Turnout is in decline in established democracies worldwide (Blais and Rubenson Reference Blais and Rubenson2013). Where, in the mid-1800s, 70–80 percent of eligible voters regularly participated in US Presidential elections, turnout has averaged just 53.7 percent since 1972. Average turnout in general elections in the UK has fallen from 76.6 percent during the period 1945–92, to 64.7 percent since 1997. And average turnout in Canadian federal elections has fallen from 74.5 percent during the period 1940–79, to 62.5 percent since 2000. For most democrats, these trends are cause for alarm. Lower turnout is associated with a range of ills, including increased inequality (Mueller and Stratmann Reference Mueller and Stratmann2003), and corruption (Birch Reference Birch2009b, 132–3). Compulsory voting is amongst the most effective means by which this decline might be arrested. When effectively enforced, compulsory voting increases turnout by 12–13 percent on average, and sometimes far more (Birch Reference Birch2009b, ch. 5). Australia, for example, introduced compulsory voting in 1924, and saw average turnout rise from 64.2 percent of registered voters to around 95 percent (Jackman Reference Jackman2001, 16316).
Compulsory voting, however, is controversial. Its opponents claim (1) that compulsory voting is coercive, (2) that coercion is morally wrong unless backed by some compelling justification and (3) that no such justification is available for compulsory voting. This article offers a two-part response to (3). First, I offer an argument from free-riding which, though gestured towards by others (e.g. Engelen Reference Engelen2007, 30–1; Lijphart Reference Lijphart1997, 11; Wertheimer Reference Wertheimer1975), has yet to receive sophisticated exposition and defence. Second, I argue that typical objections to compulsory voting rely upon unsupported empirical claims, mischaracterizations of compulsory voting, or implausible normative commitments. States, I conclude, may permissibly compel the vast majority of citizens to vote.
It is crucial to get clear immediately over the model of compulsory voting I wish to defend. Most importantly, I do not favour systems in which citizens are subject to coercible requirements to cast a valid ballot.Footnote 1 Enforcing such requirements would require some mechanism to check that citizens had voted correctly, undermining the secret ballot.Footnote 2 Instead, I favour systems in which (as is typical in the real world) citizens are required to either cast an absentee ballot, or attend a polling station on election day. Citizens, as such, remain free to abstain under compulsory voting. They can simply leave once their attendance has been registered, or submit a spoiled ballot.
I also favour regimes in which penalties for abstention are light, but effectively enforced. In Australia, for instance, abstention is (initially) punishable by a fine of $20 (Hill Reference Hill2014, 115). Severe punishments are both disproportionate, and probably unnecessary. While evidence indicates that effective enforcement is necessary for compulsory voting to impact turnout, most effective regimes do not impose heavy penalties for abstention (Birch Reference Birch2009b, 8–11, 89–95).
Further, I wish to defend compulsory voting only in the context of referenda and elections in reasonably just, established liberal democracies.Footnote 3 Any justification for a policy of compulsory voting presupposes that the state implementing that policy has the general, legitimate authority to subject its citizens to appropriate coercion. States who routinely subject large numbers of citizens to serious injustices (such as states that violate human rights on a large scale) are not plausibly legitimate in this sense. I therefore restrict the scope of my argument to reasonably just states – states who do not generally inflict such injustices upon their citizens.Footnote 4 I restrict the argument to established, liberal democracies because other contexts, though equally important, give rise to a range of complicating considerations. In some developing nations, for example, citizens who choose to vote are subject to threats of serious harm, a consideration which obviously militates against compulsory voting.
Finally, I will assume that elections are well-structured in the following three ways. First, there must be no excessive barriers to accessing the franchise. Citizens must not, for instance, be required to produce forms of identification they are unlikely to possess in order to vote. Second, voting must be low-cost.Footnote 5 Citizens must not, for instance, be required to queue for many hours or travel great distances to vote. Third, elections must be genuine, insofar as they must not be subject to undue manipulation. Compulsory voting would obviously be morally perverse were voting unreasonably difficult, excessively costly or irrelevant to political outcomes.
These clarifications in hand, we can proceed to the arguments. Though it has received very limited positive defence, the free-riding argument has been the subject of extensive criticism. My strategy, then, will be somewhat defensive. I begin by outlining an intuitively appealing, initially plausible version of the argument, together with the key considerations upon which it is reliant. The second section then refines the argument in dialogue with the most important objections raised in the literature. For reasons that will become clear, the free-riding argument depends upon the claim that compulsory voting does not entail unacceptable moral costs. The third section, then, responds to a range of arguments purporting to show that compulsory voting entails such costs. A brief conclusion follows.
Abstention as free-riding
Free-riding consists in the consumption of collectively produced public goods without making appropriate contribution to the system by which those goods are produced. Public goods are those which exhibit jointness in supply and jointness in consumption (Cullity Reference Cullity2008, 9). A good exhibits jointness in supply if supplying the good to one member of a group means supplying it to all members of that group. Providing clean air to some citizens in a particular area, for example, entails providing clean air to all citizens in that area. A good exhibits jointness in consumption where one group member’s consumption of that good does not undermine others’ consumption of that good. Buses with plenty of empty seats, for example, exhibit jointness in consumption. My travelling on the bus does not diminish the benefits others derive from bus travel.
Free-riding is objectionable because, and where, it is unfair. The unfairness in question consists in persons affording themselves a kind of objectionable privilege: arrogating certain preferential advantages to themselves, while depending upon others declining to do so (Cullity Reference Cullity1995, 22–32). Fare-evaders on public transport, for example, afford themselves the privilege of free travel, while depending upon others’ failure to exercise that same privilege. The benefits of public transport would not exist at all if no one paid the fare. Free-riding, moreover, can be unfair even where it has no non-trivial impact upon others’ interests. Suppose Eloise fare-evades on a public bus with plenty of empty seats. Eloise surely treats her fellow passengers unfairly, though her actions have no non-trivial impact upon their interests. They, after all, would be no better off had she paid her fare to the conductor. Nor do her actions plausibly affect the wider community. A bus fare is much too small an amount to make any difference to the government’s capacity to intervene upon citizens’ wellbeing, for instance.
The central claim of this article is that non-voters unfairly free-ride upon voters.Footnote 6 Demonstrating this requires us to show that voting produces some public good, the benefits of which are enjoyed by non-voters who fail to make appropriate contribution to its production.
Existing statements of the free-riding argument are almost never explicit about what this public good is supposed to be. Wertheimer (Reference Wertheimer1975, 279–82) is an important exception. He claims that competitive elections are a public good, and that ‘[n]on-voters […] free-ride upon voters’ maintenance of the electoral system’. This argument is too coarse-grained. It may well be that non-voters benefit in some sense from voters’ efforts to maintain the electoral system. However, for non-voters to count as free-riders, they must surely benefit from others’ efforts all-things-considered. John’s failure to contribute to some system which confers a $50 benefit upon him while imposing costs of $500 surely does not amount to unfair free-riding. And, as Hill (Reference Hill2014, 192) points out, many non-voters are plausibly left worse off by others’ political activity. The over-participation of some groups relative to others plausibly causes politicians to over-allocate resources to the former, and under-allocate resources to the latter. For many non-voters who belong to groups which under-participate, the costs of others’ participation will plausibly outweigh the benefits, undermining such persons’ duties.
To which public good, then, ought we appeal? Given the above concerns, the most obvious alternative might perhaps be the just advancement of interests, overall. Unfortunately, this is a non-starter. No real-world democracy ever produces the just advancement of interests, overall. Injustices of some kind are to be found in all societies. Non-voters cannot free-ride upon the production of goods which are not actually being produced. Moreover, as just noted, political participation by some groups can lead governments to unjustly over-allocate resources to those groups, at others’ expense. Such groups’ participation fails to produce the just advancement of interests, overall. Rather, it produces the unjust advancement of some groups’ interests over others. We can hardly see non-voting members of these injustice-promoting groups, then, as free-riding upon the production of the just advancement of interests, overall.Footnote 7
I suggest, instead, that we appeal to the claim that political participation by members of particular social groups produces governmental responsiveness to the legitimate interests of those groups. That is to say, voting by members of particular social groups produces a disposition on the part of elected officials to protect and promote the legitimate interests of that group in policymaking, in an attempt to win their electoral support.Footnote 8
This good is crucially important from a moral point of view. By legitimate interests, I simply mean interests that individuals have claims that the political community protect and promote (autonomy, dignity, security, for example).Footnote 9 It plainly matters, morally, that such claims be satisfied. The satisfaction of those claims frequently depends upon government action, itself a product of governments’ political incentives to respond to the legitimate interests of the various social groups to which individuals belong. It is also, in the relevant sense, a public good, insofar as it exhibits both jointness in supply and jointness in consumption. The benefits of responsiveness to the legitimate interests of particular social groups accrue to both voting and non-voting members of those groups. One group member’s enjoyment of the benefits of responsiveness to that group’s interests does not undermine others’ enjoyment of those benefits. Non-voters, thus, free-ride upon voters who belong to the same social group(s) as themselves, enjoying the benefits of responsiveness while failing to contribute, by voting, to its production. Appealing to responsiveness to the legitimate interests of particular social groups also avoids the problem faced by Wertheimer’s account. The costs of others’ political participation overall may well outweigh the benefits for many non-voters. That, however, is perfectly consistent with non-voters benefiting, all-things-considered, from political participation by members of the social group(s) to which they belong.Footnote 10 Such non-voters would plausibly be even worse off, were it not for the fact that participation by co-members of the social group(s) to which they belong causes governments to be responsive to their legitimate interests to at least some degree.
This approach also enjoys strong empirical support. There is a substantial body of evidence suggesting that political participation by particular social groups gives politicians incentives to allocate resources to those groups, to wit, to protect and promote their interests (including, much of the time, their legitimate interests). To select just a few examples, Aidt and Dallal (Reference Aidt and Dallal2008) show that the enfranchisement of women in Western Europe between 1869 and 1960 brought about increased spending (0.6–1.2 percent of GDP in the short term, with more substantial long-term increases) on social services upon which women were more likely to be reliant. Abou-Chadi and Orlowski (Reference Abou-Chadi and Orlowski2015) show that the abolition of property and income requirements on the franchise in Western Europe between 1880 and 1938, under certain conditions, resulted in increased spending on social services and public goods upon which the worse-off were more likely to be reliant. Husted and Kenny (Reference Husted and Kenny1997, 56–7, 76) show that the abolition of poll taxes and literacy tests in the Old South between 1956 and 1968 brought about enormous increases in African American participation and, subsequently, substantially increased spending on welfare programmes upon which African Americans were more likely to be reliant. Martin (Reference Martin2003, 116–23), finally, demonstrates that counties in the US with higher rates of participation, on average, receive substantially more federal grant expenditure per capita than counties with lower rates of participation, even controlling for factors like unemployment, crime rates and electoral competitiveness.
It is also worth emphasizing an important aspect of this approach. The public good to which my argument appeals is responsiveness to social groups’ legitimate interests, rather than responsiveness to their interests simpliciter. Individuals do not have duties of fairness to contribute to the production of benefits to which they are not entitled, and to which others have legitimate claims. Call these unjust benefits. Suppose that some gang steals cash from my neighbour’s safe. To apologize for any disturbance, they leave $500 on my doorstep. We should hardly wish to conclude that I have duties of fairness to contribute to the gang’s operations. Rather, I should return the $500 to my neighbour, whereupon I will no longer be a beneficiary of the gang’s activities (Pasternak Reference Pasternak2016). Social groups (especially privileged groups) often have interests in perpetuating injustices. The wealthy, for example, have interests in perpetuating inegalitarian distributions of social advantage. Such injustices, moreover, are often produced and sustained by governments’ responsiveness to the interests of those groups. Responsiveness to interests simpliciter, then, is an unjust benefit in the case of many social groups to which individuals do not have duties of fairness to contribute. Appealing only to responsiveness to groups’ legitimate interests – those interests which individuals have claims that the political community protect and promote – avoids this difficulty.
This, of course, does nothing to diminish the fact that political participation by certain groups really does promote injustice, however the free-riding argument is framed. Does that fact somehow tell against a duty of fairness to vote on the part of members of such groups?
There are two ways in which this concern might be spelled out. First, it might be argued that the injustice promoted by certain social groups’ political participation undermines the considerations upon which the purported duty of fairness is premised. This objection fails. Participation by group X might promote injustice towards group Y. It might even diminish governmental responsiveness to Y’s legitimate interests. But that in no way alters the facts upon which the duty is premised: voters from X produce responsiveness to X’s legitimate interests. Non-voting members of X fail to contribute to its production.
One might argue, instead, that such injustices generate duties on the part of members of injustice-promoting groups not to vote which outweigh the duty of fairness to vote. The simplest version of this objection would claim that, since the collective voting of group X promotes injustice, each member of X promotes injustice by voting. Each member of X, therefore, has a duty to abstain outweighing any duty of fairness to vote. This is unsatisfactory, however. In general, the claim that any individual member of some group is promoting injustice by ϕing does not follow automatically from the claim that that group’s collective ϕing promotes injustice. The amount of injustice the group’s collective ϕing causes may be unaffected by any individual’s decision as to whether to ϕ.Footnote 11 What’s needed is some more sophisticated explanation as to why individuals have duties not to vote in virtue of the injustice promoted by their group’s collective voting.
Three kinds of explanation are standardly appealed to in such cases. First, individuals’ participation in some collectively harmful activity sometimes makes that activity worse; that is to say, causes some additional actual harm. Individual voters qua individuals, however, do not cause actual harm. No voter is ever decisive over the outcome of any real-world election (Brennan and Lomasky Reference Brennan and Lomasky1993, ch. 4). Nor do individual voters make any non-trivial difference to politicians’ decisions as to how social resources are to be allocated between the various social groups. Incentives to attract support from a voting bloc of 10,242 persons in an electorate of 200,000 are no more or less powerful than incentives to attract support from a voting bloc of 10,243. Second, participation is sometimes wrong in virtue of its expected disvalue. Yet individual decisions to vote do not typically have any non-trivial expected disvalue, given the miniscule chances of individuals being decisive, the centripetal tendencies of electoral competition and the pervasive uncertainty of electoral choices (Brennan and Lomasky Reference Brennan and Lomasky2000, 65–74).Footnote 12 Finally, participation can sometimes amount to free-riding. Voters, on this view, would be free-riding upon non-voters. This gets things backwards. Free-riding implies the receipt of benefits. But, as we have argued, it is non-voters who are benefited by voters in virtue of their producing the good of governmental responsiveness, not the other way around. It is obscure, then, what the basis of this countervailing duty might be.
Moreover, even if there were such a duty, its force would be substantially undermined under compulsory voting, rendering it much less likely to outweigh the duty of fairness to vote. This is because compulsory voting undermines one of the central causal mechanisms by which such injustices are produced. This is the central claim of proponents of the egalitarian argument for compulsory voting (Birch Reference Birch2009a, 23–4; Engelen Reference Engelen2007, 25; Hill Reference Hill2014, ch. 6; Lijphart Reference Lijphart1997). It is well-accepted that voluntary voting, where large numbers of individuals abstain, leads to socio-economic bias in turnout. Disadvantaged social groups substantially under-participate, relative to other groups (Hill Reference Hill2014, 132). This, proponents of the egalitarian argument claim, contributes to elite bias in policymaking – to the unjust over-allocation of resources to more privileged social groups. Compelling citizens to vote substantially mitigates socio-economic bias in turnout (Singh Reference Singh2015). We should expect compulsory voting, then, to partially mitigate elite bias in policymaking, leading to more egalitarian, less unjust outcomes.
A growing body of empirical evidence lends support to these claims. In a cross-national study of forty-one democracies in 1991, O’Toole and Stroble (Reference O’Toole and Stroble1995) found that compulsory voting is associated with higher levels of government spending on services upon which the worse-off are more typically reliant. In a cross-national study of ninety-one countries over the period 1960–2000, Chong and Olivera (Reference Chong and Olivera2008) show that compulsory voting – when effectively enforced – leads to Gini coefficients 3.7 points lower on average. Birch (Reference Birch2009b, 131) found similar results in cross-national analyses of Western Europe and Latin America. In a study of direct referenda in Swiss cantons over the period 1908–70, Bechtel, Hangartner, and Schmid (Reference Bechtel, Hangartner and Schmid2016, 764) found that the vote share for redistributive policy measures was 11 percent higher on average in cantons that enforced compulsory voting. Carey and Horiuchi (Reference Carey and Horiuchi2017, 125–35) found that Venezuela’s abolition of compulsory voting in 1993 sharply depressed turnout (from 80 percent in 1988 to 54 percent in 1993), and led to increased Gini coefficients. Finally, Fowler (Reference Fowler2013, 173–7) found that the introduction of compulsory voting in Australia in 1924 led to increased federal spending on pensions, a policy more strongly favoured by working-class voters at that time.Footnote 13 In my view, then, we can safely set this concern aside.Footnote 14
We need only make a final clarification before stating the initial version of the argument. It is widely accepted that free-riders act unfairly only where they have the capacity to contribute. The severely disabled, for example, often consume public goods (public healthcare, for example) without contributing to their production. That, however, is surely not unfair. Some persons fail to vote out of a lack of capacity (though the vast majority do not). The free-riding argument cannot justify compelling such persons to vote. And, indeed, any reasonable regime of compulsory voting ought to make exceptions for such persons. It is for this reason that the argument, below, refers only to compelling ‘sufficiently capable persons’ to vote. Here, then, is the initial statement of the free-riding argument.
(F1) It is unfair for sufficiently capable persons to benefit from voting by members of the social group(s) to which they belong without contributing, by voting, to the political process by which those benefits are produced.
(F2) It is permissible for the state to subject persons to compulsion where doing so would substantially prevent such unfairness.
(F3) Compelling sufficiently capable persons who benefit from voting by members of the social group(s) to which they belong to vote would deter abstention, substantially preventing such unfairness.
(FC) Therefore, it is permissible for the state to compel sufficiently capable persons who benefit from voting by members of the social group(s) to which they belong to vote.
We have now laid out the central considerations upon which the free-riding argument depends. In the following section, we shall revise numerous aspects of the argument in dialogue with objections. Before doing so, however, we must address a final outstanding issue. Notice that a duty of fairness to vote of the kind for which I have argued, above, cannot be unconditional. Citizens have such duties only if they belong to at least one social group whose legitimate interests are advanced by its members’ political participation. Given that each of us belongs to many social groups – ethnic, socio-economic, geographic and so on – virtually all citizens will have such duties. However, it is at least possible that some will not. If no one from any of the groups to which Juliette belongs chooses to vote, for example, the considerations to which I have appealed obviously cannot support a duty of fairness for her to vote.
Is this a problem? Not a serious one. Such persons will be very few in number and might, perhaps, be legitimately compelled to vote on other grounds – by appeal to the egalitarian argument, for instance. If not, such persons might reasonably be exempted from the requirements of compulsory voting. The state might, for example, allow persons to appeal penalties for abstention where they reasonably believe that others’ political participation does not advance the legitimate interests of any of the social groups to which they belong. No institutional regime is perfect. It is possible that some citizens may unjustly be compelled to vote under such arrangements. That injustice, however, is overwhelmingly likely to be outweighed, all-things-considered, by the good compulsory voting is likely to promote.
Refining the free-riding argument
Two principal lines of objection have been raised against the free-riding argument. First, that duties not to free-ride are not enforceable – that is, that F2 is false. Second, that abstention is not unfair – that is, that F1 is false. Let us consider each in turn.
Is the duty enforceable?
Jason Brennan (Reference Brennan2014, 76) holds that the free-riding argument fails because duties of fairness generally are not coercively enforceable. That is much too strong. Few would deny that fare-evaders and tax-evaders act unfairly, and that the state may justifiably punish such persons in virtue of the unfairness of their conduct. That said, it is true that not all duties of fairness are coercively enforceable. Suppose Edith and Fran live in the same block of flats. Each owns a cat that is severely allergic to a patch of weeds that has sprung up on the nature strip in front of the block. Edith spends several hours one morning digging out the weeds. Unfortunately, however, the weeds will grow back in two months’ time. Fran, intuitively, would have a duty of fairness to Edith to dig the weeds out once they reappear, given that she has benefited from Edith’s doing so the first time around. However, it would obviously be inappropriate for the state to force Fran to dig out the weeds. This raises the question: are duties of fairness to vote enforceable?
We can make progress, here, by considering what distinguishes our two paradigm cases: Fran, whose duty cannot be enforced, and fare-evaders, whose duties can be enforced. The most obvious difference is that the enforcement of Fran’s duty to Edith would carry unacceptable moral costs. Any plausible enforcement mechanism would be highly intrusive. Moreover, individuals’ relationships (plausibly) will generally go better if they are required to resolve minor disputes of this kind interpersonally. Such considerations obviously do not arise when compelling citizens to pay their fare on public transport. This suggests that duties of fairness may only be enforced where doing so would not entail unacceptable mora costs. Compulsory voting does not plausibly jeopardize our privacy or interpersonal relationships. Opponents of compulsory voting, however, have claimed that compulsory voting does entail a range of other costs. I argue in the following section that none of these claims are plausible.
In the meantime, however, the question of costs does not settle the matter. Brennan (Reference Brennan2014, 77) does allow that duties of fairness may be enforced if, and only if, individuals consent to the benefits in virtue of which the duty arises. This suggests an important disanalogy between non-voters and fare-evaders. Fare-evaders consent to the benefits of public transport. Non-voters, however, do not generally consent to the benefits of others’ political participation. Brennan does not explain why this is so. The most obvious approach, however, would draw upon Simmons’ (Reference Simmons1979, 79–100) well-known critique of consent-based accounts of political authority. Genuine voluntary consent, Simmons claims, requires reasonable alternatives. If Elizabeth ‘agrees’ to give her purse to a gunman who threatens to kill her if she does not, she cannot be said to have consented to give her money away, given the unacceptability of the alternative. Similarly, citizens cannot generally be said to have given consent to the benefits the state provides, given that the only way they can avoid receiving those benefits is by the unreasonably costly act of emigration. One might, similarly, hold that non-voters do not consent to the benefits of others’ political participation since, presumably, emigration is also the only way in which one can avoid receiving those benefits.
However, I reject the idea that duties of fairness are enforceable only where individuals have given consent. Suppose Jack has recently been in an accident. The ‘meals on wheels’ service operated by his local council is his only means of getting food. Residents do not need to opt in to the service. Meals are sent automatically to residents known to be unwell. The council charges $10 a week for the service, which Jack can easily afford. Non-payment is punishable by a fine of $20. Jack takes the meals but refuses to pay. Jack has no reasonable alternative to accepting the meals. He would starve, otherwise. He does not, therefore, strictly consent. Fining Jack, however, seems totally unproblematic. Intuitively, his willing acceptance of the benefits of the service is sufficient to render him liable to proportionate coercion if he refuses to pay.Footnote 15
The intuitive core of willing acceptance seems well captured by the following: for a beneficiary to have accepted some benefit willingly, they must not have (a) attempted to reject that benefit, or (b) disgorged that benefit, or (c) disgorged something of equivalent value, or (d) preferred that the benefit not be conferred upon them.
I think that the vast majority of non-voters can be said to willingly accept the benefits of voting by members of the social group(s) to which they belong. Citizens routinely accept the benefits of government-provided public goods, insofar as they rely upon those goods (public order, roads, transfer payments, and the like) in daily life. The rate at which governments provide those goods to different social groups is a function of governments’ political incentives. These, in turn, are partially a function of the rate at which different social groups participate in the political process. In accepting the benefits of government-provided public goods, then, non-voters accept – that is, fail to reject – the benefits of political participation by members of the group(s) to which they belong. Very few non-voters take steps to either disgorge these benefits, or objects of equivalent value. And it would be simply bizarre if non-voters preferred en masse that these benefits were not conferred upon them at all. Very few, presumably, would prefer to belong to social groups to whom governments were less responsive.Footnote 16
In the absence of some further argument, then, I conclude that duties of fairness to vote are coercively enforceable. This discussion, however, has proven fruitful. It points to several revisions that need to be made to the argument. We must replace F2, F3 and FC with F2*, F3* and FC*, and add F4, as follows.
(F2*) Where persons have duties of fairness in virtue of their willing acceptance of benefits, it is permissible for the state to subject those persons to compulsion where doing so would substantially prevent those persons from violating those duties of fairness, provided there are no unacceptable moral costs associated with doing so.
(F3*) Compelling sufficiently capable persons who willingly accept the benefits of voting by members of the social group(s) to which they belong to vote would substantially prevent such persons from violating their duties of fairness.
(F4) There are no unacceptable moral costs associated with compelling sufficiently capable persons who willingly accept the benefits of voting by members of the social group(s) to which they belong to vote.
(FC*) Therefore, it is permissible for the state to compel sufficiently capable persons who willingly accept the benefits of voting by members of the social group(s) to which they belong to vote.
This narrows the scope of the argument. Some citizens – radical anarchists, perhaps – do not willingly accept the benefits of others’ political participation. The free-riding argument, as stated, cannot justify compelling such persons to vote.Footnote 17 This need not worry us. For the reasons outlined above, such persons are few and far between. They might potentially be permissibly compelled to vote by appeal to egalitarian considerations. And, if not, such persons might simply be exempted from the requirement to vote.Footnote 18
Is abstention unfair?
A more foundational line of objection challenges the claim that abstention is unfair. Three arguments have been advanced to this end.
First, Geoffrey Brennan and Loren Lomasky (2000, 75–9) and Christopher Wellman (Reference Wellman2005, 59–60) have argued that abstention is not unfair because it does not impose costs upon others. On the contrary, abstention benefits citizens who choose to vote. Brennan and Lomasky point out that shrinking the total number of voters increases each voter’s probability of decisiveness and, in turn, the expected value of voting. Wellman argues, similarly, that shrinking the total number of voters increases the amount of influence each voter can be said to exercise over the outcome. This objection can be quickly dismissed. Both versions implicitly presuppose that unfair free-riding necessarily involves imposing costs upon others (or, at least, failing to benefit others). Yet, as pointed out earlier, free-riding does not necessarily involve the imposition of costs upon, or failures to benefit, others. An individual's decision to fare-evade, for example, generally has no non-trivial impact upon anyone else’s interests. Nevertheless, it is obviously unfair.Footnote 19
Second, Jason Brennan (Reference Brennan2014, 75–82) appeals to Robert Nozick’s (Reference Nozick1974, 90–6) argument against the claim that the receipt of benefits is sufficient to ground duties to contribute to the system by which they are produced. If Anthony’s neighbours set up a public-address system that plays music he occasionally enjoys, for example, he is surely not thereby obliged to take on costs to help maintain the system. The mere fact that non-voters benefit from others’ participation, then, does not show that such persons have duties to vote.
All Nozick’s argument shows, however, is that persons who benefit from public goods do not necessarily have duties to contribute to the production of those goods. It does not show that beneficiaries never have such duties. Jack’s acceptance of the benefits of the meals on wheels service, above, clearly gives rise to a duty to contribute. Similar cases abound. If one’s neighbours were digging a well to provide safe drinking water to the street where none was otherwise available, or else eliminating wild animals that threaten human life, for example, there is a strong intuition that fairness would require one to contribute to those efforts.
As Cullity (Reference Cullity2008, 13–14) and Klosko (Reference Klosko1987, 246–53), among others, point out, what appears to make the intuitive difference in these cases is the significance of the benefits in question. The ability to listen to music chosen by one’s neighbours is trivial.Footnote 20 Food, water and personal safety are decidedly non-trivial. Nor, importantly, are the benefits of others’ political participation trivial. Consider the evidence above linking participation rates to the provision of important goods like social services and welfare. Responding to this criticism, then, simply requires us to make a slight modification to the argument, substituting F1 for F1*, as follows.
(F1*) It is unfair for sufficiently capable persons to benefit non-trivially from voting by members of the social group(s) to which they belong without contributing, by voting, to the political process by which those benefits are produced.
Finally, Jason Brennan (Reference Brennan2014, 79–82) also argues that taking the free-riding argument seriously leads to an unacceptable proliferation of moral duties. He begins by distinguishing two views of the moral requirement not to free-ride (Brennan Reference Brennan2014, 79).
(1) If you receive a good of type P, then you must provide goods of type P back in return.
(2) If you receive a good of type P, then you must provide sufficiently valuable goods of any type back in return.
For the free-riding argument to go through, we must be required to reciprocate for the benefits of voting by voting, ourselves. Under (2), citizens would be able to make up for failures to vote in other ways – paying taxes, raising children and so on. Moreover, Brennan (Reference Brennan2014, 81) claims, since almost all of us make some net positive contribution to society over our lifetimes, non-voters, under (2), would almost never amount to free-riders. The free-riding argument, then, must rely upon (1). (1), however, entails implausible conclusions. Each of us is constantly in receipt of a vast range of goods produced by others, given the division of labour characteristic of modern societies. Under (1), then, citizens would be subject to a plethora of requirements to provide a vast range of goods to a vast range of people. This radically misdescribes our moral situation. We simply are not subject to such duties. We ought, then, to reject (1) in favour of (2). As we have already pointed out, however, (2) undermines the claim that non-voters are free-riders.
Brennan, however, presents us with a false choice. Both (1) and (2) should be rejected. They are both too broad, making unrestricted claims as to what individuals are required to do where they receive goods of any kind. As a conceptual matter, however, free-riding occurs only where individuals benefit in a very particular way – that is, from others’ production of public goods. As we have just seen, duties to contribute to the production of public goods arise only where we benefit non-trivially from those goods. And our duties not to free-ride do not necessarily require us to provide goods to others, but rather to make appropriate contributions to systems for the production of public goods from which we benefit. The real choice, then, is between the following.
(1′) If you benefit non-trivially from some public good P, then you must make appropriate contribution to the system by which P is produced.
(2′) If you benefit non-trivially from some public good P, then you must make appropriate contribution to some system by which some public good is produced.
Insofar as the free-riding argument relies upon the claim that citizens who benefit from others’ contributions to the political process have duties to contribute to the political process themselves, the free-riding argument must depend upon (1′). Whether we should accept (1′) or (2′) is a matter of which best captures our intuitions. In this respect, (1′) has a clear advantage. We do not generally think that persons who have made even very substantial contributions to some system for the production of public goods are thereby ‘off the hook’ with respect to others. Georgina’s failure to buy a ticket on a public bus seems no more or less unfair if she is an average citizen or a self-made millionaire who employs hundreds of people and pays a great deal in tax. Either way, Georgina benefits from the public transport system, and ought to make appropriate contribution to that system.
Opponents are likely to protest that (1′) is vulnerable to its own proliferation objection. Notice, however, that (1′) is far narrower than (1). (1′) only requires persons to contribute to the production of public goods from which they benefit non-trivially. Many public goods provide us with trivial benefits, only. My neighbour’s well-kept garden might be a public good, insofar as anyone can take pleasure in looking at it. But, given the triviality of this benefit, (1′) gives rise to no requirement to contribute to its production.Footnote 21 Many of the other goods from which we benefit are not public goods, at all. Most goods traded on the economic market, for example, exhibit neither jointness in consumption, nor jointness in supply.
Of course, there are public goods from which we derive non-trivial benefits – public order, public education, public health, and so on – to which we are obliged to contribute under (1′). I deny, however, that this is problematic. After all, virtually all of us already contribute to very many different public goods simply by paying tax. Producing such goods is amongst the most important functions of the state, given the famously difficult co-ordination problems which attend private efforts to do so (Olson, Reference Olson1965). Governments, then, act as a ‘middle-man’, collecting contributions via taxation and distributing them to the systems by which those goods are produced. Some significant public goods, of course, are provided by non-government organizations – such as charities and churches. But in these cases, citizens generally contribute in virtue of the fact that such organizations are typically untaxed. Citizens are required to pay higher rates of taxation in virtue of the forgone revenue – effectively subsidizing such operations.
Of course, there are still cases wherein citizens derive non-trivial benefits from public goods produced by others in ways that are in no way state-supported. In such cases, however, the verdict delivered by (1′) seems intuitively correct. We should accept that citizens do, indeed, have duties to contribute to such goods. Georgina simply does seem to have a duty of fairness to pay the bus fare, no matter her contributions in other domains. Alternatively, consider again the case in which one’s neighbours are digging a well to provide safe drinking water. One simply does seem to have duties to contribute to their efforts, no matter the magnitude of one’s other contributions.Footnote 22 (1′), perhaps, requires many of us to contribute more, and in more ways, than we do at present.Footnote 23 We should embrace that conclusion. Many of us do free-ride upon others. We are rightly criticizable as such.
Brennan’s critique, however, does contain a grain of truth. (1′) requires citizens to make appropriate contributions to systems for the production of public goods from which they benefit. Yet appropriate contributions are not necessarily identical contributions. In the well-digging case, I would hardly treat my neighbours unfairly if I made the bricks for the top of the well while they dug the hole, for example. In the political case, citizens have duties of fairness to contribute to the political process by which responsiveness to their legitimate interests is produced. But citizens might discharge such duties by means other than voting. If Irene fails to vote because she is too busy ferrying dozens of members of her social group who would otherwise have abstained to the polls, for example, she can hardly be accused of free-riding.
This is an important observation. To this point, I have argued for a duty of fairness to vote. As Irene’s case makes clear, however, the unfairness of free-riding can only ground a weaker, disjunctive duty: a duty to vote, or make some equivalent contribution to the political process by which responsiveness to the legitimate interests of the social group(s) to which one belongs is produced.
Is this a problem? Again, not a serious one. To see why, we must distinguish between persons who fail to make voting-equivalent contributions, and persons who do make such contributions. The former group is by far the larger. Relatively few persons engage in political activity aside from voting. The International Social Survey Programme’s 2014 Citizenship study of thirty-four developed and developing nations, for example, found that just 3.8 percent of people overall belong to a political party and actively participate (ISSP 2016, 55). The American National Election Study found that, in 2012, just 12 percent of persons donated to a political campaign, 6 percent attended a political meeting and only 3 percent worked for a party or candidate (American National Election Studies 2012, sn. 6). The free-riding argument unproblematically justifies subjecting the vast majority of persons, then, to the requirements of compulsory voting. After all, if such persons failed to vote, they would fail to discharge even this weaker, disjunctive duty.
Compelling persons who do make voting-equivalent contributions to vote cannot be justified by appeal to the free-riding argument. Such persons discharge their duties of fairness in other ways. This is hardly a difficulty, however. As we have pointed out, this group of persons is relatively small. Subjecting such persons to the requirements of compulsory voting might be justified on other grounds – such as by appeal to the egalitarian argument. If such justifications fail, exemptions might be introduced for such persons. The state might, for instance, permit persons who fail to vote to appeal any penalty by producing evidence of having made some equivalent contribution (evidence of being active in a political party, for instance). Importantly, however, very few such persons are likely to avail themselves of such exemptions. Persons who engage in political activity of other kinds presumably have strong preferences to engage in political activity generally. They are, as such, very likely to elect to vote in spite of such exemptions.
All this notwithstanding, the argument must be modified one last time. F1*, F3*, F4 and FC* must be replaced by F1**, F3**, F4* and FC**, as follows. Since these are the final refinements, I shall now state the argument in full.
(F1**) It is unfair for sufficiently capable persons to benefit non-trivially from voting by members of the social group(s) to which they belong without either voting, or making some equivalent contribution to the political process by which those benefits are produced.
(F2*) Where persons have duties of fairness in virtue of their willing acceptance of benefits, it is permissible for the state to subject those persons to compulsion where doing so would substantially prevent those persons from violating those duties of fairness, provided there are no unacceptable moral costs associated with doing so.
(F3**) Compelling sufficiently capable persons who willingly accept the benefits of voting by members of the social group(s) to which they belong, without making some equivalent contribution to the political process by which those benefits are produced, to vote would substantially prevent such persons from violating their duties of fairness.
(F4*) There are no unacceptable moral costs associated with compelling sufficiently capable persons who willingly accept the benefits of voting by members of the social group(s) to which they belong, without making some equivalent contribution to the political process by which those benefits are produced, to vote.
(FC**) Therefore, it is permissible for the state to compel sufficiently capable persons who willingly accept the benefits of voting by members of the social group(s) to which they belong, without making some equivalent contribution to the political process by which those benefits are produced, to vote.
The defence of the argument, however, is still incomplete. F4* asserts that compulsory voting does not entail unacceptable moral costs. However, critics of compulsory voting often object that compulsory voting does entail unacceptable costs. A satisfactory defence of the free-riding argument, then, must respond to these claims.
The purported moral costs of compulsory voting
Opponents of compulsory voting have advanced three lines of argument to the effect that compulsory voting entails unacceptable moral costs.Footnote 24 Let us consider each, in turn.
Lower-quality decisions
The most popular line of argument holds that compelling citizens to vote leads to lower-quality political decisions.Footnote 25 Empirical evidence suggests that habitual non-voters are generally less informed than citizens who vote regularly (Citrin, Schickler, and Sides Reference Citrin, Schickler and Sides2003). Compulsory voting compels persons who would otherwise be habitual non-voters to participate. The median voter, then, is less informed under compulsory voting. Lower-quality decisions, critics claim, are the inevitable result.
However, the risks of lower-quality decisions under compulsory voting must surely be weighed against the chance of higher-quality decisions. As discussed above, there is substantial evidence that compulsory voting produces higher-quality decisions, at least with respect to equality, by mitigating socioeconomic bias in turnout. On the other hand, empirical evidence that compulsory voting produces lower-quality decisions is thin on the ground. There are, moreover, at least two reasons to doubt whether such evidence is likely to be forthcoming.
First, as Lijphart (Reference Lijphart1997, 10) suggests, ‘mandatory voting may serve as an incentive to become better informed’, mitigating concerns over voters’ information levels. Some studies (Birch Reference Birch2009b, 67; Loewen, Milner, and Hicks Reference Loewen, Milner and Hicks2008) find no evidence of such an effect. A number of more recent studies, however, lend support to Lijphart’s conjecture. Shineman (Reference Shineman2018), for example, surveyed two groups of around one-hundred voters prior to a San Francisco municipal election. One group was issued with a $25 gift card, to be activated only if they voted. Post-election surveys showed that the group who stood to lose $25 by not voting gathered substantially more information than the control. In a cross-national study of thirty-seven countries, Sheppard (Reference Sheppard2015) found that where compulsory voting is effectively enforced, controlling for other factors, citizens have greater average levels of political knowledge, and there is a smaller knowledge gap between well-educated and under-educated citizens. Carreras (Reference Carreras2016) found, in a study of Latin American nations, that compulsory voting increases overall levels of political discussion and attention to political news among the poorly educated. In a cross-national study of twelve European nations, Gordon and Segura (Reference Gordon and Segura1997, 137–41), and Berggren (Reference Berggren2001), found that compulsory voting has a positive effect on citizens’ ability to locate political parties on the ideological spectrum. Finally, in a study comparing Austrian provinces in which voting is compulsory to those in which it is voluntary, Shineman (Reference Shineman2012) found that compulsory voting has a significant, positive impact upon citizens’ levels of political knowledge.
Second, even if future research tells against Lijphart’s conjecture, this may not matter. A substantial body of evidence indicates that uninformed voters are often able to emulate better-informed voters by employing informational shortcuts (Lupia and McCubbins Reference Lupia and McCubbins1998). Merely knowing a candidate’s partisan affiliation, race or gender, for example, is often sufficient for relatively uninformed citizens to vote as they would have done had they been better informed. Informational shortcuts are no panacea. They can go wrong in a variety of ways. Still, they do plausibly allow relatively uninformed persons – including those who would otherwise have been habitual non-voters – to vote as though they were well informed in a wide range of cases.
I think, then, that this line of objection can be resisted. The risks of lower-quality decisions appear to be highly speculative while the benefits are demonstrable.
Valuable political effects
A second line of argument holds that optional voting has valuable political effects that would be lost under compulsory voting. Swenson (Reference Swenson2007, 540–1) appeals to Downs’ (Reference Downs1957, 119) observation that more extreme supporters of political parties can deter those parties from moving towards the median voter by threatening to abstain. Citizens would be deprived of this ability under compulsory voting. However, it is unclear whether this would always be much of a loss. Where the median voter is more reasonable than the extremes, it is surely better if parties move towards that position. Moreover, as pointed out in the introduction, compulsory voting does not deprive citizens of the ability to abstain. Citizens can leave the polling station without voting once their attendance has been registered.
Alternatively, Ciccone (Reference Ciccone2002, 348) argues that optional voting prevents unaffected or indifferent voters, who typically abstain where voting is optional, from distorting political outcomes. Voting by unaffected or indifferent citizens may lead to bad outcomes from the perspective of affected or non-indifferent citizens. Childless persons voting in school board elections, for example, may select candidates harmful to the school system. Moreover, if the number of indifferent or unaffected voters is greater than the number of affected or non-indifferent voters, politicians will have incentives to appeal to the former, at the expense of the latter.
I deny, however, that unaffected or indifferent voters pose such difficulties. Unaffected voters simply do not exist for large-scale political decisions, which typically affect all members of the relevant community in some way. There are perhaps some limited cases where groups who cannot be affected are clearly identifiable (school board elections, perhaps). In those cases, compulsory voting may be inappropriate. It obviously does not follow, however, that compulsory voting is inappropriate across the board.
‘Indifferent’ voters to whom politicians successfully appeal are, ex hypothesi, no longer indifferent. There is presumably nothing problematic about such voters exercising influence over political outcomes. On the other hand, genuinely indifferent voters can be expected to vote randomly, distributing their votes evenly over the options on the agenda, making no difference to the outcome (Goodin Reference Goodin2007, 58–9).
Finally, Blomberg (Reference Blomberg1995, 1024–5) asserts that optional voting is an efficient means of encouraging qualified individuals to run for office. Widespread abstention (impossible under compulsory voting) implies the existence of a large constituency of voters who might be mobilized by an exceptional candidate. Blomberg, however, offers no evidence in support of this claim, and there are good reasons to be sceptical. Abstention in previous elections is a strong predictor of abstention in future elections (Fowler Reference Fowler2006; Plutzer Reference Plutzer2002). Candidates who seek to mobilize large numbers of habitual non-voters can therefore expect to meet with little success. This probably deters most potential candidates from trying. Candidates who do successfully mobilize large numbers of habitual non-voters may not be of particularly high quality. And, even if they are, there is no guarantee that they will win. None of these objections, then, is particularly compelling.
The right not to vote
The final line of argument holds that compulsory voting violates the ‘right not to vote’. Some (Ciccone Reference Ciccone2002, 347–8; Katz Reference Katz1997, 244) think the right not to vote is logically entailed by the right to vote. That is much too quick. As many (e.g. Hill Reference Hill2014, 161) have pointed out, rights do not necessarily entail their inverse. A child’s right to an education, for example, does not entail a right not to be educated. Proponents of this objection, then, must offer a substantive argument for the right not to vote. Several such arguments have been offered, none of which is convincing.
First, Blomberg (Reference Blomberg1995, 1019–20, 1022) argues that citizens have a right against being compelled to vote on the grounds that abstention can be rational. This proves far too much. Much of the criminal law, after all, functions by deterring citizens from actions – for instance, theft – it might otherwise be rational to perform.
Second, Ciccone (Reference Ciccone2002, 349–50) argues that citizens have a right against being compelled to vote on the grounds that voting is supererogatory. I have argued at length, above, that voting is not supererogatory for the vast majority of citizens, and that the remainder might simply be exempted from the requirements of compulsory voting (assuming some other justification for subjecting them to such requirements cannot be found).
Third, several theorists have argued that abstention constitutes a valuable expressive opportunity. Proponents of this argument have appealed to two distinct opportunities in this respect: the ability to express dissatisfaction with, or indifference between, the options on the ballot (Ciccone Reference Ciccone2002, 347–8; Swenson Reference Swenson2007, 542), and the ability to express dissent from the democratic system as a whole (Blomberg Reference Blomberg1995, 1023; Lever Reference Lever2010, 911; Swenson Reference Swenson2007, 542–3). Let us grant, arguendo, that citizens have rights to express attitudes of both kinds, and that abstention is one means of doing so.
It doesn’t follow, however, that compulsory voting is rights-violating. This is clearest with respect to the right to express indifference between, or dissatisfaction with, the options on the ballot, by abstaining. Compulsory voting does not deprive citizens of the opportunity to abstain, and therefore does not deprive citizens of such expressive opportunities.
Compulsory voting does, however, compel citizens to attend a polling station. That, it might be argued, is a form of political participation, even if citizens go on to abstain. One might claim, then, that mandatory attendance deprives citizens of the opportunity to express dissent by declining to participate, in this more general sense. However, compulsory voting would leave open many expressive opportunities of equivalent value – posting on Facebook, writing letters to newspaper editors and so on. Indeed, compulsory voting plausibly amplifies the expressive power of refusing to attend by converting the act of abstention into a costly signal (under compulsory voting, citizens must pay a fine if they wish to express dissent by refusing to attend).
Finally, it might be argued that compulsory voting violates liberal neutrality.Footnote 26 Reasonable disagreement abounds over the value of political participation. It might be argued, then, that compelling citizens to vote violates citizens’ rights against having reasonably contested conceptions of the good imposed upon them. This, however, misunderstands the requirements of liberal neutrality. Compulsory voting certainly has effects that adherents of certain conceptions of the good will find unwelcome. Yet liberal neutrality is not ultimately concerned with ensuring neutrality in the effects of policies. If neutrality requires anything at all, it surely requires religious tolerance, which disadvantages adherents of intolerant religions, for example. Rather, liberal neutrality requires that policies not be justified solely by appeal to reasonably contested conceptions of the good (Rawls Reference Rawls1988, 260–5). Proponents of compulsory voting do not typically appeal to such conceptions but, rather, to values all citizens might reasonably be expected to share (e.g. equality, fairness).Footnote 27 It is hard to see, then, quite how compulsory voting is supposed to be objectionable on this score.
Some more compelling argument showing that compulsory voting is rights-violating might, perhaps, be available. Yet this seems doubtful. Even if there were a plausible argument for a right not to vote, it would still need to be shown that compulsory voting – wherein citizens retain their ability to abstain – would violate that right. More generally, compulsory voting is neither particularly demanding, nor particularly coercive. Elections come around only occasionally, and punishments under defensible regimes of compulsory voting are light. If laws compelling citizens to pay tax or obey the speed limit are not rights-violating, it seems unlikely that compulsory voting is rights-violating either. In the absence of such an argument, we can safely dismiss this final line of objection.
No prominent attempt to demonstrate that compulsory voting would entail unacceptable moral costs succeeds. That is of independent interest. But, more importantly for our purposes, it completes the final step in the defence of the free-riding argument. The free-riding argument, as such, appears to be sound.
Conclusion
I began this article by setting out the problem of declining turnout, and pointing to compulsory voting as being amongst the most effective remedies. Compulsion, however, demands justification. I argued that non-voters, in the vast majority of cases, are free-riding upon members of the social group(s) to which they belong who choose to vote and therefore have coercible duties to vote. Typical objections to the free-riding argument, I argued, ought to prompt us to refine the argument, rather than reject it. I also considered a range of prominent objections to compulsory voting as such, each of which purports to show that compulsory voting is unacceptably costly from a moral point of view. These objections, I argued, are uniformly unsuccessful. States, I conclude, may permissibly compel citizens to vote.
Acknowledgements
For helpful written comments and criticisms, not all of which I have been able to address, I thank Geoffrey Brennan, John Dryzek, Lisa Hill, Shmulik Nili, Philip Pettit, Nicholas Southwood, and Laura Valentini, as well as three anonymous reviewers for the British Journal of Political Science and Associate Editor Paul Bou-Habib. My thanks also to audiences at the Australian National University, University of New South Wales, University of Western Australia, and the 2018 Australian Political Theory and Philosophy Conference at the University of Sydney. My research was partially supported by an Australian Government Research Training Programme Scholarship. All errors, of course, are my own.
