Public Reason and the Legitimacy of Democratic Social Choices
Choosing Between “Populism” and “Elitism”
[This post is based on and summarizes the main points of a working paper I’ve just drafted, “Public Reason, Democracy, and the Ideal Two-Tier Social Choice Model of Politics”. You can read the full paper here. Comments are of course welcome!]
The idea that the main results of social choice theory, especially Arrow’s “impossibility theorem”, casts doubt on the very possibility of legitimate democratic social choices is not new. It has been most clearly articulated by the political scientist William Riker in his book Liberalism Against Populism: A Confrontation Between the Theory of Democracy and the Theory of Social Choice. The generic argument is that because there is no decision rule satisfying a set of plausible requirements that is not prone to strategic manipulation and/or that does not fail sometimes to identify the best alternative (because it generates a cycle), the democratic social choice cannot express the “general will”. This refutes the “populist” conception of democracy.
There is however a more constructive way to use social choice theory to reflect on the legitimacy of democratic social choices. In a review of Amartya Sen’s memoirs, the economist Partha Dasgupta sketches a normative social choice account of what an ideal democracy would look like:[1]
“Ideally, citizens would vote in line with their ‘social preferences’, not their personal interest… Arrow chose the title ‘Social Choice and Individual Values’ for his book, his intention was to draw a distinction between voting rules and directives that should guide the citizen on whom to (more accurately what to) vote for. Ethical considerations that are directed at identifying voting rules are this different from the ones citizens will wish to entertain for arriving at their social preferences over alternative policies… For a citizen to discover her social preferences over, say, economic states of affairs requires a different kind of ethical reasoning. She will, for example, want to compare people’s needs, which means interpersonal comparisons of individual well-beings would be an essential feature in her exercise.”
In other writings, Dasgupta has defended to use of the “majority rule” (i.e., the Condorcet procedure) as the best decision rule in light of the impossibility results of social choice theory.[2] This rule has two significant properties. First, it satisfies the condition known as “independence of irrelevant alternatives”. Even though the normative status of this condition for voting rules has long been controversial, its attractivity lies in the fact that it limits the range of strategic manipulation.[3] Second, of all the voting rules that satisfy the independence condition and that can be said to be “democratic”, it is the most robust one, i.e., it is the less likely to generate a cycle and thus to fail to identify the best alternative.
The crucial point here is that Dasgupta clearly separates the discussion of the voting rule and of the normative criteria that are relevant to assess it from the discussion of the ethical considerations citizens should ideally have to form their “social preferences” based on which they vote. There are therefore two distinct kinds of social choices: a “political” social choice based on a (democratic) decision rule that aggregates citizens’ social preferences, and a “normative” social choice where each citizen forms their social preferences by aggregating people’s tastes, interests, and so on, based on a set of ethical considerations captured by some “social welfare (or evaluation) function”. The articulation of these two social choices gives rise to what I call the “Idea Two-Tier Social Choice Model of Politics”. It has a simple visual representation:
How should people form their social preferences? Dasgupta briefly mentions the idea that citizens should empathize with each other and make interpersonal comparisons, for instance as modeled by John Harsanyi in its so-called “impartial observer theorem”. Another possibility worth considering is that the formation of social preferences should be made under the requirement of public reason. You find this idea for instance in John Rawls’s political liberalism. In a section of Political Liberalism aptly titled “Public Reason and the Ideal of Democratic Citizenship,” Rawls wrote[4]
“when may citizens by their vote properly express their coercive political power over one another when fundamental questions are at stake? … our exercise of political power is proper and hence justifiable when it is exercised in accordance with a constitution the essentials of which all citizens may reasonably be expected to endorse in the light of principles and ideals acceptable to them as reasonable and rational… since the exercise of political power itself must be legitimate, the ideal of citizenship imposes a moral, not a legal, duty – the duty of civility – to be able to explain to one another on those fundamental questions how the principles and policies they advocate and vote for can be supported by the political values of public reason.”
Rawls’s account of public reason notoriously leads to a very strong requirement for justification. Basically, it implies that citizens must share a (liberal) political conception and refrain from using non-public reasons to justify demands regarding “constitutional essentials and matters of basic justice”. Less demanding accounts in terms of “convergence” rather than “consensus” are also available. For instance, in the version defended by Gerald Gaus,[5] it is only required that “members of the public” agree on some moral rules, even though they do so for different reasons.
This weaker requirement can be translated into social choice theoretic terms. “Members of the public” may agree on a profile of social preferences as long as this profile results from the use of social welfare functions that no one has a decisive reason to reject. More formally, members of the public may identify a subset F* of social welfare functions that pass the test of public justification. These functions are different publicly valid ways to aggregate people’s interests and conceptions of the good. Consider a set X of social alternatives. We can construct a subset X* by picking all the alternatives that are ranked best by at least one function in F*. The rationale for the construction of this subset is that if some alternative x is ranked best by some function F belonging to F*, then F provides a public reason in favor of the social choice of x. This gives rise to a well-defined requirement of public reason applying to citizens’ social preferences: citizens must systematically socially prefer a social alternative x that is publicly supported by a function F* to any social alternative y that doesn’t have this support. Beyond that, citizens’ social preferences can largely vary. Notably, their social preferences can reflect their personal interests as well as their private reasons for preferring specific social welfare functions.
The legitimacy of the democratic social choice thus depends on the fact that it results from social preferences that are publicly justified. There is a catch, however. There is no reason to assume that the scope of public reason must stop at the formation of social preferences. Indeed, as the quote of Rawls above suggests, it should encompass the choice of the decision rule in the second tier of the social choice model. Incidentally, that individuals may also have preferences over decision rules has been suggested early on by Kenneth Arrow himself, as I have discussed at length in a previous post.
This leads to some interesting complications. It seems reasonable to suggest that members of the public have to weigh the fact that citizens are actually unlikely to vote based on publicly justified social preferences. There is indeed a growing literature on voters’ behavior that indicates that the latter reflects biases and prejudices that can hardly pass the requirement of public reason. In other words, citizens are not public reasoners. The result is a kind of stability problem for the social choice account of democratic legitimacy: what you know about voters’ behavior should surely be relevant for the choice of a decision rule. If this is not duly taken into consideration, the risk is that the democratic procedure picks an alternative that is not publicly justified. The social choice model would fail to justify the social choice on its very own terms!
The suggestion is thus that members of the public’s social preferences should be extended such as to cover decision rules. The public justification of social preferences therefore must depend on the fact that how decision rules are ordered also respond to the requirement of public reason. This disqualifies the kind of “populist” account of democracy that Riker has disparaged for different reasons. The legitimacy of a social choice is not grounded in the fact that it results from a social preference held by the majority of citizens. The alternative I propose is what I call the “Elitist View of Legitimacy”:
The Elitist View of Legitimacy. Members of the public who are aware of the fact that voters are unlikely to vote based on publicly admissible social orderings have conclusive reasons to socially prefer elitist decision rules over democratic rules.
A decision rule is elitist if at least one of the following applies: (i) its domain is restricted to the subset of social preferences that are publicly justified; (ii) some voters’ social preferences are given more weight than others’; (iii) it is “oligarchic.”
Crucially, while non-democratic in the social choice theoretic sense, some elitist decision rules are compatible with the actual political morality of liberal democracies. Consider for instance a relatively complex rule that makes the social choice depend on the unanimous judgment of an oligarchy (eventually a dictator) only in case some of the social preferences expressed are not publicly admissible and/or the social choice picks a social alternative that is not publicly justified. This is indeed nothing but a formal and simplified version of a democratic regime with a judicial review. Other elitist rules are in line with recent “epistemic” accounts of democracy, especially those that claim that legitimacy depends on the appropriate deliberative procedure. Finally, there surely are elitist rules that are plainly “epistocratic”, corresponding for instance to John Stuart Mill’s plural voting.
The main point is therefore that the conjunction of the social choice model with the idea of public reason fails to vindicate the claim that legitimacy is granted by democratic decision rules. “Democratic populism” does not pass the test of public reason. More worrying for some, public reason seems also to license a large range of elitist decision rules, some of which challenge the received democratic political culture.
[1] Partha Dasgupta, “The Perils of Cosmopolitan Intellectualism,” Society 58, no. 5 (October, 2021): 416–23.
[2] E.g., Partha Dasgupta and Eric Maskin, “On the Robustness of Majority Rule,” Journal of the European Economic Association 6, no. 5 (September 1, 2008): 949–73.
[3] Compare with the decision rule called the “Borda count” where voters rank the n alternatives and the first is given n points, the second n-1 points, and so on. This rule does not satisfy the independence condition because the social ranking between two alternatives x and y can be affected by the inclusion of a third alternative z even though the voters have not changed their relative ordering of x and y. This obviously opens the door to strategic votes.
[4] John Rawls, Political Liberalism (New York: Columbia University Press, 1993), 216-17.
[5] Gerald Gaus, The Order of Public Reason: A Theory of Freedom and Morality in a Diverse and Bounded World, Reprint edition (Cambridge New York,NY: Cambridge University Press, 2012).
I don’t think the fact that voters vote based on private reasons and biases (usually group identity, see Bartels and Achen 2016) is the justification for the elitist view of legitimacy.
Given a pluralistic society, there will be enough interests to prevent minority for obtaining significant power. Private interests will cancel each other out. This is basically Federalist 10 and the wisdom of crowds.
The issue arises when there are systemic biases that consistently mislead the democratically generated result away from public reason, in which case decision making should be outsourced to a better suited body. I wouldn’t call this theory elitist, the trustee vs delegate debate has been around for sometime. It would be decision making by experts, so not necessarily more elitist than obeying my doctor or lawyer. We can see this with systemic biases in economics (Caplan 2007) justifying central bank independence. And there is evidence showing that more independence leads to lower inflation rates (a good sign id say).
I discuss this more here https://neonomos.substack.com/p/democracy-is-a-means-to-an-end-not
Overall, while democracy is good for power balance, information conveyance, legitimacy, and shaping national values (by prioritizing and applying moral principles to national circumstances), to the extent that it’s decision making is off the mark, it’s legitimate to limit the extent of majorities’ decision making power.