In my bank’s performance context (high-level, currently), should I be comparing the geography income level of where our branches and our peer branches are located to the geography income level of the tracts within the assessment area or to the population income level, or both? For example in the assessment area, 8.75% of the tracts are low- income tracts, however the percent population that is considered low- income is 7.47%. Thanks.

Exams look at tract income class for branches. You could look at both but exams are considering how many branches are in low and moderate income tracts.

Actually if you read the large bank examination procedures you see that phrase, "distribution of the institution's branches among low-,, moderate-, middle- and upper-income geographies in the institution's assessment area(s)" but no real clear definition of how that relationship is computed.

However, if you read the CRA PE's for large banks you will see that the examiners are comparing the population distribution among census tracts by income class to the branch distribution among those tracts.

For example, the following was taken from Wells Fargo's CRA PE:
Retail Banking Services
Refer to Table 15 in the Multi-state Metropolitan Areas section of Appendix D for the facts and data used to evaluate the distribution of the bank’s branch delivery system and branch openings and closings.
The analysis of WFB’s branch distribution is primarily based on the geographic distribution of branches among low- and moderate-income geographies in relation to the proportion of the population residing in those geographies. (emphasis added)

So the examiners do compute the test results using the population distribution and there is a standard table (number 15) with the demographic data and the branch data juxtaposed in the CRA PE

However, when you read the narrative section, it states:

"The Service Test analysis focuses on WFBs distribution of branches by income level of the geography. However, there were numerous instances where a branch located in a middle- or upper-income census tract was actually right on the border with a low- or moderate-income tract; literally across the street in many cases, or within a city block in others. In those instances, the middle- or upper-income branch provides ready accessibility to the banks financial services to both income areas. Accordingly, our analysis considered the benefit these bordering branches provided to their respective communities. Finally, accessibility to branch offices as well as branch hours and services provided the most weight when determining Service Test conclusions. Changes in branch locations and community development services received a lesser amount of weight to overall conclusions."

That is exactly why I asked the question. Looking at a previous PE, the examiner compared ATM and Branch locations by geography income level to the percent of population within those geographies.

TM - that's why I cited the exams. The wording in the Large Bank Performance Examination procedures you know that the tract income class demographics are used but it is not exactly clearly explained how they are used. The only way to find out is to see what the examiners are really doing in the field - and what they are doing is not using the relative percent of each tract income class (which they do use for Lending Test 3) but rather using the relative percent of the population in each tract income class to determine how many people are located near the branches.

As cited by KB they may temper the computation with a consideration of branches that are not in LMI tracts but close by. But again, that consideration is not subject to a clear computation. It appears to be more or less an afterthought - an acknowledgement that branches not in LMI tracts may be serving the population in those tracts if the branches are close by the LMI tracts. But I have never seen that consideration translated into a calculation (although I have seen some PE's that cite the number of branches near a LMI tract and that may be serving some of that tract - but the benefit for the LMI tract population is vague and the impact on the computations in Table 15 does not seem to be affected).

Frankly, using the relative percent of the population rather than the relative percent of tracts by income class makes sense. For the Service Test (which is where this question originated) Regulators want to know how much of the population (i.e., "real people") in the AA is affected by the distribution of branches. They aren't so concerned about the impact on tracts so much for this test as they are the impact on people. Table 15 was developed to compute that impact. Since your question referred to specific computations it seemed to me that was what you were asking.

I would hazard a guess that Wells Fargo did their own analysis/performance context where the reviewed the location of each branch, especially those in middle and upper income tracts, to the proximity of low and moderate tracks, and provided that analysis to the examiner.

I sincerely doubt an examiner would go to that length on their own.

I did a Performance context and analysis for a small bank that did only one particular type of lending, but they did it nationwide. The bank had only one branch (their head office) and it was located in a sparsely populated non-MSA county which meant their Assessment Area was limited to that county. They originated less than 2% of their loans in their Assessment Area. The Bank got an Outstanding rating - the key was the Performance Context and analysis that pulled together a lot of information of the banks products, services, and community involvement.

PR - I agree that an examiner wouldn't dig that kind of information up. It really is the responsibility of the bank to present relevant performance context information to examiners. Unfortunately, too many bankers do the minimum and aren't prepared to provide this kind of helpful information.

The only time I have seen a bank pass their CRA exam with a very low AA ratio was when the OTS was around. They were the most flexible Agency with respect to CRA. On the contrary, I have seen banks threatened with a NTI rating if their business model results in an extremely low AA ratio (2% would almost always be a certainty for failure). The only concession I have seen from examiners regarding this issue is to give the bank a warning to either get an acceptable ratio or adopt a CRA Strategic Plan that reflects the non-traditional market model adopted by the Bank

This is an increasing problem in the industry since a growing number of banks are adopting non-traditional banking models marketing mortgages and SBA-guaranteed loans over wide areas far beyond the normal market for a community bank. We are advising at least 6 banks currently who are confronted with this very serious problem. Sometimes there are other options besides a CRA Strategic Plan, but I would not be comfortable advising a bank that examiners will be accommodating and understand their business model as a decisive performance context factor that would exculpate them from a very low AA ratio. A bank may get lucky and get a really understanding examiner, but I wouldn't bet the bank on it.

I recently migrated from an OCC-regulated bank to an FDIC-regulated bank.

So far, I think the FDIC has absorbed some of the "flexibility" of the OTS. For example, Title One (as opposed to the actual percent of children using the free and reduced lunch program) is not a practice the OCC would accept.

Still way too much subjectivity among the regulators.

Agree 100%. Too often banks do nothing for exam preparation. Lack of preparation is a decision and a dangerous one.