First-pass empirical findings on § 1328(f) successive-filing patterns. Dataset draws on 211,537 federal bankruptcy cases across 41 districts, 1990-2026. The full dataset is published under CC BY 4.0. Methodology and sampling caveats are documented below.
The post-BAPCPA period (October 2005 onward) is the period when § 1328(f) actually applied. Pre-2005 data are presented as counterfactuals — "would have been caught had the rule been in effect."
| Year | Ch.13 with multi-case history | Within lookback | Pct caught |
|---|---|---|---|
| 2005 (BAPCPA Act) | 683 | 52 | 7.6% |
| 2006 | 434 | 33 | 7.6% |
| 2010 | 508 | 24 | 4.7% |
| 2015 | 723 | 47 | 6.5% |
| 2020 | 496 | 29 | 5.9% |
| 2024 | 515 | 31 | 6.0% |
| 2025 | 479 | 31 | 6.5% |
Post-BAPCPA the rate hovers in the 5-7% range — lower than the pre-BAPCPA counterfactual range (often 25-30%), suggesting that the rule has produced a deterrent effect among attorney-represented cases. The remaining 5-7% are dominated by pro-se filers and cases where the prior was a transferred or rehabilitated case the debtor didn't realize counted as a discharge.
The 11.8% rate (across all years) translates to: roughly 1 in 8 successive Ch.13 filers in this dataset filed within the lookback distance from a prior discharge. For those debtors, plan completion would result in no discharge of unsecured debt (under § 1328(f)) — though they would still complete the plan and obtain its other benefits (automatic stay, secured-debt cures, priority debt payment).
This research project's per-case classification can't directly distinguish pro-se from attorney-represented filings without docket-text reading. But district-level patterns are consistent with prior empirical work showing pro-se filers disproportionately encounter the § 1328(f) bar, because pro-se debtors often don't know about the lookback rule when they file.
The dataset is assembled from PACER docket pulls focused on a research subset of districts, plus BAPCPA Table 6 aggregate data. Districts with heavy PACER monitoring (KSBK, MOWBK, FLSBK, TXSBK) are over-represented in the absolute counts. Within-sample rates (% of successive filers caught by lookback) are sampling-bias resistant; absolute rankings of districts by raw count reflect the data-collection focus, not national prevalence.
Name-based matching has known false-positive risk for common surnames; tooling improvements should produce v0.2 of the dataset with SSN-based matching where available.