The history of lending discrimination is well-documented in the United States, and the federal government employs a variety of legal mechanisms to fight its effects and promote lending equity. One such tool is the Community Reinvestment Act (CRA) which Congress passed in 1977. The CRA was designed to encourage banks to lend in the geographic areas where they raise deposits. In particular, the law empowers regulatory agencies to evaluate consumer banks on their lending to low- and moderate-income (LMI) borrowers and areas, threatening banks with sanctions if they do not pass muster.
Despite the importance of the CRA’s goal and the considerable resources that banks and regulators expend on compliance and supervision, the evidence on the efficacy of the CRA in increasing credit access has generally been mixed. In part, this owes to the difficulty of empirically studying the program’s effects: Data on consumer credit has been historically hard to come by, and it has been difficult for economists to model counterfactual outcomes for what may have happened to covered areas in the absence of the program.
In a new paper, we present solutions to these empirical challenges and offer new evidence on the efficacy of the CRA in promoting access to credit in LMI areas. We use the New York Fed/Equifax Consumer Credit Panel (CCP), which contains detailed credit information for a representative sample of the US since 1999, to delve into the credit dynamics for individuals living in the LMI areas targeted by the CRA. We leverage this data using various empirical methods to compare the borrowing of individuals in targeted areas to similar areas that are not under the purview of the CRA. All our investigative strategies come to the same conclusion: The CRA does not significantly impact household credit. Then, by looking in greater detail into the mortgage market, we document how banks purchase existing loans more frequently in CRA-eligible areas, thereby satisfying their CRA obligations without necessarily extending new credit.
How Does the CRA Work?
All depository institutions in the United States are subject to CRA regulations. Federal supervisors regularly examine these institutions and grade their CRA performance on a four-point scale. These CRA grades depend on various metrics, but regulators place significant emphasis on lending. Specifically, a bank’s lending assessment depends on its lending to the geographical area where the bank has its main office, branches, and deposit-taking ATMs, as well as surrounding geographies where it performs a large portion of its lending.
Loans are considered CRA-eligible by regulators if they are made to LMI census tracts (a small geographical unit, generally with between 1,200 and 8,000 occupants) within a bank’s assessment area. A tract is defined as LMI if it has a median family income (MFI) of less than 80 percent of the median in its larger geographic area, typically the surrounding metropolitan statistical area (MSA). Depending on a bank’s holding of loans made to these surrounding LMI tracts, they can receive scores ranging from “Outstanding” and “Satisfactory” (considered passing grades) to “Needs to Improve” and “Substantial Noncompliance.” The regulatory bite of the law comes from the fact that new branch applications and merger applications can be denied for banks receiving failing grades. However, most banks receive passing scores (97 percent).
Evaluating the Effects on Household Borrowing
We analyze the effects of the CRA by exploiting specific details of its implementation. First, we compare individuals who live in census tracts just above the income necessary to qualify as eligible to those in census tracts just below the threshold (80 percent of the surrounding MSA). The economic assumption the approach relies upon is that individuals who live very close to the income threshold, but happen to fall on either side of it, do not have different levels of credit access on average after controlling for income, other than through the effects of the CRA.
From 1999-2017, we compare the lending balances of individuals in tracts with an MFI within 15 percent of the 80 percent eligibility cutoff (controlling for the differences in income). Across a variety of types of lending (mortgages, car loans, credit card balances, and total debt), we find that individuals in just-eligible areas do not have higher balances than those in just-ineligible areas.
To supplement this exercise, we take a second approach to identifying CRA-ineligible individuals who are otherwise similar to eligible ones by comparing neighbors. Instead of examining the sharp MFI cutoff, we examine a geographic cutoff, comparing individuals on opposite sides of a census tract border, where one side is eligible, and the other is ineligible. Just as in the first analysis, neighbors that happen to differ in their CRA eligibility have indistinguishably similar levels of debt across all categories we study. This provides further evidence that the CRA has not increased credit access for individuals in LMI tracts since 1999.
We look at how individual debt balances respond to gaining CRA eligibility as an additional empirical approach. Specifically, we study areas that gain eligibility due to geographical redefinitions or data updates, comparing them to areas that remain ineligible during the entire period. To account for the potential of location-varying economic trajectories, we control for time trends in tract income using the borrowing activity of neighbors. We do not find any statistically significant evidence that debt balances increase after gaining CRA eligibility.
Evaluating Other Measures of Financial Service
Of course, the volume of debt is not the only useful measure of equity in credit access. One important secondary outcome we investigate is whether CRA eligibility seems to cause new individuals to enter the formal credit system (e.g., by taking out a first loan or a first credit card). We use the same three empirical approaches to study the number of individuals with any credit history rather than just the amount of borrowing, and similarly find that the CRA does not make a significant difference during the period we study.
Additionally important is whether the regulation may have affected adverse credit outcomes. A classical concern around programs oriented toward increasing credit access is that incremental lending is riskier and targeted toward borrowers that are unable to pay back their obligations. We use our empirical approaches to study rates of bankruptcies, foreclosures, and delinquencies, as well as changes in Equifax’s Risk Score (a measure of creditworthiness). Our empirical approaches rule out a significant effect of the CRA on any of these measures, consistent with our conclusion that the CRA does not seem to play a large role in consumer credit outcomes in general.
A Possible Explanation: Loan Churn
The fact that banks are substantially compliant with the CRA and yet consumers do not seem to have larger access to credit in eligible areas may seem at odds. However, banks have a variety of ways to comply with the CRA. While lending is the most important category of their evaluations, they can satisfy this regulatory burden by purchasing loans in addition to (or as an alternative to) originating loans. Many mortgages are made by non-depository financial entities that are not subject to CRA examinations.
Inspired by this fact, we examine loan-level data from the Home Mortgage Disclosure Act. We find that consistent with the CRA pushing banks to focus extra attention on LMI areas, banks do indeed originate more loans in CRA-eligible areas. However, we again find that total lending to these areas is no different from ineligible areas. Rather, banks purchase loans from entities such as credit unions and independent mortgage banks, which are not covered by the CRA, at significantly higher rates than in ineligible areas. They then typically resell these loans to government-sponsored entities within the calendar year. This suggests that rather than extending new lines of credit to individuals in LMI tracts, banks often temporarily own existing loans from entities that are not covered by the law as part of their CRA compliance.
Using various approaches, we examine the effects of the CRA on household borrowing and adverse credit outcomes since 1999. We do not find evidence that households have any higher access to credit due to the CRA, which rules out significant positive effects on consumer welfare through this channel. While we cannot comment on other aspects of the CRA, such as its provisions encouraging community investment in LMI areas, our evidence suggests that one of the primary directives of the law, extending credit to borrowers who would not receive it otherwise, is likely not being achieved.
We also detail one mechanism contributing to this lack of regulatory bite. We document that banks subject to the CRA purchase existing loans from unregulated entities. This highlights the pitfalls of a circumscribed regulatory regime. Current efforts by regulators seek to modernize CRA rules to help better achieve its goal of increasing credit access to low- and moderate-income communities. Our results suggest that the current implementation system may hamper the law’s efficacy and suggest possible ways to improve outcomes.
Jacob Conway is a Ph.D. student in Economics at Stanford University.
Jack Glaser is a Ph.D. student in Economics at the University of Chicago.
Matthew Plosser is a financial research advisor in Banking Studies at the Federal Reserve Bank of New York within the Financial Intermediation Policy Research Division.