A small plastic toy car sells for two dollars in a shop and leaves the factory for about forty cents. The gap is freight, duty, wholesale and retail, and it is not the interesting part. The interesting part is inside the forty cents, where a single number dominates everything: the price of the steel mould. Almost every odd feature of the toy industry — the huge minimum order quantities, the strange shapes that exist only because they are easy to eject from a tool, the fact that last year's model disappears rather than going on sale — follows from the cost of that block of steel.
The mould comes first
A mould for a simple toy car body is cut from a block of hardened steel in two halves, with a cavity shaped like the part. Molten plastic is injected into the cavity under pressure, cools, and the two halves separate along a parting line, leaving the finished part on one side. The tool is not one part: a car might need a body, two axles, four wheels, a windscreen insert and a chassis, and each is a separate cavity, often all cut into the same tool so that one shot produces a complete set.
Cutting that tool is skilled work done on computer-controlled milling machines, then finished by hand by a toolmaker who polishes the cavity surfaces until they are smooth enough that the part will release without sticking. A tool for a toy of this size costs on the order of several thousand to tens of thousands of dollars depending on complexity, cavity count and the precision of the finish. It also takes weeks to make, which is why the first unit of a new toy is expensive and the hundred-thousandth is cheap.
The tool also has a life. Every shot wears the cavity slightly, and after a few hundred thousand shots the edges that form the parting line round off and the part starts to show flash. A tool can be repaired and re-polished a few times before it must be replaced. That number — the tool life — sets the maximum total production of a design, and therefore its maximum total profit.
The arithmetic of the minimum order
Suppose the tool costs fifteen thousand dollars and can produce three hundred thousand shots before it needs replacing. If the factory sells ten thousand units, the tool cost per unit is a dollar fifty, which is more than the entire factory price of the toy. If the factory sells three hundred thousand units, the tool cost per unit is five cents.
This is the whole explanation for minimum order quantities, and it is not a negotiating tactic. A factory that accepts a small order at a normal unit price is either subsidising the buyer or planning to recover the cost somewhere else, usually by using a cheaper tool that will not hold tolerance. The number that matters is not the order size but the ratio between the tool cost and the total run. Below a certain ratio, the economics of the product collapse.
The same arithmetic explains why designs disappear rather than being discounted. Once a tool has been used for its intended run, keeping the design alive means either tying up a tool that could be re-cut for something new, or storing it in the hope of a reorder. Warehousing a tonne of steel is expensive, and fashion moves. The rational move is to scrap the tool and cut a new one, which is why toy ranges refresh so violently compared with furniture or kitchenware.
Inside the machine
Injection moulding is a cycle measured in seconds. Pellets of plastic — usually ABS for a rigid toy body, polypropylene for something that needs to flex, PVC for soft parts — are melted in a heated barrel and forced into the cavity by a screw that acts as both conveyor and piston. The pressure required is measured in tonnes of clamping force, because the plastic pushes the two halves of the tool apart as it fills. A tool for a toy car body might run on a machine with a hundred tonnes of clamp; a large toy might need a thousand.
Then comes cooling, which is where most of the cycle time goes. The part cannot be ejected until it is stiff enough to hold its shape, and the rate of cooling depends on wall thickness. This is the single most important design constraint in the whole toy, and it is invisible to the buyer. A wall that is 1.5 millimetres thick cools in a few seconds. A wall that is 3 millimetres thick takes much longer, and the difference multiplied across a production run of hundreds of thousands is the difference between a profitable toy and an unprofitable one.
Thick walls also cause sinks — a dimple on the surface where the material contracts as it cools. Designers work around this by keeping walls uniform, adding ribs instead of thickening, and placing the gate — the point where plastic enters the cavity — where a visible mark will be least noticed. Look at the underside of any hollow plastic toy and you will find a small circular scar. That is the gate, and its position was chosen so you would not see it.
The parts a machine cannot do
Ejection leaves the part with a thin ridge of plastic along the parting line, and with the sprue — the channel that fed the cavity — attached. Removing both is called deflashing, and on inexpensive toys it is done by hand with a knife. This is the most labour-intensive step in the whole process, and it is why the industry is sensitive to wage levels in a way that moulding itself is not. A moulding machine runs unattended overnight; a row of people trimming flash does not.
Decoration comes next, and there are three routes. Pad printing transfers ink from an etched plate via a silicone pad that conforms to curved surfaces, one colour at a time, and it is fast and cheap for simple graphics. Spray painting through a mask gives softer gradients and is used for anything that needs to look moulded-in rather than printed. Water-slide decals give the sharpest detail but are slow and fragile, and they survive mainly on models and collectibles where the buyer will not handle the object roughly.
Each route adds a station to the line, and each station adds handling. A toy with four printed colours on two faces needs eight passes, and each pass is an opportunity for misregistration. The painted eye that sits a millimetre off centre is not a printing error; it is a registration error, and it is the most common defect you will find on inexpensive toys.
Assembly, then testing
Many toys are assembled by hand: wheels pressed onto axles, inserts clicked into bodies, batteries fitted into compartments. This is also where the small parts that safety standards worry about get attached, which is why the assembly station is usually where quality checks concentrate. A screw that was not driven fully home is a hazard, not a defect.
Testing is a mix of sampling and destruction. Drop tests from a specified height onto a specified surface, torque tests that try to twist off any part a child could grip, tension tests that pull with a calibrated force, small-parts cylinder tests that determine whether a component would fit entirely inside a child's throat. Most of these are destructive, which is why they are run on samples rather than on the whole batch. The batch is then presumed to behave like the sample, and that presumption is the entire basis of toy safety certification.
Why the factories sit together
Almost none of this happens in one building. A toy is typically a collaboration: a design studio, a toolmaker, a moulder, a painting shop, an assembly workshop, a packaging printer and a testing lab. In mature toy regions these sit within a short drive of one another, sometimes within a few kilometres, and the reason is not sentiment. It is that a design change during development can require a new tool, a new colour trial and a new drop test, and each of those needs a conversation, not an email. When the toolmaker is ten minutes away, an engineer can walk a problem across and have it fixed the same day. When the toolmaker is on another continent, the same problem costs two weeks.
That is the real advantage of a toy cluster, and it is why clusters persist even as wages rise. The cost of a mould is not only the money; it is the time to iterate. Places that can compress that time keep the industry, and places that cannot eventually lose it — not to cheaper labour, but to faster feedback.
The two-dollar toy on the shelf, then, is not cheap because someone was paid little. It is cheap because a fifteen-thousand-dollar tool was spread across three hundred thousand identical objects, because the wall thickness was chosen to cool in four seconds, because the flash was trimmed by hand in a country where that hand was affordable, and because every supplier in the chain was close enough to argue with. Remove any one of those conditions and the price moves. That is the whole machine, and it is worth understanding before you look at the object and see only plastic.