6
2. Where applicable, the counts of mortgage, small business, and small farm loans made in each assessment
area were collected.
– Loan count data were collected from the Geographic and/or Borrower Distribution tables. In the first
vintage, low- and moderate-income loan counts were combined into one “LMI” value.
PE Vintage 2
There were several changes made to the data collection efforts for the second vintage of PE data (2018–19). In
addition to the data collected in the first vintage, the following data were also extracted:
1. Total inside and outside assessment area counts and amounts for each loan product reviewed as a part of
the performance evaluation.
2. Assessment area loan count data were collected from the Geographic and/or Borrower Distribution tables
using all of the available income breakouts (low, moderate, middle, upper, unknown, <=$1M GAR, >$1M
GAR).
3. Community development loan and investments made outside of an assessment area were recorded
separately from loans assigned to an individual assessment area. The integration of these Retail Loan and PE
data constitutes the Merged Data Table.
Sampling Procedure
For each vintage, the PEs were selected using a stratified random sample method. The first vintage of PEs was
drawn from evaluations conducted from 2005 to 2017, and the second vintage was from 2018 to 2019.
Evaluations cover bank performance over a period of years preceding the evaluation start date. The sample
therefore, covers the market during a boom period, a severe recession, and then an extended recovery.
Sampling was stratified by the existing three bank asset threshold categories: small, intermediate small, and large;
the agency which performs the performance evaluation (Office of the Comptroller of the Currency (OCC), Federal
Deposit Insurance Corporation (FDIC), or the Federal Reserve); and whether or not the bank received a rating of
“Needs to Improve” or “Substantial Non-Compliance.” Historically, ratings below a “Satisfactory” have been
relatively uncommon. Staff, therefore, oversampled banks with “Needs to Improve” or “Substantial Non-
Compliance” ratings to obtain a sufficient sample size of these performances. PEs were grouped by year,
regulator, bank size, and the performance category described above. PEs were randomly selected from each
group for data collection.
The first vintage of PEs contained approximately 22,000 PEs with an evaluation start date from 2005 through 2017
available on the websites of Federal Reserve Board (the Board), the FDIC, and the OCC. Of these, staff selected
approximately 6,300 PEs from which to collect data. In the second vintage of PEs, staff collected data from 600 of
the approximately 2,900 PEs available from the 2018–19 timeframe.
Additional Notes on Performance Evaluations
• The scope of information on performance ratings varies across agencies.
Board staff had access to ratings for each test at the state and multistate MSA level for Federal Reserve
System (FRS) banks, to ratings for each test at the individual institution level for FDIC banks, and to overall
institution level ratings for OCC banks. To address this variation in availability of digitized information, the
following evaluations were oversampled:
– FRS state member bank evaluations that received a “Needs to Improve” or worse on a lending test
rating (or overall rating, for small banks) for any state or multistate MSA;