Globalising the study of diffusion: multiple sources and the East African Community

A growing literature on di ﬀ usion shows that ideas, policies and institutions originating in the European Union shape decision-making in other regional organisations (ROs). However, a research programme which claims to examine processes of interdependent decision-making between ROs in general cannot treat a single RO as the ‘ default ’ source of this di ﬀ usion. This paper thus (1) discusses what we term the ‘ multiple sources ’ problem in di ﬀ usion research and (2) presents a model of institutional di ﬀ usion, which highlights the important role of local actors in the multiple sources scenario. Based on original data collected via extensive ﬁ eldwork in East Africa, we present a case study on the interdependent sources behind the East African Community (EAC) ’ s establishment in drawing on this model. We ﬁ nd that EAC policymakers selected between, combined and modi ﬁ ed several external institutions. These ﬁ ndings highlight local actors ’ capacity for institutional innovation.


Introduction
A broad literature on diffusion in Political Science, Sociology, Economics and Organisational Theory proposes that ideas, norms, policies and institutions developed in one political setting regularly shape decision-making in others elsewhere. This proposition underpins an emerging research programme on diffusion between regional organisations (ROs). This programme Sharing the spirit of Acharya's (2014) 'Global International Relations' for a more inclusive discipline, our analysis heeds Pinar and Futák-Campbell's plea for 'globalizing the study of regionalism in International Relations ' (2021, p. 3). Whereas the field of Comparative Regionalism has long been interested in understanding ROs in both the Global North and the Global South, the theoretical tools employed to do so have come overwhelmingly from the European experience or Western-centric International Relations theories more generally (Iroulo & Lenz, 2022). This is due, at least in part, to a side-lining of African voices in discussions on regional integrationsuch as theoretical contributions from Mazrui (1967) and Asante (1997).
The concept of 'diffusion', we believe, does not inherently have this parochial resonance, in part because it rests on a 'relational ontology' widely advocated by non-Western and postcolonial scholars to build a more inclusive discipline (Murray, 2020;Qin, 2018). Yet, the ways in which it has been employed in the mainstream literature fail to fully realise its promise, with other potential sources of diffusion beyond the EU being neglected and local actors being assigned a secondary role both prevailing tendencies here. We recognise the agency claims of actors in the Global South on both the providing and the receiving ends of diffusion processes. Joining recent scholarship which emphasises the agency of African actors in international politics (see, for example, Coffie & Tiky, 2021;Tieku, 2021), we highlight how they deal creatively with existing external institutions. Our analysis hereby takes a step towards a more inclusive diffusion perspective on ROs.
The article proceeds in four parts. The next one discusses the 'multiple sources' problem in diffusion research and proposes a simple model of institutional diffusion in this context. Then, we describe our research design and case selection. Subsequently, we trace interdependent decision-making between the potential sources of diffusion and the emerging EAC. In the concluding part, we spell out the implications of our findings for the existing literature and derive some promising avenues for future research.
Multiple sources in the study of diffusion between regional organisations Drawing on Shipan and Volden (2012, p. 788), we define 'diffusion' as the process by which institutional creation or change in one RO is influenced by the institutions of one or more others elsewhere. The distinct and powerful claim of diffusion studies is that political decisions cannot be understood by analysing them in isolation; they have to be seen as interdependent (Elkins & Simmons, 2005).
A growing number of studies have tested the claim of interdependent decision-making on the empirical terrain of ROs, giving rise to a research programme on the diffusion of ideas, norms, policies and institutions (Börzel & Risse, 2012a). Over the past decade, this research programme has amassed significant evidence to suggest that ROs cannot be fully understood if conceived of only as self-contained entities. Instead, a variety of direct and indirect interactions between them shape the political decisions they take. 2 Since, as noted, the EU is not only perceived widely to be the most prominent and successful instance of regional economic integration around the world (for example, Fioramonti & Poletti, 2008) but is also an active supporter of regional-integration processes elsewhere, the overwhelming scholarly focus on this particular RO as a provider of ideas, norms, policies and institutions has proven instructive (see, for example, Alter, 2012;Bachmann & Sidaway, 2010;Börzel & Risse, 2012b;Haastrup, 2013;Lenz, 2021;Navarro, 2010;Piccolino, 2020). Yet, this tunnel vision tends to neglect other ROs being potential sources of diffusion. As Jetschke notes, 'Diffusion processes from other regional or international organizations [than the EU] have received less attention ' (2017, p. 181).
We now discuss the biases associated with what we term the 'multiple sources problem', proposing a simple model of institutional diffusion which emphasises the important role of local actors herein as well as their capacity for institutional innovation.

The 'multiple sources' problem in diffusion research
This cited problem entails the possibility that there is more than one source of diffusion. Exclusive focus on the EU as the 'default' source thereof, we argue, generates conceptual, normative and inferential biases. Conceptually, the focus on the EU has led scholars to prioritise unidirectional spatial processes of diffusion over other existing types, even though there is nothing inherent to the concept of 'diffusion' which limits it to the spatial dimension of interdependent decision-making. In her agenda-setting piece, Solingen calls, then, for giving 'proper attention to spatial, directional, and temporal aspects of diffusion ' (2012, p. 640).
Whereas unidirectional and spatial processes of diffusion from the EU towards other ROs have received significant attention, there is a dearth of research examining those of 'multi-directional diffusion' (Zwingel, 2012) that have actors other than the EU at their core. Our empirical analysis, therefore, widens the conceptual gaze vis-à-vis the directional and temporal dimensions hereof. Regarding the former, we examine not only North-South diffusion but also South-South processes. Regarding the latter, we also consider the possibility that diffusion may not primarily be a spatial process, but also one which operates across timespecifically, by examining the influence of the first EAC on its re-established version.
This broadening of the conceptual focus also has normative implications. Diffusion research on ROs has been criticised for being overly EU-centric (Acharya, 2016;Mumford, 2020). Moving beyond this narrow view, then, involves taking seriously 'the agency claims of a broader category of actors' (Acharya, 2014, p. 651), specifically of local actors in the Global South. 3 Some of the diffusion literature has done so by emphasising how such actors localise external 'models' (Acharya, 2009;Rüland, 2014). 4 We advance this by embracing them not only as active 'takers' in transnational diffusion processes but also as sources of diffusion, in both spatial and temporal terms. This enables us to examine, as noted, not only North-South but also South-South diffusion processesthe latter an area of research widely neglected thus far (but see Kingah & Akong, 2018;Stapel, 2022). Moreover, our conceptualisation of 'local agency' goes beyond imbuing local actors with a capacity for modification, as informing the localisation literature, to highlight their capacity also for creativity and innovation. In so doing, we acknowledge for actors on both the providing and receiving end of diffusion processes similar capacities for autonomous action.
Inferentially, studies which focus solely on the EU in examining diffusion between ROs risk suffering from omitted-variable bias: that is, skewed conclusions stemming from the fact that relevant independent variables are excluded from the analysis. As King, Keohane and Verba note, if 'relevant variables are omitted, our ability to estimate causal inferences correctly is limited' (1994, p. 175). Without controlling for other potential sources hereof, scholars may mistakenly attribute outcomes to diffusion from the EUyet they may originate from other sources entirely or constitute a combination of different ones.
To be sure, scholars have sought to address this methodological problem by adducing direct evidence of EU diffusion: that is, they carefully trace how the EU shaped outcomes in other ROs. 5 This is a plausible way to substantiate claims about EU diffusion, but it cannot substitute for a more direct test of the impact of other potential sources here. In the absence of more direct ways to avoid the omitted-variable bias stemming from neglecting other potential sources of diffusion, the claim that the EU's 'impact is […] certainly not spurious' (Börzel & Risse, 2012b, p. 194) may, as such, well be incomplete.

Institutional diffusion from multiple sources, and creative local agency
Conceptualising 'diffusion' as a process of interdependent decision-making from multiple potential sources has implications for the nature of agency on the receiving side, and, accordingly, outcomes. We argue that diffusion in the 'multiple sources' scenario diversifies and extends local agency. In the 'single source' scenario, meanwhile, agency on the receiving side varies primarily according to the degree of adaptation of an external institution, spanning a continuum from full-fledged adoption to far-reaching adaptation. 6 It follows that the more an external institution is adapted to fit local contexts, the more local agency matters for understanding outcomes. Concurrently, convergence across organisations as an outcome is inversely related to the extent to which local agency informs the process. The more the adaptation of an external institution occurs in the diffusion process, the less similar the institutional outcome will be to the original source (Klingler-Vidra & Schleifer, 2014).
In the multiple sources scenario, however, local actors select between, combine and modify a variety of external institutions (see Figure 1 below). Selection involves the choosing of relevant external institutions from the menu of available alternatives (see Jupille et al., 2013, p. 29). Combination involves amalgamating the chosen external institutions into a single one by picking specific elements from each and fitting them together. This assumes that 'institutions can be broken down into smaller components that can be combined in different ways' (Beaumier et al., 2023, p. 8). 7 In this scenario, the analytical focus shifts from the degree of adaptation of single source institutions towards instead the ways in which policymakers select specific external institutions from among multiple sources and combine them to solve their own cooperation problems. 8 The selection and combination of external institutions from different sources is what Stapel (2022) discusses as a pick-and-choose strategy and Jetschke (2017) refers to as 'modular design'.
These approaches provide valuable insight into the outcomes of diffusion processes regarding specific institutional components; yet, they remain somewhat static in neglecting how local actors adapt external institutions to their own circumstances. We suggest that the latter may also play a role in the multiple sources scenario, but it refers not only to the adaptation of a single external institutionwhich the concept of 'localisation' (Acharya, 2009) capturesbut also to the modification of a combination of external institutions. 9 Similar to the single source scenario, modification generally involves taking account of specific local conditions and potentially learning from the experiencestemporal or spatialof others.
The idea of a complex selection, combination and modification of external institutions from different sources has been referred to as 'bricolage': that is, the blending of the institutional material readily available for the construction of new institutions (Campbell, 2005, p. 56). In such processes, local actorsor 'bricoleurs', as Lévi Strauss (1966, p. 11) originally dubbed themdo much of the diffusion work. Agency from this perspective involves a significant degree of creativity on the part of the actors involved and emphasises the innovative nature of the resulting institutions (Carstensen, 2011, p. 148;see also, Beaumier et al., 2023;Stapel, 2022). Figure 1 depicts the process of institutional diffusion, as well as the various tasks which local actors perform, in the multiple sources scenario.

Case selection and research approach
A rigorous test of the diffusion claim, one which takes seriously the outlined problem, must thus examine the impact of multiple sources directly. Yet, the number of contemporary and historical, international as well as regional organisations which could serve as sources of diffusion is very large indeed. We address this selection problem by applying an enhanced version of the 'possibility principle', which states that 'only cases where the outcome of interest is possible should be included [in the analysis]' (Mahoney & Goertz, 2004, p. 653).
Using this principle, we examine sources that are likely to have had an impact. 10 The first EAC, the COMESA and the EU are more likely than other potential sources to have generated diffusion effects in our chosen case for two reasons. First, and as we detail below, there are marked similarities in institutional outcome between these potential sources and the EAC's actual institutions. As is well-established in the literature, institutional similarity is an indicator that diffusion may have taken place, rendering the analysis hereof a sensible undertaking (Elkins & Simmons, 2005). Second, channels exist through which information about these sources could have plausibly reached policymakers in the EAC. The stakeholders who established the EAC were in close contact with both the COMESA and the EU and/or had overlapping memberships. 11 Knowledge about the erstwhile EAC was also extensive among the relevant actors (Biira, 2017;Mwapachu, 2012Mwapachu, , 2019. Our case selection hereby covers three broad classes of potentially influential ROs: a predecessor (EAC I); a neighbouring/overlapping RO (COMESA); Note: A, B and C denote three 'macro' institutions (e.g., a dispute-settlement mechanism) within different ROs consisting of identifiable decomposable elements, or 'micro' institutions (e.g., rules on the permanence, access to and bindingness of decisions of the dispute-settlement mechanism)designated with numerical subscripts. and a non-neighbouring RO (EU). Being located at the 'intersection' of three plausible sources of diffusion makes the EAC a useful case study. At the same time, the EAC is unique in that it has a clear predecessor organisation with a similar mandate and identical original membership, which allows us to examine cross-temporal diffusion processes; is located in a region, Africa, which has created many criss-crossing and partially overlapping ROs, which may serve as rich contemporary sources of information from contextually similar organisations; and, it has had sustained interaction with a non-African RO, the EU, widely seen as successful. This makes the EAC valuable in discerning the respective influence of different classes of ROs, and relevant beyond this case study alone (more on this in due course).
The institutional-design decisions we analyse align with our 'global' approach, which ascribes agency in diffusion processes not only to the providers but also the receivers of external institutions. We start from the perspective of the EAC policymakers confronted with the task of designing the organs listed in Article 9 of the 'Treaty for the Establishment of the EAC' (EACT): the Summit, the Council, the East African Court of Justice (EACJ), the East African Legislative Assembly (EALA) and the Secretariat. 12 Herewith, we go beyond the usual selection criteria, which tend to focus on only one or a few of these constituent institutions.
In line with the qualitative diffusion literature, we seek to detect related effects using process-tracing (George & Bennett, 2005). Tracing diffusion dynamics is demanding because it requires a careful reconstruction not only of the decision-making process itselfthat is, the participating actors, the timing of their interactions and similarbut also of the external influences on individual actors' belief systems (Starke, 2013, p. 574). Our analysis therefore relies on two key sources. First, we conducted 51 expert interviews between 2016 and 2019 during research trips to Belgium, Ethiopia, Kenya, South Africa and Tanzania, as well as online (see Annex). Interviewees were selected based on their specialised knowledge, as experts on the respective institutional decisions (Bogner et al., 2009). Second, we draw on primary documents retrieved from the EAC's archives, press accounts and from think-tank pieces. For each institutional decision, we triangulated oral and written sources to detect and correct for biases, and only categorised an institutional decision as resting on diffusion when evidence for that came from at least two independent sources.
We affirm a diffusion effect to have occurred when we find triangulated evidence suggesting that a specific institutional decision is likely to have looked different had information on our three source ROs not been available. Such an inferential decision sometimes rests on 'smoking gun'-type evidence (Van Evera, 1997, pp. 31-32), in which two independent sources directly confirm this external influence or the wordings of relevant treaty stipulations are (near-)identical between the EAC and the source treaty. In other cases, ascertaining diffusion to have happened rests on a detailed reconstruction of circumstances and evidence strongly indicative of this because they suggest, counterfactually, a decision is likely to have looked different without external influence (Starke, 2013, pp. 575-576).
This may entail the identification of direct encounters between both sides prior to a decision, implicit references to external sources or clearly identifiable similarities in institutional functions and composition. Space constraints prevent us from detailing all the evidence collected. We present instead a selection hereof to make plausible our diffusion claims, declining to discuss those institutional decisions where we did not detect any diffusion from the three identified sources. 13 Empirical analysis: multiple sources and the EAC's selection, combination and modification of external institutions We now trace the impact of three potential sources of diffusionthe first EAC, the COMESA and the EUon institutional-design decisions taken during the re-establishment of the EAC during the 1990s. We start with some background information.

EAC
Today's EAC was established by Kenya, Tanzania and Uganda in the year 2000 to foster economic and political integration in the region. When the member states decided to re-launch cooperation in the early 1990s, a committee of foreign ministers (the so-called Tripartite Committee), was appointed to work out a programme to further develop the (then-called) East African Co-operation (Secretariat of the Commission for East African Co-operation, 1993). This ad hoc arrangement was institutionalised in November 1993 with the creation of a Permanent Tripartite Commission (PTC). After a first proposal of the Treaty by the Secretariat of the PTC in consultation with treaty-making experts (Kiondo, 2002), the draft was circulated for consultation in the three countries among various stakeholders, including parliament, representatives of the private sector and lawyers associations (Ayoki, 2003;Kaahwa, 2003;Kabumba, 2010). Whereas the first EAC featured various institutions with quite specialised responsibilities, ones only loosely arranged and steered by intergovernmental principles (Adar & Ngunyi, 1992), the current EAC was designed as one, coherent regionalist scheme incorporating aspects of supranationalism and intergovernmentalism. 14 The EAC widened its membership base over the years and now includes seven states: Burundi and Rwanda (both 2007); South Sudan (2016); and, the Democratic Republic of Congo (2022). Members deepened integration with the introduction of a customs union in 2005 and a common market in 2010. According to the EACT, the next steps are a monetary union and a political federation.

EAC I
The first EAC was established in 1967 by Kenya, Tanzania and Uganda and was partially built on structures which had been created by the colonial oppressors of the regionmainly Great Britain (Mvungi, 2002). Inheriting these structures and their complex histories, EAC I incorporated the service institutions previously belonging to the East African Common Services Organization (see also, Kiondo, 2002). The first EAC was formally dissolved in 1977, although it had stopped functioning properly early on in that decade already (see Adar & Ngunyi, 1992;Shao, 2002). The first EAC and its failure were very present in East Africa during the 1990s (and continue to be so until today). Abubakar Zein, a parliamentarian in the EALA, reflected the sentiment of many of our interviewees when stating that it was a 'deliberate decision to learn from the collapse of the whole Community and based on that, that had implications in terms of design and architecture of the [current East African] Community' (Interview Zein, 2016).
COMESA COMESA (and its predecessor, the Preferential Trade Area for Eastern and Southern African States) and the EAC are closely interconnected through overlapping memberships. Moreover, the development of the initial East African Co-operation format was explicitly justified under COMESA's principle of 'variable geometry' (Imani Development, 1995;Matata, n.d.;Mwencha, 2019). Further, EAC's trade and fiscal policies were conceptualised within the COMESA framework until the late 1990s (Secretariat of the Commission for East African Co-operation, 1996Co-operation, , 1997a. In 1997, the Secretariat of the PTC undertook visits to various ROs, including COMESA, which led to the signing of a Memorandum of Understanding with the goal of harmonising integration approaches. In the late 1990s, PTC delegates were present in COMESA meetings and explicitly encouraged such interaction in their role of representing the East African Co-operation schemethat is, not as member states of COMESA (Secretariat of the Commission for East African Co-operation, 1997aCo-operation, , 1997bCo-operation, , 1998a.

EU
Since 1992, the European Commission had been invested in a technical assistance project called the 'Cross-Border Initiative Eastern and Southern Africa and Indian Ocean', which sought to strengthen policy-making and implementation at the national level, as well as regional integration (Fajgenbaum et al., 1999;Kennes, 1999). Besides, extensive information on the institutional design and other aspects of the development of the EU were made available by and through EU-funded projects, workshops and exchange formats. A technical advisor of the German Association for Technical Cooperation embedded at the EAC in Arusha described knowledge transfer here as follows: With regard to the question of whether the EAC has learned from the EU, I would say: quite a lot, but in an indirect way. It is supposed to be adapted.
[…] It is not supposed to be a copy! That was never the goal. (Fasbender, 2019, authors' own translation) In 1998, the Commission made recommendations for further institutionalising the East African Co-operation scheme, reiterating the conviction that successful integration requires a permanent institutional framework. The recommendations were integrated into the PTC's further undertakings; during negotiations, the latter's Secretariat explicitly studied the EU and its institutional design (East African Co-operation, 1999; Secretariat of the Commission for East African Co-operation, 1998aCo-operation, , 1998b.

Process-tracing of diffusion from multiple sources
The results of our analysis are summarised in Table 1, and outlined according to the EAC's main organs in due course.

Summit
Meetings between respective heads of state or government already took place under the framework of the East African Co-operation scheme during the 1990s, then being institutionalised as the 'Summit' with the adoption of the EACT in the year 2000. The Summit meets annually, while the position of Chairman, with a tenure of one year, rotates between the different heads of state or government involved. In designing the Summit, East African Within the first EAC, the intergovernmental Authority and the attached (Committee of) East African Ministers held the most sway. Even though the (new) Summit remains commanding, policymakers have sought to mitigate its influence by dividing power between various organsthe Council, the EACJ, the EALA as well as the office of the Secretary General and the Secretariat. Thus, the Summit may delegate any of its duties to one of its members, the Council or the secretary general, except for the appointment of judges to the court, the admission of new members, granting observer status and assenting to bills (EACT; Art. 11(5, 6, 9)). The Summit has the power to assent (or withhold assent) to any piece of legislation (EACT; Art. 11(9d)). Summit decisions are reached by consensus (EACT; Art. 12 (3)). The Summit's powers and decision-making structures (consensus) are still similar to those of the first EAC, yet, as noted, some of the Summit's functions can be delegated to the Council or the secretary generalwhich was not an option under the former EAC. Especially the functions and powers bestowed to the Summit are a reaction to the first EAC's perceived failures, as involving the adaptation of its respective organs (Kaahwa, 2017;Interview Mwapachu, 2019;Oloo, 2005).

Council of Ministers
Similar to the Summit, East African policymakers sought to avoid the mistakes made with the first EAC when designing the Council of Ministers. At the same time, the EU also became a reference point in these discussions since local actors perceived its Council to be a relevant model here. The Council is thus a combination of both ROs, modified to fit regional needs. We see this regarding the composition, decision-making procedures and the Council's position vis-à-vis other organs existing within the EAC's institutional framework.
The Council is constituted by the ministers responsible for regional integration in each partner state, who meet twice a year. Whereas in the first EAC each partner state nominated one parliamentarian to be an 'East African Minister' (Mvungi, 2002), the new EAC features a 'Ministry of East African Community Affairs' in each member country. 'Notably, in comparison to the Ministers of the Community, the current Council is significantly less powerful' (Kaahwa, 2017, p. 59).
The Council's chairman is determined on a rotational basis, who together with a rapporteur make up the Bureau of the Council. The Council gives directions to the partner states and all organs of the EAC, except to the Summit, the EACJ and the EALA (EACT; Art. 3c). Compared to the first EAC, the Council was set up as a separate body (from the Summit) and both were equipped with less decision-making powers than previously, which was a reaction to the erstwhile strong stance of the presidents (and ministers) herein (Interview Wambugu, 2019). However, both the Summit and the Council hold crucial functions within the institutional set-up and any one partner state can block legislation and other formal processes due to the consensus principle in decision-making.
During EACT negotiations, the decision-making procedure changed from majority to consensus (EACT; Art. 15 (4)). This is accredited to the fact that the Summit as well as the Council wanted to prevent a situation whereby one partner state would be overruled (Kaahwa, 2017;Mwapachu, 2012). Throughout the negotiation process, the Council 'gained' legislative powers in particular the competence to initiate and submit bills to the Legislative Assembly, a provision not included in the draft treaty (EACT; Art. 3b). The Council shares legislative powers with the Legislative Assembly and the Secretariat, which bears a strong resemblance to the EU. The presence of the secretary general in the EALA as well as his or her powers with regards to the initiation of bills was influenced by the procedures within the European Commissionthe latter, though, being a distinctly more powerful and differently structured body (Secretariat of the Commission for East African Co-operation, 1998b; Interview Wambugu, 2019).

East African Court of Justice
The EACJ is to 'ensure the adherence to law in the interpretation and application of and compliance with this Treaty' (EACT; Art. 23(1)). In the EACT, access to the EACJ is granted to partner states, the secretary general as well as to legal (i.e., companies) and natural persons (i.e., individuals). All can hold (other) partner states, organs or institutions accountable when they have 'failed to fulfil an obligation under this Treaty or ha[ve] infringed a provision of the Treaty' (EACT; Art. 28(1)). The EACJ also has an arbitration function, which can be called upon by the partner states, organs or other parties (EACT; Art. 32). EACJ decisions on the EACT's interpretation and application take precedence over those by national courts on similar matters (EACT; Art. 33(2)); the latter can also request preliminary rulings on issues related to the EACT (Art. 34). The EACJ only has one chamber. It is made up of six judges in total, two from each original partner statean appellate division was introduced in 2007.
With the EACJ, local actors selected from among the dispute-settlement mechanisms of the first EAC, the COMESA and the EU, combining elements of the three institutions and modifying them to fit their own needs. The EACJ was set up differently to the judicial organs of the first EAC, which were threefold in nature and had quite extensive powersespecially the East African Court of Appeal (EACA). The latter dealt with appeals relating to the rulings of relevant national courts on both civil and criminal matters with the exception of constitutional ones (Tusasirwe, 2005). As one interviewee explained: There was the EACJ, which was inherited from colonial times, so it was like the uppermost court of appeal, but it was not an East African Community court that fits into the mission of the East African Community. That one had criminal jurisdiction and what not, this one would only have a role within the confines of the jurisdiction of the East African Community. So, this again was going to be a large departure from the past. (Interview Mwapachu, 2019) Due to lobbying efforts by law societies, the introduction of a judicial organ such as the former EACA was suggested (United Republic of Tanzania, 1993). As one interlocutor recalled: The court […] is something that was demanded by the law societies, the legal fraternity wanted to go back to the former EACA.
[…] The attorneys general are the ones who drafted something, sending it back and distributed it across the region for consultation. That is how the court got included. (Interview Ruhangisa, 2019) Yet the actual EACJ differs from the erstwhile EACA, as the latter encompassed appellate functions. The former, contrariwise, does not function as the apex court of the region. The EACJ deliberately was not designed as a court of appeal for the region, and its jurisdiction was kept rather vague in the Treaty and, later on, not expanded as originally indicated, instead being restricted (Kaahwa, 2003;Interview Mwapachu, 2019;Interview Ruhangisa, 2019).
In the initial drafts of the EACT, the 'East African Community Court' (as it was called) would be almost an exact replica of COMESA's 'Court of Justice' in terms of composition, jurisdiction and access. 15 The eventually ratified EACT slightly deviates from this, with the main difference being in general jurisdiction. Here, the EACJ's jurisdiction comprises 'other original, appellate, human rights and other jurisdiction' (EACT; Art. 27(2)), which were to be determined by a protocol of operationalisation adopted by the partner states. Regarding all other functions, orders and proceedings, the EACT is a copy of the COMESA Treatybesides the option to prescribe sanctions.
According to Alter (2012), the EACJ features four characteristics which are modelled after the European Court of Justice (ECJ): a supranational commission which can raise non-compliance suits; a preliminary ruling system of national-court referrals; explicit administrative review authority; and, explicit constitutional review authority. These institutional similarities are no coincidence. The first registrar of the EACJ, John E. Ruhangisa, recalled how he and the first cohort of judges went to Europe to study both the ECJ and the European Court of Human Rights.
[O]ne of the things that I planned was to give these judges a study tour. Because having the regional court, we were not the first one. We were not reinventing the wheel, there were other institutions, similar institutions who have been successful, which were operating much as ours, collapsed, but still we had some points to refer to. (2019) In his opinion, the limited jurisdiction ascribed to the EACJ was implemented as a reaction to the ECJ's strong stance over the course of European integration. Thus it constituted a deliberate choice by East African policymakers: 'They were cautious. Of course, the decision-making bodies had learned what happened to Europe. The European Community had been influenced and the court was on the forefront, it was on the driving seat through its decisions' (Interview Ruhangisa, 2019). In conclusion, we can see that the blueprint for the EACJ was derived from the COMESA Treaty. Yet, it was adapted to include some crucial provisions inspired by the ECJ specifically; and, ultimately, the chosen design reflects lessons learned from the shortcomings of the first EAC, too.

East African Legislative Assembly
The EALA is the EAC's legislative arm, and each National Assembly indirectly elects nine persons to the EALA. In the three original partner states, the parties represented in their national parliaments nominate the prospective EALA members. The selection procedure as well as the EALA's other functions differ considerably from regional parliaments elsewhere (Schimmelfennig et al., 2020, esp. chapter 13). EALA members are elected for a period of five years and can be re-elected once (EACT; Art. 51(1)). Decision-making is based on majority voting; ex officio members do not have a vote. The EALA is responsible also for debating and approving the budget (EACT; Art. 49(2b)). With regards to the legislative processes, motions and bills can be introduced through a 'private members bill' by EALA members (EACT; 59(1)). The ultimate power to assent or withhold assent to bills lies with the Summit (EACT; Art. 63)here, the decision is based on the consensus principle. If a bill fails to be agreed upon by the Summit in its second reading, it lapses (EACT; Art. 63(4)).
In designing the EALA, policymakers selected from elements of the first EAC and of the EU, combined the respective institutions' frameworks and modified them to fit local needs. During the Treaty discussions, the speakers of the national parliaments proposed to name the legislative body the 'East African Legislative Assembly'instead of the suggested 'East African Community Assembly'as a way to invoke the legislative competences of the first EAC (Secretariat of the Commission for East African Co-operation, 1998c). The current EALA was created to foster closer links with the citizens of East Africa and thus to provide a forum embodying a more 'peoplecentred' integration process, reiterating the inclusion of more, and more diverse, stakeholders as a lesson from the first EAC (Interview Mwapachu, 2019; Interview Parliamentarian EALA (Rwanda), 2016; Interview Zein, 2016).
The former Assembly had a different role: it was closely linked to the governments, inter alia through the involvement herein of the East African Ministers. As some EALA members indicated, the legislative body was set up so as to play a more relevant role within the overall framework and, as such, stands in contrast to the original EAC parliament (Interview Parliamentarian EALA (Kenya), 2016; Interview Zein, 2016): 'This [current] parliament plays an important role in terms of representation and oversight. So, there was deliberate learning' (Interview Zein, 2016).
The current legislative procedure deviates from that of the former EAC given a joint decision-making process between the Secretariat, Council and the EALA now existsalthough, as mentioned, assent is ultimately given (or withheld) by the Summit, and the process is (still) quite strongly steered by the intergovernmental bodies. This is similar to EAC I, where the heads of state or government had to give their assent to bills. Thus, we see the selection of some of the former EAC's legislative procedures, although the final decision is (again) with the heads of state or governmentthe outcome, thus, of a modification process.
Some have compared the earlier version of the European Parliament to the EALAspecifically, in composition, formal powers and standing within the framework of the respective ROs (Mwapachu, 2012;Oloo, 2005). Bobi Odiko, EALA Senior Public Relations Officer, stated: 'We as an Assembly are also the only legislative assembly on the entire continent actually. We only mirror the EU […] as regional parliaments that have legislative powers' (2016). The process of modification of both legislative bodies, from EAC I and the EU, led to the implementation of a less-powerful and not directly electable legislative body within the current EAC. Mwapachu described this in the following way: There was a legislative assembly, under the old one [the first EAC], but without teeth. It was more of a political organ, without teeth in terms of being close to, being representatives of the national political space. So, the new treaty said: since the goal of this new EAC is political federation and those stages, we have to establish a legislative assembly that manifests the movement towards a legislature in a federal system. So, I think in a sense this legislative assembly then was inspired by the European Parliament. (2019) The idea of direct elections was discussed but ultimately dismissed due to the high costs they would entail (Mair, 2001;Mukandala, 2000). The European Parliament unequivocally served as a source of inspiration from which East African policymakers selected certain key features, combining them with aspects of the first EAC in order to modify the organ to fit their own interests.

Secretariat
The Secretariat is the executive organ of the EAC, and is directed by the Secretary General, Deputy Secretary Generals, Counsel to the Community and by other officesas appointed by the Summit (EACT; Art. 66). The secretary general's term in office is five yearsrotating with regards to their citizenship (EACT; Art. 67). The Secretariat has the following core tasks: legislative functions, such as initiating, receiving and submitting recommendations to the Council, as well as forwarding bills to the Assembly through the Co-ordination Committee (EACT; Art. 71(1a)); the researching and implementation of programmes (Art. 71(1b)); strategic planning, as well as the management and monitoring of programmes (Art. 71(1c)); and, the general promotion of the EAC to stakeholders, the public as well as the international community (Art. 71(1f)).
The Secretariat is another instance of the process of combining and ultimately modifying an organ in drawing on multiple sources. According to Mwapachu, the Secretariat's configuration would be based on learning processes which involved consciously adopting elements from both the first EAC and the EU: [I]n the first EAC, each country had ministers at the EAC Secretariat […] who had to report to their presidents, so it was like an extended cabinet […]. How was that going to be avoided? So, in the new Treaty that was removed.
[…] I mean we should follow the EU system, don't call it commissions, but the Secretariat with different departments looking at the different sectors that make the Community an important factor in integrating the economies of the region. (2019) The first EAC did not have a secretariat, although it did introduce the position of Secretary General, who was assisted by a Counsel to the Community and an Auditor-General. The current EAC deliberately included a secretariat with a secretary general and two deputies, operating under the Permanent Tripartite Commission to coordinate the cooperation. This, too, would be framed as a learning experience: [T]hey decided to have a lean secretariat, they did not want the former kind of bureaucracy in Arusha. So they started by saying: let's start with the political representatives of the three countries plus a small secretariat that should work around confidence-building measures […] and develop a framework that would lead to a treaty. (Interview Wambugu, 2019) The idea was to put in place an organisational entity which could act as a coordinator among the involved national stakeholders and ensure the coherence of the integration process (Permanent Tripartite Commission for East African Co-operation, 1997; Secretariat of the Commission for East African Co-operation, 1998a;Interview Wambugu, 2019).
The EAC does not have an organ comparable to the European Commission; the body which comes closest is the Secretariat (Interview Fasbender, 2019;Interview Mwapachu, 2019). Due to what was perceived among policymakers as overbearing (and costly) administrative entities in other ROs, not least in the EU, the EAC Secretariat was established as a small and efficient coordination body (East African Co-operation, 1999; Secretariat of the Commission for East African Co-operation, 1998b). It shares with the European Commission its direct involvement in legislative processes through its right of initiative and working with the Co-ordination Committee and the Council, as well as the (oversight over) implementation at the national level. The core functions of the EAC Secretariat are thus quite wide-reaching in light of its limited size, staff and financial equipment.

Conclusion
In this article, we have studied the process of establishing the EAC's core organs over the course of the 1990s with a view to identifying the impact hereon of diffusion from several other ROs elsewhere. We examined the impact of three likely sources of diffusion from Africa and Europethe first EAC, the COMESA and the EUin analysing whether and how policymakers adjusted external institutions to fit local circumstances. Following the spirit of Acharya's (2014) 'Global International Relations', we hence take seriously the agency of African actorsa long-standing theme in African Studies (see Bayart, 2000). Below, we spell out the theoretical implications of our findings and identify some promising avenues for future research.
Our results indicate that diffusion processes in the design of ROs and their constituent institutions are significantly more complex than recognised in the existing scholarship, and that in three key respects. First, a range of such organisations often serve as the sources of inspiration for other ROs, not just onewhat we termed the 'multiple sources' problem. The EU was indeed an important source of diffusion vis-à-vis the EAC's eventual design, yet so were the first EAC and COMESA, too. Thus, and speaking to the normative aspect of the multiple sources problem, our study contributes to debunking the perception that the EU is the only relevant source of institutionbuilding when it comes to other ROs around the globe. Moreover, we highlight the relevance of cross-temporal diffusion processes, from the first EAC to the current one . As such, one promising avenue for future research going forward would be to identify the conditions under which policymakers draw on one or another RO in institutional-design decisions.
Second, our analysis points to a more important role for those on the receiving end of diffusion processes than has been acknowledged thus far. Even if the literature on transnational diffusion widely recognises that these local actors do not simply adopt external institutions but generally adapt them, how they do so specifically and what the various roles played here by such individuals are remain overlooked. We revealed that local actors select between, combine and modify external institutions and herewith engage in a good deal of innovation based on existing institutional material. This shows that they often innovate in significant ways through diffusion. As Beaumier et al. note, 'inventors can see further […] by standing on a pile of inventions made by their predecessors […] combining design components to create a new arrangement is the essence of the inventive process ' (2023, pp. 7, 9). Whereas diffusion processes are generally depicted as contrary to innovation, future research could seek to elucidate the relationship between diffusion and innovation more fully.
We believe that the general lessons derived from the study of a single case the EACare likely to ultimately travel further. The EAC is unique in being re-established from a predecessor organisation which had been dissolved, but most ROs today have seen failures from which they seek to derive lessons while being embedded in a wider network of ROs in their immediate neighbourhood and beyond. For example, the EU's Lisbon Treaty is, in part, an attempt to learn from the failed EU Constitution, and some of these lessons have reached other world regions. The Association of Southeast Asian Nations designed its 2007 Charter in partial response to what was widely perceived as excessive detail in the EU's constitutional treaty (Wong, 2012, p. 674).
In fact, the proliferation of ROs and the resulting growing interaction between international organisations, global and regional alike, is bound to make diffusion from multiple sources more likely. That not all organisations will play similar roles in such diffusion processes from multiple sources given their relevance may vary is also a conclusion which seems applicable beyond the EAC. Finally, this more complex scenario of powerful external influence implies that deciding which organisations to draw on, and how, vis-à-vis the design of their own institutions is bound to increase local actors' impact on the diffusion process and its outcomes. Notes 1. Our focus is squarely on the design of new institutions and does not consider the ways in which local actors may imbue newly created institutions with local values and meanings.
2. Earlier studies already conceptualised diffusion dynamics between ROs but offered little systematic testing thereof (for example, S. K. B. Asante, 1982). 3. There is also a literature in Comparative Regionalism which is consciously less Euro-centric, such as the New Regionalism approach (Söderbaum & Shaw, 2003)but it has had little to say about diffusion per se. 4. This analytical move is part of a broader research programme now putting the agency of often-neglected actors centre stage (Coffie & Tiky, 2021;Tieku, 2021). 5. Scholars also seek controls for alternative explanations, but generally not for other sources of diffusion. 6. This is the focus of much of the existing work on diffusion (Börzel & Risse, 2012a;Jetschke & Murray, 2012). 7. As we will see, an RO can be broken down into several institutionsa council, a secretariat, a court, etc.and each can again be broken down into smaller components besides, such as rules on an institution's composition, functions and internal decision-making procedures. 8. In the multiple sources scenario, local actors necessarily reject most models by selecting from a large menu of existing institutions; we thus treat rejection as the other 'side of the coin' in institutional selection. We thank one reviewer for encouraging us to clarify this point. 9. We use the term 'modification' here to distinguish it from 'adaptation', which is generally associated with a single source of diffusion. 10. Apart from other ROs, domestic institutions may also provide a relevant source of inspiration here. The latter are more likely to be relevant, however, when there are few similar institutions available (see Beaumier et al., 2023, p. 9) -ROs in our case. Since this was not the case at the time of the EAC's creation in the course of the 1990s, we do not consider domestic institutions further here. 11. Kenya, Tanzania and Uganda were all members of COMESA (until Tanzania withdrew its membership in the early years of the new century). Other ROs, especially in Africa itself, may also fulfil these criteria of the 'plausibility principle'they overlap with the EAC (e.g., African Union, Southern African Development Community) and/or potential channels of communication exist (e.g., Indian Ocean Commission, Economic Community of West African States), not least via the AUbut we exclude them from our analysis since they are less empirically relevant (see Reiss, 2022). 12. We do not consider the Co-ordination Committee and the Sectoral Committees, also mentioned in Art. 9, because our empirical material on this organ is too thin. 13. We present a full summary of our results in Table 1. 14. For an in-depth discussion of developments during the EAC's establishment period, see Reiss (2022). 15. Apart from the Court, the Treaty Establishing the Common Market for Eastern and Southern Africa and the EACT share significant similarities too. The wording and content of chapters 12, 19, 22, 25, 83, 85-89, 91-93, 95 and 139 of the EACT are almost identical to the respective sections in the COMESA Treaty (see also, Mvungi, 2002).
on Europe and Regionalism of the German Political Science Association, as well as the participants of the conference 'East African Regionalism in Uncertain Times' at the British Institute in Eastern Africa for their valuable feedback. Mariel Reiss would like to thank the United Nations University -Institute on Comparative Regional Integration Studies for hosting her as a visiting researcher while she worked on this article. Tobias Lenz gratefully acknowledges funding from the Leibniz Association (grant number J31/2017).

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

Notes on contributors
Tobias Lenz is a Professor of International Relations, Leuphana University of Lüneburg, and leader of the Leibniz Junior Research Group "Legitimation Strategies of Regional Organisations" (LegRO), German Institute for Global and Area Studies (GIGA), Hamburg, Germany.
Mariel Reiss is a research fellow at the Center for Conflict Studies, Philipps-University Marburg.