Korean residential distress does not appear suddenly. It travels through a predictable statutory sequence — one that leaves traces in the public court registry months before foreclosures begin. Building a dataset to track those traces revealed a pattern consistent enough to use as a forward-looking signal: a 17-month median lag between the moment a district’s lease-lien registrations spike and the moment its forced auction registrations follow.
This post explains what the pattern is, why it exists, and what it implies for lenders now expanding into 비수도권 under the FSC’s November 2025 supervisory revision.
The five registration types
The 등기정보광장 (IROS) public OpenAPI publishes monthly counts of five registration types per 시군구 district. Each type occupies a different position in the distress timeline:
| Type | Korean | Position in cascade | Signal type |
|---|---|---|---|
| 근저당 | Collateral mortgage | Upstream — reflects debt load | Leading (structural) |
| 임차권 | Court-ordered lease lien | Active dispute — tenant asserts priority claim | Leading (distress onset) |
| 가압류 | Provisional seizure | Active dispute — creditor freezes assets | Leading (distress onset) |
| 전세권 | Jeonse deposit right | Tenant-protective registration | Mixed |
| 강제경매 | Forced auction | Creditor-initiated foreclosure | Lagging (resolution) |
A full dataset covering 228 시군구 from January 2010 to March 2026 — 222,903 rows — makes it possible to track each type independently per district per month and observe how the series relate across time.
The cascade
The statutory sequence that produces the pattern:
전세 lease signed
│ Owner over-leverages (근저당 accumulates)
▼
Lease ends — deposit not returned
│ Tenant files court-ordered lien
▼
임차권 등기 (court registry records tenant's priority claim)
│ ~17 months (median)
▼
강제경매 등기 (creditor files for forced auction)
│
▼
KAMCO disposal → 온비드The 17 months is the sum of three sequential steps:
- ~3–6 months: 임차권 registered → owner receives formal notice → opportunity to repay or refinance
- ~6–12 months: Tenant (now effectively a creditor) files 강제경매 application with the court
- ~3–6 months: Court processes and registers the auction order
These timelines are consistent with general Korean civil procedure. In years of high distress volume (2022–2024), the lag compressed slightly as courts processed higher caseloads with more established procedure. In earlier lower-volume years (2018–2020), the lag was longer.
What the numbers show
Cross-correlating the time-shifted 임차권 series against the 강제경매 series at the district level, using the full 2010–2026 dataset, produces a district-level correlation of approximately R ≈ 0.93 at the 17-month shift.
District-level variance is significant:
- High-distress districts (수원, 인천 일부, 화성): lag compresses to ~12–15 months — process volume speeds up court timelines
- Low-distress districts: lag extends to ~22–28 months — less established procedural routine
- The R ≈ 0.93 is at the district level — the 17-month-shifted 임차권 series predicts the 강제경매 series for that specific district, not a cross-district prediction
Not every 임차권 registration leads to 강제경매. The conversion rate varies: approximately 40% in recovering districts, ~75% in chronically distressed ones. The 17-month figure is the median for those that do progress.
Why this matters for 저축은행 expansion
The FSC’s November 2025 supervisory revision assigns 110% weighting to 비수도권 저축은행 loan portfolios. This creates a regulatory incentive for savings banks and mutual finance institutions to expand into non-metropolitan districts — but the credit risk landscape across 228 시군구 is not uniform.
The 임차권 trend in a district today is a statistically grounded predictor of 강제경매 activity in that district 17 months from now. A savings bank extending new residential collateral loans into a district with a rising 임차권 z-score is lending into a district where the registry is already signalling upcoming foreclosure pressure — pressure that will compete for priority against any new liens registered after the existing 임차권 filings.
This is a signal type that NICE and KCB borrower scores do not capture. Those scores reflect individual borrower creditworthiness; the registry signal reflects district-level property market distress. Both are relevant to a secured property loan. The 임차권 trend is the part that was previously unavailable in structured, time-series form.
What the dataset enables
With 16 years of monthly data across 228 districts and five registration types, it becomes possible to:
- Identify districts currently in the pre-cascade phase (rising 임차권, flat 강제경매) — forward-looking exposure
- Identify districts in the in-cascade phase (peaked 임차권 ~12–17 months ago, rising 강제경매) — active pressure
- Identify districts in the post-cascade phase (강제경매 peaking, 임차권 declining) — clearing
- Rank districts by composite distress across all five types on a rolling z-score basis
The methodology is reproducible: any analyst with IROS API access and the same 228-district codebook will derive similar results from the same public data.
The dataset underlying this analysis is described in detail on the Korean District Credit Risk Index project page. Public outputs — a sample district report and national trend chart — are available at kr-housing-risk.