The Epistemology of Democracy and the Market: Rejoinder to Elliott

ABSTRACT

Epistemic democratic theory often focuses on defending democracy from various forms of elitism, such as epistocracy. However, democracy’s informational properties may also be compared with those of the market, and not other forms of political decision-making. While Kevin Elliott’s critique of the market’s epistemic properties is a welcome contribution that broadens the range of comparisons epistemic democratic theory engages with, Elliott mischaracterizes arguments made by market theorists, overlooks their justifications for employing unrealistic assumptions, and ignores instances where they agree with his critiques and arguments. This paper examines arguments regarding the epistemology of democracy and the market, and discusses the comparative knowledge requirements each competitive system requires of its participants within the wider context of epistemic democratic theory.

Keywords:

• markets
• democracy
• voter ignorance
• epistemic democratic theory
• comparative institutional analysis

In 1998, Critical Review published several articles discussing empirical public opinion scholarship’s implications for modern politics, and by extension, democratic theory. Many of these articles examined the relationship between political knowledge and democracy, echoing themes studied by epistemic democratic theorists examining whether democratic practices, such as voting and deliberation, grants democracy informational advantages over rival forms of public decision-making (Cohen 1986; Estlund 1997; Landemore 2012; Schwartzberg 2015).

Epistemic democratic theorists often compare democracy to various forms of elite rule, examining, for example, whether diverse groups of voters may arrive at epistemically superior decisions when compared with experts or randomly selected decision-makers (Bennett 2020; Landemore 2012). However, others have sought to compare democracy and the market to illustrate the epistemic properties of rival economic and political institutions (e.g. Benson 2019; Herzog 2023). These rival forms of institutional comparison can illustrate different aspects of how social and political knowledge influence decision-making, and may also illustrate how different institutions mitigate or exacerbate the harmful effects of ignorance on human affairs.

Kevin Elliott’s (2020) publication in this journal is a welcome sign of productive exchange regarding the epistemic properties of rival institutions, and illustrates how democracy may fruitfully be compared with the epistemic properties of markets, and not just to various forms of political rule. Although Elliott recognizes different authors employ different arguments regarding the market’s epistemic properties, this paper focuses on one of the arguments he critiques (DeCanio 2014), which examined the information markets and democracy require of their participants.

Elliott’s critique claims that (1) markets cannot make collective political decisions; (2) markets cannot produce second order goods, i.e. the institutional rules governing a society; (3) democracy should not be defined in unfairly narrow terms or at the exclusion of deliberative approaches; (4) policy bundling gives policy makers incentives to do the best they can across all policies; (5) democracy can, like markets, be responsive to new circumstances; and (6) democracy should not be unfairly assessed relative to a standard of full information that cannot be met by any institution.

This paper responds to Elliott’s critique and proceeds in three sections. Section I summarizes the epistemic critique of democratic politics presented in DeCanio (2014), and explains how this argument diverges from those made by authors such as Ilya Somin, Jason Brenan, and Bryan Caplan. This section focuses on explaining how DeCanio (2014) does not focus upon the characteristics of knowledge held by voters or the incentives that may cause voters to be rationally ignorant. Rather, it emphasizes how the number of organizations populating a given system is the key feature influencing the kinds of knowledge agents must possess to make rational inferences about politics.

Section II responds to three of Elliott’s claims: that markets cannot produce certain political decisions, certain policy decisions that must be made by political decision procedures, and democracy can respond to new conditions and situations. While I agree with Elliott on these three points, I suggest the market’s epistemic function may still help illustrate specific problems caused by the singular nature of political decisions, thereby offering a helpful way to think about the challenges facing public action even when there is no alternative to producing goods through the state. While many decisions must be made politically, focusing on the problems unobservable policy counterfactuals pose helps identify a specific type of epistemic problem many public decisions inevitably exhibit and which have gone unrecognized.

Section III examines Elliott’s criticisms of my use of unrealistic assumptions and his critique of my definition of democracy. This section argues Elliott overlooks the methodological reason for why I employ unrealistic assumptions, namely how a lack of realism is employed to isolate the causal effect of specific variables in my thought experiments, and to understate the severity of the information problems democratic politics faces.

I conclude by suggesting epistemic democrats’ arguments regarding the benefits of cognitive diversity and viewpoint diversity in political decision-making can justify considering a wide range of institutional comparisons across politics, markets, and society. Although epistemic democrats often focus on comparing different forms of political decision-making, a wider range of institutional comparisons may be fruitfully employed to illustrate the epistemic properties of various institutions, organizations, and forms of competition.

Knowledge, Politics, and the Market

This section summarizes the argument in DeCanio (2014) that Elliott (2020) critiques. Doing so is necessary because, as Elliot recognizes, different authors employ different arguments regarding the causes of voter ignorance, and their institutional recommendations often follow from these theories of knowledge. One of the standard approaches begins from considering the low incentives voters have for collecting political information. For example, Somin (1998, 2013, 2020) follows Downs (1957) in arguing the low probability that any individual vote will be decisive in election outcomes causes voters to be rationally ignorant, as the costs of being informed exceed the benefits. This leads Somin (1998, 2020) to propose restricting the scope of politics and prioritizing foot voting, as he argues it is easier for voters to become politically informed when the scope of politics is limited, while foot voting better aligns individuals’ incentives with the consequences of their decisions.

Similarly, Brenan (2016) argues rational ignorance justifies restricting voting to those that are politically knowledgeable, defending a type of epistocracy, whereby well-informed voters’ preferences are empowered as a solution to the problems voter ignorance poses. Brennan (2014) alternatively suggests that one response to voter ignorance is to make decisions outside the scope of politics entirely, leaving them to be settled either within the realms of civil society or the market.

However, following Joseph Schumpeter (1952) and Walter Lippmann (1922), Friedman (1998) attributes public ignorance to the lack of information feedback from political decisions, and the inherent limits of the human mind that exist independently of voters’ incentives. Friedman (2019, Ch. 7) subsequently emphasized the consequences of ideational heterogeneity and the difficulty of predicting the effects of public policies, arguments used to justify “exitocracy” which he contrasted with technocracy in either its elitist and populist forms.

These examples illustrate the range of explanations that have been offered for the causes of voter ignorance, and the corresponding variety of institutional recommendations in light of voters’ low levels of political knowledge. I emphasize such diversity because DeCanio (2014) sought to study the consequences of public ignorance without attributing voter ignorance to any specific cause. Instead of emphasizing the incentives voters face, or the forms of information feedback public decisions exhibit, I argued a specific feature of the state’s decisions, namely their singularity and corresponding lack of observable policy counterfactuals, exacerbates the harmful effects of public ignorance.

Specifically, the state is defined as an organization that monopolizes the production of public policies, and the principal characteristic of the state’s decisions is that they are singular and lack alternatives. An unintended byproduct of this form of authority is that citizens cannot observe the counterfactual effects of alternative policies that were not implemented. For example, if a central bank cuts interest rates to stimulate the economy, it is impossible to observe whether a different interest rate decision, one that was perhaps marginally higher or lower, would have produced preferable economic consequences (lower unemployment, inflation, etc.) relative to the decision the central bank made. As multiple central banks do not produce competing interest rate decisions for society, there is simply the single decision the central bank implements and the economic conditions that are then correlated with this decision. This feature of public decisions holds across multiple policy areas, as the state makes singular decisions for issues such as foreign policy, taxation, health regulation, amongst other things.

The singularity of state decisions makes it difficult for anyone, citizens or experts alike, to determine whether social conditions are 1) caused by a specific policy decision or not, 2) would have emerged regardless of the policy the state implemented, or 3) whether better social conditions would have been produced by policies that were not implemented or had a different political party or politician been in power. Although democratic states are capable of responding to new economic conditions and the threat of electoral sanction creates strong incentives for political actors to respond to social needs and preferences, DeCanio (2014) emphasized a problem that exists independently of the incentives voters and political actors face, and which would persist even if many unfavorable aspects of democratic politics, such as shirking, rational ignorance, or elite manipulation of public opinion, were absent.

The state’s monopolization of public policy decisions ensures politics is fundamentally different from competitive markets, not due to the presence of a specific motive, such as self-interest, but because politics lacks the same forms of multiplicity exhibited by rival firms’ products that populate competitive markets. The singularity of political decisions distinguishes democratic politics from economic markets, as the former exhibits competition among groups to influence the single policy decision the state implements, while in the latter competition among rival firms produces a multiplicity of products that allows consumers to make product comparisons with a far greater degree of simultaneity than is possible in democratic politics.¹

This argument accepts that consumers are deeply ignorant about the private goods they purchase in markets just as voters are ignorant about the politicians and parties they must elect, and recognizes that, contrary to Schumpeter (1942), many public decisions exhibit forms of information feedback that are indistinguishable from private consumer goods. For example, public decisions such as the quality of customer service at the US Postal Service or the DMV exhibit direct and immediate feedback, and conversely, private decisions, such as various forms of insurance, lack immediate information feedback (Hausman 2024). This indicates that public and private decisions are not conceptually distinguished by the nature of the feedback either type of decision generates.

Instead of emphasizing differences in information feedback or incentives, politics is distinguished by the singular nature of the state’s political decisions, while “private” action in markets is distinguished by the simultaneous existence of competing organizations (i.e. firms). Since voters cannot observe the consequences of policies other than those incumbent officials or parties implement, the singularity of the state’s decisions exacerbates voters’ need for political knowledge, as voters must engage in counterfactual reasoning to assess whether different policies would have been more effective in altering social conditions at lower cost, but the accuracy of these counterfactuals can never be tested due to the singularity of the state’s decisions. This argument emphasizes the epistemic consequences of specific forms of organizational authority in politics, and not on actors’ incentives or motives.

Furthermore, this argument suggests that markets do not mitigate the epistemic problems created by the singularity of public authority because of the profit motive, information feedback, or because consumers are well-informed. Rather, markets’ epistemic function is derived from consumers’ ability to directly compare firms’ products and identify which they prefer, as such comparisons eliminate consumers’ need to understand the causal reasons why they find a product preferable to its rivals. Even if consumers are just as ignorant about firms and products as voters are ignorant of parties and public policies, equally ignorant actors have different opportunities for observing counterfactual data due to the different number of organizations populating economic and political systems, and are in different positions to make inferences regarding their satisfaction with rival organizations’ decisions.

Unlike the multiplicity of product counterfactuals consumers can observe in markets, the singularity of the state’s decisions requires voters to try to imagine the counterfactual effects produced by policy decisions political parties or the state did not implement, and compare these (imaginary) effects to those produced by the party in power. The singularity of the state’s decisions ensures voters cannot determine whether the social conditions that exist under the incumbent party were caused by the party’s policies, nor can they observe whether the incumbent party’s policies produced preferable effects relative to those that would have been implemented had a different party been in power. This indicates that the number of organizations populating economic and political systems have epistemic effects, ensuring equally ignorant actors face different inferential problems depending upon whether decisions are made in markets or democratic politics.

However, instead of arguing markets aggregate decentralized knowledge (e.g. Hayek 1945), markets’ beneficial epistemic properties are derived from the counterfactual comparisons competition facilitates, and the way such comparisons allow ignorant and misinformed actors to identify the comparative efficiency of rival responses to scarcity without having to understand why a product or production decisions is more profitable than the alternatives. These beneficial informational characteristics persist even if knowledge asymmetries exist between consumers and producers, as markets allow ignorant actors to identify efficient outcomes even if they adopt systematically incorrect ideas regarding why they prefer a given product or how a firm has outperformed its competitors (e.g. Akerlof 1970).²

Instead of grounding this theory of the market on culturally and historically contingent assumptions regarding human nature, the prevalence of self-interest, or the incentives actors face, the market’s epistemic efficiency is a consequence of the counterfactual comparisons competition facilitates, the way such inferences approximate the conditions of scientific experimentation. However, the singularity of the state’s political decisions prevents the existence of policy counterfactuals in public decision-making, exacerbating political actors’ need for knowledge in ways that are rendered unnecessary by the counterfactual comparisons markets facilitate.

Second Order Goods and the Limits of the Market

It is necessary to recognize that this depiction of the market’s epistemic properties is not equally applicable across all policies situations, and it does not justifying market production of public goods. Hence I agree with Elliot’s argument that markets cannot produce many collectively binding second order goods, nor can they produce decisions on issues such as abortion. Although this qualifies the scope of my argument, emphasizing the epistemic consequences of the singularity of state decisions helps identify a general problem confronting public decision-making. If a key characteristics of political decisions is their singularity for all members of a geographically defined society, such decisions exhibit inferential difficulties associated with unobservable counterfactuals that are well established in the methodological literature regarding causal inference and experimental research design (Holland 1986). Hence although markets are incapable of producing decisions regarding a range of political decisions, as well as various public and second order goods, the singular nature of state decisions makes it difficult for anybody to observe whether, for example, the policies used to generate a given level of equality were preferable to the alternative policies that were possible but which were not implemented.

The problems stemming from this epistemic feature of politics may help explain why social consensus on policies to reach consensual ends often remains fraught with disagreement, as the difficulties in generating experimental assessments of the goods produced through the state may make it difficult to generate the forms of consensus that are possible in other epistemic realms, such as the medical treatment of disease, where experimentation is possible. Democratic disagreement may thus persist due to the inherent nature of political decisions, e.g. their singularity, as the corresponding impossibility of conducting experimental tests of rival hypotheses regarding how to, for example, render a society more equitable or just, and voters inability to directly compare the effectiveness of parties’ policies in the same way that consumers can compare firms’ products to assess which they prefer.

Of course politics exhibits some opportunities for conducting policy experiments, and it is always possible to try to compare conditions in different states or cities, or across different countries. However, differences in individual states and local governments may frustrate efforts to use federalism to make policy comparisons, as inferences regarding the policies deemed to “work” in a given state or city may not be applicable to other states or cities that may have different characteristics.

Furthermore, on major public policy questions taken at the federal level, such as monetary policy, or decisions about foreign policy, there is no ability to compare policy counterfactuals at subnational levels. Despite recognizing markets produce a range of undesirable outcomes, if one accepts the experimental method’s validity as a unique means of generating understanding, the market’s approximation of the experimental conditions of treatment and control indicates that markets exhibit epistemic properties that are absent from politics due to the singular nature of the state’s decisions.³

The problems stemming from the singularity of political decisions has implications for Elliott’s suggestion that I overlook how democracies can respond to new issues and problems. While I agree with Elliott that democracy provides incentives for political actors to be responsive to such conditions, the issue I would emphasize is how voters are supposed to assess parties’ efforts to respond to new conditions given their inability to observe whether another party would have produced preferable outcomes.

Consider a situation where an economic crisis occurs and an incumbent party makes every effort to respond with a bundle of policies designed to combat the crisis. Following the implementation of these policies the economy exhibits a given level of unemployment, growth, inflation, etc. How are voters and political observers to judge the conditions following the parties’ response to the crisis? Were these the best outcomes that were possible given the specific features of the society in question? Were these benefits generated at lowest cost? Or would a different party or policy have produced preferable consequences at lower cost? The problem here is not that democratic states do not respond to social conditions, it is that the absence of policy counterfactuals it is difficult to assess the consequences of the policies that were implemented.

I should note that while DeCanio (2014) illustrated this argument in the context of comparing markets and democracy, this argument can also be applied to understand the internal properties of markets themselves. Indeed, examining the product counterfactuals that are possible in competitive and uncompetitive markets offers a conceptualization of markets emphasizing the inferences that different degrees of competition make possible. As the quantity of product counterfactuals may have implications for consumers’ ability to identify the products they prefer, this approach may identify problems with uncompetitive markets that are unrelated to the welfare problems typically emphasized by studies of monopoly, such as the deadweight welfare losses caused by monopolistic firms’ pricing power or barriers to entry.

However, just as there are fruitful comparisons that can be made between the counterfactual choice environments in economic and political competition, it may also be useful to compare the forms of counterfactual data generated in monopolistic markets with decentralized political systems (i.e. federalism), as voters can compare a greater quantity of political counterfactuals generated by multiple city, state, and local governments relative to the single product counterfactual that is available when a firm monopolizes a market. This type of comparative counterfactual analysis is overlooked if epistemic comparisons are only made within specific forms of economic or political competition (e.g. simply comparing democracy with epistocracy), thereby restricting the full range of comparisons that can be made across different competitive systems.

Unrealistic Assumptions and Minimalist Definitions

To illustrate how knowledge problems are not only a consequence of the incentives operating upon the aggregation decentralized information, DeCanio (2014) conducts a series of thought experiments employing deliberately unrealistic assumptions to illustrate the market’s epistemic function. Elliott (2019, 18) is critical of such lack of realism, claiming I compare imperfect democratic political institutions to an unrealistically “perfect” depiction of markets, resulting in unfairly assuming that the market “excuses any particular failure or suboptimal outcome as an error, and an error that is all but certain to be fixed in the fullness of time.”

I should note that I agree with Elliot that claiming markets will fix any error in the fullness of time is naïve. However, I used unrealistic assumptions for methodological purposes that Elliot does not recognize, and the paper went to great lengths to emphasize that my account of the market did not depend upon assuming perfect competition or information. Noting markets exhibit a range of informational problems and flaws, I stated: “it must be emphasized that I do not rely upon neoclassical assumptions regarding perfectly competitive markets or perfectly informed consumers” (DeCanio 2014, 639-640).

Hardly offering an idealized account of markets, DeCanio (2014, 640) described market competition in the following terms: “by depicting the market as a quasi-evolutionary discovery process exhibiting imperfect information and imperfect competition, I accept informational critiques of neoclassical models of the market, critiques of the first and second theorems of welfare economics … and … arguments that market prices exhibit imperfect and asymmetrical information.” Hence I did not defend an idealized view of the market, nor was I trying to deny the imperfect nature of information economic exchange exhibits.

In addition to claiming that I employ an unrealistically optimistic account of the market, Elliot also claims that I adopt an impossibly high standard for evaluating democracy, a standard where democracy “is being judged on the basis of whether it optimizes its decisions given all the alternatives open to it with respect to a complete set of preferences which fully specifies how to make all possible tradeoffs” (Elliot 2019, 18). However, this criticism overlooks how I did the exact opposite. Instead of employing high standards to assess democracy, I deliberately structured my argument to minimize and understate the epistemic problems facing democratic politics.

For example, I used Schumpeter’s minimalist definition of democracy, with its emphasis on periodic voter evaluation of elite performance, not because this definition is more realistic, but because it requires so little of voters, merely requiring that they evaluate the incumbent party’s performance instead of engaging in epistemically taxing forms of deliberation. My purpose in doing so was to make the strongest case for democracy, and then show that even under assumptions favorable to democracy, the basic epistemic problem I identified would persist.

While I find nothing wrong with Elliot’s (2019, 3) interest in bringing “a more sophisticated account of democracy to the debate,” and although I agree there are many instances where such accounts “better reflects actual democratic practice,” I never sought to maximize the realism of my argument. Rather unrealistic assumptions were used to identify a fundamental information problem democratic politics exhibits by eliminating the influence of rival causes, such as self-interest or poor information feedback, and then structuring the thought experiments to make the case for democracy as strong as possible by requiring the least of voters.⁴ The basic reason I did so was to show how other variables and explanations of the types emphasized by Brennan, Friedman, and Somin, did not matter for my argument, as one could assume voters faced no incentive problems and confronted a policy problem that had direct information feedback and still show that the basic problem posed by the singularity of political decisions would persist. Instead of structuring my argument this way to unfairly skew the analysis in favor of the market, the lack of realism was used to illustrate how even under a range of favorable conditions, politics exhibits informational challenges that markets mitigate.⁵

In addition to criticizing my use of unrealistic assumptions, Elliot (2019, 19) also claims that like others with “confidence in the progressive character of markets,” I ignore the “unnecessary ruination of people’s lives,” due to the errors and instabilities capitalism produces, such as “ameliorable financial crises.” Yet here again I disagree with how my argument is characterized. I never sought to depict markets in a utopian manner or ignore their various shortcomings, and instead of ignoring problems with financial markets I drew explicit attention to their problems, and explicitly sought to distinguish my argument from those that deny or minimize the information asymmetries Elliot emphasizes. Consider DeCanio (2014, 643) in the following passage:

It is critical to deny the ‘perfect’ functioning of market prices and to emphasize that in certain markets, such as financial markets, the information conveyed by prices is systematically less informative than prices for other goods.

As in several other cases, I believe that there is less disagreement on this point between us than Elliot recognizes, and the ultimate source of our disagreement appears to involve the justifications I employ for using unrealistic assumptions in the first place. Again, my objective was not to idealize markets, but to compare how markets and democratic politics have different implications for actors’ need for knowledge and understanding.

Conclusion

Epistemic democrats’ interest in how different political institutions process information constitutes a promising approach to studying the merits of various political institutions such as various forms of democracy or epistocracy. However, there is no reason why such comparisons must be limited to different political regimes, as they may also involve alternative social and economic institutions as well. Examining how different institutions and organizations influence actors’ need for knowledge and understanding, and how these informational characteristics influence the properties of different competitive systems, offers a potentially fruitful form of comparative analysis. When undertaken from a range of perspectives and theoretical assumptions, examining how rival institutions create or mitigate different epistemic problems constitutes a fruitful approach for a range of social scientific analysis.

However, such comparisons needn’t focus on depictions of knowledge that merely emphasize is diverse or decentralized characteristics, nor should the depiction of knowledge emphasize how explicit understanding of highly complex process through deliberation or learning is the primary means of generating rational outcomes. Indeed, the fact that experts disagree among themselves, and that the social sciences have been unable to generate forms of consensus that experimental disciplines exhibit, indicates that alternative means of conceptualizing how specific institutions may help ignorant actors to rationally orient their behavior despite the ignorance and error they exhibit. As the specific way knowledge problems are defined depends upon the way knowledge is conceptualized, one way of reframing social and political analysis would not focus on whether knowledge is decentralized among many voters, consumers, or firms, but would instead consider how different forms of social and political judgment place greater or lesser demands on actors’ need for accurate models of reality in order for them to act rationally. This approach to political and economic analysis has the additional advantage that basic aspects of its depiction of causality are already widely employed in statistical approaches to experimental inference (Rubin 1974; Holland 1986).

Hardly a consequence of actors’ preferences or maximizing objectives, market rationality is attributed to the inferences that are possible due to competition’s counterfactual choice environments, and actors’ corresponding ability to rationally orient their behavior without having to understand the causal structure of the economy they inhabit. While the market mitigates actors’ need for causal understanding due to the counterfactual comparisons consumers and producers can perform, the singular nature of public decisions exacerbates actors’ need for knowledge, requiring that they understand the causes of social problems and the efficacy of political efforts to ameliorate them.

The counterfactual comparisons rival systems facilitate, and the epistemic properties of the selection pressures these comparisons generate, may in turn help explain key features of both democratic politics and market competition. Indeed, the highly rationalized patterns certain social systems produce, such as the efficiency modern economies generate or the forms of epistemic progress that certain scientific disciplines exhibit, may have similar experimental sources. Illustrating the nature of these sources, their boundaries, limits, and their implications for different spheres of human action, may require a range of institutional comparisons, not just within politics, but across different forms of economic competition, social organization, and in the production of scientific knowledge as well.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 DeCanio (2021) connects the problems created by the unobservable nature of public policy counterfactuals with widely accepted statistical arguments associated with the potential outcomes framework for causal inference (e.g. Rubin 1974; Holland 1986).

2 This argument resembles evolutionary arguments by Alchain (1950), Friedman (1959), and Nelson and Winter (1984) regarding market selection for profitable firms. However, the explanation for how markets generate selection for efficiency, and the links between markets, knowledge, and causal inference, are distinct.

3 Political decentralization in a federal system may generate a greater amount of counterfactual data relative to a monopolized market. However, the key issue involves the counterfactual data different competitive systems generate, and the epistemic consequences such data have for decision makers in each realm.

4 For a discussion of the methodological implications of this approach see Mäki (2006).

5 For historical examples of how political elites implement anti-democratic decisions that failed to achieve their own self-interested objectives see DeCanio (2005; 2011; 2015).