POWER SHARING AS A CONFLICT RESOLUTION MODEL: AFRICA AND ELSEWHERE

Power sharing in societies which are deeply divided along ethnic lines has, as Binningsbo (Reference Binningsbo2013: 89) comments, become a ‘dominant approach to solving conflict’ and is ‘recommended as a political solution to overcome deep divisions between groups'. These are societies where there exist two key problems: first, the legitimacy of the state is questioned by a sizeable portion of the population, and second, inter-ethnic differences have the potential to ignite political violence (Guelke Reference Guelke2012).

In his detailed review of power-sharing forms of democracy, Andewag (Reference Andewag2000) highlights how these models have been a common feature of societies that are fractured by politically significant inter-ethnic differences. So, to cite a few examples, it is not surprising that power sharing is fundamental to the Northern Ireland settlement (McGarry & O'Leary Reference McGarry and O'Leary2004); that it is the core aspect of the regulation of politics in Lebanon (Kerr Reference Kerr2005); a necessary facet of political design in Afghanistan (Katzman Reference Katzman2015) and was attempted in Iraq (Philips Reference Philips2005) – although the original design is now in tatters. As a stabilising mechanism it has been used in Kenya and Zimbabwe (Cheeseman & Tendi Reference Cheeseman and Tendi2010) and more broadly in Africa (Cheeseman Reference Cheeseman2011). Certainly, power sharing is a key tool in the toolbox of democratisation, but as Lynch & Crawford (Reference Lynch and Crawford2011) point out, while progress has been made since 1990, efforts to spread democracy in sub-Saharan Africa have encountered many setbacks.

Critics of consociationalism argue that it hinders the process of democratisation by embedding ethnic differences in political structures, and indeed, this critique emerged in South Africa both during the negotiations and shortly afterwards (Taylor Reference Taylor1992; Connors Reference Connors1996). But no two consociational models are the same. It is therefore useful to consider the literature on other examples of power sharing – with a focus on Africa – in order to put South Africa's arrangement into comparative context. This will help identify the particular features that pertained to the South African case. Of course, contextual distinctions in power-sharing examples suggest that a single model of consociationalism cannot be rolled out as a remedy for ethnic conflict. Instead, power sharing is best considered a tailored response to the needs of a specific context.

Horowitz (Reference Horowitz2014) warns that power sharing is problematic because ethnic parties compete vigorously with each other in pursuit of their supporters' interests. This can lead to political instability. Horowitz's point sets limitations to what consociationalism can achieve, but interestingly, while the NP's example indicates that membership of a grand coalition can help minimise the potential for politically fragmenting and stability-threatening inter-party competition, this may not serve the long-term electoral interests of a party. Accordingly, Mehler (Reference Mehler2009: 453–4) adopts a cautious approach to the idea that consociationalism is a cure-all for conflict, arguing that in Africa ‘power-sharing agreements cannot provide sustainable solutions to all relevant aspects of complex crisis situations'. Adding to the problem of complexity is that of state weakness – a common problem in Africa – leading Sriram & Zahar (Reference Sriram and Zahar2009) to claim that power-sharing efforts are likely to falter in weak or collapsed states.

Spears (Reference Spears2000, Reference Spears2002) notes how a recurrence of violence following a settlement has not been an infrequent problem on the African continent. In a more recent article, Spears (Reference Spears2013) documents a dismal series of power-sharing failures – Sudan, Angola, Rwanda and Sierra Leone – yet comments that where there are sturdy institutional structures, for example during regime change in South Africa, it supports a successful transition. In fact, Spears points out how it was possible for the NP to walk away from the GNU because it was ‘confident that, even outside of the government, whites would have their interests protected’ (Spears Reference Spears2013: 40). While this analysis of attitudes within the NP in 1996 is accurate, Spears does not account for the evaporation of support for the party which indicated that whites were not sufficiently confident that the NP was the party best able to offer the protection sought.

Mehler, and also Sriram & Zahar as well as Spears are not alone in their power-sharing pessimism in Africa. Lemarchand (Reference Lemarchand2007: 2) recognises that ‘much of the continent has become a graveyard of consociational experiments’ but considers Burundi to be the closest approximation to Lijphart's model and an example which carries hope for the implementation of power sharing elsewhere. However, since Lemarchand's analysis, Burundian politics has taken a turn for the worse. Despite the country's laudable attempt to manage ethnic divisions via institutional scaffolding, Vandeginste (Reference Vandeginste2014) perceptively questioned the sustainability of its arrangements and problems emerged shortly afterwards. In 2015 President Pierre Nkurunziza announced his intention to run for a third term in office. This led to an eruption of political violence which resulted in multiple killings (The Telegraph 2016). According to the UNHCR (2017), since April 2015, 420,689 Burundian refugees had fled to neighbouring countries. Despite the apparent success of Burundian power sharing, it was unable to prevent this crisis. Violence has also bedevilled Africa's newest nation, South Sudan. In the relatively recent history of southern Sudan, ethnic violence has been a perennial problem since 1991 (Hutchinson Reference Hutchinson2001). However, following the creation of the new state in 2011, tribalism has continued to be a significant factor in the mobilisation of violence which has blighted the country (Pinaud Reference Pinaud2014). This brings to one's attention the insightful comments of the British anthropologist Evans-Pritchard (Reference Evans-Pritchard1940: 161) who, when addressing the potential destructiveness of tribal divisions in this region of Africa said: ‘Between tribes there can only be war, and through war, the memory of war, and the potentiality of war the relations between tribes are defined and expressed.’ It is lamentable that Evan-Pritchard's pessimism at the dawn of the 1940s has an empirical relevance in the 21st century which, unfortunately, is not confined to South Sudan but has been widespread in Africa following decolonisation.

In their review of power-sharing settlements, Ottmann & Vullers (Reference Ottmann and Vullers2015) make the point that there are major differences between the promises and practices of such arrangements. This is an interesting observation and the commentators’ argument is germane to the case of the NP in that the party experienced a disjuncture between its expectations of what sharing power would entail and what actually unfolded. The spirit of consensus, which initially had been the driving force of decision making in the GNU, had largely evaporated by 1996. This soured the relationship between De Klerk and Mandela and was a key factor in the NP leaving the unity government.

Yet, although power sharing has its challenges and drawbacks it still offers ethnic antagonists an opportunity to construct peace. Thus Hartzell & Hoddie (Reference Hartzell and Hoddie2015: 38) contend that it reduces the ‘uncertainty associated with democracy’ and persuades ‘actors emerging from civil war … to adopt at least a minimally democratic political system'. These authors argue that critics of power sharing simply underestimate the barriers which lie in the path of deepening democracy in a post-settlement context. A positive interpretation of power sharing is also embraced by Sisk & Stefes (Reference Sisk, Stefes, Roeder and Rothchild2005) who view its accommodation in South Africa's Interim Constitution of 1993 as vital to conflict resolution. Although the GNU's operation was limited by a five-year sunset clause (which for the NP did not last beyond two) they argue that this period was sufficient to build confidence amongst the country's minority groups, namely, that the ANC did not intend to establish a majoritarian dictatorship.

The leading advocate of power sharing, Arend Lijphart, has remained unfazed by his opponents’ arguments and undeterred in the face of critiques of the consociational model. Like Hartzell & Hoddie, he views power sharing as a practical compromise between competing ethnic parties – advanced by their elites – which have more to gain than lose from sharing power. Lijphart has gone so far as to claim that ‘consociational democracy is not only the most optimal form of democracy for deeply divided societies but also, for the most deeply divided countries, the only feasible solution’ (Lijphart Reference Lijphart and Reynolds2002: 37).

A major reservation harboured by critics of power sharing is the impediment it is thought to present to the development of a vibrant political opposition. Indeed, in contradistinction to the position adopted by Sisk & Stefes (Reference Sisk, Stefes, Roeder and Rothchild2005) mentioned above, Jung & Shapiro (Reference Jung and Shapiro1995) launch a major critique of Lijphart's approval of power sharing in South Africa's Interim Constitution. They comment: ‘Lijphart celebrates South Africa's new consociational institutions as “just about the best that could have been designed,” urging that they be replicated in the final constitution. This is wrongheaded advice’ (Jung & Shapiro Reference Jung and Shapiro1995: 299–300). It is ‘wrongheaded’ because these commentators consider a strong opposition to be central to a healthy and stable democracy. While critics' concerns are not without substance, it is important to bear in mind that South Africa's political negotiations were primarily driven by the need to resolve a vicious conflict which had the potential to worsen (arguably this point is not afforded the weight of consideration it deserves in Jung & Shapiro's analysis). As a result, a fixed-term power-sharing settlement was agreed.

THE NP AND ITS HISTORICAL CONTEXT

Before delving more deeply into the topic, it is useful to account for the history of the NP. Prior to the non-racial elections of 1994, the political philosophy of the NP was dominated by the importance of race. Upon coming to power in 1948 its values were manifested in a raft of laws which segregated South Africans along racial lines. Apartheid was met with international condemnation. Disapproval was expressed in a range of sanctions (Guelke Reference Guelke2005). However, following the election of F.W. de Klerk in 1989 the party adopted a reformist strategy. This paved the way toward difficult inter-party negotiations which eventually resulted in a political settlement. The settlement notwithstanding, it was not altogether easy for the NP to shake the dust of apartheid from its feet. Yet, importantly, this acknowledgement does not imply that the demise of the party can be explained solely in terms of a ‘natural’ electoral death due to the NP finding itself in a non-racial democratic environment where the taint of its racist past rendered it ill-equipped to compete. There are two reasons which support this claim. First, it needs to be remembered that apartheid South Africa was a democracy for whites and the majority were content to give their electoral allegiance to the NP and its policies. The option to vote for South Africa's anti-apartheid liberal tradition was available to whites, but as Giliomee (Reference Giliomee and Shain2006) documents, relatively few found this option appealing. Second, the party's racist history did not prevent the vast majority of whites voting for it in the 1994 non-racial elections even though the robustness of the Interim Constitution of 1993 would likely have dissipated white anxiety about ANC rule. More broadly, of course, white concern about the ANC's flirting with the international bogeyman of communism evaporated when the system disintegrated a few years earlier. Certainly, the dissolution of communism made it less difficult for de Klerk to embark on a process of reform (Guelke Reference Guelke1996).

In the light of the racial dynamics which underpinned, and continue to underpin, voter preferences (Knox & Quirk Reference Knox and Quirk2000; Ferree Reference Ferree2004, Reference Ferree2006; Guelke Reference Guelke2005; Garcia-Rivero Reference Garcia-Rivero2006; Holborn Reference Holborn2010) the NP won a respectable 20.4% of the vote in a greatly enlarged electorate. This result informs us that, at the beginning of the new era, the NP was far from being electorally doomed because of its past racism. Had it been so, then it is plausible to assume that the 1994 election results would have indicated such. Therefore, to argue that the NP's history condemned it to electoral annihilation, which is an argument some members of the scholarly community might favour, is not supported by the electoral facts. Consequently, other reasons for the party's decline need to be explored. Following the first non-racial election, however, the party's demise was relatively swift as subsequent election results confirm: in 1999 the NP's strength plummeted to 6.7% and dropped to a disastrous 1.6%. Against this embarrassing backdrop it is unsurprising that the party decided to dissolve itself in 2005 (Meldrum Reference Meldrum2005). The NP indisputably qualified as an ethnic party (Horowitz Reference Horowitz1985) in that it was systematically supported by members of the Afrikaner community but it was also an ethno-racial party due to the support it received from English-speaking whites. However, the election results make clear that a massive rift had developed between the party and its long-standing support base.

It is argued here that a major – but not exclusive – reason for the party's decline was its participation in the GNU. In this respect, power sharing had its pitfalls. These can be understood in terms of the role of oppositional politics which revolves around the hardy defence of the interests of a party's core vote. In the case of the NP, its supporters were chiefly Afrikaner and white English speakers but membership of the power-sharing government complicated its efforts to defend such groups. This aspect of power sharing has not been satisfactorily accounted for in Lijphart's model, thus a refinement of his theory is recommended. Also, scholarly critiques of consociationalism do not focus directly and in a detailed sense on the question of a party's electoral fortunes in a post-settlement power-sharing context. Hence, this paper's focus on the concrete case of the NP adds an interesting dimension to the literature on power-sharing settlements.

THE GNU AND THE OPERATION OF POLITICAL OPPOSITION

By the measure of other democracies where coalition governments are in operation, political opposition in South Africa took an unusual form after the 1994 elections. The ANC, NP and IFP participated in the GNU and between them received the vast support of the electorate. The ANC won 252 seats whilst the NP won 82. The IFP won 43 and ended up with three cabinet seats in the GNU compared with the NP's six. Both parties were overshadowed by the ANC's 18 cabinet seats. In terms of parliamentary seats this resulted in a very lopsided balance in favour of the – ANC, NP, IFP – coalition government. On this point Spence (Reference Spence1997: 529) comments: ‘The remaining political parties, with 23 parliamentary seats out of a total of 400, constituted the opposition.’ Clearly, those constituting the opposition were dwarfed by parties making up what Lijphart refers to as a grand coalition. Additionally, the opposition was fragmented. It consisted of such diverse parties as the Freedom Front (nine seats) which was somewhat unwelcoming of the new era in that it advocated territorial separatism and the creation of an Afrikaner Volkstaat; the Democratic Party (seven seats) that represented the liberal white tradition during apartheid but which had been decisively beaten by the NP; the Pan Africanist Congress (five seats) whose traditional Africanist position made the party less comfortable with the settlement than the ANC; and the newly formed African Christian Democratic Party (with only two seats and 0.45% of the vote). Therefore, a combination of ideological fragmentation and small size rendered political opposition weak in parliament.

The coalition landslide did, of course, serve the needs of political stability and this point should not be glossed over when considering the needs of conflict-ridden societies. More specifically, it assisted in terms of securing executive-level stability amidst the uncertainties of the transition. In an article which amounted to a series of power-sharing recommendations vis-à-vis its best practice in deeply divided societies, Lijphart discusses the problems which executives can encounter in parliamentary systems when the opposition becomes negative in its attitude towards the government. He says the ‘fact that cabinets depend on majority support in parliament and can be dismissed by parliamentary votes of no-confidence may lead to cabinet instability – and, as a result, regime instability’ (Lijphart Reference Lijphart2004: 103). However, with a massive 337 parliamentary seats linked to the three parties comprising the power-sharing executive, the kind of opposition which Lijphart identifies as being deleterious to cabinet stability was minimised. In a society with a history of violent racial discord this was undoubtedly beneficial.

However, although important, it is but one measure of the political arrangement. If the grand coalition is assessed in terms of its impact on the NP's electoral success then another conclusion might be drawn. To a significant degree this was because of the constraints imposed by the power-sharing settlement which complicated the role of political opposition. As will be discussed below, the NP was constricted when it came to strong representation of voters’ interests. The emphasis on coalition unity dispossessed the party of the kind of independence necessary to provide tough criticism of ANC-dominated policies. As a result the NP was unable to manage its voters' expectations.

When accounting for the nature of political opposition at this time it is important to note that the South African power-sharing settlement was a compromise. Lijphart (Reference Lijphart1998) states that it caused the ANC to concede ground on what formerly had been a position based on simple majoritarian democracy. Spence (Reference Spence1997) discusses the pragmatism underpinning the ANC's approach during the negotiations. He argues that the reason for the ANC's toleration of power sharing was due to its recognition that the NP had a crucial part to play if the transition was to begin on a more solid footing than would otherwise have been the case had it been excluded from executive-level decision making. Had the negotiations established a Westminster-like ‘winner takes all’ electoral system rather than a power-sharing one then the NP (and the IFP) would have been excluded from government. Spence (Reference Spence1997: 529) explains: ‘Nelson Mandela and his ANC colleagues were willing to accept the principle of a constitutionally prescribed coalition largely because the ANC required cooperation from the NP if the loyalty of the bureaucracy and the security forces was to be guaranteed’ – a point that will be touched upon in the next section. Hence, as Spence points out, the ANC did not have much confidence in the idea of a ‘loyal opposition’ emerging from the 1994 elections and thought it wiser to work with the NP than risk the consequences of not doing so. NP marginalisation may have resulted in wider white alienation and threatened a loss of confidence in the settlement. Also, Pottie (Reference Pottie2001) discusses the rationality underpinning the ANC's acceptance of power sharing and emphasises the economic benefits of a politically stable South Africa. But, while ANC pragmatism played a significant part in the equation, there was another side to power sharing and in this respect Spence refers to the GNU as setting an important inter-party reconciliatory example for the wider society which, of course, would benefit all parties and not only the ANC.

The power-sharing agreement also allowed the ANC to draw upon the considerable governmental experience of the NP. In correspondence with the author, former NP politician, Ray Radue, explains the usefulness of the NP to the ANC in the GNU:

This relationship did not last. The massive democratic mandate of the ANC meant that it was in a powerful position to determine the direction of South African politics. It was inevitable that, as the ANC became more familiar with running the country and implementing policy, its reliance on the NP would decline. Radue discusses the effects of this as follows:

Radue's comment that de Klerk and NP cabinet ministers became ‘uncomfortable in government’ points to the challenges senior members of the party encountered in the GNU and which led them to serve as a ‘full blown opposition'. However, the opposition space envisaged by de Klerk had increasingly been occupied by the Democratic Party (discussed below) to an advantage that was evident in the second democratic election results three years later.

THE CHALLENGES OF POWER SHARING IN THE GNU

Participating in the GNU prevented the NP from grounding itself firmly as the official opposition party (a status it had earned by winning 20.4% of the vote in the 1994 elections). Its failure to develop a distinctive party character proved costly by the time of the 1999 elections. It indicates that the NP did not strike the sort of balance which Martin & Vanberg (Reference Martin and Vanberg2008: 503) say is necessary for ‘successful coalition governance'. The balance involves making compromises which render coalition politics possible yet ‘maintaining the party's public profile and convincing supporters’ that it continues to defend their interests.

Opposition has a core role to play in the politics of a democracy. In his survey of the functions of political opposition in new nations, Apter's (Reference Apter1962) analysis was informed by three key observations: that of the opposition's ‘representation of interests’; its ‘provision of information’; and last, that of providing ‘constructive criticism’ of the government. Apter's emphasis on the importance of the representative and critical aspects of opposition are especially relevant to this article and helps create a framework within which to discuss the NP's role in the GNU.

First, despite the fact that the NP opened up its membership to all racial groups in 1990, it was still recognised as a ‘white’ party and more specifically an Afrikaner party. Its membership was predominantly white and its leadership Afrikaner – so it had a distinct interest-base to represent. That these interests emanated from a minority base, both racially and ethnically defined, is also significant. Given the momentous nature of the transition and the shift in power from one racial group to another (physically and emotionally segregated by over four decades of apartheid policies) it is unsurprising that both Afrikaners and English-speaking whites were concerned about their place in the new society. It is worth bearing in mind that research conducted in 1988 by Manzo & McGowan (Reference Manzo and McGowan1992) highlighted Afrikaner anxiety about a non-racial South Africa. Additionally, a not insignificant portion of whites had voted ‘no’ in the 1992 referendum (Strauss Reference Strauss1993) and did so because they harboured reservations about the effects that a transfer of political power would have on minorities. This rejectionist section of white society as well as whites who naturally were anxious about the implications of a transfer of power (including other minority communities who voted for the NP in 1994) are likely to have preferred focused political leadership and a committed oppositional approach post-1994. Simply put, there was much for the NP to represent at the level of minority interests.

A sturdy defence of minority interests is likely to have resulted in the NP acquiring a reputation as a strong opposition party. However, along with the IFP, the NP entered the GNU as a minor party. This coalition of the country's main parties was celebrated by Lijphart (Reference Lijphart and Reynolds1994). It was a negotiated outcome which fitted comfortably with his power-sharing theory. In Democracy in Plural Societies Lijphart (Reference Lijphart1977: 25) discussed the coalition idea as follows: ‘The first and most important element is government by a grand coalition of the political leaders of all significant segments of the plural society.’ Although it can take a number of forms the ‘grand coalition cabinet in a parliamentary system’ (25) was the version adopted in South Africa. But the power-sharing model takes on a different complexion when its mathematics are considered along with its implications for the exercise of influence: with 18 cabinet seats the ANC had triple that of the NP's six which resulted in ANC dominance (the IFP was less well off with three seats). De Klerk pointed to the marginalisation of the party as a factor in the NP's withdrawal from the power-sharing arrangement: ‘The National Party has felt for some time now that our influence within the Government of National Unity has been declining. The ANC is acting more and more as if they no longer need multi-party government’ (Deseret News 1996). De Klerk's comment does not cast a favourable light on the consensual basis upon which decisions within the GNU were to be made. This was an approach to decision making that de Klerk states was agreed between him and Mandela and an arrangement which, at least initially, had operated with relative smoothness (de Klerk Reference De Klerk1999: 270). However, power within the GNU was a numbers game and the superior number of cabinet seats occupied by the ANC rendered it dominant.

In addition to the grand coalition aspect of Lijphart's power-sharing theory he also refers to the use of a ‘mutual veto … which serves as an additional protection of vital minority interests’ (Lijphart Reference Lijphart1977: 25). However, as Guelke (Reference Guelke2012) points out, there was no veto mechanism in the GNU for the NP to use. Despite Lijphart's applause of the settlement, the absence of a veto constituted a missing strand in the application of his theoretical model to the power-sharing executive. This fact adds substance to Guelke's (Reference Guelke2012: 89) claim that the GNU amounted to a ‘very watered down version of power-sharing’ and ‘for a temporary period’ to boot. The comments of two senior NP politicians illustrate the concern felt by the absence of a veto. Tertius Delport (BBC 1995) remarked ‘so the key question now was what will they [NP ministers] do in cabinet? To what extent will they be in a position to influence or even to veto particular decisions?’ He continued:

Delport met with President de Klerk regarding the matter and stated his concern that ‘we may … end up, as the National Party being co-opted into an ANC government’ (BBC 1995). Rina Venter (BBC 1995) had similar concerns:

Clearly, neither politician was content with this aspect of the agreement but the absence of a veto avoided what Rothchild & Roeder (Reference Rothchild, Roeder, Roeder and Rothchild2005) identify as the problematic misuse of this device by politicians who represent minority groups. Yet without a veto the GNU, as a central piece of the negotiated settlement, would operate with the kind of fluidity that accommodates the emergence of ‘new power relationships’ and ‘new status positions’ (Du Toit Reference Du Toit2003: 106). Hence, veto-less, what emerged in terms of power and status between 1994 and 1996 favoured the ANC and not the NP.

Perhaps the NP's predicament within the GNU, namely, that it could be easily overshadowed by the ANC, is a simple reflection of South Africa's racial demographics and dynamics, coupled with the fact that there are particular constituencies from which parties draw their support. Basically, with 62.6% of the vote in 1994 the ANC did not need the support of the NP to form a government so the GNU was not like coalition relationships in other societies. If necessary, the ANC had the popular support and parliamentary numbers to go it alone. This led Giliomee et al. (Reference Giliomee, Myburg and Schlemmer2001: 167) to argue that because the ANC had massive black support it did not need to ‘win the vote of the white minority and can treat it with indifference or contempt’. Further, these commentators suggest that de Klerk overestimated the importance of the NP in the post-settlement period. They argue that during the negotiations the leader of the NP ‘hoped that his party would continue to play a strategic role in a … (GNU) well beyond a second free election’ (Giliomee et al. Reference Giliomee, Myburg and Schlemmer2001: 165). This hope was premised on the shaky belief that the NP would serve as an intermediary link between the ANC and vital white business interests, as well as white civil servants and the white security forces. That de Klerk was firmly of this belief was evident in a speech he gave following the election results in 1994 and before participation in the GNU: ‘Just as we could not rule South Africa effectively without the support of the ANC and its supporters’ the former President argued ‘no Government will be able to rule South Africa effectively without the support of the people and the institutions that I represent'.Footnote ¹ It was thought that the ANC could not afford to alienate these facets of white society who would insist on the NP's participation at the centre of government. This need did not materialise and the ANC was capable of both pacifying white anxieties and demonstrating, especially after the NP's departure from the GNU in 1996, that it could run the country. As mentioned above, the governmental experience of the NP undoubtedly assisted the ANC in terms of consolidating its position during the first two years of the GNU. This served the electoral interests of the ANC more than it did the NP.

Lijphart (Reference Lijphart1998: 147) accepts that in South Africa ‘consociationalism has declined since 1996’ and recognises that the NP left the GNU because of the ‘predominant power of the ANC in this power-sharing cabinet'. Despite this, he did not drill down into the dynamics of the NP's subservient role. Neither did he speculate upon the electoral challenges of being involved in the GNU. Spence (Reference Spence1997) was clear that there were potential electoral dangers in the NP becoming too aligned with the ANC and thus too uncritical of ANC-led policies. Overall, the sharp decline in support for the NP between 1994 and 2004 speaks powerfully of voter alienation whilst the significant drop in electoral backing between 1994 and 1999 suggests that Spence's analysis was perceptive.

Whilst Lijphart (Reference Lijphart2004) briefly refers to the unity government in South Africa in an article designed to offer recommendations to societies which are considering adopting a power-sharing model, he makes no mention of the possible electoral costs for participating parties. By the time Lijphart's article was written, the results of the 1999 elections provided empirical evidence that the NP had suffered a massive loss in support, thus prompting the question: did power sharing have a part to play in the party's demise? However, whilst acknowledging Lijphart's major contribution to scholarship in this area, failure to account for the possible electoral snares of power-sharing constitutes something of a blind spot in his theory which ought to be addressed.

Prior to presenting his recommendations for constitutional design in divided societies, it would have been profitable for Lijphart (Reference Lijphart2004) to reflect on the possible electoral effects of power sharing on minor parties. Accordingly, an inspection of De Klerk's comments when announcing the NP's departure from the GNU would have been worthwhile as they lacked ambiguity. For example, former NP cabinet member during the apartheid era, Leon Wessels (Reference Wessels, Kotze and Solomon2008: 57), records de Klerk's parliamentary announcement on 3 June 1996, that the NP was leaving the GNU:

Lijphart might also have considered the reflections of the NP's Roelf Meyer regarding the constraints of power sharing. Meyer, who had a cabinet seat in the GNU, said: