A closer look at proportionality and

threshold tests for RIA

Annex to the OECD Best Practice Principles on Regulatory

Impact Assessment

PUBE
This project is funded by the EU via the Structural Reform Support Programme and
implemented in collaboration with the OECD, and the European Commission
2

Acknowledgements

The paper was prepared by a multidisciplinary team, led by Daniel Trnka, Senior Policy Analyst, Eric
Thomson, Policy Analyst, Luís Baltazar, Consultant and Bruno Dámasio, Senior Consultant with the
encouragement and support of Elsa Pilichowski, Director, Public Governance Directorate and Nick
Malyshev, Head, Regulatory Policy Division, Public Governance Directorate. Claudia Paupe lent
administrative and organisation support. Paul Davidson, Policy Analyst, Regulatory Policy Division
provided comments on the first draft. The paper was prepared for publication by Jennifer Stein.
The paper would have not been possible without the generous support of the government of Portugal and
the European Union Structural Reform Support Programme. The team would like to thank in particular the
entire team from JurisAPP-UTAIL and the EU SRSP.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
 3

Table of contents

Executive summary 4
1 Introduction 6
2 Threshold tests and proportionality in OECD countries 7
Introduction 7
Description of threshold tests and proportionality 8

3 Theoretical foundations for the ex ante analysis of regulation 11

4 Conclusions 15
Suggestions for proportionality in RIA 15

References 17
Annex A. OECD countries’ practices in proportionality and thresholds 18

Figures
Figure 2.1. Requirements and practice of conducting RIA 7
Figure 2.2. Thresholds for RIA 8
Figure 2.3. Exceptions of not conducting RIA 9
Figure 3.1. Requirement to conduct RIA when transposing EU directives 12
Figure 3.2. Benefits/costs of analysing a proposed regulation 13

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
4

Executive summary

OECD countries have recognised that it is critical to develop regulations and policies using the best
available evidence. Reflecting that, every OECD country and the European Commission use stakeholder
engagement, regulatory impact analysis (RIA) and other analytical tools in some form during the
development of regulation.
Many countries have also acknowledged that not every regulation or proposal needs the same level of
consideration or scrutiny. Some proposals are procedural or have only minor effects on society, so it is not
reasonable to spend as much time on the development process. The process itself, including RIA,
stakeholder engagement and other steps should pass a cost-benefit analysis – the costs and time to
develop and analyse regulations should be outweighed by the positive effect that these have of improved
policies or regulations. Therefore, it is important the resources used to develop a policy scale with the size
of its anticipated impacts.
Many OECD countries and the EU have adopted tiered approach to RIA; a shorter or less rigorous
development process in case of regulations with lower anticipated impacts might be sufficient in some
cases. In others, policy makers are required to undertake a full RIA, if the proposal has a certain level of
impacts or meets certain qualitative criteria. In some cases, governments may exempt a proposal from
RIA or stakeholder engagement entirely, such as national emergencies. Although the impacts of a
regulation to resolve a national emergency may be high, governments may allow a policy to be developed
rapidly to avoid the negative impacts of allowing the emergency to continue.
Many countries have adopted threshold tests to determine which proposals are subject to a longer and
more thorough development process. If the anticipated impacts of a proposal are above a certain threshold,
it must undergo a longer process and a more detailed impact assessment has to be developed.
The rationale that countries have used to establish such tests is relatively weak. It appears that some
countries have set the criteria to limit the number of proposals to the capacities of the regulatory oversight
body (ROB). In at least one case, the ROB itself selects which proposals to scrutinise. In addition, the
OECD has observed that use of emergency exemptions are sometimes abused to skirt around RIA
requirements and make legislation operative more quickly, although such exemptions should only be used
when there is a significant risk to citizens if legislation is not swiftly enacted.
The ability for policy makers to change and improve a policy are also an important consideration. If
policy makers have limited flexibility in formulating the policy while developing it – for example, the minister
has a fixed idea of the regulation or policy in mind – they may see requirements for RIA or stakeholder
engagement only as an administrative burden. Policy makers also need the skills to develop regulations
and analyse them, without that higher RIA requirements will not improve policy.
OECD countries should consider the following, when developing proportionality rules or threshold tests:

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
 5

1. Determining the scope of RIA should start at an early stage when policy makers are evaluating the
problem - potentially even before considering the need for intervention - and identifying regulatory
and non-regulatory alternatives. Preferably, this process should start already in the phase of
legislative planning.
2. An oversight body should assess whether the regulator has characterised the problem correctly,
including its magnitude, when the regulator still has the flexibility in formulating a regulation or
policy1. The earlier policy makers understand the magnitude of the problem, the better the
government may target resources to developing solutions.
3. During the early stage of RIA, policy makers should begin to introduce an economic rationale and
data to determine the scope of the issue. This does not mean an in-depth analysis at an early stage
(e.g. a well-developed cost-benefit analysis). Policy makers should be broadly scanning an issue,
before undertaking an in-depth analysis.
4. The time and resources devoted to the development of regulation and its analysis should relate to
the size of the impacts, the size and structure of the economy, the impacts per capita, the flexibility
of the policy, and the relative resources of the government.
5. If a country chooses to use quantified thresholds for RIA, they should be inclusive and base the
thresholds on the size of impacts across society, rather than focusing on any specific sector or
stakeholder group. There may also be a risk in using one single value threshold that captures
impacts across society. One stakeholder group may be disproportionately affected but the total
impacts are below the threshold, so countries may wish to consider a threshold that also
incorporates a per capita or stakeholder threshold.
6. Regulations should only be exempt from completing the RIA process in genuinely unforeseen
emergencies, when a significant delay could objectively put the wellbeing of citizens at risk.
Oversight bodies should be very critical of ministries that overuse such exemptions. Ministries
should also be required to conduct an ex post evaluation to ensure that the regulation was effective
after a defined period of time.
7. Regulations with limited policy options or flexibility (e.g. transposition of EU directives or
supranational laws) might have a less rigorous process. When fewer policy options or instruments
are available, even if the impacts may be quite significant, policy makers have less flexibility to
improve a policy at this stage. Despite this, governments should be mindful that EU directives or
other supranational instruments might still have a degree of flexibility in their implementation.
8. The time and resources for regulation development and analysis should also scale with the
capacities of the government. It is important that governments continuously build the expertise of
policy makers in RIA and stakeholder engagement to make analysis more effective. Governments
must build capacities in ministries before they can require significant levels of analysis.

1
This guideline follows suggestions from Renda (2011) for the EU.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
6

1 Introduction
Regulation is a key lever for the government to change the behaviour of businesses and citizens to protect
citizens’ health, safety and wellbeing. It is systematic approach to critically assess the positive and negative
effects of proposed and existing regulations. RIA is not (or at least should not) be limited to measuring the
effects and impacts of a proposed decision, but it is the process of actually selecting the best policy option
(regulatory or non-regulatory) by thoroughly analysing the problem and by weighing various policy options.
Governments should begin early when an issue is first identified and carefully consider potential options
and their impacts to choose the decision that maximises net benefits and meets the government’s broader
policy objectives. The 2012 Recommendation from the Council on Regulatory Policy and Governance
emphasises that OECD governments should adopt ex ante impact assessment practices that are
proportional to the significance of the regulation’s anticipated impacts (OECD, 2012[1]).
Ex ante impact assessments are more than just the final RIA statement, but include the whole process and
of evaluating the regulation, contacting stakeholders, performing any supplemental analyses (SME Test,
Competition assessment, etc.), oversight and other steps.
Without an effective and proportional process, governments may produce regulation that does not meet
the needs of citizens. New regulation may not be effective in meeting its objectives or may only do so at a
high cost, placing significant unnecessary burdens on businesses and citizens. A jungle of poorly designed
regulation reduces a country’s competiveness. Businesses and citizens will also lose their trust in
government as they do not see the benefits of regulation yet experience the associated costs.
However, governments have limited resources to develop regulation may put significant burdens on
policy makers. Yes, policy makers must weigh the costs and benefits of different policy options and consult,
but the process for analysing regulations must pass a cost-benefit test itself. It makes no sense to analyse
regulations as if each proposal was identical. Besides wasting time and resources on relatively minor
proposals, there is a risk that the government will not properly assess the consequences of regulations
whose impact is potentially greater on wellbeing by giving equal importance to each regulation.
This paper is one of the first to examine how OECD countries have designed their threshold tests to target
analysis. It also considers some of the theoretical foundations for decision-making to provide some
guidance on how to develop an appropriate methodology to target analysis to regulations that have the
greatest impacts.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
 7

2 Threshold tests and proportionality

in OECD countries

Introduction

Across the OECD, governments have often put RIA at the centre of regulatory policy alongside other
programs such as:
 implementing administrative simplification and burden reduction programmes;
 promoting more transparency in regulatory decision making and consultation with stakeholders;
 introducing new regulatory oversight bodies at the centre of government;
 taking a more risk-based approach to regulating;
 and tackling compliance and enforcement issues (OECD, 2012).

Figure 2.1. Requirements and practice of conducting RIA

Note: Data is based on 28 EU Member States.

Source: Indicators of Regulatory Policy and Governance Surveys 2017, http://oe.cd/ireg.

RIA has been widely adopted and applied in OECD countries to varying degrees (See Figure 2.1).
However, most governments do not require a targeted RIA based on the expected level of impacts; only

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
8

15 jurisdictions have some sort of threshold test for RIA for primary laws or subordinate regulation, despite
a clear need to target resources effectively to high-impact policies. However, the trend is that more
countries are adopting a proportionate approach; six OECD countries adopted a threshold test between a
full and simplified RIA between 2014 and 2017 (Figure 2.2).

Figure 2.2. Thresholds for RIA

2014 2017
Number of jurisdictions
35

30

25

20

15

10
14 15
5 9
7 7 8 7 8
0
Primary laws Subordinate regulation Primary laws Subordinate regulation
Is there a threshold test to determine whether a RIA is undertaken at all? Is there a threshold to determine whether a full RIA (as opposed to a simplified
RIA) is undertaken?

Note: Data is based on 34 OECD member countries and the European Union.
Source: Indicators of Regulatory Policy and Governance Surveys 2014 and 2017, http://oe.cd/ireg.

This chapter and the annex of the paper aim to provide a mapping of the structure and resources for the
use of thresholds and proportionality for RIA in OECD countries and the EU. It does so based on the results
of the OECD Regulatory Policy Outlook and other relevant documents. This research is one of the first
organised collections comparing the experiences of different countries about their scope of the threshold
tests and proportionality rules in their regulatory processes. Both the OECD Regulatory Policy Outlook
2018 and Better Regulation Practices across the European Union report provide a summary of various
approaches.

Description of threshold tests and proportionality

OECD countries and the EU have implemented proportionality and threshold tests for a number of reasons.
First, they help to ensure that regulations with significant societal impacts are adequately assessed before
being approved. Secondly, they ensure that government resources are not unduly wasted in assessing
regulatory proposals with only minor impacts, where the costs of conducting RIA, consultation and other
development process may outweigh their benefits. In turn, this also helps to avoid consultation fatigue,
particularly with external stakeholders. It ensures that broader consultation is conducted where the impacts
of a regulation are expected to be significant and consultations will improve proposals. Finally, thresholds
help to improve the transparency of the overall development system, where there are clear criteria that are
straightforward to apply in practice.
In many cases, OECD countries require a simpler level of assessment for decisions that only make minor
procedural changes to a regulation or that have limited economic or social impacts. A full analysis is carried
out after a preliminary impact assessment indicates that the impacts require a deeper analysis.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
 9

Some countries require lighter analysis (“light”, preliminary” or “small” RIA) for all regulations and a more
thorough analysis for selected draft regulations with more significant impact. Often, governments only
require a qualitative description of potential impacts and impacted stakeholders for a lighter RIAs. This
approach might be especially useful when the country uses several stages in its legislation-making process
(e.g. first, legislative intent is approved by the government before elaborating on and developing a new
regulation or policy).
Threshold and proportionality tests also serve as a way of exempting certain regulations from some impact
assessment procedures. As can be seen from Figure 2.3, many OECD countries exempt a proposal from
the RIA process for a number of reasons. It is also worth noting that there has been a slight increase overall
in the types of exceptions available since 2014 (Figure 2.3).

Figure 2.3. Exceptions of not conducting RIA

Note: Data is based on 34 OECD member countries and the European Union.
Source: Indicators of Regulatory Policy and Governance Surveys 2014 and 2017, http://oe.cd/ireg.

More than one-third of OECD members have exceptions to conducting RIA. For example, a RIA might not
be required when an imminent natural disaster requires decisive action to protect the health and safety of
citizens. However, the OECD has found that “emergencies” sometimes made up more than 50 per cent of
avoided RIAs (see for example the Slovenia Regulatory Policy Review, (OECD, 2018[2])). Ministers and
policy makers who are under political pressure to develop a regulation could easily abuse exceptions to
RIA for emergencies.
Broadly speaking, the use of thresholds and proportionality for RIA can be broken into four types based
on their main features, although all countries use some mix of qualitative and quantitative criteria.
1. Total impact quantitative test: The level or depth of analysis is dependent on the total impacts to
society. The policy maker needs to estimate the total impacts quantitatively, usually as part of a
preliminary step in the development of a new regulation.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
10 

2. Multi-criteria test: The depth of RIA is based on a mix of quantitative and qualitative criteria in a
number of key areas. The criteria may include, for example, the number of affected businesses, a
certain level of CO2 emissions or a subjective assessment of the significance of impacts on key
sectors. Normally, these impacts are quantified, but not monetised like the other types in this list.
3. Single-issue test: The analytical depth of RIA is based on impacts to one sector or stakeholder
group. For example, a full RIA is only required when business costs exceed a certain amount.
4. Discretionary with oversight test: In these systems, generally, the policy maker undertakes a
preliminary RIA and then a regulatory oversight body (ROB) determines the level of analysis in the
next step. Alternatively, an ROB gives an opinion if the level of analysis is sufficient at the end of
the process.
Many of the thresholds in OECD countries appear to have no obvious associated economic rationality.
They have been defined directly by guidance, decree or executive order, which leaves much to be desired
in terms of credibility. None are directly tied to the size of the economy or other criteria. In cases when a
total impact measure is used, most governments have not changed or reviewed the given threshold in
decades. For example, the U.S. high-threshold of $100 million has remained unchanged since the 1980s.
In general, oversight bodies in OECD countries appear to have targeted a certain number of RIAs per
year or they target specific economic goals, like improving competition or reducing administrative burdens
(OECD, 2009[3]). The latter is certainly not in line with evidence-based policy: countries should design
processes to improve the net benefits of a policy across the whole of society, rather than specific groups
or sectors.
A haphazard approach to setting the limits of thresholds could entail serious risks to the effective and
fruitful conduct of the impact assessment process. If the threshold is too low, policies are analysed
unnecessarily, but if it is set too high, some proposals will face too little scrutiny and analysis.
Policy makers proposing regulation may also have an incentive to understate impacts so that they fall
below the threshold or to separate several related proposals in such a way as to avoid triggering the
requirements. Effective oversight is critical to the proper use of threshold tests or exemptions from RIA.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
  11

3 Theoretical foundations for the

ex ante analysis of regulation

OECD countries and the EU have apparently chosen threshold tests and proportionality rules based on
limiting the number of proposals to review per year, more than on economic rationality.
Anecdotally, the OECD has also heard of many cases when policy makers have faced significant political
pressure to decide on a policy option or regulation in a very short period. In some cases, policy makers
themselves will have decided on a policy option before they have given it due consideration.
As above, some OECD countries encourage policy makers and ministers to think about decisions more
with threshold tests or tools to control the level of analysis of each regulatory initiative. However, it is also
clear that the thresholds themselves are often not designed to consider whether additional analysis or
oversight actually improves the law.
If anything, it is in fact likely that governments in OECD countries do not deploy enough resources to
analyse regulation during its development. Even if they could, a general lack of experts in the area of
evaluation, cost-benefit analysis and policy analysis may limit the amount that any government can do to
improve the level of impact analysis in regulatory proposals.
Unfortunately, quantitative comparisons are scarce on the costs of RIA, its evaluation and oversight and
the benefits RIA has on improving regulation. As an illustrative example, Hahn and Tetlok roughly
estimated that each “major-rule” RIA review in the US costs the Office of Information and Regulatory Affairs
(OIRA) USD 700 000, compared to a minimum impact of over USD 100 million (Hahn and Tetlock, 2008[4]).
OIRA’s oversight role is likely going to improve the net impacts of a regulation by more than 0.7 per cent
during the review process. The RIA process itself, particularly in the early stages, should improve
regulatory quality even more. Different regulatory design or policy options are likely to have a greater
impact than minor changes to the design of a regulation done late in the process. Even in small countries
where net impacts are smaller, the RIA and stakeholder engagement process is clearly still valuable, given
that even a small amount of evidence is likely to significantly improve policies.
Of course, this assumes that policy makers have flexibility to choose different policy options or modify
proposed regulations. Policy makers often develop laws or policies without much flexibility. For example,
EU directives at the transposition stage give much less flexibility in how a country adopts it into their own
law, compared to a national law. However, governments must continue to be careful that any over-
implementation of an EU directive passes a cost-benefit test. In fact, almost all EU countries continue to
use the same RIA process for the transposition of EU directives as national law. RIA is also used by about
half of EU countries during the negotiation stage of EU directives (OECD, 2019[5]).

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
12 

Figure 3.1. Requirement to conduct RIA when transposing EU directives

Yes No

Number of EU Member
States
30

25

20

15
26
24
10

2
0
RIA is required when transposing EU directives The same requirements and processes for RIA apply as for domestically
made laws

Note: Data is based on 28 EU Member States.

Source: Indicators of Regulatory Policy and Governance Surveys 2014 and 2017, and the extension to all EU Member States, http://oe.cd/ireg.

On the other hand, many countries have exceptions for less flexible laws or legal instruments. New
Zealand, for example, has a provision that allows ministries to skip an impact analysis, when the policy is
necessary “in order to comply with existing international obligations that are binding on New Zealand”.
Additionally, New Zealand may not require a RIA when “the government has limited statutory decision
making discretion or responsibility for the content of proposed delegated legislation” (New Zealand
Department of the Prime Minister and Cabinet, 2017[6]).
Quantitative threshold tests have an advantage that an oversight body and regulator can relatively easily
observe which category a new proposal falls into and then guide the level analysis. Nevertheless, the
amount of effort that goes into developing and analysing a new regulation should be proportionate to the
size of the impacts. The additional benefit of more analysis declines with the amount of effort, so
policy makers should in theory stop when their analytical resources would no longer improve the policy or
could be better used in another area (Figure 3.2).
In 2016, the National Audit Office in the UK found that the government could achieve a better value for
money by increasing its rate. In 2018, the Better Regulation Executive increased its business impact
threshold from GBP 1 million to GBP 5 million for the full RIA and scrutiny threshold. In the UK, over 90%
of the GBP 10 billion regulatory cost reductions claimed during the period 2010-15 were achieved through
just 10 regulatory changes. Therefore, the NAO and suggested that the BRE and UK Regulatory Policy
Committee should focus on the few most important regulations per year.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
  13

Figure 3.2. Benefits/costs of analysing a proposed regulation

Net Benefits of the regulation Total cost of developing the regulation

Maximum benefit of analysis

Time spent on the

development and analysis

Source: Authors’ elaboration.

It might also be very difficult for an oversight body to monitor or judge the precise level of effort put into the
analysis and development of a new regulatory proposal. Despite this, the EU and many EU member states
often only recommend that policy makers have a proportionate approach to regulation without setting
concrete criteria or thresholds, leaving significant discretion to policy makers. This discretional approach
makes the targeting of RIA efforts rather flexible. Generally, the oversight body checks the quality and
scope of analysis at one or more stages during the development of a regulation. However, the regulatory
oversight body does not necessarily have sufficient information on the significance of the problem and
potential impacts to know if the policy makers have given the proposal due consideration. Thresholds may
be a way to encourage more analysis without leaving a high-level of discretion to policy makers.
Based on the OECD’s experience, policy makers often ignore or underplay a critical aspect of the RIA
process: identifying the problem. In fact, it only makes sense for policy makers to consider solutions or
policy options once the magnitude of the problem is understood, particularly its causes. Otherwise, they
may look at policy options that do not properly target the cause or select interventions that do not effectively
fix the problem. Furthermore, it is at this stage that the regulatory design process and RIA have the most
impact. Once, a policy intervention has been decided, further analysis is unlikely to change a policy
drastically and thereby generate higher net benefits. Often ministers or policy makers may be unwilling to
change a proposal that is relatively well-developed because of the delay it may cause in enacting the policy
or regulation. Therefore, the selection of the policy instrument itself at the early stage of the process (e.g.
choosing a regulation or a new program) will have significant impacts. This suggests that oversight bodies
should be intervening early to coax policy makers to consider a wide range of policy options for major
problems, before the policy makers tweak and adjust a specific regulation or other solution.
When considering proportionality for the EU, Andrea Renda suggests that the first RIA should ask the
questions: Is there a problem? Is action at the EU level needed? What alternative policy options may be
envisaged?2 (Renda, 2011[7]). Renda stresses that these questions should be analysed from a “law and

2
A version of the second question applied to national governments could be “Is action required at the federal/national
level?”

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
14 

economics” perspective, building in evidence at the earliest stage. Etzioni has suggested that a mixed-
scanning approach to public administration decision-making may be appropriate (Etzioni, 1967[8]). At the
early stage in the development of regulation, policy makers would take a broad but not very detailed scan
of the policy issue at hand and its potential options, but only focus on some of the key details of the issue.
Ideally, the existence of a problem and its significance should be developed before the stage of legislative
planning. Many countries develop annual legislative plans. One might think that at this stage, policy makers
should make a quick analysis of the scope of the issue, stakeholders affected, and the magnitude of
potential impacts. Based on this preliminary analysis, a decision could be made on the depth of RIA
necessary. The oversight body should then evaluate the legislative plan checking the decisions of
individual ministries and government agencies on the anticipated depth of RIAs necessary and potentially
suggesting adjustments.
The economic context of the country plays an important part in the depth of analysis of any particular
proposal and any thresholds. In large economies like the E.U. and the U.S., relatively minor changes in
regulations or directives can have impacts worth billions of euros or dollars. The time and resources spent
on analysis should then be significantly more sophisticated and scale to the size of the economy. Larger
economies, where small changes have large potential impacts, should apply more resources to the
analysis and oversight of regulation than small countries.
Nevertheless, small economies are not off the hook. In a small economy, regulations could have significant
impacts per capita, so proportionally more analysis may be worthwhile, if there may be significant impacts
on any one sector or stakeholder group.
Unfortunately, the relative complexity and resources to analyse a particular regulation do not necessarily
scale with the size of the economy. A small country with a relatively small public service will still need to
staff an oversight body and look at many proposals. As mentioned above, ROBs in OECD countries tend
to look at about 100-200 proposals per year, regardless of the size of the economy (OECD, 2009[3]). Given
potentially limited human resources of smaller economies’ administrations, it may make sense then for
smaller countries to target resources to fewer, but high-impact proposals.
Finally, the level of expertise available in a country or government may be a limiting factor in analysing
new regulations. A policy cannot be improved without the people who can improve it. The level of analysis
required then should also scale with the skills and capacities of the government. This does not mean
avoiding RIA or stakeholder engagement, but rather starting with a simpler or less intensive analysis at
first as the country implements RIA and stakeholder engagement. Governments that have limited
resources or capacities to assess regulations may wish to focus their efforts only on proposals with the
absolute highest impacts at first. As the country develops skills over time in RIA, then the oversight then
the regulatory development process could involve a more in-depth analysis with appropriate oversight.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
  15

4 Conclusions
OECD countries have many different proportionality tests, thresholds and exceptions. If anything, it seems
likely that the skills and resources for RIA are largely below what would be optimal. RIA and stakeholder
engagement during the development of regulations and policies has significant benefits. However, OECD
governments and the EU face budget, time and capacity constraints. Often, the thresholds and
proportionality rules for RIA do not appear to be based on an economic rationale, but rather the capacities
of the oversight body to oversee the process. Most oversight bodies review between 100 and 200
proposals per year, regardless of the size of the economy.
Critically, with limited government resources, RIA, stakeholder engagement and its oversight should focus
on regulations that not only have the greatest potential impacts on society. The process for RIA is only
valuable when it improves the decision-making of policy makers. Therefore, RIA and its oversight should
start early in the process when it has the greatest effect on proposals and therefore social welfare.
Ideally, policy makers would put effort into developing a regulation on a continuously proportional basis –
until the costs of more analysis exceed the improvements in the proposal. However, it might be difficult for
the oversight body to determine that policy makers have put in the exact level of effort. Thresholds are,
therefore, a good tool to help governments triage the relative importance (in terms of impacts) of regulatory
proposals, but they have a greater impact earlier in the process, when the government is analysing a
problem and considering options.
The depth of analysis, including the use of RIA and stakeholder engagement, should in part scale with the
size of the economy; larger economies should use more resources to analyse problems and develop
regulations because even minor changes could have large impacts. Smaller countries will still certainly
want to focus efforts on regulations with large per capita impact. OECD countries should also be cognisant
of the limitations of their own public service to analyse and develop new proposals.

Suggestions for proportionality in RIA

1. Determining the scope of RIA should start at an early stage when policy makers are evaluating the
problem - potentially even before considering the need for intervention - and identifying regulatory
and non-regulatory alternatives. Preferably, this process should start already in the phase of
legislative planning.
2. An oversight body should assess whether the regulator has characterised the problem correctly,
including its magnitude, when the regulator has the greatest flexibility in formulating a regulation or
policy.3 The earlier policy makers understand the magnitude of the problem, the better the
government may target resources to developing solutions.

3
This guideline follows suggestions from Renda (2011) for the EU.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
16 

3. During the early stage of RIA, policy makers should begin to introduce an economic rationale and
data to determine the scope of the issue. This does not mean an in-depth analysis at an early stage
(e.g. a well-developed cost-benefit analysis). Policy makers should be broadly scanning an issue,
before undertaking an in-depth analysis.
4. The time and resources devoted to the development of regulation and its analysis should relate to
the size of the impacts, the size and structure of the economy, the impacts per capita, the flexibility
of the policy, and the relative resources of the government.
5. If a country chooses to use quantified thresholds for RIA, they should be inclusive and base the
thresholds on the size of impacts across society, rather than focusing on any specific sector or
stakeholder group. There may also be a risk in using one single value threshold that captures
impacts across society. One stakeholder group may be disproportionately affected but the total
impacts are below the threshold, so countries may wish to consider a threshold that also
incorporates a per capita or stakeholder threshold.
6. Regulations should only be exempt from completing the RIA process in genuinely unforeseen
emergencies, when a significant delay could objectively put the wellbeing of citizens at risk.
Oversight bodies should be very critical of ministries that overuse such exemptions. Ministries
should also be required to conduct an ex post evaluation to ensure that the regulation was effective
after a defined period of time.
7. Regulations with limited policy options or flexibility (e.g. transposition of EU directives or
supranational laws) might have a less rigorous process. When fewer policy options or instruments
are available, even if the impacts may be quite significant, policy makers have less flexibility to
improve a policy at this stage. Despite this, governments should be mindful that EU directives or
other supranational instruments might still have a degree of flexibility in their implementation.
8. The time and resources for regulation development and analysis should also scale with the
capacities of the government. It is important that governments continuously build the expertise of
policy makers in RIA and stakeholder engagement to make analysis more effective. Governments
must build capacities in ministries before they can require significant levels of analysis.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
  17

References

Etzioni, A. (1967), Mixed-scanning: A “Third” Approach to Decision-making, Public [8]

Administration Review 27(5), https://www.researchgate.net/publication/235356796_Mixed-

scanning_A_Third_Approach_to_Decision-making (accessed on 18 December 2020).

Hahn, R. and P. Tetlock (2008), Has Economic Analysis Improved Regulatory Decisions?, [4]
http://RegInfo.gov (accessed on 18 December 2020).

New Zealand Department of the Prime Minister and Cabinet (2017), CO (17) 3: Impact [6]

Analysis Requirements, https://dpmc.govt.nz/sites/default/files/2017-06/coc_17_3.pdf.

OECD (2019), Better Regulation Practices across the European Union, OECD Publishing, [5]
https://doi.org/10.1787/9789264311732-en.

OECD (2018), Regulatory Policy in Slovenia: Oversight Matters, OECD Reviews of Regulatory [2]

Reform, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264291690-en.

OECD (2012), Recommendation of the Council on Regulatory Policy and Governance, OECD [1]

Publishing, Paris, https://dx.doi.org/10.1787/9789264209022-en.

OECD (2009), Regulatory Impact Analysis: A Tool for Policy Coherence, OECD Reviews of [3]
Regulatory Reform, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264067110-
en.

Renda, A. (2011), Law and Economics in the RIA World: Improving the use of economic [7]
analysis in public policy and legislation, https://www.ceps.eu/ceps-publications/law-and-
economics-ria-world-improving-use-economic-analysis-public-policy-and/ (accessed on
18 December 2020).

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
18 

Annex A. OECD countries’ practices in

proportionality and thresholds

The following section presents different examples of OECD countries of proportionality and threshold tests,
both with an informal proportional approach and in some cases with defined quantitative and/or monetary
thresholds.
As can be seen from the brief descriptions of selected countries, there is a wide variance in approaches.
The OECD is interested in developing this Annex and it would welcome any information to support the
summaries below. This list is by no means comprehensive.
Australia: A Preliminary Assessment determines whether a proposal requires a RIA for both primary and
subordinate regulation (as well as quasi-regulatory proposals where there is an expectation of compliance).
A Regulation Impact Statement is required for all Cabinet submissions. This includes proposals of a minor
or machinery nature and proposals with no regulatory impact on business, community organizations or
individuals. A RIA is also mandatory for any non-Cabinet decision made by any Australian Government
entity if that decision is likely to have a measurable impact on businesses, community organizations,
individuals or any combination of them.
Austria: For all new laws and regulations, an impact assessment is mandatory. The regulation
underpinning this instrument provides an explicit list of impact dimensions that have to be assessed.
Nevertheless, only impacts above a certain threshold have to be assessed in further detail.
Thresholds are mostly quantitative and vary depending on the impact dimension, e.g. for environmental
impacts it exceeds 10 000 tons of CO2 per year.
Other relevant (sub-)dimensions include impacts on the business cycle (threshold 500 enterprises) and
financial impacts like impacts on access to finance (threshold 2,5 million€ or 10,000 enterprises). In 2015,
an impact checklist was introduced, based on the aforementioned thresholds and other criteria. If all these
criteria are met, including that there are no significant impacts, no connection with budget priorities and no
costs can be expected, ministries have the option to provide a short (1-2 pages) description about the
intent of the new or amended regulation, instead of the full-scale assessment. Other than for full scale
impact assessments, in these cases, there is also no mandatory ex post evaluation and consequently no
need to come up with key indicators and target values in the RIA document. This reform was based on the
assessment that there are many minor amendments to laws and regulations that have no significant impact
that can be assessed ex ante and where impacts also cannot be properly measured ex post (e.g. changing
the formal name of an agency).
As a consequence, only for about 35% of new laws have a full scale RIA and resources can be directed to
proposals where high quality impact assessments are needed.
Belgium applies a hybrid system. For example, of the 21 topics that are covered in the RIA, 17 consist of
a quick qualitative test (positive/negative impact or no impact) based on indicators. The other four topics
(gender, SMEs, administrative burdens, and policy coherence for development) consists of a more
thorough and quantitative approach, including the nature and extent of positive and negative impacts.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
  19

Canada: Canada applies RIA to all subordinate regulations, but employs a Triage System to decide the
extent of the analysis. The Triage System underscores the Cabinet Directive on Regulatory Management’s
principle of proportionality, in order to focus the analysis where it is most needed. The development of a
Triage Statement early in the development of the regulatory proposal determines whether the proposal will
require a full or expedited RIA, based on costs and other factors:4
 Low impact, cost less than CAD 10 million dollars present value over a 10-year period or less than
CAD 1 million annually;
 Medium impact: Costs CAD 10 million to CAD 100 million present value or CAD 1 million to
CAD 10 million annually;
 High impact: Costs greater than CAD 100 million present value or greater than CAD 10 million
annually.
Denmark has a threshold for doing ex ante impact assessments using the SCM-method, which is 4 million
Danish Kroner in running costs. The threshold was made out of the experience from the baseline
measurements and expected to cover at least 95% of the overall burdens.
When they assess the consequences to be above the threshold, they use external consultants to do a
more thorough report based on interviews with businesses.
European Commission: The European Commission uses only a qualitative test to decide whether to
apply RIA for all types of regulation. Impact assessments are prepared for Commission initiatives expected
to have significant economic, social or environmental impacts. The Commission Secretariat General
decides whether this test is met based on reasoned proposal made by the lead service. Results are
published in a roadmap.
Finland uses only one threshold in its RIA process. If the impact on Government budget is more than
EUR 5 million, the bill needs to go through a special procedure. This is supervised by the Ministry of
Finance. However, this threshold does not directly affect impact assessments.
Regarding purely impact assessments, they need to be done in all primary legislation as instructed by the
Ministry of Justice. In practice, however, the quality of the assessments differs greatly. It would not be very
meaningful to spend a big amount of legislature resources into a very small matter. Ultimately, individual
ministries and even departments within ministries are rather independent to decide, how much effort they
put into individual impact assessments. The quality is supervised by the newly-founded Finnish Council of
Regulatory Impact Analysis, which selects RIAs to analyse based on their relative importance.
Germany: The Federal Government's programme considers all legislative proposals irrespective of the
anticipated amount of compliance costs for ex ante impact assessments. However, there is an informal
rule that a draft with changes in compliance cost below EUR 100 000 is considered a minor change. The
required level of detail for this kind of draft is much lower. Usually, the National Regulatory Control Council
demands some evidence that the costs are quite low but no elaborated and exact calculation.
In fact, in Germany there is no threshold. Generally, all costs to comply with a legal requirement have to
be calculated and depicted with the draft law. However, the methodology allows refraining from depicting
the compliance costs if the effects of a legal requirement are evidently small and the National Regulatory
Council (Germany’s independent advisory body) agrees.
Italy: The new Decree+ Handbook in Italy does not fix a single monetary threshold. The proportionality of
analysis is met by four conditions, which constitute the set of criteria to be used for determining the
"significance" of the expected impacts. This significance may vary. In fact, for each individual
regulation/legislation under preparation, a proposal of RIA exemption shall be presented to DAGL by the

4
The currency regarding this example is the Canadian Dollar.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
20 

individual ministry, and DAGL verifies if the decision (of not carrying out RIA) is grounded or not. In other
words, on each proposed exemption from RIA, DAGL scrutinises the evidence (a simplified RIA) that
demonstrates the low expected impacts of the new legislation/regulation.
In Ireland currently, there is no specific rule as to which approach should be used when conducting a Full
RIA. It depends on the policy area, the time and resources available and the level of costs involved.
Nevertheless, in the past, a full cost-benefit analysis was considered where it is expected that the
regulations will generate costs of €50 million over ten years.5
Mexico: Mexico specifies three levels of required RIA and distinguishes between them by a combination
of both quantitative cost thresholds and qualitative judgments as to whether the regulation would have
non-negligible impacts on employment or business productivity. For ordinary RIAs, the government
requires a second test – qualitative and quantitative – that Mexico calls a “calculator for impact
differentiation”. Based on the result of a 10 questions checklist, the regulation can be subject to a High
Impact RIA or a Moderate Impact RIA, where the latter contains less details in the analysis.
Netherlands: The Netherlands approach to “filtering” regulation for RIA purposes is different again. It has
long used a two-step RIA process, where the results of a preliminary RIA are assessed by the regulatory
reform authority to determine whether a full RIA must be completed and which RIA tests must be applied
to the proposal.
Norway: Does not have an exact threshold for conducting quantitative ex ante impact assessments. All
relevant factors of new regulations should be described and assessed, and if a major impact on any sector
(including business) is expected, a quantitative analysis is required.
South Korea: The South Korean test requires quantitative RIA to be undertaken if it affects more than 1
million people, impacts greater than 10 million Won, there is a clear restriction on market competition, or a
clear departure from international standards.
In Sweden, regulators are obliged to assess several different ex ante impacts (public administration,
environment, business, gender equality etc.) to the extent necessary in the individual case and document
this assessment in an impact analysis. but the main focus is impacts on businesses.
If a regulatory proposal is expected to have “significant impacts” on businesses, the regulators have to do
an in-depth assessment of these impacts. Sweden does not have a clear definition of the meaning of
“significant impacts”, nor have a threshold value of regulatory costs in impact assessments. They have two
reasons for this:
The experience from some other countries is that an official threshold value (or leaked unofficial threshold)
may result in regulatory cost calculations in impact assessments that are below the threshold. For instance
to avoid scrutiny from a regulatory oversight body;
However, the main reason is that a threshold value is a good indicator of what is “significant impacts” on
businesses in their daily lives. The reason for this is that a proposed regulation that impacts a lot of
businesses in total may exceed a threshold, even though the impact of each business may be slim and
not perceived as a significant burden. On the contrary, a proposed regulation that may be a really
significant burden on only a few businesses, and that maybe even threat their ongoing survival, may in
total be below a threshold. Even if the regulations do not threaten the survival of the businesses, they have
to conform to the regulation for many years to come, so it´s of great importance that the proposals take
that into consideration in order to minimise the burden before the proposal have been adopted.

5
Cf. Ireland RIA Guidelines, 2005.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
  21

Switzerland: Switzerland uses a relatively complex checklist based on a set number of conditions. When
three of the 10 conditions are met, then a more in-depth analysis is required. These conditions cover a
wide variety of measures, including:
 General economic consequences
 Main stakeholders in at least 3 of the following categories: SMEs, large enterprises, workers,
Confederation and / or cantons, taxpayers, consumers, environment, other / undetermined
 Number of enterprises affected
 Administrative burden, regulatory costs
 Competition
 Degree of international openness
 Attractiveness of the economy for investment
 Ecological sustainability
 Social sustainability
 Energy consumption
If the proposal has high administrative burdens, (at least 10 000 directly impacted businesses, increased
administrative burden), the RIA must include not only a quantitative estimate of the costs of regulation, but
also a qualitative assessment by means of a SME compatibility test with a dozen companies. If 1,000 to
10 000 enterprises are directly affected by an increase in the administrative burden and / or regulatory
costs, or if firms in at least one branch or region are heavily affected, a rough estimate is recommended
for regulatory costs or a summary SME compatibility test.
New Zealand: All regulatory proposals in New Zealand must be accompanied by a RIA, unless an
exception applies. New Zealand allows for case specific exceptions (e.g. implementing international
treaties) and discretionary exemptions that require the approval of the Treasury.
United Kingdom: The UK addressed this issue by requiring a formal preliminary assessment to be
undertaken, which provides a general understanding of the size of likely regulatory costs and forms the
basis for determining what level of assessment will subsequently be required. They have a concrete system
divided by three stages of impact: (low impact, medium impact, and high impact)6:
 Low impact: equivalent annual net direct cost to business greater than GBP 5 million but less than
GBP 10 million or net present social value less than GBP 25 million;
 Medium impact: equivalent annual net direct cost to business greater GBP 10 million but less than
GBP 50 million or net present social value greater than greater GBP 25 million but less than
GBP 50 million;
 High impact: impacts greater than GBP 50 million.
Also in UK, the impacts on business, charities and voluntary sector are considered in determining whether
the RIA threshold has been crossed.
United States: RIA is required for significant and economically significant regulatory actions as defined
under Executive Order 128667 and Executive Order 13563. An economically significant regulatory action
is one that:

6 See the United Kingdom’s Proportionality guidance for departments and regulators. Available in:
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/800603/Final_pro
portionality_.pdf.
7
Cf. Executive Order 12866. Available in: https://www.archives.gov/files/federal-register/executive-
orders/pdf/12866.pdf.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020
22 

 Likely to impose costs, benefits, or transfers of USD 100 million or more in any given year;
 Adversely affect in a material way the economy, a sector of the economy, productivity, competition,
jobs, the environment, public health or safety, or State, local, or tribal governments or communities.
The United States defined “major” rules in 1981 as those which are likely to impose annual costs exceeding
USD 100 million or those likely to impose major increases in costs for a specific sector or region, or that
have significant adverse effects on competition, employment, investment, productivity or innovation.

A CLOSER LOOK AT PROPORTIONALITY AND THRESHOLD TESTS FOR RIA © OECD 2020