The Social Structure of Mortgage Discrimination

Abstract

In the decade leading up to the U.S. housing crisis, black and Latino borrowers disproportionately received high-cost, high-risk mortgages—a lending disparity well documented by prior quantitative studies. We analyze qualitative data from actors in the lending industry to identify the social structure though which this mortgage discrimination took place. Our data consist of 220 depositions, declarations, and related exhibits submitted by borrowers, loan originators, investment banks, and others in fair lending cases. Our analyses reveal specific mechanisms through which loan originators identified and gained the trust of black and Latino borrowers in order to place them into higher-cost, higher-risk loans than similarly situated white borrowers. Loan originators sought out lists of individuals already borrowing money to buy consumer goods in predominantly black and Latino neighborhoods to find potential borrowers, and exploited intermediaries within local social networks, such as community or religious leaders, to gain those borrowers’ trust.

Racial disparities in wealth are currently at their widest levels in decades. According to the Federal Reserve (2014), the wealth of the median white household stood at $141,900 in 2013, 13 times greater than that of the median black household ($11,000) and ten times that of the median Latino household ($13,700). These gaps in wealth by race are less a product of income disparities than of differential access to good homes in high quality neighborhoods, which in turn produces racial differences in homeownership rates, home values, and the accumulation of home equity, the principal source of wealth for most American families (Oliver and Shapiro, 1995). Historically, these disparities have been driven by multiple forms of discrimination, both public and private, including white mob violence against African-Americans trying to move into formerly all-white neighborhoods, municipal segregation ordinances prohibiting residence by blacks on predominantly white blocks, racially restrictive covenants barring the future sale of a property to non-whites. One of the many forms of neighborhood-based racial discrimination that contributed to current disparities is the legacy of redlining—the denial of credit to non-white residential areas (Rothstein 2017). More recently, the rise of new lending practices that specifically target nonwhite neighborhoods for risky, high cost financial services have further widened racial disparities in home equity and wealth (Hyra et al., 2013; Lipsitz and Oliver 2010; Renuart, 2004; Ross and Yinger, 2002; Squires, 1992).

Numerous quantitative studies have found that black and Latino borrowers over the past decade were frequently charged more for mortgage loans than similarly situated white borrowers (e.g. Bayer, Ferreira, and Ross, 2015; Been, Ellen, and Madar, 2009; Bocian et al., 2011; Courchane, 2007; Rugh, Albright, and Massey, 2015). Even after controlling for credit scores, loan to value ratios, the existence of subordinate liens, and housing and debt expenses relative to individual income, Bayer, Ferreira, and Ross (2015) found that black and Latino borrowers in all of the seven metropolitan areas they studied were significantly more likely to receive a high-cost loan than others. These results held for both low- and high-risk borrowers and regardless of age. Rugh, Albright, and Massey (2015) similarly found that black borrowers in Baltimore overall ended up paying five percent more for their mortgages than white borrowers, again controlling for credit scores and other background characteristics.

Although quantitative evidence consistently demonstrates disparities in loan costs, the specific individual-level mechanisms through which discriminatory lending occurs remains largely unexplored. We do not know, for example, how the employees of bank and non-bank lenders identified black and Latino borrowers to target for high cost loans and how they gained those borrowers’ trust. Here we follow Hedstrom and Swedberg’s (2004, 1) call for the formulation of midrange models to explicate the “social mechanisms that generate and explain observed associations between events.” Specifically, we draw upon qualitative data obtained from declarations and depositions offered by borrowers, mortgage originators, and bank employees in recent civil rights cases to identify the processes through which loan originators steered black and Latino borrowers into riskier, higher-cost loans than similarly situated white borrowers.

We begin by situating our analysis in the context of the historical link between residential segregation, mortgage lending, and the dispossession of minority wealth. We then describe the qualitative data upon which we base our analysis and then draw upon these data to identify three salient aspects of the social context of the lending that emerged with the spread of mortgage securitization during the 1980s and 1990s: (1) the vertical segmentation of organizations in the lending industry; (2) the horizontal segmentation between prime and sub-prime lenders; and (3) the targeting of neighborhoods by race and the use of social networks and trusted intermediaries within racially segregated communities to cultivate and exploit borrowers’ trust. We conclude by explaining how these features of the mortgage lending industry contributed to the racialized extraction of wealth from individuals and communities of color in the wake of the housing bust (see Rugh et al., 2015).

Racial Segregation and Reverse Redlining

In the course of the 20^th century, racial residential segregation was established and perpetuated through a combination of public and private actions that made racial residential segregation a characteristic feature of urban America by mid-century (Massey and Denton 1993). Together, the spatial segregation of African Americans through real estate discrimination and the systematic disinvestment in black neighborhoods through lending discrimination made it exceedingly difficult for African American families to acquire homes and accumulate wealth during the long, postwar economic boom (Killewald, 2013; Kuebler and Rugh, 2013; Lipsitz and Oliver, 2010; Sharp and Hall, 2014). Today many of the nation’s largest historically segregated black neighborhoods, such as those in the South Bronx and South Central Los Angeles, remain severely disadvantaged and have become majority-Latino, making Latinos also vulnerable to the adverse consequences of segregated spaces (Tienda and Fuentes, 2014; Rugh, 2015; Steil, De la Roca, and Ellen, 2015).

Federal legislative changes in the 1980s and 1990s such as the Depository Institutions Deregulation and Monetary Control Act (1980), the Alternative Mortgage Transaction Parity Act (1982), the Secondary Mortgage Market Enhancement Act (1984), the Financial Institutions Reform, Recovery and Enforcement Act (1989), and the Federal Housing Enterprises Safety and Soundness Act (1992) facilitated the growth of a secondary mortgage market and contributed to a shift from direct lending by banks that held loans in their own portfolios towards the origination of loans by brokers, bankers, or non-bank lenders who then sold the loans to investment firms that, in turn, bundled them together to back bonds for sale to investors, in a process known as securitization. For securitized loans, profits for loan originators depended not so much on long-term home values or the borrower’s likelihood of default but on short term revenues from points, fees, origination charges, and especially the size of the gap between the prevailing interest rate index and the rate paid by borrowers, commonly known as the “yield spread” (Jacobson, 2010; McLean and Nocera 2010; Unger, 2008).

In this context, predominantly black and Latino communities shifted from being objects of economic exclusion to targets for financial exploitation by intermediaries seeking to expand the pool of loans available for securitization (Squires, 2005). After being denied credit for years these communities represented an untapped market with established home equity and ample room for increased homeownership populated by borrowers with little financial experience (Botein, 2013). The persistence of high levels of racial segregation combined with structural changes in the lending industry thus facilitated the development of a structurally segmented mortgage market that offered separate and unequal loan products to disadvantaged borrowers located in black and Latino neighborhoods, particularly large numbers of high-cost, subprime loans (Apgar and Calder, 2005; Hwang, Hankinson, and Brown, 2015; Engel and McCoy, 2002; Hyra et al., 2013; Rugh and Massey, 2010; Steil, 2011; Williams, Nesiba, and McConnell, 2005).

Roughly two-thirds of subprime loans in the early 2000s were made not to new home purchasers but to individuals who already owned their homes and were refinancing them (Mayer, Pence, and Sherlund 2009). During the 1990s up to one out of every three borrowers given subprime or high-cost loans were, in fact, eligible for prime loans (Mahoney and Zorn, 1996). By 2006, 62 percent of subprime borrowers actually qualified for prime loans (Brooks and Simon, 2007). Of home purchase loans made in 2006, roughly one out of every two loans made to African American (53 percent) and Latino (46 percent) borrowers were high-cost, compared to fewer than one out of five loans made to white borrowers (18 percent) (Been, Ellen, and Madar 2009). Similarly, for refinance loans made in 2006, 52 percent of black refinance borrowers and 39 percent of Latino refinance borrowers received high-cost loans compared to only 26 percent of white borrowers (Been, Ellen, and Madar 2009). Even after controlling for available loan and household characteristics, such as income, black home purchase borrowers were more than twice as likely to receive a subprime loan as white borrowers and the likelihood of receiving a subprime loan actually increased with household income, calling into question claims that subprime loans were given to riskier borrowers (Faber 2013). The systematic channeling of otherwise qualified minority borrowers into subprime mortgages carrying high costs and risks was one form of reverse redlining (Squires, 2005).¹

Data and Methods

Although quantitative studies can estimate the size of racial disparities in high-cost, high risk lending, qualitative analysis is required to identify the ways in which loan originators targeted black and Latino borrowers and succeeded in convincing them to enter into disadvantageous contracts that put their homes and wealth at risk. Policies that led to a disproportionately negative impact on non-white borrowers could have been put in place intentionally to harm borrowers on the basis of race, they could have been implemented without racial animus but nevertheless with knowledge that they would have a disproportionate negative effect on non-white borrowers, or they could have been established without any knowledge that they would have a different impact on white borrowers as compared to non-white borrowers. Regardless of the intent or knowledge, there is a pressing need to understand how policies with such consistently discriminatory effects were put into place across multiple actors in the home finance industry.

Our data come from depositions and declarations made by borrowers, mortgage brokers, loan officers, credit managers, due diligence employees, investment bankers, and others involved in subprime lending and securitization during the housing boom of the 2000s. We obtained these statements from documents publicly filed in cases brought to federal court alleging violations of fair housing and fair lending laws.

We began by seeking to identify the universe of cases alleging violations of the Fair Housing Act and the Equal Credit Opportunity Act over the past decade. Table 1 presents a list of lawsuits filed since 2006 against financial institutions for alleged reverse redlining. Suits have been filed in a range of regions against a wide variety of institutions, ranging from small local banks, such as Southport Bank in Kenosha, WI, to large financial corporations headquartered in financial centers, such as Bank of America in Charlotte, Wells Fargo in San Francisco, and J.P Morgan Chase in New York. Over 60 percent of these cases resulted in a settlement, however, leaving little in the public record aside from the alleged violations and the terms of the consent decree or settlement. Of the remaining cases, 18 percent were dismissed early in the litigation process and likewise left a sparse public record. One case went to trial, resulting in a guilty verdict against the defendants, and just over 20 percent of the cases were still ongoing at this writing.

Table 1.

Fair lending cases filed by plaintiffs alleging reverse redlining discrimination in violation of FHA and ECOA 2007–2017.

Plaintiff	Defendant	Court	Docket #
NAACP et al.	Ameriquest Mortgage Company et al.	C.D. Cal.	07-CV-00794
Zamora et al.	Wachovia et al.	N.D. Cal.	07-CV-04603
Payares et al.	Chase Bank U.S.A. et al.	C.D. Cal.	07-CV-05540
Miller	Countrywide Bank	D. Mass.	07-CV-11275
Allen et al.	Decision One Mortgage Company	D. Mass.	07-CV-11669
Alleyne et al.	Flagstar Bank, F.S.B. et al.	D. Mass.	07-CV-12128
Taylor et al.	Accredited Home Lenders, Inc. et al.	S.D. Cal.	07-CV-1732
Hoffman et al.	Option One Mortgage Corp.	N.D. Ill.	07-CV-4916
City of Baltimore	Wells Fargo Bank, NA	D. Md.	08-CV-00062
Ramirez et al.	GreenPoint Mortgage Funding, Inc.	N.D. Cal.	08-CV-00369
United States	First Lowndes Bank	M.D. Ala.	08-CV-00798
Guerra et al.	GMAC LLC, et al.	E.D. Pa.	08-CV-01297
Puello et al.	Citifinancial Services et al.	D. Mass.	08-CV-101417
Barrett et al.	H&R Block	D. Mass.	08-CV-10157
Steele et al.	GE MoneyBank et al.	N.D. Ill.	08-CV-1880
Garcia, Jenkins, Miller et al.	Countrywide Financial Corporation	W.D. KY	08-CV-448
City of Birmingham	Argent Mortgage Co., LLC, et al.	Alabama	08-CV-903691
United States	First United Security Bank	S.D. Ala.	09-CV-00644
Ventura et al.	Wells Fargo Bank, NA	N.D. Cal.	09-CV-01376
City of Memphis	Wells Fargo Bank, NA	W.D. Tenn.	09-CV-02857
United States	AIG F.S.B. & Wilmington Fin., Inc.	D. Del.	10-CV-00178
United States	PrimeLending	N.D. Tex.	10-CV-02494
Massachusetts	Countrywide Financial Corporation	Massachusetts	10-CV-1169
United States	Nixon State Bank	W.D. Tex.	11-CV-00488
United States	C&F Mortgage Corporation	E.D. Va.	11-CV-00653
United States	Countrywide Financial Corporation	C.D. Cal.	11-CV-10540
United States	SunTrust Mortgage, Inc.	E.D. Va.	12-CV-00397
United States	Wells Fargo Bank, NA	D.D.C.	12-CV-01150
United States	GFI Mortgage Bankers, Inc.	S.D.N.Y.	12-CV-02502
De Kalb County et al.	HSBC	N.D. Ga	12-CV-03640
United States	Luther Burbank Savings Bank	C.D. Cal.	12-CV-07809
Adkins et al.	Morgan Stanley	S.D.N.Y.	12-CV-7667
Consumer Fin. Prot. Bureau	National City Bank	W.D.Pa.	13-CV-01817
United States	Plaza Home Mortgage, Inc.	S.D. Cal.	13-CV-02327
United States	Southport Bank	E.D. Wis	13-CV-1086
United States	Chevy Chase Bank, F.S.B.	E.D. Va.	13-CV-1214
City of Miami	Bank of America et al.	S.D. Fla	13-CV-24506
City of Miami	Wells Fargo Bank, NA	S.D. Fla	13-CV-24508
City of Miami	Citigroup	S.D. Fla	13-CV-24510
City of Los Angeles	Wells Fargo Bank, NA	C.D. Cal.	13-CV-9007
City of Los Angeles	Citigroup	C.D. Cal.	13-CV-9009
City of Los Angeles	Bank of America	C.D. Cal.	13-CV-9049
City of Los Angeles	J.P. Morgan Chase	C.D. Cal.	14-CV-04168
Cook County	HSBC	N.D. Ill.	14-CV-022031
Cook County	Bank of America	N.D. Ill.	14-CV-02280
Cook County	Wells Fargo Bank, NA	N.D. Ill.	14-CV-9548
City of Miami Gardens	Wells Fargo Bank, NA	S.D. Fla	14-CV-22203
City of Oakland	Wells Fargo	N.D. Cal.	15-CV-04321
Cobb County et al.	Bank of America, N.A. et al.	N.D. Ga.	15-CV-04081
United States	Provident Funding Associates	N.D. Cal.	15-CV-02373
United States	Sage Bank	D. Mass.	15-CV-13969
United States	J.P. Morgan Chase	S.D.N.Y.	17-CV-00347
City of Philadelphia	Wells Fargo Bank, NA	E.D.Pa	17-CV-02203

Of the cases that did not settle and which survived a preliminary motion to dismiss to produce fuller public records, we selected four cases using an inductive strategy that deliberately sampled on the dependent variable. We chose cases in which discrimination was well documented, thereby enabling us more readily to identify the processes by which the lending discrimination was effected during the housing boom. In making our selections, we also endeavored to capture geographic and social variation among the parties and others who offered sworn testimony in the various cases.

By drawing upon multiple cases involving different originators, lenders, and financial agents, we are able to undertake a small-N comparison analysis that combines “the interpretive and narrative subtlety” of archival analysis with the “analytic strength that echoes standard causal analysis” (Abbott 2004, 58). The fact that all four cases survived a motion to dismiss means that the facts of the case were sufficient to “allow[] the court to draw the reasonable inference that the defendant is liable for the misconduct alleged” and “to infer more than the mere possibility of misconduct” (Ashcroft v. Iqbal, 556 U.S. 662, 678 2009). Our sample may therefore be skewed toward more severe cases of discrimination. In this analysis, however, we seek not to determine whether discrimination took place or how widespread it was, but to identify the institutionalized social processes and structured practices through which the discrimination occurred. Using a legal case as the unit of analysis creates the opportunity to look at the relationships among multiple actors in the mortgage process and provides a window into how discrimination took place in different contexts, from a cluster of real estate professionals working on a small scale flipping homes in Brooklyn to a large national lender in Memphis to a Manhattan investment bank securitizing subprime loans from Detroit made by a closely linked mortgage originator. As with any case study, the generalizability of the findings is limited, but the focus here is on identifying the processes through which discrimination took place, as quantitative studies have already established the pervasiveness of the discrimination.

The first of these cases, Barkley v. Olympia Mortgage, was brought in federal court by eight home-buyers in Brooklyn, New York who alleged that a real estate investor purchased properties, performed cosmetic repair work, and conspired with mortgage lenders, appraisers, and attorneys to target black and Latino first-time home-buyers in order to resell the homes using high-cost loans at far more than the properties’ true values. After a three-week trial, the jury found the real estate investor and mortgage companies guilty of fraud, conspiracy to commit fraud, and deceptive practices. The verdict was upheld on appeal by the United States Court of Appeals for the Second Circuit, yielding more than 25 depositions by key actors in the trial, with some interviews lasting for eight hours or more.

The second and third cases were brought against Wells Fargo Bank by the City of Baltimore and the City of Memphis (in partnership with Shelby County, Tennessee). Both cases alleged that Wells Fargo intentionally targeted minority communities and used discriminatory and deceptive methods to steer minority customers into high-cost loans, resulting in extraordinarily high rates of foreclosure that caused local governments to lose property tax revenue and spend additional resources to maintain vacant homes. The federal district court granted Wells Fargo’s motions to dismiss Baltimore’s complaint twice, because, in the court’s view, the city had not sufficiently established a “causal connection between the widespread damages it sought and Wells Fargo’s lending practices” (Baltimore v. Wells Fargo, 2011 WL 1557759 (D. Md. 2011)). Baltimore then focused its claims on instances in which “Wells Fargo deliberately steered African–American borrowers who qualified for prime loans into more onerous subprime loans” and on instances in which Wells Fargo targeted borrowers who already owned their homes with affordable mortgages or no mortgages at all but steered them into high-cost refinance or home equity loans. With these allegations focused on foreclosures directly caused by the discriminatory lending, the federal district court denied Wells Fargo’s motions to dismiss and the bank ultimately chose to settle both cases, agreeing to pay millions of dollars to borrowers who were overcharged and to the municipalities that brought suit. The public record in these cases includes multiple declarations made by employees from different regions and various positions at Wells Fargo and associated commercial entities.²

The fourth case, Adkins et al. v. Morgan Stanley, was brought by black residents of Detroit who alleged that they were fraudulently steered into high-cost loans by New Century Financial Corporation, the second largest originator of subprime mortgages in the United States in 2006. The plaintiffs argued that Morgan Stanley, an investment bank that acted as a warehouse lender for New Century, encouraged that firm to issue mortgages in violation fair lending laws in order to generate large numbers of high-cost mortgages that would realize greater profits from securitization. The plaintiffs also alleged that borrowers were more likely to receive such high cost, high risk loans if they were African American or lived in predominantly black neighborhoods, yielding litigation that is still ongoing.

Together, the hundreds of pages of depositions and exhibits provide a rich source of qualitative data about practices used within the mortgage industry during the housing boom, and they offer a unique opportunity to analyze the social processes behind the quantitative results developed to date. To conduct our content analysis, we randomly selected a subsample of the 220 relevant evidentiary filings yielding a dataset of 884 pages of deposition transcripts, declarations, and exhibits from individuals in different roles in the lending industry who were knowledgeable about the alleged discriminatory practices. The majority of the depositions and declarations come from employees of the defendants, such as loan officers or investment bankers, but a substantial portion come from borrowers, appraisers, closing attorneys, and other actors in the field.

To analyze these data, we began by coding the depositions, declarations, and exhibits into three categories that we derived empirically and theoretically from an initial analysis of the data and from a broader understanding of the context of subprime lending: vertical segmentation, horizontal segmentation, and institutionalized racial targeting. Data coding and analysis were conducted independently by two of the authors, with each coder labeling the data with “vertical segmentation,” “horizontal segmentation,” or “racial targeting.” Intercoder reliability was cross-checked utilizing the ReCal2 Online Utility to compute Krippendorff’s α on a 10 percent sample of the 220 texts yielding an inter-rater reliability coefficient of 0.86 (Freelon, 2013; Hayes and Krippendorff, 2007). From this 10 percent sample, we then extracted texts coded as reflecting vertical segmentation, horizontal segmentation, and racial targeting, and assembled the relevant passages for analysis and to reconstruct the structural logic of the reported discrimination.

In our analysis we seek to reconstruct the social organization of high-cost, high risk discriminatory lending as revealed by sworn statements of actors directly involved in the lending process. We explain how the securitization of mortgages institutionalized a vertical segmentation of entities separating the investors purchasing bundles of loans and borrowers seeking credit and how this segmentation contributed to the perception among upstream financial actors (such as securitization specialists and bond traders) that they had no responsibility for discriminatory or irresponsible actions of those downstream in the lending process (such as mortgage brokers) much less the borrowers themselves. We show how the mortgage lending industry also created a structure of horizontal segmentation in which prime and subprime lending operations were separated from one another organizationally. This organization made it difficult or impossible to transfer clients from subprime to prime lending officers and served as a one-way street channeling prime-eligible customers into higher priced and riskier subprime loans. Finally, we elucidate how loan originators within this segmented structure exploited racial segregation in order to target neighborhoods that had historically been denied credit for exploitative, high cost loans (Engel and McCoy, 2011; Immergluck, 2009). Taking advantage of residential segregation, originators developed specialized strategies and marketing materials aimed at identifying black and Latino borrowers as subprime lending marks. Prior research suggests that Latino immigrant households seeking to purchase a home were more susceptible, via Spanish-speaking intermediaries, to high-risk loans and subsequent foreclosures than were white, black, or Asian households (Allen 2011; Rugh 2015; Rugh and Hall 2016). Black and Latino borrowers were also vulnerable to targeted subprime refinance lending, especially older borrowers and those in older homes; overall such targeted refinance lending tended to affect those in in racially isolated African American communities the most (Botein 2013; Hyra et al. 2013). To gain the trust of those borrowers, originators worked through local social structures and interpersonal networks to enlist trusted intermediaries such as religious leaders, small business owners, and personnel in community-based organizations.

THE SOCIAL STRUCTURE OF HIGH-COST LENDING

As has been documented (e.g. Engel and McCoy, 2011; Immergluck, 2009; Newman, 2009), during the housing boom incentive structures within the mortgage finance industry were well aligned to guarantee short-term profits for the investment banks that securitized the loans and the actors who originated them, but not to assure the loans’ safety and soundness. Profits for loan originators and financiers depended largely on transaction fees and most critically on the size of the gap between the interest rate prevailing at the time of origination and that paid by borrowers.

The depositions we reviewed indicate that, unsurprisingly, this incentive structure led investment bank employees to encourage mortgage originators to generate ever more loans with high or adjustable interest rates (Kaplan, 2014a; Vanacker, 2014). Specifically, financial firms specializing in securitization sought to place the risk of future interest rises onto borrowers by steering them into adjustable rate mortgages, thereby guaranteeing investors a stable rate of return over the U.S. Treasury rate while placing individual borrowers at risk of financial stress because of increased payments (Shapiro, 2014; Vanacker, 2014).

When faced with borrowers who were unlikely to be able to repay a loan, some loan officers were encouraged by supervisors to find ways to lower the initial monthly payment through innovations such as hybrid adjustable rate mortgages. These loan packages made use of temporary low teaser rates, interest only mortgages, or mortgages with 40 year payment terms that ballooned in later years. Lenders then evaluated the borrower’s ability to repay based on the initial payment only, without taking into account the inevitable financial shock that would come when the teaser rate expired, interest payments came into effect, or balloon payments came due (Missal, 2008). Instead, lenders typically underwrote adjustable rate mortgages on the assumption that the borrower would pay the “teaser rate” for the whole life of the loan, even though they took account of higher future rates when they calculated the value of the loan itself, which of course determined the size of their commissions (Missal, 2008).

Vertical Segmentation of Lending

The demand for investment grade securities constructed from bundles of mortgages was satisfied through a hierarchically segmented lending market in which investors paid investment banks to oversee the formation of pools of loans from banks and non-bank lenders and their conversion into a security that generated a steady revenue stream and then purchased those securities. In theory, the investment banks securitizing the loans were separate from the lenders originating them. In practice, many banks established close relationships with loan originators and influenced the terms of the loans they made. This vertical segmentation between investment banks and loan originators allowed investment banks to exercise significant control over the lending process while still eschewing liability and ethical responsibility for practices with discriminatory impacts.

Although the separation of mortgage origination from mortgage investment and its implications for the stability of housing markets has been extensively discussed (e.g. Lewis, 2010; McLean and Nocera, 2010), this research highlights the way in which this segmentation was also used by investment banks to influence the types of loans that were originated while displacing responsibility for practices that had foreseeable discriminatory effects. Depositions, for example, describe how investment banks issued bid stipulations to specify the types of loans that they would buy from pools of already originated loans, thus shaping the types of loans that would be originated in the future by sending signals about what loans would be purchased (Kaplan, 2014a; McCoy, 2014). The data also show how investment banks shaped the characteristics of future loans even more directly through “forward-settle” agreements that set out in advance the terms of future loans pools they would agree to buy (Shapiro, 2014).

Some investment banks developed very close relationships with specific non-bank lending firms, which depended on the investment bank as a “warehouse lender” that would purchase the loans, thereby enabling the investment banks to get the types of loans most sought by investors. Morgan Stanley, for instance, had a close relationship with New Century Financial, which was the second largest subprime lender (by market share) in the United States in 2006. In New Century’s submissions to the SEC, it reported that:

We seek to maximize our premiums on whole loan sales by closely monitoring requirements of institutional purchasers and focusing on originating or purchasing the types of loans that meet those requirements and for which institutional purchasers tend to pay higher premiums. During the year ended December 31, 2004, we sold $14.1 billion of loans to Morgan Stanley and $5.2 billion of loans to DLJ Mortgage Capital, which represented 46.4% and 17.2%, respectively, of total loans sold.

(New Century, 2004).

Given its dominance as a buyer, Morgan Stanley was in a strong position to encourage New Century to originate high-cost loans. Indeed, investment bankers at Morgan Stanley told New Century that they would refuse to buy a loan pool if it had too many fixed rate and too few adjustable loans (Vanacker, 2014, p. 48) and they regularly required a certain interest rate spread in the loan pools they purchased (Kaplan, 2014a).

The CFO of New Century, Patti Dodge, made clear how the loans they originated were shaped primarily by the investment banks:

[L]ending criteria are very much driven by the secondary market buyers of the loans, because our financing outlet is a buyer wanting to buy those loans. And in fact the same people who buy our loans generally finance them; they provide us our short-term warehouse line. So we absolutely have to answer to them in terms of that criteria.

(quoted in McCoy, 2014, p. 23).

Proof of Morgan Stanley’s leverage came when the firm abruptly declined to purchase a set of New Century loans in early 2007 and New Century promptly declared bankruptcy.

As one of the deponents in the Adkins case revealed, investment bankers at Morgan Stanley were aware that high-risk, non-conforming loans were being made by New Century, and that many of those loans came disproportionately from cities populated by people of color, such as Detroit. Nonetheless financial officers at Morgan Stanley portrayed themselves as not responsible for any discriminatory lending that might be taking place downstream in the lending process. Morgan Stanley itself did not originate the loans even if in reality it set the terms of lending (Davis 2014, 231–232). The perception of individuals at Morgan Stanley that they were neither legally nor ethically responsible for the discriminatory lending practices was reinforced by boilerplate language in the Mortgage Loan Purchase and Warranties Agreements between it and New Century stated that New Century’s “decision to originate any mortgage loan or to deny any mortgage loan application is … in no way made as a result of [Morgan Stanley’s] decision to purchase or not to purchase, or the price [Morgan Stanley] may offer to pay for, any such mortgage loan” (Kaplan, 2014b, p. 3).

In sum, although the statements of New Century executives indicate that the terms of the loans it originated were powerfully shaped by Morgan Stanley, the standard sales contract between the two firms provided seeming independence and a warranty that any New Century loan Morgan Stanley purchased complied with “[a]ny and all requirements of any federal, state or local law” (Kaplan, 2014b, p. 3). These contract provisions bolstered the perspective of those at the top of the lending hierarchy that were not responsible for any discrimination by loan originators, even if it was foreseeable.

Horizontal Segmentation of Lending

The rise of subprime lending is well documented (e.g. Shiller 2008; Immergluck 2009; Aalbers 2012). Yet the way in which the segmentation of loan origination into separate prime and subprime lenders enabled discrimination has not been adequately addressed. As discussed below, loan originators tended to specialize in either prime or subprime loans (but not both) and some subprime lenders targeted neighborhoods with large shares of black and Latino residents. Even if a subprime lender working in a community of color was inclined to provide a prime loan to a qualified borrower, the lender was unable to. And indeed, in some cases where banks did have both prime and subprime lending operations in the same bank, internal controls were designed to drive potential borrowers towards subprime loans but not the other way around.

Compensation for loan originators was based primarily on commissions from the loans they completed and thus depended on the number of loans, their size, and the fees and interest rates that could be extracted from borrowers (Jacobson, 2010; Unger, 2008). Under these circumstances, it was more profitable for lenders and originators to place a customer in a high-cost subprime loan than in a conventional loan, even if the borrower qualified for a lower-cost prime loan (Jacobson, 2010, p. 3; Paschal, 2010, p. 6). Put quite simply, loan originators wishing to maximize profits had to convince customers with good credit to accept higher-cost, higher-risk lending products. As one Wells Fargo loan officer put it:

The commission and referral system…was set up in a way that made it more profitable for a loan officer to refer a prime customer for a subprime loan than make the prime loan directly to the customer….[S]ubprime loan officers had to give 40% of the commission to the A rep who made the referral…. Because commissions were higher on the more expensive subprime loans, in most situations the A rep made more money if he or she referred or steered the loan to a successful subprime loan officer

(Jacobson, 2010, pp. 2–3).

Whether a mortgage was new or a refinance loan, loan originators seeking to make money could do so most successfully by steering borrowers into high-cost products, regardless of their credit history or credit score. One Wells Fargo loan officer described her role in the firm in this fashion:

When I got the referrals [from prime loan officers], it was my job to figure out how to get the customer into a subprime loan. I knew that many of the referrals I received could qualify for a prime loan

(Jacobson, 2010, p. 3).

Once a loan was referred to a subprime loan officer, there was no way for that officer to make a prime loan. The organizational structure of lending operations served as a one-way ratchet pushing customers toward higher priced loans. As she noted:

Once I got the referral the only loan products that I could offer the customer were subprime loans. My pay was based on the volume of loans that I completed…. Moreover, in order to keep my job, I had to make a set number of subprime loans per month”

(Jacobson, 2010, p. 3).

In short, the horizontal segmentation of the market among—and even within the same originating or lending firm—trapped many borrowers unknowingly in high-cost loans even when they qualified for prime rates.

Institutionalized Racial Targeting

Recent quantitative research has found that metropolitan area levels of segregation in 2010 were strongly associated with higher concentrations of subprime loans because clusters of predominantly black or Latino neighborhoods created “distinct geographic markets that enabled subprime lenders and brokers to leverage the spatial proximity of minorities to disproportionately target minority neighborhoods” (Hwang, Hankinson, and Brown, 2015, p. 1081). Such quantitative data suggest that originators explicitly targeted neighborhoods with large shares of black and Latino residents for high-cost loans, yielding a very strong association between segregation and foreclosures once the market peaked (Rugh and Massey 2010). The question is how and why originators came to target these neighborhoods.

One loan officer described the mindset at his office as follows: “[t]he prevailing attitude was that African-American customers weren’t savvy enough to know they were getting a bad loan, so we would have a better chance of convincing them to apply for a high-cost, subprime loan” (Taylor, 2010, p. 2). Another subprime loan officer described the same general sentiment and set of practices:

It was the practice at the Wells Fargo offices where I worked to target African Americans for subprime loans. It was generally assumed that African-American customers were less sophisticated and intelligent and could be manipulated more easily into a subprime loan with expensive terms than white customers

(Thomas, 2010, p. 2).

In the nation’s capital region, it was no secret that Wells Fargo’s subprime lending division specifically targeted predominantly black zip codes in Washington, D.C., Baltimore, and Prince George’s County (Paschal, 2010, p. 3). In addition to using a language drop-down menu to print marketing materials in Spanish or Chinese, Wells Fargo loan officers soliciting subprime loans could also generate materials in “African American” English designed for black customers (Paschal, 2010, p. 5). One loan officer reported that Wells Fargo managers referred to majority black and Latino Prince George’s County as the “subprime capital of Maryland,” saying that they felt “so lucky” to have the county in their region because of the profits they could make through subprime lending there (Jacobson, 2010, p. 10). Another Wells Fargo loan officer described the incentive structure in the lending division as essentially putting “bounties” on minority borrowers who were then aggressively targeted by the subprime lending division (Paschal, 2010, p. 6).

To identify potential minority borrowers for high-cost home equity loans, lenders turned to data sources that were thought to indicate a lack of financial sophistication combined with a desire for credit. Loan officers were given lists of leads to solicit for subprime refinance loans, and statements by loan originators indicate that these lists did not represent a random cross-section of the local population but were disproportionately African American (Dancy, 2010, p. 2; Taylor, 2010, p. 2). Some lists were generated from current or previous borrowers with the bank, while others were obtained by purchasing lists of customers who had financed the purchase of goods, such as furniture or jewelry, at stores in black and Latino communities (Simpson, 2010, p. 2). Branch managers often used information from businesses located in minority neighborhoods to obtain lists of customers who had already taken out high-cost loans so that they could solicit them for additional high-cost refinancing (Taylor, 2010, p. 3).

Eventually, in addition to purchasing lists of borrowers who both wanted credit and had short-term financial needs, banks began to create their own lists using various exploitative techniques. One approach involved mailing “live” draft checks in the amount of $1,000 or $1,500 to prospective borrowers, targeting bank customers with credit scores in the 500–600 range (Simpson, 2010, p. 5) and non-white borrowers (Thomas, 2010, p. 3). As soon as these “checks” were cashed, they became loans with interest rates as high as 29 percent and the check cashers then became targets for home equity refinance loans (Dancy, 2010, p. 4; Simpson, 2010, p. 5; Taylor, 2010, p. 3). Similarly, branch managers were encouraged to push high-cost consumer loans neighborhoods with large shares of non-white residents, such as auto loans that enabled customers to borrow more than the car’s value with interest rates as high as 24 percent (Simpson, 2010, p. 3). Individuals who took out these high cost loans would subsequently be targeted for larger refinance loans at marginally lower rates that used the borrower’s house as collateral (Simpson, 2010, p. 3). As one loan officers explained:

The way we were told to sell these loans was to explain that we were eliminating the customer’s old debts by consolidating their existing debts into one new one. This was not really true—we were not getting rid of the customer’s existing debts; we were actually just giving them a new more expensive loan that put their house at risk

(Dancy, 2010, p. 2).

How did originators gain the trust of potential borrowers? The qualitative evidence suggests that loan originators often gained the confidence of potential borrowers through the manipulation of trusted co-ethnic intermediaries in community service organizations and churches. To gain the confidence of borrowers, brokers and originators strategically exploited social structures and interpersonal networks within minority communities. Thus promotional materials for Wells Fargo’s “emerging markets initiative” stated that as part of its effort to “further penetrate the market” of “recent immigrants, students lacking financial savvy, young families struggling to build assets, [and] victims of past redlining” the bank had “partnered with a small group of trusted local [nonprofit] organizations” which “became extensions of the bank’s organizational structure” (Wells Fargo, 2007, p. 3).

Loan originators also reported targeting church leaders in order to gain access to congregants through trusted intermediaries, with the originators often providing a donation to a non-profit of the borrower or intermediary’s choice for each new loan, further cementing the relationship between mortgage lenders and local religious and civic leaders (Jacobson, 2010, p. 10; Paschal, 2010, p. 5). As one loan officer described it:

“Wells Fargo hoped to sell the African American pastor or church leader on the program because Wells Fargo believed that African American church leaders had a lot of influence over their ministry, and in this way would convince the congregation to take out subprime loans with Wells Fargo”

(Jacobson, 2010, p. 10).

Solicitations for high-cost subprime loans in predominantly black communities were promoted through “wealth building seminars” held in churches and community centers at which “alternative lending” was discussed. No such solicitations were made in predominantly white neighborhoods or churches (Jacobson, 2010, p. 10). The experience of one of the plaintiffs in the Barkley case combines a number of these marketing techniques and illuminates the myriad ways in which real estate agents, mortgage brokers, lenders, appraisers, and others colluded in abusive lending efforts, and the way in which they used trusted intermediaries to take advantage of unwitting borrowers.

The story begins when Ms. Washington, an African-American plaintiff, was approached by Mr. Wright, a congregant of her church who was close to the pastor. He worked for a company owned by a white real estate investor who bought, then shoddily renovated and flipped over-appraised homes almost exclusively to black or Latino first-time home purchasers. Wright suggested to Ms. Washington that she might be able to buy a house (Washington, 2008, p. 6), although at the time she made only about $600 a week as a child care provider and had never contemplated buying a house before (Washington, 2008, p. 11). After she was told she needed $18,000 for a down-payment, she replied that her savings only amounted to $5,000 (Washington, 2008, p. 12). Wright nonetheless showed her one home, which she liked because it was close to the church (Washington, 2008, p. 15).

Wright, working on behalf of the seller, found her an attorney, a lender, and an appraiser and personally took her to the closing. Through a “seller’s concession,” the real estate company flipping the house put in the money to make a down payment large enough for her loan to be underwritten. Ms. Washington testified that she was told her mortgage would carry a four percent interest rate and that she had never even thought about taking on an adjustable rate loan, or had interest rates explained to her at all (Washington, 2008, pp. 29–31). In the end, she was placed in an adjustable rate mortgage with and interest rate that could climb as high as 9.5 percent. A subsequent appraisal valued the home that she had purchased for $315,000 at just $180,000 at the time of the sale.

DISCUSSION AND CONCLUSION

We analyzed the statements of borrowers, loan originators, investment bankers, and others to identify the mechanisms through which racial discrimination in mortgage lending occurred during the housing boom. Our analyses reveal structural features of the lending industry that facilitated the frequent placement of black and Latino borrowers into higher-cost, higher-risk loans than white borrowers with similar characteristics.

The vertical segmentation of entities within the home finance sector allowed actors with broad influence, such as securitizers, to see themselves as neither legally nor ethically responsible for discriminatory practices taking place at the originator level, even if they encouraged or were aware of the abuses. The horizontal segmentation of loan origination into separate prime and sub-prime sectors meant that, even if a subprime lender were willing to provide a prime loan to a qualified borrower, they generally could not. Further, because the structure of compensation rewarded steering vulnerable borrowers into riskier, higher-cost products, loan originators often targeted neighborhoods with predominantly black and Latino residents that had previously been denied credit and identified trusted intermediaries within those neighborhoods communities to gain the confidence of potential borrowers and sell them abusive high-cost loans.

Our findings are directly relevant to sociological theory on the processes through which the durable inequalities of race are created and maintained. Charles Tilly (1998, pp. 7–8) has suggested that “[d]urable inequality among categories arises because people who control access to value-producing resources solve pressing organizational problems by means of categorical distinctions.” In the present case, the valuable resources were capital and credit, which had long been denied to communities of color, a form of opportunity hoarding (Tilly, 1998). With financial deregulation and the popularization of mortgage-backed securities, the dynamics of inequality shifted from opportunity hoarding to exploitation, as individuals and communities of color came to be seen as a fruitful market for the discriminatory marketing of high-cost loans that would boost the profits of investment banks, the salaries and bonuses of lending officers, and the fees and profits of mortgage brokers.

As a result, black and Latino borrowers were frequently steered into high-cost, high-risk mortgages that later pushed borrowers into foreclosure and repossession. Instead of building wealth, these high-cost loans relentlessly stripped assets away from black and Latino communities and widened inequalities (Kochhar, Fry, and Taylor, 2011).

This analysis of discriminatory lending highlights the need to align the interests of the mortgage originators, securitizers, and investors more closely with those of borrowers. Although motivated primarily by concerns about financial stability, not civil rights, the Dodd–Frank Wall Street Reform and Consumer Protection Act in 2010 did include significant policy changes relevant to the discriminatory practices identified here. In terms of vertical segmentation, the Dodd-Frank Act did not directly address upstream responsibility for downstream discrimination but did subject securitizers to risk retention requirements. The legislation exempted from risk retention requirements securitized pools in which all of the mortgages meet the standards of a “Qualified Mortgage,” which generally means a mortgage underwritten with attention to the borrower’s ability to repay and a mortgage without interest-only periods, negative amortization, balloon payments, a term longer than 30 years, or points and fees that are more than three percent of the loan balance. The Dodd-Frank Act did not directly address the horizontal segmentation of mortgage lending, but did strengthen and consolidate consumer financial regulatory powers in the new Consumer Financial Protection Bureau and enhance federal enforcement powers over nondepository financial companies active in the home finance sector. Section 1403 of the Dodd-Frank Act directly addresses some of the discriminatory practices documented here by amending the Truth in Lending Act to prohibit mortgage originators from directly or indirectly receiving compensation that varies based on the terms of the loan (though it does allow compensation to vary based on the size of loans). Section 1403 also required the Consumer Financial Protection Bureau to amend Regulation Z (the rules implementing the Truth in Lending Act) to prohibit steering a potential borrower from a Qualified Mortgage to a higher-cost one when the consumer qualifies for Qualified Mortgage.

In the face of the higher costs of borrowing, and even with the protections of The Dodd-Frank Act, exploitative and discriminatory practices continue. For instance, with a disturbing parallel to the contract lease system that exploited African-American households’ inability to access conventional financing in the mid-twentieth century (Satter 2009), companies specializing in lease-to-own schemes are peddling substandard housing sold through high-interest installment contracts. The City of Cincinnati recently filed suit against Harbour Portfolio Advisors, one of the nation’s largest sellers of foreclosed homes, alleging hundreds of thousands in dollars in unpaid fines and fees and a failure to properly maintain dozens of homes that it purchased in foreclosure and is selling at inflated prices to low-income households on installment contracts.

Changes in lending since the foreclosure crisis raise the possibility that African Americans and Latinos will once again be marginalized from mortgage markets, returning stratification to the old days of opportunity hoarding through redlining and other forms of discrimination (Goodman, Zhu, and George, 2014; Stein and Johnson, 2013). What remains constant is the structural context of racial residential segregation and the wide gap in social distance between decision makers at mainstream financial institutions and communities of color.