As we get closer to proposed changes to the Community Reinvestment Act (CRA), a sober assessment of the effectiveness of various parts of CRA exams will help us react to the proposed changes. The discussion regarding performance measures on CRA exams becomes so overwrought at times that it almost seems like CRA exams are so subjective and devoid of performance measures that they are not useful.
The retail lending test, however, has performance measures that make sense. The shortcoming is that they are not applied consistently because of a lack of guidelines. Sharpening how banks are scored on these performance measures could very well do the trick in terms of effective reform.
Increasing clarity is preferable to a “one ratio” approach discussed by the Office of the Comptroller of the Currency (OCC) that would consider performance of all CRA activities in one performance measure. As described previously, a one ratio approach would be opaque in that a reader of an exam would not know how a bank performed in each activity. A multifaceted test with clearly described performance measures is a better way to realize CRA’s goal of measuring a bank’s responsiveness to credit needs because such a test shows in which activity a bank is meeting needs and in which it needs to improve. A multifaceted test increases the chances of collaboration with community stakeholders since it visibly shows where a bank needs help. In order to develop a robust multifaceted CRA exam, let us first consider an existing test, the lending test, that counts for half of the overall rating.
Here are two current performance measures on the retail test:
- Lending to low- and moderate-income (LMI) borrowers or very small businesses
- Lending in LMI tracts
In lending to LMI borrowers, the following is often displayed for various types of home loans like home purchase or refinance lending:
- Percent of a bank’s loans to LMI borrowers
- Percent of all lenders (in the geographical area) loans to LMI borrowers
- Percent of households in the area that are LMI
This analysis is often done separately for low-income and moderate-income borrowers.
Here are some examples from CRA exams of how this analysis is conducted:
Home lending example from a CRA exam – lending to LMI borrowers (slightly modified from the exam)
Home mortgage loans to low-income families at 20.5% is higher than the 19.8% of low-income families and significantly higher than the 6.4% aggregate distribution. The proportion of home mortgage loans to moderate-income families at 23% is higher than the proportion of moderate-income families at 18.2% and aggregate performance at 17%.
A common shortcoming as shown by this example is that a rating or score is not assigned to each result. Instead, adjectives such as excellent or good are often used. In order to increase consistency, exam guidelines can establish what score or rating corresponds to the performance. In this example, for low-income borrowers, the bank made a higher percentage of loans to low-income borrowers than the percent of low-income families and also made a higher percentage of loans to low-income borrowers than all lenders in the area (referred to as aggregate). In this case, the bank is above the demographic measure (percent of low-income families) and the peer measure (percent of all bank loans). The bank could earn an “Outstanding” on the measure of lending to low-income families.
Often times, however, a bank will issue a percent of loans to low-income borrowers that is below the demographic measure. This occurs because a portion of low-income households are in poverty or have imperfect credit and are not able to purchase homes. If a bank is above the peer measure but below the demographic measure, it could have a rating of “High Satisfactory.” In another instance, a bank could be below the demographic measure and be equal to the percent of loans made by all lenders. In this case, it could receive a rating of “Satisfactory.” The three lowest ratings that occur on CRA exams are “Low Satisfactory,” “Needs to Improve” and “Substantial Noncompliance.” A bank would receive one of these ratings if it was lower than the demographic and peer measures. A bank would earn “Low Satisfactory” if it was a couple of percentage points lower than its peers in terms of percent of loans to low-income borrowers, and could earn one of the lower two ratings if its percentage of loans to low-income borrowers was half or less of its peers.
Small business lending example in a CRA exam – lending to very small businesses
The bank’s distribution of small loans to businesses with gross annual revenues of $1 million or less is good. The bank did not collect or consider the gross annual revenues in the underwriting of approximately 35.4% of its small loans to businesses. Based on businesses with known revenues, the proportion of the bank’s small loans to businesses at 53.4% is lower than the 77.9% of businesses with gross annual revenues of $1 million or less. However, considering the bank’s distribution is higher than the 49.6% for aggregate lenders, overall performance is good.
On CRA exams, small businesses are defined as those with revenues less than $1 million. Using this definition of a small business, the exam says the performance is “good” as the bank made a percentage of loans higher than all lenders in the area but considerably lower than the percent of small businesses in the area. A rating of Satisfactory would be reasonable in this instance since the bank is within a few percentage points of its peers but both the bank and its peers are so far away from the demographic measure of the percent of businesses that are small. High Satisfactory could be reserved for banks that are perhaps within 10 percentage points of the demographic measure, and Outstanding could be reserved for banks that are closer to meeting the demographic measure. Analysis of the distribution of banks of similar asset sizes on both measures of performance could also inform the ratings awarded.
Another issue is the high percentage of loans for which the bank did not use the revenue size of the small business in its underwriting decision. Perhaps, the examiners would reserve High Satisfactory and Outstanding for the banks that have a smaller portion of loans with revenue unknown in order to encourage better data collection and prudent underwriting.
Home lending to LMI tracts in a CRA exam
The distribution of the bank’s home mortgage loans in low-income geographies at 5.7% is greater than the 3.5% of owner-occupied housing units in low-income geographies and it is higher than the 2.9% aggregate distribution. The distribution of home mortgage loans in moderate-income geographies at 18.3% is higher than the 16.1 percentage of owner-occupied housing units in moderate-income geographies and aggregate performance of 14.5%.
In this example, the percent of the bank’s loans in low-income tracts and in moderate-income tracts exceeds both the demographic measure and the peer measure. The exam called this performance excellent; I would use the word “Outstanding.” The demographic measure is the percent of owner-occupied housing units in low-income or moderate-income tracts. In other examples, the ratings would correspond to performance on the two measures in a manner similar to that described above.
Small business lending in LMI tracts – an example from a CRA exam
The proportion of the bank’s small loans to businesses in low-income geographies at 4.9% is lower than the 6.2% of businesses operating in low-income geographies, but higher than the 4.7% aggregate performance. The proportion of the bank’s small loans to businesses in moderate-income geographies at 15.7% is lower than the 18.6% of businesses operating in moderate-income geographies and lower than the 16.6% aggregate performance.
The CRA exam described this performance as good. I would use the rating category of Satisfactory instead. The bank’s percent of loans in low-income and in moderate-income tracts was lower than the demographic measure. In the case of moderate-income tracts, the bank’s percentage of loans was also about one percentage point lower than its peers, essentially a tie. In low-income tracts, the bank was also tied with its peers in terms of the percent of loans to small businesses.
Summing up the performance measures
Each performance measure such as lending to LMI borrowers has two sub-performance measures, which are lending to low-income and moderate-income considered separately. In home lending, this exercise is often executed three times, once each for home purchase, refinance and home improvement lending. Thus for one geographical area or assessment area, the home lending retail test has six scores or ratings. If a score of one to five corresponds to each of the ratings of Substantial Non-compliance to Outstanding, the score can be tallied for home lending in an assessment area to produce an overall rating for retail home lending. A similar procedure can be followed for small business lending. The scores for both home lending and small business lending can be summed to produce the overall score for retail lending. If home and small business lending are equal parts of the bank’s lines of business, a simple average can be computed, which would be the overall rating for the retail test. However, if one or more types of lending is a greater part of a bank’s business line, the overall rating can be a weighted average of the ratings for the different products. The examiner would describe any weighting in the exam narrative.
The ratings for each type of loan and overall for each assessment area should be displayed in clear tables. In this manner, the general public can clearly identify strengths and weaknesses in bank performance and help the bank address its weaknesses.
Currently, some summary performance information is provided in exams but its presentation can be improved. Here is one example:
Overall geographical distribution is good. 27 full scope Assessment Areas (AAs) – 11 are good, 15 are adequate and one is poor. Distribution among borrowers is good. One AA is excellent, 18 are good and 8 are adequate.
In exam-speak, the adjective excellent corresponds to Outstanding; good to Satisfactory; adequate to Low Satisfactory; and poor to Needs-to-Improve. What would be immensely helpful to a reader is a table showing which AAs have which levels of performance. This would make the CRA exam more useful in stimulating bank efforts to bolster performance where warranted and in motivating collaboration with community stakeholders. A major objective of CRA exams is to foster collaboration between a bank and the community to better address community needs.
Flexible and Innovative
Another criterion on the retail lending test is often referred to as flexible and innovative. This criterion measures the responsiveness and innovation of a bank’s retail lending in meeting community needs.
The following example from a CRA exam provides insight into how performance under this criterion in evaluated:
The bank further offers a proprietary affordable mortgage product targeted toward low- and moderate-income people. This in-house program is the Community Homeownership Incentive Program (CHIP). CHIP guidelines provide for low down payments, homeownership counseling, flexible credit criteria and long-term fixed rates. During the evaluation period, the bank made several enhancements to the CHIP program to make it more attractive to a broader base of applicants. In addition, the bank began allowing existing CHIP borrowers to refinance their loans through a new option, CHIP-to-CHIP. Bank management became aware that many existing CHIP borrowers were unable to take advantage of the low interest rate environment and refinance their loans due to inadequate equity to refinance to a conventional loan. Therefore, in late 2016, bank management introduced the CHIP-to-CHIP refinance option.
The number of loans originated through these programs represents approximately 18 percent of all HMDA loans originated during the evaluation period. This represents excellent performance in innovative and flexible mortgage lending.
This narrative is one of the better ones I have seen in a CRA exam. The narrative describes a bank’s own proprietary product that does not involve a government guarantee and which the bank subsequently modified in order to respond to community needs. In addition, homeownership counseling appears to accompany the origination of a CHIP loan (though we do not know the extent of counseling or its frequency from the exam narrative). The narrative also quantified the percent of products that the examiner considered innovative and flexible. While 18% seems high, CRA exams do not calculate the percentages of flexible and innovative products on a consistent enough frequency to make peer comparisons that inform judgments. In addition, an exam table reveals about 15,000 innovative and flexible loans. While that is commendable, about 80% of these loans were government-insured loans, not the bank’s CHIP product.
Overall, it would seem that a reasonable score would be Satisfactory on this performance measure. The bank developed its own product and seemed to have a high percentage of flexible and innovative loans. However, if we knew that its overall percentage of flexible products exceeded peer percentages, we could more confidently consider a higher rating. Moreover, the use of government guarantee loans does generally increase the percentage of lending to LMI borrowers but these loans are more expensive. In order to achieve a High Satisfactory or Outstanding rating on this performance measure, a bank would have a higher percentage of its own flexible and innovative products. How much higher would be informed with peer comparisons which current CRA exams do not provide. Hence, it is apparent where reform should go on this measure; we need more consistent data collection for this performance measure and ideally a creation of an industry-wide database so peer comparisons can be more readily made on CRA exams.
Currently, another shortcoming on CRA exams is that it is unclear the degree to which the innovative and flexible criterion influences the overall rating. It would seem a weight of about 20% of the overall rating on the retail lending test could be warranted. This performance measure is now considered a qualitative measure, but it can become more quantifiable as discussed here. It is important in that it can show either overall or for specific assessment areas how a bank is specifically meeting community needs. Yet, a very high weight approaching 50% is not warranted because the measure is still under-developed and may not be fully developed after this round of rulemaking. In addition, if a bank has highly innovative and flexible products, it will tend to perform on a High Satisfactory or Outstanding manner on the other criteria previously discussed.
As this analysis demonstrates, enhancing the current performance measures on the retail lending test is both feasible and informative regarding the extent to which banks are meeting needs. This incremental approach is the best at maintaining clarity and precision regarding which needs a bank responds to in a satisfactory or above satisfactory manner and which needs the bank should make more of an effort to meet.
Josh Silver is NCRC’s Senior Advisor.