Taking agent-based models (ABM) closer to the data is an open challenge. This paper explicitly tackles parameter space exploration and calibration of ABMs combining supervised machine-learning and intelligent sampling to build a surrogate meta-model. The proposed approach provides a fast and accurate approximation of model behaviour, dramatically reducing computation time. In that, our machine-learning surrogate facilitates large scale explorations of the parameter-space, while providing a powerful filter to gain insights into the complex functioning of agent-based models. The algorithm introduced in this paper merges model simulation and output analysis into a surrogate meta-model, which substantially ease ABM calibration. We successfully apply our approach to the Brock and Hommes (1998) asset pricing model and to the "Island" endogenous growth model (Fagiolo and Dosi, 2003). Performance is evaluated against a relatively large outof-sample set of parameter combinations, while employing different user-defined statistical tests for output analysis. The results demonstrate the capacity of machine learning surrogates to facilitate fast and precise exploration of agent-based models' behaviour over their often rugged parameter spaces.
KEY WORDSAgent based model; calibration; machine learning; surrogate; meta-model.
JELC15, C52, C63.
In this paperwe develop the first agent-based integrated assessment model, which offers an alternative to standard, computable general-equilibrium frameworks. The Dystopian Schumpeter meeting Keynes (DSK) model is composed of heterogeneous firms belonging to capital-good, consumption-good and energy sectors. Production and energy generation lead to greenhouse gas emissions, which affect temperature dynamics in a non-linear way. Increasing temperature triggers climate damages hitting, at the micro-level, workers' labor productivity, energy efficiency, capital stock and inventories of firms. In that, aggregate damages are emerging properties of the out-of-equilibrium interactions among heterogeneous and boundedly rational agents. We find the DSK model is able to account for a wide ensemble of micro and macro empirical regularities concerning both economic and climate dynamics. Moreover, different types of shocks have heterogeneous impact on output growth, unemployment rate, and the likelihood of economic crises. Finally, we show that the magnitude and the uncertainty associated to climate change impacts increase over time, and that climate damages much larger than those estimated through standard IAMs. Our results point to the presence of tipping points and irreversible trajectories, thereby suggesting the need of urgent policy interventions.
SignificanceObservations indicate that climate change has driven an increase in the intensity of natural disasters. This, in turn, may drive an increase in economic damages. Whether these trends are real is an open and highly policy-relevant question. Based on decades of data, we provide robust evidence of mounting economic impacts, mostly driven by changes in the right tail of the damage distribution—that is, by major disasters. This points to a growing need for climate risk management.
Since the in uential survey by Windrum et al. (2007), research on empirical validation of agent-based models in economics has made substantial advances, thanks to a constant ow of high-quality contributions. This Chapter attempts to take stock of such recent literature to o er an updated critical review of existing validation techniques. We sketch a simple theoretical framework that conceptualizes existing validation approaches, which we discuss along three di erent dimensions: (i) comparison between arti cial and real-world data; (ii) calibration and estimation of model parameters; and (iii) parameter space exploration.
We provide a survey of the micro and macro economics of climate change from a complexity science perspective and we discuss the challenges ahead for this line of research. We identify four areas of the literature where complex system models have already produced valuable insights: (i) coalition formation and climate negotiations, (ii) macroeconomic impacts of climate-related events, (iii) energy markets and (iv) diffusion of climate-friendly technologies. On each of these issues, accounting for heterogeneity, interactions and disequilibrium dynamics provides a complementary and novel perspective to the one of standard equilibrium models. Furthermore, it highlights the potential economic benefits of mitigation and adaptation policies and the risk of underestimating systemic climate change-related risks.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.