Revisiting the Nature of Regulation

Regulations are the product of negotiations and agreements.

It is widely recognized that regulation involves more than just coercion and punishment and that regulators rely on a variety of soft, flexible tools to serve the public interest, including agreements and understandings with regulated entities.

In a forthcoming article, I argue that regulator-regulatee agreements are not simply another tool in the regulatory toolkit. Instead, they constitute the dominant paradigm of regulation—its very core principle.

When we closely examine various sectors, industries, and regulated areas—such as climate change, artificial intelligence (AI), public health and safety, and gun control—examples of agreement-based regulation are plentiful. Taken together, they form a clear picture in which regulation is often based on discussion, collaboration, consensus, and agreement with regulated entities. These understandings and dialogues occur throughout all phases of regulation: They can happen before, during, or after rulemaking, supervision, and enforcement actions.

Indeed, agency agreements with regulated entities are far more common and diverse than we might think. They are used across old and new industries and large and small companies, and in social, economic, and environmental fields. For instance, consider car safety regulation. There are several laws, rules, and regulations that govern the field to prevent car accidents and reduce bodily harm. They all seem to follow a command-and-control regulatory style, relying on legal restrictions, binding technological requirements, and civil and criminal sanctions. A closer look, however, reveals that this set of laws and rules is actually based on prior agreements between the regulator––the National Highway Traffic Safety Administration—and the regulated entities, represented by groups such as the Alliance of Automobile Manufacturers and the Association of International Automobile Manufacturers Association. ‎The agreements cover most of the automobile industry, including carmakers such as BMW, Ford, and Volkswagen.

Given the widespread nature of the practice, agreement-based regulation can provide a new, unified framework for understanding how regulatory norms are shaped, applied, and enforced. Importantly, it directly challenges the conventional distinction between regulatory rule types and tools, such as strict, detailed rules versus open-ended standards; performance versus design standards; and hard command-and-control rules versus softer approaches such as self-regulation, disclosure rules, voluntary programs, shaming, and sandboxes. Since all forms of regulation are ultimately negotiated, distinguishing between them based on perceived rigidity or flexibility of specific types of rules or regulatory instruments would be theoretically misleading and of limited analytical and practical value.

The negotiation process inherent in every regulation renders the hard–soft divide, so entrenched in regulatory tools scholarship, much less relevant, because agreements accompany both “soft” and “hard” regulatory tools—as shown in the car safety regulation example—making them all inherently soft.

An example of a soft regulatory instrument based on agreements can be found in the field of AI. A few years ago, the Biden Administration negotiated a series of behavioral commitments that leading AI companies, including Google and OpenAI, agreed to apply to themselves as a form of self-regulation—a method that was itself accepted by the parties as an appropriate regulatory tool for this context. Conversely, self-regulation systems, especially enforced self-regulation, are usually understood as initiated, imposed, and designed by the regulator, who sets the system’s main principles, core values, and goals.

Moreover, the agreement-based theory of regulation could significantly affect instrument choice and rule-design policies. Typically, the conventional approach in regulatory policy treats different rule types and regulatory tools as distinct in mechanism, flexibility, social and economic justification, legitimacy, and effectiveness. In fact, the entire cost-benefit analysis process relies on regulators evaluating and comparing different applicable tools with varying levels of stringency. Furthermore, regulatory oversight bodies advise, and sometimes legislation requires, regulators to favor flexible tools over more coercive ones. Several presidential executive orders similarly encourage regulators to prefer disclosure rules and performance standards over prescriptive and rigid regulation when possible. Viewing all regulation as negotiable and based on agreements with regulatees, however, suggests that selecting a specific mechanism, tool, or legal framework is not that consequential.

For example, as part of its Cambridge Analytica settlements related to violations of Facebook users’ privacy, Meta agreed to a series of commitments, including conducting a privacy review of every new product and service, implementing a comprehensive data security program, and establishing an independent privacy committee of its board of directors.

Notably, the regulatory and legal frameworks in place before the settlement were mostly command-and-control-style rules issued under the Federal Trade Commission Act to prevent ‎unfair or deceptive practices in commerce. ‎The settlement, however, essentially rewrote this legal framework, which had merely served as the backdrop for developing and renegotiating privacy regulation at the largest social media platform. In this case, and many others, command-and-control functions as no more than a shell or a front for agreement-based regulatory actions.

The same principle applies to emerging fields such as AI as well. For example, scholars and policymakers have considered different regulatory systems, such as disclosure regulation; command-and-control with detailed rules, prohibitions, permits, and fines; and self-regulation to regulate AI platforms and companies. Given the theory of regulation through agreements, which suggests that the content of all regulatory styles is negotiable in practice, these debates may not be as critical as previously assumed‎.

Indeed, finding the right tool or rule type to address a regulatory problem has always been a key challenge in regulatory policy. Numerous books have discussed this subject, closely analyzing the benefits and drawbacks of traditional tools versus innovative methods, as well as the best mix of approaches. But if we view negotiations, consensus, and agreements as the building blocks of all forms of regulation, we might need to reevaluate these long-standing assumptions and paradigms to better serve the public interest.

To be sure, the theory of agreement-based regulation does not imply that all regulation is always ‎subject to negotiation with every regulated entity. Rather, it refers to regulation being ‎negotiable in certain segments, at certain times, including in situations that might not initially seem negotiable. Moreover, the type of rule does matter in some ways. For example, it may be relevant to regulatory reputation as some equate hard regulatory tools with a firm regulator.

The overall perspective, however, is that regulation and negotiation are not competing, complementary, or merely related concepts. A more accurate view is that regulation is based on negotiation and that agreements are an inherent, central part of regulatory life.