New Powers and the Distribution of Preferences in Global Trade Governance: From Deadlock and Drift to Fragmentation

ABSTRACT Existing theories make divergent predictions about the impact of new powers on the global political economy. Some argue that a more even distribution of power will erode international cooperation, while others argue that cooperation can continue with the help of international institutions to overcome collective action problems. We argue that this debate overlooks a critical determinant of the shape of power transitions: the distribution of preferences amongst the major powers. It is primarily in the context of divergent preferences that power transitions are likely to give rise to conflict. Moreover, even where preferences diverge, the gains of cooperation provide a strong incentive to continue to pursue goals through multilateralism. This situation leads to forms of institutional change unanticipated by established theories. These include deadlock in expansive multilateral fora, institutional drift as old rules cannot keep up with the changing political and economic context, and fragmentation as countries seek minilateral solutions that reduce preference diversity. We develop this preference-based, institutional argument by examining the distribution of preferences and institutional change at the World Trade Organization (WTO) and its Doha Round, where the power transition is relatively advanced.


Introduction
For the first time in the modern era, non-Western countries are becoming major world powers and taking on an increasingly central role in the global economy. While it is widely accepted that China, and to a lesser extent India, Brazil and other emerging economies, have shifted the distribution of international power, the implications of this shift for the global political and economic order are the subject of debate. Existing theories make divergent predictions about the impact of the rise of new powers on international order. The classical, realist view is that a more even distribution of power will erode cooperation as the hegemon is no longer able to provide public goods and cajole subordinate powers into line (Krasner 1976, Kindleberger 1981, Layne 2012. 1 World systems theory comes to similar conclusions, associating the expansion of international trade with eras of hegemonic dominance that provide geopolitical stability (Chase-Dunn et al. 2000, Kwon 2012). Some authors today therefore see a fundamental challenge to Western supremacy underway that could endanger the international order (Lemke and Tammen 2003, Arrighi 2007, Gat 2007, Mearsheimer 2010, Schweller and Pu 2011, Kupchan 2014. Others have identified logical problems in the 'hegemonic stability' thesis (Snidal 1985) and emphasise that cooperation can continue with the help of international institutions, which can disseminate information and overcome collective action problems (Keohane 1984, Milner 1992. In line with this view, many today see rising powers as rather being incorporated into existing institutions amidst continued cooperation (Ikenberry 2011a, Kahler 2013. We argue that this debate has yet to fully appreciate a critical determinant of the shape of international power transitions: the distribution of preferences amongst the major powers (Ruggie 1982, Cox 1987, Moravcsik 1997. It is primarily in the context of divergent preferences that power transitions are likely to give rise to conflict, otherwise harmony would prevail. 2 The peaceful shift of power from Britain to the United States, for example, is often explained by the similar identities and convergence of fundamental preferences between these two countries , pp. 13-15, Kupchan 2010. Likewise, Germany and Japan were incorporated into the existing order because their preferences largely converged with other G7 countries. Understanding how power transitions play out therefore requires careful empirical examination of the distribution of preferences (Kahler 2010, Kahler 2013. Moreover, we argue that, even where preferences diverge, the necessity of cooperation in light of widespread socio-economic interdependence provides a strong incentive to continue to pursue goals through multilateralism. This leads to forms of institutional change unanticipated by established theories. Deadlock in expansive multilateral fora gives rise, over time, to a growing mismatch between old rules and the changing political and economic context ('drift', see Mahoney andThelen 2010, Hale et al. 2013). Over time, countries seek minilateral solutions that reduce preference diversity, contributing to institutional fragmentation (Keohane and Victor 2011, pp. 8-9).
In this paper, we outline these theoretical claims and assess them by examining the case of the World Trade Organization (WTO), where the power transition is relatively advanced. We provide a systematic empirical snapshot of the distribution of preferences amongst the major trading powers at the time of the key negotiations of the Doha Round (the period 1996 to 2011). We find that new powers' preferences diverged significantly from the established power average, contributing to the diversity of preferences underpinning the Round and leading ultimately to the failure to come to an expansive multilateral agreement. The result is a growing gap between WTO trade rules, still largely unchanged since the Uruguay Round, and the contemporary distribution of power-weighted preferences. This gap has only grown as powerful voices in the United States now appear to question the value of the WTO in general (Swan 2018). Over time, this has increased incentives to pursue outside options via minilateral initiatives in the form of regional trade agreements. While we develop this argument in the case of the WTO, the logic of the argument applies in other areas and may indicate a likely outcome of post-hegemonic global governance.
The first section of this paper surveys existing approaches to the impact of rising powers on international institutions, and suggests an alternative theoretical approach which builds on preferencebased accounts and complementary variants of rational institutionalism. The second section applies this model to the emergence of new trade powers (centrally, Brazil, China and India, but we also include South Africa) in the WTO. In line with previous studies (Kahler 2013, Hopewell 2016, we focus particularly on these 'BIC' countries who have achieved a prominent and coordinated role in the WTO, while acknowledging that China occupies its own category in terms of material power resources. We incorporate Russia and South Africa (to make up the BRICS) into our analysis as much as our data allows. Based on primary and secondary data sources, as well as a systematic content analysis of major powers' negotiation positions in the Doha Development Agenda (DDA) round of negotiations, our evidence reveals the greater diversity of power-weighted preferences as a result of the rise of new powers. The third section shows how this new distribution of power-weighted preferences interacted with institutional features to produce institutional drift in the form of the Doha deadlock and a turn towards fragmentation via regional trade agreements, primarily in the form of 'mega-regionals'. A final section outlines how this model of preference-driven power transitions illuminates the dynamics of other institutionalised issue areas and suggests that fragmentation around likeminded minilateralism may be an expected outcome of institutionalised power transitions.

Power Shifts and Institutional Change
There is by now a considerable literature considering the roles and impact of new major powers on the international economic and political order. 3 One early line of enquiry considered the question of whether new powers, such as China, could be considered revisionist or status quo powers, that is, whether they want to change the rules governing political and economic relations (Shambaugh 2001, Johnston 2003, Chan 2004a, 2004b, Alden and Vieira 2005, Hurrell 2006, Huiyun 2009, Kahler 2010, Kastner and Saunders 2012, Wang and French 2014. This important literature was primarily focused on conceptual and descriptive issues, and relied often on single or comparative country case studies, with China receiving particular attention. This is not surprising, as China has so far been the only new power to acquire a truly systemic role, rising from around 2 per cent of the world economy in the early 1990s to 15 per cent in 2017 in market exchange rate terms. 4 Yet also other regional and multilaterally active countries, especially the five BRICS states, have received much attention (Armijo 2007, Glosny 2010, Stephen 2012, Brütsch and Papa 2013, Armijo and Roberts 2014. Moving from description to explanation, other authors sought to explain the multilateral behaviour of rising powers. These studies ranged from accounts emphasising domestic interest groups (Schirm 2012, Blanchard 2013, da Conceição-Heldt 2013, domestic political economic conditions , status concerns (Narlikar 2006, Wolf 2014, to structural material and institutional constraints (Ikenberry 2011a, Stephen 2014. Today, attention has turned to examining the consequences of rising powers for international institutions. Nonetheless, much of this debate has been framed in terms of new powers 'challenging' or 'overthrowing' existing institutions, or alternatively being 'integrated' and 'assimilated' into the status quo (Mearsheimer 2010, Schweller and Pu 2011, Ikenberry 2011a, Cai 2013, Kahler 2013, Krieger and Nolte 2016, Zangl et al. 2016). There remains a tenacious habit of depicting these changes in dichotomous terms.
Conceptually and theoretically, this paper builds on this existing literature to develop a more explicit and nuanced approach to modelling the impact of new powers on international institutions, and measuring it empirically in the crucial case of the WTO. 5 Central to this approach is the distribution of power-weighted preferences (Moravcsik 1997, pp. 520-1, Frieden 1999, Keohane 2012. 6 Institutions of global governance are likely to be challenged by international power shifts, but predominantly in areas where the preferences of the established and the rising powers diverge. This also implies that international power shifts should be understood in terms specific to particular issue areas and institutional domains; major powers' preferences may converge in one area, but diverge in others. We seek to elaborate why this is the case in theoretical terms, and to do so by drawing on preference-based and institutionalist accounts of international political change. In general, there is a need to flesh out the mechanisms linking international power shifts to changes in international institutions, and to go beyond dichotomous categories to describe these changes. Our theoretical approach can be outlined in four core propositions.
First, in common with other societal approaches to International Relations (IR), for analytical purposes the point of departure of our approach is the international distribution of preferences. From this perspective, states are not functionally undifferentiated units in an anarchic environment (cf. Waltz 1979), and the fundamental nature of international politics has not remained unchanged for millennia (cf. Gilpin 1981, p. 7). Rather, states are embedded in a domestic, transnational and global social context that shapes what interests they represent and what their goals are (Cox 1987, Moravcsik 1997, pp. 516-21, Skidmore 1997. 7 State power is used in the pursuit of sociallyderived preferences, which can both vary and are not necessarily in conflict with each other (Moravcsik 1997, Legro andMoravcsik 1999). Where state preferences converge, there is little need for international power shifts to impede cooperation. If preferences converge sufficiently, international power shifts will likely take place 'within hegemony', rather than threatening hegemony per se (Ruggie 1982, Cox 1983. In this perspective, the fundamental source of disagreement over multilateral outcomes is preference divergence (Moravcsik 1997, Frieden 1999, with negotiation strategies (Wilkinson 2009, Narlikar 2010 constituting an intervening variable. Because preferences play such an important role, they receive considerable attention in the empirical analysis that follows. Second, while preferences are critical to the results of international power transitions, we follow realist and critical approaches in allocating a large explanatory role for power-distributional factors. Not everybody's preferences are equally important: preferences need to be 'power weighted', and this inevitably directs attention foremost to the major powers in any given domain (Drezner 2007). In the context of intergovernmental negotiations, this approach emphasises that power amongst the member states is an important determinant of outcomes. The extent of the challenge to existing institutions will therefore be a function of the extent of the shift in international power, combined by the extent of preference divergence between rising and established powers. Greater systemic diversity can be expected to lead to a narrower overlap of 'win sets', which all else being equal, makes common agreements harder to achieve (Putnam 1988).
Third, in the contemporary context of 'globalisation' and deep interdependence, in cases where preferences diverge but are not diametrically opposed in a zero-sum fashion, there remains scope for international institutions to permit the realisation of joint gains via deliberate policy coordination (Keohane 1984, Koremenos et al. 2001. Nonetheless, we assume that institutions are to some extent path dependent and that there is therefore a lag between changes in underlying power and preferences and changes in the institutional form (Pierson 2000, Rixen and Viola 2014, Hanrieder 2015. Sometimes, institutional change may be blocked entirely. Along with the compromises necessary to secure institutional legitimacy and the inherent potential for 'slack' in the relationship between the states as the principals and the IO as the agent, there is the possibility for existing international rules to differ from what the current distribution of power-weighted preferences would suggest. In particular, institutions are designed in ways that privilege the most powerful actors at the time of their creation (Krasner 1985). Functional gains are often unequally distributed and reflect different preferences disproportionately. The WTO, for example, was largely a result of the initiatives of the United States and Europe, and embodied rules particularly favourable to their preferences at the time (Steinberg 2002, Kim 2010. In the absence of comprehensive institutional reforms, institutions and their rules may become incongruous from the underlying distribution of power and preferences (Krasner 1985, pp. 75-81).
This leads to our fourth proposition, which is that in the context of interdependence and a new diversity of power-weighted preferences, the likelihood of deadlock in expansive multilateral settings increases, as does the incentive to pursue minilateral solutions via selective institutions (Kahler 1992). This can reduce preference diversity and overcome bargaining problems, but it comes at the cost of cooperation's breadth. This results in forms of institutional change unanticipated by established theories. Divergent preferences are prone to deadlock in expansive multilateral fora (Keohane 2012, Hale et al. 2013. The direct consequence of this is institutional drift, where an institution is unable to be updated to reflect new circumstances (Streeck andThelen 2005, pp. 24-6, Mahoney andThelen 2010, p. 17). Countries seek minilateral solutions that reduce preference diversity. Over time, the incentive to pursue outside options increases through avenues such as regime shifting (Helfer 2009). This leads to a more fragmented institutional environment. As such, our approach suggests that the rise of new powers may produce neither institutional decline (Layne 2012) nor simply assimilation of new powers into existing institutions (Ikenberry 2011a, p. 58), but rather alters the modality of cooperation towards new institutional processes such as drift and regime shifting.
In contrast to hegemonic stability theory, this approach emphasises that preference divergence is a necessary condition for power shifts to generate institutional change, while all major powers have an incentive to maintain institutionalised cooperation in principle. Yet in contrast to a purely functionalist approach, it emphasises that institutions have effects on the distribution of power, which contains the seeds of institutional conflict in conditions of exogenous power shifts. Finally, we draw on the insights of historical institutionalism by incorporating the notions of institutional path-dependence and the processes of drift and fragmentation that can arise from increased incongruence between an institution and the power-weighted preferences of its members (cf. Zangl et al. 2016).
In the subsequent sections, we apply this model to the case of the WTO. We begin by describing systematically the new distribution of power and preferences. Our analysis reveals that the difference in the preferences of new and established powers on substantive issues is in fact considerable, which helps to explain the new contentiousness of multilateral trade negotiations. Subsequently, we trace the institutional effects of this new systemic heterogeneity in the form of deadlock, drift, and fragmentation. We argue that this model provides a more comprehensive description and explanation of the institutional consequences of the ascent of new powers than prominent alternative models. A final section concludes.

New Powers and the Distribution of Preferences at the WTO: Empirical Analysis
With Russia's accession, WTO members currently account for 97 per cent of world trade and a similar portion of world population and GDP (World Trade Organization 2014). The WTO's more than 60 legally binding agreements, and its monitoring and enforcement capacities, have contributed decisively to the legalisation of the liberal international economic order, and have served to lock-in its members' commitments to trade openness (Wade 2003, Chorev 2005, Goldstein et al. 2007, Baccini and Kim 2012. By entrenching principles of openness, non-discrimination and liberalisation, the WTO has become a keystone of the liberal international economic order that was fostered by the United States and other established powers in the period since the Second World War (Cox 1987, Gilpin 1987, Chorev 2005, Ikenberry 2011b). The Doha negotiations represent to date the most wide-ranging attempt yet to renegotiate the rules of the liberal international economic order. As such, the WTO constitutes an excellent case in which to examine the impact of new powers on the international political and economic order.
Nevertheless, the WTO and especially its rule-making pillar embodied in the DDA has for a long time faced severe and protracted negotiation deadlocks. At least since the 2003 Cancún Ministerial Conference, the DDA has seen a series of crises and revivals, and already in 2011 the former US Trade Representative proclaimed it doomed (Schwab 2011). In spite of small advances in the form of the socalled Bali (2013) and Nairobi (2015) packages, today, after 17 years of negotiations, the round is effectively dead, though it is unclear how exactly the rule-making mandate of the WTO is to continue (Financial Times, 2015).

Power Shift
As has been widely analysed, in recent years, the WTO has seen the emergence of new powers whose economic growth and institutional activism has thrust them into the core of its decision making procedures (Mattoo and Stern 2003, Narlikar 2010, Baracuhy 2012, Vickers 2012, Zeng and Liang 2013, Hopewell 2015. This power shift has occurred along three core measures of power in the international trade regime. Firstly, the traditional measure of power in trade negotiations remains market size, and the associated capacity to grant or deny access to import markets (Krasner 1976, Hirschman 1980[originally 1945], Steinberg 2002, pp. 347-8). Large economies are also typically less dependent on foreign trade due to the size of their internal markets, giving them added leeway when dealing with smaller economies. 8 Table 1 provides a summary of the core trends in market size of the BRICS and G7 countries in terms of world GDP and global imports. Most notably, between the 1990s and 2010s, China increased its share of global GDP from 2.4 per cent to 12.8 per cent; while the United States fell from 26.8 per cent to 22.7 per cent. While Japan, the four largest European economies, and Canada all experienced relative decline, India also doubled its share from 1.2 per cent to 2.7 per cent. When combined, the BRICS countries rose from less than 8 per cent to more than 21 per cent of the global economy, while G7 countries declined from 66 per cent to 47 per cent. 9 Secondly, influence in the trade regime can be linked to trade volumes (Narlikar 2010, pp. 718-19). The last three columns of Table 1 provide summary data for countries' shares of world imports volumes. As well as increasing their market shares, large emerging economies have become far more significant and central to flows of global trade (Hanson 2012), and have amassed an unprecedented share of total trade volumes. While the United States has remained relatively stable and still accounts for 12.7 per cent of global imports of goods and services, between the 1990s and 2010s, China rose from 1.8 to 9 per cent; India also expanded its imports from 0.7 to 2.4 per cent. In total, BRICS countries almost tripled their imports, while the G7 countries declined from around half of the world total to around one third.
While these shifts are significant, particularly in the case of China, they only convey a partial picture of how new powers have risen in the trade regime (Hopewell 2015). Over the course of the Uruguay Round, and in particular in the preparatory and negotiation stages of the WTO's DDA, the governments of new powers have also become central protagonists in the politics of the WTO in its rulemaking, rule-monitoring and rule-implementation stages (Narlikar 2010, Zeng 2013, Júnior et al. 2015, Hopewell 2016, Karlas and Parízek 2017. This has included a renewed attempt to forge developing country coalitions to boost bargaining power in negotiations (Cho 2004, Narlikar and Tussie 2004, Narlikar and Wilkinson 2004, Taylor 2006, Efstathopoulos 2012, Doctor 2015. While China is by far the largest new power economically, already since the 2003 Cancún Ministerial Conference, it has been Brazil and India, partly with the support of South Africa, that have led the effective opposition to the established powers in the WTO (Hopewell 2015, p. 324).
Together, changes in these three dimensions of influence have altered the power relations that previously underpinned the Uruguay Round of the GATT and the early years of the WTO. Yet, according to our model, a more equal distribution of power due to the rise of new trade powers will only generate challenges to the institutional status quo where it coincides with divergences of preferences.

The Distribution of Preferences
While new powers have risen in the global trade regime, we expect this will only generate conflict to the extent that their preferences diverge from those of established powers. To what extent, and on which issues, do the preferences of major trade powers really diverge?
Because preferences are unobservable, they can only be approximated. We measure preferences in two ways, which we cross-check against background knowledge and interviews with WTO trade representatives. The first indicator which we examine in this subsection consists of the major trading powers' trade and domestic economic profiles, encompassing both 'at the border' interface management techniques and 'behind the border' domestic institutions of capitalism. The second, discussed in the following subsection, are the preferences stated by the governments of the major trade powers during the Doha Round negotiations as expressed in statements at Doha Ministerial Conferences. We thus combine both attributional and observational approaches to measuring preferences. 10 Table 1. Share of selected countries and country groups on world market size (GDP, current USD) and imports (current USD). 38

Economic Policy Profiles
In their multilateral negotiations, states typically seek to minimise compliance costs and to extend their competitive advantages (Kahler 1995, Fioretos 2001, Hall and Soskice 2001. Because states differ in their domestic structures, they differ in their preferences regarding international institutions (Katzenstein 1977, pp. 587-8). Existing institutions tend to generate vested societal interests who gain from the existing institutional structure-elites, firms, interest groups, or voters. We therefore assume that maintaining a particular form of economic policy-such as high tariffs in a particular sector, or the level of restrictiveness of intellectual property protections-constitutes a revealed preference for that policy, at least from the point of view of government representatives. Indeed, this is typically how trade negotiations are seen to operate, with liberalising changes from a country's status quo policies considered the major currency of multilateral negotiations (Krasner 1976, Putnam 1988, Hopewell 2016. Consequently, divergent economic policy profiles are likely to generate divergent preferences over multilateral trade rules. To what extent do emerging countries 11 differ in their economic models from those of the familiar established powers? First, it is important to note that despite their embrace of economic globalisation, 12 today's emerging powers have pursued very different at-the-border trade policies to those of established powers, reflecting a new diversity of power-weighted preferences in world trade governance. This is shown in Figure 1, which depicts six different indicators describing the degree of liberalisation of countries'  [1][2][3][4][5][6][7], import taxes [0-100], non-tariff barriers [1][2][3][4][5][6][7]. Source: Tariff and tax data from the World Bank's World Development Indicators, data from 2012 or closest year. Customs and non-tariffs trade barrier index data from the World Economic Forum's Global Competitiveness Report, data from 2014. Overall Trade Restrictiveness Index of the World Bank (2014), data from 2009. 3635 trade policies. In terms of at-the-border interface management, with the notable exception of South Africa, the new powers tend to be less 'liberal' than established powers. This is true in terms of overall measures of trade restrictiveness, average applied tariffs, average applied tariffs weighted by trade volumes, the burdens of customs procedures, taxes on trade, and the prevalence of non-tariff trade barriers. Within our sample, Brazil and India retain the least liberal trade profiles. Indeed, according to the data, India remains the world's least liberalised major economy (Kowalski et al. 2008, p. 284, Safadi andLattimore 2008, p. 11), and its tariffs are even high for countries at comparable levels of national development. By retaining high tariff peaks and many sectors with no binding tariffs at all, the Indian government has retained the capacity to firmly manage its trade with the world. Brazil's integration into the liberal trading order has been similarly cautious, with effective rates of protection close to zero in its unsubsidised and highly efficient agricultural sector, but relatively high in other sectors such as manufacturing (Sally 2008, p. 122). With India and Brazil maintaining the highest levels of protection of any of the major economies, the fact that these countries became key players in the Doha round is suggestive of the new diversity in the distribution of preferences amongst the major powers.
To be sure, none of this is to say that the new powerseven if we only look at the four countries discussed aboveform a coherent block with homogenous trade policy profiles. Indeed, as an example of the differences, China's average applied tariff rate is today around half of that of Brazil (though still twice as that of the EU or the USA). 13 Yet, in the aggregate, the new powers show a systematically lower degree of the openness of their trade policy profiles.
Second, moving to the behind-the-border policy domain, the emerging trade powers also tend to be characterised by less liberal and more state-oriented forms of capitalism at home (McNally 2013, Ten Brink and Nölke 2013, Kurlantzick 2016, Musacchio and Lazzarini 2016. We observe these differences in several arenas particularly relevant to the policy field of the WTO, such as state ownership, intellectual property rights, competition policy, investment conditionality, government procurement, and administrative burdens on private enterprise. Figure 2, for example, depicts the OECD's overall Product Market Regulation index, which measures the extent of nonmarket coordination in product markets, and, secondly, highlights the state control sub-index, which assesses the level of public ownership of enterprises and the government involvement in their operations. 14 Although the differences are of degree rather than kind, the overall comparison of the two groups produces a large and significant difference between new and established trade powers. 15 National development banks, the fostering of national champions, mixed ownership models, and outright state ownership remain key elements of emerging forms of capitalism, contributing to new systemic diversity in the trade system. The diversity of economic models is also visible in other policy fields relevant to the WTO. First, national innovation and technology transfer systems in the new trade powers tend to favour weak intellectual property rights enforcement, driving their 'catch-up' modes of technology transfer (Sell 2003, Nölke 2011. Although this may now be changing in China (Serrano 2016), in contrast to established powers who share an interest in strong intellectual property rights protection, new trade powers have tended to be unenthusiastic adopters of strict intellectual property rights (Nölke et al. 2015, pp. 5-16). This suggests diverging preferences in the area of intellectual property. 16 Second, a recent study evaluating the legal and regulatory framework for competition policy also suggests significant differences between rising and established powers, although the picture was a lot more mixed (Alemani et al. 2013). Governments tend to be especially large customers in emerging markets, and they can use their procurement policies as development tools to potentially limit import competition. None of the BRICS countries is a party to the Agreement on Government Procurement, while all G7 countries are. 17 Third, there is also a new diversity in the preferences revealed by major economies' policies regarding inward foreign direct investment (FDI). While all new trade powers have welcomed FDI, they have done so selectively. Their imposition of conditions, such as enforced partnering with local firms or limits on foreign ownership, differs from the more open approach adopted by most of the established powers. According to OECD data, the G7 countries have negligible restrictions in place (averaging 0.05 on a scale of zero to one), while the average of the BRICS countries is 0.21, with China scoring 0.39. 18 New trade powers therefore differ from established powers 'at the border', but especially 'behind the border' as well. Despite the widely observed move towards international integration and market liberalisation (Becker 2009), the empirical evidence suggests that new and old powers differ strongly in the degree to which they embrace liberal trade and economic policies. A focus on the economic and trade structures of new trade powers suggests a greater reliance on interventionist interface management and a larger economic role for the state, suggesting preference for 'policy space' for the state to manage integration with the world economy.
The adoption of semi-liberal but state-guided paths of development by new trade powers suggests in particular differences with regard to the 'deep integration' (Lawrence 1996) agenda of regulatory alignment. These were to a large extent originally subsumed in the WTO DDA under the label of 'Singapore' issues: particularly the issues of the relationship between trade and investment, competition policy, and government procurement. (The label 'Singapore' issues refers to the Singapore 1996 Ministerial Conference at which working groups for these three areas were established.) This was the case also for the fourth Singapore issue of trade facilitation, which was nevertheless more oriented at administrative and technical matters, such as custom procedures, than at substantive policy agendas.
These three Singapore issues-trade and investment, competition policy, and government procurement-would have dealt with the freedom and ability of the governments to pursue their objectives through interventionist economic policies: imposing stricter conditions on foreign investment, prioritising domestic champions in competition policy, using government procurement strategically, and so on. Such measures have been a major priority of Western multinationals, pharmaceutical companies, and private financial institutions, and have little to do with trade liberalisation as traditionally understood (Rodrik 2018). If these agendas had been included within the new rules negotiated in the DDA, they would have likely extended liberal market disciplines onto important areas of domestic economic regulation (Khor 2003, Evenett 2007. Despite liberalising steps taken by new powers over the previous decades, analysis of these indicators shows that systemic diversity has increased. In the next section, we extend this analysis to cover countries' stated preferences over the course of the Doha Round.

Stated Preferences: Statements in the Doha Round
Analysing the behaviour of major trade powers is a second avenue to empirically estimating the distribution of preferences of the major trading powers. In line with our assumption that states pursue multilateral policies that reflect their domestic economic structures, we expect that, on average, the rising powers will have expounded different and less liberal positions in multilateral trade negotiations. In particular, these differences should be especially acute with regard to 'deep integration' issues that increase the potential for friction between divergent economic systems. 19 To systematically assess major trading states' preferences, we use evidence based on a new dataset of coded statements of WTO members' trade ministers at the Plenary Sessions of biennial Ministerial Conferences. At each WTO Ministerial Conference, almost all Members deliver a speech identifying the state's positions on matters related to the WTO. 20 While there is a mixed audience for such speeches-both the WTO audience and the domestic audience are addressed 21 -we consider this as valid a source of data to infer preferences as any.
Since the foundation of the WTO, one of the key themes to feature in these statements has been the state of play of the Doha round or its preparation. In assessing the states' positions on the issues under negotiation, we manually coded all the available speeches from years 1996 until 2011, identifying all the segments of text where the ministers made political claims referring to the individual areas of negotiation Statham 1999, De Wilde et al. 2014). 22 A drawback of this approach is that we cannot trace changes of preferences over time. Instead, we create 'average positions' for each state over the core negotiation period of 1996-2011. While this undoubtedly obscures important changes over time at the individual country level, our ambition is not to reconstruct the history of Doha negotiations (see e.g. VanGrasstek 2013, pp. 373-461), but to reveal clearly the broader international distribution of preferences over the form of international trade rules in this period in which new powers became central to shaping the WTO agenda.
In our coding procedure, we identify ministerial statements that articulate a position with regard to the adoption of a new regulation in a given area of trade relations. An example would be a call for the adoption of new regulation that will lead to significant trade liberalisation. Alternatively, a member may also claim the need to retain policy space and leave an area up to government discretion (leading to no additional liberalisation, compared to the status quo). The coded claims range in their length from as little as one sentence to as much as several paragraphs. In addition, one segment of text may contain two or more claims, provided that several areas become the object of the claim.
The manual coding locates the states' positions on a scale from 0 to 4. A value of '0' refers to the preference for status quo where no new policy-space restricting regulation is adopted. A value of '4', in turn, signifies a preference for complete liberalisation of the given area and hence for the creation of deep new regulation drastically reducing governments' policy options. Values '1', '2' and '3' represent compromise positions. We coded four key general areas of the negotiations. These include the three classical substantive areas of trade liberalisation talks, defined by the types of products to be addressed: (1) non-agricultural market access (industrial goods, so-called NAMA), (2) agricultural market access, and (3) services market access. In addition, (4), we also coded countries' positions on the cross-cutting Singapore issues mentioned earlier in the text, as representing the new, deep integration agenda. 23 As an example, the following statement by the French Minister for the economy, finance and industry at the 1999 Seattle Ministerial Conference was coded as pertaining to the area of agricultural market access and received a value of '1' in our coding scheme (emphasis added): As far as agriculture is concerned, we wish to keep the tools developed at Marrakesh, including the 'boxes', the special safeguard clause or the peace clause; treat all forms of support in a balanced way and take into account the multifunctional dimension of modern agriculture.
The value of '1' means it is coded as a fairly defensive statement, which nevertheless does not completely deny the possibility of (partial) liberalisation in the given area. Similarly, the following statement regarding agriculture by the Chinese Minister of Commerce received value '3' (emphasis added).
We hold that agriculture is right at the core of the new Round. The current stalemate cannot be broken nor can the overall negotiations be pushed forwarded unless developed Members with high subsidy, high support and high tariffs make major and substantial reduction commitments. Negotiations on all the three pillars should proceed in a balanced way. The value '3' signifies a fairly strong push for liberalisation, but not one that indicates an absolute and unconditional demand. In the online appendix to this article we present the coding scheme in more detail, including examples for all coded categories and the overview of the codes (Table A1) and the codes frequencies (Table A2).
Applying this procedure, we coded statements from the eight ministerial sessions for each of the eleven BICS and G7 members, obtaining proximate positions of the countries on each of the four main negotiation areas. Of course, this is not to suggest that there are no more important areas on the Doha agenda. However, instead of mapping the total of the negotiations agenda, we selected areas that have been considered both particularly significant over the course of the negotiations, and covering areas that would not systematically differentiate commitments according to developing country status, which could bias results.
Turning to the results, we first present simple averages across all four coded areas. Which powers demand the most trade liberalisation, overall? In Figure 3, we plot the position of each of the major powers on the 0-4 scale of overall support for trade liberalisation. In keeping with qualitative accounts of the Doha negotiations, it shows that emerging powers were not overall the major demandeurs of trade liberalisation, although the difference between rising and established powers is not dramatic (the average value for G7 is 2.8, compared to 2.0 for the BICS). This indicates that in our period of analysis, all major powers considered that they had something to gain from deepening trade liberalisation at the WTO. To get a more nuanced picture, we decompose the aggregate data into the individual negotiation areas of classical trade liberalisationindustrial goods, agricultural products, and services. Figure 4 indicates the positions of the G7 (left) and BICS (right) groups across these three trade sectors in box plots. 24 As we would expect, in industrial products and services the G7 members have espoused significantly more liberal positions than the BICS countries, while in trade with agricultural products, it is the BICS who favour greater openness. 25 Moreover, both groups' positions are highly convergent in their major areas of offensive interest. On the other hand, the positions of BICS are more diffuse in industrial goods and services, 26 reflecting differences in their individual trade profiles, while the G7 is more greatly divided on agriculture. Taken together in aggregate, both new and old trade powers are committed to greater trade liberalisation (this is underlined by the first boxplot of Figure 5), but their preferences have tended to diverge strongly along sectoral lines (see Figure 4).
The differences between new and old trade powers become even starker in relation to the 'deep integration' Singapore issues (see second box plot of Figure 5). Especially India, but also China and South Africa, were strictly opposed to the inclusion of the Singapore issues in the Doha agenda, and have argued throughout the Doha round that no advancement on the Singapore issues is desirable (Khor 2003, Woolcock 2003, Evenett 2007. 27 Indeed, opposition to these deep integration issues amongst the new trade powers was so intense that they were dropped from the DDA already in 2003. As liberalisation in these areas would directly influence the governments' ability to engage in activist economic policies (Woolcock 2003, Matsushita et al. 2006, they would seem to pose a challenge to more state-oriented varieties of capitalism. Because these regulatory measures are also harder to apply preferentially, in many cases the major losers would be incumbent corporations while winners will be foreign corporations and potential domestic competitors (Baldwin 2011, p. 28), which would threaten the strategy of cultivating national champions. Indeed, it is symbolic that the core complaints of the US Trump administration regarding China target precisely such matters as domestic subsidy schemes, forced technology transfer through conditions on incoming investment, and the potential non-commercial role of state-owned enterprises (Bown 2018).
Overall, then, it is clear that the rise of new trade powers at the WTO has not only created a new distribution of power, but has also led to a greater diversity of preferences amongst those powerful countries whose cooperation is necessary to produce new trade rules. This suggests that, in comparison to a scenario in which the traditional Quad powers continued to prevail, the rise of new powers has generated greater systemic diversity in the global trade regime. It is therefore not surprising that the DDA proved to be acrimonious, has stagnated for many years, and only occasionally witnessed small success in the form of mini-agreements on partial aspects of the agenda. This has given rise to several forms of institutional change that have gradually transformed the global trade regime.

Institutional Effects: Deadlock, Drift and Fragmentation
Power has not only been redistributed within the multilateral trade regime, it has also been accompanied by greater diversity in the international distribution of preferences. While a power shift can already be expected to exacerbate collective action problems and the complexity of negotiations, preference divergence additionally makes the conclusion of new encompassing, binding agreements more difficult. In contrast to traditional hegemonic stability theory or sceptical variants of power transition theory, however, the more equal distribution of power has not (yet) fundamentally eroded the multilateral trade regime represented by the WTO. But nor can global trade governance be described as a continuation of business as usual. Instead, three major institutional consequences of the new diversity of power-weighted preferences at the WTO can be identified. These can be summarised as deadlock (of the WTO's legislative leg), drift of the existing rules, and fragmentation, as major trade powers have sought minilateral alternatives to the WTO's multilateral platform. 28 In accordance with our theoretical approach, minilateralism appears to allow states to pursue the perceived gains of cooperation while reducing preference diversity through more selective forms of collaboration.

Deadlock
The rule-making pillar of the WTO, since 2001 in the form of the DDA, operates on a consensus decision-making principle. Technically speaking any member has a veto power over a negotiated outcome. Yet, in the past this has not prevented the several most important trading powers from enforcing their views and interests on the others (cf. Steinberg 2002). In practice, a negotiated deal in the WTO context effectively means a bargain acceptable to those countries able to make themselves pivotal to a final deal, whose minilateral agreements could then be multilateralised to the rest of the institutions' membership. Whereas in the past this occurred primarily amongst the old Quad of the United States, Europe, Japan and Canada, today this encompasses a far more diverse group including at least the United States, the EU, India, and China.
In light of the systems diversity revealed above, it is unsurprising that the major trade powers have largely failed to agree on new trade rules via multilateral negotiations in the DDA. Although not formally abandoned, the DDA has been effectively proclaimed dead in 2015, when the Nairobi Ministerial declaration acknowledged the lack of support for the continuation of the DDA by some members, who '[did] not reaffirm the Doha mandates' (WT/MIN(15)/DEC, art. 30; see also Wilkinson et al. 2016, pp. 253-54). While no single factor is likely to be responsible, a fundamental source of the difficulties of the DDA can be attributed to a shift in power-weighted preferences amongst the major trade powers, compared to the status quo ante of continued Quad dominance. Alternative explanations such as issue complexity, institutional design, the single undertaking norm, or the procedural organisation of negotiations (Collier 2006, Warwick Commission 2008, Odell 2009, Low 2011, Evenett 2014 have undoubtedly played a role. But a comparison to the Uruguay Round, in which these conditions also largely held, suggests that an increase in the diversity of power-weighted preferences is a fundamental factor (Steinberg 2002, Hopewell 2016. Similarly, it is difficult to ascribe the deadlock simply to an overarching game of geopolitics among the key members, as the new powers' geopolitical interests often collide, as demonstrated most vividly by India and China's rivalry in Asia (Rehman 2009, Frankel 2011. Rather, it is the forceful representation of new powers' sectorspecific trade interests that has given rise to impasses that leave WTO rules largely stuck in the 1990s, producing disillusionment for many members. Additionally, commitments to the Doha round amongst domestic constituencies in major economies appear to have waned in recent years (Evenett 2014, pp. 151-3), most recently and visibly in the election of Donald Trump in the United States, which has widened the gap in preferences even further. For champions of multilateral trade liberalisation, the failure of the Doha round is one of the most significant events in the world trading system in the last decades (Evenett 2014, p. 160, Hufbauer andCimino-Isaacs 2015, p. 696).
Deadlock in the legislative functions of the WTO does not imply that it remains unchanged, however. First, the failure to update trade rules has led to a growing incongruence between those rules and power-weighted preferences amongst WTO members. Modest agreements at the Ministerial Conferences in Bali and Nairobi on issues such as trade facilitation, agricultural export subsidies, provisions for least-developed countries, and the plurilateral Information Technology Agreement, while significant for some members, were insufficient to remobilise support from the United States for the DDA (Financial Times 2015, New York Times Editors 2016). Over the course of the DDA, especially the United States and Europe began to claim that the entire parameters of the round were challenged by the rapid economic growth in the major emerging economies, signalling a shift away from commitment to the WTO (Schwab 2011, Froman 2015a). The US Trade Representative from 2013 to 2017, Michael Froman, unfavourably contrasted the progress made in negotiations at the WTO DDA with those made in preferential trade agreements (Froman 2015a Hindu 2015). Indeed, the collapse of the DDA stems to a large extent from a major shift in the focus of the established powers towards such matters as stateowned enterprises, domestic industrial subsidy schemes, intellectual property, and in the dynamic segments of their economies, such as services and investment (Wilkinson et al. 2016, p. 524). 30

Drift
The impact of WTO rules-still mostly written in the 1980s and 1990s-has been subject to drift as the global economy changes in ways that alter their significance and impact. In historical institutionalism, drift refers to the situation where an institution is prevented from keeping pace with changes in its environment in such a way that its real role changes and its effectiveness may erode (Streeck and Thelen 2005, pp. 24-6, Rixen and Viola 2016, p. 14, Zürn 2016. There are several examples of a process of drift beginning to impact on the real-world role of the WTO.
One is the process of differential growth. At the initiation of the Uruguay Round in 1985, services accounted for around 18 per cent of the value of world trade. By 2015 this had risen to 22 per cent. As a portion of global value added, services grew from 58 per cent in 1995 to 68 per cent by 2015. 31 Likewise, global stocks of foreign direct investment have grown from around 12 per cent of global GDP in 1995 to 35 per cent by 2015 (UNCTAD Stat 2017). For developed countries these trends are much stronger. Yet WTO provisions relating to services (GATS) and investment (TRIMs) are relatively shallow, and do not constitute the core focus of its existing rules. This is contributing to shifts towards bilateral investment treaties (BITs) and the World Bank's International Centre for Settlement of Investment Disputes, the negotiation of a Trade in Services Agreement (TiSA) outside of the WTO, and new regional trade initiatives (see below). Given that major growth areas of the global economy are expected in the form of modular production and the trade-investment-services nexus, the WTO's future importance for cross-border economic activity could be facing relative decline (Evenett 2007, Baldwin 2011.
Another sign of drift for the WTO concerns the application of old rules to novel situations. A key example of this is the legal uncertainty emerging from the application of rules on subsidies and countervailing measures to novel economic systems such as China's. While the WTO has rudimentary provisions about SOEs, primarily in Article XVII of the GATT and China's accession protocol (Hufbauer and Cimino-Isaacs 2015, p. 686), the emergence of China as the second global trading power puts these provisions in a whole new light. China's status as a non-market economy, and the pervasive role of state-owned enterprises (SOEs) in its economy and export sector, leaves it open to future legal uncertainty (Du 2014, Wu 2016. Finally, drift affects even the enforcement provisions of the WTO. The WTO's dispute settlement procedure may become less attractive if the rules upon which adjudication decisions are made fail to be updated (Baldwin 2011, p. 32). Moreover, major economies may feel less obliged to comply with WTO rulings if they have already cultivated alternative dispute procedures with their major trading partners in alternative venues outside of the WTO (Stoler 2014, p. 3).

Fragmentation
Ultimately, the discordance of WTO rules with power-weighted preferences and subsequent drift has given rise to fragmentation (Biermann et al. 2009) and increased 'complexity' (Davis 2009) of the global trade regime. Fragmentation, which pre-dated the failure of Doha, has been accelerated as a result of the Doha failure (Aggarwal and Evenett 2013). 32 While preference divergence can give rise to deadlock, as long as states perceive an interest in international cooperation, deadlocked institutions provide an incentive to pursue selective minilateralism as outside options. This can narrow the distribution of preferences to make agreement more likely, and can involve seeking to shift an issue from one institution to another (Helfer 2009) or creating a new one from scratch (Morse and Keohane 2014). Fragmentation may have been one contributing factor to 'gridlock' at the WTO (Hale et al. 2013, p. 230), but it is also a critical path by which major trading countries have begun to use selective institutions to evade universal multilateralism and overcome gridlock. 33 The most salient form of fragmentation concerns the pursuit of 'mega' free trade agreements (FTAs). These are distinguished from traditional FTAs by both breadth and depth: they involve large numbers of countries from multiple regions, cover a large portion of world trade, and include 'WTO-plus' content concerning economic regulation and non-tariff barriers (Ravenhill 2016(Ravenhill , pp. 1077. Until recently, the United States had shifted its deep integration agenda to the bilateral 'Transatlantic Trade and Investment Partnership' (TTIP) with the European Union, and a 'Transpacific Partnership' (TPP) with countries in the Asia-Pacific (Hamilton 2014). China and India, by contrast, have invested in an 'ASEAN plus Six' project called the Regional Comprehensive Economic Partnership (RCEP). As we would expect based on our analysis of their attributed and stated preferences, the RCEP seems to pursue a shallower form of trade integration than the TPP or TTIP (Das 2013, Ravenhill 2016). At the time of writing, none of these mega-FTAs has been completed, and the United States has already withdrawn from the TPP. But in future, these mega-regionals have the potential to fundamentally alter the role of the WTO in global trade governance by poaching its core policy tasks of negotiated trade liberalisation, rule monitoring, and rule enforcement (Aggarwal and Evenett 2013, Stoler 2014, Hufbauer and Cimino-Isaacs 2015. These agreements challenge the principles of universalism and non-discrimination, which lie at the heart of the WTO regime. Especially where these mega-FTAs divert trade and promote different regulatory standards, they can be used as geopolitical tools to compete for global trade flows (Chow 2016b). This suggests that another form of institutional change induced by the rise of new trade powers consists of a decline of universal multilateralism and fragmentation into different, partially overlapping but potentially rivalrous institutions. These trends are now accompanied by an erosion of domestic US support for an open and multilateral trade order, and a consequent trend towards tit-for-tat tariffs between the United States and its major trading partners. Whether this adds up merely to a change within the liberal international economic order, as China increasingly positions itself as the liberal order's defender, 34 or a more fundamental erosion or decline, is hard to asses. But it does reinforce that it is hardly the power shift alone that challenges the global trade order-the societally-derived preferences of the major players play a critical role.

Conclusion
This paper has shown that the rise of new powers has increased the diversity of the distribution of power-weighted preferences at the WTO. This systemic diversity has contributed to the difficulty of coming to agreement to update the WTO and its rules, which combined with institutional inertia, has led to the trade rules overseen by the WTO becoming more incongruous with the preferences of major economic powers. This 'gridlock' (Hale et al. 2013) in global trade governance should not be mistaken for a continuation of the status quo, however. Adopting insights from historical institutionalism, this institutional stasis has led also to a growing gap between the WTO's extant rules and the changing nature and composition of international trade (drift). This, in turn, has increased the incentives to pursue minilateral alternatives that reduce preference diversity, fostering institutional fragmentation via forum shopping and alternative regime creation.
These more nuanced forms of institutional change are hard to reconcile with dichotomous analytical approaches that emphasise either rising power integration or a decline of international regimes as a result of weakening inter-state hegemony (Ikenberry 2011b, Layne 2012, Zangl et al. 2016. Moreover, to account for the forms of institutional change observed as a result of international power shifts, it is necessary to combine insights from competing institutionalist theories. Our analysis takes into account power, preferences, and institutional inertia (Zangl et al. 2016). This approach has two major advantages. First, it provides a more persuasive composite explanation of how the rise of new trade powers leads to institutional change. Second, it shows a more nuanced approach to describing the impact of rising powers that goes beyond existing categories such as 'integration' or 'overthrow'; or whether rising powers are 'status quo' or 'revisionist' powers in the trade regime.
Our argument suggests that deadlock, drift and fragmentation may characterise other policy fields and institutions in global economic governance, but especially in areas where substantive preference diverge (Hale et al. 2013, Stephen 2017. The failure of the World Bank to reflect the new distribution of power-weighted preferences also appears to have resulted ultimately in China's decision to pursue the Asian Infrastructure Investment Bank (Hamanaka 2016, Chow 2016a, leading to greater fragmentation. Likewise, dissatisfaction with representation and policy output at the IMF has contributed to pushing new minilateral initiatives such as the Chiang Mai Initiative Multilateralisation and its surveillance body the ASEAN Macroeconomic Research Office (AMRO; cf. Wade 2013, pp. 30-2). Of course, the prospects of such parallel institutions are hard to assess. At a time when the country most 'dissatisfied' with the status quo appears to be not the rising powers but the United States, this suggests that the preference divergence and the forms of institutional change we observe at the WTO may not be unique, but reflect a new normal in global economic governance.
Notes 11. Where data is available, we focus on the five 'BRICS' countries because, due to a mix of economic size and institutional activism, they have emerged as central players in global trade (Brazil, China, India) or have the capacity to do so (Russia). We include South Africa to complete membership of the BRICS group. 12. Ratios of trade to GDP for the BRICS were as follows: Brazil 27.6 per cent, Russia 50.9 per cent, India 53.3 per cent, China 50.3 per cent, South Africa 64.2 per cent. The average of 49.3 per cent is close to that of the G7 countries at 55.3 per cent (World Bank 2018). The picture for inward foreign direct investment as a percentage of GDP in the same year was as follows: Brazil 32.2 per cent, Russia 26.9 per cent, India 11.8 per cent, China 10.3 per cent, South Africa 39.7 per cent. The simple average (24.2 per cent) is fairly close to that of the G7 countries at 30.6 per cent (UNCTAD 2017). 13. The comparison is based on applied tariff rates weighted by trade volumes, indicator TM.TAX.MRCH.WM.AR.ZS of the World Bank (2018). 14. See OECD 2014. 15. Across both the indicators, the difference between the two groups is statistically significant at alpha level of 0.05 or lower, although obviously statistical inference is very problematic with as small a sample as N=11. 16. This is reflected in the data compiled by the Global Intellectual Property Center: Pugatch et al. (2015) quantify intellectual property restrictions in the major economies. 17. All G7 countries are parties to the Agreement. As of writing, China, India and Russia hold an observer status, Brazil and South Africa do not (World Trade Organization 2015). 18. FDI Restrictiveness Index, data for year 2015. Source: OECDStatExtracts 2015, accessed 22 March 2017. 19. The concept of systems friction comes originally from Sylvia Ostry (1992). 20. The data cover all G7 and BICS members. We leave Russia out of this part of the analysis as it only accessed the WTO in 2012. All data used in this section are available online at https://docs.wto.org/dol2fe/Pages/FE_Browse/ FE_B_007.aspx?MetaCollection=2&TopLevelId=7596. 21. Interview with WTO trade representative, Singapore, 2 March 2017. 22. We exclude intentionally the 2013 and 2015 Bali and Nairobi Ministerial conferences as the speeches delivered at these systematically focus more on the WTO system as a whole, and the fate of the Doha negotiations per se, as opposed to the individual points on the agenda. Hence, they only provide very shallow information. 23. We exclude from our analysis issues on which we would expect systematic differences between developed and developing countries, such as implementation, special and differential treatment, or the package for least-developed countries. Our arguments should apply regardless of level of economic development. 24. The positions indicated are not weighted by members' power (market size), they constitute the raw position values. Note that the horizontal line indicates the median position, with the distribution of the values captured by the size of the boxes and of the vertical T-shaped lines ('whiskers'). 25. See the notes below the individual charts for values of two-tailed t-tests. Some of the tests indicate a statistically significant difference between the groups, but the interpretation of the results in statistical terms is problematic due to the low number of observations. 26. Although it is notable that none of the BICS economies are party to negotiations over the plurilateral Trade in Services Agreement initiated by the US and EU in 2012. 27. The exception is the at-the-border 'trade facilitation' issue to cut red tape at the border, which was the subject of an agreement in Bali in 2013 (WTO document WT/L/940). 28. See also Faude and Stephen (2016 (17)/ST/127). 35. The order of countries in the chart reflects the overall trade restrictiveness index. Note that although commercial policy is an exclusive Union competence within the European Union, we report the G7 EU members individually as there is variation across them on some of the indexes (for instance customs procedures burden). We apply this practice of taking the four EU members of G7 individually throughout the article, also in the analysis of their statements in the WTO. 36. Data sources available at: World Bank 2014, 2018, World Economic Forum 2014. 37. Data sources available at the OECD StatExtracts website. OECD 2014. 38. Averages for the indicated periods (1990-1999; 2000-2009; 2010-2016/2017) are displayed. For 2017, part of the data for imports of services is missing. Source: World Bank (2018).