How instrument constituencies shape policy transfer: a case study from Ghana

Abstract The concept of instrument constituency provides students of public policy with a new analytical tool for the analysis of policy change. In this article, we use the example of cash transfer programs to show how this concept also makes a direct contribution to the analysis of transnational policy transfer. More specifically, the analysis shows how, over the last dozen years, actors forming an instrument constituency promoted the diffusion of cash transfers as a policy instrument from Latin America to sub-Saharan Africa and, more specifically, from Brazil to Ghana. This case study of Ghana’s adoption of a cash transfer program is grounded in semi-structured, expert interviews conducted with both domestic and transnational actors. Overall, the analysis demonstrates how the concept of instrument constituencies can enrich the literature on policy transfer, a key source of policy change in both developed and developing countries.


Introduction
The rise of the social investment approach to social policy occurred in tandem with the adoption of cash transfer programs in most developing countries (Ancelovici & Jenson, 2013;Jenson, 2010). Until then, such programs had been sparse or nonexistent in most of the poorer regions of the world. This changed in the 1990s, when Brazil and Mexico introduced conditional cash transfers an in attempt to reduce immediate poverty and to halt the intergenerational transmission of poverty (Fenwick, 2013). These policies spread rapidly in Latin American and, more recently, in Africa, a continent noted for the challenges it has faced in addressing pervasive poverty (Beegle, Christiaensen, Dabalen, & Gaddis, 2016;Schubert & Slater, 2006). The diffusion of cash transfer programs designed to alleviate poverty is puzzling, particularly, considering spatial, social and economic differences between the innovators and the countries adopting their innovations (Sugiyama, 2011).
This article uses this popular policy innovation to explore the factors that can facilitate transnational policy transfer, specifically how instrument constituencies transfer policy solutions to developing countries, beyond the use of external constraints imposed by international organizations. 1 This is not to say that international organizations do not play a key intermediary role, but rather to claim that exploring the transnational diffusion of cash transfers as a policy instrument from Brazil to Ghana, two economically and culturally distinct countries, allows us to see that constituencies, attached to a particular policy instrument, can promote diffusion through advocacy. Advocacy is often accompanied by a significant investment of technical and financial resources. Based on fieldwork recently conducted in Ghana and Brazil, this study suggests that instrument constituencies are an important missing link in the analysis of transnational policy transfer. Our central claim is that, as key actors with ties and experience in particular developing countries, instrument constituencies promote particular policy instruments while legitimizing and facilitating their transfer to particular jurisdictions.
The article is divided into four main sections. First, we review the literature on policy transfer and, second, show how the concept of instrument constituencies can enrich the study of transnational policy transfer. This is followed by the third main section, which examines Brazil's cash transfer program and how the activities of instrument constituencies contributed to the adoption of a similar program in Ghana. In the fourth main section, the role of instrument constituencies in the policy transfer process is further explored and implications are drawn for policy transfer and policy-making in developing countries.

Policy transfer
The multidirectional spread of policy innovations through policy transfer has become commonplace in a globalized era (on the concept of policy transfer, see Dolowitz & Marsh, 2000). 2 A case in point is the spread of Brazil's cash transfer program to other countries. Diverse perspectives available in the literature account for this type of policy transfer by pointing to factors such as cognitive biases, international discourse construction, external pressures through the imposition of vertical constraints and lesson drawing through the activities of policy networks or communities.
For instance, scholars have pointed to the role of international organizations in policy transfer. One stream of scholarship focuses on how powerful organizations such as the International Monetary Fund (IMF) and the World Bank (WB) can impose loan conditions requiring developing countries to implement specific policies to receive financing (Adejumobi, 2004;Boafo-Arthur, 1999). However, developing countries do not always adhere to external inducements (Orenstein, 2008). Countries have reneged on agreements with international organizations, in some cases citing bad timing or citizens' disapproval of those reforms (Whitfield & Fraser, 2010). Also, fundamental to this perspective on external inducements, is the use of coercion in the policy transfer process, which tends to neglect or downplay the agency of policy-makers in the policy process. Considering these limitations, this approach is not very insightful for our analysis, which focuses on actors and their agency rather than on coercion.
A different school of thought that is more helpful for our analysis asserts that international organizations and other transnational actors can influence domestic national policies 1 external constraints imposed by international organizations are generally known as 'conditionalities' , a term we do not use here so as to not create any confusion with the 'conditionalities' imposed on individual beneficiaries as part of Conditional cash transfers.
2 Some of the literature discussed below is associated with the closely related concept of policy diffusion, which is more commonly associated with quantitative research than the term policy transfer that we use in this qualitative article. the authors would like to thank Martin Papillon for his insight about this terminological issue.
through the ideas and discourses they help spread (Béland & Orenstein, 2013;Finnemore & Sikkink, 1998;Foli & Béland, 2014;Orenstein, 2008;Sugiyama, 2011). In this view, international conferences and summits, publications, and speeches by renowned world leaders help to draw attention to proposed policy solutions, including programs implemented in various countries around the world. Scholars note that these kinds of global policy spaces, characterized by the interaction between national and transnational actors, provide platforms for articulating new norms, generating consensus and spreading policy ideas (Shahjahan, 2012). For example, regarding cash transfers, in a speech titled 'Education for All ' , United Nations Secretary-General Kofi Annan (1997 recommended the consideration of some social policies to support children's education, including Brazil's Bolsa Escola program (Buarque, 2001;United Nations, 2000). As suggested below, international organizations can become active members of what are known as instrument constituencies.
Another key approach argues that policy transfer can be accounted for through a process of lesson drawing from other countries with the help of groups of policy actors (Rose, 2005; for different approach focusing on psychological processes, see Linos, 2006;Weyland, 2005aWeyland, , 2005bWeyland, , 2006. In fact, policy-makers can themselves learn from existing policies within their borders or from other countries by relying on epistemic communities (Haas, 1992) or policy networks in the process of drawing lessons to inform domestic policy reforms and or solve problems they confront (Rose, 1991). Actors in these communities and networks work across space bringing foreign experience to bear on domestic policy-making. Referred to as policy entrepreneurs, these actors are dynamic in how they share ideas and include individuals and groups who work increasingly across international boundaries. While their ideas on specific policy issues might shift from time to time, they also adopt different strategies in the process of diffusion (Béland & Orenstein, 2013;Orenstein, 2008). What we show in this paper is that a particular type of policy entrepreneur known as the instrument constituency plays a central role in policy transfer and deserves systematic attention. In the discussion and analysis that follows, we suggest that naming and turning to instrument constituencies can directly contribute to the analysis of policy transfer.

Instrument constituencies and policy transfer
As discussed at length in the Introduction of this special issue, instrument constituencies are a category of collective policy actors who are bound by an interest in a particular policy instrument or solution (see also, Béland & Howlett, 2016;Voss & Simons, 2014). These actors have invested significant amounts of both personal and professional time in developing the policy instruments they promote. For these reasons, they jointly engage in activities toward building, expanding and spreading knowledge of the policy instrument to which they are bound. To use Kingdon's (2003, p. 123) language, they are 'advocates of solutions' , policy entrepreneurs who identify with and promote specific policy instruments.
In contrast to Kingdon's (2003) policy entrepreneurs, 3 however, instrument constituencies are typically transnational in nature and are often government employees engaging in formal foreign policy activities, which allows them to promote their particular policy instrument across national borders (Béland & Howlett, 2016;Voss & Simons, 2014). As groups of actors that coalesce around a specific policy instrument, their activities go beyond 3 Beyond Kingdon (2003), other scholars have studied how policy entrepreneurs can be transnational in nature. See, for example, Ringius, (2000).
lesson drawing, as they constitute advocates for that particular instrument. In this context, instrument constituencies are distinct from both epistemic communities (Haas, 1992), which gather around a particular knowledge and problem area, and advocacy coalitions (Sabatier, 1988), which are held together by shared norms and values (Béland & Howlett, 2016). Unified around a shared policy instrument, instrument constituencies can help carry policy solutions from place to place, irrespective of whether or not shared policy problems have been clearly identified. Accordingly, they embody the instrument and in so doing spread it from place to place (Voss & Simons, 2014). Instrument constituencies generally arise in tandem with the emergence of a policy instrument that an agent, including international organizations or a national government, seeks to promote beyond national borders. This is the case in part because instrument constituencies 'lend agency to instruments and create a moment of "supply push"' , using different forums and strategies to facilitate policy change (Voss & Simons, 2014, p. 736). By working as advocates of a particular policy instrument, instrument constituencies, as embedded in 'social processes of knowledge production' (Voss & Simons, 2014, p. 749), are intricately involved in transnational policy transfer by helping to move policy solutions from a local context to a global one through discursive and support activities.
Gatherings of policy actors work together in the process of policy transfer. Diffusion is seen to be the result of the interaction between transnational actors and local policy processes (Gilardi, 2012;Voss & Simons, 2014). International gatherings such as policy conferences and workshops provide avenues for instrument constituencies to promote their preferred instrument, but it is not the only way they can spread an instrument from place to place. Interactions at the local or national level likewise provide opportunities for instrument constituencies to interact and for these constituencies to expand to include others from different contexts (Foli, 2016).
As suggested in this article, bringing instrument constituencies to the forefront in policy transfer analysis allows us to better identify and study the role these key actors play within the policy process. Their advocacy role in spreading policy instruments internationally has been given insufficient attention in the literature.
In conceptualizing instrument constituencies in this way, it is important to note, following scholars like John L. Campbell (2004) and Barbara Czarniawska (2012), that policy instruments are constantly adapted when they travel from one context to another. This is also what occurred with Ghana's adaptation of the Brazil-style cash transfer program, discussed below. From this perspective, instrument constituencies are a fundamental part not only of policy transfer itself, but also of policy development. The analysis below focuses on the role of instrument constituencies in both the transnational diffusion and national adaptation of a policy instrument. This analysis adds to the literature by moving policy instrument constituencies to the center of the discussion about policy transfer, where it belongs.

Policy transfer and the case study approach
Policy transfer involves the spread of policy innovations from one place to another, typically through policy learning. Instrument constituencies working across national boundaries are active in this process. They engage in bringing programmatic ideas such as policy instruments from elsewhere to the attention of other policy-makers through international and national platforms such as policy conferences, study tours and expert workshops. Using a case study approach, this article studies the trajectory of policy innovation in Brazil and Ghana and specifically how the cash transfer instrument constituency was involved in the process of bringing program information from Brazil to the attention of Ghanaian policy-makers. Our approach permits a detailed examination of Ghana's adoption of the Livelihood Empowerment Against Poverty (LEAP) cash transfer program. In addition to allowing an in-depth appreciation of the processes and the actors involved in Ghana's adoption of LEAP, our approach creates the opportunity to identify and assess various indicators that point to the concept of interest to this article, ICs. As is standard when using qualitative methods, we rely on an examination of histories, documents, interview transcripts and other sources to better understand our case of interest (George & Bennett, 2005).
Data generation involved face to face interviews with actors who were part of the policy development process. Altogether, 37 people were interviewed, including government officials from the Ghana Department of Social Welfare, the Ministry of Gender, Children and Social Protection, and the Ministry of Finance; selected development partners, including the WB, the United Nations Children's Fund (UNICEF), officials of the Brazilian Ministry of Social Development, the United Nations Development Program (UNDP); and civil society organizations. These actors were purposely selected for interviews, which were conducted in Ghana in 2013, and Brazil in 2017. In addition to these interviews, official publications including parliamentary debates, press releases, statements issued by government and reports on the program were reviewed to generate information on why and how Ghana came to adopt a cash transfer program in 2008 and the role Brazilian actors played in supporting its development.
Cash transfer programs involve regular transfers to targeted groups of people (mostly the extremely poor and the vulnerable), subject in some cases to certain conditions that are either conceived as 'soft' or 'hard' . These programs aim to meet the short-term income needs of beneficiaries, as well as to enhance beneficiaries' capabilities in the long term (Kakwani, Soares, & Son, 2006). While transfers are largely conditional in Latin America, African countries have more often pursued a mix of conditional and unconditional transfers because of their difficulties in providing access to required services (Schubert & Slater, 2006). In Malawi, for example, labor-constrained households receive regular stipends without conditions as well as extra cash for children enrolled in school (Pozarny, 2015). Conditionality in this case is indirect; though it is not stated explicitly, keeping children in school earns the household extra cash. This then becomes a motivation for school enrollment.
Officially, Ghana's transfer program has both conditional and unconditional aspects, depending on the category of recipient. Beneficiaries without productive capacity or support -people living with severe disability and those aged 65 and older -receive unconditional grants while households with orphans and vulnerable children receive conditional grants (interview with LEAP Coordinator, Accra, April 2013). For the latter group of beneficiaries, conditionalities include enrolment and retention of school-age children in school; birth registration of new born babies and their attendance at postnatal clinics; full vaccination of children up to the age of five; and non-trafficking of children and their non-involvement in the 'worst forms of child labour' . (Oxford Policy Management, 2013, p. 14) Yet, because such conditions are not always enforced in practice, 'LEAP is probably best described as "quasi-conditional"' (UNICEF, 2009, p. 75).
Ghana's adoption of cash transfers is interesting in two respects. First, LEAP is quite novel for the country because there had been no program like it since independence. Consequently, the design stage was characterized by contestation over the perceived motivations behind and impacts of the program. While some critics argued that the program would make recipients dependent and lazy, others saw it as a political tool to enhance the government's chances of being re-elected (interview with former Deputy Minister of the Ministry of Employment and Social Welfare, in charge of social development, Accra, June 2013). Second, the course of program adoption points to policy learning rather than imposition. In the African context, policy adoption or reforms instigated by transnational actors were mostly perceived as taking place under some form of duress linked to the conditions attached to loans. Yet in the case of LEAP, policy transfer did not occur through imposition but through the process of learning.
Growing concern about poverty culminated in the establishment of the Millennium Development Goals (MDGs) in September 2000. World leaders' decision to pursue the (MDGs), and now the Sustainable Development Goals (SDGs), is a tangible expression of their desire and commitment to poverty reduction. Consequently, solutions and innovative programs that have worked in some places are being promoted in others. From the mid-2000, cash transfer programs benefitted from the activities of a group of transnational actors who advocated their adoption in developing countries. This instrument constituency was comprised of various transnational actors, including the WB, the UNDP, the Brazilian Government's International Policy Centre for Inclusive Growth (IPC-IG), the UK's Department for International Development (DFID), UNICEF and the Brazilian Ministry of Social Development. Some of these actors (WB and UNICEF) also facilitate the main learning platform for cash transfers in Africa today, known as 'the community of practice of cash transfers and conditional cash transfers in Africa' (COP). Established in 2011, the Anglophone contingent of the COP has met five times since 2012; Ghana has been an active participant in all of its meetings. In March 2014, the COP was held in Fortaleza, Brazil. Thus, the adoption of a cash transfer program in Ghana, a result of this constituency's activities, provides an excellent case for the study of how instrument constituencies promote policy transfer while simultaneously promoting policy development.

Cash transfer programs: from Brazil to Africa
Poverty reduction continues to be one of the key challenges leaders in developing countries face, and it is at the core of the global consensus evident in the adoption of the MDGs in September 2000 and the SDGs in September 2015. This is even truer in Africa, where the WB reports that even though poverty rates have declined, the number of people who are extremely poor has increased (Beegle et al., 2016). Consequently, various efforts have been made toward reducing poverty in Africa. In this vein, efforts have been made to bring policy innovations with evidence of contributing to poverty reduction to the attention of policy-makers in Africa.
As discussed below, an instrument constituency emerged as a significant player in the process of spreading cash transfer programs from Latin America to Africa, and specifically to Ghana, which consolidated a formal partnership with Brazil in 2007. International development partners such as the World Bank, UNICEF, IPC-IG and DFID promoted technical cooperation. In this case of policy transfer, many actors were involved in this constituency, including the above-mentioned international organizations, Brazil's Ministry of Social Development (MDS); and research institutions such as the Carolina Population Centre of the University of North Carolina (CPC-UNC), the Economic Policy Research Institute (EPRI), the Institute of Statistical Social and Economic Research at the University of Ghana (ISSER) and many others. Often connected, actors from these institutes all provided research to support the constituency's advocacy activities. The role of these actors, who worked together to promote cash transfers as a policy instrument in Africa generally, and in Ghana particularly, is discussed below.
Brazil's cash transfer program as it operates today is the culmination of numerous discussions about social assistance policy beginning in the 1980s (Fenwick, 2013(Fenwick, , 2016Lindert, Linder, Hobbs, & de la Brière, 2007;Sátyro & Cunha, 2014). In 1995, the first two 'local' conditional cash transfers became operational. Bolsa Escola was implemented in the federal district of Brasilia by Governor Cristovam Buarque, who had first described the idea in a highly influential paper written in 1987 as a means to encouraging children to attend school. Simultaneously, the Programa de Garantia de Renda Mínima Familiar (Guaranteed Minimum Family Income Program) was launched in Campinas; it was framed more using a rights-based principle of basic income (Fenwick, 2009). By 2001, conditional cash transfer programs had spread rapidly to over 100 municipalities and states in Brazil (Lindert et al., 2007;Sugiyama, 2011). Following its successful local diffusion, Bolsa Escola was adopted at the federal level and was largely built based on the original local model focusing on education (interview with Senator Buarque, Brasilía 2017). In 2003, the program merged with four other cash transfer programs to become Bolsa Familia. Bolsa Familia embraces a more holistic approach and focuses less attention only on school attendance (for a detailed political history of Bolsa Familia, see Fenwick, 2016). Although the idea of unconditional cash transfers framed as a right to a basic income existed in Brazil, many policy actors insisted on the need for conditionality. In contrast to other programs, such Mexico's national CCTs Opportunidades-Progresa, conditions in Brazil were always conceived as 'soft' and understood as a way to reinforce the right of universal access to social services and not as a way to punish beneficiaries (personal communication with Brazilian Official MDS, August 2017).
Political support for cash transfers in Brazil was largely built on the idea that the government should play a role in reducing poverty and inequality by providing social services. Thus, Bolsa Familia has the following dimensions: regular income disbursements to beneficiaries conditioned on education (i.e. ensuring between 75 and 85% school attendance rates for children between 6 and 17 years of age) and on certain health standards (ensuring vaccinations required for children under 7 years of age have been administered; ensuring pre-natal care for pregnant women) with the aim of breaking the cycle of intergenerational poverty; and complementary actions to promote family development (Secretaria Nacional de Renda de Cidadania [Senarc], n.d.). As of 2015, the program has expanded to include 13.8 million families, each of which receive an average of R$164 ($47 USD) monthly (Senarc, 2015).
After 12 years of existence, Bolsa Familia is widely perceived as an example of a successful poverty alleviation program (Fenwick, 2016). This type of cash transfer program has spread rapidly among Latin American countries, as well as among other developing and developed countries. In the mid-2000s, few African countries had cash transfer programs, yet a study published less than a decade later identified 123 cash transfer programs in sub-Saharan Africa (Garcia & Moore, 2012;Schubert & Slater, 2006). Most of these programs were inspired by Brazil and Mexico's cash transfer programs, and many benefitted directly from Brazilian expertise in program design and implementation (Cirillo, Nogueria, & Soares, 2016). The case of Ghana is interesting in this regard as the design of its registry system and the program itself were directly influenced by Brazilian officials who formally visited Ghana three times in 2007 (personal communication with a Brazilian official MDS, August 2017). Prior to 2008, the only cash transfer program in existence in Ghana was a monthly pension, received mostly by retired formal sector workers. This changed in March 2008 however, when the LEAP program was launched.
LEAP is a social transfer program for the extremely poor and vulnerable in the following categories: orphaned and vulnerable children (OVCs), people with disabilities without any productive capacity and the elderly who are 65 years or older. The program aims at reducing poverty and promoting human development through enhancing household's caloric intake, access to health and other government services, and ensuring children's enrollment, attendance and retention in school. These are achieved through a targeted, bi-monthly cash transfer to beneficiaries that is conditional for OVCs and unconditional for other recipients. Households with eligible members received between GH 64 (USD $14.5) and GH 106 (USD $24) a month depending on the number of beneficiaries. Beneficiary households with OVCs must adhere to the following conditions: children from birth to age five are to be registered and to regularly visit health facilities for vaccinations and growth monitoring; children aged between 5 and 15 must be in enrolled in and attend school; and children cannot perform any type of child labor. The program is funded from the government of Ghana's general revenues, as well as through the support of development partners. For instance, DFID is providing over $45 million USD for 5 years (FY 2012-2017) so the government can increase the number of beneficiaries, and also increase program effectiveness and work toward sustainability. 4 In 2010, the WB approved a social opportunities project that seeks to support, among other things, the rollout and administration of LEAP with $20 million USD between 2010 and 2018. 5 The trajectory of LEAP's adoption and implementation points to the significant contribution of the transnational instrument constituency. In the mid-2000s, evaluations conducted on existing pro-poor programs in Ghana, in particular the government's subsidy program, indicated that the program's intended objectives were not being reached because the program lacked an effective targeting mechanism (interview with Kwame, policy official, Ministry of Employment and Social Welfare Accra, April 2013). Reform was necessary. Some members of the vulnerability and exclusion sector working group, which included government officials and development partners including DFID, UNICEF and the WB, proposed cash transfers as a solution to this ineffective subsidy system. This proposal led to the inclusion of cash transfers in the outline of social programs considered for Ghana's Growth and Poverty Reduction Strategy (GPRS II) document in 2005. So, how were instrument constituencies involved in the spread of cash transfers to countries like Ghana?
Positive evaluations of conditional cash transfer programs in Latin America, combined with the drive that accompanied the establishment of the United Nations' MDG on poverty reduction, convinced global supporters of cash transfers to diffuse this type of policy instrument to developing countries in other regions of the world. 'This support is not only rhetorical, but also practical, as considerable funding has been given to the dissemination of program experiences, expansion of existing initiatives and replication of similar programs elsewhere' (Kakwani, Soares, & Son, 2005, p. 12). For example, international organizations and other development partners not only supported Brazilian efforts at social development but also helped to spread these programs to other countries in the Global South (Leite, Pomeroy, & Suyama, 2015). In Africa, donors such as the WB and DFID acted as a 'global learning hub for successful initiatives' (Ragno et al., 2016, p. 148). Beyond providing technical and financial assistance, as in Brazil from 2003 to 2006 during the program's major expansion and implementation phase, international organizations were eager to promote Brazil's commitment to horizontal (south-south) cooperation (Ragno et al., 2016, p. 148). It was precisely because one of the key aspects supporting Brazil's emergence as a global player is its increasing tendency to engage in south-south technical policy co-operation that Brazilian policy experts become active members of this transnational instrument constituency in Africa (Burges, 2014, p. 356). In contrast to other countries in Latin America, there was a conscious decision in Brazil's foreign ministry to market its ability to reduce poverty and to export its technical experience to other southern countries, in particular to Africa (Itamaraty, n.d.). Brazil's foreign policy of promoting south-south cooperation through 'technical cooperation' , in conjunction with the decision of powerful international development agencies to promote CCTs as an instrument to reduce poverty, became a mechanism in creating an instrument constituency composed of transnational actors who have actively facilitated the adoption of CCTs in African countries, and particularly in Ghana. This situation points to the active role of a transnational instrument constituency in the diffusion of cash transfers to that country.
In January 2006, Brazil's MDS and the UK's DFID organized a study tour in Brasilía for officials from several African countries, including Ghana. During the tour, the Africa-Brazil Program on Social Development was launched (interview with former Deputy Minister of the Ministry of Employment and Social Welfare, Accra, June 2013). According to a Ghanaian official, DFID 'supported this trip to Brazil and they gave us the technical assistance, which enabled us to gather a number of experts' (interview with Kwame, policy official, Ministry of Employment and Social Welfare, Accra, April 2013). These efforts at promoting lesson drawing and, ultimately, policy transfer, were formalized in 2007 when Brazil's MDS, in association with the IPC-IG, and the UK's DFID, signed formal terms of reference with Ghana's Ministry of Manpower (MMYE) to provide technical assistance in designing the country's cash transfer program. This transnational cooperation involved three missions of Brazilian officials and consultants who traveled to Ghana during July, August and September 2007. These missions produced three reports for the Ghanaian government on structuring its national registry, creating the conditions to provide access to social services, and recommendations for a monitoring and evaluation system for LEAP. Most of the ideas expressed in these reports were based on Brazil's experience (Brazil, 2007a, p. 3), although in their report on 'monitoring conditions' , Brazilian officials suggested that Ghana follow the Mexican program Opportunidades' design of using community members to monitor program conditions (Brazil, 2007b, p. 8). The design of LEAP was completed in November 2007 and launched in March 2008. Ghana was the first country to benefit from the technical expertise provided by this new international agreement. For instance, Sultan and Schrofer (2008, p. 305) noted that 'the Ministry of Manpower used DFID funding to design LEAP with the help of experts who had worked on similar schemes in Brazil, Zambia and South Africa' . A government official confirmed this and indicated that transnational actors provided them with the assistance to recruit experts, 'some supported by UNICEF, some by DFID, to form a team that actually initiated the process of developing the LEAP' (interview with the LEAP Coordinator, Accra, April 2013). Additionally, Sonya Sultan, from DFID and a leading expert on cash transfers, provided significant technical advice in the policy design process. Other experts included Qaiser Khan from the World Bank and Juliana Lindsay from UNICEF (interview with policy official, Ministry of Employment and Social Welfare, Accra, April 2013), as well as Michael Samson from EPRI in South Africa. Samson was influential in advising about the difficulties of enforcing conditionalities (personal communication with Brazilian official MDS, September 2017).
Besides multi-and bilateral development cooperation that serves as a means of promoting specific policy ideas and instruments, another avenue that created a platform for the program's advocacy and knowledge sharing is the Transfer Project. The Transfer Project is a research-based program/platform instituted by DFID, UNICEF and Save the Children UK to provide 'technical assistance in design, implementation and analysis' of national cash transfer programs in Africa (The Transfer Project, n.d.). As an internet-based community of practice, the Transfer Project builds on member experiences and disseminates research information on cash transfer programs in Africa and beyond. As a member of this project, Ghana shares program experiences as well as learns from others on the platform. Other institutions such as the Carolina Population Centre of the University of North Carolina (CPC-UNC) and the Institute of Statistical Social and Economic Research at the University of Ghana (ISSER), conducted research and provided evidence in support of program design and reforms.
Following the advice of these and other members of the transnational instrument constituency, and based on their own experiences, Ghanaian policy-makers adapted the program to the Ghanaian context, in particular the fact that conditions are 'expected' , but not monitored (see The Transfer Project, n.d.). Also, the program has both conditional and unconditional beneficiary categories, and it not universal (Personal communication, Yayra Atta, Centre for Social Policy Studies, Accra, April 2013). Moreover, instead of penalizing caregivers of OVCs for not adhering to conditionalities by halting payments, which indirectly affects children in their care, beneficiaries are summoned before traditional authorities in a bid to name and shame. According to an official, we appeal to beneficiary households once you have the money send the child to school, if you don't send the child to school, we will write a letter to you, if you continue not to, we will inform the local leaders, the chiefs' they can name and shame that household, then to penalize the child. (interview with LEAP Coordinator, Accra, April 2013) In spite of program adaptation, it is noteworthy that Brazil's CCT inspired various aspects of the LEAP program. For instance, the use of conditions such as births must be registered, beneficiaries must register for free health care through the National Health Insurance Scheme and children must go to school. There are also linkages to government services that seek to enhance the productive capacity of beneficiaries. Similarly, the country adopted the concept of a single registry from Brazil's Cadastro Unico and, in October 2015, launched the Ghana National Household Registry (gnhr.gov.gh). Additionally, there is a preference for women beneficiaries, who are believed to efficiently manage the resources given to the benefit of the whole household (interview with the LEAP Coordinator, Accra, April 2013).

How instrument constituencies shape policy transfer
As suggested in the first section of this article, the influence of transnational actors and factors in shaping national social policies in the contemporary era is widely recognized (Deacon, 2007;Kaasch, 2015;Orenstein, 2008;Yeates, 2014). One route by which national policies are shaped is the diffusion of policy instruments by transnational actors. When these actors become involved in an instrument constituency, they embrace and seek to diffuse an instrument by advocating for it across borders. The following are constituency activities or roles as they relate to the diffusion of policy instruments: producing knowledge on instruments and instrument performance, framing policy discourses at international gatherings, facilitating lesson drawing, and providing resource support to countries willing to pilot a program representative of a certain type of policy instrument.
Actors that form instrument constituencies commit substantial amounts of resources to research and development. In the case of conditional cash transfers, studies conducted by organizations such as UNESCO (Brazil), the World Bank, DFID (UK), IPC-IG (Brazilian Government/UNDP) and UNICEF all helped spread the word about cash transfers as a policy instrument. The knowledge gained through such activities subsequently informed various publications about the impact of cash transfers on poverty reduction and improving health and education. The DFID's Social Transfers and Chronic Poverty: Emerging Evidence and Challenges Ahead (2005) and Using Social Transfers to Improve Human Development (2006) both point out the benefits of cash transfers in the process of poverty reduction. The evidence produced in the process is fundamental to policy prescriptions and suggestions for countries that have yet to implement cash transfers as a policy instrument. Importantly, in the process of building support for the adoption of LEAP, publications proved to be critical. As the minister in charge of social protection at the time mentioned during an interview, several publications, including by the DFID, the World Bank and other international organizations, provided critical information on cash transfers and the Latin American experience. This was used to support the argument that the program would 'contribute to economic growth and [was] not a disincentive' (interview with former Deputy Minister of Manpower Youth and Employment, Accra, June 2013). Similarly, in the formulation of the program, information IPC-IG generated both from its work in Brazil and in Ghana was fundamental to the organization's role as an advisory partner in helping mediate the relationship between the two countries (Leite et al., 2015).
Additionally, the COP and the Transfer Project, as well as other research centers located in institutions of higher learning, such as the CPC-UNC and ISSER, provided evidence to support cash transfer instrument diffusion, design and implementation. Recently, research by these two institutions, commissioned by the government of Ghana and 3iE, has supplied key information supporting increased stipends and the regularization of cash delivery to enhance predictability (Handa & Park, 2012;Handa et al., 2013). The problem of irregular payments is repeatedly flagged in formal documents as the major challenge confronting cash transfers in Africa. These studies produced evidence for reform. This example shows how instrument constituencies can serve as an intellectual hub supporting program transfer with special knowledge obtained through research. Accordingly, knowledge gained about a new policy instrument 'can then be used by other actors in governmental or party political debates' to facilitate the diffusion of such an instrument (Stone, 2000, p. 48).
Research evidence showing the positive impact of cash transfer programs proved instrumental in legitimizing this policy instrument in countries that have adopted it, as did the positive impacts it has had, as identified through research. Concerning Mexico's Progresa and Brazil's Bolsa Escola, Sugiyama (2011) notes that studies on the impact of the programs 'raised the visibility of the … program[s] and gave the program[s] greater credibility' . Knowledge generated through research informs the framing of policy discourses at conferences, seminars and workshops to give legitimacy to novel programs or policies, and to promote program implementation in other countries.
In the case of CCTs, based on evidence of the program's impact on the lives of beneficiaries, officials from Brazil's Ministry of Social Development and Fight against Hunger (MDS), as well as other international organizations, such as the DFID, attended international and regional conferences to share program experience and outcomes to promote cash transfers as a policy instrument. Numerous international and regional meetings on social development provided a platform for experts from implementing countries to share their experiences. For example, early international conferences the WB organized on CCTs were held in Mexico (2002), Brazil (2004) andTurkey (2006). These conferences directly promoted CCTs as a policy instrument (Sugiyama, 2011). As well, the African Union, 2006 intergovernmental regional conference on social protection (Zambia) and the 2008 African Union Conference of Ministers in Charge of Social Development (Namibia) offered avenues for knowledge sharing (Foli, 2016;Wright & Noble, 2010). By the end of the 2006 Zambia conference, which Brazilian MDS experts Bruno Câmara and Maria José Freitas attended, delegates unanimous supported cash transfers and called for Greater cooperation between African and other countries in the sharing and exchange of information, as well as experiences and action on social protection and cash transfers; Social transfer programmes, including the social pension and social transfers to vulnerable children, older persons and people with disabilities and households to be a more utilized policy option in African countries. (African Union, 2006, p. 5) Finally, from 2012 to 2016, the WB and UNICEF sponsored COP annual meetings, which were accompanied by bimonthly videoconferences.
Officials interviewed indicated that members of the constituency sponsored their attendance at the conferences and seminars where participants gained knowledge about cash transfers, both within the country and abroad. As stated, A number of seminars that the World Bank was organizing, series of them annually exposed the ministry to what is actually happening, and they did not do this alone, UNICEF also came in to support, likewise DFID also came in to assist the process. World Bank staff of the social protection unit -organized programs that the ministry participated in. (interview with LEAP Coordinator, Accra, April 2013) That said, Ghanaian policy actors who participated in the most recent (May 2016) COP in Arusha, Tanzania, recognize that tensions exist between the World Bank and UNICEF over their views on conditionality (interview with a participant, May 2016).
In facilitating the transnational diffusion of the policy instrument they support, instrument constituency actors encourage policy learning, development and diffusion through study tours. Study tours bring officials of implementing countries together with others seeking new policy alternatives. They also constitute an indirect form of development assistance in that they are often financed by foreign ministries, as was the case with the Brazilian missions to Ghana. These exchanges are a major promoter of south-south technical cooperation, as well as regional cooperation within Africa. Officials interviewed noted that trips to implementing countries allowed them to learn first-hand how cash transfers as a policy instrument was performing in those countries (interview with LEAP Coordinator, Accra, April 2013; interview with Coordinator of the National Social Protection Strategy, Accra, April 2013). These tours favored the diffusion of that policy instrument to Ghana since it helped national actors gain unique knowledge about how to move forward with an instrument many transnational actors had already spoken about in positive terms.
It is clear, therefore, that the advocacy for a particular policy instrument operates in tandem with the provision of varied types of additional support, including technical, financial and human resources. Recognizing that low-income countries might not have the necessary resources to support a new program, members of a particular instrument constituency complement their lesson drawing efforts with implementation support on the ground in the countries they target for instrument transfer. For instance, in 2008 the Africa-Brazil Cooperation Programme on Social Development, promoted by DFID to share Brazilian social protection experiences with African countries, provides ongoing technical expertise to aid Ghana in the maintenance and development of its own social protection initiatives by also encouraging the exchange of regional experiences with other African countries. This cooperation includes social policy areas and instrument constituencies beyond cash transfers and has largely been successful in establishing the exchange of 'know how' among countries in the African Union and with Brazil.
Ghana's particular experience with LEAP also contributes to the instrument constituency's debates over the role and impact of requiring conditions in cash transfer programs. LEAP's effect on schooling is similar to cash transfer programs in sub-Saharan Africa, all known in practice to enforce 'soft' conditions. More generally, unconditional cash transfers in Africa have a strong impact on schooling equivalent to the 'harder' conditional programs in Latin America (Handa et al., 2013, p. 30). That said, many actors within the instrument constituency are quite opposed to the idea of conditions being enforced as they often end up penalizing those who are most vulnerable. Particularly in rural and precarious urban areas, many children do not have access to the social services that are required by conditional cash transfer programs.
It is important to highlight the multidirectionality of the knowledge that exists within these instrument constituencies and also the fact that both the actors and knowledge evolves over time. For example, in 2012, Ghanaian officials visited Colombia for the first time to gain new insights into reducing poverty and to study the effect of removing fuel subsidies. 6 In a subsequent study tour to Brazil in 2014, the Ghanaian delegation included members of both the opposition and official governing party for the first time (Ragno et al., 2016). Therefore, like the cash transfers program themselves, instrument constituencies are also dynamic, contributing to the constant improvement and maturation of the policy instrument that is being promoted. 6 in 2013, when the government of Ghana finally removed fuel subsidies, UniceF argued for the expansion of the cct program. a publication in a national newspaper by Sarah Hague, the chief Social Policy analyst of UniceF (Ghana), stated that the agency and its partners 'support the scaling up of the program using the savings made from the removal of fuel subsidies' (Hague, 2013).

Conclusion
The above discussion indicates that instrument constituencies are critical in the process of policy transfer and in promoting policy development. As collective actors whose activities are defined by their support for a particular policy instrument, they have an interest in the realization of the objectives of the instruments they support. As a diverse group who develop expertise around a particular policy instrument, they lend weight to an instrument and help to spread support for it transnationally (Voss & Simons, 2014). The concept of instrument constituencies also provides an answer to the 'paradox of solutions chasing problems' encountered in the policy development process (Béland & Howlett, 2016).
The above discussion on the cooperation between Brazil to Ghana in supporting Ghana's LEAP has shown that the instrument constituencies that arose in support of cash transfers are active in facilitating both the spread and success of these programs. Various lesson drawing platforms provided practitioners from Africa and Latin America with opportunities to learn about policy developments, and to discuss the approaches various countries have taken to address common problems such as targeting, effective delivery, monitoring and enforcement, issues that exist in all implementing countries and are constantly in need of perfecting.
Overall, our analysis shows that constituencies develop around a particular policy instrument and that they can participate in its transnational diffusion through direct interaction with national and local policy-makers in potential target countries. As they embrace and adapt the policy instrument in question, these national actors can then become part of the transnational instrument constituency, both promoting and developing best practices, an issue that future research on policy transfer could look into. At the same time, future research could look into how instrument constituencies can collaborate with domestic actors to help both promote and adapt their chosen policy instrument. We hope that this article stimulates research into these questions and into the potential role of instrument constituencies in policy transfer, both in developing and in advanced industrial countries. Daniel Béland is a professor and holder of the Canada Research Chair in Public Policy (Tier 1) at the Johnson-Shoyama Graduate School of Public Policy (University of Saskatchewan campus). A specialist of comparative fiscal and social policy, he has published more than 15 books and 120 peer-reviewed journal articles.