Financial stability in a changing environment

This thesis focuses on improving the understanding of the effects of climate and pandemic-related shocks on financial sector variables. Specifically, it aims to contribute to the development of climate risk stress tests, which translate initial climate shock parameters into variables that are relevant to assess financial institutions, such as solvency and liquidity metrics. The main objective in this thesis is therefore to investigate what the relationship is between climate-related shock variables and financial variables, with a focus on financial asset value and financial institution solvency.

In the second chapter we provide a conceptual review of climate risk stress testing. In almost all cases, Climate Risk Stress Testing (CRST) combines climate, economic, and financial modelling, but not always through the macro-financial approach traditionally employed by central banks and supervisory authorities. We identify six types of climate shocks and four approaches to CRST (i.e., macro-financial, micro-financial, non-structural, and disaster risk). Findings include that existing CRST exercises may underestimate potential system-wide financial losses, due to their limited scope (e.g., not including all asset classes), incomplete modelling (e.g., lack of feedback effects), and a strong reliance on historically established relationships.

Based on the review chapter we identify a novel, micro-financial, approach to assess the effects of carbon tax scenarios on the market value of equity and debt. The third chapter develops this micro-financial approach, based on structural modelling of credit losses in different carbon tax scenarios. We develop a more tractable finance (valuation) approach at the industry-level and use a Merton contingent claims model to assess the impact of a carbon tax shock on the market value of equity and debt instruments. This model is calibrated using detailed firm level vulnerability data and apply the model to 2-digit sectoral exposures of Dutch banks. Findings include declines in the market value of banks’ assets of 3-14% of core capital for a €100 carbon tax shock, increasing to 9-32% for a €200 carbon tax shock.

The fourth chapter in this thesis builds on the structural credit modelling developed in the third chapter, but instead of linking the model to climate-related scenarios we employ the modelling to assess the economic shock related to the Covid-19 pandemic. Using a Merton contingent claims framework, the chapter develops a real time market-based approach to assess expected loan losses and apply it to euro area banks’ during the Covid-19 crisis. Although market-based indicators have improved considerably after an initial sharp downturn, they still provide warning signals for a range of (sub)sectors. During the market low point, implied losses on corporate loans amounted to 16-26% of banks’ capital. The chapter also uncovers a substantial role for monetary policy (lower discount rates) in the subsequent stock market recovery.

The fifth and last chapter develops an empirical model to assess of the impact of climate-related disasters on loan portfolios. We empirically investigate the local effects (direct and indirect effects in regions where the disaster took place) and non-local effects (indirect effects in other regions in the same country) of climate change-related disasters on non-performing loans (NPLs) in Latin America. In our baseline specification we find that NPLs as a fraction of total loans increase with 1.1-1.4 percentage points (local effect) and 0.9-1.0 percentage points (non-local effect). For national banking systems as a whole, findings suggest that NPLs on average increase between 17-46 percent after a severe disaster.