Roos de Groot

(1)

Orchestration governance beyond the orchestrator;

an empirical exploration of intermediary and target dynamics in

climate-smart agriculture governance in Tanzania

Master thesis Political Science: Political Economy Roos de Groot (12727318)

supervisor: dr. P. Schleifer

second reader: dr. C.M. Roggebrand June, 2020

(2)

Abstract

To solve complex global policy issues such as climate change, terrorism or migration, new modes of governance have emerged whereby the traditional direct and hierarchical role of the governor has been replaced by more complex governance environments and various state and non state-actors interact amongst each other on different governance levels. Orchestration governance is a new scientific concept which is characterized by a similar complex stakeholder environment as such whereas governance occurs under a three-party relationship and norm adaptation takes place without mandatory compliance. In orchestration, the orchestrator (governor) hereby mobilizes a third party (the intermediary) to execute tasks in the pursuit of aligned governance goals (Abbot et al, 2012). These intermediary actors in turn work without control from the orchestrator and attempt to influence target actors (nation-states, private entities, civil society) to certain governance behaviour. This research examines empirical dynamics between intermediary and target actors, that in contrast to orchestrator-intermediary relationship have received limited theoretical and empirical attention in scientific literature. By examining the orchestrated policy field of climate-smart agriculture (CSA) as an extreme case this thesis explores potential conditions under which intermediary-target relationships flourish. During this exploration the degree of specialization, intermediary multiplicity and representativeness are identified as potential factors for successful intermediary-target relationships. In doing so this thesis aims to contribute to further theoretical development of orchestration governance, as well as identify policy implications for orchestrated policy fields.

(3)

Foreword

This thesis was written during the second semester of the academic year 2020 for the Master Political Science at the University of Amsterdam. Writing my master thesis has been an interesting journey, although the alienating context of the Covid-19 pandemic at times did bring about some unease and insecurity.

However, I did have the opportunity to fully emerge in the CSA governance paradigm and speak to several people for which I am sincerely grateful. First of all I want to thank Ms. A.S. Natai and Mr. S. Kingazi, former and present chair of the Tanzanian Climate Smart Agriculture Alliance for taking their time to share their knowledge and respective visions on (Tanzanian) CSA governance with me. Secondly, I would like to thank my other interviewee who chose to remain anonymous, for sharing a deliberate vision on CSA from a local NGO perspective.

Furthermore, I want to mention my dear friend dr. Henry C. Umeodum who really helped me during the process of data-collection by arranging interviews. Lastly, I would like to express my sincere gratitude to my thesis supervisor dr. Philip Schleifer for the adequate assistance and concrete feedback during this thesis project. Without the assistance of Philip during some indecisive moments, this project would never came across as is has done right now.

Roos de Groot, June 2020

(5)

List of Abbreviations

ACSAA African Climate Smart Agriculture Alliance

AU African Union

ANSAF Agricultural Non State Actor’s Forum

CAADP Comprehensive Africa Agriculture Development Program

CCAFS Climate Change, Agriculture and Food Security

Research program

CGIAR Consultative Group on International Agricultural Research

CIAT International Centre for Tropical Agriculture

COMESA Common Market for Eastern and Southern Africa

CSA Climate-Smart Agriculture

CSR Corporate Social Responsibility

EAC East African Community

ECOWAS Economic Community of West African States

FANRPAN Food, Agriculture and Natural Resources Policy Analysis Network

FAO Food and Agriculture Organization (of the United Nations)

FARA Forum for Agricultural Research in Africa

GACSA Global Alliance for Climate Smart Agriculture

GCAA Global Climate Action Agenda

GCF Green Climate Fund

GFSI Global Food Security Index

INGO International Non Governmental Organization

IGO International Governmental Organization

LPAA Lima Paris Agreement Agenda

MAT Ministry of Agriculture of the Republic of Tanzania

MISP Ministry of State and Planning of the Republic of Tanzania

NATO North Atlantic Treaty Organization

NEPAD New Partnership for Africa’s Development

NAIP National Agricultural Investment Plan

OSAA Office of the Special Advisor on Africa (of the United Nations)

P-A Principal-agent

PPP Public Private Partnership

QAS Quality Assurance Scheme

SAT Sustainable Agriculture Tanzania

SUA Sokoine University of Agriculture

TAHA Tanzania Horticultural Association

TCSAA Tanzanian Climate Smart Agriculture Alliance

UNEP United Nations Environmental Program

UNFCCC United Nations Framework Convention on Climate Change

(6)

1 Introduction

1.1 Research overview

In today’s day and age governance arrangements no longer solely take place in the conventional state or interstate vacuum. Solving complex global issues requires continuous political cooperation on various governance levels whereby non-state actors convene increasing political power (Weiss & Wilkinson, 2014). To solve complex issues as such, new modes of governance have been developed whereby increased interaction takes without the use of traditional modes of governance. Orchestration governance is a relative new scientific field in global governance which differs from traditional governance modes due to its three-party relationship.

In orchestration, third parties are mobilized by governors to independently take over governance goals in the pursuit of aligned governance goals (Abbot et al, 2015a; Hale & Roger, 2013). This three-party relationship often reflected in the O-I-T model starts by governance actors (orchestrators) such as such as international governmental organization (IGO’s) who mobilize third parties (intermediaries) as go-betweens for certain governance tasks which they cannot execute themselves. Intermediaries in turn independently aim to influence targets (nation state, private entities and civil society) to certain norm-adaptation. This type of governance is thus characterized by indirectness as well as by voluntary standards, indicating that norm-standardization in orchestration takes place without formal political mandate, sanctioning or prohibited compliance from the intermediary over their target. (Abbot et al, 2012).

As orchestration governance is a relatively new field of study, empirical research has been mostly focused on the mapping of orchestration in fields such as transnational environmentalism, global labour standards and global health (Chan & Amling, 2019; Hanrieder, 2015; Bacarro, 2015). Yet, limited attention has been attributed to the policy field of sustainable agriculture governance. The mapping of new empirical fields such as climate-smart-agriculture can therefore contribute to a better understanding of the actual functioning of this soft power governance mechanism. Theoretically, the O-I-T model has furthermore only been partially researched. Even though the relationship between orchestrators and intermediaries has been described extensively, on the contrary limited research has focused under what conditions, intermediary

(7)

actors influence their designated targets actors in orchestration (Abbot et al, 2015a; Roger & Hale, 2013: Bulkeley & Kern, 2013; Havinga & Verbruggen, 2017).

By empirically examining, orchestration in the climate-smart agriculture paradigm in Tanzania, this case study aims to explore these intermediary-target dynamics in a new policy field. The CSA governance paradigm in Tanzania is therefore examined following a single case-study, based on an extreme-case model. The case of Tanzania can be considered as an extreme case model as some CSA intermediaries were highly successful in influencing their respective target-actors. Since the mobilization of CSA intermediaries in 2014, CSA policies, programs and research were implemented in a rapid period of time. National government actors, local civil society organizations and local research institutions have been increasingly involved in CSA (NEPAD, 2014). An extreme case model as such offers a great opportunity to explore new fields and trends and is therefore beneficial for the aim of this research: to understand the role of orchestration in CSA governance and consequently explore it’s intermediary-target dynamics in Tanzania (Seawright & Gerring, 2009). In doing so, this thesis likes to contribute to potential further theoretical development of orchestration theory, as well as attempt to identify policy implications for indirect soft power governance mechanisms in complex stakeholder environments.

Based on the analysis, this research indicates that successful intermediary- target dynamics were positively linked to specialized role division and multiplicity of intermediaries. Intermediaries with a high density of specialist actors, were able to increase norm-adaptation in various target groups whilst a less complex intermediary structure with limited specialization influenced less target actors. Specialization and multiplicity enabled intermediaries to pursue various governance tasks and interact which various target actors in government and civil society. This consequently increased norm-adaptation and governance effectiveness. By demonstrating the latter, this thesis likes to demonstrates that complexity and increased stakeholder interactions are mere useful tools for governance effectiveness in orchestration, as it enables participation and collaborative arrangements, necessary to solve these complex policy issues which are distinct in orchestration governance schemes.

(8)

1.2 Research questions

Within the span of this research, the core research question is set out to explore how intermediaries influence targets in orchestration governance. To examine the latter, an in-depth qualitative case study on climate smart agriculture governance in Tanzania will form the central point of analysis. The main research question in this study is therefore: under what conditions do CSA-intermediaries successfully influence target actors to climate-smart agriculture governance behavior in Tanzania? Climate smart-agriculture governance behavior can be defined as the process of adapting and adhering to CSA norms.

This main question will be followed by a descriptive sub-question, as is reflected in question one. In this chapter the emerging field of CSA governance will be shortly described. Subsequently a stakeholder mapping will be executed to determine how how actor-interactions are structured. Finally a stakeholder analysis is indicated to determine which actors engage in it, what roles do they deploy, and ultimately to define whose choices determine CSA norm adaptation (Scharpf, 1997). Consequently a more in depth analysis will take place to explore how successful intermediary- target relationships occur based on prior identified hypotheses. Lastly, sub question three will shed light on further policy implications in orchestration governance, this part will be integrated in the concluding chapters of this thesis.

An overview of underlying research questions can be found below:

I What is the role of orchestration in climate-smart agriculture governance in Tanzania? (descriptive subquestion)

II. Under what conditions do climate-smart agriculture intermediaries successfully influence target actors to climate-smart agriculture governance behavior in Tanzania? (analytical main research question)

III What policy implications can (potentially) be made based on the empirical findings between intermediaries and target actors in climates smart agriculture governance in Tanzania? (concluding paragraph)

(9)

1.3 Societal and academic relevance

A valid justification from a societal and academic perspective is needed to ensure the relevance of this study and its research questions. This chapter offers a reflection on both the social as the academic relevance of this thesis.

Societal relevance

Scientific research has indicated that agriculture is globally one of the most significant contributors to climate change. Vice versa, climate change results in fewer crop growth and failed harvests which together with global population growth implicates risks for food security, and the planet as a whole, especially for workers in subsistence agriculture (Lippler et al, 2019). To combat risks as such, the Food and Agriculture Organization (FAO) developed the climate-smart agriculture policy (CSA) paradigm to enable policy solutions to counteract climate mitigation, whilst ensuring climate adaption and food security at the same time (FAO, 2013). The CSA paradigm hereby focuses on Global South states which are highly dependent from agricultural commodity trade and above average prone to natural disasters and other climate vulnerabilities.

To enable implementation of such complex global policy issues, governance actors have mobilized independent stakeholder to raise awareness, coordinate, implement and monitor the progress of CSA policy and program integration. By empirically reflecting on how the relatively new CSA governance paradigm is orchestrated, a better and comprehensive view of their relevant activities and role division can be formulated. By examining how successful CSA relationships between intermediaries and targets function, this research can furthermore pose new policy implications for effective CSA governance globally. Not only could these newly identified conditions potentially be applicable on the whole CSA orchestration governance paradigm, but after further theoretical exploration, identified conditions could potentially similarly contribute to other policy fields which are based on indirect and soft power mechanism such as migration, global health and transnational environmentalism. In doing so described findings could enable other soft power governance mechanism in other sectors to execute their governance goals more efficiently.

(10)

Academic relevance

The academic justification for this study can amongst others, be found in the empirical exploration of new policy fields. Empirical studies on orchestration are often solely focussed on specific global policy issues such as transnational environmental policies, global health or international labour standards (Chan & Amling, 2019; Hanrieder, 2015; Bacarro, 2015). Yet orchestration characteristics are similarly present in other policy fields and empirical research within new policy fields is therefore highly relevant. The field of global sustainable agriculture governance and the climate-smart agriculture governance have not been examined yet, but for instance share extensive similarities with other indirect soft power governance fields. Given the fact Climate-smart agriculture governance arrangements and it respective institutions are quite new, this case offers high potential to explore what role orchestration plays in a new and still developing global policy field.

Besides the need for research in new policy domains, it should be noted that almost no theoretical nor empirical research has been attributed to the roles of intermediaries and their influence on targets in general (Havinga & Verbruggen, 2017). Since the establishment of orchestration theory and the construction of the O-I-T model, orchestration has significantly focused on how and why orchestrators engage third parties in the pursuit of their governance goals. Constructors of the model have amongst others focused on the reasons why orchestrators mobilize intermediaries and what their specific roles during orchestration contain (Abbot et al, 2009; Abbot et al, 2012; Roger & Hale, 2013; Abbot et al, 2015). These implications are similarly illustrated by the fact that all four key assumptions on orchestration described by Abbot and colleagues are attributed to the relationship between orchestrators and intermediaries (Abbot et al, 2015). Whilst the relationship between orchestrators and intermediaries received thus significant attention, the relationships between intermediaries and targets is neglected in the theoretical debate. In empirical studies the roles of effective intermediary-target relationships has similarly been absent. Being aware of how intermediaries influence targets to certain governance behavior is important to gather theoretical insight on how orchestration governance can be successful. The strong emergence of CSA-governance in Tanzania can hereby potentially be considered as a first step to further theoretical development of the O-I-T model of orchestration.

(11)

3 Literature review

This chapter introduces the key features and debates of orchestration governance theory and the O-I-T model, constructed by Abbot and colleagues. By examining its distinct features in contrast to other governance models such as the hierarchical unidirectional Principle-Agent model and governance model, an extensive reflection of the current status quo of orchestration theory will be visible. Furthermore there will be specific attention for empirical research of orchestration governance mechanisms in other policy fields such as transnational environmental politics and global health.

3.1 Global governance

Global governance is a subfield of international relations that focuses on the governance of policy issues that interactively move beyond the conventional state or interstate levels such as climate change, migration, and terrorism (Pattberg, 2010; Koser, 2010; Makinda, 2003). In the field of transnational governance, stakeholders from society, state and market come together and independently set and influence regulation in various roles and functions. Whilst (neo)realist deny the independent political roles of non-state actors, in global governance, there is explicit attention for this high variety of non-state actors and its interactions which influence and define the global regulatory order (Weiss & Wilkinson, 2014; Krassner, 2001; Rosenau, 1995). Since the 1980’s the emergence of this field has rapidly developed due to a significant increase in hybrid governance arrangements whereby public governance actors cooperate or independently hire private entities. Arrangements as such have instigated increasing political authority for non-state actors. The emergence of these arrangements is often based on the assumption that new modes of governance are necessary to overcome the inadequacy of both state- and market failure (Abbot & Snidal, 2010). Yet increased transnational action of non-state actors sometimes results into choices which may not totally be in the interest of the public realm. In some cases increased transnational non-state actions has implicated unequal power play, window-dressing and ineffective policies overall (Chan & Amling, 2019). In order to overcome issues as such, academic most often point to governance by global actors such as international governmental organizations whose goals should be to oversee, coordinate and align transnational governance goals into concrete action.

Similarly to state governance, global governance actors use traditional regulatory measures which are characterized by mandatory rules and a strong sense of

(12)

hierarchy. Following the conceptualization of transnational governance of Abbot and colleagues (2015) international governmental organizations (IGO’s) and supranational actors can directly establish, implement and monitor regulations and adjudicate international law, which the authors conceptualize as hierarchical governance. To establish governance goals, international actors, however often resort to alternative modes of governance. Hereby techniques such as delegation come into place whereby actors delegate specific tasks to third parties. Examples are the establishment of regulatory agencies that are in charge of implementation and compliance of standards such as occurs in food safety (Abbot et al, 2015a). Important is hereby that the hierarchical relationship between the appointed third party and the IGO remains in place. Transnational governance actors, however, often choose to ‘regulate and govern’ through soft norm adoption. Collaboration governance is a similar governance form which focusses on encouraging actors to achieve certain behavior on voluntary premises through collaborative initiatives between the traditional rule makers and rule takers. (Reinicke 1998; Rosenau 1995). The fourth and final mode of governance is characterized by both indirect and soft elements. In this case, governance actors appoint third parties to indirectly influence certain behavior on voluntary premises. The role of the intermediary is hereby independent, meaning that the rule-maker cannot enforce hard control over the intermediary. Abbot and colleagues define this form of governance as orchestration (Abbot et al, 2009). An overview of these four archetypes can be found in table 1.

table 1: Governance modes in transnational governance by Abbott, K.W., Genschel, P., Snidal, D, & Zangl, B. 1

3.2 Orchestration theory

The concept of orchestration, developed by Abbot and colleagues contrasts highly with the traditional and hierarchical views on regulation and governance (Abbot et al, 2017b). Traditionally, transnational governance is reflected in a rather direct

direct

indirect

hard

hierarchical

delegation

soft

collaboration

orchestration

Abbot, K.W., Genschel, P., Snidal, D, & Zangl, B. (2015a). International Organizations as 1

Orchestrators. Cambridge: Cambridge University Press.

(13)

relationship between rule-maker and rule-taker as is reflected in the hierarchical principal-agent model (PA-model) (Abbot et al, 2012). Orchestration governance is on the contrary defined by a three-party relationship whereby a third party besides the rule-maker and rule-taker is mobilized and independently executes tasks to achieve transnational regulatory goals of the rule-maker. The approach similarly contrasts with the P-A model as there is no immediate link between the rule-maker and rule-taker whereby activities of intermediary are executed without direct control of the rule-maker (Abbot et al, 2012). In orchestration, norm-adaptation similarly takes place under a high degree of voluntary participation, limited hierarchy nor presence of direct authority (Abbot et al, 2017b; Kern & Bulkeley 2009). Abbot and colleagues (2017a) furthermore confirm the broadness of the approach by emphasizing that orchestration approaches also include hybrid governance arrangements between public and private actors on (inter)national, transnational and local levels and both formal and informal decision-making. To reflect these governance arrangements scientifically Abbot and colleagues constructed the O-I-T model in 2012. In this model, rule-makers (the orchestrator) mobilize a third party (the intermediary) to independently influence the behavior of the rule-taker (target) based on the underlying norms of the rule-maker (Abbot et al, 2012). A distinction between this P-A model and the O-I-T can be found in figure 1.

figure 1: P-A Model and the O-I-T model based on Abbott, K.W., Genschel, P., Snidal, D, & Zangl, B. 2

Principal

Agent

Orchestrator Intermediary Targets

Besides the indirect and voluntary relationship between orchestrator and intermediaries, the O-I-T model is based on two other assumptions. The goals-seeking assumption pinpoints that orchestrators and intermediaries engage in orchestration to achieve similar goals. Underlying reasons to engage in orchestration can, however, differ from orchestrator to intermediary, but goals are aligned (Abbot et al, 2015a). Furthermore, there is the complementarity capabilities assumption embraces that

Abbott, K.W., Genschel, P., Snidal, D, & Zangl, B. (2015a). International Organizations as 2

Orchestrators. Cambridge: Cambridge University Press.

(14)

orchestrator- intermediary relationships occur based on a mutual dependent relationship. Following the model, orchestrators choose for intermediaries as they lack regulatory competence, operational capacity, and authority to effectively enforce the regulatory goals solely (Abbot et al, 2012). Vice versa, intermediaries depend on facilitation, support and government capacities by the orchestrator. This mutual dependence and specialization from all parties are crucial in the O-I-T model. Rather than the P-A model, the O-I-T model is therefore associated with increasingly complex governance tasks which require in-depth collaborative solutions and expertise. Principals in the P-A model are often not capable to solve complex transnational collective action problems solely as they require input from the agent which is not possible in a strict hierarchical relationship(Roger & Hale, 2013: Abbot et al, 2015a). In an attempt to broaden and further distinguish the field of orchestration governance, Abbot and colleagues constructed the R-I-T model in 2017 to highlight the distinctive roles which intermediaries can take in regulatory governance. In R-I-T there is more emphasis on indirectly regulating states (bypassing). Hereby this three-party model differs from the O-I-T model as it focuses on strict rules enforcement, rules standardization, enforcement and compliance rather than voluntary norm adoption. The R-I-T model hereby indicate a mere ’’unidirectional relation’’ between the regulator and the intermediary whereby the regulator remains in control over the intermediary (Abbot et al, 2017). This hierarchical relation between R and I enables distinctive roles for both actors in the model . Hereby the rule-maker (R) promulgates the rules independently whereas the intermediary (I) translates the rules in clear executable standards. Whilst intermediaries have an indirect influence on rule-standards the regulators remain in hard control over the set rule-standards. In the R-I-T model therefore provides room to solve complex regulation problem such as occur in food safety or in global value chains.

3.3 Orchestration and legitimacy deficits

Yet, although orchestration governance seem powerful in overcoming transnational collective problems, the overall effectiveness of orchestration is still considered fragile and controversial. The use of third party governance can also instigate concerns for legitimacy due to the associated risk of conflict of interest, capture and unequal bargaining power. Abbot et al (2015a) argue that orchestrators and intermediaries are often incapable of achieving governance goals which can instigate problems such as whitewashing and blue-washing. In some cases, this implicates that the use of

(15)

orchestration can be reflected as mere ’’symbolics gestures’’ (Abbot et al, 2015b). Abbot et al (2015b) assume that orchestration ensures implications for legitimacy as it does not offer public authority or electoral accountability. Due to the independence of the intermediary, orchestrators tend to have limited control. Bäckstrand & Kuyper (2017) examined the democratic legitimacy of two orchestration mechanisms in climate governance based on deliberation, participation, accountability, and transparency. The authors found huge democratic pitfalls concerning transparency and accountability standards and found that these concerns were higher for the ’’bottom-up’’ orchestration mechanism in which orchestrator and intermediary functioned on similar low-level practical levels. Whilst the high level orchestration mechanism whereby both the orchestrator as the intermediary were highly institutionalized, experienced less problems regarding transparency and accountability (Bäckstrand & Kuyper, 2017)

The preceding part, will offer a closer look on the respective actors in orchestration governance of the O-I-T model as well as the underlying dynamics between these actors.

3.4 Orchestration: orchestrator-intermediary dynamics

Following orchestration literature, orchestration is a specifically relevant mechanism for international governmental organizations (IGO’s) because these actors have limited authority to enforce mandatory regulation themselves. IGO’s lack the authority of nation-states, who have this capacity but are subsequently reluctant to disengage their hierarchical power to IGO’s (Abbot et al, 2015a). Hereby especially IGO’s with relatively limited capabilities such as UNEP adopt orchestration governance whilst authoritative international organizations such as the World Bank and NATO choose to do not. Abbot et al (2012) argue that by making use of orchestration IGO’s with a broad mandate but limited authority can partially overcome their limitations through indirect influence over domestic affairs. Roger & Hale (2013) however, reject the general implication that orchestrators in global governance are solely international governmental organizations. According to the authors, nation-states can fulfill a similar role as orchestrators themselves. The authors demonstrate the latter by referring to the quality assurance scheme (QAS) whereby the UK government mobilized a private firm to set out a voluntary global scheme for carbon offsetting (Roger & Hale, 2013).

(16)

The scientific debates on orchestration have furthermore been explicitly focused on the role of orchestrators. In terms of roles, there is an overall consensus that orchestrators have a supportive role. This role can both be material as well as ideational. Within the process of orchestration, orchestrators focus on the creation of an enabling environment for the intermediary to influence the target. Abbot al identified hereby special roles for orchestrators. Firstly, orchestrators bring necessary actors together which the authors refer as convening. Secondly orchestrators provide technical assistance in the implementation, for instance by financing programs. Furthermore actors engage in endorsement activities which provide legitimacy for intermediary organizations for instance the WTO who endorsed World Food Safety Standards (Abbot et al, 2012).

Besides the roles of the orchestrator, discussion often focuses on why orchestrators enable third parties to achieve governance goals and this way outsource their own responsibilities. Abbot and Snidal (2009) similarly examined under what theoretical conditions, orchestrators involve intermediaries. Hereby the authors emphasize that reasons to incorporate intermediaries derive mostly from the assumption that intermediaries possess more capabilities to reach the designated transnational goals. Intermediaries can regulate and facilitate norm adaptation more effective and against lower costs. One of these capabilities is the operational capacity of intermediaries whereby third parties have more expertise and know-how than the orchestrator (Abbot & Snidal, 2009). Besides these distinctive capabilities, increased independence of the intermediary is often used to ensure compliance of the target. Intermediaries are considered more politically neutral actors than the orchestrator. Finally, intermediaries can have increased legitimacy because of their current activities or distinctive organization features and procedures (Abbot et al, 2015b). In short these conditions can be referred to as operational capacity, independence and legitimacy. 3.5 Orchestration: intermediary and target dynamics

According to the O-I-T model, intermediaries are mobilized because of their specific capabilities. In contrast to orchestrators, intermediaries can involve a variety of actions whereby they can take hybrid forms. Most mentioned intermediaries are non-governmental organizations. However, in other cases, intermediaries include transnational corporations, public-private partnerships, trans governmental networks or other IGO’s. (Abbot et al, 2015a). Although the role of the intermediary is considered vital in the O-I-T model, interestingly limited theoretical research has

(17)

focussed on developing specific theoretical variables with regard to their roles. Although thus never specifically integrated in the O-I-T model, researchers of orchestration generally agree that the roles of intermediaries vary and intermediaries can take over governance roles in all transnational governance process phases (Abbot et al, 2015, Roger & Hale, 2013) Intermediaries are for instance ought to convince targets to certain norms and behaviours, by offering targets, knowledge, advice and material support (Abbot et al, 2015a). Secondly, intermediaries are tasked to turn goals in practical guidelines and policy instructions which are clear and structured hereby facilitating further implementation of correlated goals (Abbot et al, 2015). Subsequently, intermediaries often function as monitoring agents and execute evaluations to establish insights on the current achievements of established goals. Based on these roles, Bäckstrand & Kuyper (2017) make the distinction between meta-intermediaries which are organizationally more closely positioned to the respective orchestrators and focused more on high-level agenda-setting, advocacy whilst regular intermediaries function on a more practical and implementation level near target actors.

Although the relationships between orchestrators and intermediaries are strongly developed in the O-I-T framework, limited to none attention has been attributed to the relationship between intermediaries and targets. Empirical research which indicates how targets are in fact influenced by intermediaries is missing (Havinga & Verbruggen, 2018). Abbot et al (2015a) are not very focal about the influence of intermediaries on target actors either but argue that orchestration through intermediary governance is used to either manage or bypass target’s authority and political mandate. Hereby is assumed that target actors (such as national governments) would generally oppose the interference of orchestrators and intermediaries consider a threat for their political sovereignty. Tallberg (2015) in turn investigated allegations as such and found that target actors generally prefer, rather than fear, being managed by IGO’s and designated third parties. Yet, still limited research has been done on how targets perceive intermediaries and whether this influences the success of the designated intermediary-target dynamic.

(18)

4 Analytical framework and hypotheses

This chapter uses key features and debates from the literature review to formulate an analytical framework and set hypotheses. In the analytical framework, the O-I-T model will be considered as an actor-centered approach and accompanying actors will receive conceptual clarification. Furthermore, the analytical framework will identify two hypotheses on how successful intermediary- target dynamics occur and finally provide extensive reflection on why these distinctive variables are used.

4.1 The O-I-T model as an actor-centered institutionalist approach

The research in this thesis will be executed based on an actor-centered institutionalist approach. An actor-centered approach differs from regular institutionalism as it argues that the analysis of mere governance structures alone without reference for actor-interactions cannot indicate clear results about actor’s behaviour nor the governance structures in which they interact (Scharpf, 1995). Following this approach, norm adaptation, especially a complex multi-stakeholder governance environment, requires a comprehensive reflection of these designated actors interactions as well as the designated structures in which governance actors interact.

Within this actor-centered approach this thesis will follow the key features of orchestration governance which classifies and examines governance actors based on their roles as orchestrator, intermediaries and targets. In orchestration governance, orchestrator-actors mobilize independent intermediary actors to influence the governance behavior of targets actors based on voluntary norm-adaptation (Roger & Hale, 2013; Abbot et al, 2017). The literature in orchestration has often focused on the relationship between orchestrators and targets; (O —> I). For instance how and why orchestrators engage intermediaries in the pursuit of their governance goals (Abbot et al, 2015, Abbot et al, 2012).

This thesis, on the contrary aims to unravel the complex relationships between (I —> T) e.g. under what conditions do intermediaries influence targets. To address these research questions this research will follow the structure of the O-I-T model (IGO orchestration) rather than the recently developed R-I-T model (Abbot et al, 2017) In the O-I-T model the designated tasks of the intermediary are executed independently from the orchestrator who does not have hard power or control over the independent intermediary. Hereby it contrast to the newly developed R-I-T model by Abbot and

(19)

colleagues in 2017, which assumes a stricter hierarchical relationship between regulator (R) and intermediary (I). Considering that the CSA policy paradigm is characterized by voluntary and indirect norm adaptation without hard enforcement, and strict unidirectional relationships between the orchestrator and its intermediaries, the O-I-T model is most applicable to the CSA governance case.

Although the O-I-T model will be followed as the central approach during this project, the reader should take into account that both the R-I-T as the O-I-T share extensive similarities. For these reasons some key assumptions and modifications related to the R-I-T model will be used in the preceding parts of this analytical framework. All R-I-T related assumptions used, will likewise receive thorough scientific argumentation concerning its applicability to the O-I-T model characteristics.

4.2 Conceptualization

An extensive conceptualization of actors involved can be found in the preceding sub-paragraph.

Orchestrator

Within this thesis, orchestrators are conceptualized based on the basic assumptions of the O-I-T model. Abbot and colleagues reflect orchestrators as actors who actively mobilize and support intermediary actors to execute tasks which lead to common governance goals (Abbot et al, 2012). Support from orchestrators to intermediaries can hereby be both ideational as material. Although Roger and Hale (2013) instigated recent debates about the potential role of states as orchestrators, this thesis classifies orchestrators from previous definitions made by Abbot and colleagues in 2012, defining orchestrators as international governmental organizations (IGO’s) (Abbot et al, 2012;Abbot et al, 2015a). Reasons to not include state actors in the conceptualization stir from the practical applicability as limited empirical evidence suggest that states have taken orchestrator’s roles within most other empirical global governance paradigms.

(20)

Intermediary

Intermediaries are conceptualized as those actors mobilized by orchestrator to execute certain tasks based on the pursuit of aligned goals and achieving behavioral change of target actors (Abbot et al, 2015a). Similarly to the underlying assumptions of the O-I-T model this thesis does not make a strong classification in what underlying characterizations defines intermediaries. The conceptualization includes all sorts of actors and entities. Intermediaries can either be state actors, NGO’s, other INGO’s, TNC’s and other hybrid arrangements such as PPP’s whereas public and private entities collaborate. Yet, this conceptualization contrasts to the O-I-T model as it configures the potential multiplicity of intermediaries mobilized to reach a certain goal within one orchestration mechanism (Havinga & Verbruggen, 2017). Besides multiple intermediaries, this conceptualization assumes that intermediaries can interact on different governance levels e.g local national, regional and global levels. In this conceptualization a subsequent distinction is made between meta-intermediaries and sub-intermediaries which indicate a different position and role within the orchestration regime. Based on this definition by Bäckstrand & Kuper (2017) intermediaries deploy more practical roles regarding implementation derived from a certain expertise whilst meta-intermediaries’ position is closer to orchestrators as they fulfill more coordinative and organizational tasks or are active in advocacy (Bäckstrand & Kuyper, 2017; Havinga & Verbruggen, 2017).

Targets

Following the O-I-T model targets refer to the actors which orchestrators wish to indirectly influence to certain behaviour by mobilizing intermediaries. Abbot et al (2015a) argue that targets involve national governments, business, research institutions and civil society organizations. In accordance with the O-I-T model this thesis embraces a similar definition. Hereby it builds on the O-I-T model by focussing on ‘target management’ whereby targets are indirectly influenced to certain behaviour, which is line with the designated governance goals of the orchestrator and the intermediary.

(21)

4.3 Analytical framework: mapping CSA governance

A descriptive overview of how CSA emerged in Tanzania is essential to answer the first research question. The first subparagraph will therefore be attributed to the emergence of CSA governance in Tanzania. Insights on how the CSA orchestration process was set out, emerged and what actors were involved are essential to grasp a contextual understanding of the role of orchestration in CSA governance. Secondly a stakeholder mapping will be executed to gather insights on how CSA governance is structured in Tanzania and what actors are involved. A mapping of the key structures and interactions of present actors is essential to gather a complete understanding of the CSA orchestration process. The first step to explore how norm-adaptation takes place is therefore to identify the set of interactions between, orchestrators, intermediaries and targets and classify their impact on the designated governance goals (Scharpf, 1997). This mapping offers an empirical and descriptive ordering-system to classify the interactions and relationships of actors within CSA governance paradigm (van Lieshout, 2008).

To map the governance structure of CSA in Tanzania, the extended version of the R-I-T model by Havinga & Verbruggen (2017) will serve as a theoretical lens. An example of similar models can reflected in figure 1. Within this model, there is more emphasis on the plurality of intermediaries. Intermediaries can either be independent or directly related to other intermediaries, serve multiple targets or orchestrators at once or change roles over time (Havinga & Verbruggen, 2017). More importantly Having & Verbruggen agree that intermediaries can have varying positions and functions in relation to the orchestrators or targets. Similarly to the assumptions from Bäckstrand & Kuyper (2017), Havinga & Verbruggen (2017) agree that some intermediaries are more closely positioned to intermediaries (meta-intermediaries) whilst other intermediaries are closely positioned to the target: (intermediaries). This modification of the earlier R-I-T model by Abbot and colleagues in 2017 offers more opportunities to explore the hybrid and pluralistic governance arrangements as present in the current climate-smart agriculture governance paradigm. As this extension does not necessary contrast to the key assumptions of the O-I-T model namely indirect, voluntary and independent relationships between intermediary and orchestrator, the use of the similar extension in this framework is therefore justified.

(22)

Figure 1: examples of modified RIT models by Havinga & Verbruggen 3

Based on the data derived from the stakeholder mapping, the final paragraph of the mapping will contain a stakeholder analysis based on the roles and achievement of the most vital actors. An analysis as such, is necessary to explore governance patterns of norm adaptation empirically. This descriptive stakeholder-analysis will identify which actors engage and interact in what roles, with as ultimate aim to systematically unravel what designated intermediaries influenced targets’ behavior. (Scharpf, 1997). 4.4 Analytical framework: the explorative analysis

This thesis provides two hypotheses about CSA intermediaries and target relationships (I->T). In formulating these hypothesis, this thesis has made use of orchestration theory and other recent debates in global governance. Identified hypotheses will form the key structure for the explorative analysis. However, given the fact that this study is based on empirical exploration, potential newly discovered variables deducted during the analysis will if necessary, be integrated within the analysis.

Havinga & Verbruggen, (2017). Understanding Complex Governance Relationships in Food Safety

3

Regulation: The RIT Model as a Theoretical Lens. The Annals of the Academy for Political and Social Science. Retrieved via: https://journals-sagepub-com.proxy.uba.uva.nl:2443/doi/

10.1177/0002716216688872

(23)

1. Specialization hypothesis

The influence of multiple actors in a varying governance context with overlapping mandates is often considered a threat for governance effectiveness. Actor multiplicity and co-existence is by most governance theorist associated with fragmentation as would enable inconsistency, unnecessary bureaucracy and potential conflict of interest (Keohane & Victor 2011; Orsini et al. 2013). Yet in orchestration, orchestrators and intermediaries often deploy various governance tasks whilst engaging with various target actors on different governance levels. Abbot et al (2015)a argue that orchestration theory considers this complexity of interaction patterns as contributing to governance effectiveness. Especially solving increasingly complex governance tasks require in-depth collaborative solutions in which input from various actors is necessary to ensure viable results (Abbot et al, 2012; Roger & Hale, 2013). By working collaboratively on specialized tasks, orchestration enables participation and input of actors which would not have taken place in a mere strict hierarchical relationship (Roger & Hale, 2013).

In global governance, increased participation of actors is similarly related to output legitimacy as targets actors which feel involved in the creation of the norms which concern them, are more likely to comply with norms which are created. (2008; Zürn 2005; Tallberg 2002) Increased interaction between various specialized intermediaries and target actors from civil society and business can therefore contribute to norm compliance as it increases overall ownership of the target to execute aligned goals. (Zürn 2005; Tallberg 2002). Intermediary multiplicity and specialization would following this ration rational thus enable, the participation of various different Tanzanian target actors in government, business, and civil society and enable the execution of specialized governance taks which in turn will increase norm adaptation The following hypothesis will therefore be central in the first chapter of the explorative analysis

Hypothesis 1: CSA intermediaries were able to influence Tanzanian target actors due to intermediary multiplicity and specialization

2. Representation hypothesis

Representation has been a vibrant subject in global governance research as governors in global governance are often criticized for being insufficiently representative. Main

(24)

focus of critique is hereby that in it’s decision-making process, some actor’s interests values, interests and beliefs prevail over others whereby often the less powerful or vulnerable groups and individuals remain unheard (Stewart,2014). A certain degree of representativeness is in orchestration even more important than in direct or hierarchical governance as intermediaries execute governance tasks without electoral accountability nor hard political hierarchy (Abbot et al, 2015a). Without representative values, orchestration governance can instigate concerns for legitimacy.

The second hypothesis therefore relates to the representativeness of the mobilized intermediaries in relation to their respective vision, interest and beliefs of the target actor which they wish to influence. Representativeness contributes that targets perceive the process of norm creation as legitimate e.g. fair, deliberative and transparent which consequently contributes to increased norm-adaptation (Beisheim & Dingwerth, 2008). Interest and values of these target actors need to be equally represented, meaning that potential underlying asymmetric differences in power or resources between the stakeholders do not lead to underrepresentation and dis-balances of certain values, beliefs and interests (Schmitter, 2002). This conceptualization of representation hereby thus goes beyond the conceptualization of equal participation as equal access does not consequently result in equal balance of values.

Orchestration theory and the O-I-T model do not make a distinction between the choice of entity of the intermediary nor the values they present with regard to effectiveness. Yet this thesis predicts that the specific values which the intermediary represent in relation to the targets they wish to influence, affects whether norm adaptation from these intermediaries takes place (Scharpf, 1997). Intermediaries who represent a mere particular interest or who are perceived to represent a mere interest which is not aligned to the designated interests of the targets or were some interests receive more attention over others are according to this hypothesis less likely to influence targets and enable norm compliance.

Legitimate representation is therefore conceptualized in twofold. First of all legitimate representation refers to the ability to represent the various beliefs, values and interest of targets actors in a way which is considered fair, deliberative and transparent. This conceptualization refers to the objective alignment of similar goals which are representative for the goals of the target. The second variable is hereby focused on the perception of the designated target with regard to the intermediary. Based on this

(25)

conceptualization targets should actively perceive the intermediary as legitimate representatives of their designated beliefs, values and interest.

The second hypothesis treated in this thesis therefore predicts that the degree of representativeness will play a role in successful intermediary-target relationships. The following hypothesis will therefore be evaluated during the analysis:

Hypothesis 2: CSA-intermediaries are able to influence Tanzanian target actors because these target actors are legitimate representatives.

Table 2: an overview of hypotheses and identified independent variables

4.5 Reflection on hypotheses

As is reflected in the previous paragraphs of this thesis, this research will examine whether the degree of specialization, actor multiplicity and representativeness of the intermediary had positive impact on norm-adaptation of Tanzanian target actors. The selected independent variables in this research are limited due to the short time frame of this project . For these reasons only previously mentioned hypotheses will receive thorough analysis. However, this does not consequently mean that the author assumes that other variables can not contribute to successful intermediary-target relationships too.

Indirect soft power governance mechanisms are complex and often involve various stakeholders and interactions. To influence target actors in a complex global

Hypotheses independent variables

Hypothesis 1: CSA intermediaries were able to influence Tanzanian target actors due to intermediary multiplicity and specialization

• Intermediaries > 1

• Intermediaries have specialized roles

hypothesis 2: CSA-intermediaries are able to influence Tanzanian target actors because these target actors are legitimate representatives

• intermediaries represent beliefs, values and interests of target actors

• target actors perceive

intermediaries as legitimate representatives of their beliefs, values and interests

(26)

governance field many factors either contextual or institutional, indirectly or directly, influence these relationships. The author thus acknowledges that potential other structural factors such as expertise and operational capacity could significantly contributed to the influence of intermediaries on their designated targets actors. Other contextual factors such as increased political momentum or access to finance could similarly be positive factors for intermediary-target dynamics. However, the reader should taken into account that variables as such are similarly applicable in other governance contexts. In hierarchical governance, expertise and a well functioning institutional structure are important indicators for norm-adaptation too.

In order to really devote this research to orchestration characteristics and in doing so contribute to the O-I-T model theoretically, there has been decided to select determinants which are specifically applicable to these distinct features of indirect soft power governance system as occur in the O-I-T model.

For these reasons the first selected hypotheses is therefore aimed at intermediary- multiplicity and specialization. Although it should be noted that stakeholder multiplicity and specialization are not uncommon characteristics in other governance mechanisms, Abbot and colleagues do share different views on the effectiveness of the latter. In orchestration governance, complexity is seen as a distinct power whilst other governance theorists often relate multiplicity and complexity to fragmentation and governance ineffectiveness. Abbot et al (2015a), in turn argue that these distinctive feature enable increased participation of different stakeholder which are involved in all kinds of tasks. Specialization and multiplicity should contribute and is necessary to solve complex policy themes such as migration, terrorism and global health. Due to the fact that Abbot and colleagues embrace these distinctive features, there is chosen to evaluate whether intermediary multiplicity and specialization do contribute to intermediary-target relationships

Whilst specialization and multiplicity of actors are thus not totally uncommon in global governance, the voluntary and non-hierarchical relationships between orchestrators and intermediary does differ. Representativeness is in orchestration governance thus even more important as these actors operate without electoral accountability and lack political authority to execute their tasks due to these soft and indirect features of orchestration (Abbot et al, 2015a). Other than in hierarchical governance, political mandate or sense of political control on the third-party actors is missing which could instigate conflicts of interests or other problems with regard to legitimacy. Without adequate representativeness of the values, beliefs and interests of the actors whom

(27)

they wish to influence, target actors could potentially feel threatened and this could discourage them to follow these norms. For these reasons, there has been chosen to test wether the degree of representativeness positively contributes to successful intermediary-target relationships .

(28)

5 Methodology

5.1 Research Design

This thesis will be conducted following a qualitative research approach. Qualitative research is a well-suited method for an in-depth and detailed examination of a small number of cases and examination of multiple variables at once. To examine intermediary-target dynamics in CSA governance, a single-case study will be executed. Examination of one single case (N=1) is similarly useful to retrieve underlying contextual factors and give an extensive understanding of complex issues such as prevalent in the stakeholder-environment of climate-smart agriculture governance Bryman, 2012).

Within this single-case model, the case of Tanzania is adopted as an extreme-case model, meaning that the value of Y, in this case integration of CSA governance lies many standard deviations away from the mean of Y (Seawright & Gerring, 2008). Extreme case models offer a great opportunity for exploration of new fields and trends and are therefore beneficial for the central aim of this research: to examine under what conditions, CSA intermediaries influence Tanzanian target actors to climate smart agriculture governance behaviour.

5.2 Case selection

As qualitative case-selection derives from theoretical prominence rather than random sampling, case-selection in qualitative research requires scientific justification (Seawright & Gerring, 2008). Within this research, the author will empirically explore the relationship between intermediary actors and target actors in orchestration governance, hereby using the climate-smart agriculture paradigm in Tanzania as an empirical single-case study.

The choice for the empirical field of Climate Smart Agriculture movement derives from the absence of previous academic research of orchestration in global (sustainable) agriculture policies. In contrast to other policy fields such as global health and transnational environmentalism, limited to none empirical research has focused on orchestration of global sustainable agriculture (Chan & Amling, 2019; Hanrieder, 2015). The CSA paradigm is relatively new and therefore offers great opportunity for empirical exploration.

(29)

The choice for Tanzania as an extreme case model derives from the relatively high integration and adoption of CSA policies and programs in government and local level. Like most sub-Saharan African countries, agriculture is the main source of national income and therefore a highly important sector to ensure economic growth, poverty alleviation, and food security (World Bank, 2017a). Yet, the country ensures growth stagnation due to the negative consequences of climate change such as soil degradation, erosion and drought which poses threats for food security. Tanzania is currently ranked 96 out of 113 countries in the Global Food Security Index (Global Food Security Index, 2019). Measures to counteract food security such as the increased use of fertilizer and increased land use for agricultural purposes in turn further contribute to climate mitigation. To overcome issues as such national governments, NGO agencies, civil society organizations (CSO’s) have taken strong measures to integrate CSA policies in Tanzania. Examples of highly adopted measures are integrated soil fertility management, rainwater harvesting techniques and the use of drought-resistant crops (Worldbank, 2017). Furthermore the Tanzanian national government has taken a prominent stance for CSA adoption in national policies. After deliberate effort from third party organizations, Tanzania was one of the first countries to initiate a national CSA program in 2015. This consequently led to the establishment of the national CSA policy guidelines in 2017 (MAT, 2015; MAT, 2017).

This clear progressive stance towards CSA is also reflected in the CSA policy index . In 2017 the World Bank conducted cross-sectional country research on CSA policies in over 90 low- and middle-income countries. Based on readiness, coordination, services and infrastructure Tanzania ranked 16 with 76%. In doing so Tanzania had the highest integration of CSA practices of all African low-income countries (World Bank, 2017; Relief Web, 2017).

(30)

Figure 2: CSA Policy Index 4

5.3 Data collection

This research will be based on qualitative data whereby a mixed-method approach will be integrated. Within the span of this research, content analysis and semi-structured interviews will be conducted. By gathering data from different sources and applying triangulation, the reliability and validity of the thesis consequently increase. In the research both data-sources serve different purposes. Content analysis is a suited method as it enables transparency, replication and follow-up study (Bryman, 2012). To indicate whether the hypotheses are correct, this analysis will be based following the method of process-tracing which specifically focuses on analyzing processes which explain the causal relation between a certain set of variables. Process tracing is useful to explore causal relationships in deductive research as it provides specific information on contextual settings, procedures and other type of causal mechanisms (Bennet et al, 2010). In this case process tracing will be useful to analyze the interactions between intermediaries and targets, as well as their respective institutional structures. Content analysis is derived based on analysis of primary data documents, ranging from national policies, policy briefs and strategies, reports from NGO’s and research institutions. Similarly secondary data is used to create an in depth-understanding of the political and scientific field of CSA Moreover, several websites, blogs and news articles from respective stakeholders will be used to ascertain an objective and

Retrieved via World Bank (2017).

http://documents.worldbank.org/curated/en/725141507028323885/pdf/120163-4

WP-P132359-PUBLIC-ADD-AUTHORS-CSAPolicyNoteWeb.pdf

(31)

depth image of all the relevant stakeholder-interactions as well as to analyze their designated results regarding CSA integration. In the process of data-collection approximately 20 primary data document were analyzed. Other data derived from websites and blogs are similarly reflected in the bibliography

However, using sole content analysis through process-tracing could pose barriers for results as some underlying relationships between intermediaries and targets are potentially hard to retrieve by sole document analysis (Bryman, 2012). Especially whilst measuring the perceptions of representativeness, direct input from target actors is necessary to obtain valid results. For these reasons this research has conducted 3 semi-structured interviews with 2 national experts from the Tanzanian Climate-Smart Agriculture Alliance whose names in accordance with the interviewees are published. Besides the respective agents from the Tanzanian Climate-smart agriculture alliance, another interview has been executed with a respective member of a local NGO member. This respective respondent, chose to remain anonymous. In the analysis the respondent will therefore be referred to as interviewee 3. The framework for the semi-structured interviews is added in the appendix I. interviews are recored in accordance with the interviewees, names and functions are similarly known by the author to guarantee transparency.

5.4 Scientific limitations and measurement concerns

The choice of certain research designs, methods and approaches subsequently bring scientific limitations. Especially in qualitative research designs, scientific errors occur due to the interpretation of the author who classifies and structures the data itself (Hancké, 2009). For these reasons, qualitative design studies have certain limitations which should be taken into serious consideration. The following paragraph therefore presents some scientific limitations, as well as proposed solutions and consequences regarding scientific validity, reliability, replicability and generalizability of the research. Measurement concerns, regarding data, will similarly be taken into account in the final subsections.

Validity

Validity within scientific research is achieved, once the variables or concepts selected by the researcher, actually capture the causal dynamics you are measuring or are attempting to explain (Hancké, 2009). In this case, questions concerning validity can