Jump to ContentJump to Main Navigation
The Politics and Governance of Basic EducationA Tale of Two South African Provinces$

Brian Levy, Robert Cameron, Ursula Hoadley, and Vinothan Naidoo

Print publication date: 2018

Print ISBN-13: 9780198824053

Published to Oxford Scholarship Online: November 2018

DOI: 10.1093/oso/9780198824053.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (oxford.universitypressscholarship.com). (c) Copyright Oxford University Press, 2022. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.date: 02 July 2022

Context and Capability

Context and Capability

A Tale of Two Bureaucracies

(p.179) 7 Context and Capability
The Politics and Governance of Basic Education

Brian Levy

Robert Cameron

Vinothan Naidoo

Oxford University Press

Abstract and Keywords

This chapter explores how context influences bureaucracy. Bureaucratic behaviour and performance are interpreted as endogenous, shaped by decisions of political elites as to whether to direct their efforts towards providing public services or towards more narrowly political or private purposes. The chapter distinguishes among three broad contextual differences between the Western Cape and Eastern Cape—socio-economic, political, and institutional. It identifies the causal mechanisms through which these variables exert their influence, distinguishing between demand-side and supply-side influences. In the Eastern Cape, the consequence of an initially weak context is a low-level equilibrium trap in which incentives transmitted from the political to the bureaucratic levels reinforce factionalized loyalty within multiple patronage networks. By contrast, in the Western Cape, both demand-side and supply-side contextual variables support public service provision; however, weaknesses in ‘soft governance’ limit the positive impact.

Keywords:   context, bureaucracy, Western Cape, Eastern Cape, education governance, demand-side, comparative politics, patronage, low-level equilibrium trap, public service provision

7.1 Introduction

Chapters 4 and 5 examined in depth the workings of the Western Cape and Eastern Cape education bureaucracies. Even though the two provinces operate within an identical inter-governmental framework, they differ starkly from one another in their socio-economic, political and institutional contexts. They thus provide an ideal platform for exploring comparatively some fundamental questions concerning the relationship between context and bureaucracy.

In what ways does context shape how bureaucracies operate? Where bureaucracy is relatively weak, to what extent does context constrain the prospects for improving performance? Where bureaucratic capability is relatively strong but politics turns toxic, to what extent can bureaucracy serve as a brake on a downward spiral? Can participatory ‘horizontal’ governance help improve public service provision—both as a complement to a relatively strong bureaucracy, and as a substitute in settings where capability is weak? Building on the empirical platform provided by the earlier chapters, this chapter explores these questions.

7.2 Two Divergent Bureaucracies

We begin by summarizing some of the findings from earlier chapters as to the quality of the education bureaucracy in the two provinces. Table 7.1 reports the results for 2012/13 of Management Performance Assessment Tests (MPATs) for the Departments of education in the Western Cape, the Eastern (p.180) Cape and two other provinces. The MPATs, sponsored by South Africa’s Department of Policy Monitoring and Evaluation, located in the office of the Presidency, benchmark over a hundred national and provincial government departments against a variety of key performance indicators. As the table suggests, the Western Cape Education Department (WCED) was rated as the best managed of the country’s nine provinces (Gauteng rated second). The Eastern Cape Department of Education (ECDoE) rated the weakest (the Northern Cape rated second weakest (The Presidency, 2013).

Good bureaucratic processes need not, however, translate into good results. Along with managerial quality, educational outcomes comprise another relevant benchmark for assessing the performance of departments of education. Here, too, as Table 7.2 shows, the Western Cape benchmarks well against other South African provinces.

Table 7.1. MPAT Assessments of South Africa’s Education Departments (selected provinces 2012/13; distribution of scores, by level for assessed key performance areas)

Context and CapabilityA Tale of Two Bureaucracies

Notes: Level 1—non-compliance with legal/regulatory requirements; Level 2—partial compliance with legal/regulatory requirements; Level 3—full compliance with legal/regulatory requirement; Level 4—full compliance, and doing this smartly.

Source: The Presidency, 2013.

Table 7.2 summarizes some scores from standardized tests administered in 2007 to a large sample of sixth graders in fifteen countries by the independent Southern and East African Consortium for Monitoring Educational Quality (SACMEQ).1 As the table suggests, the Western Cape was the top performer among South Africa’s nine provinces, with Gauteng a close second; the Eastern Cape was the weakest performer. The ranking of provinces is similar whether one takes the median score, the score for learners at the seventy-fifth percentile of socio-economic distribution, or the score for learners at the lower, twenty-fifth percentile, socio-economic tier. As Chapter 6 explored econometrically, the Western Cape remained the strongest SACMEQ performer among South Africa’s nine provinces even after controlling for a variety of exogenous influences on performance. (Note that scores for Kenya also are included in the table and that relative to Kenya the Western Cape does not benchmark as well. Possible reasons for this, and its implications, are considered further in the final section of this chapter.)

Table 7.2. The Western Cape’s SACMEQ Scores in Comparative Perspective

50th percentile (median)

25th percentile

75th percentile

Western Cape








Eastern Cape




South Africa (overall)








Kenya (Nairobi)




Note: achievement in Grade 6 mathematics and home language by province, 2007.

Source: SACMEQ data files (2007) RSA, DBE (2010).

(p.181) Corresponding to the differences between the two provinces in performance, the modes of operation of the two bureaucracies also were starkly different. We characterize these differences using the 2×2 governance framework introduced in Chapter 1 and applied in Table 7.3 below. One dimension of the framework comprises whether governance arrangements are hierarchical (that is, organized around vertical relationships between ‘principals’ and ‘agents’), or whether they are negotiated (that is, organized around horizontal ‘principal–principal’/peer-to-peer arrangements). The second dimension distinguishes among governance arrangements according to whether they are based on impersonal rules of the game, which are applied impartially to all who have standing, or whether they are organized around personalized ‘deals’ among influential actors. This gives us four distinct (and familiar) patterns of bureaucratic operation.

The first two patterns build on a foundation of impersonal rules. The first of these, hierarchical-impersonal bureaucracy, captured in the top-right cells of Table 7.3, is the classic mode of rule-governed bureaucracy delineated by Max Weber (1922) as the mechanism through which government pursues public purposes and partially ‘modernized’ in recent decades under the rubric of New Public Management (NPM).2 The second (though not strictly speaking an example of bureaucracy), captured in the bottom-right cells of the table, comprises a form of ‘corporatism’—formal processes through which multiple principals agree collectively on codified rules of the game for policymaking and implementation.

(p.182) The third and fourth patterns operate on the basis of informal, de facto rules of the game. Hierarchical-personalized bureaucracy, the top-left cell, is similar in part to Weberian bureaucracy insofar as the mode of organization is hierarchical, and thus governed via nested principal–agent relationships—but here compliance on the part of agents follows from the personalized authority of the leadership, rather than a system of rules; the system is one of patronage. Merilee Grindle, in her 2012 book Jobs for the Boys, underscores the fatal weakness of patronage systems is not that they are inevitably incompetent, but that they are capricious’3—unconstrained by rules, and thus subject to the preferences of the hierarchical leadership.

In the final pattern, fragmented-personalized bureaucracy (in the bottom-left cells), neither formal rules nor a well-defined hierarchy of authority are in place. Appointments into public positions are politicized, with the right to appoint distributed across political factions; appointees generally focus their efforts on serving the interests of their various patrons. Where intermediate- and lower-level officials are committed to developmental goals and are skilful in nurturing alliances among internal and external stakeholders, they may be able to create ‘islands of effectiveness’. But in this pattern, overall effectiveness is low.

As Chapter 4 explored in depth, viewed through the lens of the above framework the Western Cape Education Department (WCED) emerges as disproportionately ‘hierarchical-impersonal’. It has a strong professional orientation among both the administrative bureaucracy and teaching staff. Turnover among the senior bureaucratic officials has been low, insulating the bureaucracy in significant part from the cycles of political change, including the rapid turnover of the provincial-cabinet-level appointments of political heads (i.e. the provincial ministers of education). Throughout two decades of democratic government, the WCED has endeavoured to put in place results-oriented approaches to performance management. It consistently worked to implement the ongoing stream of results-oriented systems reform initiatives which emanated from the national level, and complemented them with a variety of more home-grown initiatives (implemented especially vigorously under the Democratic Alliance (DA) administration which has governed the province since 2009). These include sophisticated computerized tools for managing budgets, staffing and procurement, school improvement plans (SIPs) for each of the provinces 1,500 schools, and online tracking systems for monitoring the progress of individual learners through the WCED system, and (on a quarterly basis), the progress of schools in implementing their SIPs.

(p.183) The left-hand side of Table 7.3 provides an heuristic summary of the Western Cape bureaucracy in terms of the two governance dimensions highlighted above, allocating 100 points across the four cells.4 For reasons delineated in Chapter 4, the Western Cape bureaucracy is depicted in the table as predominantly, but not exclusively, impersonal and hierarchical: the top leadership of the bureaucracy exerts discretionary authority over some decisions; other decisions are painstakingly negotiated between the bureaucracy and other stakeholders (including teachers’ trade unions) through tightly formalized processes. Yet overall, the Western Cape is an exemplar of a well-managed, hierarchically oriented and rule-bound bureaucracy.

Table 7.3. Characterizing Education Bureaucracies—Two Contrasting Patterns

Context and CapabilityA Tale of Two Bureaucracies

Source: chapters 4 and 5.

By contrast, the Eastern Cape Department of Education (ECDoE) resembles more closely the ‘fragmented personalized’ pattern of bureaucratic operation. As Chapter 5 detailed, it has been bedevilled by divergent and competing regional interests, organizational cultures, and patronage ties which consistently defied centralized control. Since 1994 the ECDoE has experienced repeated leadership turnover, and a general flouting of centralized authority. Obstacles to enforcing management control and sustaining leadership continuity have contributed to chronic weaknesses in both financial and personnel management. In March 2011 national government intervened, and temporarily took over administration of the ECDoE. But this did not stem the crisis. Provincial politics proved too powerful. After a few years, intervention was scaled back, having had only a limited impact.

The right-hand side of Table 7.3 characterizes the Eastern Cape pattern heuristically by allocating about two-thirds of the percentage points to the personalized column. Mirroring the fragmentation of the bureaucracy, the bulk (but, again, not all5) of these are in the negotiated rather than hierarchical cell.

(p.184) 7.3 Accounting for the Divergent Bureaucratic Patterns

This section explores how context influences bureaucracy. Bureaucratic behaviour and performance are interpreted as endogenous, shaped by decisions of political elites as to whether to direct their efforts towards providing public services or for more narrowly political or private purposes—with the incentives and decisions of political elites shaped in turn by ‘exogenous’ variables. Viewed from this perspective, the divergences in behaviour of the Western Cape and Eastern Cape education bureaucracies laid out in section 7.2 can be attributed to differences in context across the two provinces.

Applying the analytical framework laid out in Chapter 1, Table 7.4 groups the relevant contextual differences between the Western Cape and Eastern Cape provinces into three categories—socio-economic, political and institutional. The discussion which follows provides detail on each of these, and identifies four causal mechanisms which link the contextual variables identified in the table to bureaucratic behaviour and performance, distinguishing between demand-side and supply-side influences.

Table 7.4. Two Divergent Contexts

1. Socio-economic

2. Political

3. Inherited institutions

Western Cape

Diverse social composition—both ethnically and by economic class

- elites dispersed across multiple political parties, with two broad groupings

Inheritance of ‘impersonal’ bureaucracy

- competitive elections

Eastern Cape

Homogenous social composition—disproportionately poor & Xhosa-speaking

- ANC electorally dominant;

Inheritance of patronage bantustan bureaucracies

- deep intra-party fragmentation

7.3.1 Demand-Side Pressures for Performance

The first two causal mechanisms highlight the influence of exogenous socio-economic and political variables on the effectiveness of citizens’ demands on bureaucrats and politicians for decent public services. The first mechanism comprises the well-recognized6 relationship between social class and effective demand—with middle-class citizens generally better positioned than their low-income counterparts to exercise voice effectively in response to poor-quality services (and mismanagement and corruption more broadly), and thus more likely to exercise voice to pressure for better public performance.

Economically, the Western Cape was (and remains) among the wealthiest of South Africa’s nine provinces; as of 1996, its per capita income was 50 per cent higher than for the country as a whole. The Eastern Cape, by contrast, was the poorest province in the country; its per capita income has hovered at around 50–60 per cent of the national average. These differences in average income (p.185) between the two provinces translate into far-reaching differences in class composition.7

Table 7.5 reports some findings from Schotte, Zizzamia, and Leibbrandt’s (2017) disaggregation of South Africa’s income distribution into five distinct social classes. Based on this disaggregation, 49 per cent of South Africa’s total population is classified as ‘chronically poor’.8 In the Eastern Cape, the share of the population which is ‘chronically poor’ rises to 70 per cent. In the Western Cape, it is 25 per cent. The ‘modal’ social class in the Western Cape is the middle class (35 per cent of the total population), with the ‘vulnerable’ and (p.186) ‘transient poor’ accounting for a further 35 per cent. Insofar as there is a positive correlation between economic status and voice, then solely on the basis of these differences in class composition, non-elite citizens in the Western Cape generally are likely to be better positioned to demand better public services than are their Eastern Cape counterparts.

Table 7.5. Distribution of Social Classes in South Africa, 2014

South Africa (all)

Western Cape

Eastern Cape





‘Middle’ class








Transient poor




Chronic poor








Note: The distribution is based on an empirical analysis of four waves of panel data collected between 2008 and 2014 by South Africa’s National Income Dynamics Study.

Source: Schotte, Zizzamia, and Leibbrandt (2017)

The second demand-side mechanism works via the electoral process. Here, the key proposition is that citizens will be better positioned to exert demand-side pressure for decent public services in settings where elections are competitive than in those where politicians can take the support (or acquiescence) of citizens for granted, independent of how well they govern.9 The differences between the two provinces in electoral competitiveness are stark—for reasons which are rooted partly in demography, and partly in history.

In the Eastern Cape, 86 per cent of the 1996 population of 6.1 million was black African, almost all Xhosa-speaking. (For South Africa as a whole, the 1996 black African share of the total was 77 per cent.) In the Western Cape, by contrast, the largest ethnic subgroup of the population was (in the South African lexicon) coloured. In 1996, this group comprised 54 per cent of the 4.1 million people in the province; black Africans comprised only 21 per cent. By 2011, population in the Western Cape had risen to 5.8 million (with the increase driven in significant part by migration from the Eastern Cape), with the black African and coloured shares now comprising 33 per cent and 49 per cent, respectively.

The Western Cape’s relatively high ethnic diversity provided a platform for closely contested elections. To be sure, the province had been a major locus of opposition to apartheid and, as elsewhere in the country, most non-white anti-apartheid activists identified strongly with the African National Congress (ANC). But the distinctive ethnic characteristics of the Western Cape implied that among at least part of the population, loyalty to the ANC was not necessarily unequivocal—and insofar as the ANC framed the basis for its allegiance in narrowly ‘African nationalist’ terms, rather than an inclusive non-racialism, there was ample scope for shifting allegiances, and thus voting patterns, away from the ANC and towards the opposition DA, and other political parties.10

The result was a highly competitive provincial politics, with multiple competing political parties, and hotly contested elections. Over the course of the first two decades of democracy, seven different political parties/coalitions (p.187) have controlled the province. In these contested elections, non-elite swing voters (in the ‘middle’, ‘vulnerable’ and ‘transient poor’ classes) became politically central. Indeed, contestation among rival political parties for the allegiance of this demographic and economic middle has been heated.

In sum, consistent with the second causal mechanism, in the Western Cape, how a party governed while in power—whether it was perceived to use public resources well, or for more narrowly personal and political purposes—mattered for its future electoral prospects.

In the Eastern Cape, by contrast, the ANC dominated electorally. In 1994 it won 84 per cent of the vote in the province; this percentage declined subsequently, but as of 2015 had not fallen below 70 per cent. The electoral dominance of the ANC can be explained in part by the interaction between the province’s ethnic homogeneity and its distinctive historical legacy. A disproportionate number of the leaders of the struggle for liberation from apartheid—including Nelson Mandela, Oliver Tambo, Thabo Mbeki, Winnie Madikizela Mandela, Chris Hani and Walter Sisulu—were Xhosa-speaking sons and daughters of the province. This proud history translated into strong electoral loyalty to the ANC on the part of the numerically preponderant Xhosa-speaking voters.

To be sure, demand-side pressures are not the only determinants of a political party’s commitment to provide decent services. Even if middle class voices are weak, and elections uncompetitive, a commitment to better services might nonetheless derive from solidarity between a political party and its non-elite supporters. Further, it is plausible that electoral dominance could provide a ruling political party with the authority and long time horizon supportive of better service provision.11 However, for reasons which will become apparent in the next subsection, this is not what happened in the Eastern Cape.

7.3.2 Politics, Institutional Inheritances, and the Supply of Services

We turn now to the two ‘supply-side’ causal mechanisms through which politics and institutions influence bureaucratic behaviour: the role of inherited institutional legacies; and patterns of intra-elite contestation within a governing political party.

As Keefer and Khemani (2005) explore, in choosing how to seek the allegiance of non-elites, political leaders confront a fundamental choice. One option is ‘programmatic’—leaders might try to win legitimacy by promising (p.188) a specific programme of public action, and then utilize the public bureaucracy to deliver on these promises. Alternatively, they could seek to win support through personalized patronage and clientelistic vertical networks that link elite and ‘intermediate class’12 power-brokers with local communities. A key determinant of how political leaders choose between these two mechanisms is whether their promises to provide broad-based services (rather than targeted patronage) are perceived to be credible by voters. This is where inherited institutional legacies come into play.

The Western Cape inherited a bureaucracy which could straightforwardly respond to the relatively strong effective demand of citizens for services. During the apartheid era, alongside the white political and bureaucratic structures, the apartheid government had established a parallel ‘parliament’ and bureaucracy, the (coloured) House of Representatives (HoR). The white civil service and the HoR bureaucracy together were responsible for the provision of services (including education services) to the large majority of the Western Cape population.13 As numerous studies of organizational culture have shown,14 both South Africa’s white public service and the HoR bureaucracy were steeped in traditional public administration, albeit with an apartheid bent. This culture has been described as bureaucratic, hierarchical and unresponsive, aimed more at controlling rather than developing the citizens of the country—but it generally operated on impersonal and hierarchical (that is, ‘Weberian’) lines. This inherited legacy of a relatively capable bureaucracy meant that voters were likely to perceive politicians’ promises to provide public services as credible, adding to the impetus for political competition to be oriented around competing programmatic platforms, rather than patronage.

In the Eastern Cape, by contrast, the so-called ‘Bantustans’ comprised the crucial institutional legacy from the apartheid era. Two-thirds of the Eastern Cape’s total 2015 population of 6.9 million people reside in areas which formerly had been part of either the Transkei or Ciskei Bantustans. Both the Transkei and Ciskei had nominally been independent (recognized as such only by the apartheid South African government). Both had large-scale bureaucratic apparatuses which were moulded together to comprise the major15 part of the post-apartheid provincial bureaucracy.

(p.189) The two Bantustans had been organized along personalized, patronage lines.16 These patronage patterns carried forward into the workings of the Eastern Cape province via two distinct mechanisms. As per Keefer and Khemani (2005), one mechanism worked through the credibility of political promises: weakness of the bureaucracy at the outset of the democratic era meant that, even under the best of circumstances, persuading citizens that promises to provide decent services would be credible would be an uphill challenge. But the circumstances prevailing in the province were especially unpropitious. This brings us to the second supply-side influence on bureaucratic behaviour—the role of intra-elite contestation within a governing political party.

The propensity for patronage in the Eastern Cape was compounded by the way in which Bantustan political elites were incorporated politically. In the wake of the dissolution of the Transkei and Ciskei Bantustans, a large majority of their political and bureaucratic elites (and also many ordinary citizens) joined the ANC—not out of conviction, but as members of convenience. Further, the (non-Bantustan) Eastern Cape ANC was itself hardly an ideologically unified party. In its early years, the ANC (both within the Eastern Cape and nationally) reflected the aspirations of a mission-educated aspirant African middle class which increasingly was being constrained by racially discriminatory policies.17 In the decades prior to democracy, parts of the party had become increasingly militant; both Govan Mbeki (Thabo Mbeki’s father) and Chris Hani were stalwarts of the South African Communist Party. The result was that the Eastern Cape ANC was less a disciplined, programmatically oriented political organization than an overall umbrella beneath which inter-elite contestation was endemic. This continuing contestation afforded the ANC’s provincial leadership neither the authority nor the longer-term time horizon needed to translate electoral dominance into a commitment to better service provision.18

In sum, the Eastern Cape was especially poorly positioned vis-à-vis each of the causal mechanisms linking context to bureaucratic operation and performance. The middle class was weak. Elections were uncompetitive. The bureaucratic legacy was dismal. The wide diversity of ideologies and motivations resulted in ongoing internecine struggles within the Eastern Cape ANC’s leadership. Patronage, not programmatic commitments to improve services, became the default mechanism for maintaining the political allegiance of non-elites.

(p.190) 7.4 From Context to Action

What does the evidence that ‘context matters’ imply as to the prospects for improving bureaucratic performance?19 Does better bureaucratic performance translate into better service provision? Insofar as the bureaucracy is weak, what are the prospects for improvement? This final section explores these questions.

7.4.1 Cumulative Causation—Virtuous and Vicious Circles

A central goal of this chapter is to clarify and illustrate how political drivers shape and constrain the performance of bureaucracies. The interactions between the political and bureaucratic realms turn out to be more complex, with more potential for two-way causality, than is implied by simple nostrums along the lines of, say, ‘politics is trumps’.

Chapter 1 identified four distinct ‘ideal types’ of political settlement, each cascading down through multiple institutional, political, and organizational levels. While specific contexts often can be hybrid combinations of the four patterns, in practice the Western Cape and Eastern Cape turned out to be paradigmatic provincial-level examples of two distinct types:

  • The Western Cape emerged as a classic example of a ‘competitive with rule-of-law’ provincial-level political settlement; as section 7.3 has detailed, all of the exogenous variables mutually reinforced one another in a way which fostered a high-level equilibrium of a capable bureaucracy.

  • The Eastern Cape, by contrast, was a paradigmatic example of a personalized competitive (or competitive clientelist) settlement; the exogenous variables reinforced each other in the opposite direction from the Western Cape, locking in a low-level equilibrium.

The two case study provinces provide an opportunity for considering the dynamics of each of the two polar opposite ‘ideal types’. Consider, first, an Eastern Cape type starting point of an initially weak socio-economic, political and institutional setting. As the Eastern Cape case suggests, the result of this mutually reinforcing pattern is a low-level equilibrium trap. From that starting point, whether a virtuous spiral of improvement takes hold—or whether a (p.191) vicious circle of a continuing low-level equilibrium of patronage and poor service provision persists—follows directly from actions which emanate from the political realm.

A vicious circle will persist if an initially weak political context remains unsupportive (with, say, continuing inter-elite competition and continuing neglect of the development challenges confronting non-elites). In such circumstances, the incentives transmitted from the political to the bureaucratic levels reinforce factionalized loyalty within multiple patronage networks, with little incentive to improve the provision of public services.

In a context such as this, technocratic tinkering to improve bureaucratic performance is unlikely to gain traction. Rather, a necessary condition for change would be a transformation of the political context in a way which altered the incentives emanating from the political level in directions more supportive of development. Thus, a virtuous spiral might become possible if, say, the governing party were to become more developmental (perhaps as a result of internal changes within the party, perhaps the result of electoral success on the part of a newly emergent opposition). But even then, the continuing ‘stickiness’ of the other exogenous variables (weak effective demand; weak institutions), the journey of public sector transformation would be long and difficult, and risks of reversal would abound. In such contexts, policymakers and others might usefully focus on trying to improve service within ‘islands of effectiveness’ which are not dependent on broader socio-economic, political, and institutional change.

Now consider an alternative, Western Cape type, starting point where both the demand side and supply sides support the provision of decent services. This high-level equilibrium provides something of a buttress against a politically driven downward spiral: If the political leadership remains pro-developmental (with or without an actual alternation of the governing party) then good quality service provision can be sustained.

If, notwithstanding the initially strong capabilities, a new set of elites won power and sought to pursue private interests, they would be constrained (at least for a while) by the prevailing system—both the strong ‘Weberian’ culture of adherence to impersonal rules, and by the fact that within a few years they would need to defend their record in what almost certainly would be a highly contested election. Eventually, sustained predatory pressures from political leaders could breach the buttresses. However, the conflicts between predatory politics and a rule-governed bureaucracy would, at a minimum, provide time and opportunity for the political realm to self-correct, before a downward spiral could take hold.

Note that the robustness of the above two equilibria is derived from the mutually reinforcing configuration of the ‘exogenous’ contextual variables. (p.192) Had the research been undertaken in other provinces, it is likely that in at least some of them, the exogenous variables would have aligned in a more mixed way:

  • Relative to the Eastern Cape, some provinces potentially could have a stronger institutional inheritance, or a more internally coherent provincial wing of the ANC. In such settings, the challenge of turnaround might not be as daunting as it is for the Eastern Cape.

  • Relative to the Western Cape, some provinces (Gauteng is an obvious example, at least until 2016) might have had a similarly strong institutional inheritance and middle class, but less closely contested elections.

In these more mixed cases, the quality of public bureaucracy would be more of a knife edge. In the former group, were a developmentally oriented political leadership to win control of the province, the path to improving bureaucratic performance could be easier than in the Eastern Cape. Conversely, in provinces which were partial approximations of the Western Cape, because the ‘guardrails’ are relatively weaker, were control to pass into the hands of a predatory leadership, the bureaucracy could more readily slide into a downward spiral.

To make the point differently, our analysis is not intended to imply that there is no scope for provincial-level leaders (both political and technocratic) to make a difference, for good or ill. But it does imply that the broader socio-economic, political, and institutional context has a powerful influence on how much change is feasible.

7.4.2 Improving Educational Outcomes—Bureaucracy and its Limits

While context may shape bureaucratic quality, how well bureaucracy functions is not the only governance determinant of educational outcomes. To be sure, a well-functioning hierarchy can undertake efficiently many of the logistical tasks (e.g. teacher post provisioning; payroll; infrastructure provisioning and maintenance; textbook and supplies management) associated with a large public educational system. Further, a well-functioning bureaucracy potentially might function as a sort of transmission belt, communicating emerging evidence as to pedagogical approaches which improve educational outcomes throughout the school system.

Yet for all of these potential benefits, the research reported in Chapters 4–6 also underscores some of the limits of hierarchical approaches to improving outcomes. One limitation is political. As the study of the Eastern Cape bureaucracy in Chapter 5 underscores (as does, to a lesser degree, the Chapter 3 analysis of efforts to introduce performance management nationally), the (p.193) explanations for weak bureaucratic performance often are to be found, not in shortfalls of management, but in constraints which derive from the ways in which political actors interact with one another. In such settings, insofar as these constraints are decisive, it follows that efforts to improve educational outcomes might more usefully focus on approaches whose efficacy is not dependent on unachievable gains in bureaucratic functioning. To make the point differently, in such settings, horizontal governance potentially might function as a potential institutional substitute, providing a platform for school-level ‘islands of effectiveness’ to take on, reasonably effectively, some tasks which in bureaucratically better-endowed settings might be better done hierarchically.

Now consider the contrasting Western Cape setting where bureaucracy functions relatively well. In such settings, are hierarchical approaches to improving educational outcomes sufficient—or might there be case for complementing them with more horizontal initiatives? Both theory and evidence suggest that there may indeed be a case for the latter. Theoretically, as Chapter 1 details, the case revolves around the ways in which locating authority close to the service provision front-line can create opportunities for customization, for improving local-level motivation, and for utilizing local-level information.20 Empirically, the evidence from Chapters 4 and 6 on the performance of the Western Cape bureaucracy underscores the limits of hierarchy.

Consider first evidence on the contrasting educational performance of the Western Cape and Kenya. As Table 7.2 showed, even though Kenya has a per capita income less than a quarter that of South Africa’s, and thus substantially fewer resources per pupil (with not all of these differences captured in the explanatory variables), its education system systematically outperforms the Western Cape. Of course, outcomes depend on multiple influences—so focusing only on outcome measures as a basis for assessing performance is misleading.

Chapter 6 of this volume reports the results of a careful econometric effort to control for these other influences. The econometric strategy was to isolate a ‘Western Cape effect’ on educational outcomes (in the form of the coefficient of a dummy variable for the province) once other influences were controlled for—home background, socio-economic status, teacher qualifications, other teacher/classroom characteristics, plus a subset of governance indicators.21 With these controls, the coefficient of the Western Cape (p.194) dummy variable can be interpreted as being, in significant part an (unmeasured) ‘governance effect’.

Table 7.6 reports on a few of the results from the Chapter 6 effort. As the table shows, relative to Kenya22 the Western Cape effect is negative; even after controlling for a wide variety of influences. Given the evident robustness of the Western Cape bureaucracy, and the well-known unevenness of Kenya’s public management systems, it is highly unlikely that the performance gap can be explained by superior hierarchical management on the part of Kenya. Rather, the results could be interpreted as pointing to the possibility that it is in the ‘softer’ side of the governance of education—perhaps motivation on the part of teachers and other stakeholders, perhaps the patterns of participatory, horizontal governance—that Kenya has an advantage. The comparison with Kenya is explored further in Chapter 10.

Table 7.6. Grade 6 Mathematics SACMEQ Scores—The ‘Western Cape’ Effect (coefficient on Western Cape dummy variable)

Comparison country/region:



Kenya (Nairobi and Central)



Eastern Cape




Home background



Socio-economic status



Teacher test scores



Teacher/classroom characteristics



Parents contribute to school building and teaching materials


***significance at 1% level **significance at 5% level *significance at 10% level.

Source: Wills, Shepherd and Kotze (chapter 6)

The econometric comparison of the Western and Eastern Cape offers an intriguing added pointer as to the relevance of the softer side of governance, in this case as a partial substitute for weak bureaucratic capability. Unsurprisingly, as Table 7.6 shows, relative to the Eastern Cape, the Western Cape effect is significant and positive. However, the inclusion of ‘parental contribution to school building and teaching materials’ as an explanatory variable increases substantially the absolute value of the effect. Why? In the SACMEQ data series, 57 per cent of Eastern Cape parents (but only 13 per cent of Western Cape parents) assist with school building—and 65 per cent (but only (p.195) 18 per cent in the Western Cape) with school maintenance.23 Plausibly, parental participation serves as a partial institutional substitute for weaknesses in the Eastern Cape bureaucracy—with the true magnitude of the costs imposed on Eastern Cape children by weaknesses in the ECDoE only evident once the parental role is accounted for.

In sum, the econometric results suggest that bureaucracy need not be destiny—that there are other dimensions of the governance of education systems which can, as in the Eastern Cape, be partial institutional substitutes for relatively weak hierarchical bureaucratic capability, or (as seems likely in Kenya) can be complements, adding to the overall efficacy of the system. Chapters 8–10 explore the potential role of more horizontal, participatory mechanisms via an in-depth look at school-level governance dynamics, and their relation to educational outcomes.


Bibliography references:

Acemoglu, D. and Robinson, J. A. (2006) The Economic Origins of Dictatorship and Democracy. New York: Cambridge University Press.

Aghion, P. and Tirole, J. (1997) ‘Formal and real authority in organizations’, Journal of Political Economy 105, 1–29.

Cameron, R. (2009) ‘New Public Management reforms in the South African public service: 1999–2009, Journal of Public Administration 44(4.1), 910–42.

Cameron, R. (2010) ‘Redefining political-administrative relationships in South Africa’, International Review of Administrative Studies 76(4), 676–701.

Fiske, E. and Ladd, H. F. (2004) Education reform in post-apartheid South Africa. Washington DC: Brookings Institute Press.

Fukuyama, F. (2014) Political Order and Political Decay. New York: Farrar, Straus and Giroux.

Grindle, M. S. (2012) Jobs for the Boys: Patronage and the State in Comparative Perspective. Cambridge, MA: Harvard University Press.

Hughes, O. (2003) Public Management and Administration: An Introduction, 3rd edn. Basingstoke and New York: Palgrave Macmillan, 17–43.

Huntington, S. P. (1991) Political Order in Changing Societies. New Haven: Yale University Press.

Keefer, P. and Khemani, S. (2005) ‘Democracy, Public Expenditures and the Poor: Understanding Political Incentives for Providing Public Services’, World Bank Research Observer 20(1).

(p.196) Kelsall, T. and vom Hau, M. (2017) ‘Beyond Institutions: A Political Settlements Approach to Development’. University of Manchester.

Khan, M. (2010) ‘Political Settlements and the Governance of Growth-Enhancing Institutions’, mimeo.

Khan, M. and Jomo, K. S. (2000) Rents, Rent-seeking and Economic Development: Theory and Evidence in Asia. Cambridge: Cambridge University Press.

Khemani, S. (2016) Making Politics Work for Development: A Policy Research Report. Washington DC: World Bank.

Levy, B. (2014) Working with the Grain: Integrating Governance and Growth in Development Strategies. New York: Oxford University Press.

Khan, M., Hirsch, A., and Woolard, I. (2015) ‘Governance and inequality: Benchmarking and interpeting South Africa’s evolving political settlement’, ESID Working Paper No. 51. Manchester: The University of Manchester.

Lipset, S. M. (1959) ‘Some Social Requisites of Democracy: Economic Development and Political Legitimacy’, American Political Science Review 53, 69–105.

Lipsky, M. (2010) Street-level Bureaucracy: Dilemmas of the Individual in Public Service. New York: Russell Sage Foundation.

Lodge, T. (2014) ‘Neo-patrimonial politics in the ANC’, African Affairs 113/450, 1–23.

Manning N. (2001) ‘The Legacy of the New Public Management in developing countries’, International Review of Administrative Sciences 67(2), 297–312.

McLennan, A. and Fitzgerald, P. (eds) (1992) The Mount Grace papers. The New Public Administration Initiative and the Mount Grace Consultation. Johannesburg: Public and Development Management Programme, University of Witwatersrand.

Moore, B. (1966) The Social Origins of Dictatorship and Democracy. Boston: Beacon Press.

Pollitt, C. and Bouckaert, G. (2011) Public Management Reform. 2011. A Comparative Analysis, 3rd edn. Oxford: Oxford University Press.

RSA (2010) The SACMEQ III Project in South Africa: A Study of the Conditions of Schooling and the Quality of Education. Pretoria.

Sah, R. K. and Stiglitz, J. E. (1986) ‘The architecture of economic systems: Hierarchies and polyarchies’, American Economic Review 76(4), 716–27.

Schick, A. (1998) ‘Why Most Countries Should Not Try New Zealand Reforms’, The World Bank Research Observer 13(1), 123–31.

Schotte, S., Zizzamia, R., and Leibbrandt, M. (2017) ‘Social stratification, life chances and vulnerability to poverty in South Africa’, Southern Africa Labour and Development Research Unit, SALDRU Working Paper No. 208. University of Cape Town.

Schwella, E. (2000) ‘Paradigms: Context and competencies’. In F. Theron and E. Schwella (eds), The State of Public and Development Management in South Africa. The Mount Grace II Papers (pp. 33–43). Stellenbosch: University of Stellenbosch.

Scott, J. C. (1998) Seeing Like a State. New Haven: Yale University Press.

Southall, R. (2004) ‘Political change and the black middle class in democratic South Africa’, Canadian Journal of African Studies 38, 3.

Southern and East Africa Consortium for Monitoring Educational Quality (SACMEQ) (2007) http://www.sacmeq.org/.

(p.197) Streek, B. and Wickstead, R. (1981) Render under Kaiser: A Transkei Dossier. Johannesburg: Ravan Press.

The Presidency, Department of Performance Monitoring and Evaluation (2013) MPAT. Management Performance Assessment 2012/13 Final scores Education Sector Feedback. Pretoria.

Weber, M. (1922) Economy and Society. Berlin.

Wilson, J. Q. (1989) Bureaucracy: What Government Agencies Do and Why They Do It. New York: Basic Books. (p.198)


(1) The South African SACMEQ sample comprised 9,083 students drawn from 392 schools; sample size per province ranged from 900 to 1,500 observations (RSA, 2010).

(2) There is a strand of NPM which advocates giving managers greater autonomy in order to achieve results (Hughes, 2003). This version of management would perhaps fall in the hierarchical-personalized cell. It has also been argued that NPM in South Africa, particularly performance management, is less about management reform and more about asserting new forms of political control over the bureaucracy, (Cameron, 2009, 2010).

For an in-depth comparative analysis on public management NPM in high-income, OECD countries, and its limitations, see Pollitt and Bouckaert (2011). For similar studies in developing countries see Schick (1998) and Manning (2001).

(3) Grindle (2012: 32).

(4) Ranges are used to signal the heuristic nature of the allocation across cells.

(5) The ECDoE does more-or-less deliver on some core bureaucratic functions. Teachers are paid; some (but not all) schools receive textbooks; most (but not all) school buildings are maintained.

(6) For some analyses of the role of a rising middle class in pressing for greater public sector accountability, see Lipset (1959), Moore (1966), Huntington (1991), Acemoglu and Robinson, (2006), Fukuyama (2014). In recent work Tim Kelsall and Matthias vom Hau (2017) refine and extend the ‘middle-class’ hypothesis by introducing the concept of a ‘social configuration’ of power, comprising those groups that can threaten the hold on power of political leadership. They hypothesize that the broader the social foundation the more inclusive will be development outcomes.

(7) These differences in social class composition would, of course, be magnified (or diminished) if the provinces also differed in their (intra-province) patterns of relative inequality. But the limited available evidence on provincial-level Gini coefficients suggests that these differences are relatively small.

(8) The high proportion of the population that is ‘chronically poor’ is not an artifact of where the benchmark is set. For one thing, over half of the income of this group comes from social grants (pensions, child support and disability), set at less than $100 per month. For another, South Africa’s ‘chronically poor’ are notably poor when benchmarked against other middle-income countries. As Levy, Hirsch, and Woolard (2015) detail, in 2000 the poorest 40 per cent of South Africans accounted for only 5.5 per cent of expenditure, a starkly lower share than for the comparator countries (all of which have similar per capita incomes): Brazil (8.1 per cent), Mexico (11.4 per cent), Turkey (15.1 per cent) and Thailand (17.1 per cent). As of 2010, the South African share had risen to 6.9 per cent, still below the comparators.

(9) Acemoglu and Robinson (2006). Khemani (2016: chapter 5) reviews the literature on the impact of democratic political engagement on development outcomes.

(10) The DA elite comprised those in the upper tiers of business, the professions and the bureaucracy within the province. Cohesion did not come naturally among these elite sub-groups. At the dawn of democracy, there was a clear split between those with historical allegiance to the apartheid National Party, and those aligned with more liberal (but not revolutionary) white opposition parties. Over subsequent decades, these two factions largely merged—and were joined by smaller, independent (non-ANC) parties, whose allegiance was generally not ‘white’, and who historically had been very active in the struggle against apartheid.

(11) Here, again, Kelsall and vom Hau’s (2017) categories are useful. They distinguish between unipolar (hierarchically organized) and multipolar configurations of political power, and argue that a combination of unipolarity and a broad social foundation potentially can provide a strong platform for service provision. Khan (2010) makes a similar argument.

(12) The term is used in this way in Khan and Jomo’s (2000) analysis of rent networks in South Asia.

(13) Africans living in urban townships were under the control of the Department of Education and Training (DET). In the Western Cape, Africans were a minority and the number of DeT schools in the province were relatively small. There were also no Bantustans in the province. This meant that the province had to deal with only a modestly sized ‘deadweight’ of the most dysfunctional, control-oriented part of the apartheid state (Fiske and Ladd, 2004).

(14) McLennan and Fitzgerald (1992); Schwella (2000).

(15) Along with the two Bantustan bureaucracies, the third part of the Eastern Cape bureaucracy comprised some portions of the apartheid era Cape provincial bureaucracy—though the head office, and senior staff, had been located in Cape Town, and most transferred into the Western Cape’s bureaucracy.

(16) Streek and Wickstead (1981).

(17) Lodge (2014); Southall (2004).

(18) In the terminology of Kelsall and vom Hau, the Eastern Cape ANC was a multipolar rather than unipolar political party.

(19) For an in-depth analysis of how context shapes development policymaking and practice, including vis-à -vis public bureaucracies, see Levy (2014).

(20) Key contributions to literature on the value added of front-line autonomy include Sah and Stiglitz (1986), Aghion and Tirole (1997), Wilson (1989), Lipsky (2010) and Scott (1998).

(21) Each of these are composite measures, built on very detailed student, parent and school-specific information collected in the SACMEQ survey. Chapter 6 provides additional information on each of the underlying data points, including average scores for each in each of the locales they analyse.

(22) The comparison is specifically with results for Kenya’s Nairobi and Central provinces. As Wills et al. detail in Chapter 6, these provinces are reasonably similar in their SES demographic to the Western Cape. The statistical procedures break down when SES differences across the populations are too large.

(23) As Chapter 6 details, in Kenya, 55 per cent of parents assist with school buildings, and 42 per cent with maintenance.