A. Ideal theory in economics and political theory

Examples of ideal theorizing are prominent in economics and political theory. In economics neoclassical welfare analysis judges institutions against the requirements for a Pareto optimal equilibrium. The conditions for a private enterprise economy to fulfill this ideal standard are formally articulated in the first fundamental welfare theorem.^{Footnote 1} These include perfect information; perfect competition — where there are large numbers of buyers and sellers, none of whom can exert a significant effect on prices; zero transactions costs; costless mobility of resources; and the absence of externalities. Many normative debates in economics focus on whether private markets approximate these requirements or whether they depart so radically from the “ideal” that there is significant scope for a non-market mechanism to increase efficiency.^{Footnote 2}

In political theory, meanwhile, ideal analysis is frequently invoked to determine the principles and the regime types that would characterize a fully “just” set of institutions. The use of contractarian thought experiments by thinkers such as Rawls is designed to derive the principles and regime types that would emerge under conditions of “full motivational compliance.”^{Footnote 3} These idealizations are then used as benchmarks against which existing “unjust” institutions can be judged and potentially reformed.^{Footnote 4}

From the perspective of this essay, ideal theories are most useful when highlighting the institutional questions raised by “real-world” situations that deviate from the specified “ideals.” In economics, the model of a perfect market with zero transaction costs is helpful because it enables theorists to understand why hierarchical business firms and dispute resolution mechanisms such as courts may exist. With zero transactions costs all economic activities might be arranged through spot contracts, and any conflicts over resource use could be overcome through private bargaining. It is however, precisely because “real-world” transactions costs are positive that firms and courts may be necessary as institutional responses to cope with these frictions.^{Footnote 5} Similarly in political theory, the ideal of full motivational compliance can be useful when highlighting the potential necessity of the existence of states and their coercive enforcement powers in “real-world” contexts that depart from full compliance, and in providing a rationale for controlling the behavior of those who enforce coercive rules.^{Footnote 6}

Recognizing this analytical function, however, does not imply that ideal theories offer an appropriate standard of how the world can and should be made to function. Neither do such idealizations form a basis for judging between different institutions. On the contrary, in ideal conditions there would be little reason to favor one institution over another. Under the rationality and informational assumptions of neoclassical general equilibrium theory for example, there would be no difference between the efficiency of “free markets” and “socialist central planning,” with both regime types capable of maximizing social welfare under these conditions. Likewise, with full motivational compliance, justice — however defined — would be achievable under any regime type, ranging from anarchism through to democracy and even authoritarianism.

B. Robust political economy and institutional evaluation

Instead of using ideal theory, institutional evaluation should be viewed through a “robust political economy” (RPE) perspective.^{Footnote 7} In economics the analytical task is to account for the relative degrees of coordination witnessed under different regimes in the absence of the rationality and informational requirements of the first fundamental welfare theorem. Similarly, in political theory the RPE approach seeks to explain which levels of coercive state authority can sustain social cooperation when agents are not fully rational, where they disagree about justice, and where they may not comply with various rules. Institutions should not, therefore, be indicted for failing to reach perfect efficiency or full compliance with justice. Rather, what needs to be explained is why we have the level of coordination or cooperation that we do, given the decision-making traits of “real-world” actors — and what can be learned from this experience.

From an RPE standpoint, departures from ideal theoretic conditions reflect certain “non-reformable” attributes of human agents that create the frictions “real-world” institutions must address. A first set of frictions arises because people are cognitively constrained and act under conditions of uncertainty where information is often contradictory and where there is a high propensity to err. This problem is not reducible to one of imperfect information that can be remedied by searching out additional data. Human errors do not simply reflect a rational calculation where it is deemed excessively costly to become fully informed. Rather, they often arise from “radical ignorance” where agents are incapable of “knowing what they do not know.” In a properly robust analysis, therefore, institutions must be judged in terms of whether they enable people to adapt to and learn from unanticipated mistakes as well as unforeseen opportunities, and against their ability to reduce systemic or society-wide error.

A second set of frictions that an RPE analysis must consider are those arising from opportunistic behavior. In their weakest form these may reflect collective action dynamics where the lack of incentive for an individual to incur personal costs when their decisions can make no demonstrable impact on outcomes may produce undesirable results at the macro-scale. In their strongest form they may arise from the desire of some agents to abuse positions of power. Institutions should thus be judged on their capacity to address free rider problems, the generation of third party costs and to constrain power-seeking agents.

Whether one’s evaluation is in terms of efficiency or distributive justice, RPE emphasizes the importance of analytical symmetry. It will not suffice to show how ignorance, uncertainty, collective action problems, unequal power relations, and insufficient generosity lead to “failure” under one regime type while assuming away or downplaying the same problems under an alternative. To establish a robust case, the analyst must explain how and why his or her favored institutional model will be less subject to the problems concerned given its structural features and how these are likely to interact with nonideal agents.

It is important to note here that an RPE evaluative standard can only ever be one that compares different arrangements against each other in their relative propensity to cope with real-world conditions. Institutions should not be judged against a theoretical ideal exogenous to those conditions. This does not imply discarding “idealism,” understood as an account of arrangements that might improve on the status quo. Such analysis will, however, be an endogenous or imminent form of idealism. Notions of what may constitute better arrangements must arise through theoretical speculation and empirical observation of how different institutions address social frictions. The proposals that follow from an RPE analysis might therefore be considered as “nonideal” conceptions of the “ideal.”

C. Robust political economy and the priority of markets

The remainder of this essay uses an RPE standard to judge the case for giving institutional priority to markets and private contract. To clarify what is meant by institutional priority I draw on Knight and Johnson’s distinction between first-order and second-order institutional tasks.^{Footnote 8} First-order tasks refer to how an institution allocates resources. Thus, the first-order task of markets is resource allocation through decentralized private contracting and competitive supply and demand. The first-order role of democracy meanwhile is to allocate resources via deliberation and majoritarian collective action. Bureaucratic hierarchies and courts constitute other allocation mechanisms, each with its own distinctive operating features. Second-order tasks, by contrast, refer to the “meta-level” role of institutions in enabling societies to monitor, evaluate, reform, and choose between different mechanisms, and thus to alter the “institutional mix” within the social order. They allow “an ongoing process of selecting and maintaining effective institutional arrangements by identifying failures, dysfunctions, externalities and coordination problems as well as remedies to them.”^{Footnote 9}

As Knight and Johnson emphasize, an institution deemed effective in this second-order role may not be considered effective as a first-order mechanism, and vice versa. To prioritize democracy over markets and bureaucratic hierarchies as a second-order mechanism, for example, would not imply that all decisions should be made democratically, but that democracy should determine the mix between markets, bureaucracies, and democratic deliberation as allocation mechanisms. Unless one asserts that an institution has first-order priority in all tasks, then the second-order role of monitoring performance and choosing between institutions is crucial when considering the robustness of a political-economic regime.

The view defended here gives “second-order” priority to markets by advocating a basic structure that protects persons and their property against direct forms of coercion such as theft and fraud, and which secures the freedom of persons to exit arrangements they deem relatively unsatisfactory. The primary form of coordination within such an order is one of consensual exchange and freedom of contract. To prioritize freedom of contract does not, however, imply that all first-order allocative tasks should be based on individual bargaining. In some contexts it may be desirable to rely on gift-type relations of the sort found in families or charities. Similarly, there may be advantages in using organizations such as firms, cooperatives, clubs, and mutual associations whose internal practices may be based on hierarchical control, democratic procedures, or informal convention. Prioritizing markets in this context is to favor a regime where the “second-order” task of improving allocative performance and choosing the institutional mix is carried out primarily through a decentralized process of contracting which allows for a “market in institutions.” Business firms, for example, are contractual entities that suppress internal competition but are subject to external market forces, where owner managed firms, joint stock companies, worker cooperatives, and mutual associations all compete for people, sales, and investment capital. Similarly, in private residential communities, individuals contract into a higher-level organization that regulates the freedoms of those entering the governance structure by dividing property rights to address potential externalities and public goods problems. These organizations are subject to competition from agencies that have alternative governance models and/or supply different packages of public goods, and may themselves contract into higher-level structures that manage collective action problems on a larger territorial scale.

A “free market” regime, therefore, is not a “regulation free” zone but one where many of the rules governing conduct are arrived at via competition and private contracting. To prioritize markets does not imply that decentralized decision making is always superior to centralization, but is to maintain that the extent to which centralization is preferable should itself be determined via a process of contractual exchange where people retain the right to exit arrangements they deem less satisfactory and where there is open entry for suppliers of new institutions. Neither is a “free market” confined to the provision of “private” goods, but is better conceived as a tapestry of contractually formed institutions supplying a complex mix of goods with differing degrees of “private-ness” and “publicness.” The role of non-market/non-contractual institutions at the second-order level is confined to the maintenance of a constitutional structure that secures property rights and enforces contracts, to the resolution of disputes when there is doubt over ownership claims or jurisdictional authority, and to a limited number of tasks where competition may not be possible as an organizing principle.

A. Social welfare and the case for central regulation

Though the collapse of post-war socialism has led to an appreciation of market processes, it is widely held that the social welfare case for “unfettered” markets is not in fact robust. Two of the most sophisticated efficiency-related critiques have been advanced by Stiglitz^{Footnote 10} and by Knight and Johnson.^{Footnote 11} These authors suggest that real-world markets rarely meet the optimality conditions specified by neoclassical welfare theory, and while having genuine strengths in comparison to socialist structures, “free markets” are prone to numerous weaknesses creating room for widespread and feasible improvements via government intervention. To use Knight’s and Johnson’s terminology, democratic collective action should have “second-order” priority, either as a device for improving the performance of markets or in choosing to replace systems based on private contracting with alternative allocation procedures.

Stiglitz in particular suggests that the informational properties of markets are overrated and challenges what he understands to be Hayek’s argument for decentralized pricing. Stiglitz takes this to imply that prices communicate “sufficient statistics” that enable people to allocate resources efficiently without needing any information about wider economic conditions.^{Footnote 12} As Stiglitz points out, however, if markets convey this much information then no market actor would ever have an incentive to acquire information themselves preferring to “free ride” on the efforts of others by simply observing prices. Real-world prices tend to operate when participants are able to “hide” information from others, so for Stiglitz they cannot reach a fully efficient equilibrium without supplementary government action.^{Footnote 13}

Stiglitz further suggests that real-world prices are often “too coarse” to perform the indirect communication function Hayek understands them to have because traders cannot tell whether a price change is due to mistaken moves to bid up/down the price of a particular good, strategic manipulation by other traders, or reflects genuine shifts in market conditions. Similarly, absent perfect futures markets, participants cannot determine whether shifts in the terms of trade represent short- or much longer-term changes in underlying resource scarcities. On Stiglitz’s view, then, there is scope for government action to “improve” market prices via targeted measures to better reflect the data “free markets” may not themselves convey.^{Footnote 14}

While recognising that effective competition need not require costless exit and entry with many buyers and sellers, Stiglitz maintains nonetheless that strategic behavior in oligopolistic contexts impedes efficiency and that there are a large number of markets with network or returns to scale effects that render competition ineffective as a disciplinary mechanism creating significant scope for second-order regulation by the state to improve outcomes.^{Footnote 15}

Knight and Johnson concur with Stiglitz’s analysis but maintain that his neoclassical account of “market failure” does not go far enough.^{Footnote 16} Though market failure theory suggests that government regulation may improve upon decentralized competition, Knight and Johnson suggest it downplays the possibility that entirely non-market institutions might be more efficient and thus gives undue priority to markets. Just as hierarchical firms can have efficiency advantages over contractual bargaining, so non-market structures can have similar advantages over market-based supply.^{Footnote 17} For Knight and Johnson, the second-order case for using private contracting to determine the boundaries of markets defended in this essay depends on precisely the conditions required for their effectiveness as a first-order mechanism. To work effectively as a second-order procedure of institutional selection, decentralized competition and private contracting must operate under conditions that approximate a world of perfectly competitive agents with close to full information, conditions under which no actor has room to engage in strategic bargaining to distort outcomes in their favor.^{Footnote 18} “Real-world” institutional competition, however, rarely operates this way, so there can be little confidence it will select the most efficient institutional forms.

Knight and Johnson contend that when there is uncertainty over first-order effectiveness democracy provides a better second-order mechanism for experimenting with different institutional mixes and for monitoring the conditions in which they operate.^{Footnote 19} Democratic procedures, and especially those based on “democratic centralism,” facilitate experimental learning and adaptation. By forcing people to assert, defend, and revise their own views in order to get things done, they encourage a higher level of reflexivity. Decentralized competition by contrast allows agents to exit circumstances they disapprove of, thus reducing scope for persuasion, deliberation, and learning from others.^{Footnote 20}

Relatedly, Knight and Johnson suggest that democracy is uniquely placed to evaluate institutions from a meta-level perspective owing to its capacity to monitor where and when the ideal conditions required for other institutions to function well pertain. By contrast, market participants pay no heed to how their own behavior or that of actors within the wider market may lead to deviations from efficiency.^{Footnote 21} Knight and Johnson thus propose a radical centralization of decision making at the second-order level, where the entire institutional structure of society — the relative extent to which families, charities, private contracting, firms, and bureaucratic hierarchies are utilized as allocation mechanisms — is subject to a process of collective, democratic control.

B. Social welfare and market prices

The above critiques represent sophisticated variants of long-standing concerns over the social welfare properties of markets. On the view advanced here, however, they depend on an inappropriately ideal standard and/or show a failure to demonstrate the analytical symmetry required for a robust comparative evaluation.

Consider first Stiglitz’s contention that the “coarseness” of “real-world” prices leaves significant room for second-order government interventions to improve efficiency. The assumption here is that free market prices should transmit full information if they are to be favored. In conditions of chronic ignorance however, there is no way for the price system or any other institutional device to transmit full information, so why should markets be judged against this ideal?

Stiglitz partially recognizes this point when noting that: “This pervasiveness of failure, (as compared to competitive equilibrium) while it reduces our confidence in the efficiency of market solutions, also reduces our confidence in the ability of government to correct them.”^{Footnote 22}

If the requirements of perfectly competitive equilibrium cannot be achieved either by unfettered markets or government fiat, however, then why use this standard to single out “market failures”? The issue must always be one of “failure” compared to some available “real-world” alternative, and not in comparison to an unattainable and unidentifiable ideal.

The strongest argument for “free markets” is not that prices provide surrogates for perfect information, but the modest comparative institutions claim that they communicate relatively more information than a centrally directed alternative. In conditions of bounded rationality Hayek’s suggestion is that no central agency can spot and respond to as many “gaps” in the economic environment as effectively as a multitude of agents with the freedom to exit from and enter into different commercial transactions. Market prices are always generated under conditions where knowledge of exchange opportunities is dispersed in an uneven manner, but as people act on their private perceptions of these opportunities the subsequent generation of profits and losses brings about a gradual — not an instantaneous — coordination process as the various “bits” of data are communicated. The Stiglitzian account of “free riding” on equilibrium prices assumes that traders cannot profit from their information before it becomes generally available. In equilibrium, however, there is no scope for any trade to occur. On the Hayekian understanding by contrast, disequilibrium prices enable market actors to spot profit opportunities and to spread information by taking advantage of them. In other words, imperfect market prices fulfill the critical second-order task of enabling social actors to identify failures and instances of dis-coordination, facilitating incremental improvements.

Knight and Johnson share Stiglitz’s belief that the efficacy of private contracting depends on perfect futures markets populated by clairvoyant agents, but this view is mistaken. The case in their favor is that current market prices (profits and losses) make it relatively easier for people to formulate conjectures about the future than would be the case without such prices. Entrepreneurs combine current price information with specialized knowledge of non-price data in attempting to understand the direction of the market. Their subsequent forecasts are then tested against rival hypotheses through profit and loss accounting with the most successful speculators exercising more influence over the market in future rounds of investment. Entrepreneurs cannot simply “free ride” on market prices because price information, though useful, must be interpreted in light of more specialized know-how. The relative price structures emergent from entrepreneurial competition enable a “rough and ready” coordination as those lacking specialized knowledge of particular markets can adapt to changing supply and demand conditions of which they are largely ignorant. This process of coordination is “coarse,” but absent omniscience it could hardly be otherwise. The alternative of directing the market from the center is likely to be even “more coarse” because it will replace the competitive speculation of diverse entrepreneurs with legally enforced price-setting measures. If these measures are based on the erroneous conjectures of regulators, they will have negative consequences for the entire market, rather than merely a part of it.

It must be emphasized that these claims cannot be established against an objectively independent measure. It makes little sense to judge markets or non-market mechanisms in terms of “how close” they get to what would occur in a full information equilibrium because that implies prior knowledge of the optimal pattern against which the performance of “real-world” institutions must be measured. Rather, the argument follows from a structural explanation of how a system based on dispersed though unequal ownership of property may facilitate a degree of economic adjustment when knowledge of what efficiency requires is not objectively “given” to decision makers. This is combined with empirical observation of (i) the chronic surpluses and shortages characteristic of regimes that restrict private property rights and the price system compared to those that allow freer pricing systems to function, and (ii) the tendency for regimes with more liberal pricing systems to generate higher levels of economic growth.^{Footnote 23}

A robust case for second-order intervention must offer a similar structural mechanism to explain how a system regulating prices from the center can deal with real-world uncertainties more effectively than a free enterprise regime. Those positing significant scope for feasible government interventions, however, fail to explain how regulators are to overcome this “knowledge problem.” Stiglitz is forced to admit as much when noting that “a full corrective policy would entail taxes and subsidies on all commodities, based on estimated demand and supply elasticities for all commodities (and all cross elasticities). The practical information required to implement the corrective taxation is well beyond that available at the present time.”^{Footnote 24}

A “full corrective policy” at any time, it should be noted, would require omniscience when it is the problems arising from ignorance that are at stake. Even the possibility of a “partially corrective” policy assumes that regulators can judge whether their decisions are moving toward full efficiency. Real-world regulators, however, must operate with imperfect knowledge where their decisions are not subject to simultaneous competition from agents with rival interpretations and where they have no equivalent of profit and loss accounting to weed out bad decisions. Consequently, these agents may affect the operation of markets but have little endogenous measure of whether their decisions improve resource allocation in comparison to alternatives. There is, then, little reason to believe there are widespread and feasible second-order policy interventions that can “improve” on a “free market” price system.

C. Social welfare and spontaneous order

Similar problems afflict Knight’s and Johnson’s contention that markets “fail” the second-order task of holistic monitoring of societal decisions. Knight and Johnson proceed as though “ideally” people should know how their decisions mesh with macro-level outcomes, but the strongest case for prioritizing markets is that social coordination involves cognitively constrained agents who cannot be fully aware how their actions affect macro patterns. Where social wholes are more complex than the sum of their individual parts, people are necessarily ignorant of the “whole picture.” The comparative advantage of market competition as a second-order institution is that it facilitates mutual adjustment in spite of the cognitive limitations of all concerned. So long as production models that meet consumer demands make profits that signal the need for imitation, and losses signal errors, then the process of “rough and ready” coordination requires minimal holistic awareness. Similarly, so long as people can exit organizational rules that fail their requirements and enter others they judge more effective, a spontaneous process may adapt the pattern of rules and reconfigure the institutional mix, without any single authority having to be cognizant of all the margins for improvement. Adaptation to error is likely to be speedier than in a centralized equivalent because actors may exit what they perceive to be misguided structures without first having to secure approval from an overarching authority or majority.

Market processes are undoubtedly “imperfect,” but Knight and Johnson fail to explain how “democratic centralism” would generate a more holistic picture than decentralized competition. Politicians and regulators may be unable to anticipate the systemic consequences of the measures they favor but will nonetheless be empowered to enforce them on unwilling parties. Consider here, the case of financial governance. In a comprehensive account of the dynamics preceding the recent crisis, Friedman and Krauss suggest that it was precisely the inability of policy makers to understand the effects from a maze of interacting regulatory responses to democratic pressures that precipitated the near-collapse of the global financial system.^{Footnote 25} These included the decision of monopoly central banks to set interest rates below what may have been justified by private savings ratios; the regulatory and fiscal inducement of government-backed mortgage companies to relax lending requirements for home purchase to low-income families; internationally enforced capital regulations which induced banks to securitize risky mortgages; and the creation of legally protected monopolies in the credit rating business such that the financial success of these agencies was not dependent on the quality of their risk assessments but on immunity from competition. Competitive processes may themselves generate systemic failures owing to the tunnel vision of their participants, as also occurred prior to the financial meltdown. Absent any theoretical or empirical account of how “democratic centralism” would reduce rather than magnify such failings, however, then a second-order case for prioritizing these mechanisms has not been established. Empirical evidence here is far from clearcut, but longer-term analysis on the incidence of systemic financial crises (as distinct from recessions or downturns) suggests that the propensity to such events has coincided with periods of regulatory centralization and not with those of relative laissez faire.^{Footnote 26}

D. Social welfare and competition

What of the contention that second-order government regulation can improve on “imperfect” competition? Knight and Johnson maintain that “markets work at their best” when characterized by “price-taking” behavior with no scope for strategic action.^{Footnote 27} In nonideal conditions with unevenly distributed knowledge, however, strategic action should not be seen as an anti-competitive “imperfection” but as the very essence of competitive behavior. When there is uncertainty about the best production methods and organizational forms, then price-cutting, price discrimination, marketing campaigns, and other strategic activities that create advantages for some organizations over others are the competitive means by which innovations are gradually discovered and diffused.^{Footnote 28}

By contrast, “under ‘perfect conditions’, advertising, undercutting, improving (differentiating) the goods or services produced are all excluded by definition – ‘perfect competition’ means indeed the absence of all competitive activities.”^{Footnote 29}

Stiglitz maintains that in sectors with network effects or returns to scale rivalrous competition is ineffective, yet in fields such as telecommunications and information technology that have these characteristics, competition is often vigorous. That incumbent firms engage in large research and development expenditures is testament to their constant need to fend off attacks from innovative challengers.^{Footnote 30}

In nonideal conditions, the alternative to “price-making” environments is not perfect competition but “price setting” or regulation of industry structure by a monopolistic authority that may reduce competition as centralized control of prices and profits may dampen the signals and incentives for would-be challengers to enter markets. The experience of “antitrust” law, for example, suggests that rather than enhancing competition, regulation often acts as an impediment to it. In many cases, firms whose practices have lowered prices and increased output have been subjected to suits launched by rivals, and there is little evidence to suggest that antitrust works systematically to increase entry.^{Footnote 31} Where knowledge of efficiency requirements is not “given” to a regulator, it has no obvious mechanism to determine which prices and profits represent undue “market power” arising from network effects and other “imperfections,” from those that reflect better entrepreneurial foresight in conditions of imperfect knowledge.

This “nonideal” case against second-order centralization also undercuts Knight’s and Johnson’s suggestion that democracy should have priority in determining when other allocative mechanisms meet “ideal” efficiency conditions. Their argument presupposes that democratic actors can know which institutional mixes will bring about efficiency when it is the absence of such information that is the problem. A process allowing simultaneous experimentation between rules and organizational practices may be more likely to generate information about the appropriate mix of governance structures because it allows more models to be tested. Thus, deciding whether families should engage in food preparation and childcare themselves or should “contract out” these functions to commercial agencies, not-for-profits, or the communal arrangements of a kibbutz will be aided where people can observe alternative models in action and not where a majority of families are able to enforce their views on dissenting minorities. Similarly, knowledge of whether environmental goods are better supplied via individual contracting, communal control, or hierarchical management is more likely to arise in a polycentric order where people can observe differences between alternative governance arrangements.

Contra Knight and Johnson, the case for private contracting in such matters of institutional selection does not depend on full information and perfect competition between institutional models. What it requires is “freedom of entry,” where alternative governance arrangements can be offered without needing approval from an overarching hierarchy or majority. As the unit of competition increases in territorial scale (as may be the case with some larger-scale public goods problems) the range of governance structures that may compete simultaneously in supplying these goods will be reduced, and there may be lesser scope for new entrants to challenge established models. This hardly counts, though, as a decisive objection to the argument here because, absent legal barriers to entry, institutional competition will be maximal. Relative to such a process, democratic centralism will reduce competition by limiting experimentation to a consecutive process where a fixed set of arrangements is imposed on all actors over a period of time.

Knight’s and Johnson’s claim that giving second-order priority to markets requires perfect transparency and no external costs is also fallacious. Relative to democratic centralism, people in markets may have stronger incentives to reduce information asymmetries and externalities because the costs of failing to become adequately informed about purchasing or locational decisions are concentrated on them. Under “democratic centralism” these incentives may be much weaker because outcomes are determined by how everyone else votes in a context where the chance of any individual vote affecting the result is vanishingly small.^{Footnote 32} Similarly, in a competitive environment with exit options, people have at least some scope to “contract around” the external effects of other people’s behavior by entering arrangements that reduce unwanted interferences. By contrast, if people cannot, save for leaving their country, exit relationships with politicians and regulators, they will be subjected to massive collective action problems and exposure to externalities generated by the political activities of others.^{Footnote 33} In comparative terms, therefore, democratic centralist processes may be less transparent and more prone to externalities than the markets over which Knight and Johnson want them to have priority.

The forgoing analysis is well supported by a variety of evidence. Empirical work on markets and institutional competition indicates that even agents with minimal information and bargaining power are more informed about their choices than participants in most large-number elections.^{Footnote 34} Historically, a wide range of authors have suggested that it was the political anarchy between competing states and jurisdictional authorities such as churches, monarchs, and merchants that reigned across medieval Western Europe that acted as a laboratory for institutional experimentation and constrained possibilities for elite predation. Notwithstanding massive background inequalities with the vast majority excluded from political power, competitive dynamics pushed elites toward less predatory governance and facilitated the unprecedented economic expansion of the industrial revolution.^{Footnote 35} The highly centralized structures in Russia and China, however, appear to have stifled innovation and economic progress. More recently, Ostrom’s work on common-pool resources indicates that effective solutions to small- and middle-range public goods problems are more likely to be discovered and disseminated in a context of “parallel adaptation,” where no authority has the capacity to impose a particular set of governance rules. When the range of institutional devices such as individual property, communal, or club property and various mixed regimes is generated via a bottom-up process where the smallest units have the freedom to “contract up” their authority, this increases the chance of creating governance structures that internalize costs. Top-down imposition of institutional rules is, by contrast, correlated with a greater propensity to systemic failure.^{Footnote 36}

Only in a small number of instances where simultaneous competition is impossible might there be a strong case for a collective choice mechanism to have second-order priority. Obvious candidates here include large-scale transboundary issues such as anthropogenic climate change that may preclude the possibility of internalizing costs at any institutional level lower than the global scale. There is, for example, a strong nonideal case for a centralized mechanism to impose a global tax on carbon emissions which would force polluters to pay for environmental damages while retaining the scope for lower-level institutions — whether nations, communities, firms or families — to choose the means of adaptation to the tax in question.

Though the case for democratic centralization is undoubtedly strongest in these instances, it is by no means clear-cut. The tragedy of climate change is that while failing to institute a global mechanism may generate significant economic and social costs, the creation of a global enforcement institution would not be without significant risks of its own — such as, for example, mistakenly imposing a tax at levels that will generate higher economic and social costs than the effects of climate change itself. Beyond a certain point of centralization, the ability to minimize the consequences of mistakes, to learn from other decision-making nodes and to adapt speedily to changing conditions may be so blunted that the scope for systemic failure may negate any benefits that central regulation might in principle bring — even in the absence of a credible decentralized alternative.^{Footnote 37}

The foregoing analysis has presented a nonideal social welfare case for giving markets and freedom of contract second-order priority. Although few contemporary societies suppress markets in the manner of post-war socialism, many do engage in democratic centralist measures that limit competition and freedom of contract. In fields such as financial services, education, health care, energy supply, and environmental regulation, the scope for people to craft contracts and the price structures emergent from them is often constrained by top-down attempts to “improve” on the market outcome. Similarly, while few societies eliminate institutional competition, even some of the more decentralized political systems limit the scope for citizens to develop governance structures to shape markets and to address collective action problems from the “bottom up.” While it cannot be said that the current set of governance arrangements is “inefficient” judged against an objective standard, the RPE analysis presented here suggests that relative to a structure prioritizing markets, the status quo in many societies is not well placed to allow for the discovery and communication of efficiency enhancing moves. Far from improving on the status quo, the proposals of Stiglitz and Knight and Johnson for still greater central intervention seem, in the vast majority of cases, more likely to reduce social welfare.

A. Distributive justice and the case for central regulation

Ethical critics of classical liberalism do not disregard these arguments but contend that assigning them undue priority neglects the moral status of the bargaining terms on which people make their choices. I concentrate here on Rawlsian arguments and those of Knight and Johnson.

For Rawls, a just society is one whose institutional rules reflect impartial principles that people would willingly support, rather than those they may be forced to accept owing to a lack of bargaining power. Rawls deploys the theoretical device of the original position to model an impartial choice procedure that neutralizes the effects of inequalities that might bias the beneficiaries of genetic and social lotteries when judging the basic structure of social rules. Agents behind the “veil of ignorance” know nothing of their own attributes and imagine themselves to be choosing rules for a “closed society” to prevent those with potentially greater bargaining power from exercising bias in their deliberations. Rawls’s initial account suggested that impartial deliberators behind the veil of ignorance would converge on the components of “justice as fairness”; the principle of equal liberty, fair equality of opportunity, and the difference principle — the latter requiring that inequalities work to the maximum benefit of representative agents from the least advantaged class. In his later work however, Rawls abandoned the notion that deliberators would opt for justice as fairness from a common evaluative standpoint suggesting instead the possibility of an “overlapping consensus” between agents with differing though “reasonable” worldviews.^{Footnote 38}

Central to the Rawlsian account is the necessity for coercive measures to “guarantee” fairness in a society’s basic structure. This includes provision of a social minimum and the supply of publicly provided or subsidized education to ensure fair equality of opportunity. It also requires a “distributive branch of government” that preserves approximate justice by means of taxation and necessary adjustments to the rights of property.^{Footnote 39} In Knight and Johnson’s terminology, Rawls envisages a central “second-order” role for the state in monitoring and correcting the background distributive conditions in which other institutions operate. It is the absence of such mechanisms under classical liberalism that for Rawls disqualifies it from the family of “reasonable” worldviews. Under a regime prioritizing freedom of contract, the more advantaged might exit their obligations to provide sufficient support to the disadvantaged, or they might provide support in a manner that demeans the social status of the poor as free and equal citizens. By contrast, if one assumes full motivational compliance, Rawls’s favored regime-types of liberal socialism and property owning democracy, represent potentially just structures because they have an extensive coercive apparatus designed to guarantee fairness.^{Footnote 40}

In their account of the second-order role of democratic centralism, Knight and Johnson emphasize the significance of disagreement about the requirements of justice and/or the extent to which justice should be prioritized over other objectives such as economic growth and environmental protection. For Knight and Johnson, the strength of democracy rests on its ability to organize experiments with alternative social models and to reduce conflict. By ensuring that divergent voices are heard, democratic mechanisms facilitate learning and can sustain support from those losing out from any specific decision. Prioritizing private contracting, on the other hand, would produce outcomes reflecting differential bargaining power and would thus lack legitimacy in the eyes of weaker parties.

Knight and Johnson emphasize that to fulfill its potential democracy must be supplemented with “institutional guarantees” ensuring all citizens can exercise maximum equal influence. State organizations must intervene directly in private and civil associations to reduce material inequalities and power imbalances and to secure “free and equal participation.” These proposals are similar to those of a Rawlsian “property-owning democracy” which seeks to limit the scope for those with greater bargaining strength, such as private employers, to interfere with the political choices of weaker parties. If people are guaranteed sufficient income and job security such that they need not fear unemployment or loss of income, then democratic processes may approximate a free and uncoerced process of deliberation.^{Footnote 41} Income redistribution alone, however, may be insufficient if the prevailing pattern of norms, such as those pertaining to gender, prevents some people from using material resources effectively — by, for example, limiting access to high status jobs or discouraging political participation. For Knight and Johnson, therefore, guaranteeing effective participation requires intervention in the workplace, in the family, in the education of young children, and even in religious beliefs and practices to prohibit social norms that reduce prospects for democratic equality.^{Footnote 42}

B. Justice, markets, and constitutional ignorance

With its claim that classical liberalism subjects weaker parties to unjustifiable bargaining terms, the Rawlsian approach departs most significantly from the RPE framework. The assumption is that knowledge of what full compliance with justice entails is objectively attainable and can be used as a benchmark from which to construct a more just set of arrangements. The problem of “real-world” justice, however, arises because social interaction takes place in a context of fragmented and contradictory knowledge where there is no objective way to discern “fair terms of cooperation.” Consequently, it is no more plausible to evaluate institutions against full compliance with justice as fairness than it is to judge them against full compliance with efficiency. It is not clear, therefore, why classical liberal institutions or indeed any other institutions should be evaluated against this ideal.

The strongest ethical case for giving second-order priority to markets is that relative to alternative systems they may be better placed to accommodate rival interpretations of fairness. A system limiting direct coercion to the enforcement of a clear private domain where people can enter into or refrain from various exchanges is hypothesized to reduce conflict relative to alternatives where no rules are enforced, or where the rules of ownership are so uncertain that people can have no confidence their decisions will be respected. The purpose of these institutions is not to secure distributive justice per se but to provide space for the discovery and communication of alternative terms of cooperation in conditions where no central agency or group can be aware of how to deliver fair terms of exchange. The “real-world” alternative to prioritizing competition and private contracting is not a basic structure that “guarantees” fair cooperation but one that empowers a monopolistic authority to enforce what it considers the right pattern of distribution, however erroneous such a view may be.

Assuming that, for example, there is an overlapping consensus supporting the difference principle or fair equality of opportunity, the question of which patterns of resource allocation reflect these principles, remains. Does justice require cash transfers to the disadvantaged, or might it be achieved by maintaining the conditions under which the poor are more likely to secure employment? Which educational models are better placed to reveal the talents people have and to increase the opportunities available to them? If it is to discover how to achieve distributive fairness under conditions of ignorance and uncertainty, the basic structure of society must allow experimentation with and learning from, different welfare models. Though it cannot guarantee that the relevant principles will be secured, a system based on market exchange, one-to-one giving, and voluntary associations may increase the chance of securing “more just” outcomes over time. When there is limited and contradictory knowledge, the suggestion by Rawls that the absence of compulsory redistribution constitutes injustice is inappropriate. It assumes the possibility of a political-legal apparatus populated by omniscient regulators who can manipulate the basic structure of social institutions to produce “justice.” In “real-world” conditions, however, the knowledge of which practices fulfill these principles cannot be centralized in a single institution, and indeed such knowledge may not even exist outside of a competitive environment where contradictory interpretations can be tested against each other in response to shifting circumstances of time and place.

Unlike Rawls, Knight and Johnson do not assume political-legal omniscience but see democracy as an experimental second-order procedure to discover appropriate terms of association under conditions of heterogeneous and dispersed knowledge. In comparison to a constitutional structure that prioritizes markets and private contract, however, the nature of this experimentalism will be unduly constrained. Knight and Johnson believe that “democratic centralism” can mobilize dispersed knowledge by ensuring that the widest possible range of viewpoints is reflected in the resource allocation compromises that result. Yet this claim is misguided because such a process is heterogeneous only before the relevant decisions are made. The generation of knowledge and scope for learning will be stunted because once a government has been elected, people will be unable simultaneously to test and to learn about the effects of alternative viewpoints.

Now, it might be argued that scope for learning is not so limited because politicians and regulators can compare the results emergent from the policies of different nation-states. Note, however, that this response acknowledges the priority of simultaneous competition over the democratic centralism favored by Knight and Johnson. It is the absence of the simultaneous supply of alternative models that stifles the capacity to learn from a wider range of experiences. Similarly, unless people can move into those jurisdictions they find more palatable and away from those they find less so, signaling mechanisms will be blunted. Within this context, there is little reason to privilege the nation-state as the primary unit of experimentation. A model of competitive federalism may facilitate the testing of a greater range of welfare models and create more room for exit. If experimentation is enhanced under federalist structures, there is also a strong case for a more radical decentralization of distributive decisions down to individuals and voluntary associations, acting through the mosaic of institutions that constitute the market and civil society — though individuals acting through these institutions should have the liberty to “contract up” welfare provision if they judge relative centralization more efficacious.

C. Compliance, monitoring, and power relations

Discussion of these epistemic issues must be coupled with consideration of how to control those charged with enforcing coercive rules and the extent to which the relevant rules are open to abuse by opportunistic agents.

Rawls departs from the standards required for a robust evaluation by theorizing away enforcement problems with the assumption that the coercive mechanisms available to state agencies will “guarantee” justice under “full motivational compliance.” Rawls views state action as solving an assurance problem equivalent to deciding “rules of the road,” where agents follow the rules so long as they know that others will do the same. In such cases public authority may be needed to specify the rules, but once they have been chosen it will be in the interests of all parties to comply. When the rules in question involve the empowerment of some actors to redistribute resources and to regulate commercial and civil life, however, then it seems more appropriate to view adherence to Rawlsian rules as generating multiple “compliance problems.” Compliance problems arise when it may not be in the interests of an individual or a group to adhere to the rules and where there may be a need for effective monitoring procedures, not only for those subject to the rules but also for those charged with enforcing them. Rawls, however, pays little attention to whether his favored regime-types would in “real-world” conditions lead to abuses and how these abuses might undermine a cooperative social order.

The possibilities for the abuse of state power highlighted above follow from the huge informational burdens and compliance costs that regimes reserving a significant regulatory and distributive role for public bodies, place on their citizens. The lack of clarity over which policies fulfill the difference principle, for example, makes it difficult, if not impossible to hold accountable those politicians, regulators, and citizens who fail to act in accordance with this principle. It is hard to tell on balance whether the institutions of the modern regulatory and redistributive state work to improve the material position and bargaining power of the least advantaged or whether they weaken that position. Contemporary welfare state models of capitalism involve a significant amount of “downward” redistribution through direct income transfers but they may also reduce economic growth and lower the absolute incomes and bargaining power that the disadvantaged may command in the longer term.^{Footnote 43} In addition, they frequently involve large-scale “upward” redistributions such as the massive transfers to the banking sector that followed the financial crisis. Regulation in fields such as land use planning, occupational licensure, and energy markets meanwhile often reflects rent-seeking behavior and works to raise living costs for the disadvantaged while enhancing the position of incumbent property owners and incumbent firms.^{Footnote 44}

Even when the balance of interventions improves the position of the least advantaged within the boundaries of the state, it may worsen the bargaining terms of those thereby prevented from entering such a state. Evidence suggests that the extent of mandatory wealth transfers correlates with more stringent controls on entry. To avoid excessive disincentive effects on the successful and to maintain transfers to current recipients, some of the most disadvantaged agents in the world may have their option set curtailed via more restrictive immigration controls.^{Footnote 45} If, as Rawls’s egalitarian critics suggest, impartial justice requires paying due attention to the disadvantaged beyond the boundaries of any one state,^{Footnote 46} then by raising barriers to entry mandatory redistribution may be in direct conflict with this objective. Indeed, on a classical liberal view immigration controls represent one of the most egregious abuses of state power because they actively block attempts by the least advantaged to improve their position, and they prevent mutually beneficial interactions between consenting parties.

Knight and Johnson show some awareness of the problems that enlarging state power may induce, but maintain that democracy provides the best antidote to these ills. Thus,

This response is entirely unsatisfactory, however, because it is the inability of democracy to perform the second-order role of “self-monitoring” under the massive interventionist state Knight and Johnson favor, that is the issue. On the one hand, national majoritarianism will do little for the interests of those outside the boundaries of the states concerned. On the other hand, “democratic centralism” within nation-states will generate enormous monitoring costs for voters in deciphering whether redistributive and regulatory interventions are being used to public advantage or for the benefit of sectional interests. Voters will have minimal incentives to be well informed about the effects of interventions because the chance that their personal decision to become informed will affect outcomes is infinitesimally small. It will not, therefore, suffice to blame existing power imbalances on wealth inequality in contemporary welfare-regulatory states. Asymmetric information associated with voter ignorance would introduce significant bias into “democratic centralist” structures even when wealth is more evenly distributed. Evidence on political participation across contemporary social democracies shows no significant differences in participation rates between higher and lower income and wealth inequality societies.^{Footnote 48} What is consistent both within and across such societies, however, is the high level of political ignorance. Irrespective of income, educational level, and social class, the vast majority of people are ignorant of even the most basic political information, and so they are rarely able to determine how far political power is being abused.^{Footnote 49}

These problems are significant enough in the context of contemporary democratic welfare states. Regulatory/welfare state institutions have not arisen via a bottom-up process of private contracting, but through one that has prioritized the decisions of national majority coalitions that reduce exit options. Such options would, however, be further reduced under the still more centralized structures proposed by Rawls and by Knight and Johnson. Under Rawlsian “liberal socialism,” the absence of significant private ownership and the concentration of power in state agencies would further reduce exit options and produce not only greater inefficiency but also huge scope for corruption, rent seeking, and abuse of the liberties of those dependent on the relevant state bodies. The post-independence political economy of India between 1947 and 1991 is probably the closest approximation one can find to “liberal socialism” — and this was a regime characterized by predatory state licensing boards (the “permit raj”), predatory state monopolies, upward redistribution from agricultural communities to urban elites, and political corruption on a massive scale.^{Footnote 50}

Similarly, under “property owning democracy,” the powers to break up private businesses and to alter prices when these are deemed by officials as incompatible with justice would open significant opportunities for politicians and regulators to engage in redistribution and regulation that would favor their own interests and those of client groups. Far from reducing power imbalances, these structures are likely to privilege the politically connected and those rent-seeking interests better placed to overcome collective action problems. Moreover, allowing regulators to be the arbiters of which beliefs and traditions in civil society cohere with “free and equal participation” as Knight and Johnson propose, does not seem compatible with treating those who subscribe to “non-approved” beliefs and traditions as “free and equal.” Far from reducing conflict, these proposals seem more likely to intensify it, as the adherents of different traditions will compete for control of the political apparatus to ensure that their particular practices are the ones publicly approved.

From a comparative standpoint, while it will not eradicate unequal power relations, a social order that prioritizes markets and freedom of contract may reduce the scope for powerful state and private agents alike to capture and abuse positions of coercive political authority. While it would welcome voluntary “downward redistribution,” a regime giving priority to freedom of contract would limit the scope for “upward redistribution” by containing state power to the enforcement of agreements between private parties and civil associations. Correspondingly, by lowering costs of compliance, such a regime may encourage agents to open themselves to mutually advantageous cooperation with agents beyond the confines of national borders, relative to those regimes with mandatory redistribution in their basic structure. The costs of discerning whether private and public actors abide by the terms of classical liberal justice, while nontrivial, would be more limited than those required to monitor the massive state apparatus envisaged by theorists such as Rawls, and Knight and Johnson. Prioritizing markets and private contract may make it relatively easier to discern transgressions, to hold perpetrators to account, and to sustain a cooperative venture for mutual advantage, on the widest possible scale.