Cohesive Institutions and Political Violence

Being more like Norway—that’s the aspiration of many resource-rich countries. But rather than raising standards of living broadly within a society, resource wealth is often associated with instability, civil conflict, repression, and political violence. This translates into poor development outcomes, feeding the notion that natural resources can be a curse.

How can countries escape the natural resource curse? And to what extent do cohesive and democratic institutions facilitate this process? In a new CGD working paper, Cohesive Institutions and Political Violence, we look at Nigeria—often seen as the prime example of a country cursed by its wealth. We show that when political institutions are cohesive and power is shared among the diverse groups in a multi-ethnic society, political contests over resource revenues are less likely to be violent. What produces cohesive institutions? Democratic elections.

Nigeria’s unique context

The bulk of government funds in Nigeria are sourced from oil extraction (see figure 1a) and these revenues are shared according to a funding formula across tiers of government, with about 20 percent arriving at local governments. These resources are highly contested politically. While they are earmarked for public administration and the provision of public goods, there is plenty of anecdotal evidence that public funds are embezzled by local political elites. Governance and political institutions at the local level are thus key to understanding how the resource curse may or may not unfold at the local level.

In our study, we draw on unique data to observe local authorities over time across two different institutional regimes: local governments that are elected and local governments that are appointed. These appointments are often far from nonpartisan and balanced, raising the possibility that opposition groups will be systematically excluded from the governance and oversight of resource revenues. Grievances among constituencies may in turn help to mobilize fighters to engage in conflict.

So how do democratic institutions at the local government level influence the contest for resource revenues? How does the representation of ethnic groups make the government institutions more cohesive? And why do cohesive institutions make the process of resource allocation less antagonistic? The de-facto variation in local democratic institutions in Nigeria provides interesting insights. Under the Nigerian Constitution, local government councils are supposed to be elected by the people, but in many instances, they are appointed by state governors (see figure 1b). This variation allows us to study the effects of different power-sharing regimes at the local government level on political violence, using monthly tax revenue allocations as an exogenously determined measure for contested resources.

Figure 1. (a) Brent oil price and statutory allocations o the LGAs; (b) Share of elected local governments over time across Nigeria

Graphs showing brent oil price and statutory allocations o the LGAs; (b) Share of elected local governments over time across Nigeria

Notes: Panel (a) Monthly variation in overall revenue allocations made to local governments (right scale), and monthly prices of Brent Crude oil (left scale). Panel (b): Share of months with an elected local government for each local government area in the period 1999 to 2014. Panel B: Monthly time series variation in the share of local government areas with an elected government council over time.

Untangling the link between violence and resource wealth

Nigeria’s political economic set-up lets us delve into three important questions: Does resource wealth cause conflict? Do elected local governments promote peace? Do differences in the cohesiveness of local government institutions explain these effects (and if so, why)? Here’s what we find:

1. Large windfalls of politically controlled natural resource revenues encourage violence

We consider Nigeria’s system of oil revenue sharing across the three tiers of government. Revenues highly correlate with global oil prices, as displayed in figure 1 (a). We find a significant and economically sizeable link between resource rents and the incidence of political violence (this is consistent with a theoretical prediction by Besley and Persson). These effects are driven by positive shocks to resource rents. The ensuing low-intensity conflict is highly institutionalized, involving government repression and militias using targeted violence, though not broad, open rebellions involving riots or protests.

2. Elected local governments are more successful than appointed governments in discouraging violence

Looking at idiosyncratic variation in whether local governments are elected or appointed, we find that having an elected local government systematically weakens the link between rents and the incidence of political conflict. Rather than focusing on any individual election—which may be prone to violence directly affecting its outcome—we contrast consecutive periods over which local governments are elected (as opposed to appointed) and study the systematic link between shocks to rents and violence across such periods.

Our identification strategy allows us to tackle many plausible endogeneity concerns, while providing sharp results that match key theoretical predictions in Besley and Persson. We also turn to individual-level micro data to corroborate our findings, documenting that fear of political violence, actual victimization, and even engagement in conflict broadly follow the pattern suggested by the aggregate data.

3. The degree of cohesiveness generated by an elected (versus appointed) local government explains why violence is used to contest rents in one but not the other

We construct a measure of non-cohesiveness using Demographic and Health Surveys (DHS) data capturing the extent to which the ethnic makeup in an area is aligned with the ethnicity of the state governor. We provide ample anecdotal evidence suggesting that when local governments are appointed, these appointments are skewed towards the ethnicity of the state governor. We validate this measure using individual-level micro data. Our third set of findings highlights that the higher degree of cohesiveness of institutions that elections (as opposed to appointments) produce, the weaker is the link between resource rents and political violence. In other words, resource rents provoke conflicts, but only when local governments are appointed and thus ethnically aligned with the state governor, and not when they are elected. The local elections seem to improve the representation of political factions at the local government level—and help to resolve the contest for resources more peacefully.

In figure 2, we present evidence of the non-linearities in the effects. Panel (a) shows that with an appointed local government, there is an association with negative shocks to resource rents being associated with less conflict, while positive shocks to resource rents trigger conflict. Panel (b) constructs the figure using the subset of data when local councils are elected: while negative shocks continue to be weakly associated with less conflict, positive revenue shocks are not associated with conflict. This result suggests that with elected local governments, the tensions over distributional disputes arising with positive revenue shocks are resolved in a non-violent manner.

Figure 2. Positive and negative resource allocation shocks under elected and appointed regimes

Graphs showing positive and negative resource allocation shocks under elected and appointed regimes

Notes: The figure presents results from bootstrapped lowess regressions on the residuals of the dependent variable, after having demeaned the data by LGA and state-by-time fixed effects. The method first computes lowess regressions from 1000 bootstrapped samples of the demeaned data. It then calculates density estimates of the predictions from the lowess regressions for several hundred cuts along the y-axis and distributes a specified color proportional to that density estimate. The resulting figure displays the uncertainty in the regressions visually. The median value of the bootstrapped lowess predictions is indicated as a solid white line, while OLS regressions are indicated as a green line.

Whether scarce resources are allocated in a concordant and peaceful manner or invoke coercion and violence is an important question that guides a lot of political economy research. This contributes to our understanding of how a particular type of democratic institution—elected local governments—can shape and moderate the interactions between groups in a peaceful fashion. This question is particularly relevant to countries with significant resource wealth, such as Nigeria, as democratic oversight and a tight institutional framework are proposed to be critical for ensuring that countries are not cursed by their resource wealth.


CGD blog posts reflect the views of the authors, drawing on prior research and experience in their areas of expertise. CGD is a nonpartisan, independent organization and does not take institutional positions.

Cohesive Institutions and Political Violence, joint with Stephan Kyburz, Review of Economics and Statistics.

🔓 Open access


Can institutionalized transfers of resource rents be a source of civil conflict? Are cohesive institutions better at managing conflicts over distribution? We exploit exogenous variation in revenue disbursements to local governments and use new data on local democratic institutions in Nigeria to answer these questions. There is a strong link between rents and conflict far away from the location of the resource. Conflict over distribution is highly organized, involving political militias, and concentrated in the extent to which local governments are non-cohesive. Democratically elected local governments significantly weaken the causal link between rents and political violence. Elections produce more cohesive institutions, and vastly limit the extent to which distributional conflict between groups breaks out following shocks to the rents. Throughout, we confirm these findings using individual level survey data.

Press Coverage & Non-Technical Summaries


Center for Global Development (US) 🇺🇸.

Social Media Summary