The welfare-convergence dilemma: why social insurance is objectionable in the convergence conception of public justification

Abstract

Recently, convergence liberals, such as Kevin Vallier, argue that the principle of social insurance could be publicly justified. Our paper challenges this marriage of convergence liberalism and welfare state. We begin by examining Vallier’s three reasons for the principle of social insurance: risk aversion, injustice and the promotion of political trust. We then argue that all these reasons are intelligibly objectionable. After examining five possible responses that convergence liberals may offer, this paper concludes that the principle of social insurance is not conclusively justified in the convergence conception of public justification.

1. Introduction

In recent years, the convergence conception of political liberalism, which is called ‘convergence liberalism’ for short, has received increasing attention in political philosophy. Gerald Gaus and Kevin Vallier, its representative philosophers, developed this theory based on the idea of public justification, which requires that political rule must be justified to all under its authority (Gaus 2011, 2015, 2021, 2022; Vallier 2014, 2019, 2021). In contrast with Rawlsian political liberalism (Rawls 2005; Quong 2011; Watson and Hartley 2018; Neufeld 2022; Badano and Nuti 2024), convergence liberals deny that public justification must be limited within a set of shared or accessible public reasons. Rather, citizens should be permitted to reject laws and policies on the grounds of comprehensive reasons, even if these reasons are not accessible to other citizens. Consequently, many egalitarian redistributive policies would be rejected within this framework (Vallier 2021: 173–209). This causes a clear departure for convergence liberalism from Rawlsian political liberalism on the matter of distributive justice.

Convergence liberals have recently shown growing sympathy for welfare policies in their latest developments. The earlier Gaus was sceptical of welfare policies ‘because of the radical uncertainty of predicting the consequences of government policies designed to promote social goals’. Thus, a ‘liberal state … should generally eschew the field of public policy’ (Gaus 1998: 32–33). However, Gaus became more sympathetic to the welfare state in his later writings. On the one hand, Gaus admitted that convergence liberalism has a ‘classical tilt’, meaning that the convergence conception of public justification shares some similarities with a classical liberal position that favours fewer state interventions to the free market (Gaus 2010: 273). Yet, on the other hand, he did not rule out that some state interventions to provide ‘basic goods necessary for agency’ could be publicly justified (Gaus 2011: 358–359). Following the later Gaus (2021: 195), Vallier argued that the principle of social insurance could be publicly justified according to the convergence conception. This principle justifies state interventions in the free market to ensure that citizens have ‘access to goods and services vital for them to remain normally functioning persons and citizens’ (Vallier 2021: 149). Importantly, Vallier believed that even those ‘conservative and libertarian members of the public’ could not have reasons to reject the redistributive polices required by the principle of social insurance (Vallier 2021: 148). To Vallier, the principle of social insurance could be publicly justified according to the convergence conception, surviving changes of people’s justifying reasons much like other liberal rights, such as the right to private property (Vallier 2021: 139). Through various arguments, Vallier introduced a distinctive middle way between the Rawlsian property-owning democracy and the Nozickian minimal ‘nightwatchman’ state. Despite the profound diversity in modern societies, some welfare policies can still be publicly justified among citizens.

The redistributive implications of convergence liberals have often been discussed in the literature (e.g. Lister 2010, 2013: 81–104; Bou-Habib 2015). However, the discussion has predominantly centred on the theory of Gaus. More recently, Vallier has presented several new arguments to explain why social welfare could be publicly justified, but these have not yet received sufficient attention. This paper critically evaluates Vallier’s arguments for the consistency between the principle of social insurance and the convergence conception of public justification. ¹ We argue that the principle of social insurance is at most an option in the optimal eligible set in the convergence conception, but moderate idealized agents have no conclusive reasons to stably endorse it. ² Sometimes social insurance might emerge as a temporary point of convergence, but such agreements are neither enduring nor possess the moral significance ascribed by convergence liberals. This result contradicts Vallier’s ambition to make the principle of social insurance a constituent part of the convergence liberal project, which ‘requires polities to have basic social programs’ (Vallier 2021: 139; our emphasis). If Vallier succeeded, the convergence liberal project would make significant progress, as convergence liberals before Vallier, notably even the later Gaus, only hinted at the possibility of ‘not ruling out’ welfare policies, short of providing a positive justification. However, we argue that convergence liberals would then be trapped in a Welfare-Convergence Dilemma: they cannot simultaneously uphold the legitimacy of welfare state policies and the convergence conception of public justification.

In section 2, we first outline Vallier’s understanding of the convergence conception of public justification and the principle of social insurance. Then, in section 3, we argue that the reasons that Vallier offered to support the principle of social insurance are intelligibly objectionable. Thus, moderately idealized agents have no conclusive reasons to endorse the principle of social insurance. In section 4, we examine five possible replies, which are (i) to justify the principle of social insurance through a higher-order social choice mechanism, (ii) to justify the principle through the fact that it is in the optimal eligible set and would be chosen in the evolutionary mechanism, (iii) to show that the principle can be the convergence of multiple social choice mechanisms under certain conditions, (iv) to justify the principle by the value of social coordination and (v) to justify the principle by bundling it with other more widely accepted principles in public justification. We further argue that all these replies fail to rescue the principle of social insurance in the framework of convergence liberalism. The upshot of this paper is that the principle of social insurance, and therefore the typical set of policies associated with the welfare state as we know it today, is not necessarily chosen in convergence liberalism.

Before presenting our argument, we will elucidate its relationship to two prevalent criticisms of convergence liberalism: the anarchy objection and the indeterminacy objection. The anarchy objection means that, since the convergence conception of public justification permits too many reasons to defeat potential options of agreement, this ultimately leads to a situation where no principles can be publicly justified. Consequently, convergence liberals are left with no choice but to accept a state of anarchy (Eberle 2011: 301; Wall 2013: 169; Enoch 2015: 118). The indeterminacy objection, on the other hand, asserts that the convergence conception considers too many principles acceptable. Thus, moderately idealized agents face difficulty in reaching an agreement on which principle to select (Lister 2017: 73–74; Kogelmann 2017: 211–215; Kugelberg 2022: 54–61). Our objection is different from the anarchy objection. While we suggest that the principle of social insurance could be intelligibly objectionable, we also acknowledge, as many convergence liberals do, that anarchy would lead to catastrophic outcomes, prompting moderately idealized agents to make concessions to one another in the pursuit of a coordination point (Vallier and Muldoon 2021: 227). Nevertheless, in section 4, we further argue that too many principles could be seen as preferable to having none, and there is no conclusive reason to believe that the principle of social insurance would necessarily be chosen. Even if it were chosen, the outcome would be affected by many non-moral factors and could at most be considered a modus vivendi, lacking the moral qualities that the convergence liberals find satisfactory. Hence, our objection can be viewed as a version of the indeterminacy objection. Our objection, however, is narrowly focused on the principle of social insurance because we believe that this indeterminacy becomes particularly pressing when it comes to the issue of distributive justice.

Throughout this paper, we assume that basic liberal rights, such as freedom of speech and the right to private property, are more broadly acknowledged and enshrined in the constitutions of numerous contemporary societies (Neier 2012: 104–105). Unlike these issues, distributive justice has long been regarded as one of the most contentious issues in modern societies. For example, Ronald Dworkin once highlighted the intensity of this debate, observing that the topic of distributive justice stands as one of the most intricate, with a broad spectrum of differing viewpoints. ‘Our controversies about justice are too rich, and too many different kinds of theories are now in the field’ (Dworkin 1986: 74). Similarly, when Robert Dahl discusses what rights should be protected in democracy, he excludes economic rights since they are often too controversial: ‘claims to primary social rights tend to be more debatable and uncertain in democratic theory and practice than claims to primary political rights’ (Dahl 1980: 14). Therefore, this paper argues that convergence liberalism stands a greater chance of overcoming the indeterminacy objection in the context of basic liberal rights, yet this paper casts scepticism on its ability to do so within the scope of distributive justice. In sum, this paper has two contributions. First, it analyses Vallier’s latest defences of the principle of social insurance, highlighting their shortcomings in addressing the indeterminacy objection. Secondl, it critically examines several recent counterarguments to the indeterminacy objection offered by convergence liberals and shows their inadequacies (e.g. Kogelmann 2017; Schultz-Bergin 2021; Vallier 2021). Collectively, this paper presents a significant challenge to the philosophical relationship between the welfare state and convergence liberalism.

2. Vallier’s Convergence Conception and the Principle of Social Insurance

We shall present a summary of Vallier’s theory of public justification and his defence of the welfare state. Vallier justifies the need for public justification by the value of social trust. The problems of rising political polarization and declining trust have been increasingly extreme in these years, especially in the USA. In his recent two books, Must Politics Be War and Trust in a Polarized Age, Vallier argues that this division is caused by the mutual reinforcement between polarization and distrust: polarization fosters distrust, and vice versa (Vallier 2021: 6). Polarization occurs because citizens profoundly disagree with each other about morality and the requirements of justice. Due to these disagreements, mutual distrust emerges among citizens. Nevertheless, Vallier believes that citizens would trust each other when they find each other trustworthy, and citizens find each other to be trustworthy when they observe each other complying with certain social norms for moral reasons (Vallier 2019: 37; 2021: 50). So long as adequate evidence is available to every citizen that others consistently comply with social norms, which are the laws and moral rules in a society, a relationship of mutual trust can be developed (Vallier 2019: 35).

However, Vallier does not believe that all kinds of norms can increase trust in ‘the right way’ (Vallier 2019: 6). Only those norms that can be publicly justified can secure a rational basis for social trust among people with diverse views. Also, only when the state acts on those publicly justified norms can the state secure political trust from the people. Here Vallier interprets the idea of public justification as a convergence conception, formed mainly by two components: moderate idealization and the intelligibility requirement. First, Vallier does not mean to identify those norms that are publicly justified to actual people, for actual people may wrongly endorse or reject norms based on wrong information or bad reasoning. Thus, the constituency of public justification should be moderately idealized, which means that public justification only includes those reasons that people ‘would affirm if they were adequately rational and informed’ (Vallier 2019: 8). In other words, moderate idealization ‘cleans up’ the commitment of actual people. ³ Second, public justification should proceed in terms of each person’s intelligible reasons. A citizen’s reason is intelligible when others can justifiably regard this citizen as entitled to affirm this reason given this citizen’s evaluative standards (Vallier 2019: 91). For example, a Christian may reject a law that permits same-sex marriage on the grounds that God forbids it in sacred texts. Although this reason is not shared by secular members of the public, it is intelligible, given the evaluative standards of Christianity. Accordingly, many reasons that are not shared among citizens, such as religious reasons, are permitted to be used to support or reject norms in public justification, as long as these reasons are intelligible to others. Eventually, only those norms that each moderately idealized agent has different, yet conclusive, moral reasons to endorse can satisfy the requirement of public justification (Vallier 2019: 116; cf. Billingham 2017: 545–546).

In this process, agents would reject many norms for different reasons. After showing that property rights could be publicly justified (Vallier 2021: 119–138), Vallier then shows that another norm that could be publicly justified is the principle of social insurance, which means that the state should ‘provide income and resources to people who, for one reason or another, lack access to goods and services vital for them to remain normally functioning persons and citizens’ (Vallier 2021: 149). The policies implied by this principle include ‘tax-financed health insurance, unemployment insurance, disability insurance, old-age pensions, and food stamps’ (Vallier 2021: 149). ⁴ Here Vallier offers three reasons to explain why this principle is publicly justified among moderately idealized agents:

(A) First, it is widely acknowledged that humans’ wellbeing is harmed when they are unable to access food, healthcare, shelter, and other resources necessary to meet one’s basic needs. Even if one can have these goods now, one would fear that one day she may lose these goods. Hence, social insurance can ‘insure persons against the loss of these goods in order to both directly promote their welfare and reduce the fear associated with the loss of such goods’ (Vallier 2021: 149).

(B) Second, many poor and needy people suffer from misfortunes that they cannot control. For example, some of them are poor due to their social and family background. Their poverty may further affect their offspring, who also have no control over their background since their birth. Therefore, to ‘alleviate injustice against the poor and needy,’ the government should provide social insurance for these people (Vallier 2021: 149).

(C) Finally, social insurance plays a critical role in sustaining political trust. There is good evidence that the welfare state significantly affects the main determinants of political trust, which are economic performance and public corruption. The welfare state can spread the gains from economic growth across the population and thus increase the perceived economic well-being of the large majority. The increase in well-being enables citizens to judge that the state performs its expected public function of increasing public prosperity. Also, a highly economically unequal society usually has severe political corruption (Vallier 2021: 141–142). Thus, by narrowing down the wealth gap, the public perception of corruption in government is reduced.

In brief, due to the risk of poverty, injustice and political trust, Vallier argues that it would be publicly justified to have a Hayekian modest welfare state that provides a social minimum unconditionally to all poor and needy in society (Vallier 2021: 170). The principle of social insurance and the institutions demanded by it ‘will likely be part of any publicly justified polity’ (Vallier 2021: 171).

3. Can the Principle of Social Insurance be Publicly Justified in the Convergence Conception?

In this section, we shall examine Vallier’s three reasons for the principle of social insurance in turn.

3.1. Risk aversion

In general, welfare policies can be explained by the logic of insurance. Insured people make regular payments to participate in an insurance scheme. These funds are pooled by the insurance company so that, if an insured person incurs a loss, it is covered by the collective premiums of all insured people (Halliday and Thrasher 2020: 135–136). Similarly, social insurance can be compared to a mandatory, nationwide insurance programme that safeguards against the risk of losing access to resources necessary for fulfilling basic needs. For Vallier, this programme is publicly justified since it provides protection to every agent against significant losses that these agents usually fear at a fundamental level. Thus, every agent would be willing to join this programme and be protected by this safety net.

Yet, there is one crucial difference between buying insurance from private companies and being covered by the social insurance policies of a welfare state. People can arguably always adjust their private insurance policies according to their personal aversion to risk and determine the premium they wish to pay accordingly. ⁵ However, in the welfare state, one cannot choose how much tax to pay in return for one’s preferred level of insurance. In other words, social insurance in the form of the welfare state cannot always be adjusted to fit each individual citizen’s risk aversion. To implement social insurance, the state inevitably imposes the same level of risk aversion on citizens who have different conceptions of risk (or, at best, several levels of risk aversion). This difference between private and social insurance makes the principle of social insurance potentially objectionable. People’s levels of risk aversion to the loss of different kinds of necessities are closely tied to their own comprehensive doctrine. Some entrepreneurs may have a greater ambition and inclination toward taking risks, while others may prioritize stability and tend to be more risk averse. When the state imposes a unform level of risk aversion on all citizens through social insurance, some citizens may view this as imposing a comprehensive doctrine to which they may intelligibly object. The principle of social insurance, therefore, cannot be publicly justified according to the convergence conception.

One possible response from convergence liberals is that moderately idealized agents would at least agree that access to certain very basic provisions, such as food and shelter, is necessary, because such a minimum level of risk aversion is rational for everyone. Nevertheless, while meeting certain basic needs is crucial for everyone’s well-being, people may still have different opinions on how to ensure these needs are met. For instance, in setting up the social insurance system, how much responsibility to avert risk should be assumed by individuals, and how much should be assumed by the state? Some moderately idealized agents may intelligibly reject the social insurance system as overly insensitive to personal responsibility. Those who want to enjoy welfare-state benefits must do something in return, or satisfy certain eligibility requirements – for example, those recipients must prove that they are taking responsibility to find a job (Hartmann and Honneth 2006: 51–52; Shapiro 2007: 248). As David Schmidtz puts it, the point of welfare policies should be ‘asking our institutions to make people willing and able to fend for themselves’, as opposed to guaranteeing that individuals will never have to provide for themselves (Schmidtz 1998: 21–22). Similarly, Daniel Shapiro claims that people’s ‘rejection of unconditional welfare is reasonable – they believe it is unfair that they are required to provide able-bodied adults income without the latter ever being required to work, and they believe that long-term dependence on government welfare is bad for the recipients’ (Shapiro 2007: 275–276).

An example of widespread disapproval of unconditional welfare is the enactment of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (PRWORA) in the USA. PRWORA introduced the Temporary Assistance for Needy Families (TANF), which set limits on the number of years a family could claim benefits, and required welfare recipients to work in order to qualify for aid. It replaced the previous Aid to Families with Dependent Children (AFDC), which provided qualified families with income support based on the number of children they had, without a work requirement. The very introduction of PRWORA reflected many voters’ and politicians’ concerns at the time regarding the previous welfare system’s neglect of personal responsibility (O’Connor 2001). To these voters and politicians, a conditional welfare system was more preferable to a model that was insensitive to personal responsibility. After all, like the level of risk aversion, the level of personal responsibility to risk is something closely tied to their comprehensive doctrine. This belief, in turn, influences people’s opinions on how a social insurance system should be structured. Despite our disagreement with the priority given to personal responsibility as advocated by the theorists and politicians mentioned above, it is hard to deny that these reasons are intelligible reasons in Vallier’s sense. This poses a problem for the convergence conception of public justification: it cannot neglect beliefs grounded in these reasons. Consequently, it follows that the principle of social insurance could be intelligibly objectionable.

3.2. Injustice

Vallier’s second reason for the principle of social insurance is that it could alleviate the injustice of resource distribution that results from the free market. People have no control over their social and family background after all. Here Vallier seems to appeal to people’s sentiments of compassion or sympathy for others to explain why it is an injustice for some to be impoverished due to circumstances beyond their control, as some egalitarians have proposed (e.g. Anderson 1999). However, while it is reasonable to assume that moderately idealized agents have a general sense of sympathy and compassion, it is far more controversial to suggest that everyone must have the same level of sympathy or compassion toward individuals who are experiencing similar hardships. Some people may feel a stronger altruistic sense of obligation to help anyone in need, while others may prefer to reduce social insurance levels to save money for assisting their own family or community members.

Furthermore, while it is generally considered unacceptable that impoverished individuals would perish from starvation on the streets without any aid available, there is intelligent disagreement on whether such assistance should be provided through public policies enacted by government agencies. For instance, people might suggest that welfare state policies are less efficient than private charity (Shapiro 2007: 234–235; Huemer 2013: 150–151). They might also believe that the root cause of poverty lies in the economic structure, which restricts individuals’ freedom and opportunities to access markets and improve their circumstances through work (Puerta et al. 2018). As a result, some may suggest that, instead of basic welfare, it would be better to focus on infrastructure, such as reforming the criminal justice system to avoid overcriminalization and providing ample educational opportunities to everyone as more effective measures to alleviate poverty (Tanner 2020; Zwolinski and Tomasi 2023: 217–218). All these suggestions could entail an intelligible rejection of the state provision of social insurance.

Additionally, Vallier might argue that it is unjust for people to be poor due circumstances beyond their control because such individuals are undeserving of their poverty. The idea behind this is that no one should suffer as a result of conditions they did not choose voluntarily. This assumption about the grounds of desert is, however, controversial and could be intelligibly rejected. For instance, the claim that voluntary choice or control should determine one’s entitlement to material resources is debatable. It is unlikely that we would argue a person does not deserve a prize in a mathematics competition simply because her mathematical talent was inborn and not her choice. On the other hand, it can be persuasively argued that need alone may constitute a legitimate claim to resources, regardless of whether one’s state of need stemmed from personal choices (Sher 1979; Miller 1999: 131–133; Scheffler 2005: 9–10). It is, therefore, unclear whether undeserved inequality can be a compelling reason to conclude that social insurance passes the test of the convergence conception of public justification.

3.3. Political trust

Vallier could appeal to the benefits of trust to justify social insurance within the convergence conception of public justification. If the positive outcomes of enhanced political trust are significant or compelling enough, moderately idealized agents could have strong reasons, when considering all factors, to support social insurance unanimously. The persuasiveness of this argument depends on empirical evidence that demonstrates the beneficial impact of social insurance, or the welfare state, on trust. Vallier admits that the empirical literature provides limited evidence to support that the welfare state promotes social trust (Vallier 2021: 140–141). Instead, he places greater emphasis on evidence suggesting that the welfare state can promote political trust. Specifically, according to Vallier, a well-functioning universal welfare programme can boost economic performance and facilitate a more equitable distribution of resources while also reducing political corruption. As a result, people are more likely to place their trust in the state (Vallier 2021: 141–142). ⁶

The issue with Vallier’s argument is that, empirically, different combinations of welfare state policies, influenced by different social and historical contexts, can have varying impacts on the level of trust among different peoples. Therefore, it is unlikely to have a straightforward linear relationship between the level of trust and the welfare state. For example, a recent study found that Bolsa Familia in Brazil, a welfare policy used by Vallier as an example, has a 44% error rate in targeting its intended recipients. This undermines people’s political trust in the government. However, poverty alleviation schemes in South Africa that use a similar means-testing method have only an 8–13% error rate and have successfully enhanced political trust. This is because, in South Africa, the method was seen as excluding the wealthy, rather than stigmatizing the poor (Kidd and Athias 2020: 7–9).

In contrast, France initially implemented a universalist pension scheme after the Second World War, with a high degree of trust placed in the state as the protector of social solidarity. However, as economic growth decelerated and the demographic landscape aged, the burden of taxation and contributions increased. Since the 1990s, there has been growing scepticism about the state being ‘captured’ by special interest groups (Clark 2003: 66, 69–71). Pension reform became so controversial that it sparked massive protests and violent clashes in 2019, and again in 2023 (Breeden 2023). For moderately idealized agents to consider the principle of social insurance to be publicly justified according to the convergence conception, its positive impact on political trust must be clearly shown. If the positive effect of the principle of social insurance on political trust is contentious and therefore ambiguous, the principle fails to meet the criteria for public justification.

3.4. The welfare-convergence dilemma

To summarize the above discussion, it appears that moderately idealized agents in the convergence conception of public justification lack conclusive reasons to endorse the principle of social insurance; consequently, this principle fails to be publicly justified. Convergence liberals are thus trapped in a dilemma: they cannot simultaneously uphold the legitimacy of welfare state policies without compromising their commitment to the convergence conception of public justification.

Facing this dilemma, convergence liberals have two possible escape routes. The first way is to steadfastly adhere to the convergence model while conceding that welfare state policies may not be selected in public justification. Unlike basic liberal rights, the principle of social insurance fails to be legitimate because moderately idealized agents would not agree to it. The problem of indeterminacy persists in matters of distributive justice, and thus moderately idealized agents may choose other redistributive principles instead of the principle of social insurance. However, this concedes that convergence liberalism fails to achieve one of its original philosophical objectives, which is to ‘defend liberal democratic welfare-state capitalism’ (Vallier 2023: 58). The second escape route is to defend welfare-state policies by idealizing the constituency of public justification. Convergence liberals might refine the definition of moderately idealized agents by incorporating criteria that would ensure that the principle of social insurance would be eventually chosen. For example, these agents could be characterized by a more specific understanding of risk, one that attributes limited responsibility to individuals for avoiding risk. As a result, these agents would likely endorse the principle of social insurance, recognizing its role in supporting people regardless of any imprudence in their life choices. In this idealized setting, the principle of social insurance can be publicly justified among moderately idealized agents. This, nevertheless, may also not be satisfactory from the perspective of convergence liberals. For using a more substantial conception of a moderately idealized agent implies that convergence liberalism can only accommodate a less diversified range of perspectives in a free society. This consequently undermines one of the key advantages that convergence liberals frequently emphasize in their theory (Vallier and Muldoon 2021: 222). Certainly, convergence liberals might dismiss both escape routes and contend that, even if the principle of social insurance is challenged by certain intelligible reasons, it does not necessarily follow that the principle cannot be publicly justified. We shall then consider five potential responses that convergence liberals might offer in this scenario.

4. Five Possible Replies of Convergence Liberalism

Below, we shall examine five possible responses that convergence liberals would offer to defend the public justifiability of the principle of social insurance.

4.1. Higher-order social choice mechanism

Convergence liberals may concede that the principle of social insurance is objectionable from the perspective of some moderately idealized agents, but it is still publicly justified when it is chosen through a higher-order social choice mechanism. In other words, while moderately idealized agents disagree on what principle they should endorse, they may agree unanimously on the procedure to resolve such disagreement. The higher-order agreement on a particular procedure legitimizes the outcome chosen through this procedure. For example, according to Vallier, when a dispute occurs, moderately idealized agents should ‘appeal to democratic procedures to select among an eligible set of potentially publicly justified proposals’ (Vallier 2016: 259; cf. Vallier 2014: 99). The earlier Gaus also believes that the inconclusiveness of public justification should be solved by the institutions of liberal democracy that are publicly justified (Gaus 1996: 195–257). Given this view, convergence liberals could argue that, even though the principle of social insurance is objectionable to some agents, all agents could agree to resolve their disagreement by a one-person-one-vote democratic procedure. As long as the principle of social insurance can win over other proposals in the procedure, it is publicly justified among moderately idealized agents.

The problem with this solution is that, akin to the principle of social insurance, the social choice procedure might also be subject to intelligible disagreement. When considering various electoral systems such as plurality voting, runoff voting, and the Borda count, these mechanisms can produce varying outcomes despite being fed an identical set of voter preferences (Riker 1982: 22, 234–235). For example, suppose now moderately idealized agents have to choose among three principles: the principle of social insurance, Rawls’s difference principle and Cohen’s principle of community. In a plurality voting system, the principle of social insurance could win with a large minority of votes, say 40%, while the remaining 60% is split equally between the other two principles. However, if a runoff system were used and the majority of those who voted for Rawls’s difference principle preferred Cohen’s principle of community as their second choice, the Cohen’s principle of community might win the second round. Under the Borda count, if most voters rank Rawls’s difference principle consistently second, Rawls’s difference principle might accumulate the most points and win, despite never being the first choice of a majority.

This variation in outcomes poses a significant challenge in determining which system truly encapsulates the collective will, making it difficult to ascertain a universally representative voting method. Given that each of these voting systems can lead to different winners and policy decisions, moderately idealized agents might have different preferences for these mechanisms and have intelligible reasons to oppose alternative mechanisms. The imposition of a single social choice mechanism could, therefore, be seen as an arbitrary enforcement of a political rule, disregarding the intelligible objections of those it governs.

4.2. Evolutionary mechanism

Convergence liberals may use the evolutionary mechanism as another solution. This is the approach of the later Gaus, which is also sometimes mentioned by Vallier (Gaus 2011: 303–333; cf. Vallier 2019: 114–115, Vallier 2021: 235). Gaus admits that people’s evaluative standards are highly diverse, which makes it challenging for them to determine which norm they should ultimately follow (Gaus 2011: 36). To address this problem, Gaus proposes that people should compare each option with an option of ‘blameless liberty’, in which no social rules proposed by anyone could claim authority over anyone else (Gaus 2011: 322). If a member ranks a rule below blameless liberty, then it indicates that the member has no reason to endorse this rule and would rather live in a situation with no rules that could claim authority at all. By making this comparison, people will have an eligible set of rules, which include those rules on which people disagree which one is the best, but all agree that either of them is superior to no authoritative rules at all. Gaus then further narrows down this set with the Pareto criterion. If all people believe that a rule X in the eligible set is superior to another rule Y in this set, then Y should be excluded from the set. Eventually, Gaus argues, people are left with an optimal eligible set of rules: ‘no proposed rule in the set is ineligible in anyone’s ranking. Nor is it dominated by any other member of the set’ (Gaus 2011: 323).

Gaus then suggests that any rules within this optimal eligible set would be publicly justified, so long as it is selected through an evolutionary mechanism (Gaus 2011: 402). In the past, people might have lived together without any rules. They soon found this inconvenient and began to invent moral rules. Some of these rules might have been abandoned, while others were preserved. Through trial and error, society’s members may gradually come to coordinate on a set of rules. These rules are publicly justified because all people have conclusive intelligible reasons to endorse them. For if anyone were to reject these rules, the alternative would be an absence of rules – a scenario that no one prefers.

Gaus’s argument can be used to defend the principle of social insurance. Imagine there are three people, Audrey, Bruno and Chloe. Each has different preferences over principles of redistribution.

In this Table 1, Audrey, Bruno and Chloe disagree on which principle is the best. Nevertheless, Audrey and Bruno rank Cohen’s principle of community below blameless liberty, and Chloe ranks Dworkin’s principle of equality of ambition-sensitive endowments below blameless liberty. Accordingly, only the principle of social insurance and Rawls’s difference principle are included in the optimal eligible set. Suppose that, among these two, the principle of social insurance happens to be chosen in the evolution mechanism. Convergence liberals could then argue that although the principle of social insurance is not a principle that all three people have the strongest reason to endorse, it is nevertheless publicly justified because it is a principle that all people have conclusive reasons to support.

This solution, nevertheless, compromises the moral stability that convergence liberalism aims to achieve. The evolutionary mechanism may depend heavily on luck and immoral factors that undermine the ethical standing of the outcome. For instance, in the scenario mentioned earlier, Audrey and Bruno lead the two major political parties, while Chloe represents the third party. During an election, Bruno’s party is likely to win more seats in parliament. To prevent this outcome, Audrey visits Chloe and proposes that they form a coalition, which would give their alliance more seats than Bruno’s party. Despite her reservations about the principle of social insurance that Audrey advocates, Chloe recognizes the value in this opportunity to join the cabinet and enhance her political reputation. Interestingly, while Audrey is en route to visit Chloe, Bruno is also on his way, albeit slightly delayed due to a traffic accident. Had this accident not occurred, Bruno would have arrived earlier and successfully persuaded Chloe against forming a coalition. However, this was not the case. The Audrey-Chloe coalition was established, won the majority of seats in parliament, and ultimately implemented the principle of social insurance after the new cabinet was formed.

This example highlights how contingent evolutionary outcomes could be. In the story, a mere traffic accident proves to be the deciding factor in determining the result. If this accident had not occurred, Bruno would have been the victor, leading to the implementation of Rawls’s difference principle instead. To claim that the principle of social insurance is morally superior to Rawls’s difference principle seems unconvincing. The fact that Audrey, Bruno and Chloe ultimately agreed on social insurance is largely due to Audrey’s good fortune. Bruno and Chloe, despite having their own preferences, choose to reluctantly accept the result because it is at least better than a state of blameless liberty. Their decision to adhere to the status quo is primarily driven by pragmatic considerations. Additionally, a principle may be chosen in an evolutionary mechanism because of conspiracy, bribery, threat or manipulation (D’Agostino 2013: 147–148; Kogelmann 2017: 214–215). In other words, the major problem of using evolutionary mechanisms to determine the final outcome is that the process is predominantly non-moral. The result of evolution is always influenced by non-moral factors, such as luck and power. Unsurprisingly, even if the principle of social insurance is eventually chosen in the evolutionary mechanism, moderately idealized agents do not have conclusive moral reasons to endorse the evolutionary outcome over the long term. ⁸

Considering the impact of these non-moral factors, the evolutionary outcome is more like a modus vivendi among parties, rather than a publicly justified rule described by convergence liberals. A modus vivendi, as David McCabe describes, is ‘a compromise among citizens who recognize the value of ordered political life but realize that the political vision recommended by their distinct normative frameworks cannot be achieved’ (McCabe 2010: 133; cf. Horton 2010: 438, Westphal 2019: 3). In a modus vivendi, people’s compliance with laws depends on the prevailing balance of power. Should this balance shift, it is likely that people would support different laws or principles. Convergence liberals often highlight the difference between modus vivendi and publicly justified rules, for the former lacks the stability possessed by the latter. Gaus challenges modus vivendi because it fails to ‘provide a stable order’ and is merely ‘based on a shifting balance of power’ (Gaus 2003: 79). Vallier further argues that publicly justified rules can achieve ‘moral peace between persons’, more than just ‘a mere modus vivendi’ (Vallier 2019: 43). To Vallier, the key difference is that people have ‘moral reasons to adopt the [publicly justified] rules as one’s own’ (Vallier 2019: 108). Since they have conclusive reasons to endorse a publicly justified rule, they ‘internalize the rule [and] take it to be personally binding, which will often generate guilt and other moral emotions when she fails to meet the requirement’ (Vallier 2019: 108; cf. Vallier 2015: 210). Accordingly, for these people, the motivation to follow the rules is not driven by the fear of external consequences for non-compliance. Instead, they sincerely believe the rules as morally binding on their actions. In short, convergence liberals maintain that the outcomes of their models are not merely modus vivendi because people have conclusive moral reasons to endorse the evolutionary outcome, independent of changes in contingent external circumstances.

It remains unclear, however, why people would have conclusive moral reasons to endorse an evolutionary outcome, rather than replacing it with other options in the optimal eligible set. The central issue is that although the evolutionary mechanism enables people to coordinate on a principle, it does not significantly change the moral ranking of those principles in the practical reasoning of people, since this mechanism is primarily non-moral. For instance, in the previous example, Bruno and Chloe have conclusive reasons to endorse other principles other than the principle of social insurance before the election. After the election, this practical reasoning does not significantly change. While Bruno and Chloe will endorse the principle of social insurance after Audrey wins the election and implements this principle, the main reasons for them to embrace this principle have largely been pragmatic concerns: they are unlucky, and reverting to the state of blameless liberty would be even more unfavourable than maintaining the current status quo, thus they choose to endure the outcome.

In other words, Bruno and Chloe have conclusive reasons to prefer the outcome over the state of blameless liberty, but they do not have conclusive reasons to favour it over other preferable principles in the optimal eligible set. ⁹ Had the balance of power changed (say, Bruno and Chloe form a coalition in the next election), they would certainly abandon the principle of social insurance without having too much guilt. After all, the burden imposed by this principle on them is initially a consequence of their undesired misfortune. This example suggests that the evolutionary outcome, which is influenced by luck and sustained by the balance of power, is more akin to a modus vivendi because it fails to be conclusively justified to moderately idealized agents. Its stability heavily depends on contingent factors, and as a result, it does not attain the kind of stability that convergence liberals aim for.

4.3. Convergence of multiple mechanisms

Recently, Kogelmann has proposed another solution to the problem of indeterminacy. While Kogelmann agrees that people hardly agree on which social choice mechanism should be applied, they can still agree on a principle so long as this principle would be chosen no matter which mechanism is used. Kogelmann calls this the ‘convergence solution’. ‘If multiple social choice mechanisms can be applied to the optimal eligible set and still yield a singleton, then we have solved the problem’ (Kogelmann 2017: 215; cf. Vallier 2019: 187–188). Suppose, although Audrey, Bruno and Chloe prefer different social choice mechanisms, these social choice mechanisms would all select the principle of social insurance in the end. Given this convergence, governing people according to the principle of social insurance is publicly justified.

But how likely is it that different social choice mechanisms, when applied to the optimal eligible set, will choose the same principle? Kogelmann admits that normally the probability is low (2017: 222). For example, suppose there are three principles in a set (principle of social insurance, Rawls’s difference principle, Cohen’s principle of community), and thus there are six logically possible orderings (i.e. A>B>C, A>C>B, B>C>A, B>A>C, C>A>B, C>B>A). For every individual in the population, there is an equal chance of that individual having one of the logically possible preference orderings (i.e. each person has a 1/6 chance of having a particular preference ordering). In this case, the rate of convergence will be low, and people hardly agree on any principles. However, Kogelmann argues that if we introduce a certain degree of homogeneity in terms of preference distribution over the optimal eligible set, then the rate of convergence will increase dramatically. Here homogeneity means ‘the degree of consistency among voter preference’ (Kogelmann 2017: 225). For example, some people in society agree on a particular preference ordering. When homogeneity increases, ‘several major social choice mechanisms begin increasing in convergence as the level of homogeneity increases and … this convergence increases still as population size increases’ (2017: 226). According to Kogelmann, given that the population is large (i.e. the size of the population is 999) and 10% of the population agrees on a particular preference ordering, there will be almost zero deviation between most social choice mechanisms.

Kogelmann’s convergence solution sheds light on how convergence liberals could defend the principle of social insurance. Suppose that, in a society with a large population, 10% of the population (or even more) agree on the preference order ‘principle of social insurance > Rawls’s difference principle > Cohen’s principle of community’, whereas the preference orders of the other 90% of the population are random. Given this level of homogeneity, there is already a convergence on what redistributive principle should be selected. No matter what major social choice mechanisms one uses, moderately idealized agents would still select the principle of social insurance. Accordingly, despite the disagreement on which social choice mechanism is the best, moderately idealized agents would still regard a single principle as publicly justified.

The convergence solution, unfortunately, is not sufficient to rescue the principle of social insurance for convergence liberalism. Kogelmann’s solution assumes that, while 10% of the population has a homogeneous preference profile, the preference orderings of the other 90% of the population are randomly distributed. But what if the other 90% of the population’s preference orderings are not random? For example, suppose that more people join the homogeneous group, eventually causing it to make up 1/3 of the population. Meanwhile, another 1/3 of the population also unites and adopts a completely contrasting preference ordering, ‘Cohen’s principle of community > Rawls’s difference principle > principle of social insurance’. This type of culture, as defined by Hannu Nurmi, is a bipolar political culture. When the number of voters increases to 999 in such a political culture, there will hardly be any convergence of results when using different social choice mechanisms. The significant divergence of voting results remains: the voting results of any two voting systems differ from 15% to 59% in a bipolar political culture, given the different number of alternatives in the preference orders of the voters (Nurmi 1992: 471–472). It follows that having a 10% (or more) population with a homogeneous preference profile is insufficient to guarantee a convergence of results of different social choice mechanisms. To achieve convergence, additional conditions are necessary, such as the absence of an equally large opposing bloc of voters.

However, in the real world, it is unlikely to have only one group of voters sharing a homogeneous preference ordering, while most voters maintain their preferences in a random distribution. In other words, a unipolar or single-peaked political culture, even if it is possible, is hardly long-lasting in realpolitik. This is because it requires not only that 10% of total voters share a homogeneous preference order, but also that the remaining 90% of voters do not exhibit any non-random distribution of preferences. Essentially, the other 90% of voters must neither be able nor willing to form blocs to counter the dominance of the homogeneous preference order held by the 10%, even if they disagree with it (since, by hypothesis, their preference order differs from that of the 10% with the homogeneous preference order). With political parties playing a central role in the political and electoral organizations in virtually all of the real-world modern democracy, it is more likely that many among the other 90% of voters, after structured deliberation and negotiations, will revise their preference orders to form blocs around alternative, but developed and relatively stable, sets of policy preferences (White and Ypi 2011; Waldron 2016: Ch. 5). When 10% of the population unite and dominate in politics, many people in the remaining 90% will likely to align their preference orders to form alliances of the same size, if not larger.

This presents a significant challenge for convergence liberalism, as this weakens the durability of the principle of social insurance, which is a virtue that, Vallier argues, a stable liberal regime should possess. According to Vallier, a publicly justified rule should be ‘stable for the right reasons’, and this idea can be broken into two other concepts, immunity and durability (Vallier 2021: 38). Immunity means the ability of a system to resist external threats, such as the invasion by uncooperative and purely self-interested agents. Durability, on the other hand, means the ability of a system to withstand the internal dynamic and remain stable over the long term, even in the face of changing circumstances and challenges (Vallier 2019: 192–193). By this definition, the principle of social insurance lacks durability because the unipolar political culture is unlikely to be long-lasting. Initially, a homogeneous group may cause a convergence on the principle of social insurance among various social choice mechanisms. However, over time, other citizens will form groups with opposing preference orders, resulting in different social choice mechanisms yielding different principles of redistribution. Consequently, the convergence is disrupted, and the principle of social insurance is no longer the only choice that all citizens should support. In summary, the convergence on social insurance is fragile as it fails to have consistently strong support among citizens, especially during fluctuations in political opinions.

4.4. The value of social coordination

One might argue that, despite the principle of social insurance not being the optimal choice for some moderately idealized agents, they still have conclusive reasons to endorse it due to the value of social coordination. Recently, Marcus Schultz-Bergin (2021) suggested that, even if moderately idealized agents favour other political systems over the liberal scheme of basic rights, they would still endorse the liberal scheme because of the great value of social coordination. ¹⁰ The reason is that social coordination ‘creates a background of stability that enhances everyone’s ability to pursue their own life in their own way’ (Schultz-Bergin 2021: 66). Despite disagreements, it facilitates everyone to pursue one’s own life plans. Also, social coordination is morally valuable in itself, as it enables us to engage in activities we consider morally worthwhile (Schultz-Bergin 2021: 66). Given Schultz-Bergin’s argument, one might argue that just as people should endorse basic liberal rights due to the value of social coordination, people should also endorse the principle of social insurance due to the same value. ¹¹

We agree that the value of social coordination can explain why people should endorse basic liberal rights, such as religious freedom, but it fails to be a conclusive reason for people to endorse the principle of social insurance. A major difference is that basic liberal rights can enable people to pursue their own life plans, but choosing the principle of social insurance implies that people fail to implement other principles of distributive justice, such as Rawls’s difference principle and Cohen’s principle of community. Let’s reconsider the example of Bruno from Table 1. To Bruno, although the principle of social insurance is above the state of blameless liberty, it is ranked below Rawls’s difference principle. Bruno would hardly think that the principle of social insurance facilitates the pursuit of the difference principle. On the contrary, these two principles are incompatible with each other. While the principle of social insurance could allow for different levels of welfare, it still permits significant wealth disparities arising from family background differences (Rawls 1999: xv). Wealthy people can receive substantial inheritances from their kin, perpetuating social privileges across generations. From Bruno’s perspective, this clearly violates the difference principle, as it enables those wealthy individuals to benefit from the ‘natural lottery’ that is ‘arbitrary from a moral perspective’ (Rawls 1999: 63–64, 87). Furthermore, it is uncertain that embracing the principle of social insurance now will facilitate the realization of the difference principle in the future, as this presupposes a linear, smooth understanding of transition that is philosophically questionable (Barrett 2020). In brief, Bruno would likely doubt that the principle of social insurance enables him to realize the difference principle. Instead, he should view acceptance of social insurance as a tacit endorsement of injustice, which is a moral cost that the supporters of social insurance, such as Audrey, do not have to suffer.

Table 1. Audrey’s, Bruno’s and Chloe’s preferences over principles of redistribution. ⁷

We do not deny that, to Bruno, the principle of social insurance is still preferable to blameless liberty. However, given his preference for Rawls’s difference principle over the principle of social insurance, endorsing the principle of social insurance is to ask him to accept a principle that is less just within the optimal eligible set. Consequently, Bruno incurs a moral cost by endorsing the principle of social insurance, a cost he could avoid if a different evolutionary mechanism or social choice mechanism were in place. For example, revisiting the election scenario from section 4.2, had Bruno been able to ally with Chloe initially, Audrey would be the one compelled to adopt Rawls’s difference principle, facing a moral cost due to her perception of it being less just. But as it stands, Bruno and the advocates of other principles within the optimal eligible set must shoulder this moral cost. This highlights a critical flaw in convergence liberalism concerning distributive justice: while moderately idealized agents might agree on the principle of social insurance, in doing so, they forego other redistributive options and incur moral costs that not all have to share. This perceived unfairness gives some moderately idealized agents with strong reasons to seek the replacement of the principle of social insurance with other more preferrable principles in the optimal eligible set, thus disturbing the moral stability of the regime.

4.5. A ‘zoom out’ solution

Convergence liberals might argue that our critique assumes a specific order of public justification. As Lister observes, the result of public justification can vary significantly, depending on what order of public justification is chosen (Lister 2013: 89–90). In our critique, we assume a disaggregated order to public justification. The moderately idealized agents first reach an agreement on basic liberal rights such as the right to bodily integrity. Then these agents reach an agreement on private property rights. But we then argue that these agents would disagree on the principle of social insurance. However, convergence liberals might argue that an alternative aggregated order of public justification can be adopted. Suppose that the order ‘zooms out’ and the principle of social insurance is aggregated with the system of property rights to form a specific system of property distribution. In this scenario, moderately idealized agents are faced with a choice between two options: this ‘bundle’ of social insurance plus property rights or a state without both social insurance and property rights. All agents may then prefer this package over the latter state, and thus social insurance can be publicly justified. ¹² In short, by adjusting the order of public justification, convergence liberalism can potentially legitimize welfare policies within its framework.

However, this ‘zoom out’ solution is problematic for two reasons. First, the disaggregated order of public justification is not arbitrarily assumed. Rather, it is an order frequently embraced by convergence liberals themselves. As Gaus admits, the order of ‘bodily rights, property rights, social welfare’ is the ‘canonical liberal order of justification’ (Gaus 2010: 250–251). For a government that protects both bodily rights and property rights is more coercive than a government that protects only bodily rights, and a government that protects these two rights and imposes tax to support a welfare system is more coercive than the former two kinds of governments (Gaus 2010: 265). As coercion increases, individual liberty becomes more restricted, necessitating a higher level of public justification. Thus, Gaus argues that the order of justification should be arranged in a sequence of ‘bodily rights, property rights, social welfare’. People first resolve the more fundamental issues of bodily rights and property rights, and this settlement provides a background for further discussion on issues such as social insurance (Gaus 2010: 260–261). Similarly, Vallier follows a comparable sequence of public justification in his works, justifying the legitimacy of property rights (Vallier 2021: 119–138) before justifying the legitimacy of social insurance (Vallier 2021: 139–172). This order of public justification may be controversial. ¹³ Nevertheless, it is reasonable to critique convergence liberals based on an assumption they themselves employ.

Second, this ‘zoom out’ solution is ad hoc even when viewed from the perspective of convergence liberals. While Vallier admits that sometimes two issues should be considered together in public justification, he also emphasizes that we must avoid arbitrarily bundling separate issues. Accordingly, Vallier suggests that two issues can be aggregated in public justification when one is generally functionally dependent of the other. Vallier defines general functional independence as the following:

General Functional Independence: law x is functionally independent from law y if and only if (1) each member of the public believes that x will achieve at least one of her highly ranked goals [g1 … gn] with or without y based on (2) her model of the effects of x and y. (Vallier 2021: 157)

In the issues of aggregating social insurance and property rights, the law that provides social insurance is law x, and the law that protects property rights is law y. Convergence liberals might suggest that these two laws are functionally dependent according to Vallier’s definition. Most people would agree that, for instance, the goals to be achieved by the principle of social insurance can only be secured with a well-established legal system of private property that safeguards individuals’ income. Without such protection, people would lack certainty regarding the preservation of their insurance benefits. Thus, social insurance and private property could be plausibly bundled as a single policy package. Consequently, convergence liberals might argue that since no agents could intelligibly reject private property, the bundled principle of social insurance shall, in turn, be considered publicly justified.

However, the problem with this ‘zoom out’ solution is that the principle of social insurance is not the only principle that could be bundled together with private property. Many other redistributive principles could also be argued to functionally depend on private property. For example, moderately idealized agents who support Rawls’s difference principle might contend that the difference principle functionally depends on the existence of private property. For the goal of the difference principle is to ensure that ‘social and economic inequalities are to be arranged so that they are … to the greatest benefit of the least advantaged’ (Rawls 1999: 266). Without a system of private property, the least advantaged group might struggle to retain their resources after redistribution, making them susceptible to appropriation by more powerful parties. Furthermore, these Rawlsian supporters might argue that private property also functionally depends on Rawls’s difference principle. They argue for a system of private property because it safeguards each citizen’s just entitlements to their resources, and these just entitlements can only be achieved through the application of the difference principle. Once a just distribution is in place, people are entitled to what their talents and effort bring them (Rawls 1999: 74, 76; 2001: 72). Hence, according to these Rawlsian supporters, there is a stronger reason for bundling private property with the difference principle, instead of the principle of social insurance.

Our intention here is not to defend this Rawlsian bundle in the order of public justification. Rather, we merely argue that the condition of general functional dependence fails to determine an uncontroversial order of public justification among moderately idealized agents. Agents may hold divergent views on how the order of public justification should be arranged, and there are no sufficient grounds for agents to particularly favour the bundle between private property and social insurance. In sum, the ‘zoom out’ solution is objectionable for it contradicts the order of public justification that is often assumed by convergence liberals themselves, and it is also not the unique bundle that satisfies the condition of general functional dependence proposed by Vallier.

5. Conclusion

The welfare state has long been a popular model of state that is widely supported and implemented in many countries. As a result, convergence liberalism gains more moral appeal if it can demonstrate that, despite significant moral and political disagreements, a welfare state can still be publicly justified among people with diverse viewpoints. However, in this article, we critically examine the relationship between convergence liberalism and the welfare state. We argue that moderately idealized agents have intelligible reasons to reject the principle of social insurance, thus this principle fails to be conclusively justified according to the convergence conception of public justification. Thus, this casts doubt on the marriage between the welfare state and the convergence conception of public justification. Convergence liberals are confronted with a welfare-convergence dilemma: either they uphold the convergence conception and accept that the principle of social insurance could not be conclusively justified according to the convergence conception, or they defend the welfare state, give up the convergence conception, and further idealize the constituency of public justification to show that the welfare state is publicly justified. Both are painful choices for convergence liberals: selecting the former implies weakening the moral appeal of their project, but choosing the latter implies giving up the entire project itself.