The taxes and subsidies with which we shall be concerned are the corrective or Pigovian ones that could in theory be used to bring marginal private costs or benefits more closely into alignment with marginal social ones. The need for alignment arises when externalities (whether economies or diseconomies), operating at the margin, cause a divergence.

The name of Pigou is often associated with the idea, although his own statements are very cautious (1932, p. 381; 1947, pp. 99–100). That the matter is more complicated became clear from later work of Coase (1960), Buchanan and Stubblebine (1962) and others. A good summary is to be found in Turvey (1963).

Externalities may be between firms, where they are caused by technological interdependence between production functions, one firm’s containing an input or an output proper to another’s. They may be between consumers, one’s utility function containing a variable proper to another consumer. Or they may, by an obvious extension, be between consumers and firms. Examples will be given later.

Optimizing behaviour by consumers and firms implies that marginal private costs and benefits will be equalized. If these coincide with social costs and benefits, they too will be equalized; if not, there will be a measure of market failure. Market failure means that all mutually beneficial bargains have not been struck and that it is possible to make one or more participants better off without making anyone worse off. Efficient markets (or ‘Pareto efficiency’) imply the exhaustion of all such opportunities.

The purpose of a corrective tax is to bridge the gap between private and social cost (if that is the side of the market we are looking at) created at the margin by an externality, and to vary with it. (As will shortly appear, this is much easier in theory than in practice.) The idea is to bring the externality to account, as it would be brought to account if internalized by a merger, not to eliminate it. It is to make an otherwise inefficient market simulate an efficient one, achieving by fiscal intervention what might (if transaction costs had not been too high) presumably have been achieved by direct negotiation between the parties.

A simple example (still on the cost side) will help to make these ideas more concrete. Assume that factory A discharges effluent into a river, increasing the purification costs of factory B, situated lower down, where it draws its water. Assume that B’s costs of purification depend on the method chosen and vary with its product mix and levels of output as well as with A’s output and expenditure (or lack of it) on filtration. (This example is ‘simple’ because at least the causation runs one way only, from A to B. If both factories drew water from a lake into which both discharged effluent, reciprocal externalities could give rise to the sort of situation commonly encountered in Game Theory.)

The complexity of designing a tax on A that would correctly measure the costs imposed on B, and vary with them (rather then with A’s activity) as A adjusted to it (by changing the quantity and quality of water discharged) and B responded, is evident. It is also evident that, to simulate the operation of a market in which there was direct negotiation between A and B, the proceeds of the tax would have to go to B, not to the fisc. This is a simple point, often overlooked, but recognized in the older literature when reference was made to systems of taxes and subsidies. The latter were to go not only to those creating external economies but to those suffering external costs.

From the community’s point of view the most efficient (i.e. cost-effective) configuration of water treatment systems could be along the following lines. Factory A reduces its discharge of effluent (or improves its filtration) by less than it would if taxed an amount equal to the damage caused. Instead it seeks to limit the damage by sharing the costs of operating and up-grading B’s purification plant. This could be much more efficient, offering gains to both parities. It is an outcome that could be achieved by encouraging negotiation between the two firms, or allowing them to merge. It could not be achieved by a unilateral tax.

Internalizing an externality by merger is possibly only between firms, but negotiation is possible in a wider context (between consumers, or between consumers and firms) whenever the numbers involved are not large. The that the numbers often are large. Consider motorists creating congestion on a public road. Negotiation to reduce the external costs that each imposes on the others is hardly feasible. The transaction costs would be enormous. Some sort of second best solution is to be preferred. A tax on fuel, licence fees and highway rules are among the possibilities. They would at least help to reduce congestion, or limit its consequences.

Corrective taxes are best seen in this light. They are seldom a practical way of achieving Pareto efficiency. But they could be part of a second best solution to the problem of market failure. As such they are to be weighted against other second best solutions such as licences, zoning laws and outright prohibitions. In the river pollution example, if there were too many firms to make negotiation feasible, a law prohibiting the discharge of effluent, or even one prohibiting the use of river water for industrial purposes, would clearly eliminate the externality. But it would not bring about a particularly efficient state of affairs. In such circumstances the use of admittedly rough corrective taxes might be a serious alternative.

In other circumstances a combination of measures might be appropriate. A tax on tobacco is almost always designed with an eye to revenue, the demand for the commodity being inelastic, but the acceptability of the tax is probably due to widespread recognition of the external costs imposed by smokers on non-smokers. The sheer difficulty of trying to measure these costs makes a genuinely corrective tax all but impossible. So partial prohibitions in the form of non-smoking areas are common. Their creation is a much more practical approach to a common social problem than attempting to tax smokers whenever they cause inconvenience to others – or suggesting that non-smokers should pay smokers to refrain from indulging.

This brings us to a final point. In our original example, if A and B were brought to the negotiating table, the outcome of the negotiation would be very different if the laws, customs or conventions of the society gave riparian owners (a) the right to discharge effluent or (b) the right to draw clean water. The two regimes would result in two very different distributions of the potential gains. But under either regime an efficient outcome would be possible – one in which all opportunities for mutually beneficial bargains had been exhausted. Corrective taxes and subsidies are concerned with the latter, not with the distribution of the gains from trade – that is to say: with efficiency, not equity.

See Also