x

We are happy to announce that all trial registrations will now be issued DOIs (digital object identifiers). For more information, see here.
Large-scale land acquisitions and social conflict - experimental evidence from Liberia
Last registered on October 28, 2019

Pre-Trial

Trial Information
General Information
Title
Large-scale land acquisitions and social conflict - experimental evidence from Liberia
RCT ID
AEARCTR-0004792
Initial registration date
October 28, 2019
Last updated
October 28, 2019 11:14 AM EDT
Location(s)

This section is unavailable to the public. Use the button below to request access to this information.

Request Information
Primary Investigator
Affiliation
GIGA German Institute of Global and Area Studies & Universität Hamburg
Other Primary Investigator(s)
PI Affiliation
GIGA German Institute of Global & Area Studies
PI Affiliation
Universität Osnabrück
Additional Trial Information
Status
In development
Start date
2019-10-30
End date
2020-04-01
Secondary IDs
Abstract
Many large-scale agricultural investments (LSAIs) are located in direct proximity to settlements. LSAIs affect the availability and distribution of income and land. While land becoming scarcer may affect livelihoods negatively (e.g. by decreasing nutritional security), employment opportunities that have not existed prior to the investment may arise, constituting a positive shock. In these processes, it has been shown that decision-makers relatively favor specific (groups of) individuals, i.e. those with a specific ethnicity over others. This process can generate changed levels of village-level inequality in employment opportunities, assets and access to land across identity groups. These “horizontal inequalities” are particularly prone to social conflict. In this paper, we experimentally investigate how an external shock affecting the distribution of income and resources, here a LSAI, changes levels of social conflict. We conduct household surveys and “joy-of-destruction” experiments with about 1,800 inhabitants from villages that are affected or not affected from palm oil plantations in rural Liberia. Participants can destroy money of other participants who (1) belong to the chief's ethnic group, or (2) do not belong to the chief's ethnic group.
External Link(s)
Registration Citation
Citation
De Juan, Alexander, Lisa Hoffmann and Jann Lay. 2019. "Large-scale land acquisitions and social conflict - experimental evidence from Liberia." AEA RCT Registry. October 28. https://doi.org/10.1257/rct.4792-1.0.
Experimental Details
Interventions
Intervention(s)
We conduct household surveys and artefactual field experiments, namely joy-of destruction experiments, with about 1,800 individuals in villages that are affected by palm oil investments and villages that are not affected by palm oil investments.
Intervention Start Date
2019-10-30
Intervention End Date
2019-12-21
Primary Outcomes
Primary Outcomes (end points)
amount of money burnt, discrimination depending on the interaction player's identity group
Primary Outcomes (explanation)
amount of money burnt: directly observed
discrimination: differences in amount of money burnt depending on own identity group and interaction player's identity group
Secondary Outcomes
Secondary Outcomes (end points)
objective inequality, perceived inequality, conflict
Secondary Outcomes (explanation)
objective inequality: constructed from household survey data (employment, assets, income, expenditure, land use, access to land)
perceived inequality: constructed from household survey data (perceptions of inequality - questions)
conflict: constructed from household survey data (experience of ethnic and religious conflict, land conflict, conflict over employment)
Experimental Design
Experimental Design
We conduct household surveys and artefactual field experiments, namely joy-of destruction experiments, in 30 villages that are affected by palm oil investments and 30 villages that are not affected by palm oil investments. Participants are able to “burn” the money of an interaction player. We compare amounts of burnt money by individuals from villages that are affected by palm oil investments to the behavior of individuals from villages that are not affected by palm oil investments.

There are two treatments (between-subject design) with varying interaction players (within-subject design):

In treatment 1 (ethnicity salient), each participant takes two decisions: (1) how much money to “burn” if the interaction partner shares the chief’s ethnic group, and (2) how much money to “burn” if the interaction partner does not share the chief’s ethnic group.

Treatment 2 (ethnicity salient + power primed) is almost identical to treatment 1. The only difference between treatment 1 and treatment 2 is that in treatment 1, we just name the chief's ethnic group (e.g. "Bassa") whereas in treament 2, we name the chief's ethnic group and add "the town chief's ethnic group" (e.g. "Bassa, the town chief's ethnic group").
Experimental Design Details
Not available
Randomization Method
Firstly, three palm oil investments have been identified. In these areas, villages have been mapped using satellite images and secondary data. We used propensity score matching to calculate the probability of a village to be affected by a palm oil investment. The final target ("affected" and "not affected") villages have been randomly drawn by a computer.

The artefactual field experiments are conducted in all affected and non-affected villages that have been drawn. We randomly select households, but oversample households that work at a palm oil plantation and household with a minority ethnic group. Whether a participant is assigned to treatment 1 or treatment 2 and the order of interaction partners is random. It depends on a "player number" which participants randomly draw upon entering the experimental activities. This randomization is based on a randomization protocol that has been done in office.
Randomization Unit
We conduct resarch activities around 3 investment sites and randomly select 10 affected and 10 non-affected villages per investment site (randomization unit = village). Within each of the 60 villages, we randomly select household heads (with an oversample of people who work at the plantation and minority ethnic groups) who are invited to join our research activities. In each experimental sessions, participants are randomly assigned to the treatment and the order of their interaction partner (randomization unit = individual).
Was the treatment clustered?
Yes
Experiment Characteristics
Sample size: planned number of clusters
We focus on three investment sites in rural Liberia. In those three areas, we randomly select 10 villages per investment site that are affected by palm oil investments and 10 villages per investment site that are not affected by palm oil investments. In each village, we randomly assign participants to one of two treatments and randomly vary the order of their interaction partners.
Sample size: planned number of observations
1,800 individuals
Sample size (or number of clusters) by treatment arms
30 villages control, 30 villages treatment
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
GIGA German Institute of Global and Area Studies
IRB Approval Date
2019-10-28
IRB Approval Number
03/2019