Database Reference
In-Depth Information
(Tillyer et al., 2010; Engel, 2008) for extensive references. Recent additional ap-
proaches include (Anbarci & Lee, 2008; Blalock et al., 2007; Pickerill et al., 2009;
Ridgeway, 2006).
6.5.2
Credit Markets
Discrimination in the lending process may occur at several steps, from advertis-
ing, to pre-application enquires, to loan approval/denial, up to loan administration
(Turner & Skidmore, 1999). Among the various credit markets, mortgage lending
has received most of the interest. In all cases, however, the main challenge is in the
difficulty of estimating the risk of granting a loan to an applicant on the basis of her
financial capacity and her personal characteristics.
In the US, the
Home Mortgage Disclosure Act
(HMDA) requires lenders to gather
and to make available census data about their mortgage applications. Since 1990,
the HMDA has been integrated with information on discrimination grounds of ap-
plicants. One of the first relevant contribution is due to researchers at the Federal
Reserve Bank of Boston in the research work known as
Boston Fed Study
(Munnell
et al., 1996). They supplemented the original census HMDA data for Boston with
additional information on the credit history of more than 3,000 individual applicants,
including data from more than one hundred financial institutes. Several criticisms
of the Boston Fed study appeared in the literature (Ross & Yinger, 2002, Chapter
5), (Longhofer & Peters, 1999), (Turner & Skidmore, 1999, Chapter 3). Among the
problems highlighted, we mention data errors, misclassification problems, endoge-
nous explanatory variables and the omitted variables bias (e.g., loan amount and
indicator of cosigner were missing). A theoretical and empirical survey on racial
disparities in mortgage lending markets in the context of the fair housing legisla-
tion is provided in (LaCour-Little, 1999). (G. Dymski, 2006) describes the state-
of-the-art on discrimination in housing and credit markets both from a legal and an
economic perspective. A recent review has been proposed in (Yezer, 2010), which
devises three approaches of testing disparities in loan approval decisions: mortgage
rejection, pricing and defaults.
In
mortgage rejection
, the disproportionate rate of rejected decisions between
racial groups of applicants is considered
prima facie
evidence of discrimination.
Empirical studies (Clarke et al., 2009; Dietrich, 2009; Dietrich & Johannsson, 2005;
Goenner, 2010; Sanandaji, 2009) include the analysis of HMDA data at
bank level
(i.e., a model for each bank under analysis) or at a
market level
(i.e., a single model
aggregating variables for several banks). An experimental comparison of the two
approaches is reported in (Blackburn & Vermilyea, 2006). Other sources of data
range from micro-lending data (Agier & Szafarz, 2010) to on-line data derived from
a peer-to-peer lending site (Pope & Sydnor, 2011).
Mortgage pricing
concentrates on the dataset of approved loans, by considering
whether a minority group is systematically charged with the highest interest rates.
Search WWH ::
Custom Search