Open Bankruptcy Project

Bankruptcy Recidivism Patterns — v0.1

First-pass empirical view of multi-filer bankruptcy recidivism. Across 199,848 unique debtors, who comes back to bankruptcy court — how often, how soon, and where? Several findings are unique to this dataset; published academic literature has rarely measured these patterns at scale.

Headline findings

5.0%
of bankruptcy debtors are multi-filers (10,021 of 199,848 in the sample). The vast majority of debtors file once and never return to bankruptcy court.
78 days
Median interval from dismissal to next filing. Debtors whose cases are dismissed file again FAST — the 25th percentile is 33 days. This is the empirical "Chapter 20" pattern: file Ch.7, get dismissed, immediately refile (often Ch.13) to maintain the automatic stay.
1,821 days
Median interval from discharge to next filing (~5 years). Debtors who completed a prior case wait substantially longer before refiling — consistent with §1328(f) lookback (4-year Ch.7→Ch.13 / 2-year Ch.13→Ch.13) being respected.
7.5%
cross-district recidivism rate. Most multi-filers stay in the same federal district when they refile (92.5%) — substantially higher than what "venue-shopping" theories would predict.

Filings-per-debtor distribution

Filings per debtorUnique debtorsCumulative% of all debtors
1189,827189,82795.0%
28,697198,5244.4%
31,086199,6100.5%
4173199,7830.09%
542199,8250.02%
614199,8390.007%
74199,8430.002%
83199,8460.0015%
91199,8470.0005%

The "Chapter 20" pattern, quantified

The post-dismissal interval distribution is the standout finding:

Filing interval (days)p25median (p50)p75p90n
All consecutive multi-filings3298122,0573,67410,660
After prior discharge7531,8213,0714,4542,988
After prior dismissal33782461,0716,296

The contrast is dramatic. Post-discharge filers wait approximately 23x longer than post-dismissal filers (1,821 vs 78 days at the median). Two distinct populations:

Per-district recidivism concentration

Top districts by percentage of cases that come from multi-filers (within sample):

DistrictMulti-filer casesTotal cases% from multi-filers
nvbk5032715.3%
tnwbk2891,98014.6%
moebk1911,35614.1%
txsbk4,25630,41114.0%
mowbk5,45940,91513.3%
ksbk5,02538,76513.0%
paebk1581,6749.4%
caebk697818.8%
flsbk3,64543,9308.3%
almbk861,0398.3%

Districts with high Chapter 13 share (Texas Southern, Missouri Western, Kansas, Tennessee Western) cluster at the top of the recidivism rate — consistent with Chapter 13's higher plan-failure rate producing more refiling.

What the patterns mean for §362(c) and §1328(f)

The data have direct implications for two procedural-rule frameworks:

§ 362(c) successive-filing stay limits

Section 362(c)(3) limits the automatic stay to 30 days for filers with one prior dismissed case in the past year; § 362(c)(4) eliminates the stay entirely for filers with two or more prior dismissed cases. The 33-day p25 / 78-day p50 post-dismissal refiling intervals show that a substantial fraction of refilings happen within the §362(c)(3) one-year window — meaning the stay limits ARE biting frequently in practice. See § 362(c) successive-filings reference.

§ 1328(f) discharge lookback

The 5-year (~1,821-day) median post-discharge interval suggests that attorney-represented multi-filers ARE successfully waiting out the §1328(f) lookback (4 years for Ch.7/11/12 → Ch.13; 2 years Ch.13 → Ch.13). This is consistent with the §1328(f) research project's finding that ~12% of successive Ch.13 filers are caught by the bar — the population caught is the smaller subset who don't wait out the rule, often pro-se filers unaware of it. See §1328(f) findings.

Methodology and caveats

Name-based matching has known false-positive risk

Multi-filer identification is by normalized debtor name (uppercase, punctuation stripped). Common surnames (Smith, Johnson, etc.) can produce false-positive matches between unrelated debtors. SSN-based matching would be cleaner but isn't available in public PACER data.

The 10,021 multi-filer count likely includes some false positives from name collisions. Estimated upper bound on false-positive rate based on common-surname-frequency analysis: 5-15%. True multi-filer count is plausibly 8,500-9,500.

Open dataset

Three CSVs published under CC BY 4.0:

Citation

Open Bankruptcy Project (2026). Bankruptcy Recidivism Patterns Dataset, v0.1.
501(c)(3) public charity (EIN 41-5159631).
URL: https://viz.openbankruptcyproject.org/recidivism/
License: CC BY 4.0