Zambia’s Problematic Election: The Questions That Won’t Go Away

In the immediate aftermath of Zambia’s 13 August election, we raised concerns about striking discrepancies between presidential and parliamentary results. We were deliberately cautious: we did not claim that the election had been stolen, attribute responsibility, or argue that opposition leader Brian Mundubile had defeated President Hakainde Hichilema. Instead, we argued that the anomalies required investigation and that, if they could not be explained, a rerun might be needed.

Our concerns centred on three issues: whether presidential votes had been inflated, whether this was connected to the six-hour suspension of counting, and how any resulting problems could credibly be resolved.

Two weeks later, new evidence strongly supports these concerns and undermines many of the rebuttals published since. Journalistic accounts and findings from the Christian Churches Monitoring Group (CCMG) and European Union observers provide much clearer evidence of vote inflation and problems in the tabulation process, while subsequent political and legal developments have made resolving these problems harder.

In turn, this raises a broader issue: the danger of an electoral fallacy in which evidence that Hichilema’s winning margin exceeded the manipulation documented so far is taken to mean that the election was essentially sound. Elections are more than arithmetic.

International observers identified an uneven playing field and serious procedural failures, while vote inflation is simply the form of malpractice for which we currently have the clearest evidence. Excusing such abuses because they cannot be shown to have changed the winner risks normalising the incremental erosion through which democracies become less democratic.

The evidence on vote inflation

Some commentators reasonably asked whether other factors could have produced the anomalies we identified – for example, Hichilema’s greater popularity than individual UPND parliamentary candidates or split-ticket voting. Other, more partisan responses relied on flawed arguments and obfuscation to dismiss the discrepancies altogether. Our intervention was also interpreted by some as reflecting an anti-Hichilema stance, but this is not true. Indeed, our strong criticism of abuses under President Lungu’s government previously led Hichilema to thank us for our efforts to promote Zambian democracy.

CCMG’s evidence and careful analysis demonstrate that our original analysis was fair and accurate. Its nationally representative Process and Results Verification for Transparency (PRVT) estimated Hichilema’s support at 56.0% ±1.7 percentage points. His official result of around 61% therefore falls clearly outside the PVT’s margin of error. In other words, the PVT contradicts Hichilema’s tally, though does nonetheless suggest he secured the most ballots.

Just as troubling, CCMG identified 30 constituencies where the official turnout announced was significantly higher than its monitors had recorded, implying that votes had been added. Comparing its polling-station records with tally-centre figures, it found Hichilema’s vote had increased in 10 of 24 stations examined, by an average of 178 votes; Mundubile’s changed at only one, by ten votes. Split-ticket voting cannot explain votes appearing between the polling station and the constituency tallying centre.

There is therefore now compelling independent evidence that presidential votes were inflated and that this was overwhelmingly done in Hichilema’s favour. As a result, the debate has moved on: the key questions are now how the inflation occurred and why.

What happened during the suspension and why?

Our second concern was the Electoral Commission of Zambia’s decision to suspend counting and results announcements for six hours on 14 August. We did not claim that the suspension facilitated manipulation, but suggested that, if manipulation occurred, early results may have triggered concern.

Subsequent evidence backs up this hypothesis. EU EOM observers found that tabulation had generally proceeded well before the suspension but “deteriorated significantly” after it resumed. “Very good” assessments fell from 43% to 27%, while verification that figures entered into the digital results system matched the official tally sheets fell from 100% to 81%. This does not prove manipulation, but confirms that tabulation became less transparent at precisely this point.

Just as worryingly, EU EOM observers “reported extended breaks during the results tabulation process in almost half of observations, as the returning officers seemed to wait and receive instructions from ECZ headquarters before announcing results. Further, a heavy military presence in totalling centres following the suspension was reportedly intimidating to most interlocutors and resulted in the number of party agents and citizen observers present at totalling centres being lower than before.”

Combined with the CCMG findings, this shifts attention to motivation. If Hichilema’s lead exceeded the inflation identified, why rig in favour of a president who was already winning?

Perhaps the UPND underestimated his advantage and wanted insurance. But another prize was also at stake: 40 new proportional-representation (PR) parliamentary seats, introduced by the 2025 constitutional amendments and contested for the first time in 2026. Because these seats are allocated according to the presidential vote, inflating Hichilema’s tally promised to increase the number of UPND’s MPs.

Sources who wish to remain anonymous have told us that securing a strong majority was a key UPND target and – against democratic protocols – this was communicated to senior Electoral Commission figures. Opposition parliamentary victories – announced at constituency totalling centres on Election Night – would have shown that this target was in doubt.

If the party wished to ensure it had a sufficient majority to be able to change the constitution, for example, its total of 138 of 226 constituency seats, or 61%, left it short of the two-thirds majority required. This realisation may have created an incentive to compensate for lost constituency seats by maximising the government’s PR allocation through a higher presidential tally.

If it was a desire to establish a politically dominant position that motivated the vote inflation – and it remains a big if – the significance  of these efforts extends well beyond this election. Although the UPND fell short, manipulating votes to pursue a supermajority would raise concerns about whether the government is contemplating further constitutional changes.

This is particularly significant for the country’s democratic future for two reasons.

First, previous constitutional amendments under this government proved highly controversial, including proposals to increase the number of MPs appointed by the president from eight to eleven, and the introduction of the PR seats.

Second, every Zambian president who survived to the end of their second term has sought a third term or explored ways of extending their tenure. Although there are no public signs of this being on the agenda yet in Zambia, the country’s historical experiences, and the fact that the removal of presidential term limits has become a major feature of democratic backsliding across Africa in recent years, demonstrate the importance of vigilance.

Put another way, the puzzle over the motivation behind the presidential vote inflation in 2026 may contain a warning of more serious democratic challenges to come.

How can the flaws in the election be corrected?

Our final argument was that, if the discrepancies could not be explained, a rerun might offer the most credible solution, despite the many challenges involved in doing this. Subsequent evidence strengthens that case – not because we know the wrong candidate was declared the winner, but because correcting the result is more complicated than it first appeared.

Those who understand electoral processes were already well aware that a simple recount would not necessarily resolve the problem. Some ballot boxes and election materials were damaged, while the boxes have long since left polling stations where they were counted under the scrutiny of party agents. Given that the opposition has lost trust in the ECZ, and that observers found key procedures were not always followed, recounting disputed ballots would be unlikely either to answer the questions posed above or build consensus around the result.

The challenge of what to do next is compounded by issue of the 40 PR seats. Applying the PR formula to CCMG’s projection of the election outcome and comparing this to the official result suggests that vote inflation changed the allocation of two seats from the NRPUP to the UPND. Correcting the presidential tally would therefore not be enough to address the problems with the election: any credible remedy would also need to also correct the composition of parliament.

Despite its financial and logistical costs, a rerun would have provided one way of resolving both issues.

To be clear, we were not suggesting that a rerun was ever likely – they are extremely rare on the African continent – and it now appears even less feasible. Most notably, the prospects for a judicial resolution to the crisis along the lines recently seen in Kenya in 2017 or Malawi in 2019/20 have been closed by the alarming decision to shut the country’s courts on the final day for filing a presidential petition, seemingly to prevent one from being filed.

Linda Kasonde then attempted to submit one electronically, ultimately emailing it to the Chief Justice after the official “hard copy” route was undermined. Although the Chief Justice acknowledged receiving the documents and referred them to the Constitutional Court, the Judiciary subsequently stated that no presidential petition was before the Court, and thus the inauguration would go ahead as planned on 1 September.

Combined with dozens of arrests and pressure on opposition figures – Mundubile sought refuge in the United Nations compound in Lusaka, before being handed over to police for interrogation following mediation by the UN and church bodies – this has further undermined the credibility of the polls, and made an effective institutional remedy increasingly difficult to envisage. In other words, although a rerun may have offered the most effective route to a legitimate outcome, this now appears increasingly remote.

Similarly, a presidential petition – had it been filed and heard – would have given the opposition an institutional route through the post-election impasse, and helped legitimate the outcome. Instead, senior opposition figures have been detained over treason allegations – serious charges that are non-bailable but no longer carry the death penalty – that have raised concerns that claims of “treason” and “terrorism” are once again being abused for partisan gain.

Both domestic and international observers will now be waiting to see whether the government can credibly prove such serious charges in court – if not, the credibility of the process will suffer another blow.

The electoral fallacy

An additional reason we raised the possibility of a rerun was to avoid a key electoral fallacy: assuming that because the manipulation we can document was smaller than Hichilema’s winning margin, the election was essentially sound.

Our respected colleagues Michael Wahman and Lindsay Favero are right that Hichilema may have won a convincing victory without the vote inflation. But elections are more than arithmetic. International observers documented an uneven playing field, including restrictions on opposition campaigning, incumbency advantages and serious problems during tabulation. Vote inflation was therefore not the only problem with the process – it is simply the one for which we now have the clearest evidence.

More broadly, defending democracy requires major abuses to be taken seriously whether or not they changed the winner. Kenya’s 2017 annulment established an important principle: electoral integrity depends not simply on who probably received the most votes, but on whether the election met democratic and constitutional standards. Lowering that bar whenever malpractice appears insufficient to alter the outcome risks normalising the incremental abuses through which authoritarianism advances.

Zambia’s election should be judged not only by who finished first, but by whether the institutions and procedures protecting democratic choice were respected.