The domestic political economy of upgrading in global value chains: how politics shapes pathways for upgrading in Rwanda’s coffee sector*

Abstract The Global Value Chains/Global Production Networks (GVC/GPNs) literatures have become the predominant international political economy frameworks to understand the challenge of economic upgrading under 21st century globalization. However, until recently, this literature has overlooked the role of the state (outside its regulatory responsibilities) and the explanatory power of domestic political economy. Meanwhile, literature on developmental states, industrial policy and political settlements has generally taken a methodologically nationalist perspective to examine economic transformation in developing countries. This article uses insights from the political settlements literature to contribute to the growing agenda within the GVC/GPNs literature to examine how the role of the state and domestic politics shape upgrading pathways in developing countries. Using the example of the Rwandan government’s attempts to increase specialty coffee exports over the last two decades, the article shows how the public governance of the domestic value chain, combined with governance dynamics in the coffee GVC/GPN, has shaped upgrading pathways in Rwanda’s coffee sector. By developing a domestic political economy approach within the GVC/GPN tradition, this article contributes to the growing attention within international political economy to focus on how multi-scalar pressures are shaping the outcomes of economic policy in developing countries.


Introduction
The Rwandan farmer only receives a small fraction of value of the coffee that is exported. We are trying to change that … For us, in Rwanda, the coffee sector's future depends on increasing value-addition. 1 Since the 1990s, Global Value Chains (GVC) and Global Production Networks (GPN) scholarship has dominated the study of 'economic upgrading'countries and firms moving to higher-value activities in GVC/GPNs with improved skills, knowledge and technology. These influential frameworks have been used to illustrate examples of upgrading across several sectors in developing countries (Pipkin & Fuentes, 2017). As the GVC/GPN literature blossomed in its examination of the global market engagement of firms across national boundaries, it was regularly criticized for underestimating the role of the state (Cramer, 1999) and retaining an 'overly depoliticized upgrading narrative' (Vicol, Neilson, Hartatri, & Cooper, 2018, 26). Recent GVC/GPNs work has contributed to address this gap (Smith, 2015;Horner, 2017;Mayer & Phillips, 2017;Alford & Phillips, 2018).
The developmental state (DS) literature, which explained the successful catch-up development experiences of Northeast Asian states, was criticized for its state-centricity within initial GVC/GPN literature (Gereffi, Korzeniewicz, & Korzeniewicz, 1994). Similar to the GVC/GPNs literature, however, the DS literature has been light in its discussions of domestic political economy. Within the DS literature, the study of state-business relations restricted itself to highlighting the importance of 'mutuality' or 'reciprocity' in state-business relations rather than analyzing power relations (Amsden, 2001). The political settlements frameworkinitially developed by Mushtaq Khan (2010) digs deeper into domestic political economies, focusing on how politics shapes economic outcomes. Yet the political settlements framework has been criticized for failing to adopt multi-scalar levels of analysis (Hickey, Sen, & Bukenya, 2015).
Combining insights from these two literatures can help fill a gap within the literature on international political economywhere the interaction between domestic political economies and global economic change is a core analytical focus. The aim of this article is to develop a domestic political economy approach within the GVC/GPN tradition to show how the politics of public governance shapes the organization of domestic value chains in service of upgrading strategies. The article examines the case of Rwanda's coffee sector over the past two decades.
The case of Rwanda is apt for this study given that the government has been lauded for its effective policymaking, developmental state-like ambitions and economic successes (Abbott, Sapsford, & Binagwaho, 2017;Booth & Golooba-Mutebi, 2012). During 1999-2014, Rwanda's annual GDP growth was 7.7% and its annual growth in GDP per capita was 5% (Diao & Mcmillan, 2018). Rwanda is a small, land-locked country, still dependent on primary commodities but trying to access high-value segments of GVC/GPNs. The Rwandan Patriotic Front (RPF) governmentwhich has ruled the country since the 1994 genocidehas been committed to reducing its dependency on low-quality coffee exports, which has been a characteristic of most of Rwanda's independent history. Government policy has succeeded in growing the reputation of Rwandan coffee in global specialty markets. The sector's transformation has been widely applauded for its success by international financial institutions (Boudreaux, 2011), prominent economists (Easterly & Reshef, 2010) and the international press (Goering, 2006;Gambino, 2018). Yet the majority of Rwanda's coffee farmers do not produce for specialty markets and do not receive the same support as farmers producing specialty coffee.
This article begins with a discussion of the evolution of the literature on the state's role in GVC/GPNs. The next section describes how an infusion of insights from political settlements can describe how domestic political economy influences the processes and outcomes of economic upgrading. The following section details the evolution of the coffee GVC/GPN, showing how emerging global dynamics within the coffee sector set the parameters for upgrading outcomes in developing countries. Then the article details the evolution of Rwanda's coffee sector and the country's political settlement, illustrating how domestic politics has shaped the organization of the domestic value-chain for specialty coffee. The following section describes what the GVC/GPN literature explains about economic upgrading processes in Rwanda's coffee sector. Next, insights from the political settlements literature are used to inform a domestic political economy approach within the GVC/ GPN tradition where the motivations for upgrading strategies are highlighted, showing how the political settlement shapes upgrading pathways The article concludes by focusing our attention on the importance of examining the multi-scalar challenges of late development in the 21st century, hoping that this article can progress discussions in that direction.
Research for this article was conducted during 16 months of fieldwork in Rwanda between 2011 and 2018. A total of 577 semi-structured interviews have been conducted in Rwanda across several sectors, 132 of which were specific to the coffee sector, with Rwandan government and military officials, donors, private sector, cooperative representatives, journalists, consultants and some individual farmers. Several respondents were interviewed consistently across this time span. Economic data provided by the government and exporters was triangulated with interviews. Historical data was collected by accessing government archives, government data, consultancy reports, firm data and existing academic literature.
The largest share of interviews within the coffee sector were conducted with Rwandan government officials and coffee exporting companies. Thus, the article provides primary evidence of how public governance of the domestic value-chain has evolved over time. A broader understanding of Rwanda's domestic political economy has been developed through studying several economic sectors, benefiting from extensive research with government and military officials and business actors. The article also benefits significantly from the extensive literature that has developed on Rwanda's rural sector. Data about Rwanda's rural sector is largely based on existing literature and complemented by the author's own findings.
2. GVC/GPNs and political settlements: weaving together parallel literatures 2.1. GVC/GPNs and the multi-scalar analysis of economic development Since the 1990s, economic globalization has contributed to the increasing fragmentation and spatial dispersion of production activities (Dicken, 2011). During that time, the GVC and GPN frameworks have evolved with the aim of moving beyond methodologically nationalist approaches of examining economic development. The GVC and GPN approaches have been similarly concerned with developing multiscalar analyses of economic development under economic globalization. However, they differ in important respectssome of which are discussed below. GPN approaches apply a broader formulation of networks to encompass more actors and relationships in their analysis than the firm-centric 'chain' common to GVC analyses. The following paragraphs provide a brief sketch of the evolution of these literatures in relation to that challenge. For detailed discussions of these approaches, see Bernstein and Campling (2006), Bair (2005) or Neilson (2014).
In the first-full length manuscript dedicated to Global Commodity Chains (GCC) analysis, a precursor of GVC/GPN frameworks, Gary Gereffi and his colleagues outline the aim of creating a framework that 'permits us to more adequately forge the macro-micro links between processes that are generally assumed to be discreetly contained with global, national and local units of analysis' (Gereffi et al., 1994, 2). To achieve the goal of multi-scalar analysis, the GVC literature (with which Gereffi would become associated) placed emphasis on examining inter-firm relations, primarily through vertical relationships between buyers and suppliers. The initial GVC (and earlier GCC work) studies examined East Asian development. Using those examples, scholarship (Gereffi, 1999) argued that increasing participation in GVCs was a necessary step for industrial upgrading. Within GVC's multi-scalar analyses, there was recognition of 'institutional context' and a regulatory role for the state. Yet, as the literature developed, the state's role (outside being a regulator) was recognized as being inadequately understood (Neilson, Pritchard, & Yeung, 2014).
The GPN approach evolved as a critique of existing GCC and GVC work (Henderson, Dicken, Hess, Coe, & Yeung, 2002), aiming to operate as 'a relational and specifically geographical approach to the study of the global space-economy' (Hess & Yeung, 2006, 1196-1197. Building on their interpretation of East Asian development, GPN scholarship (Henderson et al., 2002;Coe, Hess, Yeung, Dicken, & Henderson, 2004) argued that successful economic development depended on the effective incorporation of territories into GPNs. These territories involved relationships incorporating firms and non-firm actors, including the state. However, within a few years, Coe, Dicken, and Hess (2008) acknowledged that there had been a limited focus on the role of state within GPN literature. Yeung and Coe (2015a, 29) developed a new framework -GPN 2.0to create 'a more dynamic theory of global production networks that could better explain the emergence of different firm-specific activities, strategic network effects, and territorial outcomes in the global economy.' In GPN 2.0, state activity is one of several factors in the 'risk environment' that shapes three competitive dynamics of cost-capability ratios, market imperatives and financial discipline. However, one recent use of the GPN 2.0 framework (Neilson, Pritchard, Fold, & Dwiartama, 2018) in the Indonesian cocoa sector criticizes Yeung and Coe (2015b) for continuing to underplay the role of the state.
The GVC and GPN frameworks are beginning to acknowledge the need to refocus attention on conceptualizing the role of the state. However, as yet, the bulk of GVC and GPN analyses have been focused on how state interventions or regulation may affect the governance of or upgrading outcomes within the chain/network. This focuses our attention on how discussions of governance have developed within the GVC/GPN literature. Gereffi (1994, 97) defines GVC governance as 'authority and power relationships that determine how financial, material and human resources are allocated and flow within a chain.' Discussions of the governance of GVC/GPNs have largely taken a 'top-down' approach. Gereffi's initial work (1994) distinguished between producerdriven commodity chains and buyer-driven commodity chains. PDCCs are characteristic of capital-intensive industries in which manufacturers control verticallyorganized suppliers, as opposed to light manufacturing industries where globally dispersed subcontracting networks are managed by designers, retailers and other brand-name firms (Gereffi, 2001). Subsequent GVC/GPNs work has found that most commodity chains have become increasingly buyer-driven since the 1970s and 1980s when American retailers joined manufacturers to search for offshore suppliers of most consumer goods. This coincided with a global trend to embrace outward-orientation in developing country economies and end import substitution policies.

Governance in GVC/GPNs
Moving beyond the buyer-driven and producer-driven binaries of chain governance, Gereffi, Humphrey, and Sturgeon (2005) introduced a new five-fold typology of governance to understand how new inter-firm linkages were being governed in GVCs, influential five-fold typology of governance to highlight five types of relationships between lead firms and suppliershierarchical, captive, relational, modular and market (Gereffi et al., 2005). Gereffi et al. (2005) theoretical advancement reduced the salience of institutional context and path-dependent dynamics, which were of central importance to context-specific GPN approaches (Bair, 2008). Since then, there have been further theoretical advances in our understandings of the governance of chain dynamics (Ponte & Sturgeon, 2014).
Part of the motivations for these theoretical advancements have been to move away from the uni-directional 'buyer-driven' focus within GVC discussions to highlight how governance could be administered by actors other than lead firms within chains and extra-chain actors including governments, standard developers or civil society organizations. Thus, rather than governance being administered in a unipolar form, chains/networks are actually governed through bipolar or multipolar forms of governance (Ponte & Sturgeon, 2014). Gereffi and Lee (2016) similarly highlight different forms of value-chain governance, differentiated in relation to the actors involved: private governance (buyers), social governance (civil society organizations) and public governance (governments and multi-lateral organizations).
Though such forms of regulation acknowledge the state's role in governance, most GVC/GPN studies have been reluctant to make the state a core analytical focus. GPN work initially employed the concept of 'strategic coupling' to describe how regional economies integrated in international trade and production networks through specific actions and practices of key actors and institutions in ways that were mutually beneficial (Coe et al., 2004). Yeung (2016) applied this concept to East Asian countries, describing how firms eventually began to act independently of the state. However, this is not the case in many developing countries where the state remains a significant economic actor. Since key literature emerging out of the GVC/GPN tradition has emerged out of East Asia, there is a tendency for assumptions about the power and capacity of the East Asian stateat its current income levelsto travel to investigations of countries in other regions at very different income levels. A more context-specific analysis yields a less pessimistic view of the extent to which developing country governments can intervene in their economies to support modest but still, important upgrading initiatives.
For most GVC/GPN scholarship, innovation is understood to be the sole responsibility of firms (Kaplinsky & Morris, 2016). Yet the tide may be turning within GVC and GPN scholarship with new work emerging that develops typologies of the different roles that states play in supporting economic upgrading (Mayer & Phillips, 2017;Alford & Phillips, 2018;Horner, 2017). Horner (2017) extended discussions of the state's role within GPNs as beyond that of a facilitator. He highlights four state roles within GPNsfacilitator, regulator, producer and buyer. Within this work, he highlights the potential activist roles of the 'state as producer' (where state-owned firms are actively involved) and 'state as buyer', which go against the market-friendly interpretations of the state's role amid globalization that dominates the GVCs and GPN literature. Mayer and Phillips (2017) built on existing governance typologies (Gereffi et al., 2005) to highlight key governance functions of the statefacilitative, regulatory and distributive. Though they aimed to focus on politics, their discussions have focused on 'understanding the role of politics and states in the construction and maintenance of a GVC world' (Mayer & Phillips, 2017, p. 136). Alford and Phillips (2018) strengthened the typology further, showing how political contestation through the Western Cape farmworkers' strike actions in 2012/2013 shaped dynamics of state governance in the context of GVC/GPNs. However, other than characterizing the roles that governments perform, there has been limited attention to how the politics of public governance interacts with GVC/GPNs to shape economic upgrading processes.

Upgrading
The primary concern of the initial GCC and GVC literature was related to the possibilities of economic upgrading in developing countries. In contrast to the initial 'top-down' focus on governance, economic upgrading allowed for a focus on 'bottom-up' possibilities where firms move to higher-value activities in GVCs to increase the benefits from participating in global production (Gereffi & Fernandez-Stark, 2011). Humphrey and Schmitz (2002) suggest that there are four types of economic upgrading: process upgrading (inputs are transformed into outputs by re-organizing the production system and introducing technology), product upgrading (moving into more specific product lines), functional upgrading (acquiring new functions to increase the skill content of activities) and chain or inter-sectoral upgrading (moving into new but related industries).
Initially, within GVC literature, it was assumed that economic upgrading would automatically lead to social gains, with workers benefiting from integration into global production. However, scholarship has consistently shown this is not the case (Gibbon & Ponte, 2005;Dussel Peters, 2008;Barrientos, Gereffi, & Rossi, 2011). There have been warnings that inclusion into GVC/GPNs result in 'immiserising growth' (Kaplinsky, 2000) and 'adverse incorporation' (Hickey & du Toit, 2007) of labor and firms into global production and distribution structures. Within the coffee sector, Neilson (2008) found that benefits from economic upgrading in the Indonesia coffee sector resulted in benefits being concentrated among a small number of actors within each node of the value-chain. Specialty coffee requires process upgrading, which is both about increasing the efficiency of the production process and meeting standards (Ponte & Ewert, 2009). Consequently, better resourced actors are more likely to invest in technological capabilities necessary to access such markets (Whitfield, 2017). A recent systematic review of the effectiveness of agricultural certification in developing countries confirms the inequities associated with economic upgrading. Oya, Schaefer, and Skalidou (2018) found evidence that though certified products were sold at higher prices, this does not translate to increasing household incomes or farmer assets and wages for workers that are not in certified production are higher than those in certified production.
Upgrading to specialty coffee production was initially conceived as a way to improve the conditions of Rwandan farmers and the economy, as a whole. Segments of the GVC/GPN literature predict that even well-intentioned ambitions of the Rwandan government would not yield equal benefits to all actors across the domestic value-chain. The next section introduces the political settlements literature, arguing that more attention to the domestic political economy can help highlight the influence of politics in Rwanda's success in accessing specialty coffee markets.

So what's missing? Politics and GVC/GPNs
The literature on the state's role in GVC/GPNs makes important contributions to our understanding of public governance in an increasingly globalized world. Where domestic political economy is discussed within this literature, it is usually restricted to discussions of how it may affect governance (Alford & Phillips, 2018). There remain few discussions of domestic political economy and state-business relations, in particular. 2 Since most GVC/GPN scholarship focuses on individual sectors, it often neglects broader issues prevailing in domestic political economies. As the GVC/GPN literature begins to focus its attention on understanding the role of the state in influencing economic outcomes, there is an opportunity to engage with the politics that informs policymaking and the organization of domestic value-chains in service of upgrading attempts.
The analysis presented in this article highlights how the upgrading opportunities shaped by GVC/GPNs require the reservation of benefits for selected groups, which is inevitably a political process that is subject to contestation. Thus, the process of upgrading is often characterized by the inclusion of some (who receive benefits) with others excluded (who fail to access GVC/GPNs). Thus, at least, in the shortterm, there may be an unequal spread of benefits when states support specific firms or farms in upgrading attempts. The political settlements framework suggests that the choice of who benefits from processes of economic transformation is a product of domestic political economy. For example, if the ruling coalition tends to centralize control of the economy and distrusts 'outsiders' (as was evident in the RPF's reign in the early 2000s), opportunities would be concentrated among loyal firms. If governments have less strength in relation to existing businesses, it may need to distribute opportunities more evenly to ensure excluded groups will not fund its rivals. Thus, the political settlements framework helps highlighting why certain groups benefited from integration into value chains, how unequal outcomes were made durable and how domestic politics affect economic outcomes.
The political settlements framework, initially conceptualized by Mushtaq Khan (2010), has become an increasing popular tool for examining how politics influences economic and social outcomes in developing countries. The political settlements literature emerged as a response to the dominance of New Institutional Economics (NIE) in the 1990s. NIE focused on the importance of formal institutions (such as secure property rights). Contrastingly, Mushtaq Khan's (2010) work on political settlements emphasized that the distribution of power in developing countries was not aligned with their formal institutions. In developing countries, powerful groups operated through informal institutionsparticularly, patron-client relationsto protect their position and income sources. Within political settlements scholarship, ruling coalitions are understood to use the distribution of rents to contribute to ensuring political stability by redistributing benefits to powerful groups who could otherwise contest the institutional structure. In contrast to NIE theorists, Khan's elaboration of the political settlements framework puts the focus firmly on power and away from just formal institutions. The focus on how context-specific power relations are maintained through different combinations of formal and informal institutions is a distinctive contribution of the political settlements approach.
Over time, different variants of the political settlements framework have developed (Khan, 2010;Putzel & Di John, 2012) and the framework has been used to study varied sectors across different countries. 3 This article retains Khan's (2018, 5) broad definition of political settlements as 'a description of the distribution of power across organizations that are relevant for analyzing a specific institutional or policy problem.' The commonality to all political settlements scholarship is its analysis of how the prevailing distribution of power across groups in developing countries motivates different outcomesin relation to management of conflict, economic or social outcomes. For example, Khan and Blankenburg (2009) describe how in Northeast Asian countries, productive rent management strategies were adopted in line with the evolving political configurations in each country whereas rent management strategies were primarily employed for unproductive rent-seeking in Latin America. Moreover, the framework emphasizes the importance of adopting a historical political economy approach given the obvious but often-forgotten truism that power is rooted in history.
For political settlements scholarship, an understanding of power is key to examining how domestic political economy shapes economic processes and outcomes. Political settlements scholarship views power relatively between ruling coalitions and other rival groups who are contesting control over polities (for macro-political analysis) or the distribution of benefits from specific policies. Khan identifies three dimensions of power in political settlements theorythe horizontal distribution of power (the relative strength of elite factions outside the ruling coalition), the vertical distribution of power (the relative strength of lower-level factions within the ruling coalition) and how different competing groups finance themselves. The concept of 'holding power' -'the capability of an individual or group to engage and survive in conflicts'collectively captures these three dimensions of power (Khan, 2010). Khan (2010) developed a typology that explores how ruling coalitions are organized, the time horizons they develop and their enforcement capabilities. If ruling coalitions had more power than competing elite factions (horizontal distribution of power) then they would have a longer time horizon. If ruling coalitions had more power than lower-level factions within their coalition (vertical distribution of power), they would have higher enforcement capabilities. Khan (2010) identified four different categories of political settlements: potential developmental, vulnerable authoritarian, weak dominant and competitive clientelist. Potential developmental coalitions referred to those coalitions that enjoyed both more relative horizontal and vertical power than rival groups. Vulnerable authoritarian coalitions referred to those coalitions that enjoyed less horizontal power and more vertical power, implying that they would be oriented towards short-term goals but would have high implementation capabilities. Weak dominant coalitions were those that enjoyed more horizontal power but less vertical power, implying that they would have longer time horizons but less implementation capabilities. Competitive clientelist coalitions have less vertical and horizontal power. The implication is that these ruling coalitions have short-time horizons and weak implementation capabilities. Khan's (2010) ideal types serve to set a foundation for construing informed hypotheses. A deeper examination of domestic political economy through a context-specific study of evolving power relations at the macro-level and in specific sectors is necessary to understand how governments choose to shape economic strategies. Their choices of business partners, as part of that strategy, are an outcome of how their relationships with domestic capitalists evolve. Outside Khan's (2010) framework, most political settlements analyses have relied on a context-specific study of power relations to show how politics inform policymaking. As the political settlements literature has begun to grapple with examining how power influences the application and outcomes of policymaking, one route has been to analyze the sources of holding power, outlined as 'economic structure, ideology, violence rights and rents' (Behuria, Buur, & Gray, 2017). These sources of holding power and their relationship to policymaking and outcomes in the coffee sector will be discussed in later sections.
The political settlements framework has been widely used to analyze Rwanda's economic recovery since the genocide. As per Khan's (2010) framework, the ruling coalition could be categorized as a 'potential developmental coalition' given that it enjoys greater relative horizontal and vertical power. Scholarship (Booth & Golooba-Mutebi, 2012;Kelsall, 2013) that has been influential both in academic and in policy circles applied the political settlements framework to Rwanda to highlight the long-horizon orientation of the country's development strategy and the effectiveness of the implementation of that strategy. Particularly focusing on the centralization of rents and power within the RPF, this scholarship argued that the use of party-and military-owned enterprises has led to productive economic outcomes.
This characterization of Rwandan political economy may be accurate at the macro-level. Scholarship (Chemouni, 2018) has continued to highlight the effectiveness of Rwandan policymaking. Critics (Reyntjens, 2013;Thomson, 2013) also agree that the RPF has increasingly centralized power but with negative implications especially in relation to human rights and prospects for democracy.
Despite the effectiveness and long-time horizon that characterizes Rwanda's economic policymaking over the last two decades, there is still limited structural transformation. Khan's typology does not adequately explain why 'potential developmental coalitions' do not achieve sustained structural transformation, with the Rwandan economy achieving only modest diversification. Attention to how domestic economic policy interacts with global economic changes would enhance the utility of political settlements analyses.
The next section illustrates the evolution of coffee politics in Rwanda before highlighting how existing GVC/GPN literature can provide insights into the evolution of Post-1994 Rwanda's coffee sector.

Governance dynamics in the coffee GVC/GPN
From 1963 to 1989, most coffee-producing and consuming countries agreed on export quotas and target price zones to stabilize coffee prices and incomes, as part of the International Coffee Agreement (ICA). The United States viewed the ICA as a means to insulate producers in developing countries, particularly Latin America, from volatility (Talbot, 2002). The ICA ended in 1989 for a number of reasons including a lack of global consensus for its continuation, opposition in the US Congress and incoherence of the Brazilian government's policy (Gilbert, 1996). The collapse of the ICA contributed to a liberalized but highly volatile global market regime, with rapid reductions in the producers' shares of the final retail price (Fitter & Kaplinsky, 2001;Ponte, 2002;Bacon, 2005).
The Coffee GVC/GPN has also become concentrated among a few retailer, roaster and international traders though there are a multitude of arrangements between producers in developing countries and smaller buyers in consumer countries. Such changes emanated from the ICA's collapse, moving from a system where producing and consuming countries shared control over the international coffee trade to an increasingly 'buyer-driven', 'trader-driven' or 'roaster-driven' valuechain (Ponte, 2002;Daviron & Ponte, 2005). This occurred alongside the dismantling of national coffee boards and the liberalization of coffee marketing systems, as part of structural adjustment policies. Booming Brazilian coffee production and Vietnam's entry as a leading coffee producer contributed to a global oversupply and a shift in the bargaining power of agents across the coffee chain.
The increasing potential of specialty coffee markets altered the Coffee GVC/ GPN. In 1982, a handful of small-scale coffee roasting companies established the Specialty Coffee Association of America and their initiatives to promote the consumption of specialty coffee led to the Northern American specialty market reaching an estimated retail value of $7.8 billion by 2001 (Bacon, 2005). Largely invisible in the 1970s, by 2001, the specialty gourmet market segment represented 17% of US coffee imports by volume and 40% of retail market by value (Giovannucci, 2001). Specialty coffee -'a term that means different things to different people'covers 'all coffees that are not traditional industrial blends', either because of their high quality, limited availability, blended flavoring, packaging or consumption experience (Ponte, 2002(Ponte, , p. 1110(Ponte, -1111. Ordinary coffee is traded on the New York Board of Trade 'C' market for use in canned pre-ground coffees and for blending. In contrast, specialty coffees are negotiated by import and export operators with producer organizations at a higher and more stable prices (Elder, Zerriffi, & Le Billon, 2012). Thus, specialty markets required roasters and traders to access producer organizations directly, bypassing the 'C' market, creating promising opportunities for upgrading within developing countries.
Sustainability initiativeslike FairTrade, Rainforest Alliance and Utz Kapehhave been central to promoting the growth of specialty coffee markets, with coffee occupying 'a flagship position in value chain organization' (Grabs, 2018, p. 2). Nearly all 'specialty coffee' is sustainably certified, with numerous certification schemes emerging as prominent private governance actors within the GVC/GPN (Giovannucci & Ponte, 2005;Auld, 2014). A growing trend of 'relationship coffee' has also emerged where roasters/retailers engage in direct trading relationships with coffee producers, bypassing traditional certification and mobilizing quality tropes (Vicol et al., 2018). The variety of relationships between producers, traders and roasters within the coffee GVC/GPN may suggest the existence of 'limited governance' (Fitter & Kaplinsky, 2001) or 'multipolar governance' (Ponte & Sturgeon, 2014). However, Grabs (2018) has recently shown that roasters and retailers like Nestle and Starbucks have begun rolling out their own certification schemes, leading to sustainability initiative adapting their strategies towards roasters and retailers from one of confrontation to competition. Though there remain numerous stakeholders occupying governance roles within the coffee GVC/GPN, there is a global tendency towards buyer-driven governance.
This tendency is sustained through the continued concentration in the trader and roaster segments of the coffee GVC/GPN. In 2012, among international traders, the two largest -Neuman Kaffee Group and Ecomhandled 28% and the largest eight traders comprised more than two-thirds of global green coffee exports (Troster & Staritz, 2015). Volcafe, Neuman Kaffee Group and Ecom control nearly 50% of the world's coffee imports. Among roasters, Nestle and Jacob Douwe Egberts enjoyed more than 40% of the global market in 2013 (Troster & Staritz, 2015). Since nearly 75% of all coffee consumed in major importing countries is bought in retail stores, retailers retain significant control over the commodity chain. 4 Despite the tendency towards buyer-driven governance, small countries like Rwandawho are aiming at economic upgrading through increasing their specialty coffee exportshave room to access the range of sustainability initiatives within the sectors, as well as new markets. Several smaller roasters have begun to emerge and new markets (East Asia and Russia) have increased their shares of global coffee consumption, signaling new opportunities associated with such 'interstices' in the value-chain (Ponte, 2002). The next section examines the evolution of Rwanda's coffee sector strategy, focusing how the political settlement has shaped the organization of the domestic value chain to access specialty coffee markets. The strategic position of coffee in national politics is common in many developing countries, with over 90% of coffee produced in Asia, Africa and Latin America (Ponte, 2002). Colonial administrations and their linked companies reaped high profits from the labor of indigenous coffee growers. The coercion that forced the organization of coffee production in these countries resulted in a reorganization of social relations within rural sectors. These efforts contributed to path dependency effects in many newly independent countries. Paige (1998) documents how some coffee elites in three Central American countries -Nicaragua, Costa Rica and El Salvadortook leadership of popular revolutions to adopt neoliberalism and protect their interests. Similar stories of how social relations around coffee impacted national politics abound in African countries including Angola, Kenya and Tanzania (Sender & Smith, 1986;Throup, 1987;Cramer & Richards, 2011).
In Rwanda, coffee occupies a similar prominenceat least, in its post-independence history. Unlike Brazil, Ethiopia or Vietnamcountries that comprise a large share of global coffee production -Rwanda has never produced more than 1% of global coffee production. Thus, even in its traditionally most important export sector, Rwanda remains a 'price-taker' in relation to the global economy. For close to a century, Rwanda has depended heavily on the coffee sector for a majority of its export revenues. Coffee and other primary commodities (tea and minerals) have traditionally comprised over 90% of Rwanda's exports. Around 99% of the coffee produced in Rwanda is Arabica coffee.
The first two independent Rwandan governments, led by Gregoire Kayibanda (1962-1973) and Juvenal Habyarimana (1973-1994, used coffee and other primary commodity sectors as avenues through which rents were distributed to their allies. Consequently, rents obtained from the sector became a source of political contestation. Major fluctuations in global coffee prices preceded periods of political instability between 1962 and 1994 (Prunier, 1995).
Initially, Kayibanda used the coffee sector as a platform to grow his national political profile. Prior to independence, he acted as TRAFIPRO's (a large coffee cooperative) president. TRAFIPRO later became the state-run marketing board for agricultural products and operated as 'the economic arm of the regime' (Verwimp, 2013). Kayibanda channeled rents to his Southern-central allies and encroached on the interests of other Hutu elites, which led to opposition (Pottier, 1993). Prior to a fall in global coffee prices in the early 1970s, Rwanda's economy weakened and Northern Hutu elites took advantage of popular grievances to mount a successful coup in 1973. According to Khan's (2010) framework, the 1962-1973 era could be categorized as a '(vulnerable) authoritarian coalition' where excluded Northern elites gradually gradually developed more relative power and mounted a successful coup.
Habyarimana was installed as President and immediately abolished TRAFIPRO, replacing it with a monopsony coffee export agency -Rwandex. In the 1970s, Rwanda experienced impressive economic growth and for a period, the country showed signs of becoming a development success story (Uvin, 1998;Verwimp, 2013). Given that Habyarimana's ruling coalition enjoyed more relative power than horizontally and vertically-excluded groups, it could be classified as a potential developmental coalition, as per Khan's framework.
For most of Habyarimana's reign, Rwandex exported around 80% of domestically-produced coffee. 5 As coffee prices began to fall in the early 1980s, the government reacted to its widening trade deficit by placing more emphasis on coffee cultivation rather than prioritizing diversification. The 1978 law on coffee cultivation made the neglect of coffee trees punishable by law. When coffee prices fell again in 1985, the area of land under coffee cultivation increased and yield dropped significantly, with evidence of increasing resistance to forced coffee cultivation (Verwimp, 2013). During the late 1980s and early 1990s, the political settlement transitioned to becoming 'competitive clientelist', as per Khan's (2010) framework. Habyarimana's ruling coalition weakened in relation to the many anti-Tutsi groups that emerged once democratic space had been opened up in the late 1980s.
Both pre-1994 governments developed peasant-centered ideologies where coffee occupied a prominent role in nationalist rhetoric, with Habyarimana calling on farmers to grow coffee for the good of the nation (Verwimp, 2013). In Habyarimana's speeches, he linked the external threat of 'Tutsis invaders' to the interests of pastoralists, which he cast as contradictory to the interests of Hutu farmers (Verwimp, 2000). The akazua group of Hutu elites, close to President Habyarimana and his wife, which controlled most economic activities and influenced policymakingalso had deep interests in the coffee sector (Des Forges, 1999;Kamola, 2007). The violence that followed in the 1990s was linked to the popular grievances that had developed because of Rwanda's economic structure, particularly coffee dependency, which left the economy vulnerable to global price fluctuations.

Post-1994 organization of the domestic value chain for specialty coffee
While the pre-1994 political settlement in Rwanda was largely organized through social relations around the coffee sectorboth ideologically and economically, the post-1994 political settlement has aimed to reduce its political and economic reliance on the coffee sector. When the RPF assumed power in 1994, Rwanda's domestic coffee sector was dealing with an altered global coffee economy. The RPF government liberalized coffee exports in 1995. The government had previously retained a Coffee Stabilization Fund and fixed the producers' price for coffee. After 1994, the government withdrew from commercial activities. By 1998, producer prices had been liberalized. The government replaced the price fixing mechanism with a progressive export tax system (taxing the exporters' profit margin at a fixed ad valorem rate of 16 per cent) (IMF, 2000). The liberalization policy meant that farmers were exposed to fluctuations in world market prices. These changes have contributed to reducing Rwanda's coffee production in comparison to the 1980s (Figure 1). Though there has been some improvement since the mid-2000s, production has remained relatively stagnant since thenat around 17,000 metric tons annually (Clay, Bro, Church, Bizoza, & Ortega, 2018). The area of land under coffee cultivation and average yield is also much lower (Figure 2).
The RPF's policies have been different from those of previous governments, which prioritized increasing the volume of coffee production and area under cultivation as part of their development strategy. Where pre-1994 governments aimed to maximize revenue and control of coffee producers through increased production of low-quality coffee, the RPF government has aimed to maximize value of coffee exported, with the goal of entering specialty coffee marketsgrowing at 15% annually in the 2000s (Ponte, 2002) when the first national coffee strategy was published in 2002. In contrast to other East African countries, Rwanda has placed a much higher prioritization on exporting washed coffee for specialty markets. Initially, the government aimed at exporting only washed coffee and aiming for all Rwandan coffee to be exported to specialty markets. 6 Over time, government officials realized that those aims were   unrealistic. 7 However, government policy continues to encourage production for specialty coffee, with the goal of reducing the export of unwashed coffee. 8 Figures 3 and 4 indicate that the value of coffee exported has increased since 1994. These improvements have been partly driven by favorable coffee prices but also by Rwandan government policy. The public governance of the domestic valuechain has been organized so Rwandan coffee can access specialty markets. Efforts began with the revitalization of coffee farms in the 1990s. Donor supportfrom USAID, IFAD and the European Unionsupported improvements in cultivation practices. Government officials publicly admit that they coerced farmers into applying fertilizer and sending coffee cherries to washing stations. 9 Ensuring more coffee was washed and processed was a priority in the first National Coffee Strategy, published in 2002. There was rapid progress in expanding coffee washing station (CWS) presence across the country. In 2000, there were only two CWS in Rwanda. By 2017, there were 301 CWS.
The rapid expansion of washing stations has contributed to the creation of differentiated value-chains (ordinary, washed, specialty and relationship coffee). See Figure 5 for an illustration of the evolving coffee value-chain in Rwanda. In the traditional value-chain for ordinary coffee, farmers pick the coffee cherries during annual harvesting periods. Cherries are then de-pulped manually and the coffee parchment dried. Farmers then sell coffee parchment to local traders (or middlemen) and sometimes, also to exporters directly. Traditionally, middle-men had sold the coffee either to the marketing board or to exporting agencies. The second value-chain is the washed coffee chain. To process and wash their coffee, farmers sell cherries to washing stations. Cherries are then pulped and fermented to remove the mucilage and beans are then washed, dried and sorted. Coffee parchment is then sent to dry mills (owned by exporting companies) to remove parchment, producing export-ready green coffee. Not all washed coffee is of sufficient quality to be sold as specialty coffee. Washed coffee may also not meet other standard requirements in line with various certification standards. Yet the government, through the National Agriculture Export Board (NAEB), has consistently prioritized ensuring its producers meet the certification standards required for access to specialty markets. Evidence of this is that Rwanda has one of the highest numbers of FairTrade certified coffee producer organizations in Africa (Elder et al., 2012).
Differentiated value-chains (ordinary, washed, specialty or relationship) should not be seen as distinct given that shifts in production of one kind of coffee has effects on the production of other kinds of coffee in a specific country. Roasters, retailers and civil society organizations govern the terms of entry into specialty markets in the coffee sector. The government governs domestic production through incentivizing (or coercing) farmers and cooperatives to change their production techniques (e.g. selling coffee to washing stations and improving production) thereby influencing the macro-level strategy of prioritizing export of specialty coffee. Crucially, the government also negotiates how roasters and retailers enter domestic value-chains and access farmers and production. 10 The Rwandan government closely directs strategic upgrading initiatives. In line with strategies in other sectors like manufacturing, the government makes risky investments itself or in partnership with close allies or strategic foreign investors, hoping the success of these investments create a 'demonstration effect' for others. This was the case in tourism sector where the government's initial investments succeeded in creating a 'demonstration effect' to other investors (Behuria, 2015). Other similar investments in the manufacturing sector including C&H apparels (a Chinese apparels company), Positivo (a computer assembly company) and Volkwagen's car assembly project are yet to achieve similar progress. Domestically, trade-and-export is liberalized but remains concentrated.
Transnational actors, which retain a preference for ordinary coffee exports, access Rwanda's domestic coffee sector through exporters. As of 2017, there were over 60 coffee exporting companies (including cooperatives) operating in the country. However, six companies export nearly 70% of Rwandan coffee. These exporterssome of whom are owned by foreign trading houses or roasting companiesbuy a mix of ordinary and washed coffee. In effect, the private governance of these exporters operates against the government's public governance strategy of the domestic valuechain, which values export of washed specialty coffee above other alternatives.
The Rwandan government's aggressive strategy of navigating the coffee GVC/ GPN to access specialty coffee markets within a liberalised domestic and global coffee environment signals the continued relevance of the role of the state in economic upgrading within GVC/GPNs. The next sections describe how the government sought access to specialty markets.
5. The politics of governing Rwanda's specialty coffee sector

Insights from the GVC/GPN literature
While the global coffee GVC/GPN has been moving towards 'buyer-driven' governance, Rwanda's public governance has been driven by the need to organize the domestic value-chain in ways that would allow increasing shares of Rwandan coffee to access high-value markets. The government perceives the future of the coffee sector to be in ensuring 'Rwanda keeps most of the value of its coffee and the way to do that is to sell to specialty markets.' 11 Characterizing the government's role as a 'facilitator' would downplay its interventionist role. The other key role that the Rwandan government has played is as a 'producer' (Horner, 2017). The GVC literature warns that while lead firms may encourage producers to undertake process and product upgrading, they were reluctant to support functional upgrading (Schmitz & Knorringa, 2000). In Rwanda, process and product upgrading has been achieved by working with donors and other leading governance actors (including Starbucks through technical support and linking foreign roasters with producers). However, even here, initial investments were made by the state or state-linked firms, showing how the state's 'producer' role has been central to Rwanda's rapid progress in accessing specialty coffee markets. The state's role as 'producer' has been even more vital in promoting 'functional upgrading' or moving into new product lines (including roasting and establishing coffee shops at home and abroad).
Rwanda's drive to access specialty coffee markets highlights the importance of the state's role as regulator, active facilitator and producer in driving economic upgrading. Success has been achieved through developing partnerships with roasters/retailers (following a large share of GVC/GPN literature) and investing itself.
Prioritizing increases in washed coffee production was the first step in accessing specialty coffee markets. Success has been rapid in this regard, with government investments ensuring a steady increase in the share of washed coffee in the country, reaching 52% of total coffee production in (NAEB, 2017. In the early 2000s, the government took a lead in investing in the construction of washing stations, ensuring it would be a 'demonstration effect' for other investors. The military, the Rwanda Social Security Board and donor-supported cooperatives contributed the first investments in washing stations. The state's active facilitative role was vital to improving the image of Rwandan coffee. Government delegations attended events held by the Specialty Coffee Association of America and Japan, the East African Fine Coffees Association, and has also held an annual national barista competition since 2008. Rwanda was also the first African country to hold Cup of Excellence (CoE) events, with producers aiming to enter such events having to meet very strict standards. CoE events have been held annually in Rwanda since 2008 and Rwandan coffee has won prizes at international CoE events regularly in recent years. Participation in global events has been instrumental in marketing the high-quality image of Rwandan coffee. 12 Several cooperatives have been successful in accessing specialty coffee markets. Donor support and the government's active facilitative role were key to their success. The Abahuzamugambi cooperative, which was founded in 1999 and supported by USAID, found a market for specialty coffees through UK-based Union Coffee Roasters and US-based Community Coffee. Their coffee is bought at premiums and exported as single-source/traceable coffee, selling at retailers including Whole Foods, Intelligentsia Coffee and Third Rail Coffee in New York (Easterly & Reshef, 2010). Starbucks-supported Rwashoscco and Buf Coffeeboth exporting companies that are owned by cooperativesmade similar breakthroughs in US markets. However, even in these investments, the government has palyed a central role, encouraging leading roasters to develop links to Rwandan coffee, provide technical assistance and even market Rwandan coffee. 13 Relationship coffee has also taken hold in Rwanda through American NGO Sustainable Harvest, which partners with cooperatives of female coffee farmers to sell their high-value coffee in the United States. Another similar example is the recent announcement (in 2018) of US-based Peet's Coffee limited edition 2018 Annual blend of Rwandan coffee produced by a female-only cooperative of coffee farmers with the promise that 5% of proceeds would provide clean water to Rwandan coffee families. 14 The majority of roasters and retailers have prefferred to buy Rwandan coffee after it has been bought by international traders and roasted abroad. Thus, toll roasting arrangements have been the dominant form of economic upgrading in Rwanda. The government worked with USAID to market specialty coffee abroad, successfully convincing roasters like Specialty Coffee Association of America and the Coffee Quality Institute to get involved in Rwanda. 15 However, President Kagame himself played an active 'facilitative' role by leveraging his personal relationship with Starbucks President Howard Schulz and others in the sector like Peet's coffee and Arkansas-based Westrock coffee to convince them to source coffee directly from Rwanda. 16 The government has also engaged directly with international buyers including the Rogers Family who bought 58 containers after visiting Rwanda in 2009. Rogers Family Company plans include a partnership with RPF-allied businessmen Alfred Nkubiri's ENAS to build washing stations. 17 Since donors were reluctant and advised against investing in functional upgrading, thegovernment took the lead in such investments. After the government's initial investments in roasting coffee domestically, by 2009, there were six main coffee roasters and by 2018, this number had increased to 18. Some roasters are individual Rwandan capitalists (although their investments are small). Some are foreign investors, targeting'interstices' (Ponte, 2002) within the coffee value-chain, selling its coffee in Middle Eastern and Ugandan supermarkets. In 2012, 116.8 tons of coffee was sold as roasted coffee. 18 As a major strategic investment within the coffee sector, the governmentthrough NAEB and BRDhas worked with partners, the Clinton Hunter Development Initiative (CHDI) and the Hunter Foundation, to create a coffee companythe Rwandan Farmers Coffee Company (RFCC)and invested in a 3 million USD coffee processing factory in Kigali. RFCC sources coffee directly from government-selected cooperatives. In 2015, the RFCC began operations and started producing under the brand -Gorilla's coffeeselling to local, African, Asian and European markets. In 2015, RFCC exported 7 tons of roasted coffee to the UK and in 2017, 4.5 tons were sold to other markets including Germany and South Korea. In 2018, 10 tons of coffeeworth $77000, with the same amount of ordinary coffee valued at $42000were shipped to the USA, as part of an agreement between RFCC and US-based Global Food, where 5,452 coffee farmers from six cooperatives would share the benefits (Nkurunziza, 2018).
The state's role as a 'producer' is also evident in setting up coffee retail outlets, through RPF-owned Bourbon coffee. Bourbon Coffee was established in 2006 by Arthur Karuletwaa Rwandan who moved to America in 1995, and worked for Proctor & Gamble, and then for the Rwandan government including as a consultant at OCIR-Caf e. Bourbon initially aimed to be the 'Starbucks of Rwanda.' Bourbon has three stores in Kigali and initially had one each in Washington DC, New York, Boston and London. Since then, its American operations have expanded but the London coffee shop has shut down. The investment in Bourbon was more a 'brand-building exercise' and it is unlikely to have made profits. 19 Bourbon's investments were an example to others, with 17 new coffee shops established across Rwanda by others. Other prominent coffee shops include Mauritian-Rwandanowned Brioche and Kaizen's Neo. Rwanda'' rapid success in accessing specialty coffee markets has only included a portion of the country's coffee producers. Nearly half of Rwanda's coffee does not reach washing stations despite the government's attempt at coercing farmers and providing incentives and the distribution of benefits through washing stations.
'Change does not come so easily. We had to use a stick and tell them (the farmers) to do things. Our farmers just used to get beans, put them in a sock and take them. We wanted to do something to change it. We built washing stations, organised cooperatives. Now, everyone is convinced. This is the effect of long-term planning. ' 20 The Rwandan government, though admitting initially coercive measures to draw farmers into producing coffee for specialty markets, hoped that the success of those who gained access to specialty markets would drive more farmers to produce for specialty markets. Yet this has not been successful, with the government mentioning that the biggest constraint in the sector is access to growth in the sector is the inadequate production of coffee cherries (Clay et al., 2018).
Washing stations have become a regulatory location for governments since seedlings, fertilizer and other inputs are often distributed on receipt of coffee cherries from farmers. In 2016, the government adopted a new zoning policy, which limits where and to whom producers can sell coffee cherries. The requirements of certification standards have also forced the reorganization of production, with specialty coffee requiring partnerships with identifiable cooperatives who can invest in meeting standards and thus, it is likely that poorer farmers will be excluded (Ponte & Ewert, 2009). The GVC/GPN literature has previously highlighted how benefits of upgrading within the coffee sector are concentrated among a limited group of actors and marginalize the majority (Neilson, 2014). There continue to be inequities within cooperatives, which is common across many African countries and the developing world (Sender & Johnston, 2004;Cramer, Johnston, Oya, & Sender, 2014). Most coffee farmers were not even organized in cooperatives. Only 8.2 per cent of coffee farmers were cooperative members in 2006 and only 14% were cooperative members in 2015 (OCIR-Caf e, 2009;NAEB, 2016). 21 Many certification schemes only apply to cooperatives and require farmers to have minimum farm sizes, which automatically excludes a large share of Rwandan coffee farmers.
The GVC/GPN literature helps clarify the different roles that the state plays in economic upgrading in the coffee sector. It also explains why the Rwandan government may have chosen to play a more active producer role to promote functional upgrading. However, it does not explain why the Rwandan government was so effective in organizing the domestic value-chain or how politics informs how the partners in upgrading attempts are chosen. The next section employs insights from the political settlements framework to explain these aspects of upgrading strategies.

Rwanda's political settlement and the coffee sector
The political settlements literature is useful to explain how the Rwandan government chose to organize the domestic value-chain to support increased access to specialty coffee markets. The RPF government's decision to prioritize economic upgrading in the coffee sector was directly linked to the historical vulnerability Rwanda has experienced because of its undiversified economic structure. Academic literature on Rwanda highlights that a clear ideological goal of self-reliance has driven policy-making (Behuria, 2016a;Reyntjens, 2016). Within the coffee sector, self-reliance is inextricably bound with the goal of economic upgrading. Fluctuations in global coffee prices were followed by periods of political struggle through the Kayibanda and Habyarimana governments, with internal power struggles intensifying as access to resources shrunk (Prunier, 1995). These examples of political turmoil and their links to coffee price fluctuations were mentioned by senior RPF officials in interviews. 22 The RPFwhich has been largely ruled by a Tutsi minority elite (with an emphasis on being broad-based)was conscious of the difficult relationship it had with the majority Hutu population, particularly those in rural areas. Most of the literature on Rwanda's rural sectors have characterized this a negative attitude towards smallholder farming (Ansoms, 2009;Huggins, 2017). In a recent review article, leading scholars on Rwanda's agriculture sector conclude that the government 'intends to co-opt farmers into a state-managed system of commercialization and to render them more dependent on state services as well as private firms and banks providing hybrid seeds, fertilisers and credit' (Ansoms et al., 2018). This article takes a less critical view, arguing that the government has developed an upgrading strategy, which reflected the difficult relationship it had with rural producers and the challenges that Rwandan coffee would have in accessing specialty coffee markets. The result was an evolving upgrading strategy, which assumed that any gains would only apply to a select group in the short-term. Acknowledging that only the most efficient producers would succeed, the government hoped their progress would provide a demonstration effect to other rural producers. 23 The RPF government has been consistent in its prioritization of altering the domestic coffee value-chain in line with accessing specialty coffee markets. The strategy that has been adopted has been implemented top-down, with a commitment to macroeconomic needs and an attitude of 'the government knows best.' 24 Without the co-optation of most rural actors, the government has been forced to rely on coercive methods. Several examples of resistance have been detailed (Huggins, 2017;Ansoms, 2013). The military's involvementat the firm level and in coordinating productionis evidence of how the state has organized coffee production, with support from violence specialists, in the language of Tilly (2003). Just within the last two years, the reserve military force has overseen the planting of over 3.7 million coffee trees on 1,515 hectares in four districts (Kirehe, Rulindo, Gakenke and Nyamagabe). The military was also involvedthrough Agro-Processing Industries, now Agro-Processing Trust Corporationas an exporter of specialty coffee and owner of washing stations.
Ideology and the management of violence and coercion was useful in organizing the domestic value-chain. The government was able to employ coercive measures because domestic opposition within the country has been unable to mount 25 concerted challenges to RPF rule. However, another key concern was how benefits (or the distribution of the rents) were organized in relation to upgrading strategies. The various upgrading strategies are outlined in Table 1. Economic and political power within Rwanda is centralized within the ruling coalition, a group comprising majority Tutsi leadership who led the RPF to victory after the 1994 genocide (and some new entrants into the ruling coalition. Changes in the broader domestic political economy mirror changes in the political economy of Rwanda's business sector. Academic scholarship has tended to focus on the use of Rwanda's party-and military-owned enterprises (Booth & Golooba-Mutebi, 2012;Behuria, 2015Behuria, , 2016b. There are several different types of RPF-linked investment groups in Rwanda. Crystal Ventures Ltd. is the party-owned conglomerate that operates in various sectors including construction, coffee, private security, agro-processing and real estate. There are also military-owned investment groups (Horizon Group, Agro-Processing Trust Corporation), government-funded investment groups managed by the Ministry of Defence (Ngali Holdings) and investment groups that are products of collective investments by local investors closely tied to the RPF (Rwanda Investment Group).
In the 1990s and early 2000s, there were prominent individual investors that owned leading businesses in certain sectors. There are relatively few prominent private domestic investors. Some domestic elites operate in value-addition segments in the coffee sector but RPF-linked investment groups are often the primary partners for large strategic investments (both because of the frictions within local state-business relationships and because few loyal investors are willing to invest in large strategic investments).
As part of the Rwandan government's strategy, the key domestic capitalist partners for upgrading attempts have been the state itself. This is especially true for functional upgrading initiatives where the government found it difficult to develop partnerships with foreign roasters. As an active regulator, the government first encouraged RPF-allied businessmen to established coffee exporting companies and then actively sought foreign investors who focused on exporting to specialty markets. However, for any risky investments, including functional upgrading attempts, the government has only relied on itself, highlighting the continued tendency towards centralizing rents to achieve long-horizon objects of diversification. The government's attempts at organizing exporters has been difficult since the early 2000s. In pre-1994 Rwanda, Rwandex (a company, which was owned by the government and British investors) exported nearly 80% of Rwandan coffee. After 1994, new RPF-allied businessmen established exporting companies. However, Rwandex continued to retain a preference for exporting unwashed Rwandan coffee. Eventually, the government's focus on specialty coffee and the company's own internal difficulties contributed to them selling their assets to US-based Rwanda Trading Company, which continues to retain a preference for exporting to specialty markets.
One of the central uses of the political settlements framework is to show how underlying domestic power relations influences economic outcomes. In this sector, the choice of partners in upgrading attempts have been a function of a vulnerable political settlement where an uneasy relationship continues to exist between the ruling coalition and private businesspeople. The RPF government's strategy to centralize rents to achieve upgrading strategies is a product of the historical context through which it has evolved. Given the uneasy relationship with rural society in the 1990s, the government embarked on an upgrading strategy that inevitably evolved selective benefits for coffee producers who were co-opted into the strategy. These methods were bolstered by an ideological basis for the strategy, linked to self-reliance, and the coercive apparatus that supported the re-organization of the domestic value-chain.
For GVC/GPN analyses, the article argues that the political settlements framework can provide benefits to help us understand the politics of public governance initiatives related to the organization of domestic value-chains in support of upgrading attempts.

Conclusion
Recent literature has made important strides in our understanding of the role of state in the governance and upgrading of GVC/GPNs. This is an important area of research in international political economy, as we aim to learn more about how developing countries are grappling with the challenges of developing strategies aimed at economic upgrading in the 21 st century.
The emerging literature on the state's role in GVC/GPNs has largely concentrated on the role of states in upgrading or on how the state policies may influence changes in governance. This article focuses our attention on how domestic political economy influences upgrading strategies and outcomes. Rwanda is an example of a country where interventionist strategies have been crucial in supporting higher shares of domestic coffee being exported to specialty markets. The tiny share of Rwanda's production in comparison to global production (less than 1%) means that there is little effect on the governance of the coffee GVC/GPN. Existing GVC/ GPN approaches tell us little about how the state was able to organize domestic value-chains to achieve successful upgrading, aside from assigning roles to state actions.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
Effective States and Inclusive Development Research Centre, University of Manchester.