Incorporating Animal Welfare into Cost–Benefit Analysis

Abstract

Since non-human animals also experience welfare, an increasing number of scholars advocate including non-human animal welfare in cost–benefit analysis. Recent proposals to achieve this through interspecies comparisons of welfare, however, are incompatible with the principles of positive welfare economics. Based on conceptual and theoretical analysis, this article argues that, to remain consistent with positive welfare economics, the monetary value of welfare changes should be set equal to the marginal costs of alternative options available to offset those welfare changes. This applies equally to human adults, small children, and non-human animals. The article further argues that monetary valuation is appropriate only in cases involving marginal changes in the risk of harm—for example, an increased mortality risk for birds and bats caused by windmills—but not in cases involving direct and certain harms, such as those inflicted on farm animals. Moreover, a key rationale behind cost–benefit analysis is that a positive outcome increases social wealth, thereby expanding the government’s capacity to enhance social welfare through redistribution or investment in public services. In the interspecies context, however, this rationale remains doubtful as long as governments fail to give equal consideration to non-human animal welfare in policy making.

1. Introduction

Whichever theory of welfare one adheres to [1], welfare is a state that not only meaningfully applies to humans but to non-human animals as well. Whether one believes the one intrinsic good consists of pleasure and the absence of pain [2], the fulfillment of desires or preferences [3,4], or flourishing [5], non-human animals have a welfare that can be either improved or worsened. For this reason, utilitarians such as Bentham [2] and Singer [3] have argued that our duty to maximize utility entails giving equal consideration to the consequences of our acts for human and non-human welfare.

Since non-human animals experience welfare as well, an increasing number of scholars advocate for including non-human animal welfare in policy making for sustainability. Treml et al. [6], for example, discuss how to add animal welfare into life cycle assessment. Visseren-Hamakers [7] has made a case for an 18th Sustainable Development Goal (SDG) on animal health, welfare and rights. Others propose to include non-human welfare in one of the central decision-making instruments in sustainability policy making: cost–benefit analysis [8,9,10,11,12,13,14,15,16,17,18,19,20]. After all, many policy decisions for which cost–benefit analysis is performed have impacts on the welfare of non-human animals as well. Examples include increased mortality risk for birds and bats due to windfarms [21,22] and airports [23], mortality risks for wild mammals due to new infrastructure such as highways that fragment habitats [24,25,26] and mortality risk for fish [27,28] and whales [29] due to infrastructures in rivers and oceans.

In such cost–benefit analysis, animal welfare cannot be equated with or reduced to the benefits that humans derive from animal wellbeing, such as warm-glow value and existence value [30]. Warm-glow value refers, for example, to the happiness humans derive from knowing that their pets are doing well [31] or that farm animal welfare is increased [32,33] and the welfare gains people experience if they know their moral commitments to non-human animals are vindicated [13] (p. 1809); see also [14]. Existence value refers to the benefits people derive from something’s mere existence, although one has no current or future plans for its active use. Thus, we may be happy to know tigers are doing well, although we have no wish to encounter them.

The question instead is how to monetize the welfare for the animal itself, or what Budolfson et al. [19] call the intrinsic value of animal welfare and Espinosa [16] calls animal welfare per se. To some extent, this animal welfare can be included in cost–benefit analysis by measuring people’s willingness to pay for such welfare out of moral or altruistic concerns. According to Sen [34], standard economic theory is too restrictive in its interpretation of willingness to pay, since not all willingness to pay originates in trying to increase one’s own wellbeing. We may also act out of preferences regarding the wellbeing of others, in which case one’s own satisfaction (the warm-glow value) is merely an unintended, though positive, side effect. Sen calls this commitment value, which, according to him, drives “a wedge between personal choice and personal welfare.” [34] (p. 329). In other words, our willingness to pay may also directly reflect benefits to non-human animals rather than merely benefits to ourselves. Nevertheless, people’s willingness to pay falls short as a true measure of animal welfare as it depends upon the contingent moral motives or commitment of people [19]. Animal welfare does not depend upon whether people care about it.

Recently, Budolfson et al. [19], see also [16], have proposed a two-step approach to monetize animal welfare for inclusion in cost–benefit analysis. In the first step, changes in animal welfare are expressed in units of human welfare by interspecies comparisons of welfare. Budolfson & Spears [35], for example, suggest relating neuron counts in certain brain regions to welfare potential, while Fischer [36] discusses the incorporation of a wide range of neurophysiological, cognitive and behavioral traits into more complex multidimensional proxies that can be aggregated to produce a single overall welfare potential score. In the second step, the changes in animal welfare are monetized by using monetary valuations of human welfare, such as monetary valuations of quality-adjusted life years (QALYs).

The approach suggested by Budolfson et al. [19] is inconsistent with the foundations of positive welfare economics underlying cost–benefit analysis, as positive welfare economics was precisely developed to avoid comparisons of welfare between different actors [37]. Although Budolfson et al. [19] do not literally spell it out in their article, their two-step approach is based upon the assumption that equal changes in welfare should be given an equal monetary value in cost–benefit analysis. Equal welfare changes should indeed be counted equally in a utilitarian calculus, but equal welfare changes do not require an equal monetary value in cost–benefit analysis.

The purpose of this article is to explain how to incorporate animal welfare in cost–benefit analysis in a manner that is consistent with the foundations of positive welfare economics. The used methodology is a conceptual and theoretical analysis. The setup of this article is as follows. In Section 2, it is first explained how changes in welfare are monetized in cost–benefit analysis by determining willingness to pay and willingness to accept. In Section 3, it is subsequently explained that, as a consequence, the value of a statistical life depends upon people’s ability to pay and thus on income and wealth. In Section 4, it is explained how to include small children in cost–benefit analysis as they lack both the financial means and the cognitive capacities to express a willingness to pay. The solution lies in determining a hypothetical willingness to accept as compensation for the negative impacts due to a policy decision on the basis of the determination of the marginal costs of alternative options available to offset the welfare changes. In Section 5, it is explained that the same methodology should be applied to non-human animals, who also lack both the financial resources and the cognitive abilities to express willingness to pay to prevent negative impacts on their welfare. Section 6 discusses the proposed approach and concludes.

2. Cost–Benefit Analysis, Kaldor–Hicks Efficiency and Willingness to Pay

Positive welfare economics is based upon the concept of Pareto improvements: changes where at least one individual becomes better off without making anyone else worse off [38]. The idea behind the concept of a Pareto improvement is that it avoids interpersonal comparisons of welfare or utility: if at least one individual becomes better off without making anyone else worse off, then it must be certain that overall welfare has improved, see, e.g., [37,39,40]. If one individual becomes better off while someone else becomes worse off, it can never be determined with certainty whether overall welfare increases, because it requires the troublesome interpersonal comparisons of welfare. Welfare, utility, ophelimity, happiness, satisfaction, pleasure or enjoyment are subjective states, and we have no objective means to compare the welfare of one person to that of the other. Please note that self-reported changes also offer no proof. Even if one person states she became very happy while another person states she hardly became happier, we have no means to objectively compare the two statements. We have no way of telling how the statements are related to how it feels from the inside, apart from assuming that the other person experiences the world exactly as we do and expresses that in the same way we do. See [41], however, on the problems with this assumption.

Pareto improvements thus explain why voluntary (market) transactions (in the absence of externalities) increase overall welfare: since no one will participate in free exchange without benefitting from the exchange, all voluntary exchange must increase overall welfare. In the case of policy making, however, such as the construction of infrastructure, it cannot be guaranteed that everyone will benefit. There may also be people who are negatively affected. Therefore, Kaldor [42] and Hicks [43] made the step from the concept of actual Pareto improvements to potential Pareto improvements, or as it is now called Kaldor–Hicks efficiency: a situation that hypothetically can lead to a Pareto improvement if the winners of a policy compensate the losers. The idea is that the winners from a change could theoretically compensate the losers and still be better off, even if the compensation does not actually occur. This idea of Kaldor–Hicks efficiency lies at the basis of cost–benefit analysis. So, in cost–benefit analysis, it is assumed that no interpersonal comparisons of welfare are required to assess whether (potentially) overall welfare improves as a result of policy plans and interventions. Kaldor [42] (p. 550) expresses this view as follows.

“The economist’s case for the policy is quite unaffected by the question of the comparability of individual satisfactions; since in all such cases it is possible to make everybody better off than before, or at any rate to make some people better off without making anybody worse off. There is no need for the economist to prove—as indeed he never could prove—that as a result of the adoption of a certain measure nobody in the community is going to suffer. In order to establish his case, it is quite sufficient for him to show that even if all those who suffer as a result are fully compensated for their loss, the rest of the community will still be better off than before. Whether … [those who suffer] should in fact be given compensation or not, is a political question on which the economist, qua economist, could hardly pronounce an opinion.”

Although we cannot make interpersonal comparisons of welfare, we can assess whether the same person becomes better or worse off. In welfare economics, this assessment is made by asking people for their willingness to pay or their willingness to accept. For example, if an airport is planned to be built with accompanying noise nuisance for a residential area, residents may be asked how much they are willing to pay to prevent the airport from being built or how much they are willing to accept as compensation to agree with the airport being built. The costs and benefits of an intervention are thus determined by willingness to pay or accept, defined as the amount that needs to be deducted from or added to a person’s income in the presence of the intervention to keep their own wellbeing constant; see, e.g., [44]. This willingness to pay or willingness to accept depends upon the alternatives available for a person to obtain equal welfare changes. If there is no such willingness, then welfare is assumed not to have changed. If costs and benefits are marginal (negligible) in comparison to income, economic rationality would require willingness to pay and willingness to accept to be equal. In reality, willingness to pay is generally much lower than willingness to accept. Respondents may, for example, consider it unfair that they are asked to pay if they feel entitled to compensation. Or people may be loss averse, meaning they want more compensation for their loss relative to the baseline (no airport being built) than they are willing to pay to improve the situation relative to the baseline (airport being built). However, there is no guarantee that welfare has not changed if willingness to pay or accept is zero, since there are other conceptions of welfare that do not take such a subjective view [1,45]. An example is objective list theory, according to which there are objective components of welfare independent of personal preferences for these components, such as friendship, knowledge or self-actualization. According to this theory, someone’s life may be improved without that person desiring the change and thus without the person being willing to pay for it.

So positive welfare economics is “positive” instead of normative in that it aims to describe the impacts of economic changes on social welfare on the basis of measurable outcomes, rather than advocating for particular policy decisions based upon value judgements. As such, the Kaldor–Hicks criterion does not prescribe whether compensation ought to be paid, nor does the Pareto criterion prescribe whether Pareto improvements are morally desirable or right. However, cost–benefit analysis may also sometimes be based upon social welfare function approaches that assume interpersonal comparisons of utility can indeed be made; see, e.g., [46,47].

3. Economic Valuation of a Statistical Life and Differences in Ability to Pay

That welfare changes are monetized in cost–benefit analysis by determining people’s willingness to pay or willingness to accept has the consequence that welfare changes for different people of which we intuitively believe are more or less the same, may nevertheless obtain a very different monetary value. The reason is that the marginal costs of the alternatives available for a person to obtain equal welfare changes generally will depend upon income and wealth. Therefore, the willingness to pay of wealthier people will generally be higher. This has the controversial consequence, for example, that welfare changes in developing countries are given a lower monetary value in cost–benefit analysis than (intuitively assumed) similar welfare changes in wealthy countries. Nevertheless, this logically follows from economic rationality, as will be explained with the concept of the value of a statistical life (VSL).

The VSL is the monetary value that individuals and societies place on reducing the risk of death [48,49,50]. On a social level, its value is determined by the marginal costs society is willing to make to prevent the loss of statistical lives, such as making a specific highway safer and thereby reducing the number of fatal casualties from three to two a year. The addition “statistical” is essential: the method does not apply when potential victims have names and faces, such as in the case of workers trapped in a mine. Instead, the method applies to situations such as road safety or preventive health care programs. On an individual level, its value may be determined by the costs people are willing to make to reduce their own risks, for example, by buying safer cars or by the higher wages people demand for riskier jobs. In the individual case, the VSL is a derivative of the costs of a partial risk reduction, for example, the costs of reducing a specific risk from occurring 1/10,000 to 1/100,000. In the United States, the cut-off limit beyond which society is not willing to pay for further risk reduction lies at about USD thirteen million per statistical life saved [51,52], which therefore is the VSL for the United States.

Viscusi and Masterman [51] show how the VSL is related to per capita income and thus lower the poorer the country. Imagine the following situation: in a wealthy country, the marginal costs to reduce risks are EUR ten million per statistical life saved, while in a poorer country, the marginal costs are EUR one million. Given the risks people are facing in different countries, with different safety standards and different levels of health care, it is simply cheaper to increase safety and save lives in poorer countries than in wealthier countries. Furthermore, imagine that a multinational company knows that for EUR eight million, it can prevent either one death in a factory in a wealthy country or five deaths in a factory in a poorer country. In that case, the logic of welfare economics states that the company can better prevent one death in a factory in a wealthy country. The reason is the following. If given the choice, the workers in the wealthy country would spend the EUR eight million on increased safety in their factory in the absence of better options outside the factory. In the poorer country, however, the workers would spend the eight million on safety outside the factory, for example, on better access to health care or safer drinking water that would save more than five people. Therefore, the workers in the wealthy country could potentially compensate the workers in the poorer country for risk reduction outside the factory, and both sides would be better off. If the workers in the poorer country are paid five million, they could pay for risk reduction outside the factory with the same welfare effect as if the company had made the investment of EUR eight million in their factory.

The example shows that equal increases in risk in a wealthy and a poor country, of which we intuitively assume they have equal welfare impacts, do not have an equal monetary value in economic analysis. To many, this may appear immoral; for further discussion, see, e.g., [53]. The immorality, however, does not lie in the different monetary valuations of different statistical lives in cost–benefit analysis itself. If there is injustice, it first originates in the wealth differences that are the cause of the different valuations, and secondly, in the fact that, in practice, often no compensation is paid to workers in the poorer country. Deviating from positive welfare economics, justice concerns may be addressed by using a social welfare function—see, e.g., [46,47]—or explicit equity weighing [54], i.e., explicitly applying multipliers to different groups based on income, vulnerability, or need.

4. Including Small Children in Cost–Benefit Analysis

A second controversial issue in cost–benefit analysis, and a next step towards the analysis of non-human animals, is how to include the welfare of small children. In contrast to adults, small children (and other humans who are “incompetent” in other ways), do not only have hardly any ability to pay but also lack the cognitive capacity to do so. A one-year-old child does not have income and cannot make considered decisions. That does not mean that changes in their welfare should be set at zero in cost–benefit analysis. The solution, however, is neither to use the parents’ willingness to pay for their child’s wellbeing, see, e.g., [55,56], nor to assume that the VSL is simply equal to that of adults [57,58]. After all, the parents’ willingness to pay for their child’s wellbeing primarily reflects its value for the parents and thus does not reflect the intrinsic value of the child’s welfare [20]. See [59] for a study trying to establish childrens’ own willingness to pay for health benefits.

The solution compatible with positive welfare economics lies in two steps. The first step consists of determining the willingness to accept a certain impact instead of the willingness to pay to avoid the impact. Using willingness to accept instead of willingness to pay solves the problem that the child has no financial means and thus no ability to pay. Even without any money, one can still meaningfully express a willingness to accept (we will return later to the child’s lack of cognitive capacities). Basing monetary valuation on willingness to accept instead of willingness to pay by no means conflicts with the foundations of welfare economics. As stated earlier (Section 3), if payments or required compensation are marginal relative to income and wealth, the two should offer the same results from a rational perspective if we ignore psychological mechanisms resulting in willingness to accept generally being higher than willingness to pay [60]. If one has no income, willingness to pay is, of course, zero, but willingness to accept does not have to be so, since it depends upon the available alternatives.

Whether the appropriate question is for people’s willingness to accept or willingness to pay is a matter of justice instead of economics. That question depends upon entitlements and property rights. If A is entitled to perform an act that harms B, such as building a fence on A’s property that blocks B’s view, then the appropriate question is for B’s willingness to pay to keep A from building the fence. If B is entitled to her view, however, the appropriate question is for B’s willingness to accept compensation for A’s fence. As Coase [61] argued, for efficiency, the allocation of rights and entitlements does not matter. Since in the case of children’s VSL, we are primarily considering the risk of bodily harm imposed by others, we may uncontroversially assume that the child holds the right not to be harmed instead of the other party holding the right to perform acts which, as a side-effect, may harm the child. Therefore, it is uncontroversial to base the VSL on willingness to accept instead of willingness to pay.

The second step consists of letting the analyst performing the CBA determine the willingness to accept instead of the child on the basis of all available alternative options that increase the welfare of the child or reduce other risks to bodily harm of the child. The CBA analyst determining willingness to accept solves the problem that the child lacks the cognitive capacities to make considered choices. However, it does raise a new problem, but one that is unavoidable: the analyst cannot take the subjective point of view that normally lies at the basis of willingness to pay or willingness to accept. That does not mean, however, that interpersonal comparisons of welfare are reintroduced. All that is required is that one compares the impacts of different choices on the welfare of the same child. Although I cannot be certain that if we both have the same illness, we experience the same welfare loss, I can be quite certain that if you experience the same illness on two different occasions, but due to different causes, you experience the same welfare loss.

The question the analyst, therefore, must answer is how much money is required for other improvements in the life of the child that make up for the increased risk due to the choice under consideration. The marginal costs of the alternatives will strongly depend upon the circumstances. Under normal circumstances, we may assume that parents will take care of their children similarly to how they handle risk for themselves, or even take more precautions in the case of their children [62,63]. For that reason, the analyst will not easily find cheap options to further improve the child’s welfare. In fact, it is likely that a VSL of the child will result that is close to the VSL of the parents, even though the child itself has no financial means. It is important to note, however, that although the VSL of a child will under normal circumstances be similar to that of its parents or even higher, the reason is not that the child’s VSL is directly based upon the parents’ willingness to pay for the child’s welfare or that the child’s VSL reflects benefits to the parents. In the case of unloving parents with little willingness to pay for the child’s welfare, the high marginal costs of alternative measures to improve the child’s wellbeing, and thus the child’s VSL, may also be the result of governmental regulations or social pressure. On the other hand, when a child is raised under really poor circumstances, there may indeed be cheap alternative options to improve the child’s welfare, and as a consequence, the VSL will also be lower. Just as in the case of lower VSLs in developing countries, the injustice here does not lie in the VSLs themselves but in the circumstances. The VSLs only mirror those circumstances.

The benefits of reduced risks to children, and thus children’s VSL, should therefore be based upon the marginal costs of available options to improve the child’s welfare or reduce other risks that offset the risk under consideration. It is important to note that although parents’ willingness to pay for a child’s welfare influences the marginal costs of available options to improve the child’s welfare, only the marginal costs of available options themselves determine the monetary valuation of the true value of the child’s welfare. Only the latter perspective is consistent with welfare economics.

5. Including Non-Human Animals in Cost–Benefit Analysis

Now that we have established how to determine children’s VSL, we also have the methodology to include the welfare of non-human animals in cost–benefit analysis. Just like small children, non-human animals lack both the cognitive capacity and the financial means to have a willingness to pay. Some have tried to attribute willingness to pay to animals by performing experiments on the trade-off animals make between food and other inputs for wellbeing, such as living space, e.g., [64,65,66,67]. Since there is no trade between humans and animals, however, such trade-offs in the non-human world have no economic meaning in the human world. The VSL of non-human animals will generally be much lower than that of humans. The reason, however, is not because of physical differences between humans and non-human animals, but because the marginal costs of reducing risk to animal welfare are generally much lower, given the fact that, in comparison to measures that are taken to protect humans, protective measures for non-human animals are almost absent. In the case of protected rare species, however, their high level of protection may, in principle, mean a higher VSL than for humans. The monetized welfare of a member of a rare species may be higher than that of a human being. Walston et al. [68], however, estimate that the costs of protecting tigers are about USD 80 million per year for a potential population size of 4435 tigers. With an average lifespan of a tiger of about 10–15 years, this would imply a VSL of more than USD 200,000, but still much lower than the VSL of humans.

A question that must be answered in the case of non-human animals is about standing. Since the methodology that is proposed depends upon willingness to accept instead of willingness to pay, the entity must have the right not to be exposed to risk. The entity must be entitled to compensation in the case of wrongful harm. Although rocks and artifacts also lack any cognitive abilities, it would make no sense for others to ask for compensation for harm on the artifacts’ behalf, apart from harm to the people themselves, if the rocks and artifacts are personal property. There is no meaningful way in which rocks and artifacts can benefit for their own sake. Moral theories about the “good” leave open the option, however, that other living entities than non-human animals are capable of having a good and therefore can be benefited for their own sake or harmed. Bentham [2] held a hedonist view on value, meaning that he believed that only the sensations of pain and pleasure were of intrinsic (dis)value. As a preference utilitarian, Singer [4] believes that the fulfillment of preferences is of intrinsic value. In both cases, it may be assumed that only sentient animals are capable of having a good. If one believes “flourishing” is the ultimate good, then plants can also be benefited and harmed [69,70,71].

Please note that although it is difficult to assess welfare impacts for someone else, particularly if the other belongs to a different species, no interspecies comparison of welfare is required. The only thing that is required is an inventory of measures with comparable welfare impacts for the same species. The marginal (cheapest) alternative measure that is capable of offering the same welfare impacts determines the monetary value of the impacts; see also [15]. Since the monetary value of welfare changes should be set equal to the marginal costs of alternative options available to offset the welfare changes, the methodology for establishing a value for the statistical life (VSL) for an animal species is similar to other frameworks to establish marginal abatement cost curves. Although providing a detailed empirical protocol for practitioners is beyond the scope of this conceptual analysis (see [72,73,74] for such protocols), the following framework provides a stepwise procedure:

• Specification of measurable welfare units. Depending on context, this may be the number of (avoided) fatalities for a specific species (e.g., bats and birds) or improvements in a welfare index based on physiological or behavioral indicators.
• Identification and inventory of all measures capable of reducing the same welfare loss. These include avoidance and minimization strategies, direct offsets, and broader conservation actions. For each measure, both scope and scalability are recorded.
• Estimation of costs and effectiveness. For each measure, full life-cycle costs are estimated, including installation, operation, monitoring, administration and enforcement. Effectiveness is expressed as expected welfare units improved per measure, based on empirical studies, monitoring data, or expert elicitation.
• Determination of marginal costs. Incremental implementations are ordered by cost-effectiveness (EUR/welfare unit), allowing for non-linearities such as diminishing returns. The resulting marginal cost curve shows the increasing cost of securing additional welfare improvements and provides the relevant monetized values for use in cost–benefit analysis. The most expensive measure in the cheapest set of measures required to offset the welfare change determines the marginal costs.

There is thus a direct link between the monetized animal welfare in cost–benefit analysis and the mitigation hierarchy framework used in environmental management and planning and environmental impact assessments to reduce negative impacts on biodiversity and ecosystems from development projects or activities [75,76]. According to this framework, negative impacts should first be avoided, then reduced and finally, when some negative impacts cannot be avoided, those must be offset or at least compensated [77]. The marginal costs of such offsets exactly offer the monetized value of animal welfare to be used in cost–benefit analysis. Although this result may come as a surprise, it should be noted that the idea that the monetized value of animal welfare is basically equal to the costs of compensation does not differ from the idea that the VSL for humans is derived from the financial compensation people require for accepting a risky job. Also in the latter case, people assess the marginal costs of available alternatives to offset the risks of accidents during work. So, to include bat welfare in cost–benefit analysis, we do not need to make interspecies comparisons. We do not need to know what it is like to be a bat [78].

It should be noted that although the marginal costs of alternative options may be the same for different cases within the same country, this will not be the case for regions with different policy standards. For the same reason, the VSL for humans differs for different regions (see Section 3). Since the marginal costs of alternative options available to offset welfare changes can vary significantly across countries and contexts, standardization will thus not always be possible.

An example where animal welfare could be included in cost–benefit analysis is the construction of (offshore) windfarms. These windfarms have the usual construction and operational costs, and the benefits of energy use and reduced climate change [21,22]. However, the windfarms also lead to bat fatalities [79]. How can one monetize the benefits or costs for these non-human animals? The value of a statistical bat life will depend upon the marginal costs of mitigating bat fatalities in other windfarms, for example, by painting the blades of existing windfarms black [80] or operational curtailment strategies [81], or by conservation and offset measures elsewhere. Peste et al. [82] offer a review of potential conservation measures in the European context to mitigate the impacts of wind farms on bats. The wide range of measures include offsets or compensation by management of autochthonous forest (such as preservation of older trees with cavities or creating small clearings or agricultural plots), diversification of forest and agriculture monocultures (such as plantation of live fences and native trees and shrubs), preservation of existing roosts and provision of new roosts (such as installing bat boxes), and the creation of ponds. Peste et al. assess the associated costs as ranging from “low” to “high”, but do not quantify them. Such quantification is necessary to monetize a bat’s VSL. EPRI [83] estimates the per bat habitat mitigation costs average about USD 13,000 and are likely to fall within a range of USD 6000–USD 20,000. These numbers are indicative, however. Some may believe that the monetary values we will find for the VSL of bats are too low, given the importance of bats in natural ecosystems and in human agriculture [84]. If bats were “paid” for the many ecosystem services they provide through protective or supportive measures, then more measures would have been implemented to reduce bat fatalities, and the marginal costs of further risk reduction would have increased as well. Consequently, the value of a statistical bat life would then have been higher as well. This is not different from the human case, however.

A more problematic case is the cost–benefit analysis of measures to improve farm animal welfare. The problem is twofold. First, the issue in such cost–benefit analysis is not about an increase or decrease in the risk of harm, but certain harm, for example, due to confining animals in smaller spaces. Welfare economics is generally not well equipped to monetize certain bodily harm, such as when people are trapped in a mine facing certain death unless saved. Please note that methods that rely on interspecies comparisons of welfare [16,19] face the same problem. The monetary valuations of human welfare on which these methods rely are obtained for situations in which welfare changes are marginal. These values cannot be used to make comparisons to situations in which welfare changes are not marginal. Economic analysis is about marginal changes, such as in the case of a small change in harm or a small change in the probability of a more substantial harm. Second, many welfare losses due to business choices in animal husbandry can hardly be compensated for by alternative measures. For example, the welfare loss experienced by broiler chickens kept indoors with nineteen chickens per square meter (the lower limit allowed by EU regulation) instead of letting them freely roam outdoors with a few square meters per chicken cannot be compensated by changes in the amount or kind of food or other goods. No money can buy the measures that are required to let the chicken have comparable welfare in both settings.

Within the framework of cost–benefit analysis, the only meaningful question that can be asked is which alternative welfare improvements could compensate a chicken to face a small probability to go from free range to a high-density factory farm setting, say, a chance of one in a thousand. By multiplying that amount by a thousand, one would obtain the monetary value of the full welfare loss. Further welfare improvements for free-range chickens could, for example, include “pecking pans” containing a particulate pecking block, and wind chimes to reduce inter-bird pecking, see, e.g., [85,86]. The difficult question then is to balance the welfare improvements for free-range chickens due to such measures against the welfare loss of a small chance to end up in an intensive factory setting. Once this question is answered, on the basis of the costs of the welfare-improving measures, the monetary value of the full welfare loss of a chicken in an intensive factory setting can be calculated. The assessment of this value is beyond the scope of the present analysis.

It might be concluded, however, that cost–benefit analysis is simply not equipped to account for substantial, rather than marginal, changes in welfare, nor can it weigh the welfare loss of humans involuntarily participating in a lethal game show against the welfare gains of millions who enjoy the program. To make such comparisons, one must either engage in a direct interpersonal comparison of utility gains and losses, or rule them out from the outset on deontological grounds—for example, by holding that violations of rights to bodily integrity and freedom of movement can never be justified by welfare gains for others. Some authors [87,88] indeed argue that the deontological concept of human rights should be extended to non-human animals as well. From that perspective, any treatment of animals as a mere means for human welfare, such as in the case of animal husbandry or animal testing, is a violation of rights. Finally, regulatory standards may be set, such as minimum welfare standards for the protection of farmed animals. Such standards, however, will generally result from a combination of scientific research, ethical considerations, government legislation, market demands, and industry-led initiatives.

6. Discussion and Conclusions

Since non-human animals have a welfare as well, an increasing number of scholars advocate for including non-human animal welfare in policy making for sustainability. As the path towards sustainable development generally requires balancing the costs and benefits of different policy options, it has been proposed to take changes in animal welfare into account in cost–benefit analysis. In recent publications, it has been proposed to do so by interspecies comparisons of welfare. Such interspecies comparisons are incompatible with positive welfare economics, however. In the present article, it has been shown that although non-human animals lack the financial resources and the cognitive abilities to express a willingness to pay, changes in the welfare of non-human animals can nevertheless be monetized in a manner compatible with positive welfare economics by calculating the marginal costs of alternative options that can offset the changes in welfare. The same methodology should be applied to monetize welfare impacts for small children.

It is important to note that monetizing changes in animal welfare in cost–benefit analysis improves their welfare irrespective of the answer to the normative question whether animals should be compensated for damage. The reason is that the primary objective of internalizing externalities in decision making by monetizing unpriced welfare changes is to improve welfare through efficient reduction in damage, not through compensation for damage. The situation does not differ from the monetary valuation of carbon emissions to incorporate climate damage to future generations in cost–benefit analysis; see, e.g., [89,90]. So, including monetized welfare effects for animals in cost–benefit analysis may imply that different technologies are applied when building windmills, leading to less harm to birds and bats, or to a reduction in industrial farming of animals, without any animal actually being compensated for harm.

Although non-human animal welfare can in principle be included in cost–benefit analysis, the case of animals shows the limits of welfare economics as well. The Kaldor–Hicks criterion on which cost–benefit analysis is based has often been criticized for neglecting distributive justice; see, e.g., [91]. Since the criterion is concerned only with aggregate benefits and costs, it can overlook significant issues related to equity and the distribution of wealth. However, the problem runs deeper, since it may be questioned whether welfare economics delivers what it promises to deliver in the first place: an increase in overall welfare [92,93,94]. The reason is that when there are large differences in ability to pay due to differences in wealth and income, there is a strong reason to doubt whether a positive outcome of a cost–benefit analysis also implies an overall increase in welfare. If, due to a policy intervention, a billionaire gains a EUR one thousand and a poor person loses EUR nine hundred, there is an improvement according to the Kaldor–Hicks criterion. If the billionaire does not compensate the poor person, however, it is highly likely that an overall welfare loss has occurred. Although interpersonal comparisons of welfare are problematic, it is generally believed that an additional dollar means more to the poor than the rich due to the diminishing marginal returns of welfare. As Marshall [95] (p. 130) noted, “a pound’s worth of satisfaction to an ordinary poor man is a much greater thing than a pound’s worth of satisfaction to an ordinary rich man.” Although this statement cannot be checked empirically, its denial would be very counterintuitive.

A possible justification for the Kaldor–Hicks criterion is that a positive outcome of a cost–benefit analysis increases social wealth, thereby increasing the options for governments to increase social welfare by redistribution or investments in public services. The assumption, in other words, is that at the end of the day, compensation does occur, although indirectly. However, this assumption is untenable in the interspecies context, as long as governments do not give equal consideration to non-human animal welfare in policy making. There is therefore strong reason to doubt whether the application of cost–benefit analysis leads to an increase in overall welfare when both human and non-human animal welfare is at stake, even when non-human animal welfare is monetized and included. Human ability to pay is, after all, incomparably much larger than the ability to pay of non-human animals.

There are two solutions to guarantee that overall welfare increases. First, by requiring that those who lose out due to a policy proposal are actually compensated, i.e., that instead of Kaldor–Hicks improvements, actual Pareto improvements occur. However, while poorer people can be financially compensated, non-human animals cannot. The latter have to be compensated in kind by offsetting. But if the project includes compensation for the loss of animal welfare by offsetting, then there is no need to monetize non-human welfare loss in the first place. The second solution to assure an increase in overall welfare is by actually performing a utilitarian calculus instead of a cost–benefit analysis. In that case, there is no need to compensate any losses, but indeed an interspecies comparison of welfare has to be performed, for which Budolfson et al. [19,35] offer suggestions. However, if a utilitarian welfare maximization is aimed at the basis of interspecies comparison of welfare, once again, it may be doubted whether there is any need for monetization at all. In that case, the calculation will be in physiological or other welfare units instead of monetary units.

Finally, it should be reemphasized that the proposed methodology to include animal welfare in cost–benefit analysis primarily applies to situations where animals face an increase in risk, such as when windmills are built on flight routes of birds and bats, similar to the changes in risk to humans in the case of cost–benefit analysis of road safety measures. When harm to animals is both substantial and certain, cost–benefit analysis is ill-equipped to deal with such harm. Moreover, such cases raise deontological concerns about whether animal rights are violated. Whether policy makers should base their decisions upon utilitarian or deontological considerations is a political matter that should be settled by lawmakers informed by moral analysis.