Ebook Credit Card Redlining
This paper evaluates the presence of racial disparities in the supply of revolving consumer credit. Disparities in access to such consumer credit as credit cards are critical to assess because this form of credit is generally the first form of credit accessed by consumers. In order to qualify for a mortgage, one typically has to "build" a credit history. This marks a significant change that has taken place over the past few decades. In the 1960s, borrowing was predominantly related to home purchases. However, households now have more access to personal loans, auto loans, educational loans, and, significantly, credit cards; building credit involves using one or more of these products to incur debt and successfully repay it. As a result, disparities in access at this stage will be magnified when consumers seek access to such products as mortgages.
To frame the research for this paper, it is useful to consider a couple of borrowers. Consider two individuals, each of whom is the same age and earns a similar salary. Our two individuals have similar credit histories in the sense that they have both obtained and used credit with similar patterns of delinquency and repayment. Thus, the two have identical credit scores. The only characteristic that will distinguish our borrowers is the racial composition of the neighborhood in which they live. Individual A lives in a predominantly White neighborhood and individual B lives in a majority Black one.
This papers principal observation is that remarkably, in spite of identical scores and identical community characteristics, our individual in the Black neighborhood receives less consumer credit (e.g. fewer credit cards) than the individual in the White area. That is, in spite of the fact that both have been assessed to have similar risks of nonpayment, as determined by the credit score, the person living in the Black area has less ability to access credit. Notice here that the example does not identify the race of the individuals, only the neighborhoods in which they live.
As is well known, there are large correlations between racial compositions and other socioeconomic factors that may be related to an individuals ability to repay debt. For example, high vacancy rates may impact home equity appreciation rates, and thus in areas with low growth, individuals may not be able to subsidize consumer spending with equity financing. Many factors indeed show a correlation between credit quality and neighborhood racial composition. For example, Panel A of Table 1 shows the results of a series of univariate regressions of "months since last delinquency" on a handful of demographic characteristics. Notice that, given the results of this table, the process of an issuer implementing some simple marketing differentiation by such location-based characteristics as crime rates could appear to an outsider as race-based outcome differences even when race is explicitly excluded from consideration. That is, the stylized fact above that individual B received less credit may simply be due to issuers avoiding lending in areas with endemic crime, vacancy, etc. That is, an issuer could argue that using race, or one of a number of other factors correlated with race and delinquency, provides information related to profitability: those living in communities with more African Americans appear to have higher default and delinquency rates than those outside. To support this type of argument, panel B of the same table shows a correlation matrix of these variables with the percentage of African Americans in a census block.
This takes us to the second stylized fact: the disparity in credit access persists even after one accounts for the socioeconomic characteristics that one might suspect are correlated with ability repay debt. In fact, it appears to survive the inclusion of numerous demographic and socioeconomic variables available in the census report.
The broad goal of this paper is to distinguish between the two stylized facts above. The first found that there are race-based differences in access at an average level, conditional on credit score. The latter found race-based differences even after conditioning on additional information potentially useful in a credit decision. That is, one wants to know whether race-based differences in credit issuance are present, even after conditioning on non^race-based profitability measures.
To answer the broad question of the presence of a location-based race coefficient, this paper uses a set of data that has not previously been applied to this topic: that is, data from a nationwide representative sample of credit reports. Used in conjunction with publicly available census and crime information, this data gives one a potential window into the methods of issuers. The paper accomplishes this in three ways. One, the data used are new to this literature and provide a number of advantages: the data include individual-level credit reports for a nationally representative and very large sample of individuals; the reports include information on total credit available and on credit used; and credit limits provide a logical proxy for supply, while the amount of credit used offers a clear interpretation for demand. This allows one to avoid the simultaneity questions that have confronted some of the mortgage literature. Two, though the quality of the supply proxy is very good and the reduced form may be sufficient for inference, the paper also uses an instrumental variables approach to account for the simultaneous determination of credit limits and utilization. And three, the instruments chosen are based on phenomena related to the correlation of consumer behavior (demand for credit) across individuals in social proximity.
Drawing on these methods, the paper finds evidence of race-based differences in the availability of credit. Though issuers marketing and underwriting decisions are not fully known, it appears likely that a race variable is included somewhere in the determination of credit availability. The remainder of this paper is organized as follows. Section 2 provides a review of related methodology and concepts from the mortgage literature. Section 3 discusses the data and section 4 describes the methodology of this paper. The results are presented in section 5, a discussion of potential confounding issues are addressed in section 6, and a conclusion is provided in section 7.
Download
PDF Ebook Credit Card Redlining
Posted in :
