costs are marginal costs—that is, the incremental costs of complying with the rule—rather than average costs. There are issues of how joint costs or products are treated in the determination of engineering cost estimates. For example, the removal of one type of contaminant may be much less costly if industries have already installed treatment processes for other contaminants; similarly, if a control technology will decrease the emissions of a number of pollutants, it is difficult to know what portion of the costs of installing and maintaining that control technology should be attributed to regulating just one of those pollutants. In such cases the marginal cost of removing the contaminant can be substantially overstated if the other pollution control activities are not considered. EPA often estimates marginal costs and accounts for spillover effects, such as joint costs, in its analyses (EPA, 2011b; NAPEE, 2008). For example, EPA uses a model that accounts for the control of multiple pollutants (sulfur oxides, nitrogen oxides, directly emitted particulate matter, and carbon dioxide) in its regulatory impact analysis for mercury (EPA, 2011b).
There is also likely to be uncertainty concerning the number of households, firms, or systems (for example, water systems)16 that may be affected by a rule and also concerning the methods that the regulated entities will use to comply with the rule. Uncertainty is even greater when EPA sets a national standard and agencies at a lower level of government, such as state agencies, implement the rule. In such instances, in addition to the issue of how firms will actually change to meet the new standards, there is additional uncertainty concerning how other units of government will implement the new standard. Once again, however, systematic inaccuracy is unlikely to occur, except in those cases in which a problem with compliance is known or anticipated.
Other sources of uncertainty that are sometimes relevant are the level of enforcement, the productivity of such enforcement efforts, and, subsequently, the compliance with the rule. Further increasing the uncertainty associated with compliance is the fact that in some cases lower levels of government enforce EPA’s regulations. The simplest approach for dealing with such uncertainty is to assume complete compliance—in other words, 100 percent enforcement. EPA’s guidelines recommend that when conducting regulatory impact analyses, analysts should, as a general rule, assume full compliance (100 percent) with EPA regulations (EPA, 2010). The guidelines recommend departure from using the “default” of full compliance only when there is sufficient data to calculate the true compliance rate (EPA, 2010). This level of enforcement may be higher than either the level that is socially optimal (that is, the one at which the marginal cost of
16 See, e.g., Federal Register, November 22, 2001, p. 47, regarding an estimate of water systems affected by a proposal rule.