a. Cost–benefit analysis

From an economic perspective, the relatively simple criterion for deciding which legal intervention is indicated concerning legal but unhealthy products or services is whether the social costs of the use of these products or services are higher than the social benefits. Of course, other criteria could be employed as well if one were to follow other approaches. A legal intervention could in that case also be based, for example, on the trade-off between, on the one hand, consumer choice and freedom and, on the other hand, the paternalistic need to protect consumers. In this contribution, we employ the law and economics perspective only. One lesson from that perspective is that even if a particular activity may create social costs (such as a health risk), this activity could, from a societal perspective, still be overall desirable on the condition that the total social benefits would outweigh the total social costs of that particular activity. A particular group of citizens could still desire to use these products or services, notwithstanding the inherent health risks related to these items, simply because they increase the utility of these people. This also implies that, from a societal perspective, imposing a general ban on the use of these products or services is not usually considered the best way. It is important to realise that the cost–benefit analysis should be based on social costs and social benefits, not on private costs and private benefits. Here the distinction between “harm to oneself” and “harm to others” is again important. In the former situation, there is only a relatively small difference between private and social costs, which results in fewer reasons to interfere with consumer choice. In the latter situation, however, the gap between private and social costs is much wider, which provides more economic reasons to interfere. Also, on the benefits-side, there can be differences between the private and the social perspective. Some products only provide benefits to the user, whereas others also provide benefits to others (so-called “positive externalities”). An example of this latter situation is vaccination against infectious diseases. The vaccine not only protects the vaccinated person, but also contributes to herd immunity, which benefits non-vaccinated people as well. Other pharmaceutical products, on the other hand, only produce private benefits, such as medicines against headache. Both vaccines and other pharmaceutical products may have negative side effects for the user.

As said, many activities create costs not only for the user but also for third parties. Those costs are not borne by the entity creating those costs. This is the issue known in economic theory as the external effect or externality.Footnote ² This is the case with, for example, smoking or the use of plastic wrapping. The more the costs of production, consumption, disposal, et cetera, are included in the price (which should be increased to include these costs), the more externalities can be internalised. The same holds for the benefits: the more the benefits to others are reflected in the price (which should then, of course, be lowered), the smaller the external effect. As long as the social benefits of an activity are larger than the social costs, it is, from an economic perspective, not desirable to totally ban the particular activity.

If the social costs are higher than the social benefits, first the question arises as to whether the activity can be controlled by taking optimal precautions. If, however, even when taking optimal precautions the social costs would be higher than the social benefits, the activity has to be considered socially undesirable and should, therefore, be controlled. One problem is that there may be a substantial difference between the social costs and benefit calculus and the calculus of private costs and benefits by the entity that creates the risk. As a result, that entity may chose different activity and care levels than are socially optimal. It is particularly in this situation that the law should intervene to internalise externalities. This will result in the risk creator adopting more appropriate activity and care levels.

These first economic considerations lead to a first observation: the mere fact that a product or service can create (health) risks is as such not yet an (economic) argument to prohibit the particular product or activity. It is, however, important to provide incentives to the actor creating the risk to adopt optimal care and activity levels.

b. Safety regulation or liability rules?

It needs to be determined which instruments can provide appropriate incentives for the actors involved to adopt those optimal care and activity levels, which comes down to trading off the effects of the safety regulation or liability rules. Safety regulation implies that the government ex ante determines a specific care level.Footnote ³ That care level, incorporated in a regulatory standard, subsequently has to be complied with by the risk creator. In a case of non-compliance, administrative or even criminal sanctions can be applied. An alternative is to use the preventative function of tort law. Via an ex post liability suit, tort law can provide the risk creator with ex ante incentives for prevention. Which is the appropriate measure in the context of unhealthy products or services?

To answer this question, the four criteria developed by Shavell in his seminal article are of importance.Footnote ⁴ A first criterion is the availability of information. In particular cases, the government may have better information to determine optimal care levels than the parties involved. The government may have the advantage of being able to use economies of scale. A general governmental assessment avoids every actor individually having to undertake the same research, thereby duplicating efforts and costs. Actors could also lack the necessary incentives to undertake that research themselves. That would especially be the case if the results could be copied by others (eg competitors). It is the well-known free-rider problem that reduces the incentives to innovate.

A second problem is that liability rules only provide incentives to employ an optimal level of care if the insolvency risk can be controlled. If the potential damage is higher than the individual wealth of a risk-taker, liability rules may lose their deterrent effect. The risk-taker will, in that case, consider the accident as one with a maximum value equal to their wealth. The potential injurer cannot lose more than their individual wealth. Under-deterrence may be the consequence. The advantage of safety regulation is that the government can impose a duty on the risk-taker to follow particular care levels that can also be enforced via public sanctions, including non-financial sanctions that are immune to the insolvency risk.Footnote ⁵

A third criterion concerns the question of the extent to which it is realistic that potentially liable actors would really be confronted with a liability suit. For instance, this may not be the case if the damage is widespread among a large number of victims. The result will be that every individual victim only suffers relatively small losses, as a result of which there are insufficient incentives to file a lawsuit. This is the well-known problem of rational apathy. Problems can also arise if victims have difficulties proving a causal relationship or if there is a long time lapse between the tort and the occurrence of the damage (so-called latency).Footnote ⁶

A final criterion advanced by Shavell relates to administrative costs. The administrative costs of liability rules are, in comparison to safety regulation, lower. This is the case because in liability law costs only arise if liability suits are really brought, but the preventative function of tort law works even without liability suits being brought.Footnote ⁷ Settlements are another cost-reducing factor. Safety regulation, to the contrary, always has high costs.Footnote ⁸ It is costly to set safety standards for all kinds of different situations. Moreover, those safety standards also need to be enforced via control and monitoring, and if a violation is established, a sanctioning mechanism has to follow.

c. Application to unhealthy products and services

When applying those criteria to unhealthy products and services, there are overwhelming reasons for an ex ante approach via safety regulation. First, determining an optimal care level can be complicated for producers. In that respect, it is important to recall that health damage cannot only be caused by large capital-intensive industries with impressive research divisions. Small and medium-sized enterprises (SMEs) can also bring particular products or services to the market that could potentially endanger health. Hence, producers might not behave optimally. As we will argue in more detail in Section II.3, under strict liability the producer decides which care measures to take, so they have to assess the optimal level. Incomplete information may result in incorrect decisions. When a negligence rule applies, the problem may even be worse because there it is not the producer but the judge who will determine the optimal care level.Footnote ⁹ In practice, judges have even less information than producers and are also not equipped to determine the optimal care level in the case of an unhealthy activity. Precisely because the government can use economies of scale (via government-funded research agencies), it could determine the optimal care level in case of unhealthy activities in a collective manner. The result of that research could subsequently be passed on to the market via safety regulation. This result generally holds true for both “harm to oneself” and “harm to others” scenarios with information about the effects on bystanders, for example, being even more difficult to obtain.

An insolvency risk is also realistic in the case of unhealthy products or services, in particular if also affected third parties need to be compensated. Think again of the SMEs. The damage they cause can cost more than the value of the enterprise itself. But even larger enterprises are regularly organised as limited liability companies. This implies that companies can use the limited liability to exercise dangerous activities without being fully exposed to the damage that they may cause. In some cases, risky activities are deliberately incorporated into a company, thus de facto creating an insolvency risk.

In addition, the chance that a liability suit will really be brought in the case of unhealthy activities might be small. The problem may arise that health damage often only appears many years after the harmful activity took place. As a result, in many cases it may be impossible to identify a liable producer. In other cases, it may appear impossible to prove causation. Especially given the long-tail character of health damage and the causation issues, there is a serious risk that liability suits will not be brought, and anticipating this may hamper a preventative effect. The danger of not confronting producers with a liability claim is particularly present in the “harm to others” scenarios. Take the example of passive smoking. There might be many passive smokers, but their harm is more remote and causality more difficulty to prove. Harm might, furthermore, spread out but become smaller as it spreads out. The smaller the loss per individual victim and/or the more difficult it is to prove causation, the more likely it is that individual victims will refrain from bringing a claim. In other words, they may stay rationally apathetic.

The only counterargument seems to be that ex ante safety regulation is relatively costly to administer compared to liability. However, given the fact that the first three criteria overwhelmingly point to the benefits of safety regulation, the conclusion nevertheless must be that health-threatening activities should primarily be subjected to ex ante safety regulation by the government. This holds even more in the “harm to others” cases than in the “harm to oneself” cases, because of the above-discussed increased challenges for liability where “harm to others” is prevalent. We write “primarily” since the priority of safety regulation does not imply that there would be no more room for the application of liability. However, before we analyse the possible role of liability, we will first briefly discuss the problem that regulation may not always serve the public interest, as we assumed above.

a. Liability from an economic perspective: prevention and risk spreading

Whereas many lawyers consider compensation of harm to be the most important goal of liability rules, economists see compensation mainly as a means to serve the goals of prevention and risk spreading.Footnote ¹⁴ Liability rules are considered to be instruments that can be employed to internalise negative externalities. They provide incentives to an actor to take optimal preventative measures; that is, measures for which the marginal costs are lower than the marginal benefits (ie the reduction of expected harm). This is the core of the preventative function of liability rules. It corresponds to the simple economic notion that liability has the advantage that it allocates the damage to the actor that created the risk, so that they internalise that damage. As said, this does not imply that the actor should necessarily ban the activity altogether, nor that they should reduce all damage at all costs. Liability rules will incentivise the actor to follow optimal activity and care levels.

In addition, liability rules can also contribute to loss spreading – both when the liable party is insured against liability and when it is able to spread the loss itself. One can think about an enterprise that would be able to pass on the costs of liability in the prices of products or services, as a result of which the liability costs would be spread over the large community of consumers.Footnote ¹⁵ Loss spreading is desirable given the phenomenon of the marginal decreasing utility of money. This implies that the extra utility from one extra euro will be reduced as someone possesses more money. The first euros will be spent on primary needs and are, therefore, worth a lot. As someone has more money, the additional euros are spent on less crucial needs and therefore deliver less utility. This implies that damage of a certain magnitude that has to be carried by one person leads to a loss of more utility than equal damage that is spread over more individuals. Loss spreading, therefore, also contributes to social welfare.Footnote ¹⁶

b. Negligence or strict liability?

At first blush, one could think that a negligence rule is inappropriate, as we refer to legal products that still cause health damage. This raises the question of the extent to which there could be negligence or wrongfulness if bringing the product or the service to the market is as such not illegal. We will argue that the mere fact that an activity is legal does not imply that the negligence rule can no longer play any role. Therefore, we will first explain how law and economics scholars view negligence and strict liability; then we will address relevant differences between both regimes.Footnote ¹⁷

From an economic perspective, liability rules are viewed as a system to provide desirable behavioural incentives. This concerns incentives to follow optimal care and activity levels. The more care an actor takes in the exercise of a particular activity, the more the expected damage will decrease. However, a higher care level also creates higher costs of prevention. As long as the marginal costs of more preventative efforts are lower than the marginal benefits thereof (in the sense of a decrease in expected damage), it is desirable that those extra preventative efforts are taken. When the marginal costs of care, however, become higher than the marginal benefits, the additional preventative efforts are no longer desirable. The turning point is referred to as the optimal care level.

The crucial difference between negligence and strict liability from an economic perspective is the following: under negligence, a particular level of care will be determined as the due care level. When the potential injurer follows that due care level, they will not be held liable. If, on the contrary, they follow a lower care level, there will be liability. Ideally, the due care level required by the legal system would equal the optimal care level. In that case, actors would be incentivised to make a socially desirable decision concerning the costs and benefits of additional care. This implies that under a negligence regime the mere fact that there is harm does not mean that there is negligence. The optimal care level is found by weighing the marginal costs of prevention and marginal benefits. Generally speaking, at this optimal care level there will still be accidents and, hence, damage. Further reducing the expected damage beyond the optimal care level would lead to higher marginal costs compared to the marginal benefits. For example, in theory, all deadly traffic accidents could be avoided by introducing a maximum 'speed of ten kilometres an hour on highways and enforcing that as well. Such a preventative measure would, however, be considered to be too expensive. The mere fact that a traffic accident leads to a fatal incident does not mean that there is negligence on the part of the injurer. Even if optimal care levels are followed, fatal accidents can still take place. When the potential injurer followed the optimal care level and nevertheless damage occurred, under a negligence regime the victim will have to bear the damage themself. This may lead to a demand for insurance.Footnote ¹⁸ The due care level required from the potential injurer will, within the framework of the application of the negligence rule, be determined by the judge. Strict liability works in a fundamentally different way: as soon as there is a causal relationship between the behaviour of the injurer and the damage, the injurer is bound to compensate, irrespective of their care level. This implies that, under strict liability, it is the potential injurer themself who determines how much care they take. Under negligence, it is in principle the judge who determines the duty of care.Footnote ¹⁹ This important difference, of course, has consequences for determining the different characteristics of negligence and strict liability.

There are several factors that determine whether negligence or strict liability is better suited to providing the desirable incentives. The first factor is information costs. It is possible that, in fixing the due care level, mistakes are made due to information problems. As a consequence, the due care level required by the judge (under negligence) or the care level chosen by the injurer (under strict liability) may deviate from the optimal care level. If the judge has the best information to determine optimal care levels, then negligence may be the optimal rule. If, to the contrary, the potential injurer is better placed to weigh marginal costs and marginal benefits, then strict liability is more appropriate. The weighing of those costs and benefits should ideally be done by the actor who possesses the best information.Footnote ²⁰

The second relevant factor for choosing between negligence and strict liability is the activity level. It is desirable that an activity is only exercised when it provides more benefits than costs. When a particular actor does not have to carry all of the costs of their activity, there is a danger that they will adopt a too-high activity level. An injurer will only take an optimal activity level under strict liability, because they bear the full costs of the activity. Under a negligence rule, the potential injurer is freed from liability if they follow the due care level, so they do not bear the expected accident losses and, hence, will choose a too-high activity level. The potential victim, on the other hand, will adopt a too-high activity level under strict liability, since in that case the injurer will be liable to compensate all damage. It is not possible to formulate liability rules that provide incentives to follow optimal activity levels for both the injurer and the victim.Footnote ²¹

The third factor is dimensions of care. Liability rules not only provide behavioural incentives for features of care that are observable and can be controlled. An advantage of strict liability is that it incentivises the potential injurer to take all possible measures into account as far as they can optimally reduce the accident risk, whether those dimensions are observable or not. As strict liability forces the potential injurer to compensate all damage, they will take all efficient measures to reduce the accident risk. Under negligence, to the contrary, the potential injurer will only take into account the measures of care incorporated in the due care level (ie those elements of the duty of care that can be observed and controlled by the judge).

The fourth factor is insolvency. Strict liability may create more problems in case of insolvency than negligence. The reason for this is that, under strict liability, there is under-deterrence as soon as the total damage is higher than the wealth of the potential injurer. Under negligence, a potential injurer will still take due care as long as the costs for taking due care are lower than their individual wealth. Following the legally required due care level is still a way for the potential injurer to avoid liability.Footnote ²² The insolvency risk under strict liability can be remedied by accompanying it by another mechanism; for example, by imposing a duty on the actor to provide solvency guarantees (eg mandatory liability insurance).

The fifth factor is loss spreading. Under strict liability, it is the injurer who compensates the entire damage; under negligence, the loss remains with the victim. Depending upon who can better spread the loss, this may give an advantage for one of the two rules.

The last factor is administrative costs. Negligence is relatively complex as it requires the duty of care to be determined, as well as whether the injurer violated that duty of care and whether they can be blamed for that violation. Strict liability, on the other hand, does not require these difficult assessments, but can lead to more court cases as victims only have to prove that the behaviour of the defendant caused their loss, not that any violation of a due care standard took place. Generally, law and economics scholars hold that strict liability will result in lower administrative costs, as the outcome of such cases is more predictable and will therefore result in (cheaper) settlements and in fewer tort cases.Footnote ²³

When applying these criteria to the case of unhealthy products or services, strict liability prevails. An advantage of strict liability is that the potential injurer (a producer of particular products or a provider of services) may have better possibilities of loss spreading than the victim. They will have the possibility to pass on the damage to all consumers via the price of the product. As was indicated in Section I, the broad concept of legal but unhealthy products and services may, in practice, cover a wide variety of different situations. However, none of the situations would be such that, for example, the risk would be created by an occasional non-professional (private) manufacturer (in which case there may also be an argument in favour of a negligence rule). Most of the cases that we envisage here (eg tobacco manufacturing) concern professional producers. Those will, as repeat players, have better information on the products and services that they bring to the market than the judge. This makes strict liability even more attractive. The argument may, of course, be stronger when damage is caused by large, well-equipped multinationals rather than by SMEs. The advantage of strict liability becomes even stronger when dimensions of care play a role that cannot easily be controlled by external parties such as a judge. This may especially be an issue in the case of services. In addition, the better activity level and the lower administrative costs are arguments in favour of strict liability.

It is important to add a defence of contributory negligence to the strict liability rule in order to also provide incentives to follow efficient care levels for the potential victim. In some cases, there may even be an information advantage on the side of the victim. The victim may know particular features of themself that may make them refrain from the use of a particular product or service. Someone with a weak back should not be able to hold the organiser of a bungee-jumping event liable for damages as a result of increased back pain after the jump. In such an example, the victim is clearly the cheapest cost avoider who can avoid the damage by not engaging in that activity in the first place. In some cases, the defence of assumption of risk could apply. This would be an issue for individuals who started smoking at a time when the dangers of smoking were general knowledge and who subsequently would wish to file a lawsuit against a tobacco producer.

The above-described framework is based on “harm to oneself” situations because it considers a producer vis-à-vis the consumers who suffer losses. A more nuanced look is necessary for the “harm to others” cases: in these cases, as we have explained above, it is doubtful whether the “harmed others” will start a lawsuit in the first place. Under strict liability, such a “low probability of being held liable” may result in the producer taking less care than is optimal, because they do not incorporate the “losses to others” if those others will not sue. Under negligence, on the other hand, the court could base the due care standard on the full losses; that is, the “harm to oneself” of the consumers and the “harm to others” of bystanders. If producers would indeed take this due care level, the outcome would be better than under strict liability where they take too little care. In order for the producers to be induced to take this due care level under negligence, the costs of taking due care have to be lower than the expected liability when not taking due care. This will only be the case if the fraction of “harm to oneself” cases is large enough to create enough expected liability to make it worthwhile to take due care. So, in extreme situations with (almost) no “harm to oneself”, neither strict liability nor negligence will work, because not enough lawsuits will be filed. In somewhat less extreme cases, negligence may already provide enough incentives to take optimal care if due care is set at the optimal level and expected liability arising out of the cases filed by consumers outweighs the costs of taking due care. Strict liability only provides the correct incentives if all losses (ie “harm to oneself” and “harm to others”) are included in expected liability.

c. Liability and safety regulation

d. Incentives for innovation

One aspect that Shavell does not deal with in his general discussion concerning the economics of accident law is that liability rules can also affect incentives to innovate. This is especially of interest in the case of products with known health risks that can be reduced via supplementary research. It could also play a role in the case of unknown health risks. Social welfare can be increased when the social costs related to products or services are reduced in the long run. The extent to which liability rules provide incentives to undertake research into (especially unknown) risks and to innovate is therefore an important criterion for determining whether liability rules or regulation are the most suited instrument.

If liability rules free producers as long as they follow the existing safety regulation or the legally required due care standard, producers will lack any incentive to undertake supplementary research into safety risks. Those incentives for innovation can only be provided if producers are to some extent held liable for those negative externalities.

If a new product appears on the market, it is possible that the related health risks will not be fully known. It is possible that the long-term effects may not have been sufficiently studied and analysed. A liability for unknown risks can increase the so-called “state of the art” as far as technology is concerned. A condition is, of course, that the potential injurer is able to exercise that research or at least is able to hire someone to do that research for them.Footnote ⁴⁰ Finsinger and Simon write in this regard related to their analysis of the development risk defence within the framework of product liability: “History teaches that it is the firms who first collect hints, and then acquire evidence of such risks. … Producers first learn about development risks”.Footnote ⁴¹

As far as the incentives are concerned, a distinction should again be made between strict liability and negligence. Under strict liability, engaging in supplementary research has two advantages for the potential injurer:

• If the research shows that the activity is less dangerous than previously thought, the injurer can manage with a lower level of care. The expected value of this advantage consists of the probability that the activity turns out to be less dangerous multiplied with the savings in prevention costs, as now less care has to be taken.

• If the research shows that the activity is more dangerous than previously thought, the actor will take more care. As a result, the primary accident costs (consisting of the costs of prevention and the expected damage) reduce as, under the old (lower) care level, too much damage was caused. The expected value of that benefit therefore includes the likelihood that the activity will be more dangerous multiplied with the reduction in primary accident costs as a consequence of the higher level of care.

If those two benefits together are higher than the costs of research, this implies that the producer concerned will effectively engage in the research. This is also a socially desirable consequence. In the case of strict liability, there is no difference between the private benefits and the social benefits, since the potential injurer bears all of the costs but also enjoys all of the benefits of supplementary research. Therefore, the potential injurer will take a socially desirable decision concerning additional research.

Under negligence, a problem may arise: if the potential injurer is only considered negligent if they take less than due care, then, given the information they have, they will receive insufficient incentives to innovate. If, on the basis of new information, the injurer establishes that the activity is less dangerous than they first thought, they will reduce their care level, which is socially desirable (as was the case under strict liability). If, however, they discovered that their activity is more dangerous than they first thought, they would have to increase their care level. This is desirable from a social perspective since the reduction in expected accident costs is larger than the increase in care costs. However, it does lead to an increase in the private costs of care, which the potential injurer could have prevented by not engaging in additional research. From this, it follows that the private benefits of information are lower than the social benefits. There can therefore be situations where the social benefits of information are higher than the information costs but the private benefits are not. In those circumstances, the potential injurers do not receive sufficient incentives to engage in the necessary research. This problem does not arise if the judge, in determining the due care level, does not examine what the injurer knew, but what they had to know. In other words, the negligence rule should be interpreted in such a way that an actor is only liable if they take less care than optimal, assuming an optimal collection of information. In that case, the private and social benefits become equal again.

If following safety regulations is a full defence in a liability claim, the same problem arises as in a case of a negligence rule that only examines what the injurer knew and not what they should have known. This implies that a rightly structured negligence rule can fulfil an important role in supplementing safety regulation. One has to avoid the situation that producers, who do their own additional research, create their own liability when they suddenly discover that their products or services are more dangerous than they originally thought. The latter could obviously reduce the incentives to engage in supplementary research. Therefore, the duty of care under negligence should be fixed according to what producers should know. This constitutes an important argument in favour of a so-called development risk defence. This defence avoids the situation where producers either do not engage in any research or do not publish the results of that research (if it was not advantageous to them). If the research costs of finding out unknown risks are higher than the previously mentioned advantages related to the adaptation of the care level and the new information, the potential injurer does not have to know the specific danger. Excluding liability for such a development risk would again align the private costs and benefits of research better with the social costs and benefits.

A smart liability system should create efficient incentives for engaging in subsequent research in combination with effective safety regulation. One of the possibilities for such a smart mix is that a producer would generate information on new risks, which they could subsequently share with public control agencies on the condition of (partial) immunity against liability for old cases.