Introduction

Because health systems are complex adaptive systems, they are widely acknowledged to be ‘path-dependent’ (Bloom, 2011; Marchal et al., 2016). Path-dependency theory says that ‘history matters’—decisions made in the past create conditions that shape contemporary decisions and therefore constrain the potential for change (Kay, 2005 p.553). As a result, stability prevails most of the time and policy change tends to be slow and incremental (Baumgartner & Jones, 2009; Gilson et al., 2018). However, path-dependency is an organising concept rather than an explanatory model—it cannot explain what happens because it does not suggest specific mechanisms that constrain change (Kay, 2005). Thus, while it is generally agreed that ‘history matters’ there is a need to better understand the mechanisms by which history matters (Pierson, 2004; Xu et al., 2019).

When dramatic policy change does occur, it is often enabled by major contextual transitions or upheavals that overcome the ‘stickiness’ of path-dependent systems (Sabatier, 1998; Weible et al., 2012). Regular political transitions (such as elections and the introduction of a new political administration) and dramatic political crises (such as coups or economic crises) help to overcome path-dependencies by introducing new actors with new ideas and delegitimizing old ideas, thereby making it easier to dismantle existing policies (Fischer, 2003; Horowitz, 1989; Kingdon, 1995; Reich, 1995). Social and political crises, in particular, spark intense ideological contestation in which actors’ perceptions of their interests can change, and new policy paradigms emerge (Hay, 2004).

Both Kingdon’s multiple streams model and Baumgartner’s punctuated equilibrium model suggest that policy change occurs as a result of contextual shifts that allow new ideas to gain primacy. The multiple streams model suggests that policy change occurs when the political reality and balance of power among decision-makers enables change, and, at the same time, the problem is well known and defined, and there is a policy idea that is recognised as a feasible solution (Kingdon, 1995). Thus, a policy window often emerges from a change in administration, or a new issue capturing the attention of officials (Kingdon, 1995). However, policy windows do not remain open indefinitely—usually because policy-makers’ attentions turn to other things (Kingdon, 1995). Similarly, the punctuated equilibrium model suggests that policy processes involve long periods of policy stability, which will be punctuated by periods of rapid change, sparked by “seemingly random initial events” such as the accumulation of unaddressed grievances, political transitions, focusing events such as a school shooting or a prominent technical study, or speeches by prominent figures (Baumgartner & Jones, 2001 p.7). The model explains stability in political systems as the result of institutional arrangements that grant certain actors access to the policy-making arena, and policy images (connecting powerful policy ideas to core political values that resonate with the public) that legitimate these institutions, creating negative feedback loops (Baumgartner & Jones, 2001, 2009). On the other hand, the punctuated equilibrium model suggests that ‘bursts of rapid change’ occur when new ideas ‘catch fire’—issues are reframed, policy-makers realise that other policy-makers have begun to understand the issue differently, and new interpretations quickly gain popularity and diffuse from one policy arena to another, creating positive feedback loops that catalyse radical change (Baumgartner & Jones, 2001; Baumgartner et al., 2008).

In the analysis of health policy processes more specifically, ideational factors—including values, ideologies, norms and beliefs—are increasingly recognised as important drivers of policy change (see for example Fox & Reich, 2015; Koon et al., 2016; Shiffman, 2009; Walt & Gilson, 1994). In this field, ideas are conceptualised as tools actors use to drive policy change—either as programmatic ideas about what the problem is and what should be done, or as frames and narratives with which to justify or legitimate policy proposals. For example, Russell et al. (2008) understand policy processes as the struggle for a dominant interpretation of policy problems, and consider how ideas are used in policy rhetoric to gain the support of other actors for a particular policy solution. Similarly, Harmer (2011), Fox and Reich (2015) and Vélez et al. (2020) demonstrate that ideas play an important role in the construction of policy problems and the creation of policy coalitions and advocacy groups, and that actors use ideas in advocacy to legitimate policy proposals and convince others to support them. Hassenteufel et al. (2010) explain health system transformation in France, Germany, Spain and the United Kingdom as a product of the influence of small, cohesive groups of policy elites, bound by a shared commitment to a programmatic idea, and able to determine the content of reforms through direct access to powerful political actors. Balabanova et al.'s (2004) analysis of health financing reform in Bulgaria suggests that the embrace of policy ideas from other health systems drove reforms. In addition, some health policy analyses explore the influence of frames and framing—ways of interpreting and making sense of the world—on how evidence is interpreted, which issues draw the attention of policy actors, and which policy solutions are considered most appropriate (Harmer, 2011; Koon et al., 2016; Parkhurst, 2012; Shiffman, 2009).

Clearly, in health policy analyses, the role of ideas is usually considered in relation to policy change. However, as discussed above, health systems tend to be change resistant, and health system reforms often fail (Villalobos Dintrans, 2019; Walt & Gilson, 1994; World Bank, 2003). This study explores the role of ideas in health systems’ resistance to change in apparent policy windows. To do so, we analyse two apparent policy windows in the South African National Health Insurance (NHI) policy process in which policy change was resisted, drawing on institutionalist perspectives to reveal how ideational factors can act as independent variables that constrain policy change and ensure path-dependence.

Background

We begin by outlining the institutionalist perspective and explaining how institutionalists account for the influence of ideas on policy processes—as necessary theoretical background—and then offering a brief overview of the South African context and the NHI policy process.

Institutionalism unpacked

As a dominant approach in political science, institutionalism explains policy outcomes with reference to how political institutions and policy procedures either facilitate or prevent different actors from exerting influence in policy processes (Hall & Taylor, 1996; Immergut & Comisso, 1992; Lieberman, 2002). Institutions are the formal and informal social rules, norms and conventions that govern individual conduct and inter-personal relations (Denzau & North, 2000; Hall & Taylor, 1996). On this account, institutions themselves are the product of ideas (of historical attitudes and policy decisions) and evolve at the hands of powerful actors who change old institutions in favour of new ideas (Béland, 2016; Steinmo, 2008; Wilsford, 1994). As Béland and Cox (2011) put it, “ideas are the foundation of institutions. As ideas give rise to people’s actions and as those actions form routines, the results are social institutions” (p.9).

The institutional environment in which actors make decisions includes organs of the state, bureaucracies, politicians and executives, industrial relations systems, financial systems, and policy legacies (Denzau & North, 2000; Hall & Taylor, 1996; Heclo, 1994). Institutions influence policy processes because, as Weir and Skocpol (1985) state, “the very theories, ideals, and goals articulated by experts or politicians are partially inspired by the administrative structures and capacities of the state structures in which they operate” (p.7). Accordingly, institutionalist approaches explain path-dependence as a product of the constraints existing social and political institutions place on contemporary policy processes (Hall & Taylor, 1996; Wilsford, 1994).

Two institutionalist approaches lend themselves to a more precise account of how ideas operate to constrain change—historical institutionalism and sociological institutionalism. Historical institutionalists emphasise the legacy of particular national experiences, and how this legacy shapes the organisation of social and political institutions, such that historical legacies constrain contemporary policy-making possibilities (Berman, 2013; Immergut & Comisso, 1992). From this perspective, institutions create path-dependencies because they shape the behaviour of actors in the system in two ways. Firstly, by setting the rules, procedures and ways of working in society, institutional arrangements shape social power relations such that some interest groups are able to participate in policy processes (usually working to preserve the status quo) while others are excluded (Hall & Taylor, 1996; Steinmo, 2008). Secondly, the institutional environment simplifies actor’s decision-making by suggesting to policy actors what kinds of policy alternatives will be feasible and considered appropriate (Campbell, 1998; Hall & Taylor, 1996; Steinmo, 2008). Thus, a key insight of historical institutionalism is its explanation of how institutions constrain the range of possible policy solutions that policy-makers consider (Campbell, 1998).

Sociological institutionalism is a constructivist approach that emphasises the social and cognitive features of institutions, which are conceptualised as a set of cultural norms and practices, shared cognitive frameworks, collective attitudes, social values, political ideologies and worldviews (Campbell, 1998; Finnemore, 1996; Hall & Taylor, 1996). These cultural ideas influence how actors make meaning of policy problems, and what kinds of policy solutions are in keeping with their self-concept and social legitimacy, thereby guiding collective decision-making and constraining the possibility for change (Finnemore, 1996; Hall, 1997; Hall & Taylor, 1996). Sociological institutionalism also helps to explain how exogenous ideas influence policy processes, because it reveals the social processes by which cultural norms and dominant ideas spread between contexts, either being transmitted by actors or diffusingFootnote 1 through epistemic communities (Campbell, 1998; Hall & Taylor, 1996). In a globalised world, ideas travel between contexts through international agencies with power to impose ideas from one context to another, or through a global policy discourse spread by academics, think-tanks and development organisations (Béland, 2009; Hall, 1997; Walt et al., 2008).

Campbell (1998) argues that comparing across institutionalist perspectives reveals a range of types of ideas at play in policy processes and presents a typology that distinguishes ideas along two dimensions: whether they are cognitive or normative ideas, and whether they are foreground or background ideas. Policy programmes and frames are (respectively cognitive and normative) foreground ideas. They are purposefully chosen and deliberately employed by policy actors to increase support for particular policy proposals (Campbell, 1998). Policy paradigms and public sentiments are (respectively cognitive and normative) background ideas. Unlike foreground ideas, background ideas are widely shared and slow to change, often to the extent that they are taken-for-granted, invisible and unquestioned (Berman, 1998; Campbell, 1998; Schmidt, 2008). While foreground ideas are used by policy actors in policy change processes, paradigms and public sentiments constrain change by delimiting the range of policy solutions that will be deemed feasible or appropriate (Campbell, 1998).

Taken together, therefore, historical institutionalism and sociological institutionalism help to explain how ideas can constrain policy change. Firstly, ideas can become embedded in tangible institutions such as bureaucracies, political parties, and organisations (Berman, 2001).Footnote 2 Secondly, ideas can become embedded in intangible institutions when they become “accepted or instinctual parts of the social world and hence are experienced as ‘natural’ or as part of ‘objective’ reality” (Berman, 2001 p.239). In this way, ideas become taken-for-granted by society or some sub-population within society, and are institutionalised in norms, cultures, and ideologies, which are widely shared and relatively durable (Berman, 2013; Bleich, 2002; Campbell, 1998). In addition, institutions have an important ‘norm-setting function’ (Rothstein, 1998). Through their daily interactions with institutions, people are habituated to certain ways of working, and ideas about what is feasible or appropriate are reinforced (Béland & Cox, 2011). In this way, institutions shape what future actors will regard as morally appropriate (Rothstein, 1998). Through institutionalisation, then, the influence of ideas outlasts the socio-political conditions that gave rise to them (Berman, 2013).

A brief history of the NHI policy process in South Africa

South Africa is an upper-middle income country with a progressive constitution that protects an expansive set of social and economic rights—however, it is also one of the most inequitable societies in the world, with half of South Africans living in poverty (Francis & Webster, 2019). In addition, the country faces a quadruple burden of disease, including HIV and TB, maternal mortality and morbidity, non-communicable diseases, and injury and violence (Coovadia et al., 2009; McKenzie et al., 2017). The country’s epidemiological, social and economic reality is shaped by its apartheid history wherein race-based geographic and political segregation and oppression were imposed through policy since the 1940s (Madore et al., 2015; McIntyre & McLeod, 2020). The African National Congress (ANC) led a programme of anti-apartheid civil-society mobilisation, which, combined with economic and international pressure, resulted in the dismantling of apartheid and the first democratic election in 1994 (Coovadia et al., 2009).

Under apartheid, as part of an effort to reinforce racial hierarchies, the health system was fragmented along racial lines, as well as between public and private sectors (McIntyre & McLeod, 2020; Pauw, 2021). Today, the public sector, funded through the general tax budget, serves the majority of the population, including those most in need of healthcare, and is severely under-resourced (Ataguba & McIntyre, 2018; Pauw, 2021). The private sector serves the wealthiest members of society (about 16% of the population) (McIntyre & McLeod, 2020). The private sector is mostly comprised of for-profit private providers (including general practitioners, hospitals), private funders (voluntary, not-for-profit health insurance associations known as medical schemes), and medical scheme administrators (for-profit companies that offer administrative services to medical schemes) (McIntyre & McLeod, 2020). Although total health expenditure as a percentage of gross domestic product is relatively high, a large proportion of expenditure is attributable to private health insurance, which excludes the poor, and government expenditure on health has remained consistently below the Abuja target of 15% (Ataguba & McIntyre, 2018). As a result, while stark inequities still characterise the contemporary health system, inequities in access and quality are increasingly related to class rather than race (McIntyre & McLeod, 2020).

Health system reform has been on the policy agenda since the early 1990s when the apartheid regime was dismantled. In the lead-up to the watershed 1994 democratic elections, there was widespread recognition that a significant overhaul of the inequitable, inefficient, fragmented and unsustainable health system was necessary, and ANC policy documents promised a National Health System that would incorporate compulsory Social Health Insurance (SHI) for the formally employed (ANC, 1992; Gilson, 2019; McIntyre et al., 2003). Since then, the policy process has been slow, contentious and highly political, and multiple iterations and variations of health system reform have been suggested (Gilson, 2019; Waterhouse et al., 2017). After an initial period of policy progress immediately after the 1994 election, the HIV pandemic pushed SHI off the policy agenda (Gilson, 2019; Gilson et al., 1999). When populist president Jacob Zuma was elected in 2009, the policy processes was reinvigorated, but once again, progress was slow (Gilson, 2019; Waterhouse et al., 2017). Most recently, the NHI Bill, tabled in Parliament in 2019, stipulated the introduction of a purchaser/provider split whereby health resources would be pooled in an NHI fund, which would purchase services from contracted providers on behalf of users (Mcintyre, 2019; van den Heever, 2019). This would drastically curtail the role of medical schemes, which would only be allowed to cover services excluded under the NHI – a provision which sparked considerable backlashFootnote 3 (Gray & Vawda, 2019; van den Heever, 2019).

Methods

In order to better understand the role of ideational factors in constraining policy change, we conducted a historical analysis of the South African NHI policy process, drawing on Capoccia’s (2015) critical juncture analysis approach. Critical juncture analysis was developed in political science for comparative-historical analysis, and provides a framework for studying moments of openness to change in path-dependent systems (Capoccia, 2015). The approach assumes that in the development of institutions, such as political regimes and public policy processes, there will be moments of openness to change (triggered by exogenousFootnote 4 shocks) in which, despite the presence of antecedent conditions influencing their decisions, actors can make choices that would set the institution on a new path or trajectory (Capoccia, 2015). Taking a broad temporal view, and analysing retrospectively, the analyst can use critical juncture analysis to explain the distal causes of the current institutional state of affairs (Capoccia, 2015) (see for example Xu et al., 2019).

The critical juncture analysis approach involves identifying apparent critical junctures by locating the exogenous shocks with which they are correlated, and ‘testing’ to establish whether major institutional change was possible in that moment—in other words whether actors had real choices and could have established a new institutional trajectory (Capoccia, 2015). Capoccia recommends in-depth analysis of candidate critical junctures to determine “whether structural antecedent conditions close off or trump choice” in these moments (Capoccia, 2015). This might be because structural conditions are such that there are no feasible policy alternatives, because structural conditions determine the actors who have decision-making power such that the outcome is a product of structural forces rather than actor choice, or because even when a choice is taken by a powerful actor, structural forces prevent actual change (Capoccia, 2015). One of the benefits of the critical juncture approach, therefore, is that it enables the analysis of ‘negative cases’ in which change was possible but did not occur.

In this study, we analysed policy windows as critical junctures in the policy process. To do so, we conducted a historical analysis, drawing on the retrospective literature review and discourse analysis presented in Whyle (2023) and Whyle & Olivier (2023). Documentary evidence was collected iteratively, beginning with a Google Scholar search for peer-reviewed literature on NHI, SHI or health system reform in South Africa to collect an initial body of secondary evidence. We then used author tracking, citation tracking and purposeful searches for material on particular events or processes referenced in the initial body of evidence to collect further evidence. During this ‘snowballing’ process we also sought primary evidence and grey literature such as policy documents, official communication, speeches by government officials, political manifestos, industry reports, submissions to parliament by industry bodies and civil society, and media articles. We continued to search for additional documentary evidence throughout the process of data analysis, until we felt that the information gathered was sufficient to develop a comprehensive timeline of the policy process and contextual influences, and to inform an analysis of policy rhetoric and the popular ideas influencing the policy process. A total of 623 items were analysed. Secondary evidence comprised 289 academic texts in fields spanning Anthropology, Development Studies, Global Health, Health Policy, Public Health, History, Economics, and Politics. Primary evidence comprised 334 items, including 176 media articles.

Primary and secondary evidence was analysed in the same way, by extracting relevant information into a data extraction sheet. The data extraction sheet enabled the extraction of data on events in the policy process, socio-political influences, health system context, and ideational variables (such as programmatic ideas, information on actor values and preferences, policy rhetoric and discourse). The data extraction sheet was organised chronologically to illuminate relationships between developments in the policy process, changing contextual realities, and ideational factors. Then, following the critical juncture analysis approach, we used the data extraction sheet to develop a timeline of the NHI policy process, identify policy windows, and explore the factors constraining change across those policy windows. The timeline captures the policy process from 1990 to 2019 in detail, but also includes the history of reform efforts since 1910. A summary of the timeline is available as supplementary material.

Following Capoccia (2015) we identified policy windows by looking for exogenous shocks and establishing which shocks had the largest impact on the power of actors to institute health policy reforms. In exploring the factors constraining change, we paid particular attention to the interests and beliefs of decision-makers, their choices, and the broader contextual (structural and ideational) factors that constrained either actor choices or actual change. The identification of the ideas at play in the policy process was based on a parallel analysis that used discourse analysis to identify the cognitive and normative ideas underlying NHI policy rhetoric from 1990 to 2019 (reported in Whyle, 2023) and informed by Campbell’s (1998) typology of cognitive and normative foreground and background ideas.

Findings

We present an analysis of two apparent policy windows in the South African NHI policy process—the first following immediately from the transition to democracy and spanning from 1994 to 1999, and the second following the election of Zuma as president of the ANC (and later the country), spanning from 2007 to 2018. Both the 1994 transition to democratic governance and the 2007–2009 election of populist president Zuma are the kinds of upheavals that might be expected to give rise to a policy window. However, neither the transition to democracy nor the tenure of President Zuma, despite spurring much policy-making activity, resulted in major progression towards the implementation of NHI. We examine these two policy windows in order to gain a clearer understanding of how ideas interact with contextual factors to resist change in complex systems.

Policy window one: 1994–1999

In 1994, the policy problem was clear. The new government inherited a health system that was under-resourced, fragmented and inequitable (Gilson et al., 1999; McIntyre et al., 1995). The inequity and fragmentation was the legacy of apartheid-era policy decisions, including the racialisation of the public health system and the deregulation and growth of the private health sector (Gilson et al., 1999). Beginning in the 1980s, the previous government, led by the National Party, had systematically privatised and deregulated the health sector and curtailed public sector spending, resulting in a ‘brain-drain’ of human resources from the public sector to the private sector (McIntyre et al., 1995; Price, 1989). In addition, deregulation meant that medical scheme contributions were increasingly unaffordable, and the 1993 Medical Schemes Amendment Act (pushed through in the dying days of National Party rule) meant medical schemes could reject applicant members on the grounds of HIV status or age, resulting in many people losing coverage and increasing the burden of care on the public sector (McIntyre et al., 1995; Price, 1994).

The ANC’s victory in the 1994 election served as an exogenous shock that opened an apparent policy window. The transition to democracy enfranchised huge numbers of working and unemployed poor that would constitute the beneficiaries of universalist health system reform and gave political power to the ANC—a party with pre-existing commitments to redistributive and socialist development policy generally, and to universalist health system reform specifically (Gilson et al., 1999; McIntyre et al., 2003; Peet, 2002). The change of government also brought the introduction of a progressive NHI advocate as the new Minister of Health, Minister Nkosazana Dlamini-Zuma. Minister Dlamini-Zuma was more radical than other newly appointed ministers in seeking redistribution and saw health system reform as a way to redress the inequities of apartheid (Bond, 1999; Gilson et al., 1999).

Under Dlamini-Zuma’s leadership, there was a flurry of NHI-related policy activity, as she steered the process towards a progressive version of NHI (Gilson et al., 1999; Thomas & Gilson, 2004). This activity included the publication of the ANC’s National Health Plan shortly after the election in May 1994, the establishment of the Health Care Finance Committee (HCFC) to examine the feasibility of an NHI in June of that year, the Committee of Inquiry into NHI in 1995, and the SHI Working Group between 1994 and 1997 (Gilson et al., 1999; McIntyre et al., 2003). The HCFC included local and international analysts and private sector stakeholders (including medical scheme industry representatives), but excluded trade unions on the basis that the mandate of the committee was ‘technical’ (Gilson et al., 1999). The Committee of Inquiry into NHI included a similar mix of local and international analysts, private sector representatives as well as Treasury officials (Gilson et al., 1999; Thomas & Gilson, 2004). The more contained Working Group included only DoH staff and local analysists (Thomas & Gilson, 2004).

The HCFC report presented three potential models for health system reform, differentiated by beneficiaries and benefit packages. The most radical, which came to be known as the Deeble modelFootnote 5 included the nationalisation of private doctors, the elimination of medical schemes and ‘tiering’ (differentiated services for the insured and the uninsured), and universal access to primary healthcare through mandatory coverage under a centralised funding mechanisms (Gilson et al., 1999). The Deeble model was debated by both the Committee of Inquiry and the Working Group, but ultimately the Working Group recommended a moderate SHI that would be restricted to the formally employed and did not involve cross-subsidisation between income groups, indicating a concession to Treasury’s concern that the middle-classes should not be ‘over-taxed’ (Gilson et al., 1999; McIntyre et al., 2003). However, the Working Group proposals did not result in any policy action (Gilson et al., 1999).

By 1999 the opportunity for radical change had passed, and the policy window was closed. In 1996 the (ostensibly) pro-poor and welfarist Reconstruction and Development Plan (RDP) macro-economic strategy was replaced by the neoliberal Growth, Employment and Redistribution (GEAR) policy (Baker, 2010; Pillay & Bond, 1995). While Treasury officials were already convinced that NHI was out-of-step with the macro-economic strategy laid out in the RDP, the adoption of GEAR solidified Treasury’s resistance to tax-funded health system reform and further stagnated public spending on healthcare (Baker, 2010; Gilson et al., 1999). In addition, around this time the HIV/AIDS epidemic began to command the attention of policy-makers and the public (Gilson, 2019). AIDS first began to emerge in South Africa in the 1980s, but by 1998 23% of the population was HIV-positive (Marks, 2002; Schneider, 1998). A series of AIDS-related corruption scandals, beginning in 1996, and the government’s persistent failure to respond appropriately to the massive public health emergency, sparked massive public and civil-society outrage (Nattrass, 2008; Schneider, 1998). Finally, In 1999, Minister Dlamini-Zuma was replaced by Minister Tshabalala-Msimang and the NHI policy process stalled (Thomas & Gilson, 2004; Waterhouse et al., 2017).

What constrained change in the 1994–1999 policy window?

Despite clear evidence of major challenges in the public sector and stark inequities between the public and private sectors, and despite a change of government that empowered a political party long committed to universalist healthcare reforms, major policy change was not achieved in the 1994–1999 period. Ideational factors—born out of the pre-1994 political and social climate—underlie many of the constraints that served to resist change and ensure path-dependence.

Firstly, at this time, neoliberal ideas were hegemonic globally (Centeno & Cohen, 2012). Neoliberalism emerged as a system of ideas, at first about economics and social welfare in the 1980s, and diffused throughout the world through the influence of international organisations like the World Bank and the IMF (Harvey, 2005; Mudge, 2008). Harvey (2005) defines neoliberalism as a political economic theory that proposes human wellbeing is best advanced by “liberating individual entrepreneurial freedoms” in an institutional context of free markets and strong property rights (p.2). This approach is based on the assumption that the state does not have the necessary information or expertise to intervene effectively, and is subject to the influence of powerful actors that bias its decision-making (Harvey, 2005). Thus, neoliberalism suggests a minimal role for the state (restricted to protecting this institutional context and providing or subsidising social services only for the poor), favours market-based solutions to social ills, and justifies the transfer of power from the state to the market and private capital (Centeno & Cohen, 2012; Harvey, 2005).

In South Africa, the apartheid government used neoliberal principles to justify privatisation of health services and a reduction in public health spending in the 1980s (Nattrass, 1994; Price, 1989). However, neoliberal ideas also had a continued influence in post-apartheid South Africa (Chipkin et al., 2018; Seekings & Nattrass, 2015). While the ANC had a long history of alliance with local and international socialist organisations and socialist-informed development and economic policies, the 1994 transition to democratic governance took place in a global context in which neoliberal ideas were omnipresent (Centeno & Cohen, 2012; Peet, 2002; Williams & Taylor, 2000). Neoliberal ideas both softened commitment within the ANC to radically redistributive policies and informed the positions and recommendations of technical experts consulted in this period. Even prior to 1994, neoliberal ideas had begun to shape the thinking of many ANC decision-makers, including through the direct influence of global actors like the IMF and the World Bank (Bond, 2014b). As a result, the ANC, which might otherwise have been ideologically unified was, in fact, divided between two schools of thought: on one hand a liberal commitment to free trade and small government, and on the other, an anti-capitalist commitment to redistribution and nationalisation (Glaser, 1997; Price, 1994). This ideological divide within the ANC meant that when options for health system reform were offered by the committees and Working Groups initiated by Minister Dlamini-Zuma, while it was clear Minister Dlamini-Zuma supported the Deeble Option, there was not consensus support from the Party.

The neoliberal ideational context also informed the recommendations of the technical experts represented in the various deliberative committees established to move the policy process forward. Thomas and Gilson (2004) suggest that the limited progress in this period was a result of a disjuncture between what was feasible and what was desirable—experts on the committees failed to consider what would be acceptable to policy-makers, and policy-makers failed to delineate for the experts what would be politically feasible. When the Committee of Inquiry into NHI was explicitly asked to consider a policy option acceptable to the Minister (an NHI), the committee demanded that the terms of reference be expanded to include more economically feasible options (Thomas & Gilson, 2004). The focus on what was economically feasible was a response to Treasury’s concern that funding an NHI through general taxation was not in keeping with the country’s macro-economic policy, that the burden on taxpayers should not be increased, and that growth in the for-profit health sector should be encouraged (Gilson et al., 1999).

Both the macro-economic policies in place in this period, RDP and GEAR, however, were products of neoliberal hegemony. While the RDP was originally influenced by trade union allies of the ANC and contained radically leftist ideas, the version of the policy codified and implemented after the election reflected the influence of global financial institutions, themselves committed to the tenets of neoliberalism, and specified a reduction in public-sector spending (Gilson et al., 1999; Pillay & Bond, 1995). GEAR was more explicitly neoliberal; it prioritised economic growth over redistributive social policy and the interests of capital over labour (Baker, 2010; Gilson et al., 1999). Furthermore, Waterhouse et al. (2017) point out that while Treasury was no doubt ideologically opposed to NHI, it is also true that GEAR reflected the broad position of the government, not only those within Treasury. Thus, ideas, here specifically the principles of neoliberalism, informed both the decisions of technical experts and Treasury members, and also informed the macro-economic policies against which the feasibility of health system reform proposals were judged.

Furthermore, the policy process was also constrained by a pervasive idea among experts in the deliberative committees, that, as a result of the size and power of the private sector, the political feasibility of health system reform depended on opportunities for the continued involvement of the private sector. Having grown steadily throughout the 1980s, the private sector was judged to have the ‘political strength’ to successfully oppose reforms, resulting in a shift in focus from purely-public, tax-funded models to mandatory insurance models that combined public and private provision (McIntyre & Dorrington, 1990; Thomas & Gilson, 2004; Waterhouse et al., 2017). The emergence of hospital groups—networks of hospitals owned by a single company—and their domination of the private hospital industry helped to consolidate their power (McIntyre et al., 1995). Post-1994 strengthening government control of the private sector was no longer considered a primary goal of health system reform and the Centre for Health Policy, a proponent of NHI, argued that “the private sector was simply too extensive to disappear and so the only politically feasible approach was to work with it” (Gilson et al., 1999 p.43; see also McIntyre et al., 2006). Thus, while the Minister opposed proposals that allowed for continued medical scheme coverage, and therefore tiering, and regarded for-profit healthcare as ‘repulsive’, reform options that would align with her views were repeatedly dismissed as politically infeasible (Gilson et al., 1999, 2003). This diversion of worldviews between experts and the Minister, while partly a reflection of the actual growth and consolidation of the private sector, can also be viewed as the influence of neoliberal ideas about the appropriateness of private provision.

Secondly, Minister Dlamini-Zuma’s beliefs had important historical tributaries of their own. The idea for universal health system reform in South Africa is rooted in a history that stretches as far back as the 1940s and includes policy proposals of the progressive anti-apartheid movement in the 1980s (Whyle & Olivier, 2023). In the 1940s, soon-to-be Minister of Health and Member of Parliament, Dr Henry Gluckman proposed health system reforms that included a dramatically reduced role for the private sector, based on a belief that it was the responsibility of the state to provide healthcare, that it was necessary for the medical profession to be socialised (Gluckman, 1946), and that private healthcare should be gradually phased out (Price, 1989; Van Niekerk, 2003). Gluckman’s proposal was influenced by the revolutionary zeal of post-war Britain, and the idea that healthcare should be ‘socialised,’ can be understood in that context (Digby, 2008; Gluckman, 1946).

Although the advent of apartheid prevented the institution of Gluckman’s National Health System, the ideas embodied in the Gluckman report informed the proposals of the progressive health movement in the 1980s (Gilson et al., 1999). At that time, the appropriate role for the private sector was a major point of contention—with some anti-apartheid allies, deeply distrustful of the for-profit health sector, drawing on Gluckman’s proposal to argue for a British-style National Health System and the nationalisation of private healthcare, while others argued that the scale and power of the private sector made a National Health System infeasible (Gilson et al., 1999; Paremoer, 2020). However, the ANC’s longstanding alliance with the South African Community Party and the Congress of South African Trade Unions (COSATU)Footnote 6 helped to ensure that the idea of socialised medicine informed the ANC’s proposals for health system reform put forward in the early 1990s (Baker, 2010; Coovadia et al., 2009). In addition to Minister Dlamini-Zuma’s views on for-profit healthcare, these ideas influenced COSATU’s stance on NHI. COSATU opposed the health system reform proposals in the 94–99 period on the grounds that multi-payer models would not ensure cross-subsidisation from the rich to the poor and would, in fact, increase the financial burden on the working poor (COSATU, 2000; Waterhouse et al., 2017). COSATU’s opposition contributed to the stalling of the policy process after 1997 (Thomas & Gilson, 2004).

In short, while the transition to democracy rebalanced the distribution of political power in the country by enfranchising many who would benefit from universalist health system reform and imbued new actors with decision-making power, ideas about the appropriate role of the private sector in health systems and the prioritisation of fiscal concerns over equity concerns constrained change. These ideas, being borne out of the pre-1994 context, ensured path-dependence across the policy window.

Policy window two: 2007–2018

The 2007 election of President Zuma—a populist president associated with the rejection of the neoliberal approach to governance that was a major constraint of health system reform post 1994 (Hart, 2014; Von Holdt, 2019) —would also have been expected to create the policy window needed to achieve implementation of NHI. By 2007, the health sector was suffering from the effects of the implementation of GEAR, which constrained health spending and hindered regulation of the private sector (Baker, 2010; Doherty & McIntyre, 2015). In 2005 GEAR was replaced by the less neoliberal Accelerated and Shared Growth Initiative for South Africa (ASGISA) (Barolsky, 2013; Francis & Webster, 2019). However, the health system was still characterised by a mal-distribution of human and financial resources that favoured the rich, including through direct subsidisation of private sector from public budget (McIntyre et al., 2006). At the same time high costs in the private sector meant that medical scheme membership was declining, and costs were continuing to rise as a result of over-servicing and high administration fees (McIntyre et al., 2006; McIntyre & Van den Heever, 2007). Attempts to control costs by implementing standardised tariffs had been stymied by the Competition Commission’s decision that the practice was anti-competitive (Berger & Hassim, 2010; Waterhouse et al., 2017). In addition, beginning in 2005, the World Health Organisation (WHO) began promoting the concept of Universal Health Coverage (UHC), including health system reform with purchaser-provider split and public–private mix in provision (Smithers & Waitzkin, 2022).

In this context, Zuma’s ascendence to the presidency of the ANC in 2007, and of the country in 2009, spurred a flurry of policy-making activity and public debate in relation to NHI (Gilson, 2019). The ANC committed itself to the urgent implementation of NHI, an NHI Task Team was appointed, and the Party’s 2009 election manifesto promised the implementation of a NHI (Madore et al., 2015; McLeod, 2009; Pillay & Skordis-Worrall, 2013). After assuming the presidency, Zuma appointed Dr Aaron Motsoaledi as Minister of Health, a passionate advocate of NHI (Gilson, 2019; Waterhouse et al., 2017). Under Motsoaledi, an NHI Ministerial Advisory Committee was established to advise the Minister on NHI policy and legislation, the NHI Green Paper was released in 2011, in 2012 ten NHI pilot sites were established, and in 2014 the Health Market Inquiry was initiated by the Competition Commission (Gilson, 2019; Madore et al., 2015; McLeod, 2009; RSA DoH, 2011). Waterhouse et al. (2017) suggests that efforts to move the policy process forward redoubled in 2015 as a result of the ANC’s poor performance in the 2014 local government elections. The Department of Health established six workstreams to provide technical support in the development of NHI policy, and drafts of the NHI White paper were released in 2015 and 2017 (RSA, 2015, 2017; Waterhouse et al., 2017). UHC was incorporated into policy documents,Footnote 7 and Motsoaledi sometimes equated NHI with UHC, using the WHO’s support for UHC to justify NHI (Madore et al., 2015; Motsoaledi, 2012). A few months after Zuma was ousted in 2018, a draft NHI Bill was gazetted (RSA, 2018). However, throughout this period, Treasury repeatedly delayed publishing funding plans for the NHI and continued to push for a multi-payer model (Madore et al., 2015; Waterhouse et al., 2017). When Motsoaledi’s tenure ended in 2019, the NHI Bill was before the National Assembly (Mcintyre, 2019; RSA, 2019). The Bill was passed by the National Assembly in June 2023 and by the National Council of Provinces in December of that year (RSA Parliament, 2023a, 2023b). In May 2024, in the context of dramatic cuts to the national health budget and just two weeks before the general elections, President Cyril Ramaphosa signed the Bill into law. However, the ANC lost its parliamentary majority in that election and was forced to create a coalition government with four other parties, all with varied—and in some cases actively oppositional—positions on the NHI (Mahlaka, 2024; van Niekerk & Pretorius, 2024). Whether NHI will be implemented in the near future remains to be seen, but by 2019 with Zuma having left office, and Motsoaledi’s tenure as Minister ended, the policy window seemed to have closed.

What constrained change in the 2007–2018 policy window?

Once again, despite clear evidence of a policy problem, and a change of government that realigned the balance of political power in favour of health system reform, ideational factors contributed to constraining change in this period. In particular, two ideas, closely linked to neoliberalism, seemed to increase contestation and constrain change in this period: firstly, the idea that the state cannot be trusted to manage healthcare resources, and secondly the idea that the interests and freedoms of the country’s tax-base (the middle-class), need to be safeguarded. By 2007, the NHI policy idea was beginning to solidify into a proposal for an NHI characterised by centralised financing and purchasing by a NHI authority, and a purchaser-provider split that would allow health services to be purchased from public and private providers, combined with a funding injection to the public sector (van den Heever, 2016). The 2019 Bill specified contracting accredited public and private providers for primary care, but leaves the role of private hospitals unclear (Gray & Vawda, 2019).

Much of the contestation in this period centred on the replacing of private medical scheme coverage with NHI coverage,Footnote 8 the role of private providers, the inadequate quality of healthcare in the public sector, and whether the state could be trusted to manage a centralised funding pool (Gilson, 2019; Madore et al., 2015; Medical Brief, 2022). While there were very real service delivery challenges in the public sector, there is also evidence to suggest that perceptions of low quality care in that sector were not based on direct experience, and quality issues in the private sector were largely ignored (CCSA, 2018; Maseko & Harris, 2018). Nonetheless, the idea that care provided by the state was of low quality informed much public opposition to NHI (McIntyre et al., 2009).

In addition, Zuma’s presidency was marred by a series of grand-scale corruption scandals and governance failures that served to further undermine trust in the state. Zuma was charged with corruption shortly after being appointed president of the ANC in 2007, but successfully pressured the National Prosecuting Authority to protect him from prosecution (Koelble, 2017; Von Holdt, 2019). In 2008, energy provision became a major issue when South Africa’s parastatal energy supplier was forced to introduce a system of planned outages, known as load-shedding, as a result of being unable to produce sufficient energy (Bowman, 2020). Eventually, it was revealed that load-shedding was, in part, a consequence of Zuma having appointed corrupt individuals to Eskom’s board for his personal gain (Bowman, 2020; Budhram, 2019). In 2012, the killing of 34 striking miners by police was broadcast on television, recalling apartheid-era police-violence and generating significant public attention (Bond, 2014a; Forrest, 2015). By 2016, the idea that the state had been ‘captured’ by corrupt officials and foreign businessmen was firmly cemented in the public consciousness (Budhram, 2019; Von Holdt, 2019). As a consequence of these, and other examples of high-level corruption, trust in the state declined significantly (Potgieter, 2017).

While amplified by contemporary events, the idea that the state cannot be trusted had historical roots. Firstly, the idea that the state lacks the capacity to adequately deliver health services and manage health resources is a central tenet of neoliberalism, as discussed above (Packard, 2016; Rushton & Williams, 2012). As such, neoliberal ideas were not only antithetical to NHI insofar as NHI would entail a larger role for the state in managing the health system and an infringement into the market for healthcare that currently operates, but also insofar as NHI requires a belief that the state can and will manage healthcare funds effectively, efficiently and impartially. Secondly, Nattrass and Seekings (2010) argue that grand-scale corruption, and the accompanying loss of trust in the state, can be understood as a consequence of the close relationship between ANC politicians and business elites that began in the late-1980s when South African business sought to position themselves as anti-apartheid allies. Thirdly, negative popular perceptions of service delivery in the public sector were inevitably rooted in the fragmentation and mal-distribution of resources of the Apartheid era, but also likely arose from policy decisions made in the early days of the new democracy—for example the decision to implement the Free Care policy in without a corresponding increase in budget or human resources (Charney, 1995; Gilson et al., 1999).

The idea that the state cannot be trusted to manage health resources complicated the position of important actors, as is evident in the subtle shift in policy position of COSATU. In the Zuma-era COSATU was generally supportive of NHIFootnote 9 (Madore et al., 2015; Waterhouse et al., 2017). COSATU championed an NHI that was a “state-mandated, state-administered system in which a single authority organises health finance aimed at ensuring that all persons, irrespective of financial status, have free access to healthcare at the point of service” (Vavi, 2008). In addition, COSATU’s, 2011 submission on the Green Paper endorsed the dissolution of medical schemes (COSATU, 2011). As such, the current NHI proposals are very close to COSATU’s preferred model.

However, during the political contestation that preceded Zuma’s eventual resignation, COSATU (possibly motivated by a desire to demonstrate support for Zuma) condemned Treasury, the Ministry of Health, and Minister Motsoaledi for supporting a multi-payer model in the case of Treasury, and for ‘selling-out’ the NHI to private interests in the case of Motsoaledi and the Ministry (COSATU, 2016; Pamla, 2016; Waterhouse et al., 2017). COSATU, viewing NHI as a mechanism for radical redistribution, had long been vocally opposed to the involvement of the private sector in the NHI (at times advocating for the incorporation of all private health resources into the public sector) and to any kind of tiering within the NHI, and committed to an expansion of public health service delivery (Thomas & Gilson, 2004; Waterhouse et al., 2017).Footnote 10 In 2016, COSATU’s statements began to reflect a stronger stance against for-profit healthcare in all its forms (Dlamini, 2017; Staff reporter, 2017). Thus, while COSATU continued to push the ANC to speedily implement NHI, it also continued to question the role of the private sector in the NHI (see for example Dlamini, 2017; Pamla, 2016), as over the course of his tenure Motsoaledi increasingly signalled willingness to engage with the private sector and accommodate the interests of private providers in the policy itselfFootnote 11 (Waterhouse et al., 2017). After Motsoaledi was replaced by Minister Zweli Mkhize in 2019, COSATU resumed its public unequivocal support for NHI, including calling for its urgent implementation (COSATU, 2019). Since then, however, even COSATU has joined the chorus of stakeholders questioning the state’s capacity to manage the NHI fund and citing dysfunctional service delivery by the state (Medical Brief, 2022).

In addition to the idea that the state cannot be trusted, ideas about what constitutes an appropriate tax-burden and an appropriate infringement on the freedoms of taxpayers continued to inform Treasury’s resistance to a single-payer NHI and constrain the potential for change in the 2007–2018 period (Gilson, 2019; Madore et al., 2015). As discussed above, in 1997 Treasury expressed concerns that an increased tax-burden on the middle-class to finance NHI would not be ‘fair’ as they were already over-burdened (Gilson et al., 2003; McIntyre & Van den Heever, 2007; Thomas & Gilson, 2004). Similarly in 2004, a Treasury official argued that a redistributive NHI could not work in the context of increasing medical scheme membership fees and declining benefits (Dawes, 2004). In addition, between 2011 and 2015 Treasury argued for a multi-payer NHI that would allow medical scheme members to retain their medical scheme membership and contribute to NHI through a ‘solidarity tax,’ on the grounds that they were already ‘accustomed’ to premium benefit packages and high per capita expenditure (COSATU, 2016; Madore et al., 2015). A senior Department of Health official suggested that there were individuals within Treasury who opposed NHI because they felt that taxpayers should not be ‘offended’ (Waterhouse et al., 2017). Paremoer (2021) points out that contemporary NHI proposals ask the minority of the population who have grown used to using the private sector to expose themselves to the ‘lived experience’ of the majority— “entrusting the state with their basic needs” (p.449). This ‘minority’ consists of those who can afford medical scheme membership, and so also represents a significant portion of the country’s tax-base. Treasury’s position on NHI proposals and public justifications thereof reveal the extent to which a certain segment of the South African population had become habituated to accessing healthcare in the private sector and suggest that the interests of this segment of the population is given priority in Treasury’s decision-making.

The dominance of the idea that the interests of the wealthy need to be protected might also have been a consequence of the hegemony of neoliberalism. As noted above, neoliberalism emphasises the individual’s freedom to pursue their interests and goals in the institutional context of the free market (Cardona, 2021; Harvey, 2005). This entails what Paremoer (2020) calls ‘economic citizenship’—citizens as self-reliant individuals “unhindered by government regulation aimed at the promotion of social welfare” (p.6). In addition, neoliberalism tends to prioritise the interests of capital and business over, for example, the interests of labour and the poor (Seekings & Nattrass, 2015). Despite the transition from GEAR to ASGISA, a neoliberal worldview continued to inform economic governance (Barolsky, 2013). Thus, in the post-apartheid state, social services, including public healthcare, were still largely considered to be for the very poor, and middle-class citizens were not considered as beneficiaries of the welfare state, and therefore should not be subject to infringement by the welfare state (Paremoer, 2020, 2021). Treasury’s hindering of the policy process, including repeatedly delaying a promised NHI funding plan, likely reflected not only economic realities, but also a general prioritisation of the concerns of the middle-class, ideas about what is ‘fair’ and appropriate vis-à-vis taxpayers and medical scheme members, and a reluctance to alienate the private sector (Surender, 2014; Waterhouse et al., 2017).

This analysis demonstrates that ideas, shaped by history, can help to constrain change, even across policy windows, or apparent opportunities for change. A summary of the key ideas, their historical tributaries and their role in constraining change is presented in Table 1. In both the policy windows described here, ideas informed by neoliberalism (such as private provision of healthcare, a limited role for government in financing and provision, and the prioritisation of markets and capital), clashed against ideas informed by socialist governance regimes, embedded in the history of the ANC (such as discomfort with for-profit healthcare, and redistribution through social welfare). All these ideas are connected to South Africa’s particular social and political history, and also to contemporary events and contextual realities. Thus, this analysis suggests that ideas help to explain path-dependence because ideas arise out of historical circumstances and continue to influence policy processes thereafter.

Table 1 Summary of ideas, their historical tributaries and their role in constraining change
Full size table

Discussion

We have demonstrated that the South African NHI policy process is an example of path-dependence and change resistance despite apparent policy windows and suggested that ideas played an important role in constraining change. In this section, we draw on institutionalism as a political science theory—including the diffusion of ideas across contexts, the (tangible and intangible) institutionalisation of ideas over time, and the power of institutions to delimit the range of policy solutions considered—to develop an explanation for the power of ideas to constrain change.

Historical and sociological institutionalism both suggest that ideas can, over time, become institutionalised—in other words, cemented in social and cultural institutions such that they become unsaid or taken-for-granted assumptions underlying policy rhetoric and choices (Berman, 1998, 2001; Campbell, 1998). In the South African NHI policy experience, these ideational dynamics are evident in two related sets of ideas: neoliberal ideas about governance, and the appropriateness of for-profit healthcare.

Institutionalism suggests that ideas diffuse across contexts through actors and epistemic communities (Campbell, 1998; Hall & Taylor, 1996). In this case, in the 1994–1999 period, neoliberal ideas were taken up both by the apartheid government and by the ANC, and cemented in tangible and intangible institutions. Neoliberal ideas were accepted by the apartheid government, which used them to justify the privatisation and deregulation of healthcare (Hilton, 1988; Price, 1994). As a result, as per institutionalist theory (Béland & Cox, 2011; Berman, 2001), neoliberal ideas about health service provision were institutionalised in private healthcare as a social institution—the network of funders, administrators and for-profit providers that would become an influential actor in the policy experience, and in the daily practices of privately delivered healthcare that would become the norm for many South Africans.

Furthermore, as sociological institutionalist theory suggests (Béland, 2009; Hall, 1997), neoliberal ideas were transposed into the ANC directly through global institutions, and indirectly through the pressures associated with operating in a global context in which neoliberal ideas were dominant (Bond, 2014b; Cronin, 2020; Pillay & Bond, 1995). As discussed, this ideological shift resulted in a lack of party support for Minister Dlamini-Zuma’s preferred policy option. In addition, however, neoliberal ideas of economic governance were tangibly institutionalised into macro-economic policy—most obviously GEAR—that severely limited public spending (Baker, 2010; Bond, 2014b; Gilson et al., 1999). In turn, GEAR justified the opposition of key actors, primarily Treasury, to health system reform (Baker, 2010; Gilson et al., 1999). Furthermore, as argued above, the hegemony of these ideas influenced the perspective of experts in the various deliberative fora established in this period, with the result that these committees conceded to Treasury’s view that that a progressive NHI was not financially feasible because increased taxation was inappropriate (Gilson et al., 1999; McIntyre et al., 2003). In the second policy window, although GEAR had given way to the less explicitly neoliberal ASGISA (and neoliberal ideas were no longer tangibly institutionalised in this way), it seems that these ideas had been intangibly institutionalised within Treasury (see Barolsky, 2013)—with the result that Treasury opposed the reform models being suggested and constrained policy change.

A second set of ideas also became institutionalised over the course of this policy experience—ideas about the appropriate role of for-profit actors in healthcare provision. The private health sector in South Africa is a social institution that is a product of policy decisions taken in the 1980s, motivated by political imperatives, and justified by neoliberal ideas about the appropriate role of the state in healthcare (McIntyre & McLeod, 2020; Price, 1994). In the early policy window, Minister Dlamini-Zuma was against for-profit healthcare on ideological grounds (Gilson et al., 2003). Conversely, the idea that the private sector was simply too big to be curtailed—which was accepted by experts in the deliberative fora of the 1994–1999 period—may have been a ‘political reality’ but also inevitably reflects the hegemony of the idea that for-profit healthcare is appropriate and inevitable. As Centeno and Cohen (2012) put it, “causality flows from the reality of economic life as well as from its interpretation” (p.328).

Furthermore, over the course of the policy experience, as private healthcare provision continued to shape the experience of healthcare for middle-class South Africans, the idea that for-profit healthcare is appropriate became intangibly institutionalised within this population. This would explain Paremoer’s (2021) suggestion that public services are perceived as being exclusively for the poor. The cultural hegemony of this idea is arguably evident in the report that Treasury felt that the standard of care and (inequitably) higher per-capital expenditure experienced by medical scheme members should be protected, as well as in the Competition Commission’s ruling against tariffs (Berger & Hassim, 2010; Madore et al., 2015; Waterhouse et al., 2017). The related and pervasive ideas, discussed above, that public healthcare is of low-quality and that the state cannot be trusted to manage funds or provide services no doubt reinforce the cultural hegemony of the idea that the private sector is the most appropriate mechanism for the delivery of health services, as does the tendency of the media to amplify the concerns of that segment of the population habituated to private healthcare provision (Gilson, 2019; Waterhouse et al., 2017). This suggests that these ideas, tangibly institutionalised in the private health sector, are becoming intangibly institutionalised—in other words are becoming background normative and cognitive assumptions, that will be very difficult to change, and will have significant consequences for future health system reform efforts—as per Campbell and the sociological institutionalists (Berman, 2001; Campbell, 1998).

Conclusion

Health systems are complex social systems, embedded in, and open to influence by, social and political contexts, which influence policy processes in a myriad of ways (Collins et al., 1999; Gilson, 2012). This analysis reveals how ideas—including normative commitments and cognitive beliefs—can constrain change and ensure path-dependence by becoming tangibly and intangibly institutionalised, such that they continue to exert influence long after the historical circumstances that gave rise to them and independently of the actors that espouse them. This suggests that ideas should be considered as elements of context, and that doing so allows for a better accounting of the role of ideas in resisting change.

In drawing on historical and sociological institutionalism to explain path-dependence in South African health system reform efforts, this paper deepens and expands existing path-dependence theory. The analysis aligns well with accounts that explain path-dependence in terms of feedback loops and increasing returns—such as Pierson's (2000),which suggests that policy choices are self-reinforcing in that once a decision is taken, the cost of switching to an alternative policy solution begins to increase, making change increasingly difficult (Fischer & Miller, 2017; Pierson, 2000). As Pierson (2000, p.252) notes, the ‘increasing returns’ conceptual model draws attention to particular factors as the source of political stability, including ideational variables like shared interpretations of policy problems and mental maps or models. This paper extends this explanation by demonstrating that the institutionalisation of ideas is one of the mechanisms or causal processes by which the cost of change increases over time.

Of course, to argue that ideas matter is not to suggest that other, more material, factors do not matter. In this case, there is no doubt that economic realities, service delivery failures and shared interests are important explanatory factors. Rather, it is to argue that any explanation for how this policy process has unfolded that neglects the role of ideas will be incomplete. This is firstly, because ideas inform actor’s judgements about their own interests (as per constructivism, see Harmer, 2011) and about whose interests ought to be prioritised—such as Treasury’s prioritisation of middle-class interests, in this case. Similarly, the judgement of the middle-classes that they are better off buying healthcare and health insurance from private actors in the healthcare market than trusting the state-owned providers and financers, reflects a particular set of ideas that are becoming institutionalised. Secondly, ideas matter because, regardless of the material forces underlying it, in order to influence policy processes, opposition or support for reform must be presented and justified publicly. This requires arguments that rely on ideas for their discursive power.

In HPSR, the significant role ideas play in policy processes is widely recognised (Gilson et al., 2018). However, as noted above, most health policy scholarship, and policy science more widely, focuses on foreground ideas—on ideas as policy proposals or policy frames—and, therefore, the focus on ideas is secondary to the interests and actions of the actors who wield them (Campbell, 1998; Fischer, 2003). From this perspective, ideas play a powerful role in policy processes, but their power depends on how much support they receive from actors, and how much power those actors have in the policy process, which, in turn, depends on the institutional context of the policy process (Campbell, 1998). In other words, ideas are a tool, wielded by policy actors to either enable or constrain policy change, and institutions mediate the influence of ideas.

However, our analysis suggests that, in addition to ideational tools wielded by actors, ideas can become institutionalised such that they operate as cognitive or practical constraints on actors. Because hegemonic ideas shape how actors make sense of the world, they are fundamental to any form of collective action, and should, therefore, be analysed as independent variables, ontologically primary to the actors that espouse them, the institutions that reinforce them, or the socio-political circumstances that gave rise to them (Berman, 2001; Hall, 1997). In other words, once ideas become institutionalised—either as culture, shared values, interests or ideology, or in the procedures and processes of institutions—they become independently influential (Berman, 2013; Hall, 1997).

In addition, to the extent that institutions and background ideas, unlike the more commonly analysed foreground ideas, are relatively stable and persistent, as Campbell (1998) suggests, they form part of the context within which actors must work and should be considered in policy analyses as such. Often, policy experts are assumed to be ideologically neutral, and above-the-fray of politics (Fischer, 1987; Rich, 2005). However, all policy actors, including experts and policy-makers, are products of social contexts and hegemonic ideas, which will influence their policy positions (Fischer, 1987; Rich, 2005; Stone, 1996). The ‘ideological terrain’ within which actors operate will determine what kinds of programmatic ideas will be considered legitimate and what sorts of frames will be persuasive (Hall, 1997). Analysing ideas as contextual factors not only more accurately reflects the dynamic nature of ideational variables—that they are programmatic ideas about what should be done, the frames by which actors justify these programmatic ideas, and are part of the stable social context within which actors must operate—but also will help analysts to better understand why ideational variables enable change in some policy experiences and constrain it in others.