Please download to get full document.

View again

of 24
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.


Publish on:

Views: 12 | Pages: 24

Extension: PDF | Download: 0

MACRO STRESS TESTING FRAMEWORK AT THE NATIONAL BANK OF SLOVAKIA 1. JÁN KLACSO POLICY PAPER National Bank of Slovakia Imricha Karvaša Bratislava October 2014 ISSN This Working Paper has not undergone a language revision. The views and results presented in this paper are those of the authors and do not necessarily represent the official opinion of the National Bank of Slovakia. All rights reserved. Reproduction of short extracts from the text, not longer than two paragraphs, is permitted without the prior approval of the author provided that the source is acknowledged. 2 Macro Stress Testing Framework at the National Bank of Slovakia 1 Policy paper NBS Ján Klacso 2 Abstrakt This paper describes the current macro stress testing framework at the National Bank of Slovakia. Stress testing is aimed at testing the resilience of the banking sector to negative developments on the financial markets and in the real economy. The paper describes satellite models and assumptions used in the framework. The results of back testing and the most actual results of stress testing are also presented. JEL Classification: E44, E47, G21 Keywords: macro stress testing, banking sector, back testing Downloadable at 1 I would like to thank Adam Geršl and Martin Šuster for their valuable comments and suggestions made on the earlier version of this paper. My special thank goes to my colleagues, Marek Ličák, Štefan Rychtárik, Pavol Jurča, Viktor Lintner, Ľuboš Šesták and Pavol Latta for their valuable comments and technical support during the built up of the stress testing framework and also when conducting regular stress testing. 2 Macroprudential policy department of the NBS, 3 1. INTRODUCTION Stress testing is a widely used tool among financial institutions, regulators, as well as central banks to test the resilience of a given portfolio, a given institution, or the entire sector to adverse developments on the financial markets and in the real economy. This paper aims to describe the macro stress testing framework of the Slovak banking sector at the National Bank of Slovakia (NBS). The term macro refers to the aim of testing the whole banking sector to adverse developments of the wider economy and on the financial markets (Geršl et. al, 2012 or Henry and Kok, 2013). The role of stress testing as a forward looking tool in the analysis of the viability and riskiness of the banking sector is gathering more and more importance after the outburst of the financial crisis. Moreover, it seems to be a useful supporting tool also for decisions in the field of macroprudential policy. Therefore, it is crucial that the stress testing framework accurately addresses the main risks banking sectors are facing and it transmits in a proper way a potential negative macroeconomic development to the development in the portfolios of banks, thus enabling capture of potential future solvency position of the banks conditional on the stress scenarios. While the feedback loop between the financial and the real sector is also an important issue stress testing should be aware of, it is still hard to incorporate this issue into the framework (see Henry and Kok 2013, Burrows et. al 2012 or Jacobson et. al, 2005). In the paper, we first give a brief overview of the basic concepts of stress testing in Section 2. Then, in Section 3 we describe the historical development and in Section 4 recent framework of the macro stress testing. We outline how scenarios are set, give a detailed description of the satellite models used in the framework and describe key simplifying assumptions that are necessary to fill in the gaps when conducting stress testing. We describe how the results are aggregated to obtain estimates of the net interest income and the solvency of the banks. Then, Section 5 presents the results of back testing, which is a useful tool in the assessment of the appropriateness of the models and assumptions used in the framework. Section 6 is dedicated to the most recent results of the macro stress test to give an overview of the risk profile of the Slovak banking sector. Finally, Section 7 concludes and outlines possible future improvements in the framework. 2. BASIC CONCEPTS OF STRESS TESTING Stress tests are used to assess the resilience of a given portfolio, an institution or a sector to adverse scenarios that have low probability, but are still plausible. It is an important tool that can help to build a more complex risk assessment. In case of banks, while expected losses are covered by provisions and regulatory capital is needed to cover unexpected losses to a given extent, stress testing can help to discover the volume of additional capital that can prevent a failure of the institution in case of a tail event. This is also one of the purposes of the stress tests conducted by supervisors, as the results can be used to decide on the volume of additional capital that can be required within Pillar II of the Basel framework. Central banks usually use stress tests in the process of the assessment of financial stability. While generally used for analytical purposes by supervisors and regulators, after the outbreak of the financial crisis in 2007 this tool proved to be useful also for political 4 purposes. The Supervisory Capital Assessment Program of the Fed in or the EBA stress testing exercise in 2011 with the consequent EBA Capital exercise in 2011/ was aimed at restoring market confidence by providing credible information about potential losses in the banking system. As stress testing is an important part of the complex risk assessment of the banks and the banking sector as a whole, it could and should be also a part of the toolkit supporting macroprudential policy. Due to the fact that stress testing is a complex process with a lot of assumptions needed, it is however disputable if it is feasible as a forward-looking indicator or an early warning device (see, e.g., Borio et al, 2012). Also, due to its highly technical nature and due to the transparency needed in case of macroprudential policy, it would be hard to use it for calibration purposes in the built-up phase. On the other hand, the results of stress testing can serve as an interesting input to the decision during the release phase, for which there is still an on-going debate and different views about the potential interpretation. In cases where the release phase focuses on the loss-absorbing capacity of the sector in case a risk has materialized or is starting to materialize, the results of stress testing can give a first view on the minimum capital that has to be released in order for the banks to cover respective losses. Figure 1 Expected, unexpected losses and the aim of stress testing Pricing and provisioning Regulatory capital Stress testing Expected losses Unxpected losses Tail events Frequency Losses The aim of stress tests can vary to a great extent. In case of a given portfolio, the resilience to the changes of one or more risk factors can be questioned (so called sensitivity tests). In case of an institution, the tests can be used to assess the overall robustness. This requires that the stress scenarios include shocks stemming from different risk factors while the correlation between these factors and their mutual dependence has to be accounted for, too. Stress tests assessing the resilience of the whole or part of the financial sector, so called 3 More details available at 4 More details available at 5 macro stress tests, can take into account also the interconnectedness between financial entities as well as possible negative feedback loops between the financial system and the real economy. While in general central banks and supervisors are interested in the impact of the scenarios on the solvency of the institutions, the financial crisis revealed that liquidity is also an important issue. Therefore, stress tests aimed at or including also the testing of the liquidity position of the institutions are now being developed (see Geršl et. al, 2011 or Schmieder et. al, 2012). In line with the goals gradually being more and more complex, the stress testing framework and the models also improved. While at the beginning models were based on simple scenarios linking historical macroeconomic development with financial variables (see, e.g., Blaschke et. al, 2001), they became much more complex integrating market risks such as interest rate risk, FX risk or equity risk, credit risk and, as a lesson from the recent financial crisis, also sovereign risk, as described also in Geršl et. al (2012). Currently, stress tests are not focusing solely on the losses from these risks, but try to incorporate them into the complex behavior of the financial institutions, which means estimation of the overall preprovision income and risk weighted assets is also an important element of the tests. The financial crisis also showed that more attention has to be paid to the design of stress scenarios since in many cases proved to be not severe enough or didn t link the shocks from different risks in a proper way (see, e.g., Ong and Čihák, 2010). While a lot of work has been done in the field of stress testing, challenges remain mainly in the way how to incorporate funding shocks that can evolve relatively quickly compared to the impact of the scenarios on the solvency of institutions, which is a much more gradual process; or how to best address the possible feedback effects to the real economy. 3. THE HISTORY OF STRESS TESTING IN THE NBS Stress testing is conducted semi-annually in the NBS since 2005 and the results are part of the Analysis of the Slovak Financial Sector published by the NBS. The first versions of stress tests consisted of sensitivity analysis of the banking sector to different risks: interest rate risk, FX risk, credit risk, liquidity risk and contagion risk. These were relatively simple scenarios, linking e.g. an increase in the default rate of the loan portfolio to the capital adequacy of the banks, or an increased outflow of interbank funds or client deposits to the liquidity position of the banks. Macro stress testing, aimed at linking an expected worsening in the development of the real economy to the development of the quality of the loan portfolio, was conducted first in The testing was based on a macroeconomic VEC model. A detailed description of the model and the results is in Zeman and Jurča (2008). Since 2009, the stress testing includes estimation of losses from corporate and household credit risk, FX risk, equity risk, sovereign risk and the estimation of net interest income. The estimation of the overall pre-provision income became also a part of the exercise. While up to 2009 the development of the domestic macroeconomic factors (GDP, inflation and unemployment) within the respective scenarios were determined by expert judgment, since 2009 the macroeconomic model of the NBS developed for medium term forecasts is used in 6 setting the expected path of these variables. It means that the assumed path of the domestic GDP, inflation and unemployment in case of the baseline scenario coincides with that of the respective medium term forecast. While up to 2011 the overall volume of the risk weighted assets was kept constant, since 2011 there are some simple assumptions linking the changes in the risk weighted assets to the expected changes of the overall loan portfolio. These simple assumptions aim to link the changes in the volume as well as in the riskiness of the loan portfolio to the risk weighted assets to give a more realistic picture about the possible development of the capital adequacy ratio. Liquidity risk is not part of the macro stress testing aimed at the solvency position of the banking sector. On the other hand, since 2009 the NBS monitors the development of the liquid asset ratio that can be explained as the short term liquidity position of banks under stressed conditions. The basic principle of this ratio is to ensure that banks are able to cover a certain percentage of the liabilities to their customers with assets that are convertible into liquid funds within a period of up to 1 month. The method of the calculation is laid down in a decree THE CURRENT FRAMEWORK OF MACRO STRESS TESTING IN THE NBS Stress testing is focusing on the assessment of the robustness of the banking sector in terms of solvency to adverse macroeconomic development and adverse development on the financial markets. It means stress testing is focusing on how the capital adequacy ratios of the banks would be affected by the baseline and the stress scenarios. Three scenarios are considered for the stress testing exercise in general. The first is the baseline scenario that is based on the medium term forecast of the NBS and is used as a benchmark when assessing the impact of the two stress scenarios. The stress scenarios are designed to involve the risk factors that are deemed relevant for the banking sector as much as possible and to reflect the current macroeconomic situation and the possible future adverse development that stems from the actual situation. The horizon of the tests is set to be two years in case the stress testing is conducted based on end-year data and two and a half year in case of end-june data. In the first round, the expected path of the global macroeconomic variables and financial indicators over the stress testing horizon is set by expert judgment. These variables are then used as an input to estimate the development of the domestic macroeconomic variables. The expected path of the global variables and the conditional estimates of the domestic macro variables are then used as an input for estimating the development of selected parts of the net profit of the sector by satellite models (net interest income, losses from corporate credit risk, household credit risk, FX risk, equity risk, sovereign risk). The remaining parts of the net profit, that are not modelled, are estimated using simplifying assumptions. The overall amount of capital of the banks is then affected by the net profit/loss after taxation and the 5 Decree No. 18/2008 of Národná banka Slovenska of 28 October expected dividend policy of the respective banks. So far, there is no distinction made between the different types of capital banks report (Tier I, Tier II). However, as the total amount of own funds of the banking sector consists mainly of the highest quality capital (Tier I capital, which after the implementation of the Basel III framework will transform in case of Slovak banks to a great extent to Common Equity Tier I capital), the lack of this distinction does not seem to be a relevant issue. Finally, the volume of risk weighted assets (RWAs) is calculated for each bank using simple assumptions linking the volume of RWAs to the development of the loan portfolio of respective banks. The capital adequacy ratio is then a simple share of the own funds at the end of each year of the stress testing horizon and the volume of risk weighted assets. Currently the capital threshold for passing the exercise is 8%. With the implementation of the Basel III framework and the possible application of several capital buffers this threshold can change and may not be the same for every bank. Figure 2 Scheme of the macro stress testing LGD: Loss given default STRESS SCENARIOS Stress testing uses three alternative scenarios: a baseline scenario that is based on the medium term forecast of the NBS and two stress scenarios. In the first round, the expected path of the global macroeconomic variables and financial indicators over the stress testing horizon is set by expert judgment. It means that there are no shocks imposed on these 8 variables, but the whole path is set for the full stress testing horizon. The expected path has to fulfil three criteria (see, e.g., Rodrigo and Drehmann, 2009). First, the development of the respective variables has to be economically reasonable, which means that the plausibility of the scenarios has to be assured. Second, the development of the variables has to reflect the actual view of the central bank about the possible risks that can affect the global macroeconomic development. Third, the stress scenarios have to reflect an economic downturn and a financial turmoil that is severe enough; it means that it deviates enough from the baseline scenario to assure a low probability of the scenarios. However, as the scenarios are set by expert judgment, it is hard to attach probability score to the manifestation of these scenarios. The development of domestic macroeconomic variables (GDP, HICP inflation and the unemployment rate) is then estimated by the macroeconomic model of the NBS used for the medium term forecasts. This macroeconomic model is a structural model of a small and open economy (Slovakia in this case) consisting of several error-correction equations describing the development of the economy. The model assumes that in the long run the economy is in its steady state determined by the supply side while in the short run there can be and indeed are deviations from this steady state due to demand, supply and price shocks. As Slovakia is a member of the Euro area since 2009, the monetary policy is assumed to be exogenous (for more details see Reľovský and Široká, 2009). The development of the exogenous variables is calibrated in such a way that the estimated path of the GDP, the inflation and the unemployment within the stress scenarios deviates sufficiently from the baseline scenario in a way to pose severe stress on the domestic banking system. While in general there can be scenarios the calibration of which is not straightforward, so far it was always possible to achieve the desired stress scenarios using this macroeconomic model. The global macroeconomic and financial variables and domestic macro variables then serve as an input to satellite models that link the development of the real economy and in the financial markets to the development in the banking sector ESTIMATION OF INTEREST RATES In case of the interest rates, it is assumed that the changes of the ECB main rate (i.e. the interest rate for the main refinancing operations) are gradually transmitted, first into the changes of the interbank and discount rates and, second, into the lending and deposit rates of banks. It is assumed that the level of interbank rates also reflects counterparty credit risk and the level of discount rates (particularly with longer maturities above 12 months) reflects, at least partially, expected future macroeconomic development. Interbank interest rates EURIBOR interest rates are estimated using error-correction (EC) models, including a cointegrating relationship between the respective interbank rate, the main rate of the ECB and the itraxx Senior Financial index that is used as a proxy for the level of the counterparty credit risk on the interbank market. This error correction approach is a widespread method in the literature when estimating the development of interest rates (De Bondt, 2002, Curcio and Gianfrancesco, 2010, Gambacorta and Iannotti, 2005). A dummy variable is also included into the cointegrating equation to reflect the possible effects of the non-standard measures of the ECB on the interbank rates. The dummy is set to be 1 from May 2009 until 9 September 2010 and then gradually deceases to 0 until December 2010 to reflect the effects of the 1Y long term refinancing operation (LTRO) of the ECB, than it is set to be 1 again from December 2011 onwards to reflect the effects of the 3Y LTRO of the ECB, and it is 0 elsewhere. Results of the unit root tests and cointegration tests are shown in Table 12 and Table 13 in the Appendix. The estimated equation has the form = , (1) = ( + + ( ) + + _ ), (2) where is the respective EURIBOR interbank rate, is the main rate of the ECB, is the itraxx index, is white noise. ( ) is the expected value of the main rate of the ECB in time 1 one period ahead
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks