Diversification effects of ESG penalties in sustainable mean–variance portfolios

We study how different ways of incorporating environmental, social, and governance (ESG) objectives into mean–variance portfolio choice affect diversification. A commonly used linear ESG adjustment tilts portfolios toward high-scoring assets but, under standard long-only constraints, systematically increases concentration as the impact of the ESG adjustment intensifies. We contrast this specification with a nonlinear ESG adjustment arising from a robust mean–variance framework, in which ESG characteristics shape uncertainty about expected returns. This approach tends to preserve diversification while still producing meaningful ESG tilts. Analytical results and numerical illustrations show that modeling choices for ESG adjustments are crucial for stable and well-diversified portfolio construction. Our findings have implications for practical ESG portfolio construction, while being derived from a stylized mean–variance setting.