The severe economic impact of the COVID-19 pandemic could threaten financial stability. Since accounting-based methods report loan losses with a delay, this column adopts a real-time, market-based assessment of the impact on corporate loan portfolios. Using European stock market data, it estimates that the market-implied losses for euro area banks could reach over €1 trillion, or, depending on the scenario, 7-43% of available bank capital.
The latest estimates of the severity of the economic downturn due to the COVID-19 pandemic suggest a contraction in global GDP of no less than 6% in 2020. The looming question is whether current bank capital buffers are sufficient to cover potential credit losses. Assessing the gravity of the COVID-19 threat to financial stability is challenging, since banks’ accounting-based loan loss provisions are sluggish (Laeven and Majnoni 2003, Benston and Wall 2005). Market-based methods provide a real-time assessment, but banks loans are not traded. The contingent claims model developed by Nobel laureate Robert Merton allows us to use stock market data to estimate the impact on the value of corporate debt by industry.
Using different scenarios for future volatility and incurred losses in case of default, we estimate that the market-implied losses for euro area banks range from 7% to 43% of available bank capital and reserves. These findings suggest that the ECB as supervisor should pro-actively manage the impact of COVID-19 on banks.
We start with a short overview of our method to estimate market-based expected losses to banks’ credit portfolios. A general definition for expected loss (EL) is:
In this formula, PD is the probability of default, LGD the loss given default (i.e. one minus the recovery rate), and EAD the exposure at default.
We use the Merton (1974) structural credit risk model to determine the change in the probability of default (PD) of individual firms that is implied by the observed changes in the value of that firms’ stock. Merton’s critical insight is that equity can be viewed as a residual claim on assets after the debt has been repaid. This implies that holders of equity hold a de facto put option on the assets of the firm, limiting their losses in case the assets of the firm become less valuable than the outstanding debt. The stock price thus conveys information about investors’ expectation of a firm defaulting (Reinders et al. 2020a).
Calibration and data
To calibrate the Merton model, we use a set of 3,867 public firms in the euro area obtained from the Bureau van Dijk Orbis database (Reinders et al. 2020b). Figure 1 shows that the stock market initially rapidly declined as a result of COVID-19, but partially recovered after the announcement of the first stimulus packages over the course of March 2020. We therefore measure equity volatility and changes in the stock price of firms not only between 1 January 2020 and 18 March 2020 (market low point), but also between 1 January 2020 and 20 April 2020 (post-stimulus shock).
Figure 1 STOXX Europe 600 price index
Notes: This figure shows the STOXX Europe 600 price index between November 1, 2019 and April 20, 2020. Data are obtained from Thomsom Reuters Datastream.
Both future volatility and post-shock loss given default (LGD) are unknown. We therefore carry out our stress test based on four scenarios, allowing for a range of uncertainty in both parameters.
- For the standard deviation of equity, we use either a constant volatility or a volatility that structurally doubles as a result of the COVID-19 shock.
- For the post-shock LGD, we use two further scenarios consisting of a lower bound and upper bound estimate of the LGD.
- For the lower bound, we assume that the LGD remains constant over time at 45% (a conservative estimate of past losses).
- For the upper bound, we assume that the average LGD during our three-year time horizon increases by 15 percentage points to 60%.
For the exposure at default (EAD), we use loan exposure data obtained from the European Central Bank (ECB) data warehouse. This dataset provides the aggregated exposure of euro area banks to industries (according to the NACE-1 industry classification) in their loan portfolios. The second column of Table 1 shows these exposures for the NACE-1 industries. Total corporate loan exposures for all euro area banks are €4.46 trillion.
Stress test results
Table 1 reports the estimated losses by 20 April 2020 (post-stimulus) for each of our four scenarios. At the lower bound (first loss column), we estimate losses of €179.34 billion. This is 4% of the corporate loan portfolio and 7% of bank capital. At the upper bound, losses amount to €1.10 trillion, which is 25% of corporate loans and 43% of bank capital. The hardest hit industry sectors are manufacturing (€192 billion), trade (€126 billion), and services (€373 billion).
Table 1 COVID-19 stress test of euro area banks’ corporate credit portfolios (by industry)
Our results are important for both financial supervisors and financial institutions. Supervisors can use our findings to compare market-based expected losses to currently announced loss-provisions by euro area banks. In some jurisdictions, such asset quality reviews may reveal gaps between the more accounting-based supervisory practices and market expectations; these could therefore be part of the rationale to implement additional measures to safeguard financial stability. Our results estimating potential declines in bank capital of up to 43% support the ECB’s ban on dividend payments and share buy-backs during the COVID-19 pandemic. For banks, our results can provide a benchmark to their own loan portfolios to estimate a plausible range of expected losses. The impact on individual banks may vary, amongst others because the industry distribution of their loan portfolio differs.
Our estimates may be on the conservative side for two reasons. First, we estimate industry impact based on listed firms that are typically larger than many firms in banks’ corporate loan portfolios and hence may be able to withstand shocks better. Second, several market commentators (e.g. Financial Times 2020a, 2020b) have questioned whether the stock market recovery in late March and April can be justified in light of the huge economic repercussions of the COVID-19 pandemic.
Benston, G J and L D Wall (2005), “How should banks account for loan losses?” Journal of Accounting and Public Policy 24: 81-100.
Financial Times (2020a), “Citi warns markets are out of step with grim reality”, 31 May.
Financial Times (2020b), “A market rally built on shaky foundations”, 9 June.
Laeven, L and G Majnoni (2003), “Loan loss provisioning and economic slowdowns: too much, too late?” Journal of Financial Intermediation 12: 178-197.
Merton, R C (1974), “On the pricing of corporate debt: The risk structure of interest rates”, Journal of Finance 29: 449–470.
Reinders, H J, D Schoenmaker and M A van Dijk (2020a), “A finance approach to climate stress testing”, CEPR Discussion Paper, DP14609.
Reinders, H J, D Schoenmaker and M A van Dijk (2020b), “Is COVID-19 a threat to financial stability in Europe?”, CEPR Discussion Paper, DP14922.