Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Exploring Equity of Climate Policies using Multi-Agent Multi-Objective Reinforcement Learning

Palok Biswas, Zuzanna Osika, Isidoro Tamassia, Adit Whorra, Jazmin Zatarain-Salazar, Jan Kwakkel, Frans A. Oliehoek, Pradeep K. Murukannaiah

TL;DR

The paper tackles inequities in climate policy guidance by integrating IAMs with Multi-Objective Multi-Agent Reinforcement Learning (MOMARL) in the Justice framework. By coupling RICE50+ components with the FAIR climate emulator and modeling 12 macro-regions as collaborating agents, Justice discovers Pareto-optimal policy sets across climate and economic objectives under SSP-RCP scenarios. Using MOMAPPO and a suite of convergence and equity metrics, the authors show Justice yields a broader, more equitable set of trade-offs than traditional single-objective IAMs like RICE50+, including a Compromise policy that mitigates inequality relative to the baseline. The open-source nature and multi-objective, multi-agent perspective offer a robust tool for policy deliberation and a benchmark for future MOMARL research in climate economics.

Abstract

Addressing climate change requires coordinated policy efforts of nations worldwide. These efforts are informed by scientific reports, which rely in part on Integrated Assessment Models (IAMs), prominent tools used to assess the economic impacts of climate policies. However, traditional IAMs optimize policies based on a single objective, limiting their ability to capture the trade-offs among economic growth, temperature goals, and climate justice. As a result, policy recommendations have been criticized for perpetuating inequalities, fueling disagreements during policy negotiations. We introduce Justice, the first framework integrating IAM with Multi-Objective Multi-Agent Reinforcement Learning (MOMARL). By incorporating multiple objectives, Justice generates policy recommendations that shed light on equity while balancing climate and economic goals. Further, using multiple agents can provide a realistic representation of the interactions among the diverse policy actors. We identify equitable Pareto-optimal policies using our framework, which facilitates deliberative decision-making by presenting policymakers with the inherent trade-offs in climate and economic policy.

Exploring Equity of Climate Policies using Multi-Agent Multi-Objective Reinforcement Learning

TL;DR

The paper tackles inequities in climate policy guidance by integrating IAMs with Multi-Objective Multi-Agent Reinforcement Learning (MOMARL) in the Justice framework. By coupling RICE50+ components with the FAIR climate emulator and modeling 12 macro-regions as collaborating agents, Justice discovers Pareto-optimal policy sets across climate and economic objectives under SSP-RCP scenarios. Using MOMAPPO and a suite of convergence and equity metrics, the authors show Justice yields a broader, more equitable set of trade-offs than traditional single-objective IAMs like RICE50+, including a Compromise policy that mitigates inequality relative to the baseline. The open-source nature and multi-objective, multi-agent perspective offer a robust tool for policy deliberation and a benchmark for future MOMARL research in climate economics.

Abstract

Addressing climate change requires coordinated policy efforts of nations worldwide. These efforts are informed by scientific reports, which rely in part on Integrated Assessment Models (IAMs), prominent tools used to assess the economic impacts of climate policies. However, traditional IAMs optimize policies based on a single objective, limiting their ability to capture the trade-offs among economic growth, temperature goals, and climate justice. As a result, policy recommendations have been criticized for perpetuating inequalities, fueling disagreements during policy negotiations. We introduce Justice, the first framework integrating IAM with Multi-Objective Multi-Agent Reinforcement Learning (MOMARL). By incorporating multiple objectives, Justice generates policy recommendations that shed light on equity while balancing climate and economic goals. Further, using multiple agents can provide a realistic representation of the interactions among the diverse policy actors. We identify equitable Pareto-optimal policies using our framework, which facilitates deliberative decision-making by presenting policymakers with the inherent trade-offs in climate and economic policy.
Paper Structure (19 sections, 2 equations, 5 figures)

This paper contains 19 sections, 2 equations, 5 figures.

Table of Contents

  1. Introduction
  2. Background
  3. Integrated Assessment Models
  4. Multi-Objective Multi-Agent RL
  5. Solution Set Concept
  6. The Justice Model
  7. High-Level Outline
  8. Multi-Agent Multi-Objective Formulation
  9. Experimental Setup
  10. Performance Indicators
  11. RICE50+ Comparison
  12. Results
  13. Convergence
  14. Solutions
  15. Comparison of Key Solutions
  16. ...and 4 more sections

Figures (5)

  • Figure 1: Overview of Justice. The main outputs of the model have been highlighted with red arrows.
  • Figure 2: Mean hypervolume and expected utility over training steps (shaded area represents standard deviation).
  • Figure 3: Pareto set of policies obtained by Justice (across 10 random seeds) with the RICE50+ polict for comparison. Arrows indicate the direction of preference for the objectives.
  • Figure 4: Performance of selected Justice policies over time (years on x-axis) for three important indicators: Global Net Economic Output, Temperature, and Cumulative Abated Emissions. The indicators show mean values and standard deviation (over 10 seeds) for selected Justice and RICE50+ policies over time.
  • Figure 5: GINI Index of Emissions Over Time for 12 Regions (years on x-axis).