Institution analysis

Financial reports, peer benchmarks, and historical trends.

Select institutions

Search all active and historical FDIC bank and savings-association names. Choose two legal institutions to compare; identifiers are resolved automatically.

Methodology

Calculation version v7.1-integrity. All interfaces use the same calculation service.

Snapshot availability. Reuse the same analysis across tabs and refreshes for up to one hour. After expiry, start a new analysis; the reporting period may change. Expired cached data is removed during later snapshot activity.

Reporting period

Overview comparisons use the latest common report when two institutions are selected; otherwise the primary institution's latest report.

Peer selection

Exclude the primary institution. Match specialization and community-bank status at 0.5–2.0 times assets; widen to 0.33–3.0, then drop community status, then use all reporters at 0.25–4.0 until at least 15 peers qualify. Order by absolute log asset distance, then FDIC certificate.

Percentiles

Available non-null observations only. Percentiles use the midpoint of tied observations; quartiles use linear interpolation. Direction-adjusted ranks reverse lower-is-better metrics.

Growth

Compare the same certificate and reporting date one calendar year earlier. Missing or zero prior values yield null. Acquisitions are not adjusted.

Units and entity scope

Balance-sheet values remain in thousands of US dollars in data. Display converts them to dollars. Ratios are percent. NUMEMP counts full-time equivalent employees of the reporting bank and its consolidated subsidiaries, not its parent company's workforce. Institutions are insured depositories, not holding companies.

Historical peer trends

Fixed sample of up to 25 nearest current peers, excluding the comparison institution. Overview peers may include that comparison institution. Historical coverage varies by quarter.

Trend ranges

Version 7 ranges select 12, 20, 40, or 80 latest primary records, or all available history since 1990. They are record counts, not calendar windows. Histories are not cut off at the overview common date.

Smoothing and regressions

Moving averages use four valid observations, including across gaps. Ordinary least squares uses elapsed calendar quarters; ratio slopes are annualized percentage points, while positive assets and headcount use log-linear annualized percent changes. At least three observations are required.

Observed changes and historical fits

Trend statistics 1.1.0-observed-change separate observed endpoint changes from fitted slopes. Year-over-year compares the latest selected bank report with the exact same quarter a year earlier in its full history. Selected-window change uses the actual first and last selected records; missing endpoints are not skipped. Ratio differences are percentage points; assets and FTE use percent change from a positive baseline. Duplicate periods and unavailable values yield explicit missing reasons. Historical fits are optional in the browser, not current growth, endpoint CAGR or forecasts. R² is measured in log space for log-linear fits and raw-value space for ratio fits; it is not a rating of predictive accuracy.

Version notes and unchanged semantics

Integrity corrections in this version

Missing assets remain null and cannot define a primary peer cohort.
Incomplete upstream results fail explicitly.
Asset-distance ties use certificate order.
Invalid comparison certificates are rejected.

Potential changes awaiting review

The following changes are not applied to these results:

Calendar-window range semantics
Common-date cutoff of full histories
Consecutive-quarter smoothing
Overview and trend comparison-cohort symmetry