should be treated as inputs to business activity rather than as an output of the economy, as is current practice.4 Similarly, commuting costs and other work-related consumer expenditures could be viewed as inputs to production rather than as outputs included in consumption (though it is not clear how these costs should be assigned for use in, say, productivity measurement). Conversely, some items now classified as intermediate inputs might better be classified as output for final demand. Researchers at the BEA have recognized this issue and changed the way they classify some market production. For instance, the BEA now classifies computer software purchases by businesses as investment rather than as an intermediate expense.
As with their market counterparts, nonmarket inputs and outputs must be properly classified for use in a double-entry accounting system and for the accounts to be useful for productivity analysis. Classification of market activities, much less nonmarket ones, is not always easy, but resolution of these classification issues will be a necessary step in the development of an expanded set of accounts.
Where the state of theoretical modeling efforts and available source data permit, it would be extremely useful if satellite accounts included estimates of externalities associated with the covered activities. In this respect, satellite accounts would differ markedly from the NIPAs. An externality is an effect from the action of individuals or businesses that either damages or creates a benefit to others with no corresponding compensation paid to or received by those who engage in the activity. The treatment of externalities is a central issue for environmental accounting. The most important applications relate to air and water pollution. Whether failing to account explicitly for the potentially large negative externalities associated with pollution distorts measures of aggregate output depends on whether only market output or at least some nonmarket output is adversely affected. In the first case, aggregate output is measured correctly, though its allocation across sectors may be incorrect; in the second case, if pollution is considered a negatively valued product, even aggregate output is measured incorrectly (Nordhaus 2004). Positive externalities are important for evaluating investments in education, which may benefit not only the individual receiving the education but also the society at large, through having a better informed citizenry, lower crime rates, and so on.
It is easy to see how information about the magnitude of externalities might be helpful to policy makers. Knowing more about the externalities associated with air or water pollution would be valuable, for example, in determining the