MUST EGALITARIANS RELY ON THE STATE TO ATTAIN DISTRIBUTIVE JUSTICE?

Abstract

It is widely accepted among political philosophers that distributive justice should be promoted by the state. This essay challenges this presumption by making two key claims. First, the state is not the only possible mechanism for attaining distributive justice. We could rely alternatively on the voluntary efforts and interactions of individuals and associations in civil society. The question of what mechanism we should rely on is a comparative and empirical one. What matters is which mechanism better promotes distributive justice. We cannot settle the question a priori in favor of the state. Second, several considerations suggest a presumption in favor of relying on civil society.

I. Introduction

Egalitarianism, as understood in this essay, entails a commitment to outcome-based principles of distributive justice. Such principles provide us with criteria to evaluate the distribution of advantages and burdens arising from social cooperation. A distributive outcome is just to the extent that it is consistent with the criteria specified by the theory. Examples of outcome-based principles include John Rawls’s two principles of justice, luck egalitarianism, and sufficientarianism.¹ In this essay I am concerned with how distributive justice, so understood, should be realized.²

The view taken for granted by the great majority of political theorists holds that distributive justice should be promoted by the state and we can establish this proposition from the philosophical armchair, without empirical inquiry. I call this the Default View.³ The state promotes distributive justice by issuing legal commands backed ultimately by the threat of coercive force. Such commands may be expressed in the form of taxation policy, labor laws, immigration controls, welfare policy, and so on.

However, in principle, distributive justice could instead be realized through the voluntary efforts and interactions of individuals and associations in civil society—or so I will argue. Charity, mutual aid, and insurance are among the mechanisms by which distributively just outcomes might be realized by private individuals and voluntary associations in civil society. I will contend, pace the Default View, that we cannot establish from the armchair whether we should rely on the state or civil society. This is an empirical question best decided on a comparative basis. What matters is which mechanism better promotes justice. ⁴ I understand “promoting” justice to entail bringing distributive outcomes in closer alignment with the conception of justice in question. I defend this claim by rebutting in the first half of this essay a series of arguments for the Default View.

I do not try to settle definitively whether the state or civil society is more effective at promoting justice, in part because I think the answer may vary depending on the society in question. However, in the second half of this essay I argue that civil society mechanisms have certain normative and practical advantages over the state. These suggest a presumption in favor of nonstate means of pursuing distributive justice. I do not here commit myself to any particular outcome-based principle. My argument is intended to apply to such principles generally. ⁵ Before beginning the substantive argument, I will make some initial remarks about how the comparison between state and civil society should be made as well as how my argument differs from traditional libertarian arguments for limited government.

A. Getting the comparison right

We should bear in mind the following considerations when comparing state and civil society actors.

(1) The focus should be on outcomes, not intentions. The intention to meet an objective and meeting it are quite different things. An actor or set of actors can fail to meet an outcome despite intending to (and vice versa). What should matter is whether the outcome is achieved. ⁶

(2) Prior to empirical inquiry, agnosticism is called for about individual motivation in different domains. We should not, for example, assume at the outset that individuals are motivated by justice in the voting booth but driven only by self-interest when acting in civil society. ⁷ Whether individuals tend to have different motivations when engaged in political activity as opposed to civil society activity or whether they tend to have consistent motivations across these domains are empirical matters.

(3) The distinction between state and nonstate action is orthogonal to the distinction between individual and collective action. There is a tendency to assume that social cooperation mediated by the state is an inherently collective enterprise while nonstate forms of social cooperation are atomized and individualistic. However, there are no necessary connections here. On the one hand, trade unions, religious associations, mutual aid associations, friendly societies, community organizations, self-help groups, charities, cooperatives, activist groups, campaign groups, and the extended family all constitute nonstate forms of collective action. On the other hand, a person whose commitment to distributive justice is limited to paying their taxes and who otherwise pursues only their private goals is engaged in collective action in only a truncated sense.

B. Libertarianism by the back door?

Libertarianism, as understood here, is not an outcome-based principle of distributive justice but a procedural one. On a procedural view, what matters is not the distributive outcome itself but the underlying social process through which that distribution arises. Libertarians hold that a distributive outcome is just so long as it arose through universal conformity to appropriate rules specifying how property is to be acquired and transferred. ⁸ The task of the state primarily is to coercively enforce these procedural rules. ⁹

To say that we might rely on the voluntary efforts and interactions of civil society actors to attain egalitarian distributive outcomes is to say that we might rely on a mechanism that libertarians consider intrinsically just—that is, a state constrained to protecting the property rights of civil society actors—to achieve egalitarian objectives. If I am suggesting that egalitarian advocates of outcome-based principles of justice should be open to relying on civil society as a means of attaining their goals simply because that would also be consistent with libertarianism, then I appear to be begging the question by implicitly presuming the truth of controversial libertarian principles. Let me stress, then, that I am not making that assumption in this essay. That is to say, I am not taking the possibility that reliance on civil society might be just from a libertarian perspective as a reason in its favor. ¹⁰ I contend only that egalitarian advocates of outcome-based principles should be open to the possibility that reliance on civil society could feasibly achieve their desired outcomes.

II. The Conceptual Argument for the Default View

What I call the conceptual argument for the Default View holds that distributive justice requires state enforcement because it is a particular kind of moral concept. This argument runs as follows:

1. (1) Distributive justice entails morally mandatory transfers.

2. (2) Morally mandatory transfers must, by definition, take place.

3. (3) State coercion ensures that morally mandatory transfers will take place, while relying on civil society entails leaving compliance to the vagaries of individual discretion.

4. (4) Therefore, morally mandatory transfers should be enforced by state coercion.

The key premise is (3). Why might coercion ensure that morally mandatory transfers will take place? Aaron James argues that the ability to coerce gives the state the power “not merely to influence but to determine, at will, what actions many others perform.” ¹¹ I understand in the following way the distinction he invokes between “determining” and mere “influencing.” Assume that Jim is morally mandated to transfer some resources to the poor as a matter of distributive justice and he is not inclined to do so of his own accord. ¹² Rosie is a dutiful fellow citizen who takes it upon herself to persuade Jim to fulfill his obligations. However, she can only try to influence Jim precisely because her powers are limited to those of persuasion. She could try to reason with him or appeal to his sympathies or even make him an offer. Still, Jim’s compliance is ultimately a contingent matter. He remains free to ignore the demands of justice if he is unpersuaded by Rosie. To determine Jim’s compliance would be to induce him to make the transfer even if he were unwilling to do so. Jim’s compliance would then be ensured. Rosie could determine Jim’s actions only if she was able to command him to comply and coercively sanction him if he refused. Unlike the state, Rosie lacks this power. The former’s ability to issue coercive commands and thereby determine the actions of others means it is the only entity that can ensure that morally mandatory transfers of distributive justice are made.

There are various difficulties with the conceptual argument. First, it is untrue that coercive commands can determine the actions of others. Individuals can still choose how they respond to those commands. For instance, an increase in the rate of taxation might bring in extra revenue or it might increase tax evasion. Coercive commands are still only attempts at influencing the behavior of others just as are the persuasive means available to non-state actors. The question in both cases is how probable it is that the attempt to influence others will succeed. Nor can we assume in advance that coercion is more effective than persuasion. We know from our personal life that the opposite is often true. ¹³

Second, it posits too sharp a distinction between the coercive means of influencing others available to the state and the persuasive means of influence available to nonstate actors. The empirical evidence on social norms suggests that state coercion is much more effective when consistent with public opinion. An informal social norm is a rule that governs certain kinds of behavior among members of a group. Unlike formal rules promulgated by the state, however, informal norms are not subject to coercive enforcement and lack the structure of secondary rules stipulating conditions under which they can be reformed or revoked. ¹⁴ There is now much evidence to suggest that informal social norms exhibit a certain priority to formal rules promulgated and enforced by the state. “Priority” here entails that if the state issues a coercive law that is in tension with a pre-existing, operative norm, then citizens will likely continue to follow the norm, in which case the law fails to secure compliance. The state would need to change social norms as well and this requires persuasion and deliberation. ¹⁵ In other words, social-scientific evidence suggests that states will ultimately have to rely on the same persuasive means as nonstate actors when influencing the actions of those who do not comply with the demands of distributive justice.

Third, even if states did have a special power to determine the actions of citizens, the basic problem of political theory is that there are no guarantees regarding how that power will be used. In democratic societies the way state power is exercised depends on how citizens vote; we cannot assume from the armchair that citizen voting patterns will necessarily induce the state to pursue just outcomes. Indeed, empirical analysis of existing welfare states often finds their benefits to be geared toward the interests of the middle class rather than the least advantaged. ¹⁶ In other words, even if the state could guarantee that mandatory transfers to the poor take place, there is no guarantee that the state would provide such a guarantee. All this is to say that the Default View cannot be inferred solely from the conceptual fact that distributive justice consists of morally mandatory demands. The comparative question facing us is whether citizens are more likely to undertake morally mandatory transfers when we rely on the state as opposed to civil society to attain justice.

III. The Legitimacy-Based Argument for the Default View

The legitimacy argument for the Default View runs as follows ¹⁷:

Citizens of a democratic state are both authors of and subject to a unified system of coercively imposed state laws—the “basic structure,” in Rawlsian terminology. The cost of exit is such that mere residence or acquiescence cannot render the coercion of such a system morally acceptable. It thus requires justification. Such a system is justifiable to all citizens only if certain conditions are met. One of these conditions is that the distributive outcomes to which the structure gives rise are just. The state should thus arrange that system in a manner that meets this condition. In other words, distributive justice should be left to the state because the very point of distributive justice is to render coercive state activity justifiable.

We can accept the substance of this argument. That is to say, we can accept that what triggers obligations of distributive justice between a group of persons is their being jointly subjected to a state-imposed, coercive basic structure and that such a structure is justified only if it does not give rise to unjust distributive outcomes.

The legitimacy argument, however, does not vindicate the Default View for the following reason. Even if state coercion gives rise to our collective obligation of distributive justice, it does not follow that we can discharge this obligation only through the state. In principle, it looks as if these could be discharged disjunctively. Imagine a society where the state is minimal in the libertarian sense. It imposes on all citizens a coercive system of laws granting each citizen extensive civil freedoms and rights to acquire and transfer property. These are negative rights. Except for some minimal public goods, the state does not grant citizens any positive rights to socioeconomic entitlements. Suppose, for the sake of argument, that this coercive scheme would be justifiable to all citizens only if each were entitled to a basic threshold of socioeconomic entitlements. Imagine now that there is a robust network of nonstate charities, food banks, mutual aid societies, and the like ensuring that no one falls below the minimum nor remains there long if they do. It looks as if citizens of this society have discharged their obligations of distributive justice to one another without recourse to the interventionist state despite those obligations being generated by state coercion. If our duties can be discharged disjunctively in this manner, then my claim that it is an empirical, comparative question as to which way they should be discharged is sustained.

IV. The Assurance-Based Argument for the Default View

The assurance-based argument for the default view is offered by Rawls. It runs as follows:

Each citizen will be motivated to sacrifice a portion of their resources for the cause of distributive justice only if they have assurance that a sufficient number of their peers will do the same. In large-scale societies, Rawls argues, the requisite assurance can be attained only if citizens know that there exists “some device for administering fines and penalties” to non-compliers and it “is here that the mere existence of an effective sovereign, or even the general belief in his efficacy, has a crucial role.” Conversely, he continues, such assurance cannot be gained through more decentralized means because “[i]n a large community the degree of mutual confidence in one another’s integrity that renders enforcement superfluous is not to be expected.” ¹⁸

The claim here is not that coercion is required because some citizens are unwilling to do their part. It is that, even if all citizens are willing to do their part, each citizen needs to be confident that a sufficient number of others will act on that willingness.

Let us take for granted this need for assurance. Robert Nozick suggests that assurance can be gained without the state via special contractual agreements. Each person would simply agree to donate a set sum if a set number of other persons agreed to do the same. ¹⁹ It does not seem inconceivable that such contracts could work on a mass scale. On the other hand, there is no guarantee that states will provide the requisite assurance. The capacity of states to enforce tax law is quite variable. Many states around the world are low in what political scientists call “state capacity,” and thus are limited in their ability to tax their citizens. ²⁰ Even if we restrict our attention to consolidated liberal democracies, there are stern limits on their ability to enforce tax laws, particularly on individuals who can move themselves and their capital. To the extent that citizens are aware that state capacity to enforce tax laws is limited, so too will the ability of states to provide assurance be limited.

A further difficulty with the assurance-based argument becomes apparent if we distinguish between three things:

1. (1) The domain population: the population in which a norm of behavior is expected to operate.

2. (2) The assurance population: the population whose rate of compliance each agent tries to ascertain to determine whether their own compliance is worthwhile.

3. (3) The compliance threshold: the rate of compliance that each agent requires as a condition for their own compliance.

The assurance-based argument assumes that the assurance population of each citizen is the whole domain population. Let me illustrate the point with an example. Suppose, for the sake of simplification, that each person in Bulgaria has a compliance threshold of 90 percent. The assurance-based argument assumes that a norm will take hold in Bulgaria only if each Bulgarian can be confident that 90 percent of all other Bulgarians will comply. When the problem is framed in this way, it follows for many that only an institution with a capacity for large-scale coercion can provide assurance, notwithstanding my earlier conjecture that nonstate mechanisms might also provide assurance on such a scale. For, of course, I cannot obtain detailed information on millions of my fellow citizens nor can I come to trust their integrity in the same way as I might that of close friends and family.

However, the assurance population of each person need not be the same as the domain population. Another, more plausible possibility is that the assurance population of each citizen is limited to those with whom they regularly interact. To adapt an above example, we might instead assume that it is sufficient for a norm to take hold in Bulgaria if each Bulgarian knows that 90 percent of the Bulgarians who they know personally comply with the norm. If the problem is framed in this way, the prospects for “bottom-up” assurance within civil society look much more promising. It is impossible to keep track of the behavior of millions of my fellow citizens, but I can gauge whether my local foodbank is well staffed with volunteers and I can come to know those volunteers fairly well.

Figure 1 illustrates how a norm might develop in this way among a group. Imagine a population of nine persons A through I. Suppose A and B start a new norm of behavior, for example, sacrificing their time on the weekend to help in a local food bank. A, B, and C are close friends of D, although C does not give up his time for the foodbank. Still, seeing two of her close friends doing that is sufficient to motivate D to do so. Observing D give up her time is sufficient for E to do the same, as is observing D and E for F. In turn, observing A, D, and F is sufficient to motivate G. Finally, observing G and F comply is sufficient to motivate I to comply, even though his other friend, H, does not. Seven out of nine members of this population come to comply with the norm, even though the assurance population of each is limited to those with whom they personally interact and not the whole group. Surely, many norms that are operative in large-scale societies but not coercively enforced arise in this manner.

Figure 1. Bottom-Up Assurance in a Nine-Person Population.

Once we account for the possibility of civil society actors generating assurance in a bottom-up fashion or through assurance contracts as well as the variable capacity of states to provide assurance through coercive means, we call into question the armchair conclusion that only states can provide the requisite assurance for distributive justice. We again face an empirical question: Which mechanism—state or civil society—is more effective at generating the requisite assurance?

V. Content-Based Arguments for the Default View

Content-based arguments for the Default View hold that if distributive justice is left to the decentralized decision-making of individuals and civil society associations, then the requirements of justice are indeterminate. We need the state to define what justice requires. I here consider two such arguments.

A. The state is needed to define justice in light of reasonable disagreement

The first content-based argument for the Default View runs as follows:

Political societies constitute a “common social world.” All citizens have equal stakes in this common world. It needs to be shaped in a way that is just, meaning that it advances the interests of each citizen equally. Yet citizens have pervasive but reasonable disagreements as to what justice—including distributive justice—requires. A duly authorized decision-making body, one in which all citizens have an equal say—that is, the democratic state—is required to determine the principles of justice that should shape our common world amid such disagreement. ²¹

Note that this argument addresses a somewhat different concern from that addressed so far in this essay. Taking the content of distributive justice for granted, I have argued we cannot assume that the state is the best instrument for realizing it. This content-based argument, however, insists that the democratic state is not merely an instrument to realize distributive justice, but also provides the mechanism by which we establish legitimately what distributive justice requires.

I can address this concern in two ways. First, I can observe that even if the state is needed to establish what justice requires, the comparative question of whether the state or civil society is better at implementing it remains a distinct matter. ²² Second, a more radical approach would be to argue that we do not necessarily need to turn to the state to determine how to live with our disagreements about justice. ²³ In adopting this latter approach, we could grant the wrongness of imposing one’s conception of justice on those who disagree with it. Our goal would then be to minimize, if not altogether eliminate, this kind of domination. Our comparative institutional question would be: “Which mechanism, the democratic state or civil society, best enables citizens with reasonable disagreements about distributive justice to live together without domination?”

Here is how we might think about the comparison between the mechanism of the state versus civil society. The democratic state solution to deep disagreement over contested issues is to vote. Members of the losing side, if nothing else, at least took part in the decision-making process and are free to contest it again at the next election. The civil society solution is to allow different understandings of distributive justice to hold sway in different associations, such as friendly societies, charities, churches, and so on. In other words, we disaggregate the common social world as much as possible into different jurisdictional spheres so that different understandings of distributive justice can live side by side rather than trying to split the difference with a uniform solution.

One worry with this approach, however, is that not all individuals have an equal say in civil society associations. The powerful in such associations, such as religious clerics, community leaders, or financial donors, might have a disproportionate say. They can impose their views on other members. This is a valid concern. Note, though, that a version of this worry also runs the other way. We have just observed how the democratic solution turns disagreement over distributive questions into zero-sum games: one side wins and the other side loses. Indeed, the electoral majority that wins and imposes its will on the rest of the population need not be, and frequently is not, a numerical majority of a democratic society. We can thus conclude that since neither the democratic nor civil society solution completely prevents some from imposing their views of distributive justice on others, we have a comparative, empirical question as to which one minimizes such imposition.

B. The state is needed to determine the content of our marginal contributions to distributive justice

A second content-based argument for the Default View holds that, even if there were agreement on the appropriate conception of distributive justice, we would still need the state to establish what our respective contributions to that endeavor would be. It runs as follows:

Individuals and civil society associations would find it very difficult to determine how to contribute to a society-wide distributive outcome. Each actor would have to understand what contributions all the other actors plan to undertake to determine how to make their own contribution and this coordination is very difficult to achieve in a decentralized fashion. The state, however, can adopt a synoptic view of the economy and intentionally allocate duties and entitlements to all citizens as necessary to attain society-wide distributive outcomes. ²⁴

It is, of course, true that state actors are typically able to adopt a more synoptic perspective than nonstate ones. Tax officials, for instance, might have data on the income and wealth of citizens across society, whereas each citizen might have knowledge of only their own income and wealth. ²⁵ This could put state officials in a better position than citizens themselves to decide, say, how much each citizen should contribute financially to distributive justice. However, such an observation alone cannot sustain the armchair presumption that only the state can resolve this kind of coordination problem. For one thing, we should not exaggerate the degree of centralized coordination the state is capable of. Coordination problems can arise within the state. Contemporary states are decentralized entities to a significant degree. Decision-making power within them is vested not just in national legislative agencies, but also in the courts, international government agencies, subnational government bodies, and administrative agencies. ²⁶ Different state agencies may have a synoptic perspective on a particular policy area—for example, the central bank on monetary policy, the finance department on taxes, or the antitrust regulator on corporate concentration—but no state agent or agency has a synoptic perspective on all policy areas. Contemporary state activity is so complex and variegated that no agent or agency can understand it comprehensively. ²⁷ The pursuit of distributive justice by “the state,” then, will require a great deal of internal coordination between myriad state agencies.

Just as individuals or civil society associations will struggle to adopt a synoptic vantage point to identify what action of theirs, given the actions of many others, will make a marginal contribution to distributive justice, so too could state agencies struggle to identify which policy initiative of theirs, given the policy initiatives of many other departments, will so contribute. Consider all the state-enforced and -promulgated rules whose coordinated reform might be necessary to meet a society-wide distributive outcome. A partial list of such policies includes national tax rates, local tax rates, education policy, regional housing policy, national infrastructure policy, immigration policy, competition policy, mental health provision, discrimination law, the regulation of lead, the regulation of product and capital markets, and carbon policy. All these policy domains have distributive implications. It is not a straightforward matter for policymakers in these areas to coordinate their reform efforts with one another.

At the same time, we should not underestimate the capacity of nonstate actors to form large-scale, centralized organizations capable of a great deal of internal coordination. It is important to recall here that collective action is not the exclusive purview of the state. Consider, as an example, the private welfare organization of the Mormon Church in Utah. In a famous article otherwise lamenting the lack of social mobility in the United States, Raj Chetty and his coauthors found that Salt Lake City had the highest rates of absolute upward mobility in the United States as well as rates of relative mobility comparable to Denmark. ²⁸ Yet, Utah is among the most frugal of U.S. states when it comes to public spending. The Mormon Church’s welfare system is one plausible explanation for this outcome—and Mormons account for 60 percent of the state’s population. ²⁹ Their organization is extensive and variegated, extending to hundreds of storehouses of food and supplies, including a 570,391-square-foot central storehouse, family counselling services, addiction recovery advice, and temporary employment services. ³⁰ It is funded, staffed, and administered by Mormon Church members, ³¹ serving predominantly (but not exclusively) the 1.62 million Mormon members living in Utah. ³² It might be objected that there are not many examples other than this one of nonstate welfare organizations operating at a large scale. ³³ However, this is arguably due to their being crowded out by state provision rather than their inherent infeasibility. In a world without state welfare, organized nonstate alternatives are likely to emerge. At the least, this is an empirical possibility that cannot be dismissed from the armchair. ³⁴

It is also worth noting that the interactions of nonstate actors might lead to or approximate a just, society-wide distributive outcome despite none of the actors intending such a result. Deliberate coordination by the state whereby each actor or civil society association is instructed as to how they should contribute to such an outcome might be unnecessary. Take the example of the friendly societies in the United Kingdom during the nineteenth and early-twentieth centuries. ³⁵ Those were self-governing associations created and managed by workers who performed many of the functions we today associate with state welfare. Members paid regular contributions that entitled them to a range of benefits. The benefits offered by those societies included sick pay in case the member was unable to work, medical care for the member and his family, financial support for widows, a grant to avoid a pauper’s funeral, and so on. ³⁶

David Green reports that by the time the British Government introduced compulsory social insurance as part of the 1911 National Insurance Act, at least 9 million of those covered were already members of friendly societies or other voluntary insurance associations. ³⁷ The rate of growth of registered membership in friendly societies rapidly rose from 2.75 million in 1877. ³⁸ According to Green, the post-1911 decline of the friendly societies was due to crowding out by state welfare. ³⁹ It seems safe to say that members of the friendly societies did not intend to contribute to the society-wide distributive outcome of an ever-increasing number of British citizens having access to an increasing array of socioeconomic entitlements. (Nor, perhaps, would they have known how directly to contribute to such a grand outcome.) The primary intention of each worker was to insure themselves, their families, and their peers against hardship, but the society-wide distributive outcome was the emergent result of their myriad individual and collective efforts over the decades. Indeed, Green documents the modest aims of the very first friendly societies and how they grew in both size and sophistication only over time and through much trial and error. ⁴⁰

Once we acknowledge that civil society actors are capable of forming large-scale organizations to pursue distributive justice, the state is not free of coordination problems, and society-wide distributive outcomes can be and have been attained by emergent nonstate processes, we can call into question the armchair conclusion that only the state can facilitate the coordination necessary for distributive justice. To reiterate the by-now-familiar theme of this essay, we have an empirical question concerning which mechanism better overcomes these issues.

VI. Presumptive Reasons in Favor of Nonstate Provision

In the first half of this essay, I argued that the Default View is false. We must, to paraphrase Ludwig Wittgenstein, “look and see” whether state or civil society actors are more effective at promoting distributive justice. I now offer some considerations suggesting a presumption in favor of nonstate mechanisms. I do not pretend that these remarks are the final word on the issue; no doubt, a reader who might be persuaded of the claims of previous sections might still think the presumption should lie with the state. Still, I offer these remarks in the hope of instigating further debate about this issue.

A. Civil society actors are better placed to promote actors’ substantive capabilities

Amartya Sen has argued famously and forcefully that the currency of distributive justice should not be the input of resources that individuals receive but the output of real freedoms to do things and pursue goals—“capabilities,” in his lexicon—that those resources can be used to secure. ⁴¹ Different individuals need differing quantities of input resources, such as income and wealth, for each to achieve a sufficient set of capabilities. In other words, they have different conversion rates of resource inputs into capability outputs. A disabled individual, for instance, would need more resources than an able-bodied one to achieve the same degree of mobility. ⁴² There might be a temptation to think that such individuals are a special case such that non-disabled persons all have similar conversion rates. However, Sen observes that disability is just one factor responsible for differences in conversion rates. Others include: “health, longevity, climatic conditions, location, work conditions, temperament, and even body size (affecting food and clothing requirements).” ⁴³ Sen’s argument is directed principally against Rawls, for whom the currency of distributive justice is an input resource—or “primary goods,” in Rawls’s terminology. Rawls’s view tends toward giving citizens the same standardized entitlement of primary goods. ⁴⁴ For Sen, Rawls’s approach is insufficiently flexible to accommodate individual differences.

Sen’s approach is compelling insofar as it tracks accurately each individual’s particular interests and needs. The difficulty with it, though, is that the informational requirements are demanding for the state to meet. In order to promote each citizen’s capabilities, the state needs a great deal of particular information about them and their idiosyncratic conversion rates. However, this is precisely the kind of information that state actors find difficult to obtain. Robert Goodin puts the point well:

[Public officials] cannot know, in full detail, the circumstances of all people whom they purport to govern … . Nor, given the clumsy policy instruments available to them, can public officials achieve the precise, fine-tuned response to particular circumstances that might be commended to (and expected of) private individuals. By its very nature, government must work through rules that are general in form … . [T]hose rules will be perfectly suited to few, if any, of the actual cases falling under them. ⁴⁵

One might wonder whether state policy can be more flexible than Goodin implies. States might be capable of tailoring their policy responses to particular circumstances by giving public officials extra discretion or by adopting mechanisms such as subsidiarity or federalism. Indeed, we observed just above that decision-making within consolidated liberal democracies is decentralized to a significant degree.

However, it is at this juncture where Sen’s approach faces a powerful challenge from Rawls, who argues that there is a stern moral limit to states’ ability to tailor their provision of distributive justice to local circumstances. This limit stems from certain publicity requirements that state administration of distributive justice must meet. Justice not only needs to be done, but also needs to be transparently seen to be done. ⁴⁶ When the state administers justice, claims of justice premised on particular information—such as an individual’s personal conversion rate—cannot meet the publicity requirement. Such facts are not transparent to impersonally related citizens of large-scale democracies. Suppose that citizen Rosie claims she is owed more than others by the public purse because of some special circumstance that involves her having a higher conversation rate than other citizens. Those reasons are opaque to Jim, another citizen who does not know Rosie. It is difficult for him to tell whether Rosie is being sincere or making a more self-serving claim. The worry here is that political trust and stability cannot be maintained if citizens are unable to publicly verify one another’s claims. By contrast, if distributive justice entails allocating to each citizen the same standardized input resources, as is more or less required by Rawls’s theory, it is much easier for citizens to verify the validity of their respective claims.

State administration of distributive justice, then, forces us into a dilemma. We are forced to choose between publicity and sensitivity to particular data. However, this dilemma can be transcended if we rely on civil society to promote distributive justice. Private individuals and civil society associations have access to localized and particular information to which state actors do not. Indeed, the strengths and weaknesses of state and nonstate actors are mirror images of one another in this regard. A state actor can gain a synoptic perspective on some aspect of society precisely because they act on the basis of standardized statistics that scope out much localized and particular information. ⁴⁷ Nonstate actors lack this synoptic perspective when they act in situ, but this also means they are better placed to offer a “precise, fine-tuned response to particular circumstances” than are state actors.

Additionally and crucially, private citizens and civil society associations do not have to worry about meeting the publicity requirement if they are using their own resources. Suppose, this time, that citizens Rosie and Jim are close friends. Jim is aware that Rosie needs temporary financial assistance because, say, a hidden disability makes it difficult for her to find work. Jim assists Rosie with his own money. In this instance, only Jim needs to be persuaded of Rosie’s veracity. Compare this with an amended version of this scenario where Jim is a state-employed social worker who has gotten to know Rosie well and feels she deserves extra funds in light of her particular circumstances. He helps her not with his own money but with public funds. Here, the demanding requirements of publicity kick in. The mass of other taxpayers lack Jim’s local knowledge and so cannot verify the cause on which their money is being spent. One reason, then, for a presumption in favor of civil society is the capacity of civil society actors to promote individuals’ substantive capacities without running afoul of the publicity constraint to which state action is subject.

B. Direct action should take precedence over political activity when we reason from the “first-personal” standpoint

Another justification for such a presumption in favor of civil society actors stems from the concrete trade-offs that real-world actors face when deciding how to pursue their goals. We can distinguish between two different evaluative standpoints: impersonal and first-personal. The impersonal standpoint evaluates institutions from a detached, social scientific perspective. This is the standpoint we have adopted up to this point. We have asked: Given compliance, assurance, and epistemic constraints on collective action, which mechanism of social organization—the state or civil society—is most effective at promoting distributive justice? The first-personal standpoint considers how a concrete individual could make a marginal contribution to justice, given the constraints they face personally.

Citizens could personally promote justice either through direct action in civil society or indirectly by trying to induce the state to promote justice. Even if a citizen concludes that the state should promote distributive justice when reasoning impersonally, she should, when reasoning first-personally, conclude that her own justice-promoting efforts should be focused primarily on direct action in civil society. For she can make a tangible difference to the lives of at least some disadvantaged citizens if she directly contributes her own time or resources, while her chances of changing state activity by, say, having a decisive vote in an election are negligible. ⁴⁸ Consider a practical example of food banks, the use of which has risen markedly in recent decades in countries like the U.K. and the U.S. ⁴⁹ Even if distributive justice would be better promoted by state action than by civil society provision through food banks, most of us could personally contribute more to distributive justice by giving our time or money to a food bank than by, say, trying to change state policy or encouraging others to vote for “food justice” candidates. ⁵⁰

C. Civil society provision of distributive justice could further other independently valuable goods

A third justification for a presumption in favor of civil society mechanisms is that they can secure values other than distributive justice, such as practical reasoning. Many political philosophers think that distributive justice is among the most important moral commitments that an autonomous actor could have. Suppose we held that we should, all other things being equal, try to exercise our capacity for practical reasoning as much as possible in the way we discharge our most important moral commitments. It would follow that, again all other things being equal, we should prefer those means of promoting distributive justice that give us greater opportunity to exercise our capacity for practical reasoning. ⁵¹ Individuals have a great opportunity to engage in practical reasoning when they are able to decide for themselves how to promote distributive justice. They would have to determine not simply how much resources in terms of time and/or money to contribute, but also more broadly how to discharge their responsibilities. Mark Pennington offers a nice list of the kind of questions an individual would have to answer:

Is a person best placed to help the disadvantaged by starting a new enterprise and employing poorer sections of the population? Would it be better for someone to take a high-paying job and contribute part of his or her income to a charitable association? Does a person have an aptitude for charitable work? If charitable activity is indeed the best way for a person to help the disadvantaged, should this take the form of monetary contributions or spending time directly with the less well off in order to provide education, advice on healthcare, or the transmission of values conducive to prosperity? ⁵²

By contrast, citizens have less of an opportunity to express their capacity for practical reasoning when they rely on the state to promote distributive justice. Citizen responsibilities under those circumstances are limited primarily to obeying state commands, for example, by paying their taxes. It might be objected that citizens could play a more active role than this. They could also vote, lobby their representatives, join a political party, and so on to ensure that the state maintains its commitment to distributive justice or better delivers on it. These methods, it might be argued, would leave citizens with ample opportunity to express their capacity for practical reasoning. ⁵³ The difficulty here, though, is that sustained political engagement will be central to the life plans of only a minority of citizens in any modern, complex society. ⁵⁴ Many citizens will have only limited engagement with politics. They might vote once every few years or perhaps attend the odd protest or political party meeting. Many others will be completely apathetic. These citizens will be positioned to properly exercise their capacity for practical reasoning with respect to distributive justice only if they bear that burden directly in their personal lives.

Another value at stake in this regard is relational in nature. As observed above, some political philosophers hold that the social process through which a distributive outcome is attained—and not simply the attainment of that outcome per se—is of moral importance. Even if we are advocates of outcome-based principles of distributive justice, we should, all other things being equal, pursue just outcomes through those social processes that uphold important social values. Solidarity and care are two such values. We should prefer those means of promoting distributive justice that provide citizens with greater opportunity to express attitudes and sentiments of solidarity and care to one another. The degree to which a person is willing to sacrifice for others is one powerful measure of the sentiments of care and solidarity they have toward them. Where individuals have the discretion to decide for themselves how much of their resources, such as time and money, to spend on distributive justice, their sacrifices are clearly willed. Conversely, where individuals are required by threat of coercive sanction to sacrifice for distributive justice, their sacrifices are not willed. If the responsibility for distributive justice falls primarily on the state, then individual sacrifices are required, not willed. Individual citizens cannot decide for themselves, for instance, how much tax to pay. Therefore, all other things being equal, we should prefer relying on nonstate mechanisms to promote distributive justice because they provide individuals with opportunities to express sentiments of care and solidarity that are lacking in state mechanisms. ⁵⁵

It might be objected that citizens can still express their willingness to sacrifice for distributive justice by voting for just parties. If I vote freely for a party that will increase my tax rate, surely that expresses willingness to sacrifice as well. The difficulty here is that my personal vote has a negligible impact on the overall result. Even if the “high tax” party wins, they would have won, in all likelihood, regardless of how I voted. My voting for them, then, does not look like a real sacrifice since my action has no practical impact. Moreover, any putative sense of sacrifice when voting for higher taxes is lessened by the fact that one is typically voting to spread the cost across many taxpayers, not just oneself.

Note that I have been careful to add the “all other things being equal” qualifier to my remarks. Thus, suppose we judge empirically that the state is very much better than nonstate actors at securing just outcomes. I am not claiming that under such circumstances we should still opt for nonstate provision because they provide actors with greater opportunity to express their rational faculties and sentiments of care and solidarity. Under those circumstances, the superiority of the state in producing just outcomes may well be decisive. But suppose, instead, that it is unclear whether the state is more effective in promoting just outcomes. The presumption should then be in favor of nonstate mechanisms in virtue of these other considerations.

VII. Conclusion

I end by offering some brief remarks as to the wider implications of this essay’s argument for theorizing about distributive justice. Five decades after the original publication of Rawls’s A Theory of Justice, the intense interest in distributive justice among political theorists appears to be cooling. One reason for this is that the connection between the various theories of distributive justice prevalent in the academy and practical political change is obscure. Existing states fall far short of the responsibilities placed upon them by these theories. Sometimes, states have not attempted to pursue distributive justice, while other times, the results have proved disappointing when they have tried. No doubt this has in large part inspired the recent resurgence of realism in political theory. If states do not behave in the real world as they do in our idealized theories of distributive justice, the thought runs, we ought to pay more attention to the political processes that affect state action in the real world. ⁵⁶ The argument of this essay suggests another response to the impasse into which theorizing about justice has fallen. We can start to inquire further into how men and women can circumvent the state and further distributive justice on their own initiative.