based on M. Gasparini et al. Nature Climate Change https://doi.org/10.1038/s41558-024-01972-w (2024).

The policy problem

As the urgency for climate action increases, financial regulators, supervisors and central banks are facing growing calls to bring financial regulations in line with a transition to net-zero carbon emissions. Various proposals have been made in the literature, such as green quantitative easing, direct central bank financing, lending quotas, including environmental, social and governance (ESG) factors in asset eligibility criteria, and differentiated prudential regulation. Financial regulators and supervisors are currently holding back on these proposals. However, the existing structure of some global financial regulations may inadvertently favour investments in carbon-intensive activities. This could negatively impact the speed of the transition to net-zero carbon emissions and increase the exposure of the financial system to climate-related risks in the long term.

Fig. 1: Change in provision coverage ratio for the 59 largest European banks.
figure 1

a, Absolute percentage increase in the provision coverage ratio (PCR) following a divestment from high-carbon assets and corresponding reinvestment in low-carbon assets, maintaining a constant level of outstanding loans by bank. Colours represent the country of the banks’ headquarters. The increase in the PCR represents the difference between the PCR required for low-carbon assets as opposed to high-carbon assets for each bank in our sample. Horizontal line represents average in basis points (bps). b, Absolute percentage increase in PCR (blue bars) ranked by gross loan exposure (red line), from largest (left) to smallest (right). Figure adapted with permission from M. Gasparini et al. Nat. Clim. Change https://doi.org/10.1038/s41558-024-01972-w (2024), Springer Nature Ltd.

The findings

We find that under the current regulations, if 59 of the largest banks in the European Union (EU) were to divest from high-carbon sectors and reinvest in other activities, they would record, on average, losses equivalent to about 15% of their previous 5 years’ profits. We show that this is due to the increase in loan loss provisions required to cover the higher estimated risk of low-carbon-intensity activities, compared with high-carbon-intensity activities. We show that the average estimate of risk (expressed in terms of the ratio between loan loss reserves and outstanding loans) among EU banks is lower for carbon-intensive activities as opposed to low-carbon activities (1.8% and 3.4%, respectively, in 2021). This is likely to be due to the backward-looking structure of model-based risk estimates that fails to adequately incorporate recent policy changes, the declining costs of low-carbon technologies and other ongoing factors. We argue that this creates disincentives for banks to invest in new low-carbon assets and exposes them to future risks from high-carbon assets.

The study

We analyse EU banks’ portfolios and associated accounting requirements (IFRS9) using data from the European Banking Authority. Specifically, we focus on the accounting rules of banks that rely on model-based risk assessments, similar to other financial regulations such as capital requirements. We classify investments into high-carbon and low-carbon emission intensity, leveraging the Climate Policy Relevant Sectors (CPRS) classification, and test the robustness of our findings with various classifications. We first investigate the relationship between the accounting requirements’ model-based risk estimates and the classification of high- or low-carbon intensity. We then use the accounting relationship between model-based risk estimates, loan loss provisions and profits to assess the potential losses emerging from an active investment strategy to shift from high- to low-carbon-intensity activities by banks. Finally, we provide evidence across various risk models of a negative correlation between risk estimates and emission intensity and analyse the possible financial drivers of such differences.