Hub-and-feeder ports

Why do most goods pass through a few distant giant ports instead of the nearest one?

A container bound for a modest coastal city may sail past that city's own perfectly serviceable harbour, continue to a giant port some hundreds of kilometres away, be lifted off, wait, be lifted onto a smaller ship, and sail back. Extra distance, extra handling, extra days.

That looks like a mistake. Shipping lines are not sentimental, so presumably it is not one. So what is being bought with all that extra movement — and why would the answer be to make some ports enormous and others small, rather than making them all much the same?

Start with the ship rather than the port. What happens to the cost of moving one container as the vessel carrying it gets bigger?

Think about what a ship costs to run: the hull, the fuel, and the crew. A crew of roughly the same size sails a vessel of two thousand boxes or twenty thousand. Fuel per container falls too, because a hull's carrying volume grows faster than the drag on its wetted surface. So cost per box drops steeply with size, and the very largest container ships now carry over twenty thousand boxes on a single sailing. That is the pull toward bigness.

Now ask what a ship that size requires. Roughly sixteen metres of water under it, so a dredged approach and berth. Quay cranes with the reach to work across a hull twenty-odd containers wide. Yard space and equipment to absorb thousands of boxes in a day, because a ship that large earns nothing sitting still. And, crucially, enough cargo in one place to fill it — an empty megaship is the worst economics in the trade.

So here is the tension worth sitting with. The cost advantage says use the biggest ship you can. The requirements say only a handful of places on any coastline can serve one. What does a network do when its cheapest unit demands conditions almost nowhere satisfies?

It specialises. A few ports take on the deep water, the giant cranes and the volume, and become hubs. The rest stay modest and are served by small ships running short legs. The long ocean crossing — where distance is greatest and therefore where cost per box matters most — is done once, on the largest ship, between two hubs. The last few hundred kilometres, where the distance is small, is done on a feeder where the size penalty is cheap to bear.

Notice that this is not each port choosing to be a hub or not. It is the network discovering that its total cost falls when its nodes stop being alike. That is what specialisation means here: differentiation of roles, driven by a threshold that only a few can cross.

But it is not free, and the limit is what makes the pattern interesting. Every transshipped box is lifted twice more than it would have been on a direct call, and it waits between ships. So hubbing pays only when the saving on the trunk leg exceeds the extra lifts plus the detour plus the delay. Change any term and the answer flips: a route with enough cargo of its own justifies a direct call, which is why large markets are served directly while thin ones are fed. This is also why the great transshipment hubs sit where they do — Singapore, Algeciras, Colombo, Tanjung Pelepas — close to the main east-west shipping lane, so the detour term is nearly zero. Singapore handles close to forty million containers a year, and a large majority of them never enter Singapore at all; they are simply changing ships.

And the honest caveat: how far this concentration goes is contested and not purely physical. Alliances between carriers, national port investment, and the strategy of terminal operators all shape which places become hubs, and a port's advantage is partly self-reinforcing once volume arrives.

Two refinements make the picture behave properly. The first is that "nearest port" is the wrong unit to optimise. The cost that matters is door to door, and the sea leg is only part of it — a box may sail past its local harbour and still arrive sooner and cheaper because the hub has daily sailings while the local port has one a fortnight. Frequency is itself a product of concentration, and it is often worth more than the distance saved.

The second is that hub and feeder are roles, not rankings. A feeder port is not a failed hub; it is a node whose traffic is genuinely better served by short legs, and many feeder ports are highly profitable at exactly that job. Some ports are both — a hub for one trade and a feeder within another.

The misconception worth gently correcting is that the extra handling is waste. It is a purchase. What the shipping line buys with two crane lifts is a berth on a vessel whose cost per box is a fraction of anything that could call locally. When the arithmetic stops favouring that — rising handling costs, a thickening local market, congestion at the hub — direct calls return, and they periodically do.

• Economies of scale at sea, and whether ship sizes have now overshot what ports and canals can economically serve.
• Port competition and the hinterland: how a hub's catchment is decided by rail and road, not only by water.
• What happens when a hub fails — congestion, strikes, or a closed canal — and how quickly cargo reroutes.

Every term the collection defines is gathered in the glossary.