# TT-002 · Corridors — method

**Grade: REPORTED for port calls, MODELLED for trade volumes.** These are different kinds of
number and are labelled separately for that reason.

## Source
IMF PortWatch (`portwatch.imf.org`), built by the IMF with the UN Global Platform on Kpler's AIS.
Licence permits commercial use with attribution:
*"Sources: Kpler; UN Global Platform; IMF PortWatch (portwatch.imf.org)."*

## What each number is
**Port calls — REPORTED.** Counted events, but the count depends entirely on how port polygons are
drawn and how vessels are classified. Counted by an algorithm, not by a person at the quayside.

**Trade volumes (`import_container`, `export_container`) — MODELLED.** Estimated from vessel type,
capacity and draft. The IMF is explicit about this. Useful for direction and shape; **never**
quotable as a measured quantity. Any publication resting on these must say "modelled, not measured."

## Coverage
20 ports · 55,100 port-days · 2019-01-01 onward. Ten US gateways, nine Indian ports, Colombo.
Refreshes weekly (Tuesday, ~09:00 ET); data lags roughly nine days.

## What it cannot do
**It has no origin–destination pairs.** PortWatch reports each port's total activity, not who
traded with whom. "India up, Colombo down" is therefore an *inference about* a corridor, never a
measurement *of* one. Every claim built this way must be written as an inference.

## Revisions
PortWatch reprocesses history. The series can change under you, so any published figure must be
pinned to the date it was retrieved. Our collector upserts on `(portid, date)` and silently adopts
revisions.

## Known failure — the Indian series is withdrawn from 2025-03
Between mid-2025 and March 2026 this source shows Indian container port calls collapsing by up to
49%, then recovering. **That collapse did not happen in the physical trade.** US customs
declarations (TT-005) show imports from India flat to strongly up across exactly that window, and
goods cannot reach the United States without leaving India.

Supporting evidence that it is an upstream artifact rather than a trade event: the damage is wildly
uneven across ports (Nhava Sheva −59%, Chennai −5%, Kattupalli **+52%**), consistent with calls
being reassigned between port boundaries; and it affects **every vessel type**, including tankers
(−30%), which no container-trade story explains.

Recorded as **F-001**. The affected series are unreachable through every outward surface.
The honest reading: this source is dependable for long-horizon shape at aggregate level, and should
not carry a port-level claim for a specific window without a second instrument agreeing.