Policy and Complex Systems
demonstrates an agent-based model of farmer decision-making on water quality in the context of first and second-generation biofuel crops and carbon trading . The ABM integrates a SWAT-based hydrologic – agronomic model .
In the bottom-up construction of an ABM , modelers need to assign decision rules to agents under specific scenarios . A major challenge lies in constructing credible decision rules for ABM ( Zenobia , Weber , & Daim , 2009 ). Most of the previous work usually assumes perfect rationality , meaning that the agents could perfectly solve for utility maximizing problems in various and complex scenarios . However , behavioral economics have repeatedly shown that human behavior is often rationally bounded at best and that individuals often use heuristics instead of optimization when making decisions . As noted by Heckbert , Baynes , and Reeson ( 2010 ), combining economic experiments with ABM offers researchers many new opportunities . Experimental economics can be used to guide calibration of ABM so that the agents ’ behaviors and decisions reflect patterns identified by actions in experiments .
Some researchers have used survey methods to develop decision rules for ABM ( Dia , 2002 ). Compared to using survey-based approaches to calibrate decision-making in ABM , we can use data collected through experiments that capture the “ interpersonal ” and “ interplayer ” dynamics that arise in experimental games ( and are overlooked by surveys ). Furthermore , Duffy ( 2006 ) pointed out that ABM projects also could facilitate researchers ’ ability to interpret the aggregate findings of an experiment involving human subjects .
Not many studies have combined experimental economics and ABM . Evans , Sun , and Kelley ( 2006 ) compared results from a spatially explicit laboratory experiment to outputs of a simulation from a land-use ABM involving utility-maximizing agents . They concluded that the participants in the experiment deviated from revenue-maximizing actions and that it was thus valuable to use non-maximizing agents in ABM . Heckbert ( 2009 ) also acknowledged the value of combining experiments and ABM , reporting a study in which a participant replaces the role of an agent and the participant ’ s behavior under several treatments can be used to recalibrate the ABM . A few studies have attempted to integrate economic experiments and ABM in NPS pollution management context . Zia et al . ( 2016a ) constructed agent-based models using an economic experiment documented in Miao et al . ( 2016 ). The agents were categorized to pursue different behavioral strategies under alternate policy and sensor information regimes , and a multi-level multinomial logistic regression model built from experimental data predicted the agents ’ type categories . Our research extends this idea by designing an experimental setting that includes technology adoption decisions and two layers of heterogeneity , meanwhile building a closer link between the experiment and the ABM .
We also include two information treatments to examine the ability of information “ nudges ” to induce desired outcomes from the participants . Originating from the social comparison theory by Festinger ( 1954 ), it has been shown that information “ nudges ” on social comparison and peer actions can promote environmental conservation behavior ( Allcott , 2011 ; Ferraro & Price , 2013 ; Goldstein , Cialdini , & Griskevicius , 2008 ). These information “ nudges ” are attractive from a policy design perspective since they are more cost-effective compared with traditional monetary-based programs . However , not much research has considered incorporating in-
167