How many FFIEC peer groups are there?

The FFIEC publishes peer group averages for roughly 38 commercial bank peer groups, plus separate groupings for savings institutions and credit unions. The exact count varies slightly by quarter as institutions move between asset bands.

How are FFIEC peer groups defined?

Peer groups are defined mostly by asset size and to a lesser extent by business model (commercial bank vs. savings institution) and geographic footprint. A $400M community bank in Missouri is grouped with other commercial banks in roughly the $300M to $1B asset band, not with JPMorgan and not with credit unions.

What is a percentile rank in a UBPR?

A percentile rank tells you what share of peer-group banks score below the value in question. A bank at the 75th percentile on ROA is outperforming 75% of its peers. A bank at the 20th percentile is being outperformed by 80% of its peers.

Why does the peer group matter?

Because comparing a $400M community bank to industry-wide averages is misleading. JPMorgan's NIM, efficiency, and capital ratios are structurally different from a community bank's. Peer-group percentiles compare each bank to other banks that look like it, which is what makes the comparison useful.

How do FFIEC peer groups work?

Short answer: The FFIEC sorts US commercial banks into about 38 peer groups, primarily by asset size. A $400M community bank in Missouri is grouped with other commercial banks in roughly the $300M to $1B asset band. Its UBPR ratios are compared to the median of that group, and its percentile rank tells you where it sits relative to peers that actually look like it.

Why peer groups exist

Banks of different sizes operate differently. A $200M community bank in rural Iowa funds itself mostly from local core deposits and lends mostly to local small businesses and farms. JPMorgan funds itself through wholesale markets and has trading desks. Comparing them on net interest margin or efficiency ratio is comparing apples to forklifts.

The FFIEC's solution is to slice banks into peer groups where the structural shape of the balance sheet is roughly similar. A bank inside a peer group still has plenty of variation around it, but at least the variation is meaningful.

How peer groups are defined

Three factors define the peer group for a typical commercial bank:

Asset size band. The dominant slicer. Bands run from under $50M to over $10B, with finer cuts at the community-bank end of the range.
Charter type. Commercial banks are grouped separately from savings institutions and credit unions.
Geography or business model (for the smallest tiers). Some peer groups separate banks by metro vs. non-metro location or by foreign-office activity.

The result is roughly 38 commercial-bank peer groups in the standard UBPR. The FFIEC publishes the full peer-group definitions at ffiec.gov/data/ubpr/peer-group-average-report.

What a percentile rank actually means

Every ratio on a UBPR comes with a percentile rank. The percentile is the share of peer-group banks whose value falls below the bank in question. So:

75th percentile on ROA = better than 75% of peers
50th percentile = the median for the group
20th percentile = better than only 20% of peers (worse than 80%)

For ratios where lower is better (like the efficiency ratio or non-performing loans ratio), the percentile is flipped in some references and not in others. Always check the column label. The FFIEC's own UBPR is consistent: percentile means percentile of the value itself, so for "bad" ratios you want a low percentile.

Why this matters for analysis

A lot of bank-comparison content online uses industry-wide averages: "the average ROA for US banks is 1.1%." That number isn't wrong, it's just not useful for analyzing a specific community bank. The mean is dominated by a handful of mega-banks.

Peer-group medians are what regulators, exam teams, and serious analysts use. They compare like to like. They're what makes the difference between "this bank has ROA of 1.05%" (so what?) and "this bank ranks at the 30th percentile of community banks in its asset band on ROA" (now it's a question worth asking).

Anywhere on BankingLENS that you see a percentile, it's a peer-group percentile, not an industry-wide one. That's the whole point.

Common gotchas

A bank's peer group can change. A bank that grows from $950M to $1.1B in assets crosses an asset-band boundary and gets reassigned to a larger peer group. Its percentile rank can shift even if its underlying performance didn't.
Peer groups for thrifts and credit unions are different. If you're looking at a savings institution or a credit union, you're in a separate peer-group system. Don't compare across charter types using percentiles.
Specialty banks distort the tails. Trust banks, credit-card monolines, and other specialty institutions get folded into peer groups that don't really fit them. Look at the absolute ratio in addition to the percentile when you suspect this.