1. Introduction
The most comprehensive methodology for assessing efficiency change in public programs or projects based on economic principles is cost-benefit analysis (CBA). CBA is an economic theoretically grounded evaluation process that quantifies in monetary units (monetizes) as far as possible the net value of the impacts of a public intervention (program, project, regulation, etc.) to a given society. CBA aspires to value all (positive or negative) impacts on all members of that society (i.e. on all individual with societal standing).Footnote 1 The economic principles underlying CBA are addressed in the spirit of Heath (Reference Heath and Heath2020b, p. 185) who summarizes CBA “as an expression of liberal neutrality” and “as an attempt by the state to avoid taking sides with respect to various controversial conceptions of the good.”
I set three tasks. First, I lay out 10 of the most important CBA principles to help clarify the basis of CBA and its practice as an aid to decision making by policy makers.Footnote 2 Second, I consider five problems related to operationalization of the principles that I believe impede wider application of CBA. Third, I consider adjustments to individuals’ valuations of impacts, in light of the principles and the current empirical evidence. All three are pertinent because most CBA analysts in democracies take underlying principles as given, largely based on “common knowledge” and on citations from the best available authoritative sources (Sunstein, Reference Sunstein2014).
Government agencies, economic think tanks and applied economists use CBA as an aid in policy assessment. It remains controversial (Frank, Reference Frank2000). Some opposition is based on a philosophical aversion to utilitarianism – or any consequentialist approach – to evaluate public programs or regulations. But some hesitation – rather than opposition – is based on a lack of clarity around some principles and on skepticism about researchers’ ability to operationalize convincing valuations of policy-relevant impacts where markets do not reveal reasonably efficient prices, particularly with respect to valuing public goods and other market failures, such as information asymmetry (Farrow & Doeringer, Reference Farrow and Doeringer2026). Some hesitation may also be based on observing disagreements or silence among CBA theorists about how to address some problems. Relatedly, some governmental guidance manuals downplay issues where CBA theorists disagree or are silent.
The 10 principles are: (i) CBA assesses allocative efficiency, not overall social welfare.Footnote 3 (ii) CBA is primarily based on individuals’ revealed or stated valuations, so on methodological individualism, rather than on the normative significance of various “collectivities” (Max Weber’s terminology). (iii) Individuals can be treated as acting in a rational manner, except with a small number of important exceptions. (iv) Property right allocations are exogenous to CBA, except where the efficiency of an allocation is itself the focus of analysis. (v) CBA treats all persons in a given society as entitled to equal standing, based on a policy or project’s demonstrable positive or negative impact on each member. (vi) In CBA, impact assessment is most directly measured by individuals’ aggregate willingness-to-pay (WTP) or willingness-to-accept (WTA) to derive consumer and producer surplus or on the best proxies for them or some mix of the two. (vii) Individuals value a unit of consumption in the current time period more than the same unit of consumption in a future time period, (but most do so too quickly for those preferences to serve as the guide to the appropriate rate of trade-off between present and future consumption for society (i.e. the social discount rate (SDR)). (viii) Envy decreases aggregate (and sometimes individual) efficiency and so CBA follows a “no-envy” rule in calculating social surplus. (ix) The nation-state is the default boundary for spatial standing based on implied consent to that state’s constitution; subnational jurisdictions with constitutional authority may also have CBA boundary legitimacy. (x) The constitutional-contractarianist principle that legitimizes national jurisdictions can also justify the expansion of spatial scope to supranational contexts where treaties and other agreements are in place, although that extension is more contestable in principle.
Many government agencies in the United States and in other developed democracies, as well as supranational quasi-governmental organizations such as multinational development banks, mandate, strongly recommend or encourage the use of CBA (e.g. DHHS, 2016; HMT, 2022; EPA, 2024). For an overview of guidelines in a range of countries, see Vining and Boardman (Reference Vining and Boardman2026). But the evidence suggests that, at least as judged by the U.S. federal courts, federal agencies’ quality of CBA has been quite variable (Cecot & Viscusi, Reference Cecot and Viscusi2015). A global trend concerning the use of CBA is an increased recognition that comprehensive monetization is not feasible in some circumstances. Another trend is more frequent recommendation to perform distributional analysis, whether as a complement to CBA or integrated into it in some way (Acland & Greenberg, Reference Acland and Greenberg2023a, Reference Acland and Greenberg2023b; OMB, 2023a; 2023b; 2023c; Kenkel, Reference Kenkel2024; Vining & Boardman, Reference Vining and Boardman2026). A requirement for integration into some broader social welfare function raises difficult theoretical and measurement issues, as recent critiques of an Office of Management and Budget (OMB) proposal on this issue shows (Viscusi, Reference Viscusi2024, Reference Viscusi2025; Blomquist, Reference Blomquist2025; Farrow, Reference Farrow2025; Kenkel, Reference Kenkel2025). Whether or not either trend leads to formal mandate changes by government agencies, both are likely to influence the way practicing CBA analysts think and behave.
In the next sections I lay out 10 principles that aspire to clarify the current scope of CBA and what it does, and perhaps equally importantly, does not encompass. At the margins, some of these principles inevitably intersect and it is useful to explore or clarify at least some of these trade-offs. Subsequently, I discuss five problems that could benefit from either further theoretical clarification or from better empirical evidence to provide clearer guidance to CBA practitioners. To encourage more transparency on positions on problems, I am explicit as possible on when adherence to principles justifies modifying subjective individual-derived valuations and when not, given the current state of evidence. To that end, the focus here is on a set of principles and problems: i.e. on the “big picture.”
2. The centrality of allocative efficiency…only
CBA assesses the efficiency of alternative policies and asserts that the “potential Pareto improvement” criterion should be used for assessing it: so the policy alternative with the largest positive net benefits is the most efficient (see, Markovits, Reference Markovits1980; Boardman et al., Reference Boardman, Greenberg, Vining and Weimer2020). Jeremy Bentham first explicitly argued that contractual agreement is the foundation of economic efficiency: “no man of ripe years and of sound mind, acting freely, and with his eyes wide open, ought to be hindered, with a view to his advantage, from making such bargain…” (Bentham, Reference Bentham1818 [1787, p. 2]; see also Persky, Reference Persky2007).Footnote 4
This scoping of allocative efficiency in CBA practice – which the OMB has labeled “traditional CBA” and others as “core CBA” – was largely unquestioned until the 1950s. Since then critics have noted that practice to that time largely ignored individual’s declining marginal utility of income (or, more comprehensively, wealth or consumption) as income rises (Harberger, Reference Harberger1984; Adler, Reference Adler2016; OMB, 2023a; b ; c ). To obtain a plausible proxy measure of individuals’ declining marginal utility of (a unit of) income, given that the expected present value of wealth is much more difficult to estimate (Catherine et al., Reference Catherine, Lu and Paron2025), valuations would be weighted using income groups, say income quintiles. The use of utilitarian weights (language) is intended to avoid an implication that employing them involves “distributional” weighting as the terms redistribution or equity are normally understood, as it pertains only to how an individual is assumed to actually value increments of a good. Utilitarian weighting would not combine efficiency with a distributional (equity) goal and therefore would not obscure trade-offs between them (see Acland & Greenberg, Reference Acland and Greenberg2023a).Footnote 5
Including an estimate of income group declining utility of income as income rises, however, would not imply extension to an inclusion of a measure of a societal redistribution of income or wealth (Harberger, Reference Harberger1984; Boardman et al., Reference Boardman, Greenberg, Vining and Weimer2020; Vining & Boardman, Reference Vining and Boardman2026).Footnote 6 Adding that kind of weighting directly to subjective valuations would require a different – more controversial – form of distributional weighting, often labeled as “inequality aversion” weighting (Adler, Reference Adler2016). This kind of weighting would reflect a “moral concern for the welfare of the poor” (Meltzer & Richard, Reference Meltzer and Richard1981; Robinson, Reference Robinson2024). Some weighting schemes would combine efficiency and some version of equitable redistribution in a unitary monetary metric (perhaps using a numerical equity coefficient greater than 1). But it is difficult to arrive at a plausible aggregate estimate across individuals because of “the complexity of individuals’ preference for redistribution, [that are] influenced by myriad factors like procedural fairness, societal norms, and ethnicity composition” (Tucker & Xu, Reference Tucker and Xu2023, p. 1529).Footnote 7 Additionally, in quite a few developed countries, transfers are more important than taxes in addressing the income, redistributional goal (Boardman et al., Reference Boardman, Greenberg, Vining and Weimer2020, p. 16–18). In sum, the empirical evidence on redistribution across countries is quite mixed and highly contextual (Pecoraro, Reference Pecoraro2014, Reference Pecoraro2017).Footnote 8 Also troubling from a principles’ perspective (as I discuss below) is that some preferences for or against specific redistributions appear to be motivated by (socially costly) envy (Sznycer et al., Reference Sznycer, Seal, Sell, Lim, Porat, Shavi and Tooby2017; Lin & Bates, Reference Lin and Bates2021). Responses to the OMB’s proposals and other evidence suggest that there are differing views among CBA scholars as to whether this latter kind of distributional analysis is better addressed by separately assessing efficiency and equity – for example, employing multi-goal analysis (Weimer & Vining, Reference Weimer and Vining2025) – or by creating a single measure that melds them (see Acland & Greenberg, Reference Acland and Greenberg2023a; Viscusi, Reference Viscusi2024, Reference Viscusi2025; Blomquist, Reference Blomquist2025).
A recent meta-analysis of utilitarian weighting studies found that the elasticity of income using an isoelastic (constant price elasticity) demand curve to facilitate comparison is 1.6, with a range between 1.2 and 2.0 (Acland & Greenberg, Reference Acland and Greenberg2023b).Footnote 9 For CBA purposes, that large a range of weights is disconcerting. Even the use of the low-end (1.2) weighting value could well change the sign of net benefits in some CBAs, a mid-value could change the sign in quite a few analyses and a high-end (2.0) weighting value would likely do so in many cases. It is worth noting that the isoelastic function emphasizes the role of income differences in driving declining marginal utility. Another review of survey responses of the declining marginal elasticity of income did so in over 50 countries (conducted between 1972 and 2005) and found an average elasticity of 1.24, with a range of 1.19 to 1.34 (Layard et al., Reference Layard, Mayraz and Nickel2008). These estimates are considerably smaller than 1.4, but are not trivial. Neither study considers more complex, but plausible, utility functions that have concave and convex inflection points over income changes (Friedman & Savage, Reference Friedman and Savage1948; Lawsky, Reference Lawsky2011).
The role of utilitarian weights in CBA is not a purely theoretical issue. In 2023, the OMB in the United States issued proposed guidance on CBA practice that permitted (but did not require) an adjustment of impacts across income groups and suggested a coefficient of 1.4. Acland and Greenberg (Reference Acland and Greenberg2023a; Reference Acland and Greenbergb ) present the case that it is quite feasible to implement this approach. Regardless, the Trump administration rescinded this draft provision in 2025.Footnote 10 Given the empirical uncertainties, however, it is preferable to present utilitarian weights or qualitative assessments alongside estimates without them, rather than combining them in a single metric. A number of CBA scholars that commented on the OMB 2023 proposal on this issue appeared to reach that conclusion as well (Dudley and Kniesner, Reference Dudley and Kniesner2025).
3. Methodological individualism: Only the minimalist version
Political philosophers over the ages have posited units of analysis that transcend the individual, such as class, gender, caste, property ownership, the “Volk,” religious rank or technical expertise (a “benevolent social planner”), all varieties of “holism.”Footnote 11 CBA, in contrast, normally eschews distinctions between individuals impacted by a policy except for informational and accounting purposes; for example, for distinguishing between the recipients of consumer surplus versus producer surplus. Some commentators conflate methodological individualism with capitalism which seems like overreach with respect to this principle, although one could argue that it has more force with respect to deference to property rights (although private property rights are not directly privileged in these principles) (see Selznick, Reference Selznick1952) and maybe other principles. But following this criticism generally would mean that CBAs of early childhood interventions would be deemed illegitimate because most are embedded within somewhat capitalist systems.Footnote 12
CBA requires adherence only to a minimalist meaning of methodological individualism, as introduced to sociology by Max Weber and to economics by Joseph Schumpeter (for extended discussions, see Heath, Reference Heath and Zalta2020a; Hodgson, Reference Hodgson2007; Neck, Reference Neck2021). Weber emphasized the limited scope of methodological individualism noting that “it is a tremendous misunderstanding to think that an ‘individualistic’ method should involve what is in any conceivable sense an individualistic system of values” (Weber Reference Weber, Roth and Wittich1986, p. 18, as quoted in Heath, Reference Heath and Zalta2020a). Similarly, Schumpeter emphasized its core focus on individual utility: “Almost every modern writer starts with wants and their satisfaction, and takes utility more or less exclusively as the basis of his analysis… it unavoidably implies considering individuals as independent units or agencies. For only individuals can feel wants” (Schumpeter, Reference Schumpeter1909, p. 214). In sum, methodological individualism is a central CBA concept and in combination with allocative efficiency and universal personhood makes up an underlying core trinity of CBA (see Neck, Reference Neck2021).
4. Individuals mostly make rational choices
In applying CBA, analysts generally assume that individual preferences can and should be derived from the choices or decisions that they make (Adler & Fleurbaey, Reference Adler and Fleurbaey2016), or in Bentham’s words, assuming that individuals choose with their “eyes wide open.” More broadly, Falk et al., (Reference Falk, Becker, Dohmen, Enke, Huffman and Sunde2018, p. 1646) note that: “Many theories of human behaviour, in economic and neighbouring disciplines, assume that a set of preferences drives individual decision making.” So while preferences are intrinsically subjective, CBA mostly assumes them to be substantively rational (Shogren & Taylor, Reference Shogren and Taylor2008; Smith & Moore, Reference Smith and Moore2010; Gillingham & Palmer, Reference Gillingham and Palmer2014). This is the case despite the reality that the economic parameters of rationality are actually somewhat vague and so contestable on the margin (e.g. Lieder et al., Reference Lieder, Griffiths and Hsu2018). Indeed, Jones (Reference Jones2021) (surprisingly?) shows that most economic textbooks hardly mention the meaning or scope of rationality.
Despite the usual assumption of rationality in CBA, approximately since the beginning of the 21st century, psychologists, economists and cognition theorists have questioned the rationality of individual choice in a variety of contexts. The most influential is prospect theory (Kahneman & Tversky, Reference Kahneman and Tversky1979; Kahneman et al., Reference Kahneman, Knetsch and Thaler1991; Rachlin, Reference Rachlin2007; Knetsch et al., Reference Knetsch, Riyanto and Zong2012). Adopting it, or close alternatives, results in thinking of eyes as “half-shut” rather than “wide open.”Footnote 13,Footnote 14 There is certainly strong evidence of individual behavioral bias in choice contexts involving significant uncertainty and longer time horizons (Tversky & Kahneman, Reference Tversky and Kahneman1992; Kahneman, Reference Kahneman2003; Bernheim & Rangel, Reference Bernheim and Rangel2009; Vining & Weimer, Reference Vining and Weimer2010; Bernheim, Reference Bernheim2016; Maxted, Reference Maxted2025).Footnote 15
Following Stango and Zinman (Reference Stango and Zinman2023, p. 4), however, it is preferable to limit the scope of CBA-relevant behavioral bias to those that are both reasonably stable and can be shown to exhibit “departures from clear classical benchmarks (e.g. time-consistent discounting, dominance avoidance, mean-zero expectation error).” In the absence of an observed behavior meeting that hurdle, it is more sensible to assume rational behavior, at least in terms of using a given bias as an adequate rationale for adjusting subjective preferences. That hurdle is equally relevant in assessing whether individuals’ bias offset in CBA aggregation: i.e. whether there is evidence of an absence of a zero (or sometimes 1) mean (usually after some normalization; see Stango et al., Reference Stango, Yoong and Zinman2017) or the presence of skewness for each specific type of bias (hereafter, simply “net bias”) when aggregated across individuals for a particular impact.
I consider the issue of adjusting individual subjective valuations based on net behavioral bias initially in the section that considers individual intertemporal consumption choice behavior (as against the SDR), and then specific particular adjustments in the problems section.Footnote 16 I conclude that the evidence for the presence of net bias, and so the rejection of individual behavior as a guide for (at least) the SDR, is strongest with respect to “present-bias.” Below, I also make a case that behavioral bias is an inadequate model for the analysis of addiction resulting from the consumption of many drugs, as this consumption leads to physiological – and more specifically, neurological – change, that is beyond the reasonable scope of “behavioral” change as the word is normally used (Lefreve et al., Reference Lefreve, Adamo, Biron, Misse, Hughes and Thomas2009; Adamo, Reference Adamo2013).
5. The ownership (and so distribution) of property rights is mostly exogenous
Identifying the owners of assets or resources is a prerequisite for CBA (Browne & Ji, Reference Browne and Ji2023; Zerbe, Reference Zerbe2023). In a given CBA, it is necessary to determine whether WTP or WTA is a more accurate characterization of the relevant property rights. A regime does not, per se, need to consist solely, or even primarily, of private property ownership. So, CBA can be performed in the presence of either collective or fractionalized ownership, although the accounting is more complex (Demsetz, Reference Demsetz1967; Vining & Weimer, Reference Vining and Weimer2016). Property rights are a bundle of tangible and intangible rights that can be disaggregated in a variety of ways (Barzel, Reference Barzel1989).
CBA is pragmatic in recognizing that rights can either be legally sanctioned through state institutions or accumulated through custom (e.g. “time immemorial”) and shared norms (Sender & Wood, Reference Sender and Wood2021). From an efficiency perspective, hopefully a given property rights regime creates incentives for owners to maximize the net social benefits of ownership and so fosters economic efficiency (Demsetz, Reference Demsetz1967; North & Weingast, Reference North and Weingast1989). The presence of transaction costs, path dependency and the use of force (conquest) usually means that actual property right regimes fall short of this goal to a lesser or greater degree (Libecap & Lueck, Reference Libecap and Lueck2011; Ayres et al., Reference Ayres, Edwards and Libecap2018). But the de facto acceptance of current allocations certainly does not imply or require that, an existing property rights regime is economically efficient or equitable. Given comprehensive property rights, however, CBA normally treats the in situ regime as exogenous; it does not normally refight or relitigate the Norman Conquest despite the hopes of a displaced Anglo-Saxon elite.
This is not a claim that property rights cannot be changed. New laws and regulations in democracies alter property rights on an almost daily basis, usually on the margin, but occasionally wholesale (Acemoglu et al., Reference Acemoglu, Cantoni, Johnson and Robinson2011; Dahl, Reference Dahl2016). To take an example, over the last decade, a number of nations have restored some property rights of indigenous groups that were “alienated” by “settler” governments over several centuries (Sanchez et al., Reference Sanchez, Leonard and Edwards2023). To take another example, the property rights in, and of, animals are changing (being changed) across western democracies almost on a yearly basis (Weimer & Vining, Reference Weimer and Vining2024). So while normally exogenous, when the purpose of a particular CBA is to analyze the efficiency of the allocation of property rights (often in terms of relative transaction costs), it is legitimate to treat property rights as endogenous and to analyze their evolution, changes in allocation and efficiency (Demsetz, Reference Demsetz1967; Fitzpatrick, Reference Fitzpatrick2006). However, in these cases, that CBA should ideally be explicit on the alternative property rights allocations (the counterfactual). So CBA, with this exception, treats the current property rights as the basis for analysis (Coase, Reference Coase1960).
6. All individuals (with standing) are equal in counting and aggregating impacts
A fundamental principle of microeconomics – and of CBA – is the equality of all individuals qua persons. The history – and evolution – of the meaning of “personhood” shows that equality has not often been the norm or even aspiration (Mill, Reference Mill1869). Adam Smith came close to an economic philosophy of universalistic personhood (Paganelli, Reference Paganelli2018; Easterly, Reference Easterly2021).Footnote 17 Widespread personhood in political and economic terms (economists sometime summarize this as “a property right to political and economic representation”), however, only emerged in the latter part of the 20th century (Weimer, Reference Weimer1997; Benati et al., Reference Benati, Guerrieo and Zaina2022). Globally, there are still many exceptions to the operationalization of generalized personhood. Broad societal personhood, however, underlies property rights (e.g. intellectual property rights) and standing. The right to be counted (equally) may seem irrelevant to CBA, but “stakeholder-weighted” analysis, “economic multiplier” analysis and most versions of “multiple accounts” analysis are usually not consistent with this or other related principles (e.g. Bradshaw, Reference Bradshaw2020). The equality of those affected by policy strongly complements methodological individualism.
7. WTP and WTA are the best measures of valuations but credible proxies also work
Welfare economics assumes that individual preferences can be derived from the choices that individuals make (Adler & Fleurbaey, Reference Adler and Fleurbaey2016). In CBA, benefits are best measured by either an individual’s WTP or WTA compensation for policy impacts (Whynes & Sach, Reference Whynes and Sach2007). If an individual does not own some fraction of the good (and so does not have a property right), then WTP is the appropriate framing and measure. If an individual possesses the property right (“owns the good”), then WTA is the conceptually correct framing and measure. Particularly in the context of public goods, sometimes it is not obvious who should be viewed as the property right holder (Brown & Gregory, Reference Brown and Gregory1999; Knetsch, Reference Knetsch2020). If an individual is deemed to have a property right, that individual’s decision to forgo it is not directly constrained by their income or wealth: one can demand an infinite price (treat it as being “priceless”). In contrast, if an individual does not have the property right, willingness to pay is constrained by his or her income/wealth (i.e. they directly face a budget constraint), unless some adjustment to valuation is introduced. Also, because WTP and WTA are fundamentally about individual valuations of an impact (whether perceived as a cost or a benefit), even when averaged across all impacted individuals, this principle is also consistent with methodological individualism (Heath, Reference Heath and Zalta2020a).
This principle implies that if individuals have preferences for – and demonstrated WTP for – redistribution, those valuations should be included in CBA (Zerbe, Reference Zerbe1998; Alesina & La Ferrara, Reference Alesina and La Ferrara2005). However, there are several complicating factors in deriving a plausible “redistribution coefficient” when aggregating individual valuations, as I discuss in more detail later. To illustrate the complexity here, I make two points. First, evidence shows that those lower in the income distribution in some countries are less likely to favor redistributive policies (Hoy and Mager, Reference Hoy and Mager2021). Second, many people (usually in taxpayer surveys) wish to redistribute only to the “deserving” and not to others; this kind of preference is often hard to match up with other income- or wealth-based redistribution preferences (Alesina & Giuliano, Reference Alesina and Giuliano2009).
When observation of WTP or WTA (and so of relevant surpluses) of the relevant good in markets is not feasible, CBA requires the use of credible proxies, if it is feasible to posit one (Farrow & Doeringer, Reference Farrow and Doeringer2026). These proxies include valuations in related markets, such as some individuals’ willingness to accept higher levels of job risk, to wait in queues, to purchase or not purchase safer but more expensive products or by surveying residents about their valuations of public goods or various amenities (contingent valuation) (Banzhaf, Reference Banzhaf2021). Relying on observed WTP or WTA or close proxies flows naturally from the assumption that individuals or their agents (e.g. in the case of minors) make rational choices. Continuing progress on monetizing impacts and credible proxies is important because considerable evidence suggests that impacts that are not measured and monetized are often undervalued or ignored (Moore et al., Reference Moore, Corona, Griffiths, Heberling, Hewitt, Kaiser, Kling, Massey, Papenfus, Phaneuf, Smith, Vossler and Wheeler2023; McDonald & Powdthavee, Reference McDonald and Powdthavee2025). Ultimately, the question for any proposed proxy for WTP/WTA is: Is that proxy consistent enough with principle(s), including, for example, methodological individualism, (at least constructive) Paretian improvement, property rights, etc. to be normatively legitimate?
8. Individuals prefer present consumption to future consumption (often by too much)
Most CBAs evaluate government projects or regulations that incur costs and provide benefits that occur over quite long temporal horizons. Ultimately, public investment defers the consumption of some individuals to a future time period (hereafter a year) and normally represents an opportunity cost. At the same time, individual saving and investment (say in one’s own human capital) is obviously not irrational (Wang, Reference Wang2006). Ideally, based on several CBA principles, we would prefer to derive both a social rate of time preference (the “pure” rate of time preference is usually assumed to be close to zero) and the SDR based on individuals’ rates of time preference. To do so, however, requires that individuals evaluate their rate of consumption over time periods in a reasonably rational manner, given that these decisions are multidimensional, contextual and probabilistic. Considerable evidence shows, however, that individuals’ intertemporal decisions are particularly prone to behavioral bias (Laibson, Reference Laibson1997; O’Donoghue & Rabin, Reference O’Donoghue and Rabin2015; Stango et al., Reference Stango, Yoong and Zinman2017; Cohen et al., Reference Cohen, Ericson, Laibson and White2020; Maxted, Reference Maxted2025).
Perhaps most importantly for CBA purposes, the evidence shows that “people are on average present-biased” (Imai et al., Reference Imai, Rutter and Camerer2021, p. 1788).Footnote 18 The “on average” shows that present-bias dominates future bias and so its aggregate impact of this bias is asymmetric. The (greater) presence of present-bias is distinct from a presence of considerable heterogeneity around consumption preferences in general (Falk et al., Reference Falk, Becker, Dohmen, Enke, Huffman and Sunde2018). Still, Stango and Zinman (Reference Stango and Zinman2023) do find that “bias is more norm than idiosyncrasy.”Footnote 19 One would expect that both heterogeneity and bias to be most pronounced around the valuation of public goods (or “bads”) and other market failures that are more readily perceived as ambiguous or hypothetical.
A range of empirical evidence suggests that an “exponential growth” model explains present-bias reasonably well: when individuals discount consumption, not only do they undervalue year 2 consumption relative to year 1 consumption, but they undervalue year 10 consumption relative to year 9 consumption at a much greater rate (i.e. exponentially). Maxted (Reference Maxted2025) reviews the empirical evidence and provides a range of estimates of the significant welfare (utility) losses associated with the exercise of present-bias.
To quantify and monetize the value of an impact over time from a societal perspective, it is necessary for someone (more on that below) to weight the factors (“discount factors”) to apply to any impact that occurs in different years (these factors are the inverse of the SDR) (Moore & Vining, Reference Moore and Vining2018). For many projects, it is reasonable to assume an intragenerational time frame; (i.e. the same notional person at both temporal endpoints), but for many environmental and social projects it is not.Footnote 20 For determining the level of the SDR, it is important to distinguish between consumption flows versus monetary flows in terms of assessing utility loss and so the cost of individual present-bias. As O’Donoghue and Rabin (Reference O’Donoghue and Rabin2015, p. 274) emphasize “present bias operates on utility…in the absence of liquidity constraints and uncertainty, choices between different streams of monetary payments are driven entirely by maximizing the present discounted value of wealth, and preferences are irrelevant.”
The strength of present-bias among other factors suggests that individuals’ intertemporal preferences should not be used to derive discount factors or a credible SDR rate (Moore et al., Reference Moore, Boardman, Greenberg, Vining and Weimer2004; Tol, Reference Tol2024).Footnote 21 Indeed, many people find it difficult to conceptually even distinguish between the (real) SDR and the (nominal) interest rates that they face in financial and related markets (e.g. Freeman et al., Reference Freeman, Groom and Turk2020). But CBA theorists (either behind a veil-of-ignorance or, I admit, acting as benevolent social planners) have been able to derive discount factors and to come up with a reasoned, if contested, SDR level (for a recent review of the issues and national levels of the SDR, see Groom et al., Reference Groom, Drupp, Freeman and Nesje2022).Footnote 22 One might ask: Exactly what principle is involved here? Essentially, it is a negative one: individual intertemporal choice behavior cannot be used to arrive at a credible SDR, so it must come from somewhere else. Many democratic governments mandate or recommend a (real) rate in the 2 to 3 percent range, although some still use considerably higher rates (Moore et al., Reference Moore, Boardman and Vining2020; Groom et al., Reference Groom, Drupp, Freeman and Nesje2022).
9. “No-Envy” applies to surplus assessment
In economics, the achievement of a no-envy outcome is often seen as a – perhaps the central – principle of fairness, rather than of efficiency, as canonically represented by the two-person (“I cut, you choose”) cake-cutting exercise (Varian, Reference Varian1974). The evidence, however, shows that many individuals display envy (or more broadly a concern for their “relative position”): they are willing to reduce their own pay out to reduce it to others (Frank, Reference Frank2005). In both ultimatum and dictator games, the evidence shows participants’ sense of (un)fairness often dominates economic self-interest. Envy, though, is often costly for individuals (an actual negative internality?) and almost always for society (Mujcic & Oswald, Reference Mujcic and Oswald2018; Knoll et al., Reference Knoll, White and Meloy2022). In sum, “resources are expended in unproductive, from a societal perspective, attempts to harm the envied” and does not pass a Paretian improvement test (Fehr & Camerer, Reference Fehr and Camerer2007; Fischer & Torgler, Reference Fischer and Torgler2013; Grossman & Komai, Reference Grossman and Komai2013, p. 2). A no-envy principle is also consistent with minimizing the likelihood of conflating efficiency with equity (Kniesner & Viscusi, Reference Kniesner and Viscusi2003).
10. Constitutions can substitute for individual contracts and direct agreements
The constitutions of (somewhat) democratic nations (and other jurisdictions with constitutional authority) can be interpreted as a legitimate substitute for a contract between all the individuals of the society, in practice, their representatives with on-the-ground negotiating authority, such as the “founding fathers” or Nelson Mandela. This assumes that behind a “veil-of-ignorance” as to one’s ex post position (i.e. in an “original position”), we can treat constitutions as efficient, on the assumption that the negotiators shared an ex ante expectation of mutual gain when they ratified (Harsanyi, Reference Harsanyi1953; Buchanan, Reference Buchanan1990; Vanberg, Reference Vanberg2020; see Beeson et al., Reference Beeson, Chilton, Metcalf and Nielsen2024 for a discussion of a number of specific behind-the-veil utility functions).
The crucial point is that national constitutions, one-way-or-another, have overcome a high adoption threshold (usually a supermajority) and so more closely approximate an actual Pareto improvement, as in contracts between individuals (see Zerbe, Reference Zerbe2020). All residents are deemed to have acceded to such a constitution behind a veil-of-ignorance (Munger & Vanberg, Reference Munger and Vanberg2023). Implied consent by residents to a constitution gives them the greatest legal, economic and political normative force and legitimacy and is a valuable property right in itself (Marshfield, Reference Marshfield2013; Zerbe, Reference Zerbe2020, Reference Zerbe2023; Cope, Reference Cope2023). The principle does not require that everyone regards every constitutional provision as personally improving, only that overall compliance makes them better off (Samuelson, Reference Samuelson1956; Buchanan & Tullock, Reference Buchanan and Tullock1965; Buchanan, Reference Buchanan1975; Vermeule, Reference Vermeule2001). In essence, this principle adopts a portfolio approach to public programs, which is consistent with at least the Hicksian view of Kaldor–Hicks potential Pareto improvement (Persky, Reference Persky2007). Based largely on this principle, the nation-state is the default CBA spatial scope boundary.
An impartial observer might argue that the “we would have all agreed to it” assumption is a stretch because few historical constitution makers represented the racial, economic or social makeup of their society or universal personhood. It is impossible to argue that the U.S. Constitution passed the test if one regards slaves as constructive citizens, as the personhood principle requires (Mill, Reference Mill1850). If there had been an operational veil-of-ignorance/uncertainty or a strong Smithian “impartial spectator” assessment (see below) – i.e. if the founders had not known whether they would be slaves or slave-owners – they would have written the U.S. Constitution differently (see original constitution, Article I, sections 2 and 9 and Article IV, section 2, clause 3). Smith argued strongly against slavery, partly I surmise on impartial spectator and veil-of-ignorance grounds. In spite of these historical realities and limitations, national constitutions of – now – universal personhood democracies have the highest CBA boundary legitimacy.
Application of the constitution-as-contract principle implies that it could also apply to subnational entities with constitutional authority – essentially those in federal nations – and to indigenous groups (nations) that concluded nation-to-nation treaties with settler governments (Marshfield, Reference Marshfield2013; Burgess & Tarr, Reference Burgess and Tarr2021; Boardman et al., Reference Boardman, Greenberg, Vining and Weimer2022). Across the globe, many subnational jurisdictions and a few other entities cross this constitutional threshold, either generally or in specific policy domains (Watts, Reference Watts1998; Burgess & Tarr, Reference Burgess and Tarr2021; Cope, Reference Cope2023). They almost always did so through explicit or implicit super-majoritarian processes (Boardman et al., Reference Boardman, Greenberg, Vining and Weimer2022; Sharma, Reference Sharma2022). In many cases, constitutional authority derives from the fact that the subnational entity preceded a subsequent national government and conditioned their consent to it (Kaye, Reference Kaye2012). This set of circumstances applied to the 13 American colonies that had constitutions at accession to the U.S. Constitution, Swiss cantons and some jurisdictions in Africa, Asia and the Americas (Gardner, Reference Gardner2008). However, I am aware of no cases (with the possible exception of Belgium) where constitutional status extends to metropolitan regions, cities or counties, regardless of their population size, as they do not face a constitutional threshold test. As a result, economic multiplier studies that focus on local or regional benefits do not qualify as CBAs (Taks et al., Reference Taks, Kesenne, Chalip, Green and Martyn2011). Even studies that critique multiplier studies may not qualify as CBAs, even when claiming to be so (e.g. Sepuvelda, Reference Sepuvelda2022).
11. Supranational agreements can substitute for contracts (and affect standing)
The same logic can be applied to treaties between nations: i.e. “International Paretianism” (Steele, Reference Steele, Budolfson, Pherson and Plunkett2021). (For comparative statics and dynamics on some of the efficiency consequences, see Colliard and Steinbach (Reference Colliard and Steinbach2025)). However, this is a contestable extension because the potential for contract enforcement, and so an assessment of constructive contract reality, is more speculative in binational contexts. Applying this logic, however, provides an economic rationale for giving citizens of the other treaty nation(s) standing in some, otherwise, national-level CBAs.Footnote 23 This would have a significant impact on the net benefits of CBAs where costs are borne within one nation, but benefits are spread across multiple nations (as is the case with, for example, climate-warming abatement policies).
This principle does not imply adoption of global standing, as some respected CBA analysts advocate in some circumstances (Whittington et al., Reference Whittington, Carson and Sterner2023). The logic for global standing appears to rest either on a global human right to the ecosystem or Anthropocene or on a market failure argument that the global atmosphere is a pure public good, and ignoring property rights or contractual agreement logic (Springmann et al., Reference Springmann, Godfray, Rayner and Scarborough2016; Whittington et al., Reference Whittington, Carson and Sterner2023). But others argue there should be very few or no exceptions to a national standing perspective (Rowell, Reference Rowell2015; Fraas et al., Reference Fraas, Lutter, Dudley, Gayer, Graham, Shogren and Viscusi2016). In 2023, OMB did raise the potential application of supranational or global standing, but was obscure as to the permissible circumstances for doing so. I suggest that a principle-based contractarianism rationale for standing requires an explicit agreement and at least a notional reciprocal enforcement mechanism. Meeting both conditions is most plausible in the presence of somewhat symmetrical cross-border externalities, and in other treaties where the threat of withdrawal signals credible potential enforcement (Kumm, Reference Kumm2004; Mendez, Reference Mendez2017).
These principles do not represent an argument that analysis of distributional impacts of policies or regulations with respect to income is not valuable or appropriate, either independently or as a complement to a particular CBA (see Weimer & Vining, Reference Weimer and Vining2025), but only it is problematic to conflate it with efficiency. Additionally, recognizing the legitimacy of distributional policy analysis is not the same as concluding that there is convincing evidence of inequality aversion redistributive preferences that justifies adjustments to individuals’ subjective preferences.
12. Issues related to more use of CBA: Yes, no, maybe?
CBA requires aggregating (and averaging) the valuations of impacted members of a society in order to calculate gross, and then net, benefits (or costs). At least based on this set of principles, adjustments to subjective preferences and valuations are justifiable if there is evidence of substantial net asymmetry – whether because of a net overestimation or net underestimation of value – in aggregating valuations to the societal level. So, for example, with respect to behavioral bias, the issue is not the existence of substantial bias; the evidence is clear on that. Rather the issue is whether there is net bias after aggregation. In the absence of that evidence, an adjustment is not justifiable until it emerges.
Apart from the SDR (where I argue the evidence shows that that individuals’ behavior cannot be used as an adequate or appropriate guide to social choice), there are five issues or problems that might justify modifications to subjective valuations.Footnote 24 Before addressing them, though, there is a necessary precursor question. Assuming that it is appropriate on social efficiency grounds to alter subjective valuations in some contexts, on what basis could one assess modifications that can still claim to derive from methodological individualism and related principles? A “benevolent social planner” stance, for example, cannot easily claim to be consistent with several of the individualistic principles proposed above (Zuber, Reference Zuber2011; Sugden, Reference Sugden2013).
There are two stances that could plausibly claim to retain some degree of inferred methodological individualism in the absence of individual revealed choice.Footnote 25 Adam Smith essentially described both. The first is a person contemplating choice “in the mild sunshine of undisturbed tranquility” (Smith TMS III.3). This is a minimalist version of Smith’s “impartial spectator.” This person is self-reflective and thoughtful, but socially embedded in (in Smith’s case the milieu) of 18th- century Britain (see also Sen, Reference Sen2010; Robinson & Hammitt, Reference Robinson and Hammitt2011; Paganelli, Reference Paganelli2016).Footnote 26 The second, more abstract, impartial spectator is a person behind a veil-of-ignorance as to her future economic and social status or, indeed, whether she would be born in Britain, China or India (see Harsanyi, Reference Harsanyi1975; Huang et al., Reference Huang, Greene and Bazerman2019; Beeson et al., Reference Beeson, Chilton, Metcalf and Nielsen2024).Footnote 27 Some principles rely on the reasonableness of a veil, for example, for thinking about constitutions as contracts. But, individuals behind a veil-of-ignorance might well reach quite disparate conclusions about optimal redistribution and might well display the same dispersion of valuations as the recent empirical evidence shows not-behind-the-veil.Footnote 28 For the purpose of “normalizing” individual “in-the-moment” choices, either stance is potentially defensible on methodological individualism grounds, but that of a tranquil individual in mild sunshine appears much closer to a sovereign consumer. The minimalist stance seems most appropriate for many CBA purposes, particularly where market prices are operational, if not socially efficient because of market failures or other market imperfections (Weimer & Vining, Reference Weimer and Vining2025). Indirectly, for example, it is the rationale for mandating the provision of product information to consumers prior to the purchase of post-experience goods (Vining & Weimer, Reference Vining and Weimer1988).
There are three measurement or evidentiary problems that might justify adjustments to individuals’ valuations of specific impacts: an absence of an existing plausible and evidence-backed monetary estimate for an impact, the legitimacy of aggregating the heterogeneous preferences across individuals and the presence of discrepancies between WTP and WTA valuation for the same good.
12.1. The problem of incomplete monetization
As most analysts are well aware, an important barrier to comprehensive monetization in CBA is the absence of plausible and defensible monetary measures for specific impacts (Masse & Barnett, Reference Masse and Barnett2002; Vining & Weimer, Reference Vining and Weimer2010, Reynolds et al., Reference Reynolds, Temple, White, Ou and Robertson2011; Kuklinski et al., Reference Kuklinski, Fagan, Hawkins, Briney and Catalano2015; Vining et al., Reference Vining, Moore and Boardman2023; Vining & Boardman, Reference Vining and Boardman2026). This occurs most frequently in “soft” social policy arenas or for projects with long (sometimes multi-century) impact horizons, for example, a measure of the value of preventing below-legal-age pregnancy (Rosenthal et al., Reference Rosenthal, Ross, Bilodeau, Richter, Palley and Bradley2009). Comprehensive monetization is simply too costly in some cases, especially when it requires general equilibrium analysis (GEA) to pick up important impacts, for example the benefits of vaccination policy (e.g. Jit et al., Reference Jit, Hutubassy, Png, Sundaram, Audimulam, Salim and Yoong2015). However, there has been progress over the last 20 years in monetizing (“monetary conversion factors”) many impacts despite the absence of prices or (socially) efficient prices (e.g. McCollister et al., Reference McCollister, Yang, Sayed, French, Leff and Schackman2017; Farrow & Doeringer, Reference Farrow and Doeringer2026). Federal and state agencies and regulatory oversight bodies all recognize that not all impacts can be monetized or even quantified. So, practicing CBA analysts must monetize where credible, quantify where possible and go qualitative assessment when there is no alternative. An inability to monetize is clearly a problem for CBA practice, but given ongoing development of credible proxies, it is not a fundamental barrier to wider use. In operational practice, this encourages the use of multi-goal analysis as a complement to CBA, rather than as a substitute for it (Vining & Boardman, Reference Vining and Boardman2026). As discussed earlier, this is consistent with the current guidance of many national governments.
12.2. The problem of aggregating individual preferences
A second measurement problem is whether and how to aggregate heterogeneous individuals (and their preferences) into “a representative agent.” The problem of the extent of the actual representativeness of the notional agent is one shared with other applied economic fields, such as behavioral finance (Blackburn & Ukhov, Reference Blackburn and Ukhov2013). Given the principle of universal personhood, CBA requires preference aggregation to a societal level (however spatially defined) for accounting and informational purposes, rather than for prioritizing the valuations of some individual or groups over others. Some libertarian commentators argue that any aggregation of preferences contravenes methodological individualism (Adler & Posner, Reference Adler and Posner1999). That is not a necessary reading of minimalist methodological individualism as advanced by Weber or Schumpeter (Arrow, Reference Arrow1963; Sen, Reference Sen1970). Treating individual indifference curves and so elasticities as additive (homothetic preferences) is a stronger assumption that is analytically very useful for CBA purposes, as it allows one to treat Marshallian demand curves as linear in income. Heufer and Per Hjertstrand (Reference Heufer and Per Hjertstrand2019a, Reference Heufer and Per Hjertstrandb ) show that it is reasonable to follow the additive rule in general (see also Chambers et al., Reference Chambers, Echenique and Shmaya2017); it is certainly standard CBA practice.
There is certainly compelling evidence of heterogeneous preferences in some CBA-relevant contexts (e.g. Stango et al., Reference Stango, Yoong and Zinman2017; Alberini et al., Reference Alberini, Scasny and Bigano2018; Weimer, Reference Weimer2020). However, the issue for CBA is not the existence of heterogeneity, but (as with bias) rather it is the net degree of asymmetry around some specified variable mean (Weimer, Reference Weimer2020) to assess its CBA importance. Currently, the commonest disaggregation is to estimate age or life-year cohorts rather than using a more inclusive “all-individuals” mean or median to facilitate more accurate estimates of WTP valuation (Aldy & Viscusi, Reference Aldy and Viscusi2007; Keller et al., Reference Keller, Newman, Ortmann, Jorm and Chambers2021).Footnote 29 This is particularly useful for impacts that either decline or increase over the temporal horizon or that only occur after decades (as in, for example, concussion recovery analysis (Giza et al., Reference Giza, Kutcher, Ashwal, Barth, Getchius, Gioia, Gronseth, Guskiewicz, Mandel, Manley, McKeag, Thurman and Zafonte2013). Increasing the accuracy of prediction around an impact by disaggregation contravenes no principle.Footnote 30
12.3. The problem of WTP versus WTA discrepancies
A clear problem for CBA principle is evidence of divergence between individuals’ valuation of required WTA compensation for a perceived loss or offered gain, as against the more usually offered WTP valuation (Kahneman, Knetsch, and Thaler Reference Kahneman, Knetsch and Thaler1991; Tuncel & Hammitt, Reference Tuncel and Hammitt2014). Neoclassical economics assumes that these valuations are approximately equal. The empirical evidence, however, shows that discrepancies can be quite large and vary depending on whether a given impact is presented to respondents as a benefit increase or as a reduced loss. The differences appear to derive most frequently from idiosyncratic differences in individuals’ reference points (Whitecliff & Howley, Reference Whitecliff and Howley2025), which are, in turn, often based on their current (status quo) position (Knetsch et al., Reference Knetsch, Riyanto and Zong2012; Knetsch, Reference Knetsch2020); as a result these points are essentially endogenous (O’Donoghue & Somerville, Reference O’Donoghue and Somerville2018). This issue is not directly relevant to most observed choice behavior, such as valuations derived from employment choice, so it is mostly limited to stated choice methods, such as in contingent valuation surveys. The use of WTP when WTA is the more accurate framing (or vice versa) based on who actually possesses the property rights is incorrect, but sometimes unavoidable because of a lack of evidence. Similarly, the framing of a given valuation change as an increased loss versus a reduced gain often cannot be reliably matched to existing property rights and so is not (yet) a basis for an adjustment. Sometimes the only practical answer is to “split the difference.”Footnote 31
In sum, none of these problems justify systematic adjustments to individual valuations. There are two other problems relating to the use of individual valuations in particular evaluation contexts that do justify adjustments.
12.4. The problem of behavioral bias: Is there evidence of net bias in aggregate valuations?
As already pointed out in principles, a potential problem for CBA derives from the differences between the neoclassical and behavioral views of the efficacy of individual rational choice. The differences are important because they challenge the reliance on subjective preferences for impact assessment. The problem arises particularly in making intertemporal valuation of the same good, in individuals’ valuations of (abstract) public goods or in the presence of addiction that develops temporally. In these contexts, CBA analysts have to choose between various proxies for revealed preferences, based on either an individual’s stated valuations or on the individual’s valuation as inferred by a third party mimicking individual preferences. That mimicry could be based on simply supposing a more reflective, and so rational, individual or by inferring individuals’ preferences indirectly behind a forward-looking (possibly either self-interested or altruistic) veil-of-ignorance. Alternatively, analysts (or government mandates) could abandon an individualistic stance for specific impacts or types of impacts (and at least some principles on some margins) and adopt some version of the benevolent social planner stance (see Sugden, Reference Sugden2013).
Neoclassical economics treats individual behavior as consistently rational with respect to axiomatic microeconomic precepts (transitivity, separability, etc.).Footnote 32 In contrast, behavioral economics treats behavior as more limitedly rational when judged against the neoclassical standard (Weimer et al., Reference Weimer, Vining and Thomas2009; Vining & Weimer, Reference Vining and Weimer2010; Cutler et al., Reference Cutler, Jessup, Kenkel and Starr2016; Weimer, Reference Weimer2017; Bernheim & Taubinsky, Reference Bernheim, Taubinsky, Bernheim, DellaVigna and Laibson2019; Massin & Miera, Reference Massin and Miera2020; Rose, Reference Rose2022; Maxted, Reference Maxted2025). For CBA, the differences between the two goes to the normative validity of individual behavior with respect to either revealed or stated WTP and WTA and to other proxies for them. If significant or frequent adjustments are deemed necessary to (then so-called) individual, and subsequently aggregate, WTP and WTA, the claim that CBA is based on allocative efficiency, methodological individualism and Pareto-based efficiency is seriously eroded. For a few theorists and analysts, adjustments will be based on a binary choice between the neoclassical and the behavioral, while for many others it will be based on assessing the consequences of reduced rationality and devising a continuum of the required degree of adjustment. For practicing analysts, the question is usually an instrumental one: “Can you give me a reasonably comprehensive set of justifiable adjustments for various kinds of impacts”?
What is the evidence of net bias in preferences that would justify an adjustment to aggregate valuations for a given impact?Footnote 33 I conclude that the best evidence of net bias pertains to present-bias, which biases choice in favor of present, as against deferred, consumption. Convincing evidence of net bias, however, does not extend to loss aversion because of the endogeneity of reference points. Present-bias is mostly relevant in CBA to intertemporal consumption choices beyond a few years because of the nature of discounting mechanics. What is the empirical evidence on the asymmetric impact of other biases on aggregate valuations? Stango et al., (Reference Stango, Yoong and Zinman2017) provide the most pertinent, if indirect, evidence based on a multiyear representative U.S. survey sample. Their main focus is on the temporal stability of within-individual bias and on the degree of correlation across biases, but they do indirectly analyze 17 biases that they conclude display bidirectionality. My reading is that their analysis does not provide unambiguous evidence of much net bias despite the presence of endemic bias (see especially their tables 5–7 and appendix tables 1–10). If this assessment is accurate, a theory-derived SDR is the most justifiable adjustment based on bias, as it implicitly corrects for the impact of individual present-bias.
12.5. The problem of physiological addiction
The presence of widespread addiction is problematic for almost any theory of rational individual economic behavior in the context of CBA. There are a number of theoretical approaches. One is to treat addictive behavior as primarily a manifestation of behavioral bias in individual decision-making. I argue below that “behavioral” bias cannot comprehensively explain at least some addiction. An alternative approach, more in the neoclassical spirit, is to explain the addiction process as being largely rational and not amenable to efficiency-oriented policy prescription, which I address below. A third approach is to view addiction as some mix of market failures, including as an internal cognitive struggle (I have used this explanation myself). I argue here that all three are inadequate as explanations of neurophysiological addiction, and so as guidance on the efficacy of adjustments to subjective preferences. Instead, I propose that parasitism provides a more plausible model, at least of some kinds of addiction.
In common usage, addiction covers activities ranging from intermittent gambling to (frequently fatal) fentanyl ingestion (McConnell et al., Reference McConnell, Brandner, Blank, Kearns, Koob, Vendruscola and Tunstall2021). So, the label covers both consumptive and non-consumptive actions and behaviors. That conflation is unhelpful for analysis because the consumption of some substances, particularly synthetic opioids and methamphetamine stimulants (such as cocaine), is difficult to explain using either neoclassical or behavioral economics assumptions alone (see Wise & Robble, Reference Wise and Robble2020). Most illegal drug consumers are initially unaware of the longer-run consequences of consumption (Wise & Jordan, Reference Wise and Jordan2021; Koijam et al., Reference Koijam, Singh, Nameirakpam, Haobam and Rajashekar2024), so this aspect of addiction is consistent with standard welfare economics in the form of (post-experience) information asymmetry. However, it is inadequate as an explanation of continuing consumption as consumers become aware of some consequences.Footnote 34
Recent neurobiological research demonstrates that drug-induced changes to the brain occur at three distinct levels: the molecular, the cellular and at neuron and synaptic connections. So, “addiction…is at its core biological: changes that a physical substance (drug) causes in vulnerable body tissue (brain)” (Nestler, Reference Nestler2005, p. 9). Understanding the neurological consequences at all three level has progressed rapidly over the last two decade, primarily through the use of animal models, human brain imaging and cadaver biopsies (Tomkins et al., Reference Tomkins, Hall, Chen, Li, Wu, Hsu, Couture and Riggs2012; Koob & Volkow, Reference Koob and Volkow2016; Volkow & Blanco, Reference Volkow and Blanco2021). The brains and physiologies of infants, children and adolescents are particularly susceptible to the harmful consequences (Poikolainen, Reference Poikolainen2023). Koijam et al., (Reference Koijam, Singh, Nameirakpam, Haobam and Rajashekar2024, p. 1) summarize the range of neurophysiological consequences as follows:
Drug addiction is a complex disease affected by numerous genetic and environmental factors. Brain regions in reward pathway, neuronal adaptations, genetic and epigenetic interactions causing transcriptional enhancement or repression of multiple genes induce different addiction phenotypes for varying duration.
As this quote emphasizes, there is increasing recognition of the importance of epigenetic impacts. Furthermore, “The definition of epigenetics has evolved to include both heritable and stable changes in gene expression within mature, post-mitotic cells without alterations in the DNA sequence” (Vassoler & Sadri-Vakili, Reference Vassoler and Sadri-Vakili2014, p). The multiplicity of change pathways that come into play over time, combined with variable genetic predispositions, means that, while epigenetic impacts are theoretically reversible, in practice many are not (Neilsen et al., Reference Neilsen, Utrakar, Reyes, Simons and Kosten2012; Muller, Reference Muller2019; Kaplan et al., Reference Kaplan, Xu, Abreu and Feng2022). The probability of irreversibility weakens a normally reasonable assumption of stable preferences (as in rational addiction) and is qualitatively and categorically different from most drivers of behavioral change. An implication to be drawn from Koijam et al., Reference Koijam, Singh, Nameirakpam, Haobam and Rajashekar2024, p. 1) is that there are deleterious intergenerational epigenetic impacts from some drug ingestions:
Addictive drug use causes epigenetic alterations and similarly epigenetic changes induced by environment can promote addiction. Epigenetic mechanisms include DNA methylation and post-translational modifications like methylation, acetylation, phosphorylation, ubiquitylation, sumoylation, dopaminylation and crotonylation of histones, and ADP-ribosylation. Non-coding RNAs also induce epigenetic changes.
I argue that it is more evidence-based to frame the neurophysiological addicting process as (intraspecific) parasitism: human agents motivated by both individual and organizational (self) interest foster harmful consumption and dependence (Ruxton & Broom, Reference Ruxton and Broom2002; Beekman & Oldroyd, Reference Beekman and Oldroyd2008; Korb & Heinze, Reference Korb and Heinze2016; Kneiss, 2024).Footnote 35 Parasitism is ubiquitous in nature and is even a force in major evolutionary changes (Maynard Smith & Szathmáry, Reference Maynard Smith and Szathmáry1999).Footnote 36 So, its presence in hominid contexts should not be surprising, although intraspecific manifestations, particularly, appear to surprise or even shock some analysts and policy makers. In the human addiction context, they might experience this process as being (somewhat?) more palatable if it is analyzed as an “exploiter/victim” or “predator/prey” interaction, rather than as parasitism (see Brander & Scott Taylor, Reference Brander and Scott Taylor1998; Brockhurst et al., Reference Brockhurst, Buckling and Rainey2006; Negus, Reference Negus2024).
Addiction is different from other distortions posited to arise from bias because of the degree (not necessarily the existence) of human intentionality and “innovation” to foster and maintain addiction (e.g. Wagner & Rosen, Reference Wagner and Rosen2014; Coccia & Watts, Reference Coccia and Watts2020; Maclean et al., Reference Maclean, Mallett, Ruhm and Simon2022). As Volkow et al. (Reference Volkow, Michaelides and Baler2019) explain: “Ingenuity has enabled humans to extract and refine highly reinforcing stimuli against which naturally occurring reinforcers cannot easily compete” (see especially their figures 1 and 2 for diagrammatic summary of neurological impact sites). Many of these induced changes may appear to be, and in some senses are, “behavioral” responses (and this is common in a “host’s” reaction to parasitism, for example, infected fish swimming closer to the surface), but the causal mechanisms are primarily biological.
What are the other proposed alternatives to bias theory as the driver of addiction? Some CBA theorists accommodate addiction within market failure theory and welfare economics in one of two ways. First, some do so by framing some addiction as a result of information asymmetry (Levy et al., Reference Levy, Norton and Smith2018). A weakness of an information asymmetry framing alone with respect to addiction, however, is that it ignores the inability of many addicts to learn from provided information or from the disutility of early, or even multiple, consumption episodes for certain drugs. (Persistent graphic information provision, though, does appear to have played a role in the decline in smoking and nicotine addiction in many countries over decades). Second, other theorists do so by framing some addiction as a negative externality and explaining the addictive behavior as an internal-to-the-addict dysfunctional cognitive process and relabeling it as an “internality” (a la Jekyll and Hyde) (Weimer et al., Reference Weimer, Vining and Thomas2009; Kenkel, Reference Kenkel2025). Obviously this process is posited using economic, rather than neurobiological, terminology and analysis. Apart from just positing information asymmetry, Allcott et al., (Reference Allcott, Lockwood and Taubinsky2019, p. 210) also argue “consumers may face problems of self-control and time-inconsistency” and so essentially revert to treating addiction as a severe form of behavioral bias.Footnote 37
There are, however, a number of weaknesses with either internality or information asymmetry explanations of addictive “behavior.” First, the addiction label, as commonly used, covers a very wide spectrum of activities, behaviors, responses and consequences, as discussed above. The inherently wide range of any realistic addiction causal taxonomy suggests that the internal cognitive struggle framing is over inclusive, and ultimately unscientific.Footnote 38 Second, it is doubtful that the “externality-to-an-internality” concept is fit for this particular CBA purpose.Footnote 39 An economic externality is inherently based on contracting parties imposing costs on some third party. An internal-to-the-addict “contract” breech is an unconvincing extension to many addiction phenomena. Furthermore, the internality frame does not deal with the temporal process of many addictive progressions, nor the role that external actors with harmful (negative externality-inducing) incentives play in the “addicting” process.
Parasitism can be applied in the addiction context as a process where “drug dealers” (broadly understood as multiple actors along a supply chain) derive profit from strategically targeting those predisposed to addictive substances (Bichler et al., Reference Bichler, Malm and Cooper2017; Moeller, Reference Moeller2018; Hughes & Libersat, Reference Hughes and Libersat2019).Footnote 40 This addiction-inducing process is consistent with a model of drug consumption that consists of a reinforcing sequence of consumption choices over time, rather than as a single choice (Hyman, Reference Hyman2005; Rachlin, Reference Rachlin2007; Negus, Reference Negus2024). This form of parasitism has some similarity to brood (or “social”) ectoparasitism found in avian species (such as the cuckoo) that search for weakened victims (Ewald et al., Reference Ewald, Strack and Orsini2019). However, as chemicals are ingested, drug parasitism also has some similarity to classic endoparasitism (Olano et al., Reference Olano, Weller, Guerrant and Walker2011).
While some may view a parasitic model of addiction as too metaphorical, I posit it is less so – and more consistent with neurological science, evolutionary biology and some economics (Mehlum et al., Reference Mehlum, Moene and Torvik2003) – than either the “internality” or “rational addiction” framings discussed above. Indeed, a parasitic framing is essentially the one that many public plaintiffs have relied on in successful briefs against pharmaceutical companies over the opioid crisis (Mayor & City Council of Baltimore v. Purdue Pharma L.P. et al., 2018).Footnote 41
The rational addiction model preceded, and so did not confront, the issue of neurological change in the brain (Volkow et al., Reference Volkow, Fowler and Wang2003). But, it has also been criticized on several other theoretical and empirical grounds, particularly the infinite lifespan assumption (Gruber & Koszegi, Reference Gruber and Koszegi2001; Rogeberg, Reference Rogeberg2004; Clarke & Danilkina, Reference Clarke and Danilkina2006; Rogeberg, Reference Rogeberg2020). However, rational addiction theory appears to have had limited policy analysis impact, especially within health policy evaluation. This is not to claim that a parasitic or exploiter/victim frame is necessarily the most appropriate one for all addictive substances or activities, for example, for gambling or even nicotine (but see Zendle & Cairns, Reference Zendle and Cairns2019; Zendle, Reference Zendle2021). I argue that economic analysis of addiction, including in CBA, should be based on the specific neurological, behavioral and intergenerational (epigenetic and socioeconomic) evidence on impacts related to each particular substance or behavior.
How would this framing of addiction affect revealed subjective valuations? The most consequential adjustment would be to treat the notional consumer surplus of opioid and stimulant users as zero in CBA, except for the first, or first few, ingestions of these drugs. This rejects an argument that doing so would turn “welfare economics on its head” (Lemieux, Reference Lemieux2017). But I believe that on the basis of current evidence, it is a “bridge too far” to treat the expected health benefits potentially flowing from “now avoided or reduced” addiction as a benefit category of some policy alternatives. However, epigenetic treatments or genetic editing policies may justify an inclusion of those benefits in future analyses (Levy et al., Reference Levy, Norton and Smith2018).
13. Conclusion: CBA, play your assessment of (social) efficiency position
The principles of CBA both describe and circumscribe what it can, and cannot, provide to, and for, policymakers. But there is disagreement on the usefulness of some of these principles. Current CBA practice concerns potential Pareto efficiency (admittedly sometimes stretched), rather than some broader notion of social welfare. Some regard this scoping as a weakness of CBA, but I rather regard it as strength: as in team sports, playing your analytic position is almost always the best strategy. Analysis of efficiency and distributional impacts of policies are both extremely valuable, but more so in parallel rather than combined is the most informative (e.g. Dennin & Muller, Reference Dennin and Muller2025).
At its most basic level, CBA is about analysis and advice, rather than decision-making. To the extent that CBA does influence legislative or regulatory decision-making, its influence is best described as a “drip, drip, drip,” of liberal neutrality rather than claiming any degree of dominance (e.g. Black, Reference Black2013; Eliasson et al., Reference Eliasson, Borjesson, Odeck and Welde2015; Friedman, Reference Friedman2017; Volden, Reference Volden2019). Given the underrepresentation of social efficiency in legislative and regulatory decision-making, the drip is good, and hopefully useful, in spite of remaining barriers to more use of CBA (Vining & Boardman, Reference Vining and Boardman2026).
Acknowledgments
I thank two anonymous reviewers for very constructive comments and criticisms. I thank David Weimer and Don Kenkel for many helpful comments on earlier drafts. The usual caveats apply. I dedicate this paper to the memories of David H. Greenberg and Jack Knetsch. Both were inspirational scholars and leaders in this field.
Open access
