The role of mutual insurance network in smallholder farmers' cooperation under drought uncertainty: a dynamic irrigation game

How to cite

Siltanu, Gebretsadik Teshager; Gugsa, Gebreegziabher Fentahun; Kahsay, Goytom Abraha; Belay, Dagim. 2025. "The role of mutual insurance network in smallholder farmers' cooperation under drought uncertainty: a dynamic irrigation game." AEA RCT Registry. AEARCTR-0015313. https://doi.org/10.1257/rct.15313-1.0

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

Gebretsadik Teshager Siltanu* , Gebreegziabher Fentahun Gugsa , Goytom Abraha Kahsay , Dagim Belay

* Primary investigator

Registry ID
AEARCTR-0015313
DOI
https://doi.org/10.1257/rct.15313-1.0
First registered
Last updated
Status
On going
Trial dates
2025-02-05 – 2025-04-15
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/15313

Abstract

Climate change presents a tangible and pressing reality, with its observed and projected impacts disproportionately affecting nations and sectors worldwide. Smallholder farmers in low-income countries are particularly vulnerable due to their limited resilience, constrained adaptive capacity, and dependence on climate-sensitive livelihoods. In our study areas, these farmers operate within already degraded landscapes, further impeding their adaptive capacity and contributing to the low adoption of climate change adaptation strategies. We employ a dynamic irrigation game to examine the role of mutual insurance network on collective action in integrated watershed management as a pathway for climate change adaptation.

Experimental design

The study involves a lab-in-the-field experiment (an irrigation game) to examine watershed management and river water appropriation for irrigation. The experiment is framed to minimize participants' computational burden in the game and imitate real-world scenarios. The experiment will be conducted with smallholder farmers in Ethiopia engaged in irrigation agriculture and have experience in campaign-based watershed management activities, such as tree planting and soil and water conservation. The design combines within- and between-subject components to estimate the effects of different intervention treatments under dynamic scenarios. Participants will be randomly assigned into one of the four experimental arms: (i) drought incidence, (ii) mutual insurance, (iii) drought incidence and mutual insurance, and (iv) control group. Participants in the experiment play in groups of 5 in 2 stages, a total of 7 rounds. The five players are randomly assigned to positions 1, 2, 3, 4, and 5, ranging from upstream (position 1) to downstream (position 5). The groups and positions in the group remain fixed throughout the game. In stage 1, all participants play a classical irrigation game for 2 rounds. In the second stage, participants play the irrigation game with one of the assigned experimental arms, three treatments, and a control group. The control group continues to play the classical irrigation game as in stage 1.

Drought incidence: Participants randomly assigned to this treatment will face shared vulnerability to drought, with a known risk of 60%. This indicates that in five rounds of the second stage, drought events will occur in three rounds. Subsequently, a reduction in water flow that has a potential effect on output level will have the probabilities of 10%, 30%, and 50% with a 50%, 30%, and 10% of water loss to show the unpredictability of drought incidences. However, although participants have complete information on the risk of droughts and the severity of possible drought effects, information on the severity of an impact at the time, t, remains incomplete increasing the uncertainty. The uncertainty may directly affect participants' decision to invest in the watershed. Proactive watershed restoration in earlier rounds increases the maximum water generation potential, mitigating current and future drought effects. Accordingly, water availability at period t is determined by drought incidence, the collective efforts of participants, and their past actions.

Mutual insurance network: We will introduce a community-based social insurance scheme in this treatment. Participants can allocate a portion of their endowment to a shared fund in addition to investing in the watershed project and their private account. Participants experiencing severe drought and crop losses due to water shortages can draw from this shared fund to recover their endowment. A significant loss occurs when participants fail at least to recover their initial endowment, drawing from the mutual fund solely for recovery purposes. We hypothesize that this insurance mechanism may incentivize downstream participants to invest more in watershed management, thereby enhancing aquifer levels.

Mutual insurance network during drought scenarios: In this treatment, we will examine the interaction between drought scenarios and mutual insurance mechanisms on participants' collective action. Specifically, we aim to investigate whether the presence of mutual insurance influences participants' efforts to invest in watershed management and their strategies for allocating irrigation water under drought conditions.

Each round follows two steps: first, each player receives an initial endowment of 30 Ethiopian Birr (ETB), which they can either invest in watershed structures or save in their private account. Investing in watershed structures improves the future water retention potential. Therefore, the hydrological system becomes dynamic, and previous group contributions and depreciation of the previous watershed structures now influence the maximum water generation potential of the aquifer. Second, players extract water from the river for irrigation in the order of their assigned position, i.e., a player assigned to position 1 will be the first to extract water, and a player assigned to position 5 will be the last to extract water.

Design details

Randomization method
We conducted randomization through a public lottery.
Randomization unit
An individual level.
Clusters
no clusters
Observations
3430 observations: 490 individuals, each playing 7 rounds
Arms
840 observations (120 individuals playing 7 rounds) for drought incidence, mutual insurance, and control groups, and 910 observations (130 individuals playing 7 rounds) for drought incidence interacted with mutual insurance
Primary outcomes
Watershed investment and water extraction
Secondary outcomes
Inequality of investment and extraction

Cited by

No citation data available for this registration yet.

Keywords