Association for Public Policy Analysis & Management

The Community Reinvestment Act instructs banks to serve the credit needs of the markets in which they do business. I study how banks have responded to CRA incentives in residential mortgage lending through different periods in the credit cycle from 1990 through 2017. Residential mortgages qualify as eligible for CRA credit in three ways: the property secured by the mortgage is in a low-to-moderate income (LMI) census tract (tract median income is less than 80 percent of area median income), the borrower’s income is less than 80 percent of area median income, or the tract is a distressed or underserved nonmetropolitan tract.

I use a regression discontinuity design to estimate the impact of the median tract income on originations using tracts just below and above the LMI cutoff. Prior to 2002, more mortgages were originated in census tracts with median income just below the LMI threshold than would otherwise be predicted. This increase disappears during the housing boom of the early to mid-2000s and becomes negative during the housing crisis. From 2012 onward, the bump in CRA eligible tracts returns and more mortgages are originated just below the threshold than otherwise predicted.

Next, I will use Call Report data to study whether liquidity constraints help predict whether a bank pulls back on CRA-eligible lending. Following liquidity constraints literature, the greater the ratio of cash plus federal funds sold plus securities to total assets, the better a bank should be able to overcome shocks to sources of credit. I hypothesize that as short-term debt markets collapse in 2007 and 2008, those banks that have a smaller liquidity buffer will pull back on CRA-eligible lending the most. To test, I will estimate bank level regression discontinuity designs, again focusing on census tracts with median income just below and above the LMI threshold. I will then regress the coefficient estimates from each bank-year regression against the liquidity ratio described above. A positive relationship would support the hypothesis that banks with lower liquidity buffers decreased CRA-eligible lending more.