As we gear up for federal agency reform proposals regarding the Community Reinvestment Act (CRA) regulations, Penn Institute for Urban Research has released a useful collection of academic studies regarding the effectiveness of CRA and proposals for reform.
Called “The Future of the Community Reinvestment Act,” the publication is a series of articles examining retail lending, community development financing, and branching of banks. When contemplating the future of CRA, a good place to start is a baseline of current bank activities, which this provides.
The data presents a mixed picture, neither uniformly backing CRA’s effectiveness nor concluding that CRA has failed. As the National Community Reinvestment Coalition (NCRC) has long suggested, incremental reforms targeted toward areas in clear need of fixing, rather than a tear-down-and-rebuild approach, is the best way to address this mixed picture.
In “Is the CRA Still Relevant to Mortgage Lending?” Paul Calem, Lauren Lambie-Hanson, and Susan Wachter describe trends in large bank, regional bank, community bank, and nonbank lending to low- and moderate-income (LMI) borrowers and communities. The large banks have decreased their home lending to LMI borrowers and communities in large part because they have retreated from Federal Housing Administration (FHA) government-insured lending. Non-banks, which are reliant on government-insured lending, have filled the gap left by the big banks in the LMI market.
Before critics decry the ineffectiveness of CRA, they should be reminded that non-banks rely on another public sector intervention, FHA lending, to serve LMI markets. The largest banks have said that the False Claims Act and stringent requirements to document FHA lending or else face costly buybacks of loans from Ginnie Mae were responsible for their abandoning FHA lending. Large banks have other lines of business that they can increase in lieu of FHA lending. In contrast, non-bank mortgage companies do not have other lines of business and thus have gobbled up FHA lending. Solving this FHA issue for the largest banks is an issue that is beyond CRA reform.
However, an issue that is under the domain of the federal bank agencies is the long lag time between CRA exams for the largest banks. NCRC has documented that while the federal agency CRA examination guidelines generally stipulate a two- to three-year CRA exam cycle, successive CRA exams for the largest banks can be six or seven years apart.
Such a long period between CRA exams has probably contributed to the large banks’ decrease in lending to LMI borrowers and communities by not stimulating them to increase conventional lending to these populations in the wake of their retreat from government-insured lending. Thus, the Calem et al. findings suggest an incremental reform: fix the long lags between CRA exams for the largest banks rather than throwing out the current structure of their exams.
The other major finding by Calem et al. is that the smaller banks—community banks—have largely maintained their performance in lending to LMI borrowers and communities. In contrast to large and regional banks, community banks “consistently have originated a comparatively large percentage of their conforming conventional and FHA mortgage loans to LMI borrowers, including in recent years.” For community banks, it is likely that more frequent CRA exams have kept them on their toes and provided stronger incentives to lend to LMI borrowers and communities compared to the largest banks. Future research should include focus on groups of banks of different sizes to ascertain more of the reasons behind the variations in performance.
In “The Community Reinvestment Act: What Do We Know, and What Do We Need to Know?” Laurie Goodman, Jun Zhu, and John Walsh of the Urban Institute appear to conclude that CRA has been more effective in stimulating multifamily lending in LMI tracts than single-family lending for LMI borrowers and communities. CRA exams designate “assessment areas,” geographical areas with bank branches, in which lending activity is scrutinized. Intuitively, lending to LMI borrowers and communities should be higher in assessment areas than areas beyond bank branches since banks are rated in their assessment areas. Interestingly, the Urban Institute found this was the case for multifamily lending but not for single-family lending.
In a simple world, a supporter of CRA would like a slam dunk result, in this case, that bank LMI lending was much higher in assessment areas for both single-family and multifamily lending. Alas, life is often more complex than the descriptive data analysis presented in the Urban Institute paper.
Other research, by Lei Ding and Daniel Ringo of the Federal Reserve Board, found that after controlling for demographic and economic factors, adding or taking away assessment area status or whether census tracts are considered LMI under CRA definitions did influence how much lending took place. Thus, the conclusion should be that CRA does have an assessment area impact on LMI lending, but that reform should strengthen this impact by combating grade inflation in which 98 percent of banks currently pass their exams. On another topic in their paper, the Urban Institute authors were absolutely correct that multifamily lending is a critical and overlooked part of CRA and that efforts by the industry to delete multifamily lending reporting in the Home Mortgage Disclosure Act (HMDA) must be resisted.
Kevin A. Park and Roberto G. Quercia, in “Who Lends Beyond the Red Line? The Community Reinvestment Act and the Legacy of Redlining,” identify a clear weakness in CRA exams that is best addressed by an incremental reform rather than a wholesale revamping. Park and Quercia, in an approach building upon NCRC’s research from last year, examined lending trends in neighborhoods that were identified by the Home Owners Loan Corporation (HOLC) during the 1930s as high risk. These were so-called category C and D neighborhoods that were heavily populated by minorities and recent immigrants. Park and Quercia documented that they are still predominantly minority. Furthermore, banks have significantly lower market share in them than non-banks. Overall, however, banks have higher market shares to LMI borrowers than middle- and upper-income borrowers in the 14 cities they examined.
The contradiction described by Park and Quercia is puzzling, but an explanation is that CRA is propelling banks to do well in LMI markets overall but not in the most distressed urban neighborhoods that are predominantly minority. CRA exams are conducted at a metropolitan level, and lending overall to LMI borrowers and neighborhoods is examined, not lending in specific neighborhoods or distressed neighborhoods. The current CRA analysis therefore can allow banks to focus on LMI neighborhoods that are gentrifying and leave behind LMI or communities of color that are distressed. To remedy this, NCRC has advocated for adding a criterion to the lending test and possibly the other component tests that would consist of lending and other activity in distressed census tracts that could be identified by low levels of lending per capita.
In “The Community Reinvestment Act and Bank Branching Patterns,” Lei Ding and Carolina Reid shift focus from retail lending to branching. Their paper is an important contribution to the literature because while several studies have scrutinized CRA home and small business lending, none to my knowledge have examined CRA’s impact on branching. Overall, the number of branches has declined significantly, from 88,022 in 2009 to 79,872 in 2018. The decline has been steeper in LMI tracts, at 11 percent, than non-LMI tracts, at 9 percent, during this same time period. Nevertheless, Ding and Reid employed an econometric analysis that controlled for demographic and economic factors in census tracts examined under CRA compared to census tracts with income levels just a little higher than CRA income levels. They found that CRA has reduced the number of branch closures by 11 percent in LMI tracts. Importantly, CRA’s impact is the strongest preventing closure in LMI tracts with just one branch, thus preventing banking deserts. Also, CRA is preventing substantial declines in small-business lending that occurs when branches close.
Some critics of CRA assert that CRA impedes branching by creating barriers to entry and exit. According to this analysis, banks are reluctant to open branches because they figure they will be subject to burdensome CRA exams in the areas with the new branches and that CRA will penalize them harshly if they close branches. However, in the more than 40 years of CRA’s history, no such impacts on branch opening have been documented. On the contrary, the FDIC found that branching per capita has actually increased in recent years compared to previous decades despite the periodic booms and busts in bank branching. Furthermore, Ding and Reid’s study did not say that banks are compelled to keep branches open in LMI areas where branching is unprofitable due to low population or little economic activity. Instead the study said that when all other factors are equal, LMI status alone helps motivate banks to keep branches open.
This is an important study arguing for the preservation of the service test that examines bank branching patterns in LMI areas. CRA reform should retain the criterion of branching and add more data that includes how many deposit products and other bank services are reaching LMI borrowers and communities. Current CRA exams mostly lack this data, including CRA exams of the newer banks that have few or no branches. There is no way to assess the effectiveness of these banks without data on deposits and services used by LMI populations.
Carolina Reid’s paper, “Quantitative Performance Metrics for CRA: How Much ‘Reinvestment’ is Enough?” is a masterful quantitative and qualitative analysis of the likely harm of the Office of the Comptroller of the Currency’s (OCC) one ratio approach, which would reduce bank responsiveness to local needs. Under this approach, a ratio would be computed for each assessment area and for the bank performance overall that would compare the dollar amount of CRA activity to the dollar amount of a bank’s deposits. In order to increase the numerator of the ratio, banks would favor larger deals and financing in more expensive areas. Reid’s analysis illustrates this in California with high-cost areas like San Francisco being favored over inland areas in California that have higher rates of unemployment and economic distress. In a particularly disturbing part of the report, Reid calculates that in San Francisco, 92 percent of lending in LMI tracts is for middle- and upper-income people enjoying the benefits of gentrification and high-cost housing. A city official interviewed by Reid complained that banks do not have an interest in participating in a safe and sound city affordable homeownership program. This is not surprising and it would seem that a one ratio would exacerbate the dysfunctional outcomes of inequalities among geographical areas, and non-responsiveness to real needs in violation of the letter and spirit of the CRA statute. This is powerful evidence supporting incremental reform instead of the so-called transformational approach of the OCC.
In “Updating CRA Geography: It’s Not Just About Assessment Areas,” Mark Willis describes an approach for examining both branchless banks and hybrid banks that use branches and non-branch means to serve customers. He gets it about half right. He is correct that banks should be assessed in geographical areas where they have significant numbers of retail loans but no branches. However, he overlooks the importance of insisting on banks meeting local needs in the case of community development financing for affordable housing and economic development. Retail loans are for individual homeowners or small business owners whereas community development refers to larger-scale investments or loans that finance multifamily affordable housing, sizable economic development, or neighborhood revitalization projects.
Willis suggests that hybrid or branchless banks be examined on a national basis for community development financing with extra credit awarded if they focus on community development in geographical areas receiving large amounts of their retail loans. The extra credit could end up inflating CRA ratings by awarding higher ratings for lower levels of activity. In addition, the national reach of community development could end up diverting community development financing from areas with significant retail lending. The chance of successful neighborhood revitalization increases when a bank focuses retail and community development financing in the same areas. In contrast, NCRC would build upon the current regulatory approach that allows banks to serve a statewide or regional area outside of their assessment areas with community development financing provided they satisfy needs in their assessment areas. In addition, NCRC would allow banks to receive credit if they conduct community development financing in underserved counties that are identified as economically distressed areas with low levels of lending and investing.
The Penn Institute papers suggest that careful reform of the CRA regulations can build upon the progress in lending and investing if the reforms are incremental instead of “transformational,” which would feature a one ratio that would distort CRA outcomes. These incremental reforms plug gaps in the existing CRA regulation and updates CRA to take into account changes in the industry. In contrast, a one-ratio-focused CRA exam would take CRA in the wrong direction and render it ineffective in motivating banks to meet local community needs, which is the purpose of the statute.
A version of this article originally appeared on the NCRC blog.