Elite Conflict,Compromise,and Enduring Authoritarianism: Polarization in Zimbabwe,1980–2008

Prelude to polarization

Zimbabwe became independent on April 18, 1980, but it was bound by the terms agreed to by the nationalist parties and the Zimbabwe Rhodesia transitional government at the 1979 Lancaster House Conference. The agreement required that ZAPU and ZANU gather their armies at designated assembly points prior to the February elections, and it established a parliamentary system of government with twenty reserved seats for Zimbabwe’s white population, which composed at the time only about 8 percent of the population, and eighty “common roll” seats to be filled through proportional representation. ³ Though ZANU (led by Robert Mugabe) and ZAPU (led by Joshua Nkomo) had negotiated as a united Patriotic Front (PF) at Lancaster House, they contested the elections as separate parties. Both parties’ fighters gathered at assembly points, and ordinary voters were unsure that the ceasefire would hold. This affected the climate in which the 1980 elections took place. Election observer reports indicate widespread voter intimidation, as well as threats by ZANU cadres of a resumption of war should the party lose the elections (Kriger 2005, 4–6; Laakso 2002, 329–30).

The results reflected the different regional bases of the two parties and, for the most part, reflected the ethnic geography of Zimbabwe. Thus, ZANU dominated the provinces where their army had been operative, winning more than 80 percent of votes in Mashonaland East and Central, in Masvingo, and in Manicaland (Cliffe et al. 1980, 46). ZAPU, on the other hand, won 86 percent of votes cast in Matabeleland South and 79 percent of votes in Matabeleland North. Put differently, ZANU did well in areas with Shona-speaking voters, and ZAPU performed strongly in Ndebele-speaking areas. Though observers often present the 1980 elections as an ethnic census, especially given what was to follow, this is not exactly the case. Despite its ethnic core constituency, ZAPU won more than 27 percent of the vote in the ethnically mixed Midlands province, and it won a parliamentary seat and 13 percent of the vote in the overwhelmingly Shona-speaking Mashonaland West.

These ZAPU victories seemed, at least initially, to be accepted by ZANU. The party invited ZAPU to join it in a coalition government, and four ZAPU Members of Parliament joined the government as ministers, including ZAPU leader Joshua Nkomo. By July, only a few months after the unity government was announced, Finance Minister Enos Nkala declared at a rally that ZANU’s goal and desire “from now on [was] to crush Joshua Nkomo and forget about him.” ⁴ In November 1980, ZANU elites continued their attacks on ZAPU by declaring to their supporters at a rally that “from today the PF[ZAPU] has declared itself the enemy of ZANU.” ⁵ At this rally, Nkala went on to urge ZANU supporters in Bulawayo to form vigilante groups so that they could “challenge [ZAPU] on its home ground” (Alexander et al. 2000, 186).

Conditions were equally tense at the assembly points, where the ZANU-linked Zimbabwe African National Liberation Army (ZANLA) forces were taking control over the new Zimbabwe National Army. Some members of ZAPU’s army, the Zimbabwe People’s Revolutionary Army (ZIPRA), refused to come to the assembly points, and others left after arrival. ⁶ In late 1980, the first large-scale clashes between ZANLA and ZIPRA fighters broke out, and conflict spilled from the assembly point camps into urban townships. ZANU rhetoric likely played a role in inflaming tensions. ZANU-PF politicians insinuated that ZAPU politicians were involved in the violence, and, in February 1982, ZANU-PF Prime Minister Robert Mugabe announced that arms caches had been discovered on lands belonging to high-ranking ZAPU politicians.

The discovery of the arms caches brought the emerging conflict between ZANU-PF and ZAPU elites and fighters to a head. Accusing ZAPU of plotting to overthrow the government, Mugabe fired most ZAPU ministers from his coalition government and detained several senior ZIPRA commanders. ⁷ In dismissing his coalition-partners, Mugabe declared that: “ZAPU and its leader, Dr. Joshua Nkomo, are like a cobra in a house. The only way to deal effectively with a snake is to strike and destroy its head.” ⁸ Both the language and the strategies employed by elites were consistent with pernicious polarization. The dismissal of Nkomo and other senior ministers from the coalition government prompted another wave of ZIPRA defections from the new national army and from the assembly points. Nkomo fled Zimbabwe and remained in exile until 1985.

Civilians and the ZANU–ZAPU conflict

Did this elite confrontation penetrate to the societal level, and in what ways? Though marginal groups of violent ZIPRA dissidents posed little threat to the regime (see Alexander et al. 2000), the ZANU government mobilized significant military force to the affected regions, with disastrous impact on civilians. ⁹ The operation was termed Gukurahundi, which roughly translates from Shona as “the storm that washes away the chaff,” and its political or partisan nature was clear. The thousands-strong Fifth Brigade was drawn exclusively from former ZANLA combatants and was under the direct control of the Office of Prime Minister Robert Mugabe. In a report published in 1997, the Catholic Commission for Justice and Peace (CCJP) estimated that the Fifth Brigade was responsible for the following casualty toll in the single district of Tsholotsho alone: the killing of 1,500 civilians; the disappearance of another 350; the detention of close to 3,000 civilians, of whom roughly 380 were tortured; the sexual assault of 250 women by troops; and 1,500 injuries of other kinds (LRF 1997).

Consistent with other episodes of polarization, violence was accompanied by attempts to shape loyalties and enforce ideological compliance. Those caught in the brigade’s dragnet were often marched to schools or other public places at gunpoint and forced to sing Shona songs in praise of ZANU in what was imagined as a re-education campaign of sorts. Sources are in agreement that the intent of the operation was less tracking down dissidents than creating a climate of intense fear among presumed ZAPU loyalists in the run-up to the 1985 elections. Also consistent with other episodes of polarization, the conflict allowed for no intermediate ground between “us” and “them.” In practice, ZANU-PF politicians and military personnel conflated partisan and ethnic identity. In their rich account of the civilian experiences of the conflict in the 1980s, Alexander, McGregor, and Ranger reveal a clear pattern of targeting Ndebele civilians on the presumption that all Ndebele were ZAPU supporters and were therefore dissidents: “the child of a snake is also a snake” (Alexander et al. 2000, 222). Fifth Brigade commanders were also reported to have explained their actions as meting revenge on the Ndebele for the nineteenth-century raids of their ancestors on the Shona: “you have been killing our forefathers, you Mandebele.” It is therefore unsurprising that Alexander at al. conclude that one of the brigade’s most notable impacts in the conflict was “in hardening ethnic prejudice, and in bolstering a strong identification between ethnicity and political affiliation” (2000, 224). This was not entirely one-sided. To the present, some Zimbabweans mention roadblocks, forced evictions, and even killings of Shona-speaking residents of western Zimbabwe, though these attacks by ZIPRA or other armed actors were rare in comparison to the government’s attacks on Ndebele-speaking victims.

Estimates of the total number of civilians killed vary significantly, from the 2,000 deaths directly documented by the CCJP to the more than 20,000 total deaths estimated by the CCJP and Amnesty International. ¹⁰ To some extent, the violence can be viewed as imposing an ethnic cleavage from above, with the clear intent of shaping election results.

The military occupation of western Zimbabwe continued through the 1985 elections, and both atrocities against civilians and direct repression of ZAPU made campaigning impossible. Kriger reports that more than eighty ZAPU officials were abducted or disappeared during the 1985 election campaign, and, by the time of polling, “outside Bulawayo, virtually every urban and rural ZAPU office had been closed or burned down” (Kriger 2005, 8). During the campaign, as Kriger documents, ZANU-PF politicians consistently used inflammatory rhetoric that may have encouraged violence (Kriger 2005, 8–13). The results, remarkably, indicate some degree of robust support for ZAPU: ZAPU retained 88 percent of the votes that had been cast for the party in 1980, and it lost only 5 of its 20 parliamentary seats.

“Power-sharing,” party fusion, and the end of polarization

Shortly after the 1985 polls, perhaps convinced of its inability to eliminate its political rival via force, ZANU opened negotiations with ZAPU. Though violence and atrocities continued in western Zimbabwe, Nkomo returned from exile. In December 1987, ZANU and ZAPU signed the Unity Accord, which formally merged the two parties into ZANU-PF. Senior ZAPU members were given prominent cabinet posts, and, after constitutional changes adopted in 1990, Joshua Nkomo became one of two national vice presidents. The division of cabinet and other political positions was written into the new ZANU-PF constitution, as was the commitment to pursue the establishment of a one-party state. Although ZANU-PF did not succeed in creating a de jure one-party state, unification with ZAPU eliminated any potential for viable multi-party competition and laid the groundwork for the consolidation of a competitive authoritarian regime. In August 1987, ZANU-PF abolished the twenty seats reserved for white Zimbabweans and replaced them with candidates nominated and elected by parliament. ZANU-PF was left in control of 99 of 100 seats in parliament. In October 1987, the constitution was again amended to create an executive president with extensive powers. Robert Mugabe was the only presidential candidate nominated by parliament and was swiftly elected by the assembly. Zimbabwe transitioned from a parliamentary to a presidential system, but the centralization of powers that accompanied the party merger was much more important in shaping the post-1987 regime’s character and conduct. ¹¹

Surprisingly, ethnic polarization at the elite level faded quickly. The Unity Accord substantially increased Ndebele representation in the cabinet, resulting in more ZAPU ministers than in the first 1980 postindependent cabinet (Laakso 2003, 132). Since the signing of the unity accord, former ZAPU members have always held at least one of the vice presidential positions. Even after Mugabe’s overthrow in the “soft coup” of November 2017, this tradition continues. The accord also required the two parties to merge their entire organizational structures to the branch and cell levels. Given the geographic division of the two parties’ support bases, this may have required less merging than relabeling, and the task may have been helped by the decimated nature of many ZAPU district, branch, and cell structures by 1987. But, more fundamentally, the ideals and basic political attitudes of ZANU and ZAPU activists and politicians were not terribly different. Both parties espoused varieties of socialism, viewed the ruling party as the vanguard of the people, and were sympathetic to strands of traditional cultural revival. ¹² More importantly, the two parties’ elites believed that the liberation war should be central to party ideology and central to all party attempts to mobilize mass constituencies.

Nor were there indications that ethnic polarization reshaped political loyalties. Despite the toll that the Fifth Brigade exacted in western Zimbabwe, there were few signs that Ndebele citizens were necessarily hostile to ZANU-PF or opposed to the Unity Accord. In a snap poll conducted at a joint party rally in western Zimbabwe in January 1988, The Herald, Zimbabwe’s state newspaper, reported that few attendees knew about the alliance but cheered when informed that ZAPU members had been appointed ministers in the ZANU-PF government (quoted in Laakso 2003, 132). Voting patterns in western Zimbabwe do not suggest simmering resentments or an unwillingness to vote for ZANU-PF or even for former ZAPU cadres who had “sold out” the party and joined government. In the 1990 elections, voter turnout was marginally lower in Matabeleland than it was in the rest of the country (47 percent compared with 56 percent elsewhere), but there were few signs of a voter revolt. ZANU-PF won all twenty-four parliamentary seats in Matabeleland North and South in 1990 and 1995, and the party consistently won well over 80 percent of votes cast in rural constituencies nationwide.

Nor did party campaigns suggest that ethnicity was an issue. In 1990, the newly unified ZANU-PF did face a robust challenger. The Zimbabwe Unity Movement was formed by Edgar Tekere, a member of ZANU’s high command during the war and a popular former secretary-general of the party. Tekere and his party won 17 percent of the presidential vote and roughly 20 percent of the national parliamentary vote, mostly in urban areas. The Shona-speaking Tekere made no special attempts to politicize the prior violence against ZAPU supporters known as Gukurahundi, and his party’s program focused on economic conditions, corruption, and centralization of executive power—with no mention of human rights abuses or the atrocities in Matabeleland (Sylvester 1990, 389-392). ¹³ Ethnicity—or, indeed, any aspect of politics in Matabeleland—did not seem important enough to merit more than passing mention in academic articles on the election (Sachikonye 1990; Sylvester 1990).

This is not to suggest that Gukurahundi did not or does not matter to Ndebele Zimbabweans. Ethnic resentment over the lack of justice for the victims certainly lingers, and there are occasional suggestions that there exists a subtle, embedded political tribalism in Zimbabwean electoral politics (e.g., Muzondidya and Ndlovu-Gatsheni 2007). But our general point is that—in comparison with the polarization discussed below—the 1980s elite-level polarization was limited. Despite the rhetorical hardening and violence associated with this confrontation, there are few signs that polarization penetrated to the mass level. It did, however, result in a speedy closure of democratic possibility in Zimbabwe. From the mid-1980s to 2000, there was no space for viable political challenge in Zimbabwe. A new period of polarization was required to force such an opening.

Emergence of polarization

ZANU-PF’s response to electoral challenge and regime threat in 2000 was three-fold. First, it launched a series of mass land invasions of white-owned farms, in an attempt to shore up its popular support base among rural Zimbabweans. Second, it launched a campaign of state-sponsored violence that targeted MDC candidates, party organizers, and those suspected to be supporters of the party, such as teachers residing in rural areas. Up to 12,000 MDC supporters and activists fled the rural areas for the relative safety of the cities, and thousands more were displaced by the farm invasions. Finally, the ruling party deployed a violent rhetoric aimed both at discounting political opposition as treasonous and at reinforcing the liberation war as the central lynchpin of the country’s politics (Ranger 2004).

The direct costs of polarization were not borne by MDC supporters and activists alone. In an eloquent demonstration of polarization’s flattening of all politics into a single us/them distinction, formal and informal restrictions were placed on civil society organizations, and journalists and lawyers also became targets of violent attacks, arbitrary arrest, and abduction. Actions taken by government to target perceived opposition supporters also harmed economic activity and remade urban spaces. Businesses in the formal sector that did not express explicit support for the government found themselves subjected to electricity load-shedding during business hours, and others found their workplaces invaded and their unionized workers harassed by a newly formed, ZANU-PF-sponsored splinter union. In late May 2005, the Zimbabwean government implemented Operation Murambatsvina, an “urban beautification” campaign that forcibly displaced up to 700,000 Zimbabweans. As Potts underlines, this was not merely the removal of “squatters” or the clearance of illegal settlements; instead, the bulk of homes destroyed were backyard shacks constructed in formal, legal high-density areas (Potts 2006). The Zimbabwean government’s own statistics suggest that 570,000 Zimbabweans were displaced during the operation (Tibaijuka 2005, 32). Victims of Murambatsvina were disproportionately opposition supporters.

But the process at work should not be seen as simply state repression or the imposition of a disloyal or treasonous identity on any who voted against ZANU-PF. Instead, as LeBas has argued elsewhere, it is best to see this entire period as a reciprocal process of polarization that reinforced the solidarity and commitment of constituencies on either side of the political divide (LeBas 2006; LeBas 2011). In 2000 and in subsequent years, violence was disproportionately associated with state security forces, ZANU-PF party militia, and war veterans who were loyal to the ruling party (Kriger 2005). But the MDC and its civil society allies deployed their own polarizing rhetoric, reinforced us/them boundaries, and, especially within the grassroots structures, rejected any movement toward moderation or compromise. As time passed, the MDC created its own parallel structures that did not shy away from using violence and coercion to enforce party aims (LeBas 2006), and a culture of aggressive “hardcore activism” took root in some of the MDC’s affiliated structures (Hodgkinson 2013).

Polarization stabilized support for the two parties. This is perhaps the strongest evidence that polarization was reciprocal, as one-sided repression would have yielded a collapse of opposition support. Instead, we see continued loyalty and even the hardening of support among MDC activists and supporters, despite a consistent campaign of repression by ZANU-PF against both. In 2000, the MDC ran candidates in all 120 parliamentary constituencies and won 57 of those seats. Its support was concentrated in Zimbabwe’s cities and in ZAPU’s former strongholds in western Zimbabwe, but it performed well, and won parliamentary seats, in rural Shona-dominated constituencies as well. Though the party did not perform as strongly in the 2002 presidential and the 2005 parliamentary elections, it still won roughly 40 percent of the overall vote in both contests, despite rigging, intimidation, and violence. In the 2008 presidential and parliamentary elections, the MDC had split into two rival factions, but they collectively won more than 50 percent of the national vote and 110 of Zimbabwe’s 210 parliamentary seats. The MDC’s Tsvangirai won the presidential election outright, but ZANU-PF insisted that he fell short of the 50 percent needed to avoid a run-off. The period before the presidential run-off was associated with the worst spell of violence since Gukurahundi: 200 died at the hands of state security forces, and 36,000 were displaced.

Polarization and civilians

The penetration of political polarization to the mass level is also evident in public opinion data collected during this period. In the Afrobarometer’s second round of public opinion polling in 2004, roughly 48 percent of ZANU-PF supporters said they were “very” or “fairly satisfied” with democracy in Zimbabwe, while only about 10 percent of MDC supporters said the same. ¹⁴ Thirty-five percent of MDC supporters said they trusted the president “not at all” versus only 2 percent of ZANU-PF supporters; trust in other institutions, ranging from the police to courts to parliament, followed the same pattern. In the multiple rounds of Afrobarometer polling since then, these stark differences between MDC supporters and ZANU-PF partisans have remained remarkably constant. To a large extent, the partisan boundary in Zimbabwe shapes citizens’ attitudes toward democracy, toward their country’s institutions, and even toward the acceptability of political violence. In the most recent Afrobarometer polling, conducted in the run-up to July 2018 elections, voting intentions were almost entirely monopolized by the two poles with less than 2 percent of voters expressing a preference for parties other than MDC Alliance and ZANU-PF (Bratton and Masunungure 2018).

How can we know whether this polarization is partisan, as we argue here, or ethnic? In a study of Zimbabweans’ attitudes toward transitional justice and punishment of offenders for past human rights abuses, Bratton (2011) finds evidence consistent with our argument about both periods of polarization. As we would expect, MDC supporters are 36 percent more likely to demand justice than are supporters of the ruling party, but Ndebele-speakers—the victims of the first episode of polarization—are no more likely to demand justice than any other Zimbabweans (Bratton 2011, 372–74). Political polarization, yes; ethnic polarization, no.

We argue that elite confrontation and polarization in the 1980s did not have the effects on mass attitudes suggested by this public opinion data. Why might that be the case? First, to a greater extent than in 1980 to 1987, political conflict from 2000 to 2009 had an impact on ordinary Zimbabweans regardless of where they lived. Because rural teachers were often viewed as “closet” MDC supporters, political polarization took a particularly heavy toll on education in Zimbabwe. Attacks on teachers continued after the 2000 elections were over, and the first wave of forced rural school closings started in 2001 (Zimbabwe Human Rights NGO Forum 2002). In the run-up to the second round of the 2008 elections, the Progressive Teachers Union of Zimbabwe (PTUZ) reported violent attacks on teachers in schools, the use of schools for political meetings and militia bases, and forced student attendance at political meetings. What is notable is the geographic reach of these incidents: the schools affected in the first two weeks of May 2008 spanned twenty-three districts across the entire country. PTUZ reported that 65,000 to 75,000 teachers had been displaced from rural areas by 2010, and UNESCO reported in 2008 that 94 percent of rural schools had shut down due to political persecution. ¹⁵

These events took a toll on basic services and economic conditions nationwide. Schools operated with fewer teachers, clinics suffered staff shortfalls (especially later, with the collapse of professional salaries in 2002–2003), and other government programs lost skilled staff. As some rural areas became “no-go” areas for opposition, many Zimbabwean nongovernmental and community-based organizations ceased operations, and access to international food assistance was politicized as well (Human Rights Watch 2003). The collapse of the agricultural sector, price controls, and the resulting hyperinflation devastated the Zimbabwean economy, giving rise to kukiya-kiya (make-do) economy in which ordinary people turned to hoarding, trade in contraband, fraud, and even crime to make ends meet (Jones 2010).

Power-sharing and the continuation of mass polarization

As in 1987, the path out of polarization was power-sharing. In late 2008, under extreme pressure from South Africa and other regional actors, ZANU-PF signed a power-sharing agreement that named MDC leader Morgan Tsvangirai as prime minister, distributed ministerial positions evenly between the two MDC factions and ZANU-PF, and created a new decision-making body that was intended to give opposition a voice in the executive. The Global Political Agreement resembled past political settlements in Zimbabwe, including the 1987 Unity Pact. Like them, it did little to depoliticize the state security forces or carry out the institutional reforms needed to push Zimbabwe toward democracy (Bratton 2014). The GPA was also immediately undercut by public statements by ZANU-PF’s leading politicians and by the military heads who suggested continuity with past polarizing rhetoric. On the eve of signing the power-sharing agreement, Mugabe said that the MDC and ZANU-PF were “as different as fire and water” (quoted in Cheeseman and Tendi 2010, 220). Zimbabwe’s military chiefs refused to salute Tsvangirai at his inauguration, citing his lack of liberation war credentials, and high-ranking military officials referred to him as a security threat or as “taking orders from foreigners” throughout the power-sharing period. ¹⁶ The contrast with the period following the 1980 Unity Accord, during which there seemed to be near-seamless military and political integration of the two nationalist parties, is notable. Most troubling, violence continued throughout the power-sharing agreement. This included waves of new farm invasions, armed clashes between MDC and ZANU-PF youth, continued state targeting and detention of MDC officers and elected officials, and violence and intimidation during by-elections.

Though power-sharing seemed a continuation of the politics of polarization for ZANU-PF, it had striking effects on the MDC and, especially, on its electoral support (LeBas 2014). The party abandoned the confrontational rhetoric that had shored up its identity as distinct from ZANU-PF. MDC party leaders—now flush with the benefits of ministerial positions—allowed grassroots structures to erode, failed to maintain close relationships with civil society, and seemed to abandon their commitment to internal democracy. By 2012, grassroots MDC activists were disillusioned by allegations of corruption, the remote nature of party leadership, and what seemed an abandonment of the party’s commitment to internal democracy. In 2013, MDC voters seemed to come out for the party once more but to little effect. Though roughly the same number of votes were cast for the MDC in 2013 as in 2008, these translated into only 34 percent of the overall vote and just over 25 percent of parliamentary seats. This was partly due to the redrawing of electoral constituencies, which diluted MDC votes and made winning in a single-member-district more difficult. But it was also due to an incredible expansion in the size of the electorate and the overall number of votes cast for ZANU-PF, which the MDC and some civil society observers saw as a further indication of a fraudulent voters’ roll. The extent of rigging is unknown.

ZANU-PF viewed this defeat as an end to polarization, and both overt repression and polarizing rhetoric diminished. Both ZANU-PF and the MDC became more consumed with internal squabbles and purges than with elite-level confrontation. Does this mean polarization has diminished? As noted above, partisan loyalties and vote intentions in the months before the recent elections remained sharply polarized between ZANU-PF and a newly reconstituted MDC Alliance, which together monopolize voting intentions. In the 2015 Afrobarometer poll, patterns of distrust and disapproval also seemed largely unchanged from earlier rounds: 21 percent reported that they were either “not at all satisfied” with democracy in Zimbabwe or volunteered that Zimbabwe is not a democracy, and a further 18 percent refused to answer the question. Twenty-five percent of Zimbabweans reported that they fear violence and intimidation at the polls “a lot,” and this number rises to 40 percent of MDC supporters versus 19 percent of ZANU-PF supporters. Finally, a majority of Zimbabweans (63 percent) think the country is going the wrong way, but the gap in belief between MDC supporters and ZANU-PF supporters was 40 percent in 2015. ¹⁷

In 2018, we see similar polarization on party lines around beliefs about media access and freedom (Bratton and Masunungure 2018, 16). To a large extent, even though elite polarization has diminished, ordinary Zimbabweans’ attitudes remain polarized.