The Potential of Agent-Based Modelling for Performing Economic Analysis of Adaptive Natural Resource Management

Abstract

This paper explores how individual agent-based modelling can be used by economists and others to evaluate the economic aspects of adaptive management of natural resources. To date, economists have had few tools to perform economic analysis on adaptive natural resource management strategies and there has been limited economic analysis of adaptive management. Part of the reason for this situation may be the inherent nature of adaptive management which involves a series of ‘if-then’ in situ experiments in which expected outputs are not known with certainty, and future management actions depend on the outcome of the experiments. Individual agent-based modelling allows simulation of system wide emergent ecosystem properties that can reflect adaptation of individual agents with bounded rationality to their environment and to the interaction with other agents. These simulations can be used to mimic adaptive management experiments when insufficient information is available for more structured equation-based simulation models. The distribution of simulated outputs from individual agent-based models along with costs of the management actions may give economists the ability to better apply economic analysis to evaluate alternative adaptive management strategies.

Agent-Based Modeling of Harvest Decisions by Small Scale Forest Landowners in Maine, USA

“…. computationally intensive dynamic simulation model(s) of how individual agents (typically using simple behavioral rules) interact with their environment and each other, giving rise to system-wide macro patterns or emergent properties which cannot be deduced from the individual agent's rules" [25]. The interaction between agents and their environments makes ABM a valuable tool to assess repeated decisions of individual landowners.…”

On the potential use of adaptive control methods for improving adaptive natural resource management

“…AM techniques feature the resolution of uncertainty through closed-loop experimental learning on a relatively smaller scale than the grand project. Experiments are used to test hypotheses on the response of ecosystems to intervention, then monitored, and 57 continued/expanded if the hypothesis regarding the response is not rejected, but subject to revision if rejected [16]. The resultant management path is thus path dependent and depends on both the value of information and the costs of acquiring it [3].…”

“…As such, ex ante valuation of information collection is completely ignored using this technique. In AM, however, experimental learning is a key component of the overall management strategy, and thus the use of information becomes paramount [16]. AC offers a way to decompose the expected value of alternative types of system information into its component parts, thus providing a significant amount of insight (information) to a decision maker.…”

“…As such, given alternative future simulated stochastic realities, entire AM plans can be developed and/or evaluated for planning and other purposes, as opposed to the planning of just one experiment in the immediate future. Similar to the 'information exploitation' of [25] and the agent-based modeling approach described in [16], researchers and managers have the ability to use the AC model as a virtual lab by exploring the implications of certain future outcomes on the optimal adaptive path, and planning accordingly. In particular, this could conceivably include the circumstances under which it is optimal to stop a given experiment and start another, even if the full resolution of the uncertainty has not yet occurred.…”

Using Numerical Dynamic Programming to Compare Passive and Active Learning in the Adaptive Management of Nutrients in Shallow Lakes

“…As such, under AM, management plans are path dependent and not a priori defined over the entire length of the planning horizon. Despite adoption of this management technique for several high profile projects, including endangered species management in the Grand Canyon, Everglades restoration, and management efforts in the Missouri river and Klamath River basin (Bureau of Reclamation 1995;DeAngelis et al 2000;Prato 2003; USDA Natural Resources Conservation Service 2004), natural resource economists have largely been silent in documenting the trade-offs involved in AM (Milon et al 1998;Loomis et al 2009). Traditional benefit-cost analysis generally does not take the potential of learning into account, even though information collection may be implicitly valuable (see, e.g., Graham 1981;Miller and Lad 1984;Fisher and Hanemann 1987;Hanemann 1989;Chavas and Mullarkey 2002), and thus is not particularly useful in evaluating AM strategies.…”