CSYS5010 Introduction to Complex Systems
Assessment 1 – Article Review of a paper using agent-based modelling
Article Info: Cai, J., & Xiong, H. (2017). An agent-based simulation of cooperation in the
use of irrigation systems. Complex Adaptive Systems Modeling, 5(1), 9. doi:10.1186/s40294-
017-0047-x
Literature Review
Length: 497 words
The large-scale irrigation system is considered a dynamic complex system due to a massive
reliance on the cooperation among various agents, such as water users, the local community,
government, etc. Cai and Xiong’s article focuses on the factors that affect the cooperation of
using the irrigation system where agents’ behaviour and their interaction is critical so that the
agent-based modelling(ABM) approach is used in this study.
In the article, the authors indicate a micro-level mechanism of forming the cooperation in an
irrigation system based on previous studies and their field research. Considering a farmer
household as an individual agent, the feasibility to join the cooperation is determined by its
willingness to pay versus the cost. Meanwhile, the government support, seed participant
influence as well as neighbourhood effects are also nontrivial, found in their field research.
Cai and Xiong come up with two hypothesises to establish the ABM: 1) the velocity of
joining the cooperation is positively associated with the proportion of seed participants; 2) the
velocity and the coverage of cooperation both are positively associated with the
governmental subsidies.
The model comprises four types of entities: social system (i.e. a village, as a network),
irrigation resource, water users(households) and the government. At the initial period, a small
portion of households in the village are set to participate in the cooperation, called seed
participants. In each iteration of the process, non-participants firstly check if the cost of
cooperation is within the willing-to-pay limit then make their own decision under a random
cooperation propensity. Additionally, the environment of the model is initialized by creating
a random or scale-free network that consists of households and an irrigation resource, which
can be represented explicitly by a set of parameters. Finally, the results examined in NetLogo
show that the participation rate and speed increases with the increase of the ratios of
government subsidies, respectively, however, does not vary significantly with that of seed
participants, which means 2) is verified while 1) is not supported by the outcome. A robust
regression analysis echoes the above findings regarding parameters at the 1% significance
level. Therefore, the authors conclude the study that the government support plays a critical
role in the formation of irrigation cooperation.
Overall, Cai and Xiong’s study is an icebreaker among rather limited studies done in its field.
It's a remarkable example of ABM referring to Sayama’s modelling loop, for instance, some
reasonable hypothesises are obtained from observation and a considerable amount of
experiments (3,840,000 sets of running) to ensures the objectivity and variety. Nevertheless,
the limit of the study where I raise my concern is that an insufficient quantity of surveys was
collected in the pre-research (only 20% of 20 households in average answered), and also as a
result of refuting a hypothesis, there’s no further evaluation or adaption of the model.
Besides, the ODD protocol is employed to describe the model, which provides a standard
format of a systematic and visible structure to help with documenting any large, complex
ABMs.














































学霸联盟