Portfolio — The renewal right and the 5% cap
4.9 vs 3.9
The median jeonse renewal rose 4.9% when the tenant invoked their renewal right and 3.9% when they did not. On the averages the cap appears to do nothing. What it does is not in the middle — it is in the ceiling: nine in ten invoked renewals stayed under 5.0%, and for the rest that line sits at 14.6%.
Deposit increase · by whether the renewal right was invoked
Two distributions across one axis instead of side by side. Side by side they are two similar-looking shapes a reader has to overlay in their head; across one axis they are a single silhouette — a wall on one side, a tail on the other — and the comparison happens in the eye. The two groups differ by more than ten thousand contracts, so each is drawn as a share of itself, and both use the same pixels per point: a bar twice as tall is twice as common. Giving each half its own scale would have filled the picture more evenly and made the 47% spike look like the 30% one.
185,956 renewals · jeonse to jeonse · filed 2025-09 – 2026-08
- Right invoked · 94,671
- Not invoked · 91,285
Summary statistics — the first two rows are the trap
| Right invoked | Not invoked | |
|---|---|---|
| Median increase | 4.9% | 3.9% |
| 90th percentile | 5.0% | 14.6% |
| Share above 5% | 0.2% | 27.8% |
| Share held flat | 17.3% | 27.3% |
What this figure does not hide
- Jeonse only
- Of 333,408 renewals, 185,956 were used. A deposit and a monthly rent are not one number, and combining them needs a conversion rate this page would have to invent. The 146,440 renewals with a monthly rent on either side were dropped. What was dropped is counted rather than quietly disappeared.
- "Not invoked" may mean "not recorded"
- A blank renewal-right field is counted as not invoked. The filing does not distinguish a right that was not used from one that was not written down. If blanks are common, the lower distribution contains some invoked renewals — and that error makes the two distributions look MORE alike than they are. It weakens the finding rather than manufacturing it.
- Filed, not signed
- A transaction is filed within a window after the contract is signed, so the month here is the month it was reported, not the month it was agreed. Across a twelve-month window that lag does not change the shape of the distribution, but no single month should be read off it.