Returns to Scale Versus Diminishing Returns

Adding one more worker to a factory with a fixed number of machines feels different from doubling the whole operation because they are literally different experiments. Hire worker after worker onto the same ten machines and eventually each new hire adds less output than the last one, since they are crowding around scarce equipment. That is diminishing marginal returns, and it requires a fixed input to bite against. Returns to scale asks a completely different question: if you scaled up every input together, say doubling both workers and machines, does output more than double, exactly double, or less than double? There is no fixed input in that experiment, because everything moved together.

That is why returns to scale is fundamentally a long-run idea. The short run is defined as the period where at least one input, usually capital, is stuck. Diminishing marginal returns lives entirely inside that constraint. The long run removes every fixed input, so firms can rescale their entire operation, and the relevant question becomes how output responds to proportional changes in the whole input bundle, not to one input at a time.

This distinction shows up directly in cost curves. Diminishing marginal returns in the short run is what makes short-run marginal cost eventually rise. Returns to scale in the long run is what shapes the long-run average cost curve: increasing returns to scale produces falling long-run average cost, constant returns produces a flat stretch, and decreasing returns produces rising long-run average cost. The U-shape many firms show in the long run traces out exactly this sequence of scale effects.