The Impact of Macroeconomic Fluctuations on the Likelihood of African American Female Homeownership

Abstract

Homeownership, a primary component of household wealth, confers benefits not just in terms of the value of home equity itself, but also the tax benefits and inflation protection associated with owning property. The 1990s represented a booming time in the economy and record low interest rates which allowed homeownership to become more available to more people than ever. As a result, the US housing market experienced significant growth and home values began to soar in part due to rising incomes. However, this period of rapid expansion in the housing market was followed by a rapid decline, precipitated by the recession, as home values began to plummet and foreclosures steadily increased. This study examines the impact that recent macroeconomic fluctuations had on the likelihood of homeownership for African American women compared to their white counterparts. Using data from the American Housing Survey (1997–2009), this tenure spatial analysis used a logistic regression model to examine the odds in favor of homeownership during economic fluctuations along with taking into consideration other important determinants. The empirical results show that there are significant differences in the likelihood of homeownership between African American women and white females.

Keywords

Homeownership Foreclosure Macroeconomic fluctuations African-American

Introduction

In addition to being an integral part of the American dream (Leigh and Huff 2007a, b), research has consistently demonstrated that homeownership is an acceptable indicator of social integration and economic well being. Homeownership, being a primary component of household wealth, confers benefits to the homeowner not just in terms of the value of home equity itself, but also the tax benefits and inflation protection associated with acquiring a property (Kain and Quigley 1975). Along with the private economic benefits, there is recent evidence that homeownership may have a positive effect on the owner's behavior and on the children of homeowner (DiPasquale and Glaeser 1999). Additionally, a homeowner's neighborhood choice has been found to create positive neighborhood externalities (Estrada 2005).

The prevalence of homeownership however, is not universal. Across different demographic groups and even within different regions, homeownership rates vary widely. Many of these demographic inconsistencies are longstanding, resulting in significant differences in access to the social benefits of homeownership from community to community (Sykes 2005). For instance, the Department of Housing and Urban Development (n.d.) reported that despite increases in housing sales there were differences in homeownership rates for households of different ethnicities. Specifically, from the first quarter in 2007 to the first quarter in 2010 the African American homeownership rate dropped from 48% to 45.6% (2.4% decrease), while whites homeownership dropped by 0.8% from 75.3% to 74.5% (Jourdain-Earl 2011). In essence, African Americans are three times more likely than whites to lose their homes.

While there are a number of demographic and socioeconomic factors that have been identified as determinants of homeownership, research demonstrates that race differences in homeownership rates persist even after accounting for life-cycle, resource, preferences, and regional differences across groups (Myers and Chung 1996 and Alba and Logan 1992). For example, Leigh and Huff (2007a, b) reported that the homeownership gap between the lowest income blacks and lowest income whites is the greatest gap between black and white homeowners at any income level and this gap has increased from 22% to 25.8% between 1970 and 2001. Because neighborhood quality, including schools, may have disproportionate effects on children, racial gaps in homeownership and home values may tend to perpetuate themselves and to reinforce gaps in other social outcomes such as income, employment, and criminal behavior.

Past studies used the various twentieth century Integrated Public Use Microdata (IPUMS) to study the long-run evolution of racial differences in homeownership and housing values among male household heads. However, the proportion of black households headed by females has increased over time relative to the proportion of whites. Consequently research should be expanded to include female households to avoid giving a misleading portrait of racial change over time. In addition to the paucity of studies that examine the effect homeownership inequality among women. William Darity (2005) challenges economists to develop and / or adopt a theoretical framework to understand the persistence of ethnic and racial inequalities of well-being, income, and wealth. To this end, this study seeks to examine the impact that cyclical fluctuations in the economy have on the likelihood of homeownership for black and white women.

Disparities in Wealth and Homeownership

The large disparities in wealth between black and white American families have been the subject of much discussion and research. Given the strong historical correlation between homeownership and wealth, it is likely that the racial gap derives, at least partially, from the large observed differences in housing wealth (Charles and Hurst, 2001). But the racial differences in wealth are even more pronounced among those who have a significant share of their net worth in financial assets. While virtually every racial and minority group in the United States has experienced discrimination in terms of wealth accumulation, recent studies demonstrate that racial and ethnic groups are not and have not been equally disadvantaged (Shapiro 2001; Sherraden 2001; and Stern 2001). The explanations offered for the differences are as diverse and include the following theoretical perspectives: (1) the role of financial inheritance; (2) discrimination in mortgage lending; (3) human and social capital; and (4) contemporary social policies (Sykes 2005).

Studies have shown that non-whites are less likely to possess wealth than their white counterparts at every age and as such are less likely to either give or receive intergenerational transfers of wealth (Sykes 2005). Whites are more likely to receive intergenerational transfers and are also more likely to receive a significantly larger sum than that of their non-white counterparts (Sykes 2005). In addition to the lack of intergenerational transfers as a means to accumulate wealth, blacks have to grapple with higher costs of borrowing to purchase homes in an attempt to accumulate wealth. In 2006, 52.9% of African Americans and 47.3% of Hispanics who acquired home-purchase loans had subprime (high cost) loans. Conversely, 26.1% of their white counterparts acquired these high cost loans (Leigh and Huff 2007a, b). These disparities in mortgage lending and at other stages in the home buying process has lead to lower rates of homeownership among non-whites (Yinger 1999 and Sykes 2005) and to lower returns on the housing investment for non-whites relative to whites (Turner and Skidmore 1999 and Sykes 2005).

Racial differences in housing values and intergenerational wealth transfers have also been linked to racial differences in human capital and social capital (Sykes 2005). Whites have a preference of living, learning, and working amongst each other rather than with non-whites. This is one of the more prevailing theoretical perspectives used to explain racial and ethnic differences in human capital and social capital (Emerson and Yancey 2001 and Sykes 2005).

Racial and ethnic differences in wealth, especially in the area of homeownership and housing values, have been examined in the context of the effectiveness, or lack thereof, of contemporary social policies (Sykes 2005). The government program, Homeowners’ Loan Corporation (HOLC) established in the 1930's, was significant because it served as a model for other lending institutions which lead to a macro-level exclusion of many racial and ethnic groups from one of the largest single opportunities for mass accumulation, homeownership (Sykes 2005). Cottrell (2010) noted that the suburban boom era was made possible by government-backed mortgages. But those mortgages mostly went to people who were white because the federal government saw mortgages for black people as to risky to insure. Mortgage insurers would draw red lines around black communities, indicating areas where they wouldn't grant homeowners FHA mortgage insurance. It was this redlining, as it is called, that resulted in “predatory lending aimed at racially segregated minority neighborhoods” (Carey 2010). It has been speculated that the HOLC program institutionalized the practice of redlining which undervalued racially and ethnically mixed neighborhood (Massey & Denton, 1993; and Sykes 2005).

The importance of race as a factor in homeownership is by no means new. However, very little is known about how women belonging to particular racial and ethnic groups differ in terms of homeownership. Even less is known about what makes women in different racial and ethnic groups more susceptible to losing their houses. Sykes (2005) attributed this phenomenon to the fact that housing and homeownership studies often are studied within the confines of the marital unit where women are rarely the substantive population.

Past studies have shown that non-whites are less likely to own homes compared to whites. However the extent to which macroeconomic cyclical fluctuations have differential impact on different racial and ethnic groups has not been demonstrated. So the following questions arise: Are blacks more likely to lose their homes during a recession? Are the relative odds of homeownership gains greater for black during economic booms? While there are sociodemographic variables that predict the relative odds of homeownership, we do not know which racial or ethnic group is more at risk during a recession or which group experiences improved odds in homeownership as the economy improves.

The Model

The model used in this study is similar to the model used by Conrad and Alleyne (2011) to determine the likelihood of homeownership.¹ In this model, utility (U) is defined as a function of goods available throughout the metropolitan area (X), centrally located goods and services (Z) and housing services (H). Equation 1 displays the utility function where the amount of housing services influences utility through the factors described.

U=f {X,H (s, dist, mhi, mhv, asst, htype, edu, age, race, sex, emp),Z}

(1)

Conrad and Alleyne (2011) built on the models first put forth by Li (1977) and Gujarati (1995).

subject to:

Y=pXX + pZZ + pHH

(2)

where pX, pZ and pH would be the price of X, Z and H respectively.

Therefore, the amount of housing services consumed can be influenced by factors such as the number of rooms (s), median household income (mhi), the median house value (mhv), the receipt of public assistance (asst), the type of household (htype), the educational attainment of the head (edu), the age of the householder (age), the race of the householder (race), the sex of the householder (sex), and the presence of workers in the household (emp) and the distance between the area of economic development and the adjacent areas (dist), which influences their consumption of centrally located goods and services.

The probability of homeownership is calculated as the proportion of all families that are homeowners. Subsequently, the odds ratio as P_i/(1-P_i) where P_i is the probability of owning and 1-P_i is the probability of not owning, and is a function of income (X_i). Gujarati (1995) expanded the linear probability model, so that

P_{i} = E ({Y=1/X}_{i}) = β_{1} + β_{2} X_{i},

(3)

where P_i= E(Y = 1/X_i) is the conditional probability

So, the probability of homeownership as a cumulative logistic distribution function such that

P_{i} = 1/ (1 + e^{- Zi})

(4)

(1 - P_{i}) = 1/ (1 + e^{Zi}),

(5)

Where

Z_{i} = β_{i} + β_{2} X_{i} .

(6)

P_i is nonlinearly related to Z_i. Since P_i is nonlinear in X, the odds ratio is the log of (1-P_i) which is linearly related to Z_i and by default X_i. So, the odds ratio is given as P_i/(1– P_i), such that

P_{i} / (1 - P_{i}) = (1 + e^{Zi}) / (1 + e^{-Zi}) = e^{Z i}

(7)

In order for the odds ratio to be linear in the independent factors or determinants of homeownership and in the parameters, the natural log of the odds ratio (L) is taken.

Therefore the log of the odds ratio is

L= In (P_{i} / ({1-P}_{i})) =In (e^{Zi}) {=Z}_{i} {=β}_{1} {+β}_{2} X_{i} .

(8)

The goal is to assess the significance of the relationship that exists between the odds in favor of homeownership and those factors that influence ownership tenure. This analysis suggests that the odds in favor of homeownership is a function of income, which is a key determinant of homeownership; along with other factors such as the ratio of the median house value to the median monthly rent that assesses the cost of owning relative to renting, race, and the receipt of public assistance.

Estimation equation

Following from the model, the estimation equation is specified below:

In ((P/ ({1-P}_{i}) = \begin{array}{l} β_{1} + β_{2} (Year Built) + β_{3} (Age) + β_{4} (Education) + β_{5} (Black) + \\ β_{6} (Public Assistance) + β_{7} (Income) + β_{8} (Value of the Home) + u . \end{array}

(9)

Where ln(P/(1-P_i) is the natural log of the odds in favor of homeownership. The continuous variables in the equation are Year Built, Income and Value of the Home. The variable Year Built captures the year in which each house was built. The Income variable is the natural log of the head of household's estimated yearly income. The Value of the Home is the natural log of the estimated land and unit value of the home included in the study. The Age and Education variables represent the female's age and highest degree earned, respectively. The Public Assistance dichotomous variable awards a one to a woman who receive public assistance in the form of rent subsidy food stamps, etc and a zero if they receive no form of public assistance. It is predicted that the Public Assistance variable will result in a lower likelihood of homeownership.

Lastly the key variable for the equation is a dichotomous variable labeled Black. The Black variable will receive a one for all black female head of households and a zero for all white female head of households. The parameters β₂, β₃, β₄, β₅, β₆, β₇, and β₈ measure the change in the log of the odds ratio due to a unit change in each of the independent variables. This provides the elasticity of the likelihood of homeownership to changes in the independent variables.

Data

This study utilizes data for 1999, 2003 and 2009 from the National American Housing Survey (AHS). These years were selected to capture the effect of macroeconomic fluctuations on black and white female homeownership trends. The year 1999 represents a period of strong economic growth within the U.S. economy. During this time the unemployment and poverty rates were low by historical standards. The 1990s also represented a decade of higher wages and a historic drop in poverty rates among for blacks (Austin 2008). Data from 2003 will capture the impact of the 2001 recession, which began the halt of economic gains that blacks made over the 1990s (Austin 2008). Finally, the 2009 allows us to evaluate how the economic decline due to the foreclosure crisis affected black women in comparison to their white counterparts.

The AHS, conducted by the U.S. Census Bureau, is a large, national housing sample survey that provides detailed information of housing characteristics, including housing costs, size of housing unit, and housing and neighborhood quality. It also contains extensive economic and demographic information about the householders and their family members, which makes it an ideal dataset for analyzing the individual demand for housing characteristics, including the housing space. National data are collected biannually and the sample covers an average of 55,000 housing units.

The authors took the National American Housing Survey data for each year and extracted heads of household by gender and race, leaving a database encompassing approximately 18,000 observations of white and black female head of household economic and housing characteristics.

Table 1 displays the marital status of the respondents for the three study periods. For each period the statistics show that black women have the largest percentage of women in the status never been married. White women surveyed had the largest proportion of being married with a spouse present in the household. In 1999 and 2009, the second largest marital status group for both white and black females was divorced. In 2003, the second largest group for white females was a marital status of widowed.

Table 1

Marital status of black and white female head of household respondents

	1999		2003		2009
	Black	White	Black	White	Black	White
Married spouse present	16%	29%	16%	33%	19%	42%
Married spouse absent	2%	1%	3%	2%	2%	2%
Widowed	19%	26%	18%	23%	12%	16%
Divorced	21%	22%	21%	22%	19%	20%
Separated	9%	4%	8%	3%	8%	4%
Never married	33%	18%	34%	17%	40%	17%

Results

Table 2 displays the results of the logistic regressions for the years of 1999, 2003 and 2009. The Nagelkerke R shows an explanatory factor of 77 percentage points or higher for all three periods. The signs of the coefficients are consistent for each time period reported.

Table 2

Logistic regression results of homeownership among black women

	1999		2003		2009
	B	Exp B	B	Exp B	B	Exp B
Intercept	-37.334	0.000^***	-19.559	0.000^***	-19.559	0.000^***
Year built	0.016	1.016^***	0.007	1.007^***	0.005	1.005^***
Age	0.056	1.059^***	0.045	1.046^***	0.026	1.026^***
Education	0.047	1.048^***	0.024	1.024^***	0.040	1.041^***
Black	-0.383	0.682^***	-0.417	0.659^***	-0.425	0.654^***
Public assistance	-1.027	0.358^***	-0.956	0.384^***	-0.523	0.593^***
Income	0.121	1.129^***	0.482	1.619^***	0.568	1.764^***
Value of the home	0.011	1.012^***	0.011	1.011^***	0.009	1.009^***
“-2 log L ”	5189.761		883.053		12777.218
Nagelkerke R Square	0.792		0.772		0.801

*p<.05; **p<.01

***

p<.001

The majority of the variables in the study were statistically significant thereby making the results robust for analysis. The variable Year Built was positive and significant, indicating that newer homes had a higher likelihood of homeownership. The Public Assistance variable was found to be negative, indicating that individuals receiving public assistance are less likely to be homeowners. The Income variable was positive, indicating a direct relationship with homeownership. In addition, the Value of the home variable was also positive, signifying that based upon the data found in the American Housing Survey higher valued homes are more likely to be owner-occupied than renter-occupied dwellings. For the variable Age, the results indicate that as white and black women age, they have a higher likelihood of becoming a homeowner. The Education variable also shows that as a woman's education level increases, she is more likely to own a home. The key variable in this study is the variable labeled Black. The Black variable is a dichotomous variable that offers a one for black females and a zero for white females. The likelihood of homeownership for black females in comparison to white females has declined over the three time periods.

To better understand the odds of homeownership for black women in comparison to white women, the authors transformed the Exp B values into the following equation:

[E x p (B) - 1] * 100 = p e r c e n t c h a n g e i n o d d s

(10)

Table 4 displays the percentage change in the odds of homeownership for black women. The results indicate that in 1999 black women had a 32% lower probability of homeownership compared to their white counterparts. The results also show that the odds of homeownership for a black woman declines as the time periods progress. These findings indicate that black female homeownership compared to white female homeownership is not only less likely but fell by 3 percentage points between 1999 and 2009.

Table 3

Logistic regression results of homeownership white women

	1999		2003		2009
	B	Exp B	B	Exp B	B	Exp B
Intercept	-37.334	0.000^***	-23.014	0.000^***	-19.984	0.000^***
Year built	0.016	1.016^***	0.007	1.007^***	0.005	1.005^***
Age	0.056	1.059^***	0.045	1.046^***	0.026	1.026^***
Education	0.047	1.048^***	0.024	1.024^***	0.040	1.041^***
White	0.383	1.467^***	0.017	1.517^***	0.425	1.530^***
Public assistance	-1.027	0.358^***	-0.956	0.384^***	-0.523	0.593^***
Income	0.121	1.129^***	0.482	1.619^***	0.568	1.764^***
Value of the home	0.011	1.012^***	0.011	1.011^***	0.009	1.009^***
“-2 log L ”	5189.761		883.053		12777.218
Nagelkerke R Square	0.792		0.772		0.801

*p<.05; **p<.01;

***

p<.001

Table 4

Percentage change in the odds of homeownership: black and white women

	1999		2003		2009
	Black	White	Black	White	Black	White
Percentage change in odds of homeownership	-32%	47%	-34%	52%	-35%	53%

Table 3 depicts the results of the logistic regressions for the years 1999, 2003, and 2009. Once more, the variables are all statistically significant suggesting that the model is a good fit and useful for analysis. However, the key variable in this data is the variable labeled White. The data shows that homeownership for white females compared to that of black females increased over the three time periods. Additionally, the age and education variables were positive indicating that as women age and their education levels increase they have a higher likelihood of becoming a homeowner. A positive, direct relationship exists with the variables Income and Value of the home as it relates to homeownership. Public assistance for this data was negative, indicating that persons in need of public assistance are less likely to be homeowners.

Table 4 illustrates the probability of homeownership for white women in comparison to black women. In 1999, whites were 47% more likely than their black counterparts to be homeowners. By 2009, this rate had increased by 6 percentage points to 53%. This shows that whites were not only more likely than their black counterparts to be homeowners but their likelihood is increasing over the time periods.

The findings reveal that black women are less likely to own homes and more likely to lose their homes during macroeconomic fluctuations than white women. While black women homeownership rates were significantly lower than white women for each time period, the 1999 results for black female homeownership displayed the smallest margin of the three time periods. During the late 1990s, the unemployment rates were significantly low and the economy was experiencing an economic boom within the business cycle. The Bureau of Labor Statistics reports that the December 1999 unemployment rate was 4.0%, which were a 1.7 percentage points lower than 2003 unemployment rate and 5.9 percentage points lower than 2009 unemployment. For black women the unemployment rate increased by 2.4 percentage points from 1999 to 2003 while for white women the unemployment rate only increased by 1.3 percentage points from the same time period. These findings indicate that black women began to lose their jobs due to the early 2000s economic recession at a faster rate than white women indicating a causal effect for the decline in black female homeownership rates from 1999 to 2003.

From 2003 to 2009 the US economy's national homeownership rates went from 68.3% to 67.4%. Based upon data from the Department of Housing and Urban Development the national homeownership rates peaked in the year of 2004 then begin to decline. For whites the decline in homeownership from 2004 to 2009 was 1.6 percentage points but for blacks the homeownership rate declined by 3.8 percentage points during the same time period. Foreclosures begin to rise in 2005 from 1.5% to 6% due to the subprime adjusted rate mortgage crisis. According to the Pew research center, blacks entered into subprime mortgage loans at a significantly higher rate than their white counterparts (Kochhar et al. 2009). Leigh and Huff (2007a, b) found that many of the subprime loans taken out by blacks were not only for home purchases but for refinancing and home improvement. In 2006, blacks purchased 52.9% of the high priced loans while whites only purchased 26.1% of the same loans (Leigh and Huff 2007a, b). The results from the study displayed that black female homeownership rates declined significant more than white female rates. Based upon the macroeconomic analysis we can attribute the decline from the 2003 to 2009 time period to an increase in foreclosures among blacks due to their significant acquisition high priced risky loans. Future research is needed to explore whether black women were in fact more likely to undertake the high price loans compared to black men.

Conclusion

Essentially, the purpose of this study was to examine the relative odds of homeownership for African American women and their white counterparts during macroeconomic cyclical fluctuations. The findings revealed that black women are more likely to lose their homes than white women. Additionally, the likelihood of homeownership for white women has increased while the likelihood for black women has decreased over the time period. This indicates that the gap between the likelihood of white and black female home ownership has increased. While black women homeownership rates were significantly lower than white women for each time period, the 1999 results for black female homeownership displayed the smallest margin of the three time periods. Whether or not black women were more likely than their white counterparts to acquire subprime loans is an area of study that needs to be addressed in order to assess the reasons behind the plummeting homeownership rates of black women.

References

Alba

, Logan

Assimilation and stratification in the homeownership patterns of racial and ethnic groups.

Int Migrat Rev. 1992; 26(4): 1314–41.

Austin

Reversal of fortune: economic gains of the 1990s overturned for African-Americans from 2000–07. Washington DC: Economic Policy Institute Publication; 2008.

Carey

Race-based predatory lending fueled housing crisis; 2010. Retrieved March 2011 from http://www.capitolhillblue.com/node/33344.

Charles

, Hurst

The transition to home ownership black- white wealth gap; University of Michigan and University of Chicago Working Paper; 2001.

Conrad

, Alleyne

The spatial impact of revitalization on the likelihood of homeownership: a look at Washington DC.

Commun Dev: J Commun Dev Soc. 2011; 42(1): 34–55.

Cottrell

How segregation caused the housing crisis; 2010. Retrieved March 2011 from http://wwwchicagonow.com/blogs/chicago-muckrakers/2010/10/how-segregation-actually-caused-the-housing-crisis.html.

Darity

“Stratification Economics: The Role of Intergroup Inequality.”

Journal of Economics and Finance. 2005; 29(2): 144–53.

DiPasquale

, Glaeser

E.L.

Incentives and social capital: Are homeowners better citizens?

Journal of Urban Economics. 1999; 45(2): 354–384.

Emerson

, Yancey

Does race matter in residential segregation? Exploring the preferences of white Americans.

Am Sociol Rev. 2001; 66: 922–35.

10.

Estrada

Getting and keeping a home: a life table analysis of homeownership transitions in the US, 1969–2003. Los Angeles: International Sociological Association Research Committee on Social Stratification and Mobility; 2005.

11.

Gujarati

D.N.

Basic econometrics. 3rd ed. New York: McGraw-Hill; 1995.

12.

Jourdain-Earl

The foreclosure crisis and racial disparities in access to mortgage credit 2004–2009; 2011. http://www.compliancetech.com/files/The%20Foreclosure%20Crisis%20and%20Racial%20Disparities.pdf Retrieved March 2011.

13.

Kain

, Quigley

Housing market discrimination, homeownership and savings behavior.

Am Econ Rev. 1975; 62(3): 263–77.

14.

Kochhar

, Gonzaolez-Barrea

, Dockterman

Through boom or bust, minorities, immigrants and homeownership; 2009. Pew Research Center Publications. www.pewresearch.org. Retrieved February 2010.

15.

Leigh

, Huff

African Americans and homeownership: separate and unequal, 1940 to 2006. Washington, DC: The Joint Center for Political and Economic Studies; 2007a.

16.

Leigh

, Huff

African Americans and homeownership: the subprime lending experience, 1995 to 2007. Washington, DC: The Joint Center for Political and Economic Studies; 2007b.

17.

M.M.

A logit model of homeownership.

Econometrica. 1977; 45(5): 1081–97.

18.

Massey

D.S.

, Denton

N.A.

American apartheid: Segregation and the making of the underclass. Harvard University Press, Cambridge, Mass; 1993.

19.

Myers

, Chung

Racial differences in homeownership and home equity among pre-retirement aged households.

Gerontologist. 1996; 36: 356–60.

20.

Shapiro

The importance of assets. In: Shapiro

, Wolff

, editors. Assets for the poor. New York: Russell Sage; 2001.

21.

Sherraden

Asset-building policy and programs for the poor. In: Shapiro

, Wolff

, editors. Assets for the poor. New York: Russell Sage; 2001.

22.

Stern

The Un(credit) worthy poor. In: Shapiro

, Wolff

, editors. Assets for the poor. New York: Russell Sage; 2001.

23.

Sykes

A home of her own: an analysis of asset ownership for non-married black and white women.

Soc Sci J. 2005; 42: 273–84.

24.

Turner

M.A.

, Skidmore

Mortgage lending discrimination: a review of existing evidence. New York: The Urban Institute; 1999.

25.

Yinger

Sustaining the fair housing act.

Cityscape. 1999; 4: 93–106.