Making sense of the comprehensive assessment
The ECB conducted a comprehensive assessment of banks and identified capital shortfalls for 25 banks, totalling €25 billion. In this column, the authors provide a number of benchmark stress tests to estimate capital shortfalls. The analyses suggest possible capital shortfalls between €80 billion and more than €700 billion depending on the model. They find a negative correlation between their benchmark estimates and the regulatory capital shortfall, and a positive one between the benchmarks and the regulatory estimates of losses. This suggests that regulatory stress test outcomes are potentially affected by the discretion of national regulators.
The ECB has finalised its assessment of the largest banks in the Eurozone before it commences their regulatory oversight in November 2014. It has now disclosed its own assessment about the solvency of the banking sector (ECB 2014).
In an earlier piece (Acharya and Steffen 2014), we have estimated capital shortfalls of European banks that are going to be part of the Single Supervisory Mechanism (SSM) using ‘benchmark’ stress tests. We documented that the comprehensive assessment might reveal a substantial lack of capital in many peripheral and core European banks. How do our benchmark capital shortfalls compare to the regulatory shortfall estimates?
The ECB included 130 banks in the comprehensive assessment. Of these, it will eventually supervise 120 banks directly. This set of banks includes 39 publicly listed financial institutions for which supervisory as well as our benchmark stress test data are available. We use balance sheet data from SNL Financial as of 31 December 2013, which is also the starting point of the comprehensive assessment.
Table 1 shows that these banks have €12.5 trillion in total assets and a market capitalisation of €539 billion. Table 1 also provides an overview of the mean regulatory capital ratio Core Equity Tier 1 (C Tier 1) as well as equity/asset and market-to-book ratios. The mean C Tier 1 capital ratio is 11.68%, the mean equity/asset ratio is 5.3% and the market-to-book ratio is 0.84 and well below 1, i.e., markets are substantially discounting banks’ assets and Cyprus, Italy, and Germany lead the table with the banks that show the lowest market-to-book ratios.
Note: This table reports descriptive statistics of the publicly listed banks included in the comprehensive assessment conducted by the ECB in 2014. C Tier 1 is the Core Tier 1 ratio and is Core Tier 1 Capital divided by risk-weighted assets (RWA). Equity/assets is book equity over total assets. Market-to-book is market value over book value of equity. Market cap is the market value of equity measured in million euros. Assets are total assets and measured in million euros. Banks are the number of public banks in each country. All data are as of 31 December 2013 and aggregated at the country level. Ratios are weighted using total assets.
Benchmark stress test results (‘SRISK or capital shortfall in a systemic crisis’)
We assume a systemic financial crisis with a global stock market decline of 40%. SRISK 5.5% VLAB is our measure for a bank’s capital shortfall in this scenario, assuming a 5.5% prudential capital ratio with losses estimated using the VLAB methodology to estimate the downside risk of bank stock returns. While this scenario and the resulting SRISK measure uses market data and market equity (instead of book equity) in determining leverage, the approach is conceptually similar to that of the EU stress tests, which is to estimate losses in a stress scenario and determine the capital shortfall between a prudential capital requirement and the remaining equity after losses.
Supervisory stress test results
We use the following outcomes from the comprehensive assessment to relate to our benchmark stress tests:
- Capital shortfall of banks to a threshold of 5.5% Common Equity Tier 1 (CET1) in the adverse scenario in million euros.
- Three-year cumulative impairment losses on financial and non-financial assets in the banking book (‘loan losses’) – the cumulative impairment losses are measured in the adverse scenario.
- Three-year cumulative losses from the stress in the trading book (‘trading losses’) – the cumulative impairment losses are measured in the adverse scenario.
- Three-year cumulative total losses – total losses are the sum of loan losses and trading losses.
We analyse and compare the benchmark regulatory capital shortfalls along two dimensions: The absolute size of the shortfalls and the rank correlation of banks that incur shortfalls.
The calculation of capital shortfalls considers the losses banks incur in the banking and trading book. The ECB then calculates shortfalls using a regulatory capital ratio (the CET 1 ratio). This ratio incorporates risk-weighted assets in the denominator. Moreover, the numerator is Common Tier 1 capital introduced by the Basel III framework and implemented in the EU in the CRR/CRD IV. These choices are problematic for two reasons:
- The use of risk-weights is questionable as they are based on internal models for banks using the basic or advanced internal ratings based modelling approach (IRB banks).
Even in the standardised approach, risk-weights are not necessarily reflecting the true risk of the banks’ assets, e.g., sovereign debt still has a zero risk-weight.
- Common Tier 1 is a ‘new’ measure of regulatory capital and incorporates a substantial number of transitional arrangements until it is fully implemented, i.e., a number of regulatory deductions from capital are going to be phased-in over time.
Subtracting goodwill and other intangible assets is one example; the treatment of deferred tax assets (DTA) is another example. Recognising these items, however, can be decided by the national competent authorities and thus gives them considerable discretion. This discretion was heavily used in the comprehensive assessment as reported by the ECB.
We thus compare the benchmark stress test results both to the capital shortfalls as calculated by the EBA/ECB and to the actual losses in the banking and trading book under the adverse scenario that forms the basis of these shortfalls. The advantage of using these losses is that they are unaffected by risk-weights or regulatory discretion.
Comprehensive assessment outcomes:
- The regulatory capital shortfall as estimated by the ECB is €19.8 billion. Public banks thus account for more than 80% of the total capital shortfall reported by the ECB (€24.6 billion) (see Table 2).
- Losses in the banking book (loan losses) and in the trading book (trading losses) are large and amount to €275 billion and €37 billion, respectively (Table 2).
Note: This table reports capital shortfalls under our benchmark stress test as well as based on the supervisory stress test as disclosed by the ECB. SRISK 5.5% VLAB is calculated assuming a 5.5% prudential capital ratio (which is the measure available on the NYU Stern Volatility Lab website) as of 31 December 2013. Shortfall 5.5% CET 1 is the shortfall to the 5.5% common equity Tier 1 capital ratio in the adverse scenario. Loan Losses are 3-year cumulative impairment losses on financial and non-financial assets in the banking book. Trading losses are 3-year cumulative losses from the stress in the trading book. Total losses are the sum of loan losses and trading losses. Losses are incurred in the adverse scenario. Banks are the number of public banks in each country.
Comparison of our benchmark capital shortfalls with comprehensive assessment shortfalls
- Our benchmark capital shortfall estimates amount to €450 billion for the 39 publicly listed banks. The countries with the largest expected shortfalls in a systemic crisis are France (€189 billion), Germany (€102 billion), and Italy (€76 billion). Malta and Slovakia (whose banking systems are among the smallest in the Eurozone) have no capital shortfalls under our benchmark estimates (see Table 2).
- The size of the regulatory capital shortfalls is less than 5% of the estimates using our benchmark stress test (Table 2).
- While the 5 largest banking systems (measured by total assets of banks in our sample), i.e. France, Germany, Italy, Spain, and Belgium, have an estimated capital shortfall of €432 billion using our benchmark stress test, they have less than €8 billion shortfall in the adverse scenario of the regulatory assessment (Table 2).
- Capital shortfalls estimated under our benchmark stress tests are weakly but in fact negatively correlated with the supervisory shortfalls (Figure 1). The rank correlations reported in Table 3 support this negative association.
Note: SRISK as of 31 December 2013; shortfall estimates are in million euros and aggregated over all public banks within each country.
Note: This table reports rank correlations of regulatory stress test capital shortfalls and losses with SRISK 5% VLAB. SRISK 5.5% VLAB is calculated assuming a 5.5% prudential capital ratio (which is the measure available on the NYU Stern Volatility Lab website) as of 31 December 2013. Shortfall 5.5% CET 1 is the shortfall to the 5.5% common equity Tier 1 capital ratio in the adverse scenario. Loan Losses are 3 year cumulative impairment losses on financial and non-financial assets in the banking book. Trading Losses are 3 year cumulative losses from the stress in the trading book. Total Losses is the sum of Loan Losses and Trading Losses. Losses are incurred in the adverse scenario. ** indicates significance at the 1% level.
Comparison of our benchmark capital shortfalls with comprehensive assessment losses
- The capital shortfalls estimated under our benchmark stress tests are highly correlated with the actual losses under the adverse scenario both in the banking book (rank correlation of 0.761) and the trading book (rank correlation of 0.937) (see Figure 2 and Table 3).
Note: SRISK as of 31 December 2013 against 3 year cumulative loan losses, trading losses and total losses (loan losses + trading losses) in the adverse scenario. SRISK and loss estimates are in million euros and aggregated over all public banks within each country.
Acharya and Steffen (2014) provide a number of benchmark stress testing models to estimate capital shortfalls during a systemic crisis. The analyses suggest possible capital shortfalls between €80 billion and more than €700 billion depending on the respective model. The regulatory capital shortfall disclosed by the ECB on 26 October, 2014 reveals a capital shortfall under an adverse scenario of €24.6 billion, of which €19.8 billion can be attributed to publicly listed banks.
The negative correlation between our benchmark estimates and the regulatory capital shortfall, but a positive correlation between our benchmark estimates and regulatory estimates of losses, suggests that regulatory stress test outcomes are potentially heavily affected by the discretion of national regulators in measuring what is ‘capital’, and especially by the use of risk-weighted assets in calculating the prudential capital requirement.
This highlights the importance of using multiple benchmark leverage ratios, such as the market-based approach we employ, and simple leverage ratio (which is not affected by regulatory risk weights).
Moreover, the differences between the shortfalls estimated in Acharya and Steffen (2014) and the ECB’s estimates appear to be driven by the large banks in large countries such as France and Germany. No capital shortfall was identified for these banks during the comprehensive assessment. This is possibly due to the fact that systemic risk and feedback effects from the financial sector in the real sector, which are captured in the market data, have been completely ignored in regulatory assessment (Steffen 2014).
 Bank of Cyprus was taken private in 2014. Banco Espirito Santo (a Portuguese) lender, failed in August 2014. While the ‘bad bank’ will be wound down, the viable part of the bank has been transferred into a new entity which was not included in the stress test exercise due to time constraints.
 This capital shortfall measure has been implemented based on Acharya at al. (2012) and Brownlees and Engle (2013). The data are provided by New York University’s VLAB (http://vlab.stern.nyu.edu/welcome/risk/). The theoretical motivation for the measure can be found in Acharya et al. (2010). SRISK has been documented to be a comprehensive measure that includes losses due to both a bank’s investments in assets and its exposure to fragile liabilities, which in the current European context relate, respectively, to holdings of peripheral sovereign bonds and (short-term) funding risk such as US money market fund withdrawals and other wholesale investors (Figure 5).
 The removal of the prudential filter on unrealised gains or losses on sovereign exposures held in the available-for-sale (AFS) portfolio is a notable exception. EBA-defined harmonised rules require a transitional phase-in of gains or losses (2014: 20%; 2015: 40%; 2016: 60%) [ECB 2014]. The ECB recognises that “there is a need to improve the consistency of capital and in particular the treatment of the deductions and the related quality of CET1 capital. This will be an issue for the SSM to address as a matter of priority“.