Why Size Matters: Property-owning Democracy, Liberal Socialism, and the Firm*

GIANT firms like Walmart and Amazon occupy a central role in modern capitalist economies. Generally, observers have viewed the rise of these firms as the triumph of more efficient technologies and techniques that increase consumer choice and reduce prices. But recently there have been growing concerns with aspects of these firms, in particular the highly unequal way in which their immense capital is owned, and the lack of power and status they afford their workers. Drawing on John Rawls’s later work, several liberal and republican political theorists have argued that these types of problems require the radical solution of replacing welfare-state capitalism with an alternative system of ownership. However, there is ongoing disagreement about what that alternative should be, and the types of firms it should include. Some call for a ‘property-owning democracy’ (POD), in which there is widespread and roughly equal private ownership of capital, claiming that this would increase workers’ ability to exit firms, thus rendering persisting workplace hierarchies benign. Others argue for

a 'liberal socialism' (LS), in which firms are owned (or leased) and democratically controlled by their workers. 6 Despite these disagreements about the ownership and control of firms, contributors to this debate all share one surprising thing in common: they either ignore or reject distinct concerns about the size of firms. 7 In this article, I argue that this consensus overlooks something fundamental: regardless of whether firms are owned by a large number of small investors (as in a POD), or owned/leased and controlled by their workers (as under LS), and regardless of their internal structure, giant firms like Walmart and Amazon pose a credible threat to democratic principles that both liberals and republicans should endorse. Moreover, this threat persists even if laws prevent such firms from raising consumer prices, and we assume favourable motivational conditions. Consequently, the existing normative debate about ownership requires supplementation by a more detailed comparison of strategies to deal with the political power of giant firms.
To make this argument, I shall integrate insights from two empirical literatures into the normative debate about ownership. First, I will draw upon hierarchical theories of the firm from institutional economics. Although some political theorists have begun to utilize these theories, they have done so primarily to evaluate the internal structure of firms. 8 In contrast, I will combine them with recent work by neo-Brandeisian scholars that highlights how firms with large market shares can have troublesome power, even if they cannot raise consumer prices. This work has so far received little attention from political theorists. 9 For rejections, see Hsieh, 'Work, ownership, and productive enfranchisement', p. 152; Schemmel, 'How (not) to criticise the welfare state', pp. 404-5; Schweickart, 'Property-owning democracy or economic democracy?', p. 205; Williamson, 'Who owns what?', pp. 445-6. Other theorists remain silent, with two partial exceptions: Thomas and Edmundson reference antitrust measures among their solutions to firm power; e.g. Edmundson, John Rawls, pp. 152-4; Thomas, Republic of Equals, pp. 153, 157, 161. However, they also retain a role for large-scale firms; e.g. Edmundson, John Rawls,pp. 42,153; William Edmundson, 'What are the "means of production"?', Journal of Political Philosophy, 28 (2020), forthcoming; Thomas, Republic of Equals, pp. xxi, 331, 363-5. It is therefore unclear whether their antitrust measures follow the mainstream approach of tackling market power, rather than firm size (and its associated problems) as I define it in Section I. Hence, it is an open question whether their overall policy proposals offer a viable solution to the central problem raised in this article. 8  Hence, although liberals and republicans have the normative resources to accommodate this analysis of giant firms, they have not yet traced its institutional implications for a POD or LS.
My argument proceeds in four steps. Section II argues that giant corporations would enjoy significant political power in a POD. Section III highlights three ways in which giant corporations would pose a credible threat to democratic principles in a POD. Section IV argues that similar problems would apply to giant worker-managed firms in a POD or under LS. Section V considers two objections, before Section VI concludes by highlighting several potential policy solutions that deserve further examination.

I. KEy CONCEPTS
Before embarking on my argument, I must clarify two key concepts: 'giant firm' and 'democratic principles'. By a 'giant firm', I mean a firm that (1) enjoys the type of market share ordinarily associated with monopoly or oligopoly, and (2) has a large financial value in absolute terms. The second criterion excludes firms that have insufficient resources to engage in meaningful political activity regardless of their market share. The first criterion requires two elaborations.
First, whether a firm satisfies this criterion will depend on how one defines the relevant market. For example, the market for private healthcare might be diluted at the global level, oligopolistic in certain countries, and monopolistic in certain localities. Given these complexities, antitrust authorities generally define a relevant market as one where goods are substitutable from the consumer's perspective, because this indicates that firms are in direct economic competition. 10 However, because my focus is political rather than economic competition, I view a relevant market as one that is typically governed by distinct laws. To see how the two standards differ, imagine a world in which private healthcare markets were governed only by laws that apply at the national and international levels. In this world, if a provider enjoyed a monopolistic market share at a local level, then they would be a concern for antitrust authorities, but would not necessarily satisfy my first criterion.
Second, my focus on market share, rather than market power, draws on recent work by neo-Brandeisian scholars. Mainstream economics views a firm as having market power if it can raise its prices for a sustained period without losing all of its sales. 11 Since the 1980s, the dominant view in both US and European Union competition law has been that a high market share is only problematic if it leads to exercises of market power in this sense because this decreases consumer 10 Lorenzo Coppi and Mike Walker, 'Substantial convergence or parallel paths? Similarities and differences in the economic analysis of horizontal mergers in U.S. and EU competition law', Antitrust Bulletin, 49 (2004) welfare. 12 However, neo-Brandeisian scholars have detailed how firms with large market shares can have troublesome power even if they decrease consumer prices. 13 For example, Walmart and Amazon both gained large market shares by introducing new technologies and production techniques that lowered consumer prices. yet, over time, they amassed resources that furnish them with troubling power over smaller (potential) economic competitors 14 and smaller or more diffuse (potential) political competitors. 15 Although competition law aims to prevent these firms from exercising market power in the mainstream sense of raising prices, it often neglects these other troubling forms of power. I shall employ these neo-Brandeisian insights to argue that giant firms would pose a credible threat to democratic principles in a POD or LS even if laws prevented them from raising consumer prices. 16 The democratic principles I have in mind come from an understanding of democracy as a collective decision procedure that instantiates political equality and popular citizen sovereignty. 17 16 An anonymous reviewer suggested that many liberals and republicans would already view their theories as condemning firms that advantage consumers while disadvantaging producers via lower wages or higher inequality. I agree that, if the economic actions of giant firms in a POD/LS had these distributive consequences, then this would raise an instability concern that complements my argument. However, my argument focuses on the likely political, rather than economic, actions of giant firms for two reasons. First, my conclusions hold even if we grant that a POD/LS could maintain a roughly equal distribution of wealth, as proponents of these systems suppose. Second, some liberals and republicans might be less concerned about the abovementioned economic actions of firms provided they conform to rules authorized by a suitably democratic procedure. My argument precludes this response because the actions of giant firms that I focus on compromise the democratic process. 17 For the pedigree of this view, see Albert Weale and Hugh Ward, 'Is rule by majorities special?', Political Studies, 58 (2010), 26-46, at pp. 29-31. collective decision making: pre-voting deliberation, agenda-setting, voting, and post-voting implementation. Accordingly, it covers not only one's formal opportunities (as determined by voting rules), but also one's informal opportunities (as determined by the resources at one's disposal). 18 Popular citizen sovereignty is required to show why situations where all citizens have equally little political power are undemocratic. For our purposes, it is sufficient to note that this principle prohibits situations where non-citizen agents enjoy a significant proportion of political power. Following Rawlsian and republican thought, I view citizenship as a particular role in which individuals apply principles of justice or the common good to their shared institutions. I therefore include other roles that individuals occupy, such as their economic roles, in the category of 'non-citizen agents'. 19 My argument addresses all those who view these democratic principles as lexically prior to whatever principle regulates economic inequalities above a sufficiency threshold. One need not view them as lexically prior to whatever principle secures basic liberties. On my reading, this includes the vast majority of liberals and republicans in the contemporary debate about systems of ownership.

II. THE POWER OF GIANT CORPORATIONS IN A POD
Given their democratic credentials, why do contemporary property-owning democrats appear relatively relaxed about firm size? I suggest that it follows from a tendency to equate egalitarian capital ownership with dispersed private power, at least under certain conditions. On this view, provided each shareholder in each giant corporation owns a small and roughly equal share, and competition law prevents such firms from raising consumer prices, and citizens generally aim to promote justice, there is little reason to be concerned about the political power of firms with giant market shares.
This understanding of the politics of a POD is not without support. It conforms to Mancur Olson's seminal small-beats-large theory of collective action, which explains why free competition between interest groups often leads to smaller groups having more political power than larger groups. 20 Olson's theory distinguishes three types of groups. 21 First, privileged groups have at least one member for whom the anticipated benefits of achieving the group's common interests outweigh the anticipated costs of acting alone to achieve them. for whom the anticipated benefits of achieving the group's common interests outweigh the anticipated costs of acting together to achieve them. Olson's theory applies principles of rational choice to explain why non-intermediate latent groups will typically remain immobilized due to free-riding effects, whereas privileged and intermediate groups will typically mobilize. This should generally translate into smaller groups enjoying greater political power than larger groups. 22 Drawing on this theory, one might conclude that the many small shareholders of a giant corporation in a POD would comprise a non-intermediate latent group that would fail to mobilize. Specifically, one might reason that, although each shareholder would have an interest in achieving policy changes that increase the profitability of their corporation, as a group they would struggle to overcome free-riding effects. This would appear to justify being relatively relaxed about giant firms under the abovementioned conditions. However, it overlooks the relevance of another element of Olson's work.
Olson's by-product theory explains why non-intermediate latent groups sometimes mobilize. Briefly, when group members enjoy private excludable goods, each member has a selective incentive to commit resources to the group, a proportion of which can fund their common interests. Therefore, when non-intermediate latent groups mobilize, this is a by-product of their primary purpose: to provide members with selective benefits. 23 Olson applies this theory to trade unions and professional associations; however, in my view, it applies equally well to giant corporations in a POD, given the incentives that small capital owners would have to invest in them. These incentives are illuminated by institutionalist analyses of the corporation.
Let us begin with the incentives small capital owners in a POD would have to invest in corporations generally. These follow from four features of this institutional form: 24 (1) the separation of ownership (shareholders) and control (directors who task managers) allows small investors to diversify their portfolios, thus spreading their financial risks; (2) 'limited liability' shields investors from corporate debts, making this type of investment less risky for passive investors; (3) 'asset lock-in' means that once an investor buys shares, they cannot withdraw their investment, but can sell it on; and (4) 'entity shielding' means that a shareholder's personal creditors may only take their shares, but no further corporate assets, as repayment. These final two features mean that corporations do not have to keep assets in liquid form to finance investor withdrawal or debt, and can therefore specialize their assets, which allows for higher returns. Moreover, the tradability of shares and the voting rights they confer also lower the risks of corporate investments. 25 22 I assume some familiarity with Olson's argument to this conclusion. 23 Olson, The Logic of Collective Action, pp. 132-5. 24 Here I follow David Ciepley, 'Beyond public and private: toward a political theory of the corporation', American Political Science Review, 107 (2013), 139-58, at pp. 143-4. 25 Abraham Singer highlights this advantage and explains why many firms would finance their operations through equity rather than debt, under the transaction cost assumptions adopted below; The Form of the Firm, p. 67.
Importantly, small capital owners in a POD would also often have strong incentives to prefer giant corporations. Neoclassical economics recognizes that giant firms would offer investors higher returns in industries that enjoy increasing returns to scale or network economies. However, if we turn to institutionalist theories of the firm, we can see why giant corporations would offer higher returns in many other industries in a POD.
Broadly speaking, these theories developed in response to Ronald Coase's observation that neoclassical economics cannot explain why firms exist, nor why there is not just one big firm. 26 Coase argued that we can only answer these questions by recognizing the costs of using the market, such as discovering prices and negotiating contracts. Given these 'transaction costs', it is often more efficient to leave contracts incomplete and give authority to one party to fill in the gaps later. 27 This authority relationship is constitutive of the firm, and it is therefore efficient for any given firm to expand 'until the [bureaucratic] costs of carrying out an extra transaction [hierarchically] within the firm become equal to the costs of carrying out the same transaction by means of an exchange in the open market.' 28 Oliver Williamson influentially developed Coase's approach by emphasizing that market transaction costs are only discernible if we relax certain unrealistic psychological assumptions underpinning neoclassical models of perfect competition. Specifically, the assumptions of 'perfect information' and 'honest dealing' should be replaced with 'bounded rationality' and 'opportunistic deceit'. 29 As Williamson notes, this more realistic approach provides reasons to think that giant corporations will often emerge as attractive investment options, even in industries that are not reliant on scale or network effects. First, certain firms will begin with superior business acumen, which can lead to giant size due to transaction costs in the market for managers. 30 Second, certain firms will benefit from repeated poor decision making by their competitors due to transaction costs in capital markets, allowing them to attain giant size. 31 Third, conglomerate firms of large absolute size will emerge to economize on transaction costs in capital markets, and, due to their efficiency advantages in capital investment, will be able to create giant firms. 32 This analysis reveals why Olson's by-product theory would apply to many giant corporations in a POD. Briefly, the small shareholders of such corporations would often enjoy higher returns than are available elsewhere, and, given these selective incentives to invest, the managers of such corporations should be able 26 Ronald Coase, 'The nature of the firm', Economica, 4 (1937), 386-485, at p. 388. 27 Ibid., pp. 391-2. 28  to use some of their shareholders' substantial combined resources to pursue the group's shared political interests. Importantly, on this analysis, giant corporations are often able to offer higher returns than are available elsewhere because of their lower costs of production (due to economizing on scale and transaction costs), so small shareholders would often have incentives to invest in them even if competition law prevented them from raising consumer prices. This is potentially troubling; however, it does not necessarily mean that giant corporations would undermine democratic principles in a POD. After all, other groups might plausibly exercise sufficient countervailing power in each relevant dimension of politics.

III. THE LACK OF COUNTERVAILING POWER IN A POD
In this section, I highlight three dimensions of the politics of a POD, in which the power of giant corporations poses a credible threat to democratic principles due to a presumptive lack of countervailing power. 33 In each dimension, we could understand giant corporations as undermining political equality or popular citizen sovereignty, depending on the role we view them playing in politics. First, we might view each corporation as a non-citizen actor, or an aggregate of noncitizen actors, that aims to influence policy outcomes at the non-voting stages of politics. 34 This would make them political competitors to citizens and a potential threat to popular citizen sovereignty. Second, we might view the political activities of corporations as providing certain citizens with greater opportunities to influence policy. For example, citizens who happen to share a corporation's policy preferences might enjoy free use of its public research. This would make corporations issue-subsidizers for certain citizens and a potential threat to political equality.

A. Giant Corporations versus Small and Medium-Sized Enterprises
Let us begin by examining policy issues where the interests of giant corporations are likely to conflict with the interests of small and medium-sized enterprises (SMEs). To capture this dimension of politics, imagine a POD in which the vast majority of SMEs support a ban on heavy goods lorries using high streets because the noise and pollution they create lowers footfall in these areas, and this reduces the profitability of their outlets. However, there is a giant grocery corporation-PODmart-whose outlets are based outside town centres, and its transportation costs would increase if heavy goods lorries are banned from using high streets; hence, it opposes the ban. Suppose that whenever a representative sample of citizens is polled under conditions suited to the articulation of authentic 33 A threat is credible when undemocratic interference is probable; hence, it should concern liberals and republicans. The final paragraph of this section, along with Section V, explain why my conclusions remain presumptive. 34 My argument is compatible with realist or reductivist conceptions of corporate agency.
judgements about the common good, there is a clear majority in favour of the ban. How should we expect the democratic process to play out on this issue? We should begin by distinguishing six main interest groups: (1) citizens in favour of the ban; (2) citizens against the ban; (3) consumers for the ban (perhaps on product choice grounds); (4) consumers against the ban (perhaps on product price grounds); (5) shareholders and employees of SMEs in favour of the ban; (6) shareholders and employees of PODmart. We know that group (1) is larger than group (2), but the crucial question, from a democratic perspective, is whether group (1) will get its way.
To answer this, it does not matter whether we suppose group (3) is larger than (4), nor whether (5) is larger than (6). Nor does it matter whether we suppose individuals belong to more than one group. All that matters is that each group is a non-intermediate latent group, so, according to Olson's small-beats-large theory, they should all fail to mobilize. As such, one would expect to see the option selected by group (1) at the voting stage implemented. However, this analysis is incomplete because Olson's by-product theory also applies to group (6). Therefore, the managers of PODmart should be able to use some of their shareholders' resources to pursue the group's policy interests, giving them a significant resource advantage over their political opponents, including group (1), at the non-voting stages of politics.
Admittedly, we have so far ignored another important group: PODmart's market competitors. To remedy this, suppose that PODmart's sales comprise 20 to 25 per cent of the domestic grocery market, and another four supermarkets enjoy 10 to 15 per cent each. 35 Suppose also that this level of market concentration is efficient in terms of scale and transaction costs. In this scenario, it is reasonable to assume that all five giant grocery corporations would oppose the ban for the same reason, so they should fail to mobilize due to free-riding effects. However, on an Olsonian analysis, the group would be able and willing to monitor one another's political efforts to minimize free riding because there are so few of them. 36 This remains the case even if competition law prevents them from sustainably coordinating on pricing. Accordingly, they would plausibly comprise an intermediate latent group that is able to mobilize (hereafter PODmart et al.).
Drawing on the experiences of capitalist democracies, we can distinguish various ways in which PODmart et al. could use their resource advantage to undermine democratic principles at the non-voting stages of politics. 37 First, at the agenda-setting stage, they could employ expert lobbyists, particularly at the 35 This is the approximate structure of the UK and French grocery markets, left ambiguous between monopoly and oligopoly concentration. In the US, the market leader (Walmart) has a similar market share, but only one other corporation enjoys a share above 10 per cent. In Australia, two corporations each enjoy a 30 to 40 per cent market share. 36  committee stage of politics, which has lower public salience. Consequently, by the time of the vote, the policy options favoured by SMEs may well have slipped off the agenda. 38 Second, at the pre-voting deliberation stage, PODmart et al. could exercise 'ideational power' by 'flooding the relevant marketplace of ideas'. As Thomas Christiano notes, in large-scale societies, most citizens have limited time to discover and comprehend political arguments, so the wealthy can dominate the 'socially limited cognitive space' of popular media. 39 PODmart et al. could achieve this in various ways: by funding large-scale advertising campaigns; 40 by funding grassroots campaigning; 41 and by funding or amplifying research organizations that either already share their policy preferences or are expected to promote them thereafter. 42 Finally, at the post-voting implementation stage, PODmart et al. could direct sophisticated lobbying strategies at bureaucrats, 43 or flood the sub-marketplace within which bureaucrats consume ideas about policy implementation. Additionally, if a society allows judicial review, PODmart et al. could raise the cost of implementing policies by challenging them in the courts. 44 At each of these stages of politics, we could view PODmart et al., or their members, as political competitors to citizens. Understood as such, they would pose three credible threats to popular citizen sovereignty. First, economic actors would be deciding much of the policy agenda on certain issues. Second, economic actors would be drowning out citizens' voices during pre-voting deliberation on certain issues. Third, economic actors would be exercising undue control over the final shape and (perceived) viability of certain policies. To reiterate, this threat persists even if the individuals playing these economic roles also belong to group (1).
Alternatively, we could view PODmart et al. as issue-subsidizers for certain citizens. Understood as such, they would pose three credible threats to political equality. First, politically active citizens who oppose the ban would be able to utilize PODmart et al.'s lobbying strategies and materials to influence the policy agenda. Second, such citizens would be able to utilize PODmart et al.'s public arguments on the issues at hand to influence pre-voting deliberation. Finally, these citizens would be able to utilize PODmart et al.'s lobbying strategies, public arguments, and judicial efforts to influence the implementation of policies.
In sum, we have identified a dimension of politics in a POD in which giant corporations pose a credible threat to democratic principles.

B. Giant Corporations versus Non-Economic Interest Groups
Let us now consider policy issues where the interests of giant corporations are likely to conflict with the interests of non-economic interest groups. By a noneconomic interest group, I mean a group of individuals who share a political interest that is not a common economic interest. For now, it does not matter whether they are self-interested or morally motivated. 45 To capture this dimension of politics, imagine that each of the corporations comprising PODmart et al. realize that they could lower their transportation costs without using high streets if they switched to smaller lorries that fit on country lanes. However, these smaller vehicles release harmful pollutants, so the vast majority of countryside residents want them banned. They are also unable to refrigerate larger food items; consequently, consumers are divided on the ban: some prioritize product choice, and others lower prices. Suppose that representative polling once again shows a clear majority of citizens in favour of the ban. How should we expect the democratic process to play out on this issue?
The main interest groups in this issue domain are the same as in the previous scenario with the exception that group (5) now contains countryside residents in favour of a ban for non-economic reasons. Nevertheless, this group remains a non-intermediate latent group. Therefore, the shareholders and employees of PODmart et al. would once again enjoy a coordination and resource advantage over their political opponents, including group (1), at the non-voting stages of politics. This highlights a second dimension of the politics of a POD in which giant corporations pose a credible threat to democratic principles.

C. Super-Managers versus Shareholders of Giant Corporations
Finally, let us turn to policy issues where the interests of the super-managers of giant corporations are likely to conflict with the interests of their shareholders. 46 To accommodate this dimension of politics, we must relax an assumption made so far about the relationship between shareholders and managers in a corporation. We have assumed that managers always act in the interests of shareholders by maximizing profits. This view of the corporation as a 'unitary rational actor' follows from neoclassical economics, where the firm is a 'black box'. 47 However, it is not necessarily consistent with institutionalist theories, which illuminate 45 I return to their motivations in Section V. 46  relations within the firm. Indeed, once we adopt a transaction-cost analysis, there are good reasons to worry about managers acting in their own interests, rather than the interests of their shareholders.
This type of concern was first popularized by Adolf Berle and Gardiner Means, who argued that, in a society of small shareholders, each shareholder has neither the incentive nor expertise to hold corporate managers to account. Accordingly, the modern corporation severs the historical link between ownership and control, allowing for problematic managerial discretion. Furthermore, large corporations are particularly problematic, because their more complex bureaucracies are harder for shareholders to navigate. 48 If this analysis of modern capitalism is correct, then the problem would only be exacerbated in a POD defined by smallscale ownership.
Admittedly, some contemporary theorists of the firm reject this analysis of modern capitalism. They argue that competitive markets for managers and capital protect shareholders from managerial discretion (the latter via takeovers). 49 But, as Abraham Singer points out, these theorists reconceptualize the firm as a nexus of complete contracts, which conflicts with the basic Coasian insight that firms only exist insofar as authority relations economize on market transaction costs. 50 Williamson preserves this insight by consistently assuming bounded rationality and opportunistic deceit in economic life. But this implies significant transaction costs in the markets for managers and capital; hence, Williamson takes Berle and Means's analysis seriously. 51 We should therefore expect the politics of a POD to include a range of issues where the interests of super-managers conflict with the interests of their shareholders. For instance, imagine a POD in which super-managers support deregulation of the market for managers because they think their remuneration will be higher in a regime that rewards short-term profits and allows them to leave organizations before longer-term problems materialize. However, the vast majority of shareholders in giant corporations oppose deregulation, worrying that it would leave their investments vulnerable, given their limited ability to monitor managers. Suppose that representative polling shows a clear majority of citizens against deregulation. How should we expect the democratic process to play out on this issue?
We can distinguish four main interest groups: (1) citizens against deregulation; (2) citizens in favour of deregulation; (3) shareholders in giant corporations; (4) super-managers of giant corporations. 52 As before, groups (1) and (2)  intermediate latent groups, so they should fail to mobilize. However, in contrast with previous scenarios, the by-product theory does not apply to group (3). This is because the shareholders of each giant corporation cannot rely on their managers to pursue their political interests, so they must organize independently. But, in doing so, they would encounter the coordination problems facing all nonintermediate latent groups, and therefore typically fail to mobilize. In contrast, group (4) is an intermediate latent group, as are the super-managers from each giant corporation within each industry. Therefore, they would typically mobilize, highlighting a third dimension of politics in a POD in which giant corporations pose a credible threat to democratic principles.
Before moving on, we should consider two possible sources of countervailing power to corporate super-managers in a POD. First, some theorists stipulate that shareholders in a POD would invest in large institutional funds, the managers of which would have sufficient expertise and incentives to discipline corporate super-managers. 53 However, this analysis has two limitations. First, institutional investors generally spread their investments across many corporations, weakening their incentive to hold any given corporation to account. 54 Second, the problem of asymmetric information looks likely to re-emerge between small shareholders and managers of institutional funds; hence, fund managers and corporate supermanagers often form powerful coalitions on certain issues. 55 Alternatively, one might argue that widespread share ownership would significantly increase shareholder activism in a POD, and, consequently, group (3) would effectively discipline corporate super-managers. This increased activism might involve shareholders advancing their economic interests or broader social goals, and it might occur directly or via 'meso-level' institutions, such as the abovementioned institutional funds or a trust that invests on behalf of all citizens. 56 Regardless of the specifics, this argument relies on an empirical premise: that widespread share ownership would be an especially effective form of 'motivation-conscious institutional engineering'. 57 Though not implausible, further research is required to substantiate this premise if it is going to play this major role in justifying giant corporations. Hence, this potential solution to the problem of giant firms deserves further examination alongside the other approaches highlighted in Section VI.

IV. SIMILAR PROBLEMS WITH GIANT WORKER-MANAGED FIRMS
So far, we have focused on how giant corporations would threaten democratic principles in a POD. But many theorists envision a significant role for workermanaged firms in a POD, 58 while others advocate a LS that mandates such firms. 59 So, we should consider whether giant worker-managed firms avoid the problems raised thus far.
Of course, worker-managed firms can be owned in different ways. Unsurprisingly, property-owning democrats tend to prefer a worker-owned model, 60 whereas liberal socialists tend to prefer a publicly owned and workerleased model. 61 For this reason, my argument is designed to apply to any firm where the 'direction and management [is] elected by, if not directly in the hands of, its own workforce' 62 according to the principle of 'one person, one vote', regardless of its ownership structure. Importantly, in any such firm workers will typically aim to maximize average income per worker, rather than profitability. 63 To see whether worker management avoids the problems raised thus far, let us reimagine the first two scenarios from Section III. This time, group (6) contains the worker-managers of a giant worker-managed grocery firm-WMFmartrather than the shareholders and employees of PODmart. How should we now expect the democratic process to play out on each issue?
The main interest groups in each issue domain remain non-intermediate latent groups, so they should fail to mobilize. The crucial question, then, is whether Olson's by-product theory continues to apply to group (6). In my view, it often would, because giant worker-managed firms would offer investors higher returns than are available elsewhere for two familiar reasons. First, certain industries would continue to supply goods with increasing returns to scale or network economies. Second, in other industries, giant worker-managed firms would often economize on market transaction costs more efficiently than smaller firms would. Hence, in practice, a range of small to large worker-managed firms have emerged in both private ownership and nationalized economies. 64 A possible objection to my second line of reasoning is that the costs of carrying out an extra transaction democratically (in a worker-managed firm) would equal the costs of transacting through the market much sooner than the costs of carrying out an extra transaction hierarchically would. Therefore, we cannot rely on institutionalist theories of the firm to show that the by-product theory applies to giant worker-managed firms as it does to giant hierarchical corporations.
A full response to this would require further research. For now, I offer two conjectural replies. First, proponents of worker-managed firms generally accept that, under conditions of bounded rationality, they will be representative (not direct) democracies. However, insofar as they retain an element of hierarchy to overcome problems of inefficiency, the by-product theory becomes applicable. Second, there is evidence to suggest that democratic workplaces motivate workers to be more productive. 65 If this is because democratic relations of production are less vulnerable to opportunistic deceit than market or hierarchical relations, then the by-product theory becomes applicable.
In sum, it is plausible that giant worker-managed firms would emerge and mobilize in the first two dimensions of politics considered in Section III. However, it is unclear whether the threat they pose to democratic principles is equivalent to that posed by giant hierarchical corporations, particularly once we also consider conflicts between owners and managers. Consequently, further research is required, comparing the prevalence of giant firms and managerial discretion under a POD and LS, before an informed choice can be made between these systems.

V. OBJECTIONS
The main objections to my argument are empirical criticisms of my use of collective action theory, and normative criticisms. I consider each in turn.

A. Collective Action Theory Objections
One might first object that my argument applies Olson's by-product theory to giant firms, but does not consider whether it applies to other diffuse groups, like associations of SMEs or environmental groups. yet Olsonians acknowledge that these groups sometimes mobilize and provide their members with selective benefits, such as insurance. 66 In reply, I note that the selective benefits these groups offer are fragile, because private firms that do not fund political activities can offer them at lower prices. 67 Conversely, the selective benefits provided by giant firms (higher returns) result from their having lower production costs than other firms. Hence, the by-product theory does not apply equivalently across these groups.
Nevertheless, one might argue that Olson overlooks other reasons why large diffuse groups sometimes mobilize. Indeed, the contemporary collective action literature identifies six: 68 (1) group members sometimes overestimate their contribution to the group's efforts; 69 (2) each policy domain has its own communication network, which no group can easily control, thus lowering the coordination costs for diffuse groups; 70 (3) political entrepreneurs sometimes take on (non-financial) mobilization costs; 71 (4) social movements spearheaded by social entrepreneurs sometimes take on (non-financial) mobilization costs; 72 (5) a wealthy subgroup of members sometimes operates as an intermediate group; 73 (6) wealthy patrons sometimes take on the costs of mobilization. 74 I accept that the first four reasons provide grounds to be more optimistic about the political opponents of giant firms sometimes providing countervailing power than my earlier arguments suggest. This undermines the strict Olsonian thesis that large diffuse groups will never mobilize. But it remains consistent with the weaker Olsonian thesis that concentrated groups and diffuse groups that provide (non-fragile) selective benefits enjoy coordination advantages, because other groups generally rely on more fragile incentives or motivations to mobilize and remain mobilized. This weaker thesis, which can accommodate the historical emergence of organizations that provide some countervailing power to giant firms, is widely accepted in the collective action literature, 75 and remains sufficient for my argument. Moreover, the fifth and sixth reasons do not apply to a POD/ LS because there would be no suitably wealthy people in these societies. So the political rivals of giant firms would plausibly provide less countervailing power in a POD/LS than in existing capitalist democracies.
Gunnar Trumball poses a different challenge to my use of Olsonian analysis. 76 He accepts the weaker Olsonian thesis that giant firms enjoy certain coordination advantages, but argues that large diffuse groups nevertheless enjoy a greater advantage: perceived legitimacy. Consequently, giant firms can only succeed politically when their interests (appear to) align with those of larger diffuse groups, like consumers. 77 If Trumball is right, then Olsonians have things back to front, and my concerns about corporate power in a POD/LS are unwarranted.
In my view, Trumball highlights additional constraints on the political actions of giant firms, but underplays a form of power that they retain due to their coordination advantages by implicitly portraying it as a form of luck. To elaborate, he argues that when large diffuse groups fail to mobilize it is typically due to a lack of perceived common interests. 78 To overcome this, they require a public narrative defining their common interests. 79 Conversely, concentrated interests find it easier to mobilize, but require a public narrative that aligns their interests with those of larger diffuse groups in order to appear legitimate. Trumball presents this as a constraint on corporate power: giant firms can only succeed when their interests luckily align with those of larger diffuse groups. But this underplays the ideational power that giant firms have to shape large diffuse groups' awareness and understanding of their interests. Admittedly, this power operates within constraints, but there is still considerable room for manoeuvre, as Trumball seemingly accepts. 80 Trumball might reply that giant firms are trusted so little that the reputational risks of getting caught trying to exercise this type of power generally outweigh the potential benefits. 81 However, there are four reasons why this reply does not undermine my argument. First, giant firms sometimes reach the opposite conclusion, as documented by other scholars, 82 and acknowledged by Trumball. 83 Second, even if giant firms exercise such power infrequently, this does not mean it is less troubling. Indeed, there is evidence that firms ration their exercises of power to secure strategic victories. 84 Third, giant firms need not manipulate other people's views directly in order to violate my principle of political equality; it is enough that they provide some citizens with greater resources to influence policy, which is doable at lower risk. Fourth, even if we accept Trumball's view, it simply confirms that, on issues where the interests of giant firms align with those of large diffuse groups, interest group politics can draw policy outcomes away from those that would emerge from a process that treats citizens as equal and sovereign, thus violating my democratic principles. Hence, suitably caveated, I maintain that giant firms would pose a credible threat to democratic principles in a POD or LS.

B. Non-Ideal Theory Objections
A different objection to my argument is that it relies on pessimistic motivational assumptions that are justifiably idealized away by other normative theorists in the debate about ownership. Specifically, one might argue that most theorists in this debate follow Rawls's approach of engaging in 'ideal theory' where the aim is to identify the best institutions for a 'well-ordered society' in which: 1. individuals accept and know that everyone else accepts the same principles of justice; 2. the major institutions together satisfy, and are publicly known to satisfy, these principles; 3. individuals have a normally effective sense of justice that allows them to understand and apply the publicly recognized principles of justice; 85 4. individuals are motivated to comply with and further just institutions provided there is sufficient assurance that others will likewise do so. 86 In contrast, the scenarios that I rely on appear to include undemocratic behaviour that conflicts with condition (4). Consequently, my argument looks like a contribution to non-ideal theory: that is, how best to deal with avoidable moral failings.
I have two replies to this objection. First, my argument does not rely on office-holders in giant firms intentionally exercising political power in ways that undermine democratic principles. Rather, it is sufficient for my purposes that they might unintentionally do so while pursuing permissible apolitical goals. For example, the managers of PODmart might spend large amounts on advertising the benefits of having a variety of low-cost items available in one location. Their sole aim in doing so might be to maximize profits by 'flooding the marketplace of commercial ideas'. Nevertheless, this campaign might accidentally spill over and flood the 'marketplace of political ideas', thus undermining democratic principles despite condition (4) being satisfied.
Second, there are two reasons why we should expect office-holders in giant firms sometimes to undermine democratic principles intentionally in a wellordered society. The first follows from the italicized portion of condition (4), which highlights that individuals in a well-ordered society are motivated by reciprocity. As Paul Weithman notes, this means that assurance problems will continue to arise: that is, individuals will only restrict their pursuit of self-interest in politics if they have sufficient assurance that others will. 87 Weithman argues that condition (1) solves this problem, 88 but under conditions of bounded rationality, we cannot simply stipulate condition (1). Rather, to overcome this assurance problem, we require laws that publicly assure mutual compliance with democratic principles throughout the political process. 89 However, for reasons I set out shortly, any such laws risk being either ineffective or unacceptably illiberal. Accordingly, we should expect individuals, including office-holders in giant firms, sometimes to undermine democratic principles intentionally in a well-ordered society. (This argument also explains the existence of self-interested non-economic interest groups.) A second argument to the same conclusion notes that, for Rawlsians, a full evaluation of institutional arrangements must consider how likely they are to generate interests and norms that ensure their ongoing stability. Rawls brackets this issue in his work, 90 but, as others note, this abstraction can only be preliminary. 91 yet, once it is relaxed, it becomes relevant that many think the norms internalized in hierarchical workplaces 92 and market competition 93 lead citizens to privilege their economic group interests over the common good. The former practice persists in a POD and the latter under LS. Hence, we have another reason to expect office-holders in giant firms sometimes to undermine democratic principles intentionally in such societies. (This also explains the existence of morally motivated non-economic interest groups: they form to counteract corrupt economic actors.) A different form of 'non-ideal theory objection' to my argument is that it relies on pessimistic institutional assumptions that are justifiably idealized away by other normative theorists. Specifically, certain theorists argue that we could hone the Scandinavian strategy of insulating politics from economic power using measures such as publicly funded elections and media regulations. 94 Consequently, my argument is only relevant if we assume such measures fail, which contradicts condition (2).
I accept that my argument applies less forcefully to societies that insulate their political process from economic power more thoroughly. However, there are two reasons why it does not therefore apply only to societies that exhibit avoidable moral failings in this regard. First, even if a society completely insulated its political decision makers from corporate lobbying, giant corporations could unintentionally compromise democratic principles by 'flooding the marketplace of commercial ideas', as previously described. One might reply that an insulation strategy could encompass this kind of commercial speech. However, this highlights a more general dilemma for well-ordered societies under conditions of bounded rationality: any attempt to insulate politics entirely from economic power risks 90  being either ineffective or unacceptably illiberal. For instance, notice how difficult it would be to assure citizens that sophisticated strategies are not being used by the economically powerful to undermine democratic principles, particularly if issue-subsidization alone undermines political equality. 95 Such assurance would seemingly require a public authority to monitor and prevent many interactions between citizens in ways that restrict basic freedoms of speech and association. 96 So, although a society with a comprehensive insulation strategy would not require an additional strategy to deal with corporate power, the most attractive blend of strategies is a topic that requires further examination.

VI. CONCLUSION
I have raised doubts about whether the versions of a POD and LS proposed by contemporary theorists would be capable of securing democratic principles that they should endorse. These doubts stem from the neo-Brandeisian insight that firm size matters, because giant firms furnish certain groups with greater power than their political rivals, allowing them to divert policy outcomes away from the authentic preferences of equal citizens.
Accordingly, the contemporary normative debate about ownership requires supplementation by a detailed comparison of strategies to deal with giant firms. The existing literature suggests some promising options. For example, some theorists advocate an associative POD in which the state uses fiscal policy and motivation-conscious institutional engineering to incentivize a balanced ecology of groups, 97 while others advocate a corporatist POD in which the main groups in each issue domain enjoy public authority. 98 We might view these as different forms of a countervailing power strategy that encourages the formation of groups that exert equal opposing influence in each issue domain.
It is also worth considering a different, and potentially complementary, approach advanced by neo-Brandeisians in the US, and ordoliberals in Europe. They place less emphasis on counterbalancing concentrations of private economic power, and greater emphasis on preventing such concentrations in the first place. We might label this the fragmentation strategy, insofar as it aims to confront certain interest groups with collective action problems that they might otherwise 95 Similar concerns are shared by Edmundson, John Rawls, pp. 135-6; Freeman, 'Property-owning democracy and the difference principle', p. 16; Krouse and McPherson, 'Capitalism, "propertyowning democracy," and the welfare state', pp. 86-7; Thomas, Republic of Equals, pp. 107-9; Williamson, 'Who owns what?', pp. 437-8. 96 For a similar concern, see Schemmel, 'How (not) to criticise the welfare state', p. 403. 97 See, e.g., Thomas, Republic of Equals, pp. 274-7. An anonymous reviewer highlighted state counter-speech as another viable means to this end. 98 A corporatist POD is recommended on grounds of stability, rather than democracy, in Waheed Hussain, 'Nurturing the sense of justice: the Rawlsian argument for democratic corporatism', O'Neill and Williamson, Property-owning Democracy, pp. 180-200. overcome. Interestingly, this intellectual tradition seems to have inspired Rawls's initial preference for a POD over welfare-state capitalism. 99 It deserves further examination, alongside other approaches, in order to determine the form(s) of political economy that best serves liberal and republican values.