This Article claims that jurisdictions do not necessarily select their regulatory topics in a rational manner. It introduces criteria and a reflection framework to decide whether a specific topic should be subject to regulation.

In the early years of the Regulatory Institute, it developed a matrix with more than 1,000 jurisdictions adopting laws (states and intra-state entities like Tamil Nadu in India, Patagonia in Argentina and Colorado in the U.S.) on one ax. On the other ax, it listed topics covered by legislation. Looking at the matrix, the Regulatory Institute discovered that very few topics were covered by all jurisdictions. For most of the topics, there were only some erratic dots here and there.

There was no logic detected behind the pattern of topics or sectors being covered or not, except that some topics, such as the fight against money laundering, were pushed by international conventions into almost all jurisdictions while some other topics or sectors, like penal law, are so ancient and essential that we find them everywhere. However, generally, and for the bulk of the topics, the Regulatory Institute could not identify any reason why topic X was relevant for jurisdiction B and not for its neighbor jurisdictions A, C and D[1] with similar conditions. Even paramount topics like the prevention and sanctioning of child abuse have not found a regulatory response in all neighboring jurisdictions, though the sociocultural factors and, therefore, the prevalence of child abuse are similar. Thus, the Regulatory Institute, at the time, concluded that there is, in quite some jurisdictions, a certain lack of rationality regarding the selection of topics or sectors deemed necessary to be regulated.

This “lack of rationality” hypothesis has been to a modest extent confirmed by episodic insights into law-makers’/regulators’ considerations. According to what we read and hear, most jurisdictions “tumble” into law-making for a certain topic without any comparative analysis of costs and benefits[2]. Even where there is a comparative approach, it is often incomplete. E.g., in a 2023 public consultation, a government seemed to seek confirmation for its “risk-based approach” regarding the selection of products to be covered by regulation. According to what we understood from the necessarily simplified presentation in the respective policy paper, the government wished product categories to be prioritised in accordance with their respective risks or hazards and adopt regulation step-by-step from top-risk/top-hazard downwards[3]. However, is it meaningful to adopt regulation on high-risk/high-hazard products e.g. where…:

• …the industry already have minimised risks?
• …guidance, an industry standard, implicit control by liability insurance, a quality rating or a voluntary quality label would suffice to minimise risks?
• …the industry consists predominantly of actors unable or unwilling to fulfil legal requirements whilst the necessary enforcement capacities are lacking so that, whatever regulation is adopted, nothing changes in reality?[4]
• …meaningful legal requirements would necessarily imply a certain technical solution (the current state-of-the-art) and thus hamper more radical risk reduction/elimination by another technical solution (the potential future state-of-the-art)?
• …the risk can be much more appropriately and cost-efficiently addressed by use instruction, training and use control?
• …the high-risk products (e.g. para-gliders) are only rarely used by a few hundred persons, whilst slightly less risky products (e.g. electric scooters) are used by millions and where there is not enough legislative and enforcement capacity available to cover both categories of products?
• …with its available human and financial resources, the government can either save 500 lives per year by regulating a very high risk product category or 300 + 400 lives per year by regulating two less risky, but much less resources-demanding product categories?

These questions lead to a series of factors disturbing the logic of a pure risk-based or hazard-based approach. We claim that there can be many more situations and factors destroying the superficial convincing logic of a pure risk-based or hazard-based approach for the selection of products to be covered by regulation. More fundamentally, we claim that any single-criterion selection of topics to be regulated is short-sighted.

We seem to need a reflection framework which covers appropriately the situations listed above as well as other situations where the application of one single criterion does not fit. Before stepping into a regulation for topic X, we should compare its cost-benefit with the cost-benefit of other regulatory measures for topic X[5] and with the cost-benefit of both regulation and other regulatory measures for topics Y and Z that could be taken with the same available human and financial resources. This is necessary because law-makers or regulators can assign resources to different regulations and other regulatory measures regarding a variety of topics. To optimise their effects, the law-makers/regulators need to compare different scenarios in which the available human and financial resources are distributed to different possible combinations of measures regarding different topics; in the example case: product categories. To be able to compare, they need to evaluate the effects of these possible combinations of measures according to some criteria or goals, whilst each criterion or goal might have a different value for the respective law-makers or regulators so that the criteria need to be weighed.

When developing reflection frameworks, we face a trade-off. The more detailed the framework, the more precise and reliable the result. But the more detailed the framework, the more the respective government or parliament will need to invest in establishing underpinning data. The trade-off needs to be assessed by each law-maker/regulator individually. To respond to widely varying demands, we offer in the following a basic reflection framework that can be specified in different ways, e.g. in the ways presented further below. We know that the sharp eyes of a mathematician will detect flaws in the presented reflection framework, and even we ourselves spot some. It is our hope that the presented reflection framework is the best possible compromise between preciseness and readability for non-mathematicians.

Basic reflection framework

1. List the policy goals for your field of responsibility and attribute a total of 10 points to the various policy goals G^a to G^z (so that all goals together sum up to 10). The points serve later as weighing factors for your policy goals.

2. List the different topics to be potentially targeted by measures. Try to develop measures for these topics. At this stage, each measure should only deal with one potential topic.

3. Describe the key elements, including mechanisms for implementation, of the different measures M¹ to M^x (regulation and other regulatory measures as listed in the Handbook, Chapter 4), for the different potential topics to be targeted and estimate the respective human and financial resources needed; sum-up the value of both human and financial resources needed for a certain measure to create a single figure for the overall cost of each measure.

4. Evaluate the effects of the various measures M¹ to M^x with regard to the various policy goals G^a to G^z: to what extent do the measures improve the current situation with regard to the different policy goals for your entire field of responsibility? Attribute numbers from 1 to 10 to the respective improvement for each policy goal, 10 being the maximum. 10 stands for a 100 fulfilment of the respective policy goal, and 1 for “no change”. The number stands for the “improvement value”, thus the extent to which the fulfilment of a certain goal is improved by the measure. It is possible that the effects will be negative, in which case you need to attribute a negative number.

5. For each measure M¹ to M^x, please fill in the following efficiency formula:

M¹:         Improvement value for G^a x Points for G^a + Improvement value for G^b x Points for G^b …           Cost

          = E¹ (overall efficiency of M¹)

M²:       Improvement value for G^a x Points for G^a + Improvement value for G^b x Points for G^b …               Cost

          = E² (overall efficiency of M²)

         … (to be continued for each measure)

 

6. Now please also calculate the improvement values of the measures M¹ to M^x:

M¹:   Improvement value for G^a x Points for G^a + Improvement value for G^b x Points for G^b  …

         = MIV¹ (measure improvement value of M¹)

M²:   Improvement value for G^a x Points for G^a + Improvement value for G^b x Points for G^b  …

        = MIV² (measure improvement value of M²)

       … (to be continued for each measure)

7. Assess which costs your administration can cover in total. Try to find a combination of measures that:

• stays below your cost limit, but
• has the highest overall improvement value (or overall benefit). The overall improvement value is the sum of the measure improvement values of the selected measures (MIV¹ to MIV^x).

It is probably best to select the most efficient measures (see results of Step 5) first and go down to integrate more measures into your combination of measures. Only exceptionally and to absorb otherwise useless resources, the optimal combination will skip one of the most efficient measures.

8. You should now check whether the available resources are sufficient for the combination of measures selected. The first criterion is: Is there full convertibility of human and financial resources so that one can replace the other? If there is no full convertibility of human and financial resources: check whether the distribution of available resources (e.g. 50 full-time staff and 1.200.000 $) fits with the combination of measures retained under 7. If not, look for the next best combination of measures under 7 that fits with the distribution of available resources.

9. You might also have different types of financial resources. Please check whether the types of financial resources you have available fit with the combination of measures retained under 7 – there might be budgetary rules that create an obstacle. If the combination does not fit, look for the next best combination of measures under 7 that fits with the budgetary rules.

10. Human resources (or better: humans) are not identical and convertible either. Check whether the qualification of available or recruitable staff fits with the combination of measures retained under 7 – you might need specialists for some of the measures that you cannot find. If the combination does not fit, look for the next best combination of measures under 7 that fits with the staff you have available or whom you can recruit.

You have now identified a combination of measures that produces very good results in view of your or your masters’ policy goals. Each measure targets only one topic. Implicitly, you have also determined the topics to be addressed by measures. Thus you have selected topics by the mega-meta-criterion: how best to pursue your policy goals? This is already a huge step that brings you, in terms of rationality, into the top 3% of law-makers/regulators worldwide, if not even the top 1%; we hardly ever observe any kind of comparative considerations when it comes to the selection of regulatory topics.

Still further fine-tuning and thus possibly even slightly better results can be obtained by the following specifications, depending on the case.

Specification 1: Merger of measures.

So far, we have assessed the efficiency and improvement value of each measure, whilst each measure dealt only with one single topic. But you might have some measures in your list that are similar by nature, though regarding different topics (e.g. guidance for manufacturers of medicines and guidance for manufacturers of medical devices). Check now whether a merger of measures is possible and whether the merger reduces costs (e.g. because the manufacturers are partly identical) or improves the results for the various policy goals. Insert the modified figures in the calculation above and check whether another combination of measures is now the best.

Specification 2: Uncertainties

You might find it difficult to assess the effects given that you do not know how the situation will evolve in various regards. If so, please develop different scenarios and make your assessment based on each scenario individually. As a second step, please attribute likelihoods to the different scenarios. As a third step, multiply the improvement values of Scenario 1 by the likelihood of Scenario 1, the improvement values of Scenario 2 by the likelihood of Scenario 2 and so on. Insert the sum of these products into the baseline framework.

Where exactly? This depends on the scope of your application of scenarios. The scenario technique can be used at different levels:

• the same scenario is applied to all measures;
• the same scenario is applied only to certain measures (whilst others are possibly applied to other measures);
• the same scenario is applied only to a single measure (whilst others are possibly applied to other measures);
• Each scenario is applied only for the assessment of one improvement value for one specific goal.

Apply scenarios as limited as possible. If there is uncertainty only about one improvement value for one specific goal, limit the scenario building to cover this uncertainty alone.

Specification 3: The time dimension of effects

The effects of a measure on the fulfilment of the G^a to G^z might vary over time. To get a grip on this issue, estimate the effects for the different years (or even months) to come and calculate the average as improvement value for Step 4. If you think that time-wise, remote results are less reliable or valuable, e.g. because the situation will have changed or new/alternative regulatory measures will become available, apply a devaluation factor for the remote years. The devaluation factor may even increase with each additional year, e.g. by 10%.

Specification 4: The time dimension of resources

So far we have assumed that you can distribute the available resources freely over the years. However, this is not realistic in quite many jurisdictions and government departments. To remedy this planning deficiency, check whether the selected measures can be executed as planned with the distribution of resources over the years which is the most likely/realistic one. If not, use two approaches and compare the results thereof:

• adapt the measures so that they fit with the most likely distribution of resources over the years and reevaluate the measures in their adapted form. Recalculate the various improvement values, including the overall improvement value (overall benefit);
• try to find the best combination of unchanged measures that fit to the distribution of resources over the years and calculate the overall improvement value.

Choose now the combination of measures which has the best overall improvement value. Is it the one developed by the first approach? Or the one developed by the second approach?

Specification 5: What if efficiency counts, not overall effects?

To ensure the best possible attribution of resources among departments, your department might be subject to comparative efficiency assessment. Therefore, or simply for regulatory policy reasons, your department might be called upon to choose measures according to efficiency alone, top down, possibly until you reach a certain minimum efficiency. This approach implies that you will not necessarily spend all the resources attributed to your department. As a consequence, the overall improvement value might be lower than what it could be.

If you have to follow this approach, go back to Step 5. Place the measures in the order of their overall efficiency and select them top-down until you reach your budget limit or the applicable minimum efficiency threshold, whatever comes first. The efficiency measurement of your inter-department management might of course deviate from our Step 5.

Specification 6: Criteria, not goals

Your jurisdiction or department might be used to apply criteria instead of goals. If so, simply replace goals with criteria and assess the measures by criteria. The reflection framework should be operational in very much the same way, as long you respect a ceiling for the points to be attributed under each criterion (e.g. 10)[6].

[1] However, it has to be conceded that the similarities among neighbors were stronger than between remote jurisdictions. Thus, there is also some spreading of ideas from neighbor to neighbor.

[2] Admittedly, quite many jurisdictions apply a non-comparative analysis of costs and benefits, but this only eliminates totally inefficient measures.

[3] Text of the respective policy paper: “We will examine approaches where products are categorised by their hazards and consequent risks, falling into one of several defined risk levels. Categorisation criteria could include: the likely impact should harm be caused, the expected user group, the likelihood of harm being caused, the environment it is likely to be used in, and the cumulative effect of risks. For example, where products, or an element in them, could cause death or serious injury, a higher category would be allocated. The system would be agile and responsive to changes in a product’s risk level over time and allow for re-categorisation of a product where evidence suggested this was appropriate. It would also encourage innovation to ‘design out’ hazards, with manufacturers thereby potentially benefiting from reduced regulatory requirements.”

[4] In such a situation, the few compliant operators are unfairly penalised.

[5] The first chapters of the Handbook “How to regulate?”, namely Chapters 4 and 5, provide an analytical framework for this comparison.

[6] Not to respect the ceiling would mean that you torpedo the weighing of the criteria.