the impacts of remittances, residency status and financial attachment on housing tenure for...

28
2007 V35 4: pp. 451–478 REAL ESTATE ECONOMICS The Impacts of Remittances, Residency Status and Financial Attachment on Housing Tenure for Mexican-Heritage Americans: Inferences from a New Survey Donald Bradley, Richard K. Green ∗∗ and Brian J. Surette ∗∗∗ Immigration has and will continue to alter the composition of housing de- mand in the United Sates. In this article, we analyze results from a new sur- vey of Mexican-heritage households to draw some inferences about tenure choice within that group. Some measures of attachment to the United States— residency status and the amount of money sent to relatives and friends in Mexico—suggest that, among Mexican immigrants, permanence is a key deter- minant of homeownership in the United States. More specifically, being a citizen increased the probability of ownership, whereas being undocumented reduces the probability. Surprisingly, after controlling for residency status, length of tenure in the United States does not predict tenure status, except that those who refused to report length of tenure were more likely to have higher tenure status. Those who sent remittances home to Mexico were less likely to become homeowners. Introduction Immigration has and will continue to alter the composition of America’s popu- lation and its demand for housing. Twenty-seven percent of household growth in the United States between 2000 and 2010 will come from legal and undocu- mented immigration (Masnick and Di 2003, Joint Center for Housing Studies 2005). Most immigration will come from Hispanic countries; most Hispanic immigration will be from Mexico (Chiswick and Miller 2003). Although a large number of detailed studies have focused on the impact of im- migration on the labor force, “relatively few studies have attempted to determine the impact of immigration on the housing market” (Borjas 2002, p. 2). Existing studies have largely been limited to enumerating the demographic demands Bradley Consulting, 12135 Preston Drive, Lusby, MD 20657 or Bradley381@ comcast.net. ∗∗ Department of Finance and George Washington Institute for Public Policy, The George Washington University, Washington, DC 20052 or [email protected]. ∗∗∗ Freddie Mac, McLean, VA 22102 or Brian [email protected]. C 2007 American Real Estate and Urban Economics Association

Upload: donald-bradley

Post on 20-Jul-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

2007 V35 4: pp. 451–478

REAL ESTATE

ECONOMICS

The Impacts of Remittances, ResidencyStatus and Financial Attachment onHousing Tenure for Mexican-HeritageAmericans: Inferences from a New SurveyDonald Bradley,∗ Richard K. Green∗∗ and Brian J. Surette∗∗∗

Immigration has and will continue to alter the composition of housing de-mand in the United Sates. In this article, we analyze results from a new sur-vey of Mexican-heritage households to draw some inferences about tenurechoice within that group. Some measures of attachment to the United States—residency status and the amount of money sent to relatives and friends inMexico—suggest that, among Mexican immigrants, permanence is a key deter-minant of homeownership in the United States. More specifically, being a citizenincreased the probability of ownership, whereas being undocumented reducesthe probability. Surprisingly, after controlling for residency status, length oftenure in the United States does not predict tenure status, except that thosewho refused to report length of tenure were more likely to have higher tenurestatus. Those who sent remittances home to Mexico were less likely to becomehomeowners.

Introduction

Immigration has and will continue to alter the composition of America’s popu-lation and its demand for housing. Twenty-seven percent of household growthin the United States between 2000 and 2010 will come from legal and undocu-mented immigration (Masnick and Di 2003, Joint Center for Housing Studies2005). Most immigration will come from Hispanic countries; most Hispanicimmigration will be from Mexico (Chiswick and Miller 2003).

Although a large number of detailed studies have focused on the impact of im-migration on the labor force, “relatively few studies have attempted to determinethe impact of immigration on the housing market” (Borjas 2002, p. 2). Existingstudies have largely been limited to enumerating the demographic demands

∗Bradley Consulting, 12135 Preston Drive, Lusby, MD 20657 or [email protected].

∗∗Department of Finance and George Washington Institute for Public Policy, The GeorgeWashington University, Washington, DC 20052 or [email protected].

∗∗∗Freddie Mac, McLean, VA 22102 or Brian [email protected].

C© 2007 American Real Estate and Urban Economics Association

452 Bradley, Green and Surette

for housing and comparing the homeownership rates in immigrant and nativepopulations.

The tenure choice literature includes the seminal papers by Hendershott andShilling (1979), Haurin, Hendershott and Kim (1994) and Laidler (1988). Thesepapers focus on the financial determinants of owning versus renting. Since then,the literature has examined in depth the financial, sociological and demographiccontributors to the tenure choice decision. This article owes much to that lit-erature, and its modeling strategy will be informed largely by that literature.However, this article uses a custom-designed survey that focuses on Mexicansand Mexican Americans. Although others (Painter, Gabriel and Myers 20011,Haurin and Rosenthal 2006) have focused on Latino Americans, they have beenforced to use data from the census and the American Housing Survey. Thesedata sets do not contain many of the variables needed to understand tenurechoice among people of Mexican heritage in America.

We address this deficiency using retrospective survey data that report howhouseholds go through the planning process necessary to buy a house. TheTomas Rivera Policy Institute conducted the survey. Buying a house is muchdifferent than buying a nondurable good and is even different from buyingother types of durables. A house purchase often requires months, and some-times years, of planning. This purchase often represents both the single largesteconomic expenditure and chief means of wealth accumulation of a household.Homeownership also is a major commitment to a geographic place and to alifestyle. For us to understand why households buy a house, we must under-stand what leads households to plan actively to buy a house. The survey usedhere reports responses that help us understand the home-buying process.

We focus on Mexican-heritage Americans for three reasons: (1) advocates haveworried about the homeownership gap between whites and minority groups;(2) Mexican-heritage households as well as other Hispanics are expected tobe the largest contributors to household growth over the next two decadesand (3) the impact of immigrant status, length of stay in the United States andcontinuing connections to home country are aspects of tenure choice that are notyet well understood. It is also fair to note that immigration has become an issueof considerable policy interest. We limit our analysis to Mexican Americansliving in the metropolitan areas of Los Angeles, Houston and Atlanta. Thesethree places represent distinct types of housing markets (Los Angeles has ahigh-cost housing market, Houston is an affordable housing market and Atlanta

1 Painter, Gabriel and Myer’s (2001) paper focuses on African Americans as well asLatino Americans, but it contains particularly important findings on the impact of nativityon Latino tenure choice.

Housing Tenure for Mexican-Heritage Americans 453

is an intermediate market) and they have large or growing Mexican-heritagepopulations.

We find that, after controlling for typical socioeconomic variables, continuingattachments to Mexico—as measured by regular remittance payments, resi-dency status and various other measures of integration into the U.S. finan-cial system—are statistically significant predictors of whether people becomehomeowners. Migrants to the United States who send remittances to Mexico areless likely to be homeowners than those who do not. Those who have a measur-able formal attachment to the financial sector are more likely to be homeowners.Finally, those who say they always pay their bills on time are more likely tobecome homeowners.

Residency status has a statistically significant influence on homeownershipamong the study population. This could imply: (1) renters disproportionatelyplan to return to Mexico and therefore indicate an insufficient planned tenurein the United States to amortize the fixed costs of becoming an owner or (2)those who are not permanent residents or citizens have limited access to tra-ditional mortgage financing. Our results suggest that some of the unexplaineddifferences in homeownership rates between non-Hispanic whites and Latinosreported by other researchers may be attributable to residency and integrationinto the financial system among both immigrants and native-born Latinos. Theseare characteristics that should be explored in greater depth in future research.

The remainder of the article is organized as follows. We begin with a brief reviewof the tenure choice literature and then discuss more explicitly our contributionto that literature. We follow with a description of the survey data that we use.Next, we present our estimation model and the results of that model. Finally,we draw some inferences and present some implications.

The Literature on Tenure Choice

The theory of housing tenure choice is straightforward: households have a util-ity function that they are seeking to maximize over housing and a numerairegood, subject to a budget constraint. The choice between owning and rentingis a function of the relative financial benefits of each, as well as relative nonpe-cuniary benefits. Owning involves transaction costs that renting does not. Suchcosts get amortized over time. Households that expect to live in a place fora long time are more likely to wish to become homeowners than householdsthat expect to move quickly. It is possible that transaction costs for owning aregreater for some groups than others; for example, it is possible that the transac-tion costs of owning for nondocumented aliens are higher than for permanentresidents.

454 Bradley, Green and Surette

Tenure is also influenced by institutional phenomena, such as mortgage under-writing standards (see, e.g., Linneman and Wachter 1989, LaFayette, Haurinand Hendershott 1995, Haurin, Hendershott and Wachter 1997). As such, theoptimization problem with respect to tenure choice is a problem subject toinstitutional constraints: the absence of funds for a down payment or of a well-established credit history could prevent households who would like to chooseowning over renting from buying.

The transaction costs of owning along with institutional constraints becomeparticularly important when one considers one variable that is a focus of thisarticle: remittances. Nearly half the households in our study group send remit-tances to Mexico.2 As we shall see, remittances are a significant predictor oftenure choice: those who remit are less likely to become owners. Remittanceshave at least one important impact on tenure choice, and perhaps more. Theycertainly make it more difficult for households to accumulate a down payment,thus postponing the ability to become an owner. It is possible that they placefinancial stress on the household, making it more difficult to establish a strongcredit history. At the same time, remittances might reflect something aboutunobserved characteristics of households: attachment to a native country andfinancial discipline. Attachment might be a proxy for the expected length oftenure, which in turn influences the relative attractiveness of owning. On theother hand, to the extent remittances reflect financial discipline, they may be aproxy for the ability of a household to make the sacrifices necessary to becomehomeowners.

In a world of representative agents, the costs of owning and renting shouldequilibrate, and everyone should be indifferent between the ownership forms.As Rosen’s (1974) innovative paper on hedonic pricing and housing demandpointed out, however, the representative agent model is not close to being suffi-ciently rich to characterize housing demand: heterogeneous households matchthemselves with heterogeneous housing. In the context of Rosen’s paper, thismeans that households with more children will value living area more thanhouseholds with fewer children and will therefore be willing to pay more forspace, ceteris paribus. Goodman (1988) applied this argument to tenure choice.Household idiosyncrasies can have important impacts on the relative costs ofowning and renting, as well as motivations for owning and renting.3 We maythink about this from the perspective of taxes, the difference between transitoryand permanent income, demographic characteristics and other characteristics

2 According to the Inter-American Development Bank (IADB) (2007), approximately70% of Latino immigrants in the United States at least occasionally send remittanceshome.3 Goodman (1988) notes that renters have only consumption motives, whereas ownershave investment and consumption motives.

Housing Tenure for Mexican-Heritage Americans 455

that might have an impact on expected length of stay in a house. Moreover, onecannot discuss tenure choice in the U.S. market without considering the impactof race.

We may expect income taxes to segment the housing market both within andacross tenure types (see Capozza, Green and Hendershott 1996 on capitalizationof tax preferences). The relative cost of owning versus renting is a function ofhow a house is financed and the tax code. Because imputed rent is not taxedand mortgage interest is deductible for determining taxable income, itemizersin high marginal tax brackets will outbid itemizers in lower tax brackets forhousing, and itemizers with mortgages will outbid nonitemizers with mortgages.Differences across states in marginal tax rates have allowed us to determine theimpact of tax status on tenure choice controlling for income (Mathur and Rosen1975, Rosen, Rosen and Holtz-Eakin 1984, Green and Vandell 1999).

Expected length of stay also is an important determinant of tenure status. Pur-chasing a house involves incurring fixed costs (title insurance, lawyer’s fees, animplicit share of the real estate brokerage commission and moving expenses)absent from renting. To the extent one expects to spend a long time in one place,these costs will be amortized over a long period. On the other hand, for peoplewho expect to stay in a dwelling for a short period, these costs are large anddiscourage homeownership.

Short stayers also are less likely to be concerned with one of the most impor-tant nonpecuniary benefits of owning—security of tenure. Therefore, estimatesof tenure choice equations in the past have found that some obvious proxiesfor expected length of stay—age (young people move more often than olderpeople) and marital status (singles are more mobile than married couples)—areimportant predictors of tenure (see Goodman 1988, Green 1996).

Goodman (1988) showed that current income is not by itself a strong predictorof housing demand or tenure. Households do a reasonably good job of sizing uptheir lifetime income prospects and consume housing accordingly. For instance,households with high levels of education expect high levels of permanent in-come, controlling for age and current income, and consequently consume morehousing.

The most recent literature shows that race remains an issue in the housingmarket in general and for tenure choice in particular. Ondrich, Ross and Yinger’s(2003) and Farley and Krysan’s (2002) recent works use matched pairs todemonstrate that discrimination remains in the housing market, although it is notas prevalent as it once was. Interestingly, the ultimate impact of housing marketdiscrimination is not altogether clear—although it is the case that discriminationremains in the owner market, DeRango (2001) shows that it is even more

456 Bradley, Green and Surette

common in the rental market. One could imagine that discrimination, perniciousas it is in any market, actually tends to make African Americans less likely torent.

On the other hand, we know that the ownership rates among Latino Americansand African Americans are lower than for white Americans; for African Amer-icans, this is true even after controlling for the commonly used socioeconomicvariables (income, education, age, marital status, tax status, etc.). This outcomearises in part from continuing housing market discrimination, but it could alsoreflect differences in wealth, credit history and other variables that are hard tocome by in the microdata typically available in tenure choice studies. We doknow that African Americans and Latinos have substantially less wealth thanwhites and are far less likely to have an account with a depository institutionthan whites (Bucks, Kennickell and Moore 2006).

Recent matched-pair work by Turner et al. (2002) suggests that Hispanics facediscrimination about as often as African Americans. Moreover, the patternsare remarkably similar to African Americans; renters are more likely to faceunfavorable treatment than owners. Yet the share of respondents in our surveywho said they were discriminated against was roughly half the share that Turneret al. (2002) found received unfavorable treatment. Perhaps this reflects a dif-ference in the method for finding discrimination: Turner et al. (2002) observedbehavior, whereas the Tomas Rivera Policy Institute just asked people whetherthey felt they had faced housing discrimination.

Finally, the literature has shown that immigrant status has a major impact onthe likelihood of homeownership. Painter, Gabriel and Myers (2001) show thatonce one controls for the usual variables plus nativity, citizenship status andlength of stay in the United States, there is no statistical difference in ownershiprates between Latinos and Anglo Americans.

Our survey allows us to determine whether immigrant status in and of itselfcreates a barrier to homeownership or whether those immigrants who do notbecome homeowners do so out of choice. Indeed, we have a good proxy for tiesto home country to investigate the correlation between tie to the home countryand the desire to become a homeowner. The proxy is remittances. Mexicansliving in the United States returned roughly $17 billion in 2006 (IADB 2007).4

The survey also provides us with data on the relationship between Mexicans andthe formal U.S. financial system, which allows us to estimate the impact thatfinancial institution relationships have on the propensity to purchase a home.

4 The IADB reported total remittances to Mexico of $23 billion; it also reported thatthree-fourths of remittances to Latin American countries come from the United States.

Housing Tenure for Mexican-Heritage Americans 457

Survey Design and Data

Although Latinos share a common linguistic background and many culturalsimilarities, the Latino population is fairly diverse, representing many differentcountries and regions.5 Among foreign- and native-born Latinos, Mexico is themost common country of ancestry or origin, accounting for 67% of all Latinosresiding in the United States. Puerto Ricans and Cubans are, respectively, thesecond and third largest identifiable heritage groups. This study focuses onLatinos of Mexican origin, because of their current preponderance as well astheir expected contribution to future U.S. population growth.

The Tomas Rivera Policy Institute collected data during the spring of 2003from a telephone survey of 1,400 Mexican-heritage renters and recent first-time homebuyers in Los Angeles, Houston and Atlanta. The sample is a ran-dom sample of households with a Latino surname with a phone number listedin the directory. Thus we exclude households without any phone, without alandline and those not listed in a directory. As such, the sample is surely notrepresentative. However, budget issues prevented field interviews.

On the other hand, we were able to capture households that are nonpermanentresidents. In fact, the random sample produced a higher share of immigrantsthan the 2002 Current Population Survey.6 The telephone survey enabled usto create a quantitative profile of prospective and recent homebuyers and toidentify specific barriers to homeownership.

Each of the three areas in our study represents a different type of housingmarket.7 Los Angeles has the largest Mexican-heritage population in the UnitedStates and is the second largest ethnic Mexican city in the world (it is exceededby only Mexico City). It includes the largest Mexican American barrio (EastLos Angeles) and is home to one in five of the nation’s Latinos of Mexicanorigin. This group constitutes 30% of the total residents and 75% of Latinoresidents in the greater Los Angeles area (Table 1). Los Angeles is also amongthe country’s most expensive housing markets.

5 Latino immigrants are by far the largest immigrant group residing in the United States.According to the 2000 Census, persons born in Mexico, Latin America and the Caribbeanconstitute 51% of all immigrants. Persons born in Asia and Europe constitute, respec-tively, 28% and 16% of immigrants.6 Tomas Rivera developed sample weights to account for nativity. We found, however,that the in-sample fit of regressions based on unweighted data was superior to regressionsbased on weighted data. We consequently report an unweighted regression. Results fromall model specifications are available upon request.7 We included only those counties with more than 20,000 residents of Mexican origin.It would have been logistically difficult and cost prohibitive to conduct case studyinterviews throughout such a large geographical area.

458 Bradley, Green and Surette

Table 1 � Population of study areas.

MexicanNon- Residents Percent

Total Latino Latino of Mexican of TotalResidents Residents Residents Heritage Population

Los Angeles CMSALos Angeles County 9,519,338 5,277,125 4,242,213 3,041,974 32.0%Orange County 2,846,289 1,970,710 875,579 712,496 25.0%San Bernardino County 1,709,434 1,040,047 669,387 532,186 31.1%Riverside County 1,545,387 985,812 559,575 463,465 30.0%Ventura County 753,197 501,463 251,734 211,925 28.1%Total 16,373,645 9,775,157 6,598,488 4,962,046 30.3%

Houston CMSAHarris County 3,400,578 2,280,827 1,119,751 814,693 24.0%Fort Bend County 354,452 279,581 74,871 51,447 14.5%Brazoria County 241,767 186,704 55,063 42,422 17.5%Galveston County 250,158 205,219 44,939 34,670 13.9%Montgomery County 293,768 256,618 37,150 27,845 9.5%Total 4,540,723 3,208,949 1,331,774 971,077 21.4%

Atlanta MSAGwinnett County 588,448 524,311 64,137 36,016 6.1%Fulton County 816,006 767,950 48,056 32,476 4.0%DeKalb County 665,865 613,323 52,542 30,741 4.6%Cobb County 607,751 560,787 46,964 29,833 4.9%Total 2,678,070 2,466,371 211,699 129,066 4.8%

The Houston Consolidated Metropolitan Statistical Area (CMSA) also has alarge Mexican American barrio and also attracts a large share of immigrantsfrom Mexico. It has large numbers of both recent immigrant and native-bornMexican-heritage households. Unlike Los Angeles, however, Houston has arelatively inexpensive housing market. In the Houston CMSA, three countieswere excluded from the study because of small Latino populations. Included inthe study are the five counties that contain 99% of the area’s Mexican population.

The Atlanta Metropolitan Statistical Area (MSA), although having a relativelysmall Latino and Mexican population (Table 1), has had rapid Latino and Mex-ican population growth in recent years. As an emerging port of entry for newimmigrants, it is an example of a homeownership market in a location thatis only recently accommodating immigrants. We surveyed Mexican Americanhouseholds in 4 of the 20 counties in the Atlanta MSA. Seventy-eight percent ofLatinos of Mexican origin residing in the Atlanta MSA live in the four selectedcounties.

Although all of these markets have substantial Mexican or Mexican Americanpopulations, they also differ in important ways. The Los Angeles housing marketis one of the most costly in the United States and consequently one of the mostdifficult for first-time homebuyers. On the other hand, housing in Houston

Housing Tenure for Mexican-Heritage Americans 459

and Atlanta is relatively affordable. According to the National Association ofRealtors, in 2003 the median house price in the Los Angeles MSA was $355,000,whereas in Houston and Atlanta, it was $130,000 and $157,500, respectively.

Another difference is that Los Angeles is an “old” Latino city, Atlanta is a “new”city and Houston is somewhere in between. By mixing three very different citieswith large Mexican American or Mexican populations, we hope to have surveydata that are close to representative of the nation’s Mexican-heritage population.

Samples for the telephone survey were drawn from directory-listed householdswith Spanish surnames within the selected counties. The survey containedquestions on homeownership expectations, experience and knowledge of therespondents. The telephone survey was conducted in the language of respon-dent’s choice—either English or Spanish—over a 3-week period, from May6 to 29, 2003. Interviews were conducted with adults of Mexican origin whowere the household’s decision maker on important financial decision such aswhat apartment to rent or house to buy.

We divided our respondents into three mutually exclusive groups based ontenure and their plans for buying a home in the United States. The three groupsare:8

Not Actively Looking for a Home—Renters who expressed a desire to owna house someday, but have not taken any specific steps for doing so.

Actively Looking for a Home—Renters who state they are in the process oflooking for a house to buy.

Recent Owner—Households who have bought a house within 2 years of theinterview.9

We first examine descriptive characteristics of these groups, and we then per-form a multivariate analysis of the survey data using an ordered logit regression.The ordered responses for the logit are: not actively looking for a home, ac-tively looking for a home and recent buyer of a home. We had 1,182 usableobservations in the data, of which 363 were“not actively looking for a home,”101 were “actively looking for a home” and 718 had “bought a home withinthe past 2 years.”

8 Some households indicated that they had no plans to buy a house in the United States.The funder of the survey, Freddie Mac, allowed surveyors to ask these households onlya limited set of questions.9 Their survey was designed to elicit information about those considering owning andthose who have recently become owners. Households that had owned for more than 2years were excluded from the survey.

460 Bradley, Green and Surette

Table 2 � Stages of homeownership of Mexican-heritage rental households.

Expected Ownership Timeline Los Angles Houston Atlanta Total

Not interested in ownership 11.2% 9.4% 19.3% 13.5%5 years or more 23.7% 20.0% 20.7% 21.4%Within 5 years 33.6% 31.4% 28.5% 31.1%Actively engaged in buying 8.1% 7.4% 6.4% 7.3%New homeowners 23.5% 31.8% 25.1% 26.7%

100.0% 100.0% 100.0% 100.0%

Past research points to four groups of determinants of housing tenure—demographic, financial, information and residency factors (Lopez 1986, Krivo1995, LaFayette, Haurin and Hendershott 1995, Gyourko and Linneman 1996,Haurin, Hendershott and Wachter 1996, 1997, Coulson 1999, Painter, Gabrieland Myers 2001). In our work, we have the following specific explanatoryvariables within each category:

� Demographic variables: age, marital status, number of children in thehousehold and area of residence.

� Financial variables: household income, financial conditions and formalattachments to the financial system (bank account, car loan, credit card).

� Information variables: familiarity with how to qualify for a mortgage,difficulty in finding an advisor.

� Residency and assimilation variables: nativity, length of stay in theUnited States, immigration status, formal relationship with U.S. finan-cial institutions, remittance to Mexico and discrimination in the housingmarket.

Characteristics of Sample Households10

More than 86% of the Mexican-heritage households surveyed are either own-ers or wish to become owners at some point. Many are recent immigrants andeconomically constrained. Table 2 shows that the time by which these house-holds think they will become owners varies. For over 21% of the households,homeownership is a distant dream that they think will take at least 5 years to ac-complish. Another 31% feel that they will be ready to attempt homeownershipwithin 5 years. Thirty-four percent are either actively engaged in buying a home(in contact with a real estate agent or in the process of mortgage financing) orhave recently bought a home (within 2 years of the interview).

10 Our sample is more heavily immigrant than we expected based on the immigrantshare of renters of Mexican heritage in our study areas reported in the 2002 CurrentPopulation Survey.

Housing Tenure for Mexican-Heritage Americans 461

Table 3 � Reasons for not wishing to be a homeowner.

Reasons Percent of Cases

I plan to return to Mexico 48.8%Insecure financial or legal status 15.1%I may not stay in this town 10.2%I just prefer to rent 9.6%Cannot afford to own 8.4%I do not want the responsibility 4.8%Life is too unpredictable 3.6%Not a good investment 2.4%I live alone 2.4%I do not want a mortgage 1.8%

Do not know/refused 12.2%

A small minority of households (13.5%) showed little interest in homeowner-ship. Table 3 displays the reasons offered by these households for not beinginterested in homeownership in the United States. The primary reason is theexpectation of returning to Mexico (48.8%). For these households, the UnitedStates is viewed as a temporary residence in which employment is available.Focus groups conducted by the Tomas Rivera Policy Institute, however, sug-gested that many of these households will end up staying in the United Statespermanently and that their U.S. homeownership expectations may change. Whatis significant is that few of these households could be classified as per se prefer-ring renting to homeownership. Ideally, we would include these households inthe regression analysis below. Unfortunately, the organization that funded thesurvey, Freddie Mac, was not interested in the characteristics of those who hadlittle or no interest in ownership, and so we do not have regressors for thesehouseholds.

Table 4 describes the demographic makeup of the sample population.11 TheMexican-born respondents represent 86.7% of the respondents. A higher shareof U.S.-born respondents than immigrants are engaged in homeownership ac-tions, whereas among immigrants, renters are less likely than owners to bepermanent residents and U.S. citizens (Figure 1). Not only are the renters inthese three cities overwhelmingly born in Mexico, a majority of them havecome to the United States within the last 10 years. Their foreign-born statusand recent entry into the United States is further reflected in both their lowincome and their relative youthfulness. Seventy-one percent of the households

11 This survey was a telephone survey based upon listed Spanish surnames. As a con-sequence, the selection bias should be for the more established, economically secure,native-born Hispanic households.

462 Bradley, Green and Surette

Table 4 � Demographic characteristics of Tomas Rivera Policy Institute sample

Nativity

U.S. Born Foreign Born Refused Total

Noninterested 4.8% 84.1% 11.1% 100.0%Ownership dreams 7.8% 89.5% 2.7% 100.0%Ownership actions 12.8% 83.4% 3.8% 100.0%

Total 9.1% 86.7% 4.2% 100.0%

Length of Residency

U.S. 5 Years 6–10 11–20 Over 20Born or Less Years Years Years Total

Noninterested 6.8% 34.1% 22.0% 22.7% 14.4% 100.0%Ownership dreams 8.5% 21.2% 24.8% 34.3% 11.3% 100.0%Ownership actions 14.6% 11.7% 16.9% 36.3% 20.5% 100.0%

Total 10.4% 19.3% 21.8% 33.7% 14.8% 100.0%

Total Household Income

Under $35,000– $50,000 Don’t Know/$35,000 $49,999 or More Subtotal Refused

Noninterested 83.5% 12.6% 3.9% 100.0% 45.5%Ownership dreams 79.7% 14.8% 5.5% 100.0% 18.9%Ownership actions 56.1% 27.6% 16.3% 100.0% 18.7%

Total 71.6% 19.2% 9.2% 100.0% 22.4%

Age

Under 25–34 35–44 45–54 55 or25 Years Years Years Years Older Total

Noninterested 9.4% 44.5% 25.0% 7.8% 13.3% 100.0%Ownership dreams 10.1% 45.9% 30.8% 10.0% 3.3% 100.0%Ownership actions 6.9% 40.2% 30.1% 15.3% 7.4% 100.0%

Total 8.9% 43.8% 29.9% 11.6% 5.7% 100.0%

in the survey have incomes of less than $35,000 per year; 52.7% are less than35 years old.

Annual remittances to relatives and friends still living in Mexico are an impor-tant feature of Mexican-heritage households’ family finances. Nearly half of allrespondents reported sending some remittances in the previous year. This canbe a significant drain on financial resources, further stretching comparativelylow incomes among those in the sample. Remittances can make it more chal-lenging to save for a down payment, and they may make monthly paymentsmore difficult than they otherwise appear. Indeed, we see that renters are more

Housing Tenure for Mexican-Heritage Americans 463

Figure 1 � Renters less likely to be U.S. citizens or permanent residents.

Preliminary draft: Do not quote without permission 2

7.76

29.52

7.48

46.53

8.71

12.82

33.19

17.65

25.63

10.71

0

10

20

30

40

50

US born US Citizen Permanent resident Something else Missing

PercentRenters Owners

Figure 2 � First-time homeowners less likely than rentersto send remittances to Mexico.

Preliminary draft: Do not quote without permission 3

51.7

5.443.13

29.39

10.34

64.5

3.78 4.41

25.53

3.78

0

10

20

30

40

50

60

70

None less than $500 $500-$1000 $1000-$5000 more than $5000

PercentRenters Owners

How much last year did you send to relatives and friends still living in Mexico?

Mean amount remitted among renters: $4434 among owners: $3115

likely than owners to send remittances to Mexico (Figure 2). Although this maybe a consequence of their more recent arrival to the United States, remittancesturn out to have substantial predictive power in the ordered logit model aftercontrolling for residency status in the United States and other variables.

Numerous past studies, including Haurin, Hendershott and Wachter (1996),have shown marital status to be an important predictor of tenure choice, partlybecause it often is a precursor to having children and partly because it increasesthe financial resources of the household. As shown in Figures 3 and 4, home-owners in our sample are more likely than renters to be married and are morelikely to have children.

464 Bradley, Green and Surette

Figure 3 � First-time homeowners are more likely than renters to be married.

Preliminary draft: Do not quote without permission 3

63.67

4.92.72 1.22

16.33

8.57

2.59

77.73

1.891.89 1.26

6.51 6.933.78

0

10

20

30

40

50

60

70

80

Married Separated Divorced Widowed Never Married Partnered Missing

PercentRenters Owners

Figure 4 � First-time homeowners have more children than renters.

Preliminary draft: Do not quote without permission 4

7.35

18.91

45.17

25.99

2.59

7.77

13.87

45.59

28.36

4.41

0

10

20

30

40

50

Missing No Children 1 or 2 Children 3 or 4 Children 5 or more Children

PercentRenters Owners

The survey results also find that respondents who reported a willingness tomove to a nearby neighborhood when buying a home are more likely thanothers to be first-time homeowners (Figure 5).12 The high cost of housing inseveral of the counties studied makes willingness to move especially impor-tant. Gabriel and Painter (2003) discussed how African Americans’ path tohomeownership in Southern California often involved moving away from veryexpensive Los Angeles County to less expensive Riverside and San BernardinoCounties.

12 For first-time homeowners, the question asks about what the respondent actually didwhen they bought their home.

Housing Tenure for Mexican-Heritage Americans 465

Figure 5 � First-time homeowners more mobile than renters.

Preliminary draft: Do not quote without permission 5

45.71

25.44

17.01

8.44

3.4

41.18

36.34

14.08

6.93

1.47

0

10

20

30

40

50

Current neighborhood Nearby neighborhood Nearby city Somewhere e lse Missing

PercentRenters Owners

[Just before you became a homeowner…] Which of the following best describes your intentions regarding [where you will buy a home]?

Econometric Model for the Transition to Owning

We model the homeownership decision as an ordered logit regression on ourcross-sectional sample of Latino renters and recent first-time homeowners. Thethree categories of response variables are: not engaged in homebuying, engagedin homebuying and recent homebuyer. We would model these three outcomesmost optimally in a multinomial logit framework; unfortunately, we do not haveenough households in the“engaged in homebuying” category to fully exploit therich set of explanatory variables. The ordered logit results give us the relativelikelihood that a household will be in a higher category relative to a lowercategory. In our context, the “lowest” category is renter, the next highest isactively pursuing buying a home and the highest is homeowner. The orderingseems natural.

We report results based on an unweighted ordered logit model that includes in-formation variables. We specified remittances in three ways: a dummy variableindicating whether a household actually sent a remittance home, the naturallog of remittances sent home and remittances in thousands of dollars. Thislast specification leads to an intuitive odds ratio interpretation, and so it is theone we report. Both remittances in thousands of dollars of remittances and adummy variable for remittances had coefficients that were significant at the95% confidence level. The natural log of remittances did not.

A category we include for each question is “missing response.” Householdsthat refuse to answer a particular question may well be systematically differentfrom the others, and we want to avoid losing degrees of freedom by discarding

466 Bradley, Green and Surette

such households from our data. With two exceptions—a question about lengthof tenure and a question about the ability to survive an interruption in income—we do not report these parameters so as to save space.13

Endogeneity could influence parameter estimates for three types of vari-ables: household composition variables, information variables and attachmentvariables.

Over the life cycle, it is entirely possible that tenure choice influences thedecisions to have children and to get married. One could imagine, for example,that the security of tenure created by owning could make households morecomfortable about having children. In one study (Green 1996), tenure choiceand marriage are empirically modeled as a joint-decision process. The equationwe report may therefore have biased coefficients for the household type and thepresence of children variables, as well as those variables that would be in a latentsimultaneous equation explaining household type and presence of children.14

On the other hand, work by Haurin, Hendershott and Wachter (1997) shows thatthe decision to change household type usually precedes the decision to changetenure type. This finding suggests that the biases arising from the endogeneityof household type and children may be small.

More troubling are the information variables and the attachment variables.One would expect those who have succeeded at becoming owners would bebetter informed about such things as credit and the home-buying process. Weattempted to ameliorate this problem by asking owners to remember how wellinformed they were before they purchased their home. This is a reason weonly interviewed owners who had purchased their house within the previous 2years—so that they could remember the transaction process reasonably well.As we shall see, the questions asking people about their level of knowledge hadno significant predictive power about tenure choice.

The attachment variables, especially remittances, present us with a more seriousproblem. Those who send remittances home to Mexico may be more attachedto their country and may therefore be less likely to buy a home.15 But those whoare owners may feel greater financial strain and may therefore be less able tosend remittances home. Because remittances and ownership are both functionsof financial and planned length of stay in the United States, it is hard to think

13 All results are available upon request. We include the missing response variable forthese questions, because they are the only responses that are statistically significant.14 See Goldberger (2001), pp. 369–372.15 They also may be more financially disciplined, and hence more likely to buy a home.

Housing Tenure for Mexican-Heritage Americans 467

Table 5 � Tests of statistical significance for variables explaining tenure status.

Variable DF Wald χ 2 Pr > χ 2

Income 5 19.3 0.002Age of household head 5 9.28 0.10Number of children 4 9.66 0.05Marital status 6 21.5 0.002Location 13 50.4 <0.0001Mobility 4 11.7 0.02Trustworthy advisor question 4 25.6 <0.0001Familiarity about mortgage process question 4 4.03 0.41Has a bank account 1 15.8 <0.0001Has a credit card 1 .50 0.48Has a car loan 1 6.12 0.01Pays bills question 1 7.4 0.007Survive income interruption question 5 11.12 0.05Help with payment question 1 1.60 0.21Discriminated against? 1 6.09 0.01Tenure in United States and nativity 5 10.5 0.06Residency status 3 8.30 0.04Remittance 1 3.96 0.05

Notes: There were 1,182 observations. Concordance between outcomes and predictionsof outcomes is 80.2%.

of an instrument that explains remittances that does not also explain tenurechoice. Given that we find in a reduced-form regression that households whosend remittances are significantly less likely to become owners, this issue isworthy of more data collection and study.

A Note on How Results are Reported

Because almost all of the explanatory variables are categorical, we have a verylarge number of coefficient estimates in the ordered logit model. Consequently,we will report our results in two stages. The results in Table 5 report on whetherthe various answers to a particular question are jointly significant. Only whenthe answers to a question are jointly significant at the 90% level of confidencedo we report individual coefficient estimates, standard errors and odds ratiosfor each individual answer in Table 6. We report point estimates of odd-ratiosand their 95% confidence band because these give more intuitive measuresfor understanding the impacts of the explanatory variables on tenure than thecoefficients and standard errors alone.

Because of its length, we divide Table 6 into four panels. The sections belowalign with the panels of Table 6.

468 Bradley, Green and Surette

Table 6 � Ordered logit models of tenure choice.

Panel 1: Income and Demographic Variables: Ordered LogitDependent Variable = 2 if Owner, 1 if Buying, 0 if not Buying

Coefficient Lower 95% Upper 95%(Standard Odds- Bound BoundError) Ratio Odds-Ratio Odds-Ratio

Intercept 1 −0.57(0.76)

Intercept 2 −1.06(0.76)

Relative to < $35kIncome $35,000–49,999 0.70∗∗∗ 2.01 1.36 2.98

(0.20)Income $50,000–64,999 0.93∗∗∗ 2.53 1.32 4.84

(0.33)Income $65,000–$79,999 0.73 2.07 0.67 6.39

(0.58)Income $80,000+ 1.45 4.25 1.08 16.7

(0.70)∗∗

Relative to never marriedPartner 0.96∗∗∗ 2.54 1.26 5.38

(0.37)Separated 0.03 0.90 0.37 2.83

(0.52)Widowed 1.10 1.37 0.70 12.9

(0.74)Divorced 0.60 1.49 0.64 5.25

(0.53)Married 1.04∗∗∗ 2.68 1.61 4.86

(0.29)

Relative to zero childrenOne or two children 0.21 1.23 0.79 1.95

(0.23)Three or four children 0.47∗ 1.60 0.97 2.62

(0.25)Five or more children 1.05∗∗ 2.85 1.20 6.72

(0.43)

Relative to > 55Age < 25 −1.03∗∗ 0.36 0.14 0.91

(0.48)Age 25–35 −1.09∗∗∗ 0.34 0.15 0.73

(0.40)Age 35–44 −1.06∗∗∗ 0.34 0.16 1.21

(0.40)Age 45–54 −0.61 0.54 0.24 1.06

(0.41)

Notes: ∗significant at 90% level; ∗∗significant at 95% level; ∗∗∗significant at 99% level.There were 1,182 observations.

Housing Tenure for Mexican-Heritage Americans 469

Table 6 � continued

Panel 2: Location and MobilityDependent Variable = 2 if Owner, 1 if Buying, 0 if not Buying

Coefficient Lower 95% Upper 95%(Standard Odds- Bound BoundError) Ratio Odds-Ratio Odds-Ratio

Relative to Harris County, TXTexas counties

Brazoria 0.74 1.52 0.54 8.12(0.69)

Ford Bend 0.30 1.09 0.29 4.07(0.62)

Galveston −0.06 1.05 0.18 6.16(0.94)

Montgomery 0.62 1.88 0.54 6.55(0.60)

Georgia countiesCobb 0.42 1.51 0.81 2.82

(0.31)Dekalb 0.10 1.10 0.58 2.07

(0.32)Fulton −0.55 0.58 0.29 1.16

(0.35)Gwinnet 0.93∗∗∗ 2.56 1.52 4.30

(0.27)California counties

Los Angeles −0.77∗∗∗ 0.46 0.30 0.72(0.23)

Orange −0.60∗∗ 0.55 0.30 1.00(0.31)

Riverside 0.82 2.29 0.79 6.65(0.54)

San Bernardino 0.42 1.53 0.68 3.42(0.41)

Ventura 0.39 1.48 0.62 3.52(0.44)

Relative to remain in neighborhoodMove to new neighborhood 0.46∗∗∗ 1.58 1.13 2.22

(0.17)Move to nearby town −0.24 0.79 0.50 1.23

(0.23)Move to far-away town −0.08 0.92 0.50 1.69

(0.31)

Notes: ∗ means significant at 90% level; ∗∗significant at 95% level; ∗∗∗significant at 99%level.

470 Bradley, Green and Surette

Table 6 � continued

Panel 3: Perceptions of Financial Position and Attachment to Financial InstitutionsDependent Variable = 2 if Owner, 1 if Buying, 0 if not Buying

Coefficient Lower 95% Upper 95%(Standard Odds- Bound BoundError) Ratio Odds-Ratio Odds-Ratio

Relative to Relative to Relative toneither neither neitherdifficult difficult difficultnor easy nor easy nor easy

Trustworthy advisor,very easy

−0.31 0.73 0.31 1.70(0.43)

Trustworthy advisor,easy

−0.28 0.75 0.33 1.72(0.42)

Trustworthy advisor,difficult

−1.05∗∗ 0.35 0.15 0.80(0.42)

Trustworthy advisor,very difficult

−1.05∗∗ 0.35 0.15 0.84(0.44)

Relative to no bank accountHas a bank account 0.76∗∗∗ 2.14 1.47 3.11

(0.19)

Relative to no car loanHas a car loan 0.42∗∗ 1.52 1.09 2.13

(0.17)

Relative to always pays bills strongly agreeDoes not strongly

agree with “Alwayspays bills on time”

−0.51∗∗∗ 0.60 0.41 0.87(0.19)

Relative to very unlikely tosurvive interrupted income

Survive interruptedincome very likely

−0.06 0.94 0.52 1.69(0.29)

Survive interruptedincome likely

0.003 1.00 0.57 1.76(0.29)

Survive interruptedincome neitherlikely nor unlikely

−0.69 0.50 0.10 2.48(0.78)

Survive interruptedincome unlikely

0.28 1.32 0.68 2.56(0.33)

Survive interruptedincome missing

1.36∗∗∗ 3.93 1.39 11.1(0.53)

Relative to no discriminationEncountered

housing marketdiscrimination?

−0.72∗∗ 0.48 0.27 0.86(0.29)

Notes: ∗ means significant at 90% level; ∗∗significant at 95% level; ∗∗∗significant at 99%level.

Housing Tenure for Mexican-Heritage Americans 471

Table 6 � continued

Panel 4: Nativity, Tenure and AttachmentDependent Variable = 2 if Owner, 1 if Buying, 0 if not Buying

Coefficient Lower 95% Upper 95%(Standard Odds- Bound BoundError) Ratio Odds-Ratio Odds-Ratio

Relative to permanent residentU.S. citizen 0.45∗ 1.57 0.97 2.57

(0.25)Neither citizen nor

permanent resident−0.34∗ 0.71 0.47 1.06(0.21)

Refuse to answer −0.15 0.86 0.42 1.77(0.36)

Relative to moved to United States since 1990Native to United States −0.39 0.68 0.34 1.34

(0.35)Not reported 0.81∗∗ 2.24 1.10 4.58

(0.36)Moved to United States

before 19700.72 2.06 0.69 6.20

(0.56)Moved to United States

between 1970 and 1980−0.15 0.86 0.47 1.58(0.31)

Moved to United Statesbetween 1980 and 1990

0.16 1.17 0.78 1.77(0.20)

Per 1,000Remittance (000) −0.04∗∗ 0.96 0.92 1.00

(0.02)

Notes: ∗ means significant at 90% level; ∗∗significant at 95% level; ∗∗∗significant at 99%level.

Income and Demographic Variables

We begin by discussing the variables that have always been found to drivetenure choice: income, age and family status (Table 6, panel 1). Not surpris-ingly, income is important to the explanation of homeownership. The estimatesindicate that a respondent with family income in the $35,000–$49,999 range istwice as likely to be in a higher tenure category than a respondent with incomeless than $35,000, holding constant other characteristics, and that a respondentin the $50,000–$64,999 range is 2.5 times as likely to be in a higher tenurecategory. The parameter estimate for the $65,000–$79,000 income category isalso always positive, but observations are relatively sparse, and so we have largestandard error around the parameter. Those in the highest income category aremore than four times as likely as those in the lowest to be at a higher level oftenure.

472 Bradley, Green and Surette

Marital status, presence of children and age also are important. Those who aremarried, have a partner or are widowed are significantly more likely to be at ahigher tenure level than singles. Those who are separated or divorced behavelike singles. Those who are married are 2.7 times more likely than singles tobe owners; those who have a partner are 2.5 times more likely, and those whoare widowed are three times more likely. The likelihood of ownership risesmonotonically in number of children, although the impact does not becomesignificant at the 10% confidence level until households have at least threechildren. Households with three or four children are 1.6 times more likely tobe owners than households without children; those with five or more children are2.9 times more likely. Consistent with other tenure choice studies, householdswith older heads are more likely to be owners than those with younger heads.Roughly speaking, those who were surveyed and were under the age of 45 wereabout 35% as likely to own as those over 55.

Location and Mobility

The single most important determinant of tenure status appears to be whererespondents live (Tables 5 and 6, panel 2). Relative to living in Harris County,Texas, families living in the more expensive Los Angeles or Orange Countiesin California are significantly and substantially less likely to be homeowners(46% as likely in Los Angeles, 55% as likely in Orange), whereas those livingin Gwinnet County in Georgia are 2.6 times more likely to become owners.Although our data show that incomes are modestly higher among the householdswithin our samples that live in the highest house price counties, relative houseprices in these counties are exceptionally expensive, meaning families are ingeneral more constrained in these counties.

To test our assumption that differences in ownership are a function of cost dif-ferences, we ran a simple correlation of the coefficients on the various countiesin the logit and median house price for the county based upon the 2000 Census.We found a correlation of –0.31 between the coefficients and median houseprices.

We find that willingness to move can attenuate the adverse effect of house prices.Our analysis shows that families that intend to move to a nearby neighborhoodwhen they are ready to buy a home—perhaps to a lower cost area—are 1.6 timesmore likely to become homeowners than families who do not intend to move.In contrast to Gabriel and Painter (2003), who found that African Americanhouseholds would move long distances within the Los Angeles metropolitanarea to purchase homes, the respondents in the survey who moved or planneda move to a different town were no more likely to plan to become or actuallybecome a homeowner than those who stayed in their own neighborhood.

Housing Tenure for Mexican-Heritage Americans 473

The Roles of Information, Financial Knowledge, Formal Attachmentto the Financial System and Discrimination

Discussion of integration into the U.S. financial system leads naturally to therole of information and financial knowledge. Survey respondents displayeda considerable lack of familiarity with the financial system—particularly ofthe mortgage process. This parallels Haurin and Morrow-Jones (2006), whofound that African Americans were not as well informed about the real estateand mortgage markets as white Americans and that this lack of informationhad an impact on the ability to become an owner. We thus sought to examinewhether financial knowledge and access to advice affected homeownership.This exercise is complicated by the cross-sectional nature of our sample. Recentsuccessful homebuyers have actually experienced the homeownership processand so are likely to be more familiar with it, leading to a spurious correlationbetween our measures of knowledge and homeownership outcomes. To mitigatethis, we asked recent first-time homebuyers to recall their knowledge as of thetime “you started looking for a home.” This retrospective phrasing may mitigatethe endogeneity problem.

The questions asked include: “How familiar are you with what it takes to qualifyfor a mortgage” and “How easy is for you to find someone you can trust to adviseyou on the home-buying process?” Surprisingly, answers to the question aboutfamiliarity with the mortgage process were not significant predictors of tenure(Table 5). The question about the ability to find a trusted advisor, however, wasa very significant predictor; those who said it was somewhat or very difficult tofind an advisor were only 35% as likely to be at a higher level of tenure thanthose who did not find it difficult or who were neutral (Table 6, panel 3).

Not surprisingly, the presence of bank accounts and car loans were significantpredictors of tenure status; those who had bank accounts were more than twiceas likely to be at a higher tenure status than those who did not, and those whohad car loans were 1.52 times as likely to be at a higher tenure status. Aftercontrolling for the presence of a bank account and a car loan, we could notidentify any impact of the presence of credit cards on tenure status (Table 5).This is likely a function of the close correlation among these three variables.With respect to these questions about attachment to the financial system, ownerswere asked whether they had such attachments before they became owners. Tosome extent, having a bank account is a necessary condition for ownership, but italso reflects attachment to the formal financial system. According to tabulationsfrom the 2004 Survey of Consumer Finances, slightly less than one-quarter ofLatino households are “unbanked” and therefore unattached.

We also have findings on self-assessed credit worthiness, ability to pay billswhen facing stress and external financial resources. The first question asked

474 Bradley, Green and Surette

respondents to strongly agree, somewhat agree, neither agree nor disagree,somewhat disagree and strongly disagree with the statement, “I always paymy bills on time.” Ninety-four percent of respondents placed themselves inone of the first two categories. Because of this, we modeled this variable asdichotomous: the households who “strongly agreed” and everyone else. Theself-separation is highly revealing; those who do not “strongly agree” are only60% as likely to be in a higher tenure category relative to those who “stronglyagree.” The next question asked, “If you or your immediate family faced a majorunexpected loss in income, how likely or unlikely is it that you could pay yourbills for the next three months.” Answers ranged from “very likely” to “veryunlikely.” As Table 5 shows, this question was statistically significant at the95% confidence level in explaining tenure. However, it was those householdsin the “missing or refused to answer” category that were statistically differentfrom the rest; these households were almost four times as likely as those whoanswered “very unlikely” to this question to be owners. A look at the raw datashows that 59% of the 32 respondents in the “missing” category were owners,as opposed to 30% of the entire sample. We are not sure what to make of thefact that those who have become owners are less likely to be willing to answera question about financial stress.

Finally, the survey asked, “Other than your spouse or domestic partner, wouldyou have someone else contribute to the down payments or monthly paymentsto your home” (or, for homeowners, do you have someone else contribute). Thiscoefficient for this variable was not statistically different from zero (Table 5).

Only 9% of the households in our sample felt that they had faced discriminationin the housing market, but it mattered to those that did. Those who felt that theyhad faced discrimination were only 48% as likely to be at a higher tenure statusthan those who did not.

Attachment to the United States

The financial wisdom of owning a house is strongly influenced by expectedlength of stay in one place. Mexicans planning short tenures in the UnitedStates have less incentive to own than those planning to stay.

Two indicators of expected length of stay are residency status and tenure inthe United States, and we use answers to questions about both as predictors oftenure. We find, not surprisingly, that residency status is a significant predic-tor of housing tenure. U.S. citizens are 1.57 times as likely to be in a highertenure category relative to permanent residents, whereas those who have nei-ther citizenship nor permanent residency status are 71% as likely to be in a

Housing Tenure for Mexican-Heritage Americans 475

higher category (Table 6, panel 4). Although the answers to the residency sta-tus questions are jointly significant at the 95% confidence level, the individualcoefficients are significant only at the 90% level. Nevertheless, the signs andmagnitudes of the coefficients are not surprising.

Once we control for residency status, length of tenure (which subsumes nativity)in the United States is a marginally significant predictor (Table 5). Strangely,the only individual coefficient that appears to influence tenure is for householdswho were not born in the United States and did not report their arrival time tothe United States. It is not surprising that we cannot identify the effects of bothresidency status and length of stay in the regressions (length of stay is corre-lated with residency status), but the apparent importance of the “nor reported”coefficient suggests that Mexican-heritage Americans may be sensitive aboutbeing asked the length-of-stay question.

The Role of Remittances

A large proportion of households in the survey (44%) reported that they rou-tinely send money to relatives and friends still living in Mexico. In light ofhow little income many of these households earn, we find this large percentageremarkable. This leads us to consider whether these remittances affect home-ownership. Ex ante, it is unclear what effect remittances will have on homeown-ership. By diverting a portion of their income to Mexico, remittances may slowthe respondents’ accumulation of funds for a down payment and may makemonthly payments more burdensome. On the other hand, the ability to sendremittances could positively affect homeownership if it proxies for unobservedwealth or financial discipline.

The model provides evidence that remittances play an important role in home-ownership decision among nonnative-born Mexican-heritage households. Thecoefficient on remittances is significant at the 95% confidence level, and it iseconomically important. For each $1,000 in remittances sent to Mexico, theprobability of owning falls by about 4%.

Participants in the Tomas Rivera Policy Institute Focus Groups indicted thatlarge portions of the remittances are used for constructing or improving housingin Mexico. If this is true, remittances may either lead Mexican-heritage familiesto postpone homeownership due to the added financial burden or signal anexpectation of a shorter length of stay in the United States, which would makehomeowning in the United States less financially attractive. In any event, thisissue is well worth pursuing as we attempt to better understand the determinantsof immigrant housing tenure.

476 Bradley, Green and Surette

Conclusion

The Mexican-heritage population in the United States is primarily an immigrantpopulation. Some measures of attachment to the United States—residency sta-tus and the amount of money sent to relatives and friends in Mexico—suggestthat, among Mexican immigrants, permanence is a key determinant of home-ownership in the United States. More specifically, being a citizen increases theprobability of ownership, whereas being undocumented reduces the probabil-ity. Surprisingly, after controlling for residency status, length of tenure in theUnited States does not predict tenure status, except that those who refused toreport length of tenure were more likely to have higher tenure status. Indicatorsof financial attachment, such as whether a household has a bank account ora credit card, are also important predictors, as is self-reported credit history.Those who can find a trusted advisor are more likely to achieve higher tenurestatus.

We find the remittance results particularly striking and interesting, in part be-cause so far as we know, no previous work has found a correlation betweenthe presence of remittances and tenure status, after controlling for other house-hold characteristics. Given how rapidly remittances are growing (according tothe IADB, remittances to Latin American countries tripled between 2001 and2006), and given their relative importance (according to the IADB, remittanceflows are greater than foreign direct investment flows in Latin America), theimpact of remittances on a large variety of economic outcomes is well worthstudying. There is also a relationship between remittances and attachment tofinancial institutions; those who are formally attached can send money homemore cheaply than those who are not. About 70% of remittances are wired witha typical transactions cost of 8% (IADB 2007). The transfer of funds from bankto bank is much less expensive.

The results in the article are based on one survey of 1,182 heads of Mexican-heritage households. Although this survey size is sufficient for drawing someinferences, the particular answers to some questions are sparse enough thatit is not possible to identify the influence of all answers on tenure status. Itis also the case that some variables (residency status and length of tenure inthe United States, for example) are highly correlated. Consequently, we shouldnot make too much of the fact that certain variables that we would expectto have an influence on tenure choice are not significant predictors of tenurechoice.

That said, we have evidence that variables not present in other data sets, such asthe ability to find a trusted advisor, self-assessed credit history and remittances,do add to the ability to predict tenure choice, after controlling for variables

Housing Tenure for Mexican-Heritage Americans 477

commonly used in tenure choice studies. As such, we see this article as ajustification for developing better and larger data sets to better understand thepaths to homeownership for immigrant households.

The views expressed in this article are ours alone and do not necessarily reflect theviews of our respective organizations. We thank Don Haurin, Pat Hendershott and ananonymous referee for detailed comments. Remaining errors are our own.

References

Borjas, G.J. 2002. Homeownership in the Immigrant Population. Working Paper No.02-01. Research Institute for Housing America.Bucks, B., A. Kennickell and K. Moore. 2006. Recent Changes in US Family Finances:Evidence from the 2001 and 2004 Survey of Consumer Finances. Federal ReserveBulletin: A1-A38. Available at: http://www.federalreserve.gov/PUBS/bulletin/2006/financesurvey.pdfCapozza, D.R., R.K. Green and P.H. Hendershott. 1996. Taxes, Mortgage Borrowing,and Residential Land Prices. H. Aaron and W. Gales, editors. Economic Effects ofFundamental Tax Reform. Washington, DC: Brookings Institution Press, 171–198.Chiswick, B. and P. Miller. 2003. Issue Paper on the Impact of Immigration for Housing.Issue Papers on Demographic Trends Important to Housing. Washington, DC: U.S.Department of Housing and Urban Development, 1–79.Coulson, N.E. 1999. Why Are Hispanic- and Asian-American Homeownership Ratesso Low? Immigration and Other Factors. Journal of Urban Economics 45: 209–227.DeRango, K. 2001. On the Determinants of U.S. Black-White Racial Segregation. UrbanStudies 38: 1521–1529.Farley, R. and M. Krysan. 2002. The Residential Preferences of Blacks: Do They ExplainPersistent Segregation? Social Forces 80: 937–980.Gabriel, S. and G. Painter. 2003. Paths to Homeownership: An Analysis of the ResidentialLocation and Homeownership Choices of Black Households in Los Angeles. The Journalof Real Estate Finance and Economics 27: 87–106.Goldberger, A. 2001. A Course in Econometrics. Cambridge, HA: Harvard UniversityPress.Goodman, A.C. 1988. An Econometric Model of Housing Price, Permanent Income,Tenure Choice, and Housing Demand. Journal of Urban Economics 23: 327–253.Green, R.K. 1996. Should the Stagnant Homeownership Rate Be a Source of Concern?Regional Science and Urban Economics 26: 337–368.Green, R.K. and K.D. Vandell. 1999. Giving Households Credit: How Changes in theU.S. Tax Code Could Promote Homeownership. Regional Science and Urban Economics29: 419–444.Gyourko, J. and P. Linneman. 1996. Analysis of the Changing Influences on TraditionalHouseholds’ Ownership Patterns. Journal of Urban Economics 39: 318–341.Haurin, D., P.H. Hendershott and D. Kim. 1994. Housing Decisions of American Youth.Journal of Urban Economics 35: 28–45.Haurin, D., P.H. Hendershott and S. Wachter. 1996. Wealth Accumulation and Hous-ing Choices of Young Households: An Exploratory Investigation. Journal of HousingResearch 7: 33–57.

478 Bradley, Green and Surette

——. 1997. Borrowing Constraints and the Tenure Choice of Young Households. Journalof Housing Research 8: 137–154.Haurin, D. and S. Rosenthal. 2006. The Influence of Household Formation on Home-ownership across Time and Race. Working Paper, The Ohio State University.Haurin, D.R. and H. Morrow-Jones. 2006. The Impact of Real Estate Market Knowledgeon Tenure Choice: A Comparison of Black and White Households. Housing PolicyDebates 17: 625–653.Hendershott, P.H. and J. Shilling. 1979. The Economics of Tenure Choice, 1955–79.C.F. Sirmans, editor. Research in Real Estate. Greenwich, CT: JAI, 105–133.Inter-American Development Bank. 2007. Migrant Remittances as a Development Tool.Available at: http://www.iadb.org/mif/remittances/index.cfmJoint Center for Housing Studies. 2005. State of the Nation’s Housing, 2005. Cambridge,MA: Harvard University.Krivo, L.J. 1995. Immigrant Characteristics and Hispanic-Anglo Housing Inequality.Demography 32: 599–615.LaFayette, W.C., D.R. Haurin and P.H. Hendershott. 1995. Endogenous MortgageChoice, Borrowing Constraints and the Tenure Decision. NBER Working Papers 5074.Laidler, D.E. 1988. Income Tax Incentives for Owner-Occupied Housing. NeoclassicalMicroeconomics. 2: 343–369.Linneman, P. and S. Wachter. 1989. The Impact of Borrowing Constraints on Home-ownership. AREUEA Journal 17: 389–402.Lopez Garcia, M.A. 1986. Sobre los Efectos de Diferentes Sistemas de Pensiones de laSeguridad Social. (With English summary.) Revista Espanola de Economia 3: 69–94.Masnick, G.S. and Z.X. Di. 2003. Projections of U.S. Households by Race/HispanicOrigin, Age, Family Type, and Tenure to 2020: A Sensitivity Analysis. Issue Paperson Demographic Trends Important to Housing. Washington, DC: U.S. Department ofHousing and Urban Development, 79–124.Mathur, V.K. and H. Rosen. 1975. Regional Employment Multiplier: A New Approach:Reply. Land Economics 51: 294–295.Ondrich, J., S. Ross and J. Yinger. 2003. Now You See It, Now You Don’t: Why Do RealEstate Agents Withhold Available Houses from Black Customers? Review of Economicsand Statistics 85: 854–873.Painter, G., S.A. Gabriel and D. Myers. 2001. Race, Immigrant Status, and HousingTenure Choice. Journal of Urban Economics 49: 150–167.Rosen, H.S., K.T. Rosen and D. Holtz-Eakin. 1984. Housing Tenure, Uncertainty, andTaxation. Review of Economics and Statistics 66: 405–416.Rosen, S. 1974. Hedonic Prices and Implicit Markets: Product Differentiation in PureCompetition. Journal of Political Economy 82: 34–55.Turner, M., S. Ross, G. Galster and J. Yinger. 2002. Discrimination in MetropolitanHousing Markets. Urban Institute Working Paper. Available at: http://www.urban.org/UploadedPDF/410821 Phase1 Report.pdf