Data sources
- Federal Judicial Center Integrated Database (IDB). Case-level federal bankruptcy data published by the Federal Judicial Center. The IDB is the authoritative count of US federal bankruptcy filings; OBP uses a local copy refreshed from FJC's periodic releases. Chapter mix, filing volumes, prior-filer flags, pro se flags, and dispositional outcomes all derive from IDB fields.
- Free Law Project RECAP Archive via CourtListener. Docket-level coverage on OBP district pages measures RECAP's share of FJC-counted filings. Docket metadata comes from CourtListener; full filings are linked to CourtListener, not rehosted.
- Federal Judicial Center 2024 Unredacted SSN Study. National privacy-rule context on every district page cites Garri, Germano, Cantone, and Laks (April 2024), the Federal Judicial Center's empirical study of unredacted Social Security Numbers in federal PACER documents.
- Court websites (
.uscourts.gov). Judge rosters, courthouse addresses, and local-rules landing pages are scraped once per district with a polite 10-second crawl delay and a descriptive user-agent. Scraped data is cached; re-scraping is gated on staleness.
Metric definitions
Total cases
Count of records in FJC IDB fjc_cases where district matches the district's 2-character FJC code. Includes all chapters and all fiscal years in the IDB release.
Latest-FY filings
Count of IDB records for the most recent complete fiscal year. The current federal fiscal year (Oct–Sep) is excluded when filings are still arriving.
Chapter mix
Share of filings by chapter (7, 11, 12, 13) computed over records with a non-blank chapter code (crntchp). Chapter 15 cases are folded into a single bucket; Chapter 9 appears only where municipalities have filed.
Disposition mix
d1fdsp) is populated only when a case was assigned a terminal disposition at the time of the FJC extract. Historical cohorts typically have 30–50% coverage because the IDB is a periodic, not continuous, dataset. OBP reports disposition percentages over the classified subset (i.e., cases with a recorded outcome) and surfaces the unclassified count explicitly. We do not infer outcomes for unclassified cases.
Disposition codes are grouped into four buckets: discharged (code A), dismissed (codes D, E, F, G, H, I, J, O, T), converted (code K), and other (revocations, administrative closures, transfers, and other terminal codes). Bucket membership follows the FJC IDB code sheet.
Prior-filer rate (Chapter 13)
Share of Chapter 13 cases with the prfile='Y' flag (prior bankruptcy filing of record), computed over the classified subset (prfile in {'Y','N'}). This metric is relevant to 11 U.S.C. § 1328(f) discharge eligibility and has been the subject of OBP's § 1328(f) research.
Pro se rate
Share of filings without an attorney of record, from the FJC IDB d1fprse field. Computed over the classified subset.
RECAP coverage
Count of dockets for a district in the Free Law Project's RECAP Archive, divided by the district's total FJC IDB case count. This measures how much of the district's public docket is available free of charge; coverage varies widely by district and by time period.
National rank and percentile
Districts are ranked 1–94 on each metric, and a percentile is computed as (1 − (rank − 1) / 94) × 100. A district at the 75th percentile on total filings is in the top quarter of the country.
Peer districts
For each district, three peer districts are identified as the same-circuit districts closest in total case volume. The goal is to give readers a comparative frame without introducing editorial judgment about comparability.
Privacy-rule (Rule 9037) context
The statistics in the Privacy Rule section of every district page are national figures from the Federal Judicial Center's 2024 study of unredacted Social Security Numbers in PACER. OBP does not publish district-level compliance or non-compliance rates: the FJC study itself does not break its findings out by district, and the difference between "appearance of non-compliance" (which the FJC measured) and "adjudicated Rule 9037 violation" is a legal distinction OBP respects. Per the FJC authors' own note, "researchers interpreted the exemption provisions broadly" — the 72% non-compliance figure is a floor, not a ceiling.
Judicial information
Judge rosters are drawn from each court's public website. OBP publishes names and roles (e.g., Chief Judge, Judge) as they appear on the court's roster, and nothing else. Qualitative characterizations of judges — whether they are “strict,” “lenient,” or anything similar — do not appear anywhere on this site. Case statistics by judge are not currently published on the district pages.
Entity-naming policy
Creditors, law firms, and individual attorneys are not named on the district profile pages. OBP's research publications aggregate. Questions about specific attorneys should be directed to the relevant state bar association or federal district court. Questions about specific creditors should be directed to the consumer bureau with jurisdiction (e.g., the Consumer Financial Protection Bureau) and, where applicable, the trustee or US Trustee supervising the relevant case.
Reproducibility
The scripts that build these pages are open-source. Each district profile page can be regenerated from source data using the tools in pacer/scripts/ in the OBP GitHub organization. Specifically: district_metadata.py (the canonical 94-district registry), scrape_court_site.py (the court-website scraper), build_district_profile_data.py (the preprocessor that produces per-district JSON), and render_district_profile.py (the HTML renderer). Every metric on a rendered page can be traced back to a specific SQL query or source file.
Updates
District profiles are regenerated on a rolling schedule. The footer of each profile page shows when its data was last refreshed. FJC IDB refreshes follow the FJC's own release cadence; RECAP docket caches refresh no more than once per 24 hours.
Corrections
If you find an error in the data, a broken link, or an inaccurate description, please submit a correction. Every correction is logged and the relevant page is rebuilt; significant corrections are noted in the page’s data-updated stamp.