Advocacy Coalition Constellations and Norm Collisions: Insights from International Drug Control, Human Trafficking, and Child Labour

ABSTRACT To date, there has been little research on how advocacy coalitions influence the dynamic relationships between norms. Addressing norm collisions as a particular type of norm dynamics, we ask if and how advocacy coalitions and the constellations between them bring such norm collisions to the fore. Norm collisions surface in situations in which actors claim that two or more norms are incompatible with each other, promoting different, even opposing, behavioural choices. We examine the effect of advocacy coalition constellations (ACC) on the activation and varying evolution of norm collisions in three issue areas: international drug control, human trafficking, and child labour. These areas have a legally codified prohibitive regime in common. At the same time, they differ with regard to the specific ACC present. Exploiting this variation, we generate insights into how power asymmetries and other characteristics of ACC affect norm collisions across our three issue areas.


Introduction
Decades of IR research on transnational civil society actors have confirmed that in many cases these actorsand particularly the networks and coalitions they formplay a decisive role "in teaching governments what is appropriate to pursue in politics" (Price 1998, 639). But what if there are competing coalitions with diverging views on what is appropriate?
Looking at norm collisions in three international issue areas (international drug control, human trafficking, child labour), we identify a research puzzle that merits comparative research: why, despite shared characteristics of the three issue areas, do we see differences in how norms are brought into collision? In our article, we aim to provide explanations for this puzzle by looking at constellations of advocacy coalitions and the ways in which these constellations affect conflictual relationships between international norms. We understand norm collisions as the perceived incompatibility between two or more norms. Our overall theoretical starting point is the assumption that norms do not collide by themselves but always require agents that articulate norm collisions in international debate.
Our concept of norm collisions speaks to but differs from related concepts such as norm contestation (Wiener 2018;Deitelhoff and Zimmermann 2020), norm sabotage (Schneiker 2021), or norm disappearance (Panke and Petersohn 2012). We share the agency-orientation of the literature on norms as well as its emphasis on studying the development of norms as a dynamic process. Yet we depart from this literature's focus on contestation, sabotage, or "death" of a particular international norm. Most importantly, our approach to norm collisions is decidedly distinct from the widely used notion of norm contestation. Norm collisions presuppose that the validity of two or more colliding norms in a given situation is not contested; otherwise, norms and their behavioural prescriptions could not be identified as being in conflict with each other. Instead, we propose to examine how actors relate various norms to one another in a given situation and focus on perceived incompatibilities between at least two norms that actors articulate.
While we have assessed a variety of scope conditions and their explanatory potential for the emergence and dynamic of norm collisions elsewhere (Gholiagha, Holzscheiter, and Liese 2020), in this article we focus on constellations of advocacy coalitions as a potentially powerful explanatory factor for both the emergence and the dynamics of norm collisions. Our focus on constellations rests on the core observation that particularly since 1990 there has been an unprecedented increase not only in non-state actors but also in larger transnational networks and coalitions surrounding international institutions (Keck and Sikkink 1998;Scholte 2004). It is thus plausible to assume that policy-fields that can be described as "regime complexes" (Alter and Meunier 2009) with intersecting rule-systems are also characterised by multiple advocacy coalitions.
Following established theories on transnational advocacy and norm dynamics (Wiener 2018;Sandholtz 2008;Panke and Petersohn 2012;Winston 2018;Lantis and Wunderlich 2018;Fehl 2018;Hahn and Holzscheiter 2013;Keck and Sikkink 1998) we define transnational advocacy coalitions as connected groups of actors, both state and non-state, on the international level who work together towards a shared goal informed by common values and beliefs.
Moving beyond studying single advocacy coalitions, thus, we are interested in advocacy coalition constellations (ACC), which we define as relations among advocacy coalitions within a specific issue-area. These constellations vary in terms of the number of advocacy coalitions as well as their underlying values, their position towards the dominant discourse in a given issue area, and their relative power.
Our principal argument is that ACC are decisive for norm collisions to come to the fore. The constellation we find in a specific issue area influences whether, how, and when norms come into collision in international fora. Our argument is based on, but seeks to extend, the major scholarly contributions to the advocacy coalition literature referred to above. By linking this literature to the contemporary study of how different norms relate to each other, we claim, it is possible to harness the explanatory potential of coalitions of state and non-state actors not only with regard to the emergence, diffusion, consolidation, and contestation of norms but also with regard to how these coalitions may bring about conflicts between norms, leading to norm contestation and transformation. We thus assume, in the first place, that ACC influence when and how norm collisions come to the fore in international debatesusing the term "activation" for this process. Secondly, we contend that variation in advocacy coalition constellations may explain the difference in how such norm collisions evolve in otherwise similar cases.
In our empirical analysis, we look at three issue areasdrug control, human trafficking, and child labour. All of them share a number of important characteristics: a prohibitive regime at their core, the existence of norm collisions, and the existence of advocacy coalitions. At the same time, they vary in regard to the specific constellation of advocacy coalitions. Choosing this case selection strategy, we exploit this variation in order to understand how these constellations matter for the activation of norm collisions.
The remainder of our paper is structured as follows: Section 2 provides a literature review. Section 3 develops the framework we use to theorise and analyse ACC and their effect on norm collisions. Section 4 presents the research design we employ for our within-case and across-case comparative study, provides an overview of the three issue areas, and contains the substantive analysis of the effect of ACC on norm collisions in each issue area. Finally, Section 5 offers a brief conclusion, discusses our results, and provides an outlook on future research.

Advocacy coalitions and normative change in IR scholarship
Scholars of International Relations (IR) have studied both advocacy and norms in global politics for at least 20 years. Keck and Sikkink's seminal work on transnational advocacy networks (1998) has been taken up in several scholarly works (Bloodgood and Clough 2016;Carpenter 2007;Cogburn 2017;Farquharson 2003;Rodrigues 2011). Over time, constructivist IR scholars have developed a number of models and concepts to analyse how transnational advocacy coalitions influence the emergence, diffusion, implementation, localisation, and contestation of norms (Acharya 2004;Risse, Ropp, and Sikkink 2013;Wiener 2018;Zimmermann 2017;Berger 2017;Deitelhoff and Zimmermann 2020;Homburger 2019;Finnemore and Sikkink 1998).
Based on the seminal literature that bridges work on advocacy coalitions and norms, we assume that advocacy coalition constellations affect whether norm collisions are debated in forums such as international organisations. Several authors have studied the intertwinement and relationship between norm change and advocacy coalitions (Carpenter 2007;Hahn and Holzscheiter 2013;Epstein 2008;Almagro 2018;Belloni 2014;Carpenter 2014;Ciplet 2019). A shared insight from these studies is that transnational advocacy coalitions matter for and strongly affect global politics and international law. Cases such as the landmine ban or the prohibition of whaling have been prominently used and analysed in order to demonstrate and explain success cases of transnational advocacy coalitions (Epstein 2008;Anderson 2000;Price 1998). By contrast, examples such as the small arms movement (Bob 2010;Brem and Rutherford 2001) or global climate politics (Hadden 2015) have been used in order to expose competition between advocacy coalitions and the way it obstructs normative consensus and the swift adoption of new international legislation. Overall, there is by now ample evidence for variation in the impact that advocacy coalitions have on the emergence and legalisation of international norms.
A shared feature of the studies referred to above is their analytical focus on the process of negotiating international norms. They suggest that independently of their outcome and long-term success, the actions of advocacy coalitions across a vast number of issue areas have resulted in normative tensions, opened up new discursive spaces, or created at least the possibility of reconsidering moral values and appropriate behaviour. In our view, however, greater attention needs to be paid to the very moments of open conflict, not only between certain state and non-state actors, but also among different advocacy coalitions. Building on the insight that advocacy coalitions matter when it comes to normative change, we assume that situations in which norms are perceived to be incompatible will most likely involve more than one advocacy coalition. Hence, we contend that it is the constellation of advocacy coalitions in a specific issue area that matters for the emergence and dynamic of such collisions.
As we have demonstrated elsewhere, other scope conditions such as the legalisation of the norms involved or power-shifts in the international realm also affect whether and how norm collisions come to the fore (Gholiagha, Holzscheiter, and Liese 2020). However, in the cases we examine in this article, these alternative scope conditions can hardly explain the observed outcome. We observe great variation in the constellations of advocacy coalitions but relative stability concerning alternative scope conditions. Hence, we focus our analysis on these constellations. We develop this argument against the backdrop that ACC change over time and that advocacy coalitions in 2019 are markedly different from those that inspired the first wave of theorising on advocacy and norm emergence in IR in the 1990s (Zajak 2017 provides a recent analysis of current structures of transnational activism). Hence, a closer look at the relationship between these coalitions, their constellations, and dynamic norm relations is necessary. Noteworthy developments, such as the opening up of international organisations (Tallberg et al. 2013) and the growth and diversification of global civil society (Scholte 2004) suggest that advocacy coalitions have not only become an omnipresent feature of international politics but also continue to contribute to the transformation, democratisation, and politicisation of international politics. At the same time, seminal contributions on the dynamics of power and competition among civil society actors and advocacy coalitions in international politics have shown that, in many cases, advocacy organisations and coalitions compete for visibility, influence, and resources in densely populated issue areas of global governance (Stroup and Wong 2017;Cooley and Ron 2002;Hadden 2015).

Norm collisions and advocacy coalitionsextending the debate
Constructivist norms research has always been marked by a strong interest in advocacy coalitions, non-state actors, and so-called "norm entrepreneurs" (Finnemore and Sikkink 1998, 893). It regards them as decisive for the creation, diffusion, and contestation of international norms (Keck and Sikkink 1998;Risse, Ropp, and Sikkink 2013;Wiener 2018). Our paper thus relates to this traditional interest in the role of advocacy and norm entrepreneurship. However, it extends this popular field of inquiry to situations in which different norms are applicable but potentially conflicting with each other. As we argue below, situations of norm collisions are related to more complex constellations of advocacy coalitions.

Key concepts: norm collisions and advocacy coalitions
For some time, constructivist norm researchers have focused on norm contestation (Wiener 2008;Liese 2009), norm disappearance (Panke and Petersohn 2012), and norm change (Sandholtz and Stiles 2009)often focusing on single norms. Most recent constructivist work, however, has started to dissect the dynamic relationship between norms (Lantis and Wunderlich 2018;Winston 2018;Müller and Wunderlich 2013;Fehl 2018;Sandholtz and Stiles 2009;Fehl and Rosert 2020). It is this latter strand of work that our article contributes to, taking an interest in situations in which actors perceive two (or more) norms as incompatible with each other: a phenomenon we have conceptualised as norm collisions (Gholiagha, Holzscheiter, and Liese 2020). Norm collisions are instances in which actors perceive the behavioural prescriptions of two or more norms as incompatible although these norms can equally be applied to the situation the actors are in. Assuming that norms do not collide by themselves but that collisions occur when actors articulate their concerns with incompatible behavioural expectations, we argue that norm collisions need to be activated. Activation goes beyond subjective perceptions of norm incompatibilityit requires intersubjective exchange on this incompatibility in order to be socially consequential. We do not deny that actors may articulate such normative incompatibility for a variety of reasons, including instrumental, self-interested, or normative, altruistic intentions. As our article is interested in the effects of ACC on the emergence and development of norm collisions, though, actors' motives are irrelevant. It is the act of articulating incompatibility between two or more norms that constitutes the starting point for our analysis.
Following Winston's conceptualisation (Winston 2018), we define norms as valuebased collective expectations for the appropriate behaviour of governments and other actors in the transnational realm. As a consequence, not only norms but also values must be seen as points of reference that bind actors together within a specific advocacy coalition. Importantly, values are part of the complex structure of norms, but values and norms are not identical, as only norms include a formulation of what the appropriate behaviour for an actor is. For the purpose of this article and its focus on norm collisions, we zoom in on collisions between different standards of appropriateness that actors express in communicative interaction. These are open to various norm types (Wiener 2014) and may apply to social and legal norms equally (Finnemore 2000). Our perspective on norm collisions is actor-centred, which implies that we identify a norm collision when actors perceive two or more norms to be incompatible and articulate this perception in international forums. We thus endorse an "internalist" perspective on norm collisions, as opposed to externalist perspectives in which norm collisions are identified by scholars or legal experts (Kreuder-Sonnen and Zürn 2020; Gholiagha, Holzscheiter, and Liese 2020).
Our strategy for identifying a norm collision rests on articulations that relate to multiple norms and the (contested) hierarchies between them (Lantis and Wunderlich 2018;Fehl 2018). Colliding norms can thus reflect discrepancies between values or shared values but differing hierarchies between norms. We may find that different advocacy coalitions are united by a consensus on certain core values and related normsi.e. the necessity to preserve whale stocks, or to protect children from exploitation and societies from harmful drug usebut may still argue over the appropriate behavioural expectations to realise these values. Coalitions may also concur on the fact that multiple international norms are relevant to the same situation, but they may hold opposing views on which norms should enjoy priority. Situations in which actors perceive multiple norms as relevant but differ regarding their prioritisation can be observed for example in debates on human trafficking, where we find diverging views on regulating or abolishing sex work/prostitution (Doezema 2005), or in debates about public health protection and individual liberty rights in the treatment of drug users (Wolfe and Saucier 2010).
Shared social expectations and beliefs matter greatly for the very definition of what constitutes an advocacy coalition (Jenkins-Smith and Sabatier 1993). In the policyoriented literature on such coalitions, the analysis of competing coalitions has been a focal point in understanding policy change (Blomquist and Schlager 1996, 6). We propose to extend these assumptions by focusing more strongly on values as constitutive characteristics of advocacy coalitions. While the IR literature has used different concepts most notably "advocacy coalition" (Sabatier and Jenkins-Smith 1993) and "transnational advocacy networks" (Keck and Sikkink 1998)we contend that all definitions rely on the "centrality of principled ideas or values" (Keck and Sikkink 1998, 1) to identify such actors (Díez García 2017, 62). Drawing from these conceptualizations of norm advocates in national and international politics, and on our own work (Hahn and Holzscheiter 2013, 497), we define transnational advocacy coalitions as connected groups of actors, both state and non-state, who promote shared norms and norm hierarchies and work together towards a shared goal informed by common values and beliefs.

Advocacy coalitions and their constellations
In order to assess the effect of ACC on norm collisions, we make use of the insights provided by the advocacy literature on the conditions that determine success and failure of advocacy coalitions and networks. In doing so, we transfer these insights and assumptions from intra-to inter-coalition dynamics, assuming that they enhance our understanding of how ACC affect norm collisions. Keck and Sikkink (1998, 25-29) provide a general discussion of these conditions. Advocacy scholars have argued that a constellation of competing advocacy coalitions may lead to failure in policy change, with small arms norms being a case in point (Grillot 2011, 531). Grillot (2011) and Clarke (2008) compare the International Campaign to Ban Landmines with the small arms movement to identify respective strengths and weaknesses. They identified both structural (leadership, level of organisation) and issue-related (clear causal claim, direct responsibility) factors. The International Campaign to Ban Landmines has often been portrayed as a successful model of civil society influence and engagement in global politics, and a blueprint for other campaigns (Price 1998;Clarke 2008). In contrast, the small arms movement has been used to demonstrate the negative effect of weak leadership and structure on advocacy work (Clarke 2008, 7).
An essential factor for the functioning of advocacy coalitions is the relationship between local actors and globally operating (often "western" based and originated) NGOs. As Hertel (2006a, 265) demonstrates: "local activists and their international supporters do not always agree on the nature of human rights at stake". If local and global actors disagree, local actors can either block a global campaign to change the normative frame of the campaign or use back-door strategies to add different norms to the existing framing of the campaign, as was the case in anti-child labour advocacy in Bangladesh for example (Hertel 2006a, 267-268).
While these insights refer to all constellations equally, they may be differentiated from each other on the basis of a number of important characteristics. From our perspective, variations in ACC can be inferred by looking at the number of coalitions involved, their underlying values, their relative power, and their position towards the dominant discourse in a given issue area. We identify these characteristics on the basis of secondary literature and our own analysis of the respective issue area. The number of coalitions is determined by counting those groups that promote specific norms and work together towards a shared goal informed by common values. Relative power in our understanding refers to the greater or lesser capacity of advocacy coalitions to influence both agendas and frames in international debates vis-à-vis other coalitions and vis-à-vis state actors. Underlying values and a coalition's position towards the dominant discourse can be inferred from references to norms and positions articulated during international negotiations. We now turn to the question of how these coalitions and their constellations affect norms and norm collisions.

How do advocacy coalitions and their constellations matter for norm collisions?
Some of the literature on advocacy coalitions is cognisant of the relation between different advocacy coalitions but focuses mostly on the policy issue at hand, such as health rights (McDougall 2016), tobacco control (Farquharson 2003), or small arms control (Grillot 2011). However, overall the relation between different coalitions is a topic which is discussed more in passing. That is, scholars often mention relations between coalitions and study the effect on questions of norm development or legalisation (Bob 2010;Hertel 2006b), but they do not engage in a substantive discussion on how these relations between coalitions matter for understanding dynamic norm relations, such as collisions.
What we can take from this literature is detailed knowledge and conceptual offers with regard to the strategies actors use in their advocacy work. As our approach is decidedly internalistidentifying relevant norm collisions through the ways in which state and non-state actors articulate themtwo key strategies by state and non-state actors appear noteworthy: framing (Keck and Sikkink 1998, 17;Locher 2007, 67;Joachim 2003, 250) and agenda-setting (Carpenter 2007;2014, 662;Price 1998;Joachim 2003, 268;Ciplet 2019;Luxon and Wong 2017). Both of these strategies or "tactics" have been identified as vital in garnering attention, highlighting new issues and shifting problem definitions and the meaning of norms. Even though the advocacy coalition literature and the transnational advocacy network literature point to the importance of norms, they have, so far, not focussed on dynamic relationships between norms, including norm collisions. We expect advocacy coalitions' tactics and strategies, though, to be equally relevant for the study of norm collisions.
The core claim of our article is that the analysis of ACC is a promising route in explaining why norm collisions become a matter of international debate. While there are additional scope conditions which may facilitate the activation of norm collisions (see section 2), in the cases we study, these scope conditions remain relatively stable.
We do find variation with regards to ACC and therefore contend that this variation may explain our research puzzle, i.e. why norm collisions play out very differently in otherwise very similar issue areas.
The policy-oriented advocacy coalition literature assumes that cooperation between different coalitions, brokered by other actors, leads to policy change (Sabatier and Jenkins-Smith 1993). More recent work on advocacy coalitions points out that both policy core beliefs, which refer to a single policy field and deep core beliefs, which are valid across various policy fields of advocacy coalition members matter for advocacy coalitions (Matti and Sandström 2013, 248). It is, however, the policy core beliefs that "constitute the defining element of coalitions" (Matti and Sandström 2013, 253). Building on this insight, we assume that when the core and policy beliefs amongst advocacy coalitions vary, norm collisions are more likely.
Grillot offers another interesting insight into competing advocacy coalitions. She argues that the primary explanation for weak small arms norms is a competitive normative environment that is facilitated and perpetuated by: (1) competing coalitions that promote opposing norms and ideas and (2) a great-power consensus that works against stronger arms control norms (Grillot 2011, 531, our emphasis).
For Grillot, opposing coalitions promoting either free trade norms or the right to live in a secure and safe environment thus explain weak (i.e. not legalised) and non-existent small arms norms (Grillot 2011, 540). We assume that another effect of competing coalitions is that they facilitate norm collisions by promoting opposing norms. At the same time, it seems plausible to expect that competing coalitions lead to a protracted debate on such norm collisions and obstruct norm change or the resolution of such conflicts. Thus, competing coalitions provide actors with the possibility to activate norm collisions. Another potential source of norm collisions is constituted in tensions between human rights and development NGOs claiming to represent a certain group and representatives of that group (Hudson 2001), something we find for example in debates about human trafficking and child labour (Hahn and Holzscheiter 2013, 502).
Inspired by this literature, we trace the framing and agenda-setting activities of advocacy coalitions. We take these activities to expose the values promoted by coalitions and their position towards the dominant discourse. This allows us to examine the specific ACC in each case, enabling us to study its effects on whether and how norm collisions come to the fore. Concerning framing, different actors may frame an issue with reference to norms they deem compatible with each other or with reference to norms actors perceive to be in collision with each other. Concerning agenda-setting, we assume that when different advocacy coalitions exist, they may pursue the same issue but for different normative reasons or pursue different issues for the same normative reasons. In both cases, the normative reasons and more importantly the norms the actors invoke in their arguments may be perceived as either compatible with each other or in collision with each other.
Building on these arguments and assumptions, we go on to empirically assess cases in the aforementioned issue areas. While we both observe norm collisions and identify advocacy coalitions, the cases vary in terms of the advocacy coalition constellations that have been present in an issue area over time. We thus go beyond the dominant focus of the advocacy literature which mostly deals with the emergence of advocacy actors (Keck and Sikkink 1998), their internal structure and organisation (L. Jordan and van Tuijl 2000), their legitimacy (Hudson 2001;Hahn and Holzscheiter 2013), or the conditions under which they fail or succeed (Keck and Sikkink 1998;Farquharson 2003). Instead, we assume that these constellations are a plausible explanatory factor for the dynamics and trajectories of norm collisions.
4. Comparing the effects of ACC on norm collisions in international drug control, human trafficking, and child labour In order to analyse the relationship between advocacy coalition constellations and norm collisions, we study three issue areas: international drug control, human trafficking, and child labour. All three of our issue areas had a prohibitive regime (Nadelmann 1990) at their centre prior to the activation of norm collisions, build on codified international law in the form of treaties and conventions, and have advocacy coalitions working within them. However, they differ with regard to ACC. This design allows us to examine the effect of different constellations.
In the first part of our analysis, we map the three issue areas in terms of prohibitive norms, advocacy coalitions, and norm collisions (4.1-4.3). In the second part (4.4-4.6), we look at the effects of advocacy coalition constellations on the norm collisions identified. We do so by analysing meeting records and statements of advocacy coalitions. We also look at debates within international fora by studying official records, government statements, and more informal records such as NGO reports. In addition, we draw on secondary literature. In the third part of the analysis (4.7), we compare our findings: we examine the number of coalitions, their relative power, their values, and their position towards the dominant discourse in a given issue area. The primary and secondary sources that our analysis is based upon allow us to draw conclusions on the type of constellations between advocacy coalitions; in particular, the relative power of some coalitions vis-à-vis others concerning their access to those international forums in which relevant norms are (re-)negotiated and their opportunities for active participation in those debates and negotiations.
The issue areas we explore share three important characteristics in terms of the type of regime, the presence of advocacy coalitions, and the existence of norm collisions. In each issue area and the specific norm collision, the respective constellation of advocacy coalitions differs. In the area of international drug control, advocacy coalitions have challenged dominant norms of criminalisation and prohibition by bringing in human rights norms such as the right to health and indigenous rights. In the area of human trafficking, we find two equally powerful advocacy coalitions with contradictory positions concerning the relationship between prostitution and human trafficking. In the area of child labour, an influential advocacy coalition supports the dominant view of the prohibition of child labour while a vocal, but still inferior, coalition of grassroots organisations attempts to shift the debate from abolitionism to regulation and respect for economic rights.
We assume that different constellations of advocacy coalitions affect norm collisions. To probe this assumption, we provide an analysis of norm collisions, and the role advocacy coalitions and their respective constellations have played in activating those norm collisions for each issue area.

Prohibitive norms as a common denominator of drug control, trafficking and child labour
All three issue areas can be characterised as being heavily influenced by prohibitive norms which, for a long time, constituted the uncontested normative core of each regime. The international drug control regime aims to prevent all non-scientific and non-medical use of substances classified as drugs (Article 4(c) of the Single Convention on Narcotic Drugs (1961)). It obliges States Parties to criminalise any form of production, possession, and trade of any drugs outside of medical and scientific use (Article 36 of the Single Convention). The human trafficking regime with its core treaty, the 2000 Protocol to Prevent, Suppress and Punish Trafficking in Persons Especially Women and Children as part of the UN Convention against Transnational Organized Crime, aims to prevent and combat human trafficking (Article 2). In the issue area of child labour, the International Labour Organization (ILO) Convention 138 (1973) obliges States Parties to prohibit child labour and light work under a certain age and to determine such a minimum age (Article 2). In addition, ILO Convention 182 (1999) prohibits the worst forms of child labour, including slavery, prostitution, and illicit and harmful activities (Article 3). States Parties are obligated to eliminate those forms of child labour for all persons under the age of 18.

Variation of ACC in our three prohibitive international regimes
In the following, we describe the advocacy constellations in each issue area and apply the conceptual framework developed above. In the issue area of international drug control, we find one advocacy coalition that is formed by a number of mainly European countries, Australia, Canada, and several NGOs and non-state alliances, inter alia the International Harm Reduction Association, the Drug Policy Alliance, and the International Harm Reduction Development Program (2004,383). The coalition came together "under the harm reduction banner" (Jelsma 2003, 184). This advocacy coalition opposes the dominant discourse in international drug control, i.e. a prohibitionist and criminalising approach. Instead, it supports harm reduction, which it derives from human rights such as the right to health. Harm reduction became an issue during the HIV/AIDS crisis in the 1980s, and the idea was further pushed during a 1993 General Assembly meeting. Advocates and scholars argue that drug users' fundamental rights such as the right to health should be protected and that the dominant drug criminalisation norm has led to a violation of this right (Burke-Shyne et al. 2017).
In the issue area of human trafficking, we find two advocacy coalitions. They formed during the negotiations of the aforementioned protocol on human trafficking. The Human Rights Caucus, with the Global Alliance Against Trafficking In Women as one of its key members, holds "that legalizing prostitution would help reduce trafficking" (Andreas and Nadelmann 2006, 36), while The International Human Rights Network, with the US-based Coalition Against Trafficking in Women at its centre, has taken "a more prohibitive approach" (Andreas and Nadelmann 2006, 36). The coalitions saw each other at least as competitors, even perhaps as opponents. While both shared the goal of stopping human trafficking (and thus shared the dominant position in the discourse), they had strongly diverging views on what human trafficking entailed. More specifically, the differences lay in the question of whether human trafficking applies to all forms of prostitution or whether one should distinguish between voluntary and forced prostitution (Jakobi 2013, 165). These diverging views led to fierce battles on the definition of human trafficking, which, as a compromise, leaves it up to states whether to regulate or prohibit prostitution (A. D. Jordan 2002, 32).
A side effect of the conflict between the coalitions was that issues where they could have lobbied together, such as human rights protection for trafficking victims, received much less attention (Ditmore 2012, 112). In terms of values, both coalitions identified themselves as defenders of human rights of trafficking victims and believed that human trafficking needed to be prevented. Their strongest difference lay, as mentioned already, in the question of whether one could distinguish voluntary forms of prostitution from forced ones. Underlying these core beliefs were different value-sets pertaining to questions of self-determination and consent (Siller 2017, 423;Abramson 2003, 480;Wylie 2016, 61-62).
In the issue area of child labour, we also find two advocacy coalitions. Yet their power varies considerably. They enjoy different opportunities for access and participation in international debates. Discussions on child labour in the context of the ILO were to a large extent shaped by a long-standing and powerful advocacy coalition of state and non-state actors, including ILO member states, the International Labour Office, and organisations from the Global South associated with the Global March against Child Labour. These endorsed an abolitionist agenda and were hence in line with the dominant position in the discourse. The opening of the ILO Child Labour Conferences to representatives of transnational and regional networks of working children and former child workers enabled more contentious debates on child labour policy, albeit to a lesser degree and for a shorter period. The first and longest-standing coalition advocated for strict minimum wages and abolitionist policieswhich was justified, for example, by the argument that such labour prevented children from attending school (IPEC, n.d.). The child worker coalition instead advocated for an improvement of working conditions (health & safety, wages, working hours) and for a perspective on child work that would see it not as obstructing education and schooling but as vocational training and the acquisition of work and life skills. Hence, they held a position that was not in line with the dominant position in the discourse. In terms of underlying values, the abolitionist coalition based their position on such values as the protection of children from harm and exploitation and the fulfilment of the right to education. The child worker coalition based its position on values such as self-determination and the importance of good working conditions. Rather than seeing work and education as mutually exclusive, it understands work as a means of gaining necessary knowledge and skills.

Dynamics of norm collisions in drug control, trafficking and child labour
In the issue area of international drug control, we focus on a norm collision between prohibitive and criminalising norms on the one hand and the right to health on the other hand (Burke-Shyne et al. 2017). The right to health first appeared in debates about the right to health for drug users. More specifically, the rise of HIV/AIDS and attempts to ensure fewer infections gave rise to this debate (Levine 2003, 148). In the issue area of human trafficking, we have studied a norm collision between an absolute ban of prostitution on the one hand and regulated prostitution on the other hand, with both sides in the conflict referring to various human rights norms (Doezema 2005;Lobasz 2019). More specifically, those arguing for an absolute ban refer to the prohibition of sexual exploitation in the Palermo Protocol, arguing that any form of prostitution equates to sexual exploitation. Those arguing for regulating prostitution hold that there should be a distinction between forced and voluntary prostitution. They acknowledge that despite structural factors that "may lead to women's involvement in sex work, including poverty, gender inequality and global restrictions on migration for work", women can decide to engage in sex work (Wylie 2016, 56). In the issue area of child labour, we look at a norm collision between the prohibition of child labour and the right to work. The working children advocacy coalition furthermore claims violations of procedural rightsthe right to participation and "to be heard"as laid down in Article 12 of the Convention on the Rights of the Child (CRC). Before providing a detailed analysis of the ACC and their effect on norm collisions, the following table (Table 1) provides a summary of the issue areas and their characteristics.

International drug control: one coalition, successful advocacy for harm reduction
The drug control issue area has seen advocates of harm reduction come together in an advocacy coalition to promote respective measures such as needle exchange programmes or substitution therapy. While this has been of limited success so fartheir view has not informed outcome documents of international meetingsharm reduction has at least

Drug Control
Prohibition and criminalisation of drugs except for medical and scientific purposes.
One advocacy coalition, challenging the dominant position of criminalisation and prohibition.
Drug criminalisation and prohibition vs harm reduction and the right to health.

Human Trafficking
Prohibition of human trafficking.
Two coalitions with opposing views on prostitution.
Banning all forms of prostitution vs regulating voluntary prostitution (prohibition of sexual exploitation vs acknowledgement of the independent decisions by women to engage in sex work).

Child Labour
Prohibition of (the worst forms of) child labour. Prohibition of child labour under the age of 15 (14 in some cases).
Two coalitions, one supporting the dominant ILO position and one challenging the prohibition with reference to right to work and participatory rights.
Right to work vs prohibition of child labour.

Source: Authors.
been put on the agenda. The following paragraphs show how the debate on harm reduction developed within the relevant IOs in the field. In a joint report from 1998, the Joint United Programme on HIV/AIDS (UNAIDS) and the World Health Organization (WHO) supported harm reduction measures to make the use of injection drugs safer (UNAIDS and World Health Organization 1998, 36-39). The International Narcotics Control Board (INCB) 1 had always been critical of respective measures, such as needle exchange programmes or safe injection rooms, and criticised such measures for violating the drug prohibition norm (Bewley-Taylor 2004, 485). In addition, actors such as the WHO have pointed out that the drug prohibition regime has led to limited access to relevant medicines for HIV/AIDS patients, thus violating the right to access adequate medicines and the right to health (World Health Organization 2011).
In 2009, the Human Rights Council passed a resolution focussing on harm reduction and HIV/AIDS. The Council noted that there was a need to ensure access to relevant medical services for drug users (Human Rights Council 2010, 55-56). Yet, the term harm reduction was not included in the final version of the Commission on Narcotic Drugs (CND) 2009 Political Declaration (Rushton 2018, 272-273). 2 Moreover, in the same year, during a CND Session, a group of states made an "interpretative statement" (CND 2009, 119): "The statement interprets the term 'related support services' used in the 2009 Political Declaration and Plan of Action as 'harm reduction measures'" (Jelsma 2017, 16 at footnote 50). This interpretation was immediately opposed by delegates in the assembly hall. As scholars have noted, "[w]hile some met this [statement with applause], others rose to declare that they could not agree with this statement" (Bewley-Taylor 2009, 9;Horwood 2009, 12).
The issue of harm reduction also resurfaced during the UN General Assembly Special Session on Drug Control in 2016 (Rushton 2018, 267). In both forums, the CND session and the UNGA Special Session, transnational advocacy coalitions were able to put a perceived norm collision, in this case between prohibitive measures of drug control and the right to health, on the agenda. In addition, the case provides evidence of the transnational advocacy coalitions' attempt to frame drug prohibition policies as violating drug users' fundamental human rights such as the right to access healthcare and the related practice to be provided with safe injection facilities. Establishing harm reduction as a policy grounded in human rights norms thus led to the activation of a norm collision in the international drug control regime that had hitherto been marked by prohibition of drug use and criminalisation of drug users.

Human trafficking: opposing coalitions, unresolved norm collisions
The human trafficking issue area is characterised by two strong opposing advocacy coalitions with clearly diverging if not opposing policy beliefs regarding the question of how to understand and conceptualise human trafficking. Human trafficking as an issue dates back to the early twentieth century, where a number of conventions against trafficking signed by European states addressed the issue of human trafficking of European women for the purpose of sexual exploitation. Apparently, these conventions and the subsequent practice were more concerned with fighting prostitution than human trafficking (McClean 2007, 15;Shin 2018, 17-18). In 1949 the UN General Assembly passed a Convention against Human Trafficking, which expanded the issue for the first time to all genders and all forms of human trafficking. In the 1970s, global movements for women rights put the issue of human trafficking on the international agenda again (Bruch 2004, 11). The topic was further discussed during various international conferences in the 1980s and 1990s, where it was often framed as a social issue related to discrimination. In 1994, a World Ministerial Conference on Organized Transnational Crime mandated a group of experts to put forward an instrument against trafficking (Charnysh, Lloyd, and Simmons 2015, 326-327). Many saw the 1988 Drug Trafficking Convention as a useful model (McClean 2007, 11;Jakobi 2013, 99).
The negotiations of what later would become the Palermo Protocol took place within an Ad-Hoc Committee between 1998 and 2000. While there was a consensus that human trafficking had to be tackled as an issue of international crime and that an international instrument was necessary, the issue of prostitution became the subject of "heated debates" (Chuang 2006, 438). During the negotiations of the Palermo Protocol as the most recent and central international treaty to address trafficking in human beings, two opposing advocacy coalitions put the question of how to deal with sex work/prostitution in the context of human trafficking on the international agenda. Due to different normative positions on whether human trafficking applies to all forms of sex work/prostitution, they fiercely debated with each other and focused their advocacy on the definition of human trafficking and, relatedly, the question of how to address prostitution. In this debate, the norm collision between prohibiting any kind of sex work/prostitution as human trafficking on the one hand, and differentiating between voluntary prostitution (sex work) and forced prostitution on the other, came to the fore. More specifically, this was about the question of whether all forms of sex work/prostitution constitute sexual exploitation (which was to be prohibited) or whether women can choose to engage in sex work/prostitution (Wylie 2016, 44, 56).
This example underscores our assumption that differences in core and policy beliefs facilitate norm collisions. Scholars, as well as practitioners, argue that human rights were left out of the debate on human trafficking because the two advocacy coalitions diverged on the issue of sex work/prostitution (Ditmore 2012, 112;Locher 2007;Bruckmüller and Schuhmann 2012, 114). This was especially the case when it came to victim protection, which was only discussed late in the negotiations (Shin 2018, 20) and ultimately appears in the protocol only in weak and non-binding terminology. From this observation, we can draw the tentative conclusion that the presence of two competing advocacy coalitions with opposing views on one issue, may also lead to a situation where other potential norm collisions, here between the prosecution of human trafficking vs victim protection, do not come to the fore. Even though both of these groups perceived the lack of victim protection as problematic, a lack of joint agenda-setting prevented open discussion. Yet, both coalitions framed the issue of sex work/prostitution and human trafficking in terms of human rights, "[…] either with a focus on the protection of bodily integrity of vulnerable persons or by referring to socio-economic rights" (Hahn and Holzscheiter 2013, 504). 4.6. Child labour: power asymmetries between abolitionist and pro-work coalition, a short-lived norm collision The field of cooperation against child labour was, from the onset, shaped by a strong and rather unified advocacy coalition. Some of the oldest transnational civil society organisations such as the Anti-Slavery Society founded in 1839 (today Anti-Slavery International) and also workers' associations like the "International Trade Secretariats" have been traditional partners or members of the ILO (Myrstad 1999). Their abolitionist position on child labour was supported by states, trade unions, women's organisations, and charities. Thus, from the very beginning of international cooperation on labour standards, a number of non-state actors contributed to international law-making to regulate economic exploitation and activities of children with child labour prohibition being described as the "easiest question to agree upon" (Dahlén 2007, 112). When the first ILO convention on a minimum age in certain sectorsadopted in 1919no longer corresponded to the economic and social conditions in many countries, the ILO adopted a General Convention on minimum age for entry into employment (No. 138) in 1973. Aiming at "achieving the total abolition of child labour" it demands a gradual rise of the minimum age for employment to a level guaranteeing the full physical and intellectual development of young people (Dirks, Liese, and Senghaas-Knobloch 2002). Until the end of the 1980s, advocates promoted a modern notion of childhood as a protected, happy, sorrow-and responsibility-free phase of life. This notion turned into a largely undisputed value in international law with the adoption of the UN Convention on the Rights of the Child in 1989 (Boyden and Levison 2000).
Over time, however, actors became increasingly polarised. An anti-abolitionist coalition emerged, which held alternative interpretations of global childhood and surrounding values (Holzscheiter 2016(Holzscheiter , 2018. It challenged the line of division between harmless "child work" and harmful "child labour". Prohibitionist norms were seen as incompatible with progressive understandings of children's rights as well the cultures, traditions, and "realities" of developing states and indigenous communities (Myers 2001;Smolin 1999;White 1994). The increasing transnationalisation of new and highly visible non-state actors such as child worker associations, indigenous peoples' representations and, in general, voices from the Global South, led to a diversification of nonstate actors in the child labour regime. Eventually, the International Labour Office started to admit that child work contributes to the development of qualifications and skills (International Labour Office 1996).
It was in the context of the setting of new standards on child labour in the late 1990s that norm collisions in this area of international cooperation came to the fore. Conflicts between the pro-abolitionist and the anti-abolitionist advocacy coalition arose, particularly during the discussions preceding the unanimous adoption of ILO Convention 182 on the Worst Forms of Child Labour. The International Labour Office, ILO Member States, and what may be called the first truly global network of CSOs advocating for an end to child labourthe Global March against Child Laboursupported the new ILO Convention.
The new instrument was strongly opposed by several trade unions who feared that it would undermine a strict prohibition of child labour (Fyfe 2007, 26, fn 55). At the same time, it triggered criticism by child worker associations who claimed their right to be heard. During the 1997 Child Labour Conference in Oslo, they demanded a right to decent, regulated work instead of outright condemnation of child work. Their position e.g. that vocational training during working hours was often more useful than education in poorly run schoolssat uncomfortably with the beliefs and norms endorsed by many Northern delegations and provoked strong and hostile reactions by some state delegations who claimed that it was unacceptable to defend the idea that children should have a right to work (Miljeteig 2000, 18-19).
A strong power asymmetry prevails between the traditional, prohibitionist advocacy coalition on the one hand and the coalition of child workers and Southern NGOs on the other. While many members of the pro-prohibitionist coalition enjoyed privileged access to high-level intergovernmental conferences, the advocacy coalition for a right to work had only restricted access. In fact, at the ILO Conference in Den Haag in 2010, a pro-abolitionist Global March spokesperson was the only CSO speaker on the conference's high-level panel discussions. 3 And during the Child Labour Conference in Brasilia in 2013 child participation was relegated to young people's media reporting on the conference and an exclusive web space for "youth" called Child Labour Dialogues. During this conference, some participants explicitly criticised the fact that working children were not meaningfully included in the international debate on child labour. 4 Most recently, the ongoing denial of access to high-level meetings of ILO Member States resulted in a complaint of child worker associations in front of the Committee on the Rights of the Child (MOLACNATS 2017).
Despite its less privileged access and limited possibility for articulation, the pro-work advocacy coalition was most successful in agenda-setting by triggering an international debate on children's economic, social, and political agency coupled with a renewed attention to international norms on child participation and self-determination. The pro-work advocacy coalition, therefore, extended the international debate during the 1990s on how to differentiate between harmful child labour and acceptable child workwhich also involved a debate on the role of education and schooling in defining "harmful" work (i.e. as all forms of work obstructing school education). The ACC in the case of child labour also shows that child workers were successful in challenging the privileged status of long-standing ILO partner organisations (such as the International Trade Union Confederation or Anti-Slavery International) by claiming voice and agency. The transformative effects of the norm collision articulated by the pro-work coalition are evidenced by ensuing changes in the policies of child protection NGOs working on child labour, which increasingly embraced the idea of working children as competent social agents.
What is most interesting in the child labour case is that both advocacy coalitions framed child labour in human rights terms and heavily drew on the same set of norms, i.e. the general rights of the child as laid down in the UN Convention on the Rights of the Child. However, while the prohibitionist advocacy coalition was capitalising mainly on rights to protection (from work, exploitation), the pro-work AC privileged the political rights to self-determination in the UN CRC. We therefore conclude that the ACC exhibits normative inconsistencies and tensions already inherent in international law on children's rights (Holzscheiter 2018).

Discussion of results
The previous section provided an analysis of transnational advocacy coalition constellations and their effect on norm collisions in three different issue areas. We now return to our assumptions on the relationship between ACC and norm collisions. As is evident from our three cases, the mere existence of one or more advocacy coalitions in an issue area is not sufficient for a norm collision to come to the fore. Rather, our approach suggestsfollowing well-established insights on the importance of both framing and agenda-setting in advocacy workthat both of these activities matter in explaining how and when ACC affect the activation of norm collisions. Examining framing activities, we were able to observe that it mattered whether a coalition supported or opposed the prevailing discourse. The case of harm reduction vs criminalisation in the context of drug control provides an excellent example of a coalition opposing a hitherto uncontested prohibitive and criminalising approach. By framing drug control as an area that needed to be more responsive to human rights normsand specifically the rights of drug usersthe advocacy coalition was successful in problematising a norm collision and provoking a contentious debate on the appropriateness of conservative drug policies. In the issue area of child labour, we find that despite the existence of two advocacy coalitions with opposing positions, the more powerful coalition supported the predominant prohibitive discourse on child labour.
Examining agenda-setting, we find the relative power of the coalition and the relationship between multiple coalitions to be relevant. Our empirical analysis of ACC in child labour exemplifies the relevance of power asymmetries between multiple coalitions, with the more powerful coalition dominating the negotiations over a new ILO Convention on the "worst forms of child labour". By contrast, in the issue area of human trafficking, we find two equally powerful coalitions. They put the norm collision between regulating sex work/prostitution and abolishing all forms of sex work/prostitution on the agenda by framing the question of how to deal with sex work/prostitution as a question of selfdetermination and women's rights. It seems plausible to conclude that their opposing views on the matter coupled with the power balance between the two coalitions explains the fact that their respective goals were not included in the Palermo Protocol. Rather, a technical compromise was found, leaving the question to be decided by States Parties. At the same time, the conflictual relationship between both coalitions had the unintended consequence that questions of human rights and victim protection remained marginalised during negotiations and in the actual protocol.

Conclusion
This article has put forward the proposition that advocacy coalition constellations matter in explaining the emergence and unravelling of norm collisions in international politics. We thus extend research on the interplay between advocacy and international norms by looking at the fundamental role that constellations of advocacy coalitions play not only in processes of norm emergence or norm diffusion but also with regard to dynamic and often contentious relations between different norms. By activating norm collisions, ACC not only contribute to new norms, but they also destabilise a previous consensus or reinforce positions that had previously been silenced. Our argument speaks to an increasing literature that emphasises the centrality of actors and agency in norm dynamics, especially with a view to the role of transnational advocacy coalitions (Baumgarten and Amelung 2017). Our research results are also compatible with the view that "the multi-layered nature of the existing governance architecture also provides opportunities for transnational mobilisation" (Zajak 2017, 125-126). Finally, we also build on and expand the literature on norm development, norm contestation, and norm dynamics by providing a detailed analysis of how ACC affect norm collisions.
The core argument of this article is that the number of advocacy coalitions, their relative power, their underlying values, and their stance towards existing international rules and policiesin short, the specific constellation of advocacy coalitionsinfluences whether a norm collision comes to the fore. We find that advocacy coalitions contribute to the activation of norm collisions in at least two ways: advocacy coalitions bring in new norms that have not been applied to a particular issue before, or they emphasise norms that have increased in strength to overthrow a former consensus. Our comparative empirical analysis also evidenced that the existence of several advocacy coalitions is not a sufficient condition for the activation of norm collisions, as advocacy coalitions can hold similar positions and share core values regarding a prohibitive regime and its underlying norms. Rather, our analysis has demonstrated that it is the specific constellation of coalitions that matters for explaining how and when norm collisions come to the fore.
Building on these insights, there are several promising avenues for future research: studying power asymmetries between coalitions in a more systematic fashion (Blomquist and Schlager 1996, 7), investigating whether advocacy coalitions remain normatively consistent across forums and venues (Keck and Sikkink 1998, 18), or assessing whether our argument also holds in a most different systems design, for example when comparing prohibitive and permissive international regimes or regimes with low and high involvement of civil society organisations. Linking our arguments with the burgeoning literature on regime complexity or governance complexes, we would also suggest extending research towards the study of ACC as a result of complex institutional landscapes in which some actors may wield considerable power not even in multiple advocacy coalitions but also, potentially, across different issue areas (Alter and Meunier 2009). Several studies on the power of philanthropies have shown that there are powerful private actors that are equipped with considerable material (money, staff) and immaterial (networks, knowledge, data) resources and thus find themselves in a privileged position to navigate complex landscapes (Rushton and Williams 2011;Youde 2013;Seitz and Martens 2017). A closer analysis of the types of non-state actors involved in advocacy coalitions and the influence of private funding institutions and business actors in forging coalitions seems to be appropriate in light of the contentious politics surrounding the term "non-state actors", in academia as well as within the confines of international organisations.