Sustainable investing under uncertainty: A dual-criterion probabilistic framework

Abstract This study proposes an ambiguity-aware framework for optimal asset allocation under uncertainty in firms’ ESG scores. Unlike traditional sustainable investing models that assume fully reliable ESG information, the proposed approach explicitly accounts for imperfect and noisy ESG signals by jointly integrating financial and sustainability criteria in a probabilistic setting. Portfolio weights are modeled as random variables, and low informativeness in ESG scores, measured through a signal-to-noise ratio, is penalized by reallocating assets toward financially robust investments when ESG information is weak or ambiguous. Empirical results show that the proposed framework promotes greater diversification and delivers enhanced robustness to parameter uncertainty, exhibiting lower sensitivity to model assumptions and more stable out-of-sample performance compared to standard expected-utility-based ESG portfolios.