Introduction

If regulation is often perceived as a tool to increase welfare, it is not infrequent that it leads to unintended consequences; examples are numerous. After all, regulating in an evolving world of inherent complexity (and interdependence between economies) and uncertainty is not an easy task; it is the reason why it is bound to be a never-ending process. Nevertheless, it is not because perfect regulations are not from this world that it is not possible to develop tools to reduce the amount of inefficient regulations, especially when these regulations could be expected to impact many domains of a national or a regional economy. Regulatory impact assessments (RIA) are precisely trying to contribute to this aim by requiring regulators to assess the probable consequences of their regulation for the economy using models, data, and evidence. These RIA are also supposed to help political decision-makers to identify whether and how to regulate while also defining a conception of the public interest that could then be debated. It is not then surprising that the practice of RIA is highly recommended by the OECD (1997) and part of its indicators to measure regulatory performance (see Radaelli 2004, for an explanation of the diffusion of RIA).

From the point of view of law and economics, RIA is both the recognition of its practical relevance and its inherent limitation for regulators and policy makers. Moreover, it helps to understand the real value of law and economics.

The Economic Logic Behind Regulatory Impact Assessment

In traditional law and economics, regulations are required in two situations: when the regulation can reduce transaction costs (defining the basic entitlements and the way to exchange them) or when it is supposed to reduce market failures (externalities, asymmetrical information or market power) (see Baldwin et al. (2010)). Nevertheless, as public choice emphasized, unproductive profit seeking by special interest groups to secure favorable regulation is definitional of the regulatory process (Tullock 1967), so there are no reasons why rational policy makers should implement only public interest regulation (Stigler 1971; Posner 1974; Peltzman 1976) without external constraints (public opinion’s pressure or procedural requirement). Hence, regulations could merely be designed to redistribute wealth, and, when it does, these regulations should be considered as highly suspicious. This does not mean that, from an economic point of view, it is not possible to engage in policy choices but that these choices should be made sufficiently clear in their consequences to avoid the lurk of blindfold populism and moralism, an unavoidable risk in any democracy.

The mere need of RIA for enacting “better and smarter” regulations cannot be explained if it is possible to consider that regulations are only driven by the public interest and efficiency requirement. RIA is thus a tool to overcome some regulatory failures. It is not a tool to make sure that regulations are efficient but a tool to ensure that they are not blatantly inefficient because it is forcing policy makers and regulators to justify their policy not only through a political rhetoric but through an explanation of the trade-offs they are engaging in and the requirement to quantify and identify the public interest based on the best data available (of course the quality of the data will be crucial, Torriti 2011).

Of course, since RIA are costly (administrative cost and delay are necessarily creating such costs), their necessity and precision are dependent on the regulation at stake (Rose-Ackerman 2016). It is only because RIA are supposed to lead to some social benefits (avoidance of bad measures, welfare gain from enacting what appeared to be efficient) that its costs could be justified. It is then to be anticipated that RIA requirement should be higher when the regulation at stake, either by its breadth or magnitude, is significantly impacting the economy. The fact that this requirement appeared in the field of environment and health should not then be a surprise (see also Arrow et al. 1996; Livermore and Revesz 2013).

Regulatory Impact Assessment as a Process

A. Three stages of RIA could be distinguished: description, identification, and quantification (see also European Commission 2009; for an overview of the methods and domains, see Dunlop and Radaelli 2016).

At first a RIA should explain what is the problem (identification and definition) the regulation is supposed to curb and why an action is required. At this stage, this is the need for an intervention that should be assessed. Using economic theory, this need should be either a market failure or a means to reduce transaction costs.

The second stage is inquiring into the regulatory options. Indeed, a same problem could be addressed using a variety of regulatory tools from command and control to economic incentives to mandatory disclosure or nudges. It is not only the “specie” of regulatory option that should be considered but also the variety within the “specie.”

The third stage is dedicated to the quantification of the direct and indirect benefits and costs of each option (including the administrative and information regulatory burden; see also, more generally, Marneffe and Vereeck 2011; Hahn and Tetlock 2007). Of course, the benefits and costs can only be assessed through a set of hypotheses or choices regarding, more especially, the valuation methodology and baseline scenario. These hypotheses are crucial to assess the subjective value of the RIA for policy making (see, e.g., Ogus 1998; Rose-Ackerman 2011a; Hahn and Litan 2005). It will then be possible to compare options and eliminate obviously relatively inefficient options.

Being transparent about all these stages could contribute to a better accountability of regulators (since trade-offs are made clear) and to an increased consistency in regulations (since the need of a new regulation is explained). Moreover, with experience in RIA, better data and analytical methods should be expected to emerge. RIA is, from this point of view, a learning by doing a process.

B. RIA is not only a document (or a set of documents) providing information regarding the expected consequences of a regulation (an end result); it should be conceptualized much more as a process (Adler and Posner 2006). Indeed, as an “end result,” it could be considered as a mere new rhetoric to justify regulatory choices; the hypotheses would then be chosen ex post to justify the desired regulation which will then be submitted to some formal or informal political debate – it is quite often this instrumental use of RIA that is criticized (Renda 2006). Conceptualized as a process, RIA exhibits two interesting features. First, it makes clear that RIA is a way to inquire into a regulatory problem: neither the problem nor its options are facts; they are identified and assessed differently depending on the state of knowledge. RIA, once made, is then not an end result but a “first draft” that will evolve with the comments and criticism of stakeholders. Second, RIA is not a stage after drafting a regulation; it should be developed in parallel of it to reduce the likelihood of an instrumental use. This also requires a complete transparency of the process.

The Rise of Regulatory Impact Assessment

The rise of RIA began in 1974, in the USA. President Ford’s executive order 11821 established procedures for preparing inflation impact statements that were designed to illuminate the economic impact of regulatory proposals, specifically their effects on productivity and competition. Nevertheless, it is at the beginning of the 1980s that it became the mainstream in the USA. Indeed, President Reagan’s executive order 12291 of 1981 states that “Regulatory action shall not be undertaken unless the potential benefits to society from the regulation outweigh the potential costs to society.” Another 20 years was required for RIA to spread among OECD countries. During the 1990s and before 1997 (year of the first OECD publication on the subject even if RIA were recommended by the council of OECD in 1995), less than 10 OECD countries had comprehensive RIA program, by the end of 2000, 14, and since 2008, more than 30. Nowadays, almost all OECD countries have such programs despite differences (among which the definition of the type of regulation and intervention submitted to RIA). For example, in France, only primary legislation prepared by the government is subjected to RIA to the exclusion of all secondary legislation.

There are very few studies analyzing the potential of RIA for developing countries (Kirkpatrick and Parker 2003); Korea and Mexico are often mentioned as good examples of RIA design. However, in most developing countries, when RIA are present, they are not systematically used, and the methodology developed remains largely incomplete (Kirkpatrick et al. 2003). Rodrigo (2005) highlights the political requirements allowing RIA to prosper.

Conclusion

Even if RIA are often targeted as promoting “conservative” regulations and “deregulation,” it is hard to deny the necessity of assessing the consequences of regulations. Once again, the purpose is not to make sure that only good regulations will be enacted or to avoid all negative unexpected consequences but to reduce the likelihood of bad regulations by clarifying available regulatory options. RIA should remain a tool to inform policy makers and regulators (see also Jasanoff 1990; Rose-Ackerman 2011b); it can be used adequately if and only if its limits are well identified (regarding model hypotheses and data quality and aggregation).

Cross-References