This was due to be the last in a series of four articles on the Kenyan general elections of 2017. The first three looked at the campaign, the state of play between the main alliances and the capabilities and activities of the Independent Elections and Boundaries Commission and made a series of predictions about the likely results of the 8 August poll at presidential, gubernatorial and parliamentary levels. This article looks at what happened next: the results, where those predictions were right and wrong, what we can deduce about the conduct of the electoral process in the light of the Supreme Court’s invalidation of the presidential poll on 1 September and what lessons there may be in the first presidential poll for the second.
The Presidential Results
In the Presidency, as predicted in all three articles, according to the Form 34Bs which record the 290 constituency results, Uhuru Kenyatta won a clear victory; winning 54% of the vote to 45% for his main challenger Raila Odinga. This was the result of an electoral process which initially pleased almost everyone. The procedures on polling day worked well, the electronic voter identification and tallying systems mostly functioned as intended (or at least as predicted), there was no military intervention, no mass failure of the electronic voter verification system and counting at the polling stations was mostly uneventful. The presidential results were (mostly) logical and consistent with previous elections and with the parallel elections taking place and there were no excesses of votes in the Presidency compared to the other polls. The overall process was given the support both of domestic and international observers (with qualifications as the results had not yet been declared at that point).
For now, this analysis is based on the opinion – which I hope to explain – that while there were material administrative issues sufficient in the minds of the Supreme Court to invalidate the election, the evidence strongly suggests that the presidential results announced by IEBC were not “cooked” or “computer generated”.
That is not the view of a large number of Kenyans who supported the NASA coalition however, nor of the Supreme Court, and we will look in more detail at their concerns later. For now, this analysis is based on the opinion – which I hope to explain – that while there were material administrative issues sufficient in the minds of the Supreme Court to invalidate the election, the evidence strongly suggests that the presidential results announced by IEBC were not “cooked” or “computer generated”. Many of the complaints raised relate to the IEBC’s partial migration to an electronic tallying system, which as predicted was a key source of confusion.
Overall, the IEBC results showed that Kenyatta and William Ruto had won a decisive victory, by a greater margin than most had predicted. They won 26 counties to Odinga’s 21. Uhuru won three counties I thought he would lose – Garissa, Narok and Nyamira – and lost one, Tana River.
I got closest in my article in June, which predicted a 55-45% victory, In fact, the closer to the election we got and the more information I acquired, the less accurate my predictions were. In fact, I had begun to doubt my own numbers and modified my eve-of-poll prediction from 53-47% (which the spreadsheet suggested) to 52% to 48%. I left however the predicted votes for each candidate the same, and there I was pretty close: the official constituency Form 34Bs show that Kenyatta beat Odinga by 8.2 million to 6.8 million votes, compared to which I had predicted 8 million to 7 million.
Regionally, Kenyatta and Odinga (and their respective Vice Presidential candidates William Ruto and Kalonzo Musyoka) won all their Kikuyu, Kalenjin, Luo and Kamba “heartlands” as expected, and by huge margins. The two internal “insurgencies” in Bomet (Isaac Ruto for NASA) and Machakos (Alfred Mutua for Jubilee) both had little impact on the presidential votes. I had expected Ruto to bring more voters to Odinga than he in fact did. Little will change here in a rerun. As predicted, Kenyatta won most of the north and North east, Odinga most of the Coast and Western. Nairobi (on the far left of the chart below) was narrowly pro-Odinga (51% to 48%), much closer that opinion polls had predicted, a source of some surprise. Ipsos for example had run a survey in Nairobi just pre-poll which predicted a 56% Odinga vote with a margin of error of +-2.7%. The Kisii and Nyamira result (on the far right) were also a surprise, as most commentators, myself included, had given the region to NASA as in 2013. Explanations given afterwards included the heavy investment Jubilee had made in the region, the defection of virtually all ODM MPs to Jubilee and the influence of Fred Matiang’i as cabinet secretary.
Note: Orange throughout is NASA or Odinga; Blue throughout is Jubilee or Kenyatta. I use blue rather than red, the “Jubilee colour”, because red and orange look similar in some display formats, and because blue is a more “conservative” colour in most political systems than red, which tends to be associated with socialism and communism, and Jubilee is definitely a more conservative alliance.
As expected, all the other candidates were irrelevant, except for Joseph Nyagah (small spread votes) and Mohammed Dida (in green above), who polled creditably in the north and north east. Rejected and otherwise inadmissible votes were reasonable, down on 2013 at 0.5% overall (based on the Form 34Bs).
When I summed them manually, the 34Bs added up to almost exactly the same results as IEBC had announced around 8pm on 11 August (which they had done with a couple of seats still missing, as they were entitled to do).
These Presidential results are taken directly myself from the 34Bs, when they were published in a repository by IEBC, which were the only formal and legal basis for announcing a result. When I summed them manually, the 34Bs added up to almost exactly the same results as IEBC had announced around 8pm on 11 August (which they had done with a couple of seats still missing, as they were entitled to do). There were three Form 34Bs missing from the Forms repository (a different result had been uploaded instead), so I used the 34C national summary for them. The results in the IEBC real time portal (initially fed by the KIEMS system and then corrected and topped up later manually with missing results) were similar, though not identical, with the main difference being the spoilt votes, where – as in 2013 – there appeared to be an glitch which led the number of rejected, disputed and objected votes to be far larger electronically than in fact it turned out to be (something the IEBC has never explained).
Comparing the now invalidated presidential results against those for 2013 (easy with the same constituencies and candidates) we can see clear trends. Kenyatta did better in most areas, picking up votes especially in the north and North East, the Coast, Western and Kisii/Nyamira. Odinga did better in Bomet, some northern Kalenjin seats, most of Western (where he took the majority of Mudavadi’s 2013 vote) and Meru.
Turnout was substantially down on 2013. This was as predicted: the 2013 election had been fought on a new register, which had been only incrementally and partially updated since then, leaving at least a million dead voters still registered, so turnouts were inevitably going to be lower. In addition, the electronic voter identification system, with id cards, photographs and fingerprints combined, and (uneditable) tallies of voters maintained electronically by the KIEMS systems, deterred or prevented some “top up voting“ (officials voting for missing voters at the end of the day) which occurred in 2013.
In summary, if the Presidential result was substantively rigged or the result otherwise affected by the issues found, it is near certain that all the other elections must have been rigged or affected in the same way, as they involved the same voters, method for voting, technology for voter identification and results transmission (KIEMS), the same real-time results display portal, the same voting and counting processes, the same election officials and almost the same end results.
The turnout pattern (in black below) matched very closely that of previous polls, highest in the Luo and Kikuyu homelands, lowest on the coast. Turnouts exceeded 85% in 35 mostly Kikuyu, Luo and Kalenjin constituencies, a sign of some forced voting, top ups or stuffing, but exceeded 90% nowhere, and nationwide were a very reasonable 78% (compared to the 76% that a long-term weighted average of the last five elections suggested). The change in turnout on 2013 (in green below) was mostly consistent, as would be expected if dead voters were the main reason. Turnouts rose slightly in a couple of Kilifi seats where they had been depressed by the Mombasa Republican Council violence in 2013, and in Tharaka in Tharaka-Nithi (unexplained so far).
In the 47 gubernatorial races, the results followed a similar pattern to those for the Presidency. Again, Jubilee won decisively, by a greater margin than predicted. Here too, I underestimated the scale of Jubilee’s victory (though I got the winner right in 40 of 47). I predicted that Jubilee and their KANU, MCC, FAP, PNU, DP, NARC-Kenya and independent allies would win 21-28 Governorships, but they ended up with 29. As expected, they won their homelands, and Mike Sonko won Nairobi. Jubilee also won four counties where I had them as marginals (Narok, Kwale, Lamu and Wajir) and four (Garissa, Kajiado, Bomet and Machakos) which I had given to NASA. Across the nation, only 21 of the 47 incumbent governors returned to office.
New Governors included three Kenyatta first-term cabinet secretaries, all dropped from their posts for various alleged misdeeds: Anne Waiguru in Kirinyaga, Joseph ole Lenku in Kajiado and Charity Ngilu in Kitui – plus retired Kibaki-era Secretary to the Cabinet Francis Kimemia. This reaffirmed the illusory nature of the distinction between senior non-partisan state officials and politicians. If they were not in active politics when they entered office, they certainly were by the time they left.
For many Kenyans, the local races for MP and MCA were just as important as those for the President and Governor. There too, the same pattern was seen – Jubilee successes across the board.
NASA did not petition the governorship elections collectively, though they made allegations that some results were “computer generated” and initially, nor did most losing gubernatorial candidates. There seemed a general assumption that the non-presidential polls were not systematically rigged until the Supreme Court’s judgement, which immediately opened the floodgates for petitions by defeated candidates, including losing gubernatorial candidates, in Embu, Siaya, Kirinyaga and Machakos, with more to come.
The Parliamentary Races
For many Kenyans, the local races for MP and MCA were just as important as those for the President and Governor. There too, the same pattern was seen – Jubilee successes across the board. In the National Assembly, for the 290 constituency MPs my prediction of a 54% pro-Jubilee to 46% pro-NASA win turned out again to be a slight underestimate of the size of Jubilee’s victory. In fact, Jubilee and allies won roughly 60% to just under 40% for NASA. Jubilee did well in Bungoma and Kakamega (where ex-New FORD Kenya members formed the core of their victors), Kisii and Maasailand, and even won a couple of seats in Kitui and Machakos. ODM swept Luo areas and most of the Coast and Wiper most of Ukambani, while Mudavadi’s ANC, FORD-Kenya and ODM competed for the non-Jubilee western seats. Nairobi split 9 seats to Jubilee to 8 to NASA. The majority of MPs were newcomers, with voters clearly demanding change at the local level, particularly in the Kikuyu and Luo homelands, where few incumbents were re-elected.
The pattern was similar amongst the elected county Women’s MPs (with 31 for Jubilee and its allies versus 16 NASA and one independent) and in the Senate, where Jubilee and allies won 27 elected seats to NASAs 20). Overall, Jubilee won (initially) the presidency, the National Assembly, the Senate and most of the Governorships, the most decisive victory since the NARC wave of 2002.
Contrasting Perspectives and NASA’s Concerns
In general, the elections appeared to have been smoothly run, the results consistent, the electronic portal reporting convincing and the IEBC appeared comfortable in delivering its mandate. Observers commended the process as “peaceful, fair, and transparent”. Believing it had lost its ability to validate and correct constituency errors after the Maina Kiai et al case, IEBC headquarters limited itself – for the presidential election victory announcement – to a process of extraction, verification and entry of the 290 constituency Form 34B returns, the summing of these results and the announcement of the winner. As there remains dispute on this, the key decision summary is reproduced here from the Kiai judgement (http://kenyalaw.org/caselaw/cases/view/133874/):
The results Chebukati announced from the 34Bs (acknowledged by all to be without a complete set of 40,000 matching polling station Form 34As) matched closely with the parallel returns coming from the polling stations via the electronic KIEMS system in real-time to Bomas. From close of poll on the 8th, the parallel result stream from KIEMS soon showed a lead for Kenyatta and that lead grew over the next 48 hours as more and more of the electronic kits reported in.
The independent Parallel Vote Tabulation conducted by the ELOG domestic observer network and announced on 12 August validated the results almost precisely (its sample-based prediction gave 54% for Kenyatta to 45% to Odinga with a 1.9% margin of error). This was crucial because it provided independent verification to observers and the media that their perception of a well-run election was matched by independent assessment. Of course, this could have been faked, but there is no evidence yet offered that it was.
A macro-level comparison of voters cast and results between elections in fact shows that Odinga did better presidentially than his candidates in general. A re-tallying of the 15.3 million gubernatorial votes by constituency gives 5.7 million votes to ODM, Wiper, CCM, ANC, FORD–Kenya and allied candidates, far less than Odinga’s 6.8 million (in red). Thus Odinga did better in the cancelled presidential elections than did his gubernatorial candidates. The same pattern is seen in Parliament – again, Jubilee candidates polled more than 2 million more than NASA, though results are incomplete become 18 seats still don’t have full results on the Portal (https://public.rts.iebc.or.ke).
Jubilee = Jubilee + KANU + FAP + MCC + EFP + DP + PNU + NARC-Kenya plus defectors from the above after losing primaries, where known
NASA = ODM + Wiper + CCM +ANC + FORD-K + CCU + NARC plus defectors from the above after losing primaries, where known
Jubilee’s victories in the annulled presidency matched well with its victories in parliament and the Governorships. Comparing the Presidential, Gubernatorial, Senate and Women’s Representative results against each other by winner, in only nine counties did voters switch tickets: Nairobi, Machakos, Lamu, Tana River, Kwale, Taita-Taveta, Turkana, Narok, Trans-Nzoia and Nyamira.
Of those, Odinga won every one except Nyamira. In summary, if the Presidential result was substantively rigged or the result otherwise affected by the issues found, it is near certain that all the other elections must have been rigged or affected in the same way, as they involved the same voters, method for voting, technology for voter identification and results transmission (KIEMS), the same real-time results display portal, the same voting and counting processes, the same election officials and almost the same end results.
Rather than conceding once the trend was clear, Odinga rejected the presidential results outright (though not the other results) and accused the IEBC of a “complete fraud”. NASA’s impassioned follow up allegations were more specific, claiming form substitution, un-gazetted polling stations and administrative chaos in the IEBC and castigating the IEBC for releasing the presidential results without all the Form 34As. The sometimes-contradictory and implausible hacking claims made by senior politicians including Odinga, James Orengo and Mudavadi on 9-10 August raised the political temperature sharply, as intended, but also distracted attention for a while from real issues which were emerging relating to the IEBCs handing of the Form 34As. Despite widespread scepticism and challenge from the international observers, who had all judged the polls so far (before results had been announced) to be free and fair, NASA’s leaders refused to accept the results, claiming they were “cooked” or “faked” and demanded – even before all form 34B were in – that IEBC declare Raila as President (at one point using a faked NASA parallel count document as supporting evidence).
Unexpectedly abandoned by the international observers, who they had previously seen as allies, they lashed out at them as well. A few NGOs including the Kenya Human Rights Commission backed up NASA’s allegations to varying degrees, which then raised further fears of state repression (and generated further bad press internationally) when the state briefly tried to shut them down immediately the result was announced.
However, Odinga and the other NASA principles came under intense domestic and international pressure to take the constitutional path, as their ambivalent, partial move to “the streets” to protest during Wednesday 9th – Sunday 13th August was escalating and several people (probably at least 28) had been killed, mostly by the security forces.
NASA followed up their allegations with a petition against the presidential election, filed just within the one-week deadline on 18 August. Until the 16th, they had told Kenyans that “filing a petition at the Supreme Court to challenge the results was out of the question” because of CORD’s difficult experience in 2013 in crafting a case in one week, and the high burden of proof then demanded. However, Odinga and the other NASA principles came under intense domestic and international pressure to take the constitutional path, as their ambivalent, partial move to “the streets” to protest during Wednesday 9th – Sunday 13th August was escalating and several people (probably at least 28) had been killed, mostly by the security forces. Fears of broader communal violence in Nairobi were growing, fuelled by a series of fake media photographs, pretending to be current and of Kenya, designed to incite hatred. The decision to petition offered a temporary release for that tension.
For just one week (extraordinarily brief because of the two-week end to end deadline for concluding presidential cases, which the judiciary had already asked unsuccessfully to be extended) the Supreme Court heard the NASA case and responses from the IEBCs lawyers and other interested parties, with the verdict announced 1 September. NASA’s case focussed on five main areas – the electronic vote transmission system and its potential hacking (with the extraordinary claim that the portal results were a mathematical calculation unrelated to the actual votes cast); the missing form 34As and whether some were invalid or had been faked or substituted and errors in the KIEMS data entry which sent some of the results to the tallying centres; whether the IEBC Chairman should have declared without all the form 34As in his possession; examples of tallying errors between form 34As and Bs and possible malpractice in particular constituencies; and the pre-poll electoral environment including campaigning by Cabinet Secretaries for the ruling alliance.
The two dissenting judges’ Ndung’u and Ojwang’s opinions on the case were brutal – that the petition was without merit, devoid of evidence and that any transmission irregularities did not and could not have affected the outcome of the actual election at the polling stations or the count at constituency tallying centres.
To some surprise, by a 4-2 majority verdict the Supreme Court led by Chief Justice Maraga nullified Kenyatta’s re-election, because the poll was “not conducted in accordance with the Constitution”, and specifically the IEBC had “committed irregularities and illegalities inter alia, in the transmission of results”. The detailed grounds for that decision are not yet known, as the formally argued verdict will only be issued in 21 days (as it was “not humanly possible” in the words of the CJ to prepare the report in the time available). The court found no evidence of misconduct by Kenyatta (which had been one of Odinga’s petition grounds), though again we do not yet know their reasoning. It ordered another “fresh” presidential poll to be held in 60 days.
The two dissenting judges’ Ndung’u and Ojwang’s opinions on the case were brutal – that the petition was without merit, devoid of evidence and that any transmission irregularities did not and could not have affected the outcome of the actual election at the polling stations or the count at constituency tallying centres. Justice Ojwang argued that “there is not an iota of merit in invalidating the clear expression of the Kenyan people”. Kenyatta’s lawyers were furious, with one calling it “a political decision that is absolutely devoid of an iota of legal reasoning”, but the Supreme Court is Kenya’s final court and there is no further appeal.
Where were the Real Issues?
The single most vexed element of the whole election proved to be the electronic vote tallying and reporting, which had been introduced in the 2016 and 2017 Elections Act amendments. The unsolved murder of the IEBC expert responsible for KIEMS just before polling day (the reasons for which have still not been explained, though at least one person is still in custody) added fear and uncertainty to an already confusing situation. Most of this was unnecessary, as the election results used to calculate the Presidential winner should always and only have been those from the form 34Bs. The electronic results which came direct from the 40,833 polling stations to the portal were unofficial, incomplete (because they would and could never get 100% electronic results in a country so large and diverse economically as Kenya) and would inevitably differ (as they in fact did) from the 34Bs prepared at constituency level (mostly due to data entry errors into KIEMS by officials when transcribing manually from the completed forms). Repeated NASA allegations of hacking of the central IEBC server did not make great sense once it was clear that the central IEBC system was only being used for parallel presentation of polling station results from KIEMS. The actual presidential result came from the 290 constituency Form 34Bs. And the allegedly hacked portal had almost exactly the same result (8.2 m to 6.8 m) as that produced by adding the Form 34Bs.
The second significant concern was the delays in obtaining and then displaying the form 34As in IEBC headquarters. These were not (in the IEBC’s view) required for the central presidential announcement, but were still essential in order to determine whether the overall election was free and fair. No constituency RO should have announced their winners without all their form 34As, yet a week after they had finished, thousands were missing. The IEBC originally promised that “The results for the presidential election will be transmitted together with an image of the polling station tally sheet”. Then two days before polling, they announced what had already been widely suspected – that 11,000 polling stations did not have sufficient wireless network coverage – so the results from KIEMS would either come later or minus the scanned Form 34A copy. The whereabouts of these 11,000 forms became a huge problem. The IEBC was ambivalent and even misleading at times in its reporting. It seems they had not initially realised that the ‘one-time use’ model for KIEMS devices meant that for the polling stations where the system could not send the image but could send the results online, the scan of the form 34A would have to be provided much later by other means. These trickled in over the next 1-2 weeks, electronically or by hand. The IECB’s ambiguity over the 34As and the portal cost them dearly in perceptions of their competence and credibility.
Their failure to provide a display portal for the Form 34As and Bs was a mistake which was rectified, quickly for the Form 34As, and then grudgingly, a week after the vote, for the 34Bs. However, once done, it exposed gap between image and reality, when huge swathes of form 34As were found to be missing and some to be illegible. Those which were in the system matched well with the results in the online portal, but some were unsigned, unstamped or in a different format, and no-one knew what had happened to those which were missing. Some reports suggested the gaps were politically material (e.g. disproportionately from Odinga’s homelands).
It now appears that some media houses were ordered not to report on constituency contests, which might lead to suspicion that something deeper was amiss.
This linked to a more systemic concern – the back office operation of IEBC headquarters. While on the face of it, Wafula Chebukati, Ezra Chiloba and other commissioners maintained a relaxed face, and the portal and forms systems worked well, exactly where the portal results were coming from and why so few Form 34As were available has never been fully explained. It seems that administratively things were far from smooth in the back office. Basic security controls were lax, with IEBC staff frantically updating systems with whatever data they could get using various userids, some of their much vaunted document security features were invalid, key constituency documents were duplicated or unsigned and some officials were not even gazetted. There are still no published results apart from those on the portal for any of the other elections – no Form 35,36 ,37 and 38 for the parliamentary, gubernatorial, women representative or senatorial results have been published anywhere. The IEBC portal has results, but they are still incomplete nearly a month after the election, and differ from the (fragmentary) official results gazetted by IEBC on 18 August. In general, the results reporting and display process was unclear and IEBC did not always follow the procedures it had promised pre-election to ensure transparency and build confidence. The evidence from NASA’s petition showed numerous data and quality integrities, which while they were modest in individual impact and probably affected all candidates (and therefore would have limited material effect on the election result) certainly led many to question what was happening behind the scenes.
Another concern (less widely known) is the way in which the Kenyan media focused entirely on the electronic portal for their results, making no effort to report the actual constituency results. No independent tally was maintained and for the first time ever the press did not report any Constituency presidential, parliamentary or other results as announced. Initially I has thought that was simply practical laziness – since the portal was available and online – but it seemed inexplicable that the media were not reporting any of the announcements at all. It now appears that some media houses were ordered not to report on constituency contests, which might lead to suspicion that something deeper was amiss.
Still more concerns existed as to how individual presiding and returning officers behaved during their counting and tallying. Some Presiding Officers (for example in Mandera) were replaced the night before polling for unclear reasons. In some stations in pro-Jubilee homelands, NASA agents were not admitted and there was evidence in some stations of “top up” marking of unused ballots after polls closed. Many of the Form 34As had arithmetical issues or were not appropriately signed. It seems from NASA’s petition that some 34As may have been substituted with new (fake) documents or amended after counts finished (though KIEMS should prevent that, KEIMS didn’t work everywhere). In 13 per cent of polling stations, ELOG reported that Form 34A results were not displayed publicly as required by law. Some Form 34Bs show basic mathematical errors. There is also statistical evidence that (as in previous polls) presidential tallies were somehow inflated in the homelands (though there were few public protests at the time). For example, work in progress by Raiya Huru looking at the statistical distribution of Form 34A numbers suggests that in Murang’a, Nyeri, Nyandarua, Siaya, Kisumu and Homa Bay, the polling station results had been tampered with by someone (http://raiyahuru.com/Analysis.pdf). This matched well the NASA petition analyst’s view that something was amiss statistically with many of the results. The IEBC admitted that there were errors in the forms, but claimed they were not substantial enough to affect the outcome of the election.
The Presidential Election Part II
As the petition proceeded, life had begun to return to normal. The new MPs had been sworn in, governors had mostly completed their handovers, and for most Kenyans, the lengthy, expensive, diverting election was becoming a thing of the past. However, with the Court’s announcement we are now in uncharted waters, with the IEBC required to rerun the presidential poll within 60 days, for reasons which are not yet clear.
The IEBC should have been prepared for a runoff, so in theory all should be ready for a rerun. However, whether the IEBC can put together the temporary staff, the KIEMS devices, the logistics and the ballot papers in time for 17 October we do not yet know, especially as the IEBC itself is now under threat. So far Chebukati is staying put rather than resigning, but Chiloba has been side-lined entirely, as have several other officials (putting further stress on those who remain). But NASA is already objecting to the Supreme Court’s order that IEBC conduct a fresh poll in 60 days (because IEBC must be reconstituted), and IEBC has already decided not to conduct a full presidential poll anyway but only a second round runoff, based on the judgement in the 2013 petition [para 291] that “If the petitioner was only one of the candidates, and who had taken the second position in vote-tally to the President-elect, then the “fresh election” will, in law, be confined to the petitioner and the President-elect.”. And the precedent set in the Presidential petition would appear to allow every loser in the other five elections to annul every winner’s election on the same basis, if they can file a petition in time. So, more court cases loom while time runs out.
How effectively the two alliances will respond – without much time to raise money – to the need to do it all again no-one knows, but Jubilee are now grim, angry and spoiling for a rematch, which may well be dirtier than the first. My first guess would be that the result of the second election, if actually held, will be similar to that in the last, and in all the other “down ballot” elections, but until we know the real reasons why the Court annulled the vote, we do not know how much impact the irregularities they found may have had on the first presidential result. Victory in the courts may give the NASA camp fresh impetus and mitigate the pro-Jubilee bandwagon effect of incumbency, but Jubilee have a huge regional advantage (as they always did), more money and no intention of losing.
I had thought this would be my last piece, but perhaps we will need one more.
Support The Elephant.
The Elephant is helping to build a truly public platform, while producing consistent, quality investigations, opinions and analysis. The Elephant cannot survive and grow without your participation. Now, more than ever, it is vital for The Elephant to reach as many people as possible.
Your support helps protect The Elephant's independence and it means we can continue keeping the democratic space free, open and robust. Every contribution, however big or small, is so valuable for our collective future.
Counterfeiting, War and Smuggling: British American Tobacco Dirty Games in the Sahel
Billions of cigarettes, most made by BAT, are smuggled north through Mali every year on their way to the gray markets of the Sahel and Northern Africa.
Stashed inside pickup trucks and guarded by armed militias and jihadists, every year billions of illicit cigarettes wind their way through the lawless deserts of northern Mali bound for the Sahel and North Africa.
The profits from their long journey fuel north Mali’s many armed conflicts, lining the pockets of offshoots of al-Qaida and the so-called Islamic State (IS) group, as well as local militias, and corrupt state and military officials. This violence is now spilling out across West Africa, displacing more than two million people in Burkina Faso, Chad, Mali, and Niger.
Cigarettes made by one of the world’s largest tobacco companies, British American Tobacco (BAT) and distributed with the help of another major, Imperial Brands, through a company partially owned by the Malian state, dominate this dirty and dangerous trade.
Now an investigation by OCCRP can show this is no accident.
Secrets contained in leaked documents, backed up by trade data and dozens of interviews with insurgents, former BAT employees, experts, and officials, show BAT started to oversupply Mali with clean-labelled cigarettes soon after the north fell to militants, knowing that its product would be fodder for traffickers.
The profits of cigarette smuggling fuel the bloody struggle between jihadists, armed militias, and corrupt military officers that has turned northern Mali into a lawless warzone.
For years the company partnered with Mali’s state-backed tobacco company, a subsidiary of Imperial Brands, to distribute cigarettes in regions controlled by rebel militias and throughout the country. Sources say these cigarettes, trucked north with the help of the military and police, then fall into the hands of jihadists and militias. An internal document suggests BAT used informants in West Africa to keep abreast of the workings of the illicit trade.
The dirty business goes well beyond the desert. OCCRP’s reporting found the Malian government not only helps to distribute BAT’s cigarettes, but also apparently turns a blind eye to gross accounting irregularities at its partner Imperial and even possible trade fraud.
And it continues today. Public trade data and expert analysis show BAT and Imperial continue to oversupply the country with billions more cigarettes than it needs. Meanwhile, BAT’s annual revenue in 2019 alone exceeded the total GDP of Mali and Burkina Faso.
The Malian case is the latest to show the world’s leading tobacco companies are not always abiding by the terms laid out in a series of historic agreements between 2004 and 2010 with the European Union (EU), in which they agreed to prevent their cigarettes from falling into the hands of criminals by only supplying legitimate demand. The agreements were concluded in the wake of legal disputes between three companies and the EU over cigarette smuggling.
“This is their playground,” Hana Ross, a University of Cape Town economist who researches tobacco, said of the industry.
“They know they can get away with stuff. It’s much easier to bribe. It’s much easier to cheat the system,’’ she said. “Governments here are generally weak. This is where they do things that they don’t dare to do in Europe anymore.”
A spokesperson said BAT was opposed to the illegal trade in tobacco, which the company called a “serious, highly organized crime.”
“At BAT, we have established anti-illicit trade teams operating at global and local levels. We also have robust policies and procedures in place to fight this issue and fully support regulators, governments and international organizations in seeking to eliminate all forms of illicit trade.”
BAT started to oversupply Mali soon after the north fell to militants, knowing its product would be fodder for traffickers, according to dozens of interviews.
Imperial said it is committed to ensuring high standards of corporate governance and “totally opposed to smuggling which benefits no-one but the criminals involved.”
The Malian government did not respond to requests for comment for this story.
The Tobacco People
In the deserts of northern Mali, cigarette smugglers are called “kel tabac,” the tobacco people.
Illicit cigarettes from the capital, Bamako, and ports in Guinea, Benin, and Togo are loaded into convoys with armed guards and driven north along thousands of kilometers of winding roads and desert tracks to Libya and Algeria, and as far east as Sudan.
Smuggling has long been a part of life in the vast and largely empty Sahel region, where armed insurgents claim a patchwork of ever-shifting territories. Jihadist movements linked to al-Qaida and IS, Tuareg separatist forces, and local ethnic militias take turns controlling roads and checkpoints along the way.
Moving illegal tobacco is a difficult and dangerous job, with trips taking between three and 10 days. Many truckers are killed by military or armed groups along the way. But it is well-paid: In a country where most people live on less than $1.90 per day, drivers can expect to earn between 6,000 to 10,000 euros for moving a load of contraband cigarettes.
It is also a lucrative trade for the drug lords and corrupt local officials in Mali’s restive northern regions.
Hama Ag Sid Ahmed, spokesman for the National Movement for the Liberation of Azawad (MNLA), an armed Tuareg independence movement that has controlled much of northern Mali on and off, said state officials and organized crime work together to profit from smuggling.
“Certain military officers, members of the intelligence services, heads of military zones in the northern regions are approached by drug lords,” he said.
“Large sums of money are paid for a contract related to a service rendered or to be rendered.”
A former tobacco industry insider said various militant groups, from the Tuareg separatists who have been fighting the Malian state for decades to the more recent offshoots of IS jihadists, also take a cut along the way.
“Product is escorted north by the Malian army or the gendarmerie [police], to protect it from so-called bandits,” said the former official, who would only speak on condition of anonymity due to safety concerns. “It would be given to the Tuareg for the trip onwards near Timbuktu, and then the Tuareg looked after paying IS in the Sahel.”
With the continuing violence and lawlessness, Malian customs have abandoned much of the north. Samba Ousmane Touré, an ex-employee of BAT’s distributor in Mali who is now a member of the country’s tobacco control committee, said armed groups have become the gatekeepers of the smuggling routes towards Algeria, Libya, and Niger.
“Armed groups play the role of customs,” he told OCCRP. “Yes, [BAT] knows.”
One of the most high-profile jihadists in northern Mali, an al-Qaida operative known as Mr. Marlboro, is thought to have financed his jihad by smuggling cigarettes.
The one-eyed Mokhtar Belmokhtar allegedly orchestrated terror attacks, including one in Algeria in January 2013 that killed more than 35 people. He led the so-called Those Who Sign in Blood Battalion. In June 2013, U.S. authorities offered a reward of up to $5 million for information leading to Belmokhtar’s location.
His battalion had ties to key Malian armed groups, reportedly providing crucial military assistance to the terrorist group MUJAO against the MNLA during the battles of Gao and Timbuktu. A senior U.S. official said in July 2013 that Mr. Marlboro “has shown commitment to kidnapping and murdering Western diplomats and other civilians.” One such hostage was the former U.N. Niger envoy Robert Fowler.
Sid Ahmed, the spokesperson for the MNLA, said many terrorists like Belmokhtar started out trafficking cigarettes before moving onto harder substances, and then to violent jihad.
“The Arab drug barons created armed militias to protect their drugs and which later developed into the terrorist organizations that are present today in the Sahel region,” he said.
Research from The Global Initiative Against Transnational Organized Crime argues the long established smuggling networks in Mali and the Sahel evolved “first to move illicit cigarettes, later hashish and then, most profitably, cocaine.”
A 2017 KPMG report agrees, noting that the region’s cocaine trade overlays routes originally used to smuggle cigarettes, and that illicit trade “can also intersect with the operations of terrorist groups.”Illicit trade is “an important component of the local political economies” of Mali and other countries in the Maghreb, said the report, which was sponsored by Philip Morris, though it claims the trade is fueled by illicit cigarettes from free-trade zones in the United Arab Emirates.
Raoul Setrouk, who is pursuing a court case against BAT competitor Philip Morris in the state of New York for intellectual property theft, said that illicit tobacco in the region has consequences that go far beyond health and tax issues.
“I hope we don’t have to wait for a new Mr. ‘Marlboro’ like terrorist Mokhtar Belmokhtar to raise our consciousness,” he told OCCRP.
Multiple sources, from soldiers and U.N. employees to businessmen, and armed militia members, told OCCRP that brands made by BAT and Philip Morris dominate the illicit trade.
Most common are Dunhills, produced in BAT’s factories in South Africa, and Philip Morris’ flagship brand Marlboros, which are handed to smugglers linked to armed groups by PMI’s politically connected representative in Burkina Faso, along with American Legends.
“Those which transit through are mainly three brands: Dunhill, American Legend and Marlboro,’’ said Hama from the MNLA. “It is the same thing also in northern Niger and not far also in the south of Algeria.”
Mohamed Ag Alhousseini, an independent researcher in the region, said much the same: “Even in Algeria, the trafficking is encouraged by the need of Marlboro and Dunhills, because they have other brands in the country.”
It’s hard to determine exactly how many illicit cigarettes are smuggled through Mali.
Trade data, information from customs officials, leaked BAT documents, and industry experts indicate there may be up to 4.7 billion surplus cigarettes in Mali every year — the equivalent of around 470 shipping containers of extra cigarettes. Some of them are produced in the country, but more are imported, almost all of them from South Africa.
Mali’s government has ignored years of blatantly false tax figures from Imperial Brands, a shareholder of the state tobacco company that distributes Dunhills in militant-run areas.
It’s also tricky to determine how much profit BAT makes because the company doesn’t separate out country figures in its annual reports. A company presentation from around 2007 estimates BAT’s market value in 18 “operational markets” in West Africa at 201 million British pounds (about US$394 million), and its market share in Mali at 61 percent. Another document, from 2012, gives gross turnover for Mali of 52.06 million British pounds ($84.6 million).
A BAT source, by contrast, estimated the company had a gross turnover of over $160 million in Mali in 2019 alone.
Imperial said SONATAM’s sales are “commensurate with the legitimate demand of the Malian population” and the company operates a stringent sales monitoring system.
“All cigarettes imported by SONATAM into Mali are done so legally under synallagmatic contracts with other commercial operators,” the company said in a statement.
Understanding Mali’s illicit cigarette trade is a messy business — and that includes the data behind it. Because the illicit market is so opaque, many of the calculations rely on educated guesswork.
Euromonitor International, a strategic market research company, estimated the country’s retail volume at 3 billion cigarettes in 2016, rising to nearly 3.2 billion in 2020.
Leaked documents obtained by the University of Bath and shared with OCCRP show that in 2007, BAT estimated the country had demand for 1.9 billion cigarettes. In 2011, the company upped the estimate to 2.4 billion. Both these figures are lower than independent projections for the same years.
After northern Mali became a war zone, however, BAT’s calculations changed, with documents from 2013, 2014, 2015, and 2017 estimating the market as significantly larger than Euromonitor’s figures, at between 3 to 3.8 billion sticks.
The reason behind these high figures is unclear, as the same documents contain estimates of Mali’s smoking prevalence that are below the WHO’s. Experts have varying estimates for smoking rates. In 2011 BAT pegged it at 9.5 percent. The World Health Organization, by contrast, says 12 percent smoked in 2017, a rate that has remained steady over the past decade.
Yet data shows that every year since 2016, the first year after Mali’s 2012 rebellion for which trade figures are available there may have been up to almost 8 billion cigarettes in Mali.
Exact figures are hard to determine. A Malian customs official estimated an annual total of 4.6 billion cigarettes based on adding imports (2.6 billion in 2018 and in 2019 each year) with local production (around 2 billion in 2018 and in 2019 each year).
U.N. Comtrade data, however, shows between an estimated 3.4 billion to 5.9 billion cigarettes were exported to Mali per year from 2016 to 2019, nearly all of them from BAT’s regional hub, South Africa. Adding in local production, that could mean as many as 7.9 billion cigarettes are available in Mali each year.
Officials in Mali and South Africa confirmed the accuracy of the Comtrade numbers, which closely match regular reports on the value of tobacco imports released by the Malian government.
Hallmarks of an Illicit Trade
In Gao, a city in northern Mali that has long been under the control of armed groups, a warehouse that distributes BAT’s cigarettes does a brisk trade.
Ahmoudou Ag Attiane, a local automotive dealer, told OCCRP that 20-ton tractor-trailers stocked with cigarettes commonly arrive at the warehouse. Many of the cartons are then trucked 10 hours north to Kidal, which is controlled by al-Qaida in the Islamic Maghreb (AQIM).
“The law is the [AQIM group] that has the most power — the terrorists, the jihadists — and they banned smoking and also alcohol. So you see, someone can’t show off too much by opening up a place where everyone knows this is where cigarettes are stored, this is where cigarettes are sold.
“All these big traders have relations with the big boss of Kidal,” he said, “which means that they are protected.”
Sid Ahmed, the MNLA spokesperson, added to this point, saying: “The traffickers make a large order with a merchant in Gao or Timbuktu. The traders transport [product] from Bamako to Gao and or Timbuktu. From Gao it goes to Algeria [and] Libya and from Timbuktu it goes to Mauritania and Algeria.”
The company that runs the warehouse, SONATAM — the state tobacco company whose shareholders include Imperial and the Libyan Arab African Investment Company — has been BAT’s distributor in Mali for years. Many of the cigarettes that pass through its warehouse in Gao are Dunhills from BAT’s plant in Heidelberg, near Johannesburg, which have accounted for up to 37 percent of South Africa’s total cigarette exports in recent years.
Unlike locally produced brands, the South African Dunhills come in packaging covered with health warnings in a major European language, French, known in the industry as a “clean label,” meaning they can be sold on the gray market.
David Reynolds, who built Japan Tobacco International’s program on countering the illicit tobacco trade, said BAT in South Africa is “notorious” for oversupplying the region.
“The rule is always the same: Oversupply plus lack of local controls leads to gray trade. That’s been a big part of BAT’s — and other cigarettes companies’ — business model for years,” he said.
“If you combine a major, high-end international brand, plus oversupply in a marginal market, such as Mali, with a clean label, you have all the hallmarks of intentional diversion into the parallel [illicit] trade.”
Documents obtained by OCCRP shed further light on how BAT’s Dunhills fall into the hands of armed groups in northern Mali.
A document from 2013 show SONATAM distributes between 25 percent to 75 percent of the three brands of BAT’s cigarettes sold in Mali. Three of its warehouses and distribution points are in rebel-controlled areas, including Gao, as well as Timbuktu and Mopti in the north of the country.
One BAT presentation from 2013 calls northern Mali a “war zone,” but notes that BAT has nonetheless identified future stockists and networks in Gao, Timbuktu, and Kidal. Another from 2017 highlights the “extremist insurgency” in eight of Mali’s regions, noting that three of them “remain completely dangerous to operate within owing to terrorist activities.”
However, an internal strategy memo from 2015 shows BAT planned to increase its business in these regions. The plan, called “Desert Storm” in an apparent reference to the U.S.-led military operation during the Gulf War, discusses how to reach “full potential” for their brands in Mali by incentivizing SONATAM to meet sales targets in areas including insurgency-run regions.
“As we know, in a dark market, the war is won on the battlefield with no pity for our competitors,” said the memo.
A 2007 presentation echoes the language of Europe’s colonial-era Scramble for Africa to describe the contest for the “crown jewels” of Mali and Ghana, casting West Africa as a battleground and speaking of “fighting ITG [Imperial Tobacco Group] to the death” and a “PMI [Philip Morris International] attack.”
“Mali was such an important market that BAT undertook a two-pronged strategy,” said Andy Rowell, a University of Bath researcher working with anti-tobacco watchdog STOP.
“The company set out to secure a ‘license to operate’ by schmoozing government officials. At the same time, the company sought to ‘delay and disrupt’ the operations of the opposition.”
Other BAT documents lay out its strategy to increase its market share against lower-cost cigarettes in Bamako and “UPC” — jargon for “Up Country” — including detailed analysis of the competition. They also show the company’s fine-grained ability to map and track contraband in West Africa: One presentation from around 2006 lists BAT’s “informants” in Mali and Niger.
Telita Snyckers, a lawyer who previously held senior positions at the South African Revenue Service and author of the book Dirty Tobacco: Spies, Lies and Mega-Profits, called the operation “corporate espionage stuff.”
The slides of the 2007 presentation discuss BAT’s strategy for West Africa, including Mali, stressing the need to “Grow VFM in Freedom Markets and Mali.” Snyckers said that VFM, or “Value For Money,” is a euphemism for smuggling and illicit channels.
In another presentation from 2009, a group of legal and security officials from BAT was told that “Mali, as the principal market which has the highest volume of illicit trade, is where we have the most to gain by increasing contestable market space.”
A BAT spokesperson declined to comment on the documents without seeing them before the publication of this article, but added, “we are not aware of the phrases ‘dark market’ or ‘value for money brands’ relating to illicit trade.”
Extraordinary Mistakes or Barefaced Lies
The rampant tobacco smuggling in Mali isn’t only down to the cigarette companies. OCCRP’s reporting indicates there is little state oversight of the industry.
For one thing, the government has overlooked blatant inaccuracies in figures from BAT’s distribution partner, Imperial, which for two consecutive years stated in its public accounts that SONATAM paid 5.5 million euros in taxes more every year than its total turnover.
West African financial analyst Oumar Ndiaye called the numbers “impossible.” Some former tobacco executives in Mali dismissed the SONATAM turnover figures as deliberate lies to fiscal authorities.
Imperial attributed them to an error in currency conversion, with West African CFA francs mistakenly not converted into euros. The company declined to provide documentation, however, and referred reporters to the Malian government, which did not respond to several requests for comment.
Alex Cobham, the chief executive officer of the Tax Justice Network and an expert on tax avoidance by multinationals, said Imperial’s explanation “doesn’t stand up,” and that repeating the same numbers over multiple years is “implausible.”
“Whoever wrote these numbers down thought nobody would ever look at them,” he said. “They’re either making extraordinary mistakes, year after year, or they’re telling you barefaced lies, or both.”
He also faulted the company’s auditor, PricewaterhouseCoopers, for apparently accepting the shoddy accounting.
“The idea that one of the world’s leading accounting firms, that prides itself on the auditing of multinationals to ensure they’re behaving as they should do, would not have picked up any of this in their rigorous annual audit process is difficult to square with any claim that corporate tax is being paid or audited on an appropriate basis,” he said.
It’s unclear who put together the “impossible” numbers.
Imperial inherited much of West Africa’s tobacco business from Bolloré Group, a giant in France’s former colonies which operates a number of ports across Africa and logistics companies worldwide.
The tobacco purchase bought Imperial a stack of elite connections. The directors of SITAB, an Imperial subsidiary in Ivory Coast, included a relative of former President Felix Houphouet-Boigny. Lassine Diawara, the chairman of the board of directors of MABUCIG, a Burkina’ cigarette manufacturer. His online biography says he is a Knight of the National Order of Merit in France. He has traveled with Blaise Compaoré, the ex-president of Burkina Faso. SONATAM was run for a number of years by Cissé Mariam Kaïdama Sidibé, who became prime minister of Mali for a short period in 2011.
Ross Delston, a U.S.-based lawyer and anti-money laundering compliance expert who has worked in West Africa, said the Malian government could well have an incentive to overlook years of obvious errors.
“Any governmental authority that has a monopoly over a given commodity also has a high degree of risk for corruption,’’ he said after discussing SONATAM figures with OCCRP. “It’s just too easy to skim off a bit, or more than a bit, for the people at the top.”
Touré, the ex-employee of BAT’s agent in Mali, agreed, saying that the state shared in the responsibility for the bad accounts, adding, “I think that [in] corrupt states like ours, the tobacco industry has a lot of power over their leaders.”
Mali’s government declined to comment.
U.N. trade figures also indicate years of discrepancies equaling millions of dollars in the price of the country’s cigarette imports.
Mali imported more than 3 million kilograms of cigarettes from South Africa annually in both 2016 and 2017, representing around 95 percent of the country’s cigarette imports. An ex-BAT official said that the only cigarettes Mali imports from South Africa are BAT’s Dunhill cigarettes, a point confirmed in an earlier BAT document.
If the former employee is correct, BAT reported to the government of South Africa it sold the cigarettes for under $7 per kilogram, while SONATAM reported it bought the cigarettes for $15 per kilogram in 2016 and 2017, the years for which U.N. trade data is available for Mali. The discrepancy amounts to between $29.1 million and $32.8 million per year, and appears to have continued afterward, according to Malian government data available for 2018.
It’s unclear exactly what is behind the difference.
A Malian customs official dismissed the numbers as a likely lag in reporting shipments.
Two former tobacco industry insiders told OCCRP that trade mis-invoicing, a method for moving money across borders that involves deliberate falsification of the volume or price of goods, is common practice in the company’s dealings with Mali.
“Mis-invoicing, under- and over-invoicing, and invoicing direct to the U.K. instead of in the delivered country were all used at one time or another,” one of them said.
Cobham, of the Tax Justice Network, said SONATAM’s overpayment is “very much consistent with the longstanding history of commodity trade price manipulation for profit-shifting purposes.”
That’s apparently not unusual for BAT. In 2019, Cobham’s organization authored a report that found BAT used various methods to shift profits out of poorer countries, at a scale that could deprive eight countries in Asia, Africa, and South America of nearly US$700 million in tax revenue until 2030.
“The bottom line is BAT is manipulating the price of the same commodity and the transaction in a way that can’t be justified by any possible transport costs, and any auditor worth their salt should have picked that up,” he said.
SONATAM did not respond to requests for comment.
Imperial did not respond to several OCCRP requests for clarification, saying only that the company “is committed to high standards of corporate governance” and “totally opposed to smuggling which benefits no one but the criminals involved.”
A BAT spokesperson said the prices of its tobacco “are in line with what external, independent parties would charge,” which is documented in the company’s tax strategy.
“BAT entities … comply with all applicable tax legislation and regulations in the countries where we operate,” he said.
PricewaterhouseCoopers and its French partner Xavier Belet, who audits the SONATAM accounts, ignored several requests for comment by OCCRP.
Friends on the Ground
From warehouses in Gao, Timbuktu, and Mopti, Dunhills flow north largely unchecked by Malian regulators.
“With the insecurity, the customs abandoned an important part of the north because of the narco-traffickers,” said Aboubacar Sidiki Kone, a Malian customs official.
Even if customs did man Mali’s lonely desert posts in the north, it’s unclear what they would do. An internal document obtained by OCCRP shows Malian customs and police were sponsored by BAT.
In a 2013 presentation, BAT lays out an “action plan” for a series of scheduled raids to be carried out by Malian customs and police in collaboration with company agents, tallying seizures of illicit cigarettes made by its competitors. A mission order and a protocol agreement in the presentation show BAT was supposed to pay for these raids.
Internal documents show BAT used informants in West Africa to keep abreast of the illicit trade.
A former BAT employee described staffers in Mali feeding intelligence on contraband to customs agents, helping them to seize the brands of other manufacturers.
Sory Coulibaly, a former sales executive for a BAT distributor in Mali, added that BAT has sweetened the deal, equipping customs agents and police with motorcycles and small patrol boats. Touré added that BAT has given customs several new cars every year.
The cooperation between Mali’s customs and BAT was formalized further in 2019, when local media reported Malian customs’ announcement of a memorandum of understanding (MoU) with the tobacco company.
Deals with customs agencies are a longtime tobacco industry strategy, detailed in a paper published by the BMJ’s journal Tobacco Control the same year. Eric Crobie, Stella Bialous, and Stanton A. Glantz found that there are more than 100 such MoUs around the world, that they violate the World Health Organization’s international tobacco control treaties, and are ineffective at reducing smuggling.
Memoranda of Understanding (MOUs) were seen by transnational tobacco companies as “useful to provide access to decision makers and promote the image of [tobacco companies] as government partners,” the authors wrote.
In Mali’s case, the details of neither its deal with BAT nor an MoU it signed with SONATAM are easy to find. Abdel Kader Sangho, director of the customs’ training center, ignored several inquiries from reporters.
Touré, the Malian tobacco control expert, said the country’s tobacco laws are weak and there is little enforcement of them on the ground. “Our anti-smoking texts are not strong and most of our leaders are corrupt,” Touré said. “The texts exist, but it remains to apply them in the field.”
Today, SONATAM’s statistics claim Mali’s contraband levels are at an all-time low, while BAT continues to flood the country with cigarettes far exceeding demand.
Anecdotal evidence suggests the flows of smuggled tobacco may even be increasing. Touré said he has observed that the amount of Dunhills moving to the north, have recently been on the rise.
“I’m sure these cigarettes are destined for other countries, Niger, Algeria and others,” he said.
Meanwhile BAT and the Malian government are planning to make more cigarettes in the country. In 2017 they partnered up to build a new $18.2 million factory, according to local media reports. It is expected to open this year with the capacity to produce 3 billion Dunhills per annum.
Sandrine Gagne-Acoulon contributed reporting.
Who Are Kenya’s 42+ ‘Tribes’? and Should We Be Asking?
Asking whether or not the census should continue to count ethnic groups is one way into the difficult conversation about how to reckon with the legacies of colonial weaponisation of ethnicity.
It was a hot and dusty day in January 2019. Sam had been driving me around Nairobi since my first visit, ten years earlier. I often float ideas past him as we endure interminable traffic. “Sam, how many tribes are there in Kenya?” I knew there was no definitive answer but I wanted to know his thoughts. “Well, now we are . . . is it . . . 46? Or 47? We used to be 42 but some new ones were recently added. Makondes. Asians. Who else was it? Nubians . . .” “And where is the list?” I probed. “Oh that one . . . is it gazetted somewhere? I don’t know.” Later that day, while he refined my left hook, I asked my boxing trainer. Embarrassed, he laughed and said “You know . . . I’ve not brushed up on my tribes lately . . .” “Just roughly . . . how many?” He replied after some thought, “I think . . . well . . . I know that we used to be . . . is it 41? Or 42? 42. We used to be 42. But now, I don’t know.”
In multiple interviews with various government officials I was repeatedly told there were 42+ tribes, but nobody could tell me the nature or location of the list. “Do you know?” one official asked me. Ten years earlier, I had asked members of the minority Nubian community too: “Forty-two tribes. And we will be the 43rd.” They even had a letter from a Minister declaring they would, indeed, be counted as such in the 2009 census. But I struggled to find the list. Who is on it? Does it even exist? And if so, who controls it, and how? Why does nobody know? And does it matter?
In my research, this idea of “the 42” kept coming up over and over again. I have been conducting academic research in Kenya since 2009, mostly with the minority Nubian community which has long sought recognition as Kenyan, and has had considerable success in recent years in getting it. It was my first interviews with Nubian elders in 2009 that made me start wondering about this idea of “the 42”, where it comes from, why it matters.
So why does it matter?
Being recognised as a “tribe of Kenya” is important to people. It’s important symbolically as it makes people feel like legitimate citizens. And it is important materially, or at least there is an anticipation that it is. There is a belief that if you are one of the tribes of Kenya, then you can access the state’s resources. The exact mechanisms through which this is expected to happen include, for example, revenue sharing to the counties, drawing of administrative and electoral boundaries, and accessing special provisions like the Equalisation Fund. There is a popular belief that these are somehow connected to ethnicity, even though many Kenyans will point out they mostly shouldn’t be.
Counties, wards and so on are often treated as if they “belong” to a particular group. So, the idea is you have to be a recognised group to get your hands on government resources. Whether this is true or not, the perception that it is matters a lot for how people feel they belong, and how they might feel they are in competition with each other for resources. Plenty has been written on inter-ethnic competition and tribalism in Kenya. That’s not my focus here.
There is a belief that if you are one of the tribes of Kenya, then you can access the state’s resources.
At another level, the idea of “the 42(+)”, or the idea that there is or could be a list somewhere, matters for debates – prominent here in The Elephant among other places – about what it might mean to decolonise identity. On one hand, I’ve heard some Kenyans suggest that Africans should abandon ethnicity altogether, as it is a colonial construct used by the British and other imperial powers to conquer; to divide and rule. On the other hand, there is an argument that ethnicity is an important facet of African identities, and that these days “the West” has turned around and wants to eradicate it, especially around elections; therefore, the anti-imperial thing to do would be to affirm ethnicity. Both arguments have merit. My proposition here is not to take a strong position on either side, but to look at this idea of “the 42(+)” and its bureaucratic origins as a way of thinking through this debate. Decolonising identity is not only a personal thing – it is also a bureaucratic thing.
The title of this essay, and the academic research article on which it is based, is, then, deliberately provocative. I never thought – and my research confirmed this – that there would be a clear answer to these questions. I have never even been sure that “who are the tribes of Kenya?” is quite the right question to be asking. It carries some very politically loaded assumptions: that “tribe” is an appropriate term (more on this below); that there is a clear-cut way to determine who is and isn’t Kenyan based on their ethnic identity; that there are only 42 (or 43, 44 or 45) ethnic groups which can call Kenya home. My suggestion here is that asking why we ask this question is more important than the question itself.
The census is the only official list of “all” ethnic groups, and the only official tool to count the population by ethnicity. And 1969 is the only year that 42 ethnic groups were counted. Voter rolls prepared by the Independent Electoral and Boundaries Commission do not record ethnic identity.
Electoral boundaries do not involve listing ethnic groups. Boundaries are connected to the census – insofar as they draw on population data – but before the 2010 constitution they bore no official relation to ethnic data. The 2010 constitution allows for a possible use of ethnic data. Under chapter 7, one explicit consideration for boundary redrawing is “community of interest, historical, economic and cultural ties”, which could potentially be interpreted to mean ethnic communities. However, the exact role this clause – or ethnicity more generally – now plays in boundary drawing is not clear.
The civil service doesn’t list ethnic groups. Civil service employment records routinely record and make public how many people are employed in the civil service from each ethnic group, but that only captures, of course, civil servants. To establish the “fairness” of each ethnic group’s share of civil service jobs, that data is compared to census data, but only at the national level or by problematically inferring ethnicity by location – for example, by assuming that if you live in Ukambani you must be Kamba.
Identification cards do not record ethnicity.
Nor, contrary to popular understanding, does the Kenya Gazette (the government’s official announcement record) list ethnic groups, although it was used as if it does when Asians were gazetted as the 44th Tribe of Kenya in 2017, despite no identification of the preceding 43.
So that leaves us with only the census.
In my research, I compared all ethnic classifications in all Kenyan censuses from 1948 to 2019. I looked at every census report, but also, where available, all the questionnaires used by enumerators when visiting households, instructions to enumerators about how to record “tribe”, explanations made by the Bureau of Statistics and its predecessors for what “tribe” means and why they chose the lists they did, and archival material (for 1948 and 1962) where colonial administrators debated in letters and meetings how they would conduct the census.
The list of “tribes” has changed in every single census, and since the first census in 1948, 150 different groups have been named. Of those, there are only 14 ethnic groups which have been named and counted exactly the same way in every census. The others have all changed, sometimes multiple times, for example by adding or deleting “sub-tribes”, by moving from a “sub-tribe” to a “main tribe” or vice versa, or by appearing or disappearing altogether. There are also some instances where a “tribe” was listed on the questionnaire but didn’t make it to the final census report, or where – curiously – they were not listed on the questionnaire but did make it to the final report.
You might recognise your ethnic group(s) in this list, possibly in multiple forms as some groups have changed names over time (e.g. Sudanese to Nubi), or even – unfortunately – in a derogatory form (such as Dorobo, which was only removed in 2019 because it refers to having no cattle, suggesting some form of inferiority). Some groups included on the list for “the tribe question” aren’t even really tribes: for example “Stateless” in 2019, or “Kenyan” in 2009.
So, how are these lists determined? There is no transparency on how these lists are decided, or what it means to be “coded”.
The first census in Kenya was carried out in 1948 and was part of an East African census that included other British territories in the region. More interested in the European population than the Indigenous one, the “non-native” census was extremely thorough, and the “native” one much more basic. Whereas all kinds of details that are useful for development purposes were gathered for the white population, the only three statistics gathered for the African population in every household were age, sex and – you guessed it – “tribe”. For Census Superintendent C. J. Martin, it was so obvious that you would count “tribe” that, in his extremely detailed report on the census, he didn’t even bother to explain why. Other factors that are much more useful in making sense of a population’s development needs, like fertility, education and occupation, were only counted for 10% of the African population in a sample census, and then generalised.
The list of “tribes” has changed in every single census, and since the first census in 1948, 150 different groups have been named.
The actual list of “tribes” that enumerators were given in 1948 was also, for Martin and the other census organisers, self-evident. The British authorities acted as if it was obvious which ethnic groups should be counted, but it clearly wasn’t, because there were differences between the list provided on the questionnaire, and that which appeared in the final report. We can only assume that any range of factors may have shaped the final 1948 list, including self-identification by householders, initiatives on the part of the enumerators or District Commissioners who compiled the returns, or maybe even political lobbying. In other words, determining the tribes of Kenya was not as self-evident as Martin imagined. Decisions about which ethnic groups, what names they use, how they are spelled, what and whether “sub-tribes” are counted and so on, always have to be made by someone.
But the thing about a census, as with so many official tools, is that it gives off an air of authority. When a list like that of “Kenya’s tribes” is made in this way, it comes to feel as if it is definitive, even when it never can be. Even though every census after 1948 has changed the list, it always builds on that first list made by British administrators, some of whom had very little understanding of the communities they were counting and classifying.
In 1962, the list was very similar to the one of 1948, but it dropped most of the ethnic groups which mostly live in other parts of East Africa (Tanzania, Uganda) and added some from the North and East of Kenya. By this time, the British authorities had established much more administrative control in those regions and had learnt of new groups not included in that earlier census, showing again major gaps in their knowledge of the people they had colonised. Morgan, another colonial administrator, this time involved in the 1962 census, later admitted that the concept of “tribe” was a bit arbitrary, but stuck to it anyway, stating:
[Tribe is] a unit which evades satisfactory definition but which was widely recognised. It may be said to be a group to which the individual feels a strong sense of belonging and which is usually distinguished by a common language and culture and, since marriages are mostly within it, may have inherited traits. […] For this study we have to accept the classification used in the census, for which no justification was published. The ascriptions were those routinely used by the administration and which appear to have presented few problems to those recording or those being recorded. They were the socio-political groups encountered by the colonial power upon its entry and with which it had to deal. Administrative boundaries were normally constructed to contain them and this probably increased the sense of tribal identity at that level.
Though he admits some arbitrariness, Morgan goes on, in this passage, to suggest again that it was obvious, uncontroversial and accepted by everyone – African and colonial administrator alike – who the “tribes” of Kenya were. If this was really the case, why then would it have changed?
The 1969 census, the first one conducted by the first post-colonial government, used the same list as was used 1962, but added two more Somali groups, without really explaining why. The 1979 census used the same list again, but collapsed a number of groups into “Kalenjin”. It is likely no coincidence that this happened the year after Moi became President, and Gabrielle Lynch has done some great research about the creation of the Kalenjin identity around this time. In 1989 there were only a few small changes. In short, with the exception of the introduction of Kalenjin as an ethnic group rather than just a linguistic group, the list remained pretty similar to the colonial-created one for the first three decades of Kenya’s independence, but not similar enough to agree with colonial officials Martin and Morgan that it was ever truly “obvious” which ethnic groups should be counted.
By 1999, with the politics of democratic reform in full swing, and the effects of Moi’s majimboist politics being felt across the country, no results were published on ethnicity from that year’s census. It was too sensitive.
Then, come 2009, only eighteen months after the post-election violence of 2007-08, the list of ethnic groups in the census underwent its first radical change since independence, with the number of groups skyrocketing to well over a hundred. This included long lists of “sub-tribes” for groups such as Swahili, Kalenjin, Mijikenda and Luhya, as well a considerable number of newly recognised ethnic groups, including Nubians (last counted in 1948 as “Sudanese”). The political mood was an inclusive one, seeking peace and inter-ethnic harmony. It felt right at the time to generously offer recognition. And it didn’t hurt that chopping up the population into lots of small groups might help cool the temperature on inter-ethnic competition between the larger groups. The 2019 census added yet more sub-tribes and new tribes, moved some around from one category to another, and renamed a few.
The only thing the history of the census classifications shows conclusively, then, is that there cannot be any conclusions. The census, though it has an air of officialdom, is really just a result of layer upon layer of bureaucracy, politics and coloniality. Politicians and civil servants might want to bed this down and make it feel certain, but they can’t. It changes every decade. They also can’t, practically, start from scratch either. The lists they have built are based on everything that came before – both colonial and postcolonial. They bear the markings of all the political moments in which censuses were conducted, and the particular concerns of politicians and statisticians at those times. And this is true of every census, everywhere in the world. They are not foolproof. They are not certain. They are not conclusive or definitive. The idea of the 42(+) is just that – an idea – however widespread and deeply believed.
The only thing the history of the census classifications shows conclusively is that there cannot be any conclusions.
The reality is that there is no definitive list of Kenya’s ethnic groups. That is, there is no list that does (or could) state with certainty and finality who the ethnic groups of the nation are. But there are official lists – those in the census – that are often perceived as certain, and those have to be reckoned with.
How colonial is ethnicity?
From one perspective, the story of ethnic classifications in the census is interesting as a puzzle. Working out who got added, who got removed, when, how and why is fascinating. There is a lot to be learnt about Kenyan history and ethnicity by looking at the details.
But from another perspective there is a bigger question to be considered here, and that is about whether, how, to what extent or in what ways ethnicity is colonial. The Elephant and other discussions in various forums are increasingly – and rightly – working through what it might mean to decolonise African identities. From renaming streets to pulling down monuments to pushing back against arbitrary determination of one’s identity by another, Kenyans and other Africans are questioning why ethnicity is such a strong form of identity; in what ways it was imposed by the colonial experience; and in what ways it has changed or should change form, or maybe even be abandoned.
Terence Ranger, a keen scholar of Kenya but also a former colonial official, coined the term “invention of tradition” to explain how the British came, saw, and invented ethnicity or – more specifically – “tribe”. Seeing Africans as being defined first and foremost by tribe allowed the British to divide and rule, and to imagine they were not just extracting and exploiting, but also civilising. The roots of ethnicity, in this sense, are problematic. The concept itself as well as the specific ethnic groups the British identified and made names and Native Reserves for, were fundamental tools of colonial control. Ethnicity kept Africans divided from each other and in a supposedly inferior place on the hierarchy of civilisation that justified British colonial authority. To the British, at least.
It is this history that makes the word “tribe” a problematic one for many people. Ngugi has written compellingly about how the word – the whole concept – should be abandoned because of its role in colonisation. Nonetheless, it remains the word used by KNBS to ask the ethnicity question in the census, which is why I have used it in this piece. It is something to think about.
Ethnicity kept Africans divided from each other and in a supposedly inferior place on the hierarchy of civilisation that justified British colonial authority.
This history of ethnicity gives cause to ask some critical questions about what to do with ethnicity in any project aimed at decolonising identity. It is indisputable that ethnicity has been – at least partly – invented by colonialism. We must, therefore, be attentive to ways in which some of the projects of colonialism – divide and rule, hierarchies of civilisation, extraction – are perpetuated by ethnic identification today. But I think it would be a mistake to reduce ethnicity to this.
How postcolonial is ethnicity?
Ranger, and others after him, including myself, have also shown that Africans also participated – and continue to participate – in the construction of ethnic identities. And this is not necessarily a terrible thing.
During the colonial period, some ethnic groups had special favour with the British and so it suited them to identify ethnically. Intermediaries like African teachers, missionaries, soldiers and so on, benefitted from colonial patronage. If a man (never a woman, of course) could position himself as a leader of his tribe, he could gain from that. So, he needed the tribe to exist. On the concerning side of the ledger, this kind of patronage politics and the inter-ethnic competition it led to are not such great outcomes.
On the more positive side of the ledger, though, many Kenyans have also come to identify with their ethnic group in more positive ways, as many did before the arrival of the British as well. Most obviously, the cultural practices and community connections that make people feel safe, secure, valued and which give many people’s lives meaning and structure, are not bad.
Then there are dimensions of ethnic identity that are more ambiguous. Many, including Rasna Warah, believe – for better or worse – that to belong to Kenya, you have to belong to a Kenyan ethnic group. This is why the announcement that Asians are the 44th tribe was so significant, even though most people wouldn’t have used the word “tribe” to describe this community in the past. Warah laments, “What makes me uneasy about the designation of Kenyan Asians as one of Kenya’s 44 tribes is that it reinforces the idea that one must belong to a tribe to be recognised as a bona fide Kenyan citizen.”
Seeing Africans as being defined first and foremost by tribe allowed the British to divide and rule, and to imagine they were not just extracting and exploiting, but also civilising.
In my book on the marginalisation of Kenya’s Nubians, I made a similar argument – that ethnic identity, and specifically recognition as being an ethnic group of Kenya, was necessary to feel one belonged to the nation. I showed how it was a source of pride and security for Nubians to identify ethnically. It has been the only way they can imagine securing a place for themselves in Kenya. When the Nubians were recognised in the 2009 census, it felt really very good for them. It has for many different groups. That can’t be disregarded, even though it might be questioned.
The postcolonial history of ethnicity, therefore, raises some additional questions for those interested in decolonising identity, questions about whether or not there might be “good” aspects of ethnic identity that are worth retaining, even if they contain shadows of the colonial past. Perhaps it is transformation, rather than abandonment, that is needed in a decolonial project?
Decolonising identity in the census?
The census is a key tool in the maintenance of ethnic identities. Any discussion about what it might mean to decolonise identity really must think through the role of the census in sustaining ethnic codes first invented by the British, but also actively continued and transformed by the postcolonial government and its citizens. Indeed, bringing the abstract conversation about decolonising identity down to the level of this very concrete list is both a challenge and an opportunity to explore and test ideas and emotions related to ethnicity.
Asking whether or not the census should continue to count ethnic groups is one way into the difficult conversation about how to reckon with the legacies of colonial weaponisation of ethnicity, as well as what it means to people today. Such a conversation needs to consider the varied effects of counting, both good (recognition for minority groups) and bad (competition and posturing based on group size). I wonder if there is a way that ethnicity can be recognised without reproducing the negative effects that first arose under colonial authorities. It is a genuine question – I don’t know the answer. Any such system of recognition, though, would have to be carefully thought through with respect to who gets to determine which groups are recognised, through what processes, with what official outcomes, and with attention to how the inevitable changes in how people identify ethnically will be accommodated. Reflecting on how you, as the reader, feel about how your ethnic group has been counted, or not, in the census, can be a useful entry point to clarifying where you sit on this question of what it might mean to decolonise identity.
Editors note: This essay is based on the author’s article ‘Who are Kenya’s 42(+) tribes? The census and the political utility of magical uncertainty’ published in The Journal of Eastern African Studies. To see the full table of all codes, click on the link, then on ‘Supplemental’. The first 50 readers can access the full article for free here. If these are all used up, Africa-based readers can access the full article for free by signing up to the STAR program.
Congo-Brazzaville Strongman Buys Secret Weapons Haul from Azerbaijan
Congo-Brazzaville’s repressive government has quietly bought an arsenal from Azerbaijan. Opponents of President Denis Sassou-Nguesso say one recent cache is designed to tighten his grip on the nation.
In January 2020, at the Turkish port of Derince on the eastern shores of the Sea of Marmara, a huge cache of weapons was loaded onto the MV Storm. Registered in the tax haven of Vanuatu, the ship set sail with an arsenal of mortar shells, multiple launch rockets, and explosives, en route from Azerbaijan to the Republic of the Congo, better known as Congo-Brazzaville.
In total, more than 100 tons of weaponry wound its way to a building that appears to be the headquarters of Congo-Brazzaville’s elite Republican Guard, according to a confidential cargo manifest obtained by OCCRP. The cargo, estimated to be worth tens of millions of dollars, was just the latest in a series of at least 17 arms shipments sent by Azerbaijan’s Ministry of Defense to the regime of President Denis Sassou-Nguesso since 2015, according to flight plans, cargo manifests, and weapons inventories obtained by OCCRP.
Saudi Arabia was listed as the “sponsoring party” on several of the cargo manifests reviewed by reporters. It’s unclear what that sponsorship entailed, but it could mean that Riyadh paid for the weapons or the cargo deliveries.
There are no public records of Azerbaijan exporting these weapons, and no similar records of Congo-Brazzaville importing them. The latest transfer has sparked opposition concerns that Sassou-Nguesso is prepared to use force if necessary to maintain power as the country’s March 21 election nears.
His well-armed security services are a key reason he has ruled the Central African country for 36 years, split between two separate terms, making him one of the world’s longest-serving leaders. His party looms large over parliament, which recently changed the constitution to allow Sassou-Nguesso to run for office again, sparking local and international condemnation. The move means the 77-year-old could, in theory, run in every election for the rest of his life.
OCCRP has obtained confidential documents showing that in the eight months preceding the March 2016 election, and for over a year after it, Sassou-Nguesso’s security services bought more than 500 tons of arms from Azerbaijan in 16 separate shipments. Just weeks after the vote, the government began a brutal campaign against a militia from an opposition stronghold that lasted for more than a year.
Opposition leaders claim the Republican Guard used the Azerbaijani weapons in that post-election conflict, spurring a humanitarian emergency which the United Nations said affected around 140,000 people in the region of Pool, in the country’s south. Satellite imagery obtained by international media outlet The New Humanitarian appears to show widespread destruction caused by weapons like rocket launchers and explosives. (There is no way to be certain that these weapons were from Azerbaijan, since Congo-Brazzaville does not declare its arms imports.)
Since 2015, Congo-Brazzaville has bought a huge weapons stockpile from Azerbaijan, with over 500 tons of weapons delivered to the country in multiple shipments.
Sassou-Nguesso’s regime is facing one of Africa’s most severe debt crises, raising questions about how these arms shipments have been financed. Documents show that at least two consignments delivered between 2016 and 2017 were sponsored by Saudi Arabia, at a time when Riyadh was vetting Congo-Brazzaville’s application to join the Organization of the Petroleum Exporting Countries (OPEC). Given Congo-Brazzaville’s significant oil reserves, the kingdom had an incentive to have a compliant Sassou-Nguesso government in the Saudi-dominated club, according to leading arms expert Andrew Feinstein, author of The Shadow World: Inside the Global Arms Trade.
The world’s biggest arms importer, Saudi Arabia is also an unremorseful supplier of weapons to global conflict zones including Yemen, where it is fighting Iranian-backed Houthi rebels.
Flight manifests list Saudi Arabia as a “sponsoring party” on multiple arms shipments to Congo-Brazzaville, dispatched in 2016 and 2017, as Congo-Brazzaville was on the verge of OPEC membership.
Described by critics as an oil cartel whose members must be compliant with Saudi output demands, OPEC helps the kingdom dominate global oil supply. The effect this has on oil prices, in turn, can boost petroleum revenues in member states.
OPEC’s 13 members include Africa’s biggest producers, Nigeria, Angola, and Algeria. Congo-Brazzaville, which eventually joined OPEC in 2018, would have been seen as a coveted member because it is one of the continent’s top oil producers, which gives OPEC even more heft.
Azerbaijan is not a full OPEC member but it is a significant oil producer.
Feinstein added that the latest Azerbaijan shipment could have been intended to give Sassou-Nguesso the arms to enforce his political will.
“The timing of this shipment is extremely suspicious, given Sassou-Nguesso’s previous crackdowns around elections,” he said. “The government is likely preparing to quash any dissent around the polls.”
A spokesman for Congo-Brazzaville’s government did not respond to multiple requests for comment. Azerbaijan’s Ministry of Defence did not respond to a reporter’s email seeking comment, and neither did a ministry representative listed on multiple documents. Saudi Arabia’s Ministry of Defense did not respond to questions about the nature of their sponsorship of the arms deals.
Boulevard Denis Sassou-Nguesso
The most recent weapons load, addressed to the Republican Guard at 1 Boulevard Denis Sassou-Nguesso in Brazzaville in January 2020, included 775 mortar shells and over 400 cases of rockets designed to be launched out of Soviet-era trucks, the confidential cargo manifest shows. The consignment from Azerbaijan was loaded onto the MV Storm at Derince, about 1,000 kilometers southeast of Istanbul.
The exact price paid by the Congolese regime for the arms shipment could not be verified, although an expert who examined the cargo manifests said it would be worth tens of millions of dollars. A former senior diplomat with access to information about arms inventories, who asked to remain anonymous for fear of reprisal from authorities, confirmed the authenticity of the cargo manifest and other documents and noted the sale price for the arms was likely well below market value.
The documents included end-user certificates, which are issued by the country importing the arms to certify the recipient does not plan to sell them onward.
In January 2020, more than 100 tons of weaponry was sent from Azerbaijan to Congo-Brazzaville’s Republican Guard, including 775 mortar shells and over 400 cases of rockets designed to be launched out of trucks.
Pieter Wezeman, a senior researcher at the Stockholm International Peace Research Institute, said arms received at a discount are often either surplus weapons or those produced in Bulgaria or Serbia, which are both known for their cheap ordnance.
“It would be less likely that Congo-Brazzaville would be able to buy some of this equipment from … other European countries which have more restrictive arms export policies,” he said.
The Pool Offensive
The 100-ton shipment from Derince was significant, but separate documents reveal another arsenal sent from Azerbaijan between 2015 and 2017 that dwarfed it — and may have had terrifying consequences.
In total, over 500 tons of weapons, including hand grenades, mortar systems, and millions of bullets, were sent to Congo-Brazzaville in 16 shipments during those years, according to documents including inventories, end-user certificates, and cargo manifests obtained by reporters.
One end-user certificate shows five thousand grenades imported for the purposes of “training, anti-terrorism, security and stability operations.” It was signed by a special adviser to President Sassou-Nguesso on March 3, 2016, just days before the election.
After the vote, the opposition claimed the government had rigged the election in favor of Sassou-Nguesso, and unrest broke out in the capital, Brazzaville. The government blamed the unrest on a militia known as the Ninjas, made up of people mainly from the Lari ethnic group and based in the Pool region, which partially surrounds Brazzaville.
The weapons from Azerbaijan were then used, an opposition leader claims, to help fuel a prolonged armed conflict in Pool targeting the Ninjas. Amnesty International condemned the offensive as “an unlawful use of lethal force by the country’s security forces.” As the government pursued the Ninjas, witnesses to the carnage told Amnesty that dozens of bombs were dropped from helicopters, hitting a residential area and even a school.
“During the violence in Pool, the regime deployed a scorched earth strategy,” said Andréa Ngombet Malewa, leader of the Incarner l’Espoir political party. “The weapons that they bought from Azerbaijan went straight to that operation.”
The Baku-Brazzaville Connection
Azerbaijan has emerged as a key foreign ally of Congo-Brazzaville, providing its regime with discount arms and, perhaps more importantly, secrecy.
Buying from Ilham Aliyev, strongman of the notoriously opaque South Caucasus nation, Congo-Brazzaville could do so in the knowledge that the sales wouldn’t be reported.
Congo-Brazzaville has not reported any arms imports for more than three decades, and since there’s no arms embargo in place against the country, it isn’t required to do so. Nonetheless, a trail exists, with disclosures by other countries showing Sassou-Nguesso has been active in the arms market. In 2017, Serbia reported exporting 600 assault rifles to Congo-Brazzaville. Bulgaria sent 250 grenade launchers.
Opposition figures claim that previous shipments of weapons from Azerbaijan were used to fuel a brutal post-election offensive in 2016 that led to a humanitarian crisis.
But the Azeri weapons shipments have never been publicly reported, even though documentation seen by OCCRP shows Azerbaijan has been exporting lethal weapons to Sassou-Nguesso since at least as far back as September 2015. Some of the weapons were sourced from Transmobile, a Bulgarian company authorized to trade weapons for Azerbaijan, while others were bought from Yugoimport, a Serbian manufacturer. Neither company responded to requests for comment.
The first shipments of arms arrived in Brazzaville on Azerbaijani Air Force planes, but starting in 2017 a private carrier, Silk Way Airlines, began flying the weapons in instead. As a private carrier, Silk Way would have likely received less scrutiny than its military counterpart.
Silk Way is registered in the British Virgin Islands, a tax haven, and was previously linked to the Aliyev family. As well as previously winning lucrative contracts with the U.S. government to move ammunition and other non-lethal materials, Silk Way was found, in leaked correspondence reported by Bulgarian newspaper Trud, to have used flights with diplomatic clearance to secretly move hundreds of tons of weapons around the world, including to global conflict zones, between 2014 and 2017. The airline did not respond to a request for comment.
Braced for a Crackdown
As his regime heads to the polls on March 21, strongarm tactics mean Sassou-Nguesso is expected to win. He will reportedly face Mathias Dzon, his former finance minister from 1997 to 2002, and Guy-Brice Parfait Kolélas, who finished second in the 2016 presidential election, among others.
Saudi Arabia was listed as a “sponsoring party” in at least two arms consignments sent in 2016 and 2017, around the same time Congo-Brazzaville’s admittance to OPEC was being negotiated.
In 2016 he claimed 60 percent of the vote, with Kolélas securing just 15 percent. The U.S. slammed the government for “widespread irregularities and the arrests of opposition supporters.”
Experts don’t believe the opposition will fare any better this time around. Abdoulaye Diarra, a Central Africa Researcher for Amnesty International, said the government is carrying out a pre-election campaign of intimidation, harassment and arbitrary detention against its political opponents.
Fears that press freedom could be under threat ahead of the polls have risen after Raymond Malonga, a cartoonist known for satirical criticism of the authorities, was dragged from his hospital bed by plainclothes police at the beginning of February.
And now, the weapons haul from Azerbaijan has the opposition concerned about the prospect of violence around the polls.
“We are worried that the weapons that Sassou-Nguesso’s regime bought from Azerbaijan could be used to crack down on the opposition during the upcoming election,” said opposition leader Ngombet.
“They don’t want the world to see how much the Congolese people are eager for political change.”
Simon Allison, Sasha Wales-Smith, and Juliet Atellah contributed reporting.
Long Reads1 week ago
The Anatomy of Kenya Inc: How the Colonial State Sustains and Re-Creates Itself
Politics1 week ago
A Class That Dare Not Speak Its Name: BBI and the Tyranny of the New Kenyan Middle Class
Op-Eds1 week ago
Is Balkanisation the Solution to Somalia’s Governance Woes?
Satire1 week ago
UK Corruption Cover-Up Exposed As US Marks Grim Milestone
Politics1 week ago
Congo-Brazzaville Strongman Buys Secret Weapons Haul from Azerbaijan
Op-Eds1 week ago
The Danger of the Single Story and Africa’s Refugee Equilibrium
Op-Eds1 week ago
Lava Jato: The CIA’s Poisoned Gift to Brazil
Op-Eds2 weeks ago
The G20 Is Gathering. Debt Justice Is Our Demand