Methodology and ethics

The ethnographic data that informs this article dates from my doctoral fieldwork, conducted between 2014 and 2016, on the role of religion in local associational life. This fieldwork involved long-term participant observation in churches, families, groups of financial mutuals, community fundraising ceremonies and fundraiser organizing committees. A habitus of patronage-based electoral mobilization featured across such contexts, particularly as the 2017 elections drew nearer. I observed and eventually interviewed incumbents and aspiring candidates along with their local intermediaries, all the while comparing my findings with diverse interlocutors’ views on local politics. Between July and September 2023, I conducted follow-up interviews with local intermediaries as well as with their neighbours and family members to gain diachronic insight into agent–candidate alliances as they played out in the 2017 and 2022 elections.

My long-term embeddedness within multiple households, churches and groups of financial mutuals meant that my network of interlocutors was broad and dense enough to obviate the risk of responder bias. As a white researcher associated with a British university, assumptions about my own views on the ethics of electoral patronage could encourage interlocutors to censor their engagement in electoral politics. Incidentally witnessing electoral patronage being given and requested at a variety of public gatherings allowed me to first identify alliances between candidates and local intermediaries, and subsequently approach them to discuss their alliances with open reference to the brokerage of patronage. In addition, I triangulated individual intermediaries’ accounts of their engagement in local electoral mobilization with those of their spouses and children, neighbours and fellow church members, as well as with voters identifying with different families, houses and sub-clans.

There are ethical limits to participant observation, especially in studies of electoral patronage. Accordingly, I declined candidates’ invitations to campaign for them or ‘escort’ their contributions to public fundraisers with contributions of my own. I joined three aspiring candidates – two of whom were direct competitors – on the campaign trail for no more than a day per candidate. I did so after several months of Egesa – a Gusii radio station – replaying an interview with me in which I described my research. As in other studies of illegal behaviour (Bourgois Reference Bourgois2003; Sanchez Reference Sanchez2010), which stress both the cultural and the material circumstances that motivate such behaviour, my interlocutors did not see electoral patronage as inherently unethical but rather as a rational and culturally acceptable mode of engaging in and benefiting from electoral competitions. However, since electoral patronage constitutes an offence under Kenyan law, I have been deliberately vague about where in Gusiiland I conducted my research to protect the anonymity of interlocutors who shared sensitive materials that – if traced back to them – could have material consequences for their future relationships with patrons and electoral candidates.

Political competition in Gusii history

Gusii history defies simplistic understandings of African and Kenyan politics. Here, sub-ethnic social identities remain key structuring factors in electoral competitions. In most Kenyan settings the salience of such local-level distinctions and identities waned with the reintroduction of multiparty electoral competitions in 1992, when national-level inter-ethnic cleavages and alliances appeared to all but determine voting patterns (Posner Reference Posner2005: 264–9). Not so in Kisii and Nyamira counties, which make up Gusiiland; and especially not when compared with other settings in post-devolution Kenya. Voter coordination based on shared ethnic identity in presidential contests is rare. Only once did the Gusii people vote as an ethnic bloc in a presidential contest (Ocharo Reference Ocharo, Maupeu, Katumanga and Mitullah2005). Bloc voting along clan and sub-clan lines is the norm, elections are highly competitive, and the outcomes typically bespeak political fragmentation. Few incumbents make it to a second term. In the 2007 parliamentary elections, six out of ten Kisii constituencies were among the 12 per cent of Kenyan constituencies where seats were won with less than 29 per cent of the vote (Wolf Reference Wolf2007, cited in Boone Reference Boone and Boone2013). Since devolution in 2013, as Edmond Were notes,Footnote ¹ few other counties in Kenya have witnessed their top-level positions politicized quite as sharply in terms of longstanding inter-clan and sub-clan rivalries. A resurgently clannist form of ‘negotiated democracy’ has emerged in recent years: engagement in electoral competition means negotiating and building alliances between hierarchically positioned clan and sub-clan kinship groupings, all the while accounting for, rectifying or maintaining historical grievances and privileges (Masese Reference Masese, Thibon, Fouéré, Ndeda and Mwangi2015).

The social organization of electoral competitions around inter-clan rivalries reflects popular ideas about shared kinship as a preferred basis for political trust and cooperation. Voters favour ‘our own person’ (oy’ominto) – that is, someone who identifies with and was born into the same ‘family’, i.e. the same ‘clan’ (eamate) and corporate ‘house’ or sub-clan (enyomba). Although voters and candidates alike describe this tendency as ‘natural’, the ethnographic and historical record suggests that this tendency is anything but. Rather, kinship-based trust has been ‘naturalized’ as clan and sub-clan kinship identities were constructed over the past two centuries.

To an extent, the kinship identities that now structure Gusii politics were reified and, in that sense, ‘invented’ with the imposition of British indirect rule. Colonial administrative units were drawn up to correspond to the distinct territories associated with putatively discrete descent groups, which were governed with impunity and favouritism by local ‘chiefs’ (abagambi). This institutionalized hierarchies and conceptions of political constituency as proxies for specific descent groups and kinship networks in a patrilineal and patrilocal society that historically did not recognize absolute distinctions of rank and status. Social identities that had hitherto been circumstantial, fluid, open to revision and enactment through subjunctive fiat rather than physiological descent per se were thus redefined as indexing essential units of political competition and cooperation (cf. Mayer Reference Mayer1965; Lonsdale Reference Lonsdale1977).

However, the Gusii people did not simply take on clan and sub-clan identities as new instruments of life in a colonial social order. Like other African identities, they did not come out of nowhere; they have a history that extends beyond twentieth-century colonialism (cf. Nugent Reference Nugent2008). In Gusiiland, known precolonial precedents extend to the eighteenth century, when exogamous clans competed to the point of ‘bitter conflict’ (Ochieng’ Reference Ochieng’ and Ogot1978: 103) for the flattest ridges and the most auspicious highland areas after being displaced from the Kano plains by Luo, Maasai and Kipsigis populations through cattle raiding, military attacks and assimilationist migration (see also Akama Reference Akama2017). In the latter half of the nineteenth century, following a succession dispute over Sweta leadership, individuals and networks identifying as Nyaribari – a then subordinate collective ‘house’ (enyomba) of ‘real’ (bori) and classificatory kin – separated from their encompassing lineages, claimed and cleared land elsewhere, and came to constitute a clan of their own, on a par with both their parent lineage Getutu as well as Machoge, the other large lineage constituent of the yet more encompassing former Sweta identity (Ochieng’ Reference Ochieng’1974: 112–33, 152–71). While most clanspersons (abaamate) today understand themselves as interrelated and descended from a common male ancestor, the history of clan identities is therefore not one of straightforward unilineal descent, but rather one of relational fictions and frictions precipitated by inter-clan and sub-clan rivalries.

The hierarchies and inequalities that such rivalries generated gained a more permanent character as new elites and authority figures emerged under colonial rule. A pivotal figure was Musa Nyandusi. As a chief in the colonial administration, he concentrated enough power and wealth not just to mobilize Nyaribari solidarities in opposition to other clans (Maxon Reference Maxon1981: 118), but also to redefine who identified as ‘Nyaribari’. Nyandusi ‘bought’ people from other clans (Getutu) and ethnic groups (Kipsigis), giving them land strategically placed at the frontiers of Nyaribari territories, at the front line of conflicts and cattle raiding with Maasai and Kipsigis populations as well as Getutu and Bassi clans. These ‘bought’ and partially assimilated groups of ‘dwellers’ (abamenyi) had a distinctive experience of colonial and postcolonial governance and administration, rooted in a rivalry over state resources with their adoptive lineages. A new identity – Kambanane (lit. ‘eight ropes’, a designation interlocutors associated with the eight-stringed Gusii harp or obokano) – emerged to distinguish the eight adoptive lineages as interrelated and as unrelated to Abatondo (lit. ‘people who have been bought’) and Abaguche. The history of Nyaribari sub-clan identities continues to reverberate in contemporary politics.

Elderly and middle-aged male Tondo and Guche interlocutors recalled at length, often with a measure of sarcastic glee, that Nyandusi – in a bid to protect Kambanane land for agriculture – inadvertently favoured Tondo and Guche interests by concentrating land-intensive government school construction in Tondo and Guche territories. Better and more accessible schooling gave Tondo and Guche houses an advantage in electoral competitions, formal employment and government tenders in the late colonial and postcolonial eras. Hence the subsequent and current preference among Kambanane voters for candidates who are ‘really’ Nyaribari rather than classificatory clanspersons. In short, then, fellow kin and clanspersons are not ‘naturally’ more reliable in state power. Rather, a history of uneven development and identitarian competition has recreated the conditions of possibility for shared kin to be reimagined and recognized as the more trustworthy patrons and allies. This dynamic endures even as candidates seek to broker alliances that mobilize votes across clans, families and social classes.

Buying votes and building trust

In scholarship on the role of money in politics and electoral competitions, the distribution of money and material resources to clients is understood first and foremost as a transaction of money for votes that is inherently inimical to democratic ideals. According to this understanding, gift-giving norms on the campaign trail reflect public mistrust in state institutions and their representatives; promote exclusionary forms of cooperation among elites; and reproduce – in Kenya’s case – inter-ethnic structures of agonistic, amoral and zero-sum political competition (Ng’ethe Reference Ng’ethe2010; Bratton Reference Bratton2008; Reference Bratton2013; Vicente and Wantchekon Reference Vicente and Wantchekon2009). Other analyses point out that electoral patronage enables pro-social acts of trust in ways that cannot be easily divorced from notions of moral virtue, long-term reciprocity and democracy (Björkman Reference Björkman2014; Piliavsky Reference Piliavsky2014). Indeed, recent ethnographies of political mobilization in Kenya suggest that the politics and solidarities which both candidates and voters pursue and act on when giving and receiving electoral handouts cannot be assumed in advance (Rasmussen and Van Stapele Reference Rasmussen and Van Stapele2020; Kilimo Reference Kilimo2022). Patronage and sheer self-interest have both ethnographic and analytic limits, which can be overcome by attending to voters’ moral evaluations of candidates (Lockwood Reference Lockwood2019). Insofar as electoral handouts do ‘buy’ votes, it is less because they are enforceable and time-bounded transactions, and more because they enable voters to evaluate competing aspirants by conveying information about candidates’ moral values and personal character, their viability as contenders, and the credibility of their promises to act as conduits of welfare and development (Kramon Reference Kramon2018). Material inducements might play a decisive role in political competition, but whether or not they buy votes is a moot point. Rather, the question in patronage-based political competition is how material resources are converted into moral legitimacy and symbolic capital (Nugent Reference Nugent, Basedau, Erdmann and Mehler2007).

In Kisii’s fragmented constituencies, local intermediaries play a pivotal role in facilitating this exchange. As their candidate’s associates and ‘agents’, they distribute electoral patronage and ‘create a name’ (okorosia erieta) for their patron in their networks, often through public contributions on behalf of their candidate at community fundraisers for church construction, university fees or funeral costs. Although this makes up the bulk of their work, it is by no means their only work on the campaign trail. In the run-up to election days, agents may buy votes in what is widely acknowledged as ‘direct’ buying: travelling to their candidate’s rival home base and temporarily confiscating ID cards in exchange for cash. On election day, some agents supervise election proceedings within polling stations on their candidate’s behalf; agents who did so reported opportunities for enforcing at least some exchanges of cash in return for votes (for example, by pretending to help elderly illiterate voters read the ballots and place their vote). One veteran agent developed strategies for reading body language through ballot box curtains, thus producing educated guesses of who voted for whom. Such knowledge is critically insightful for candidates, as it provides a benchmark in relation to which suspicions of other kinds of electoral fraud can be verified.

By and large, however, election-time vote buying is difficult to enforce, and relatively strong ballot secrecy laws make electoral handouts an unwieldy – albeit no less necessary and widely expected – campaign tool (Kramon Reference Kramon2018). Money can be and certainly is exchanged for votes, but this exchange is not straightforwardly effective. This is partly because intermediating agents’ own actions and reputations influence how electoral patronage is received, and partly because – as one former agent observed – ‘today’s poverty’ compels voters to ask for money from wherever they can, which makes the exchange even more uncertain. Some voters, especially those in positions of class privilege, spoke of themselves as unpurchasable while pointing to low-income individuals and families, the youth and the destitute as more likely to be impressed, indebted, and ‘bought’ by whatever money they receive. Conversely, young and poorer voters called out such assessments as hypocritical, pointing out that ‘middle-class people’ also accept and request patronage. Indeed, most voters I knew – regardless of socio-economic background – received money and, in their own words, ‘ate from’ multiple candidates. Doing so, voters explained, enabled them to evaluate different options as one might in a marketplace, and thus participate in an exchange that is at once transactional and moral: ‘Food is for eating; on voting day it is I who will decide.’ Importantly, voters and agents described votes – their own and those of their families, neighbours, colleagues and associates – as earned rather than simply bought, and the money they received as rightfully theirs. In part, these claims helped rationalize and justify exchanging votes for money, as well as actively comparing candidates in terms of how much money they handed out. However, such claims also speak to the kinds of cooperation and mis/trust that typically arise among candidates and their agents.

Unlike ordinary voters, agents enter alliances with candidates. The terms of cooperation between candidates and agents are highly unequal. Mutual trust is tentative at best, and cooperation mistrustful. Each candidate entrusts elements of their campaign (money, intelligence) to multiple agents, who may or may not act in reliable ways or be able to influence other voters, usually in return for some (often very limited) immediate access to material benefits along with promises of future meaningful help such as jobs or procurement tenders. Agents do not entertain these promises naively and may switch sides either following or during electoral periods, according to their own shifting assessments of the likelihood that such promises may materialize. Experience has taught them that ‘politics is not an industry’; ultimately, elections secure benefits for candidates alone (who may leverage a good run in their favour even if they do not win). By contrast, for agents, agreeing to publicly campaign for a candidate means entering an exclusive form of cooperation with strictly one candidate at a time. This limits agents’ capacity to coax patronage out of multiple candidates, as they become known as a particular candidate’s person. In effect, agents become doubly vulnerable: on one level vis-à-vis the candidate they are allied with; and on another level vis-à-vis other local voters who may come to suspect and contest the agent’s supposedly privileged access to a candidate’s campaign finances.

Rivalries between voters for agent positions, along with competition between candidates for a select few agents, consolidate material inequalities between agents and further complicate agent–candidate cooperation. Agents hail from a variety of ‘levels’ (ebiwango) or socio-economic classes, but leadership positions within campaign teams reflect class privilege. Candidates entrust chairperson, treasurer or secretary positions on campaign committees to individuals with substantial salaries, business revenues, wealth they might contribute to a candidate’s campaign finances, or assets they can sell if they misappropriate campaign money. Since wealth and privilege can translate into both fame and popularity, agents from well-to-do backgrounds are most likely to find favour with candidates. Skilled orators and locally respected farmers or ‘hustlers’ in the informal economy can also become agents. However, they face intense pressures from rival and wealthier voters competing for agent positions on the wealthiest candidates’ campaign teams. Gossip and ‘words’ are ‘thrown’ around (okoruta amang’ana), as agents and prospective agents discredit and ‘cut’ each other’s ‘legs’ (okonacha amagoro), most often by drawing an equivalence between economic status and social influence or moral authority. Moreover, agents from low-income backgrounds typically receive less patronage in return for their campaigning work than wealthier agents, some of whom further benefit from inter-candidate competition for their loyalty and allegiance. Thus, agent–candidate cooperation, just like the patronage-based competition between candidates that such cooperation makes possible, generates classist forms of trust. Relative wealth stands reaffirmed as a measure of trustworthiness, such that – in one agent’s words – ‘those who have money are the ones who are added more money’.

Such classist forms of trust and cooperation in turn reproduce the conditions of possibility for enacting clannist ideologies of trustworthiness. Consider Jackson’s account of his involvement in the 2022 elections as an agent for two direct competitors vying for the local MP seat. He had initially supported candidate X, whose patronage on the campaign trail evidenced not only great wealth but also, in contradistinction to his competitors, the absence of ‘hatred’ and jealous greed (uchoyo). The door to his home was always open and food always available; everyone walked away both fed and with 1,000 Kenyan shillings (KES) in their pocket. This candidate promised Jackson’s eldest daughter a job as a ‘soldier’ and security guard, a promise he had made to other agents, but which nevertheless sounded credible enough given the many petrol stations he owned. Yet cooperation faltered; mutual mistrust arose. At one point, when Jackson relayed information about local fundraisers where they could campaign, candidate X would neither show up nor send money.

Jackson saw the candidate’s apparent mistrust in him as the consequence of envious classist gossip. He referred to ‘people around here’, his neighbours and relatives, and especially those in hierarchically super-ordinate positions, of envying down. ‘They saw us with candidate X; maybe they thought we will also develop,’ Jackson remarked. The rumour that he was visited by multiple candidates, or simply being noted as ‘small’ and ‘lower class’ and therefore uninfluential, could have been all it took to undermine Jackson’s credibility. So intense did the discursive pressure created around him become that he left candidate X’s team, even though its prospects were promising. His alliance with candidate X had become an undesirable and immoral exchange of money for votes, one that could not maintain hopes for meaningful future help for him and his family. He then joined the campaign of candidate R, a relative and fellow house member.

Remarkably, Jackson’s decision to ally with a fellow clansperson marks a departure from a type of alliance that cuts across kinship identities. Candidate X identified with a minority house that – within its encompassing sub-clan – ‘had never ruled’ (njana kogamba) or had one of their own in power. So, campaigning for X meant negotiating a cross-clan buy-in to affirmative action. Accordingly, Jackson reasoned as follows with his neighbours and fellow house members in a way that whipped up local anti-elite sentiments: given that their own past leaders had proven so wanting, and since the house that candidate X represented had never ruled, why shouldn’t the MP seat go to that house too, and to a truly capable, generous and caring leader no less? This reasoning proved popular among Jackson’s neighbours and relatives, including those who contested his local area agent position, and whose classist politics of envious discredit had undermined Jackson’s personal alliance with X. For Jackson, though, trusting one’s closest kin became the only sensible, morally worthwhile, perhaps – as per candidate R’s suggestion – even the ‘natural’ alliance to enter. This suggests that even when mobilizing a variety of intersecting class- and kinship-based solidarities, agent–candidate cooperation and the rivalries that shape such alliances contribute to the naturalization of clannist forms of trust.

Family politics

When Jomo Kenyatta, Kenya’s first president, reimagined Kenyan society as more than just the sum of its ethnic groups, he posited ‘the family’ as the cradle of moral virtues and the source of expansive mutual trust in a multi-ethnic Kenya. In part, this marked a projection of mid-twentieth-century anthropological theories of kinship as reducible to blood relations onto precolonial Gikuyu thought. But calling for trust while identifying ‘the family’ as its self-evident source was also consistent with patrilineal and specifically patriarchal ideas about social organization, common in Gikuyu postcolonial history. Inclusion in the ‘trust’- and ‘family’-based polity that Kenyatta called for was not just tethered to personal reputations earned through family relations and the fulfilment of normative kinship ideals; it was also assumed to be a male prerogative: ‘The male individual only could be both a family man and a man of power’ (Angelo Reference Angelo2020: 57). Similar preoccupations with personal reputation and moral authority as mediated by gendered family relations and household reproduction consistently feature in the history of political movements in central Kenya over the past century (Lonsdale Reference Lonsdale, Atieno Odhiambo and Lonsdale2003).

Anti-colonial movements and other large-scale political associations did not emerge in colonial Gusiiland (Maxon Reference Maxon1981). However, here, just as in central Kenya, moral and political authority ‘begins at the household gate’ (Lockwood Reference Lockwood2019: 366). Whether or not a candidate has a wife and children; the way he relates to agnates and affines from different social classes; where his ancestral home (sobo) is; whether he has built a house there and what kind of house; whether he built one for his parents – all are key considerations when voters evaluate candidates’ moral character. The credibility of agents as reliable conduits of a candidate’s patronage is similarly evaluated. Agents who are locally known as poor managers of their own households, or as cruel, arrogant, greedy or unkind towards neighbours and relatives, are unlikely to earn their family’s votes for the candidate they represent. Beatrice, Jackson’s wife, explained this very point to a candidate at a meeting at his house during the 2022 electoral season, warning him that the agent he had partnered with from the local area was widely discredited:

The moral reasoning in Beatrice’s account evidences a well-documented Kenyan, if not African, capacity of finding political virtue in the distribution of patronage through hierarchical networks of clients and dependants (cf. Scherz Reference Scherz2014; Ferguson Reference Ferguson2015; Haynes Reference Haynes2017). Yet the emphasis on family conflicts and disputes as consequential for the mediation of patronage also points to the special significance of family relations to political mobilization in Gusiiland. Both of its constituent counties – Kisii and Nyamira – are impoverished, densely populated and socio-economically stratified. Here, the lived experience of scarcity, precarity and inequality hides their historical constitution under colonial and postcolonial rule. Discourses on land scarcity and associated conflicts are ‘framed in terms of the interpersonal or family crimes of dishonesty and betrayal, rather than in terms of collective grievances of wider social and political scope’ (Boone Reference Boone and Boone2013: 189). A micro-politics of exclusive mutual aid among the wealthy reproduces the material conditions of possibility for accusations of witchcraft, rumours of malign intent, as well as ascriptions of greed, jealousy and envy (Zidaru Reference Zidaru2019). These developments resonate with other ethnographies that evidence kinship as defined by uncertainty and mistrust born out of rivalries and competing interests (Geschiere Reference Geschiere2013; Lockwood Reference Lockwood2020).

Although long-running, the significance of family relations and associated politics of envious mistrust in electoral mobilization has shifted in post-devolution Kisii. This is partly because devolution created new scales of political competition, which interlinked with older ones. Hyper-localized alliances vying for Member of County Assembly (MCA) seats typically object to clannist discrimination in patronage networks at broader county- and national-level government bureaucracies while rallying votes for their ‘own people’. This cross-scalar clannist calculus means that rivalries at lower and lower levels of kinship relations are now more readily made visible at broader scales, through inter-agent competition, often in ways that undermine agents’ credibility in agent–candidate alliances. Nyagarama, a long-time agent for Livingstone – the Guche politician introduced at the beginning of the article – noted that ‘jealousies [chindamwamu] have always been there, but devolution has exposed them’. Nyagarama spoke from personal experience.

A well-educated, Nairobi-born and formerly employed farmer hailing from the Mwaboto sub-clan, Nyagarama had campaigned for Livingstone at almost every election since 2007. By 2022, their alliance was dead in the water. According to Nyagarama, their alliance did not end because Livingstone exploited him. Albeit rare, he did receive meaningful help from Livingstone. He recalled how, when he confronted Livingstone over his meagre if not non-existent ‘salary’ as his campaigner, Livingstone paid him KES 30,000 for his son’s university fees. Instead, Nyagarama was displaced as Livingstone’s agent by another faction within the Mwaboto sub-clan, who ‘wished to build their own leader’. This turn of events reflected the way in which the rivalry between Mwaboto houses for the local MCA seat in the 2017 elections played out, and Nyagarama’s shifting position within this rivalry.

Two candidates who shared a great-grandfather but had different grandfathers were the favourites in that race. Although he traces his descent to candidate A’s grandfather, Nyagarama supported candidate B, out of a stated desire to punish the former for not helping him with his son’s university fees, even though years ago Nyagarama had helped A migrate to the USA. Tellingly, Nyagarama had also contemplated running for the MCA seat himself after the 2013 elections, and arguably saw candidate A as an unwanted rival challenging his limited class privilege. Nyagarama’s immediate neighbours, with whom he shared a closer kinship identity, resented Nyagarama’s alliance and questioned why he didn’t support ‘one of their own’, as if that were the appropriate and obvious course of action. Nyagarama appeared to observe this normative prescription when, in the run-up to election day, he switched sides and publicly supported candidate A. Nevertheless, Nyagarama’s neighbours and relatives continued to criticize him. They not only ridiculed him for supporting an ‘outsider’, but – like voters from their rival house – also for opportunistically coaxing money from both candidates. They thus emasculated him as a man who lacks a ‘stand’ (msimamo), who is fickle and unreliable, and therefore an untrustworthy broker of his patriline’s interests. Such talk was a far cry from the way in which people had previously celebrated Nyagarama for calling on patronage from multiple politicians to help finance a relative’s funeral during a drought season, when money is hard to come by. Agents switching or playing multiple sides was thus hardly new, and not inherently immoral. After all, as one interlocutor intimated, ‘[Politicians] use us; why shouldn’t we use them?’

What was new was competition over the local MCA seat, along with the discourse that this competition incited among and beyond Mwaboto families. Nyagarama’s alliance with Livingstone, which had endured despite Livingstone’s suspicions and Nyagarama’s track record of conceited opportunism, became untenable. According to Nyagarama, this was purely ‘because of numbers’: other Mwaboto houses have more people, so it made sense for Livingstone to partner with them, considering that – as a Guche person – he needed to secure as many votes as possible in Mwaboto and other Kambanane territories if he was to win the Nyaribari vote. Nyagarama’s son, however, pointed out that ‘word went round that [Nyagarama] is untrustworthy’ after the 2017 local MCA election, and that this rumour best explained not just why Livingstone terminated his alliance with Nyagarama, but also why Nyagarama ‘struggled finding anyone to support in 2022’. These rumours reflected badly on Nyagarama’s wife and adult children as well, souring relations between them and other relatives – so much so that even they were forced to recognize, in Nyagarama’s son’s words, ‘etamaa nero ekomochanda’: ‘greed is what challenges Nyagarama’. Thus, local post-devolution politics have made it more likely for hitherto private suspicions and ascriptions of ‘ugly emotions’ (Hughes et al. Reference Hughes, Mehtta, Bresciani and Strange2019) to feed into broader dynamics of moral policing.

The gender of political mis/trust

Clearly, the deepening entanglements between electoral mobilization and family rivalries has further undermined agents’ viability as agents. However, these entanglements have animated evaluations of trustworthiness that extend beyond singular individuals’ emotions, values and character. Increasingly, it is the credibility of male agents as men that stands challenged. Beatrice, for example, warns candidates not just about individual agents they have partnered with – such as the teacher and union leader who hadn’t supported his late brother’s children – but also about partnering with men in general. In electoral meetings, she cites a now commonplace discourse that casts men as unreliable, disloyal, incapable of ‘having a stand’. Like other women, she accuses men – young and old – of withholding electoral handouts from their mothers and wives, whose votes they are supposed to broker.

Beatrice herself said so at a meeting with candidate X towards the very beginning of his campaign, when his campaigners were mostly young men. She went on to note: ‘[Male] youth are after money; you give him 500, another gives him 1,000, he will compare and vote for who gave the most.’ Instead, Beatrice advised X, he should trust and ally with women, because ‘when a mother says yes she means yes’. Subsequently, X called all mothers forth when he ‘milled’ (ogosera) or handed out money, reportedly telling them: ‘My sweethearts! You are my mothers. Will you give me your votes?’ He was careful to enact gendered norms of intergenerational respect and obligation, projecting an image of the good son who lives up to an ideal of the mother–son relationship as defined by giving back to one’s mother. Moreover, he included Beatrice and other women as campaigners and mobilizers on his team.

Beatrice, a middle-aged mother of five, had campaigned for multiple politicians over the years, alongside her husband Jackson. In the alliances they entered, Beatrice and Jackson sought money to feed themselves, educate their children, and ideally find a job for themselves or their eldest daughter who had finished school. While they did find money to eat, and occasionally even to pay school fees, jobs were never forthcoming. Since Beatrice lost her job as a school librarian over a decade ago, they depended solely on brickmaking and ‘hustling’ their limited crops on market days. Their children were regularly sent home from school on account of their fee balance. Accordingly, they actively strive to extract as much money as possible from candidates during the campaigns, knowing that candidates – regardless of whether they win or not – are unlikely to help them personally. This, for example, is how Beatrice responds when candidates have other people call her to ‘test’ her support after publicly taking sides:

The way in which Beatrice negotiates more substantial patronage, with a view to contribute to the continuity of the family beyond immediate consumption, bespeaks the increasingly matrifocal character of social relations in postcolonial Kisii. Here, the fickle fortunes of rural market capitalism and wage-labour migration have transformed gender relations, generating ‘competition for control over scarce resources and … new roles and obligations’ (Silberschmidt Reference Silberschmidt1999: 101). With men often away in large urban centres, women took on new roles as providers for their households. As in other matrifocal contexts, bonds between mothers and their children became central not just within households, but also in the relations between households and families (cf. Lockwood Reference Lockwood2019; Neumark Reference Neumark2017). Women’s widespread participation in rotating contribution clubs and other ‘groups’ (ebiombe) of financial mutuals is a case in point. Membership in such groups, and especially in groups large enough to leverage their strength in numbers, provides women with a means to participate in electoral politics and gives them access to flows of patronage from which they are otherwise excluded (such as the patronage proffered at church fundraisers or funeral ceremonies).

Partnering with women as agents and vote mobilizers has become a critical strategy on the campaign trail. Clannist rivalries and political fragmentation came to a head in the 2017 elections, which people remembered as ‘tough’; the margins were small, the competitors many, allegations of fraud common, and ‘nobody would accept [defeat]’. In the 2022 elections, it became even harder to play the shared kinship card. Those who favoured their ‘own people’ and managed to get them in power in the 2013 and 2017 elections accused their leaders of ‘house-ism’ (obonyomba), of privileging strictly ‘their own families’ (chifamilia chiabo), ‘in their own absolute internal house’ (enyomba yabo ime kabisa). Such interlocutors voted for more distant clanspersons and even non-clanspersons in the 2022 elections – not so much because they saw such candidates as any more reliable or desirable, but rather to punish a ruling kinsman and their immediate family for their greed and broken promises. In the context of this anti-elite backlash at the grass roots, the candidates and agents who brokered the broadest coalitions of voters were those who mobilized votes based on gendered identities just as much as through more old-fashioned appeals to shared kinship and age-based or cross-class solidarities. One agent referred to several races in which, all things being equal, ‘the reason why [the candidate] won is the mothers’ – i.e. the women who mobilized votes as well as female voters writ large. Another agent recalled how Simba Arati, now governor of Kisii County, milled money during the 2022 elections:

Arati’s campaign had other strengths than its carefully gendered and age-based style and structure. He had been nominated by Azimio la Umoja, the party led by Raila Odinga, who was at the time widely popular in Kisii, especially among Gusii youth. It also bears noting that Arati’s main rival, Ezekiel Machogu, was burdened both with financing current president William Ruto’s campaign and with a public display of family and intra-clan rivalries. Machogu shared the Nyaribari clan identity with several other gubernatorial contestants, including a member of his own inner house and family. This not only fragmented the Nyaribari vote but also seemed to confirm a widespread stereotype of Nyaribari competitors as mutually envious ‘vote spoilers’ (abasarerie chikura). Machogu’s problematic entanglement in clannist politics was further complicated by his church-based campaign strategy. He was popular among Catholics and Seventh Day Adventists, deftly deferring to – and securing subtle expressions of support from – predominantly wealthy and male church elders. However, since the membership of individual churches shares kinship identities, voters interpreted his campaigning through the idiom of clannist just as much as faith-based allyship. Machogu’s patriarchal kinship- and faith-based patronage networks made him the darling of the elderly and the economically established. In contrast, Arati’s campaign spoke directly not just to the youth but also specifically to women, young and old, making special provision for the brokerage of female-gendered patronage. So, while Machogu partnered mostly with men who promised votes from within their families and congregations, the prominent presence of women in Arati’s campaign directly undermined the alliances Machogu forged. In effect, Arati negotiated alliances that not only crossed his own Bassi clan boundaries but also cultivated gendered forms of kinship-based trust in a matrifocal idiom that recognizes women’s importance in family as well as electoral politics.

Kinship sentiments and identities

Over a decade has passed since Kenyans first voted in county-level elections. The expectation that a devolved institutional structure of political competition would assuage local grievances and promote impersonal forms of political trust based on bureaucratic procedures and technocratic rationality has proven sorely off the mark (Chome Reference Chome2015). Post-devolution elections were widely welcomed as ‘everyone’s turn to eat’; older forms of personalistic, identitarian, patronage-based cooperation and trust financed by rent seeking thrived (D’Arcy and Cornell Reference D’Arcy and Cornell2016). This politics of personalistic trust based on shared kinship or ethnicity plays out in familiar ways, not least through cross-class alliances that lend moral justification for what – viewed from global, national and structural perspectives – constitutes ‘corruption’ (Berman Reference Berman, Berman, Eyoh and Kymlicka2004). For students of Kenyan as well as African politics more broadly, the resilience of clannist forms of political trust in Gusiiland may be familiar enough to come across as a straightforward continuation of the structural legacies of authoritarian governance under colonial rule. Such governance, having been based on arbitrary decisions and favouritism regardless of competence and due process, gave way to a ‘widespread lack of trust in the play of rules’ and a tendency to rely on personal ties among kin and co-ethnics to mediate desirable outcomes (Táíwò Reference Táíwò2011: 161, 156–85 passim).

An ethnographic approach that documents electoral patronage from within the interactions between candidates, intermediaries and voters in Gusiiland suggests that the persistence of personalistic and identitarian trust is itself a dynamic phenomenon, rather than a static holdover. Agent–candidate alliances over the past decade have not so much leveraged pre-existing shared kinship ties as remade them anew. Previous gendered hierarchies of credibility stand contested. Ethical judgements and ascriptions of negative emotions such as greed, jealousy and envy – typical of everyday life in socially stratified families – influence the terms and outcomes of agent–candidate collaboration and mis/trust. In the process, new scope has opened up for older normative ideals of kinship-based trust to shape how the competitive exchange of money for votes and rivalries for brokerage rights play out, and the forms of political trust or mistrust that emerge as a result. Trusting one’s fellow kin may be undemocratic, but it endures precisely because of how it plays out and emerges as preferrable at the small scale of social interaction among neighbours, families and agent–candidate alliances – where political mobilization is hardly distinguishable from the struggle over the material conditions of life itself. Such normative trust is realized in shifting ways that can at least appear to meaningfully respond to longstanding tensions, conflicts and experiences of exclusion. A case in point are the novel opportunities that devolution has created for women to participate in electoral politics – not just as politicians and ‘matrons’ (Kilimo Reference Kilimo2022), but also as mothers and influential voters who can broker allegiances in their families and networks. In effect, trusting one’s fellow, wealthier kin to make available misappropriated rents remains and feels unproblematic and even somehow fair and democratic.

These findings highlight the significance of documenting alliances between candidates and local intermediaries in analyses of patronage-based political competition in Africa and elsewhere. This methodological strategy offers three analytical opportunities. First, it picks up on the mobilization of personal ties that cross established kinship or ethnic identity categories, thus keeping assumptions about the prevalence of identitarian politics in Africa at bay (see also Koter Reference Koter and Koter2019). Second, it illuminates the ongoing construction of identity categories as ‘natural’ sources of trust. This sidesteps the assumption that trust among kin or co-ethnics is a given. Rather, candidate–intermediary alliances generate circumstances and bonds that not only remake senses of shared kinship or ethnicity, but also naturalize these sentiments and identities as normative bases of trust. Third, it brings to the fore the emotions that electoral competition and everyday rivalries animate and amplify across intersecting scales of social life, from ‘the domestic’ to ‘the political’. In doing so, this strategy obviates the tendency to erase and hide the role of emotions in social and political life. This is a common tendency in liberal knowledge making, not least among the middle-class experts, professionals and activists who envisioned post-devolution Kenya and drafted its 2010 constitution (Murungi Reference Murungi2013: 16, 157–90). Yet emotions – not just material interests – continue to have a bearing on how alliances and rivalries play out, on who mistrusts whom or entrusts what to whom, and the forms of competition and mis/trust that emerge as a result.