Research Papers

Giving Credit where Credit is Due? The Community Reinvestment Act and Mortgage Lending in Low Income Neighborhoods

The Community Reinvestment Act (CRA) is a major credit market intervention intended to expand credit access in low income communities.  I estimate the effect of CRA on mortgage credit flow by exploiting knowledge of the selection rule that determines which census tracts CRA encouraged banks to lend in.  Using a comprehensive source of micro data on MSA mortgage applications, I implement a regression discontinuity design that compares lending in census tracts just above and below the CRA-induced regulatory line.  I find that CRA affects bank lending primarily in the largest MSA’s, where enforcement is most intense.  The analysis indicates that CRA’s effect on bank originations was about 4% between 1994 and 1996, and expanded to 8% in 1997-2002, consistent with the timing of a reform strengthening CRA.  The results also suggest net "crowd-in": lending to targeted tracts by unregulated institutions rises in post-reform years, in particular to those areas that have had relatively low home purchase volume in the recent past.  This evidence is consistent with a model of information externalities in credit markets.

 

The Impact of the Community Reinvestment Act on Mortgage Lending to Low-Income Borrowers

Between 1994 and 2005 the U.S. homeownership rate rose faster than at any period since the 1950’s, led by a rise in homeownership amongst low-income households. This paper tests whether the Community Reinvestment Act (CRA), a law mandating that banks lend to low-income families, may have contributed to the rise in homeownership. While coincident changes in the mortgage market make isolating CRA’s effects challenging, I exploit a discontinuity in CRA’s income eligibility rule to identify CRA’s causal impact. Using local regression discontinuity methods similar to those described in Imbens and Lemieux (2008), I find CRA’s impact is concentrated in the largest MSA’s during a 5-6 year period immediately following enactment. The RD estimates indicate that CRA caused a 6% increase in home purchase lending by banks at the regulatory cutoff. These results are robust to various bandwidth choices and control function specifications, and tests for discontinuities away from the cutoff and for unregulated lenders suggest the finding for banks is not spurious.

 

 

Regression Discontinuity Estimates of the Effects of the GSE Act of 1992

In this paper I estimate the effect of the Underserved Areas Goal established under the “GSE Act”, a 1992 law mandating that the housing government-sponsored enterprises Fannie Mae and Freddie Mac help promote credit access and homeownership opportunities for low-income households and in low-income and minority neighborhoods. I identify the goal’s impact by taking advantage of a discontinuity in the census tract eligibility rule. Employing local linear and non-parametric regression discontinuity methods, I find that this goal has had a direct effect on GSE purchasing activity of 3-4% and increases overall GSE-eligible originations by 2-3% on average at the cutoff. Unlike previous research, I find no evidence of an offsetting reduction in FHA and subprime lending. The results imply a lower bound on the aggregate impact of the goal on credit in targeted neighborhoods of about $2.4 billion between 1997 and 2002.