Banking Sector Risk in the Aftermath of Climate Change and Environmental-related Natural Disasters

This cross-country evaluation estimates the aggregate impact of natural disasters on banking stability and finds persistent negative effects on loan performance and other key variables. The authors call for financial regulators to implement policies that improve climate risk assessments. They also say that more granular data would improve risk monitoring and better safeguard the solvency and profitability of financial institutions.

The research was produced by three World Bank financial economists – Owen Nie, Martijn Regelink and Dieter Wang – and the results show that natural disasters pose risks to financial system stability, for example through non-performing loans (NPL) and reduced asset values.

Disaster episodes increase the ratio of system-wide non-performing loans by between 0.5% and 0.6% two-to-three years after a disaster. The effects are found across multiple regions, for most income groups and after four different types of disaster. The ratio reaches 0.72% in high-income countries after three years and 5.8% in low-income countries after one year.

This NPL ratio also increased to 2.3% after two years in the paper’s study of the effect of typhoons on the Philippines, which suggests that aggregate analysis may result in an underestimation of costs.

The authors used a large cross-country panel data set containing 184 countries from 2018-2019. This allowed them to generate a generic estimate using the largest possible sample. However, a significant limitation of this approach identified by the authors is the lack of forward-looking data, which may result in a further underestimation of risk by neglecting non-linear impacts.

Based on the findings, the authors urge financial authorities to improve the monitoring and mitigation of climate risk. In particular, they recommend policies that:

1. proactively fill data gaps;
2. require detailed assessment of risks, especially diverse risk channels;
3. integrate the paper’s findings by using them to guide model parameters in stress testing;
4. tackle methodological constraints (for instance, by considering forward-looking, non-linear and indirect impact channels).

The authors say that future research should use a global sample with balance sheet-level loan data and geographical NPL data, to allow for “more detailed exploration of [the] particular transmission channels” at work. World Bank research already undertaken for the Latin American and Caribbean region by Miguel Liriano is highlighted as promising in this regard.