The democratic deficit of the G20

ABSTRACT In the last few decades, the democratic credentials of global governance institutions have been extensively debated in the fields of international relations and political philosophy. However, despite their prominent role in the architecture of global governance, club governance institutions like the Group of Seven (G7) or the Group of Twenty (G20) have rarely been considered from the perspective of democratic theory. Focussing on the G20, this paper analyses its functions in international political practice and discusses whether, in exercising these functions, the G20 exhibits a democratic deficit. As a standard of democracy, the analysis uses the all-affected principle, according to which all those who are affected by a policy decision should be given the opportunity to participate in decision-making. This paper identifies several democratic shortcomings of the G20, for instance related to the exclusion of citizens of non-member states and a lack of parliamentary and public control. By describing realisable reforms that could to some degree alleviate these shortcomings, it is shown that more democratic institutional alternatives are feasible. Thus, the ascription of a democratic deficit to the G20 is warranted.

The last few decades have seen a broad debate in the fields of international relations and political philosophy on democracy in political institutions beyond the state. First, scholars have, on a general level, discussed the justification for democratic institutions in global governance, what democratic decision-making procedures in this context can look like, what preconditions particular forms of democracy have and whether these conditions are met in the global realm (Held 1995;Marchetti 2008;Goodin 2010;Archibugi and Held 2011;Christiano 2012). Secondly, scholars have investigated the extent to which democratic values are realised in specific political institutions beyond the state and, if deemed as necessary, how these institutions can be democratised. Very common in this literature, in particular with regard to the European Union (EU), is the diagnosis of a 'democratic deficit', which might call into question the legitimacy of the institutions considered. In this context, critics have, for example, pointed to the lack of parliamentary control and opportunities for public deliberations and contestation (Norris 1997;Zürn 2000;Nye 2001;Howse 2003;Follesdal and Hix 2006). Others have challenged such diagnoses by questioning the standards of democracy applied by the critics (Majone 1998;Hunter 2002;Moravcsik 2004).
A type of global governance that has rarely been discussed with respect to its democratic credentials is the so-called club governance. 1 Club governance is a weakly institutionalised form of cooperation in a selected group of states that builds on regular meetings of government officials. Prominent club governance institutions include the Group of Seven (G7), the Group of Twenty (G20) and the meetings of a group comprising Brazil, Russia, India, China, and South Africa (BRICS). 2 Such clubs are arguably exercising significant influence in global governance and their summit meetings have received a lot of public attention and contestation in recent years. This is true in particular for the G20, which had, for example, a considerable role during the financial crisis in 2008.
This paper discusses whether club governance exhibits a democratic deficit. As democratic values are widely shared and democracy is often taken to be crucial for political legitimacy, an investigation of this question seems to be a valuable endeavour. 3 Due to its prominent position, this paper focuses on the G20. However, most arguments presented also hold true for other club governance institutions, which often have similar working procedures, although the different membership of the clubs could certainly justify different judgements with respect to their degree of conformity to democratic values.
Section 1 presents the so-called all-affected principle as a central normative standard for the assessment of the democratic credentials of political institutions. The principle states that all those affected by the decisions of an institution should have effective opportunities for political participation. Before this standard can be applied to the G20, it is necessary to understand the nature of the institution and its functions in international politics. Thus, Section 2 analyses the G20 in light of governance theories developed in the field of international relations in recent years. Two overarching functions of the institution in political practice are identified: first, the coordination of national policies of its member states; and secondly, the provision of guidance to other global governance institutions and the advancement of international regulations.
In Section 3, the exercise of the G20's two functions in political practice is discussed in light of the all-affected principle. Shortcomings are identified, for instance, with regard to the exclusion of most states from the group, the predominant role of executive bodies in the G20 process and the lack of opportunities for participation of citizens of non-democratic G20 member states. Finally, Section 4 discusses directions for feasible institutional reforms that could, to some degree, diminish the identified democratic shortcomings.
In sum, the arguments presented in this paper warrant the judgement that the G20 exhibits a democratic deficit and suggest that this conclusion also holds true for other club governance institutions. This might not be a surprising result. After all, defenders of this form of governance believe that legitimacy can be created by 'effective problem-solving', which should be facilitated by a limited group size. Club governance is thus by intention exclusionary. However, that does not compromise the relevance of the arguments presented. First, to ensure that the argumentation of this paper is politically relevant, the paper considers the ascription of a democratic deficit only as warranted if a more democratic alternative is feasible. That this holds true for the G20 might already be a less obvious claim.
Secondly, it is beneficial to analyse exactly what constitutes the democratic shortcomings of the G20 and to discuss what reforms might reduce them. Establishing some clarity on the concrete nature of the democratic deficit and potential reforms allows an assessment as to whether these reforms would substantially compromise the presumed benefits of the G20 in its current form. This paper tentatively argues that these reforms are not necessarily associated with a decline of effectiveness. However, even if this were the case, a clear account of the concrete nature of the democratic deficit would still be beneficial. Only if we have such an account can we make an informed decision as to whether the presumed benefits outweigh the democratic shortcomings. 4

Democracy and the all-affected principle
Prior to addressing the question of whether an institution has a democratic deficit, it must be clear what standard of democracy is being applied. This paper applies a general principle that is often conceived as a central requirement of democracy, the so-called allaffected principlestating that all those affected by a policy decision should be given opportunities for participation in decision-making. Versions of this principle have repeatedly been defended on theoretical grounds (Pogge 1992;Goodin 2007;Fung 2013;Arrhenius 2018) and are often used in more empirically oriented assessments of existing political institutions (Woods 2001;Zürn 2004;Steffek and Nanz 2008;Koenig-Archibugi 2017). 5 In the way the all-affected principle is understood here, the opportunities for political participation that have to be provided to meet the principle can take very different forms, including opportunities to participate in plebiscites or the election of representatives, access to fora that enable public deliberation and contestation, and opportunities to hold policy-makers retrospectively accountable for their actions. As different forms of participation can be feasible and justifiable in different contexts, the all-affected principle should, as a general standard, leave open what forms of participation have to be enabled.
In addition, this paper assumes that opportunities for participation should be equal for all affected persons if there are no compelling reasons to differentiate in this regard. Such reasons not to provide everyone affected with the same opportunities for political participation can be that persons are affected in different ways and to different degrees. Especially affected persons (or social groups) might, for example, be consulted in a hearing before a decision is taken.
Furthermore, the all-affected principle requires that those affected not only have formal rights to political participation, but that these persons must also be able to effectively exercise these rights. There are different reasons why formal rights to participation might not translate into effective opportunities for participation. Citizens might lack resources necessary to effectively exercise their rights, for example with regard to education or time. A very different kind of obstacle to effective political participation can be constituted by external constraints. For instance, states might not be able to implement the policies favoured by their citizens because this would negatively affect the competitiveness of their economy (Zürn 2000, 189). Due to problems of collective action, this might even hold true if the population of all states favoured such policies. Even if the relevant states were to give their citizens equal opportunities in their domestic decision-making, the citizens might in this case have more effective opportunities for democratic participation if the competence to decide on the respective policy issues was shifted to a higher level political institution.
As the all-affected principle sets the bar for political institutions very high (Goodin 2007), one might wonder if it can play a useful role in the assessment of real-world institutions. Certainly, as policies usually have far-reaching implications, no existing institution fully satisfies the principle. However, the all-affected principle can still be used in comparative judgements of whether a particular institution does better in terms of the allaffected principle than an alternative institution (Koenig-Archibugi 2017, 178). We can always ask if one alternative offers effective opportunities for political participation to a higher proportion of those affected by its policies than another.
If the all-affected principle is to be used in comparative judgements, the question arises as to what is the relevant scenario of comparison for the diagnosis of democratic deficits of political institutions. There are three obvious options in this regard: First, the situation in which the institution exists could be compared with the situation in which the institution does not exist (everything else kept constant). However, that the existence of the institution in question is beneficial in terms of the all-affected principle compared to its non-existence can hardly be enough for it to be considered sufficiently democratic. If there are preferable alternatives, the institution in question is deficient.
Secondly, one could identify an ideal institution that satisfies the all-affected principle to the highest conceivable degree and use this best alternative as a scenario of comparison. Leaving aside the question how the ideal institution could be identified, it would in any case be a trivial finding that real-world institutions fall short in comparison to the best conceivable institution. That is not to say that ideal-type investigations are necessarily misguided or uninteresting. However, if they are meant to have practical implications for the actual institutional choices we face, diagnoses of democratic deficits of political institutions have to take into account the feasibility constraints we encounter in our world.
The third option is thus to compare the institutions in question with feasible alternatives. In this paper, a democratic deficit is attested if there is a feasible alternative that meets the requirement stated by the all-affected principle to a higher degree than the institution in question. 6 A lot depends then on what is deemed as feasible. Candidates for relevant feasibility constraints in the context of this investigation are, for example, limits in the availability of resources or the existence of power relationships that are hard to change. Applied in this type of comparative judgement between feasible alternatives, the all-affected principle can be a useful standard for the assessment of the democratic credentials of political institutions.

Nature and functions of the G20
Before we can assess the G20 against the all-affected principle we need to know more about the nature and functions of this institution in international political practice. Only then are we in the position to assess who is affected by the exercise of these functions and whether the G20 provides effective opportunities for political participation to those people.
The G20 is part of a long tradition of weakly institutionalised clubs of states whose government officials meet regularly (the most prominent example is arguably the older G7). The G20 was founded in 1999 as a meeting of finance ministers and central bank governors of 19 countries and representatives of the European Union. A motivation for the formation of the group was that the G7 had to notice during the Asian financial crisis that they alone were not able anymore to make and effectively implement decisions concerning the global economy (Martin 2013, vii). In 2008, again in response to a financial crisis, the G20 started to meet additionally once a year at the level of heads of state and government. As outcomes of these summits, the group publishes communiques on which all member states have agreed.
In addition to the meetings of heads of state and government, the G20 has an extensive working process. In this process, representatives of different government departments meet repeatedly during the course of the year in a number of thematic working groups and discuss policy issues, formulate common goals, develop joint initiatives and prepare the summits. At G20 working group meetings and summits, representatives of various international and regional organisations and guest countries are also usually present.
Policy-makers often associate with the G20 and other club governance institutions the hope that the limitation to a relatively small group of states will facilitate substantial agreements on important issues of international concern, even though decisions are only taken by consensus. Substantial agreements are taken to be facilitated by the small size of the group as well as the fact that negotiations at summits and in preliminary working processes are conducted by relatively high-level government representatives who wield some discretion. That the countries involved occupy powerful positions should ensure that these agreements are effective. In defending the legitimacy of the G20, policy-makers usually refer to this alleged capacity of the institution to make progress with respect to pressing global issues. In this line, the former Canadian finance and prime minister, Paul Martin, claims that 'the ultimate justification for the G20 […] is to make globalization work' (Martin 2018).
For the purpose of this investigation, it is useful to distinguish two major overarching functions of the G20: First, the coordination of national policies of its member states; and secondly, the provision of guidance to other global governance institutions and the advancement of international regulations. National policies are coordinated in the way that governments commit in the group to implement certain policies. In addition, the exchange in the G20 can influence national policies by furthering the spread of norms and practices among governments and by framing the way in which policy-makers think about certain issues (Luckhurst 2019, 105ff.).
Certainly, this is not meant to imply that a large share of national policies of G20 member states are decisively influenced by G20 discussions. Indeed, many observers believe that the effectiveness of the G20 in the coordination of national policies strongly declined after the end of the financial crisis (Ocampo and Stiglitz 2011, 1f.). Although the degree of influence of the G20 on policies of member states is hard to assess, it seems reasonable to assume that the G20 still exercises some influence on them.
The second function of the G20 is the provision of guidance to other global governance institutions and the advancement of international regulations. Viola (2015) explicates this function of the G20 by applying the orchestrator-intermediary model of governance (presented in Abbott et al. 2016) to the G20. Orchestration is a form of governance in which an institution (orchestrator) mobilises other agents (intermediaries) to perform certain governance tasks because it assumes that these other agents are better able to perform these tasks than itself. However, governance tasks are not formally delegated to the intermediaries as they are not under the direct and strict control of the orchestrator, who only provides material or ideational support. The intermediaries perform their activities voluntarily because they share the governance goals of the orchestrator.
Viola argues that the G20 acts as orchestrator with regard to a number of intermediaries, like the Organisation for Economic Co-operation and Development (OECD), the International Monetary Fund (IMF), the World Bank, and the Financial Stability Board (FSB). 7 As the G20 does not have its own bureaucracy, in many cases it draws on the expertise of these institutions or relies on their capacity to implement policy measures. In turn, the G20 can provide political endorsement of the activities of the intermediaries and member states can commit in the G20 to provide financial resources. 8 Intermediaries can also allow the G20 to reach out to states that are not members of the group and advance international regulations that extend beyond the borders of its member countries. The G20 can initiate negotiation processes within other institutions and build consensus among its members. If G20 member states have reached consensus on some international regulations, it is often hard for any other agent to effectively promote a divergent proposal. Many of the international regulations advanced by the G20 over time become relevant not only to G20 member states (and are enforced not only by these states) but also to other countries.

The G20 in light of the all-affected principle
This section investigates the effects of the exercise of the two functions of the G20 identified in the previous section on the opportunities for effective and equal political participation of the citizens of G20 member states and non-member states. The most obvious shortcoming of the G20 with respect to the all-affected principle concerns the citizens of non-member states. About one third of the world's population does not have any representative at the G20 table, while it could be heavily affected by G20 decisions. First of all, this is relevant with respect to the second function of the G20, i.e. the provision of guidance to other global governance institutions and the advancement of international regulations. The activities of global governance institutions, under the guidance of the G20, often directly target countries that are not G20 member states. Moreover, as described, many of the international regulations advanced by the G20 become relevant to and are enforced by non-member states.
To some degree, the lack of representation of citizens of non-member states is also problematic with respect to the first function of the G20, i.e. the coordination of national policies of the member states. National policies such as environmental regulations or tax policies of powerful countries like the G20 member states often have external effects in other countries. This can be problematic from a democratic perspective because those citizens of non-member countries who are affected by such external effects did not have a voice in decision-making with regard to the respective policies. Thus, although the G20 constitutes a step forward with respect to its inclusiveness compared to the G7 (Goodin 2010, 195ff.), due to the lack of representation of the citizens of nonmember states, the exercise of both of its functions by the G20 violates the all-affected principle.
With regard to the citizens of the G20 countries, G20 decision-making might appear at first glance as unproblematic because these citizens are represented by their national governments in the group. However, also in this respect, G20 decision-making suffers from serious democratic shortcomings.
First, the working arrangements of the G20 bring together representatives of the executive of member states. 9 Thus, the G20 suffers from the shortcomings that are described by Zürn (2004, 264) as related to 'executive multilateralism'. If the G20 is indeed influential, parliaments can only subsequently approve and realise decisions taken by governments (Cooper and Pouliot 2015, 343). Although the G20 process does not affect the formal competencies of parliaments, some legislative decisions are de facto prepared in meetings of G20 working groups (i.e. in technocratic meetings of bureaucrats from national government departments). Parliamentary control of the executive is complicated by the privileged access of governments to information with regard to negotiation processes within the G20. In addition, governments can put pressure on parliaments by highlighting the need to comply with commitments made by governments in the G20 because otherwise international cooperation and the status of the state as a credible negotiation partner might be undermined. This can be seen as a problematic shift of power to the executive and as reducing opportunities for democratic participation and surveillance in general (Beisheim and Brunnengräber 2008).
Certainly, in democratic states governments are also democratically legitimised. However, chains of delegations between individual citizens and the government officials that take decisions in G20 working groups are longer than between citizens and parliamentarians. As some discretion exists at every stage of delegation, longer chains of delegation are usually associated with less opportunities for effective political participation (Buchanan and Keohane 2006, 414f.).
Secondly, policy-making by the G20 often suffers from a lack of transparency (Bendiek and Neyer 2016, 99). G20 negotiations take place behind closed doors. While there is considerable public attention on G20 summits, the working process is characterised by a low level of media coverage and public control. The lack of transparency can allow governments to claim successes for themselves and blame other agents for less popular decisions (Bendiek and Neyer 2016, 96). The first and the second shortcoming both contribute to the general problem that the behaviour of governments in the G20, even of democratic states, is not subject to much public control and participation.
Thirdly, as in other international negotiation processes, due to power asymmetries and differences with regard to financial capacities and access to expertise, the influence of member states on G20 decisions is highly unequal. In general, more powerful countries have the capacity to put pressure on other countries to agree on the measures they prefer. In addition, issues on the G20 agenda are sometimes discussed beforehand at G7 or BRICS meetings without representatives of many G20 member states. If the representatives of these groups jointly advance a proposal in the G20, it is hard for other states to push through alternative suggestions.
Furthermore, while in the case of wealthy states each G20 working group meeting is prepared and attended by a delegation of several people, in the case of other states a single representative has to cover more than one of the G20 working groups. Such differences exist also with regard to access to academic research and policy advice. This limits the ability of the representatives of some countries to fully assess the consequences of potential G20 decisions as well as their opportunities to engage in the process with their own proposals. If differential capacities of states to influence G20 decision-making do not follow population size, these inequalities are contrary to the requirement of giving everyone affected equal opportunities to participate in political decision-making (as long as there are no compelling reasons, such as differences in degree or kind of affectedness, that could justify differentiated opportunities for participation).
Fourthly, not all G20 member states are democratic. The governments of countries such as China, Russia, Saudi Arabia or Turkey are not democratic representatives of their citizens, who can thus hardly be described as having effective opportunities to participate in G20 decision-making. 10 Fifthly, in general with a growing number of people living in countries that coordinate their policies, the degree of influence each individual can exercise on the results of coordination diminishes. If policy decisions are not made by national-level political institutions, but by an institution such as the G20, this can thus be associated with a shrinking space for meaningful political participation of individual citizens (Dahl 1999, 22). Again, the G20 does not formally assume any competence that could alternatively be held by lower level political institutions, but if the group is indeed influential, its influence has to be associated with a shrinking de facto decision-making power of such institutions. In the case of Germany, an individual is then not 1 of about 82 million inhabitants of Germany, but 1 of about 5 billion inhabitants of all G20 countries who could try to influence decisionmaking. This might call into question whether the individual really has opportunities for effective participation. I will call this the problem of growing constituency.
The existence of these democratic shortcomings does not show, though, that, in principle, the exercise of the two functions of the G20 by a well-designed institution could not also have positive impacts on the provision of opportunities for effective democratic participation. Obviously, national policies of G20 members do not only have external effects in non-member states (as mentioned above) but also in other member states. If such national policies are coordinated by a global governance institution, that could give the representatives of the citizens of a member country that is affected by a policy of another member country some voice in decisions relevant to that policy, at least on a very general level. The coordination of national policies can thus be beneficial with respect to the realisation of the all-affected principle.
In addition, as mentioned in Section 1, if states act in an uncoordinated manner, they often face collective action problems. For instance, the concern of each state for the competitiveness of domestic industries can lead to a race to the bottom with regard to environmental standards, tax rates or workers' rights. International coordination to solve such collective action problems is crucial to enable effective democratic participation with regard to some policy issues.
The exercise of the second function of the G20 could, in principle, also have positive impacts. Decision-making in many global governance institutions that are under the influence of the G20 suffers from similar democratic shortcomings to those discussed here with respect to the G20 itself. In many intergovernmental organisations, like the IMF or the World Bank, voting rights and even more so actual opportunities to influence decisions have highly unequal distributions. Decision-making is often not very transparent and leaves little room for public deliberation and contestation. 11 A reasonably democratic international institution fulfilling an overarching coordinating function could provide a platform that facilitates transnational public debates and provide better opportunities for democratic participation in global governance than most of the existing highly technocratic international institutions (Benson and Zürn 2019, 8).
In addition, an institution that provides guidance to other global governance institutions can be beneficial because of the fragmented nature of global governance. In many policy areas, institutions with overlapping mandates exist that act in parallel. If different institutions pursue conflicting activities, this can, obviously, hamper coherent policies. Furthermore, the various global governance institutions that focus on their own thematic remit are often not able to adequately address interlinkages among different policy areas. In light of the requirement that all affected persons should have an effective say in policy-making, fragmentation can thus constitute a democratic shortcoming. The potential negative effects of fragmentation can be reduced if the relevant institutions are coordinated in some way (Zürn and Faude 2013). An institution that, at least to some degree, conforms with the all-affected principle and provides guidance to such other global governance institutions might thus be desirable (Caney 2005, 162;Benson and Zürn 2019, 7f.).
Obviously, the claim that the two functions of the G20 do not necessarily reduce opportunities for democratic participation, does not imply that the exercise of these functions by the G20 is desirable (and even less that the G20 is the best feasible institutional alternative in this regard). The democratic shortcomings of the group identified above give us reasons to doubt that this is the case. I will discuss in the next section how the G20 could be reformed to provide more effective opportunities for political participation to those affected by its policies.
Beforehand, though, I should address two objections. First, the G20 might not be the kind of political institution that ought to be democratic. Some might grant that the G20 does not live up to the all-affected principle but object that the G20 just does not make policy decisions that have to be taken democratically. Certainly, discussions in the G20 influence policy decisions but it cannot be required that all affected persons are able to participate in all causal processes that contribute to a policy decision. Opportunities of political participation in the final decisions (of the member states or international organisations that implement G20 commitments) might be provided via other channels than via participation in G20 decision-making.
However, we should ask in what other ways those who do not have opportunities to participate in G20 decision-making could influence the relevant policies. As already mentioned, many global governance institutions that act as intermediaries of the G20 do not provide effective opportunities for political participation. In national-level political institutions at least those that are affected by external effects of policies in other states usually do not have opportunities for participation. 12 One might object that this is not a deficit of the G20 but rather a deficit of decision-making in these other institutions or even of the global governance system as a whole. But still, we have to ask what the G20 contributes to this deficit and what could, in principle, be the contribution of an alternative institution in providing all those affected opportunities for political participation. As argued above, it is not inherent in the two functions of the group that their exercise harms opportunities for political participation. In addition, if the democratic deficit is ascribed to the global governance system as a whole, every single institution can argue that other parts of the system are responsible for providing the opportunities for political participations that the all-affected principle calls for. We should not let institutions such as the G20 so easily off the hook.
Others might accept that the G20 has to conform to democratic values, but question the all-affected principles as suitable standard. Some scholars defend, for instance, the socalled all-subjected principle as alternative standard of inclusion (Miller 2009;Fraser 2010). According to (one version of) this principle, only those who are subjected to laws are entitled to opportunities for participation in decision-making. For limitations of space, I cannot make a case in this paper for the all-affected principle vis-a-vis the all-subjected principle. 13 And I also cannot comprehensively discuss what parts of my argumentation would still hold if the all-subjected principle is used as standard of democracy. However, it should at least by indicated that some of the shortcomings described above also seem to be problematic from the perspective of this alternative principle. After all, in the G20 commitments are made whose implementation requires legislative changes by the member states. Certainly, the G20 itself does not formally decide on any laws. But as I argued above, if parliaments only subsequently approve and realise decision prepared in the G20, this can reduce opportunities for democratic participation of citizens.
The reference to the all-subjected principle might make a greater difference with respect to the entitlement to opportunities for participation of citizens of non-member states. In many cases, citizens of non-member states are affected by G20 decisions implemented by international organisations or the G20 states, but not subjected to any relevant laws. However, even with respect to the citizens of non-member states the all-subjected prinicple can be violated by the G20 because these states can come under pressure to adopt regulatory measures that go back to G20 decisions. 14

Institutional reforms
This section presents a number of reform proposals that could, to some extent, alleviate the democratic shortcomings identified above. In Section 1, I argued that the diagnosis of a democratic deficit is warranted if an institutional alternative meets the requirement stated by the all-affected principle to a higher degree than the institution in question. As the all-affected principle allows for different ways in which affected persons can be provided opportunities for political participation, I will describe in the following different channels of participation in G20 decision-making that could be improved. By developing realisable reform proposals, it is shown that there are feasible and more democratic alternatives to the G20 in its current form. This warrants the diagnosis that the G20 has a democratic deficit.
This paper focuses on reform proposals that retain, to some degree, the general nature of the G20 as an institution whose decision-making takes place in meetings mainly of representatives of states. This is done to increase the likelihood that the proposals made are in fact feasible. It should not be denied, though, that some of the described shortcomings are inherent to the summit format. More far-reaching reform proposals that might realise the all-affected principle to an even higher degree would question this format and suggest, for instance, the establishment of an independent global parliamentary assembly of representatives of national parliaments or even a directly elected world parliament that could take over the functions of the current G20 (Falk and Strauss 2001;Marchetti 2008, chapter 7). However, the establishment of such institutions does not seem to be very probable in the near future and it is controversial as to whether the conditions under which they are justified are currently met in the global realm (Christiano 2010).
The suggestions made in the following do not constitute fully developed proposals but should only point to potential directions of progress, which are worth a more comprehensive investigation. I first consider potential reforms meant to diminish democratic shortcomings concerning the citizens of countries that are not members of the G20 before discussing suggestions aimed at those shortcomings that concern citizens of member countries.
With regard to the exclusion of many states from G20 decision-making, reforms of different levels of ambition could be pursued. The reform that would require the smallest amount of institutional change concerns the relationship of the G20 with other, more inclusive international processes. When it exercises influence on other global governance institutions and on international regulations the G20 could adhere to goals and principles that have been agreed on in more inclusive political processes. The 'Universal Declaration of Human Rights' from 1948 or the more recent '2030 Agenda for Sustainable Development' include such goals and principles. 15 If the G20 consequently oriented its activities according to such frameworks, the democratic shortcomings caused by the lack of representation of the part of the world's population living in countries that are not members of the group would be somewhat attenuated. Certainly, in this case, the citizens of non-member states would still have no voice in G20 decision. This might be somewhat less of an issue, though, if they, at least, would have a voice in the formulation of the goals and principles that could guide G20 activities.
But even if the G20 put itself in service of these goals and principles, it is doubtful whether this would be an adequate response to the described shortcomings. The 2030 Agenda, for instance, describes a number of goals to which a set of indicators is attached that are used to follow-up on these goals. It leaves open what specific policies should be taken to realise these goals. Decision on these policies should, from the perspective of the all-affected principle, not be left to an exclusive group in the form of the G20. Therefore, more ambitious reform proposals should tackle the current membership of the group.
In this line, Ocampo and Stiglitz (2011) suggest replacing the G20 with a Global Economic Coordination Council (GECC) within the UN system. 16 According to their proposal, the GECC should, like the current G20, meet at the level of heads of state and government, be supported by the relevant international organisations and decide usually in consensus. The GECC should include representatives of all systemically relevant countries, while groups of smaller countries would share a representative in the council. Ocampo and Stiglitz (2011, 10) suggest that economic weight and, eventually, population size should be used as criteria to determine which countries have a seat on the council and which countries have to share a representative. However, of these two criteria only population size is in line with the all-affected principle. Such a transformation (or replacement) of the G20 would be a democratic achievement because the citizens of those countries that are currently excluded from the group would gain a voice.
With regard to the democratic participation of the citizens of G20 member states, a number of problems were identified in the last section: the predominant role of executive bodies in the G20 process, a lack of transparency, the unequal capacity of member states to influence G20 decisions, the non-democratic nature of some member states and the problem of growing constituency. In general, these democratic shortcomings speak in favour of limiting the array of national policies that are coordinated by the G20 to those policies that have significant external effects on other member countries. As described above, in these cases international coordination can have a positive impact on opportunities for democratic participation because those affected by the external effects gain a voice in decision-making about these issues. If no external effects are at stake, the all-affected principle speaks against a coordination of national policies in an institution such as the G20.
The problems related to the character of the G20 as a process limited largely to executive bodies and its lack of transparency can partly be addressed by changes in the working procedures of the G20. In this regard, the establishment of an independent work stream with meetings of representatives of national parliaments or, preferably, the participation of representatives of the thematically relevant parliamentary committees in G20 working groups might be desirable. This would also to some degree respond to the suggestion of Forst (2015, 103) and others to open up international fora to opposition parties to facilitate the contestation of policies by different political communities. Certainly, it would be hard to reach consensus on such reforms because parliaments do not have the same central role in all G20 member states and governments in general will hesitate to weaken their position in favour of their national parliaments. However, as a start, individual G20 member states could decide for themselves to include parliamentarians in their negotiation delegations.
With regard to the second problem, the working arrangements of the G20 could be changed to improve transparency. For instance, the publication of agendas, lists of participants and minutes of G20 summits and working group meetings would increase the amount of publicly available information without creating any significant costs. G20 meetings might also include public sessions where media representatives could observe the discussions. Transparency could also be increased by a more systematic, coherent and accessible reporting of the G20 working groups on their activities. 17 Furthermore, the engagement of G20 working groups with civil society organisations could be strengthened to enable more effective political participation (Scholte 2016, 24). If decision-makers have to justify G20 policies to outsiders and listen to their proposals, this can additionally improve the role of deliberation in the G20 process, which is often seen as a democratic virtue (Slaughter 2017). Being aware of the democratic potential of engagement with civil society, the G20 undertakes dialogue processes with several so-called engagement or outreach groups that are formed by representatives from business (B20), trade unions (L20), civil society organisations (C20), youth organisations (Y20), think tanks (T20), academies of science (S20), and women's organisations (W20). The engagement groups prepare recommendations for the G20 governments and organise workshops and conferences where representatives of the respective stakeholder groups and political decision-makers meet and discuss G20 policies.
However, it is debatable how much influence the engagement groups currently really have. For instance, while they also emphasise the potential of the G20 engagement group process, Chodor (2020) and Slaughter (2019) argue with respect to the C20 that the design of the process constraints the ability of civil society to contest G20 policies. One reason for this is that the G20 exercises influence on who participates in the C20 (e.g. the G20 presidency usually mandates a civil society organisation from its own country to organise the process) and controls the access of the group to the official G20 working groups. In addition, the C20 can usually only react to the agenda of the G20 instead of raising their own topics. These limitations arguably also hold true for the other engagement groups. Furthermore, many of the groups only have access to rather limited financial means for their working process.
A step forward with respect to the engagement of the G20 with civil society could be to give the exchange between G20 policy-makers and representatives of engagement groups a more continuous and formalised basis. For instance, while meetings between representatives of engagement groups and participants of the official G20 process occasionally already took place in the past, G20 working groups could invite representatives of civil society and especially affected groups according to a fixed schedule to their meetings and thereby give them the opportunity to express their views on G20 plans and present their own proposals. Such exchanges should take place at an early stage of each year's G20 working process to facilitate its impact on decision-making. At the same time, it is crucial that the engagement groups operate as independently as possible from national governments to avoid 'the danger of civil society being co-opted and managed' (Slaughter 2019, 46) in the engagement process.
In general, it is not easy to ensure, though, that the organisations that have the capacity to make use of such opportunities for engagement do not also primarily represent privileged segments of the societies of a limited group of countries and that dialogue processes with them go beyond mere elite participation (Scholte 2016, 24f.). For instance, it is often argued that business organisations have, partly by making use of the B20 process, a great influence on G20 decision-making, which damages political equality (Martens 2017;Chodor 2020). But also civil society organisations that represent non-commercial interests are often dominated by rather privileged groups. It is thus an important challenge for the engagement groups to ensure a diverse membership, which should go beyond organisations from G20 member states.
Regarding the inequalities between G20 member states with respect to their capacities to influence G20 decision-making, measures to decrease knowledge asymmetries can be beneficial. In this regard, research institutions in less wealthy G20 member states could be promoted. Furthermore, a newly established G20 secretariat might be able to provide independent research and policy advice to all member states (Ocampo and Stiglitz 2011, 9). 18 Since the G20, as yet, has no secretariat, it often builds on expertise provided by international organisations that participate in G20 meetings (e.g. the OECD, IMF and World Bank). However, these international organisations are usually also controlled by rich countries.
In addition, less wealthy member states could receive financial support from a fund financed by other member states to allow them to increase their staff occupied with the G20 process and attend G20 meetings with larger delegations. These measures seem to be feasible steps forward, although they would obviously not completely remove differences in the capacities of member states to influence G20 decisions.
As regards the problem that not all member states are democratic and the population of these countries therefore lacks proper representation in the group, there does not seem to be a good remedy with a significant chance of implementation. To exclude the respective countries would, first, not solve the problem because the citizens of these countries would then still not have a voice in decision-making. Secondly, effective international coordination can usually hardly be achieved if all countries are excluded that are not fully democratic. However, a mechanism leading to the exclusion of governments in cases of grave human rights violations might be an option. This would ensure that rulers that clearly turn against their own populations are not recognised as their representatives in the group. The feasibility of such proposals is reduced, though, by the challenge of achieving agreement on the exact conditions of such a mechanism.
Although I cannot comprehensively argue for this claim here, it seems unlikely that the implementation of such reforms would substantially reduce the ability of the G20 to reach agreements and facilitate effective policies. 19 For instance, creating more transparency can increase the pressure on governments that block certain policies because transparent decision-making procedures allow governments to be publicly held to account. 20 Engagement with civil society organisations might sometimes help to break gridlocks because transnational civil society can provide additional dialogue fora for participants from G20 states. Strengthening the role of parliaments in the process could improve the implementation of G20 policies in member states because parliaments would then have a greater ownership of G20 decisions. In addition, a stronger role for parliaments in the process might also increase continuity because opposition parties that might take over government after the next election would already have been included in the process. The more fundamental transformation of the G20 to a GECC would also not necessarily be detrimental to its ability to reach agreement because such a council would not have to include a higher number of members.
Certainly, there might also be mechanisms that lead some of the described reforms to have a negative impact on effective decision-making. For instance, greater transparency could sometimes make it harder for governments to make compromises because they might fear being portrayed at home as weak leaders that are unable to assert their position in international politics (Stasavage 2004). However, it is by no means clear that the impact of the reforms described on effective policy-making would be negative overall.
Finally, it needs to be emphasised again that the democratic shortcomings of the G20 process would not be removed but only alleviated by the described reforms. However, assuming that many of these reforms are indeed feasible, the discussion in this section already shows that, in its current form, the G20 does significantly worse in terms of the all-affected principle than feasible alternatives and can therefore be judged to have a democratic deficit.

Conclusion
This paper assessed the G20 with respect to the realisation of the all-affected principle as a central democratic requirement. In light of the global impacts of G20 decisions, the exclusion of the citizens of non-member states in G20 decision-making is a serious violation of the all-affected principle. With regard to the opportunities for political participation of citizens of G20 member states, democratic shortcomings include the predominant role of executive bodies in the G20 process, a lack of transparency, the unequal capacity of member states to influence G20 decision-making and the lack of democracy within some G20 member states. Although this paper could not argue for this claim, the decision-making procedures of other club governance institutions seem to be deficient in very similar ways. The G7, for instance, shares many of the deficits of the working procedures of the G20 with respect to transparency and parliamentary control. The exclusive nature, which becomes problematic if decisions are taken that affect people living in nonmember states, is also shared by other club governance institutions.
Feasible reforms could at least alleviate the democratic shortcomings of the G20 described by improving different channels of political participation. For instance, the membership of the G20 could be changed so that larger countries are included while smaller countries share a representative in the group. In addition, the role of parliaments in the G20 could be enhanced, working processes could be designed to be more transparent, engagement with civil society organisations could be strengthened and knowledge asymmetries between governments could be reduced.
Even if more far-reaching proposals that question the general format of the G20 as a summit process of states were currently not be feasible, these reform options would already show that the G20 in its current form does significantly worse in terms of the democratic requirement expressed by the all-affected principle than feasible institutional alternatives. This warrants the conclusion that the G20 exhibits a democratic deficit.

Notes
1. The only exceptions in this regard that I know are Slaughter (2017), Bendiek and Neyer (2016) and Scholte (2016 3. The diagnosis of a democratic deficits is also relevant in the context of recent research that links the rise of authoritarian populism in many countries to shortcomings of the existing democracies. According to Zürn (2021), these shortcomings are, among others, related to increasing competencies of international organisations that are not directly democratically accountable to the citizens. 4. As argued in Section 1, a lack of capacity for effective governance can also be a democratic shortcoming because democracy requires opportunities for effective democratic participation. Therefore, if there really were a trade-off between effectiveness and some democratic requirements in the design of club governance, strictly speaking this choice could sometimes also be described as a choice between different democratic requirements. 5. The principle is not undisputed, though. I briefly come back to the objection that the G20 should not be assessed against the all-affected principle in Section 3. 6. This is very much in line with the way in which Moravcsik (2004) frames the question of whether an institution has a democratic deficit. He states: 'Any democratic metric derived from ideal theory must therefore be "calibrated" in order to assess whether the current arrangements are the best that are feasible under "real-world" circumstances' (Moravcsik 2004, 337). However, Moravcsik opines that one should make comparisons with existing democratic states, while I believe that comparisons with other existing political institutions, and even currently not existing but presumably feasible institutions, should also be taken into account. 7. However, the G20 has sometimes not only the weak relationship with other global governance institution that is highlighted in the orchestrator-intermediary model. Even if the G20 as institution has no direct control over intergovernmental organisations, its member states often do have relatively direct control over them. 8. However, the pledging of financial resources is less common in the G20 than in the G7. 9. In the past, consultations of G20 parliaments were also held occasionally (for instance, the Parliamentary Forum and Speakers' Summit in 2018). However, such events only have a marginal role within G20 processes. 10. Democratic procedures in other G20 states also suffer from serious shortcomings, for example with regard to high economic inequality translating into differential capacities to influence public policies (Gilens and Page 2014;Schäfer 2015). 11. In fact, compared to the decision-making in some other global governance institutions, G20 decision-making might, despite the shortcomings described, sometimes already be less technocratic. In the G20, decisions are at least not left entirely to the bureaucracies of international organisations, but elected politicians are brought into the process at the summits. Summit meetings can also lead to a politicisation as public attention can bring issues of global governance into the public debate and allow for public contestation and resistance, at least if the summits and their preparatory processes are conducted in a reasonably transparent way. Counter events and demonstrations can give a voice to groups that might otherwise not be heard in global governance and can thereby be a side effect of summits, which is valuable from a democratic perspective (Scholte 2016, 19). 12. Furthermore, I argue above that the G20 to some degree constraints the de-facto competencies of national parliaments and thereby has a negative impact on opportunities for political participation in its member states. 13. For critical discussions of the all-subjected principle, see, for instance, Arrhenius (2018, 105ff.) and Valentini (2014, 792f.). 14. An example in this regard is the Basel III regulatory framework for banking supervision that was decisively advanced by the G20 (though developed by the Basel Committee on Banking Supervision). For the question of how these standards spread to many countries of the Global South, see Jones (2020). 15. One could also question, though, whether such frameworks are developed through sufficiently democratic processes, as they often also suffer from the shortcomings related to executive multilateralism, the existence of power asymmetries and the participation of non-democratic governments. However, the democratic deficit that only a very limited number of states participate in G20 decision-making does not apply to the adoption of these frameworks. Efforts of the G20 to align its activities with the 2030 Agenda led to the adoption of the 'G20 Action Plan on the 2030 Agenda for Sustainable Development' (G20 2016) under China's G20 presidency in 2016 and regular 'updates' of this plan in the following years. Nevertheless, the G20 is still far from comprehensively putting itself in service of 2030 Agenda (Martens 2016). 16. According to the proposal of Ocampo and Stiglitz (2011) the new GECC should not evolve from or replace the Economic and Social Council (ECOSOC), which is currently responsible within the UN system for economic and social issues and the coordination of the specialised agencies of the UN (although it is highly questionable whether it is capable to exercise this coordinating role). The GECC would instead be a new body that would also provide direction to ECOSOC (Ocampo and Stiglitz 2011, 11f.). 17. Some G20 working groups (e.g. the G20 Development Working Group) already publish accountability reports. For a discussion of deficits of the G20 with regard to transparency and accountability, see Hilbrich and Schwab (2018). 18. Similar proposals have been made with regard to knowledge asymmetries within the WTO (Birkbeck 2009, 28). 19. However, the main conclusion of this paper that from a democratic point of view the G20 has important shortcomings because more democratic alternatives are feasible does not depend on the claim that these alternatives will to the same extent be able to generate agreement and be as efficacious as the current G20.
20. In this line, Benson and Zürn claim that '[s]ecret handshakes and backroom deals are not only bad for global democracy, they also create bad policy' (Benson and Zürn 2019, 6).