Singapore port feels transshipment strain as Red Sea reroutes pile up

The Analysis

The BRF Logistics note on Singapore runs to four sentences and contains not one number. Transshipment volume is up. Vessel arrivals are late. Yard density is higher. Connection times are longer. That is the entire quantitative payload, and it is enough to cost you money, provided you do the division the note declines to do.

The phrase that hides the damage is longer connection times. A relay hub does not slow cargo smoothly. It slows it in whole feeder cycles. Singapore's transshipment function runs on a weekly pulse: mother vessel in, discharge, stack, feeder out on its window. When the mother is two days late you do not arrive two days late. You either make the window or you miss it, and a miss costs a full cycle, not the two days you lost.

That is the difference between a two-day problem and a three-in-ten problem, and it is what I want to price out here.

BRF Logistics reports rising transshipment volume, delayed vessel arrivals and higher yard density at Singapore, and attributes all three to Red Sea diversions that have pushed more cargo through the hub. The note adds that Yantian and Ningbo are handling strong South China export demand. Read the three indicators as one mechanism rather than three symptoms, because that is what they are. Diversions add boxes that were never planned to relay here, late arrivals compress the discharge work into shorter windows, and yard density is simply what happens when both land on the same terminal in the same week. The note's advice, build extra buffer for transshipped cargo and watch feeder schedule reliability, is correct and unusable at the same time. Extra buffer is not a number.

Take a weekly feeder window and a mother vessel arriving late by an amount that varies from sailing to sailing. A box makes the connection if it clears discharge, gets stacked and is nominated before the feeder cut-off. Treat the delay as spread evenly across the seven days of the week, purely as a modelling convenience and explicitly my assumption rather than a figure from the source, and the share of boxes that miss the window is roughly the delay divided by seven. Two days of average arrival delay puts close to 29% of boxes onto the following sailing. Average transit time then rises by about two days, which is exactly the number that will appear in a monthly reliability table and exactly the number that conceals what happened. Nearly three boxes in ten are seven days late. The mean looks mild because the distribution has gone bimodal: one cluster on time, one cluster a full cycle behind, and very little in between.

A bimodal distribution cannot be hedged with an average. A one-day buffer does nothing for the 29%, because those boxes are not one day short, they are a cycle short. Buffer has to be sized in feeder cycles. If your downstream commitment can absorb three days and not seven, a four-day buffer is worse than no buffer at all: it costs inventory and still fails.

The cargo that hurts first is relay cargo with a hard downstream date: retail launches, promotions, line-side parts for a plant running a thin buffer. Commodity cargo with thirty days of float feels almost nothing beyond extra paperwork and a later invoice. Direct-call cargo out of Yantian and Ningbo is largely protected on transit time, but it is exposed on a different axis, because those ports are absorbing the same South China export surge and the substitute route is therefore not idle capacity. Scale changes the arithmetic less than it changes the options. An importer moving twenty FEU a month through a relay at a 29% miss rate has roughly six boxes a month arriving a week late. An importer moving two hundred has fifty-eight. Same percentage, but the small importer cannot reallocate stock between distribution centres and the large one can. Percentage losses are democratic; the ability to absorb them is not.

There is a structural reason relay cargo absorbs the pain first, and it is not bad luck. Transshipment boxes have no local consignee chasing them. Nobody calls the terminal about a box that is only passing through, so when a yard is dense and a berth window is short, the box with no local owner is the box that waits. Congestion at a relay hub is not distributed evenly across cargo. It is distributed by who is shouting.

The date on the note is 4 September, and that matters more than it looks. A missed connection on a 4 September sailing lands in the first half of October, not in September. If your fourth-quarter allocation is already committed on relay routings, the financial effect arrives after the point at which you could still change the routing. That is the uncomfortable part of the calendar: the signal is dated in September and the cost is booked in October, which is precisely the gap in which nobody acts.

Here is the part the coverage does not say, and it is the reason I bothered with the arithmetic. At a stressed relay hub the binding constraint is not vessel arrival delay. It is the number of feeder windows available per week on your lane. Singapore relay runs on a weekly pulse. If congestion consumes one of the two weekly windows a lane normally offers, your effective capacity on that lane halves and your average delay roughly doubles, while every arrival statistic still looks like a delay of a day or two, because the mothers keep arriving on schedule and keep discharging. The schedule is intact and the service is not. Watch windows, not arrivals. Ask your carrier how many feeder departures per week your connection depends on, and what happens to the second one when the first is missed. Most shippers cannot answer that about their own main lane, which is why congestion arrives in their numbers as a mystery.

The counterfactual decides whether you sign anything long, so state it plainly. If Red Sea routings normalise, relay load at Singapore reverts within one to two schedule cycles, and any premium you paid for a direct call or a non-Singapore relay becomes a permanent cost against a temporary problem. The condition that flips the conclusion is duration: if the diversions outlast your contract renewal date, the relay premium is not temporary and belongs in the base rate, not in an exception clause nobody remembers.

Put money on it with round numbers I am choosing for illustration, not figures from the release. A 40-foot box with a declared value of USD 180,000, carrying cost of 10% a year covering capital, insurance and markdown risk, costs about USD 49 a day to hold; seven days is USD 343. Detention beyond free time at USD 40 a day per box for seven days adds USD 280. Call a missed connection USD 620. A shipper moving twenty FEU a month through the relay at a 29% miss rate loses about 5.8 boxes a month, roughly USD 3,600. That is not the number that should worry you. The number that should worry you is the two boxes a year that miss twice, take fourteen days, and cost you a promotion or a line stop. Reserves are sized for the tail. Averages are for the report.

The break-even for abandoning a relay is straightforward once you have a measured miss rate. A direct call costs some premium P per box over the relay. Switching is rational when the probability of a missed cycle times the cycle length times your daily carrying cost exceeds P. With the numbers above, at a 29% miss rate the expected delay cost is about 0.29 times USD 620, near USD 180 a box. If the direct-call premium on your lane sits below that, take the direct call now. If it is double, stay where you are and buy buffer instead. The trap is turning a congestion headline into a routing decision when what you actually hold is a probability you have never measured.

The mistake I see most often is re-routing the whole book after one late shipment, then discovering the alternative is loaded too, because Yantian and Ningbo are absorbing the same export surge that pushed volume through Singapore in the first place. Measure at port-pair level across at least six sailings before you move anything. Lock the direct-call differential in the first week of October, while the peak is still being priced as temporary. After that you are not buying certainty. You are buying the description of it.

↑ Back to the key points

— By Vivian Zhao