This repository contains the code for the ABM described in E.K. Burchfield and J. Gilligan, "Dynamics of Individual and Collective Agricultural Adaptation to Water Scarcity," accepted for publication *Proceedings of the 2016 Winter Simulation Conference," T.M.K. Roeder, P.I. Frazier, R. Szechman, and E. Zhou, eds.

Drought and water scarcity are growing challenges to agriculture around the world. Farmers can adapt through both individual and community-based collective actions. We draw on extensive field-work conducted with paddy farmers in rural Sri Lanka to study several adaptations to water scarcity, including switching to less water-intensive crops, farming collectively on shared land, and individually turning from surface water to groundwater by digging wells. We explore how variability in climate affects agricultural decision-making at the community and individual levels using three types of decision-making, each characterized by an objective function: risk-averse expected utility, regret-adjusted expected utility, and prospect theory loss-aversion. We also assess how the introduction of individualized access to irrigation water with wells affects community-based drought mitigation practices. Preliminary results suggest that the growth of well-irrigation may produce sudden disruptions to community-based adaptations, but that this depends on the mental models farmers use to think about risk and make decisions under uncertainty.

The main model is contained in agrowell_abm.py. Code to generate the payoff table is contained in payoff.py Code to generate the figures from the paper is contained in read_abm_data.R (Fig. 2--3) and survey_analysis/log_model_ABM_stan_no_mlm.R. The results of the Bayesian regression are contained in lfit_no_mlm.Rda (Fig. 1)

The command to reproduce the model runs described in the paper is:

python parallel_simulation.py