Command and Control ABM

The command and control policy in natural resource management, including water resources, is a
longstanding established policy that has been theoretically and practically argued from the point
of view of social-ecological complex systems. With the intention of making a system ecologically resilient, policymakers apply the top-down policies of controlling communities through regulations. To explore how these policies may work and understand whether the ecological goal can be achieved via command and control polices, this research uses the capacity of Agent-Based Modeling (ABM) as an experimental platform in the Urmia Lake Basin (ULB) in Iran, which is a social-ecological complex system and has gone through a drought process.

Despite the uncertainty of the restorability capacity of the lake, there has been a consensus on the possibility to artificially restore the lake through the nationally managed Urmia Lake Program (ULRP). To reduce water consumption in the Basin, the ULRP widely targets the agricultural sector and proposes the project of changing crop patterns from high-water-demand (HWD) to low-water-demand (LWD), which includes a component to control water consumption by establishing water-police forces.

Using a wide range of multidisciplinary studies about Urmia Lake at the Basin and sub-basins as well as qualitative information at micro-level as the main conceptual sources for the ABM, the findings under different strategies indicate that targeting crop patterns change by legally limiting farmers’ access to water could force farmers to change their crop patterns for a short period of time as long as the number of police constantly increases. However, it is not a sustainable policy for either changing the crop patterns nor restoring the lake.

Thank you for the presetation Farzaneh. Since you started the presentation on Urmia Lake Basin, I wondered how you will use empirical data from your case from your study in the modeling effort. At the moment the model is rather abstract. Do you plan to have an empirically validated model, or do you have broader research questions related to irrigated agriculture that could be applied to other parts of the world.

Thanks Marco for your attention to the Urmia Lake Basin case. Your questions about the input data, validation of the model, and the plan for a broader research on irrigated agriculture are very valuable. And they present the great opportunity that I can elaborate some of the unique challenges that I faced for this project.

As the background of the model, I have to state that, as you may know, Urmia Lake Basin received a great international attention in both academic and practical research and more than ten international universities carried out deep studies in collaboration with top universities as well as organizations at national and local levels in Iran. I have studied the Basin with different purposes in last three years. For example, I reviewed Agricultural Land Use Scenario Development Method in the Basin to understand whether the applied method was suitable for the complexity of social-ecological system of Urmia Lake. Or I studied the probability that Urmia Lake could be restored through the Restoration Program and modelled the adaptability of social-ecological complex system of Urmia Lake. However, in my studies I found out that there was nearly no studies at individual behavioral level. Only recently a narrative study was carried out by Sharif University with a very specific objective that made it difficult to use for different purposes. Even though the Command-And-Control ABM doesn’t have any direct input data, it is built on all these studies plus my own personal informal studies in the area. Therefore, I believe I have used empirical data from my studies. I would be curious if you can suggest any other ideas about the empirical data?

Regarding the empirically validating the model, besides the internal validation of the model as an exploratory that can be found in the presented paper, I would like to explain two types of information that I believe can be considered as the informal empirical validation of the model. One of them is the support that the model received in the presentation in Iran. Besides the model objectives, my personal objective of programing the Command-And-Control ABM was to present the capacity of Agent-Based Models to policy makers, academia, as well as communities in Iran in order to encourage them to carry out a participatory Agent-Based Modeling for resilience building in the Basin because based on my earlier studies the successfulness of the Urmia Lake Restoration program was sensitive to early participation and resilience building process. When I presented the Command-And-Control model in several universities and organizations at the national level it was praised. The most interesting reactions were observed at the local level. When the model was presented at the local meetings, where the experts from different communities and organizations were attending, the participants started to support the model by giving the information. For example, they explained that some of the water-polices that were recruited from the same communities ignored that people were taking extra water or digging wells. Or, some of those who were recruited from outside of the community accepted bribes to ignore the illegal water pickup. Also, it was told that some of the water-polices didn’t work effectively. All of them have been seen in the model as the water- police efficiency and whether farmers are at the vision of water-polices. Also, the photos and data were published and showed how farmers took surface water even in the places where smart control systems were installed and how many times wells were closed, yet farmers tried to dig wells again, all supported the relationships within the model. The second type of information that I believe practically validated the model was some of the published information about the result of water-police policy. For example, according to Urmia Lake Restoration Program (ULRP), in 2014 the numbers of water-police stations, which were established in three provinces of the Basin, were 9 in East Azerbaijan, 30 in West Azerbaijan, and 22 in Kurdistan. The latest report indicated that the number of police stations had been increased to 280. In the roundtable discussion with IRNA news agency, the head of the Water Department in East Azerbaijan said that they had started with 12 water-police in 2012 but their numbers had to be increased to 320 in 2018. As you may have noticed, in the model I started with 10 water-police stations and water-polices are deployed one by one randomly from each station until the policy goal is met. Therefore, the model starts with 10 water-police. In the satisfactory policy option, the number of water-police increases up to 300-400 and plays around that numbers. In the extreme policy, the number of water-polices increases up to several times of farmers and makes the farmlands a battle field, still the objective of the policy cannot be met. I am curious to know if you also consider these the empirical validation of the model. Do you have any other suggestions?

For the matter of broader research questions, I would like to say that I had expanded the model by adding the agricultural extension network as one of the other policy component of the ULRP. Moreover, I have been working to model the resilience process through learning and self-organizing. I wish I could have programed it as a participatory ABM, but unfortunately due to high political, social, and economic tension in the region and the country, I am programing it in abstract form and hoping to use it to encourage for participatory model later.

I hope I was able to answer your question and I look forward to further discussions.

Interesting research! I read through the model code and documentation, and would like to challenge you with two comments.

First, your model investigates the effect of monitoring and sanction intensity (as measured by the relative number of police force) on resource use sustainability. However, such policy is not limited to command-and-control economies. Command-and-control has many manifestations, e.g., in the pre-1978 China, it was more about the problem of distorted and inefficient resource allocation by a central planner due to imperfect information (about production and consumption). Therefore, when interpreting the present model results for discussing command-and-control, we will benefit from making model assumptions explicit to avoid over-generalization.

A following comment for making model assumptions more explicit is on the initialization of model parameters and inputs. Some of the initial values are given in Table 1 of your model documentation, yet in the code, there are more hard-coded parameters not included in the table or shown in the Interface (e.g., radius set to 2 and the parameter settings of the increase/decrease police sub-model). With more model parameters, the generalizability of the model will be lower. Moreover, parameters in the Table 1 with non-random values can be better grounded on empirical findings (e.g., add a column to provide the references) or justified by sensitivity analysis.

Looking forward to your paper on this model.

Hi Bing-bing Zhou!

Thanks for your interest in the Command and Control ABM model. Your comments and carefully review of the model are very valuable. Here is my short response but I would like to continue discussion after you read the full paper.

The model definitely supports your definition of command and control in natural resources as a top-down policy making, specifically land use planning. This model is not about just monitoring and sanction, as you noted. It is a simulation of one of the smallest component of the command and control policy of the Urmia Lake Restoration Program (ULRP). Methodologically, the land use planning of ULRP demonstrates a top-down method, and in the planning, itself, all the high-water demand crops and low water demand crops are listed and according to the plan the crop pattern has to be changed from high water demand crops to low water demand crops without taking into consideration of farmers or local capacity for this decision. The water-police forces have been introduced as the controlling instrument of the command to change the crop type.

Regarding the initialization parameters, such as setting the radius 2 for the water-police vision, you can find the explanations of the parameters and full sensitive analysis for each of them as well as in combination in the paper. Regarding three policies, the calculations are based on the available information, which have been given in the paper.

I like the idea of adding references in the table 1 to show some of the parameters are grounded on the studies, but they can be found in the paper.

Your engagement in the discussion made me to review the parameters and their definitions once again and I highly appreciate it.

Hi and thanks for your presentation, it was instructive to see how an issue like this could be broken down into an agent-based model. You say in your presentation that this program targets areas currently dominated by high water demand crops and I was curious how your model would differ if you wanted to model these dynamics in regions dominated by low water demand crops (or if that would even be a relevant thing to model). Also, when your farmers decide to seek extra water, does or should their success in doing so depend on whether their neighbors are also seeking extra water?

Thanks Matthew for listening to the presentation and writing to me. Your both points are very valuable and the model can be extended to carry out those experiments.

For the first question, as you noted, my questions and experiments are just for the situation where the high water demand crops are dominant. When I received your note, I started to run the model for the situation where low water demand crops are about 80% of the farmlands. I found the pattern of crop changing very interesting and I guess comparing two situations would be a good research agenda to be explored by extending this model.

Regarding the second question, the number of neighbors with high water demand crops is a variable that each agent takes into consideration before deciding to take action for moving to take extra water. The reason is the security that agent feels when its neighbors do the same. In the real world, the law enforcements rely on the neighbors reports. However, the number of neighbors with high water demand crops doesn’t affect in the success of agent to pick water. I have been working on an ABM in resilience building mechanism in conflict situation, in which community engagement and support is important and the agent’s neighbors take group action to defend their group decisions if the agent faces legal problem. Your point is valuable to be explored by extending this model.

Once again, thanks for your thoughtful questions and if you have more ideas for extending this model I would like to learn about them.