I have described only one of several approaches to using Pigouvian taxes to control pollution--an effluent fee, a fixed price per unit of pollution produced. At least two other ways of doing essentially the same thing have been discussed in the literature and employed in practice: auctioning pollution permits, and granting firms transferable pollution permits.

With an auction system, the EPA or equivalent first decides how much pollution it is willing to tolerate--say a million pounds per year of SO₂.It then auctions off permits for that amount of pollution, letting the market set the price. A firm that wants to emit a hundred pounds a year must buy permits for that amount. So the price of the permit plays the same role in this system as the effluent fee in the system described in the text.

Using an auction instead of an effluent fee makes sense if the EPA has better information about what the optimal quantity is than about how much damage one more unit of pollution does.Imagine a situation where a particular sort of pollution produces negligable effects below some level, and very serious effects above it, perhaps because that is the level that can be adequately dealt with by natural processes. By auctioning a number of permits slightly below the critical level, the EPA gets roughly the efficient amount of pollution, without having to somehow measure the amount of damage done by one more pound of pollution.

Under the third alternative, the EPA creates permits but instead of auctioning them off, it allocates them to someone else--typically to the firms currently doing the pollution. Suppose, for example, the EPA decides that a particular pollutant ought to be reduced to 90% of last year's level. Further suppose it knows how much of that pollutant each firm produced last year. It issues each firm permits for 90% of last year's pollution. The firms are then free to trade permits among each other. So if my firm finds that it can easily reduce its pollution to 50% of last year's level, it does so, and sells the surplus permits to other firms that find it hard to cut back even to 90%.

Here again, the price of the permit plays the same role as an effluent fee. This is obvious in the case of a firm that wants to emit more of the pollutant than it has permits for, and must buy additional permits in order to do so. It is also true, although less obvious, for a firm that is selling permits. The more it pollutes, the fewer permits remain to be sold, so part of the cost to it of an additional unit of pollution is the lost revenue from not being able to sell the permit for that unit to someone else.

Like the auction, a system of transferable permits makes sense if we have better information about how much pollution we want than about how much damage an additional unit of pollution causes. Its advantage is that the firms, which are made worse off by an effluent fee or an auction, are compensated by having the permits allocated to them, which is likely to reduce the political opposition to pollution control. One disadvantage is that firms that anticipate such a system coming into existence have an incentive to produce lots of pollution now, in order to be allocated lots of permits when the system is established.

This is one example of a more general issue: When legal rules change, how should the associated costs be distributed? At one extreme, the change can be designed to minimize its effect on the distribution of income--as with transferable permits. At the other extreme, it can be designed to punish past instances of the behavior that is now being penalized--for instance, by making the rules retroactive, requiring firms to buy permits to cover past as well as present pollution.

Consider the same issue in the context of one of the most controversial legal changes in American history--the abolition of slavery. One approach, actually followed to some degree in a number of other countries, was for the government to buy the slaves from their owners, and frree them. That has the advantage of reducing opposition to the change. The opposite approach is to not only free the slaves without compensating their owners--what actually happened in the U.S.--but to also make the owners liable to the freed slaves for the cost imposed on them by being enslaved in the past. This has the advantage of giving a slaveowner who anticipates the change an incentive to free his slaves before he has to, in order to reduce future liability. I discuss these issues in "Choosing Metarules for Legal Change."