Creating a Scorecard for the CRA Service Test: Strengthening Banking Services Under the Community Reinvestment Act

POLICY BRIEF #96

Despite the longest economic expansion on record during the 1990s, 10 percent of all American families—including 25 percent of African-Americans and Hispanics, a quarter of all families with incomes under $20,000, and nearly half of all families moving from welfare to work—have no bank accounts. Without bank accounts, families often pay high fees to check-cashing services and other "fringe bankers" to conduct basic daily financial transactions. Banking status also has profound implications for families' long-term self-sufficiency and access to a broader set of financial services. Low-income people with bank accounts are more than twice as likely as their unbanked counterparts to hold savings or have a home mortgage, and are six times as likely to own a credit card.

But having a checking account is not the same as using credit wisely. Nationwide, millions of "underbanked" families who do have checking accounts, but live from paycheck to paycheck, frequently pay a high price when conventional banks are either unwilling or unable to meet their acute credit needs. The vacuum in consumer credit created by the withdrawal of most mainstream lenders from the small loan market is being filled largely by a growing number of companies offering high-cost "payday loans"—one- to two- week cash advances on a portion of the borrower's paycheck. Because of their short terms, triple digit interest rates, and ease of access, payday loans are turning financially fragile families into chronic debtors.

In a recent book, Savings for the Poor, (Brookings, 1999) Michael Stegman discusses how technological advances and public policies—particularly the delivery of government benefits through electronic funds transfer (EFT) and the creation of individual development accounts (IDAs) to help lower-income families build assets—have combined to create new market opportunities for financial institutions to profitably serve unbanked and marginally banked populations. Yet despite the dramatic cost-cutting potential of electronic banking and the availability of public funds for basic account and financial education programs, many banks have been slow to reach out to potential new customers.

The 1977 Community Reinvestment Act (CRA), originally enacted by Congress to combat credit "redlining" in low- and moderate-income neighborhoods, includes a "service test" that could provide powerful incentives to financial institutions for improving retail banking services for low-income unbanked and underbanked families who have difficulty accessing the types of financial services they need. But due to inadequacies in its implementation, the test has fallen far short of its original promise. Our statistical analysis suggests that pressure to help institutions earn satisfactory overall CRA ratings may serve to boost service test scores for "borderline" banks. Creating a more objective scorecard for measuring performance under the CRA service test could reduce grade inflation and increase banks' responsiveness to the needs of the banked and underbanked.

A Primer on the CRA Service Test

CRA establishes an obligation on the part of federally insured depository institutions to help meet the credit needs of their entire communities, including low- and moderate-income communities, consistent with secure banking practices. An inadequate record under CRA may be grounds for denying or conditioning an application, for example, to merge with another depository institution. Financial modernization legislation enacted in 1999 provides that financial holding companies and depositories are prohibited from entering new lines of business if the depository or any of its affiliates fail to maintain a satisfactory CRA record.

Under the regulations that implement CRA, federal regulators grade large banks (those with at least $250 million in assets) on three measures of compliance: lending, investment, and service (see "The Grading System," p. 3). Despite the fact that the service test rating receives significant weight (25 percent) in determining a bank's total CRA score, its impact to date on bank activities is unclear. This is partly because service test criteria cover a broad range of issues, and because performance data are not as readily available for statistical analysis as are performance data on bank lending activities. In addition, examinations to date indicate that the service test requirements are the easiest to fulfill; only eleven banks of more than 1,500 received less than "satisfactory" ratings on the service test between 1996 and early 2001.

The creation of the service test in 1995 was part of a larger overhaul of CRA regulations designed to address widespread complaints that examination standards were vague, burdensome, and left too much room for inconsistency among examiners. Of the three component tests, the service test remains the most eclectic and the least clearly defined. The lending test requires extensive quantitative spatial analysis of bank loans as to number and amount, geographic distribution, and borrower characteristics. The investment test focuses primarily on the amount of qualified community development investments. The service test, however, covers several different subjects and includes a broad mix of both quantitative and qualitative criteria. There are two prongs to the service test—retail banking services and community development services.

Retail Banking Services Examiners consider the "availability and effectiveness" of a bank's systems for delivering retail banking services, specifically:

• The current distribution of branches among low-, moderate-, middle-, and upper-income neighborhoods;
• The bank's record of opening and closing branches, particularly those serving low- and moderate- income (LMI) neighborhoods and individuals;
• The availability and effectiveness of alternative delivery systems serving LMI neighborhoods and individuals; and
• The range and degree of tailoring of services provided in low-, moderate-, middle, and upper-income neighborhoods.

Community Development Services Examiners focus on large banks' provision of financial services that have community development as their primary purpose. These may include the provision of low-cost deposit accounts, technical assistance to housing and development organizations, credit counseling and financial education, Individual Development Accounts (IDAs) and other savings initiatives, and electronic benefits transfer programs. For each such activity, examiners must determine: the extent of community development services provided, including customers' usage; and the innovativeness and responsiveness of the services. Innovation includes a determination of whether the activities serve low- and moderate- income populations in new ways or serve groups of new customers.

The Grading System

As in the lending and investment tests, service test ratings are divided into five categories: outstanding, high satisfactory, low satisfactory, needs to improve, and substantial noncompliance. On the service test, to receive an "outstanding" rating generally, an examiner must judge a bank to:

• Have delivery systems that are "readily accessible" to communities and individuals of different income levels;
• Have a record of opening and closing branches that has "improved the accessibility" of its delivery system, particularly for low- and moderate-income neighborhoods and individuals, to the extent that changes have been made;
• Provide services that are "tailored to the convenience and needs" of its assessment areas, particularly for LMI neighborhoods and individuals; and
• Be a "leader" in providing community development services.

However, none of the phrases in quotes are defined in any regulatory materials. Moreover, a bank's performance need not actually meet all of the criteria for a particular ratings category; exceptionally strong performance in one area may compensate for weak performance in another as long as overall performance is consistent with safe and sound practices and with the appropriate rating "profile."

In contrast to the individual component tests, the computation of banks' overall CRA ratings depends on a mathematical formula. Once a large bank has received its lending, investment, and service test scores, examiners use a point system to determine whether the institution receives an overall CRA rating of outstanding, satisfactory, needs to improve, or substantial noncompliance. The lending test is worth twice as much as the other exams, so that banks can earn a maximum of twelve points on lending and six points each on investment and service. Activities directly related to the provision of credit are further emphasized by requiring that institutions earn at least a "low satisfactory" (worth six points) on the lending test, in addition to receiving at least eleven total points, in order to receive a satisfactory CRA rating overall (see table 1).

<>Determinants of Service Test Scores

To learn more about how examiners grade under the service test, we analyzed scoring patterns on almost 2,000 CRA examinations conducted over the last five years under the new CRA regulations, and reviewed in-depth more than 100 of these examinations. Extremely high passage rates, and often inconsistent and superficial examinations that lead to unsupportable scores, suggest that the service test creates only minimal incentives for banks to increase their service performance, especially with respect to the provision of basic banking products, financial literacy programs, and other community development services.

Banks tend to earn the highest number of "outstanding" and "high satisfactory" ratings on the lending test, but the service test is easiest to do well on (i.e., it is associated with the highest proportion of satisfactory or better ratings). Specifically, only eleven banks out of 1,500 have received a rating of "needs to improve" on the service test in the past five years, and no bank has ever earned a "substantial noncompliance" rating on service activities. This 0.7 percent failure rate is slightly better than for lending, where about 1 percent of institutions have earned "needs to improve" or "substantial noncompliance" ratings since 1996, and significantly better than the 17.5 percent failure rate on the investment test.

Our comparison of the distribution of overall CRA ratings by regulator—Federal Reserve Board (FRB), Office of the Comptroller of the Currency (OCC), Federal Deposit Insurance Corporation (FDIC), and Office of Thrift Supervision (OTS)—suggests that not all agencies place the same weight on the various criteria used to develop the scores. For instance, while all four agencies are very unlikely to give out "needs improvement" ratings, the OCC tends to award fewer outstanding scores than the others.

We analyzed CRA service test scores using a multivariate statistical technique known as ordered logistic regression, or logit, to measure the likelihood that a bank's service test rating will improve by one grade level—for instance, from "needs to improve" to "low satisfactory" or from "high satisfactory" to "outstanding"—with a unit change in a given factor, such as asset size, while controlling for other variables, such as location, or regulator.

The results of our statistical analysis indicate that under-performing banks—those on the border between a "needs to improve" and a "satisfactory" rating overall—are more likely to receive higher service test scores than other institutions, including some with significantly better lending and investment records (and the higher scores that go with them). In fact, when banks performed so poorly on the lending and investment tests that they were in danger of receiving a "needs to improve" rating overall (e.g., a combined point score of seven), their service test scores were much higher than would otherwise be expected. The higher-than-expected service test scores often gave banks just enough cumulative points (11) to eke out a "satisfactory" rating overall. This suggests that some CRA examiners, consciously or not, inflate the service test scores of under-performers in order to help them get an overall passing grade.

This peculiar situation is depicted graphically in chart 1 for a typical bank. The predicted probability of receiving a "high satisfactory" service test rating (worth four points) jumps from below 5 percent for banks that earn a combined six points or less on their lending and investment tests, to approximately 80 percent for banks that earn seven points (specifically, a "low satisfactory" on the lending test and a "needs to improve" on the investment test). The probability then drops back down again as the combined scores on lending and investment increase beyond the point where a "high satisfactory" score on the service test is needed to pass.

Policy Recommendations

Given the demonstrated anomalies in service test scores and the subjective and varying criteria used to assess bank performance on the test, policymakers and regulators should carefully consider the following seven proposals to make the service test more performance-based. These reforms can be implemented under existing law, pursuant to the argument that stronger retail services help families enter the pipeline for mainstream credit services, which represent the primary focus of the CRA statute.

Evaluate alternative delivery channels based on effectiveness and availability. Banks should have the freedom to select among different delivery channels based on popularity with customers, competition with other financial service providers, costs, and other factors. Once a choice has been made, however, CRA regulations require these channels to be evaluated as to their availability and effectiveness in serving poorer neighborhoods and customers. The alternative channels should be evaluated based on actual usage rates that measure effectiveness and extent of service, and not be credited simply because they are in place.

Analyze consistently the size of low-cost account programs and other community development retail services. While federal regulators refused in 1995 to base the service test on general deposit growth, the CRA regulations state that community development services should be evaluated as to (1) the extent of the bank's service provision and (2) the innovation and responsiveness of the service in meeting community needs. Ensuring consistent application of these standards is particularly critical given widespread complaints by community groups that banks with low-cost products sometimes refuse to promote them or even to inform consumers of their existence. Simply inquiring as to the number of low-cost accounts maintained or opened by the institution, or the number of checks cashed for non-accountholders, during the current examination cycle relative to the number of LMI households in the assessment area would be a significant improvement over current examination methods, and would allow examiners to award more credit to banks that provide extensive services.

Evaluate transaction services for "unbanked" and marginally banked populations and traditional low-cost account programs using the same standard. Although there is some debate among community advocates over whether banks should receive CRA credit for check cashing and similar activities, examiners are in fact directed by federal examination procedures to consider whether particular community development services target new groups of customers. Giving greater attention to low-cost check cashing, money orders, bill payments, non-predatory, short-term credit products that would compete with payday loans, and other stand-alone transaction services could encourage banks to compete more directly with fringe bankers.

Develop specific benchmarks for the most common types of community development services to ensure consistency in evaluations. As with low-cost banking products, CRA regulations regarding the provision of non-retail community development services should be made more consistent. Too many evaluations simply provide a laundry list of activities without any analysis. Admittedly, using a rigid mathematical approach similar to that for branch distribution would be difficult with regard to community development services because there is no clear system for weighing one type of activity—an IDA program, for example—against another, such as technical assistance to a small business. However, implementing standard measures for common activities, such as hours of staff assistance to community organizations, or the number of financial education seminars offered and number of attendants at such seminars, would be relatively simple and would eliminate much guesswork for both banks and examiners.

Give service test credit to banks that sponsor independent evaluations of the effectiveness of their community development service programs. With regard to financial literacy programs, for instance, examiners sometimes report data on the number of seminars or participants but almost never provide information on how many "graduates" of those programs actually obtained loans, opened accounts, or otherwise improved their financial status. Tracking seminar graduates' behavior or otherwise evaluating the effectiveness of particular activities may not be feasible for every examination cycle, but it is important from a policy perspective to encourage institutions to structure their programs to have real impact on enhancing credit access. Providing credit for bank-financed evaluation programs would further this goal.

Base examinations of large banks on the number of accounts per census tract, as determined by the mailing address for account statements. Federal regulators rejected a proposal during the 1995 rulemaking process to base the service test on deposit growth, both because they feared it would require burdensome geocoding of deposit accounts and because they argued that such a requirement would not be consistent with the CRA statute's focus on credit services. Because banks have made extensive technological improvements in response to Year 2000 concerns and as a means of improving their own marketing and customer relationship programs, the compliance burdens related to the spatial reporting of account owners are far less taxing now than they were six years ago. Adding a deposit analysis would make the service test more performance-based by focusing attention on the level of financial services actually being provided rather than continuing to concentrate predominately on the channels available to deliver such services.

Change large bank examinations to make it easier for institutions with limited investment opportunities to shift their focus to providing more basic banking services. Adjusting the balance between service and investment activities is important for several reasons. First, many large banks are finding it hard to do well under the investment test, a fact that has been confirmed by former FDIC Chairwoman Donna Tanoue. Second, providing more flexibility in the tradeoff between investment and services would allow banks to improve their competitive positions by concentrating on their core operations of providing basic retail products and increasing deposits. Finally, greater weight for the service test would strengthen the incentives for banks to provide accounts, check cashing, financial education, and other services directly to unbanked and underserved populations. In return for greater service test weight, we propose that banks be expected to demonstrate significant outreach and actual service to unbanked and underserved

Conclusion

As federal regulators embark on a comprehensive review of CRA regulations in the next year to determine the effectiveness, consistency, and compliance burden of current examination standards, our research suggests that the service test should be strengthened. Creating stronger incentives for banks to reach out to underserved populations is consistent with the statutory purposes of the CRA, because basic accounts, savings, and financial education are critical for helping families achieve financial stability and prepare themselves to qualify for consumer, home, and business credit over the long run. Just as CRA lending requirements have helped banks recognize the market potential for home loans in low- and moderate-income areas, a strengthened service test would eliminate the kind of undeserved "grade inflation" illustrated here, and facilitate the development of new markets, products, and technologies to help banks provide profitable basic banking services for underserved populations.