Federal regulators periodically examine CRA-regulated depository institutions’ (banks, hereafter)1 performance in lending and financial services. Banks’ CRA ratings are considered when regulators approve bank mergers, acquisitions, and branch openings.
In 2013, the Office of Management and Budget published new metropolitan statistical area (MSA) and metropolitan division (MD) boundaries, which influenced the neighborhoods that were considered LMI and, therefore, the neighborhoods in which banks could receive CRA credit for mortgage lending.
In a study2 recently released by the Federal Reserve Bank of Philadelphia, we examined the effectiveness of the CRA and the role of nondepository institutions (nonbanks) in mortgage lending in LMI neighborhoods by focusing on those neighborhoods in the previous five-county Philadelphia MD that became eligible and ineligible for CRA credit as a result of the new MSA/MD boundaries. We present a few findings from the full report here.
We found that the growth in home purchase loans made by banks following the boundary change was restrained in neighborhoods that lost their CRA eligibility and that this slowdown in lending provided an opportunity for nonbanks, which are not subject to the CRA, to expand their market share. Purchase lending by banks increased in newly eligible neighborhoods, but the increase was not significant when compared with lending in nearby similar-income neighborhoods with unchanged CRA-eligibility status.
Unintended Consequences of Changes in Metropolitan Division Definitions
Updates to metro area definitions can affect the designation of LMI neighborhoods. In 2013, the former five-county Philadelphia MD was split into two: a new Philadelphia MD consisting of Philadelphia and Delaware counties and a second MD consisting of Montgomery, Bucks, and Chester counties (MBC MD). Because LMI neighborhoods are identified by comparing the income level of a census tract with the income level of the corresponding MD, the split of the former Philadelphia MD changed which neighborhoods were considered LMI by CRA standards.3 There were a total of 102 census tracts exclusively in the new Philadelphia MD that became middle-income tracts and lost CRA eligibility status; we termed these tracts “newly ineligible” tracts. There were a total of 80 census tracts exclusively in the MBC MD that became moderate-income tracts and gained CRA eligibility status; we termed these tracts “newly eligible” tracts. Figure 1 shows the location of the newly eligible and newly ineligible tracts.
Figure 1. The Location of Newly Eligible and Newly Ineligible Census Tracts
Putting this into context, about one-third (34.5 percent) of previously CRA-eligible tracts in the new Philadelphia MD became CRA ineligible after 2014, whereas the number of CRA-eligible tracts in the MBC MD tripled from 2013 to 2014. Across all the major metropolitan areas in the U.S., the Philadelphia area observed the largest prevalence of neighborhoods with changed LMI designations from 2013 to 2014. This definitional change provided a unique opportunity to determine the impact of the CRA on mortgage lending in LMI neighborhoods.
To isolate the impact of the CRA, we constructed two control groups: one for newly eligible tracts and another for newly ineligible tracts (the “treatment” groups); each control group was composed of nearby tracts whose CRA eligibility status had not changed and whose income was similar to that of the respective treatment group. To capture banks’ responses to the geography change, we aggregated mortgage applications, originations, and volume for the two-year period before and after the MD split occurred (2012 to 2013 compared with 2014 to 2015). By comparing lending activity between the treatment and control tracts and including other controls in a regression analysis, we were able to isolate the effect of the change in metro area boundaries on CRA lending.4
The Effect of Losing CRA Eligibility
The loss of CRA eligibility inhibited the growth in home purchase loans made by banks in newly ineligible neighborhoods in the new Philadelphia MD. Home purchase loans made by banks increased by 6.2 percent in the newly ineligible neighborhoods but grew by 21.7 percent in the control neighborhoods (Figure 2). This difference was found to be statistically significant in the tract-level regression analysis. Was this lending gap between newly ineligible and control neighborhoods due to tighter underwriting standards or to less outreach and marketing to those neighborhoods that lost their CRA eligibility status? Denial rates declined similarly in both the newly ineligible and control neighborhoods, while purchase applications submitted to banks declined in newly ineligible neighborhoods. Therefore, the inhibited purchase lending growth in newly ineligible neighborhoods was likely due to reduced outreach and marketing efforts than to tighter underwriting standards.
Figure 2. Percent Change in Home Purchase Loans Made by Lenders Subject to the CRA, Philadelphia MD
Nonbanks partly filled the void created by banks in the newly ineligible neighborhoods. Nonbanks originated 46.6 percent of all home purchase loans in newly ineligible tracts between 2012 and 2013. During 2014 to 2015, nonbanks originated 53.1 percent of all home purchase loans, an increase in market share of 6.5 percentage points, which was larger than the increase of 3.3 percentage points for nonbanks in the control group.
The Effect of Gaining CRA Eligibility
If banks reduced their home purchase loans in neighborhoods that lost their CRA eligibility status, then conversely, did banks increase their lending activity in neighborhoods that gained CRA eligibility status? The answer is not straightforward. Banks increased their home purchase loans in newly eligible neighborhoods in the MBC MD by 5.4 percent and decreased their purchase lending by 2.0 percent in neighborhoods in the control group (Figure 3). While this suggests that CRA coverage translates into a 7.4 percentage point boost in purchase lending between newly CRA eligible and control neighborhoods, this difference becomes insignificant in the tract-level regression. At least two factors could help explain the lack of significance in the CRA effect in the MBC MD. First, residents of the MBC MD have higher incomes on average than do Philadelphia MD residents, so it is possible that banks were serving these neighborhoods even before they became CRA eligible. Second, the current lending environment is characterized by significant regulatory change and heightened lender caution, both of which could explain why banks might be quicker to reduce lending in newly ineligible neighborhoods than pick up lending in newly eligible neighborhoods.
Figure 3. Percent Change in Home Purchase Loans Made by Lenders Subject to the CRA, MBC MD
Overall, our findings suggest the CRA plays an important role in expanding access to credit to LMI communities. While revising MDs is important for accurately representing metropolitan areas as they undergo change, our research indicates that these types of changes can yield unintended consequences for CRA lending activities.
Editor’s Note
The Federal Reserve Bank of Philadelphia received a commendation from the Philadelphia City Council for its study on the CRA.
The views expressed here do not necessarily represent the views of the Federal Reserve Bank of Philadelphia or the Federal Reserve System.
[1]For simplicity, we use the term “banks” in this article to refer to the subset of depository institutions that have a local branch located in the same county as the mortgaged property. These banks are likely subject to the CRA.
[2]Lei Ding and Leonard Nakamura, “’Don’t Know What You Got Till It’s Gone’ — The Effects of the Community Reinvestment Act (CRA) on Mortgage Lending in the Philadelphia Market,” Federal Reserve Bank of Philadelphia Research Department Working Paper 17-15 (June 2017); available here. Lei Ding and Kyle DeMaria, “A Practitioner’s Summary: The Effects of the Community Reinvestment Act (CRA) on Mortgage Lending in the Philadelphia Market,” Federal Reserve Bank of Philadelphia (June 2017); available here.
[3]The CRA defines LMI neighborhoods as those census tracts that have a median family income (MFI) that is less than 80 percent of the MSA or MD in which those families reside. The MFI decreased from $76,400 in 2013 to $54,200 in 2014 in the new Philadelphia MD, moving 102 additional tracts above the 80 percent threshold. The MFI increased from $76,400 in 2013 to $95,400 in 2014 in the MBC MD, moving 80 tracts below the 80 percent threshold.
[4]To control for unobserved differences across neighborhoods, we also performed a number of difference-in-differences regressions based on data aggregated to the census tract level in the paper.