Systemic risks

Stress tests are an important tool for banking supervision. Experts conduct stress tests to find out whether banks will be able to remain stable and resilient even if an extreme economic shock occurs. For large banks, the ECB and the European Banking Authority (EBA) carry out stress tests. Stress tests for smaller banks in Austria are conducted by the OeNB.

The starting point of each stress test is the definition of adverse scenarios like a sharp increase in inflation or a slump in economic growth. On this basis, experts calculate the impact of these developments on banks’ capital and liquidity, comparing hypothetical losses under the stress scenario with the capital (the buffer for unexpected losses) actually maintained by the banks. 

Stress tests can also help determine how developments at one bank may affect others. In other words, these tests show how the entire banking system might react in a crisis situation. Such comprehensive analyses are known as macro stress tests.

Stress test findings can contribute to banks substantially raising their capital levels, thereby also enhancing banks’ risk-bearing capacity. However, stress tests are not a silver bullet; they are just another tool of quantitative financial market analysis.
 

What are the roles of the EBA, the ECB and the OeNB?

National and European supervisory authorities cooperate closely and perform complementary roles in stress testing: The EBA and the ECB are responsible for the largest banks in the EU: These banks submit their risk data and results, and European supervisors review and consolidate them (“bottom-up stress tests”). 

The OeNB conducts stress tests for smaller banks in Austria and for the Austrian banking sector as a whole. These stress tests follow a “top-down” approach. This means that OeNB experts use existing data for their calculations rather than collecting additional information from banks. This efficient approach enables the OeNB to respond flexibly and independently to emerging developments. 
 

What scenarios are used in stress tests?

Severe, but plausible – this is the guiding principle stress testers follow when preparing the stress scenarios on which the tests are based. Stress tests should simulate a serious threat to banks’ viability while being realistic and credible at the same time. At the OeNB, the banking supervision, economics and financial stability functions develop scenarios in close cooperation. They are not limited to typical economic crises, but also reflect shocks like the COVID-19 pandemic or the effects of carbon taxation, and are used as a basis for calculating risk factors like credit defaults or house price shocks. 
 

Where are stress test results used?

Stress tests provide insights into the vulnerability of individual banks. Supervisors base their expectations that they communicate to banks on stress test results (for example the expectation that a bank improves capitalisation to address its vulnerabilities). As stress tests also take a system-wide perspective, they help assess the stability of the Austrian banking sector as a whole. The OeNB publishes aggregated stress test results in its Financial Stability Report and on its interactive Web portal on an annual basis.
 

How did stress tests become a standard supervisory tool?

Originally, stress tests were used in banks’ (market) risk management functions. 

The International Monetary Fund (IMF) promoted the use of stress tests worldwide on a broader scale as part of its Financial Sector Assessment Programmes (FSAP). The OeNB conducted its first large-scale stress test under the 2003 FSAP. 

The broader general public may have first taken note of this supervisory tool during the 2008–09 global financial crisis, when large US banks had to undergo stress tests. In Europe, the European Banking Authority (EBA) and the European Systemic Risk Board (ESRB) jointly coordinate stress tests comparable to those conducted in the USA. The ECB and the national supervisory authorities – in Austria, the OeNB and the FMA – are responsible for performing these tests. Such an international exercise requires a large amount of coordination. However, the effort pays off: Supervisors, market participants and the general public benefit from comparable data and consistent risk assessments.

In 2007, the OeNB for the first time involved major Austrian banks in the calculation of macro stress tests. Between 2007 and 2014, these banks used to run bottom-up stress tests whereas the OeNB would run top-down scenarios, both using the same scenarios. This approach has since become international best practice. The ECB has conducted bottom-up stress tests for large banks since it took over the direct supervision of significant institutions, while the OeNB has continued to run top-down stress tests for all Austrian banks. For this purpose, the OeNB implemented ARNIE (Applied Risk Network and Impact Assessment Engine) in 2013, which is still in use today.

At the international level, there is a growing trend towards the development of stress test models that base their calculations on dynamic bank balance sheets. This makes it possible to simulate how banks might react to new developments. At the OeNB, it has been possible to use this approach with existing tools since 2024. The OeNB publishes aggregated stress test results in its Financial Stability Report. In addition to its annual solvency stress test, the OeNB also publishes analyses covering specific focus areas, such as climate risk stress tests or calculations based on the dynamic balance sheet model.
 

To effectively reduce risks to the financial system as a whole, we need detailed guidelines and special – macroprudential – tools. Buffers are a key instrument in the macroprudential toolkit. They are additional capital cushions that banks may be obliged to hold. In addition, the supervisory authority may also impose other, in particular borrower-based, measures. The OeNB provides the analyses that underpin the implementation of macroprudential measures by the competent authority in Austria, the Financial Market Authority (FMA).

Macroprudential measures must be based on sound evidence and analysis. Before a measure is implemented, OeNB experts assess whether a severe risk to the financial system exists and whether a particular macroprudential measure would be effective in addressing this risk. Only if this analysis shows that a measure is justified may the Financial Market Authority (FMA), as the competent authority, impose it.

However, while macroprudential measures (i.e. measures to mitigate systemic risks) can reduce the probability of financial crises, they may also lead to higher interest rates, thereby dampen economic growth. Therefore, it is crucial to weigh their social benefit against their social costs. Only if the expected benefits exceed the expected costs will a measure be implemented. 

OeNB experts deliver the necessary analyses as required by law (financial market omnibus act, Act on the Oesterreichische Nationalbank as amended). These analyses include an assessment of the impact a given macroprudential measure has on the real economy. They also examine the potential consequences for the economy if the measure were not implemented.

Ex ante impact assessments analyse potential costs. The methodology used varies depending on the type of macroprudential measure. Still, for both macroprudential buffers and borrower-based measures, the OeNB must estimate the social costs their implementation potentially cause, including, in particular, the effects on economic growth. The advantages that the prevention or mitigation of a systemic banking crises brings should clearly outweigh these social costs in the long term.

Banking crises always cause high social costs. A study based on data from 151 crises worldwide put the public costs to around 6.7% of GDP. Public debt rises by around 21% of GDP, and economic losses over the entire duration of a crisis amount to around 35% of GDP, according to this study. Moreover, banking crises often have long-lasting effects: In high-income countries, more than half of all crises last five years or longer (Laeven and Valencia, 2018).

The chart below shows the international median of these social costs for high-income countries, including Austria.

After a macroprudential measure has been implemented, the OeNB conducts ex post analyses to find out whether it had the intended effect. Therefore, the OeNB examines those indicators that have been used in the ex ante risk analysis and in setting the buffer. In addition, the analyses take into account other regulatory developments, macroeconomic changes and the assessments of international institutions (such as the ECB or the IMF) and rating agencies.

Impact assessment of macroprudential buffers

Ex ante impact assessments of capital-based macroprudential measures (i.e. capital buffers such as the O-SII buffer, the SyRB or the CCyB) follow a multi-stage process. In short, experts analyse how the “private costs” incurred by banks through the implementation of a macroprudential measure affect the pricing of loans. As a result, financing costs in the real economy may increase, which potentially dampens overall economic activity. These social – that is, macroeconomic – costs may also translate into subdued investment and consumer spending and result in slower economic growth.

The analysis takes place along the following methodological stages:

• First, the capital requirements of the banks which will be requested to maintain higher capital buffers are identified by comparing actual capital levels with future regulatory capital requirements. At the same time, the capital requirements of all banks are calculated by assuming that the management buffer (i.e. capital in excess of the current capital requirement) remains unchanged.
• In the impact assessment, experts estimate the additional “private costs” banks incur as they have to raise capital (“opportunity costs”). It is assumed that banks replace the most expensive part of debt capital with equity capital to meet buffer requirements (with total assets and profit forecasts remaining unchanged). Empirical evidence shows that banks have numerous options for responding to capital requirements. They choose the options that keep additional costs as low as possible. Therefore, the costs across all measures are expected to be below the costs of a capital increase. The use of the latter is hence a very conservative assumption.
• Next, it is assumed that banks pass on additional costs (opportunity costs) in full and exclusively to borrowers in the real economy. This means that banks compensate for additional capital costs by raising interest rates on new loans to non-banks. 
• Finally, the impact of this increase in interest rates on the whole economy is estimated. These calculations are based on the OeNB’s tried-and-tested forecasting model. The elasticities calculated represent the growth effects of a rise in interest rates on macroeconomic variables (including GDP growth, changes in gross fixed capital formation and private consumption). Two channels are particularly important: First, interest rates enter the equation as cost factors (“cost of capital channel”). All other things being equal, an increase in interest rates reduces the optimal stock of capital and the demand for investment. Second, higher interest rates reduce private consumption, as they raise the saving ratio on the one hand and, due to lower consumer demand, reduce employment and, hence, real disposable household incomes (“substitution channel”).
• In addition, the analysis also covers the imputed perspective (capital tied up).

Impact assessment of borrower-based measures 

The impact assessment examines economic developments under various crisis scenarios with and without borrower-based measures, using the results of the TUI model in combination with the OeNB bank stress test results.

The social benefit of borrower-based measures is the fact that property market and banking crises can be averted. Banking crises may occur even when probabilities of default (PDs) seem to be low. The past has shown that even PDs between 2% and 4% are associated with property market crises. Banks’ funding costs are an important factor for the real economy. They are influenced by rating agencies, among others, who also take into account macroprudential measures. The Austrian banking system is currently in the third-highest rating category in Standard & Poor’s global ranking of banking systems (no country is in the top rating category). Unexpectedly high losses from real estate lending during an economic crisis are generally associated with a deterioration in balance sheets and credit ratings, leading to significantly higher refinancing costs for banks (Schmitz et al., 2019). These higher costs are passed on to customers taking out new loans (households, businesses, the government). Furthermore, increased refinancing costs may result in assets already held in the portfolio making a negative contribution to earnings, thereby causing banks’ equity capital to fall further (Leika et al., 2019). There is a risk of a downward spiral caused by rising refinancing costs, falling bank capitalisation and a further loss of confidence in the banking system, which in turn would drive up refinancing costs for banks and borrowers even further. In short, a crisis in the real estate market has negative consequences not only for banks but also for the real economy. And the social costs are even higher when property market and banking crises lead to bank failures (like, for example, in Ireland or in Spain 2008–14).