The externality is in some ways a straightforward concept: yet, in others, it is extraordinarily elusive. We know how to take it into account in our analysis, and we are aware of many of its implications, but, despite a number of illuminating attempts to define the notion, one is left with the feeling that we still have not captured all its ramifications. Perhaps this does not matter greatly. The definitional issue does not seem to have limited seriously our ability to analyze the problem, and so it may not be worth a great deal of effort. Certainly, we do not delude ourselves that this discussion will be the last word on the subject.

The literature has also offered distinctions among a number of different classes of externalities. Some of these distinctions have been illuminating; others (including one proposed by the authors!) have been the source of some confusion. Consequently, it is useful to explore the different kinds of externalities and their implications for Pareto-optimal pricing. In particular, we shall examine with some care Viner's distinction between technological and pecuniary externalities; although economists have generally accepted Viner's distinction and its implications, propositions that overlook this point still appear periodically and yield misleading conclusions. We will encounter such a case later in this chapter.

We shall also examine the distinction between public-goods and private-goods externalities, which has its source in the seminal work of Bator and Head.