Precaution Against Environmental Accidents and Liability Rules for Damages

Abstract

Economic activities sometimes cause accidents that lead to environmental pollution and serious harm to society. Firms should take precautions against accidents. In this chapter, we explain how liability rules for damages give firms incentives to adopt precautionary measures. We also discuss how the socially optimal level of measures against accidents should be set and what is the optimal liability rule.

The stranding of the Exxon Valdez in Alaska in March 1989 resulted in the largest oil spill in U.S. history, and the losses caused by the accident were extensive; it polluted about 2,000 km of coastline, killed hundreds and thousands of wild animals including seabirds, sea otters, and seals, and contaminated marine products, severely affecting the regional economy. According to Wan and Chen (2018), at least 80% of oil-tanker accidents involve human errors due to, for example, fatigue from overworking and lack of expertise.

Human factors also cause environmental accidents other than oil spills. The accident at Fukushima Daiichi nuclear power plant owned by the Tokyo Electronic Power Company (TEPCO) and subsequent radioactive contamination in 2011 were triggered by the tsunami after the Great East Japan Earthquake. The tsunami hit the plant and damaged its equipment, resulting in the loss of power at the plant which caused the contamination. It should be noted, however, that even before the earthquake, a large-scale earthquake had been predicted to hit the country, possibly causing a nuclear accident.

Proper countermeasures had been called for. Nevertheless, they had not been implemented where necessary. For example, emergency power supply facilities and fuel tanks were installed in coastal areas 4 m above sea level while countermeasures against a tsunami of up to 6.1 m had been assumed to be necessary. It is pointed out that a major factor of the nuclear accident is the failure to take sufficient precautionary measures. The effects of radioactive contamination from the accident are still present, incurring enormous damages.

As the TEPCO’s accident demonstrates, economic activities sometimes cause accidents that lead to environmental pollution and serious harm to society. Firms should take precautions against accidents. At the same time, institutional designs are necessary to give them incentives to adopt precautionary measures, given that the adoption incurs a significant financial burden to firms.

Liability rules for damages, which establish the liability of the injurer who caused an accident, give firms incentives for precaution against accidents. What liability rules for damages are socially desirable? To answer this question, in this chapter, we consider: (1) How should the socially optimal level of measures against accidents be set? and (2) What institutional designs are desirable to achieve that level of measure?

1 Optimal Precaution and Optimal Liability Rules for Damages

For a more detailed discussion, see Shavell (2004).

1.1 Social Costs and Optimal Precautionary Measures

Avoiding accidents incurs various costs. For example, accidents due to machine malfunction can be prevented by the installation of safety devices that notify the operator when an abnormality occurs, but this countermeasure requires installation costs. The more precaution is taken to reduce accident probability and the harm caused by the accident, the greater the costs for implementing countermeasures.

What if firms are mandated to reduce the probability of an accident occurring to zero, and to take measures to reduce the damages to zero even when an accident occurs? The regulation would result in huge costs, making it difficult for firms to continue their production activities. Given that the production of goods has positive aspects that benefit society, it is not socially desirable to reduce harms caused by accidents to zero.¹

The social costs of an accident are the sum of the amount of damages and the cost of measures against accidents. Because accidents are stochastic phenomena, the expected value of damages is used as an index for the damages caused by an accident. Even when damages caused by two accidents are the same amount, the accident damage that occurs with a 10% probability and the accident damage with a 1% chance of occurring are not the same because the latter is smaller than the former. That is: (the expected value of damages caused by an accident) = (the probability of the accident) × (damages caused by the accident). Using this equation, let us consider the following example to examine the level of measures that minimizes social costs, i.e., the socially optimal level of measures.

Table 6.1 shows the cost of measures, accident probability, and expected damages caused by an accident and its social costs, and they are shown in accordance with three levels of measures that the firm can take: none, medium, and high. Here we examine the case in which the probability of an accident changes depending on the level of measures taken by the firm. We assume for simplicity that damages caused by an accident are constant irrespective of the level of measures that the firm takes. For example, if the firm takes the “medium” level of measures, the cost of measures is $220,000 and expected damages are $2 million (10 million × 0.2), resulting in social costs of $2.22 million.

Table 6.1 Firm’s precautionary measures and social costs

As shown in the table, the socially optimal level (i.e., the level that minimizes social costs) is the “medium” level of measures. Raising it to the “high” level will lower expected damages by $1 million while increasing the cost of measures by $1.03 million, resulting in an increase in social costs by $30,000. This means that it is socially not optimal for the firm to take the “high” level of measures; the additional cost of measures exceeds the expected damages that will be reduced, resulting in an increase in social costs.

1.2 Firm’s Precautionary Measures and Cost Under Liability Rules

Three liability rules can be applied to firms: no liability rule, strict liability rule, and negligence rule. Under the rule of no liability, the firm is not liable for damages caused by the accident. Strict liability is a rule that requires the firm to compensate for damages when an accident occurs regardless of whether the firm is found to be at fault. Under the negligence rule, minimum standards for the measures that firms must take are established by law in advance. If an accident occurrs because a firm fails to meet the standards of measures, then the firm is found to be at fault and obligated to compensate for damages. However, the firm is deemed to be nonnegligent and exempt from liability if it satisfies the standards.

Based on Table 6.1, we computed the firm’s expected cost under these liability rules and presented in Table 6.2. The firm is not liable for damages under no liability, so it just bears the cost of measures as shown in the second column of Table 6.2. Given that the firm has an incentive to choose the level of measures whose cost is minimum, it chooses to adopt “none” of measures, which means that the socially optimal level of measures is not achieved. Under the strict liability rule, the firm’s cost is the sum of the cost of measures and expected damages, which corresponds to social costs in Table 6.1. Strict liability gives the firm incentives to take the “medium,” i.e., the socially optimal level of measures, as shown in the third column of Table 6.2. Here, choosing “none” is not desirable for the firm; by switching from “none” to “medium,” the firm can reduce the total cost, as expected damages decrease by $500,000 even though the cost of measures increases by $220,000.

Table 6.2 Firm’s precautionary measures and cost under the liability rules

The firm’s cost under the negligence rule is shown in the fourth column of Table 6.2. In making this computation, we assume that the legal standard of negligence is set at “medium,” i.e., the level that minimizes social costs. If the firm chooses to take “none” of the measures, it will be found negligent and bear the cost of measures and expected damages. The firm will not be found negligent, however, if it chooses either the “medium” or “high” levels of measures. Then, the firm only bears the cost of measures. Given that the firm chooses what minimizes its cost burden, the firm takes the “medium” level ($220,000). The socially optimal level of measures is achieved also under the rule of negligence rule.

1.3 Which Liability Rule is Socially Desirable?

So far, we saw that the socially optimal level of measures is achieved both under the strict liability rule and the negligence rule. Recall, however, that the firm is liable for all damages under the strict liability rule whereas the negligence rule does not impose damages on the firm if it is found nonnegligent. In that case, the victim is not entitled to compensation for damages. Therefore, compared to the strict liability rule, the negligence rule tends to incur more cost to the victim and less cost to the firm. The strict liability rule may thus be considered more desirable than the negligence rule from a fairness perspective. The strict liability rule is superior to the negligence rule from perspectives other than fairness as well. Below we explain the advantages of the strict liability rule compared to the negligence rule.

• (1) Under the negligence rule, the government needs to estimate the socially optimal level of measures that should be adopted by firms and set standards of negligence accordingly in advance. To do so, the government needs to know possible damages caused by an accident as well as accident probability when firms take precautions, which is not an easy task. In addition, the more accurate predictions they attempt to obtain, the more information they need, which incurs more costs. Such administrative costs do not incur under the strict liability rule, because the government does not need to make the computations.

• (2) Damages caused by an accident usually vary depending on the circumstances. For example, suppose that a plant located in a highly populated area had an accident that caused environmental pollution. The accident and subsequent pollution would affect many people and cause substantial damage. However, damages would be smaller if the plant was located in a less densely populated area. If expected damages are large, stringent measures are desirable from the standpoint of minimizing social costs, whereas measures need not be so stringent if expected damages are relatively small.

  Because the scale of damages differs depending on the circumstances, socially optimal measures vary as well. As such, standards of optimal measures must be established in accordance with different situations. This is practically not feasible as it would incur enormous costs. The optimal level of measures set under the negligence rule is not tailored to accommodate different circumstances, which means adopting the negligence rule results in large social costs. This sharply contrasts with the case of the strict liability rule under which the government does not need to establish standards of negligence in advance; the firm bears all social costs under the strict liability rule, so they choose the level of measures that minimizes their own cost (i.e., social costs) depending on the situation.

• (3) Under the negligence rule, the court must judge if the firm is negligent (i.e., whether the firm satisfied the minimum standards of measures imposed by law), which incurs costs since evidence and investigation are necessary to make conclusions about the firm’s negligence. These costs would not be incurred under the strict liability rule, as there is no need to judge the firm’s negligence.

• (4) Failure to gather sufficient evidence² may result in an erroneous judgement of the firm’s negligence.³ The firm is unlikely to be held liable for damages even when it is negligent if proving the negligence is difficult. In such a case the firm's expected cost becomes lower and he is not likely to take proper measures. Below we explain this point further with an example.

Based on Table 6.1, we show in Table 6.3 the case in which the court incorrectly determines the level of measures to be one level higher with a 95% chance (e.g., the firm adopted “none” of measures, but the court determined that it adopted the “medium” level of measures). The probability of an accident is 25% when the firm chooses “none.” When an accident occurs, then the probability that the firm is held liable is 5%. Hence, the expected damages the firm bears are: 0.25 × 0.05 × 1000 = $125,000. The firm will choose to take “none” of the measures because the total cost (i.e., the sum of the cost of measures and expected damages) is lower than that of choosing the “medium” or “high” level of measures. This means the socially optimal level of measures will not be achieved.

Table 6.3 Erroneous judgement of negligence and changes in the firm’s expected cost

2 The Judgment Proof Problem

This section is written by simplifying formal analysis by Beard (1990)

Damages caused by environmental pollution can be tremendous. If damages exceed the assets of the responsible firm, the firm is unable to pay fully the damages for which it is liable, and the maximum payment that the firm bears will be limited to the firm’s assets. This is known as the “judgment-proof problem” (Shavell 1986). Below we consider how the firm’s incentives for precaution are affected by being “judgment-proof,” i.e., being unable to compensate for all of damages due to insufficient assets.

2.1 Cases Where Firms Take Insufficient Precaution

In Table 6.4, we calculate the firm’s cost under the strict liability rule using the example in Table 6.1. Suppose that the firm’s assets are $3 million, meaning that the maximum amount the firm can pay is its entire assets and it cannot pay fully the amount for which it has been found liable. Let us consider the firm’s expected payment for damages at the three levels of measures. By taking “none” of measures, the firm bears the expected payment of $750,000 (0.25 × 3 million = 750,000).

Table 6.4 Judgment proofing and the firm’s cost under the strict liability rule

When the firm adopts the “medium” level of measures, the maximum amount of payment it bears is $2.78 million (i.e., the firm’s assets minus the cost of measures, $220,000). So, the expected payment is $556,000 (0.2 × 2.78 million = 556,000), and the firm’s total cost is $776,000 (556,000 + 22,000 = 776,000).

When the firm adopts the “high” level of measures, the firm’s expected payment will be capped at $1.75 million, which is the balance of its assets after paying $1.25 million for adopting measures. So, the expected payment is $175,000 (0.1 × 1.75 million = 175,000) and the firm’s total cost is $1.425 million (175,000 + 1.25 million = 1.425 million). Consequently, the firm will choose to take “none” of the measures and the socially optimal level of measures (which is “medium”) will not be achieved. In this way, under the strict liability rule, the firm is not likely to take sufficient precautions when its assets are less than the damages it may cause.

Would we see the same outcome under the negligence rule? Table 6.5 presents the firm’s cost under the negligence rule. When the firm takes “none” of precaution, it will be found at fault and its payment will be $3 million, the assets it owns. Hence, the firm’s expected payment will be $750,000, and the amount will be the same as in the case under the strict liability rule. However, when the firm takes either the “medium” or “high” level of measures, it will be found nonnegligent and thus not be obliged to compensate for the damages; the only cost the firm bears is that of adopting measures. The firm will hence choose to take the “medium” level of measures, which means that the socially optimal level of precaution will be achieved.

Table 6.5 Judgement proofing and the firm’s cost under the negligence rule

As can be seen from the discussion above, the firm’s incentives for precaution diverge under the two liability rules when the firm is judgment-proof; the socially optimal level of measures is not likely to be achieved under the strict liability rule, whereas the negligence rule is likely to give the firm incentives to take measures at the socially optimal level.⁴ For this reason, the adoption of the negligence rule is socially more desirable in situations where firms with insufficient assets may cause accidents that result in enormous damages.

2.2 Cases Where Firms Take Excessive Precaution

Let us turn to the case in which the strict liability rule gives the firm incentives for excessive precaution. Table 6.6 presents the firm’s cost under the strict liability rule when the assets owned by the firm are $9.5 million. When the firm chooses to take “none” of precaution, the maximum amount of payment of the firm for damages is $9.5 million and its expected payment is $2.375 million (0.25 × 9.5 million = 2.375 million). When the firm takes the “medium” level of measures, the maximum amount of payment is $9.38 million (i.e., the assets minus $220,000 for implementing measures), and thus, the expected payment is $1.856 million (0.2 × 9.38 million = 1.856 million) and the firm’s total cost is $2.076 million (1.856 million + 220,000 = 2.076 million).

Table 6.6 Judgment proofing and the firm’s excessive precaution under the strict liability rule

When the firm takes the “high” level of measures, payment of the firm for the damages will be capped at $8.25 million, that is, the assets minus $1.25 million for adopting measures. The expected payment is therefore $825,000 (0.1 × 8.25 million = 825,000) and the total cost is $2.075 million (825,000 + 1.25 million = 2.075 million). Consequently, the firm will choose to take the “high” level of measures, which is higher than the socially optimal level, resulting in taking excessive precautions.

2.3 Judgment Proofing and the Firm’s Incentives for Precaution

We observed above that when the firm’s assets are less than the damages it may cause, the strict liability rule gives the firm incentives for excessive precaution at times and insufficient precaution at others. Below we explain why.

Let x be the cost of measures, and let accident probability be denoted by p(x), a function of x. Let us assume that the larger x is, the higher the safety and the lower the probability of an accident. That is, p(x₀) ＞ p(x₁) when x₀ ＜ x₁. Suppose that the amount of damages is D and constant when the firm causes an accident. Then, when the cost of measures is x, social costs SC(x) are:

$$\text{SC}\left(\text{x}\right)=\text{ x }+ \text{ p}\left(\text{x}\right)\text{D}$$

where the second term on the right-hand side of the equation represents the expected damages caused by the accident. The change in social costs when the cost of measures increases by one unit is:

$$\text{SC}\left(\text{x }+1\right)-\text{SC}\left(\text{x}\right)=1-\left\{\text{p}\left(\text{x}\right)-\text{p}\left(\text{x}+1\right)\right\}\text{D}$$

(6.1)

where the first number “1” on the right-hand side of the equation represents the increase in the cost of measures.

Since p(x) − p(x + 1) (>0) denotes the decrease in accident probability due to the increase in the cost of measures, the second term on the right-hand side of the equation represents the decrease in expected damages due to the decrease in accident probability. Hence, the change in social costs (i.e., the left-hand side of the equation) equals the increase in the cost of measures (i.e., the first term on the right-hand side of the equation) minus the decrease in expected damages. If the increase in the cost of measures is smaller than the decrease in expected damages, the change in social costs will be negative, and social costs will decrease by increasing the cost of measures. Hence, it is socially desirable to increase the cost of measures.

When the firm owns assets W that are smaller than D, the firm’s cost of choosing x under the strict liability rule, C(x), is:

$$\text{C}\left(\text{x}\right)=\text{x}+\text{p}\left(\text{x}\right)\left\{\text{W}-\text{x}\right\}.$$

Here, if the firm implements x, its assets will be W − x, so payment of the firm for the damages in the event of an accident is W − x. Therefore, the second term on the right-hand side of the equation denotes the expected payment of the firm for the damages when an accident occurs, which is smaller than expected damages p(x)D. The change in social costs as the firm increases the cost of measures by one unit is:

$$\text{C}\left(\text{x}+1\right)-\text{C}\left(\text{x}\right)=1-\left\{\text{p}\left(\text{x}\right)-\text{p}\left(\text{x}+1\right)\right\}\left\{\text{W}-\text{x}\right\}-\text{p}\left(\text{x}+1\right)$$

(6.2)

The sum of the second and third terms on the right-hand side of Eq. (6.2) represents the reduction in expected payment due to the increased cost of measures. The second term represents the reduction in expected payment for the damages resulting from the decrease in accident probability. p(x + 1) in the third term represents a decrease in the upper limit of the expected payment for damages⁵ since an increase in the cost of measures reduces firm's assets by one unit (Let us call this the “subsidy effect”).

Since W < D, it always follows that {p(x) − p(x + 1)}{W − x} < {p(x) − (x + 1)}D. If the subsidy effect is small, then:

$$\left\{\text{p}\left(\text{x}\right)-\text{p}\left(\text{x}+1\right)\right\}\left\{\text{W}-\text{x}\right\}+\text{p}\left(\text{x}+1\right)<\left\{\text{p}\left(\text{x}\right)-\text{p}\left(\text{x}+1\right)\right\}\text{D}$$

(6.3)

By comparing Eqs. (6.1) and (6.2), we obtain:

$$\text{SC}\left(\text{x}+1\right)-\text{SC}\left(\text{x}\right)<\text{C}\left(\text{x}+1\right)-\text{C}\left(\text{x}\right)$$

If SC(x + 1) − SC(x) < 0 < C(x + 1) − C(x), then social costs will decrease by increasing the cost of measures. However, the firm will not do so because it is not willing to accept an increase in its cost burden. As a result, the cost of measures taken by the firm will be lower than the socially optimal level, which is what we observed in Table 6.4. Thus, the firm takes insufficient precaution.

Now, when p(x + 1) is large, the inequality sign in Eq. (6.3) will be reversed:

$$\left\{\text{p}\left(\text{x}\right)-\text{p}\left(\text{x}+1\right)\right\}\left\{\text{W}-\text{x}\right\}+\text{p}\left(\text{x}+1\right)>\left\{\text{p}\left(\text{x}\right)-\text{p}\left(\text{x}+1\right)\right\}\text{D}$$

By comparing Eqs. (6.1) and (6.2), we obtain:

$$\text{SC}\left(\text{x}+1\right)-\text{SC}\left(\text{x}\right)>\text{C}\left(\text{x}+1\right)-\text{C}\left(\text{x}\right)$$

If SC(x + 1) − SC(x) > 0 > C(x + 1)  − C(x), then social costs will increase due to an increase in the cost of measures; the firm will nonetheless increase the cost of measures because doing so will reduce the firm’s total cost. As a result, the cost of measures will be higher than the socially optimal level, which is what we observed in Table 6.6. Thus, the firm takes excessive precautions.

The conclusions we reached can be summarized as follows. In circumstances where a firm is judgment-proof, it is socially more desirable to adopt the negligence rule than the strict liability rule. The socially optimal level of precaution is not achieved under the strict liability rule; the firm will take insufficient (excessive) measures when the subsidy effect of increasing the cost of measures is small (large). In contrast, the negligence rule leads the firm to take precautions at the socially optimal level even when the firm is judgment-proof but its assets are larger than a certain threshold. Given that the negligence rule does not give the firm incentives to take optimal measures when its assets are below the threshold, it may be desirable to introduce regulations on firms to allow only those with assets above a specific threshold to engage in production activities.

Box 6.1 Responding to the Risks Associated with Consuming Food and Using Products: Risk Information Disclosure and Consumer’s Self-responsibilities

Various risks are associated with food and product consumption. For example, consuming food items containing soy or other allergenic ingredients without knowledge of the ingredients can be harmful to human health. Moreover, genetic modification contributes to higher-yield varieties and makes crops more resistant to diseases and pesticides, thereby increasing productivity and lowering food prices, but excessive consumption of genetically modified (GM) foods may also lead to health risks. Additionally, damages may occur owing to improper use or malfunction of products.

Air pollutants emitted by production or consumption activities harm other parties who are neither producers nor consumers of the products (i.e., third parties such as residents). Firms and consumers usually do not consider the health hazards or other adverse effects on third parties. Hence, policy instruments such as environmental taxes are necessary to control pollutant emissions and promote environmental conservation efforts. However, food consumers are affected by food risks. As the damage is not borne by a third party, no external costs are incurred. Therefore, imposing environmental taxes on contaminated food is not necessary.

If consumers are provided with accurate information on the health hazards involved (e.g., which food is contaminated, the probability and degree of contamination, and possible health risks of consuming the food), they will make the right decisions and avoid foods with a high risk of health hazards. People can choose which products to consume based on their risk level for certain allergies. For example, people will buy foods, if they are not allergic to any ingredients in those foods, while they will not buy them, if they are allergic to some ingredients. So if people are allergic to certain ingredients, producers can anticipate consumer behavior and stop producing high-risk foods. The consumer’s decision does not affect the health of the other parties. Therefore, market failure does not occur, even in the absence of environmental taxes. It is essential for the government to provide consumers with accurate information about food safety; otherwise, consumers will not be able to make the right decisions.

However, it is not desirable to ban the overall production of high-risk foods to reduce risk to zero as it would lead producers to use higher-cost ingredients, resulting in higher production costs. Low-income households will be forced to buy higher-priced foods and, consequently, cannot afford other foods and fail to receive adequate nutrition. If GM foods are banned, the risks associated with their consumption will be eliminated, but it may lead to other health risks due to malnutrition. It is more beneficial for consumers to have access to more diverse product choices and to take responsibility for their own decisions.

It is essential for the government to provide consumers with accurate information about food safety; otherwise, consumers will not be able to make the right decisions.

However, it is not desirable to ban the overall production of high-risk foods to reduce risk to zero since it would lead producers to use higher-cost ingredients, resulting in higher production costs. Low-income households will be forced to buy higher-priced foods and, consequently, cannot afford other foods and fail to receive adequate nutrition. If GM foods are banned, the risks associated with their consumption will be eliminated, but it may lead to other health risks due to malnutrition. It is more beneficial for consumers to have access to more diverse product choices and to take responsibility for their own decisions.

It should be noted that, even if accurate information is provided, people with limited cognitive abilities to understand information, such as children and the elderly, may still make inappropriate decisions. This is particularly true when information is complicated or voluminous. They may consume high-risk products that they would not consume if they had sufficient abilities. Therefore, if such concerns are serious, the government should consider establishing food safety regulations and standards and providing easy-to-understand information through a system of labeling the presence or absence of certain ingredients and their level of risk.