2017-28124. Adoption of Recommendations

AGENCY:

Administrative Conference of the United States.

ACTION:

Notice.

SUMMARY:

The Administrative Conference of the United States adopted five recommendations at its Sixty-Eighth Plenary Session. The appended recommendations address Plain Language in Regulatory Drafting; Marketable Permits; Agency Guidance Through Policy Statements; Learning from Regulatory Experience; and Regulatory Waivers and Exemptions.

FOR FURTHER INFORMATION CONTACT:

For Recommendations 2017-3 and Recommendation 2017-7, Frank Massaro; for Recommendations 2017-4 and 2017-5, Gisselle Bourns; and for Recommendation 2017-6, Todd Rubin. For each of these actions the address and telephone number are: Administrative Conference of the United States, Suite 706 South, 1120 20th Street NW, Washington, DC 20036; Telephone [phone].

SUPPLEMENTARY INFORMATION:

The Administrative Conference Act, 5 U.S.C. 591-596, established the Administrative Conference of the United States. The Conference studies the efficiency, adequacy, and fairness of the administrative procedures used by Federal agencies and makes recommendations to agencies, the President, Congress, and the Judicial Conference of the United States for procedural improvements (5 U.S.C. 594(1)). For further information about the Conference and its activities, see www.acus.gov. At its Sixty-Eighth Plenary Session, held December 14-15, 2017, the Assembly of the Conference adopted five recommendations.

Recommendation 2017-3, Plain Language in Regulatory Drafting. This recommendation identifies tools and techniques agencies have used successfully to write regulatory documents (including rulemaking preambles and guidance documents) using plain language, proposes best practices for agencies in structuring their internal drafting processes, and suggests ways agencies can best use trainings and other informational resources.

Recommendation 2017-4, Marketable Permits. This recommendation provides best practices for structuring, administering, and overseeing marketable permitting programs for any agency that has decided to implement such a program.

Recommendation 2017-5, Agency Guidance Through Policy Statements. This recommendation, formerly titled Agency Guidance, provides best practices to agencies on the formulation and use of policy statements. It lists steps that agencies can take to remain flexible in their use of policy statements and to encourage, when appropriate, public participation in the adoption or modification of policy statements.

Recommendation 2017-6, Learning from Regulatory Experience. This recommendation, formerly titled Regulatory Experimentation, offers advice to agencies on learning from different regulatory approaches. It encourages agencies to collect data, conduct analysis at all stages of the rulemaking lifecycle (from pre-rule analysis to retrospective review), and solicit public input at appropriate points in the process.

Recommendation 2017-7, Regulatory Waivers and Exemptions. This recommendation provides best practices to agencies in structuring their waiver and exemption procedures for regulatory requirements. It encourages transparency and public input by asking agencies to consider establishing standards and procedures for approval of waivers and exemptions and to seek public comments in developing standards and procedures and in approving individual waivers and exemptions.

The Appendix below sets forth the full texts of these five recommendations, as well as a timely filed Separate Statement associated with Recommendation 2017-5, Agency Guidance Through Policy Statements. The Conference will transmit the recommendations to affected agencies, Congress, and the Judicial Conference of the United States, as appropriate. The recommendations are not binding, so the entities to which they are addressed will make decisions on their implementation.

The Conference based these recommendations on research reports that are posted at: https://www.acus.gov/​68thPlenary.

Dated: December 22, 2017.

Shawne C. McGibbon,

General Counsel.

APPENDIX—RECOMMENDATIONS OF THE ADMINISTRATIVE CONFERENCE OF THE UNITED STATES

Administrative Conference Recommendation 2017-3

Plain Language in Regulatory Drafting

Adopted December 14, 2017

For decades, agencies have worked to make regulatory requirements more comprehensible to regulatory stakeholders and the public at large, including by using “plain language” or “plain writing.” ^[1] Clearly drafting and explaining regulations facilitates the core administrative law goals of public participation, efficient compliance, judicial review, and the protection of rights. Numerous statutory and executive requirements direct agencies to draft rules and guidance plainly.

Plain Language Legal Requirements

The Plain Writing Act of 2010 (PWA) ^[2] and Executive Order 13,563 ^[3] require agencies to use plain language in various public-facing documents.^[4] Plain writing, as defined by the PWA, is “writing that is clear, concise, well-organized, and follows other best practices appropriate to the subject or field and intended audience.” ^[5] The Plain Language Action and Information Network (PLAIN) ^[6] further explains that “[w]ritten material is in plain language if your audience can find what they need, understand what they find, and use what they find to meet their needs.” ^[7] As such, writing in plain language does not mean abandoning complexity or nuance, nor does it mean omitting technical terms.^[8] For the purposes of this recommendation, writing that is “plain” conveys the intended meaning in a way that the intended audience can easily understand.

The PWA requires agencies to use plain language in all “covered documents,” which are: Documents necessary “for obtaining any Federal Government benefit or service or filing taxes;” documents that “provide information about any Federal Government benefit or service,” such as pamphlets; and documents that provide recommendations on “how to comply with a requirement the Federal Government administers or enforces,” such as guidance documents.^[9] Although the PWA does not cover regulations, Executive Order 13,563 requires them to be “accessible, consistent, written in plain language, and easy to understand.” ^[10] The Office of Management and Budget (OMB) interprets the PWA to apply to “rulemaking preambles,” ^[11] because a “regulation,” as exempted by the PWA, is a “rule carrying the force of law,” ^[12] but a preamble explains a rule's basis and purpose ^[13] and is not binding.

The PWA further directs agencies to: Designate “senior officials to oversee . . . agency implementation”; communicate PWA requirements to employees and train them in plain writing; maintain a “plain writing section of the agency's website”; and issue annual compliance reports.^[14] Finally, the Act precludes judicial review of agencies' compliance with its terms.^[15]

Agency Plain Language Practices

The PWA formalized and expanded a decades-long internal administrative effort to promote plain language in rules and guidance documents.^[16] For instance, many agencies have provided trainings and other resources on plain writing since the 1970s ^[17] —a practice codified by the Act.^[18] Some agencies make their trainings and related resources publicly available. Trainings may cover the PWA's requirements and plain writing techniques, including the use of organization and formatting to guide readers through a document; the use of bullet points, lists, and other visual aids; and the use of simple rather than complex vocabulary, if doing so will not alter the intended meaning. Additionally, trainings may focus on meeting the needs of the agency's various audiences, such as regulated small businesses.

Agencies must also designate officials to oversee compliance with the Act's requirements, such as by delivering trainings.^[19] Agencies may designate plain language officials in a number of different kinds of offices, such as media, executive correspondence, or public outreach. These officials can provide a valuable coordination function when the agency is communicating with the public.^[20] In some agencies, plain language officials may be well positioned to support agency staff during—not just after—the drafting process.

Rule and guidance drafting processes may directly incorporate other efforts to promote plain writing. Agencies' internal drafting manuals, which provide style and formatting guidelines, often encompass plain writing techniques. Agencies also have guidelines specifying how offices within the agency should coordinate when drafting rules or guidance. These practices have important implications for how agencies implement plain writing, though divergent approaches may be equally successful. For example, one agency's practice is to assign each office involved in drafting the responsibility for reviewing documents based on its expertise; this can include reviewing documents for plain language, in addition to reviewing them for technical sufficiency. In this agency, edits or comments on a document marked as within an office's assigned responsibilities must be either accepted or resolved in consultation with that office. Thus, a regulatory attorney may flag text that could be interpreted in multiple ways as an issue of both plainness and legal ambiguity. Similarly, program staff, economists, and engineers may be responsible for ensuring that text involving their areas of expertise is not only accurate, but plain to relevant audiences. Other agencies may not assign such formal responsibilities to particular offices; rather, the program office originating a rule or guidance may be in charge of reviewing the whole of the document and working with other participating offices to ensure text is plainly written.

Each of the above practices structures how an agency drafts rules and guidance, both of which may inform an agency's audiences of regulatory requirements or benefits.^[21] For instance, a final rule may target an audience of legal professionals and industry experts who expect to see certain terms of art, whereas a guidance document may walk a small business through the process of filing financial forms. Though it is appropriate to tailor guidance to a specialist audience, sometimes tailoring documents to particular specialist audiences runs the risk of obscuring or glossing over important information for other audiences. In certain circumstances, some commentators have raised concerns that guidance may omit salient information, leaving non-specialist parties at a disadvantage compared to experts.^[22] Crafting guidance carefully can ensure it is fully explanatory while remaining comprehensible—though this may come at the cost of brevity.^[23]

Finally, though agencies have worked to implement plain writing for rules and guidance both prior to and since the PWA's enactment, challenges remain. Inter- and intra-agency coordination in drafting is inherently difficult. Additionally, departing from language that external stakeholders expect to see, or that has required significant negotiation, may be costly. And, due to ever-present resource constraints, agencies must prioritize investing in plain writing when audiences will most benefit.

* * * * *

This Recommendation identifies tools and techniques agencies have successfully used to facilitate plain language drafting in rulemaking and guidance documents. Additionally, this recommendation proposes best practices for agencies' internal drafting processes, makes suggestions to maximize the value of trainings and related resources, and notes special considerations for drafting rulemaking preambles and guidance documents.

Recommendation

Plain Writing Practices in General

1. Agencies should follow the plain language best practices and writing techniques documented in the Federal Plain Language Guidelines.

Agency Internal Drafting Processes

2. Agencies should consider directing one or more offices involved in drafting rules and guidance to review them for plain language.

Agency Plain Language Officials, Trainings, and Related Resources

3. To improve the accessibility of rules and guidance, agency drafting staff should consider soliciting guidance or input from senior officials responsible for overseeing an agency's compliance with the Plain Writing Act (PWA).

4. When delivering trainings on plain writing techniques and the requirements of the PWA and related executive guidance, agencies should ensure appropriate focus on how plain language promotes the core administrative law goals of public participation, efficient compliance, judicial review, and the protection of rights. Agencies should additionally consider offering trainings to their technical experts to help them understand their role in the regulatory process and how they can draft technical text plainly for both specialist and non-specialist audiences.

5. In their PWA compliance reports, agencies should consider highlighting rulemaking preambles and guidance documents that exemplify plain language best practices.

Plain Drafting in Rulemaking Documents

6. To support plain drafting, internal agency rulemaking guidelines should include:

a. A requirement that rule drafters write documents in terms that the relevant audience can understand.

b. Information on plain language techniques and reference materials that the agency considers most relevant to its rulemaking practice. Such techniques include omitting excess words; using active voice, headings and other formatting techniques, such as bullet points, lists, Q&As, and other visual aids, to organize documents; and replacing complex vocabulary with simple words by, among other things, providing examples of substitutions that would be appropriate.

c. Examples of how the agency's rules, guidance, or other documents have implemented these techniques.

d. In addition to accounting for the needs of each relevant audience in any given document, at a minimum:

i. The preambles to proposed rules should include a summary of the rule that non-specialists and the general public can understand. Such summaries may be those already required by the Administrative Committee of the Federal Register or applicable executive guidance. Other subparts of the preamble should include language that is plain for specialist audiences if it is not practicable to describe the rule's purpose, reasoning, or requirements without legal or technical language, although these subparts may benefit from brief introductory summaries directed at non-specialists.

ii. The preambles and text of final rules should be written in language that reviewing courts and attorneys inside and outside the agency can easily understand.

7. Agencies should consider including in each notice of proposed rulemaking a request for comments on whether the regulation's purposes and requirements are clear and understandable. Agencies should also consider specifying topics or questions on which the agency would most benefit from feedback from non-specialist stakeholders and the general public.

Plain Drafting in Guidance Documents

8. When drafting guidance documents, agencies should tailor the guidance to the informational needs and level of expertise of the intended audiences. Audiences that are particularly likely to benefit from tailored guidance include: Regulated small business; regulatory beneficiaries, e.g., benefit recipients, consumers, and protected classes; and private compliance offices, e.g., human resources departments. For audiences that may find complex technical and legal details inaccessible, plain language summaries, Q&As, or related formats may be especially helpful.

9. When drafting guidance documents, agencies should strive to balance brevity, usefulness, and completeness. One way to help strike this balance is for guidance documents to include citations, hyperlinks, or other references or points of contact enabling readers to easily locate underlying regulatory or statutory requirements.

Administrative Conference Recommendation 2017-4

Marketable Permits

Adopted December 14, 2017

Marketable permits are a type of government-created license that regulates the level of a particular activity.^[1] Often, they ration the use of a resource (for instance, clean air by limiting pollution, fisheries by limiting fish catch, or the electromagnetic spectrum by allocating it among various uses), but they may also be used to satisfy affirmative obligations to engage in an activity (such as requirements to produce renewable energy). Marketable permits are distinguishable from other regulatory permits in that they can be bought or sold independently of any real property or other interest.^[2] Because marketable permits are alienable, it is particularly important to define their longevity and the privileges conveyed by their ownership, so that parties will understand exactly what it is that they are purchasing.

Marketable permitting programs generally fall into one of three types.^[3] In “cap-and-trade” programs, regulators set a limit, or cap, on the total amount of activity that can take place. For example, the cap could be total tons of a pollutant, total number of fish that can be caught, or total number of airport landing slots. A “rate-based trading” program is similar, but instead of capping the total amount of a regulated activity, agencies limit the relative amount of activity per regulated entity or unit of regulated activity. For example, a rate-based air pollution permit market may limit the amount of pollution power plants can emit per unit of electricity generated, and fuel efficiency standards set limits on the acceptable amount of fuel required to drive a mile. Finally, in “credit trading” systems, regulators set a relative goal (e.g., no net emissions increase or no net increase in property development), and then any covered entities seeking, for example, to increase emissions or develop property must purchase offsetting credits that are sold by third parties and verified by regulators. Credits can be earned when parties limit their level of the regulated activity by more than the required amount. Credit systems can also be combined with cap-and-trade or rate-based programs. For example, in a greenhouse gas cap-and-trade program, unregulated sources may be allowed to reduce their emissions voluntarily and sell verified credits on the market. In a property development setting, a party could decline to develop a particular parcel of land to generate a credit, and then sell that credit to another party.

Establishing a Marketable Permitting Program

Like other agency activities, marketable permitting programs must be within the agency's statutory authority. But even when an agency has statutory discretion to use a marketable permitting program, such a program may not be the most suitable regulatory tool to achieve an agency's goal. Marketable permitting programs are more likely to be suitable when:

• The agency can clearly define the privileges or obligations to be assigned by the program and has the necessary information to set the level of regulated activity.
• The agency has sufficient resources to design and administer the program and is capable of reevaluating the appropriate target level of activity over time.
• The agency finds it difficult or expensive to discern compliance costs for individual regulated parties. This often occurs when the activity to be regulated is conducted by numerous heterogeneous or small sources, or when there are as yet unrealized opportunities for significant technological developments by actors other than those upon whom the regulatory obligations fall.
• The agency is reasonably confident that a robust market is feasible. This requires interest and participation by regulated entities that have, or are capable of developing, sufficient knowledge to make efficient decisions in the market.
• Regulated parties have sufficiently differing compliance costs, such that the savings from trading are likely to be greater than transaction costs.
• The agency determines that the overall level of an activity is more significant than the identity or location of the actors engaging in the activity. Alternatively, a marketable permit system could take locational differences into account in its structure, by, for example, setting prices so that it costs more to buy permits in a place where the marginal benefits of cutbacks are high.^[4]

Marketable permitting programs are less likely to be suitable when:

• The balance of factors listed above is not favorable.
• The risk of unintended consequences from trading, such as the potential for localized problems,^[5] is difficult to manage.

Once an agency has decided to create a marketable permitting program, it must consider how to establish it. Many agencies have used notice-and-comment rulemaking when creating a marketable permitting regime.^[6] In a handful of instances, agencies have established marketable permitting programs through guidance documents.^[7] Since agencies cannot impose legally binding obligations through guidance documents,^[8] this latter approach can lead to some uncertainty among existing and prospective permittees and even agency officials as to the permanence of the program.^[9] While notice-and-comment rulemaking has costs, it also has the virtue of soliciting stakeholder input while a rule is being shaped.^[10] Public input can be beneficial in determining whether a particular activity lends itself to regulation via a marketable permitting regime and, if so, how the program should be designed so as to best serve the public interest.

Allocating Permits

Once a marketable permitting program has been established, permits will need to be distributed. The initial allocation of permits is referred to as the “primary market” for permits.^[11] Agencies typically develop systems and regulations to allocate and keep track of permits and to verify their ultimate retirement, under their authority to implement the underlying permitting program.

Agencies predominantly follow one of two approaches in distributing permits: Historical-based allocations and auctions. Historical-based allocations distribute permits based on historical use of the regulated activity. This method is typically used to avoid disruptions to the status quo, to protect returns on past investments, and to ease tensions with the regulated industry and gain political support. However, it may also reward parties for engaging in activity that the agency now wants to curb, increase the risk of monopolies in the permit market, reduce the incentive to innovate, and incentivize undesirable strategic behavior, like a firm artificially inflating its use of a resource ahead of an allocation benchmark to increase its share of allocated permits.^[12]

By comparison, distributing permits through auctions reduces the barriers to entry to the regulated activity. Auctions also tend to lower the risk of monopolies and strategic behavior, facilitate price discovery, and prevent undue windfalls. However, auctions can be challenging to administer, especially for agencies without prior experience in doing so, and may require significant resources upfront to design and implement.^[13]

There are also several other, less common ways of conducting initial permit allocation that may be useful in certain specialized contexts. These include output-based allocations,^[14] allocating permits to particular communities,^[15] or allocating permits based on other policy objectives.

In deciding how to allocate permits, agencies must make two additional important decisions. The first is to decide who is eligible to purchase permits. Some agencies restrict the buying and selling of permits to regulated entities, whereas others allow non-regulated parties—such as brokers, speculators, market facilitators, or the general public—to purchase permits. Allowing access to the market for permits to a wider range of parties can promote market liquidity and facilitate efficient price discovery, though it also increases the risk of market participants trying to “corner the market” (amassing permits to control prices). Allowing unregulated parties to buy permits and retire them also allows the public to decrease the level of the cap.

The second is whether to hold a pool of permits in reserve for future entrants. Once the initial allocation of permits has been made, in the absence of competitive markets, permit holders may have an incentive to impede purchases from potential new competitors.^[16] Agencies have sometimes addressed this barrier to entry by creating a reserve pool of permits for new entrants. Some agencies have also instituted similar mechanisms for introducing permits into the market in the wake of large economic changes or emergencies that heavily drive demand for permits.

Overseeing a Marketable Permitting Program

Once initial permit distribution has occurred, agencies will want to ensure that parties comply with any obligations that arise under their permits. Monitoring ongoing performance is essential to achieving compliance with permit obligations. This includes tracking ownership of permits through their lifecycle, tracking the amount of regulated activity by permit holders, and verifying that credits represent real offsets of regulated activity. Agencies often conduct compliance monitoring themselves, but sometimes rely on self-verification by regulated parties or use third parties to verify compliance.^[17]

In the event that regulated parties engage in more of the regulated activity than their permits allow, agencies have several enforcement tools.^[18] For instance, agencies can require parties to buy additional permits until their use is in compliance with the number of permits they possess and can require parties to develop plans to ensure future compliance. Agencies can also impose sanctions. There is evidence that compliant parties are more supportive of enforcement in marketable permitting programs because noncompliance by other parties lowers the value of their allowances.^[19]

Compliance monitoring and enforcement are important aspects of ensuring the integrity of a marketable permitting program. Another involves overseeing secondary and derivative markets that may emerge, with or without government assistance, following the initial allocation of permits. The secondary market for permits involves transactions in which permits are bought and sold following their initial entry into commerce in the primary market. This is in contrast to derivative markets, which are primarily risk management and price discovery markets in which actual transfer of permits might not occur.^[20] Trading in secondary and derivative markets can be accomplished through (1) negotiations between buyers and sellers—which may or may not be facilitated by third parties (these are known as over-the-counter transactions)—or (2) exchanges, which match buyers and sellers in standardized transactions.^[21]

The authority to oversee trading on secondary markets is somewhat fragmented, and authority over marketable permit programs is not always well defined and would benefit from clarification. The Commodity Futures Trading Commission (CFTC) has broad enforcement authority to pursue manipulation of the price of a commodity in interstate commerce.^[22] It also has the authority to surveil spot trading (sales for the immediate delivery of a commodity) conducted on exchanges.^[23] However, the CFTC only rarely brings enforcement actions for fraud in spot markets. The Federal Trade Commission (FTC)—under its authority to act against unfair, anticompetitive, and deceptive practices affecting commerce—and the Department of Justice—under its antitrust authority—also have some authority over secondary permit markets, though they have had limited involvement with marketable permitting programs to date. An individual agency's ability to oversee secondary markets will depend on its statutory authority, but even when it does have such authority, it may lack the expertise or resources to routinely monitor trading in these markets.

Authority to oversee derivative markets is largely vested in the CFTC.^[24] It oversees derivatives traded in exchanges, which must publish certain kinds of trading information that would allow the CFTC to detect fraud and manipulation. The CFTC also has authority to oversee over-the-counter transactions. The CFTC's authority over derivative markets, and particularly over-the-counter derivative transactions, was strengthened by the Dodd-Frank Wall Street Reform and Consumer Protection Act.^[25]

Agencies with authority to oversee permit markets have various tools to combat fraud, manipulation, and price volatility, all of which can undermine economic efficiency and erode confidence in permit markets. Fraud and manipulation can be addressed through various mechanisms, such as position limits, accountability triggers, market surveillance, and reporting requirements. Position limits can be used to ensure that no single party or combination of parties can control the supply of permits to the point of dictating prices. Position accountability triggers, which require permit holders wishing to exceed a certain threshold of permits to submit to additional reporting and oversight, can likewise be used to prevent hoarding of permits. Effective surveillance of markets and robust reporting requirements also discourage fraudulent activity.

Price volatility can occur in marketable permitting programs even without fraudulent activity, particularly in smaller, less robust markets with fewer participants, due to unexpected increases in demand or the costs of compliance. Volatility increases the risk of noncompliance and decreases confidence in the market system. Tools to address volatility include circuit breakers, which limit how much prices can rise or fall in a given period, and safety valves, which can set maximum or minimum prices or release reserve credits into the market in case of emergencies or demand spikes. Another way to reduce volatility is to issue permits with different durations. Finally, by defining a broader program that covers more entities under a single market, agencies can diversify the portfolio of permit seekers, reducing the risk of unexpectedly high cost in an isolated sector. Any individual regulated sector can experience unexpected compliance costs as economic conditions change; a broader market offers more flexibility, better absorbs price volatility, and so increases certainty for regulated parties and investors.

Because permit markets rely heavily on the decisions of both the agency and permit buyers, facilitating the flow of information is an extremely important part of a marketable permitting program. Making data on permit transactions, prices, and holdings publicly available can help the agency and the public assess the efficacy of the program. It also enables smooth operation of the permit markets by enabling permit buyers to better evaluate the value of the permits. Having clear communication policies for announcing policy changes or enforcement actions that could influence the market prevents pre-publication leaks and information asymmetries that could unjustly benefit some parties and undermine the permit market.

* * * * *

This Recommendation does not address whether agencies should increase or reduce their usage of marketable permitting programs or speak to the substantive areas in which such programs may be desirable. Rather, the Administrative Conference acknowledges that agencies have been directed to consider marketable permits, consistent with statutory authorization and any applicable statutory requirements, as one possible mode of regulation and seeks to identify the key considerations in assessing marketable permits as a potential alternative.^[26] This Recommendation highlights best practices that agencies should consider in designing a marketable permitting program.

Recommendation

Establishing a Marketable Permitting Program

1. When designing a marketable permitting program, agencies should carefully consider whether such a program will best achieve their policy objectives, and, if so, whether the agency's goals would be better served by using a cap-and-trade, rate-based, or credit trading system or a combination of the above.

2. Agencies should establish and publish clear guidelines containing all of the features of marketable permit programs, including expectations as to the longevity of marketable permits and the precise obligations or authorizations that they convey.

3. Agencies should generally consider using notice-and-comment rulemaking when creating a marketable permitting regime, both in order to reduce uncertainty as to the permanence of the program and to gather public input that may prove beneficial in shaping the program.

4. Agencies should consider whether to allow non-regulated parties to buy and sell permits. Allowing a broader range of parties to trade permits can promote market liquidity and facilitate efficient price discovery but may increase opportunities for manipulation in thin markets.

5. Agencies should explore agreements with other appropriate agencies and authorities to allocate responsibilities for developing standards or policies, where appropriate. These actions may include addressing compliance enforcement and market manipulation.

Overseeing a Marketable Permitting Program

6. As with other types of permitting programs, when designing a marketable permitting program, agencies should include mechanisms to ensure compliance with the program. Agencies should monitor performance by tracking ownership of permits, tracking regulated activity, and verifying that credits represent real offsets from regulated activity. Depending on feasibility and efficiency, agencies should consider verifying compliance directly, making use of self-verification, or engaging third parties to verify compliance. Self-verification tends to be a useful option when verification procedures can be standardized or when legal remedies are available to aid in enforcement. If an agency chooses to use third-party credit verifiers, it should set standards to ensure that they are qualified, insured, and free from conflicts of interest.

7. As with other types of permitting programs, in designing a marketable permitting program, agencies should require noncompliant parties to come into compliance and should include sanctions with sufficient deterrent effect to discourage noncompliance.

8. Agencies should coordinate with other appropriate agencies and authorities to identify which oversight tools are appropriate to prevent fraud and manipulation.

9. Agencies should address extreme price volatility by creating broad markets, issuing permits with different durations, or using circuit breakers, safety valves, or reserve pools, as necessary. Agencies should also consider using reserve pools to facilitate new parties entering the market.

Information Management

10. Subject to other agency priorities and applicable legal requirements, including the Paperwork Reduction Act (PRA) and e-Government Act, agencies should collect data on the operation of marketable permitting programs and consider periodically assessing both the policy effectiveness and economic efficiency of existing marketable permitting programs. Agencies should be cognizant that some of the data collected may be confidential and protected against disclosure by law.

11. To the extent practicable, agencies should release data on permit transactions, prices, holdings, compliance rates, and other data to help the public gauge a market's policy effectiveness and to help parties make efficient decisions in the market.

12. Agencies that manage marketable permitting programs should coordinate with other agencies and authorities that have expertise to improve marketable permitting programs.

13. In order to minimize information asymmetries, agencies should develop communication policies for announcing policy changes or enforcement actions that could influence the market.

Administrative Conference Recommendation 2017-5

Agency Guidance Through Policy Statements

Adopted December 14, 2017

General statements of policy under the Administrative Procedure Act (hereinafter policy statements) are agency statements of general applicability, not binding on members of the public, “issued . . . to advise the public prospectively of the manner in which the agency proposes to exercise a discretionary power.” ^[1] Interpretive rules are defined as rules or “statements issued by an agency to advise the public of the agency's construction of the statutes and rules which it administers.” ^[2] Both policy statements and interpretive rules are exempt from the APA's requirements for the issuance of legislative rules (including notice and comment) ^[3] and are often referred to as “guidance” or “guidance documents” (although usage varies). This Recommendation, however, covers only policy statements, not interpretive rules; nevertheless, many of the recommendations herein regarding flexible use of policy statements may also be helpful with respect to agencies' use of interpretive rules.

Over the years, the Conference has issued several recommendations pertaining to policy statements. Recommendation 76-5 states that agencies should provide for public participation in the formulation of policy statements (and of interpretive rules) depending on the impact of the statement in question and the practicability of participation.^[4] Recommendation 92-2 recognizes the value of policy statements but expresses concern about policy statements “that are intended to impose binding substantive standards or obligations upon affected persons” notwithstanding the legal requirement that they be nonbinding on the public, and it advises agencies to establish flexible procedures that allow members of the public a fair opportunity to argue for approaches different from those set forth in a policy statement.^[5] The Conference has now decided, twenty-five years after Recommendation 92-2, to update its recommendations on the formulation and use of policy statements in light of current administrative experience.^[6]

Policy statements are important instruments of administration across numerous agencies, and are of great value to agencies and the public alike. Compared with adjudication or enforcement, policy statements can make agency decisionmaking faster and less costly, saving time and resources for the agency and the regulated public. They can also make agency decisionmaking more predictable and uniform and shield regulated parties from unequal treatment, unnecessary costs, and unnecessary risk, while promoting compliance with the law.^[7] Compared with legislative rules, policy statements are generally better for dealing with conditions of uncertainty and often for making agency policy accessible, especially to regulated parties who lack counsel. Further, the provision of policy statements often takes less time and resources than legislative rulemaking, freeing up the agency to, for instance, take other action within its statutory mission. In pursuit of benefits such as these, agencies may use policy statements to bind some agency employees to the approach of the policy statement,^[8] so long as such employees are not bound in a manner that forecloses a fair opportunity for the public or employee to argue for approaches different from those in the policy statement or seek modification of the policy statement.^[9]

Despite their usefulness to both agencies and the public, policy statements are sometimes criticized for coercing members of the public as if they were legislative rules, notwithstanding their legally nonbinding status. Recommendation 92-2 defined this problem in terms of an agency's intent to use policy statements to bind the public, which may imply that the problem is one of agency bad faith. While agency intent to make a policy statement binding, if shown, would deserve criticism and correction, a focus on intent is often inadequate for understanding and addressing the phenomenon of binding policy statements. This Recommendation supplements Recommendation 92-2 by addressing other reasons why members of the public may feel bound by what they perceive as coercive guidance.

There are several kinds of reasons why members of the public sometimes find they have no practical escape from the terms of a policy statement. First are those that are not of the making of an agency or its officials. Specifically, modern regulatory schemes often have structural features that tend to lead regulated parties to follow the policy statement's approach even if in theory they might be legally free to choose a different course, because the costs and risks associated with doing so are simply too high. This is often the case if statutes or regulations (a) require a regulated party to obtain prior approval from an agency to obtain essential permissions or benefits; (b) subject a regulated party to repeated agency evaluation under a legal regime with which perfect compliance is practically unachievable, incentivizing the party to cultivate a reputation with the agency as a good-faith actor by following even non-binding guidance; or (c) subject the regulated party to the possibility of enforcement proceedings that entail prohibitively high costs regardless of outcome, or can lead to sanctions so severe that the party will not risk forcing an adjudication of the accusation. Meanwhile, a policy statement can operate on beneficiaries of a statute or legislative rule as if it were a legislative rule by effectively depriving them of the statute or legislative rule's protection. This can occur if the policy statement promises to treat regulated parties less stringently than the statute or legislative rule requires, effectively freeing those parties to shift their behavior in a direction that harms beneficiaries. Similarly, in its focus on regulatory beneficiaries and regulated parties, an agency policy statement may induce conduct harmful to other interested parties.

Second, there are a number of reasons why agencies themselves may naturally tend to be somewhat inflexible with respect to their own policy statements. Even though these reasons are more within an agency's or its officials' control than those discussed above, this lack of flexibility may often stem from causes other than agency intent. Officials who behave inflexibly may be seeking to balance the importance of being flexible against stakeholder demands to honor other, competing values that officials would be remiss to ignore. For example, if one regulated firm argues for a different approach from that in a policy statement and the agency approves, this may prompt other firms to criticize the agency for not keeping a level playing field among competitors; may cause other firms to lose faith in the agency's consistency and predictability, which may render them less likely to trust and cooperate with the agency; and may open the agency to accusations of favoritism from non-governmental organizations (NGOs), the media, and congressional overseers.

In principle, one way an agency might reconcile these understandable pressures would be to prepare and disseminate written reasons when it approves an approach different from that in a policy statement, thereby making the same reasoning available to all similarly situated parties going forward. This transparency helps level the playing field, makes agency behavior more predictable, and diminishes concerns about favoritism. But agencies might still find inflexibility the easier course and adopt it by default, because reason-giving requires agency resources.^[10] Besides this, there are additional organizational reasons for inflexibility: Some agency offices, by reason of their usual day-to-day business, are socialized to be less receptive to stakeholder requests than others; higher-level officials have institutional reasons to back the decisions of their subordinates; and the distinction between binding and nonbinding policies is counter-intuitive for many officials, at least without substantial training.

These various pressures tend to give at least some policy statements a quasi-binding character in fact regardless of their legal status. That said, there are important steps that agency officials can take to mitigate these legislative-rule-like effects of policy statements by stating that they are not binding ^[11] and by remaining flexible in their use of such statements by offering members of the public a fair opportunity to argue for other approaches. What steps to take and when is the focus of paragraphs 4 through 8 of this Recommendation. Agencies should also, in appropriate circumstances, use appropriate tools to enable public participation in the formulation of policy statements before these statements are adopted. This is the focus of paragraphs 9 through 11 of this Recommendation.

First, flexibility often requires managerial initiative and resources to foster and maintain. This Recommendation identifies concrete organizational measures that agencies may take to foster flexibility: Low-cost measures that agencies should take at a minimum and additional measures with higher cost that agencies should consider in light of resource limitations and competing priorities.

In addition, public participation at the time of a policy statement's adoption may be of value to the agency, regulated parties, regulatory beneficiaries, and other interested parties. Such public participation may be especially valuable to parties that lack the opportunity and resources to participate in the individual adjudicatory or enforcement proceedings to which a policy may apply.

Choosing a level and means of public participation that is appropriate to a policy statement's likely impact and is practicable requires consideration of several factors. Given the complexity of these factors and their tendency to vary with context, it is appropriate to make decisions about whether or how to seek public participation on policy statements on a document-by-document or agency-by-agency basis.^[12] A government-wide requirement for inviting written input from the public on policy statements is not recommended, unless confined to the most extraordinary documents.^[13] This is a function both of the complex cost-benefit considerations noted above and the fact that broad mandates for written public input on policy statements can result in two additional unintended consequences. First, a broad mandate applied to a resource-strapped agency may cause the agency to fail to process and incorporate comments and instead leave many policy statements in published “draft” form indefinitely, which may at least partly defeat the purpose of participation and cause stakeholder confusion. Second, a broad mandate may so legitimize policy statements in the eyes of the agency that such statements could end up largely supplanting legislative rulemaking.

Recommendation

Policy Statements Should Not Bind the Public

1. An agency should not use a policy statement to create a standard binding on the public, that is, as a standard with which noncompliance may form an independent basis for action in matters that determine the rights and obligations of any member of the public.

2. An agency should afford members of the public a fair opportunity to argue for lawful approaches other than those put forward by a policy statement or for modification or rescission of the policy statement.

3. Although a policy statement should not bind an agency as a whole, it is sometimes appropriate for an agency, as an internal agency management matter, and particularly when guidance is used in connection with regulatory enforcement, to direct some of its employees to act in conformity with a policy statement. But the agency should ensure that this does not interfere with the fair opportunity called for in Recommendation 2. For example, a policy statement could bind officials at one level of the agency hierarchy, with the caveat that officials at a higher level can authorize action that varies from the policy statement. Agency review should be available in cases in which frontline officials fail to follow policy statements in conformity with which they are properly directed to act.

Minimum Measures To Avoid Binding the Public

4. A policy statement should prominently state that it is not binding on members of the public and explain that a member of the public may take a lawful approach different from the one set forth in the policy statement or request that the agency take such a lawful approach. The policy statement should also include the identity and contact information of officials to whom such a request should be made.

5. A policy statement should not include mandatory language unless the agency is using that language to describe an existing statutory or regulatory requirement, or the language is addressed to agency employees and will not interfere with the fair opportunity called for in Recommendation 2.

6. The agency should instruct all employees engaged in an activity to which a policy statement pertains to refrain from making any statements suggesting that a policy statement is binding on the public. Insofar as any employee is directed, as an internal agency management matter, to act in conformity with a policy statement, that employee should be instructed as to the difference between such an internal agency management requirement and law that is binding on the public.

Additional Measures To Avoid Binding the Public

7. In order to avoid using policy statements to bind the public and in order to provide a fair opportunity for other lawful approaches, an agency should, subject to considerations of practicability and resource limitations and the priorities described in Recommendation 8, consider additional measures, including the following:

a. Promoting the flexible use of policy statements in a manner that still takes due account of needs for consistency and predictability. In particular, when the agency accepts a proposal for a lawful approach other than that put forward in a policy statement and the approach seems likely to be applicable to other situations, the agency should disseminate its decision and the reasons for it to other persons who might make the argument, to other affected stakeholders, to officials likely to hear the argument, and to members of the public, subject to existing protections for confidential business or personal information.

b. Assigning the task of considering arguments for approaches other than that in a policy statement to a component of the agency that is likely to engage in open and productive dialogue with persons who make such arguments, such as a program office that is accustomed to dealing cooperatively with regulated parties and regulatory beneficiaries.

c. In cases where frontline officials are authorized to take an approach different from that in a policy statement but decline to do so, directing appeals of such a refusal to a higher-level official who is not the direct superior of those frontline officials.

d. Investing in training and monitoring of frontline personnel to ensure that they (i) understand the difference between legislative rules and policy statements; (ii) treat parties' ideas for lawful approaches different from those in a policy statement in an open and welcoming manner; and (iii) understand that approaches other than that in a policy statement, if undertaken according to the proper internal agency procedures for approval and justification, are appropriate and will not have adverse employment consequences for them.

e. Facilitating opportunities for members of the public, including through intermediaries such as ombudspersons or associations, to propose or support approaches different from those in a policy statement and to provide feedback to the agency on whether its officials are giving reasonable consideration to such proposals.

Priorities in Deciding When To Invest in Promoting Flexibility

8. Because measures to promote flexibility (including those listed in Recommendation 7) may take up agency resources, it will be necessary to set priorities for which policy statements are most in need of such measures. In deciding when to take such measures the agency should consider the following, bearing in mind that these considerations will not always point in the same direction:

a. An agency should assign a higher priority to a policy statement the greater the statement's impact is likely to be on the interests of regulated parties, regulatory beneficiaries, and other interested parties, either because regulated parties have strong incentives to comply with the statement or because the statement practically reduces the stringency of the regulatory scheme compared to the status quo.

b. An agency should assign a lower priority to promoting flexibility in the use of a policy statement insofar as the statement's value to the agency and to stakeholders lies primarily in the fact that it is helpful to have consistency independent of the statement's substantive content.

Public Participation in Adoption or Modification of Policy Statements

9. When an agency is contemplating adopting or modifying a policy statement, it should consider whether to solicit public participation, and, if so, what kind, before adopting the statement. Options for public participation include outreach to selected stakeholder representatives, stakeholder meetings or webinars, advisory committee proceedings, and invitation for written input from the public with or without a response. In deciding how to proceed, the agency should consider:

a. Existing agency procedures for the adoption of policy statements, including any procedures adopted in response to the Office of Management and Budget's Final Bulletin for Agency Good Guidance Practices (2007).

b. The factors listed in Recommendation 8.

c. The likely increase in useful information available to the agency from broadening participation, keeping in mind that non-regulated parties (regulatory beneficiaries and other interested parties) may offer different information than regulated parties and that non-regulated parties will often have no opportunity to provide input regarding policy statements other than at the time of adoption.

d. The likely increase in policy acceptance from broadening participation, keeping in mind that non-regulated parties will often have no opportunity to provide input regarding policy statements other than at the time of adoption, and that policy acceptance may be less likely if the agency is not responsive to stakeholder input.

e. Whether the agency is likely to learn more useful information by having a specific agency proposal as a focal point for discussion, or instead having a more free-ranging and less formal discussion.

f. The practicability of broader forms of participation, including invitation for written input from the public, keeping in mind that broader participation may slow the adoption of policy statements and may diminish resources for other agency tasks, including the provision of policy statements on other matters.

10. If an agency does not provide for public participation before adopting or modifying a policy statement, it should consider offering an opportunity for public participation after adoption. As with Recommendation 9, options for public participation include outreach to selected stakeholder representatives, stakeholder meetings or webinars, advisory committee proceedings, and invitation for written input from the public with or without a response.

11. An agency may make decisions about the appropriate level of public participation document-by-document or by assigning certain procedures for public participation to general categories of documents. If an agency opts for the latter, it should consider whether resource limitations may cause some documents, if subject to pre-adoption procedures for public participation, to remain in draft for substantial periods of time. If that is the case, agencies should either (a) make clear to stakeholders which draft policy statements, if any, should be understood to reflect current agency thinking; or (b) provide in each draft policy statement that, at a certain time after publication, the document will automatically either be adopted or withdrawn.

12. All written policy statements affecting the interests of regulated parties, regulatory beneficiaries, or other interested parties should be promptly made available electronically and indexed, in a manner in which they may readily be found. Written policy statements should also indicate the nature of the reliance that may be placed on them and the opportunities for reconsideration or modification of them or the taking of different approaches.

Separate Statement for Administrative Conference Recommendation 2017-5 by Senior Fellow Ronald M. Levin

Filed December 20, 2017

The accompanying Recommendation observes that “[t]his Recommendation . . . concerns only policy statements, not interpretive rules; nevertheless, many of the recommendations herein regarding flexible use of policy statements may also be helpful with respect to agencies' use of interpretive rules.” This remark is well taken as far as it goes, but in another respect it is notably cautious. Other governmental bodies that have adopted procedures or guidelines regarding the same general subject during the past two decades have each used only one framework to address all guidance—that is, both policy statements and interpretive rules.^[1]

In adopting the Recommendation, the Assembly of the Administrative Conference was generally sympathetic to the stance taken by the groups just mentioned, but it concluded that it did not have enough information to take a firm stand. The research for its project had focused primarily on policy statements. Thus, the Assembly opted for a relatively narrow recommendation for the present, but it also adopted a “sense of the Conference” resolution envisioning a follow-up study that would lay the groundwork for a subsequent recommendation on interpretive rules. The Assembly's caution is understandable, but I will use this separate statement to emphasize that its ancillary resolution has pointed in the right direction.

The basic problem that Recommendation 2017-5 seeks to redress is that regulated persons sometimes feel that they have no choice other than to comply with a policy statement's position, even if they disagree with it. The Recommendation seeks to mitigate that problem by suggesting ways in which an agency can give those persons a fair opportunity to ask the agency to reconsider and perhaps change its position. At the same time, the Recommendation's solutions are made “subject to considerations of practicability and resource limitations,” so as to avoid deterring agencies from giving advice that the public desires.

Essentially the same analysis can also be applied to interpretive rules: The relative proportion of law and policy in the document has little or nothing to do with either the agency's interest in giving advice or the private party's interest in being able to induce the agency to reconsider it. Moreover, in practice, law and policy blend together in many guidance document; thus, procedures that speak to one and not the other are bound to prove somewhat artificial.

Why, then, wouldn't one urge agencies to apply the same principles to interpretive rules? It may be thought that, in contrast to its handling of policy statements, an agency will naturally treat an interpretive rule as binding, because it concerns binding law. But that is a non-sequitur. An agency should, of course, be free to state and act on its position that a statute or regulation, as construed in an interpretive rule, is binding. However, the very purpose of issuing such a rule is to specify which of various imaginable readings of the statute or regulation the agency considers correct. Persons who may believe that a different interpretation is correct should have what Recommendation 2017-5 calls a “fair opportunity” to try to persuade the agency to adopt their preferred view—just as the Recommendation contemplates with respect to policy statements. For an agency to assert that, because the underlying text is binding, the interpretation that the agency happens to have chosen must also be binding is to beg the question that ought to be the subject of that dialogue.

The Assembly was mindful that opinions have differed on the question of whether, for procedural purposes, interpretive rules can be binding in a sense that policy statements cannot be. As just suggested, I myself believe the answer is no, but some agency lawyers think otherwise. Ultimately, however, that divergence in opinion should not prevent the Conference from moving forward with a recommendation in the next phase of its inquiry. As with most Conference pronouncements, the principal goal should be to articulate recommended practices, not to opine about the law.

I hope that a project of the kind contemplated by the sense of the Conference resolution will be pursued in the near future. I trust that it will culminate in broad recognition that most, if not all, of the advice in the present Recommendation can and should be applied to interpretive rules as well.

Administrative Conference Recommendation 2017-6

Learning From Regulatory Experience

Adopted December 15, 2017

Making sound regulatory decisions demands information and analysis. Several Administrative Conference recommendations encourage agencies to gather data when making new rules and when reviewing existing rules.^[1] These recommendations reinforce analytic demands imposed on agencies by legislation,^[2] executive orders,^[3] and judicial decisions.^[4]

Agencies need information about the problems that new rules will address, such as the risks involved and their causes. But agencies also need information about potential solutions to these problems. What possible alternative rules or rule designs might help solve the problems? How effective are these alternatives likely to be in addressing the underlying problems? Are there constraints, barriers, or unanticipated consequences that arise in the use of these different alternatives? In terms of understanding possible alternatives and how well they might work in practice, agencies benefit from having information from experience with different solutions. Learning from experience is the focus of this recommendation.

Learning From Regulatory Experience

No uniform or tidy formula exists as to how agencies should generate, gather, and analyze the data necessary to support sound regulatory decisions. A variety of well-accepted and widely-used methods exist from which agencies may choose, with the appropriate choices often varying agency by agency and even from situation to situation. Practical considerations such as resource and data availability will affect the choices agencies make about the methods of learning used to support regulatory decisionmaking.^[5] Still, it is possible to identify some of the main methods for learning that agencies should consider using at different stages of the rulemaking lifecycle. These methods, which are not necessarily mutually exclusive, can be used before or after a rule is adopted, and they may be considered on occasion as part of the final rule itself, which might be structured to facilitate future learning by agency officials.

Variation is the key to agency learning. In this context, “variation” can refer to differences among jurisdictions ^[6] or across time,^[7] with some jurisdictions or time periods having in place a version of a rule and others having in place a different version of the rule (or no applicable rule at all). It can also refer to differences among regulated entities or people within the same jurisdiction, with some entities or people subject to a version of a rule and others subject to a different version of the rule (or no applicable rule at all).

An agency can learn from all of these kinds of variation. For example, a regulation that goes into effect in 2017 leaves the agency with two distinct time periods to compare: The years before 2017, and 2017 and beyond. A rule that applies in jurisdictions X and Y but not in jurisdictions A and B leaves the agency with the ability to compare outcomes in X and Y with those in A and B, assuming the jurisdictions are comparable or that differences can be statistically controlled. The agency can then learn whether outcomes are improved in those time periods or jurisdictions with the regulatory obligation. However, agencies must be careful not to assume automatically that any differences in outcomes that they observe have been caused by the intervention of the regulation. Other factors that correlate with the observed outcomes might also vary across the same time periods or jurisdictions.

Using Observational or Randomized Methods To Learn From Experience

To learn from experience, agencies should seek methods that allow them to draw valid inferences about whether a particular regulatory intervention causes (or will cause) improvements in the desired outcomes. Concern about the validity of such causal inferences generally takes two forms. The first of these—external validity—refers to the extent to which the inferences from a study situated within a particular time period or setting can apply to other time periods or settings. In other words, an agency should consider to what extent the results of a study focused on entities or individuals in one period or setting are generalizable to entities or individuals in other times or settings. The second type of validity—internal validity—refers to the extent to which the outcomes observed in a study can be said to have been caused by the intervention rather than by potential confounders.^[8] In other words, an agency should consider whether what might appear to be a relationship between a regulation and changes in outcomes truly derives from the regulation. For example, if a study shows that accidents from a particular industrial process have declined following the adoption of a regulation intended to reduce those accidents, concern about internal validity would lead agency officials to consider the possibility that the observed decline might have arisen from market or technological factors that led to changes in the relevant industrial processes around the same time as the regulation but which came about for reasons entirely unrelated to the regulation. An agency may wish to learn whether the observed decline came from the regulation or from other factors so as to know whether to redesign the regulation if further improvements are warranted.

To isolate the true effects of a regulation on relevant outcomes, such as risk reduction, agencies can use randomized approaches or observational approaches. Both of these approaches have advantages and disadvantages, and choosing between them will depend on a variety of contextual factors.

Randomized approaches promise to generate results with a high level of internal validity because, by making a random assignment of individuals or entities subject to a regulatory intervention, any other factors that might lead to changes in the relevant outcomes should be distributed randomly between the group subject to the regulatory intervention and the comparison group. Of course, randomized methods can also have their limitations. There is always a question as to whether the results of a randomized experiment are externally valid. For example, a perfectly designed randomized experiment may indicate that exposure to an intervention generates particular outcomes in a laboratory setting but may not mean that those same outcomes will occur outside of the laboratory. In addition, the results of randomized methods may lack validity if individuals, knowing that their behaviors are part of a randomized experiment, behave differently from how they would otherwise act. Researchers try to limit this particular threat to validity by using double-blind, or even just single-blind, study designs.^[9] However, it is possible that, in many regulatory contexts, regulated parties will know they are subject to a randomized study and may engage in strategic behavior that may skew the results of the study.

In addition to these methodological challenges, randomized study methods may present legal, policy, and ethical concerns. From a legal standpoint, subjecting similar parties to different rules may be thought to raise concerns under the Equal Protection Clause of the Constitution or the arbitrary-and-capricious standard of the Administrative Procedure Act.^[10] Of course, an agency might present a legally valid argument that the rational basis, or non-arbitrary reason, for its action is to generate information necessary to make an informed decision.^[11] From a policy standpoint, if some entities are subject to regulation and others are not, an agency may well risk artificially distorting a market, depending on what a rule requires or how the study is designed. From an ethical standpoint, if a rule specifically sets up an experiment with the idea that, after the experiment, the agency may change the rule, a concern may exist if some regulated entities will by then have invested heavily in capital-intensive equipment required by the rule. Another concern might be with varying levels of health or safety protection to different members of the public. In the absence of countervailing considerations, legal, policy, and ethical challenges such as these may mean that regulatory agencies should use randomized study methods only under limited circumstances.

If randomized study methods are either unavailable or inadvisable, agencies can use a broad range of opportunities to learn from observational studies. Sometimes these studies are called “natural experiments,” as they seek to draw inferences based on variation that naturally arises over time or across settings in the absence of randomization. For this reason, observational studies lack some of the methodological advantages of randomization. Internal validity is generally a more present concern with observational studies, as other factors may confound a study's results. In other words, other factors may also vary naturally with the intervention under study and affect the observed outcomes. An example of a potential confounding factor is when an intervention is accepted voluntarily; those individuals or entities who voluntarily choose to adopt a new practice may be different from the individuals or entities to whom a mandatory requirement would apply.

The possibility of such confounding factors should be accounted for when conducting observational studies and can be effectively addressed by using various methods that attempt to mimic statistically what occurs with randomization.^[12] Assuming the potential threats to internal validity can be addressed, observational studies may in some circumstances lead to results with stronger external validity than randomization. As a general matter, observational studies will also not raise the same legal, policy, or ethical concerns as randomization. With observational studies, the agency is either exploiting natural variation that would have arisen from the rule anyway or allowing for learning from other existing variation, such as state-by-state variation.

Opportunities for Learning From Experience Throughout the Rulemaking Lifecycle

Agencies have opportunities to learn from experience throughout the rulemaking lifecycle. For example, one stage of this cycle occurs before a rule is adopted, as agencies are focused on a problem to be addressed and are considering potential regulatory solutions. Learning from experience at this early stage can help inform an agency of how a rule should be designed. Another stage of the cycle lies with the design of the rule itself. At this stage, as an agency writes a rule, it may design it in a way that can facilitate the type of variation needed to promote learning. Finally, yet another stage arises after the agency has promulgated the rule. At this stage, agencies can consider actions, such as waivers, that can facilitate learning from experience.

Learning Before Adopting a Rule

Prior to adopting a rule, an agency should gather information using appropriate methods to help inform the regulatory action it plans to take. An agency may wish to consider randomized or observational methods.

Randomized Methods. Agencies can analyze existing peer-reviewed studies that incorporate a randomized design. They can also initiate or support new pilot programs that produce randomized study data. For example, if an agency were trying to determine whether a certain default rule related to saving for retirement should be required of all employers offering 401(k) plans, it might, if consistent with applicable law, seek the cooperation of some large employers to see whether they would assign randomly some of their employees to a company policy that requires them to opt into a retirement savings plan and other employees to a company policy that defaults employees into the plan but then allows them to opt out. Such action would be voluntary by the company but random (and effectively involuntary) by the individual. The agency might be able to learn better which default rule will yield greater savings and then use these results to inform a decision about a regulation that would apply to all companies.

Observational Methods. Agencies can also undertake observational studies prior to creating new rules. An agency might, for example, employ a cross-sectional research design by looking at variation in existing policies at the state level (or perhaps in other countries), taking to heart Justice Louis Brandeis's observation that “a . . . state may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country.” ^[13] In fact, Congress has, on numerous occasions, directed agencies to analyze state-by-state variation to help determine optimal policies.^[14]

Designing a Rule To Facilitate Learning

An agency can write a rule to facilitate future learning or to enable it later to take advantage of variation that stems naturally from the rule.^[15] Again, an agency may wish to consider randomized or observational methods.

Randomized Methods. When appropriate, an agency might consider structuring a rule to allow for learning through a randomized method.^[16] This could entail writing a rule in such a way that some entities or people that fall within the agency's regulatory scope are subject to one version of the rule and some are subject to another version of the rule or not subject to the rule at all. The agency's decision as to who falls within each category could be made on a random basis. For example, Michael Abramowicz, Ian Ayres, and Yair Listokin use as an example a test of speed limits in which the posted limits on different roads are randomly increased or decreased.^[17] Drivers on these roads are informed of the regulatory intervention (i.e., the speed limit on that road) without necessarily knowing that they are participating in a randomized experiment. Although this example falls outside the realm of federal rulemaking, agencies at the federal level may have similar ways to structure the timing or application of a rule using randomization. Assuming any potential methodological, legal, ethical, and policy concerns about randomization can be addressed, there may be some circumstances in which randomization will be an appropriate way for an agency to generate variation that will facilitate learning from experience.

Observational Methods. For the reasons discussed above, agencies will generally find it more feasible to use observational approaches than randomized ones. In any rulemaking, there will be variation from observing the world before the rule went into effect and comparing it to the world after the rule has taken effect. Further, in the case of a rule that an agency has rescinded, there will be variation in three conditions: the world before the rule went into effect; The world in which the rule was in effect; and the world after the rule was rescinded. Such variation can present rich opportunities for observational studies, especially when a satisfactory baseline or control group can be identified. Agencies may well decide, at the outset when promulgating a new rule, to commit to setting up a longitudinal study. In doing so, they would need to collect data from regulated parties before the rule goes into effect and then collect data once the rule has taken effect, keeping in mind potential confounders and using statistical techniques to control for them.^[18]

Additionally, agencies may consider deliberately introducing or allowing for some non-random variation in response to a rule by allowing for flexibility by states in the implementation of the rule. For example, variation can occur if the agency sets a federal minimum standard and permits states to exceed that standard. Agencies then can commit to using the resulting state-by-state variation to compare firms separated by a very short distance in neighboring states that have adopted different rules. Using the statistical technique known as regression discontinuity, the agency may be able to approximate randomization (i.e., the “assignment” of firms to a state with one rule versus another would be effectively random).^[19]

Learning After Promulgating a Rule

An agency can also use either randomized or observational methods to take advantage of variation once a rule has been put into place.

Randomized Methods. An agency might choose, only if appropriate, after taking into account all legal, ethical, practical, and fairness considerations, to vary the application of a rule on a randomized basis to learn from variation.^[20]

Observational Methods. In addition to varying the application of a rule on a randomized basis, agencies can achieve variation once the rule is in place by considering conditional waivers and exemptions. For example, if a regulated entity can present some evidence to suggest that it can meet the purpose of the regulation using an alternative approach, the agency might grant a waiver to that entity with the condition that the entity uses that alternative approach.^[21] After granting a certain number of waivers, the agency could then test the effectiveness of its rule by comparing entities that have selected different approaches. The agency would likely find it necessary to use statistical techniques to control for potential confounders. Over time, these kinds of studies may provide the agency with retrospective information that justifies amending an existing rule. Fairness, legal, and ethical concerns might be minimized when using conditional waivers if the agency permits all regulated entities to seek a waiver based on presentation of evidence and the agency widely publicizes its waiver availability.^[22]

Table 1 summarizes the main methods of learning discussed in the preceding sections.

Table 1—Examples of Methods for Regulatory Learning

	Randomized	Observational
Learning before adopting a rule	• Randomized voluntary pilot programs	• Pilot programs where intervention is not assigned randomly (such as with voluntary programs).
	• Studies that rely on randomization	• Analysis of regulatory approaches in different jurisdictions, including countries.
Designing a rule to facilitate learning	• Randomized assignment of different regulatory obligations	• Rules that allow for state implementation and variation (e.g., cooperative federalism).
		• Analysis of temporal differences (i.e., “before and after” comparisons).
		• Creation of regulatory thresholds that will facilitate later comparisons of entities above/below a threshold.
Learning after promulgating a rule	• Randomized application of rules in appropriate circumstances	• Granting of waivers or exemptions that allow for the adoption of alternative approaches that can be studied.

Common Issues in Learning From Experience

As noted, each stage of the rulemaking lifecycle allows agencies to learn from variation. Agencies can learn from both randomized and observational methods, keeping in mind the virtues and challenges of each. Whichever method an agency chooses, at least two additional issues should be considered: Data collection and public input.

Data Collection

Collecting data is essential. Only with information can agencies hope to learn from analyzing regulations. When collecting data, though, agencies must be mindful of the Paperwork Reduction Act (PRA), which can constrain their ability to send a survey instrument to ten or more parties.^[23] As part of agencies' data collection efforts, it may be helpful for agencies to work closely with the Office of Information and Regulatory Affairs to ensure proper use of available flexibility in accordance with the PRA and the Office of Management and Budget's implementing regulations.

Public Input

Best practices generally call for some opportunity for the public to learn about and comment on the design and results of studies an agency undertakes. For pre-rule learning, the notice-and-comment process provides the required minimum process by which agencies should engage the public, but there are other methods of public input that might be useful, even at the pre-rule stage, for public input beyond just notice and comment.^[24] If an agency is planning to revise a rule, a subsequent notice-and-comment rulemaking will provide an additional opportunity for public input. If an initial rule provides for its expiration on a certain date, that may also help ensure that the public has the opportunity to offer input on a future notice-and-comment rulemaking to keep or modify the rule. Even rules not subject to notice-and-comment procedures can benefit from subsequent opportunities for public comment.^[25]

But even in situations in which the agency does not undertake a new notice-and-comment rulemaking or otherwise leaves a rule “as is,” the agency may benefit from outside input on the systematic learning effort it has undertaken, whether through a peer review process, advisory committees, public hearings or meetings, or just a supplemental solicitation of comments. The decision as to which approach to use to solicit public input will turn on numerous factors, including resource constraints.^[26]

Recommendation

1. Agencies should seek opportunities to collect data to learn the most effective way to design their rules and analyze the effects of their rules. They can learn from experience at one or more stages of the rulemaking process, from pre-rule analysis to retrospective review. Before adopting a rule, agencies can learn from pilot projects, demonstrations, and flexibility among states or regulated entities. After promulgating a rule, agencies may, where legally permissible, use waivers and exemptions to learn. As agencies seek out such learning opportunities, they should give due regard for legal, ethical, practical, and fairness considerations.

2. When agencies analyze variation to learn more about the effectiveness of policy options, they should make every effort to collect data and conduct reliable analysis. Only where appropriate, agencies should consider creating variation through a randomized control trial.

3. To inform the learning process, agencies should consider soliciting public input at various points in the rulemaking lifecycle. This can include input on the design and results of any learning process. In addition to the public input required under 5 U.S.C. 553(c), agencies should consider, as time and resources permit, the use of supplemental requests for public comment, peer review, advisory committee deliberation, or public hearings or meetings.

4. When gathering data, agencies and the Office of Management and Budget (OMB) should seek to use flexibilities within the Paperwork Reduction Act and OMB's implementing regulations (e.g., a streamlined comment period for collections associated with proposed rules) when permissible and appropriate.

5. Agencies, as appropriate, should seek legal authority from Congress to take advantage of this recommendation.

Administrative Conference Recommendation 2017-7

Regulatory Waivers and Exemptions

Adopted December 15, 2017

Individuals and entities regulated by federal agencies must adhere to program-specific requirements prescribed by statute or regulation. Sometimes, however, agencies prospectively excuse individuals or entities from statutory or regulatory requirements through waivers or exemptions.^[1] The authority to waive or exempt regulated parties from specific legal requirements affords agencies much-needed flexibility to respond to situations in which generally applicable laws are a poor fit for a given situation.^[2] Emergencies or other unforeseen circumstances may also render compliance with statutory or regulatory requirements impossible or impracticable.^[3] In such instances, requiring strict adherence to legal requirements may not be desirable.^[4] This is particularly true when the recipient of a waiver or exemption demonstrates that it intends to engage in conduct that will otherwise further the agency's legitimate goals.

Yet, waiving or exempting a regulated party from a statutory or regulatory requirement also raises important questions about predictability, fairness, and protection of the public. For instance, when an agency decides to waive legal requirements for some but not all regulated parties, the decision to grant a waiver or exemption may create the appearance—or perhaps even reality—of irregularity, bias, or unfairness. Waiving or exempting a regulated party from a legal requirement, therefore, demands that agencies simultaneously consider regulatory flexibility, on the one hand, and consistent, non-arbitrary administration of the law, on the other.

Agencies' authority to waive or exempt regulated parties from legal requirements may also intersect with other principles of administrative law. When agencies frequently issue waivers or exemptions because a regulation is outdated or ineffective, for example, amending or rescinding the regulation may be more appropriate in some circumstances, despite the necessary resource costs.^[5] Such revisions can enhance efficiency and transparency. The requisite notice-and-comment procedures can also foster public participation and informed decisionmaking.

The following recommendations offer best practices and factors for agencies to consider regarding their waiver and exemption practices and procedures. They are not intended to disturb or otherwise limit agencies' broad discretion to elect how to best use their limited resources.

Recommendation

Scope of Waiver and Exemption Authority

1. When permitted by law, agencies should consider creating mechanisms that would allow regulated parties to apply for waivers or exemptions by demonstrating conduct that will achieve the same purpose as full compliance with the relevant statutory or regulatory requirement.

2. When consistent with the statutory scheme, agencies should endeavor to draft regulations so that waivers and exemptions will not be routinely necessary. When an agency has approved a large number of similar waivers or exemptions, the agency should consider revising the regulation accordingly. If eliminating the need for waivers or exemptions requires statutory reform, Congress should consider appropriate legislation.

Exercising Waiver or Exemption Authority

3. Agencies should endeavor, to the extent practicable, to establish standards and procedures for seeking and approving waivers and exemptions.

4. Agencies should apply the same treatment to similarly situated parties when approving waivers and exemptions, absent extenuating circumstances.

5. Agencies should clearly announce the duration, even if indefinite, over which a waiver or exemption extends.

Transparency and Public Input in Seeking and Approving Waivers and Exemptions

6. Agencies should consider soliciting public comments before establishing standards and procedures for seeking and approving waivers and exemptions.

7. Agencies should endeavor, to the extent practicable, to make standards and procedures for seeking and approving waivers and exemptions available to the public.

8. Agencies should consider soliciting public comments before approving waivers or exemptions.

9. Agencies should provide written explanations for individual waiver or exemption decisions and make them publicly available to the extent practicable and consistent with legal or policy concerns, such as privacy. Further, agencies should consider providing written explanations of representative instances to help illustrate the types of activities likely to qualify for a waiver or exemption.

Footnotes

1.  These terms carry the same meaning and are used interchangeably here.

2.  Public Law 111-274, 124 Stat. 2861 (2010) (codified at 5 U.S.C. 301 note).

3.  Exec. Order No. 13,563, 76 FR 3821 (Jan. 18, 2011).

4.  Executive guidance issued prior to the PWA's enactment also directs agencies to use plain language. Executive Order 12,866 provides that “[e]ach agency shall draft its regulations to be simple and easy to understand.” Exec. Order No. 12,866 § 2(b), 58 FR 51,735, 51,737 (Oct. 4, 1993). President Clinton's 1998 Plain Language Memorandum further requires agencies to “use plain language in all new documents, other than regulations, that explain how to obtain a benefit or service, or how to comply with a requirement [the agency] administer[s] or enforce[s],” as well as “all proposed and final rulemaking documents published in the Federal Register.” Memorandum on Plain Language in Government Writing, 63 FR 31,885 (June 10, 1998).

5.  5 U.S.C. 301 note sec. 3(3).

6.  PLAIN grew out of early, informal agency efforts to share plain writing tools and techniques, and has served as a hub for such resources since its establishment during the Clinton Administration. About Us, Plain Language Action & Information Network, https://plainlanguage.gov/​about/​.

7.  What is Plain Language?, Plain Language Action & Information Network, https://plainlanguage.gov/​about/​definitions/​.

8.  For guidance on writing plainly without compromising nuance or avoiding important technical terms, consult the Federal Plain Language Guidelines, a resource compiled by PLAIN, which both the PWA and executive guidance direct agencies to use. Plain Language Action & Information Network, Federal Plain Language Guidelines (Rev. ed. May 2011), http://www.plainlanguage.gov/​guidelines/​.

9.  5 U.S.C. 301 note sec. 3(2)(A).

10.  Exec. Order No. 13,563 § 1(a), 76 FR 3821, 3821 (Jan. 18, 2011).

11.  Office of Mgmt. & Budget, Exec. Office of the President, OMB Mem. M-11-15, Final Guidance on Implementing the Plain Writing Act of 2010 5 (2011).

12.  See United States v. Mead Corp., 533 U.S. 218, 226-27 (2001).

13.  5 U.S.C. 553(c).

14.  Id. § 301 note sec. 4(a).

15.  Id. § 301 note sec. 6.

16.  See Cynthia Farina, Mary J. Newhart, & Cheryl Blake, The Problem with Words: Plain Language and Public Participation in Rulemaking, 83 Geo. Wash. L. Rev. 1358, 1367-79 (2015).

17.  Blake Emerson & Cheryl Blake, Plain Language in Regulatory Drafting 33 (Dec. 8, 2017) (report to the Admin. Conf. of the U.S.), https://www.acus.gov/​report/​plain-language-regulatory-drafting-final-report.

18.  5 U.S.C § 301 note secs. 4(a)(1)(A), 4(a)(1)(C).

19.  Id. § 301 note sec. 4(a).

20.  Emerson & Blake, supra note 17, at 32-33.

21.  Some envision rulemaking and guidance documents as situated along a “continuum” ranging from more “complicated” documents like the rule itself to simpler documents that digest the material for non-specialist audiences. Complicated documents can be written plainly, but may require greater resource investment.

22.  Joshua D. Blank & Leigh Osofsky, Simplexity: Plain Language and the Tax Law, 66 Emory L.J. 189, 193 (2017).

23.  For a closer examination of guidance practices, see Nicholas R. Parrillo, Federal Agency Guidance: An Institutional Perspective (Dec. 1, 2017) (report to the Admin. Conf. of the U.S.), https://www.acus.gov/​report/​agency-guidance-final-report.

1.  See Jason Schwartz, Marketable Permits: Recommendations on Application and Management i (Dec. 11, 2017) (report to the Admin. Conf. of the U.S.), https://www.acus.gov/​report/​marketable-permits-final-report.

2.  In 2015, the Administrative Conference issued recommendations on the design and tailoring of regulatory permits generally, which are defined as “any administrative agency's statutorily authorized, discretionary, judicially reviewable granting of permission to do something which would otherwise be statutorily prohibited.” Admin. Conf. of the U.S., Recommendation 2015-4, Designing Federal Permitting Programs, 80 FR 78,164 (Dec. 16, 2015).

3.  Many of the examples in this Recommendation are drawn from marketable permitting programs in the environmental context because a significant amount of the experience and writing to date regarding marketable permitting programs stems from the environmental area. This is not meant to imply that marketable permits are not suitable in other contexts, nor that they are always useful in environmental contexts.

4.  For example, as with sulfur dioxide emissions from the Midwest which affect the East Coast and emissions from the East Coast which mostly blow out to sea.

5.  See, e.g., Exec. Order No. 12,898, § 1-101, 59 FR 7629, 7629 (Feb. 16, 1994) (requiring each federal agency to “identif[y] and addres[s], as appropriate, disproportionately high and adverse human health or environmental effects of its programs, policies, and activities on minority populations and low-income populations”); see also Clean Air Act, 42 U.S.C. 7491(a)(1) (2016) (noting with respect to “Class I” areas (primarily national parks) that “Congress hereby declares as a national goal the prevention of any future, and the remedying of any existing, impairment of visibility in mandatory class I Federal areas which impairment results from manmade air pollution.”).

6.  Schwartz, supra note 1, at 27.

7.  Id.

8.  Chrysler Corp. v. Brown, 441 U.S. 281, 301-02 (1979).

9.  Schwartz, supra note 1, at 27-28.

10.  The Administrative Conference has long advised use of notice-and-comment even when it is not legally required. See, e.g., Admin. Conf. of the U.S., Recommendation 2012-2, Midnight Rules, 77 FR 47,801 (Aug. 10, 2012); Admin. Conf. of the U.S., Recommendation 92-1, The Procedural and Practice Rule Exemption from the APA Notice-and-Comment Rulemaking Requirements, 57 FR 30,101 (July 8, 1992); Admin. Conf. of the U.S., Recommendation 82-2, Resolving Disputes Under Federal Grant Programs, 47 FR 30,701 (July 15, 1982).

11.  See Interagency Working Grp. for the Study on Oversight of Carbon Mkts., Report on the Oversight of Existing and Prospective Carbon Markets Carbon Study 12 (2011) (describing the primary market as the entry point for permits, whether entry occurs as a result of the government distributing permits directly to market participants, auctioning permits, or some combination of the two).

12.  T.H. Tietenberg, Emissions Trading: Principles and Practice 138-39 (2d ed. 2006).

13.  Peter Cramton & Jesse Schwartz, Collusive Bidding: Lessons from the FCC Spectrum Auctions, 17 J. Reg. Econ. 229 (2000).

14.  Often proposed in marketable permitting programs that regulate electricity generators, output-based allocation distributes permits for pollution based on the amount of electricity produced by a given party, as opposed to the historical amount of pollution that party generated. This results in awarding permits to some of the cleanest producers of electricity, like renewable energy, rather than disproportionately to the most heavily polluting producers. Project on Alt. Regulation, Marketable Rights: A Practical Guide to the Use of Marketable Rights as a Regulatory Alternative 14 (1981).

15.  For instance, tradable fish catch shares are sometimes allocated directly to native communities to enable them to protect their interests.

16.  For example, airlines in possession of valuable landing slots have an incentive to retain the slots for possible future ridership, rather than deciding to sell the slots to a potential new competitor.

17.  In some marketable permitting programs, monitoring has been accomplished by spot checking only a small percentage of permit holders. On the other end of the spectrum, some programs require extensive measures such as third-party audits of all permits or credits annually or every few years.

18.  An example of a program that has achieved near perfect compliance is the acid rain market. It features a sophisticated monitoring system that tracks pollution allowance holdings and compares them at the end of the compliance period to total emissions registered in an emissions monitoring system. It also includes stiff penalties fixed to inflation per excess ton of pollutant discharged and imposes a requirement to submit a plan for how excess emissions will be offset in future years. Schwartz, supra note 1, at 65.

19.  For example, in many fishery and catch share programs, fishers are reportedly more cooperative with enforcement officials after the introduction of a marketable permitting program, recognizing that illegal fishing reduces the value of their quota. Tom Tietenberg, Tradable Permits in Principle and Practice, 14 Penn. St. Envtl. L. Rev. 251, 260 (2006).

20.  Derivatives are contracts or instruments based on the value of another financial or economic interest or property and are used for hedging and speculation. A derivative of a marketable permit would be a contract or instrument based on the value of the permit. Hedging allows the transfer of market risks to parties more capable of assuming it. Speculation involves attempting to earn profit by anticipating price movements or taking advantage of a perceived mispricing. Commonly traded types of derivative contracts include futures, options, and swaps.

21.  Interagency Working Grp. for the Study on Oversight of Carbon Mkts., supra note 11, at 14.

22.  See id. at 43 (“Because the CFTC has broad enforcement authority to pursue manipulation of a commodity's price in interstate commerce, the agency would have the authority to bring actions against individuals or entities believed to be involved in the price manipulation of allowance and carbon offsets.”).

23.  For example, the CFTC oversees trading of permits for the Regional Greenhouse Gas Initiative and the acid rain market on exchanges like the Chicago Climate Futures Exchange.

24.  Interagency Working Grp. for the Study on Oversight of Carbon Mkts., supra note 11, at 44, 51. The Securities and Exchange Commission has authority over securities and securities based swaps.

25.  See Dodd-Frank Wall Street Reform and Consumer Protection Act, Public Law 111-203, 124 Stat. 1376 (2010). Certain activities involving derivatives may be exempt from CFTC oversight, but CFTC has the statutory authority to eliminate many of those exemptions and to provide comprehensive oversight of derivatives in permit markets. Schwartz, supra note 1, at 76.

26.  Exec. Order No. 12,866, 58 FR 51,735 (Oct. 4, 1993). Other examples of regulatory tools drawing on economic incentives include fees, penalties, subsidies, changes in liability rules or property rights, required bonds, insurance, and warranties. Office of Mgmt. & Budget, Exec. Office of the President, OMB Circular A-4, Regulatory Analysis (2003).

1.  Attorney General's Manual on the Administrative Procedure Act 30 n.3 (1947).

2.  Id.

3.  5 U.S.C. 553(b)(A).

4.  Admin. Conf. of the U.S., Recommendation 76-5, Interpretive Rules of General Applicability and Statements of General Policy, 41 FR 56,769 (Dec. 30, 1976). Additional prior Conference recommendations pertaining to policy statements and agency guidance more broadly, apart from others referenced specifically in this preamble, include Recommendation 2015-3, Declaratory Orders, 80 FR 78,163 (Dec. 16, 2015); and Recommendation 2014-3, Guidance in the Rulemaking Process, 79 FR 35,992 (June 25, 2014).

5.  Admin. Conf. of the U.S., Recommendation 92-2, Agency Policy Statements, 57 FR 30,103 (July 8, 1992).

6.  The Conference commissioned a study that resulted in interviews with 135 individuals across agencies, industry, and non-governmental organizations (NGOs), which are the basis for this Recommendation. See Nicholas R. Parrillo, Federal Agency Guidance: An Institutional Perspective (Oct. 12, 2017) (report to the Admin. Conf. of the U.S.), https://www.acus.gov/​report/​agency-guidance-final-report.

7.  See id. at 28-30; see also Admin. Conf. of the U.S., Recommendation 71-3, Articulation of Agency Policies, 38 FR 19,788 (July 23, 1973) (“Agency policies which affect the public should be articulated and made known to the public to the greatest extent feasible. To this end, each agency which takes actions affecting substantial public or private interests, whether after hearing or through informal action, should, as far as is feasible in the circumstances, state the standards that will guide its determination in various types of agency action, either through published decisions, general rules or policy statements other than rules.”).

8.  See Recommendation 92-2, supra note 5; Office of Mgmt. & Budget, Exec. Office of the President, Final Bulletin for Agency Good Guidance Practices, 72 FR 3432, 3436 (Jan. 25, 2007) (“[A]gency employees should not depart from significant agency guidance documents without appropriate justification and supervisory concurrence.”); id. at 3437 (“[W]hile a guidance document cannot legally bind, agencies can appropriately bind their employees to abide by agency policy as a matter of their supervisory powers over such employees without undertaking pre-adoption notice and comment rulemaking.”).

9.  See Final Bulletin for Agency Good Guidance Practices, supra note 8, 72 FR at 3440.

10.  Another difficulty with giving reasons is a potential tension with agency policies on the protection of confidential business or personal information. This Recommendation is not intended to alter existing agency policies on such protection.

11.  See, e.g., About Guidance Documents, U.S. Food & Drug Admin., https://www.fda.gov/​RegulatoryInformation/​Guidances/​default.htm#about (“Guidance documents represent FDA's current thinking on a topic. They do not create or confer any rights for or on any person and do not operate to bind FDA or the public. You can use an alternative approach if the approach satisfies the requirements of the applicable statutes and regulations.”).

12.  Some agencies have adopted procedural rules requiring solicitation of written input from the public for large and well-defined categories of their policy statements, whereas others have undertaken such solicitations on a decentralized, ad hoc basis. Parrillo, supra note 6, at 167-68.

13.  The Office of Management and Budget's Good Guidance Practices calls for pre-adoption public comment on “economically significant” guidance documents, but this appears to cover only a very small number of documents. See id. at 167-71 (citing Final Bulletin for Agency Good Guidance Practices, supra note 8, 72 FR at 3439-40).

1.  See, e.g., Prohibition on Improper Guidance Documents, (DOJ, Nov. 16, 2017), https://www.justice.gov/​opa/​press-release/​file/​1012271/​download;​; Final Bulletin for Agency Good Guidance Practices, 72 FR 3432 (OMB, Jan. 25, 2007); FDA Good Guidance Practices, 21 CFR 10.115 (2017) (issued Sept. 19, 2000).

1.  See, e.g., Admin. Conf. of the U.S., Recommendation 2014-5, Retrospective Review of Agency Rules, 79 FR 75,114 (Dec. 17, 2014); Admin. Conf. of the U.S., Recommendation 85-2, Agency Procedures for Performing Regulatory Analysis of Rules, 50 FR 28,364 (July 12, 1985); Admin. Conf. of the U.S., Recommendation 79-4, Public Disclosure Concerning the Use of Cost-Benefit and Similar Analyses in Regulation, 44 FR 38,826 (June 8, 1979).

2.  See, e.g., Data Quality Act, Public Law 106-554, 515, 114 Stat. 2763A-153 (2001).

3.  See, e.g., Exec. Order No. 12,866, § 5, 58 FR 51,735, 51,739 (Oct. 4, 1993) (“[T]o . . . improve the effectiveness of existing regulations . . . each . . . agency will periodically review its existing significant regulations to determine whether any such regulations should be modified or eliminated so as to make the agency's regulatory program more effective in achieving the regulatory objectives.”); Exec. Order No. 13,563, § 6, 58 FR 3821, 3822 (Jan. 21, 2011) (requiring agencies to “consider how best to promote retrospective analysis of rules that may be outmoded, ineffective, insufficient, or excessively burdensome, and to modify, streamline, expand, or repeal them in accordance with what has been learned”); Exec. Order No. 13,771, § 2, 82 FR 9339 (Feb. 3, 2017) (requiring the repeal of two existing regulations for each new regulation proposed, and leaving in place prior analytical requirements); Exec. Order No. 13,777, § 3, 82 FR 12,285, 12,286 (Mar. 1, 2017) (requiring the establishment of regulatory reform task forces that “shall evaluate existing regulations . . . and make recommendations to the agency head regarding their repeal, replacement, or modification, consistent with applicable law”).

4.  See, e.g., Motor Vehicle Mfrs. Ass'n v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 43, 52 (1983) (explaining that the agency must show that its action was the result of “reasoned decisionmaking” consistent with “the evidence before the agency”).

5.  A general discussion of factors to consider in choosing methods and measurements in regulatory learning can be found in Cary Coglianese, Measuring Regulatory Excellence, in Achieving Regulatory Excellence 291-305 (Cary Coglianese ed., 2017) [hereinafter Coglianese, Measuring Regulatory Excellence].

6.  Cross-sectional analysis means analysis of data collected across at least two groups or jurisdictions, with one that is subject to the intervention (such as regulation) and one that is not. See Cary Coglianese, Empirical Analysis and Administrative Law, 2002 U. Ill. L. Rev. 1111, 1117-19.

7.  Longitudinal analysis is a research design that involves repeated observations of the same subjects over a period, where variation in the intervention occurs over time (i.e., data before and after an intervention is introduced). See Cary Coglianese, Measuring Regulatory Performance: Evaluating the Impact of Regulation and Regulatory Policy, Organization for Econ. Co-Operation and Dev. [OECD] Expert Paper No. 1 39 (Aug. 2012) [hereinafter Coglianese, Measuring Regulatory Performance].

8.  In this context, “confounders” refer to changes in outcomes that may appear to have been caused by the regulation but are actually caused by other factors. See Coglianese, Measuring Regulatory Performance, supra note 7 and accompanying text.

9.  “Blindness” in this context means subjects are not aware of whether they are in the treatment or comparison group. “Double blindness” means neither the subjects nor the researchers know which subjects received the treatment, and which received the placebo. See Michael Abramowicz et al., Randomizing Law, 159 U. Pa. L. Rev. 929, 948-50 (2011).

10.  See 5 U.S.C. 706(2)(A).

11.  See Abramowicz et al., supra note 9, at 968.

12.  Examples of such statistical methods include: difference-in-differences, propensity score matching, instrumental variables, and regression discontinuity. See Coglianese, Measuring Regulatory Performance, supra note 7, at 39-42.

13.  New State Ice Co. v. Liebmann, 285 U.S. 262, 311 (1932) (Brandeis, J., dissenting).

14.  See, e.g., Energy Policy Act of 2005, Public Law 109-58, 139, 119 Stat. 594, 647 (2005) (“[T]he Secretary . . . shall conduct a study of State and regional policies that promote cost-effective programs to reduce energy consumption (including energy efficiency programs) that are carried out by utilities that are subject to State regulation.”).

15.  These features can facilitate retrospective review. See Admin. Conf. of the U.S., Recommendation 2014-5, Retrospective Review of Agency Rules, 79 FR 75,114 (Dec. 17, 2014).

16.  See generally Abramowicz et al., supra note 9.

17.  See id. at 951.

18.  See Admin. Conf. of the U.S., Recommendation 2014-5, ¶ 7, Retrospective Review of Agency Rules, 79 FR 75,114, 75,116-17 (Dec. 17, 2014).

19.  See Jonah B. Gelbach & Jonathan Klick, Empirical Law and Economics, in The Oxford Handbook of Law and Economics (Francisco Parisi ed., 2017).

20.  In 2004, the Securities and Exchange Commission (SEC) varied the application of its “Uptick Rule.” See Order Suspending the Operation of Short Sale Price Provisions for Designated Securities and Time Periods, Exchange Act Release No. 50,104, 69 FR 48,032 (Aug. 6, 2004). Market observers characterized the SEC's conclusion to be that the rule did not substantially increase market efficiency. The SEC rescinded the rule. See Zachary Gubler, Regulatory Experimentation 42 (Nov. 17, 2017) (report to the Admin. Conf. of the U.S.), https://www.acus.gov/​report/​regulatory-experimentation-final-report.

21.  See Admin. Conf. of the U.S., Recommendation 2017-7, Regulatory Waivers and Exemptions, 82 FR ___(approved Dec. 15, 2017); see also Aaron Nielson, Waivers, Exemptions, and Prosecutorial Discretion: An Examination of Agency Non-Enforcement Practices 30 (Nov. 1, 2017) (report to the Admin. Conf. of the U.S.), https://www.acus.gov/​report/​regulatory-waivers-and-exemptions-final-report.

22.  See Admin. Conf. of the U.S., Recommendation 2017-7, Regulatory Waivers and Exemptions, 82 FR ___(approved Dec. 15, 2017).

23.  See 44 U.S.C. 3502(3)(A)(i).

24.  See, e.g., Admin. Conf. of the U.S., Recommendation 2017-2, Negotiated Rulemaking and Other Options for Public Engagement, 82 FR 31,039 (July 5, 2017); Admin. Conf. of the U.S., Recommendation 2013-5, Social Media in Rulemaking, 78 FR 76,269 (Dec. 17, 2013).

25.  Admin. Conf. of the U.S., Recommendation 95-4, Procedures for Noncontroversial and Expedited Rulemaking, 60 FR 43,110 (Nov. 8, 1995).

26.  See Gubler, supra note 20, at 54.

1.  Agencies may also retrospectively decline to bring an enforcement action once a legal violation has already occurred. This recommendation, however, is confined to the agency practice of prospectively waiving or exempting regulated parties from legal requirements.

2.  The terms “waiver” and “exemption” carry various meanings in agency practice. For the purposes of this recommendation, when Congress has expressly authorized an agency to excuse a regulated party from a legal requirement, the term “waiver” is used. If an agency is implicitly authorized by Congress to excuse a regulated party from a legal requirement, “exemption” is used. These definitions stem from the report underlying this recommendation. See Aaron L. Nielson, Waivers, Exemptions, and Prosecutorial Discretion: An Examination of Agency Nonenforcement Practices (Nov. 1, 2017) (report to the Admin. Conf. of the U.S.), https://acus.gov/​report/​regulatory-waivers-and-exemptions-final-report. Some agencies may also derive authority to grant waivers or exemptions from presidential delegations under Article II of the Constitution. That category of waivers and exemptions is outside the scope of this recommendation.

3.  See, for example, the Stafford Act, 42 U.S.C. 5141, authorizing any federal agency charged with the administration of a federal assistance program in a presidentially declared major disaster to modify or waive administrative conditions for assistance if requested to do so by state or local authorities.

4.  Of course, agencies cannot issue waivers or exemptions unless authorized by law, and even when authorized by law, agencies must not issue them in an arbitrary fashion.

5.  See Admin. Conf. of the U.S., Recommendation 2014-5, Retrospective Review of Agency Rules, ¶ 5, 79 FR 75,114, 75,116 (Dec. 17, 2014) (identifying petitions from stakeholder groups and members of the public and poor compliance rates as factors to consider in identifying regulations that may benefit from amendment or rescission).