A variety of policies for constraining the private costs of government regulation have been put forth in recent years, and a few have been put in place. The most prominent are the executive branch programs, established by President Ford and expanded by President Carter, aimed at encouraging the regulatory agencies to pay greater heed to the costs their decisions impose on the economy. Under these programs the agencies are required to analyze the costs and benefits of major new regulations, and their analyses are reviewed and criticized by two supervisory groups operating out of the Executive Office of the President—the Council on Wage and Price Stability (CWPS) and the Regulatory Analysis Review Group (RARG). The regulation‑review programs have elicited numerous proposals for increasing the President's role in individual regulatory proceedings even further—notably, by making the Carter program statutory and by giving the President explicit authority to revise or veto regulations after they are published.

The first article in this series reviewed the regulation‑review programs and the proposals to strengthen them. The article suggested that the unique problem of regulatory costs lies in the circumstance that these costs—unlike the costs of government policies pursued through direct expenditure—are unconstrained by systematic mechanisms of public finance such as taxation, appropriation, and annual budgeting. Of course, regulatory policies must be established according to the strictures of administrative law—notice, opportunity for interested parties to be heard, and written decisions whose logic may be scrutinized and rejected by a court—and these may be thought of as regulatory analogs to the budgetary restraints that limit the authority of program managers and grant givers. It is doubtful, however, that the two methods of restraint are even approximately commensurate, and the regulation‑review procedure seems poorly suited to closing the gap. For legal and political reasons, the review pro­grams stop short of requiring the agencies ac­tually to base their decisions on the results of cost/benefit analyses—requiring only that the analyses be performed. The review process cannot touch more than a small portion of the government's regulatory activities or match the agencies' technical and political mastery of any individual proposal, without a huge in­crease in the size and resources of the review­ing agencies. Most important, that process by its very nature cannot affect the rate at which regulations are generated, and thus their aggre­gate economic impact, any more than selective government jawboning can affect the rate of inflation.

The regulation‑review procedure, however it might be strengthened by making it statutory or increasing the role of the President, is based on the unpromising idea that regulatory deci­sions will be improved by centralizing decision making within the government. A contrary, de­centralizing approach, based on the analysis of the problem sketched above, would be a budg­eting process applicable to regulatory costs. One of the earliest exponents of this idea was Robert W. Crandall, who described it in a 1978 article as follows (from A. M. Okun and G. L. Perry, Curing Chronic Inflation):

The most practical possibility for con­fronting regulators with the costs of their actions would be to construct a shadow budget to cover the resources that the agency requires private agents to consume in the pursuit of the regulatory goal. An agency such as Office of Management and Budget would have to administer such a system, but the Congress could specify the size of the budgets for each agency or even each program. Administrators would not be told how to value the benefits of the dangers they presumably would reduce.

Instead, there would be a limitation on the cost that they could impose on firms in a given year; their function would then be to maximize the benefits of their regula­tions subject to that limitation. These cost limitations would be publicized each year in the budget message and would there­fore be more visible to the public, which, in turn, would be less likely to assume that regulatory costs are minimal.

The idea of a regulatory budget attracted considerable notice in official Washington. By late 1978 a number of informal papers on the subject were circulating within the executive branch. In 1979 two members of the Joint Economic Committee of the Congress, Sena­tor Lloyd Bentsen (Democrat, Texas) and Congressman Clarence J. Brown (Republican, Ohio), introduced a bill to establish a regula­tory budget, and the committee's annual report—its first unanimous report in twenty years—recommended the measure. The 1980 Annual Report of the Council of Economic Advisers discussed the proposal, noting the difficulties of estimating regulatory costs precisely but concluding that "tools like the regulatory budget may have to be developed" to encour­age the agencies to set priorities—"to make certain that the first problems addressed are those in which regulations are likely to bring the greatest social benefits." In early 1980 the Office of Management and Budget (OMB) cir­culated among the agencies a proposed "Regu­latory Cost Accounting Act" which would, in effect, establish a regulatory budget. There ap­pears to be no immediate prospect of enacting the proposal, but its rapid ascension within the federal establishment suggests that it has wide enough appeal to be taken seriously as an eventual possibility.

How a Regulatory Budget Would Work

The regulatory budget would operate by close analogy to the conventional fiscal process. Each year (or at some longer interval), the federal government would establish an upper limit on the costs of its regulatory activities to the econ­omy and would apportion this sum among the individual regulatory agencies. This would pre­sumably involve a budget proposal developed by OMB in negotiation with the regulatory agencies, approved by the President, and submitted to Congress for review, revision, and passage. Once the President had signed the final budget appropriations into law, each agency would be obliged to live within its regu­latory budget for the time period in question. The budget would cover the total costs of all regulations past and present, not just new ones. Thus, for a given budget period, an agency could issue and enforce new regulations only to the extent that the costs imposed, when added to the current costs imposed by regulations issued in previous years, were within the agen­cy's total budget. The policies of individual regulatory statutes would have to be imple­mented within this budget constraint, just as they now must be implemented within the con­straint of the expenditure budget.

A regulatory budget would not require re­vision of the primary regulatory statutes to permit or require the agencies to take any par­ticular account of costs—a major stumbling block of current efforts to insinuate cost/bene­fit analysis into individual regulatory decisions. If, in the extreme case, an agency found itself compelled by its primary statute to take action more costly to the economy than its remaining regulatory budget permitted, the agency would have to obtain a supplemental appropriation, just as if it had exhausted its fiscal appropria­tion; more manageably, the agency might ob­tain a "carry‑forward" from OMB against its budget for the next period. Such a case would be the Food and Drug Administration's dis­covery of a new carcinogenic food additive that would cost more to ban (as required by the Delaney Amendment) than remained in its budget. The economic effects of most other nondiscretionary statutes (such as the man­dated standards of parts of the Clean Air Act) would be more predictable, and presumably Congress would take them into account in de­termining the agency's regulatory appropria­tion in the first instance.

Of course, the regulatory budget would not be allocating actual government revenues, only authority over private revenues. Two unique institutional mechanisms would therefore be required. First, it would be necessary to meas­ure regulatory "outlays" on an estimated basis when regulations were first issued and on an actual basis after the regulations had taken hold. The actual costs of regulations would be determined by a retrospective accounting every (say) five years, and carry‑forwards among budget periods would be used to reconcile esti­mated and actual costs (as well as to accommodate unforeseen regulatory actions).

Second, there would need to be an office, presumably in OMB, responsible for certifying the agencies' calculations of regulatory costs. Agencies would have strong incentives to over­state estimated costs in order to obtain large initial appropriations, and later to understate actual costs in order to increase their discre­tion within budgeted amounts. Regulated firms and other private parties would also be in­clined to misstate costs one way or another in order to affect agency discretion. It would be nice if some institutional mechanism could be designed to induce perfectly truthful cost esti­mates at every stage of the budgeting process (one such mechanism is proposed by Lawrence J. White in "Truth in Regulatory Budgeting," page 44, this issue). Still, numerous honest dif­ferences of opinion and technique would re­main to be resolved authoritatively, such as the proper treatment of joint costs. The certifica­tion procedure would not be part of the budg­eting process itself, but would serve to estab­lish official outlay figures (both initial "esti­mated" costs and subsequent "actual" costs) for purposes of monitoring and enforcing budget compliance.

Costs, Not Benefits

Two related and important features of the reg­ulatory budget—familiar in the case of the ex­penditure budget but less obvious here—are that (1) it would be concerned with costs in the sense of expenditures and (2) it would not be concerned with benefits at all. First, the measure of cost budgeted to the regulatory agencies would be something less than the eco­nomic concept of social opportunity cost. For reasons of administrative practicality discussed later in this article, costs other than measura­ble expenditures by organizations and indi­viduals would generally be excluded. Second, benefits would be excluded altogether, which means that expenditures in the form of trans­fers from one group to another would be counted as regulatory costs, although they are not economic costs at all. Thus, in the case of environmental regulations, regulatory costs would include the expenditures of firms on pollution controls, and these would not be off­set by resulting benefits—even those in the form of unambiguous increases in wealth else­where in the economy, such as increased prop­erty values or reduced expenditures on medi­cal care. In the case of economic regulations, regulatory costs would include the additional expenditures of consumers due to monopoly pricing, with no account taken of the corre­sponding additional revenues collected by reg­ulated firms, nor of the benefits to some con­sumers resulting from intensified product­ quality competition among firms. In the case of product safety regulations, the costs of com­plying with safety standards would not be off­set by the benefits of increased product availability to consumers.

The exclusion of benefits would apply re­gardless of whether the beneficiaries them­selves paid none, some, or all of the costs of a regulation. It would be immaterial whether the beneficiaries were pure third parties, as in the case of a pollution control whose exclusive beneficiaries are nonconsumers of the products whose manufacture is being regulated, or pure first parties, as in the case of a safety regula­tion whose exclusive beneficiaries are consum­ers of the product whose design is being regu­lated. It would also be immaterial whether first‑party beneficiaries were individuals or regulated firms. Firms may, of course, benefit from health and safety regulations as well as economic regulation. For example, product safety standards may yield significant benefits to producers by eliminating uncertainty over how to respond to changing consumer tastes. A plausible instance is the Consumer Product Safety Commission's design standard for baby cribs, which was established in the wake of widespread publicity of the hazards of tradi­tional cribs and which appeared to enable all crib manufacturers to react in lockstep to a change in consumer preference. Even where safety regulations oblige firms to make invest­ments they would not make otherwise, the firms will typically publicize these investments in order to increase demand for their products; for example, pharmaceutical manufacturers aggressively advertise their FDA‑required safe­ty and efficacy testing programs, which are more extensive than they would undertake on their own. Robert Leone of the Harvard Business School has demonstrated that regulations which force major capital investments, such as air and water pollution control standards, may generate large short‑run benefits (which econ­omists call "quasi‑rents") for firms with rela­tively low compliance costs, since post‑control prices may be determined by the costs of firms with relatively high compliance costs. Producer benefits such as these would be ignored in a regulatory budget, just as would benefits to consumers. If a regulatory agency concluded that the benefits to firms of complying with (say) a safety regulation exceeded the firms' compliance costs, the agency would simply withdraw the regulation. The firms would con­tinue to produce at the desired safety level and the agency's budget would be increased by the extent of the "saved" compliance costs. Com­plaints that regulated firms try to take advan­tage of regulatory standards, and that govern­ment officials try to take credit for develop­ments the market would have occasioned any­way, would be undercut by the dynamics of the budget process.

"Market Failure" and "Regulatory Failure"

Many readers will probably be troubled by the regulatory budget's exclusive focus on costs. Would this not be a sharp and inappropriate departure from the application of cost/benefit analysis on a case‑by‑case basis, as in the cur­rent regulation‑review program? The short answer is that benefits would indeed be taken into account—but early in the process, when the President and Congress determined the size of each agency's budget. Cost budgeting and cost/benefit analysis are by no means the same thing, however, and it will be useful to compare the two in some detail. To do so we must first elaborate briefly on the notion of "market failure" and the formal economic function of regulation.

Market failures—such as natural monop­oly, externality, and consumer uncertainty over product quality—are a consequence of the costs of voluntary market transactions, such as the costs buyers and sellers incur in finding each other and negotiating and enforcing agree­ments. In the usual formulation, a market failure is said to exist when these kinds of costs prevent mutually beneficial transactions from taking place. Of course, one could say that where the costs of engaging in a transac­tion are greater than its benefits to the indi­viduals involved, ipso facto the transaction is not "mutually beneficial" and it ought to be left unconsummated. But some rearrangement of legal or social institutions might reduce transaction costs, or the government might use its coercive powers to effect certain transac­tions more cheaply than the private market. For example, a regulation might obtain the re­duction in pollution that would have resulted, in the absence of transaction costs, from vol­untary agreements between polluting firms and their neighbors. Strictly speaking, to say that a market failure exists is to assert that some such corrective possibility would actually im­prove matters: otherwise there is no point in talking about the market having failed. The correction of market failure is the only non­paternalistic rationale for government regulation, and largely for this reason it permeates the official rhetoric of most regulatory pro­grams as well as the logic of most regulatory decisions.

If regulation in fact operated purely to correct for market failure, there would obvi­ously be no occasion for a regulatory budget. In such a world any limitation on regulatory costs would only reduce the efficiency of the economic system. But in the real world regu­lation departs from the economic ideal in two related ways‑which may be called "regula­tory failures." First, regulation frequently op­erates not to improve the efficiency of markets, but rather to redistribute income and wealth in the direction of politically effective groups, typically at the expense of economic efficiency. Second, regulators have skewed incentives re­sulting from their institutional responsibility to pursue a single policy single‑mindedly, which frequently lead them to go far beyond the point of zero marginal returns in correct­ing market failures. Fearing above all a con­spicuous disaster within their jurisdiction—-deformed babies or a major accident—they issue regulations requiring (say) the elimina­tion of 95 percent of some risk where a 90 per­cent reduction would cost only half as much and would be the equilibrium point in a per­fectly functioning market. The first "regula­tory failure" is most often associated with eco­nomic regulation and the second with health, safety, and environmental regulation, but the two overlap insensibly. A regulator may go be­yond the efficient control point in drafting a regulation in part because influential political groups benefit disproportionately from the ex­cessive margin and in part because bureau­cratic incentives generate extreme risk aver­sion.

It may appear that the most direct ap­proach to correcting both of these regulatory failures would be simply to require that all regulations pass a cost/benefit test. This might be accomplished by a beefed‑up version of the regulation‑review program, by a revision of regulatory statutes, or by further development and strict application of the judicial doctrine that "reasonable" regulations (which almost all regulatory statutes presently require) are those whose expected costs can be shown to be "reasonably related" to expected benefits. The situation is more complex than this, how­ever. In theory, neither cost budgeting nor cost/benefit analysis is a precise approach to con­straining the costs of regulatory redistributions, but (again in theory) the two approaches in tandem are superior to either taken alone in constraining the costs of regulatory overreach­ing.

Constraining Regulatory Redistributions

Consider first the problem of regulatory redis­tributions. Here one could argue that the eco­nomic problem consists not of the redistribu­tions per se, but only of the true social costs ­the lost opportunities or "dead‑weight losses" that arise when redistributions are accom­plished by tinkering with market prices and product quality. Thus, when the Interstate Commerce Commission (ICC) fixes motor­carrier rates above competitive levels, trucking firms are enriched at the expense of those consumers who continue to employ their serv­ices. But the transfer of wealth from consum­ers to firms is not itself an economic cost, since the increased expenditures of consumers are exactly matched by increased revenues to firms. The only immediate loss to the economy is the lost utility to consumers who would have pur­chased trucking services at competitive rates but do not at the higher fixed rates: their lost opportunity is not offset by gains to trucking firms or anyone else. In any such case one might be indifferent or even favorably disposed toward the redistribution itself. (If you don't own an ICC trucking certificate you probably don't like the distributive effect in the exam­ple given, but you may own a similar certifi­cate from a professional association, or you may enjoy the distributive effects of telephone or insurance or environmental regulation.) In no case, however, could one be indifferent to the attendant dead‑weight losses: minimizing such losses is in everyone's interest, including the recipients of redistributions.

A regulatory budget, however, could not possibly distinguish finely between dead‑weight losses and mere transfers, and would be cer­tain to capture transfers more completely than losses. Restricting the government's authority over gross private expenditures would be a dis­criminating means of reducing dead‑weight losses only to the extent the two were strongly correlated; but the correlation might not be strong, since dead‑weight losses are a function not only of the size of redistributions but also of the elasticities of demand and supply in af­fected markets. The budgeting process would not, in other words, distinguish between effi­cient and inefficient regulatory redistributions. Moreover, we could not be content that the budgeting process would reduce dead‑weight losses crudely, simply in the course of reduc­ing the level of regulatory activity. Because the measure of costs subject to budgetary control would necessarily be less than complete (for reasons explained below), regulators would be inclined to employ regulatory strategies that relied less on readily observable budgeted costs and more on subtler unbudgeted costs. An agency might, for example, prohibit new in­dustrial facilities in a given area rather than require all facilities to install expensive pollu­tion controls, or it might prohibit strip mining altogether rather than require extensive recla­mation. Depending on the cost standard em­ployed and the regulators' ingenuity in work­ing their way around it, the budgeting process could conceivably increase the inefficiency of regulatory redistributions.

This difficulty is a serious one, but it is mitigated by two further considerations. First, the distinction between transfers and dead­weight losses is less exact than it at first ap­pears. As Richard Posner has argued, reg­ulation‑induced (and other) redistributions tend to be dissipated—transformed into dead­weight losses—as groups compete to obtain the redistributions. In the example of ICC rate-­fixing, transfers from consumers to trucking firms will be wasted as the firms invest in ob­taining the transfers: lobbying to maintain political support for the regulatory program, litigating to secure the ICUs approval of higher‑than‑competitive rates, and competing among themselves in service quality to in­crease their individual market shares at the established rates. Obviously, the objection that a regulatory budget would primarily limit transfers rather than true social costs is weak­ened to the extent that transfers should be reckoned as costs in an ultimate accounting.

Second, the focus on transfers rather than underlying dead‑weight losses is also a charac­teristic of the expenditure budget. If the pur­pose of a regulatory budget is to establish a general constraint on the government's regula­tory endeavors, comparable to that which the expenditure budget places on taxing and spend­ing endeavors, then limiting redistributions is fully as important as limiting dead‑weight losses. This is especially so if we believe, as suggested in Part One of this series, that regu­latory redistributions tend to be both politi­cally and economically perverse as compared to direct redistributions.

It is nevertheless clear that a regulatory budget would not be a precise tool for elimi­nating the costs of regulatory redistributions and confining regulation to the correction of market failure. It is therefore important to note that even a thorough application of cost/benefit analysis to individual regulations would also fail to accomplish this feat. A market‑correcting regulator asks whether trans­action costs are preventing some group of indi­viduals from making a mutually beneficial transaction with another group, and if the an­swer is yes he undertakes to effect the trans­action through an administrative rule. A cost/benefit regulator asks whether some administrative rule would benefit one group of individuals more than it would cost another group—the two groups may overlap or even be iden­tical—and if the answer is yes he goes ahead. The difference in approach is critical. If one is merely trying to approximate the results of a market unfettered by transaction costs, one must attend to both the preferences and the economic resources of the individuals involved. The cost/benefit regulator is not so confined. He is free to make transfers that would not be made in a perfect market because of the budget constraints of the individuals involved, and his stratospheric perspective encourages a relaxed attitude toward the problem of comparing the utility to different individuals of identical goods. Thus, he may restrict air pollution be­low the level a perfectly functioning market (in air) would produce by giving decisive weight to a generous estimate of "aesthetic benefits"‑benefits which seem nice in the ab­stract but which very few people (possibly in­cluding himself) would actually purchase at cost. Or he may be satisfied with a market­wide product safety standard on grounds that its costs are no greater than the accident costs it eliminates, although its effect is to oblige careful individuals to subsidize careless indi­viduals (in such a case, a perfect market would produce more‑safe and less‑safe versions of the same product). As these examples suggest, cost/benefit analysis is potentially a powerful tool of redistribution.

A revealing feature of discussions of cost/benefit analysis in regulatory policy is the ubiquitous caveat that benefits are harder to measure than costs. The argument is that it is relatively easy to tote up the costs of complying with regulations, even if this requires estimation of unobservable opportunity costs, but damnably difficult to calculate the value of such benefits as prolonged human life, smog‑free cityscapes, and the mental calm that comes from knowing the caribou are roaming unper­turbed. Observe, however, that the asymmetry would vanish if regulation consisted solely of overcoming the deterrent effects of transac­tion costs in private markets; it is no more dif­ficult to estimates whether one group would make a purchase than whether another group would make a sale.

The asymmetry arises because the goals of regulatory programs are more ambitious than this—and also more ambiguous. Is the goal of pollution regulation to improve health, or to improve aesthetics, or to increase the relative influence of government in determining the economy's output? It is all of these goals and some others as well, and even the narrowest goal of improving health is not confined to ben­efits people would actually pay for if they were as well‑informed as regulators. Moreover, it is not the same goal or mix of goals to any two participants (legislators or lobbyists or execu­tive officials) in the political process. The heart of the problem is that cost/benefit analysis is an internal decision procedure for a single or­ganization—an exercise in which the organiza­tion attempts to sea out with exactitude its pur­poses and what it must give up to achieve them. But the political process is not a single organization and it has no "purpose." It con­sists of a multitude of organizations with con­flicting purposes—purposes that are persistent­ly obscured because of the need to maintain political support while achieving practical compromise. This is why no single organization (RARG or CWPS or a federal court) could ever apply a strict cost/benefit test to the decisions of another organization (a regulatory agency). The most that can be done is to require some unspecified "reasonable relationship" between benefits and costs.

Constraining Regulatory Overreaching

Consider next the second "regulatory failure," that of overreaching—which, as described ear­lier, consists precisely of giving too much weight to regulatory benefits relative to costs. Here one is tempted to argue that the imposi­tion of a regulatory cost constraint, in contrast to the imposition of cost/benefit analysis, would leave the problem untouched: agencies would be induced to choose the least costly means of reaching a given target, but they would not be induced to choose the most cost-­effective target in the first place. Other things being equal, they would choose regulation A, with low costs but proportionately lower bene­fits, over regulation B, with high costs but pro­portionately higher benefits.

The flaw in this argument is that other things are not equal. Regulators already have strong political incentives to adopt policies with the largest benefits that fall upon identi­fiable cohesive groups. A budget constraint would provide a counterbalance, obliging reg­ulators to weigh costs against benefits even when costs are widely dispersed. Agencies would, moreover, be required periodically to entreat a budget authority and a congressional committee for regulatory appropriations, and this would involve showing as convincingly as possible that past and proposed regulatory ac­tions are on balance beneficial. Regulatory agencies would have greatly enhanced incen­tives for calibrating costs and benefits at the margin; if an agency was "spending" $50 mil­lion a year to eliminate 90 percent of some risk, it would be a poor argument for another $50 million that it would reduce risk another 5 per­cent.

Looking at the matter in this way brings out the advantages of the regulatory budget over attempts to impose cost/benefit analysis from the top down. In contrast to the current trend under the regulation‑review programs and the proposals to strengthen them, a regula­tory budget would prompt a vast devolution of responsibility for cost/benefit judgments from the White House to the regulatory agencies. Just as economists prefer that safety and en­vironmental regulations subject firms to "per­formance standards" rather than "engineering standards" in order to harness private incen­tives, a regulatory budget would subject the agencies themselves to the performance stand­ard of a budget limitation rather than the cur­rent engineering standard of simply perform­ing cost/benefit analyses. Faced with a budget constraint, the agencies would measure the costs and benefits of individual regulatory pro­posals in order to further their own organiza­tional interests rather than to satisfy the mini­mum requirements of an executive order or judicial review. The need for a central office to match the agencies' technical competence on a case‑by‑case basis would be greatly reduced. Some amount of cost/benefit analysis would continue in one part or another of the Execu­tive Office of the President, but as an adjunct to the budgeting process rather than its driving force, just as in expenditure budgeting. (OMB examiners evaluate the costs and benefits of various weapons systems, but the defense budg­et is not simply the sum of their conclusions.)

The Political Logic of the Regulatory Budget

The logic of the regulatory budget is ultimately political rather than economic: it would ac­knowledge explicitly the political nature of reg­ulatory benefits and permit the President and Congress to make political judgments in light of more thorough information about economic costs. The analogy of the expenditure budget is instructive here, for an expenditure budget is as far removed from the pure economics of public expenditure as a regulatory budget is from the pure economics of regulation. One could say that the activities of the Weather Bureau are undertaken because of "public goods" problems in the provision of informa­tion about the weather, and that the appropri­ate level of the bureau's activities should there­fore be determined by calibrating the point at which its marginal costs equal its marginal benefits to the nation's economy, rather than by the crude imposition of a budget constraint that takes no formal account of benefits. But of course this is a superficial criticism. The eco­nomic benefits of government provision of weather information are impossible to measure with any precision, and they are taken into ac­count in the budget process, as funds are allo­cated to the Department of Commerce as op­posed to Defense and Transportation, and with­in the Department of Commerce as funds are allocated to the Weather Bureau as opposed to the Census and Standards Bureaus. Similarly, the benefits of regulatory programs are diffi­cult to measure precisely and inevitably involve a large measure of political judgment: it is ap­propriate that regulatory benefits too should be accounted for by allocating the costs of achieving various goals among‑the regulatory agencies according to the political judgments of the President and Congress.

To say that regulatory appropriations should be determined by "political judgments" is not to give up on economic rationality as hopeless at the highest policy‑making levels of government. In setting an agency's regulatory budget, the President and Congress would weigh costs not just against commensurate benefits as they see them, but also against con­current costs in all other regulatory (and ex­penditure) programs. There is an implicit eco­nomic logic to this procedure. Comparing par­tial costs with concurrent total costs is a casual exercise in marginal analysis, of a sort that nev­er occurs in most practical applications of cost/ benefit analysis where consideration of costs outside the project being analyzed are deliber­ately excluded. It is also a means of overcoming the uncertainty over benefits, similar to the ap­proach people employ routinely in personal life in deciding whether they can afford a particular purchase given their income and their other current expenditures. In strictly economic terms one can know very little about whether a given level of federal (that is, federally im­posed) expenditure on pollution control is too little, too much, or just right. Under the cir­cumstances, an official who must decide how much the government should cause to be spent—even if he is determined to minimize political considerations—will wish to compare pro­posed spending levels with gross national product, with total federal spending, and with how much the government is spending (or causing to be spent) on medical care, work­place safety, housing, defense, and other en­deavors.

The most attractive feature of the regula­tory budget is that it would establish a clear upper limit on the government's regulatory ac­tivities and clear priorities among its various health, safety, environmental, and economic ventures. The increasing tendency of govern­ment to pursue its objectives through regulation rather than taxing and spending—even when regulation is otherwise less desirable­ because regulation is less constrained would be reduced. At the same time, a good deal of the steam would go out of the current political dispute between the President and Congress over the President's proper role in individual regulatory decisions. The President and Con­gress would be jointly responsible for estab­lishing systematic bounds on regulatory policy, and both would be less inclined to be drawn into sporadic, unproductive forays over the de­tails of this or that regulatory proceeding.

Some Practical Difficulties

The regulatory budget is an idea of consider­able theoretical appeal; it has, however, a num­ber of serious practical shortcomings. First and most obvious is the problem of collecting and analyzing the vast quantities of cost informa­tion that would be needed to establish and en­force the budget. Indeed, when one contem­plates the variety and extent of federal regula­tion of so many aspects of contemporary life, one is liable to despair of the possibility of any kind of comprehensive cost accounting. Two considerations suggest, however, that this problem may not be as serious as it appears. First, a regulatory budget, unlike the expendi­ture budget, would not be a system of revenue accounting, and so would not have to be ap­plied with penny‑wise precision. For purposes of setting an overall restraint on regulatory ac­tivities and obliging regulators to recognize the trade‑offs among policies, agency budgets ap­plied only to the nearest $10 million would be more than sufficient:. Second, a great deal is al­ready being invested in measuring regulatory costs for both private and public purposes and, given the growing importance of regulation, this investment will continue regardless of whether a regulatory budget is established. Even apart from the regulation‑review pro­gram and the routine industry and agency cost­accounting procedures that have accompanied it, large industrial firms must measure regula­tory compliance costs for their own financial and planning purposes. The Securities and Ex­change Commission has recently begun to re­quire firms to make precise disclosures to in­vestors of the costs of complying with major regulatory requirements. The additional costs of systemizing and supplementing the cost in­formation already being collected may not be great.

A more serious problem is the measure of costs to be employed in the budgeting process; as mentioned earlier, budgeted costs would of necessity be something less than total social opportunity costs. The reason is that measure­ment of some types of regulatory costs, such as the lost consumer surplus resulting from re­tarded innovation or premarketing regulatory delays, is inherently speculative and inevitably hedged about with ifs, ands, and buts. Argu­ments over elasticities of demand and supply, adjustments to account for risk aversion, ex­ogenous variables insufficiently accounted for, and the reality of the economist's fundamental assumptions could swamp the budgeting proc­ess in controversy and destroy its program­matic neutrality. This problem would not be so serious if there were some narrow, unambigu­ous category of regulatory costs. Unfortunately there is none—indeed there is no basis in eco­nomic principle for distinguishing categories of costs at all. And, as we have seen, any limitation on budgeted costs would create unfortunate in­centives for regulators to rely on kinds of costs outside the limitation. Clearly it would be unacceptable to limit budgeted costs to the expenditures of business firms and other large organizations, which was the approach of the recent study of the Business Roundtable (see Regulation, July/August 1979). A regulatory budget so limited in scope would be easily evaded. Agencies would be inclined to prohibit certain (new) activities outright rather than require that all activities (new and old) be un­dertaken in a certain manner—thereby rein­forcing the already pervasive tendency of reg­ulation to freeze production technology at the level assumed as the basis of regulation. And such an approach would, of course, place no re­straint at all on programs whose costs fall upon unorganized consumers, such as traditional controls on pricing and entry in common‑car­rier industries. At the same time, limiting budg­eted costs to the expenditures of firms would encourage the perception—probably a misper­ception—that the "tax incidence" of regulatory redistributions is generally progressive.

But once we move beyond the direct ex­penditures of a manageable number of large organizations, we immediately plunge into the world of imprecise estimates acutely sensitive to assumptions about behavior and technology—in particular, assumptions about the respon­siveness of firms and consumers to changes in price. The difficulties to be encountered here are suggested by recent studies of the costs of railroad regulation. Studies conducted in the 1960s by Ann F. Friedlaender, Robert W. Harbe­son, and others found that the ICUs policy of discouraging railroad rate reductions on truck­ competitive freight imposed large costs on shippers and ultimate consumers by causing much of this freight to be transported by the higher‑cost mode—by truck rather than rail­road. The range of the cost estimates was, how­ever, extremely wide—from Friedlaender's $150 million to Harbeson's $1.1‑2.9 billion. Then two more elaborate studies in the 1970s by Kenneth D. Boyer and Richard C. Levin found the costs of minimum‑rate regulation to be far lower (Levin estimated $53‑135 million).

Obviously there would be no point to a budgeting system whose components could be established, even by disinterested scholars, only to a degree of magnitude. Yet this is the situation today in many areas of regulatory study; even the less comprehensive cost studies conducted by regulatory agencies, firms, and CWPS under the regulation‑review programs have often varied by factors of two or three.

Clearly, a workable budgeting system would have to rest on a practical compromise‑some measure of "expenditures by firms, consumers, and third parties" that was narrow enough to facilitate general agreement in particular cases but not so narrow as to stimulate massive cost ­substitution strategies by the agencies. Such a compromise may simply be infeasible. The problem is, in any event, serious enough that any budgeting program would need to be pre­ceded by several years of effort to develop a uniform methodology of cost measurement ap­plicable on a program‑by‑program basis.

A related problem is joint causation and the pattern of regulatory costs over time. The findings of the studies on the costs of railroad regulation can be reconciled by observing that, by the time the later studies were conducted in the 1970s, costs may hove fallen consid­erably as railroad and trucking firms ad­justed their operations in response to regula­tion; railroads, for example, may have redi­rected their efforts away from carrying truck ­competitive traffic since it was foreclosed to them in any event. This possibility illustrates the important point that costs, being ultimate­ly a measure of alternative opportunities, change over time with the alternatives avail­able in the market. (A further, often‑noted ex­ample is that the costs of pollution‑control technology usually end up being much lower than originally predicted, as a result of unanticipated innovations stimulated by the very costliness of the initial control technology.) A budgeting procedure such as that described here—covering present costs of old as well as new regulations and featuring periodic account­ing to reconcile actual with projected costs ­would mitigate this problem, but in the process would introduce the further problem of joint causation. Except in cases where regulations certify market forces (as in the baby crib ex­ample), there is no causal ambiguity to ex­penditures made in compliance with new regu­lations: the change in expenditures at the time the regulation is introduced provides a con­vincing market test. Over time, however, regulations may become partially redundant with market forces, and compliance costs will then be joint with other production costs in some degree. But the degree of jointness and hence the true marginal cost of the regulation will then be discoverable only by actually removing the regulation so as to provide a market test comparable to that which existed at the outset. One could eliminate this problem by assessing all projected costs (including discounted fu­ture compliance costs) to a regulation during the budget period when it was introduced and eliminating costs of past regulations from all current budgets. But this procedure simply re­introduces the problem of the uncertain pat­tern of costs over time. The budget would misstate regulatory costs—typically overstating ongoing compliance costs relative to initial costs, so that the direction of regulatory policy would be artificially skewed away from policies with ongoing costs.

The various difficulties of measuring regu­latory costs, and the incentives problems cre­ated by attempts to simplify cost measurement and accounting, are simply extensions of the general problem of the government's relying too heavily on regulation in order to move the costs of its activities off the measured public budget. A regulatory budget would not end the existing bias in favor of regulation: it would only reduce it, and only to the extent the meas­ure of budgeted costs could be made complete.

A separate problem is the source of the budget constraint under a regulatory budget. Even assuming that the problems of measuring regulatory costs could be resolved in a satisfac­tory manner, the budgeting process would re­main a rather abstract exercise. In the case of the expenditure budget, the government's ac­tivities are limited by its tax revenues plus the extent of its willingness to go into debt or to inflate the supply of money. But the regulatory budget would be a one‑sided ledger: there would be no tax receipts with which to compare spending levels, and no familiar norms for rhetorical support such as the analogy of the fam­ily budget and the prudence of budget balanc­ing. The payless payday would not even be a hypothetical possibility for the regulators themselves. Regulatory agencies might assume that their appropriations were continuously negotiable, guessing that when the issue of budget compliance came down to the decision of a particular regulatory controversy, the po­litical pressures surrounding the controversy itself would secure the requisite supplemental appropriation or budget carry‑forward. If even a few agencies were successful in this gambit the rest might quickly follow suit. Supplemen­tal appropriations and carry‑forwards would become routine and the budget process would lose its bite, as the President and Congress in­creasingly found themselves, at budget time, ratifying past regulatory decisions rather than limiting prospective ones.

It is easy, however, to overstate the differ­ence between the expenditure and regulatory budgets. If regulatory costs are more hypothetical than fiscal costs, so are regulatory ben­efits. It is instructive that, in the cotton dust case discussed in Part One of this series, Presi­dent Carter and Secretary of Labor Marshall were able to postpone the imposition of the final regulatory standards for four years with­out creating major political difficulties for themselves. Certainly they could not so easily have postponed a spending program compara­ble in size to the very large benefits the Occupa­tional Safety and Health Administration pro­jected for its cotton dust rule. Moreover, the government's fiscal expenditures are con­strained at the margin not by revenue limita­tions but only by elected officials' perceptions of the political costs of higher interest rates and lower currency values in the private sec­tor. The marginal constraint under a regulatory budget would be equivalent. In both cases, poli­ticians would be dealing with tables of abstract numbers and economists' conflicting interpre­tations of what they portended, not with the cold bottom line of a bank account. Of course, the evidence of recent decades is that the mar­ginal fiscal constraint is not very effective against short‑term political pressures, which is one of the reasons for recent efforts to insulate budgetary decisions from decisions about in­dividual spending programs, such as the estab­lishment of the congressional budget commit­tees and the proposals for constitutional spend­ing limitations. But if a regulatory budget can be no more effective than the expenditure budg­et with these new procedural accoutrements (which could be extended to the regulatory budget as well), it at least might be nearly as effective—thus reducing the present policy imbalance which exists because political officials never have the occasion to consider regulatory costs taken as a whole.

The Psychology of Government Control

A final, deeply ironic difficulty of the regulatory budget is its potential to legitimate government control of private expenditures, and thus to weaken rather than strengthen the boundaries on the government's appropriate functions. The growth of federal regulation has already brought about a remarkable erosion of the tra­ditional distinction between the public and private sectors. It is now, for example, custom­ary in political debate to assume that the returns earned by petroleum companies belong to the federal government, as in President Car­ter's frequent complaint that these firms "mis­use" their money by investing in hotels, depart­ment stores, and other ventures whose returns are higher than the government permits of petroleum investments. In some cases, such as that of the railroads (especially in the North­east), the financial sources of regulatory redis­tributions have been so depleted by the process that they have eventually required outright government subsidization—leading, of course, to outright government control. This sequence of events threatens to become routine. When the Chrysler Corporation recently requested direct public subsidies for its costs of meeting federal automobile regulations, the important political question was not whether or how much the government would pay, but what manner of control it would thereby assume over the firm in the form of government‑ap­pointed directors, stipulated wage settlements, agreements to invest in certain political dis­tricts, and explicit or implicit promises to acquiesce in further auto design requirements.

If the regulatory expenditures of General Motors were part of a formal public budget ­and especially if they were itemized as such, it might come to seem equally appropriate to treat them as public expenditures, with conse­quences that are difficult to foresee. A full-­fledged regulatory budget program might as­semble cost data by firms, industries, and other groups in the society, as well as by regulatory agency; indeed there would be good reason for doing so, in order to measure aggregate regula­tory impacts on particular sectors of the econ­omy. What, then, would the federal Office of Regulatory Budget require of firms in order to certify their costs as official costs for budgeting purposes? At first, no doubt, nothing more than that they comply with official federal account­ing policies. But later, in the heat of some political controversy, it might seem a small fur­ther step to require compliance with addition­al government policies, and soon it might be­come a fine legal question whether the activi­ties of certified firms were or were not "state action" for constitutional purposes.

This difficulty brings us back to the view, asserted at the outset in a matter‑of‑fact way, that regulation differs from other government ventures only in that the expenditures involved do not wend their way through the public treasury. The comparison proved useful for an­alyzing the programmatic details of the regula­tory budget proposal. But there is, perhaps, a deeper political logic in treating regulation in the traditional fashion as a corrector of ineffi­ciencies in private markets and a surrogate for private legal institutions. If an industrial facil­ity pollutes the air around it, its neighbors may bring an action in nuisance and, in appropriate circumstances, oblige the firm to reduce the pol­lution or compensate them for their damages. If a firm's employees or customers are injured in accidents involving the manufacture or use of its products, it may be obliged to pay their damages, and this prospect influences the firm's investments in product design and production techniques. If consumers can show that a firm or cartel of firms has engaged in monopoly pricing or related practices, the consumers can obtain compensation for their losses due to the monopolization. In none of these cases are the firms' costs of adhering to environmental, safe­ty, and economic standards deemed to be pub­lic expenditures; the expenditures and receipts involved are, indeed, constituent threads of the fabric of rights and obligations that is private property itself. According to the market failure justification, public regulation is simply a mat­ter of obtaining adherence to private standards such as these in circumstances where they would otherwise be ignored because of trans­action costs. If this is the ideal, then the result­ing expenditures ought not to be considered public business, no matter that we have strayed far from the ideal. The notion of market failure, for all its defects in theory and practice, at least has the virtue of keeping private matters pri­vate. The notion of private‑expenditures‑for­-public‑purposes dissolves the distinction.

The regulatory budget, by incorporating a large additional fraction of the economy's pri­vate expenditures into the public budget, could make it seem more natural to treat them as public expenditures outright, and easier to cali­brate them according to the good behavior of the groups and individuals affected. This possi­bility—easy to dismiss, hard to evaluate—could dwarf the problems of cost measurement and other technical aspects of implementing a regulatory budget. It is worth pondering at length before we invest too much effort in the details of implementation. If there is anything to it, the regulatory budget might join a long list of government programs which, for all of their abstract appeal, end up achieving nearly the opposite of their intended results.