research


G. Crippa. The Journal of Impact and ESG Investing 5.4 (2025): 52-96.

M. Alaluf, G. Crippa, S. Geng, Z. Jing, N. Krishnan, S. Kulkarni, W. Navarro, R. Sircar, J. Rang.

Risk & Decision Analysis



Working Papers


From Constraints to Rational Preferences [link]

Under Review Management Science

Standard approaches to portfolio optimization often rely on constraints to proxy investor preferences, assuming equivalence with utility-based formulations. This paper challenges that assumption by developing a general framework that characterizes conditions under which constraints can rationalize preferences. We derive necessary and sufficient conditions for equivalence, revealing that constraints can misrepresent investor behavior or overlook heterogeneity. To overcome these limitations, we propose a multi-objective optimization framework that allows trade-offs to be determined endogenously rather than fixed ex ante. We extend the classical mean-variance model to sustainable investing, and derive closed-form equilibrium strategies in complete markets. Using simulations and U.S. equity data, we construct a three-dimensional mean-variance-sustainability efficient frontier  that captures the trade-offs among the different objectives. Our framework offers a flexible foundation for portfolio design and sustainable investing, with practical relevance for asset managers and policymakers.


Corporate Omissions: Correcting the Bias in Carbon Reporting. Joint with R. Rigobon, F. Berg [link]

Corporate carbon disclosure is critical for accountability and effective policy design, yet emissions data are often incomplete and selectively reported. This paper addresses the problem of missing Scope 1 emissions among U.S. public companies by developing an imputation framework under a missing not at random (MNAR) assumption. Using firm-level data from 2012 to 2023, we combine outcome modeling, response mechanism estimation, and exponential tilting to account for strategic non-disclosure. Our results show that non-reporting firms tend to have higher emissions, with MNAR-corrected estimates increasing average Scope 1 emissions by 15–19\% relative to unadjusted values. Sectoral and size-based patterns in disclosure further support the endogeneity of missingness. We extend the analysis to estimate carbon damages using social cost of carbon benchmarks, finding that under-reporting leads to substantial downward bias in aggregate impact estimates. These findings highlight the importance of modeling disclosure behavior in emissions accounting and provide a robust framework for improving climate-related financial analysis.


Informations & Factors 

ESG Mutual Funds: Marketing or Impact? Joint with M. Gasparini, C. Foroni


Preprints


Machine Learning Reveals Intrinsic Determinants of siRNA Efficacy. Joint with C. Mandelli