Supply Chain Action Points
Read this first — the conclusion, and the moves to make:
- By 24 July, put the published wait for every load and discharge port on your strings on one page, flag anything above three days, and use that list as the week's routing agenda instead of reading it as news.
- Shift 10 to 20 percent of August volume off any single-string port whose wait exceeds three days, and confirm in writing that the alternate booking sits on a vessel already worked, not a proforma.
- Re-quote destination free time on Manila-bound cargo before 31 July and pre-agree either a seven-day extension or an inland staging plan, so bunching does not convert into demurrage.
- Write one standing trigger this week, two consecutive updates above three days at the load port moves fifteen percent of next month's volume, and back-test it against July's figures.
- Track your own ninetieth-percentile transit rather than the published average, and reconcile it against port dwell on the first Monday of each month.
Kuehne+Nagel's mid-July update showed Shanghai and Ningbo recovering from Typhoon Bavi backlogs, Qingdao improving, and Singapore stabilising with terminal waits of 0.5–1 day, though Manila remained strained at a 4.48-day average. Carriers are still working through vessel bunching as East Asian ports resume normal rotations.
The Analysis
The mid-July port update landed this week and reads, at first glance, like the end of a weather story. Shanghai and Ningbo are working through the backlog Typhoon Bavi left behind. Qingdao is improving. Singapore has settled at terminal waits of half a day to a full day. Then, at the bottom of the same table, Manila is still averaging 4.48 days.
That last number is the reason to read the whole note twice. One line says the storm is over. Another says this network is not yet back to itself, and that one node in it never was.
What follows is about the gap between a storm passing and a network recovering, which are not the same event, and about what an importer or exporter should actually do in the two or three weeks after a headline says normal.
Start with the sequence, because the sequence is the argument. A typhoon takes berth-hours out of a port. Vessels wait at anchor or slow down. When the port reopens, the ships do not come back politely spaced; they come back together, and the terminal's cranes, yard slots and gate lanes are fixed while the arrivals are not. That is what the note means by vessel bunching, and it is the part of a typhoon that outlives the typhoon. The storm might cost three days at the quay. The bunching costs two to three weeks of schedule, paid unevenly, string by string and carrier by carrier.
Recovery is a rate, not a state. A terminal clears a backlog at the speed of its daily berth capacity minus the day's normal arrivals; whatever is left over eats the queue. Two ports with identical backlogs recover at different speeds if one has more crane density, more yard headroom, or a gate that can work a second shift. That is why the same note can call Shanghai and Ningbo recovering, Qingdao improving and Singapore stable without contradicting itself. Those are three different slopes, not three different moods.
The map matters more here than the adjectives. Shanghai, Ningbo, Qingdao and Singapore are not four equivalent dots; they are four nodes with different jobs. Ningbo and Shanghai are load ports for the same export hinterland, so a disruption in one is partly absorbed by the other, at a price. Singapore is a transshipment hub, which makes its half-day-to-one-day wait a statement about connections rather than about origin cargo. Manila is a destination and a domestic distribution gateway, and its 4.48 days sits at the end of the leg, where the cargo has nowhere else to go.
Which is the first layered effect, and it has less to do with the size of a shipper than with redundancy. A company with four or five weekly services spread across two ports will barely notice: some boxes roll, some shift, the month closes. A company with one string out of one port feels all of it, and that company is not necessarily small. A single-product exporter shipping forty containers a week on one service is structurally fragile in exactly the way a thirty-container-a-month trader on three services is not. Volume buys options. Concentration takes them away.
Direction flips the meaning of the same headline. Exporting out of Shanghai or Ningbo, the worst is behind you and what remains is arithmetic: how many boxes are still in the backlog, how many berth windows per day, when the queue reaches your booking. Importing into Manila, the number is not a residue of the storm. It is the port as it currently is, and a plan built on a two-day wait is a plan built on a port that does not exist this month.
Timing carries a lag that catches people. The port says normal; berth windows normalise five to ten days later; the rollovers happen in the gap between the two. That gap is where the money goes, because it is where a booking that looked confirmed becomes a box sitting on the quay while the vessel sails. A second clock runs at the far end: destination free time starts at discharge, not when your customer is ready, and a bunching event compresses arrivals at destination into a shorter window than the yard can absorb.
Here is what the update does not say, and it decides whether the next three weeks are cheap or expensive. The announcement that a port is recovering is not a description. It is an instruction. The moment the note says normal, every forwarder holding cargo releases it, and the terminal takes a second wave at the exact hour it is trying to drain the first. The queue is re-loaded by the news that the queue is clearing. Nobody is behaving irrationally; everyone is optimising against the same published signal, which is precisely how a published signal becomes a fresh source of congestion.
The other thing the note does not say is that it is averaging two different kinds of event into one table. A typhoon is a pulse. Manila's 4.48 days is a trend that the storm made visible. Average a pulse and a trend and the result describes neither, and a reader who treats the table as a single recovery story will mis-time both ends of it. The honest reading is that East Asia is in the tail of a pulse while one node is in the middle of a constraint that predates the weather. Those two facts ask for opposite responses: the first says wait three weeks, the second says route around something for a year.
Run a small account, with the assumptions stated. Take an exporter moving 200 FEU a month out of Ningbo across two strings; that volume is my assumption, not a figure from the update. If bunching adds six days to the average cycle for three weeks, 150 FEU carry an extra six days, which is 900 FEU-days of inventory parked inside the system. At an assumed all-in carrying cost of USD 30 per FEU per day, also my number and not the note's, that is roughly USD 27,000 which never appears on a freight invoice and therefore never appears in a review. Move a fifth of that volume to the secondary string and you cut about 180 FEU-days, at the price of a transshipment and a longer nominal transit. The trade is not about freight rate. It is about which number you would rather explain.
The counterfactual worth running is not what would have happened without the typhoon. It is what would have happened if you had a written trigger before it: when the published wait at the load port exceeds three days on two consecutive updates, fifteen percent of next month's volume moves to the alternate string, with documentation pre-cleared for the move. Under that rule the decision happens in week one, while alternatives still exist, instead of in week three, when everyone is calling the same three forwarders. Rules written in calm weather are the only ones anyone follows in a queue.
Pitfalls, since they are predictable. Free time and demurrage clocks start at discharge whether or not your inland leg can move, and bunching turns that into a real cost at destination. Rollover fees are usually charged per box per event and are almost never in the quoted rate. Port dwell KPIs exclude boxes still on the water, so a port can report improving dwell while your cargo has not moved a metre. And a booking confirmation is not a berth: a slot confirmed against a vessel that has not been worked is a hope with a reference number.
A typhoon in the East China Sea once was a local event with local consequences. The storms have not grown. The network has tightened, and tight networks transmit. What you control is not the weather but the number of paths you hold when it arrives. Manila's figure this week is a reminder that a node with no alternative is not a node; it is a dependency.
↑ Back to the key points