How Context and Reform Align (or Misalign): New Evidence from the Education Sector

Why have increased access to schooling and discoveries about learning not translated into improved learning outcomes? One answer is that knowledge about learning needs to be connected to knowledge about governance, policymaking, and implementation.


Image of Brian Levy

Brian Levy

School of Advanced International Studies, Johns Hopkins University

Recognition that ‘context matters’ for development policymaking and implementation has become commonplace, almost to the point of cliché—but going beyond the general nostrum to something practically useful continues to challenge. This post draws on a set of ambitious new RISE-commissioned case studies of the politics of policymaking in education to illustrate the practical potential of an approach that takes seriously the ways in which power and institutions shape context, and thus reform opportunities and constraints. (A related post lays out the approach’s theoretical underpinnings.)

On the surface, global gains in educating children have been remarkable. Access has expanded enormously. So, too, has education-sector-specific knowledge about how students learn and successful teachers teach. Yet the combination of access and knowledge has not translated into broad-based gains in learning outcomes. Why?

To better understand the reasons for this disconnect, and to help uncover new ways of improving learning outcomes, the RISE programme sponsored a set of country studies of the politics of education policy adoption. In early 2022, I was commissioned by RISE to write a synthesis overview of the individual studies. The synthesis paper provided an opportunity to explore further some questions left over from an earlier round of research in the education sector.

A disconnect between knowledge in education and knowledge in governance, policymaking, and implementation

Back in 2012, I launched an in-depth research project on the politics and governance of basic education, centred around case studies in two provinces of South Africa. The project built on decades of work as a researcher-practitioner at the interface between governance and economic development across a wide range of sectors (though never before on education), I came away from the research enormously impressed by the rigor of specialised, education-sector scholarship and, more broadly, by the knowledge and commitment of many in the education-sector-focused policy and research community. But I was also struck by how little progress has been made in linking this knowledge to broader findings on interactions between governance, policymaking, and implementation.

In seeking to account for this disconnect, a useful point of departure is the 2018 Learning World Development Report’s distinction between proximate and underlying causes of learning shortfalls. Proximate causes include the skills and motivations of teachers, the quality of school management, the available of other inputs used in schools, and the extent to which learners come to school prepared to learn. Underlying these are the governance arrangements through which these inputs are deployed. Specialist knowledge on the relation between the proximate causes and learning outcomes can straightforwardly be applied in countries where governance works well. However, as the RISE-commissioned political economy case studies detail vividly, in countries where the broader governance context is less supportive, specialist sector-specific interventions to support learning are less likely to add value.

How to move forward in the latter contexts?  “Focus not only on sector-specific technical interventions, but also on improving governance” is a seemingly obvious answer. That answer is not wholly wrong—but it can all-too-readily be interpreted in ways that lead reformers down counterproductive dead-ends. To see why, consider the definition of governance offered by an influential World Bank report:

 “Governance is the process through which state and nonstate actors interact to design and implement policies within a set of formal and informal rules [institutions] that shape and are shaped by power”.

As this definition signals, governance processes are embedded within broader contexts shaped by power and institutions. Further, as voluminous research has shown (see the books Violence and Social Orders and In the Shadow of Violence: Politics, Economics, and the Problems of Development, and another blog post of mine), over the medium-term (in most countries, most of the time) these broader contexts change only on relatively small margins. Viewed from the perspective of sector-level decision-makers, the broader political context is exogenous. What can be done to improve outcomes in messy governance contexts?

A typology of political-institutional contexts

One useful way to move forward is to construct a typology, organised around a small number of distinct contexts, each characterised by distinctive configurations of power and distinctive institutional forms – and thus distinctive patterns of incentive and constraint (and possibilities for improving development outcomes) within which sectoral governance plays out. With a set of distinct types in hand, a key next step is to find ‘good fit’ ways forward, by identifying a variety of potential entry points for improving outcomes, and clarifying how they align with the different contexts.

The synthesis paper is organised around three distinct (heuristic) political-institutional types, each resonant with a familiar ‘real-world’ pattern. (See the related blog post for a discussion of the theoretical rationale for focusing on these three heuristic political-institutional contexts.) The three are:

  • Context A (‘dominant’)—in which power is highly concentrated, and exercised top-down.
  • Context B (‘personalised competitive’)—in which authority is fragmented, with multiple centres of power, limited capacity for co-operation, and limited compliance with formal rules.
  • Context C (‘impersonal competitive’), characterised by strong formal ‘rules of the game’ that are intended to provide a platform for resolving conflict among stakeholders and their goals, and for implementation—but, insofar as political contestation remains unresolved, can result in a combination of exaggerated rule compliance and/or isomorphic mimicry.

This piece (and the RISE synthesis paper) uses this three-fold typology to organise, and analyse comparatively, the country cases studies on the politics of education policy. As an initial step, the case study countries were grouped into the three types. This was done using three V-DEM governance indicators—the extent of electoral democracy, the quality of the rule of law, and the pervasiveness of clientelism. The resulting categorisation is shown in Table 1. (See the synthesis paper for details).

Table 1

Dominant Personalised Competitive Impersonal Competitive

Ethiopia (1941 - 2016)

Indonesia (1871 - 1988)

Nigeria (1970 - 1999)

Tanzania (1970 - )

Vietnam (1970 - )

Bangladesh (1991 - 2012)

Ghana (1993 - )

Kenya (2003 - )



Chile (1990 - )

India (1978 - )

Peru (2001 - )

South Africa (1994 - )


The synthesis paper builds on the individual case studies to lay out some distinctive, within-type patterns of education sector governance:

  • In dominant contexts, with power centred around a political leader and a hierarchical governance structure, the education sector’s goals are largely shaped by the leaders.  As the Vietnam case details, top-down leadership potentially can provide a robust platform for improving learning outcomes.  However, as the case studies of Ethiopia, Indonesia, Nigeria and Tanzania illustrate, all too often dominant leaders’ goals vis-à-vis the education sector can veer in other directions.
  • In impersonal competitive contexts, a combination of strong formal institutions and effective processes of resolving disagreements can, on occasion, result in a shared commitment among powerful interests to improve learning outcomes—but in none of the case studies was this outcome evident. Instead (as discussed further below), the case studies showed a combination of unresolved political contestation over the education sector’s goals, exaggerated rule compliance, and performative isomorphic mimicry.
  • Personalised competitive contexts such as Bangladesh, Ghana and Kenya lack the seeming strengths of either their dominant or their impersonal competitive counterparts; there are multiple politically-influential groups and multiple, competing goals—but no credible framework of rules to bring coherence either to political competition or to the education bureaucracy.

As the case studies detail, these political and institutional realities rendered ineffective many specialised sectoral interventions intended to improve learning outcomes.

Finding entry points

What might be some context-aligned entry points for improving learning outcomes in the midst of this messiness? (A RISE insight note addresses this question in depth; the paragraphs that follow summarise the argument in that note.) Key is to open up space in a way that enables sector professionals to bring their specialist knowledge to bear. The rows in Table 2 highlight four ‘soft governance’ entry points with space-expanding potential.  Each entry point is (loosely) aligned with a distinct level in a chain of governance processes that link politicians, policymakers, public officials and citizens:

  • The leadership level - purpose: What are the goals of the education system? How to strengthen leaders’ orientation towards learning?
  • The bureaucracy level - mission: How to empower mission-oriented public officials within the education system?
  • The stakeholder level - alliances: Which influential stakeholders champion learning? How to foster co-operation among these stakeholders, thereby strengthening their collective influence?
  • The citizen-level - expectations: How do the expectations of parents, communities and citizens influence the extent to which the education sector is learning-oriented? How might learning-oriented influences be strengthened?

As the cells in Table 2 suggest, and the paragraphs below detail, the potential for each of these ‘soft governance’ entry points to improve learning outcomes varies systematically across the three types.

Table 2: Improving learning outcomes when governance is messy—top priorities across contexts

Leadership level: Purpose Top priority: influence ideas of political leaders vis-à-vis education    
Bureaucracy level: Mission   Top priority: Foster socially-embedded bureaucratic autonomy
Stakeholder level:  Alliances Top priority: Foster local (including school level) alliances
Citizen level: Expectations  

Source: Author.


For dominant contexts, the top ‘soft governance’ priority is to engage with leaders as to the purpose of education. As noted earlier, Vietnam was alone among the case studies of dominant countries in consistently having improvements in learning outcomes as the sector’s principal goal. In Indonesia and Tanzania the principal goal was to align education with a distinctive set of ideas about the nation and its collective identity; in Ethiopia under the military Derg regime (and in Nigeria, too, at least for a time) it was to expand access to historically excluded groups, with little attention to quality. (How) can leaders be persuaded to prioritise learning, and to take the steps needed to improve learning outcomes?

For personalised competitive contexts, contestation among stakeholders invariably leads to policy incoherence, bureaucratic fragmentation, and high risks of predation—and thus little prospect that efforts to strengthen public systems can gain traction. Yet as Table 2 suggests (and as analyses of Bangladesh and Ghana detail), fragmentation can have a silver lining—it can create space for alliances of developmental stakeholders to successfully push back against predatory pressures, and eke out islands of effectiveness at local levels (sometimes even as localised as an individual school). More ambitiously, insofar as a societal expectation of “all for education” can take hold—that parents and communities, especially, have an active role to play in supporting a learning-oriented education system—then, even in personalised competitive contexts, far-reaching national gains in learning outcomes can be achieved.  As the synthesis paper details, Kenya offers an example of what is possible.

In impersonal competitive contexts, there is (as noted earlier) a clear normative vision of how a learning-oriented education system should function. In practice, however, things fall short of that vision in all four of the impersonal competitive case study countries:

  • In Peru, there was ongoing conflict over purpose between politicians on the one hand, and sectoral stakeholders and experts on the other.
  • In India, there was a large disconnect between (national) policymakers and (state-level) implementers.
  • In Chile and South Africa, there was an ongoing pre-occupation with formal systems, with correspondingly less de facto attention on how to improve learning.

Lessons from two RISE case studies

A comparison of the Chilean and Peruvian case studies offers some striking insights as to both the challenges confronting impersonal competitive contexts, and a promising way forward.  In Chile, interactions among stakeholders largely were top-down and systematically managed. Peru, by contrast, was characterised by ongoing back-and-forth jockeying among stakeholders, messy compromises with the teachers’ union, and multiple policy reversals. Insofar as better-aligned institutional arrangements and systematic, consistent policies are likely to be more effective than ‘messier’ ones, learning outcomes would be expected to show more improvement over time in Chile than in Peru. Yet, as the synthesis study details, between 2000 and 2018 Peru achieved very large gains in learning outcomes, while the gains in Chile were modest. Why?

The Chilean approach to sector governance was, from a technocratic perspective of governance, “best practice”. Yet the (not yet published) case study concludes that:

“Good intentions to improve educational quality, resources and carrots and sticks have not been enough to move the Chilean educational system in the direction that its political authorities wanted…. The top down character of Chilean educational policy making and the insufficient use of institutional voice mechanisms might backfire as the mounting social tensions and the 2019 social movement casts some doubts about its survival” (p.47)

By contrast, Peru’s messier, less formalistic and more iterative process of policy formulation and adaptation helped build broad legitimacy among stakeholders—importantly including strengthening trust in the technocrats and professionals responsible for its formulation—thereby enhancing their ability to push back against idiosyncratic initiatives proposed by political appointees. As the Peru country case study put it:

“Civil society organisations – NGOs, universities, think tanks and research centers – have also played a key role in defining policy agendas [and….]  in the development of education policies and reforms. Though not always able to contain either technocrats’ or other policy makers – agreements are often ignored by ministerial administrations and political parties - they have certainly contributed to the continuity of agendas and to the advancement, through piecemeal, of reforms.”

The contrasting trajectories of Chile and Peru point to the importance, in impersonal competitive contexts, of not seeking to govern education solely within the strictures of an autonomous bureaucracy, but rather to open up space by embracing “social embeddedness”, working to build developmental alliances with a sense of shared purpose. Indeed, the point applies broadly. Across the range of less-than-perfect governance contexts, rather than focus narrowly on technocratic (governance or sector-specialised) initiatives, foreground attention to the question of “commitment to learning”. Especially in competitive contexts (both personalised and impersonal), cultivate the idea that improving learning outcomes is everybody’s business, and create opportunities for engagement—invite citizens to become active participants in a shared endeavour to equip coming generations with the capabilities they will need to be part of a vibrant, thriving society.


This blog originally appeared in the author’s website.

RISE blog posts and podcasts reflect the views of the authors and do not necessarily represent the views of the organisation or our funders.