EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Carbon–Cost–Resilience–Fairness Trade-offs in Green Public Procurement: A Multi-Agent Reinforcement Learning Approach

Colin Hayes

International Journal of Business and Social Science Studies · 10.09.2026 · 6 (2) · 172-183

DOI: 10.66795/ijbsss70 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Business and Social Science Studies, online erschienen am 10.09.2026. Verfasst von Colin Hayes. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Green public procurement (GPP) increasingly requires public buyers to reconcile environmental performance with fiscal efficiency, continuity of supply, and access for smaller suppliers. Existing learning-based green supply-chain models typically emphasize carbon, cost, and service resilience, while distributive fairness remains external to the operational policy. This paper formulates procurement as a constrained multi-agent reinforcement-learning problem in which one public coordinator and heterogeneous supplier agents adapt sourcing and commitment decisions under stochastic demand surges and capacity disruptions. We introduce Fair-MARL, a lightweight tabular method that combines an adaptive Lagrange multiplier for a minimum small-supplier participation floor, disruption-tail oversampling, and shock-weighted service rewards. The experiment intentionally uses a transparent synthetic simulator because directly observed transaction-level supplier embodied-carbon labels are not available from the public sources considered; consequently, no real supplier outcome is imputed or presented as observed data. Across five random seeds, 150 training episodes per seed, and 60 held-out episodes per method and seed, Fair-MARL obtains a 37.7% small-supplier share and a mean fairness violation of 0.0002 while maintaining 99.13% mean service and 0.973 resilience. Relative to Static-Cost, normalized carbon falls by 1.5% at a 1.4% cost increase. Relative to unconstrained MARL, the fairness violation falls by 99.4%, with only a -0.25% change in resilience. The study shows that fairness can be treated as a controllable operational constraint rather than an ex-post audit statistic, while making the carbon–cost–resilience trade-off explicit.

Zitieren

APA

Hayes, C. (2026). Carbon–Cost–Resilience–Fairness Trade-offs in Green Public Procurement: A Multi-Agent Reinforcement Learning Approach. International Journal of Business and Social Science Studies, 6(2), 172-183. https://doi.org/10.66795/ijbsss70

Vancouver

Hayes C. Carbon–Cost–Resilience–Fairness Trade-offs in Green Public Procurement: A Multi-Agent Reinforcement Learning Approach. International Journal of Business and Social Science Studies. 2026;6(2):172-183. doi:10.66795/ijbsss70

BibTeX
@article{Hayes2026,
  author = {Hayes, Colin},
  title = {Carbon–Cost–Resilience–Fairness Trade-offs in Green Public Procurement: A Multi-Agent Reinforcement Learning Approach},
  journal = {International Journal of Business and Social Science Studies},
  year = {2026},
  volume = {6},
  number = {2},
  pages = {172--183},
  doi = {10.66795/ijbsss70}
}
Kostenlos starten →