| 摘要 | Government programs are expected to bring benefits; they also have costs. For many of these programs, the total (or social) costs imposed on society are largely the same as their administrative costs and thus are largely measured by the fiscal budget—over which there are direct legislative controls. But this is not true for regulatory programs. In their case, most of the social costs are not reflected in the fiscal budget. Instead, they are borne by the private and public organizations being regulated and are, therefore, not subject to legislative controls. This situation has become a matter of growing concern. With increasing frequency in recent years, legislation authorizing a new regulatory program has stated broad goals but then given the agency broad discretion on implementation. There are no direct constraints on the magnitude of the cost burden that can be imposed on society to achieve these goals.
The regulatory budget is one proposal for dealing with this phenomenon. It has been suggested largely in the context of health, safety, and environmental regulation and takes its cue from the normal fiscal budget for government. Just as Congress authorizes broad fiscal programs but then allocates specific spending budgets for each agency for each fiscal year, Congress could pass broad regulatory programs but then place annual limits on the costs that each regulatory agency could impose on the sectors it regulates. Thus, each regulatory agency would have its own regulatory budget, and there would be a total regulatory budget for the entire federal government. The legislative process for this new budget could parallel the legislative process for the existing budget. (For greater detail on many aspects of the regulatory budget, see the preceding article by Christopher C. DeMuth.)
The Major Problem
There are, unfortunately, a number of problems with the concept of the regulatory budget. For example, to administer the proposal, either the management and oversight capability of the Office of Management and Budget would have to be greatly enlarged or some new budgetary agency of at least equal size would have to be created. Also, the budgetary burdens on Congress—apparently onerous even now, judging by the delays that plague the appropriations process—would become much heavier.
Perhaps the most serious impediment to the operation of a regulatory budget, however, lies in the determination of the costs that the regulatory agencies impose on the sectors they regulate. In specific instances of health, safety, and environmental regulation, the regulators’ and the regulatees’ estimates of these costs have often differed by a factor of five or ten. This should not be surprising. Frequently, the cost estimates apply to technologies that have not yet been developed or perfected, and great uncertainty prevails. Furthermore, the regulators have an incentive to underestimate costs, so as to understate the cost burden they are imposing on society; while the regulatees, on the other hand, have the opposite incentive, so as to convince the regulators (or, ultimately, Congress) to reduce the stringency of the regulations. Even in retrospect, the costs of regulation may be difficult to determine, since the consequences of forgone opportunities may be difficult to measure: How much consumer welfare was lost because a potential product was not produced? What are the true social costs of building a factory at a third-choice location?
Thus, in contrast to the case of the normal fiscal budget, in which a dollar is a dollar and there are few disagreements on actual costs, it is likely that wide disagreements on regulatory costs would bedevil efforts to draw up a regulatory budget in practice. This problem would probably persist even if a standardized methodology were adopted for determining such things as discount rates for future costs, wage rates, the value of individuals’ time, and so on. Moreover, agencies would have strengthened incentives to understate the costs they were imposing—so they could carry out more regulatory activities within their given budgets—while the regulatees’ incentive to overstate these costs would remain. The monitoring and adjudication burden of the central budgetary agency would be likely to be enormous.
A Solution
One way to solve this problem would be to devise a “truth in regulatory costs” scheme that created incentives for regulators and regulatees alike to state accurately their best estimates of the real social costs of the regulations being imposed. For one important class of regulations—those that impose standards, like emission standards for automobiles or safety standards for work places—such a scheme is possible. It requires the institution of noncompliance fees—that is, penalties paid by regulatees if they are not able to comply with the regulation. Noncompliance fees are becoming accepted as part of the enforcement strategy of regulation. They are required, for example, for stationary sources of air pollution and for heavy-duty truck exhaust emissions under provisions of the Clean Air Act Amendments of 1977.
With the principle of a noncompliance fee established, we need three further conditions for our “truth in regulatory costs” scheme to work: (1) the noncompliance fees should be set at levels that are roughly equal to the marginal (or incremental) costs of compliance; (2) the marginal costs and the total costs of compliance should in actual practice bear some reasonable relationship to each other; and (3) the regulators should prefer that the regulatees meet the standards rather than escape them by paying the noncompliance fees.
The first requirement would have to be established by legislation or by administrative rulemaking. It is worth nothing that the Environmental Protection Agency has established the principle of noncompliance fees equal to the marginal cost of compliance for heavy-duty truck exhaust emissions. If regulatory standards are set at roughly the point at which marginal social benefits are equated to marginal social costs, a noncompliance fee would have most of the properties of a properly structured effluent fee.
The second requirement implies a two-fold condition. First, the marginal costs of compliance should in practice be constant or rising as standards become more stringent—in other words, each increment in stringency should cost at least as much as, or more than, the previous increment. Most regulatory standards appear to generate costs that fit this pattern. Second, the fixed costs of compliance should not be dominant—in other words, the initial costs of compliance, irrespective of stringency levels, should not be relatively large. If these two conditions are met, projections or extrapolations of the total costs of compliance from the marginal cost estimates would provide roughly accurate estimates of the true total costs of complying with the regulations.
Finally, the requirement that regulators prefer regulatees to comply with the standards rather than pay the fees would be easily met: It certainly characterizes the current and likely future preferences of regulators as a class.
Let us now examine how and why this “truth in regulatory costs” scheme would work. Recall that in a regulatory budget environment a regulatory agency would want to understate the costs of regulation. But if it did so under our scheme, it would have to set noncompliance fees that were too low—which would, in turn, allow the regulatees to escape compliance too cheaply. Typically, however, because regulators are more interested in gaining compliance than in collecting penalties, they would not want this last action to occur. Thus, the regulator would have an incentive to avoid underestimating the costs of regulation. At the other extreme, if the agency were to overestimate the costs of regulation, it would then have to set high noncompliance fees—which would, of course, eliminate cheap avoidance, but the overestimate would absorb more of the agency’s regulatory budget and leave less room for other regulatory actions. Again, a self-correcting mechanism would be introduced. (It is not clear, though, whether overestimation of regulatory costs would be a means of obtaining increased budget allocations for future years. Would the claim that a regulatory program is more costly lead to greater or smaller budget allocations?)
As for the organizations being regulated, clearly they would no longer wish to overestimate regulatory costs, because they would prefer to have low noncompliance fees. But might they have the opposite incentive—to underestimate regulatory costs so as to bring about cheap noncompliance fees? Probably not, because underestimating might also lead to added and more stringent regulation. So, again, a self-correcting mechanism would be created.
The self-correcting scheme for “truth in regulatory costs” outlined here (and there are probably other similar schemes that might be devised) is somewhat analogous to the scheme sometimes suggested for the self-assessment of property taxes: Individuals or companies, it is argued, should be allowed to assess the value of their own property for tax purposes, but must be willing to sell the property to anyone prepared to buy it at that price. Given these conditions, taxpayers would not want to set too high a value, since they would then pay high taxes, but also would not want to set too low a value, for fear of having to sell the property at too low a price.
Finally, aside from being a key element in the “truth in regulatory costs” scheme, the incentive for regulatory agencies to estimate the marginal social costs of compliance in a nonbiased fashion should also have beneficial consequences for the regulatory process generally. Sensible decisions on how stringent a standard should be require accurate estimates of the marginal social benefits and marginal social costs of different levels of stringency. Or, alternatively, if the situation is one in which benefits are difficult or impossible to measure, cost-effectiveness calculations—comparing the marginal costs at which different regulations achieve a specified improvement toward a given goal—provide a means of improving regulatory efficiency. Accurate estimates of the marginal social costs of compliance should help improve decisions on the stringency of the standards.
Conclusions
The “truth in regulatory costs” scheme developed here would, of course, be unsuited to regulatory situations in which noncompliance fees were not applicable. Such would be the case, for example, if a regulatory agency considered a product wholly unsafe and were intent on banning it, no matter what the cost. Other examples of regulations that can have large social costs but no direct compliance aspects are the national ambient air quality standards established under the Clean Air Act (in this case, the compliance requirements are created by the subsequent regulations setting emissions standards for specific sources) and rules specifying certain regulatory procedures (say, testing procedures for vehicle emissions). In cases of these kinds, a different scheme would have to be devised.
Nevertheless, the number of regulations involving standards or other requirements that could carry noncompliance fees appears to be large enough so that the scheme proposed here would have substantial utility. Anyone interested in a regulatory budget should also be interested in noncompliance fees. ■
Lawrence J. White is professor of economics at New York University. |
