2. Essential and Discretionary Public Goods

The distinction between essential and discretionary public goods structures the philosophical debate about distributive justice in coercive contributions to public goods. This debate takes the economist’s framework as its point of departure, in which public goods are identified as central sources of market failure: their non-excludable and non-rival nature invites strategic free riders to withhold from contributing, thereby undermining the market mechanism. But while philosophers agree with welfare economists that non-excludability and non-rivalry imply that some public goods can potentially be more efficiently secured through state coercion than through voluntary market exchange,Footnote ⁸ they claim that superior efficiency does not generally suffice to vindicate such coercion. This point has arguably been made most forcefully by Nozick (Reference Nozick1974: 90–95), who argues that the fact that one benefits from a public good such as a broadcasting service does not render it legitimate for others to demand a contribution unless one has voluntarily consented to it.

The problem is that state coercion does not only solve the problem of strategic free-riders, who are insincere in their preference revelation for public goods. It also coerces citizens whom Schmidtz labels honest holdouts (Reference Schmidtz1991: 84; see also Mack Reference Mack and Lee1986: 513–514). In contrast to strategic free-riders, honest holdouts sincerely claim not to derive benefits from the public good, or even dismiss a public good as a public bad.Footnote ⁹ Rawls (Reference Rawls1971: 283) captures the unease of resorting to state coercion for public goods when he claims that ‘there is no more justification for using the state apparatus to compel some citizens to pay for unwanted benefits that others desire than there is to force them to reimburse others for their private expenses’. While it may in practice be difficult to ascertain who is an honest holdout and who is a strategic free-rider, the mere existence of honest holdouts reveals important shortcomings of the efficiency rationale for public goods. It prompts us to consider for which public goods citizens have legitimate standing to withhold their consent to mandatory contributions. The economic framework, with its emphasis on superior efficiency of state coercion compared with voluntary market exchange, is too indiscriminate in this regard. As Cordelli argues, the framework overreaches the scope for legitimate state coercion, failing to discriminate, for instance, between national defence and fireworks (which both exhibit non-rivalry and non-excludability): ‘a government lacks the legitimate authority to force people into forms of cooperation, including producing discretionary goods like fireworks, just because those forms will benefit other people, by satisfying preferences that would otherwise remain unmet’ (Cordelli Reference Cordelli2021: 27). The problem of the honest holdout underscores that public goods can generate unwarranted asymmetries in benefits and burdens (resulting in distributive unfairness). This motivates a distinction among two types of public goods: goods for which coercive contributions are justified because they are a necessary requirement of justice, and goods that are more optional and against which the honest holdout can thus legitimately object.

The distinction between these two types indeed underpins the debate about distributive justice and public goods, where each type has been given a specific label:Footnote ¹⁰ essential public goods are defined as ‘those required by justice’, whereas discretionary public goods are those whose supply is optional, since they are ‘not required by justice’ (Miller and Taylor Reference Miller, Taylor and Olsaretti2018: 562).Footnote ¹¹ For a good to be required by justice, according to Miller and Taylor, means that it is ‘required for social cooperation to take place at all’ (Reference Miller, Taylor and Olsaretti2018: 562) (their example is law and order), that it is ‘needed to ensure that the value of fair shares is not undermined’ (Reference Miller, Taylor and Olsaretti2018: 562) (their example are zoning restrictions, which specify for which purposes land can be used and thus which externalities others may impose) or that it is needed to ‘meet (at least some of) the basic needs of fellow citizens’ (their example is access to drinking water) (Reference Miller, Taylor and Olsaretti2018: 564).

According to the accepted view (see e.g. Klosko Reference Klosko1990; Murphy and Nagel Reference Murphy and Nagel2001; Miller Reference Miller, Dowding, Goodin and Pateman2004; Claassen Reference Claassen2013; Miller and Taylor Reference Miller, Taylor and Olsaretti2018), essential public goods do not give rise to distinct questions of distributive justice: they are required for justice, but for this reason, how to divide the costs of their supply is immediately determined by whatever theory of justice one commits to.Footnote ¹² Discretionary public goods, by contrast, are seen to give rise to an independent question of justice: because the benefits and costs obtained through coercive contributions can be distributed in a variety of ways, and ‘efficiency alone will not dictate a choice among them’ (Murphy and Nagel Reference Murphy and Nagel2001: 61), we must ask what justice requires. In response to this question, authors have proposed several principles. On the one hand, some authors have proposed purely procedural principles, such as the unanimity requirement (Rawls Reference Rawls1971; see also Wicksell Reference Wicksell, Musgrave and Peacock1958; and Buchanan Reference Buchanan1969) or a decision by majority vote (Rawls Reference Rawls2001). Others have insisted that we confront a substantive question of justice (see e.g. Miller Reference Miller, Dowding, Goodin and Pateman2004: 133–134) and have proposed substantive principles that prescribe a particular distribution of net benefits, like a proportional benefit principle that requires net benefits to be proportional to contributions (Murphy and Nagel Reference Murphy and Nagel2001: 62) or an equal benefit principle that requires net benefits to be equalized (Miller Reference Miller, Dowding, Goodin and Pateman2004: 142–143). Yet others have advocated for a combination of procedural and substantive criteria to ensure justice in securing discretionary public goods: Claassen (Reference Claassen2013: 281–285) proposes to rely on a simple majority vote while at the same time restricting the choice to options that fall in-between ensuring equal benefits and equal overall welfare.Footnote ¹³

As this brief overview reveals, the presumption that we can draw a clear-cut distinction between essential and discretionary public goods structures the philosophical debate about public goods. It narrows down the challenge of the honest holdout, who is taken to enjoy adequate standing to object only to coercive contributions to public goods that are not independently justified as necessarily required by justice. And it motivates the two-stage approach in assessing the distributive justice of the supply of public goods: at a first stage, we identify the set of essential public goods and ensure government secures them in accordance with our preferred theory of justice; at a second stage, and assuming an otherwise just society, we then apply a supplementary principle that governs the distributive outcomes of the supply of discretionary public goods.Footnote ¹⁴

3. The Dependence on Background Conditions

This section puts forward our first challenge, the argument from background conditions. This argument shows that we can at best hope to classify a good as essential or discretionary relative to some background conditions. To put it differently, contrary to what the literature suggests, we cannot compile a list of public goods and then simply divide them into two columns, essential and discretionary. Instead, whether a good qualifies as essential will typically depend on (i) the social background conditions, and in particular the state of social development, and on (ii) the current existence of injustices, including non-material ones. While this finding is relevant, it leaves open the possibility of distinguishing between essential and discretionary public goods in a given context and is thus one that proponents of the distinction could in principle accept.

According to the accepted definition, public goods qualify as essential if their provision is required by justice. As Miller and Taylor (Reference Miller, Taylor and Olsaretti2018: 564) note, ‘many theories of justice … hold that there should be a social minimum that no individuals should fall below’ where this ‘minimum is usually cashed out in terms of basic needs that must be met for an individual to live a minimally decent life (in their society)’. It is for this reason that they hold that public goods may be required by justice, and may thus qualify as essential, even if they are not needed to secure the basic conditions for social cooperation, nor for the protection of private resource claims. Yet, it is hard to deny that what is required for a person to live a minimally decent life depends, at least partly, on what is required for participation in society and therefore on the state of social development. Consider, for example, access to the internet: while access to the internet can hardly be said to have been a basic need in the early 1990s, when such access was rather exceptional even in wealthy countries, it arguably qualifies as a basic need in the early 2020s. The reason for this is that access to the internet has become a prerequisite for fully participating in most societies today.Footnote ¹⁵ If this is true, then it shows that whether universal access to the internet qualifies as an essential public good depends on the state of social development.Footnote ¹⁶ But even if one disputes the claim, in this specific instance, the case nonetheless points to the possibility that whether a specific public good qualifies as essential depends on the state of social development. This shows that we cannot simply compile a list of public goods to then divide them into essential and discretionary ones. Instead, whether a public good qualifies as essential needs to be evaluated by reference to a given society.Footnote ¹⁷

Next, consider the relevance of (ii) injustices that exist prior to the supply of public goods. Discretionary public goods are typically examined under the assumption that an initial state of justice has already been achieved: all public goods required for justice have been secured and the distribution of privately owned resources is just (see e.g. Murphy and Nagel Reference Murphy and Nagel2001: 57; Miller Reference Miller, Dowding, Goodin and Pateman2004: 113, note 4).Footnote ¹⁸ Yet, where these conditions fail to hold, the existence of injustices may affect which public goods are required by justice. In particular, or so we think, there can be public goods whose supply becomes a requirement of justice only because there are existing injustices. Suppose that access to critical healthcare is considered a basic need. Suppose, moreover, that we have a society where everyone initially has the means to secure adequate access to critical healthcare in a market, be it directly or through insurance. In this case, there is no problem: the basic need is met. Now, suppose that as material inequalities grow, some individuals no longer manage to secure access to critical healthcare through the market. In this case, it would seem, securing healthcare as a public good becomes a requirement of justice.Footnote ¹⁹ But it is important to be careful here: after all, it is conceivable that there are other ways of ensuring universal access without resorting to provision by government, such as via redistribution of private resources. So, one may ask: does a public good qualify as essential if there are alternative means of meeting a certain basic need?

This concern echoes an issue noted by Miller and Taylor, who ask whether a good should qualify as essential ‘when it provides the only feasible way of meeting individuals’ basic needs, or when it is the most efficient way’ (Reference Miller, Taylor and Olsaretti2018: 573)? While this is an important question, we need not take a definitive stand on it here. For making our case, it suffices to show that sometimes the provision of a public good (which would otherwise have been optional) may be the only feasible or most efficient way of addressing an injustice, in which case whether this public good qualifies as essential depends on whether an injustice exists.Footnote ²⁰

Consider, for example, a society in which some groups are excluded from access to certain market club goods based on their gender or race, as has historically been true of country clubs (Lenkiewicz Reference Lenkiewicz2011).Footnote ²¹ At first sight, redistribution of private resources, alongside the design and enforcement of anti-discrimination laws to ensure equal access, seem more straightforward ways of addressing the injustice.Footnote ²² However, these measures may not offer a feasible, let alone the most efficient, way of addressing the injustice. Anti-discrimination laws, for one, need to be balanced against rights to freedom of association (in exclusive clubs), and the costs of enforcing such laws may be high. And redistribution of private resources to create additional market club goods for those subject to exclusion need not be the most efficient solution to address the injustice. After all, the imperfect rivalry of these goods remains a potential source of market inefficiency. Therefore, if unequal access persists as an injustice despite redistributive and regulatory measures, the additional provision of goods such as country club facilities, public beaches, and parks in non-exclusionary form may sometimes be an essential requirement of justice for which coercive contributions are warranted on grounds of feasibility and efficiency.Footnote ²³

To summarize: the argument from background conditions shows that the armchair is not a position from which we can determine, for every public good, whether it qualifies as essential. This is relevant because it implies that we can at best hope to compile a list of essential goods relative to some specific society. But the argument explicitly leaves open the possibility of drawing this distinction for a given society, and at a given moment in time – taking account of the needs that qualify as basic and of any injustices that exist in this society.

4. From Individual Public Goods to Configurations of Multiple Public Goods

This section puts forward our second challenge: the argument from competing configurations of public goods. So far, our arguments have shown that we need to be cautious in drawing a distinction between essential and discretionary public goods because we cannot categorize a public good without taking account of the social context. Our second argument is more fundamental: contrary to what is commonly assumed, it claims, the distinction between essential and discretionary public goods cannot meaningfully be applied to public goods that are considered in isolation.

To motivate this idea, note first that whether a given public goods qualifies as essential typically depends on which other public goods are provided. This is evident from Miller and Taylor’s (Reference Miller, Taylor and Olsaretti2018) discussion of access to fresh drinking water as a basic need. Miller and Taylor (Reference Miller, Taylor and Olsaretti2018: 565) discuss whether the creation of a national company that delivers drinking water would be required by justice only if it is the only feasible way of meeting the basic need, or also if it is merely the most efficient way of doing so.Footnote ²⁴ In their scenario, the national water company is a public good or, more accurately, provides the public good of universal access to drinking water. What they overlook, however, is that the supply of other public goods can affect whether the basic need of access to drinking water can be satisfied, at all or more efficiently, in the absence of the creation of the public water company.Footnote ²⁵ The alternative mode of provision they consider is the digging of private wells. But other public goods might affect whether this is a feasible and an efficient mode of provision. It might, for instance, make a difference if the state subsidizes free wells digging courses or a costless water quality check service, which similarly qualify as public goods.Footnote ²⁶ This shows that, even if we concede that an unregulated market will on its own fail to satisfy a basic need, we cannot infer that the public water company qualifies as an essential public good. On the one hand, if the supply of other public goods could, perhaps in combination with a market, also satisfy everyone’s basic need of access to water, then the creation of the public water company would turn out to not be the only feasible way of achieving justice. On the other hand, if the availability of other public goods, such as courses or water quality checks, affects the efficiency of other solutions, then whether the creation of the public water company is the most efficient way of meeting the basic need may also depend on which other public goods are provided. We cannot, then, determine whether the public water company is an essential public good, unless we know which other public goods are available – goods for which it may in turn likewise be impossible to ascertain, when considered in isolation, whether they are essential.

Admittedly, the case of access to drinking water may appear to be rather specific and the market solution of digging private wells somewhat artificial. But the point generalizes: for the satisfaction of most basic needs, we can choose not just among a market and a public goods solution, but among several public goods solutions, including combinations of various modes of provision. For another example, consider access to nature for the purpose of physical and mental well-being, which arguably qualifies, at least in some limited form, as a basic need. When several countries prohibited people from going outside at the outset of the recent pandemic, essentially blocking parks, meadows and beaches, the effect on people’s well-being was quite severe.Footnote ²⁷ Now, freely accessible nature parks, beaches, forests and city gardens are all public goods that can arguably satisfy this basic need for access to green space in nature. But if various public goods could satisfy this basic need, then none qualifies as essential or as discretionary per se. Instead, whether a particular good qualifies as essential may depend on which other public goods are already secured or which other, discretionary public goods will be available. If, for instance, a lot of public beaches and nature parks already exist, then public city gardens qualify as discretionary public goods because the basic need is already met; if, by contrast, no public beaches and nature parks exist, they may qualify as essential. But which public good is most efficient at satisfying our basic needs may in turn depend on the provision of what under any plausible description qualify as optional goods: if, for instance, the state arranges good signposting and free maps in a nature park, or takes care of picnic spots and rubbish collection, or instals a shuttle service from the city, then this may render such parks more efficient at satisfying the basic need of access to green space than they would be without provision of these services. It appears clear that none of these additional public goods would be required by justice, but whether they would be provided might nonetheless affect which other public goods (beaches, nature parks, city gardens etc.) qualify as essential. This shows that we cannot in general categorize specific public goods as essential or discretionary in isolation but must take account of the entire set of public goods that the government secures. The finding that, for any individual public good, whether it qualifies as essential depends on which other public goods are secured, is significant. It shows that, even if we take account of social background conditions, we are not able to meaningfully assess, for any specific public good, considered in isolation, whether it qualifies as essential. Because the supply of many different configurations of public goods could in principle achieve the satisfaction of our basic needs, no specific public good would come out as essential on its own: what is required is not that any specific public good is made available, but that some set of public goods, or a bundle, is available that is able to jointly satisfy people’s basic needs. This does not imply that the distinction between essential and discretionary public goods is ill-conceived; but it highlights that this distinction should be applied from a holistic instead of an isolated perspective. Taking account of all the goods secured, we can identify those public goods that – given current configurations – are indeed necessary to satisfy basic needs, and thus qualify as serving essential requirements of justice. We cannot, however, make this determination when we look at certain public goods in isolation. This implies that any characterization of a public good as, in part, essential will typically be contingent on the presence or absence of other public goods.Footnote ²⁸

5. The Problem of Specification

This section puts forward our third challenge, the argument from specification: even if we consider public goods in a particular society at a specific moment in time conditional on the other public goods that are secured, we confront a difficulty of specifying the form and extent of their required supply. The abstract classification of public goods as essential or discretionary masks the need to determine how, more concretely, a good is actually supplied. But once we look at the concrete shape that provision can take, the distinction between essential and discretionary goods quickly becomes fuzzy.

To see how this specification problem destabilizes the conceptual distinction, consider the allegedly essential public good of national defence. National defence serves as a textbook example of an essential public good (Sekera Reference Sekera2019; Reiss Reference Reiss and Zalta2021). It meets the economist’s criteria of non-excludability and non-rivalry in consumption, and its provision by government is not only warranted on grounds of efficiency: it instead seems a hard requirement of justice, since foreign invaders typically breach many individual rights and liberties, and countries afflicted by war are rarely places where one can lead a minimally decent life. In abstract terms, it therefore seems relatively uncontroversial to categorize national defence as essential, and to argue that justice is enhanced when its supply is regulated by core principles of justice. Informed by these principles, we could for instance fund national defence budgets through progressive taxation and dismiss the claim of holdouts who oppose coercive contributions as such.

Yet as soon as we try to specify the form and extent of supplying this public good, problems emerge. Consider first the form or shape that national defence can take. Is this good primarily secured by maintaining high-quality armed forces and joining, say, a defensive alliance such as NATO? There is ample scope for reasonable disagreement here. One might argue that national security should not (or not only) be pursued by military but rather by diplomatic means, or that a policy of demilitarization or armed neutrality is preferable. It is thus far from self-evident that the maintenance of armed forces is a necessary public good while, say, a diplomatic service that spends a large chunk of its resources on promoting intercultural ties and dialogue is merely optional for a just society.Footnote ²⁹ This problem is brought into sharper relief if we also consider the extent of securing defence. One might argue that governments fulfil their obligation to secure the essential public good of national defence when they subscribe, say, to NATO’s norm of spending at least 2% of GDP on one’s armed forces. But just as arbitrary as this threshold itself appears, it similarly seems arbitrary to imply that any expenditure exceeding this norm is merely optional under otherwise peaceful international conditions. And it also seems difficult to regulate essential military expenditures up to NATO’s 2% norm by core principles of justice, while letting an auxiliary distributive principle govern any additional expenditures. Intuitively, it might seem fair to send fellow citizens who derive disproportionate benefit from non-essential military expenditures an additional tax bill – for instance, because they visit annual military parades or rejoice at the sight of officers wearing the latest military gadgets. But those citizens might object that a discretionary military parade also serves essential training purposes, and that high-tech gadgets are not merely nice to have but critical to effectively defending a country.

A further way of articulating the specification problem is to point out that the availability of public goods may be subject to various levels of risk. Choices about the form and extent of supplying a public good will impact the likelihood that the public can adequately enjoy it. Arguably, public goods such as healthcare and public health turned out to be subject to high levels of risk during the COVID-19 pandemic, as shortages of ICU beds and personal protective equipment inhibited both the effective supply and enjoyment of these goods. Countries with stronger reserve capacities were more resilient in securing them. Similar questions arise as to how much and what type of ammunition armed forces should hold in stock. Insofar as a public good is deemed essential for a just and orderly society, the intuitive answer appears to be that any significant level of risk is problematic. Yet, it also seems impossible, or prohibitively costly, to reduce the level of risk that a public good will turn out to be unavailable to zero. So, we confront the question of how much risk is acceptable when a public good serves a basic need: e.g. how many weeks or months of reserve supplies of ammunition or medicine or emergency food rations are required, and when do further reserves become discretionary?Footnote ³⁰ Here again we find that the essential-discretionary distinction becomes blurry once we try to specify how government should ensure the good in question, since various levels of risk at which a good is made available convey how essential it really is for a just society.

Thus, even if there are plausible ways of designating which public goods count as essential or discretionary in abstract terms, the distinction becomes much less clear-cut once we consider how government should secure them in practice.Footnote ³¹ Indeed, it may not only turn out that upon further specification an essential public good appears discretionary (or vice versa), so that its supply should not enhance but rather preserve a just pattern of distribution (or vice versa). It is also to be expected that, for numerous citizens, certain specifications of the form and extent of their availability would pervert the public good into a public bad, prompting them to take the stand of an honest holdout. For instance, a reasonable citizen who adheres to a pacificist or ecologist conception of the good will object that the maintenance of armed forces with nuclear capabilities is neither justice-enhancing (essential), nor justice-preserving (discretionary), but instead justice-impeding. She might consider such maintenance a public bad that not only achieves the opposite of national or global interests in security but also unduly forces her to fund this bad through coercive taxation. And the specification problem manifests itself not only in reasonable expressions of moral and political disagreement over the form and extent supply can take. It also has a clear epistemic and practical dimension: if we are not in the epistemic position to draw a clear distinction between essential and discretionary uses, then we will have a hard time applying this distinction in practice. In the case of military expenditures, as for public expenditures more generally, it oftentimes appears impossible to forecast in advance, or determine in retrospect, if they are spent on essential or discretionary public goods and how their social surplus is distributed over the population. This opens up the possibility of legitimate complaints by honest holdouts and we cannot be sure whether core or auxiliary distributive principles should be at work in arranging coercive contributions.

The specification problem shows that even if we deem a particular public good essential at a high level of abstraction, this may be of little guidance in assessing whether concrete supplies – their form and extent – are really required by justice. The argument thus puts stress on the categorical distinction presupposed in the debate surrounding the honest holdout and the related question of distributive justice in the supply of public goods. Note, moreover, that this concern cannot be alleviated by drawing a distinction between a services-conception of public goods – which defines them by reference to their specific mode of supply, such as a lighthouse – and a benefits-conception of public goods – which defines them abstractly by reference to a benefit they bring about, such as orientation. While the philosophical literature tends to conflate these two conceptions, they could in principle be kept apart, and one might argue that only the benefit-conception is adequate: for the good that is non-rivalrous and non-excludable really is the benefit, like orientation, not the entity that ensures it, like a lighthouse, which could be operated in a variety of ways (see Coase Reference Coase1974). But on the benefits-conception, one might then note, we can keep essential and discretionary public goods apart: some benefits are fundamental enough to be required by justice whereas others are not (in given circumstances).Footnote ³²

This line of reasoning has some appeal, and indeed raises an important question about whether philosophers should restrict themselves to a benefits-conception of public goods only; regardless of the view one takes on this question, however, it does not offer a solution to the issue at hand. This is for two reasons. First, note that the benefits-conception allows us to identify particular goods as essential only at the cost of rendering them highly abstract. At a practical level, though, once we decide to secure orientation, we must choose a concrete means – be it lighthouses, a GPS-system or fog horns. And if any of these goods also serve certain discretionary needs, then we cannot infer that the distribution of costs should be treated according to the rules for essential public goods. So, even if restricting the notion of public goods to the benefits-conception would yield a number of essential public goods, this would only kick the problem of specification down the road: this problem would not go away, and neither would the question of how to justly allocate costs and benefits incurred in securing essential public goods. Second, even if one were to reject this argument, another problem would remain: on the benefits-conception of public goods, we still confront the question to what extent, and subject to what level of risk, specific benefits are to be secured.

We thus take our third objection to show that we indeed cannot clearly distinguish between essential and discretionary public goods – even relative to a particular society and conditional on which other public goods are secured. This does not imply that the distinction is meaningless: we may still seek to distinguish aspects of supply that meet a basic need and thus qualify as essential from aspects that do not – and we may seek to put to work the different distributive principles for dealing with both aspects. What it shows, however, is that rather than conceiving of public goods as exclusively essential or discretionary, it will often be more accurate to speak of them as partly essential and partly discretionary – or, even more accurately, as partly serving essential and partly serving discretionary needs – and to thereby acknowledge that the distinct principles apply, within practical limits, to both essential and discretionary uses of the same goods.

6. Implications: Reconfiguring Essential and Discretionary Public Goods

The arguments put forward in this paper show that the distinction between essential and discretionary public goods is not as straightforward as is generally assumed in the philosophical debate. We cannot simply proceed by compiling two lists of goods, one of essential public goods, and the other of discretionary public goods. This is for several reasons. First, social background conditions generally affect whether the supply of a given public good is required to achieve initial conditions of justice, so these must be considered. Second, basic needs can typically be satisfied by different sets of public goods; so, we cannot meaningfully proclaim a public good to be essential while considering it in isolation. Third, the actual delivery of public goods can take a variety of forms, and as we move from the level of abstract designation to that of concrete means, the line between essential and discretionary goods becomes increasingly blurry: because most public goods end up serving more fundamental as well as more elective needs, they qualify partly as essential and partly as discretionary.

These arguments are of immediate relevance to the debate surrounding public goods. Concerning the honest holdout problem, they show that, at first sight, this problem might be more contained than is often assumed. More specifically, they show that the set of public goods for which the honest holdout has the required standing to object is smaller than is often assumed. This is not only because the supply of many public goods may be justified to address existing injustices; it is also because many public goods serve fundamental needs alongside discretionary ones. This entanglement of essential aspects of public goods with discretionary purposes means that even the honest holdout is confronted with a non-refutable reason in favour of coercive contributions to public goods. For instance, holdouts will have to accept that some coercive contributions to the maintenance of city parks, from which they may honestly derive zero or even negative utility, are nevertheless justifiable, since those goods are needed by others to meet basic needs of mental and physical health and not for mere discretionary recreational purposes.

At the same time, our findings also show that those who employ the essential-discretionary distinction cannot take the justifiability of coercing holdouts for granted while directly proceeding to the analysis of distributive fairness in the supply of discretionary public goods. Note that even if one adopts a preference-independent theory of justice, these problems – of justification and distribution – remain linked. The grounds upon which holdouts object to coercive contributions need not be dishonest, irrational or ill-informed. Holdouts who are honest, rational and well-informed do exist.Footnote ³³ They can have legitimate complaints against coercive contributions, not least because these contributions can translate into unjust distributive outcomes.

Consider again the case of citizens who on pacifist grounds object to coercive contributions to the public good of security by military means. Pacifism may be an integral part of these citizens’ conception of the good. If one adheres to a preference-independent theory of justice, one cannot therefore safely ignore their pacifism as a mere preference, let alone dismiss it as an ill-informed, irrational or dishonest one. But while pacificist beliefs may be central to their conception of the good, these beliefs also translate into preference orderings and utility levels. When pacifists are coerced to contribute to the purportedly essential public good of an army, not to mention allegedly discretionary expenditures such as ceremonial military parades, they can legitimately object on the ground that they are subject to distributive unfairness. To be sure, even if a pacifist were to receive a tax deduction for unwanted military expenditures or an exemption from conscription, so that distributive unfairness would be avoided or at least reduced, they may still object to government supplying security by military means on their behalf as well. But when exemptions and deductions like these are neither desirable nor feasible (for instance, because of the difficulty of distinguishing honest holdouts from insincere free-riders), pacifist holdouts do become subject to distributive unfairness.Footnote ³⁴ In fact, the ensuing distributive unfairness may itself be a central reason for justifiable complaints against coercive contributions.Footnote ³⁵ So, the concerns of distributive fairness and of justifying coercive contributions are ultimately linked.

Yet, while our discussion explains why, if one prioritizes justice over efficiency (as most theories of justice do), the honest holdout problem is less severe, it also shows that public goods pose a more complex challenge to theories of distributive justice than previously realized. The existing debate treats discretionary public goods as a blind spot of theories of justice: while essential public goods are required by justice, and thus covered by existing principles, discretionary public goods are not so covered, and prompt the need for a complementary principle that governs the distribution of any surplus. But if we in fact cannot draw a clear divide between essential and discretionary public goods, then this two-step approach becomes infeasible. That various sets of public goods can satisfy fundamental needs shows that we often have a choice concerning which of different sets of public goods satisfies them most efficiently; and that even where they serve fundamental needs, public goods typically also satisfy additional, discretionary needs. The two sets of principles – those governing essential and those governing discretionary public goods – are not easily kept apart.

Bicycle lanes illustrate this point. Moral disagreement may arise as to whether bicycle lanes count as essential. Cyclists may use them to bike to their job and, absent access to public transport or more expensive means of transportation, they may require bicycle lanes to enjoy fair equality of opportunity.Footnote ³⁶ Yet bicycle lanes are also intensely used for purposes of recreation and leisure, and hence may be qualified as discretionary public goods, since a just society is not required to satisfy the recreational preferences of its members through coercive contributions. Now suppose we come to agree, morally, that they are essential public goods for workers yet discretionary ones for recreationists. In that case we still face a substantial epistemic burden of tracing essential and discretionary uses by bikers, but we can seek to proportionately assign costs for the construction and maintenance of bicycle lanes according to both core and auxiliary principles of distributive justice.Footnote ³⁷

The distinction between essential and discretionary public goods is not, then, as innocuous as is assumed within the philosophical debate that relies on it, but this does not show it to be ill-conceived, let alone meaningless. On the one hand, the distinction can still be applied at the margins: fireworks, for example, plausibly qualify as a discretionary public good as they serve no fundamental interests (bar in extraordinary circumstances). On the other hand, the attempt to distinguish public goods that are required as a matter of justice from those that are not has a compelling motivation: it appears plausible that different distributive principles are applicable in each instance. But even where we cannot draw a clear line between essential and discretionary public goods, distributive fairness still requires that we at least try to apply core and auxiliary principles of distribution in a manner that tracks their dual purpose. We might, for instance, insist that to the extent to which a public good serves only discretionary needs, the costs of access should be proportionately borne by the beneficiaries. And this happens: in the context of school swimming lessons, children typically access public swimming pools for free (enabling the teaching of vital skills, they function as an essential public good), but in the context of birthday parties, access to a public swimming pool will come at a cost (here, the pool serves as a discretionary public good, facilitating recreation). In this respect, our analysis gives reason to endorse Taylor’s (Reference Taylor2021: 572–573) recent call to seek approximations of distributive fairness in the costs and benefits of public goods. However, since our findings show that the essential-discretionary divide is often intractable, there is also reason to be more sceptical about the likelihood that government officials can actually succeed in this task.Footnote ³⁸ Still, practical and epistemic challenges provide no reason to not at least try to approximate distributive fairness, regulated by context-sensitive applications of both core and supplementary principles.

Our analysis shows that no neat and final resolution of claims to justice in accordance with either core or supplementary principles can be expected at the level of ideal theory. This is not only because basic needs are subject to change, and depend on which other goods are available, but because public goods can be provided in a variety of forms, with greater or smaller resilience, and there is consequently space for reasonable disagreement over which of the benefits they generate qualify as essential, and which as discretionary. Yet, the distinction nevertheless aids in diagnosing social injustice and distributive unfairness in the current configuration of public goods that rest on coercive contributions – from policing hooligans in the billion-dollar industry of football to the distance between poor neighbourhoods and high-quality city parks. More generally, the distinction retains important value as a heuristic, and perhaps even as a regulative ideal, in addressing stakes of social justice in actual policymaking under non-ideal circumstances: politics involves, and should involve, the ongoing reconfiguration of which public goods count as essential and which ones as discretionary.