Fundamentals

Regulatory Impact Assessment: What It Is, How It Works, and Its Limits

By Daniel Sardá · Published on October 13, 2026

7 min read1,374 words

Regulatory impact assessment examines a problem, compares responses, and makes their likely effects visible before a rule is adopted.

How can a government justify that a new rule is necessary, addresses the right problem, and does not impose disproportionate burdens? A Regulatory Impact Assessment (RIA) is meant to bring order to that question before a decision is made.

It is not a formula that automatically produces perfect regulation. It is a process for defining a problem, gathering evidence, comparing alternatives, and setting out the foreseeable effects of each option. Its chief value lies in requiring the public authority proposing an intervention to explain its reasons in a way that can be examined.

Key idea: An RIA is not a seal of approval for a rule. It is a discipline of prior justification that makes it possible to ask whether intervention is warranted and which instrument should be used.

An assessment before regulation

An RIA is carried out before a regulation is adopted. It begins with a proposal, rather than a rule whose results can already be observed. According to principles promoted by the OECD, it should be integrated into decision-making and be proportionate to the importance and risks of the measure under review.

This matters because regulation can rest on an incomplete diagnosis. It can also settle on an instrument—a prohibition, a license, or a disclosure requirement—before clarifying the conduct or harm it is meant to address. When that happens, the assessment risks becoming a retrospective defense of a decision already made.

A serious RIA reverses that order: it first identifies the problem and its causes, then states the objective, and finally compares ways to achieve it. Those options should include not adopting a new regulation, which serves as a baseline for understanding what would happen without the proposed intervention.

How regulatory impact assessment works

While specific requirements differ across jurisdictions, the reasoning usually follows five steps.

1. Define the problem. Distinguish its symptoms from its causes, and show its scale with the available evidence. An undesirable situation alone does not establish that a new rule can improve it.
2. Specify the objective. The intended result must be concrete enough to compare options and, later, to assess whether the measure worked.
3. Develop alternatives. Consider regulatory instruments, non-regulatory responses, better information or oversight, changes to existing rules, and the option of not intervening.
4. Estimate impacts. Identify benefits, costs, risks, and indirect effects. Quantitative and qualitative information can be combined when not everything can reasonably be expressed in monetary terms.
5. Explain the choice. The authority should show which option it prefers, which assumptions it used, and why it considers that response proportionate to the problem.

Cost-benefit analysis can be part of this process, but it does not exhaust it. An aggregate figure can conceal decisive questions: who pays, who receives the benefits, which rights are at stake, or how much the result depends on uncertain assumptions.

An example: licenses for a low-risk activity

Imagine that an authority identifies misleading information in a low-risk commercial activity. Its first response might be to require every provider to obtain a license in advance. An RIA requires a pause before that intuition becomes a rule.

First, it would be necessary to establish how frequent and harmful the practice is, as well as its causes. Do consumers lack information? Is it difficult to bring a complaint? Or does the problem arise from non-compliance with rules that already exist? Several responses could then be compared: banning certain practices, requiring a license, creating a simple register, strengthening complaint mechanisms, imposing disclosure duties, or maintaining the current situation.

A license might make oversight easier, but it could also raise entry costs and reduce competition. A register might make responsible parties easier to locate with a lighter burden. A clear disclosure duty might address the problem more directly, although it would require credible monitoring and penalties.

The example does not determine which option is correct. It shows why comparison should come before the decision, and why it is sensible to seek the least restrictive effective means. From a liberal perspective, that requirement is especially important when the state restricts peaceful choices or conditions access to an economic activity.

Key idea: Comparing alternatives does not mean always choosing the cheapest option. It means justifying why the burden imposed is necessary and proportionate when other responses are available.

Which impacts should be made visible?

A useful assessment looks beyond the public budget or the average cost. Depending on the proposal, it may consider:

• compliance costs, paperwork, and time;
• expected benefits and avoided risks;
• effects on prices, competition, innovation, and market entry;
• social or environmental consequences;
• the real capacity for enforcement and compliance;
• different effects on small businesses, consumers, or vulnerable groups; and
• uncertainty in the data and assumptions used.

Distribution matters as much as the aggregate balance. A measure with broad but modest benefits can impose highly concentrated costs on a group with little capacity to adapt. Conversely, a general burden can benefit incumbent actors disproportionately. Making these effects visible does not settle the political or legal judgment, but it prevents that judgment from being presented as a purely technical conclusion.

Consultation, transparency, and public reasons

Consultation allows affected people, specialists, and organizations to challenge data and contribute information that the administration does not possess. It should take place early enough to influence the assessment, not only once the decision is closed. Publishing the alternatives, assumptions, and responses received also makes public accountability easier.

That does not turn consultation into a vote or guarantee neutrality. Participants have unequal information and interests, and better-organized groups often find it easier to take part. The authority therefore remains responsible for assessing the evidence, explaining disagreements, and justifying its choice under standards consistent with the rule of law.

Caution: If an RIA is prepared to defend an option that has already been chosen, it can become an expensive ritual. Publication, early consultation, and quality control can help reveal that problem, but they cannot eliminate it on their own.

What an RIA is not

Several related tools serve different purposes:

• Ex post evaluation: examines an existing regulation and compares its actual effects with its objectives. An RIA estimates beforehand; ex post evaluation checks afterwards.
• Cost-benefit analysis: compares costs and benefits, sometimes in monetary terms. It can be part of an RIA, but an RIA also addresses alternatives, distribution, legality, and uncertainty.
• Environmental impact assessment: focuses on the environmental consequences of projects or decisions under the applicable framework. It does not replace a general assessment of a regulatory intervention.
• Legal review: asks whether a measure respects powers, procedures, and rights. A favorable RIA does not make an unlawful rule lawful.
• Compliance plan: explains how a regulation will be applied and supervised. It is an important part of design, not a complete assessment of necessity.

In Mexico, for example, the federal authority uses the term Manifestación de Impacto Regulatorio within its own system, according to CONAMER. That name and its requirements should not be treated as a universal legal model: each jurisdiction establishes different procedures, thresholds, and exceptions.

Limits: data, bias, and difficult decisions

An RIA works with estimates. Information may be missing, indirect effects may appear late, and some consequences—such as restrictions on rights or institutional changes—cannot be adequately expressed in money. Quantifying what is available should not make what can only be assessed qualitatively disappear.

There is also a question of administrative proportionality. Not every minor rule needs an extensive study; the depth of the assessment should match the decision's risk and scope. In urgent situations, completing the ordinary analysis may be impossible, but that exception makes subsequent review even more important.

Ex post evaluation closes the cycle: it compares what happened with the forecasts, identifies unexpected effects, and makes it possible to correct, simplify, or withdraw the rule. In that sense, an RIA stops being merely an advance snapshot and becomes a basis for institutional learning.

Regulatory impact assessment does not replace democratic deliberation, legal limits, or the responsibility to decide. Its contribution is more modest and, for that reason, valuable: it makes it harder to regulate without defining the problem, comparing less restrictive means, and recognizing who will bear the consequences.