Macroprudential Stress Testing

Transcription

1 Macroprudential Stress Testing The foundations of Stress Testing: An introductory overview of the ECB approach Riccardo Tedeschi Senior Specialist

2 1 George E.P. Box was a British statistician, pioneer in process quality control, analysis of time series, design of experiments and Bayesian inference All models are wrong, but some are useful. George E. P. Box ( ) 1 The most original innovations often stem from necessity or are born for the purposes of overcoming unforeseen difficulties. This is the case for macroprudential stress tests, namely a set of analysis techniques used by financial supervisory authorities since the great global financial crisis in This approach has resulted in the achievement of significant advances in the grasp of the mechanisms governing the interactions between the financial system and the real economy on the one hand, and between the former and international relations on the other. This article proposes to address three issues: when were stress tests introduced in Europe; what microprudential stress tests are and their purpose; and what are their most recent developments. ) 2 (

3 The introduction of stress tests in Europe 2 That is to say, those banks whose assets and equity were aligned with the requirements set out by the very same supervisory authorities 3 Although finalised since 2004, the Basel II agreements became active only in January 2007 W ith the great financial crisis in , supervisory authorities worldwide unexpectedly realised that a number of banks, while well capitalised 2, were facing serious difficulties within a very short time due to losses accumulated on specific held assets (Asset Backed Securities, ABS) or on the grounds of liquidity. The numerous ensuing banking failures and the subsequent financial crisis are common knowledge. In Europe, the impact of the crisis further expanded with the so-called sovereign debt crisis that, between 2011 and 2012, affected countries of which the governments were most heavily indebted in relation to GDP. The reasons and root causes, as well as the transmission channels of the great financial crisis will not be addressed herein. The latter triggered a deep change in banking supervisory regulations on a global scale, with the transition from the so-called Basel II agreements that had just come into force in , to the subsequent Basel III agreements in 2010, reinforcing the capital requirements of banks and introducing a wealth of restrictions and controls relating to their degree of liquidity. Since 2009, the supervisory Authorities of various countries, such as the USA, Great Britain and the European Central Bank in the Euro area, started to internally conduct a series of stress tests on the monitored banks following a top-down approach, in order to better orient their monetary policy and supervisory decisions. In 2014, prior to the launching of the Single Supervisory Mechanism (SSM) and upon the initiative of the European Banking Authority (EBA) in collaboration with the ECB, an extensive and accurate analysis of the assets held by the 130 main European banks to be monitored by the Frankfurt authorities was conducted (comprehensive assessment). This analysis also included a stress test at the level of each individual bank with a bottom-up approach. The EBA EU-wide stress tests, that is to say extended to larger European banks (significant institutions) were then replicated in 2016 and will be repeated every two years. The next will be conducted in At the end of 2014, the entire Supervisory Review and Evaluation Pro- ) 3 (

4 cess (SREP) was reformed. This is the annual process in which supervisory authorities assess the situation of each individual bank: the business model, the governance of equity risks and liquidity risk. Stress tests therefore became part of the instruments used for the risk assessment of individual banks on a permanent basis. Stress tests and reverse stress tests were also introduced within the context of the recovery plans under the new European legislation on banking resolution (BRRD). It is no exaggeration to say that today, ten years after the onset of the Great Crisis, all risk managers within medium-large banks carry out at least one stress test per year for management and regulatory purposes. Microprudential testing by the EBA T he purpose of microprudential testing for any bank may be explained by drawing the following parallel example. Just as athletes taking part in sporting competitions are required to undergo ECG monitoring on a regular basis to detect any potential cardiovascular anomalies arising only upon exertion (that, conversely, would not emerge at rest), banks aiming to compete in financial intermediation must undergo tests to appraise their resistance to periods of prolonged recession in particularly adverse scenarios. When a bank fails to adequately pass stress tests, the supervisory authorities may take action by imposing measures of increasing intensity, such as organisational or strategy changes, an increase in liquidity buffers, the suspension of payment of dividends or an increase in capital. The EU-wide stress tests organised by the EBA are mandatory for the larger banks monitored by the ECB and consist of simulations conducted at the level of each individual bank, with the primary aim of evaluating their financial soundness. In brief, the characteristics of the EBA stress tests are clarified as follows: A dedicated European Systemic Risk Board (ESRB), which relies on data and resources provided by the ECB, develops two macro-financial scenarios identical for all banks: a baseline scenario expressing the consensus forecasts and an ad- ) 4 (

5 verse scenario representing the rather unlikely, albeit not unrealistic, negative evolution of the economic and financial situation. These scenarios are delivered to all participating banks. The banks undergoing stress testing should simulate the evolution of their budgetary developments over a three-year time horizon: capital situation and income statement; regulatory capital and the so-called Risk Exposure Amounts, REA, also referred to as Risk Weighted Assets, RWA. There is a special focus on solvency indicators: CET1 capital ratio, which is the ratio between the best liable equity capital (the Common Equity Tier 1, CET1) and the RWA. For the test to be successful, this ratio at the end of the simulation horizon must be above a specific threshold. In the first stress-testing edition in 2014, the CET1 capital ratio thresholds were respectively equal to 8% for the baseline scenario and 5.5% for the adverse scenario. Banks falling below one of the thresholds were required to submit an action plan for the recapitalisation of the business. By contrast, in the second stress-testing edition in 2016, the EBA did not make the thresholds explicit, even if the substance of the test remained unaltered. Figure 1 illustrates the aggregated results of the 2016 EBA stress tests: the evolution of the aggregated CET1 capital ratio percentage that, from an initial 13.2% in 2015, drops to 9.4% at the end of 2018: a 380-basis-point negative impact within the European average. Figure 1 Evolution of the aggregated CET1 capital ratio % and impact (bps) in relation to the initial figure in 2015 (bps) Source: EBA, ST 2016 Results ) 5 (

6 4 The impacts are detailed in the two scenarios, phased-in or transitional CET1 capital ratio, calculated on the basis of a gradual transition provided for by the Basel III regulations until 2018, and fully loaded CET1 capital ratio, namely according to the comprehensive regulations in force after 2018 Instead, Figure 2 illustrates the impacts of stress testing on the CET1 capital ratio over the three years of simulation between 2015 and 2018 in the adverse scenario per individual bank 4. As shown, there is a great diversity of results across the different banks. Figure 2 Impact on the CET1 capital ratio between 2015 and 2018 in the adverse scenario per individual Source: EBA, ST 2016 Results ) 6 (

7 In its present configuration stress-testing, while based on macro scenarios developed by the ESRB, is conducted according to a microeconomic approach. The calculations are made by each individual bank and, as mentioned, follow a bottom-up basis: banks use their own management systems and usually summarise the data of their managed portfolios from a very granular database at the level of each individual operation. The preliminary results processed by the banks are then forwarded to the supervisory authorities and are subject to a careful quality assurance process, with a traffic-light returned report system: green light : go-ahead; yellow light : request for further clarifications and information; red light : blocked and returned to sender for new processing. In order to formulate their own opinions and identify any potential abnormalities, the supervisory authorities rely on different systems. The most important is an actual top-down budget simulation model of the individual bank, that uses aggregated budget data and the reports received for monitoring purposes and a sophisticated benchmarking system, comparing the performance of similar size and with similar business models. The main characteristic of the EBA EU-wide stress tests is that they are based on the so-called static balance sheet approach: in the course of the three-year simulation period, any variation in the volumes of the masses managed are considered in relation to those at the start of the period. In other words, no bank can grow with new loans to customers, deposits or bonds issued to customers, or an increase in the volume of managed savings, even indirectly (assets under management). As the lending and funding transactions expire, they are reinvested in the same type of original transactions. This scenario aims to equate all banks taking part in the stress testing (playfield levelling), irrespective of the differences in growth provided for by the strategic plans of the individual businesses. The sole budget item that is allowed to grow during the simulation is the one relative to non-performing loans, that shift their status from performing loans to defaulted on the basis of the probability of default as estimated by the bank and that then determine the value adjustments on loans, according to the estimated loss rates (Loss Given Default, LGD). The probability of default and the LGD are estimated by the individual banks via own satellite models, based ) 7 (

8 5 Typically, in the real world, banks recover a significant portion (on average 40%) of the defaulting loans over a number of years (for Europe, on average 3 years, for Italy 5 years) 6 The compression of the interest margin is all the more intense the lower the initial rating of the bank itself 7 Recently, some authors have raised the issue about how to render stress tests more accurate and in a way more severe owing to the market risks embedded in the so-called level 3 assets, namely the unlisted securities that are evaluated according to bank s internal models on the economic scenarios provided by the ECB and especially in the adverse scenario can reach very high levels as compared to the current ones. The rules of the game of such testing include some particularly stringent ones for commercial banks. First, the bank cannot recover any amount on assets in a state of insolvency (no workouts on defaulted assets): this circumstance is all the more burdensome the greater the value of the non-performing loans 5. During the simulation, a deteriorated credit rating of the bank is also hypothesised, to generate an increase in the cost of borrowing of the bank itself that cannot be recovered via an increase in the rates for loans to customers and the ensuing compression of the interest margin of the bank 6. The stress testing part concerning the liquidity risk is, on balance, relatively limited. However, a strong negative impact is taken into account on the value of Government securities and other securities in the portfolios of financial assets held for sale and negotiation purposes 7. Macroprudential testing by the ECB 8 Vítor Constâncio: The role of stress testing in supervision and macroprudential policy. Keynote address by Vítor Constâncio, Vice-President of the ECB, at the London School of Economics, London 29 October 2015 (see R. Anderson Ed. (2016), Stress Testing and Macroprudential Regulation: A Transatlantic Assessment, CEPR Press) 9 ECB Financial Stability Review December 2009, Special Features B E U-wide stress tests of the EBA chiefly represent a microprudential and severe assessment of the solvency of the individual banks monitored by the ECB, useful for preventive actions in specific situations. This is an extremely useful test, although insufficient per se from the perspective of the supervisory authorities. As explained exhaustively by the Vice-President of the ECB, Vítor Constâncio, in a speech given at the London School of Economics in October on the role of stress testing and macroprudential policy, from the viewpoint of those having to make economic and financial policy decisions aimed at monitoring and limiting the so-called systemic risk, this test presents an assortment of limits. Systemic risk is defined by the ECB itself as the risk that financial instability may significantly damage the supply of financial products and services by the banks and financial institutions to such an extent that economic growth and general welfare might be seriously affected 9. The basic idea is that financial crises imply considerable costs for the real economy; therefore, the supervisory ) 8 (

9 10 ecb/tasks/stability/ html/index.en.html (ECB, web page dedicated to Financial stability and macroprudential policy). 11 STAMP, Stress- Test Analytics for Macroprudential Purposes in the euro area, Edited by Stéphane Dees, Jérôme Henry and Reiner Martin ( eu/pub/pdf/other/ stampe en.pdf) authorities should do their utmost to prevent them. There are essentially three key sources of systemic risk: macroeconomic shocks on aggregate demand or supply; imbalances stemming from public or private over-indebtedness; contagion risks resulting from high levels of interconnection and herding behaviour. The macroprudential policies adopted by central banks pursue myriad objectives along diverse dimensions 10 : time dimension: to prevent an excessive accumulation of risks, stemming both from external factors as well as from any potential market failures, and to smooth the fluctuations over the financial cycle; transversal dimension: to render the financial sector more resilient to shocks and to limit contagion effects; structural dimension: to encourage the use of a broad perspective at system level in financial regulations for market operators in order to create a set of correct incentives for market participants. Examples of macro prudential decisions include the variation in the socalled capital buffers, the additional layers of minimum capital that banks must hold in a specific period. Alternatively, the introduction of binding rules in the allocation of funding such as a ceiling for the ratio between the mortgage on a property and its value, namely the loan-to-value ratio. In financial literature since 2008, various ratios have been proposed for measuring the current level of system risk, as well as the marginal contribution of each individual financial institution to the overall systemic risk. Nevertheless, from the standpoint of the supervisory authorities, none of the indicators suggested so far has proved useful, in practical terms, to guide the decisions for intervention on the financial system in order to prevent crises from forming or to guide it out the quicksand of an existing crisis. For this reason, the ECB has decided to invest in the development of a set of its own upgraded macro prudential models that leverage the set of data and instruments used for microprudential stress testing on the solvency of banks. The approach employed by the ECB and the set of instruments developed, Stress-Test Analytics for Macro prudential Purposes in the Euro area (STAMP ), is outlined in detail in a publication dated February What are the characteristics of this new approach? ) 9 (

10 Based on the adverse scenario, identified as a potential threat to the stability of the financial system, the balance sheets of the individual financial institutions are simulated as in the EBA EU-wide stress tests. This time, however, the topdown and dynamic balance sheet approach is used. Whereas in the EBA microprudential stress testing no adjustment by the bank s management of the adverse scenario is permitted in any way, in this type of simulation conducted independently by the ECB, the possibility of dynamic adjustment by the management to the variations in the macroeconomic framework or to the actions of the supervisory authorities is introduced. Banks may react in various ways: deleveraging their riskiest activities, proceeding with capital increases, if the conditions of the stock market so allow, or accelerating the recovery of non-performing loans also via transfers. A combination of the various alternatives is also feasible. The impact resulting at system level shall differ according to the behaviour most pursued by the individual banks. For instance, let us assume that, following a slowdown in the economic cycle and a subsequent increase in the capital ratios imposed by the supervisory authorities (or required autonomously by the market), banks react by reducing the loans granted to their customers, specifically those considered the riskiest. In so doing, the credit rationing to the economy would trigger a vicious circle that could engender a recession even more severe than the one originally conceived. Figure 3 The structure of the ECB modules for the top-down analysis of the solvency of the banks Source: ECB, publication STAMP feb. 2017, chapter 2 ) 10 (

11 12 The first is a Dynamic Stochastic General Equilibrium (DSGE) model and the second is a Global Vector Autoregressive (GVaR) model The behaviour of the individual banks is simulated via models for the re-composition of loans and their collection, based on criteria for optimising the risk/ return profile. The results of the simulations conducted by the ECB tend to confirm the widespread opinion that in response to negative shocks entailing a deterioration of the credit risk and tight constraints on collection, banks tend to divest assets as opposed to proceed with increases in capital and maintain the existing leverage. Once the reactions of the banks has been defined, it is crucial in macroprudential terms to grasp the interconnection between financial economy and real economy. To this end, the ECB uses two different macroeconomic models 12, tailored to each individual European country, that adopt the behaviour hypotheses of the banks and allow for the development of the subsequent impacts on GDP, consumption and investment, and different spill-overs across the diverse economies. For instance, in Figure 4 illustrates the aggregate impact (estimated on 2013 data) on the CET1 capital ratio of a stress test in different cases: static, at variable trends, with dynamic managerial actions, with macroeconomic feedback. Figure 4 Macroeconomic feedback effects of the financial economy on the real economy in the STAMP System Source: ECB, publication STAMP feb. 2017, chapter 2 Another significant limit of microprudential EBA EU-wide stress testing is represented by the complete lack of controls on the so-called second round effects. An important source of second round effects that should not be over- ) 11 (

12 13 Integrated Dynamic Household Balance Sheet (IDHBS) model of the euro area household sector 14 Household Finance and Consumption Survey (HFCS) looked stems from the high degree of interconnection across banks, which may entail contagion effects able to amplify the initial effects of a financial shock. For instance, the insolvency of a bank may bring about losses for other strongly exposed financial intermediaries. To properly address the dependencies that interbank deposits and derivative contracts imply across banks, especially those of greater systemic proportions, representation models archetypal in the analysis of social networks have been developed (network analysis). Such models make it possible to know in advance, given the existing connection network, where and how the shock wave caused by the failure of a bank would have a more significant impact (financial contagion). A further extension of the simulations of microprudential stress testing is to consider the interactions between the banking sector and the rest of the economy, from households and enterprises to other non-bank financial institutions. In order to take into account in an increasingly realistic manner the second round effects, the ECB has introduced among its instruments new models that appraise such interactions. For instance, a household-specific modelling has been developed 13 based on the results of an array of sample surveys on household budgets 14 and extended to 15 European countries, to obtain a projection and an independent estimate of the demand for credit and of the probability of default of the private sector, specifically in the residential mortgage segment. The possibility of submitting household budgets to stress testing is especially relevant in macro prudential terms, to guide the supervisory authorities in imposing caps to indicators such as loan-to-value, debt-to-income and debt-service-to-income. The imposition of caps to such indicators in granting bank credit indeed affects the actual demand for mortgage loans by households, the level of debt of the economy and, accordingly, the entire evolution of the financial and economic cycle. Phenomena of diffusion of a crisis may also spring from the plain loss of value of equity securities issued by banks, such as shares or bonds, on the portfolios of other financial institutions. Specialised models have been developed for the simulation and stress testing of balance sheets of non-bank financial segments: insurance companies, pension funds and social fund schemes, investment funds and other financial institutions (that constitute the so-called shadow banking system ). ) 12 (

13 Lastly, explicit macro liquidity stress tests have been introduced that attempt to grasp in macro terms the two dimensions of the issue: the correlation between liquidity and solvency at the level of the individual bank and the repercussions of a liquidity crisis on the financial system in its entirety. Figure 5 displays an overview briefly summarising the various instruments making up the new system of the ECB because of the macro prudential extension of stress testing. Figure 5 Structure of the macroprudential extension of the EBA monitoring stress tests Source: ECB, publication STAMP Feb. 2017, chapter 3 Those responsible for the macroprudential tests of the ECB describe STAMP as an evolving, complex system, indicating its development and optimisation guidelines. This fascinating challenge shall pave the way for new paths and new questions. No doubt, it represents a step forward in understanding complex economic and financial systems. The American economist John Kenneth Galbraith, in the final part of his book A History of Economics (1987), on the topic of future developments in economic theory, predicted that in future: The distinction between microeconomics and macroeconomics will blur and disappear. With the evolution of the instruments and techniques of macroprudential stress testing, this time is fast approaching. 15 Alan Jay Perlis was an American computer scientist, pioneer in computer science and in the first programming languages Simplicity does not precede complexity, but follows it. Alan Jay Perlis ( ) 15 ) 13 ( ) 13 (

14 track record for ECB Stress Testing P rometeia is a global provider of consulting services and FinTech solutions focused on Risk & Performance Management. Since 1974, the firm supplies highly specialized advisory, analytical tools and research programs, integrating quantitative models, market and customer data, financial and economic scenarios. With 750+ industry experts, serves over 200 financial institutions in 20 different countries, through a consolidated network of foreign branches and subsidiaries located in Europe, Africa and the Middle East. With particular reference to the European market, collaborates with over 15 banking groups operating within the SSM regulatory space, in particular in the areas of: Asset Liability Management Liquidity Risk and Interest Rate Risk in the Banking Book Capital Management and Optimization Compliance with ICAAP and ILAAP requirements Regulatory Stress Testing In the specific field of Regulatory Stress Testing, the company has assisted a large numbers of clients in performing the exercises required by the European regulator in 2014 and 2016, leveraging the capabilities of ERMAS, the integrated SW platform developed by to support Holistic Balance Sheet Management. ERMAS v.5. has been enhanced to cover the extensive requirements set by the European Central Bank for integrated stress testing: the solution supports the projection of net interest income, REAs and loss provisions, under multiple scenarios and consistently with new IFRS9 principles. Source Systems & Data Feeds ERMAS Data Management ERMAS Analytical Modules ERMAS Reporting Source Systems Contractual Data Counterparty Data Market Data Default and Recovery Data Rating input Extraction Transformation - Loading Data Quality Assurance ADB Common data layer Financial Datamart Counterparty Operations Transactions credit risk repository Clients Collaterals Market data Financial statements Behavioural data Market data, etc. Credit Datamart Clients Exposures Collateral Ratings. Common Parameters and Rules Balance Sheet Risk Capital Management ALM & Market Risk Managerial Liquidity Regulatory Liquidity Enterprise Stress Testing IFRS9 Compliance FTP and Margin Analysis Risk Data Analytics / «in memory» reporting Dashboards Slice & Dice ) 14 (

15 About us We are a leading provider of consulting services and software solutions focused on Enterprise Risk & Performance Management Founded in 1974 as an independent institute for economic research by a group of young university professors in Bologna, in 1981 began offering analysis services to businesses and financial intermediaries. Since the 90s, the company s activities have focused increasingly on the integration of research, analysis, consultancy and software system development. This distinctive mix of services has made a leading European company in risk and wealth management solutions, business consulting and advisory services for institutional investors. collaborates closely with its clients to help them maximize their performance, be they banks, insurance companies, institutional investors, businesses, or public authorities. Our combination of tech proposition, quantitative advisory, training and economic research makes our business model unparalleled in today s market s approach to Enterprise Risk Management is based on the development of quantitative models and analysis methodologies. The production of highly specialized software applications leverages leading technologies, the knowledge of our subject matter experts and our ability to successfully respond to the growing demands of regulation. Our deep understanding of international markets, derived from proprietary economic research, adds a unique element to s business model and value proposition. ) 15 (

16 -> with over 750 industry experts -> more than 40 years of economic expertise -> we serve more than 200 financial institutions, including 50 primary banking groups in EMEA -> in 20 different countries -> through a network of branches in Europe, Russia, Turkey, the Middle East and Africa Our unique business model combines RISK SOLUTIONS RISK ADVISORY FINANCIAL AND ECONOMIC RESEARCH KNOWLEDGE TRAINING ) 16 (

17 Linkedin: prometeia YouTube: prometeia Facebook: prometeiagroup (0) ) 17 (