Defining the problem
A negative externality of production is a spillover cost that the act of producing a good imposes on third parties, a cost that neither the producer nor the consumer pays for. The firm decides how much to make by looking only at its marginal private cost (MPC), the labour, materials and energy it actually pays for. But when production also pollutes, the true marginal social cost (MSC) lies above the MPC by the value of that external cost, so MSC equals MPC plus the external cost. There is no offsetting externality on the benefit side, so marginal private benefit and marginal social benefit coincide along the demand curve.
Left alone, the firm produces where MPC meets demand, at a free-market output that is larger than the social optimum where MSC meets demand. For every unit between the social optimum and the free-market output, the cost to society exceeds the benefit, so these units destroy welfare even though they are privately profitable. The market over-allocates resources to the polluting good, under-prices it, and leaves a deadweight welfare loss. That loss is the case for doing something.
Why intervention can be justified
A Pigouvian or carbon tax
An indirect tax set equal to the marginal external cost at the social optimum internalises the externality. It raises the firm's private cost so the firm now faces the full social cost of its decision, shifting the supply curve leftwards and pulling output back towards the social optimum. The revenue can fund the clean-up of the harm or compensate those affected. A carbon tax on a polluting industry works in exactly this way: it puts a price on each tonne of emissions so that abatement at the margin becomes worthwhile, and the output of the dirty good falls towards the level society would choose.
Evaluation. The decisive weakness is information. Governments rarely know the exact value of the marginal external cost, so the rate is likely to be set too high or too low, leaving the market away from the optimum. If demand is price inelastic, as it often is for energy and fuel, the tax cuts output only modestly while raising costs for consumers, which can be regressive and politically fragile. France's attempt to raise fuel taxes for environmental reasons in 2018 triggered the gilets jaunes protests and was shelved, a reminder that the distributional impact can sink an otherwise sound tax.
Tradable permits and regulation
An alternative is a cap-and-trade scheme, where the government fixes the total quantity of the harm and lets firms buy and sell permits. This guarantees the environmental outcome, because the cap sets the quantity, and it lets the market find the least-cost way to meet that cap, since firms that can abate cheaply sell permits to firms that cannot. Command-and-control regulation, such as a legal emissions limit, is simpler to grasp and signals strong disapproval, but it is blunt: it ignores differences in abatement costs between firms and can be costly to monitor.
Evaluation. The EU Emissions Trading System is the developed example here. Launched in 2005, it caps emissions from power stations and heavy industry across member states and lets them trade allowances. Its early years showed the central risk of permit markets: the cap was set too generously, the carbon price collapsed, and the scheme delivered little abatement. Only after the cap was tightened and surplus allowances were withdrawn did the price rise enough to change behaviour. The episode shows that a permit market guarantees the environmental outcome only if the cap is credible and binding, and that regulators face the same measurement and political pressures that complicate a tax.
Judgement
The impact of a negative production externality is real and persistent: the market over-produces, under-prices the harm and leaves a welfare loss that will not correct itself, so doing nothing is rarely defensible. Intervention is therefore justified in principle. But no single instrument is best in every setting. A tax suits a harm whose external cost can be roughly valued and whose source is diffuse, while tradable permits suit a harm where the total quantity matters most and abatement costs vary widely between firms; regulation suits the most dangerous harms that society wants stopped rather than merely reduced. The strongest response is usually a considered mix, matched to how measurable the external cost is, to the price elasticity of demand, and to the structure of the market, and reviewed as better data arrive. Intervention is justified, but it should be judged on design and information, not assumed to be optimal in itself.