In much of the world, including much of Africa, politics is organised less around competing policy programmes than around the targeted exchange of benefits for support — and often along ethnic lines. This module explains clientelism and ethnic politics not as cultural facts but as equilibria produced by weak institutions and the collective-action logic of earlier modules. The evidence base here is unusually strong, including some of the cleanest empirical work done on African politics.
Programmatic vs clientelistic politics
The distinction
Programmatic politics: parties compete by offering policies that apply impersonally to everyone who qualifies (a universal pension, a tax cut, a health scheme). Your benefit does not depend on how you voted. Clientelistic politics: politicians (patrons) deliver targeted, individualised benefits — a job, a bag of fertiliser, a cash handout, a borehole — in direct or implicit exchange for political support, channelled through brokers who know who voted how. The benefit is contingent on loyalty.
Clientelism is not merely vote-buying on election day. It is an ongoing relationship: the patron provides a stream of help (school fees, hospital bills, funeral contributions, jobs) and the client provides reliable support. Susan Stokes and co-authors (Brokers, Voters, and Clientelism, 2013) emphasise the broker — the local intermediary who solves the patron's information problem by monitoring who is loyal and targeting benefits accordingly. Without brokers, the patron cannot tell whom to reward, and the exchange breaks down.
Why clientelism persists
Clientelism is a trap in the game-theoretic sense. For a poor voter, a guaranteed bag of fertiliser now from a patron who will deliver is worth more than a promised universal programme that may never materialise from a state with low capacity and low credibility. For the politician, targeted benefits buy support more cheaply and verifiably than broad public goods. So both sides rationally choose the clientelistic exchange — even though everyone might be better off under a programmatic, public-goods equilibrium. It is a collective-action and credibility failure, and it reproduces the low state capacity that makes the programmatic alternative non-credible in the first place.
Ethnicity as the organising axis
Why does the targeting so often run along ethnic lines? The leading explanations are instrumental, not primordial. (1) Information and enforcement: ethnic networks lower the cost of identifying members, monitoring reciprocity, and excluding outsiders — they make the clientelist exchange workable, much as Olson's selective incentives make organisation workable. (2) Coordination: in a context where the state targets benefits to groups, voting with your ethnic bloc is a rational way to be in the winning coalition that gets the roads and jobs; expecting others to vote ethnically makes it your best response too. Daniel Posner (Institutions and Ethnic Politics in Africa, 2005) showed that which ethnic identities become politically salient depends on the institutional arena — the same Zambians emphasise different cleavages under different electoral rules — strong evidence that ethnic politics is strategic, not fixed.
The Kenyan road-building evidence
Burgess, Jedwab, Miguel, Morjaria & Padró i Miquel (2015)
In one of the cleanest tests of ethnic favouritism, the authors examined road-building across Kenyan districts from 1963 to 2011. Under autocracy, districts that shared the ethnicity of the sitting president received roughly twice the development spending and around five times the length of paved roads of other districts — a stark measure of ethnic targeting. The striking second finding: during the democratic period, this favouritism essentially disappeared. Electoral competition, by forcing leaders to build a broader winning coalition, disciplined the ethnic bias. The paper is the empirical bridge between selectorate theory and ethnic politics: democracy enlarged W, and the targeting shrank.
Preferences or coordination?
Does ethnic politics reflect a genuine preference for one's own group, or just the equilibrium of a coordination problem? Habyarimana, Humphreys, Posner, and Weinstein (Coethnicity, 2009), using experiments in Kampala, found little evidence that people simply favour co-ethnics out of taste. Instead, co-ethnic cooperation was driven by stronger norms of reciprocity and easier sanctioning within ethnic networks — a technology for solving collective-action problems, not an end in itself. This matters for reform: if ethnic politics is a coordination-and-enforcement equilibrium rather than fixed animosity, then institutions that change the coordination problem (electoral rules, cross-cutting coalitions, programmatic public goods, credible universal provision) can shift behaviour — exactly what the road-building evidence shows democracy did.
Exercise
A reformist governor wants to move her county from clientelistic to programmatic politics — from targeted handouts to universal public services. (1) Explain, using the credibility and collective-action logic, why individual voters might rationally prefer the clientelist's handout even though they would collectively be better off under universal provision. (2) Using the Burgess et al. finding and selectorate theory, explain what institutional condition made ethnic favouritism decline in Kenya, and what that implies for the governor's strategy. (3) The Coethnicity research suggests ethnic cooperation is about reciprocity and enforcement, not taste. How does that finding make the governor's goal more achievable than a 'primordial' view of ethnicity would suggest? (4) Propose two concrete institutional moves that attack the clientelist equilibrium at its roots.