The alternative and novel approach presented here, to represent energy economic agents that are heterogeneous, diverse, evolve in space and time, and make decisions under exogenous constraints, is based on (i) the progress on the combination of Geographical Information Systems with Agent-based modelling, GIS-ABM, suggested by Crooks, et al. [1] and the theoretical foundation from the so-called heterodox school of thought, such as (ii)  the Theory of Bounded Rationality initially described by Simon [2], [3] , discussed and expanded by Petracca [4], (iii) the Theory of Real Competition by Shaikh [5], which is based on Karl Marx’s Theory of Competition, and (iv) the Post-Keynesian theoretical foundations of agent-based modelling by Lavoie [6].

Framing the discussion and using the residential sector as an example

The current path of anthropogenic greenhouse gas emissions will lead to dangerous climate change this century. Despite government efforts, existing global pledged contributions to emissions reduction are still insufficient to meet Paris Agreement mitigation objectives. The adoption of new low-carbon and clean energy technologies will play a key role in limiting the global average temperature increase to 1.5 °C, significantly reducing emissions from fossil fuel consumption. Disrupting the current emissions path will require high capital investment to accelerate the phase-out of fossil fuels. Unfortunately, countries are still pursuing a climate, political, and economic agenda on the use of natural resources leading towards disastrous 2.7 degrees of global heating in the coming decades. This temperature increase scenario will affect people, species, and crops, impacting the whole of society.

The scientific community across the world has been working on understanding the impact of energy consumption on climate change and vice versa. The energy consumed in residential buildings particularly comprises between 30% and 40% of overall energy consumption of a country or region depending on the level of socioeconomic development, geographical features, cultural aspects, among others. Heating and cooling can represent up to 80% of the total energy consumed in houses. Related global greenhouse gas emissions represent 17% of global emissions. Thus, the assessment of the sustainable transition of this sector is key to meet mid-century net-zero emission targets.

Models lack of representing the human dimension

Integrated Assessment Models have emerged to help the research community and stakeholders to build self-consistent scenarios and simulate the complex transition of the global climate-energy-economy system. However, developing climate-energy-economy models for the long-term sustainable transition of the residential sector is still a classic case of all the challenges of modelling. Four challenges are still pending to be addressed in contemporaneous models; (i) capturing the spatio-temporal dimension; (ii) addressing transparency and reproducibility; (iii) capturing complexity across sectors; and (iv) integrating the human dimension. Most conventional models disregard these challenges. We intend to address these challenges in this research.

Key assumption in traditional models

At the basic level of most climate-energy-economy models, a main assumption rules input treatment, calculations, and analysis of results. Millions of consumers are deliberately represented as a single agent that takes prices as given, making rational choices with perfect knowledge of the market under rational expectations to maximize welfare, subject only to budget constraints [7], also called a hyperrational representative agent [5]. To overcome the limitations of representative homogenous hyper-rational agents in traditional climate-energy-economy models – so-called the mainstream – the representation of the human dimension requires the use of empirical, historical, and analytical data along with the appropriate theoretical foundations.

What is GeoAI-ML, Big Data Analytics and Agent-based Modelling

Tools such as Geospatial Artificial Intelligence (geoAI), Machine Learning (ML), Geospatial big data analytics (combination of Geographical Information Systems, GIS, and Big Data Analytics) and agent-based modelling (ABM) present a potential opportunity to introduce the human dimension into the analysis in a more realistic manner. These tools can capture the complexities of heterogeneous shaping structures and the diverse shaping attributes of agents that evolve in space and time, which are driven by bounded rational expectations and exogenous factors. These complexities do not always allow agents to maximise their decisions (e.g., budget or economic expectations), however, complexities representation presents an opportunity of more realistic assessments.

The framework to represent human dimension

This research aims to provide a systematic framework of geospatial agent-based modelling, with a spatially-resolved and temporally-explicit agent definition, using historical, empirical, and analytical high-resolution gridded data and non-gridded data. No such framework currently exists in the literature for climate change mitigation analysis.

The methodology applies a combined geospatial data-driven, agent-based, technology-rich, bottom-up approach for capturing the human dimension in climate-energy-economy to model the decarbonisation of the residential sector globally. Fundamentally, the methodology starts with collecting and handling data, and ends with the application of the MUSE (ModUlar energy system Simulation Environment), ResidentiAl Spatially-resolved and temporal-explicit Agents (RASA) model. The MUSE-RASA model uses geospatial big data analytics to define eight agent-based scenarios to explore long-term climate-energy-economy transition pathways towards the net-zero emission targets by mid-century.

Concluding remarks

Results show the relevance of representing agent disaggregation capturing a range of attributes when assessing the long-term sustainable transition of the residential sector globally. The agent heterogeneity, diversity, evolution, decision-making process, and exogenous constraints have been captured to define twenty agents, disaggregated in 28 regions of the world. In 8 systematically defined scenarios, for each agent and region, five outputs of the MUSE-RASA model have been produced: service demand with a focus on space heating, heating supply by technologies, fuel and electricity consumption, global emissions, and transition costs. Recommendations for energy policy design have been described with reference to four aspects: household budget limitations, quantification of the total demand, carbon tax schemes and heat density restriction. Considering these aspects in policy design will introduce realism in the evidence base, thus supporting most robust policy design.

Takeaways

More realistic assessment requires theoretical foundations and tools beyond traditional ones in the mainstream. Our research explores the combination of geoAI-ML, Big Data Analytics with ABM under alternative theoretical foundations from the heterodox school of economic thought where consumers are heterogenous, diverse, evolve in space and time, and make decision under exogenous constraints.

The provided datasets are spatially resolved and temporarily explicit, which also serve to capture the spatiotemporal dimensions in global model simulations. A range of agents are systematically defined. It is suggested that these datasets be used as inputs in future research on the decarbonisation of the energy system when considering the human dimension for more realistic results.

Decision-makers, policy-makers, firms, civil society, social organisations, and researchers can identify four potential applications of these datasets in the context of assessing climate-energy-economy transition paths: (i) Considering consumer budget limitations in the assessment of the energy transition; (ii) Targeting consumers under specific socioeconomic characteristics that drive the energy transition; (iii) Defining carbon tax schemes implementation, avoiding the regressive impact on poorer households; and (iv) Research and development prioritisation based on heat density demand as an assessment metric to analyse the technical feasibility of low-carbon solutions.

For further reading, please explore the complete study here:

Diego Moya, Dennis Copara, Alexis Olivo, Christian Castro, Sara Giarola, & Adam Hawkes. MUSE-RASA captures human dimension in climate-energy-economic models via global geoAI-ML agent datasets. Scientific Data, 2023, https://doi.org/10.1038/s41597-023-02529-w

References

[1]        A. Crooks, N. Malleson, E. Manley, and A. Heppenstall, Agent-based modelling and geographical information systems: a practical primer. Sage, 2018.

[2]        H. Simon, "A Behavioral Model of Rational Choice," The Quarterly Journal of Economics, vol. 69, no. 1, pp. 99-118, 1955, doi: 10.2307/1884852.

[3]        H. A. Simon, "Bounded Rationality," in Utility and Probability, J. Eatwell, M. Milgate, and P. Newman Eds. London: Palgrave Macmillan UK, 1990, pp. 15-18.

[4]        E. Petracca, "Simulating Marx: Herbert A. Simon's cognitivist approach to dialectical materialism," History of the Human Sciences, p. 09526951211031143, 2021.

[5]        A. Shaikh, Capitalism: Competition, conflict, crises. Oxford University Press, 2016.

[6]        M. Lavoie, Post-Keynesian economics: new foundations. Edward Elgar Publishing, 2014.

[7]        A. Nikas, H. Doukas, and A. Papandreou, "A detailed overview and consistent classification of climate-economy models," Understanding risks and uncertainties in energy and climate policy, pp. 1-54, 2019.

[8]        D. Moya, C. Castro, D. Copara, A. Olivo, S. Giarola, and A. Hawkes, "MUSE-RASA captures human dimension in climate-energy-economic models via global geospatial agent datasets using AI-ML," Figshare Data Repository, 2023, doi: https://figshare.com/s/734d5c001df8d7d3cb14