Exploiting Multi-Design Space for Hedging Purpose: the Case of Climate Risk

Abstract

This paper proposes a novel, data-driven framework for constructing climate risk hedge portfolios by combining a two-step factor-mimicking approach with systematic multi-design optimization. The method first estimates asset-level climate exposures (climate betas) and then translates these betas into portfolio weights. Recognizing that both steps involve multiple design choices—including the decomposition of climate betas into frequency- and horizon-specific components and the selection of investment universes—a third step optimizes across these designs through a validation procedure to select the most robust portfolios. Empirical results, based on over 3,000 models, show that the framework substantially improves out-of-sample hedging performance relative to standard factor-mimicking portfolios and performs comparably to more sophisticated approaches.