RCT of Farmer-Herder Conflict Resolution Mechanisms in Upper West, Ghana

How to cite

Haugan, Gregory; Ome, Alejandro; Bukari, Kaderi. 2024. "RCT of Farmer-Herder Conflict Resolution Mechanisms in Upper West, Ghana." AEA RCT Registry. AEARCTR-0012874. https://doi.org/10.1257/rct.12874-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Gregory Haugan* , Alejandro Ome , Kaderi Bukari

* Primary investigator

Registry ID
AEARCTR-0012874
DOI
https://doi.org/10.1257/rct.12874-1.0
First registered
Last updated
Status
On going
Trial dates
2023-11-20 – 2025-01-31
Countries
Private
JEL codes
D74, D02
Registry record
http://www.socialscienceregistry.org/trials/12874

Abstract

Conflicts between Fulbe (also known as Fulani) pastoral herders and settled farming communities in Northern Ghana and the wider Sahel have grown more frequent and violent in recent years. While these conflicts are due in part to competition for scarce resources, exacerbated by climate change and increased migration from the Sahel, complex social forces are also at play, escalating conflicts into violence through interactions between and within social networks and key actors. Within this context, the USAID/OTI Littorals Regional Initiative (LRI) will implement a three-pronged intervention to address tensions between Fulbe herders and non-Fulbe farmers in a set of communities across the Upper West region of Norther Ghana: (1) a facilitated dialogue session between farmers and herders; (2) inter-ethnic mediation committees to resolve disputes; and (3) integrated Village Savings and Loans Associations (VSLAs) to increase contact and cooperation between the two groups.

The evaluation of the intervention is conducted as a cluster-randomized control trial in 46 communities across the region, where 23 are assigned to receive all three components of the three-pronged intervention, and 23 are assigned to a control condition. Key outcomes will be: perceptions among Fulbe and non-Fulbe households that disputes are resolved fairly, presence of recent violent conflict, perceived likelihood of future uncompensated crop destruction among non-Fulbe households, and perceived tenure security of Fulbe households. These outcomes are measured through baseline and endline household surveys with both men and women in Fulbe and non-Fulbe households, as well as surveys with community leaders.

Intervention

USAID/OTI Littorals Regional Initiative (LRI) will implement a three-pronged intervention to address tensions between Fulbe herders and non-Fulbe farmers in a set of communities across the Upper West region of Norther Ghana: (1) a facilitated dialogue session between farmers and herders; (2) inter-ethnic mediation committees to resolve disputes; and (3) integrated Village Savings and Loans Associations (VSLAs) to increase contact and cooperation between the two groups.

Experimental design

We conduct baseline and endline surveys (approximately one year apart) with households in 46 communities to form a two-period household-level panel. After the baseline survey, communities are matched using 1-1 matching without replacement, and then one community in each matched pair is randomly assigned to treatment and one to control. See pre-analysis plan for additional details.

After endline, we use an ANCOVA regression to test for the difference in outcomes between treatment and control groups at endline, controlling for baseline values of the outcome variable, fixed effects for district, and household- and respondent-level controls: gender, age, education, dummy for being born in the community, number of household members, dummy for all school-age children in school, number of cattle owned, and farm plot size.

As a robustness checks, we also run a Dif-in-Dif model with community fixed effects, and Dif-in-Dif with household-level fixed effects. The purpose of these is to see if the ANCOVA results remain robust to more demanding models with less statistical power.

Design details

Randomization method
Following baseline data collection, NORC conducted a community-level statistical matching exercise. The matching created a set of 23 blocked pairs of communities, where each blocked pair included a set of two statistically similar communities. NORC then randomly assigned one community in each blocked pair to treatment using a program run in an office by a computer. See pre-analysis plan for additional details.
Randomization unit
Community
Clusters
46 total - 23 treatment and 23 control.
Observations
10 Fulbe and 10 non-Fulbe households are surveyed in each of the 46 communities, for a total of 460 Fulbe and 460 non-Fulbe households, and a final sample of 920 total households. Hypothesis tests are run separately for Fulbe and non-Fulbe.
Arms
23 treatment communities (with a total of 230 Fulbe and 230 non-Fulbe households), and 23 control communities (with a total of 230 Fulbe and 230 non-Fulbe households).
Primary outcomes
(1) Tenure security: (a) Percentage of Fulbe respondents who say it is unlikely they would be asked to leave the community in the next three years. (b) Percentage of non-Fulbe respondents who say it is unlikely someone could destroy crops without adequate compensation in the next three years. (2) Perceptions among Fulbe and non-Fulbe households that disputes are resolved fairly (3) Presence of violent clashes during most recent rainy season.
Secondary outcomes
See attached pre-analysis plan.

Cited by

No citation data available for this registration yet.

Keywords