Learning in Lending: The World Bank’s Organisational Learning Journey in Albanian Health Financing Reform Projects, 1994–2022

The role of aid in the development of low- and middle-income countries (LMICs) has been extensively studied, with many scholars adopting a critical viewpoint. For instance, Easterly (2002; 2006) argues that despite substantial investments and efforts, donors, including the World Bank (WB), have often fallen short of achieving meaningful impacts. Similarly, Sachs’s (2015) findings suggest a tenuous relationship between official aid and economic growth. Within this discourse, the role and effectiveness of the WB in fostering development have come under intense scrutiny.

While the WB aims to address global challenges, promote sustainable development, and reduce poverty through its interventions, concerns have been raised regarding its effectiveness in addressing the unique needs of different countries and varied sectors (Heyneman, 2003; Mosley et al., 1995). A critical determinant of its effectiveness lies in the organisational commitment to continuous learning (Ravallion, 2016). However, the extent to which the WB adheres to this criterion in practice, particularly in social sectors, remains uncertain. In these sectors, “where lending and policy operations have had to work hard to justify themselves”, the WB encounters challenges compared to other sectors, such as infrastructure (Ravallion, 2013, p. 27).

This article explores the WB’s lending projects for health financing reforms in post-communist Albania from 1994 to 2022 and addresses the research question: How does the World Bank learn during its health lending projects?

There are two compelling justifications for focusing on the WB’s health projects. First, the WB is the largest funder of health-related projects in LMICs (Ruger, 2005). This significant financial contribution underscores the WB’s influence on health policies (Noy, 2017; Sridhar et al., 2017), making it a crucial area to examine. Second, the WB is also the second-largest funder of global health initiatives (Lancet, 2023). This highlights the WB’s substantial involvement in shaping global health policies and programmes, making it imperative to understand its approach and learning activities in the context of health projects.

Additionally, paying attention to post-communist Central Eastern Europe (CEE) – by focusing on Albania, an under-researched country in the region – is noteworthy due to the WB’s involvement in the political, economic, and social transformation of the CEE alongside the criticisms it has faced. Scholars studying the post-communist healthcare reforms in CEE have highlighted the WB’s “politico-ideological” approach and its tendency to overlook domestic contexts when shaping policy changes (Nemec & Kolisnichenko, 2006; p. 24; Deacon & Hulse, 1997; Wedel, 2015).

The selection of lending projects in Albania’s health financing reforms is particularly relevant, as the WB has been involved in four health projects in the country from 1994 until 2022, offering a rich and comprehensive basis for deep analysis. The Health Services Rehabilitation project (1994–2001) initially focused on infrastructure reconstruction and healthcare service rehabilitation, with limited attention to health financing policy reform. The Health System Recovery and Development project (1998–2005) continued this focus on infrastructure while also targeting institutional capacity building. In contrast, the Health System Modernization project (2006–2012) marked a shift towards structural healthcare reforms, moving away from infrastructure reconstruction. The Health System Improvement project (2015–2022) focused extensively on health financing reforms, addressing issues in all financing policies: revenue collection, pooling, purchasing, coverage, and benefit entitlements.

For detailed information on these policies, see Kutzin et al. (2010).

The analysis employs a sociological lens on International Organisations (IO) by incorporating insights from management studies into International Relations (IR) scholarship. The interest is in scrutinising documented knowledge, conceptualised as a form of explicit and collective knowledge that is codified and communicated using language (written documents) and deposited in a collective or public knowledge repository (WB repository’s online platform), making it easily shareable and assessable over time (Nonaka, 1994; Duncan & Weiss, 1979). In the context of WB lending projects, these documents primarily comprise project management reports, with the majority being monitoring and evaluation reports, as detailed in the method section. These reports epitomise a model of result-based project management that considers accountability and learning as inseparable entities, as noted by Kogen (2018), “two sides of the same coin: that by addressing one we are necessarily addressing the other” (p. 101). The implications of this conflation are discussed in the final section. The focus on documented knowledge means that I will not look at individual learning, while recognising that all organisational processes involve individual action. Instead, I am considering organisational learning as “a social process […] of exchange, evaluation and integration of knowledge … that is extraindividual”, one intended to produce “a general organisational knowledge base” that is crucial for organisational effectiveness (Duncan & Weiss, 1979, p. 89).

The qualitative analysis of data obtained from the WB project documents revealed three types of learning during lending operations in Albania: learning into lending, learning while lending, and learning from lending. Additionally, findings showed that the WB’s country-level interventions in health financing reforms were influenced by prior knowledge and lessons learned during and after project completion. However, these processes resulted in adjustments to activities and were intended to achieve project aims and goals, mainly by detecting and swiftly correcting errors and project bottlenecks rather than leading to learning.

The article elucidates the challenges the WB faced in supporting Albanian health financing reforms, thus enriching debates on the effectiveness of the WB’s lending activities and its ability to meet the specific needs of the healthcare sector (Ravallion, 2013). It offers insights into the WB’s involvement in post-communist healthcare reforms in CEE, particularly focusing on health financing, an area marked by inconclusive findings (Kaminska et al., 2021). It also sheds light on an under-researched country, Albania, within this context (Druga, 2022). Moreover, the study advances our understanding of organisational learning in IOs, an area underexplored in existing literature (Lindenthal & Koch, 2013; Koch, 2015). Notably, it empirically illustrates IOs’ challenges in balancing standardised approaches and customisation, particularly in operational knowledge (rather than policy recommendations). What sets this study apart is its unique perspective. While past research has primarily focused on examining the impact of IOs’ activities on national contexts, this study addresses a notable gap (Zapp, 2020, p. 1038) by exploring IOs’ responses to country-specific demands and contexts.

The article is organised into six sections. The following section outlines the state of knowledge in the current literature, analysing both WB lending projects and WB learning processes. It underscores the limitations of existing studies and informs the conceptual framework applied in the analysis. Next, the method section explicates the multiple case design, the strategy for document collection, and how data analysis proceeded. Based on the conceptual framework, the fourth section examines the WB’s learning processes in the four projects. Section five discusses the findings, and the paper concludes with final remarks in the sixth section.

The study employed a descriptive and exploratory approach, utilising qualitative research methods (Yin, 2017, p. 39). A multiple case study method (Stake, 2013) examined four WB health projects in Albania between 1994 and 2022, as depicted in Table 1, to understand the commonalities and differences in learning across these cases. First, each case was analysed individually to gain insights into the specific learning processes within their contextual setting. Subsequently, through cross-case analysis, the study explored the commonalities and differences in learning across these cases, offering insights into how learning manifested in different temporal contexts (p. 27).

Before elucidating the criteria employed to select documents for qualitative content analysis and elaborating on the analysis process, it is essential to offer an overview of the WB’s lending instruments and associated processes. This overview facilitates reflection on the nature of the project documentation retained for analysis, recognising that lending project documents encompass various types, each characterised by specific attributes and tailored to address distinct issues (like presentations to governing boards).

The WB employs three instrumental approaches in its lending projects: policy-based lending, investment project loans, and programme-for-results lending. Policy-based lending involves conditionality, requiring borrowers to implement specific policies. Investment project loans are tied to particular projects, with the WB providing funding based on the agreed plan. Introduced in 2004, programme-for-results lending emphasises borrower needs, local contexts, and aid coordination (Cormier, 2015; Mosley et al., 1995).

The lending project cycle consists of three overarching phases: preparation, implementation, and evaluation.

For detailed information on the project cycle framework used by the WB, see https://projects.worldbank.org/en/projects-operations/products-and-services/brief/projectcycle

While the roles and contributions of local and international project team members are significant in shaping project outcomes, the research interest in this article is on examining the WB’s organisational learning process by specifically focusing on documented knowledge rather than tacit knowledge held by WB staff or their Albanian counterparts. This limitation, however, presents an opportunity for future research to explore the interplay between documented and tacit knowledge in WB lending projects, offering a deeper understanding of organisational learning dynamics.

Documentation plays a crucial role in each stage of the WB’s project management, serving as a resource for learning and accountability. Various reports are produced at different stages, including Staff Appraisal reports and Program Documents during preparation, Implementation Status Results reports and Restructuring reports during implementation, and Implementation Completion and Results reports, Implementation Completion Report Reviews, and Project Performance Assessment reports during evaluation.

The logic of the WB project cycle (preparation, implementation, and evaluation) and the respective line of documentation, Staff Appraisal reports and Program Documents (preparation), Implementation Status Results reports and Restructuring reports (implementation), and Implementation Completion and Results reports, Implementation Completion Report Reviews and Project Performance Assessment reports (evaluation) for the four WB health projects in Albania, comprise the first set of criteria for the document collection and data analysis. Next, following the Learning in Lending framework, the second set of criteria consists of documents that contain knowledge (and learning) either from other WB lending activities with some components in the healthcare reforms or from analytical work, known as Economic and Sector Work (ESW) (see Bazbauers, 2018, pp. 175–6) produced for the Albanian healthcare reforms during the period under investigation.

To collect relevant documents for analysis, a two-step approach was employed. The initial step involved gathering health project-related documents using the WB’s projects and operations database,

https://projects.worldbank.org/en/projects-operations/projects-home, accessed on March 21st, 2023

Fig. 2:

Data analysis was undertaken manually. The process began with screening all documents and excerpting the relevant text passages about lesson learning and lesson drawing, which were usually distinct sections or paragraphs in the WB reports. For instance, Staff Appraisal Reports, as documents drafted before starting a lending operation, contain data on what previous knowledge is used in designing these operations and contextual – political, economic, and social – data and information enclosed in a subsection entitled “Lessons learned and reflected in the project design” (for instance, see WB, 2006b). These data were later analysed to explore learning into lending. The other documents, such as Restructuring Reports and Implementation Status and Results Reports, produced during the revision of project implementation depict knowledge and learning gained during this project stage, that is, learning while lending. Finally, documents such as Implementation Completion and Results Reports, Implementation Completion Report Reviews, and other evaluation reports issued after the project’s completion or termination portray learning from lending. The last group of documents also contributes to corroborating data. Finally, an additional analysis of forthcoming projects was conducted to check whether they would implement the knowledge and learning gained. This step provided evidence for learning into lending. Additionally, Implementation Completion Report Reviews were the document source for performance ratings after project completion.

The WB’s learning process showed progression over time. The initial project in 1994 relied on limited knowledge. Between 1998 and 2005, a notable shift towards a “learning by doing” approach led to more experience and lessons. This growth trend continued with further enhancements in the WB’s learning from 2006 to 2012 and 2015 to 2022. In addition, as presented in Table 1, the four projects underwent various changes during implementation. These processes are analysed in the following subsections.

The study investigated the WB’s learning processes in lending in the context of health financing reforms in Albania from 1994 to 2022. Through a multiple case study approach, the analysis examined the unfolding of these processes within four specific health projects. The findings revealed the applicability of the Learning in Lending framework developed by the IEG and identified a consistent pattern of three types of learning that occurred during lending operations: learning into lending, learning while lending, and learning from lending.

The findings demonstrate the type learning into lending, in which previous knowledge and lessons were incorporated into the design of subsequent projects. For instance, the second project incorporated lessons from the first one by adopting a modest design and considering the challenges of the Albanian context in setting expectations. The third project focused on implementing structural reforms based on insights from previous projects and sector analysis, prioritising them over infrastructure investments. The fourth project integrated inputs from previous projects, including pilot programmes, improved monitoring and evaluation systems, and comprehensive technical assistance activities.

The findings also show that not all sorts of learning gained from previous projects were considered during the design of the subsequent project. For instance, a key piece of learning from lending in the third case was the importance of the political economy analysis of the Albanian health financing reform. Despite recognising its significance, this valuable lesson was overlooked, resulting in a lack of planned analytical and advisory technical work to address it in the design of the fourth project. Consequently, the failure to incorporate this indispensable lesson into the project design ultimately resulted in extensive changes during implementation and the complete exclusion of the project component associated with the health financing reform in 2018.

The type learning while lending consisted of adjustments in project scope, activities, and timelines to address challenges and ensure progress in implementation. Examples include adapting activities in the first project to avoid duplication of technical assistance and revising the project scope and activity structure in the second project based on analytical work outside the health sector. The third project underwent significant changes during implementation to address weaknesses within the Ministry of Health and enact reforms. The fourth ongoing project has experienced more modifications, even dropping key project components.

The final type, learning from lending, highlighted lessons related to the importance of understanding contextual factors, effective project management, and the influence of political and governance factors on implementation progress. Other additional lessons included the significance of timely information dissemination, accurate assessments of domestic capacity, and the need for robust monitoring and evaluation systems.

Across cases, the findings highlight a striking pattern in the WB’s organisational behaviour – increased supervisory and reporting activities – indicating that adjusting activities during project implementation was a prevalent practice. In the first and second cases, these activities included mid-term reviews, with one in the former and two in the latter. In the third case, five supervisory and reporting activities took place in addition to mid-term reviews. Finally, nine such activities have been conducted for the ongoing fourth case up to 2018.

These results indicate an adaptive behaviour of the WB during project implementation (Fiol & Lyles, 1985, p. 811) and highlight the WB’s attention towards project oversight and reporting activities. As a result, short- and medium-term modifications to project implementation plans followed. Nonetheless, while identifying and correcting errors, the findings show that various planned activities changed significantly (in scope and financing) and were even dropped. In line with Argyris and Schön (1978), in each case (project), the WB’s learning processes followed a single-loop approach characterised by the tendency to identify and correct errors by adjusting activities and aiming for immediate outcomes and quick fixes. Across cases, however, repeating-the-same-error patterns were observed, such as concerning the need for technical assistance activities or (inaccurate) estimations of domestic capacities in policy implementation and policy change.

These arguments lead to the conclusion that even though the most frequently manifested behaviour in the WB’s organisational learning was in the form of adaptive learning (Siebenhüner, 2008, p. 96), the adoption of new lessons in future activities failed in some instances. Furthermore, a “no learning form” (ibid.) due to not addressing new challenges (one was the political economy analysis of the health financing reform) demonstrates another pattern in the WB’s organisational behaviour.

Did the WB’s organisational learning processes in Albanian health projects imply a “learning to learn” (Meyer, 2002, p. 464) approach? Not at all. On the contrary, in some cases, the organisation even failed to learn. Did the WB’s learning enhance the project’s implementation? The present analysis, although not delving deeply into this aspect, is able to provide a modest answer. Effectiveness, that is, the fulfilment of the project’s purposes and objectives, was achieved, sometimes with changes in the activities’ scopes and funding with deadline extensions, and sometimes with terminating activities and even entire project components. This organisational behaviour aligns with Meyer’s statement that “effectiveness … is a good criterion for single-loop learning” (2002, p. 465). Further studies are needed to investigate and quantitatively assess the effectiveness of the WB’s health projects in Albania, as this was beyond the scope of this article. Future studies could also explore the potential for the WB to engage in double-loop learning processes that lead to critical reflection and challenge fundamental assumptions underlying the WB’s current practices (Argyris & Schön, 1978; Siebenhüner, 2008).

This study has provided valuable insights into the WB’s learning processes in an operational context. However, it is essential to acknowledge its limitations. By focusing on documented knowledge, the analysis has overlooked, first, bureaucracies’ adeptness at reporting (change) despite the tangible results on the ground, and second, the tacit knowledge, the unspoken knowledge held by individuals that cannot be formally documented. In these regards, interviews with WB personnel involved in the project design, implementation, and evaluation and their Albanian counterparts involved in project implementation would have provided a more profound insight into learning dynamics. Future research is encouraged to employ diverse data sources and approaches to enhance the analysis of organisational learning processes in the WB and other IOs.

The importance of learning in aid and development organisations has long been recognised as a critical concern (Cassen Report, 1986). In line with this recognition, the WB has emphasised the role of learning in enhancing the effectiveness of its lending operations. However, one issue that warrants attention in these remarks is the relationship between accountability and learning. For instance, an evaluation report on the WB’s lending projects (IEG, 2014) sheds light on weaknesses in the organisation’s operational culture concerning the role of learning in achieving projected results, as it criticises the tendency of using accountability mechanisms, such as supervising, monitoring, and evaluating activities, which are critical tools for organisational management, to improve project ratings rather than as valuable sources for learning.

In other words, although effective learning requires robust feedback loops that allow the organisation to gather input from all stakeholders involved in a lending project, evaluate the project results, and adjust project activities accordingly, accountability mechanisms focused on compliance with predefined targets may, however, not prioritise the proper collection and analysis of feedback from beneficiaries and partners. As the IEG report reveals, lessons learned from previous projects are not consistently integrated into subsequent project designs, and preparing appraisal documents for new projects is sometimes a mere copy–paste process, disregarding the lessons drawn (IEG, 2014).

Balancing accountability and learning within aid and development organisations is paramount, especially amid ongoing debates on aid effectiveness. While accountability often revolves around meeting predetermined objectives and demonstrating results, learning involves a deeper understanding of effectively promoting development. Kogen’s analysis sheds light on the prevailing tendency to conflate these two concepts, where accountability is sometimes prioritised at the expense of fostering learning. She emphasises that learning should not be seen as a byproduct of accountability practices but as the primary objective of evaluation. Rather than viewing success solely through achieving predetermined targets, Kogen advocates prioritising learning to understand how best to promote development effectively. According to her, the focus should shift from whether aid works to the more critical questions of why and how aid works, particularly when it works (Kogen, 2018, pp. 101–2; p. 105).

This is a noteworthy conclusion when discussing the distinction between evaluation practices and accountability mechanisms in relation to aid organisations, particularly the WB or organisations in general. While both types of organisations strive for efficiency and effectiveness in achieving their goals, aid and development organisations, such as the WB (also UN agencies), have a unique mandate for “the betterment of the world at large as their task” (Forss et al., 1998, p. 52). Learning is, therefore, vital for such IOs to effectively address complex development challenges and fulfil their mandate.