|Remarks by Governor Edward M. Gramlich
At Widener University, Chester, Pennsylvania
November 6, 1998
Examining Community Reinvestment
The Federal Reserve does many things, some well known and some not so well known. Our well known function is to set monetary policy, or the overall level of interest rates. That is not what I plan to talk about today. What I would like to talk about is one of the lesser known things we do, which is to administer the Community Reinvestment Act (CRA).
This act was passed back in 1977 with relatively little fanfare. Based on the charters of banks and savings associations, which say that these financial institutions should meet the convenience and needs of the communities they serve, the CRA encourages these financial institutions to help meet the credit needs of their entire communities, including low and moderate income borrowers.
While for many years the CRA was little known to those outside the banking business, recently it has become the source of much more attention and controversy. Some see CRA as one way to aid low and moderate income groups living or working in poor areas. Others see the same act as an overly-bureaucratic and distortionary credit allocation scheme. Provisions involving CRA became a major source of contention in this year's political debates on credit union membership, financial modernization, and regulatory relief. What I would like to do today is provide a summary of evidence to date regarding CRA issues and propose several avenues of further study.
The CRA is administered by the regulatory agencies already responsible for the relevant financial institutions --the Federal Reserve Board (FRB), the Office of Comptroller of the Currency (OCC), and the Federal Deposit Insurance Corporation (FDIC) for banks and the Office of Thrift Supervision (OTS) for savings and loan associations. These agencies give financial institutions performance ratings ranging from "outstanding" to "substantial noncompliance," based on their provision of lending, investment, and retail banking services to low and moderate income groups in their assessment areas. The agencies administer common CRA regulations and meet periodically to try to standardize their examination criteria, and indeed the distribution of ratings looks remarkably similar across the regulatory agencies. All ratings are made public and the ratings are an important component of regulatory assessments of proposed mergers and acquisitions. It is this possibility that forms the real teeth of the CRA law.
Some evidence shows that CRA appears to be a highly effective federal government program in dealing with the credit needs of low and moderate income groups. For small business loans, in 1997 the 1,900 large banks and savings and loan associations subject to detailed CRA reporting requirements made $159 billion worth of new small business loans, two-thirds of the national total for such loans. A significant share of these loans went to low and moderate income areas or groups. There were another $11 billion for small farm loans, one-fifth of the national total for these loans, and another $18 billion for community development loans.1 While many of these loans would presumably have been made without CRA, the size of the gross loan numbers and their distribution across geographical areas suggest the importance of CRA in the process. There is also a great deal of anecdotal evidence, contained in periodic reports of the Federal Reserve Banks, of the success of various CRA community lending programs.
CRA also covers mortgage lending and there is evidence that it contributed to a big increase in mortgage lending to low and moderate income groups in recent years. Mortgage lending has its own detailed government reporting requirements. Most major mortgage lenders are subject to these requirements, and they can be prosecuted for persistent patterns of discrimination under the fair lending laws. The Home Mortgage Disclosure Act (HMDA) requires financial institutions to report the race, income, gender, and census tract of all applicants and borrowers, along with details of the disposition of the application. Using these and other data, the Justice Department can bring discrimination suits. In addition, the institution's mortgage lending practices are considered in its CRA exam.
In 1997, the 8,000 banks and savings and loan associations subject to these more detailed mortgage lending reporting requirements made over 700,000 mortgage loans to low and moderate income individuals. For a time in the early 1990s, the growth in mortgage loans to low and moderate income homeowners and in low and moderate income areas was far greater than for other mortgage loans. In recent data the differences in growth rates have narrowed, but there has still been a sizeable upward shift in the share of mortgage loans received by low and moderate income groups and by minority groups.2 As with small business lending, it is likely that CRA played an important role in bringing about this shift.
But CRA does have its critics. Banks have complained about getting "held up" by community groups when they try to merge or acquire other banks. In the early 1990s a Senate Task Force named CRA as one of the ten most burdensome regulations. One observer has detailed the case, arguing that the program consists of a costly set of regulatory procedures that seem unwarranted in terms of basic lending discrimination, and that impose big reporting and other costs on financial institutions, whether they discriminate or not.3 While many CRA critics would not go this far, even the strongest advocates of the program admit that systematic evaluative evidence on CRA is very thin.
There are at least four important evaluation questions that could be raised about CRA.
Is there Systematic Lending Discrimination?
But there could be several different types of distortions in credit markets. There could be plain old lending discrimination against racial groups, ethnic groups, women, or lower income groups. There could also be what is known as statistical discrimination against these groups, under which lenders without full information on the repayment prospects of borrowers use racial, gender, or income information as proxies for loan repayment possibilities, hence inappropriately denying credit to some worthwhile minority or low income borrowers. There could be what are known as cultural affinity problems, under which loan officers are less likely to approve loans to minorities or low income groups. Or, there could be redlining and other forms of geographical discrimination under which lenders feel that loans in certain neighborhoods will racially "tip" the neighborhoods, lower property values, and lower repayment prospects for these and other loans, thereby reducing access to credit in these neighborhoods.
If the problems in credit markets were based on outright discrimination, cultural affinities, or some kind of misinformation, either about the repayment prospects of borrowers or the tipping propensities of neighborhoods, the CRA could in principle help remove these distortions by encouraging financial institutions to make loans they otherwise would not. In this event both the equity and efficiency of credit markets would improve, the latter because both lenders and borrowers might benefit from the removal of arbitrary but misinformed distortions.
Similarly, if there were externalities that impeded lending in certain areas, CRA might also generate joint benefits to borrowers in these neighborhoods, others with a stake in the neighborhoods, and to the creditors themselves. CRA might ensure that many lenders would lend to an area, keeping the area economically viable and assuring any one financial institution that its own loans would have reasonable repayment prospects. The business lending reported under CRA might do likewise, opening up new business possibilities in urban or rural areas that otherwise would have remained blighted, raising property values, and creating new jobs.
A necessary condition for CRA to prove beneficial is the existence of some or all of these types of distortions, either in mortgage lending or in business development lending. Due to the paucity of data, a point I will discuss shortly, very little is known about business lending. For mortgage lending, most existing information is about racial discrimination, which raises slightly different questions than the income discrimination that is the predominant focus of CRA. On racial discrimination, Ladd summarizes many years' past work that finds substantial evidence of such discrimination, though with large uncertainties and research gaps.4
A great deal of attention was given to a study organized by the Federal Reserve Bank of Boston that supplemented the normal HMDA data with a series of other variables on the characteristics of mortgage applicants and neighborhoods. This study found mortgage denial rates in 1990 that were higher for minority groups than for whites, though gross disparities in denial rates are cut substantially when researchers control for a long list of applicant characteristics not reported under HMDA, and also for interactions between financial characteristics and race.5 The most recent HMDA data continue to show large disparities in denial rates. The 1997 data give a ten point racial disparity in denial rates for conventional home purchase loan applications. For prime lenders such as banks, denial rates are on the order of 20 percent for blacks and 10 percent for whites. For subprime lenders, denial rates are on the order of 65 percent for blacks and 55 percent for whites.6 Presumably controlling for other characteristics would cut into, but perhaps not eliminate, these disparities.
It is also possible to identify discrimination through repayment records. As the theory of discrimination suggests, if only the more qualified minority borrowers get credit, minority repayment records should exceed those of whites. On a cross section basis, minority repayment records are not quite as good as for whites, though the disparities are small and are again cut further by careful controls for the characteristics of borrowers.7 Because it does not suggest that only the most worthy minority borrowers receive credit, these cross section results do not provide much evidence of discrimination. But time series evidence may argue the other way, since the rapid growth in loans to minority borrowers has not been accompanied by a rise in delinquency rates or a drop in lender profit rates. This time series evidence suggests that CRA or other mortgage lending restrictions may have broken down some discriminatory distortions in credit markets without hurting lenders.
Is the CRA Law Structured Appropriately?
The proof of the pudding is in the eating. How much leverage does this provision give the regulatory agencies?
Over the 1993-97 period, one regulatory agency, the Federal Reserve Board, considered an average of 1,100 merger cases a year where a merging institution was subject to CRA. Based on the above ratings data, one would expect about 90 percent of these mergers to pass (the square of the above implied pass rate) without a detailed CRA review. In fact about 93 percent passed.
But in about 70 cases per year, potential CRA problems were raised. Sometimes there were CRA protests raised in writing or at public meetings held by the Fed to discuss these mergers. Sometimes the issue involved the CRA ratings themselves. Occasionally the issue was an alleged fair lending problem or some other compliance matter.
The Board staff analyzed the cases, submitted a recommendation to the Board, and the merger was either approved or denied. During this period there was an average of only one denial per year based primarily on CRA, but in another seven cases per year, the merger application was withdrawn, sometimes for CRA related reasons. In still other cases indifferent CRA ratings could have been a contributing factor to a merger blockage.
On the surface it seems like this CRA test is not very strict. In less than ten of 1,100 cases was a merger disapproved because of CRA. But this superficial impression may be incorrect. Many banks with low CRA ratings might not even have applied for a merger, and CRA could have contributed in subtle ways to other blockages. Beyond that, the CRA law might be very effective in turning the attention of financial institutions to serving their low and moderate income groups in a way that can never be captured by statistics. Combining these numbers with the significant loan figures given earlier, this method for enforcing the CRA law might in fact be very effective, at least for the large number of banks intending to participate in mergers.
The Properties of CRA Loans
To investigate the first question, it would be desirable to have time series data for a series of financial institutions. The growth of a particular bank's lending to low and moderate income individuals could then be measured before and after the introduction of CRA, or in comparison to lending patterns of institutions not subject to CRA. There are no data for conducting such studies, but LaCour-Little has an approach that comes close. He analyzed confidential data for one large but anonymous mortgage lender.8 This lender engaged in credit scoring, a statistical method for screening loan applicants, and this enabled LaCour-Little to tell, with reasonable precision, what loans would have been made with and without CRA. His analysis indicated that just about half of the mortgage loans to low and moderate individuals living in low and moderate income census tracts would not have been made under the more rigorous credit-scoring standards used for other mortgage loans. In this sense about half of the loans credited to CRA should in fact be so credited. Were this propensity to be applied to other lenders, it suggests that the gross loan figures that might be attributed to CRA are overestimates, but that the law may still be quite important in breaking down discriminatory distortions.
LaCour-Little can also measure other magnitudes with the same data. He finds a slightly higher foreclosure rate for likely CRA mortgage loans than for non-CRA mortgage loans with similar loan-to-value ratios, 2.3 percent for CRA loans as opposed to 0.4 percent for non-CRA mortgage loans. This finding echoes that from a survey of its members conducted by the Consumer Bankers Association.9 It is consistent with earlier evidence against wide-scale discrimination, but not very emphatically because these foreclosure rates are still low and not dissimilar racially.
Other relevant data come from a survey given by the Consumer Bankers Association. The survey shows that an average of 93 percent of respondents had Affordable Mortgage (AM) programs in the 1992-95 period. Of those financial institutions that had AM programs, 93 percent worked through lower downpayments than would otherwise be required, 72 percent waived mortgage points and 46 percent subsidized the mortgage rate. To the extent that these reported implicit subsidies lowered mortgage loan profitability, 79 percent of the institutions said that the subsidies were funded internally. These data suggest pretty strongly that many of these mortgage loans would not have been made without CRA.
A final way of checking these results is by asking banks explicitly about the profitability of CRA mortgage loans. This was done by Meeker and Myers who conducted a survey of large residential mortgage lenders.10 Considering everything, the higher risk of nonperformance, the lower risk of prepayment, and the higher administrative costs, the Meeker-Myers survey indicates that 98 percent of mortgage lenders found CRA loans profitable, though only 2 percent of mortgage lenders found CRA loans more profitable than other mortgage loans and another 22 percent found CRA loans as profitable as other loans. Thus while not all lenders found CRA profitable in an opportunity cost sense, virtually all found it profitable in an absolute sense. There remain the sizable community relations benefits for banks in participating in CRA loans.
The upshot of all of this information is that CRA does seem to have generated a large amount of new loans, though less than those officially recorded. Lenders report that there does seem to be some subsidy on some CRA loans, partly because of the slightly worse repayment rate, partly through explicit subsidy terms such as lower downpayments but they also say that these loans are profitable in an absolute sense. There seems to be little doubt that most of these outcomes would not have occurred in the absence of CRA and other fair lending laws. All of these conclusions, however, are based on fragmentary data, and it would certainly be productive to conduct some new surveys to verify the results.
The Long Term Impact of CRA
One can imagine several ways of attacking the issue. For one thing, some lenders are subject to CRA and some not. It should be possible to compare time series data for the different types of lenders, to measure gross differences in loans. Schwartz has conducted this type of test, by looking for differences between banks that have and have not signed CRA agreements with local community groups.11 While this test may suggest especially good performance by banks that have gone beyond minimal CRA requirements, it is not a true test of the CRA requirements themselves because all banks are subject to CRA, whether or not they have signed loan agreements.
Moreover, the fundamental evaluative questions do not involve gross loans, which seem to be substantial for CRA, but the true economic impact of the loans. For urban development loans many CRA projects have followed the practice of saturating particular areas with loans. Visually, urban areas with many CRA investments simply look different from other areas, with nicer living quarters, more commerce, better parks. But these superficial impressions are just that. It would be desirable to buttress them with formal comparisons of property values or other measures of urban vitality, across census tracts that have and do not have CRA exposure.
Similar comments could be made about mortgage lending. The HMDA data could be studied and expanded on further, as was done in the Boston study. There could also be a tighter geographical examination of mortgage data, to see whether CRA and other mortgage-inducement programs have had an impact on the distribution of mortgages by neighborhood or census tract.
Another potential area of inquiry involves financial strategies. One of the welcome innovations with CRA is that the various financial institutions and community groups have devised new lending arrangements, featuring lending consortia, loan pools, specialized lending units, micro-enterprise loan funds, and many other types of arrangements. Each of these types of arrangements differs in the responsibilities of entrepreneurs and the risks borne by lenders. Undoubtedly some arrangements are more successful than other types of arrangements. One could study these arrangements, their strengths and weaknesses, and perhaps learn about more and less successful ways of intervening.
1 Press release on the availability of 1997 Reinvestment Act data, Federal Financial Institutions Examination Council (FFIEC), August 24, 1998.
2 Memorandum, Robert Avery and others to the Board of Governors of the Federal Reserve System, August 21, 1997.
3 George J. Benston, "Discrimination in Mortgage Lending: Why HMDA and CRA Should be Repealed," Journal of Retail Banking Services, vol. 19 (Autumn 1997), pp. 47-57.
4 Helen F. Ladd, "Evidence on Discrimination in Mortgage Lending, Journal of Economic Perspectives, vol. 12 (Spring 1998), pp. 41-62.
5 Raphael W. Bostic, "The Role of Race in Mortage Lending: Revisiting the Boston Fed Study," Finance and Economic Discussion Series 1997-2 (Board of Governors of the Federal Reserve System Divisions of Research and Statistics and Monetary Affairs, January 1997).
6 Press release on availability of data on 1997 mortgage lending activity, FFIEC, August 6, 1998.
7 Ladd, 1998.
8 Michael LaCour-Little, "Does the CRA Make Mortgage Credit More Widely Available? Some New Evidence Based on the Performance of CRA Mortgage Credits" (unpublished paper, Citicorp Mortgage, Inc., May 4, 1998).
9 Consumer Bankers Association, Affordable Mortgage Program: An Analysis of the July 1995 Survey (1995).
10 Larry Meeker and Forest Myers, "Community Reinvestment Act Lending: Is it Profitable?" Journal of Financial Industry Perspectives (December 1996), pp. 13-35.
11 Alex Schwartz, "Bank Lending to Minority and Low-Income Households and Neighborhoods: Do Community Reinvestment Agreements Make a Difference? Journal of Urban Affairs, vol. 20 (1998), pp. 269-301.