doi.org/10.71734/DP‑2026‑33

As the financial sector grapples with climate-related risks, central banks and regulators are increasingly using climate stress tests (CSTs) to assess banks’ resilience and encourage climate-sensitive lending practices. This study investigates whether the European Central Bank’s (ECB) 2021–22 CST influenced how euro area banks price loans based on borrowers’ carbon intensity. The findings reveal that while the CST did not uniformly alter loan pricing, it triggered significant portfolio reallocation and differential pricing strategies, shedding light on how supervisory tools can shape the financial sector’s role in the low-carbon transition.

How do climate stress tests influence banks’ lending behavior?

Introduction

Banks are pivotal in financing the transition to a low-carbon economy, but they also face significant risks from climate change. These include transition risks, which arise from policy changes and shifts in market preferences, and physical risks, stemming from climate-related disasters. To address these challenges, the ECB conducted its first EU-wide CST in 2021–22, aiming to assess banks’ preparedness for climate risks and encourage better integration of these risks into their governance and lending practices. This paper examines whether the CST influenced banks’ loan pricing and credit allocation, particularly with respect to borrowers’ greenhouse gas (GHG) emissions intensity.

Methodology

The authors use granular loan-level data from the Eurosystem’s AnaCredit database, which provides detailed information on bank loans to non-financial corporations. Borrower-level GHG intensity is estimated using sectoral emissions data and firm turnover, allowing the study to capture transition risk exposure across a broad range of firms, including small and medium-sized enterprises. The analysis employs a difference-in-differences (DiD) framework, comparing changes in the sensitivity of loan spreads to borrower carbon intensity for banks subject to the CST (treated banks) versus those not subject to it (control banks). The study also decomposes the observed effects into intensive (within-relationship repricing) and extensive (portfolio reallocation and entry/exit) margins.

Theoretical Framework

The paper builds on a theoretical model in which a CST acts as an information-revelation shock, shedding light on transition risks and increasing supervisory scrutiny. This shock alters banks’ perceived risk distribution, influencing their lending decisions. The model predicts that CSTs can steepen, flatten, or leave unchanged the relationship between loan spreads and borrower carbon intensity, depending on how banks adjust their pricing and portfolio composition in response to the new information.

Key Findings

1. Aggregate Effect on Loan Pricing: On average, the CST did not significantly change the sensitivity of loan spreads to borrower carbon intensity. The post-CST shift in the spread–intensity slope was small and statistically indistinguishable from zero.
2. Portfolio Reallocation: Treated banks reallocated credit away from more carbon-intensive incumbent borrowers, reducing their share of new lending to these firms. This reallocation was particularly pronounced among the most carbon-intensive borrowers.
3. Entry/Exit Pricing: Treated banks charged higher spreads to new borrowers with higher carbon intensity compared to incumbents, about 10 bps more for one SD increase in carbon-intensity. This is evidence of more stringent pricing of transition risks on the extensive margin.
4. Intensive Margin: Within existing bank–borrower relationships, there was little evidence of systematic repricing based on carbon intensity. This suggests that the CST’s impact operated primarily through portfolio composition rather than within-relationship adjustments.
5. No significant Heterogeneity in Treatment Intensity: Among significant institutions (SIs), banks that participated in the more demanding bottom-up module of the CST did not exhibit stronger changes in carbon-sensitive pricing compared to those that only participated in lighter modules. This suggests that the main behavioral adjustments stemmed from CST participation itself, rather than the depth of the exercise.
6. Asymmetric and Nonlinear Effects: Treated banks responded more strongly to decreases in borrower carbon intensity than to increases, and the CST’s impact was more pronounced for borrowers in the upper tail of the emissions distribution.

Conclusion

The ECB’s CST influenced banks’ lending behavior primarily through portfolio reallocation and differential pricing of new relationships, rather than through uniform changes in loan pricing. Treated banks reduced their exposure to carbon-intensive borrowers and applied more stringent pricing to new high-intensity relationships, signaling a shift toward climate-sensitive credit allocation. However, the CST’s impact on within-relationship repricing was limited, highlighting the importance of relationship dynamics and informational frictions in shaping banks’ responses to climate risks.

From a policy perspective, these findings suggest that CSTs can serve as effective tools for encouraging banks to integrate climate risks into their lending decisions, even in the absence of direct capital requirements. By revealing transition risks and increasing supervisory scrutiny, CSTs can drive portfolio adjustments that align with broader climate goals. Future research could explore the real-economy implications of these adjustments, such as their impact on firm-level emissions and investment, as well as the interaction between CSTs and other climate policies.

De Bandt, O., D. Nefzi, J.-G. Werner (2026), Stress testing European banks’ transition risk and lending behavior, Bundesbank Discussion Paper, No 33/2026.