Christopher Carrigan is Assistant Professor at the Trachtenberg School of Public Policy and Public Administration at George Washington University and a scholar at the GW Regulatory Studies Center.
Executive agencies have been required to perform regulatory impact analyses (RIAs) as part of the rulemaking process in the United States since the Reagan Administration. These analyses, which provide a detailed account of the anticipated consequences of a regulatory action, are prepared whenever an agency intends to promulgate a rule that would have a “significant” effect on the economy. They then must submit their RIAs, along with the proposed rules themselves, to the White House Office of Information and Regulatory Affairs for review before taking any action.
Despite the fact that the RIA requirement has been in existence for 35 years, the practice of preparing RIAs remains a target of criticism from across the ideological spectrum. Those opposed to the requirement assert that RIAs are inherently biased against regulatory initiatives. Those in favor of the requirement view its implementation as having been woefully inadequate, as regulations continue to be adopted that do not maximize net benefits.
But critics from both sides do agree on one thing: RIAs are too often used not to inform regulatory decisions but to justify regulatory decisions that are made for other reasons. When RIAs are intended to justify pre-made decisions, agencies have little incentive to present them in ways that encourage feedback from interested parties. Thus, a related, widely-shared concern is that RIAs are becoming more complex, making them less accessible to potential commentators who would otherwise be able to provide input to make better rules. This latter criticism is not without merit: between 2000 and 2012, RIAs grew in average length from 31,000 words to 128,000 words.
In a recently published paper, we propose a solution both to the unnecessary complexity with which RIAs are sometimes presented and to their alleged use as window dressing for pre-determined, political decisions.
Our proposal requires agencies to produce a simpler analysis than they currently do, and to produce that “back-of-the-envelope” analysis much earlier in the regulatory process, making it available for public scrutiny at that time. As it currently stands, RIAs are published concurrently with proposed regulations. Considerable research has shown that agencies often make key decisions prior to the issuance of their proposed rule. Thus, if an analysis is to inform agency decisions, it stands to reason that it must be performed prior to when those decisions are made.
But merely moving the timing of the analysis is insufficient to ensure that it meaningfully shapes agency decisions. Proponents of benefit-cost analysis have long asserted that it has the potential to improve the transparency of regulatory decision-making. Yet, if the analysis is hundreds of pages long and filled with complex technical analysis and jargon, then the benefits of transparency are not achieved. Any change to the RIA requirement must correct this problem, as well.
In our paper, we argue for a new visioning of RIAs, in which the analysis is based on the following three principles.
Of course, even with our re-visioning of benefit-cost analysis with these principles in mind, the devil may lie in the details. Implementation of any new requirement for agencies to engage in early, back-of-the-envelope analysis is bound to be tricky. Agencies could “game” such a requirement by making the analysis more complex than we intend or by choosing alternatives that are not realistic.
But the three components of our requirement have the potential, at least under many circumstances, to reinforce each other. For example, if the agency now has to consider a broader set of alternatives, then analyses may themselves have to be more streamlined as a result, allowing the reader to identify controversial assumptions or methods more easily. In such cases, the agency will be less able to implement its preferred choice if that choice does not stand up to a benefit-cost test.
However, agencies, in some situations, may be able to avoid performing a genuine analysis of multiple alternatives by copying and pasting an unduly complex analysis of one alternative while incorporating only small changes. As a result, we offer possible “carrots” and “sticks” that can be used to encourage agencies to adhere to our more simplified analysis requirement. These include varying the stringency of subsequent steps in the rulemaking process. For example, agencies that produce good back-of-the-envelope analyses might be exempted from producing more detailed RIAs later, particularly since RIAs are not necessarily used to inform some of these rulemakings anyway.
In addition, we identify two potential candidates to be used to assess agencies’ adherence to a new, back-of-the-envelope requirement. First, since rules are already subject to judicial review, the courts could evaluate back-of-the-envelope analyses more easily than they can full-scale RIAs—which, under current law, are usually exempt from judicial review. Still, judicial review of any back-of-the-envelope analyses would have to be carefully structured, as we do not want courts to start requiring agencies to add complexity to a back-of-the-envelope analysis out of fear that the regulatory action might be deemed arbitrary and capricious.
Second, the Office of Information and Regulatory Affairs within the White House has incentives both to ensure that analysis is conducted early in the process and that it is done in such a manner that encourages public input.
After thirty-five years, benefit-cost analysis has become a well-established part of the U.S. regulatory process. But there is no doubt that its application to regulatory questions could be significantly improved. We believe that simplicity and timing are central to making this improvement possible. If agencies use analysis to offer realistic policy alternatives to the public at an early point in their decision-making processes, then that analysis might fulfill its potential to encourage better policy decisions as envisioned by its proponents, while not crippling the regulatory process in a way feared by its opponents.