When the new CRA rule was published in late 2023 it contained some very serious flaws that need to be corrected. In fact, some of those flaws are so bad that they left the Trump Administration with only 2 choices, significantly fix the flaws or repeal the 2023 Rule altogether. To those alternatives I add a third, simultaneously revert to the legacy CRA rule but with some minor changes that could vastly improve the effectiveness of the CRA performance evaluation system.
In my first article on the need to fix (or repeal) the 2023 CRA Rule I focused on the serious flaws in the CRA assessment area delineation requirements, explaining how they can lead to meaningless and even misleading performance ratings. In the second article, I point out important and substantial volumes of different types of credit important to the community that are excluded in the 2023 rule thereby undermining the rule’s value as a measure of “meeting the community’s need for credit”. In this article I will comment on how complicated (actually “convoluted” would be a more accurate description) the performance rating system in the 2023 rule is. This subverts one of the stated major goals of the 2023 rule, more transparent and objective performance metrics that would be understandable and useful to all users alike, the community, banks and examiners.
The implementation of an objective and quantitative performance rating system is highly commendable. But the performance rating system under the “Retail Lending Test” adopted in the 2023 CRA rule involves an exceptionally large number of data points, data calculations, data conversions, decision trees and multiple data aggregations that make the ratings derived from the foregoing activities difficult to understand. To be useful, a ratings system should not just deliver performance ratings, it should facilitate recognition of the underlying dynamics behind the results and allow a bank to determine how to “fix” any performance shortcomings. The complexity of the 2023 ratings scheme can be overwhelming and make interpretation of the results difficult if not impossible.
Overview of the ratings system in the 2023 CRA Performance Evaluation Scheme
The 2023 implements a performance rating system for large banks and Intermediate-small banks measuring (1) the Retail Lending Test, (2) the Community Development Financing Test or Community Development Test (at an ISB’s option), and for large banks only (3) the Retail Services Test, and (4) the Community Development Services Test. By far the most significant test is the Retail Lending Test that constitutes 40% of a bank’s “composite” performance rating. Of the 4 tests, it is the only test on which a bank must score at least a satisfactory performance rating. The other 3 tests are important, but for those tests a less than satisfactory performance rating does not dictate an overall less than satisfactory performance rating, unlike the Retail Lending Test. The Retail Lending Test is not only the most important test it is the most complex performance rating scheme in the new rule.
The Complex Computational and Data Conversion Burden
The computational burden based on the numerous types of steps, and sheer number of calculations, and many data conversions (after quantitave scores are calculated the results must be converted to various “supporting conclusions”), which are then converted back to weighted scores that are then converted to more “major product line” supporting conclusions, which are once again converted back to numbers and then weighted to produce a “recommended conclusion” for an assessment area.
All of these compuations and conversions are magnified by the number of markets for which the exercise must be applied. In the 2023 CRA rule there are 3 types of markets for which “large” (assests >=$2 billion) banks will be evaluated: Facility-based assessment areas, Retail Lending assessment areas, and the Outside Retail Lending Area, or “ORLA”. A bank may have 10 FBAAs and 5 RLAAs and an ORLA with 8 or 9 “component geographic areas”.
Adding to the complications is the requirement to assign scores based on best results when comparing loan activity to market and community benchmarks. A lender that compares poorly to the market can receive a satisfactory rating based on comparisons to the community benchmarks. Effectively, this involves a decision tree.
A review of the various retail lending tests indicates that, when there are 2 “major product lines”, there is a minimum of 160 steps, computations, and data conversions for each facility-based assessment area and each retail lending assessment area. The volume of computations can be substantially more for the ORLA because the ORLA can include a significant number of “component geographic areas” each of which includes computations identical to those done for the FBAAs and RLAAs. It’s very possible that more than 1,000 data manipulations will be needed for the ORLA itelf for some banks!
Potential Confusion
Adding to the potential confusion is the fact that some loans that don’t get measured in Retail Lending tests are included in the the volume metric to determine if a bank’s loan volume passes the Retail Lending Screen test. For example, all multifamily mortgages count for Retail Lending Screen Test but MF mortgages are not included in the “geographic” and “borrower” performance measurments in the Retail Lending Test. Open-end mortgages are included in the Retail Lending Screen Test but not in any of the geographic and borrower performance tests. It will be easy to confuse what loan types are included in one test but excluded from another test.
Another factor adding to the work and confusion is the aggregation of all the recommended conclusions at different geographic levels, including the FBAA, RLAA results at the state level and the institution level which requires more calculations, weighting and data conversions.
When a performance rating scheme is so complex the results can be difficult to understand for professionals, let alone the public that’s suppose to benefit from the “transparent” rating system. This is contrary to one of the main goals of the 2023 CRA rule.
Repeal and Replace would be the best approach
Repealing the 2023 CRA rule should be the first step on updating the CRA. The regulators would be best advised to simultaneously propose a modified CRA rule. Many of us know that only a few minor tweaks in the legacy CRA rule could vastly improve the CRA performance rating system, making performance evaluations more consistent and more comprehensive. Aside from vastly improving the CRA performance system, simultaneously proposing and adopting simple changes to the legacy rule will show the public and critics of the repeal that the Agencies are for improving the CRA, not destroying it. This is very important in gaining public support for the Agencies’ action.
Recap of the 2023 Rule flaws in terms of the performance rating system
- Many different types of data manipulation: data numerical comparisons, data conversions (from scores to “supporting conclusions” then back to scores and weighed scores back again to “recommended conclusions”, then aggregating scores over different markets to arrive at performance rating)
- The need for decision trees when computing performance (choosing best scores based on market benchmarks versus community benchmarks)
- More than 160 computations and data conversions in simple situations to hundreds and sometimes thousands of data manipulations and calculations
- Inconsistent use of different types of loans for various tests in the Retail Lending Test
When combined with the serious assessment area delineation problems and the omission of important loan types the incredible complexity of the Lending Test and Community Development Financing Test (the latter of which is not explained in this article), the case for repealing the 2023 CRA rule is overwhelming.
However, it is highly recommended that the proposed repeal of the 2023 Rule be matched with proposed improvements in the legacy CRA rule. It is not enough to be against something unless you have something better to replace it.
What can be done to improve and modernize the CRA?
There are some very simple improvements to the legacy rule that will be the topic in my fourth article about the 2023 CRA rule.
[View source.]
