Integrating households climate change adaptation in a complex evolving economy: the role of different behavior assumptions – presenting at the Social Simulation Conference 2022

by Alessandro Taberna

How do behavioral biases, heterogeneity, and social interactions affect the diffusion of adaptation actions against flooding among households? What are the distributional and indirect economic consequences of such bottom-up actions? These are some research questions I tackled during my presentation at the Social Simulation Conference 2022, Milan, where I had the opportunity to present some of my ongoing work.

How humans respond to climate change has profound implications for well-being, equity, and societies’ ensuring resilience. Traditionally, models that support policy design in nature-society systems omit social complexity, imposing oversimplified assumptions about human behavior, like rationality and lack of diversity. To fill this gap, we employ a novel approach and combine evolutionary economics agent-based models with rich behavioral data collected by SC3.

Leveraging various decision-making processes that range from a representative rational agent (RA) to a heterogeneous population of boundedly-rational agents (BA), we find that adaptation diffusion among households is significantly overestimated when assuming a representative perfectly-rational agent compared to boundedly-rational choices based on the survey data. This “adaptation deficit” (Figure 1) exhibited by a population of empirically-informed agents is explained by economic heterogeneity and alternative decision heuristics that account for behavioral diversity and social influences driving private adaptations. Varied adaptive capacity and uneven adaptation uptake aggravate inequality and undermine resilience when households with low capacity and less-effective undergone adaptations experience long recovery periods following floods.

Figure 1. Adaptation deficit across four behavioral strategies (panel a-c): representative rational agents (RAhom), rational households heterogeneous in incomes, education, and damages (RAhet), representative (BAhom), and heterogeneous (BAhet), boundedly-rational agents. The reported values are averages across 100 Monte Carlo runs for each of the four behavioral framings, with the shaded areas denoting the standard deviations. Panel d shows the regional damages household experience in case of a hazard, in hundreds of millions.

Our results highlight the need to exploit the power of computational models and primary rich behavioral data to explore and gain robust insights from complex social phenomena. Moreover, they show the importance of a unified framework that integrates multiple stakeholders and encompasses monetary damages, focusing also on indirect risk and recovery process.

A special mention to the SSC conference, which has been able to gather many inspiring presentations but also created a convivial atmosphere beyond science that made our stay in Milan both useful and fun (the social dinner in the 16th-century palace definitely helped).