A plastic toy looks like the output of a single factory: a box, a brand, an address on the back. In Chenghai, on the coast of Guangdong, it is more accurate to say the toy is the output of a neighbourhood. One workshop cuts the mould. Another runs injection machines of a given size. A third sprays and pads the details. Somewhere else a bin of gears, axles, battery contacts and tiny screws sits on a shelf, and a few doors down a room of people put the pieces together and drop the result into a printed carton. Nothing about this arrangement is picturesque. It is a way of making the distance between a mistake and its correction as short as possible.
The reason is not cheap labour, and it is not a mysterious local genius for toys. It is that a new toy is a guess. A drawing that looks right on a screen becomes a mould, and the mould produces the first real object, and that object almost never closes properly, stands up, or fits the other parts. So the mould goes back to be cut, welded, polished and tried again. Each round trip consumes hours. Put the mould shop across town and you get one round trip a day; put it two doors away and you get three. Over a fortnight of development, that difference is not a convenience. It is the difference between a toy that ships this season and one that does not.
The mould is the real product
A steel mould is a negative of the toy, split into two halves that meet along a parting line. That line is why almost every plastic object you own has a faint seam you can find with a fingernail. The mould decides wall thickness, where the ribs go, whether a screw boss will sink or hold, and how much the plastic shrinks as it cools. It also decides how many pieces can be made before it wears. All of the decisions that matter in a toy are taken before a single one is sold, and they are taken in steel, which is slow to change and expensive to get wrong.
This is why the mould shop sits at the centre of the street rather than at its edge. Everything downstream is comparatively easy to rearrange. A different paint mask, a different carton, a different motor: these are afternoon decisions. A different wall thickness is a machine-shop decision. The geography of the district is really a map of which decisions are hard to undo.
Nobody owns every machine
Injection presses are rated by clamping force, and a toy with a large body shell needs a much bigger machine than a tiny wheel does. A workshop that bought every size would spend most of the year with most of them idle. So the district shares them implicitly: shops specialise in a range of sizes and take work from whoever needs that range. During a busy season the queue at one machine sets the pace for a dozen brands at once.
The same logic applies to the small parts. Gears, springs, axles, transparent windows, decals and battery contacts are made in volume by specialists and kept as stock. A new toy is therefore assembled mostly from parts that already exist. That is the quiet secret of the apparently endless variety on a shelf: the variety is in the arrangement, not in the inventory. A different shell over a familiar chassis is a new product with almost no new tooling risk.
Why small orders are possible at all
Most manufacturing gets cheaper when you make more, because a long uninterrupted run spreads the setup cost. A district like this breaks that rule in a narrow but useful way. Setup is cheap because the mould is nearby, the machine time is rented by the hour, and the parts are already sitting on a shelf. The fixed cost of trying something is small, so a short run is not punished. A design can be made in a modest quantity, watched, and adjusted before the next batch.
That has a consequence for the object in your hands. Details change without the box changing. A later run may have a slightly different screw, a revised axle, a small rib added inside the shell. Collectors sometimes treat this as sloppy inconsistency. From the workshop's point of view it is the whole point of being located where it is: the second batch is the first batch with what everyone learned in between.
What the street cannot do
It is worth being plain about the limit. This system is extremely good at short cycles and shared tooling, and much less good at anything that needs one huge dedicated line, a long qualification process, or a material nobody nearby stocks. It also concentrates risk: when a single shared process is busy, everything waits, and when a rule changes about a coating or a small part, a hundred small workshops have to learn it at once. The same density that makes a correction fast makes a disruption common.
There is another cost that is easy to miss. Because the parts are shared, different brands can arrive at very similar objects. The street is a machine for producing variation cheaply, not for producing difference deeply. If you have ever compared two unbranded toys from different boxes and found the same seam in the same place, you were looking at the same mould family, not a coincidence.
How to read the back of the box
Turn a toy over and look for the parting line, the screw bosses, the way the wheels sit on their axles, and whether the painted details stop cleanly at an edge. These are not finishing details. They are a record of which decisions were made in steel, which were made in a spray booth, and which were made on an assembly table. Then look at the address. A district that fits on one street is not a romantic survival of old craft. It is a piece of equipment: a machine whose main moving part is the time it takes a person to walk a mould down the road and back.