The global transition to net-zero carbon emissions necessitates substantial investment, largely channeled through the financial system. While various policies aim to assess and mitigate climate-related financial risks, the potential negative impact of existing financial regulations on this transition remains understudied. This paper addresses this gap by examining whether widely used model-based risk regulations, specifically financial accounting rules, create disincentives for financial institutions to divest from high-carbon assets and invest in low-carbon alternatives. These regulations, such as capital requirements (Basel III/IV) and accounting rules (IFRS9), rely on statistical models to assess financial risk and influence key financial metrics, management incentives, and resource allocation. The study focuses on loan loss reserves (LLR), a key measure reflecting potential future losses from outstanding loans, and its impact on banks' profitability and investment decisions. The core question is whether the model-based risk assessments embedded in these regulations inadvertently favor high-carbon investments, hindering the green transition. The urgency of climate action and the growing concerns about climate-induced financial instability and stranded assets underscore the importance of this research.

Existing literature highlights the significant risks associated with climate change for the financial system, including the potential for stranded assets and climate-induced financial instability. Several studies have explored policies to catalyze the green transition, such as climate stress testing and climate-related risk disclosure. However, a critical gap in the research is the assessment of whether current financial regulations might inadvertently hinder the shift towards low-carbon investments. This paper builds upon prior work that examines the financial risks of the net-zero transition and the impact of capital requirements on banks' lending behavior, extending the analysis to the specific influence of accounting rules on investment choices in high-carbon versus low-carbon sectors.

The research leverages data from the European Banking Authority (EBA) transparency exercise, which provides loan loss reserves (LLR) and outstanding loans of supervised banks in the EU, categorized by economic sector (NACE rev2 level 1). This data is combined with the results of the EBA risk assessment exercise to classify sectors as 'high-carbon' or 'low-carbon' based on their share of climate policy-relevant sectors (CPRS). The provision coverage ratio (PCR), the ratio of LLR to outstanding loans, serves as a proxy for banks' estimates of expected credit losses. The empirical analysis compares PCRs for high-carbon and low-carbon sectors. To assess the implications of divesting from high-carbon assets, the authors simulate the effect on key financial metrics, including PCR, loan loss provisions (LLP), and net profits, assuming a shift of lending from high-carbon to low-carbon sectors while maintaining constant overall loan amounts. The simulation utilizes accounting relationships among these metrics to determine the potential cost of such a divestment. Additionally, the authors analyze a dataset of oil and gas and renewable energy firms to compare financial ratios commonly used in risk assessment, providing insights into the origins of the observed differences in risk estimates between high- and low-carbon sectors. Robustness checks are conducted to address potential limitations in sector classification and data coverage.

The analysis reveals a significantly lower average PCR for high-carbon sectors (1.8%) compared to low-carbon sectors (3.4%) in 2021 across 59 major EU banks, representing 93% of total banking exposure. This difference holds consistently across banks of varying sizes and headquarters locations, with the exception of Italy. The simulation of a divestment strategy indicates a substantial impact on banks' financial metrics. A complete shift from high-carbon to low-carbon lending would necessitate a loan-weighted average increase of 35% in LLR across EU banks. This could lead to a more than doubling of provisions for some banks and a significant reduction in net profits, estimated at 15% of the cumulative profits over the past five years on average across the sampled banks, translating to a potential total loss of approximately €28 billion. This significant financial cost could create perverse incentives, disincentivizing banks from divesting from high-carbon assets. The analysis of financial ratios for oil and gas and renewable energy firms further supports the findings, showing historically lower risk estimates for high-carbon sectors due to the backward-looking nature of the models used. Simulations considering a carbon tax or oil reserve write-offs suggest that forward-looking models are necessary to adequately capture the changing risk landscape.

The findings suggest that current model-based financial regulations, particularly accounting rules, might unintentionally hinder the transition to net-zero emissions by creating disincentives for banks to divest from high-carbon sectors. This bias originates from the use of backward-looking risk models that rely on historical data, which may not accurately reflect the future risks associated with high-carbon assets in a rapidly changing climate. The high costs associated with shifting investments from high- to low-carbon assets, as demonstrated by the simulation, highlight a critical policy challenge. While the results are specific to the EU banking sector, the underlying mechanisms and insights likely apply more broadly. This highlights the need for a more forward-looking approach to risk assessment that incorporates scenario analyses considering climate-related risks and the energy transition. The potential for significant financial losses resulting from a rapid shift away from high-carbon assets underscores the importance of addressing the regulatory framework's unintended consequences.

This study reveals a critical unintended consequence of current model-based financial regulations: they may inadvertently hinder the transition to net-zero carbon emissions by disincentivizing banks from divesting from high-carbon assets. The significant financial costs associated with such a shift, as highlighted by the simulation, create a strong disincentive. The backward-looking nature of risk assessment models is identified as a key driver of this bias. This research emphasizes the need for regulatory reforms that incorporate forward-looking climate risk assessments and scenario analyses to ensure that financial regulations effectively support the green transition without creating perverse incentives. Further research should explore the presence of this bias in other model-based regulatory frameworks, such as capital requirements, and investigate the broader systemic implications of these regulatory challenges.

The study's findings are based on data from the EU banking sector and may not be directly generalizable to other regions or financial systems. The sector classification used, while robustly tested, relies on available data and may not perfectly capture the full spectrum of high- and low-carbon activities. The simulation of a divestment strategy assumes a complete and immediate shift in lending, which might not fully reflect the reality of a gradual transition. The analysis focuses primarily on the impact of accounting rules, while other aspects of financial regulations could also contribute to the observed bias. Future studies could investigate these limitations and explore additional factors influencing investment decisions in low-carbon assets.