• focus its attention on substantive regulatory legislation,
• not renew the attempt to pass an omnibus regulatory reform,
• evaluate proposed regulations against a broad range of standards in addition to the benefit/cost standard,
• broaden the guarantee of just compensation to all property owners who are required by regulation to provide a public benefit, and
• reassert its authority to approve all pending regulations before they become law.

Federal regulations now impose direct compliance costs of about $500 billion a year. In recent years those costs have been a roughly stable share of gross domestic product, but that apparent stability masks two contrary trends. Federal economic regulation has been declining since the late 1970s. Over the same period, however, the cost of federal regulation of health, safety, and the environment has been increasing sharply. A reduction of the relative burden of federal regulation will require some combination of reducing the remaining economic regulations; major changes in the legislative authority for the regulation of health, safety, and the environment; and much more effective administration and congressional review of both existing and proposed new regulations.

The Clinton Record at Halftime

The regulatory record of the Clinton administration (so far) has been better than the Bush record, primarily because relatively little new regulatory authority has been approved on Clinton's watch. That is the good news. You already know the bad news: the Bush record was awful. The Bush administration endorsed more costly new regulatory legislation than any administration since Nixon. (Yes, dear reader, the modern regulatory state has largely been created during Republican administrations.) The Clinton record could have been much worse; Clinton's health plan of 1993 would have been the largest single expansion of regulatory authority since the New Deal, but that plan never reached a floor vote in a Congress controlled by the Democratic Party. And the Clinton record could have been much better if the administration had recognized that "smart'' regulation, more often than not, means less regulation.

As it turns out, Congress has had a full regulatory agenda during the Clinton years without much input from the administration. Most attention has been focused on the older forms of economic regulation. Congress initiated and approved the most important agricultural and telecommunications deregulation bills in 60 years. Other changes included ending the restrictions on interstate banking, deregulating intrastate trucking, and terminating the Interstate Commerce Commission. The only new laws that significantly increased regulation were the Family and Medical Leave Act and the minimum wage increase. The Safe Drinking Water Act and the comprehensive pesticide legislation were reauthorized without much change. The major regulatory controversy between the Clinton administration and the 104th Congress involved proposed changes in the regulatory review process. Clinton opposed a comprehensive regulatory reform bill but accepted most of its provisions as parts of other legislation.

The record of administrative regulation on Clinton's watch is more complex. Clinton issued a new executive order on regulation in September 1993 that is very similar to the two Reagan orders that it replaced, and in January 1996 the Office of Management and Budget issued more detailed guidelines on how to conduct economic and risk analyses consistent with the executive order. Those administrative measures would have provided an adequate basis for review of agency-proposed rules if reinforced consistently by the White House.

At the same time, however, several regulatory agencies aggressively pressed the limits of their statutory authority with the apparent approval of the White House. The Environmental Protection Agency sought authority to set pesticide standards without regard to the economic benefits of using pesticides and general authority to set cancer risk standards without a test of statistical significance. The Occupational Safety and Health Administration issued draft guidelines on how to reduce violent crime at retail establishments that are open at night, claiming that the "general duty'' clause of its enabling legislation provides sufficient authority even without promulgation of a formal regulation. Most recently, the Food and Drug Administration announced major restrictions on tobacco marketing, the authority for which is now being challenged in court. The general lesson from those examples is that neither good executive guidance nor clear statutory language is sufficient to constrain an aggressive regulatory agency unless both the president and Congress reassert their joint authority to approve final rules. In the meantime, agencies are setting new records for excess regulation: a recent study estimates that the median cost per cancer averted by the Superfund program, for example, is $3.6 billion.

Next Steps

In the subsequent chapters, my colleagues summarize the many substantive changes in regulation that should be considered by Congress and the administration. My suggested next steps focus on the standards and process for reviewing and approving federal regulations. For years politicians and regulation analysts have been groping for some "bright line'' standard, some procedure, some "silver bullet'' to stop excessive regulation. For the most part, I suggest, that is wishful thinking; there is no substitute for the hard work necessary to revise the substantive regulatory legislation. Some changes in standards and procedures, however, can be helpful.

Standards

Scientists and economists, not surprisingly, have long promoted good science and good economics as the standards against which regulations should be evaluated. The scientists think that regulation should be considered only if there is a high level of statistical confidence in a scientifically plausible relation between cause and effect. The economists think that a regulation should be approved only if it generates the highest positive net benefits of any mutually exclusive alternative. Those standards have guided the White House regulatory review staffs beginning with President Ford and have been prescribed by executive order beginning with President Reagan.

As an economist, a former editor of the Benefit/Cost Annual, and the editor of Regulation magazine, I have also sung in that choir. Over time, however, I have come to believe that those standards are sometimes misleading and are seldom a sufficient screen against bad regulation. The crusade to regulate "by the numbers,'' I suggest, is similar to the crusade for scientific socialism--well-meant, naive, and ultimately futile.

Both the scientific and the economic standards are sometimes misleading. Careful scientific and statistical analysis is generally valuable, but it is not always appropriate to insist on a high level of statistical confidence. (The conventional standard is to reject any finding for which the probability of a zero relation is more than 5 percent.) If the cost of acting on false information is low relative to the benefits of acting on good information, it is rational to accept higher risks. In other words, the appropriate statistical standard is situation specific depending on the benefits and costs of the decision considered.

But the maximum net benefit standard itself is not a sufficient guide. Most important, the net benefit standard does not provide a rationale for a coercive transfer from some people to other people. That standard is appropriate, thus, only if its application over a set of rules generates expected net benefits for (virtually) everyone. Second, the net benefit standard does not provide a rationale for regulating the behavior of adults who bear the full marginal cost of their choices. More often than not, changes in personal behavior would increase safety at a far lower cost than would changes in environmental conditions, but this observation does not provide a basis for shifting the focus of regulation from environmental conditions to personal behavior. The net benefit standard may be the best basis for evaluating the regulation of risks to which people are involuntarily exposed. But the personal behavior of adults who bear the costs of their own choices should not be regulated at all, whatever the estimated net benefits.

And, for several reasons, those standards are seldom a sufficient screen against bad regulation. The authorizing legislation for much regulation preempts the standards by setting some other performance standard such as "reasonable certainty of no harm'' or by directly setting technical standards that preclude the opportunity to choose the most efficient means to meet a performance standard. In several cases, the Supreme Court has overruled the net benefit standard when that standard was not specifically required by Congress. In those cases, Congress must bear the responsibility for the hard work to amend the authorizing legislation.

Moreover, the regulatory agencies have learned to play the numbers game. A pattern of potentially exaggerated estimates of physical effects, benefits, and costs is difficult to check because the regulatory agencies generate most of the relevant data. The draft EPA report on the costs and benefits of the Clean Air Act is only the most egregious recent example. There are still major disputes about the basic science on which much risk regulation is based; for example, there appears to be no nonarbitrary way to extrapolate from the carcinogenic effects of very high dose rates on test animals to the effects of very low dose rates on humans, but many billions of dollars have already been spent to reduce those potential effects.

For those reasons, even with the authority of an executive order that endorses the net benefit standard, the regulatory review agencies have had little success in rejecting proposed rules that do not meet the standard. The scientific and economic standards should be supplemented by different standards; one or two standards that have proved easy to evade are not enough. A proposed federal regulation, I suggest, should meet each of the following standards:

• Does the activity by some individual or firm impose significant (nonpecuniary) adverse effects on other parties?
  
  If not, no regulation of any type is appropriate. This standard alone would rule out all regulation of activities for which people bear the full cost of their own choices, all regulation of activities that have only pecuniary effects on other parties, and all regulations that require people to provide benefits to other people.
  
• Would regulation of activities with these adverse effects be more efficient than reliance on contract and tort law?
  
  If not, the proposed regulation should be rejected in favor of the common law. This is an important question to ask, because of the frequently too casual assumption that regulation is the only instrument for reducing adverse interpersonal effects. At the same time, one should recognize that the transactions costs of the tort law are now very high, especially when there are numerous tort feasors and tort victims.
  
• Does the conduct of this activity impose significant adverse effects on people in other states?
  
  If not, state governments have sufficient incentive and authority to control those effects, by either regulation or the common law, and no federal regulation is appropriate. The fact that activities with adverse effects may be nationwide is not a sufficient basis for federal regulation; potential multistate effects, not the multistate source of those effects, should be the focus of this standard. Not all nationwide problems require a federal response.
  
• Does federal statutory law provide authority for the proposed regulation?
  
  If not, the proposed regulation should not be approved even if potentially desirable. The Constitution vests all legislative powers in Congress, and regulatory agencies should not be allowed to define their own powers. If an agency contends that some new regulatory power is desirable, they should make their case to Congress.
  
• Does the Constitution provide authority for the proposed regulation?
  
  If not, of course, the proposed regulation should be rejected; the federal government does not have the authority to define its own powers. This is an awkward issue, however, because the federal government has effectively defined its own economic powers for over 60 years. The problem is that there is no effective procedure for challenging the authority of the federal government, given the Supreme Court's generally elastic interpretation of the Constitution.
  
• Does the proposed regulation generate the highest positive expected net benefit of any mutually exclusive alternative?
  
  If not, the proposed regulation should be rejected. It is most important to recognize that this net benefit standard is relevant only if the proposed regulation clears each of the prior five hurdles.

In summary, this approach does not replace the net benefit standard, but it focuses that standard and the necessary quantitative analysis on only those proposed regulations that meet five independent either/or tests. One side effect of this approach is that it shifts much of the burden for constraining federal regulation from scientists and economists to lawyers. So be it; there are more of them than there are of us.

Procedures

The regulatory reform movement has been dominated by the quest for some "silver bullet,'' some set of standards and procedures that would stop bad regulation. For several reasons, that goal has been elusive. The net benefit standard is not a sufficient basis for a redistribution of income, for the taking of private property to provide a public benefit, or for restricting the activities of individuals and firms that bear the full cost of their choices. Judicial review provides no protection against bad analysis; the courts will not accept the role of evaluating scientific and economic studies. Congress will not accept the regimen of an automatic sunset rule.

There is only one effective solution to this problem: Congress must take much more responsibility for the rules that are made with its authority. The necessary first step is careful drafting of the substantive legislation; much, maybe most, bad regulation is a faithful interpretation of bad legislation. If Congress is the problem, only a political or constitutional challenge can stop bad regulation.

One general rule would make both Congress and the agencies more responsible: The Fifth Amendment guarantee of just compensation should be broadened to include all property owners who are required by regulation to provide a public benefit. No compensation would be required, of course, for the costs of meeting regulations to reduce a public harm originating on the property. The distinction between providing a public benefit and reducing a public harm is one that courts made for many years and should be restored. A requirement to compensate property owners who provide habitat for endangered species, for example, would enormously improve the incentives of both property owners and the government.

In many cases, however, final rules go well beyond the intent of Congress. And Congress now has no effective procedure for vetoing those rules. After a brief preamble, the first words of the Constitution are "All legislative Powers herein granted shall be vested in . . . Congress.'' For 60 years or so, however, Congress has delegated the authority to approve final rules to regulatory agencies, subject only to the constraints of the substantive legislation and the Administrative Procedures Act. Moreover, since the 1983 Chadha decision, Congress may veto an agency rule only by a new law; agency-made rules, thus, become law even if endorsed only by the president and one-third of either house. The Constitution has been turned upside down.

One way or another, Congress must restore its authority to approve all final rules. A little-noticed amendment to the 1996 debt limit bill permits Congress to delay a final rule for 60 legislative days, but overriding the rule still requires a new law. The 105th Congress should first test this new authority by a very careful selection of pending rules to delay, with the objective of using the delay to organize a sufficient coalition to override them. Congress should also consider a resolution opposing the Chadha decision and some instrument to challenge that decision; the reasoning of the Court in that decision was about as dumb as that of any decision in the past 20 years, and there is reason to hope that the Court would change its position.

Congress may soon be ready for a more radical reassertion of its constitutional authority to approve all pending rules before they become law. Several bills with that objective were introduced in the 104th Congress and should be reconsidered. Congress could continue to delegate the drafting of rules to the regulatory agencies, but an affirmative vote would be required to approve a final rule. The regulatory agencies, thus, would be transformed from rule-making and rule-enforcing agencies into rule-drafting and rule-enforcing agencies. And the constitutional separation of powers would be restored.

On the basis of the reasonable expectation that Congress would not be willing to address as many new rules as are now approved every year, a case has been made that this procedure would overload Congress. But the number of new rules is neither optimal nor exogenous. New regulations, like any other types of new laws, should be limited to those that Congress is willing to address. More congressional responsibility will both improve the quality of regulation and reduce the number of new regulations.

McKinley, Vern. "Sunrises without Sunsets: Can Sunset Laws Reduce Regulation?'' Regulation, no. 4 (1995).

Niskanen, William. "Clinton's Regulatory Record.'' Regulation, no. 3 (1996).

"Is Regulatory Reform Dead? Should Anyone Care?'' Regulation, no. 3 (1995).

"Regulating by the Numbers.'' Regulation, no. 2 (1996).

Shapiro, Martin. "A Golden Anniversary? The Administrative Procedures Act of 1946.'' Regulation, no. 3 (1996).

--Prepared by William A. Niskanen