Latino immigrants.
Data and Model
In order to accomplish this analysis, multiple research methods have been used.
Initially, demographic analysis of metropolitan areas was used to determine the
affordability of the metro for the general population and for Latino immigrants. American
Community Survey data for 2006 through 2011 was used to provide a baseline and to
determine the changes immediately after the economic recession and the effects during the
recovery. In addition to a demographic approach, a probit regression model was used to
estimate the likelihood of migration after the housing bust as a result of the cost of housing
using the alternative measure of affordability.
The dataset used for this analysis is drawn from the 2006 - 2011 American
Community Survey (ACS) microdata set. The data were obtained from Integrated Public
Use Microdata Series (IPUMS) provided by the Minnesota Population Center (Steven
Ruggles) and include data at the national, state, metro summary level, county and at the
89

PUMA level. The data included household and individual level characteristics that provide
detailed information on demographic, economic, and mobility attributes.
The specific attributes used in this study are the following: household and individual
income, age of the head of household and individual age, mobility within the past year,
employment, education, marital status, length of time in the United States (immigrant or
non-immigrant and length of time), and ethnic background (Latino or non-Latino).
Additionally, metropolitan level characteristics included in the analysis are: length of time
Latino-immigrants have been in the United States, overall employment in the metro, and
the affordability of the metropolitan area based upon the alternative measure of
affordability). Ethnic background was determined from the “Hispanic Origin” variable
provided by the ACS. People self-declaring that they are “Not Hispanic” were considered to
be not Latino; all others were considered to be of Latino origin. Immigrant status was
determined using the variable YRSUSA2 coded by IPUMS and based upon the ACS variable
“year of immigration”. This is coded with “N/A”- not applicable (native born) and 5-year
increments with the final group being “21+ years”.
Additionally, metropolitan level characteristics are included in the analysis, such as:
Percent of Latino-Immigrants who have been in the country less than 10 year and Overall
employment in the metro. The metro area of origin was used for the regression analysis for
the change in overall employment and the level of affordability between metros (refer to
table A-6 in the appendix).
Group home and incarcerated people were excluded from the dataset as motivations
for moving, in these cases, vary from the motivations of the general population. Individuals
were classified as Latino-Immigrant and Other. The Other category includes all native born
90

people and all people who are immigrants, but not Latino. Variables, such as housing
quality, are not included in the analysis since this data is not thoroughly provided in the
ACS data. Additionally, since government mandated improvements in housing quality are
not within the control of individuals, there is little justification for including it as part of the
choice of individuals. Many of the head of household characteristics were assigned to all the
members of the household for classification and analysis purposes.
This study aggregates data at the metropolitan level. The metropolitan areas which
are selected do not align directly with the immigrant gateway typology outlined by Audrey
Singer in her 2004 paper. However, there are many of similarities in the metropolitan areas
selected for this analysis. For consistency, the 25 metropolitan areas selected in this
analysis are the same as those presented in the first chapter of this dissertation co-
authored with Gary Painter, PhD.
The following section provides summary statistics on the changes in the
employment rate, the change in the percent of Latino-immigrants and recent Latino-
immigrants. The regression model uses a probit framework to test what are the factors that
would influence the probability of moving out of the metropolitan area. This is done with
three models that provide first, a base model then the addition of the single income
measure for affordability, followed by the substitution of the dual income measure for
affordability.
Summary Statistics
The recession started in the housing market and moved to the overall economy. The
increase in unemployment and underemployment that resulted forced many people to
rethink their housing options. Additionally, many people lost their homes during the
downturn due to lack of ability to pay or loss of willingness to pay. The movement to new
91

housing was partially the result of the housing market bust and the subsequent loss of
existing housing. It was also the result of people losing jobs and seeking out new housing
options where employment potential was better.
Figure 17 shows the percent of recent Latino immigrants as a percent of all Latino
immigrants in each metro area, and highlights the areas where this population changed
dramatically over the study period. Recent Latino immigrants are identified as people who
have been in the US for less than 10 years. The reduction of recent Latino immigrants was
caused by many factors, including changes to laws in places such as Arizona that made it
less hospitable to Latino-immigrants, especially younger, non-citizens, and greater
deportation rates. The downturn in the overall economy was likely a strong contributing
factor as well. The eight metro areas with the highest (top third) percent of recent Latino
immigrants did not all change at the same rate. Some of the greatest percentage point
losses are from metro areas with average percentages of recent Latino immigrants in 2006.

92

Figure 17 - New Latino Immigrants as Percent of All Latino Immigrants

The percent of Latino immigrants overall changed very little during the 5-year
period with Phoenix being the outlier with a 3 percentage point reduction. The major
change in the Latino immigrant population was in the percent of recent immigrants. Nearly
every metro, with the exception Baltimore, Columbus, Richmond, and San Antonio, had a
loss in the actual number of recent Latino immigrants. Part of this is explained with the
transition from being considered a “recent” Latino immigrant (in the country for fewer
than 10 years). However, five metros had a net loss in the number of Latino immigrants
between 2006 and 2011, meaning that the number of all Latino immigrants, not just recent,
had dropped.
% Point Change
2006 2007 2008 2009 2010 2011 2006 to 2011
Atlanta,GA 67% 62% 58% 54% 51% 38% (29)
Austin,TX 57% 50% 53% 47% 43% 34% (23)
Baltimore,MD 55% 57% 50% 52% 52% 49% (5)
Charlotte-Gastonia-Rock Hill,NC-SC 68% 67% 63% 57% 49% 35% (33)
Columbus,OH 69% 67% 65% 55% 64% 53% (16)
Dallas-FortWorth,TX 49% 45% 46% 41% 37% 32% (17)
Detroit,MI 49% 50% 46% 49% 39% 24% (24)
Greensboro-WinstonSalem-HighPoint,NC 62% 58% 57% 56% 51% 31% (31)
Indianapolis,IN 62% 65% 63% 65% 59% 44% (18)
KansasCity,MO-KS 65% 62% 57% 52% 49% 50% (15)
LasVegas,NV 46% 42% 43% 40% 33% 29% (18)
LosAngeles-Long Beach,CA 28% 27% 27% 26% 23% 19% (9)
Memphis,TN/AR/MS 73% 66% 63% 57% 55% 46% (27)
Nashville,TN 67% 64% 59% 61% 63% 49% (19)
Phoenix,AZ 53% 50% 46% 38% 29% 22% (31)
Raleigh-Durham,NC 71% 63% 63% 59% 50% 44% (28)
Richmond-Petersburg,VA 68% 55% 54% 62% 62% 51% (17)
Riverside-SanBernardino,CA 29% 30% 27% 23% 23% 17% (13)
Sacramento,CA 40% 34% 42% 37% 30% 25% (15)
SanAntonio,TX 35% 33% 28% 29% 32% 29% (6)
SanFrancisco-Oakland-Vallejo,CA 39% 37% 36% 33% 33% 29% (10)
SanJose,CA 41% 39% 37% 35% 33% 25% (16)
Seattle-Everett,WA 54% 52% 50% 51% 46% 40% (15)
Washington,DC/MD/VA 50% 50% 48% 46% 43% 37% (13)
WestPalmBeach-BocaRaton-Delray Beach,FL 50% 47% 46% 44% 37% 35% (14)
New Latino Immigrants as a percent of Latino Immigrants
93

The movement of people between areas is the result of many factors, such as
employment opportunities and cost of housing. Employment opportunities are commonly
thought of as a strong driver in the decision to move and the location selected. The
employment rate for Latino immigrants in the metropolitan areas changed at different
rates during the recovery period, as shown in Figure 18 . Both Las Vegas and Phoenix had
high single digit declines in the percentage of Latino immigrants working. This differs from
the changes in the overall population, Figure 19, which show losses in all metros.
Figure 18 - Employment Rate of Latino Immigrants

In 2011, the employment rate of Latino immigrants is higher than the overall
population in over half of the metros. While this is a change from 2006, there does not
% Point Change
Latino Immigrant 2006 2007 2008 2009 2010 2011 2006-2011
Atlanta,GA 78% 76% 79% 73% 70% 73% (5)
Austin,TX 76% 77% 80% 75% 71% 71% (4)
Baltimore,MD 78% 75% 82% 77% 82% 76% (3)
Charlotte-Gastonia-Rock Hill,NC-SC 76% 75% 78% 70% 71% 73% (3)
Columbus,OH 69% 74% 77% 68% 70% 74% 5
Dallas-FortWorth,TX 75% 76% 77% 73% 71% 71% (4)
Detroit,MI 62% 58% 73% 60% 68% 62% 1
Greensboro-WinstonSalem-HighPoint,NC 78% 74% 75% 75% 69% 78% 1
Indianapolis,IN 75% 76% 77% 70% 69% 70% (5)
KansasCity,MO-KS 72% 72% 75% 75% 71% 74% 2
LasVegas,NV 76% 74% 76% 70% 69% 66% (9)
LosAngeles-Long Beach,CA 72% 71% 73% 71% 69% 68% (4)
Memphis,TN/AR/MS 75% 66% 83% 71% 69% 70% (4)
Nashville,TN 77% 76% 77% 75% 74% 71% (6)
Phoenix,AZ 72% 71% 71% 63% 63% 64% (8)
Raleigh-Durham,NC 77% 78% 78% 77% 69% 75% (2)
Richmond-Petersburg,VA 71% 81% 79% 65% 73% 76% 4
Riverside-SanBernardino,CA 68% 68% 68% 65% 63% 63% (6)
Sacramento,CA 68% 71% 71% 67% 66% 65% (3)
SanAntonio,TX 67% 68% 68% 72% 69% 68% 1
SanFrancisco-Oakland-Vallejo,CA 73% 73% 76% 74% 73% 72% (1)
SanJose,CA 73% 74% 74% 72% 71% 72% (1)
Seattle-Everett,WA 77% 80% 80% 71% 70% 75% (1)
Washington,DC/MD/VA 80% 80% 81% 78% 81% 80% 1
WestPalmBeach-BocaRaton-Delray Beach,FL 75% 76% 75% 74% 72% 72% (3)
94

seem to be a clear relationship between metros with higher Latino immigrant employment
in 2006 and those in 2011.
Figure 19 - Employment Rate of Total Population

In contrast to the similar declines in overall population and the Latino immigrant
population as a whole, the recent Latino immigrant population had much larger losses in
many metro areas over the five years, Figure 20. There were double digit losses in 10 of the
25 metros, with Phoenix being the second highest at 15 percentage points. Phoenix’s loss of
the highest percent of recent Latino immigrants coupled with the second lowest
employment rate in 2011 may be partially explained by changes in the political climate as it
% Point Change
All Population 2006 2007 2008 2009 2010 2011 2006-2011
Atlanta,GA 74% 74% 76% 71% 69% 69% (5)
Austin,TX 76% 77% 78% 76% 74% 75% (2)
Baltimore,MD 76% 75% 78% 75% 74% 73% (3)
Charlotte-Gastonia-Rock Hill,NC-SC 75% 75% 77% 72% 69% 70% (5)
Columbus,OH 75% 76% 77% 74% 72% 74% (1)
Dallas-FortWorth,TX 75% 75% 77% 74% 73% 72% (2)
Detroit,MI 69% 67% 69% 63% 63% 64% (5)
Greensboro-WinstonSalem-HighPoint,NC 74% 74% 74% 71% 68% 69% (6)
Indianapolis,IN 76% 75% 76% 73% 71% 71% (5)
KansasCity,MO-KS 77% 76% 78% 75% 75% 75% (2)
LasVegas,NV 75% 74% 75% 71% 68% 67% (8)
LosAngeles-Long Beach,CA 72% 72% 73% 70% 68% 68% (4)
Memphis,TN/AR/MS 71% 73% 74% 69% 68% 68% (3)
Nashville,TN 75% 75% 76% 74% 72% 72% (3)
Phoenix,AZ 75% 73% 75% 70% 69% 69% (5)
Raleigh-Durham,NC 76% 76% 77% 75% 72% 74% (3)
Richmond-Petersburg,VA 76% 76% 78% 74% 72% 73% (3)
Riverside-SanBernardino,CA 69% 70% 69% 65% 62% 62% (7)
Sacramento,CA 71% 73% 73% 68% 65% 65% (6)
SanAntonio,TX 71% 71% 73% 71% 70% 69% (2)
SanFrancisco-Oakland-Vallejo,CA 74% 74% 75% 72% 71% 71% (3)
SanJose,CA 73% 73% 74% 71% 70% 71% (2)
Seattle-Everett,WA 76% 77% 78% 75% 73% 73% (3)
Washington,DC/MD/VA 79% 79% 80% 78% 78% 77% (2)
WestPalmBeach-BocaRaton-Delray Beach,FL 74% 73% 74% 70% 70% 69% (6)
95

relates to the acceptance of Latino immigrants. Interestingly, Baltimore and Richmond had
increases in the number of recent Latino immigrants and the percent employed of the same
group. Recent Latino immigrants showed gains in employment rate in only six of the
metros. These gains were uneven over the recovery period and are influenced by the small
size of the population in most of the metros.
Figure 20 - New Latino Immigrant Employment Rate

During the recovery period, there were also differences in the changes in the cost of
housing with some areas recovering faster than others. This meant an increase in the cost
of owner occupied housing as well as the already high cost of rental housing. Additionally,
% Point Change
New Latino Immigrant 2006 2007 2008 2009 2010 2011 2006-2011
Atlanta,GA 79% 74% 79% 74% 69% 68% (11)
Austin,TX 75% 75% 80% 71% 66% 65% (10)
Baltimore,MD 80% 71% 78% 73% 83% 69% (10)
Charlotte-Gastonia-Rock Hill,NC-SC 75% 73% 76% 66% 65% 65% (10)
Columbus,OH 66% 73% 78% 69% 67% 69% 4
Dallas-FortWorth,TX 73% 74% 74% 69% 67% 66% (7)
Detroit,MI 58% 47% 76% 54% 70% 67% 9
Greensboro-WinstonSalem-HighPoint,NC 80% 74% 74% 74% 66% 69% (11)
Indianapolis,IN 79% 73% 72% 65% 64% 61% (17)
KansasCity,MO-KS 68% 70% 72% 75% 67% 72% 4
LasVegas,NV 75% 71% 72% 69% 67% 65% (10)
LosAngeles-Long Beach,CA 71% 71% 71% 69% 68% 67% (4)
Memphis,TN/AR/MS 76% 63% 81% 66% 63% 72% (4)
Nashville,TN 75% 77% 77% 74% 71% 65% (11)
Phoenix,AZ 73% 69% 72% 58% 59% 57% (15)
Raleigh-Durham,NC 77% 79% 81% 77% 64% 73% (3)
Richmond-Petersburg,VA 73% 81% 74% 57% 69% 74% 1
Riverside-SanBernardino,CA 67% 66% 65% 58% 59% 54% (13)
Sacramento,CA 63% 72% 70% 68% 67% 61% (2)
SanAntonio,TX 67% 68% 68% 66% 64% 64% (3)
SanFrancisco-Oakland-Vallejo,CA 70% 70% 74% 70% 73% 69% (0)
SanJose,CA 74% 75% 70% 63% 66% 70% (3)
Seattle-Everett,WA 77% 83% 77% 71% 64% 82% 4
Washington,DC/MD/VA 76% 77% 78% 73% 78% 80% 3
WestPalmBeach-BocaRaton-Delray Beach,FL 77% 74% 74% 71% 71% 72% (5)
96

the metro areas had varying decreases in the cost of housing as a result of the bust, so
increases over the recovery period are relative. Using the alternative measure of
affordability defined in the previous chapter, the percent of households that can’t afford
housing costs without spending more than 30% of the single highest income in the
household, Figure 21, was created.
Figure 21 - Change in Percent of Households that cannot afford housing with 30% or less of a SINGLE Income

This figure has the ten metros with the highest percent of households that can’t
afford housing costs with 30% or less of income in gray. Palm Beach and Riverside had the
highest at 78% and 76%. This means that 78% of households pay more than 30% for
Point Change in %
Single Income 2006 2007 2008 2009 2010 2006 to 2010
Atlanta,GA 63% 64% 62% 65% 66% 2.9
Austin,TX 63% 63% 64% 65% 67% 4.5
Baltimore,MD 64% 66% 66% 68% 67% 2.8
Charlotte-Gastonia-Rock Hill,NC-SC 61% 59% 59% 62% 63% 2.2
Columbus,OH 62% 63% 61% 62% 63% 0.5
Dallas-FortWorth,TX 65% 64% 64% 64% 65% (0.3)
Detroit,MI 66% 66% 66% 68% 67% 1.4
Greensboro-WinstonSalem-HighPoint,NC 65% 64% 62% 64% 65% 0.0
Indianapolis,IN 59% 58% 58% 60% 59% (0.2)
KansasCity,MO-KS 61% 60% 60% 60% 63% 2.3
LasVegas,NV 72% 73% 73% 73% 72% (0.1)
LosAngeles-Long Beach,CA 74% 74% 75% 75% 76% 2.0
Memphis,TN/AR/MS 66% 65% 64% 67% 68% 1.5
Nashville,TN 64% 62% 64% 63% 66% 2.0
Phoenix,AZ 66% 68% 68% 68% 68% 2.1
Raleigh-Durham,NC 58% 59% 58% 59% 62% 3.6
Richmond-Petersburg,VA 62% 64% 65% 65% 66% 4.3
Riverside-SanBernardino,CA 76% 77% 77% 77% 77% 0.8
Sacramento,CA 73% 73% 73% 73% 73% (0.0)
SanAntonio,TX 64% 62% 66% 65% 64% 0.3
SanFrancisco-Oakland-Vallejo,CA 71% 70% 70% 70% 71% 0.4
SanJose,CA 69% 69% 68% 69% 70% 0.7
Seattle-Everett,WA 66% 66% 67% 66% 67% 1.1
Washington,DC/MD/VA 63% 63% 63% 64% 64% 0.6
WestPalmBeach-BocaRaton-Delray Beach,FL 78% 80% 79% 80% 79% 0.8
97

housing when only one income is considered. Some of the metro areas show signs of
improvement after the housing bust, but most continue on the same trajectory. Even with
the extreme drop in the prices of homes, the cost, as a percent of income, was still a burden
for most households when only one income is considered. It should be noted that Figure 21
is a combined chart that includes both renters and owner occupants and that rents
increased in many metros, even as home prices declined.
The metropolitan area with the greatest percentage increase in households that
can’t afford housing at 30% or less of income using only the single highest income was
Austin, followed closely by Richmond. It should be noted that Austin’s overall Latino
immigrant population makes up just over 9.25% of the total population in 2011, after the
loss from 10.08% in 2006. The drop in recent Latino immigrants was 23 percentage points
over this same period.
The above figure shows that only three metros improved in affordability when
considering only a single income and this improvement was marginal. This is not as telling
as the figure below, which represents the percent of households that have to pay more than
30% of the two highest incomes to housing costs (Figure 22 ). With the exception of 6
metros, all have more than 50% of the households paying more than 30% of the two
highest incomes to housing costs in 2006. By 2010 all of the metros had over 50% of
households paying more than 30% two incomes to housing costs, with the exception of
Indianapolis and Raleigh, which were both very close. None of the metros, except San
Antonio, had improvement at the end of the period, even though housing prices dropped
dramatically. This is partially a result of rental rates increasing as housing prices decreased
in most metro areas.
98

Figure 22 -Change in Percent of Households that cannot afford housing with 30% or less of Two Incomes

The changes in affordability, the percent of recent Latino immigrants in the
population, and those employed declined in most metro areas. That data in Figure 23
highlights that nearly every metro area had a decline in the percent of recent Latino
immigrants as a percent of the overall Latino immigrant population. It also highlights the
loss in employment by recent Latino immigrants and the increase in the percent of
households that cannot afford housing costs using less than 30% of two incomes.
Point Change in %
Dual Incomes 2006 2007 2008 2009 2010 2006 to 2010
Atlanta,GA 51% 51% 50% 52% 55% 3.9
Austin,TX 51% 51% 50% 51% 54% 3.4
Baltimore,MD 51% 53% 54% 55% 54% 3.1
Charlotte-Gastonia-Rock Hill,NC-SC 48% 47% 46% 50% 52% 3.7
Columbus,OH 49% 49% 47% 49% 51% 1.6
Dallas-FortWorth,TX 52% 51% 51% 52% 53% 0.4
Detroit,MI 56% 56% 56% 58% 58% 2.6
Greensboro-WinstonSalem-HighPoint,NC 53% 52% 51% 53% 55% 2.0
Indianapolis,IN 47% 46% 47% 48% 48% 1.9
KansasCity,MO-KS 48% 47% 47% 48% 50% 2.3
LasVegas,NV 58% 59% 60% 60% 59% 1.4
LosAngeles-Long Beach,CA 63% 64% 64% 64% 65% 2.5
Memphis,TN/AR/MS 56% 53% 53% 56% 57% 0.9
Nashville,TN 52% 50% 51% 51% 53% 1.7
Phoenix,AZ 53% 55% 57% 57% 57% 4.1
Raleigh-Durham,NC 47% 47% 45% 47% 50% 2.8
Richmond-Petersburg,VA 48% 50% 51% 52% 54% 5.9
Riverside-SanBernardino,CA 65% 65% 66% 66% 66% 0.8
Sacramento,CA 61% 61% 60% 62% 62% 0.8
SanAntonio,TX 53% 51% 55% 54% 53% 0.0
SanFrancisco-Oakland-Vallejo,CA 60% 60% 60% 59% 60% 0.1
SanJose,CA 58% 59% 58% 58% 58% 0.5
Seattle-Everett,WA 53% 53% 54% 54% 55% 1.8
Washington,DC/MD/VA 50% 50% 51% 51% 50% 0.7
WestPalmBeach-BocaRaton-Delray Beach,FL 69% 72% 71% 72% 71% 1.9
99

Figure 23 - Change in Measures for New Latino Immigrants

The comparison for Latino immigrants that are not “recent” is slightly different, only
in respect to the change in the size of the population as a percent of the overall population,
Figure 24. Las Vegas had a decline in population in all three groups and a decline in
employment; however, the affordability for dual income households remained fairly stable.
In contrast, Washington DC had an increase in the percent of employment for recent and
established Latino immigrants, and has a slight uptick in the population of Latino
New Latino
Immigrant %
New Latino Immigrant
Employment
Affordability
Dual Income
Atlanta,GA (29) (11) 4
Austin,TX (23) (10) 3
Baltimore,MD (5) (10) 3
Charlotte-Gastonia-Rock Hill,NC-SC (33) (10) 4
Columbus,OH (16) 4 2
Dallas-FortWorth,TX (17) (7) 0
Detroit,MI (24) 9 3
Greensboro-WinstonSalem-HighPoint,NC (31) (11) 2
Indianapolis,IN (18) (17) 2
KansasCity,MO-KS (15) 4 2
LasVegas,NV (18) (10) 1
LosAngeles-Long Beach,CA (9) (4) 3
Memphis,TN/AR/MS (27) (4) 1
Nashville,TN (19) (11) 2
Phoenix,AZ (31) (15) 4
Raleigh-Durham,NC (28) (3) 3
Richmond-Petersburg,VA (17) 1 6
Riverside-SanBernardino,CA (13) (13) 1
Sacramento,CA (15) (2) 1
SanAntonio,TX (6) (3) (0)
SanFrancisco-Oakland-Vallejo,CA (10) (0) 0
SanJose,CA (16) (3) 0
Seattle-Everett,WA (15) 4 2
Washington,DC/MD/VA (13) 3 1
WestPalmBeach-BocaRaton-Delray Beach,FL (14) (5) 2
Percentage Point Change in:
100

immigrants while the affordability remained fairly stable. The overall population increased
in size in all metros, with the exception of Detroit. The employment outcomes, however,
declined in all metros for the overall population.
Figure 24 -Change in Measures for "Aged" Latino Immigrants

The evidence above suggests that all groups had decreases in overall employment at
a time when overall affordability was also becoming harder to achieve. The impacts on
Latino immigrants, recent Latino immigrants, and the overall population differed in degree,
but the groups had similar outcomes in some respects.
Moving to employment opportunities has historically helped to mitigate the income
deficits that result from the trough of the business cycle. Had housing costs remained stable
Latino Immigrant % of Total
Population
Latino Immigrant
Employment
Affordability with
Dual Income
Atlanta,GA 0 -5 4
Austin,TX -1 -4 3
Baltimore,MD 1 -3 3
Charlotte-Gastonia-Rock Hill,NC-SC 0 -3 4
Columbus,OH 0 5 2
Dallas-FortWorth,TX -1 -4 0
Detroit,MI 0 1 3
Greensboro-WinstonSalem-HighPoint,NC 0 1 2
Indianapolis,IN 1 -5 2
KansasCity,MO-KS 0 2 2
LasVegas,NV -1 -9 1
LosAngeles-Long Beach,CA -1 -4 3
Memphis,TN/AR/MS 0 -4 1
Nashville,TN 1 -6 2
Phoenix,AZ -3 -8 4
Raleigh-Durham,NC 0 -2 3
Richmond-Petersburg,VA 1 4 6
Riverside-SanBernardino,CA -1 -6 1
Sacramento,CA -1 -3 1
SanAntonio,TX 1 1 0
SanFrancisco-Oakland-Vallejo,CA 0 -1 0
SanJose,CA -1 -1 0
Seattle-Everett,WA 0 -1 2
Washington,DC/MD/VA 1 1 1
WestPalmBeach-BocaRaton-Delray Beach,FL 1 -3 2
Percentage Point Change in:
101

during the recession and recovery, the effect of moving to employment may have assisted
with a speedier recovery. In order to determine the effects of affordability on the decision
to move, a regression analysis was conducted as described in the following section.
Regression Results
The above evidence suggests that there is no clear connection between affordability
in a metro area and the likelihood of greater employment. In order to gain a better
understanding of the role that affordability played in the decision to move, a multivariate
probit model was used. Below is evidence on the predictors of the decision for households
to move.
First, the factors used to predict mobility are focused on household characteristics
to determine what characteristics increase the likelihood of a person to remain in the same
metro area or move to a different metro area. In addition to individual and household
characteristics, the change in the overall employment in the metro of origin, and the
network ties that Latino immigrants may have are included.
The network ties of the Latino immigrant community influences their mobility
choices and the stronger ties should reduce mobility out of the area. In order to estimate
the network strength, the percentage of the Latino immigrant population that has been in
the country for more than 10 years (C. Liu) (Liu and Painter) is used as a proxy for the
strength of network tie.
The final regression models include the percentage point change in affordability of
the metropolitan area of origin using first the single income measure and then the dual
income measure. It is expected that the affordability of a metro will have a greater impact
on the likelihood of moving than the network ties. This differs from previous research that
focuses on the cost of housing to the households rather than the ratio of affordability to the
102

individual income earners in the household. A variable representing the median housing
cost for the metro is included to compare to the regressions including the affordability
measure.
The first model representing the household level and basic metro variables for the
Latino immigrant population is provided in the addenda as figure A-1. The likelihoods are
in relation to the omitted: stayed in the same metro, native head of household, married
male head of household, and a head of household with less than a high school education.
The impacts of these variables on the likelihood of moving is as expected with
people being less likely to be employed if they moved, people who have been in the country
longer are also less likely to move between metro areas. The variables for marital status
and gender of the head of household remain fairly consistent and show slightly greater
likelihood for female headed households, with greater likelihood for households headed by
single-males.
A college education for the head of household seems to have a stronger influence in
moves between metro areas. This could be an indication of moving job opportunities. The
metro level variables change through the years as the overall economy changed. Overall
employed has an inconsistent probability, this is likely a result of the changing
circumstances of the economy and the evolving recovery. Consistently through the years,
the Latino immigrant network tie variable shows that metros with a higher percent of
established Latino immigrant communities decrease the likelihood of moving between
metros.
103

Figure 25 combines the two separate regressions with the variables for Single and
Dual income affordability added to the base regression presented. Full regression
summaries are in the addenda as figures A-2 & A-3.
Figure 25 - Summary of Regression Results

In the SINGLE income regression model, the variable representing the point change
in the percent of households that can afford housing while spending less than 30% of the
SINGLE highest income on housing is added to the regression. It is expected that metros
with higher percentages of people who cannot afford housing on a single income will be
less attractive to people and may encourage moving to employment opportunities.
The result of the addition of the single income affordability measure shows that
likelihood of a move between metros is stronger in the early years of the recession as the
percent of people in the metro who cannot afford housing at 30% or less of income with
only a single income increase. The strength of the likelihood increased at the point
following the beginning of the recovery, in 2011. This is often identified as the period of
transition from falling housing prices to increasing housing prices.
This summary also shows the use of the variable representing the point change in
the percent of households that can afford housing while spending less than 30% of the
TWO highest incomes on housing is added to the regression. It is expected that metros with
higher percentages of people who cannot afford housing on two incomes, will be less
attractive to people moving to employment opportunities. However, the effect of this
dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err.
Pct Point change in % Afford with Single Income 9.471 *** 0.031 7.140 *** 0.033 7.311 *** 0.035 7.779 *** 0.040 13.174 *** 0.038
Pct Point change in % Afford with Dual Incomes 10.394 *** 0.029 6.384 *** 0.029 6.866 *** 0.033 8.745 *** 0.037 11.490 *** 0.035
Observations 7,258,352 7,138,151 7,231,099 7,280,319 7,202,236
Pseudo R2 (Single Income) 0.057 0.053 0.051 0.051 0.099
Pseudo R2 (Dual Income) 0.073 0.053 0.052 0.061 0.090
*p < .05; **p< .01; ***p< .001
2007 2008 2009 2010 2011
104

variable seems to be strong than the single income measure at the beginning of the
recession, but is slightly weaker in 2008 and 2009. It is also expected that the effect of the
dual income variable will be greater and more consistent than that of the single income
variable used in the previous regression since the inability to afford housing using two
incomes would likely be a greater deterrent for people when considering a move to a new
metro. However, the single income variable may be more representative of those who
moved since there is a stronger likelihood that a mover is a single male.
The variables for marital status and gender of the head of household remain fairly
consistent and in the expected directions, with single male head of households more likely
to move both within and between metro areas than other gender/statuses. A college
education for the head of household seems to have a stronger influence in moves between
metro areas. This could be an indication of moving job opportunities.
The addition of the affordability measures to the original regression increased the
likelihood of the impact of changes in employment opportunities and made the direction of
the effect more consistent. The results appear to validate the original proposition that
decisions to move are influenced not only by the possibility of employment opportunities,
but also by the expected cost of living.
For the purposes of comparison, an additional regression analysis was performed,
using the median housing cost as a variable instead of the affordability measures. The full
results of this analysis are present in table A-4 in the addenda. It shows that housing costs
had little impact on the likelihood to move and the addition of the variable resulted in
inconsistent signs in the network ties and overall employment change variables.
105

Addenda A-5 includes the regression using the traditional measure of housing
affordability instead of the alternative measure proposed in the previous essay. The
substitution of the traditional measure results in a coefficient that makes sense for the
affordability. It shows that as the percent of people that can afford housing at 30%
increase, the likelihood of them moving is diminished. However, the use of the variable
causes the overall employment coefficient to take on a positive sign which would indicate
that people are more likely to leave an area if employment increases in the area. This is
counter to the expected result that people would be less likely to leave an area of the
employment situation is improving.
Comparing the traditional measure and the median housing cost to the alternative
measures provides greater support for the use of the alternative affordability measure
rather. The regression results of the alternative measures fall more in line with the
expectations that Latino immigrants move from places that cost more, stay in places that
have higher employment, and stay in places that have greater network ties. Further
research using panel data or interviews would need to be completed in order to absolutely
determine that employment opportunities in one metro were forgone in order to pursue
lower paying employment opportunities in an area that has a lower cost of living.
Conclusion
Using data from the ACS and an alternative measure of housing affordability, this
research documents the changes in population and employment outcomes for Latino
immigrants from 2006 to 2011, and explores the potential impacts of housing affordability
on the decision to move for potential employment opportunities. The 25 metro areas were
selected to correspond to the metro areas selected in the first essay. The recent recession
provided an opportunity to capitalize on a natural experiment of a specific shock to
106

housing related employment and to the general downturn of the economy to measure the
outcomes for Latino immigrants.
The evidence suggests that recent Latino immigrants were particularly hard hit as a
result of the recession. The absolute numbers and relative percent of recent Latino
immigrants declined in most metropolitan areas. Simultaneously, the affordability of
housing got worse in most metro areas. The regression models confirm that the effects of
increasing housing costs and loss of affordability tend to encourage the likelihood of people
moving to other metro areas.
The regressions also highlight the benefits of using the alternative measure to better
capture the effects of affordability on Latino immigrants. The use of the single income
measure provides insight into the likelihood of moving and is especially important for
mobile Latino immigrants, who in many cases, are single males. The effect of the dual
income measure was also consistent with the expected outcomes.
One of the benefits of using the two alternative measures is that they capture
slightly different degrees of likelihood. As expected, basing the analysis strictly on the
single income measure gives a stronger likelihood of movement in most years. Having this
disaggregated information can assist policy-makers in the better determining the extent of
the affordability problem since the level of affordability is not muted by the changing
composition of households. Both measures will be useful in the measurement of outcomes
for non-Latino immigrants as well for Latino immigrants and will prove most useful as
more households add earners beyond two.
107

References
Bogdon, Amy S. and Ayse Can. "Indicators of local housing affordability: Comparative and
spatial approaches." Real Estate Economics (1997): 43-80.
Chapple, Karen and T. William Lester. "The resilient regional labour market? The US case."
Cambridge journal of regions, economy and society (2010): 85-104.
Coulson, N. Edward and Lynn M. Fisher. "Housing tenure and labor market impacts: The
search goes on." Journal of Urban Economics (2009): 252-264.
Dunne, Timothy. "Household Formation and the Great Recession." Staff report. 2012.
Ellen, Ingrid Gould and Samuel Dastrup. "Housing and the Great Recession." Policy Brief.
2012.
Elliott, Diana B., Rebekah Young and Jane Lawler Dye. "Variation in the formation of
complex family households during the recession." Symposium paper presented at
National Council on Family Relations’ 73rd Annual Conference. Florida, 2011.
Head, Allen and Huw Lloyd-Ellis. "Housing liquidity, mobility and the labour market." The
Review of Economic Studies (2012).
Hurd, Michael D. and Susann Rohwedder. "Effects of the Financial Crisis and Great
Recession on American Households." Working Paper. 2010 .
Kochhar, Rakesh, C. Soledad Espinoza and Rebecca Hinze-Pifer. After the great recession:
foreign born gain jobs; native born lose jobs. Washington D.C.: Pew Hispanic Center,
2010.
Kothari, Siddharth, Itay Saporta-Eksten and Edison Yu. "The (un) importance of
geographical mobility in the Great Recession." Review of Economic Dynamics (2013):
553-563.
Liu, Cathy. "Employment Concentration and Job Quality for Low-Skilled Latino
Immigrants." Journal of Urban Affairs (2011): 117-142.
Liu, Cathy Yang and Gary Painter. "Travel behavior among Latino immigrants: the role of
ethnic concentration and ethnic employment." Journal of Planning Education and
Research (2011).
Mulliner, Emma, Kieran Smallbone and Vida Maliene. "An assessment of sustainable
housing affordability using a multiple criteria decision making method." Omega
(2013): 270-279.
108

Orshansky, M. "Counting the poor: Another look at the poverty profile." Social Security
Bulletin (1965): 3-29.
Painter, Gary and Ray Calnan. "The Response of Latino Immigrants to the Great Recession:
Occupation and Residential (Im)mobility." Forthcoming.
Painter, Gary and Zhou Yu. "Caught in the housing bubble: Immigrants’ housing outcomes
in traditional gateways and newly emerging destinations." Urban Studies (2013).
Rhee, Serena and Fatih Karahan. "Geographical reallocation and unemployment during the
great recession: The role of the housing bust." Staff report. 2013.
Schwartz, Mary and Ellen Wilson. "Who Can Afford To Live in a Home?: A look at data from
the 2006 American Community Survey." 2006.
Singer, Audrey and Jill H. Wilson. "The Impact of the Great Recession on Metropolitan
Immigration Trends." Metropolitan Policy Program at Brookings (2010).
Steven Ruggles, J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder,
and Matthew Sobek. "Integrated Public Use Microdata Series: Version 5.0 [Machine-
readable database]." Minneapolis: University of Minnesota, 2010.
Stoll, Michael A. "Great Recession spurs a shift to local moves." 2013.
Stone, Michael E. "The residual income approach to housing affordability: the theory and
the practice." Community Studies Faculty Publication Series. 2011.
Stone, Michael. "What is housing affordability? The case for the residual income approach."
Housing Policy Debate (2006): 151-184.

109

Addenda
A-1 – Base regression for Latino Immigrants

dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err.
Moved between Metro Areas
Employed -0.084 *** 0.002 -0.059 *** 0.002 -0.086 *** 0.002 -0.124 *** 0.002 -0.082 *** 0.002
Length of Time Head of Household has been in the US
0-10 years -0.031 *** 0.003 -0.091 *** 0.003 -0.110 *** 0.003 -0.093 *** 0.003 -0.189 *** 0.003
11-20 years -0.093 *** 0.003 -0.179 *** 0.003 -0.251 *** 0.003 -0.160 *** 0.003 -0.274 *** 0.003
21+ years -0.226 *** 0.003 -0.288 *** 0.003 -0.248 *** 0.003 -0.201 *** 0.003 -0.302 *** 0.003
Marital Status of the Head of Household
Female HH Married 0.047 *** 0.002 -0.010 *** 0.003 0.052 *** 0.003 0.114 *** 0.003 -0.037 *** 0.003
Male HH Single 0.215 *** 0.002 0.142 *** 0.002 0.136 *** 0.003 0.222 *** 0.003 0.164 *** 0.003
Female HH Single 0.042 *** 0.002 0.092 *** 0.003 0.095 *** 0.003 0.120 *** 0.003 0.033 *** 0.003
Education level of Head of Household
HS Diploma 0.088 *** 0.002 0.029 *** 0.002 0.062 *** 0.002 0.110 *** 0.002 0.070 *** 0.002
Some college 0.180 *** 0.003 0.148 *** 0.003 0.238 *** 0.003 0.148 *** 0.003 0.177 *** 0.003
College degree plus 0.276 *** 0.003 0.207 *** 0.003 0.224 *** 0.003 0.325 *** 0.003 0.301 *** 0.003
Overall Employment Chg 1.073 *** 0.064 -3.077 *** 0.077 -2.790 *** 0.054 0.338 *** 0.036 0.454 *** 0.083
Pct of Latino Immig 10+ yrs -0.395 *** 0.008 -1.398 *** 0.009 -1.254 *** 0.008 -1.338 *** 0.009 -1.382 *** 0.009
_cons -1.497 *** 0.006 -0.786 *** 0.007 -1.093 *** 0.006 -1.010 *** 0.007 -0.795 *** 0.007
Observations 7,258,352 7,138,151 7,231,099 7,280,319 7,202,236
Pseudo R2 0.018 0.030 0.030 0.032 0.034
*p < .05; **p < .01; ***p < .001
2007 2008 2009 2010 2011
110

A-2 – Regression for Latino Immigrants with Percent of Households in Metro including the Percentage Point change in Percent of housing
Affordable with a Single Income.

dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err.
Moved between Metro Areas
Employed -0.090 *** 0.002 -0.062 *** 0.002 -0.085 *** 0.002 -0.120 *** 0.002 -0.085 *** 0.002
Length of Time Head of Household has been in the US
0-10 years -0.034 *** 0.003 -0.121 *** 0.003 -0.111 *** 0.003 -0.110 *** 0.003 -0.212 *** 0.003
11-20 years -0.080 *** 0.003 -0.190 *** 0.003 -0.248 *** 0.003 -0.181 *** 0.003 -0.283 *** 0.003
21+ years -0.204 *** 0.003 -0.290 *** 0.003 -0.229 *** 0.003 -0.205 *** 0.003 -0.288 *** 0.003
Marital Status of the Head of Household
Female HH Married 0.056 *** 0.002 -0.011 *** 0.003 0.054 *** 0.003 0.118 *** 0.003 -0.003 0.003
Male HH Single 0.234 *** 0.002 0.138 *** 0.002 0.139 *** 0.003 0.222 *** 0.003 0.156 *** 0.003
Female HH Single 0.060 *** 0.003 0.102 *** 0.003 0.109 *** 0.003 0.124 *** 0.003 0.060 *** 0.003
Education level of Head of Household
HS Diploma 0.102 *** 0.002 0.038 *** 0.002 0.068 *** 0.002 0.096 *** 0.002 0.064 *** 0.002
Some college 0.187 *** 0.003 0.163 *** 0.003 0.254 *** 0.003 0.149 *** 0.003 0.196 *** 0.003
College degree plus 0.321 *** 0.003 0.214 *** 0.003 0.245 *** 0.003 0.313 *** 0.003 0.302 *** 0.003
Overall Employment Chg -2.847 *** 0.067 -2.976 *** 0.077 -3.598 *** 0.054 -0.648 *** 0.036 -2.334 *** 0.082
Pct of Latino Immig 10+ yrs -0.864 *** 0.008 -1.424 *** 0.009 -1.818 *** 0.009 -1.635 *** 0.009 -1.762 *** 0.010
Pct Point change in % Afford with Single Income 9.471 *** 0.031 7.140 *** 0.033 7.311 *** 0.035 7.779 *** 0.040 13.174 *** 0.038
_cons -1.212 0.006 -0.779 0.007 -0.779 0.006 -0.799 0.007 -0.641 0.008
Observations 7,258,352 7,138,151 7,231,099 7,280,319 7,202,236
Pseudo R2 0.057 0.053 0.051 0.051 0.099
*p < .05; **p < .01; ***p < .001
2007 2008 2009 2010 2011
111

A-3 – Regression for Latino Immigrants with Percent of Households in Metro including the Percentage Point change in Percent of housing
Affordable with Two Incomes.

dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err.
Moved between Metro Areas
Employed -0.090 *** 0.002 -0.062 *** 0.002 -0.085 *** 0.002 -0.120 *** 0.002 -0.079 *** 0.002
Length of Time Head of Household has been in the US
0-10 years -0.033 *** 0.003 -0.118 *** 0.003 -0.113 *** 0.003 -0.115 *** 0.003 -0.206 *** 0.003
11-20 years -0.075 *** 0.003 -0.188 *** 0.003 -0.250 *** 0.003 -0.187 *** 0.003 -0.284 *** 0.003
21+ years -0.199 *** 0.003 -0.288 *** 0.003 -0.229 *** 0.003 -0.205 *** 0.003 -0.290 *** 0.003
Marital Status of the Head of Household
Female HH Married 0.057 *** 0.002 -0.012 *** 0.003 0.055 *** 0.003 0.118 *** 0.003 -0.008 ** 0.003
Male HH Single 0.239 *** 0.002 0.136 *** 0.002 0.139 *** 0.003 0.223 *** 0.003 0.156 *** 0.003
Female HH Single 0.065 *** 0.003 0.097 *** 0.003 0.112 *** 0.003 0.125 *** 0.003 0.053 *** 0.003
Education level of Head of Household
HS Diploma 0.102 *** 0.002 0.037 *** 0.002 0.065 *** 0.002 0.094 *** 0.002 0.065 *** 0.002
Some college 0.186 *** 0.003 0.162 *** 0.003 0.253 *** 0.003 0.148 *** 0.003 0.191 *** 0.003
College degree plus 0.322 *** 0.003 0.211 *** 0.003 0.245 *** 0.003 0.309 *** 0.003 0.295 *** 0.003
Overall Employment Chg -5.245 *** 0.068 -3.170 *** 0.078 -4.159 *** 0.055 -1.365 *** 0.037 -3.314 *** 0.083
Pct of Latino Immig 10+ yrs -1.014 *** 0.008 -1.455 *** 0.009 -1.925 *** 0.009 -1.826 *** 0.009 -1.984 *** 0.010
Pct Point change in % Afford with Dual Incomes 10.394 *** 0.029 6.384 *** 0.029 6.866 *** 0.033 8.745 *** 0.037 11.490 *** 0.035
_cons -1.085 0.006 -0.745 0.007 -0.733 0.006 -0.683 0.007 -0.431 0.008
Observations 7,258,352 7,138,151 7,231,099 7,280,319 7,202,236
Pseudo R2 0.073 0.053 0.052 0.061 0.090
*p < .05; **p < .01; ***p < .001
2007 2008 2009 2010 2011
112

A-4 – Regression for Latino Immigrants with Percent of Households in Metro including the Median cost of Housing.

dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err.
Moved between Metro Areas
Employed -0.072 *** 0.002 -0.046 *** 0.002 -0.084 *** 0.002 -0.106 *** 0.002 -0.065 *** 0.002
Length of Time Head of Household has been in the US
0-10 years 0.001 *** 0.003 -0.071 *** 0.003 -0.105 *** 0.003 -0.079 *** 0.003 -0.179 *** 0.003
11-20 years -0.060 *** 0.003 -0.145 *** 0.003 -0.237 *** 0.003 -0.134 *** 0.003 -0.266 *** 0.003
21+ years -0.194 *** 0.003 -0.267 *** 0.003 -0.226 *** 0.003 -0.164 *** 0.003 -0.274 *** 0.003
Marital Status of the Head of Household
Female HH Married 0.059 *** 0.002 -0.015 *** 0.003 0.054 *** 0.003 0.107 *** 0.003 -0.039 *** 0.003
Male HH Single 0.238 *** 0.002 0.154 *** 0.002 0.145 *** 0.003 0.242 *** 0.003 0.172 *** 0.003
Female HH Single 0.066 *** 0.003 0.113 *** 0.003 0.104 *** 0.003 0.132 *** 0.003 0.043 *** 0.003
Education level of Head of Household
HS Diploma 0.110 *** 0.002 0.041 *** 0.002 0.068 *** 0.002 0.114 *** 0.002 0.076 *** 0.002
Some college 0.216 *** 0.003 0.181 *** 0.003 0.258 *** 0.003 0.166 *** 0.003 0.189 *** 0.003
College degree plus 0.344 *** 0.003 0.263 *** 0.003 0.276 *** 0.003 0.367 *** 0.003 0.357 *** 0.003
Overall Employment Chg -4.775 *** 0.064 -6.246 *** 0.078 -1.841 *** 0.053 1.180 *** 0.035 0.564 *** 0.080
Pct of Latino Immig 10+ yrs 0.277 *** 0.008 -0.719 *** 0.009 -0.455 *** 0.009 -0.600 *** 0.009 -0.795 *** 0.010
Median Cost of Housing -1.069E-04 *** 3.540E-07 -7.540E-05 *** 3.340E-07 -6.980E-05 *** 3.430E-07 -7.410E-05 *** 3.430E-07 -6.010E-05 *** 3.100E-07
_cons -0.085 0.007 0.161 0.008 -0.371 0.007 -0.270 0.007 -0.261 0.008
Observations 7,258,352 7,138,151 7,231,099 7,280,319 7,202,236
Pseudo R2 0.061 0.057 0.053 0.059 0.055
*p < .05; **p < .01; ***p < .001
2007 2008 2009 2010 2011
113

A-5 – Regression for Latino Immigrants with Percent of Households in Metro including the Traditional Measure of Affordability

dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err. dF/dx Std. Err.
Moved between Metro Areas
Employed -0.085 *** 0.002 -0.059 *** 0.002 -0.086 *** 0.002 -0.123 *** 0.002 -0.083 *** 0.002
Length of Time Head of Household has been in the US
0-10 years -0.037 *** 0.003 -0.103 *** 0.003 -0.117 *** 0.003 -0.097 *** 0.003 -0.190 *** 0.003
11-20 years -0.102 *** 0.003 -0.192 *** 0.003 -0.263 *** 0.003 -0.167 *** 0.003 -0.274 *** 0.003
21+ years -0.238 *** 0.003 -0.299 *** 0.003 -0.263 *** 0.003 -0.217 *** 0.003 -0.311 *** 0.003
Marital Status of the Head of Household
Female HH Married 0.044 *** 0.002 -0.007 ** 0.003 0.053 *** 0.003 0.117 *** 0.003 -0.033 *** 0.003
Male HH Single 0.208 *** 0.002 0.139 *** 0.002 0.123 *** 0.003 0.213 *** 0.003 0.161 *** 0.003
Female HH Single 0.042 *** 0.002 0.086 *** 0.003 0.087 *** 0.003 0.115 *** 0.003 0.030 *** 0.003
Education level of Head of Household
HS Diploma 0.084 *** 0.002 0.025 *** 0.002 0.057 *** 0.002 0.103 *** 0.002 0.067 *** 0.002
Some college 0.177 *** 0.003 0.143 *** 0.003 0.231 *** 0.003 0.142 *** 0.003 0.177 *** 0.003
College degree plus 0.277 *** 0.003 0.203 *** 0.003 0.213 *** 0.003 0.316 *** 0.003 0.300 *** 0.003
Overall Employment Chg 2.229 *** 0.067 2.454 *** 0.095 0.151 * 0.060 0.695 *** 0.037 1.928 *** 0.086
Pct of Latino Immig 10+ yrs -1.292 *** 0.013 -2.556 *** 0.015 -3.278 *** 0.019 -2.330 *** 0.014 -2.264 *** 0.014
Traditional Measure of Affordability -2.404 *** 0.029 -3.352 *** 0.034 -4.186 *** 0.035 -2.226 *** 0.026 -2.071 *** 0.025
_cons 0.398 0.024 1.686 0.026 2.625 0.031 0.898 0.023 0.937 0.022
Observations 7,258,352 7,138,151 7,231,099 7,280,319 7,202,236
Pseudo R2 0.021 0.035 0.038 0.036 0.038
*p < .05; **p < .01; ***p < .001
2007 2008 2009 2010 2011
114

A-6 – Diagram of Employment Change and Housing Affordability Variable Assignment

Did you move?
Yes
Within the SAME metro?
Use the Housing
affordability/Overall
Employment Change from
the EXISTING metro
To a DIFFERENT metro?
Use the Housing
affordability/Overall
Employment Change from
the PREVIOUS metro
No
Use the Housing
affordability/Overall
Employment Change from
the EXISTING metro
115

Conclusion
This dissertation set out to show the effects of the Great Recession on the mobility
and employment outcomes of Latino immigrants following the shock of the housing bust
and the dramatic changes in employment in construction and employment generally.
Additionally, an analysis of the existing measure of housing affordability was conducted.
This existing measure does not provide an accurate picture of the true dynamics occurring
in housing and therefore an alternative measure that better captures these changes was
created. Finally, an analysis of the effect of housing affordability of the decision to move
was conducted to show that the alternative measure more accurately describes the effect of
affordability on the likelihood of moving.
The Great Recession provided an opportunity for a natural experiment that resulted
in a shock to employment in different sectors at different times. There were distinct
housing and labor markets that were particularly hard hit. This was primarily due to the
fact that the housing industry had fueled much of the recent economic growth and a major
part of this economic downturn was the result of housing and housing related
unemployment. Combining the outcomes of Latino immigrants with the alternative
measure of housing affordability provide a unique opportunity to assess the ability of a
vulnerable group to the dramatic changes in the economy.
The importance of studying Latino immigrants, in particular, is that this is a growing
group in the United States and the people fill many of the basic jobs that facilitate the
growth and recovery of the economy. Additionally, Latino immigrants and their children
are a growing part of the future home buying and renting population. The ability for these
116

people to assimilate and move through “upward mobility” life cycle is often heavily
dependent on the cost of housing.
The first essay, co-authored with Gary Painter, PhD, focused on identifying the
occupations that Latino immigrants tend to concentrate in and the effects of the network
for Latino immigrant communities. This paper takes advantage of the shock to the
construction industry in order to investigate the responses of Latino immigrants in
metropolitan areas that were most heavily concentrated with Latino immigrants in the
construction industry. The essay measured the likelihood of Latino immigrants to move
given the construction employment changes, and the likelihood of obtaining employment if
a move has occurred.
Evidence from the regression models confirmed that job losses in construction
spurred exit from 2007 to 2008, and that job losses at the metropolitan level, more
generally, dominated the reasons for exit. We also found evidence that those who moved
out of the metropolitan area were less likely to be employed. Many of these immigrants
would likely not have been employed in the previous metropolitan area and this may be the
impetus for the move. The permanence of the immigrant population in the metropolitan
area had two effects which might be related. Immigrants were less likely to move when
living in a metropolitan area with more immigrants who had been in the country for more
than 10 years, and less likely to work in those same areas. We posit that the resources
within the immigrant community may have contributed to these two findings. This may
also be connected to the income levels and ability to afford to move as has been shown in
previous research related to the manufacturing job losses in past decades.
117

This helps to fill a gap in the existing literature that has focused on outcomes for
racial groups and changes in employment, but has not considered the direct impact of the
shock to construction related occupations on the employment and mobility outcomes of
Latino-immigrants and how they differed from the population in general.
The second essay addresses the complex issue of housing affordability. It is often
considered from a single vantage point of affordability of an overall household. Various
measures and models have been created to help further refine the concept of affordability;
however, most do not adequately consider the changes in the household itself. The changes
in the make-up of the households are considered in order to identify an alternative
measure for housing affordability. The convenience of the long established measure of
affordability is hard to overcome due to its simplicity. However, the argument against its
use is that the existing measure does not account for changes in the number of people in
the household that are required to work in order to maintain the same level of affordability.
The alternative measure proposed showed that the change in affordability over the
years is much more dramatic when consideration is given to the number of earners in a
household and owners without mortgages are removed from the analysis. The major
advantage of this alternative over other methods, including residual methods, is that it is
still simple to understand and use. The simplicity is part of the reason the existing measure
has become so entrenched and it is also a reason why this alternative should be easier to
adopt than more complicated residual techniques. This essay showed that household
make-up should be considered in the affordability equation in order to overcome the
changing household make-up that makes the existing measure of affordability less useful.
118

The difference between the proposed alternative measure and other potential
alternatives found in the existing literature is that this measure is similar to the existing
measure in that it is simple to understand. It does not change the basic principle of how to
measure affordability and does not require the use of a residual method or the
incorporation of local amenities, which are difficult to quantify. It also provides insight into
the differing affordability issues for single and multiple income households.
In the third essay, the questions from the first two essays are tied together and the
impacts of housing affordability, using the alternative measure, on the decision of Latino
immigrants to move following the housing bust are considered. Latino immigrants, being a
vulnerable group, are more likely to move as a result of job loss and could potentially move
in together in order to pool funds and reduce overall individual costs. Since part of the
decision to move should be based not just on the employment potential, but on the change
in the cost of housing, it is expected that areas that are less affordable will become less
desirable to people as income decreases or becomes less certain.
The analysis showed that the likelihood of a person moving increases as the percent
of affordable housing decreases. This follows the premise in the second essay that
established a measure of the percent of households that pay more than 30% of a single or
two incomes. The application of the single and dual income measure in this analysis had
similar, but slightly different degrees in the results and could be in response to the likely
mover being a single male, when Latino immigrants are considered. A final comparison
between the use of the alternative measure and the median housing costs was completed
and the results show that the affordability measure provided more predictable and
119

consistent results. This is true for the comparison to the traditional measure of housing
affordability.
Overall, this dissertation explored the outcomes of Latino-immigrants following the
Great Recession and proposed an alternative measure of housing affordability. The
ultimate finding in this analysis is that housing affordability does alter the likelihood of
Latino immigrants to make the decision to move to a different metro area and changes in
construction employment had less impact on the outcomes of Latino immigrants than
changes in employment overall. The alternative measure provides a better picture of the
true changes in housing affordability.
As demonstrated, housing affordability must be considered by policy-makers, even
in periods when the price of housing is declining. The alternative measure of housing
affordability can provide policy makers with insight into how well an area is performing
and better target policies to particular areas that have higher rates of unaffordability for
single and dual income households. In the future, as the third income becomes more
important, the new measure will help policy-makers identify areas where the affordability
issue is worsening faster than would be discerned from the traditional measure. The
adoption of this alternative measure would greater increase the understanding by policy-
makers of how to craft policies that address the affordability problem.
This research also highlighted the need for policy-makers to address employment
overall, rather than focus on one sector. Historically, the focus has been on trying to
preserve jobs in the industry that has been most recently and dramatically affected by
economic changes, in this case, construction. Evidence from this research shows that Latino
120

immigrants are most hurt be losses in jobs overall, since they are able to adapt and changes
jobs if alternative employment options are available.
The Great Recession has had dramatic impacts on the labor market in general and
on Latino immigrants specifically. As the recovery continues, additional research will need
to follow those who moved in order to observe whether they eventually improved their
labor market outcomes and whether Latino immigrant employment in construction is able
to return to the pre-recession levels as quickly as other groups. Additional research on the
impacts of changing immigration laws, such as those in Arizona and the increased rate of
deportation, would also shed light on the changing immigrant landscape and migration
patterns.

Abstract (if available)

Abstract During the Great Recession in the US, there were distinct housing and labor markets that were particularly hard hit. This was primarily due to the fact that the housing industry had fueled much of the recent economic growth. This paper takes advantage of the shock to the construction industry in order to investigate the responses of Latino immigrants in metropolitan areas that were most heavily concentrated with Latino immigrants in the construction industry. While declines in construction jobs did predict moving out of a metropolitan area, decline in the overall job market had a larger impact on mobility. The shifts in employment and the dramatic changes in housing costs, following the economic boom, provide a unique opportunity to assess the impact of housing affordability on the decision to move. The analysis shows that affordability based upon single and dual incomes differs mainly in degree. Areas with high unaffordability for two incomes are less likely to have in-migration. The measure created helps to highlight the changes in affordability to households if the make-up of households remained constant through the years.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Beyond spatial mismatch: immigrant employment in urban America

PDF

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

PDF

Living arrangements of young Latino couples: getting by in high-cost Los Angeles

PDF

Reshaping Los Angeles: housing affordability and neighborhood change

PDF

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

PDF

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

PDF

Housing market crisis and underwater homeowners

PDF

The spatial economic impact of live music in Orange County, CA

PDF

Urban spatial transformation and job accessibility: spatial mismatch hypothesis revisited

PDF

Latina elected officials in California: a call to action to prepare and pipeline Latinas into the political process

PDF

Where there is discretion, should law enforcement officers at the local level be involved in enforcing federal immigration law? a study for consideration

PDF

Health care for all? Anti-Latino and anti-immigrant attitudes, health care policy, and the Latino community

PDF

The use of mobile technology and mobile applications as the next paradigm in development: can it be a game-changer in development for women in rural Afghanistan?

PDF

Evergreen economies: institutions, industries and issues in the green economy

PDF

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

PDF

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

PDF

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

PDF

The interactions between housing and business

PDF

The impact of social capital: a case study on the role of social capital in the restoration and recovery of communities after disasters

PDF

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

Asset Metadata

Creator Calnan, Raymond (author)

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

School School of Policy, Planning and Development

Degree Doctor of Philosophy

Degree Program Policy, Planning, and Development

Publication Date 07/17/2015

Defense Date 06/24/2015

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

Tag great recession,housing affordability,Latino-immigrants,mobility,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Painter, Gary (committee chair), Myers, Dowell (committee member), Suro, Roberto (committee member)

Creator Email calnan@usc.edu,raycalnan@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-597838

Unique identifier UC11299531

Identifier etd-CalnanRaym-3629.pdf (filename),usctheses-c3-597838 (legacy record id)

Legacy Identifier etd-CalnanRaym-3629.pdf

Dmrecord 597838

Document Type Dissertation

Format application/pdf (imt)

Rights Calnan, Raymond

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