# Gaps — RCEP preference margins SEA (`rcep_preference_margins`)

**As of:** 2 October 2026 (BST)

## Intentional empties / caveats

| Gap | Detail |
|-----|--------|
| **TLS preferential cells** | Timor-Leste is **not** an RCEP Party — `base_duty_free`, `day1_duty_free`, utilisation, EIF empty / `not_applicable`. Applied/MFN joins from `tariffs_rcep` (2021) remain for dual-story contrast only. |
| **Per-AMS ultimate duty-free %** | APEC locks **aggregate** 89.7% full liberalisation after 21 years; EnterpriseSG ~92% of goods traded. **No verified per-AMS ultimate % panel** locked — omitted rather than invented. |
| **Country-level MFN−RCEP simple average margin** | ERIA Hayakawa publishes charts / bands (forerunners ≤~0.6 pp vs non-AMS; CLMV larger) — **not** a clean SEA-11 numeric margin column. Scoreboard uses `eria_margin_band` categorical only. |
| **Export utilisation coverage** | Verified published figures locked only for **VNM 1.26%**, **THA 2.7%**, **MYS 0.07%** (2023, CIRD EN summary). BRN/KHM/IDN/LAO/MMR/PHL/SGP/TLS export util = **GAP**. |
| **Import utilisation (IDE)** | Partner panel covers importers **Japan, Korea, Philippines, Thailand** only — not a full SEA-11 importer matrix. |
| **HS-line preferential rates** | Out of scope — do not invent. Point researchers to MTI/EnterpriseSG Annex I schedules. |
| **Do not overwrite `tariffs_rcep`** | Applied/MFN/EIF cells here are **joins/crosswalks**. Canonical applied/MFN remain in `topics/tariffs_rcep/`. |
| **Myanmar political caveat** | Day-1 schedule share (30%) is a legal-text metric; utilisation/conflict context not invented here. |
| **Definition mismatch risk** | Day-1 = % of **tariff lines**; utilisation = % of **eligible trade** using RCEP; applied SM = WDI trade-policy average. Never average across these. |

## Not gaps (do not invent)

- Singapore 100% Day-1 is real (near-zero MFN hub).
- Philippines high Day-1 share with low observed util is the dual story, not a data error.
- Empty utilisation cells stay empty — not zero.