The Community Reinvestment Act: A Regression Discontinuity Analysis

The Community Reinvestment Act (CRA) is the most important federal legislation addressing the credit needs of low-income families and neighborhoods. Enacted in 1977 in response to concerns about redlining, CRA established "continuing and affirmative obligations" for lenders to meet the credit needs of low- and moderate-income (LMI) neighborhoods. Advocates point to nearly $1 trillion in CRA-eligible lending and rapid growth in credit for minority and LMI borrowers as evidence that the act has had a significant impact on its intended communities. However, critics counter that other factors, such as technological changes and market restructuring, account for the growth in lending to LMI markets, making CRA irrelevant. Existing empirical studies have not disentangled convincingly the effects of the law from other social and economic trends, and the controversy over CRA continues. In this paper, we apply regression discontinuity analysis, a design that arguably is stronger for causal inference than any other except a randomized experiment, to identify the effects of CRA on lending to LMI individuals and neighborhoods. Specifically, the regulation uses a strict cut-off of 80 percent of metropolitan area (MSA) median income to identify LMI neighborhoods and individuals, loans to which count toward a lending institution's CRA performance rating. Using this RD design and a data set comprising millions of mortgage applications from 1993 to 2003, we are able to distinguish the causal effects of CRA from other (observable and unobservable) variables correlated with residential lending rates. To preview our results, we find vanishingly small effects of CRA on the outcomes we examine.