A bilevel approach to ESG multi-portfolio selection

We rely on bilevel programming to model the problem of financial service providers that, in order to meet stakeholders' demands and regulatory requirements, aim at incentivizing accounts' holders to construct ESG-oriented portfolios so that the overall ESG impact of the firm is optimized, while the preferences of accounts' owners are still satisfied. We analyze this complicated framework from a theoretical point of view and identify sufficient conditions that make it numerically tractable via a novel, specifically tailored algorithm, whose convergence properties are studied. Numerical testing on real-world data confirms the theoretical insights and shows that our model can be solved even when dealing with considerable problem sizes.

Automated summary

In this paper, we investigate the problem faced by financial service providers in incentivizing account holders to construct ESG-oriented portfolios. By using bilevel programming, we propose a novel algorithm that can optimize the overall ESG impact of the firm while satisfying the preferences of account owners and meeting stakeholders' demands and regulatory requirements. Theoretical analysis and numerical testing on real-world data demonstrate the effectiveness of our approach even for large-scale problems.