Scaling service delivery in a failed state: cocoa smallholders, Farmer Field Schools, persistent bureaucrats and institutional work in Côte d’Ivoire

ABSTRACT The increased use of sustainability standards in the international trade in cocoa challenges companies to find effective modes of service delivery to large numbers of small-scale farmers. A case study of the Sustainable Tree Crops Program targeting the small-scale cocoa producers in Côte d’Ivoire supplying international commodity markets shifts attention from mechanisms of private governance to the embedding of service delivery in the institutional dynamics of the state. It demonstrates that, despite a recent history of violent conflict and civil unrest, the introduced Farmer Field Schools programme achieved a surprising scale in terms of numbers and geographical spread. The analysis of this outcome combines political science and anthropological studies of effective and developmental elements in the state with the interest in institutional work found in organization science. The scaling of a new form of service delivery is explained by the skilful practices of institutional work by managers of a public–private partnership. They have been professionally associated with the sector for a long time and had the capacity to embed new forms of service delivery in persistent pockets of bureaucratic effectiveness in a failed state.


Introduction
At the beginning of the new millennium, the cocoa sector in Côte d'Ivoire, its cocoa organizations and the country itself could not have had a worse reputation. In 2001, reports of a 'child slave ship' in trouble off Africa's west coast (Lamb, 2001) sparked allegations of child slavery on cocoa plantations, causing an international outcry. By 2002, the country had entered another period of violent conflict and economic crisis, and in the following years, funds destined for cocoa farmers were used to fuel the fighting (Klaas, 2008). In 2005, the country ranked highest on the 'Failed States Index' (FFP, 2014) and political elites maintained their status quo in the cocoa sector (Woods, 2007). Côte d'Ivoire was generally considered, under the rule of President Gbagbo, to be a failed state with clearly predatory traits. This context seems not very conducive for endeavours to implement sustainability standards for cocoa and the related new forms of service delivery to large numbers of small-scale cocoa farmers. Nevertheless, a public-private partnership (PPP) programme targeting cocoa farmers was able to organize a growing number of Farmer Field Schools (FFS), and consequentially reach a substantial number of smallholder cocoa farmers and even transform a topdown extension approach into a participatory discovery learning approach. Alignment of the PPP with the Ivorian extension system and the mobilization of unprecedented national and international private sector support appear as crucial preconditions for this outcome. This reflects a puzzling outcome when seen in the light of the discussion on failed states: what generated this scalable form of service delivery in a context that appears unconducive?
For explaining the observed transformation and scaling of service delivery to small-scale cocoa farmers in Côte d'Ivoire, this paper investigates the sequence of activities set in motion by the career professionals managing the Sustainable Tree Crops Programme (STCP), a regional PPP programme in West Africa (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011). STCP introduced the FFS method and connected it with the ongoing reform process of the public agricultural extension infrastructure in the country. The analysis of the specific case of scaling FFS in Côte d'Ivoire aims at tracing pathways through which project or company-based sustainability initiatives merge with the continuously reproduced and gradually transformed modes of operation of specialized groups of career professionals within the state bureaucracy.
The case study central to this paper addresses the question of how the practices and strategies of career professionals, both within and close to the bureaucracy, generated the scaling of a novel form of delivering extension services. Instead of looking at the state as an unwieldy vehicle, the paper focuses on functional units within the state. Analysing this connection made between the practices of PPP management and the institutions in the cocoa sector may offer clues to how effective and scalable transition strategies towards agricultural and rural sustainability are brought about.
Understanding how FFS, a new bottom-up form of service delivery and knowledge exchange, reaches large numbers of smallholder cocoa farmers is relevant for various sustainability initiatives. Today, the private sector has a leading role in these initiatives, often in partnerships with NGOs and international development agencies. Apart from addressing the issue of child labour in cocoa farming, such partnerships usually concentrate on training producers and ensuring compliance with sustainability standards (Bitzer, Glasbergen, & Leroy, 2012). However, in strategic sectors such as the cocoa economy in Côte d'Ivoire or Ghana, the public sector will be part of these sustainability initiatives, anchored in the private sector and civil society (Vellema, Laven, Ton, & Muilerman, 2016). Hence, it is important to explore in what ways the public sector accommodates or adjusts to these private initiatives and what this implies for the embedding of global standards in locally established modes of partnering (Vellema & van Wijk, 2015).
The FFS approach promoted by STCP in Côte d'Ivoire was not new, but it was innovative for cocoa in West Africa, offering a nine-month-long training not on a food crop but on a perennial crop. A successful small-scale initial experience in Indonesia led to the co-designing and piloting of the cocoa FFS approach in Central and West Africa in collaboration with the national research and extension institutions. In contrast to the classic transfer-of-technology approaches, cocoa FFS showed measurable impact on farmers' productive capacity (Gockowski, Asamoah, David, Gyamfi, & Kumi, 2010). FFS entails a shift from abstract, academic, specialized and slow-to-change knowledge production to more localized multidisciplinary knowledge production in a context of application (Gibbons et al., 1994;Nowotny, Scott, & Gibbons, 2001. FFS is focused on locally adapted participatory, flexible knowledge production by groups of smallholder farmers, and can be translated differently by actors manoeuvring in the context of institutional change (Sherwood, Schut, & Leeuwis, 2012).
The paper starts by explaining the design of the case study and our conceptual lens. Next, it describes the context of the introduction of STCP FFS and presents the outcomes in terms of achieved scale of novel modes of service delivery (e.g. Nowotny, Scott, & Gibbons, 2003;Wigboldus & Leeuwis, 2013). The analysis detects four plausible processes generating the scaling of FFS in the specific context of Côte d'Ivoire. By offering a more contextual assessment of the outcomes of induced modes of service delivery to small-scale farmers, the paper adds to the emerging area of research on the African state and the dynamics of its bureaucracies. The discussion builds on Booth et al. (2015) who argue that not compliance with 'good governance' principles, but rather a deep understanding of the functioning of regimes explains differences in development performance. A combination of the concepts of 'institutional work' found in organization science (Lawrence, Hardy, & Phillips, 2002;Lawrence, Suddaby, & Leca, 2009) and 'pockets of effectiveness' used in political science as well as anthropological studies of effective and developmental elements in the state (Bierschenk, 2010;Bierschenk & de Sardan, 2014;Roll, 2011Roll, , 2014 is proposed to unravel the situated processes of diffusion and scaling.

Data collection
The case study central to this paper documents and analyses the sequence of events and organizational actions contributing to the scaling of FFS in Côte d'Ivoire. The STCP partnership was considered to have pioneered the cocoa FFS approach in West Africa. The implementing partners themselves, including the International Institute of Tropical Agriculture (IITA), recognized the need to learn from the experience (STCP, 2009, p. 23).
Main data sources were project documents, internal discussion papers, Memoranda of Understanding and contracts between STCP and partner organizations, compiled from the archives. From this broad set, 24 key documents (mainly annual reports, newsletters, programme overview papers, pilot and transition reports and progress reports) were selected for in-depth analysis of the scaling process. Key statistics came from the partnership's M&E and training databases, and from financial administration.
For further insight, additional qualitative data were collected through in-depth, semi-structured (group) interviews in June 2010, one year before STCP became defunct. Interviews purposely targeted specific high-level officials in government institutions Agence Nationale d'Appui au Développement Rural (ANADER) and Fond Interprofessionel pour la Recherche et le Conseil Agricoles (FIRCA); the ANADER staff operating within STCP; partner cooperative and NGO staff; STCP management staff at the national and regional levels; and a local FFS project expert. Field research at the cooperative level in Côte d'Ivoire was limited to the Abengourou area owing to the continued socio-political crisis, unrest and insecurity.
Short follow-up interviews were held with partnership staff over the course of 2010 and 2011. Interviews were recorded and transcribed and an initial synthesis was shared with the STCP's country manager for validation and for guidance on additional sources to corroborate data.

Data analysis
The data were analysed through inductive content and discourse analysis focussing on interactions, courses of events and the evolving interactions among partners. This was done after STCP ended and all of its IITA-affiliated management moved on in their careers. Figure 1 presents the conceptual framework used for analysing the management interactions within the PPP, and interactions of the proposed strategic intervention (FFS) with specific groups of state bureaucrats, functioning in an unconducive institutional and socio-political context, aiming to achieve effective public service delivery.

Limitations of the study
The principal researcher was externally recruited, but undeniably linked with STCP's key partner IITA. This may have affected the perceptions and expectations of interviewees. Also, due to the retrospective nature of the study andfor the interviewsa dependence on informants' memory, some details and interpretations related to decision-making are likely to be missing or skewed by hindsight. At the same time, this is inevitable when the study aims to look through the eyes of the key actors to understand its outcome. The research documented on specific practices of career professional managing interactions within the partnership and the state bureaucracy, reported in reviewed documents and observed by respondents, in order to understand a specific outcome: the number and spread of FFS. The research did not assess key respondents' backgrounds or their actual practised norms and values (including levels of personal and fiscal integrity).
Resultsa case study of scaling extension models and service delivery in Côte d'Ivoire Public programmes on service delivery in cocoa From around 1999, governments in West Africa, their partners in development and the global chocolate industry committed to developing a healthier and more responsible cocoa value chain (Shapiro & Rosenquist, 2004). The realization that increases in yields could play a significant role in the fight against poverty was reinforced by international consumer, media (Hawksley, 2001) and political pressure, especially on the eradication of (the worst forms of) child labour and by upcoming certification schemes focused on social and environmental sustainability. A key constraint to achieving success in this area was the near absence of agricultural extension and the ineffectiveness of service delivery. An STCP baseline (Kouadjo, Keho, Mosso, & Toutou, 2002) showed that in 2001, only 19% of Ivorian cocoa farmers had been in contact with a government extension agent within a period of three months.
In an attempt to address the weak cocoa extension infrastructure in West Africa, STCP was launched in 2000. Conceived as a PPP, STCP was predominantly funded by the global chocolate industry, represented by the World Cocoa Foundation (WCF) and the United States Agency for International Development; it was hosted by IITA and operationally supported by the national governments of Cameroon, Côte d'Ivoire, Ghana, Nigeria and Liberia. Over the course of a couple of years, the programme adapted, tested and validated the FFS method for training cocoa farmers on integrated crop and pest management (Asare & David, 2011;David & Cobbah, 2008).
A country manager and technical field staff were responsible for the programme's direction and embedding in each country. Bitzer et al. (2012) describe how, by 2010, STCP had turned into a regional hub coordinating many of the partnerships active in West Africa. It served as a platform for technology dissemination and research and concentrated resources and expertise on farmers' training. Through access to its country-level infrastructure and institutional network, STCP took up a broker's function among researchers, local communities, donors and companies. The expectation was that the programme's brokerage would sustain the mode of service provision and create leverage, resulting in more partnerships being established and more resources invested (STCP, 2010a, pp. 10-13).
The case study of the scaling of STCP FFS in Côte d'Ivoire first describes the context wherein this took place. It then presents the outreach and scale of the STCP programme and later of the service provision by the public agricultural extension agency, ANADER, and introduces the different groups within the cocoa bureaucracy. Next, it maps the connections between the intervention programme and the particular context in Côte d'Ivoire.

Context: state reform and conflict
After independence, Côte d'Ivoire experienced a period of stability and economic success aided by an extensive bureaucracy based on a network of political patronage (Bonjean & Chambas, 2001), the increasingly important export of cash crops, mainly cocoa, and foreign investments (Boone, 2007;Romani, 2003). After the 1960s, for nearly 30 years, cocoa production developed with a central role for the Caisse de Stabilisation et de Soutien des Produits Primaires Agricoles (CAISTAB), with roles including external commercialization and the assurance of price stability for cocoa producers.
In the 1980s and 1990s, this 'Ivorian miracle' was affected by fluctuations in the international commodity markets and the imposed structural adjustment programmes deregulating the cocoa sector (Gilbert, 2009;Grossman-Greene & Bayer, 2009;Ruf, 2009). Boone (2007) shows that the break-up of state control over the cocoa export circuit created spatially dispersed points of corruption, private accumulation and even political mobilization all along the coffeecocoa marketing network across southern Côte d'Ivoire. Towards the end of the 1990s, the political situation became unstable. This had already started in 1993 after the death of Houphouët Boigny, the long-reigning president, and culminated in protracted political turmoil with an ethno-nationalistic character. Finally, a coup d'état in 1999 created deep ethno-political divisions in society (Daddieh, 2001;Mitchell, 2012). A series of manipulated elections, widespread protests and a failed attempt at a coup in 2002 resulted in a civil war with ties to conflicts in Guinea, Liberia and Sierra Leone and split the country along ethno-regional lines in 2003 (Chauveau & Richards, 2008;ICG, 2010). When, in 2007, a peace was brokered, the agreement did not address the root causes of the conflict (Ayangafac, 2007). These tensions resurfaced during the tumultuous 2011 presidential elections, in which Laurent Gbagbo refused to admit defeat and finally was arrested and sent to the International Criminal Court in The Hague.
In 1999, the World Bank withdrew its funding for the national agricultural research and extension institutions Centre National de Research Agronomique (CNRA) and ANADER (Stads & Doumbia, 2010). A myriad of newly emerging or provisional regulatory agencies and service delivery organizations in the cocoa sector created competition distortions and added administrative charges (Bonjean & Chambas, 2001;Gilbert, 2009). Transnational companies, supported by fiscal incentives, increasingly influenced the dynamics in the sector at the expense of national exporters. The sector was ill-equipped to deal with hard-bargaining commodity buyers; cocoa farmers were additionally levied through the system installed by the central Ivorian government to collect taxes (Gilbert, 2009;Grossman-Greene & Bayer, 2009;Ruf, 2009). The public sector and politicians were notorious for their lack of transparency and for misappropriation (Dévérin, 2006;GlobalWitness, 2007;ICG, 2009). Ayangafac (2007) emphasizes that the crisis needs to be understood as a struggle for control over the institutions that distribute resources, most notably, cocoa. These political developments shaped the conditions for service delivery and thus for the STCP programme.

Outreach and scale of service delivery
In this context of Côte d'Ivoire, with political turmoil, violent conflict and a failed state with clearly predatory traits, it seemed unlikely that STCP would achieve its targeted outcomes in terms of the numbers of FFS organized and farmers reach and simultaneously achieve an exit strategy that would materialize in public and private actors taking over the new mode of service delivery. Ten years later, contrary to this initial expectation, ANADER, Côte d'Ivoire's public agricultural extension organization, reported that 879 FFS had been organized in the 2011/2012 cocoa season (ANADER, 2013) directly after STCP stepped back, using national and international private sector support ( Figure 2). By 2010, the programme reached more smallholders in Côte d'Ivoire with FFS than in three other West African countries combined ( Figure 3). If we take a realistic average number of about 30 participants for a cocoa FFS in Côte d'Ivoire (Muilerman & David, 2011), this reported figure would mean that approximately 26,000 people graduated from cocoa FFS in 2011/2012, or roughly 3-4% of the estimated 600-900,000 cocoa smallholders (Ingram et al., 2014).
STCP temporarily halted activities on the ground after another episode of socio-political instability and violence in early 2011, though work continued into 2012 (STCP, 2011a). However, during this time, ANADER led the organization of FFS, which took off in a great flight under the national 'Quantity, Quality and Growth Programme' (2QC) (STCP, 2011c). The private sector and international donors increasingly aligned with 2QC in collaboration with the fully overhauled national cocoa institutions now managed by the Comité de Gestion de la Filière Café Cacao's (CGFCC) successor, the Comité Café Cacao (CCC) in which the former STCP country manager today fills a key position (STCP, 2011b).
Besides ANADER, the public agency for extension, the private sector contributed significantly to the diffusion of FFS. According to the PPP's annual reports, initially, the private sector's initiatives under the PPP focused on the training of FFS facilitators coming from cocoa cooperatives (63%) or communities (33%) and did not focus on those from ANADER (2%) or NGOs (2%) (STCP, 2010b). Eventually, it was mainly through independent national funding awarded to the PPP that the number of ANADER agents trained on cocoa FFS tripled to around 200, thus allowing for collaboration at scale (STCP, 2010b).
From the interviews, it emerges that especially towards the end of the PPP programme, ANADER increasingly moved into the driver's seat, despite the fact that these semi-privatized bureaucrats were part of what can be called a dysfunctional state apparatus. By 2010, the FFS had taken off on a large scale and also continued to grow after 2011 when STCP became defunct in Côte d'Ivoire. The private sector kept up its collaboration with ANADER and built on new initiatives such as the WCF's Cocoa Livelihoods Program (CLP) (Figure 2). This outcome for FFS in Côte d'Ivoire is even more striking if we consider the challenging socio-political and institutional context under the reign of President Laurent Gbagbo (2000Gbagbo ( -2011. During the electoral crisis of 2011, even ANADER's headquarters in Abidjan were pillaged, reported ANADER-seconded STCP staff. The work in 2012, 'happened in an appeased socio-political context which favoured a good working climate' (ANADER, 2013, p. 4, author's translation), although Ruf, Bourgeois, Kiendré, and Bebo (2014) point to a lack of capacity within ANADER and CLP after the departure of IITA in 2011.
Since the end of the STCP programme in 2011, Côte d'Ivoire's regulatory cocoa organizations have become more proactive. In 2014, the Ivorian government and the WCF announced an unprecedented strategy to accelerate public-private collaboration. According to Barry Parkin, WCF Chairman and Chief Sustainability Officer at Mars Inc.:  This agreement represents one of the most significant steps the sector is taking to make cocoa sustainable. (WCF, 2014) In the following sections, the paper first introduces the central elements of the cocoa bureaucracy that still existed despite the context of conflict and state reform. Subsequently, it traces the different processes that defined the interactions of groups of bureaucrats within the STCP programme, in order to better understand service delivery and the scaling of FFS.

Persistent elements of the cocoa bureaucracy
Four cocoa-related bureaucratic organizations played a principal role in the process through which the PPP firmly fixed its approach to service delivery in the organizational infrastructure in Côte d'Ivoire until 2011: ANADER, CNRA, FIRCA and CGFCC.

ANADER and CNRA
Cocoa extension (training, advice and research) in Côte d'Ivoire has been the mandate of ANADER and the CNRA. Both were launched in the 1990s during the reform processes controlled by the International Monetary Fund (IMF) and World Bank. ANADER is the coordinating agency for agricultural extension and functions as an intermediary between the national agricultural research centres (mainly CNRA) and the farmers. Throughout the country, a large network of agents and agronomists stayed in villages and provided professional services to farmers, including the introduction of innovations and formation of farmers' groups. (Romani, 2003) showed how, around 2000, ANADER's extension focus and impact were mainly on food crops and less on coffee and cocoa.
ANADER and CNRA could not work in large parts of the country in the first decade of the new millennium. CNRA was heavily affected by the crisis, and today, its capacity is greatly reduced in comparison to when it benefitted from World Bank-funded programmes, or with the research budget of its largest competitor, Ghana. Today, CNRA is funded directly by the private sector, through the producer organizations of different agricultural value chains, managed by FIRCA (Stads & Doumbia, 2010). It is essentially a federation of over a dozen associations from industry including producers and processors (Byerlee, 2011) and a financing body for agricultural research, extension provision and capacity building of agricultural producers and their professional organizations. Not only the state but also all producers participate financially through professional savings schemes per agricultural value chain.
This system is not fully in place for all commodities and most notably, not for cocoa where funds are provided through levies on exports (Byerlee, 2011). From the interviews, it becomes clear that the cocoa operators and associations are represented in the advisory body Comité National des Sages (CNS) and in FIRCA's General Assembly with representation according to the importance of the agricultural product and the funds contributed by that particular value chain. The associations have a majority vote (73%) in both the General Assembly and the Executive Board. The other members are representatives from agro-industry and academia, with a remaining 5% made up of government officials. At least 75% of the professional savings mobilized by a specific value chain will be used to that particular value chain. The rest is principally dedicated to solidarity with other underdeveloped and difficult-to-auto-fund value chains (mainly of food crops). A percentage is reserved for FIRCA's administration (Stads, 2011;Stads & Doumbia, 2010).
Despite its tendency to under-invest in agricultural research, largely due to its non-exclusive and non-rival nature (Byerlee, 2011), the private sector has taken over the role of funding agricultural capacity building, extension provision and research. It should be noted that involvement by the private sector has always been more 'intimate' in Côte d'Ivoire than in other cocoa economies . FIRCA's professional savings schemes and innovative funding of commercial research are unique in Africa (Byerlee, 2011;Ragasa & Byerlee, 2013;Stads, 2011;Stads & Doumbia, 2010).

CGFCC
In September 2008, President Gbagbo announced another overhaul of the management of the cocoa sector with the formation of the provisional CGFCC (Grossman-Greene & Bayer, 2009). This move, perhaps surprisingly, lessened direct political control over the cocoa sector (Gbagbo, 2008b). By the end of 2011, all the cocoa institutions established since 1999 were united under the permanent CCC, the successor to the CGFCC.

Embedding FFS and service delivery in the cocoa bureaucracy
This study primarily focuses on the embedding of new forms of service delivery for an explanation of the observed scaling, rather than explaining the achieved scale from the nature of the FFS approach itself. The scaling of the service delivery introduced by the PPP co-evolved with institutional reforms in the cocoa sector in a troubled socio-political context (Table 1). Deliberately making connections with the bureaucratic career professionals in the state appeared to be an important characteristic of the work of those managing the PPP. Our conceptual lens and method of inductive analysis and discourse analysis of the documents and interviews helped to identify four emerging purposive actions, which elucidate how STCP management aligned interventions with the remaining bureaucratic autonomy in the cocoa sector. The section that follows traces these actions. It starts at the farmers' level and then looks for linkages between the PPP and bureaucrats at the levels above that of the farmers. These examples of purposive and situational actions by the PPP unpack the observed scaling 'against the odds' amidst a turbulent stream of dramatic socio-political developments within the Ivorian institutional cocoa landscape.

Flexibly modifying extension modalities
The first action reveals flexibility and the willingness on the part of the PPP to step away from dogmatic thinking about how higher and lower level extension staff could link with farmers and their organizations. The PPP, with its aim of scaling and institutionalizing a specific approach (FFS) to extension, had to navigate cautiously between its technical partnership with the national public extension services and the private sector-driven demand to reach as many cocoa producers as fast as possible.
The first entry point in 2003 was working with cooperatives through an initial partnership with the NGO Canadian Cooperation Society for International Development (SOCODEVI), which evacuated that same year. Unorganized groups of farmers were also targeted but only within private sector collaborations. At any rate, according to the PPP's country manager, the original idea was certainly not to introduce separate cooperative extension services.
We are lucky to have a fund that is funded by the farmers' organizations, so why should I give another burden to a farmer to fund extension? (INT04, Abidjan, 07-06-2010) Although the PPP primarily focused on ANADER agents, it also experimented with an approach, innovative for Côte d'Ivoire, of engaging farmer-facilitators or 'cocoa relay farmers' (Paysan Relais or PR). Within FFS, the competent PR functioned as a linchpin between the ANADER agents and the cooperative farmers and assured a constant presence and the continued training of peers. While the incorporation of the PR in itself was perceived as a success by STCP management, it proved a formidable hurdle in the relationship with the bureaucratic partners and in going to scale.
On the one hand, ANADER initially respected the idea of the PR in the pilot. It observed that the farmers' response to FFS was positive and participation markedly higher than under traditional approaches. The PPP's management firmly believed that ANADER's rural development agents, or Agents de Développement Rural (ADR), could never reach enough producers without village-level support, also during the scale-up. At the cooperative level, judging from the interviews, appreciation and social recompense were high for the PR, but these quickly felt overburdened and under-remunerated.
On the other hand, ANADER opposed the 'innovative' PR system from the start, arguing in the interview that there would be no sustainable funding available for it. The PPP's stance that FIRCA might be approached only confirmed ANADER's fears that national extension funding could be reallocated, reported ANADER/STCP staff. The PR, better integrated, connected and less hindered by bureaucratic ego, did have a number of benefits as compared with the generalist ADR, and their good performance did indeed undermine the ADR's position. A good proportion of ADR were less trusted and less participatory; they often refused or were unable to speak anything other than French, thus increasing dependence on the PR. Some cooperative members stated: This leads to a peculiar situation, confirmed by a regional ANADER director: The ADR and PR work side by side because ANADER can never reach all the farmers. The national discourse is different from the regional and local. (INT09, Abengourou, 09-06-2010)  1999 Total liberalization of cocoa sector, release cocoa price, liquidation of CAISTAB. World Bank pulls out of long-term programmes, drastically reducing funds ANADER, CNRA. Transnational companies enter Ivorian markets and strengthen positions over the next 5 years.
Military coup against Bédié led by Robert Guei. President Bédié flees to France.

2000
Nouveau CAISTAB restructured and dissolved into ARCC, BCC, FRC, FDPCC and GFGCCC Guei proclaims himself president, but flees the country after popular uprising against perceived election rigging. Gbagbo proclaimed winner and president. Ouattara, excluded, calls for new election.

Fighting erupts between Gbagbo in (Christian) South and Ouattara in (Muslim) North. Gbagbo wins parliamentary elections. 2001
Harkin-Engel protocol on eradication of child labour rocks cocoa sector. Private sector declares no trust in bureaucracy but STCP decides to work with them anyway.
Attempted coup fails Reports of child slave ship off Africa's west coast. Allegations of child slavery in cocoa cause international outcry and force the government to take action. FIRCA supported the PR system only for a short while, which led to these comments by FIRCA management: [The PR] start to look more and more like our agents. Especially since we give them work and means, we are deviating a little. They start to show the same behaviour as the technician.
[…] We have found the weaknesses. Now in the second contract [with STCP] we are correcting them. (INT06, Abidjan, 08/06/2010) By doing so, FIRCA effectively reoriented the PPP's role to backstopping ANADER. The PPP could have held on to its beliefs but decided to agree to take a backseat in the process. In the last two years of its existence, the PPP focused on quality control by its seconded ANADER staff, increasingly getting less support for research-driven modes of data collection, or as FIRCA representative stated: We don't need useless theoretical calculations. (INT06, Abidjan, 08/06/2010) The country manager of the PPP disagreed: We need to let the cocoa industry understand that, so that there will not be any conflict any more. That means [the industry partners] go directly to ANADER and make sure that ANADER accepts that we ensure the quality control. It is the best way for us, it is the best way also that our efforts are going to be sustainable and institutionalized. (INT04, Abidjan (07/06/2010) Eventually the PPP, at the expense of seeking full compliance with the 'ideal type' FFS, chose to be accommodating, not dogmatic, in order to allow the bureaucratic cocoa career professional to effectively align their strategy and mode of working, thereby, to some extent, retrogressing to classic top-down approaches (Ruf et al., 2014).

Resisting short-termism and engaging with national bureaucracy
Secondly, the actions and strategies of the PPP management showed resistance to a private sector reflex to concentrate on immediate results and short-term concerns. Although this created some initial tensions in the partnership, it opened opportunities for engaging with groups of bureaucratic career professionals.
The PPP programme selected the STCP's country manager from the ranks of a national think-tank on public expenditure. He was deeply embedded in and knowledgeable about bureaucratic architecture and procedures. Several respondents point to his previous involvement in the analysis of the World Bank and IMF-led reforms of the agricultural sector in the 1990s (STCP, 2003a). This appears to have been a decisive reason for his opposition to the belief that the failing government institutions should be discounted as implementing partners. He argued that a longterm perspective was absolutely necessary and was in favour of intimate collaboration with the professionals within the national bureaucracy. This course was eventually agreed upon with the private sector partners. Moreover, in a somewhat desolate NGO landscape due to the crisis, no operator other than the semi-privatized ANADER and its 6000 agents could potentially reach substantial numbers of cocoa producersa key objective for the private sector partners.
The national organizations co-invested in the partnership (STCP, 2006a, p. 11). ANADER seconded some senior specialized cocoa agents, or Techniciens Spécialisés (TS), at no cost even before the start of STCP, and with CNRA contributing resource persons. One of the TS took up the role of Participatory Extension Specialist (PES) and became integrated into the PPP management team, acting as a linchpin by being active, diplomatic and communicative. This helped the PPP to connect proactively with other agencies. All of this was considered a major task, as explained by the country manager: It is not the ANADER facilitators who facilitated the adoption [of FFS] by [ANADER] Headquarters, it was the STCP management; meetings, workshops we invited them to see. We did our job. (INT04, Abidjan, 07/06/2010) An additional approach, reports the country manager, was the cautious but firm refusal to involve regular extension agents during the first phase. With the full integration of high-level subject matter specialists (TS), instead of the rural development agents (ADR), the management of the PPP had two key objectives: (i) to ensure the proper testing, fine-tuning and validation of the FFS approach and (ii) to create a group of high-level extension workers ready to go back to ANADER to train other extension workers. The initial group grew into eleven 'master trainers'; half of them returned to ANADER after the pilot, explained the PES, after which ADR were also trained. This conscious focus on high-level technical agents in the agricultural extension bureaucracy was linked to a deep belief in the career professionals and their capacity to construct new practices and rules, rather than in the capacity of the organization per se, according to the country manager: Because in this country we havethe public institutions are very heavyvery competent human resources but sometimes the administrative procedures can prevent them from having the expected impact. (INT01, Abidjan, 07/06/2010) ANADER reports that it embraced the FFS programme because it matched their objective to move away from the superseded 'Training & Visit' approach. The success of this move was demonstrated by increasingly louder criticism from the ANADER directorate that STCP was getting all the credit for work performed by ANADER agents.

Anticipating the direction of institutional reform
A third feature of the work and strategy of the PPP is an accurate anticipation of the direction of the institutional reform set in motion during the liberalization processes. The programme manager was determined to link programmatically to a new but not yet functional national cocoa sector-led funding mechanism. Large multinational agro-dealers, donors, as well as the government of Côte d'Ivoire were integrated into the PPP right from the start. However, management of the PPP focused mainly on interfacing with a nascent configuration for the national funding of agricultural professional service delivery to rural operators. In particular, the country manager and the ANADER-seconded PES gave their support to the not yet functional FIRCA, effectively taking a gamble on developments, years down the line: [FIRCA] did not 'come from heaven'. I was aware. It was already a process since 1999. I was part of the team that … . The FIRCA director clarifies: This will also mean that ANADER will no longer get money automatically for cocoa; they will have to respond to a tender just like anyone else. (INT06, ANADER, 08/06/2010) FIRCA evolved in the middle of a political storm, especially until 2008. During this period, the cocoa sector was provisionally managed by CGFCC, an interim cocoa and coffee marketing board, dethroning duplicitous older institutions (Gbagbo, 2008b;Yapo, 2010), and CNS, an interim consultative body to ensure proper representation of the agricultural sector operators and guarantee the moral and stability of the coffee and cocoa sectors (Gbagbo, 2008a).The government had limited influence over technical decisions made in CGFCC and CNS. However, CGFCC still allocated some government funds directly to FIRCAcontrary to the situation with other crops where sector-led saving mechanisms emerged.
The management of PPP made an effort to get the sector's buy-in, particularly of FIRCA. It set up a PPP steering committee and national forum with representatives from all key actors. This committee advised the PPP and had considerable influence over the programme's direction. Close relations with the Ivorian government were cultivated, also with President Gbagbo. The President was introduced to the PPP as early as 2003, and the management of STCP appeared in joint television interviews with ANADER and CNRA in 2005(STCP, 2003b, 2005. The tight partnership with ANADER's strategic direction and the non-salaried affiliation of its agents were played out during an organized high-level field visit with officials from FIRCA, World Bank and ANADER to showcase the FFS approach. This political alignment encouraged FIRCA to embrace the FFS approach, which opened the opportunity for STCP to be funded as a service provider, not only by the old funding mechanism, for example, Fund for the Development and Promotion of activities of Coffee and Cocoa Producers (FDPCC) (STCP, 2004(STCP, , 2006b, but also by the new funding mechanisms that involved support from the sector operators. These were early signs of the high potential for scaling that encouraged the (international) private sector in 2005 to support another phase of the regional PPP (STCP, 2006b). The experience also materialized in a new national CGFCC cocoa programme in 2008. The new role for the PPP became to backstop FFS programmes involving ANADER agents. In 2009, a World Bank team wrote that any cocoa programme in Côte d'Ivoire should include the STCP's approach and achievements (STCP, 2010b). In 2010, FIRCA staff, confident of the sustainability of the mechanism, suggested that STCP had become increasingly superfluous.
Keeping the private sector on board and connecting them with the bureaucracy Finally, STCP management made considerate efforts to keep the private sector on board. Participation in the PPP offered the private sector guidance and opportunity for experimentation on how to connect with groups in the bureaucracy, explained the STCP manager. Specific projects in collaboration with the private sector were set up, and much work focused on integrating bureaucratic career professionals from different institutions into FFS activities. Working with a functional merit-based cell of bureaucratic professionals did much to 'teach' the private sector how to successfully engage with the cocoa bureaucracy, especially after the reform.
The PPP particularly included multinational agrodealers that entered the national market around 2000. In 1999, the World Bank and IMF induced liberalization and institutional reform of the entire cocoa sector. This policy gave more influence to private operators at the expense of government bodies and the political elites.
Initially not much confidence existed between the private and public sectors. Within the PPP, this was exemplifiedaccording to the country managerby a high-level MARS Inc. representative in 2003 plainly pointing out that, to their knowledge, no organization existed in Côte d'Ivoire to properly implement any extension activity at a large scale. There was a strong call to work instead with NGOs. The PPP's management resisted this pressure and continued to focus on ANADER and a national scaling mechanism as the best outcome of the reform process. Even though political resistance caused long delays, the PPP increasingly adopted a focus on building effective alliances between private sector operators and the bureaucratic career professionals.
The work of the partnership altered relationships between the private and public sectors. By 2010, for example, MARS Inc. had changed its beliefs, as can be read on its corporate blog (Morgan, 2010): [we have] been determined to work in collaboration with the leadership of the Ivorian cocoa sector, because meaningful change in the sector […] requires the support of the cocoa institutions that are already in placeespecially the research structure CNRA and the extension programme ANADER.
In this blog, the company also announced its intensified and groundbreaking collaboration with government: While we have been engaged in a number of industry efforts in West Africa involving the training of cocoa farmers [STCP and CLP] -I think the signing of the MOU between Mars and the Côte d'Ivoire Ministry of Agriculture sets the stage for a new and different kind of collaboration.
Companies were linked with the pilot programme of cocoa FFS that the PPP implemented in collaboration with ANADER. The PPP half-heartedly employed the services of two reputed national NGOs in private sector-funded sister projects, but the reports show limited scaling impact. Companies requested support for similar FFS programmes. Archer Daniels Midland was the first one (STCP, 2005) and soon other companies followed, all incorporating ANADER agents. This movement was amplified by the 'Harkin-Engel Protocol' (CMA, 2001) on the eradication of the worst forms of child labour in 2001, which raised pressure from consumers and caused the industry to focus on farmers' training. A representative of ANADER understood the shift in the private sector as follows: They feel the outside pressure to do something in return for the farmers. That is why they contact us. (INT05, Abidjan, 08/06/2010) FIRCA guided government agencies such as ANADER to develop their own agenda outlining priorities for the sector. In 2008, the Ivorian national cocoa programme '2QC' started, which took inspiration from the PPP's achievements. The CGFCC actively invited multinationals to discuss the alignment of programmes. Without relinquishing any control over funds and thereby effectively still bypassing the national funding mechanism, multinationals increased collaboration with ANADER, CNRA and CGFCC (later CCC); this led to the much-desired opportunities for substantial scaling of impact and cost reduction opportunities for FFS (Muilerman & David, 2011). By 2011, the PPP was increasingly left out of these negotiations and agreements. The country manager explains this as success: If they keep on coming to you, then your project is not sustainable. (INT04, Abidjan, 07/06/2010)

Discussion
As a PPP, the STCP developed a substantial regional cocoa FFS programme in four West African countries. The case study of the functioning of the partnership in Côte d'Ivoire and the resulting scaling of FFS opens a discussion on how to explain this remarkable achievement, especially because of the difficult and unpredictable circumstances in the country and the cocoa sector. FFS is generally considered an effective and attractive mode of service delivery, although it is resource-intensive and therefore difficult to scale (David, 2007). Hence, the quality of the intervention cannot be credited for the scaling outcome observed in Côte d'Ivoire. Explaining the achieved scale by the design properties of the FFS programme itself seems to contradict the substantially smaller scale realized by sister cocoa FFS programmes in neighbouring countries in West Africa with a similar design (Muilerman, Vellema, Schut, & Leeuwis, 2016).
The analysis in this paper shifts attention from the nature of FFS to the nature of managed interactions of the programme with the cocoa bureaucracy within a particular institutional context. These interactions stimulated and even catalysed the shift to a new form of extension services. The interactions between the pockets in the cocoa bureaucracy, the partnership and the intervention programme were purposefully constructed and managed. The country manager, with a long history in the cocoa sector, combined technical and political skills while navigating through a reality of corruption, patronage, inflexibility and unresponsiveness, resembling the work of the 'tecnico' described by Grindle (1977).
The situated action and the constant efforts of the country manager to embed the PPP in specific groups in the cocoa bureaucracy can be understood as 'institutional work', as conceptualized by Lawrence et al. (2009). They understand institutional work as the purposeful practices of professionals (in organizations) aimed at creating, maintaining and disrupting institutions. The insight gained from this case study is that institutional work of this kind is a pre-condition for scaling an intervention or, more specifically, a new form of service delivery.
The PPP's management succeeded to embed the FFS approach into a bureaucratic 'pocket of effectiveness' (Bierschenk, 2010;Bierschenk & de Sardan, 2014;Roll, 2011). Roll (2014) describes the concept of a pocket of effectiveness as relatively effective public service delivery within a context and environment in which this is the exception rather than the norm. This, of course, does not necessarily mean that these bureaucratic professionals are model citizens. Also, the term 'bureaucratic' needs to be understood inclusively as anyone professionally engaged in delivering public services based on certain competencies, regardless of the contractual arrangement. Hence, also STCP's management and particularly the country manager are part and parcel of the pockets of effectiveness.
Analysis of the scaling of FFS in Côte d'Ivoire shows how an intervention programme consistently aligned with a pocket of career professionals. This pocket, with a long track record in the cocoa sector, teamed up with the PPP and the private sector to make service delivery work. Crook (2010) relates the high levels of delivery of public services in Côte d'Ivoire until the late 1990s to both the existence of a clear mandate by the political elite and to the performance of bureaucratic career professionals in organizational 'islands of effectiveness', similar to Roll's concept of 'pockets'. This 'guild' of career professionalsas we could call itseems to have survived the turbulent period during which the partnership implemented the FFS programme (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011).
The evolving presence of the PPP in Côte d'Ivoire includes a selection of intentional and serendipitous generative and enabling practices through which the PPP management embedded the FFS intervention in the work and procedures of these persistent groups of bureaucrats. Following Roll (2014), alignments with these groups create the institutional conditions that explain the difference in performance and scaling among the countries where the STCP programme was active. Though this interaction did not emerge in Cameroon and Ghana (Muilerman, Vellema, Schut, et al., 2016), Nigeria showed similar productive interactions with 'pockets', but here, the federal governance system meant a slower State-by-State scaling outcome (Muilerman, Vellema, & Dietz, 2016). The deliberate making of connections with persistent cocoa bureaucrats enabled the partnership to actively navigate the violent Ivorian torrent of sociopolitical events and anchor its service delivery model in an ongoing process of institutional reform.
Studies of African states stress the dysfunctional nature of the post-colonial state and propose an almost deterministic political view on the functioning or failure of the state (e.g. Anders, 2010;Chirot, 2006). The discussion in this paper makes it relevant to relate the obvious symptoms of state failure and predatory behaviour to exploring what elements or groups in the state have a more developmental agenda. Following Evans (1995), these groups are essential for understanding emerging functional interactions between public and private sectors. The case study describes how the partnership generated conditions under which the private sector was able to engage with specific groups within the bureaucracy. The intervention programme brokered connections to persistent bureaucrats with a long-standing career and professional network in the cocoa sector. The approach attempted to use formal competence and problemsolving capacities, rather than clientelistic ties or traditional loyalties, as a requirement for entry into the network (cf. Evans, 1995).
The case study accentuates the analysis of Crook (2010) of the state in Côte d'Ivoire. He argues that public services are not overstaffed or lavishly overresourced. Instead, the public services will be able to do a good job if, in this case, a PPP constructs and legitimizes a public funding mechanism (FIRCA), builds collaborative organizational cultures and encourages staff with the capacity to make a difference at local and even regional levels.

Conclusions
The scaling of induced and novel modes of service delivery needs to be understood as an emergent outcome of the interaction between an intervention programme and the institutional realities in public and private realms. The reported scale of service delivery via FFS realized by a public partnership in the turbulent cocoa sector in Côte d'Ivoire can be explained by the institutional work done by the PPP's management connecting an intervention programme to specific groups of career professionals within the state's bureaucracy. These purposeful actions silenced, to some extent, those mechanisms obstructing effective service delivery which were also present in the violent and conflictual cocoa sector and society in Côte d'Ivoire.
These insights propose to strengthen the growing interest in research tracing the precise processes of embedding interventions in both the public and private sectors, without perceiving the state and business as communicating vessels. Analyses and assessments of endeavours aimed to enhance the sustainability of international commodity chains, such as the introduction of standards and the accompanying training and service delivery models, need to look beyond the intrinsic qualities of these interventions for explaining under what conditions scale is achieved. The paper proposes further development of the interest in the competences and capacities of partnership managers, bureaucrats and business leaders to construct effective spaces for problem-solving. This agenda implies the support of natural experiments and evolving collaborations that target the making of institutions situated in unconducive contexts.

Disclosure statement
No potential conflict of interest was reported by the authors.