EmmanuelJDP/nearshoring-latency-feed

LexMetrix Nearshoring Latency Feed (NSLF)

The first open weekly operational latency index for the Caribbean Nearshoring → US East Coast corridor.

Built by LexMetrix · Updated weekly · License: CC BY 4.0

---

What this dataset is

The NSLF measures how smoothly cargo moves through the two main Dominican Republic ports — Puerto Caucedo (DP World) and Puerto Río Haina (HIT) — relative to their historical baseline, and benchmarks them against four regional competitors every week.

As US manufacturing nearshores to the Caribbean Basin under CAFTA-DR, port operational latency becomes a critical variable for supply chain planners. A single congestion event like the January 2025 Caucedo scanner failure (anchorage duration spiked to 11.6 days vs. a 0.6-day baseline) can cascade into missed delivery promises across dozens of manufacturers.

This feed turns that risk into a quantified, trackable, weekly number.

---

Latest reading — Week 2026-W23

Port	NSLF Index	Signal	Corridor Rank
Caucedo (RD)	97.40	🟢 STABLE_CORRIDOR	2 / 5
Haina (RD)	101.51	🟢 STABLE_CORRIDOR	—
Kingston (JM)	94.08	🟢 STABLE_CORRIDOR	1 / 5
Balboa (PA)	100.99	🟢 STABLE_CORRIDOR	3 / 5
Cristóbal (PA)	101.39	🟢 STABLE_CORRIDOR	4 / 5
Cartagena (CO)	105.82	🟡 WATCH_CORRIDOR	5 / 5

Base 100 = normal historical conditions (CEIC 2022–2025). Lower = less congestion.

---

Files

output/
├── weekly_feed.parquet        ← Primary dataset. One row per port per week.
│                                Includes NSLF index, component scores,
│                                and all 4 regional comparative columns.
├── comparatives_feed.parquet  ← Regional ports table (Kingston, Cristóbal,
│                                Balboa, Cartagena). Same schema as weekly_feed.
└── disruption_log.csv         ← Documented port disruption events with cause,
                                 duration, and NSLF impact. The differentiator
                                 no competitor publishes.

---

Schema

weekly_feed.parquet

Column	Type	Description
week	string	ISO week, e.g. 2026-W23
port	string	caucedo or haina
nslf_index	float	Composite latency index. Base 100 = normal
nearshoring_signal	string	FAVORABLE_CORRIDOR / STABLE_CORRIDOR / WATCH_CORRIDOR / CONGESTED_CORRIDOR
anchorage_days	float	Mean days vessel waits at anchor before berthing
als_score	float	Anchorage Latency Score (40% weight)
port_stay_days	float	Mean days vessel stays from berth to departure
psds_score	float	Port Stay Duration Score (35% weight)
vessels_waiting	int	Vessels at anchor at time of snapshot
vqp_score	float	Vessel Queue Pressure (20% weight)
ecs_score	float	External Congestion Signal from GoComet/Portcast CSVs (5% weight)
comp_kingston_nslf	float	Kingston (JM) NSLF that week
comp_cristobal_nslf	float	Cristóbal (PA) NSLF that week
comp_balboa_nslf	float	Balboa (PA) NSLF that week
comp_cartagena_nslf	float	Cartagena (CO) NSLF that week
rd_advantage_vs_kingston	float	Caucedo minus Kingston NSLF. Negative = RD has less congestion
rd_advantage_vs_balboa	float	Caucedo minus Balboa NSLF
corridor_rank	string	Caucedo's ordinal rank in corridor, e.g. 2/5
data_mode	string	DEMO (synthetic calibrated) or LIVE (real AIS)

disruption_log.csv

Column	Description
date	Event date (YYYY-MM-DD) or ISO week
port	Affected port(s)
event_type	EQUIPMENT_FAILURE / CONGESTION_PEAK / INVESTMENT_ANNOUNCEMENT / INFRASTRUCTURE / ECS_AUTO_SPIKE
severity	LOW / MEDIUM / HIGH
duration_days	Estimated operational impact duration
description	Human-readable event summary
source	Original source (news outlet, market report, etc.)
nslf_impact	Quantified effect on NSLF index where documented

---

Index Methodology

NSLF Index = ALS_score × 0.40
           + PSDS_score × 0.35
           + VQP_score  × 0.20
           + ECS_score  × 0.05

Where each score = (observed_value / historical_baseline) × 100

Signal thresholds:

Range	Signal	Interpretation
< 90	FAVORABLE_CORRIDOR	Below-normal congestion. Competitive advantage over Asia.
90–110	STABLE_CORRIDOR	Normal operations. Predictable for planning.
110–125	WATCH_CORRIDOR	Elevated congestion. Monitor, adjust safety stock.
> 125	CONGESTED_CORRIDOR	Crisis-level. Alert clients, review shipment promises.

Historical baselines (CEIC 2022–2025):

Port	Anchorage (days)	Port Stay (days)	Vessels Waiting
Caucedo	0.6	1.0	21
Haina	0.4	1.0	15
Kingston	1.8	2.1	28
Cristóbal	0.9	1.4	35
Balboa	1.5	1.6	40
Cartagena	0.7	1.2	18

---

Documented Disruptions (selected)

Date	Port	Event	Peak NSLF Impact
2025-01-08	Caucedo	Scanner failure + post-holiday backlog. Protests by transporters.	Anchorage spike to 11.6 days (CEIC, baseline 0.6)
2025-05-15	Caucedo	Caribbean-wide congestion peak. 91% yard utilization. 2–6h vessel wait times.	VQP estimated >130
2025-05-26	Caucedo	DP World + RD Gov sign US$860M expansion MoU. Target: 1.2M → 2.5M+ TEU.	Structural signal. Construction disruption risk 2025–2027.
2026-02-23	Nacional	National blackout (transmission fault). NAVIS SPARCS N4 impact probable.	PSDS disruption, duration ~hours

---

Data Sources

Layer	Source	Frequency	Coverage
Port congestion baselines	CEIC Data via MarineTraffic	Weekly	Caucedo, Haina 2022–2025
AIS vessel arrivals/departures	VesselFinder	Real-time → weekly aggregate	All 6 ports
External congestion signal	GoComet / Portcast exports	As available	RD ports
Disruption events	Diario Libre, K+N Market Updates, Business Insider	Event-driven	Manual curation
Regional comparatives	CEIC Panama, K+N Seaexplorer, Panama Ship Service	Weekly	Kingston, Cristóbal, Balboa, Cartagena

---

Current Status: DEMO Mode

All rows with data_mode = "DEMO" use synthetic data calibrated to CEIC historical baselines. The index structure, schema, comparative methodology, and disruption log are production-ready.

Transition to LIVE mode requires:

1. VesselFinder API key (free tier: 100 req/day — sufficient for weekly pipeline) → Register: https://www.vesselfinder.com/api/plans
2. Run: python pipeline.py --week 2026-W24 --apikey YOUR_KEY
3. data_mode field changes automatically to LIVE

---

Why this dataset exists

Dominican Republic's free zone sector hit 200,134 direct jobs in December 2025 — a historic milestone. DP World's US$860M expansion targets 2.5M+ TEU capacity to compete directly with Kingston. ITEK, AAFA apparel brands, and medical device manufacturers are actively nearshoring to RD under CAFTA-DR duty-free access.

Yet no public weekly index existed to tell supply chain planners: how congested are Caucedo and Haina right now, and how does that compare to the alternatives?

This feed is that index.

---

Related datasets

• LexMetrix Anti-Monopoly Feed RD — HHI concentration index for Dominican Republic markets

---

Citation

@dataset{lexmetrix_nslf_2026,
  author    = {LexMetrix AltData},
  title     = {Nearshoring Latency Feed (NSLF): Caribbean Port Operational Intelligence},
  year      = {2026},
  publisher = {HuggingFace Datasets},
  url       = {https://huggingface.co/datasets/lexmetrix/nearshoring-latency-feed},
  note      = {Weekly. Covers Puerto Caucedo (DP World) and Puerto Río Haina (HIT),
               Dominican Republic, with regional comparatives for Kingston (JM),
               Cristóbal and Balboa (PA), and Cartagena (CO).}
}

---

License

Creative Commons Attribution 4.0 International (CC BY 4.0)

You may use, adapt, and redistribute this dataset for any purpose, including commercial, provided you credit LexMetrix AltData.

---

Maintained by LexMetrix · Santo Domingo, Dominican Republic