Papers

Banks are the primary source of external funding for small businesses. Integrating sustainability into small business lending can therefore support broader sustainability goals — yet the sustainable finance literature has focused almost entirely on large, listed corporations. This paper assesses the current state of sustainable small business lending practices among German banks through a survey of 62 institutions. Results show that banks are implementing sustainable finance practices more quickly for large corporate clients than for SMEs. They emphasize sustainability risk management over business model transformation and prefer hard, quantifiable sustainability data over soft information gathered through relationship lending, even though both have relevant applications.

ESG ratings from different providers diverge widely — the same firm may be rated highly by MSCI and poorly by S&P. This "aggregate confusion" undermines the informational value of ESG data for investment and capital allocation. The EU Taxonomy, as a harmonized classification system for sustainable activities, could reduce this divergence. Using Taxonomy alignment data from ISS ESG combined with ratings from MSCI, S&P, Refinitiv, and V.E (Moody's), the authors use tobit regressions to show that Taxonomy alignment is significantly positively related to environmental (E) ratings from three of four providers. However, the potential for Taxonomy-driven harmonization has not yet fully materialized.

Current policy discussions assume that better corporate sustainability disclosure leads to better-informed investment decisions. This paper questions that assumption. In a pre-registered survey experiment with 3,437 participants (including representative US and UK samples, students, retail investors, finance professionals, and sustainability experts), participants assessed the "greenness" of firms based on carbon emissions data. Results show a systematic central tendency bias: individuals overestimate the greenness of brown firms and underestimate it for green ones. This attenuates firms' financial incentives to invest in emissions reductions. Some disclosure design features — such as relative benchmarking — significantly improve differentiation ability.

Does uncertainty about future carbon prices affect how equity markets price firm-level carbon exposure? Using a monthly panel of STOXX Europe 600 firms (2013–2022) and the option-implied Carbon VIX (CVIX), this paper shows that the carbon return premium is state-dependent. High-emission firms earn modestly lower returns in low-volatility periods, but a 10-point CVIX increase raises monthly excess returns for top emitters by approximately 0.30 percentage points — sufficient to flip the sign. The result reconciles conflicting evidence in the carbon premium literature and demonstrates that carbon price uncertainty, not just price levels, should be incorporated into asset pricing models and policy design.

Firm-level sustainability data — carbon emissions, biodiversity footprints, water usage — is critically important but largely incomplete. This paper presents a machine learning framework for estimating sustainability metrics from readily available financial data, without domain-specific modeling for each metric. The framework is adapted to handle multiple data types, quantifies prediction uncertainty via conformal prediction, and uses prediction-powered inference to support downstream empirical research. Applied to corporate carbon emissions and waste-water discharge, the results show superior out-of-sample performance relative to regression benchmarks.

Article 2.1c of the Paris Agreement calls for financial flows consistent with low-carbon, climate-resilient development. But what does this mean in practice? Using global institutional ownership data, this paper examines whether Responsible Investors (UN PRI signatories, who hold roughly one-third of global equity) contribute to real-economy decarbonization. Results show that Responsible Investors allocate less capital to high-emission companies and more to already-green ones. Their ownership is significantly associated with firms committing to carbon reduction targets — but not with actual emission reductions. Instead, higher Responsible Investor ownership is associated with improvements in ESG ratings, suggesting a focus on perceived sustainability over fundamental decarbonization.

Climate stress testing — assessing how bank portfolios would fare under severe climate scenarios — is increasingly required by financial supervisors. But existing frameworks face a critical data limitation: most firms, especially SMEs, do not disclose emissions. This paper develops a micro-prudential stress testing framework that uses machine learning to estimate firm-level emissions and predict default probabilities under carbon price shocks. Applying the framework to a EUR 100 per tonne carbon price shock, the authors find that the number of loss-making firms nearly doubles and loan defaults increase by 7.9%. SMEs are disproportionately affected. Prediction uncertainty remains substantial due to limited SME disclosure.

Sudden and irregular climate-related shocks — such as abrupt policy changes, extreme weather events, or rapid technological disruption — pose significant threats to financial system stability that current risk models are ill-equipped to handle. This paper introduces C-STRIKE (Climate Shock Transmission and Risk Inference for Kredit Evaluation), a modular, bottom-up modeling framework designed to simulate how climate shocks transmit through firms' balance sheets and into bank credit portfolios. The framework captures transmission through three layers: direct firm effects, supply chain linkages, and macroeconomic demand feedback. Three case studies are developed: an emissions price shock, a 1-in-100-year flood in Northern Europe, and a European heatwave.