enforcement equals the marginal benefit of enforcement) or the level that occurs in practice. Estimates made with this assumption, therefore, should be considered to be high estimates of enforcement cost. Alternatively, EPA could estimate a range of costs using different percentages of firms complying with the laws. Whatever level of enforcement is assumed should be assumed throughout the analysis, including in the computation of benefits.
A few studies have compared the compliance costs estimated in regulatory impact analyses to estimates of actual compliance costs incurred after a regulation has been put into effect (Harrington, 2006; Harrington et al., 1999; OMB, 2005). Those comparisons indicate that compliance costs are often overestimated. OMB (2005) also concluded, however, that benefits are overestimated to a greater extent than costs, so that economic analyses typically predict that the performance of a regulation will be better than actually occurs. Harrington (2006) also found that although total costs were overestimated, unit costs were not, and he found “no bias in estimates of benefit–cost ratios.” Regardless of which analysis is more accurate, all of those analyses demonstrate the uncertainty that is inherent in predictions of compliance costs (and benefits). And as discussed by Harrington (2006), these experiences also demonstrate the importance of conducting analyses after the implementation of a regulation (so-called ex post analyses) to evaluate and improve the methods used for predicting costs and benefits.
When a rule is to be implemented over a number of years, additional uncertainties arise. For example, input prices (that is, the price of inputs to a process, such as the price of low-sulfur coal) might vary with time. Moreover, as discussed earlier, the costs of the technological changes and equipment necessary for a plant to comply with the rule might change. For example, the promulgation of a rule on a national basis might increase the market size for an innovation that improves environmental quality. Such new technologies may be more productive in achieving a particular environmental goal, and in some cases the purchase prices of equipment incorporating the new technologies might be lower. At the time the rule is being considered, however, there is considerable uncertainty about how innovators will respond to the rule (that is, the amount of investment in research and development that will be forthcoming in response to promulgation of the rule), the yield from such investment, the time frame within which any yield will occur (that is, when the innovation will occur), and at which price the new technology will be marketed. A practical solution is to assume a worst case in which no innovation takes place, but such an assumption might underestimate the net benefit of innovation to the extent that there is an overestimation of the costs.
Estimates of compliance costs often must be made by estimating the number of facilities currently not in compliance with a proposed standard or rule, the magnitude by which those facilities would be out of compliance,