Towards a Regulatory Budget:
Jim Tozzi, ed. 1979.
PART 3: ALTERNATIVE CONCEPTS OF A REGULATORY BUDGET
Data on the costs of regulation would be a necessary raw material for a regulatory budget, but a regulatory budget would differ from a periodic compilation of cost estimates in two respects. First, to the extent possible, uniform methods and definitions would be employed, so as to facilitate comparison of the costs of various regulatory programs. Second, and much more important, a regulatory budget system would be designed to present and/or use data on the costs of regulation as a management tool. This does not mean that a regulatory budget would necessarily reduce the costs of individual regulatory programs, or prevent these costs from growing. Like the expenditure budget, a regulatory budget would have considerable flexibility, and resources could be shifted between regulatory programs and perhaps between other areas of government activity and regulatory programs. Nevertheless, a regulatory budget would in some way be directed to the problem of controlling the costs of regulation.
It is equally important to recognize that the term regulatory budget doesn't describe a single idea but, instead, covers a range of fundamentally different systems. For example, budget ceilings might be for informational purposes only or legally binding; a regulatory budget might include all costs due to regulation or only direct compliance cost; and the budget might include all regulatory agencies or only selected regulatory activities.
This paper discusses the options on three major aspects of a regulatory budget (RB):
Table 1 summarizes the options that are described under each of these headings. The discussion first goes through each of the topics shown in this table (which is effectively a table of contents of the first three sections). Section 4 then turns to the problem of combining options on the various aspects of an RB into complete packages, whichleads to a summary of three broad alternative conceptions of an RB.
OPTIONS FOR AN RB
a. Informational - RB divorced from review process
2. Cost Coverage
a. Direct and Indirect Compliance Costs
3. Scope and Structure
a. All Agencies
It is often assumed that an RB would be "like the expenditure budget," and, granting this assumption, the role that an RB would play seems clear. In particular, budget ceilings would presumably be proposed by the President; would be subject to Congressional approval; and once set, would be binding on the agencies. In short, on this view, the RB would itself act as a control mechanism. But, control in only one of the roles that a budget can play. Another, and basically different role, is that of a planning mechanism. That is, a budget may serve only as a means for identifying broad policy options and informing decision makers, while operating decisions are not directly based on the budget, or controlled through the budget system. It is between these two poles--information and control--that the specific function of an RB would be assigned.
An informational RB would be defined by two features, both negative in character:
Put in positive terms, an informational RB would be a tool for characterizing the size of regulatory programs, the costs of regulation, and means for improving regulation.
An informational regulatory budget would have two great advantages. First, it would avoid virtually all of the very difficult legal and political problems that arise to the extent that an RB would be designed to have a direct role in effecting individual regulations. Second, an informational budget would be relatively easy to do, since the task could initially be defined as that of drawing together available information on the costs and benefits of regulatory programs.*/ The principal disadvantage of an informational RB would lie in uncertainty as to its impact. The key questions in this respect are, whether provision of coherent information on costs would play a substantial role in obtaining solutions to the perceived problems of regulation, and whether a budget format is appropriate to oversight of regulation.
*/ The most important of these from the standpoint of an RB are the data on costs compliance with environmental regulations compiled by BEA; CEQ's estimates of the projected costs of environmental regulation; and the analyses of major regulations required by Executive Order 12044.
Between an informational RB and an RB designed as a control mechanism, there is another option which will be referred to as an "advisory" RB. In this option, the budget ceilings would not be legally binding, but the RB would be assigned a specific advisory role. For example, the RB might be transmitted annually by the President to the Congress, together with recommendations for legislative action on regulatory programs.*/ Alternatively, the RB might be given a place in 0MB budget reviews or the RARG process redesigned to make use of the RB.
The differences between the pros and cons of an advisory and an informational RB are matters of degree. Depending on the nature of the role assigned to it, an advisory RB might present legal and procedural issues, and it would probably have to observe higher standards of accuracy and uniformity than would be required of an informational RB. Against these points can be set the likelihood that an advisory RB would be more effective than an informaitional RB.
The remaining option on function is that of a "control" RB. In this case, the budget ceiling would be legally binding. The budget ceilings, along with the means for enforcing the ceilings, would then constitute a mechanism for controlling regulation.
The advantages and disadvantages of a control RB are the reverse of those of an informational and advisory RB. A control RB could be very effective but, in terms of legal, political, organizational and technical issues, it is by far the most difficult of the three options identified here.
The most important aspect of the design of an RB is the decision on the function I would serve. This is also the most basic issue, as it would guide decisions on other aspects of an RB. However, the decision on the function of an RB leaves open some on the costs included in budget ceilings and the structure of an RB. Consequently, these topics are considered separately before going into a discussion of complete options for an RB.
*/ An informational RB would resemble the reports by the President to the Congress are typically a part of sunset proposals.
Direct and Indirect Costs
The costs of regulation can be divided into direct and indirect compliance costs. Direct compliance costs include capital costs, operating costs, paperwork costs, and other costs (such as R& D) incurred in complying with regulations. Indirect costs are all other costs to consumers and sellers of individual products attributable to regulation.
It is necessary to recognize two broad options on the treatment of costs:
It is generally agreed that the second of these options is preferable in principle, as indirect compliance costs are as real as direct compliance costs, and in some cases are the predominant element of total cost. Moreover, there is a question as to whether indirect costs could be excluded, as doing so might lead to unacceptable distortions in decisions based on an RB.
The considerations on the other side of the issue--which are disputed--go to a range of technical problems and related organizational issues. The pivotal issues raised by indirect costs, then, are not matters of principle, but of practicality:
Without attempting to fully resolve these issues, the discussion which follows points out the major considerations involved.
*/ Government expenditures on regulatory programs would also be considered in some in an RB system, but how this would be done is not a key point.
Measurement Problems: The definition of direct compliance costs would involve a number of problems; e.g.:
These problems are not intrinsically different, or more difficult, in the context of a regulatory budget than they are in other contexts, for example, the tax laws. Consequently, it is reasonable to assume that established accounting procedures would provide a basis for defining the direct compliance cost of regulatory programs.*/ It is also relevant to note that direct compliance costs are expenditures of identifiable organizations or individuals and so could be compiled through conventional reporting and accounting techniques.
Indirect compliance costs present a different situation in two respects. First, indirect costs do not show up as identifiable expenditures required by regulation. Indirect costs are, instead, costs due to various consequences of regulation, (e.g., effects on prices and their measurement typically requires use of complex econometric models. Second, the nature of indirect costs varies from one situation to the next, and depends on the details of the regulation and the facts of the economic activities affected.
It is necessary before pursuing these points to distinguish the indirect costs of regulation from the effects of regulation on the inflation rate, unemployment, productivity, and other macro-economic magnitudes. Just as the macro-economic impacts of the expenditure budget are estimated, so, presumably, would those of an RB.
*/ The fact that public utility regulation amounts to the construction of a type of RB (which includes only direct compliance cost) for the regulated firm is a strong argument in favor of this assumption. For example, it is relevant to know the effects of regulation on measured inflation rate and on productivity in various sectors of the economy. However, indirect costs as defined here refers to specific costs (caused by regulation) borne by individuals and firms, which is different from the macro-economic impact of regulation.*/ For the sake of clarity, then, it is best to maintain the macro-economic impacts of regulation and the indirect costs of regulation as separate categories.
No standard breakdown of various types of indirect costs of regulation has ever been developed because, as was noted above, that the character of indirect costs varies from one situation to the next. However, for the purpose of gauging problems of measurement, it is informative to distinguish five categories of indirect costs of regulation:
*/ Macro-impacts can perhaps be reduced to dollar measures of real resource cost. However, such an exercise would be contentious because of the difficulties of defining and measuring the real resource cost of a given impact. Suppose, for example, that it is established that some (large) set of regulations increases the inflation rate in some year by 0.5 percent. The question then is: what is the real resource cost of a 0.5 percent increase in the inflation rate? This is obviously a difficult question both in principle and in practice. It would also not be to a good point to address such questions as it is macro-impacts (not dollar measures of cost) that are used in stabilization policy.
**/ For example, EPA's Prevention of Serious Deterioration (PSD) program seeks to prevent deteriorations in air quality in parts of the country that are in compliance with natural ambient air quality standards. The PSD program has the potential for causing plants to be located on sites more costly than would be chosen in the absence of the PSD program and, in some instances, plants will not be built at all because of the PSD program.
***/ Delays as such are not necessarily a real resource cost, but delay can have real resource costs; especially it may lead a firm to invest in an interim facility that would not be required if decisions could be made more expeditiously.
5. Negative benefits; e.g., costs to patients due to the unavailability of drugs eventually proved to be safe and efficacious.
It should be stressed that, in individual cases, there may well be indirect costs that cannot be appropriately placed under one of these headings.
A brief discussion of the first of the categories just listed is sufficient to suggest the magnitude of the problems involved in measuring indirect costs. Part of the direct compliance costs are typically passed on in the form of higher prices, which implies that these costs are partly borne by regulated firms, and partly borne by their customers. Another point--which is particularly relevant to the definition of indirect costs-- is that the higher prices mean that the amounts sold (and consumed) are lower than they otherwise would be. The value of the foregone output to consumers plus the loss in profits to firms due to the reduction in output measures the total costs of the regulation to buyers and sellers. Indirect costs can, then, be defined as total cost less direct compliance costs. This measure is analogous to the deadweight loss due to a tax. If the industry is purely competitive and the industry's long-run supply curve is perfectly elastic, then indirect costs so defined are zero. However, if these assumptions do not hold, then indirect costs due to direct compliance costs are positive, and their measurement requires estimates of both the industry's supply and demand curves.*/ Moreover, to the extent that downstream industries are not perfectly competitive, direct compliance costs lead to second, third, and so on, round indirect compliance costs.
As these comments suggest, measuring the indirect costs of a regulation would typically be a major undertaking. However, there are examples of studies that contain estimates of each of the types of indirect costs identified here. Consequently, it cannot be concluded that it is simply impossible to measure indirect compliance costs.
One part of the problem presented by indirect cost is a set of organizational issues, which stem from technical requirements. Studies of indirect costs must be "one of a kind," in the sense that their specific features must vary from one case to the next.
*/ If the direct compliance costs are relatively small, it might be sufficient to have estimates of the elasticities of supply and demand.
They would also have to be updated periodically, and the costs concepts, methods and basic assumptions used would obviously have to be consistent across studies of different regulations. The first of these considerations implies that studies of indirect costs could not be routinized, while the second points to the necessity of codifying the methods to be used in measuring costs and benefits. Without further elaborating these points, it is clear that including estimates of indirect costs in an RB would involve substantial questions of organization staffing and authority.
A much more difficult question concerns the accuracy of economic measurements of costs. Indirect compliance costs cannot be simply counted, but must be inferred economic models to complex situations and the resulting estimates will invariably contain errors, which will often be large. The question that must be asked, then, is "h good enough"?
This question cannot be answered by reference to some abstract standard of accuracy. what is "good enough" for one purpose is not necessarily "good enough" for another is to say that whether the errors in estimates of indirect costs are tolerable c. determined with reference to the purpose that the RB is intended to serve. As i further below, from this perspective, it seems reasonable to assume that standards for accuracy and clarity of cost estimates must be higher to the extent that the RB a means of control, rather than only for informational purposes.
Problems of Excluding Indirect Costs: These comments have been directed to the question of whether, as a practical matter, estimates of indirect costs could be included in a regulatory budget. An equally pressing question is whether indirect costs could be excluded, as their exclusion might systematically distort decisions based on an RB. A similar problem is present in the expenditure budget, which suggest that it would be unreasonable to evaluate this aspect of an RB in terms of some unobtainable standard of perfection. The proper test is whether exclusions of indirect costs would lead to unacceptable distortions in any decisions that would be based on an RB.
The answer to this question turns, in part, on how wide a range. of different regulatory programs would be included in an RB. This consideration is important because the major costs of economic regulation tend to be indirect costs, while social and environmental regulation typically involve substantial direct compliance costs. Clearly then, comparisons of, for example, the ICC with OSHA on the basis of direct compliance costs would be grossly misleading. Similarly, comparison of the ICC and the FMC--again, on the basis of direct compliance costs--would be to no good point, as the bulk of the costs in case are indirect.
An RB limited to regulatory programs that work only through direct compliance costs would escape this objection, as the only. indirect costs involved are those due to direct compliance costs, and omission of these is unlikely to be a problem.*/ But this point is not very helpful, because a number of important regulatory programs involve both substantial direct and indirect compliance costs. Consequently, it is necessary to assume that an RB that excluded indirect costs would seriously understate the costs of some programs and would present dangers if the system involves making trade-offs between regulatory programs.
One option is to exclude indirect costs and accept these dangers. Another is to include estimates of indirect costs in the budget ceilings, which would present the very formidable organizational and technical problems outlined above. A third option would be to present "special analyses" of the indirect costs of those programs included in the system that impose large indirect costs. Estimates of indirect costs would, in this case, not be included in the budget ceilings because the estimates would be made only for selected regulations and might be updated less frequently than the figures for direct compliance costs.
The option of estimating indirect costs for only selected regulations (and not including the estimates in the budget ceilings) would have three great advantages. First, it would provide a reasonable response to the problems noted above; i.e., serious understatement of the costs involved in some regulatory programs and biases in comparison of programs. Second, it could greatly reduce the amount of work entailed by an RB.
*/ Indirect costs due to direct compliance costs are likely to be very roughly proportional to the direct compliance costs. To the extent that this is the case, omission of indirect costs due to direct compliance costs will not bias comparisons of regulatory programs. Moreover, indirect costs due to direct compliance costs are likely to be a fairly small fraction of direct costs. It should be stressed that the suggestion here is only that omission of indirect costs due to direct costs would probably not involve gross error, and not that it would involve no error at all.
Finally, use of "special analyses" would provide a means for identifying and appraising the uncertainties present in estimates of indirect costs that would not be available if such estimates were only stated in budget ceilings.
These comments suggest that, subject to two qualifications, it would not be necessary to include indirect costs in budget ceilings. The first qualification is simply that it would be fruitless to include an economic regulatory program in an RB system unless the indirect costs of the program are estimated. Second, while estimates of indirect costs might be excluded from budget ceilings, provisions should be made to include in an RB document analyses of indirect costs for those regulatory programs which impose 1arge indirect costs.
Scope and Structure
The expenditure budget is set up on an agency basis, and all agencies are included. An RB might follow this same pattern, but this is not necessarily the case. There are choices on both what agencies would be included, and on the way in which the RB would be structured.
Scope: One possible approach to the problem of specifying what programs to include an RB starts from the premise that an RB is a tool for controlling the costs that regulation imposes on the economy. Granting this point, it seems that coverage should be as broad as possible. Specification of the scope of the budget then becomes a matter of excluding agencies on the basis of practical considerations.
Many agencies could be excluded from an RB simply on the grounds that the costs associated with their regulations are small. Beyond that, there are two sorts of regulatory activities which clearly should be excluded from a regulatory budget:
The first of these criteria would exclude the IRS and the Customs Bureau. The second would exclude the Social Security Administration (and other income transfer programs), government procurement, and similar activities in which the agency's functions are largely executed through some means--making transfer payments, buying, etc.--other than regulation.
A fourth criterion for exclusion--suggested by the preceding discussion--is agencies or rules that primarily impose indirect costs and do not involve direct compliance cost. The basis for this exclusion would be the difficulties involved in measuring indirect compliance costs.
In summary, four types of agencies (or programs) that might be excluded from a regulatory budget have been mentioned:
Taken together, the criteria implicit in these items would provide a manageable way of determining what agencies or programs should be included in an RB. Moreover, the 1ist of regulatory programs included would probably be fairly short and primarily make up programs directed to social regulation.
An alternative way to go about specifying what agencies to include in an RB is ask what type of regulatory programs are adapted to control through a budget. It seems obvious that the answer is: those regulatory programs whose purposes are achieved by spending and which present genuine choices between more and less spending.
Economic regulation does not meet these tests because it does not work by requiring expenditures. Instead, economic regulation operates through policies directed to restrictions on entry, exit, prices, service levels, mergers and acquisitions, and so on. Attempting to budget the costs of these policies would be a very difficult undertaking. It would also probably not be very informative, as the issue presented by economic regulation is characteristically that of whether the regulations in question are needed at all and not of more or less spending on some reasonably well-defined goal.
Social regulation, in contrast, generally attempts to force economic agents to internalize elements of costs of their activities such as those of environmental pollution, which would not be fully recognized in the absence of the regulation. Regulation in these cases amounts to requiring expenditures on public goods (e.g., on "production" of cleaner air), and the appropriate level of expenditure is a real question, as in a world of scarcity there is no escaping trade-offs between alternative uses of resources. It seems, then, that a RB would be an appropriate way of approaching regulatory programs that require that costs be internalized, which are those programs that impose substantial direct compliance costs.
A final consideration relevant to the question of what agencies would be included is the nature of the RB's legal basis. If the RB were created by statute, the statute presumably would specify what agencies or programs would be included. However, an RB created by Executive Order could be enforced only on Executive Branch agencies. It is relevant in this respect to note that most agencies concerned with social regulation are in the Executive Branch, while most independent agencies are involved in economic regulation.
The upshot of this discussion is that how to go about specifying what agencies should be included in an RB is probably not a key issue. Whether the issue is approached in terms of criteria for exclusion or in terms of an affirmative principle describing what types of regulatory programs are suitable for control through a budget, the prime candidates for inclusion in an RB are those regulatory programs that involve substantial direct compliance costs.
Structure: An RB, like the expenditure budget, might be set up on the basis of agencies. Alternatively, an RB might be structured on a "functional basis"; that is, programs on a given topic (e.g., air pollution, traffic safety, carcinogens, etc.) grouped together.
Much of the economic rationale of a budget lies in the role that it can play in organizing decisions on trade-offs between programs. From this point of view, the key requirement is that programs be ranked in terms of net benefits. All marginal programs can then be compared so as to establish which should be cut back, with the resources freed thereby going to programs with higher net returns. The required rankings and comparisons could, in principle, be made under either an agency-based or a functionally-based RB. In short, the relevant economic principles do not favor either of these two options over the other, so the choice between them must be made on some different basis.
A functionally-based RB would have the advantage of dealing directly with the most sensible trade-off s. For example, DOT has regulatory authority in several areas, two of which are traffic safety (NHTSA) and oil spills (Coast Guard). In an RB done on an agency basis, DOT would be required to make its own rankings of such disparate programs. Furthermore, relationships between similar programs (e.g., the Coast Guard's responsibilities for oil spills and EPA's for spills of hazardous substances) might be easily overlooked in an agency-based system. The argument for an agency approach is relative ease of administration, as fewer actors would be involved at each part of the budget review. The fact that the legal authority to impose rules is granted to specific agency officials may also argue for an agency-based system.
The importance of the choice between an agency-based and a functionally-based RB depends in part on how extensive the system is. If most agencies were included, there would be a genuine choice on whether to use a functionally- or an agency-based system. However, if an RB were designed to serve some limited objective, such as control of direct compliance costs, and the number of regulatory programs included correspondingly limited, it might be clear that an agency-based system is the only real choice.
Another option is to base an RB on neither a functional or an agency basis, but simply on new regulations. In such a "rule-based" system, an overall ceiling on the additions to prevailing costs of regulations would be established. All proposed major rules proposed during the budget year would then be reviewed in the light of this ceiling. Such a system could amount to an extension--although a very considerable one--of the existing RARG process.
Table 1 above listed the options that have been discussed here on each of three aspects of an RB. If the choices made on one aspect of an RB did not limit those available onothers, the possibilities listed in Table 1 contain 81 different types of RB. In fact, however, choices made on some topics do constraint the range of choices that are available on others and there are other considerations that serve to prune the range of internally consistent options.
First, it would probably not be possible to include indirect compliance costs in an RB system in which the ceilings were legally binding. This statement does not reflect any judgment on what would be legally permissible, but rather what would be appropriate as a matter of broad policy. In a system based on negotiation and persuasion, considerable error and ambiguity in estimates of indirect cost would be tolerable. But a formal system, with legally-binding ceilings, probably could not accommodate in a sensible way, estimates known to contain errors of some substantial, but unknown, and often nearly unknowable, magnitude.
Two other points mentioned above are also relevant. First, if a wide range of regulatory programs is included in the RB, it would probably be necessary to include both direct and indirect compliance costs. Conversely, if the programs included are limited to those which impose substantial direct compliance costs, it would be less important to formally include both indirect and direct compliance costs in RB ceilings. Second, a broadly inclusive RB would be best organized on a functional basis, while a more limited RB that included only direct compliance costs, should be cast on an agency basis. Taken together, these points suggest two internally consistent packages:
For the reasons stated above, the latter system could perhaps have legally-binding ceilings, while the former probably should not.
The three broad alternative conceptions of an RB which emerge from this discussion are summarized in Table 2. These are not necessarily the only options,*/ and it is not claimed that any of the options shown are definitely feasible. The three options shown are offered only as examples of alternative concepts of an RB which are internally consistent and consistent with the considerations brought out in the preceeding discussion.
A control RB--i.e., an RB with legally-binding ceilings--apparently offers no range of choice in terms of the factors considered here. This reflects the conclusion that in an RB system with legally-binding ceilings, indirect costs should not be included in the budget ceilings. If so, a control RB would be limited to those agencies or programs that impose substantial direct compliance costs. There would then probably be little analytical gain to putting the RB on a functional basis, but attempting to do so would involve substantial organizational costs. Consequently, the control RB is assumed to be structured on an agency basis.
In one sense, an informational RB offers very broad ranges of choice, as the task would be to provide the best and most extensive information on costs of regulation as is possible within practical limitations. However, in terms of RB principles, the range of choice is limited to the two options shown--the broad-based and the narrow-based systems. Given that in both cases, the RB would only be used for informational purposes, the choice between these two options would presumably turn on judgments concerning the value of the information.
The broadest range of choice is available for the advisory RB. In this case, three distinct options are shown: a broad-based system, a narrow-based system, and a system based neither on agencies or functions, but on proposed major regulations. Moreover, there would be options (not shown in Table 2) as to the nature of the advisory role assigned to the RB.
*/ Nor are the options fully stated. For example, no mention has been made of how the various options could be implemented or of the question of Congressional involvement.
ALTERNATIVE RB SYSTEMS
1. Control RB: budget ceilings are legally binding.
Narrow-based system: direct compliance costs only, agency basis, agencies that impose substantial direct compliance costs.
2. Advisory RB: ceilings not legally binding, RB has advisory role.
a. Broad-based system: most agencies included, direct and indirect compliance costs, functional basis.
b. Narrow-based system: direct compliance costs only, agency basis, agencies that impose substantial direct compliance costs.
c. Rule-based system: super RARG.
3. Informational RB: ceilings not legally binding, RB has no role in oversight.
a. Broad-based system: most agencies included, direct and indirect compliance costs, functional basis.
b. Narrow-based system: direct compliance costs only, agency basis, agencies that impose substantial direct compliance costs.
The five options listed in Table 2 illustrate the statement made at the onset of the discussion, that the term regulatory budget covers a range of fundamentally different systems. This point is important for what it says about the problems of creating an RB system. If an RB is assumed to be basically "like the expenditure budget"--the control option in Table 2--it would be necessary to face up to some very difficult issues. But an RB might bear very little resemblance in form, organization or legal status to the expenditure budget. And, looking at the end of the scale in Table 2, an informational RB might be a relatively modest undertaking.