2.1 Positive criteria

Proponents of BCA tout its ability to support “rational decision making” (Posner, Reference Posner2004) and “intelligent priority setting” (Sunstein, Reference Sunstein2005) and to promote bureaucratic accountability. Two qualities of BCA that underpin these features are the objectivity and commensurability of the value estimates that inform the analysis. That BCA is objective – that it utilizes verifiable data and well-established methods consistent with positive economics to characterize costs and benefits as individuals perceive them – helps guard against bias on the part of the analyst or decisionmaker. That the costs and benefits considered are commensurable enables a clear-cut, objective assessment (conditional, of course, on the normative choice to utilize the benefit–cost principle).

2.2 Normative criteria

For decisions guided by BCA to be not merely rational but also good, the object of quantification and maximization must have moral salience.Footnote ¹ While some philosophers challenge the notion that preference satisfaction has moral salience (e.g., Sagoff, Reference Sagoff1986, Reference Sagoff1994), the key question for the present assessment is what makes preference satisfaction morally salient to those who see it as such. Proponents of BCA argue that policy alternatives should be evaluated based on their consequences for human welfare. They accept (often implicitly) that the satisfaction of economic preferences, as reflected in measures of compensating or equivalent variation, constitutes an increase in human welfare. They take as axiomatic that individuals seek to maximize their own welfare and that each is the best judge of how to manage his or her private affairs with that aim. Concern about antisocial preferences is mitigated so long as resources are allocated on a lawful and mutually voluntary basis.

Thus, at a minimum, WTP estimates used in BCA must accurately reflect preferences; and the act of satisfying preference must plausibly lead to welfare gain. At another level, for BCA to be perceived as good, Zerbe et al. (Reference Zerbe, Bauman and Finkle2006a) suggest that “the information provided is thought useful by those who experience its effects.” The information used must be seen as legitimate and appropriate to the analysis.

There are several important challenges to the inclusion of existence value in BCA in a manner consistent with meeting the positive and normative criteria set out above. These challenges include a variety of interrelated aspects of human cognition and ethical reasoning as they apply to non-use goods. In the following sections, I summarize these challenges (see also Johansson-Stenman, Reference Johansson-Stenman1998) and then spell out how they apply to the BCA criteria.

3.1 Attitudes versus preferences

Departures from standard theory present several related challenges. The first challenge is the possibility that respondents to SP questions express the strength of their attitudes rather than preferences for environmental preservation (even if they communicate rational preferences for other goods). As explained by McFadden (Reference McFadden1999), “[a]ttitudes are defined as stable psychological tendencies to evaluate particular entities (outcomes or activities) with favor or disfavor” (p. 74), while “[p]references are viewed as constructed from more stable attitudes by a context-dependent process that determines the prominence given to various attitudes and the tradeoffs between them” (p. 81). Kahneman, Ritov, and Schkade (KRS; Reference Kahneman, Ritov and Schkade1999) argue that contingent valuation (CV) survey responses represent attitude expressions rather than expressions of preference. They note that responses vary with the elicitation method and are subject to insensitivity to scope: respondents’ stated WTP increases little or not at all with the extent of preservation (Kahneman & Knetsch, Reference Kahneman and Knetsch1992).

Some economists dismiss these observations of scope-insensitivity as attributable to methodological defects (Carson, Reference Carson2012; Kling, Phaneuf, & Zhao, Reference Kling, Phaneuf and Zhao2012) or assert that they in fact may be compatible with standard consumer theory (Amiran & Hagen, Reference Amiran and Hagen2010). The potential compatibility of scope insensitivity with theory does not rule out that stated WTP for preservation reflects attitudes rather than preferences. KRS argue that CV responses exhibiting scope insensitivity are “inevitable manifestations of known characteristics of attitudes and attitude expressions” (p. 204), which Kahneman and Knetsch (Reference Kahneman and Knetsch1992) see as arising from the nature of the pure public good of preservation or “existence,” for which private purchase is not conceivable. The latter notion, which appears consistent with the nuanced scope study by Heberlein et al. (Reference Heberlein, Wilson, Bishop and Schaeffer2005), allows for the possibility that improved SP study design could mitigate scope insensitivity for goods with use value, as suggested by more recent, favorable assessments of CV (Carson, Reference Carson2012; Kling et al., Reference Kling, Phaneuf and Zhao2012), even if individuals have only attitudes and not preferences regarding “existence.”

It may be possible to avoid the attitude expression problem even for non-use public goods by employing alternative SP methods. KRS and Kahneman and Knetsch focus on CV; perhaps well-structured contingent choice methods with consequentiality could avoid attitude expression by eliminating the need for respondents to directly “assign a price.” As noted by Stevens et al. (Reference Stevens, Belkner, Dennis, Kittredge and Willis2000, p. 64), “from a psychological perspective, the process of making choices in the [choice experiment] format may be quite different from that associated with making decisions about WTP … [R]espondents may react differently when choosing among commodities that have an assigned price as compared to making dollar valuations of the same commodities,” especially if those choices are consequential. Evidence for this hypothesis is tied to a behavioral model of contingent voting,Footnote ² in which a respondent votes in favor of the referendum if the expected value of her vote plus the immediate benefits of voting in the affirmative are positive:

(1) $$ P\left(V-T\right)+E\ge 0, $$

where P is the probability the vote is pivotal and acted on effectively by policymakers, V is the WTP for the mooted change in environmental quality, T is the tax the respondent would have to pay if the policy is adopted, and E the inferred monetary value of her attitude expression or warm glow.

Under the (dubious) assumption that the value of the attitude expression is independent of the expected value of the vote, solving the equation for the T such that the left-hand side equals zero (voter indifference) yields

(2) $$ eWTP=T=V+E/P, $$

where eWTP is the WTP inferred from the vote. It includes both “true” WTP and the value of the attitude expression. If attitude expressions are operative, eWTP should fall as P rises, in contrast to the invariance expected under consequentiality in the absence of attitude expression. Experimental studies (e.g., Carson et al., Reference Carson, Groves and List2014) provide evidence in support of invariance (i.e. no attitude expression) for the referendum-based allocation of private goods. But the attitude expression problem is likely to be particularly acute for the provision of non-use, public goods. In a survey-based study, Herriges et al. (Reference Herriges, Kling, Liu and Tobias2010) find invariance of WTP for water quality improvement, the value of which may be dominated by use benefits. In contrast, a survey-based study of WTP for climate change mitigation, which is more likely dominated by non-use value, showed statistically significant correlation between WTP and an index measure that captured respondents’ assessments of the probability that the referendum would influence policy and the probability that policy would be effective (Nepal et al., Reference Nepal, Berrens and Bohara2009).

3.2 Constructed preferences

Even if survey respondents do express clearly defined preferences among alternatives (rather than attitudes), such preferences might not exist prior to the survey. While most economists assume that “preferences are there just to be uncovered” (Vatn, Reference Vatn2004), “an increasing number of researchers believe that the assumption of well-articulated preferences is tenable only when people are familiar and experienced with the preference object” (Payne et al., Reference Payne, Bettman and Schkade1999). In other cases, preferences are constructed at the time that individuals face a new choice situation. This constructive view of preferences may engender further skepticism that respondents express preferences, rather than attitudes, when faced with unfamiliar choices regarding non-use goods. More fundamentally, if preferences are only constructed at the time of the valuation exercise, then it is dubious that any WTP reported could be representative of the broader population or that preservation could make non-respondents better off in the economic sense of preference satisfaction.

3.3 Rule-based decisionmaking and lexicographic preferences

Another, related challenge is that unfamiliar choice situations may lead SP respondents to engage in rule-based decisionmaking rather than assess tradeoffs (McFadden, Reference McFadden1999). In this type of decisionmaking exercise, the individual seeks to identify analogous choice situations, studied or experienced, in which a single choice rule clearly applies. The apparent expression of preference in this context may not reflect a reasoned ordering of options in terms of their contribution to the individual’s well-being.

Moreover, individuals may use rule-based decision criteria even in familiar or well-considered choice situations, as a matter of ethical commitment. For Sen (Reference Sen1977, p. 328), commitment entails “choosing an action that yields a lower expected welfare than an alternative available action.” Such behavior is consistent with a deontological ethics, based on which individuals act according to what they believe is right rather than what is good for them considering only the particular case. A representative sample of Swedes suggests approximately 17 % of the population ascribes to such ethics (Johansson-Stenman, Reference Johansson-Stenman2012). This is not necessarily irrational: the benefits from honoring a commitment on the whole may be expected to exceed the sum of the welfare costs associated with following the rule in each case. Nevertheless, commitment “drives a wedge” between individual choice and individual welfare (Sen, Reference Sen1977; Sagoff, Reference Sagoff2006) as they relate to the options in a given choice instance.

There is also the possibility that deontological ethics motivate lexicographic or modified lexicographic preferences over income and environmental preservation (see Rekola, Reference Rekola2003, for a theoretical treatment). Rekola (Reference Rekola2003) notes that the prevalence of lexicographic choices in SP studies varies widely and has been found to be as high as 79 %. However, lexicographic choices may reflect survey shortcomings – an insufficiently wide range of payment values or a degree of complexity that requires respondents to use simplified decision rules, for example – rather than actual lexicographic preferences (Sælensminde, Reference Sælensminde2006). Veisten et al. (Reference Veisten, Navrud and Valen2006) find very few survey respondents that consistently exhibit choices reflecting lexicographic preferences. Of course, it only takes one person with non-compensable preferences to undermine application of the Kaldor compensation test to the results of a BCA.

3.4 Warm glow

Assuming there does exist a well-defined, compensable preference for the non-use good, it must be somehow disentangled from the “warm glow” derived from contributing to its provision or sticking to one’s principles. For Kahneman and Knetsch (Reference Kahneman and Knetsch1992), that’s a big “if.” They hypothesize that, consistent with observed insensitivity to scope, stated WTP for non-use goods purely reflects the purchase of moral satisfaction. Hanemann (Reference Hanemann1994), in contrast, sees warm glow as a “red herring.” He argues that warm-glow effects seem unlikely if an appropriate payment vehicle is used: “I have seen no evidence that people get a warm glow from voting to raise their own taxes, whether in real life or in a contingent valuation study” (p. 33). It is not clear, however, why the proposed payment vehicle should affect the warm glow some respondents might experience from exercising their agency to choose what they see as a morally good option (Johansson-Stenman, Reference Johansson-Stenman1998). While experimental studies have shown that “taxes” seem to remove warm glow motivation for charitable giving (Eckel et al., Reference Eckel, Grossman and Johnston2005), the study design does not give participants any agency regarding the question of whether or not there will be a tax, and thus their results provide limited insight regarding the question of warm glow in referendum-based SP studies using a tax payment vehicle.

3.5 Altruistic motivation

Finally, if all the foregoing challenges could be overcome to identify a meaningful economic preference for the non-use good, the analyst must still discern whether it is self-interested or altruistic: Do individuals benefit directly from the preservation of the non-use good, or do they benefit from the belief that preservation enhances others’ welfare? It is not obvious that motivation should matter; it does not matter for the valuation of market goods. However, Milgrom (Reference Milgrom and Hausman1993) shows that the BCA is invariant to pure altruism for non-use goods, implying that adding altruistically motivated non-use value to use-oriented WTP for preservation would bias the analysis. McConnell (Reference McConnell1997) confirms Milgrom’s conclusion regarding pure altruism but shows that it does not extend neatly to cases of paternal altruism. If some individuals care about others’ consumer surplus from the preserved environmental good, or its availability to them, rather than caring for others’ utility in general, the paternal altruists’ WTP to satisfy those preferences could affect the BCA. This would happen, for example, if preservation did not otherwise pass the benefit–cost test, but paternal altruists’ WTP was sufficient to cover the net costs to the “beneficiaries” and to the pure altruists who value the beneficiaries’ overall utility. As McConnell (Reference McConnell1997, p. 33) notes, “the problem of the extent of the market is severe [with two types of altruists]. Counting an altruist as paternalistic when he or she is really non-paternalistic means entering as negative a number that should be positive.”

The problem may run even deeper. It seems likely that pure altruism and paternal altruism are not competing motivations so much as they are the ideal and practicable versions, respectively, of the same motivation. Pure altruism as modeled in the literature requires that altruists are knowledgeable of the utility of beneficiaries. This is dubious even in the familial context and manifestly implausible when the beneficiaries could be anyone. Altruists’ assessments of others’ utility must be purely speculative prior to the economic valuation exercise. Precisely because they are incapable of the assessment necessary for pure altruism, pure altruists might behave like paternal altruists, using the quantity or quality of the resource as a proxy for what they really care about. They could be made better off with the knowledge that the resource is preserved (for others’ sake) but worse off upon learning that the benefits to others do not actually compensate the costs.

4.1 Objectivity

“The scientific problem with [non-use value] is that there is no direct evidence for its magnitude that can be refuted with commonly accepted tests based on observable behavior” (McConnell, Reference McConnell1997, p. 35). Existence value is not objectively knowable; it cannot reliably be distinguished from respondents’ “warm glow” or arbitrary quantitative expression of environmental attitudes; and it may not even exist as an economic value for many citizens who have not yet “constructed” clear preferences regarding the relevant tradeoffs. Moreover, there is no obvious method for a benefit–cost practitioner to determine what proportion of the public has well established preferences for the non-use good and is likely to obtain knowledge of the satisfaction or thwarting of those preferences by the regulatory action under consideration.

“A second concern is that the range of possible existence values may well be limitless and certainly extends far beyond the current application of the concept. Because existence value refers to any non-use-related change in the state of the world that affects utility, anything that shapes perceptions of the world becomes potentially eligible for estimation as an existence value. This vastly increases the possible scope of economic analysis and the potential burden of analytical work. If no practical way of limiting the range of existence values can be developed – and none has been established to our knowledge – then there may be no closure to the analytical efforts required of economists and the very concept becomes infeasible” (Rosenthal & Nelson, Reference Rosenthal and Nelson1992, p. 117). While not a sufficient justification to abandon non-use value in principle (Kopp, Reference Kopp1992), it is a possibly insurmountable challenge in practice for the analyst or agency seeking to maintain objectivity. The exercise of trying to determine which non-use goods merit inclusion would seem ripe for bias and political manipulation.

4.2 Commensurability

Stated WTP for the preservation of non-use goods is not necessarily commensurable with the value of typical economic goods. Lexicographic preferences, of course, would be a clear indication of incommensurability, though the mere fact that motivation matters so much for the incorporation of non-use value (but not for the value of other goods) in BCA is also suggestive of non-use benefits’ distinctiveness. They are obviously unlike most goods and services, which are traded in markets, and differ categorically from other non-market goods in that they leave no behavioral evidence of their value. Non-use value is not even theoretically well-defined: Mathematically, the distinction between use and non-use values depends on the specification of the utility function (Freeman et al., Reference Freeman, Herriges and Kling2014).

There are also philosophical and legal considerations. Many would argue that the non-use cost of, say, driving a species to extinction is not an economic cost but a moral one. According to Nozick (Reference Nozick1981, p. 489), “Moral cost differs from other costs – it is not appropriately plugged into maximization principles that focus only on the differences between costs and benefits.” Nevertheless, some proponents of BCA argue strongly for the inclusion of moral “sentiments” (Zerbe et al., Reference Zerbe, Bauman and Finkle2006b) or “commitments” (Posner & Sunstein, Reference Posner and Sunstein2017). Consistent with Krutilla’s contention, Posner and Sunstein argue that there is a real psychic welfare cost associated with violating the moral commitment. They argue further that it may be arbitrary to omit WTP for commitments and thus a violation of the U.S. Administrative Procedure Act. But moral sensibilities are not legally protected interests: one does not have standing to sue simply because she finds an agency action morally objectionable (Sierra Club v. Morton, 405 U.S. 727, 1972). This legal distinction would seem not only to militate against the concern of Posner and Sunstein but to further set non-use values apart from other values included in BCA.

4.3 Connection to economic welfare

Many of the same issues with existence value that pose a challenge to objectivity and commensurability also pose a challenge to the normative salience of stated WTP for non-use goods. Because WTP is only meaningful insofar as it reflects a change in well-being, and because the objectivity of the analysis requires a consistent and exclusive application of economic theory to measure changes in well-being, the eye of the needle that existence value must thread is vanishingly small. The wellbeing – broadly understood – of those with pro-environmental attitudes may well diminish upon learning of the loss of a species or pristine landscape. They may experience negative “affect,” in the psychological jargon, or in Krutilla’s words “acute distress.” But psychological measures of attitudes or affect are not compatible with economic analysis, and stated WTP only dubiously corresponds to economic preference for species preservation or avoiding disutility due to species extinction. At the same time, stated WTP almost certainly does capture warm glow that does not relate to the actual welfare effect of the policy. The normative content of estimated existence values is thus at best ambiguous. An additional challenge arises in the case of lexicographic preferences, in which case the positive Kaldor compensation test no longer reliably indicates even the potential for a Pareto improvement.

4.4 Perceived legitimacy

Challenges to the perceived legitimacy of including existence values in BCA stem from its possible lack of commensurability with other economic values. These challenges may come from proponents and opponents of preservation alike. Individuals with ethical commitments to, or lexicographic preferences for, preservation might object to the use of stated WTP in BCA. While their primary objection might be to the use of BCA as a decision tool when preservation is at stake, they might specifically object to the inclusion existence values in the BCA. They might find objectionable in principle the inclusion of values they see as incommensurable, even if inclusion would tend to favor preservation, and would advocate for decisionmakers to consider environmental values outside of the BCA. To the extent that agency decisions turn on the outcome of a BCA, some citizens might object to the inclusion of possibly paternalistic values. While paternalistic preferences might motivate the purchase of private goods, they do not lead to an imposition on adult “beneficiaries,” who cannot be forced to consume the good or bear its cost. The same cannot be said of paternalistic preferences for non-use goods if they are determinative in the BCA. Thus, even as (or because) they embrace BCA as a guide to allocate resources in a way that is analogous to the market, some will see the inclusion of paternalistically motivated non-use values as illegitimate.

The problems highlighted above do not arise from general inadequacies of SP methods for the estimation of non-market goods, nor are they likely to be resolved by methodological improvements. Rather they are fundamental to the nature of existence value. As Quiggin (Reference Quiggin1993, p. 197–198) notes, “Benefit–cost analysis can provide useful information on the effects of policy proposals, insofar as these effects involve changes in individual consumption of goods and services.” However, “[t]he moral beliefs underlying existence value can be neither consistently incorporated within benefit–cost analysis nor legitimately excluded from consideration.”