Suggested citation: FRAIT, Jan, ed. Systemic Risk in Post-Crisis Financial Markets. Praha: VSFS, 2019. SCIENCEpress. ISBN 978-80-7408-200-9. Edited by Jan Frait Systemic Risk in Post-Crisis Financial Markets Publisher Vysoká škola finanční a správní, a.s. (University of Finance and Administration) Edition SCIENCEpress, supervised by PhDr. Jan Emmer Estonská 500,101 00 Praha 10 Tel.:+420 210 088 862 www.vsfs.cz Reviewers: doc. Ing. Jaroslav Brada, Dr., University of Economics Prague doc. Ing. Zuzana Kučerová, Ph.D., Mendel University in Brno This publication was supported by The Czech Science Foundation with project no. 16-21506S New Sources of Systemic Risk in the Financial Markets. Publishing Editor Mgr. Petr Mach First edition, Prague, 2019 Print dům tisku, s.r.o. This publication has not undergone language editing, for the content and linguistic aspects of the text take responsibility the authors. © Vysoká škola finanční a správní, a.s., 2019 ISBN 978-80-7408-200-9 https://doi.Org/10.37355/03.2019/2 All rights reserved. No part of this publication maybe reproduced and used in electronic form, dubbed or recorded without the prior written permission of the publisher. Edited by Jan Frait SYSTEMIC RISK IN POST-CRISIS FINANCIAL MARKETS PRAGUE, 2019 Authors: Jan Frait, chapters 1, 2 (editor and corresponding author) University of Finance and Administration, Czech National Bank* jan.frait@cnb.cz Michal Bezvoda, chapter 7 University of Finance and Administration Petr Budinský, chapter 7 University of Finance and Administration Liběna Černohorská, chapter 4 University of Economics Prague Jaroslav Daňhel, chapter 9 University of Economics Prague Eva Ducháčková, chapter 9 University of Economics Prague Jan Hájek, chapter 2 European Central Bank, Czech National Bank* Hana Hejlova, chapter 3 Charles University Prague. Czech National Bank* Mojmír Helísek, chapter 6 University of Finance and Administration Narcisa Kadlčáková, chapter 5 Czech National Bank* Luboš Komárek, chapter 5 University of Finance and Administration, Czech National Bank* Zlatuše Komárkova, chapter 3 University of Finance and Administration, Czech National Bank* Miroslav Plašil, chapter 2 Czech National Bank* Jarmila Radová, chapter 9 University of Economics Prague Marek Rusnák, chapter 3 Charles University Prague, Czech National Bank* Bohumil Stádník chapter 8 University of Economics Prague Petr Teplý, chapter 4 University of Economics Prague * The views presented are those of the authors and not necessarily those of the Czech National Bank or other non-academic institutions. II Foreword The global financial crisis that started in 2007 represented a striking example of underestimating risks of systemic nature. Both the academic research and the financial market regulation had overly focused on individual institutions' risks in pre-crisis years. However, they had underestimated the risks across individual markets and financial institutions as well as their potential impact on asset prices. Moreover, risks were underestimated in terms of the macro-economically and financially stable economic environment, i.e. low-volatility and prosperous economic environment. This had resulted in strong risk appetite, and low risk aversion leading to overly low risk premiums, consequently causing optimistic expectations with regard to the future development of returns and general economic performance. Both international and national authorities responded to global financial crisis by plethora of regulatory changes. Systemic risk was acknowledged and macroprudential policies were instituted (for details see Frait et al., 2016). The contribution of this book is the proposal for the framework for the application of macroprudential policy tools in practice. It applies primarily to the countercyclical capital buffer (Chapter 2) and liquidity regulations (Chapter 3). Next, the book focuses on the new sources of systemic risks that emerged pro the policy responses to the global financial crisis aftereffects. It concerns primarily accommodative monetary policies of central banks. These new sources of systemic risks have become apparent only recently and new research on their substance and the ways of mitigation is needed (Chapters 1,4, 5, 6). And finally, although the regulatory and economic/political response to the crisis addressed number of issues, it likely failed to address the sources of systemic risk stemming from the non-banks. We start from the observation that the regulatory overhaul focused primarily on the banking sectors opening thus avenues for systemic risk in other sectors. We therefore investigate in to risks in insurances, pension funds and investment funds (Chapters 1, 7, 8 and 9). The book is structured as follows. Chapter 1 describes the sources of systemic risk associated with indebtedness of both private and public sectors and with cyclical patterns in provision of financial services. Chapter 2 focuses on the ways of detecting credit cycles and on approaches to setting countercyclical capital buffer in banking sector. Chapter 3 turns attention from credit risk to liquidity risk in banks and the approach to its testing. Chapter 4 describes developments of monetary policies in advanced economies after the crisis that contribute to sources of systemic risk. Chapter 5 looks at the risk of contagion in the foreign exchange market while Chapter 6 explores risks associated with the transition to fixed III exchange rate regimes. Chapter 7 investigates into the measurement of credit risk using the data from capital markets. Chapter 8 studies the behaviour of insurance-linked securities and Chapter 8 sums up the sources of systemic risk in insurance sector. This book can be of interest of researchers, university teachers, financial analysts and policy makers. It is the output of research activity supported by The Czech Science Foundation with project no. 16-21506S New Sources of Systemic Risk in the Financial Markets. Some chapters deliberately provide simplified and less technical presentation of research outputs so that we can address broader group of readers. We note that everything contained in this book represents their own views and not necessarily those of the institutions where they are employed. All errors and omissions remain entirely the fault of the authors. Jan Frait (editor) Prague, 1st December 2019 IV Content Foreword................................................................................................Ill Content....................................................................................................V Chapter 1 Debts, financial cycles and systemic risks............................................................1 1.1 Introduction....................................................................................................1 1.2 Low interest-rate environment and prospects of Japanisation........................3 1.3 Risks stemming from low interest rate environment......................................7 1.4 Debts and savings after the GFC..................................................................11 1.5 Conclusions..................................................................................................13 Chapter 2 Detecting credit cycle and setting countercyclical capital buffer.......................15 2.1 Introduction..................................................................................................15 2.2 CCyB essence and the BIS/ESRB guidelines...............................................16 2.3 Key indicators for the CCyB rate setting......................................................23 2.4 Deciding upon the CCyB rate.......................................................................25 2.5 Deciding upon the release of the CCyB........................................................30 2.6 Conclusion....................................................................................................31 Chapter 3 A liquidity risk stress-testing framework with Basel liquidity standards...........32 3.1 Introduction..................................................................................................32 3.2 Related literature...........................................................................................34 3.3 The concept of the approach.........................................................................34 3.4 Application of the model to selected Czech banks.......................................41 3.5 Conclusion....................................................................................................45 Appendix............................................................................................................47 Chapter 4 Monetary policy after the 2007-2009 global financial crisis in the context of systemic risk...............................................................................................48 4.1 Introduction..................................................................................................49 4.2 The basic principles of monetary policy before the GFC.............................50 4.3 The impact of the GFC on monetary policy.................................................51 4.4 The effectiveness of unconventional monetary policy in the selected countries.....................................................................................................53 4.5 Systemic risk in the context of current monetary policy...............................57 4.6 Conclusion....................................................................................................58 Chapter 5 Foreign exchange market contagion in Central European countries...................60 5.1 Introduction..................................................................................................60 5.2 Contagion and extreme value theory............................................................61 5.3 Exchange rate developments and crisis episodes in CEE.............................64 V 5.4 Loss absorbency in resolution as a major potential threat............................67 5.5 Empirical findings........................................................................................70 5.5.1 Unit root tests.............................................................................................70 5.5.2 Cointegration.............................................................................................71 5.5.3 Extreme value theory.................................................................................73 5.6 Conclusion....................................................................................................74 Chapter 6 Risks associated with the transition to fixed exchange rate regimes..................75 6.1 Introduction..................................................................................................75 6.2 Literature review...........................................................................................76 6.3 Fixed exchange rate and currency crisis.......................................................78 6.4 Currency crisis in ERM II - empirical experience........................................81 6.5 Theoretical models of currency crisis and their relevance in relation to ERM II.......................................................................................................84 6.6 Risk of currency crisis in ERM II.................................................................87 6.7 Involving the Czech koruna into ERM II......................................................90 6.8 Conclusion....................................................................................................92 Chapter 7 Measuring credit risk based on CDS and bond spreads......................................94 7.1 Introduction..................................................................................................94 7.2 Development of CDS....................................................................................96 7.3 Model of implied ratings...............................................................................97 7.4 Comparison of selected countries...............................................................101 7.5 Conclusion..................................................................................................106 Chapter 8 Insurance linked securities and their future research........................................107 8.1 Introduction................................................................................................107 8.2 Valuation oflLS.........................................................................................112 8.2.1 Calculations on 3D tree............................................................................113 8.2.1.1 Results on 3D tree.................................................................................114 8.2.2 Simulation on 3D Tree.............................................................................115 8.3. Valuation using 2 scenarious model..........................................................116 8.3.1 "Cyclone" Style Catastrophic Process.....................................................116 8.3.2 "Earthquake" Style Catastrophic Process................................................117 8.3.3 Earthquake + cyclone style components..................................................118 8.4 Possibilities of the future research..............................................................119 8.5 Conclusion..................................................................................................120 Chapter 9 Insurance and systemic risk..............................................................................121 9.1 Introduction................................................................................................121 9.2 Reasons for the varying resistance of banks and insurance companies to cyclical fluctuations..................................................................................121 9.2.1 Methodological differences.....................................................................121 VI 9.2.2 Differences in the character of business risks, consequences for the reduction of systemic risk by the insurance sector...................................123 9.2.3 Differences in information asymmetry....................................................126 9.3 The effectiveness of special built-in stabilizers to stabilize the economy of the insurance business and reduce the possible transfer of systemic risk............................................................................................................127 9.4 Current changes in the tax deductibility of a built-in stabilizer of insurance technical reserves in the current Czech environment...............129 9.5 Conclusions................................................................................................130 List of Abbreviations...........................................................................131 References............................................................................................133 List of tables.........................................................................................150 List of figures.......................................................................................151 Summary..............................................................................................153 VII VIII Chapter 1 Debts, financial cycles and systemic risks By Jan Frait In response to the global financial crisis (GFC) that started in 2007, both international and national authorities initiated number of regulatory changes. These were addressing primarily the risks generated in the banking sectors and to some extent at the insurance industry and capital market. In particular, the Basel Committee on Banking Supervision (BCBS) introduced number of reforms to the international framework for measuring and mitigating solvency, liquidity, and market risks. Besides regulatory changes for functioning of individual institutions, macroprudential policies were instituted to address systemic risks (for details see Frait et. al (2016). The GCF has had dire consequences for macroeconomic dynamics and stability of global economy, the advanced economies in particular. Weak demand, partially associated with high indebtedness in number of economies, contributed to strong disinflationary pressures. Central banks have responded by exceptionally accommodative policies that created environment of exceptionally low interest rates (Section 1.3). Despite it, economic activity in most advanced economies remained subdued and disinflation pressures persisted. This chapter deals with the potential of adopted policies to create potential sources of systemic risk. It also discusses the risk of Japanisation of European economy and its financial sector (Section 1.2). 1.1 Introduction Besides setting monetary policy rates to levels close to zero or even below, some central banks responded to post-GFC environment by quantitative easing that resulted in depressing long-term interest rates and yields of financial assets to exceptionally low levels (Chapter 5). Credit spreads and risk margins have been often depressed too. What originally appeared to be a temporary environment related to the necessary response to the GFC has become a longer-term structural factor. Liquidity injections, asset market purchases of public and private debt by central banks, exceptionally low nominal interest rates, and often negative real 1 interest rates, supported global liquidity generation and "search for yield" (Shin, 2013). The outcome has been strong demand for riskier financial assets, residential and commercial real estate. Higher demand for foreign assets in advanced economies enabled large nonfinancial companies from advanced and emerging economies to tap large funds through corporate bond issues. Easy access to, and the low cost of, loans for house purchase, coupled with expectations of continued growth in house prices, have created a potential for spiralling between property prices and loans for house purchase. All this contributed to spreading of systemic risk in global financial system. Figure 1-1 Historical development of world nominal interest rates (in %) tc-j-iiuMfi^-í-ii-í-í-^-o-^-j-c-^-ni-ů-vi-io-^-j-v; r-c-í-fl-vir-^^r^Tf^iN^Lň-ůíC^ — n ■» in r- x 5 - rT\ír--^:rifi/;^í:ř-'fj-MrJr-íĚ*- — — _ _ — _ _ _ _ _ _ _ _---________--__----_ _ _ _______ řjrj Bond butt market -Risk-fee rate, % -3YMA —8Y1LA Source: Schmelzing (2017)1 The understanding that the interest rates could stay very low for a long time globally owing to strongly accommodative monetary policies of the major central banks and other major central banks lead to intensive considerations how to combine monetary policy instruments with macroprudential policy tools so as to attain its price stability and financial stability objectives simultaneously. A fierce debate on the interaction of monetary and macroprudential policies erupted first in 2013 in connection with the accommodative monetary policy being pursued by the Federal Reserve, the ECB and the Bank of England coupled with a strong recovery in property markets and some financial market segments. Official commentaries were published on the contribution of the sustained easy monetary 1 This study provides a new and highly valuable dataset for the estimates of annual risk-free rate in both nominal and real terms going back to the 13th century. It also comes out with a bit speculative conclusion that there is a declining trend of nominal interest rates lasting centuries. The discussion of the conclusion goes beyond the scope of this chapter. 2 conditions to inflated prices of houses and some other assets, the greatly increased activity on the corporate bond market, inadequate risk assessment and the compression of yields on debt securities (BIS, 2014, p. 3). The prevailing conclusion of this debate was that the potential undesirable effects of easy monetary policy on the risks to financial stability could be largely mitigated by applying suitable macroprudential tools in good time. However, concerns were voiced that more aggressive use of such tools could neutralise the effects of accommodative monetary policy and foster deflationary pressures. The debate has not been resolved up now. 1.2 Low interest-rate environment and prospects of Japanisation Nominal interest rates went to historical low levels in recent years (Figure 1-2). As to the real interest rates, there where periods characterized by lower rates compared to recent years. These were, however, present in war-time or post-war periods (owing to high inflation) or during the oil-induced inflationary shock in 1970s. In other words, we currently have post-war interest environment without a war. Figure 1-2 Historical development of world real interest rates (in %) 1676 1886 1696 1906 1916 1926 1936 1946 1956 1966 1976 19S6 1996 2006 2016 — Short-term nsal rtrtes — Long-term real rates Source: Borio et al. (2017) Sharp decline in government bond yields occurred over summer 2019.2 The whole yield curve of German government bonds slipped to negative territory. Both government and private bonds equivalent to 18 bn. USD globally recorded negative yields. There were also plus expectations of short rates staying negative for a long time. All this lead to fears of Japanisation (or Japanification) of European economy and its financial sectors. Despite some increase in yields since such fears persist. According to popular and a bit misleading view "Japanification is the term economists use to describe the country's nearly 30-year battle against deflation and anaemic growth, characterised by extraordinary but ineffective monetary stimulus propelling bond yields lower even as debt burdens balloon" (Financial Times, 27 August 2019). 2 The ECB's September 2019 statements raised expectations of a continued policy of zero or negative monetary policy rates in the years ahead and increased quantitative easing. 3 Figure 1-3 Yield curve of German Figure 1-4 Stock of bonds with Bunds in August 2019 negative yields (in %) (in bn. USD) Source: ECB Source: ECB Our comparison of Japan (1980s onwards) and euro area (2000 onwards) presented below shows that there are some common trends as well as major differences. Japan went through major bubbly boom in 1980s followed by spectacular bust. Drastic downward correction of perceived wealth3 owing to asset prices collapse (Figures 1-5 and 1-7) contributed to decline of private sector demand. Firms and households responded to the shock by attempt to correct their balance sheet by saving more and lending less leading to balance sheet recession (a term popularized by economist Richard Koo). Monetary policy response initially cautions with no intention to restore previous asset prices levels and bailout investors. Highly accommodative fiscal policy maintained aggregate demand. 3 Wealth losses up to 2007 achieved 300% of initial yearly GDP. 4 Euro area enjoyed bubbly boom after 2003 followed by a major shock to some economies after 2008. Asset prices were in most countries affected partially and temporarily (Figure 1-6). Monetary policy responded to the shock vigorously, private investors were often supported by public policies. Aggregate demand management has been done primarily through monetary policy while fiscal policies involvement only selective. Figure 1-7 Stock prices in Japan Figure 1-8 Stock prices in EA 1984 1988 1993 1997 2002 2006 2010 2015 —NIKKEI 225 STOCK AVERAGE —NIKKEI 500 STOCK AVG BANKING (rhs) 6000 5000 3000 2000 600 2000 2002 2004 2006 2009 2011 2013 2015 2018 -EURO STOXX 50 ^—EURO STOXX BANKS E (rhs) Source: Bloomberg Source: Bloomberg Figure 1-9 Credit to private non-financial sector in Japan 1985 1 989 1 993 1 997 2001 20D5 2009 201 3 2017 -Total credit -Total credit-to-GDP Source: -Bank credit Bloomberg Figure 1-10 Credit to private non-financial sector in EA 2000 2002 2004 2006 200S 2010 2012 2014 2016 2016 -Total aetu -Total ae<»Uo-G0P — I^Hiik r.:rtsJil Source: Bloomberg In Japan, property prices as well as stock prices in Japan never got back close to boom levels (Figures 1-5 and 1-7). In euro area current property prices often exceed peak levels in most cases thanks, among other things, the ECB policy. Stock prices recovered quite well (Figures 1-5 and 1-7) with the exception of the 5 bank stocks that scored badly like in Japan. Japan experienced strong deleveraging with the non-financial firms' net borrowings negative for a long time (Figure 1-9). Deleveraging in euro area occurred in some countries only and was only mild (Figure 1-10). Japan's current consumer price level is only marginally higher than in 1990s owing to deflation in number of years. The euro area price level has been steadily growing, just slowly than implied by the ECB target. Figure 1-11 Nominal interest rates in Japan iL Figure 1-12 Nominal interest rates in EA 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015 2018 ■ Overnight "Long-term GB Source: Fred 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 ■ Overnight ■ EA long-term GB ■ DE long-term GB Source: Fred Figure 1-13 Long-term real interest rates in Japan 3,5 3 2,5 2 1,5 1 0,5 0 -0,5 iL Figure 1-14 Long-term real interest rates interest rates in EA 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015 201 Source: Fred lllllillJli.il!, 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 ■ EA long-term GB ■ DE long-term GB Source: Fred Note: Real interest rates based on CPI inflation, measured on ex post basis. What is striking that interest rates and government bond yields were positive or close to zero most of the time in Japan while in the euro area the interest rates and some government bond yields (such as German) turned negative earlier than in Japan Figures 1-11 and 1-12). And Japanese real long-term bond yields were positive even during deflationary period, and turned negative after the GFC only 6 (Figures 1-13 and 1-14). In euro area the real rates and some bond yields were particularly low even during the pre-GFC boom. It could thus be justified not to talk about Japanisation of the European economy, but about Euroisation of Japan economy. Unfortunately, despite all differences, both economies slipped to sustained slow output growth. The "low-for-long" scenario used in international institutions' analyses in previous years can thus continue to be materialising.4 1.3 Risks stemming from low interest rate environment The environment of low interest rates and yields is fostering a reduction in the risks stemming from the economic slowdown in the short term. In the medium term, however, the risks to financial stability - especially in the form of overvalued prices of market assets due to reduced risk premia and increasing indebtedness -are growing in this environment (Section 1.4). The low interest rates are creating favourable conditions for borrowers, as reflected in a drop in debt service costs. However, the low rates are simultaneously giving an impression of easy debt service and encouraging the acceptance of higher levels of debt. Both government and private sector debt are at historical highs in many countries. Significant debt growth has been recorded by emerging economies, China and selected euro area countries. High debt and a potential sizeable rise in losses on loans taken out in the optimistic phase of the cycle currently represent the main risk to global financial stability.5 Potential economic slowdown and drop in borrowers' incomes, and the related deterioration in their ability to repay accumulated debts, could be the primary trigger of the materialisation of this risk. An environment of very low interest rates can jeopardise the financial stability of individual financial market segments via two channels. The first is its negative effect on the profitability and, in turn, the resilience of financial institutions. The second is the resulting search for yield, reflected in investment in riskier assets, growth in leverage and concentration, more intense sectoral interconnectedness and hence greater vulnerability of the financial sector. Both channels may create - to a certain degree spurious - impulses for a shift from a banking-based financial system towards capital markets and to migration of financial activities into less regulated segments, which are generally more sensitive to market shocks. This interest rate environment - reflected in a flat yield curve - squeezes banks' profitability via a decline in net interest rate margins. This applies especially to countries in which interest rates on client deposits are zero or even negative. In such a situation, banks often cannot respond to a drop in interest rates on loans by lowering their deposit rates. For many European banks, the room for cutting funding costs will be limited by the obligation to acquire further eligible liabilities to comply with the MREL requirement and raise regulatory capital. 4 The ESRB's November 2016 report Macroprudential policy issues arising from low interest rates and structural changes in the EU financial system and the July 2018 analysis Financial stability implications of a prolonged period of low interest rates prepared by the Committee on the Global Financial System operating under the BIS in Basel. 5 IMF (October 2019): Global Financial Stability Report: Lower for Longer. 7 Banks can try to reduce their operating costs, but this has its limits. Another option is to increase the volume of remunerated assets. However, this is difficult to do in economies where sectors are so indebted that demand for further loans is weak. Banks can partially relax their credit standards and invest in more profitable but potentially riskier assets. However, holdings of such assets are significantly limited by regulatory measures. The lower incentive to write off problem loans has an indirect negative impact on banks' profitability in the longer term. Cheap financing allows banks to hold such loans in their balance sheets for longer, but at the expense of new and potentially more lucrative clients. Life insurance companies and pension funds providing defined benefit pension plans with guaranteed returns face similar risks. Their assets may have a shorter maturity than their liabilities, so a decline in interest rates increases the present value of liabilities more than that of assets. Specifically, this means that plans concluded 20 years ago, for example, were based on an assumption that government bond yields would fluctuate around 4%. If the environment of very low yields persists for an extended period, some providers of defined benefit pension plans may run into solvency problems, although more probably in the longer term. This risk is marginal for some European countries (such as the Czech Republic) but high for others. If it were to materialise, the problems could spill over to other sectors and cause a lack of confidence in the stability of the financial system as a whole. In addition, the current interest rate environment is discouraging financial institutions from providing products with guaranteed returns and generating incentives to transfer risks to clients. The low interest rates are forcing insurance companies and pension funds to make riskier investments. As higher-yield bonds mature, they are having to choose between safe assets with low yields and assets with higher yields but riskier profiles. The share of funds invested in property, especially commercial property, is rising in many countries. The increased property exposures, which can be observed in almost all sectors, also mean higher exposures to credit, market and concentration risks. Property exposures are sensitive to changes in economic activity, interest rates and market sentiment. They are often subject to an increased risk of price overvaluation and a subsequent marked correction. If institutions were to suffer substantial losses, this could create a need for support using public finances. The costs would be borne mainly by the younger generation. Investment funds are reacting to the decline in yields on safe assets by purchasing higher-yield investment-grade or speculative-grade corporate bonds and investing in alternative assets (property, private capital funds, funds investing in infrastructure assets etc.). The market liquidity of these assets is often low or uncertain. The share of highly liquid assets in portfolios is thus decreasing. In the event of a strong financial shock, many funds would probably have problems redeeming shares. Pressure on central banks to stabilise the situation could be expected to emerge. If investment funds were forced to react to requests to redeem shares by selling off assets on a larger scale, this could cause a drop in their prices and a spillover to other sectors. This risk also pertains to life insurance companies. 8 The existing recommendations for mitigating these risks are rather general. They focus mainly on strengthening the resilience of the riskiest segments through higher capitalisation and various types of buffers. In the case of banking, they concentrate on the use of macroprudential instruments to prevent a build-up of systemic risk. However, instruments for mitigating specific risks in the non-bank sector are mostly in the initial discussion phase. This is rather problematic since even after the GFC, the global financial system remained highly complex and interconnected. These qualities arise from mutual relations among individual entities of the given system that act as financial transaction counterparties. The complexity of the system translates into systemic risk, since interconnections within such system increases and individual effects may be mutually reinforcing. The substance of systemic risk consists in simultaneous adverse development of constituents within a financial system and subsequently within the real economy. The systemic risk arises from co-dependence6 of potentially feasible individual risks and/or idiosyncratic events of the system agents (financial institutions, markets, products, and infrastructures).7 6 Co-dependence in linear perception could be described as correlation; however, to consider co-dependence in linear perception would be inaccurate when talking about systemic risk. 7 Ceccheti et al. (2010) liken systemic risk to (negative) externality, such as pollution, as it occurs due to activities of certain entities and is transferred to other entities. It may take two forms: (i) joint bankruptcy of institutions within a certain period of time, due to their joint exposure to the same risks or due to mutual ties of intermediaries; (ii) procyclicality - it can be described as mutual interaction of the real economy and the financial system, multiplying each other, thereby contributing to cyclicality (boom and bust cycles) -potentially endangering stability of both the financial sector and the real economy. 9 Figure 1-15 Non-financial sector debt after 2000 (% of GDP) Figure 1-16 Changes in non-financial sector debt (2008-2018, p.p.) 2003 2006 -All countries •Emerging economies ■Euro area 2012 2015 201 -Advanced economies -USA Source: BIS Note: The solid lines denote total debt (including the government) and the dashed lines private non-financial sector debt. Figure 1-17 Total general government debt (in % of GDP) 120 110 100 90 80 70 60 50 40 30 20 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 -All -AEs -EMs -G20 US -EA Source: BIS Note: Calculated indirectly as difference between total NF All countries Advanced Emerging economies economies ■ Total debt USA Euro area Private sector debt Source: BIS Note: The data are for 43 countries covered by credit statistics available on the BIS website. Figure 1-18 Changes in general gov. debt levels (in p.p.) AEs EMs ■ 2008-2000 G20 US 2018 vs. 2008 Source: BIS Note: Calculated indirectly as difference between total NF The financial system structure is not static constant, but it changes and forms over time - also as a consequence of financial innovations. There surely could be some welcome innovations even in the financial sector. Nevertheless, one can also find plenty of bad innovations in the financial industry in this and previous centuries. One particular area of systemic importance is mortgage financing. Dubious approaches like interest only schemes, deferred payment of interest and principle, extensive stretching of maturities and even not amortized mortgages. Their final outcome was that people that were buying houses and flats were taking much higher loans leading to high indebtedness and vulnerability to increase in 10 debt-servicing costs. Altogether it created systemic risk with dire consequences in some countries. The high indebtedness of private sector, especially households, may have created a very challenging environment for monetary policy or other policies, or these policies even could be jointly trapped. 1.4 Debts and savings after the GFC It was assumed that accommodative monetary policies would enable the private sector in particular to reduce its debt, which was regarded as one of the major causes of the crisis. Ten years on, we can say that no overall decrease in debt has been observed in advanced or emerging economies (Figures 1-15 and 1-16).8 Advanced economies have mostly seen an increase in government debt, which in most of the countries has exceeded the stagnation or decline in private debt (Figures 1-17 and 1-18). In emerging economies, private sector debt has increased significantly. A reversal of the rising debt trend has occurred in recent years owing to higher economic activity being reflected in faster nominal income growth. Figure 1-19 Changes in the debt of households and non-financial corporations (2008-2018) All countries Advanced Emerging economies economies ■ Non-financial corporations ■ Households Source: BIS Note: The data are for 43 countries covered by credit statistics available on the BIS website. Figure 1-20 Changes in the debt and debt service of the private non-financial sector (2008-2018) Source: BIS Note: The data are for 32 countries covered by debt service statistics available on the BIS website. Debt and debt service are expressed as a percentage of GDP. The Czech Republic is plotted in red. The debt trends have differed substantially across the countries monitored and between the sectors of households and non-financial corporations. In many advanced countries, including the USA, write-offs of non-performing loans after the global crisis, falling property prices and macroprudential measures focused on mortgage loans have been reflected in a decline in household debt (Figure 1-19). The data are from credit statistics available on the BIS website https://www.bis.org/ statistics/ about credit stats.htm. 11 By contrast, cheap and available funding has motivated the corporate sectors in many countries to increase their leverage (Figure 1-19). In addition, private sector growth was replaced by public sector debt (Figures 1-17 and 1-18). Differently from the pre-GFC decade, the rise in private non-financial sector debt has often not been accompanied by a rise in debt service (Figure 1-20). On the contrary, debt service has fallen significantly in some of these countries (Figure 1-20, bottom-right quadrant). The main reason is a decrease in interest rates. The low level of real interest rates and associated rise in debt levels is often explained by global savings glut (Bernanke, 2005). Should it be the case, the recent developments should not be viewed as an important source of risk. The data on household savings are generally available with some delay and are more reliable that the data on corporate savings. One of the reasons is that the household savings are typically the main domestic source of funds to finance capital investment, which is a major impetus for long-term economic growth. Still, saving rate indicators have to be assessed with caution since they often have residual features, may not be comparable across economies and are subject to major revisions. Looking at the data on households' gross rate of saving from disposable income (the portion of their disposable income that is not used for consumption), the European countries do not fit well into the story above (Figure 1-21). Figure 1-21 Gross saving rates of households in EU countries (% of gross disposable income) 25 -i ■ 2005 "2009 «2017 Source: Eurostat The decrease in saving rates in many countries in recent years can be considered risky from the long-term perspective. Household savings are one of the main sources of financial investment in the real economy, which is a key factor of 12 economic growth. Besides their wealth, households generally use a substantial portion of their savings to buy and maintain owner-occupied housing. In particular, if house prices rise faster than households' income, the investment rate in the household sector may exceed the saving rate for a time. In such a case, households draw on the savings they have accumulated in the past, the savings of other sectors, or foreign sources. In cases where households invest largely in "overpriced" properties, inefficient utilisation of savings may occur. This, in turn, is usually reflected in sharp macroeconomic volatility and structural distortions in the economies concerned. Correctly conducted macroprudential policy helps to prevent such episodes from becoming excessive. 1.5 Conclusions The response of most central banks in advanced economies to weak demand and disinflationary pressures after the GFC has created the environment of ultra-low interest rates. The side effect of this environment consists of the growth of asset prices and creation of incentives to invest into real estate and other risky classes of both real and financial assets. Asset price boom associated with increase in private and public sectors' debt could under some circumstances lead to bust ending up in massive loss of perceived wealth, balance-sheet recession and further drop in demand. Aggressive monetary policies attempting to avoid Japanese-lie scenario may thus at the end produce it. Having this in mind, there is growing reliance of advance economies on the active use of macroprudential policies. Some countries set relatively high countercyclical buffers and use the LTV, DTI and DSTI limits for limiting the risks associated with mortgages. The capital add-ons for systemically important banks have also become standard. Central banks have price stability and financial stability objectives. For meeting them they have monetary policy and macro prudential policy tools at their disposal. Number of people in the central banking community and even more in academia think that each policy should be used fully separately to meet its specific objective. Nevertheless, once we start to think about transmission mechanisms of these policies, we will likely conclude that these policies are not independent, they are interlinked. Anything that affects the availability and price of credit (or assets in general) also affects the growth rate of these assets, their quality and profitability. Changes to both monetary policy tools and macroprudential tools act via channels working through credit supply and demand, the risk-taking of economic agents, the asset prices, the perceived actual and expected bank credit risk or the banks' profitability. In some situations, the two policies can come into conflict because of a need for them to work in opposite directions, while in other situations it may be desirable for them to act in the same direction. Being aware of such connection, some central banks such as the Czech National Bank decided to set a framework for coordination of these policies (Frait et al., 2014). However, there are some other situations in which it is not that easy to coordinate and even find a proper mix. It is because the mix depends on the properties of two cycles, the financial cycle and business cycle. And because these cycles are rather different, sometimes 13 is very difficult to tell what is the best approach or least problematic approach (Malovaná and Frait, 2017). The typical situation of this sort is right now in Europe. The prevailing view in the central banking community these days is that ultra loose monetary policies are necessary. As they produce risks, the macroprudential policy tools should do the job, prevent excessive credit growth and asset prices misalignments. The calls for a more holistic approach were played down for years. However, under the volatility and uncertainties that emerged in 2019, this view has become to be adopted more generally. The complicating factor for number of economies is the high level of private and public debt. Very low interest rates help to keep debt servicing cost easily manageable. However, even relatively small increase in the level of lending rates could lead to much higher default rates, decline of consumer lending and disinflationary pressures. This also means that the protracted period of low interest rates could be self-enforcing. If this environment enables emergence of high debt, the central banks will get under pressure to keep policy rates low, because otherwise their economies would find at a risk. Also, in countries with high level of debts, macroprudential policy could become unwittingly a substitute for monetary policy. And if the central bank is not the macroprudential authority, and if such authority is constrained by political considerations, a trap may emerge. And finally, the macroprudential policies could mitigate some risks, not all. Current macroprudential tools are developed for building buffers in banks and limiting residential real estate risks. Tools for mitigating the commercial real estate risks or the risks from other asset classes are untested or non-existent. Tools for life insurances, pension funds and investments funds are at the initial stage or developments. Further research in these areas is therefore much needed. 14 Chapter 2 Detecting credit cycle and setting countercyclical capital buffer By Jan Frait, Jan Hájek and Miroslav Plašil In the aftermath of the global financial crisis, the importance of the objective of financial stability across the world increased dramatically. Besides increased interest in financial stability analyses, the whole institutional framework of maintaining financial stability was strengthened by instituting the macroprudential policy. The main aim of this policy is to mitigate systemic risk, i.e. the risk of instability of the financial system as a whole. Basel III regulatory framework established as one of the key macroprudential instruments in the banking sector a countercyclical capital buffer. This instrument is designed to reduce the consequences of worsened access of firms and households to banking credit in bad times. This chapter proposes comprehensive approach to the countercyclical capital buffer using the experience of the Czech National Bank. It describes its decision-making process from assessing the position of the economy in the financial cycle through detailed analysis of particular risks to setting the buffer rate. The approach that can be labelled discretion guided by multiple-factor analysis builds upon the signals from both individual and composite indicators of financial cycle and systemic risk. The chapter then describes the factors that the macroprudential authority takes into account when setting the specific countercyclical capital buffer rate. 2.1 Introduction The countercyclical capital buffer (CCyB) is a pure macroprudential policy tool. It is designed to protect the banking sector against risks arising from its behaviour through the financial cycle, and in particular from excessive credit growth, which generates systemic risks and increases the potential for sharp swings in economic activity. A macroprudential policy authority should ensure that banks create a capital buffer during the financial expansion to enable them to absorb losses in the event of an adverse shock accompanied by elevated financial stress and growth in loan defaults. Use of the buffer at such a time should prevent 15 a fall in the supply of credit to the sound part of the economy and stop the shock spreading from the financial sector to the real economy and causing the banking sector further losses. At first glance, the CCyB is a very simple tool. In reality, though, setting the CCyB rate is a complex task in terms of both decision-making and communication. It can be particularly difficult to justify the specific level at which it is set. This chapter aims to present key aspects of the CNB's approach to setting the CCyB rate, contribute to better formation of expectations about the future path of the rate and thereby facilitate capital planning for credit institutions. The chapter is structured as follows. Section 2.2 summarises the essence and purpose of the CCyB, describes the BCBS/ESRB methodology and points out some issues with its application to the Czech economy. Section 2.3 introduces the main indicators used to determine the position of the economy in the financial cycle. Section 2.4 details the CNB's approach to setting the CCyB rate and discusses its decisionmaking process, which draws on stress test results and known facts about the morphology of the financial cycle. Section 2.5 briefly discusses the approach to releasing the CCyB. The section 2.6 concludes. 2.2 CCyB essence and the BIS/ESRB guidelines The recent financial crisis revealed that stress in the financial sector can easily spread to other sectors of the economy. Faced with capital shortages due to losses, banks in some countries severely curtailed the supply of credit even to sound non-financial corporations (a situation generally referred to as a "credit crunch"). In response to these funding constraints, some firms had to cut their production substantially. This led to rising unemployment, falling household incomes and, in turn, to a deepening recession. Inadequate capital creation by banks in the upward phase of the financial cycle was thus reflected in a downward spiral where falling aggregate demand due to difficulties in raising funds for viable projects led to further credit losses and further lending constraints. In some countries, public money had to be used to resolve the crisis in the banking sector. This was reflected in growth in long-term interest rates and also adversely affected the real economy. To avoid a repeat of the spill-over effects of such shocks from the financial sector to the real economy, a countercyclical capital buffer (CCyB) has been incorporated into the macroprudential policy toolkit (BCBS, 2010). The CCyB is aimed at "protecting" banks against excessive impacts of the financial cycle, which banks themselves are involved in creating. In the spirit of this regulation, banks are meant to set aside a sufficient buffer in good times - characterised by rapid credit growth accompanied by relaxation of credit standards and growth in property prices - to cover losses arising from the switch to the downward phase of the financial cycle. 16 Figure 2-1 Countries with non-zero CCyB rates (% of total risk exposure, October 2019) 3,0 i 2,5 - SE NO HK IS SK CZ DK UK IE LT BG FR LV BE DE LU ■ 10/2017- Pending rate ■ 10/2017-Applicable CCyB rate ■ 10/2019- Pending rate ■ 10/2019-Applicable CCyB rate Source: BCBS, ESRB The buffer should be released when risk materialises, so banks should be able to apply a reduced capital requirement to maintain the supply of credit to the sound part of the real economy. As adverse shocks can occur unexpectedly, the macroprudential authority can set a new CCyB rate with immediate effect when deciding to release the buffer.9 The addition of a CCyB rate to the overall capital requirement may help tame credit growth in the expansionary phase of the financial cycle; however, this can be regarded only as a positive side-effect of the CCyB and is not the main purpose of creating the buffer. The primary objective is still to boost the banking sector's resilience to adverse shocks at times of financial instability and to ensure smooth funding of the real economy through the financial cycle. 9 There is no clear consensus across the economic community on whether the creation of a capital buffer will give rise to a reduction in the supply of credit by banks. Financial sector representatives often assert that higher capital requirements lead to a decrease in the supply of loans (see Admati et al., 2011). Based on an analysis of data for advanced countries, however, Gambacorta and Shin (2016) find that better capitalised banks have lower funding costs and are capable - especially in worse times - of lending more to the economy than banks with lower capitalisations. For that reason, efforts to constrain credit growth should not be the main motivating factor in CCyB rate decisions. 17 The CCyB is a new macroprudential tool and there is limited experience with its use so far. A limited number of countries have already set non-zero CCyB rate (Figure 2-1). A universally shared approach to the introduction of non-zero CCyB rates and the setting of their specific level has yet to emerge in the international regulatory community. Some macroprudential authorities view the CCyB as a tool that should only be applied in a strongly expansionary phase of the financial cycle when systemic risks are already clearly visible and tangible. Other macroprudential authorities prefer a more prudent approach in which the CCyB should be created right at the start of a credit recovery or at a certain level even in the neutral phase of the cycle.10 Such approach is applied by the CNB as well. It repeatedly communicated that it was desirable to set a non-zero CCyB rate when cyclical financial risks are still close to their usual, standard levels and have not yet become significantly elevated. The aim of the standard rate concept is to ensure that the banking sector's resilience starts to be supported in a timely manner after the acute phase of a cyclical contraction, or even a financial crisis, has subsided. The CNB's detailed approach to setting and calibrating the standard CCyB rate is described in Plasil (2019). The basic framework for applying the CCyB was formulated by the Basel Committee on Banking Supervision (BCBS) and subsequently introduced into EU regulatory practice through the CRD IV directive and its transposition into the Member States' national legislation. The European Systemic Risk Board (ESRB) further developed the core principles of the original framework in the form of a recommendation (ESRB, 2014). From the operational macroprudential policymaking perspective, though, the BCBS/ESRB methodology still represents only a very rough guide to when to introduce a buffer rate and what rate to set. For this reason, it needs to be further elaborated and tailored to the specifics of each national financial sector. The BCBS/ESRB methodology can be summarised into four main steps (see the dark blue boxes in Figure 2-2). The first involves determining the deviation of the credit-to-GDP ratio from its long-term trend using the Hodrick-Prescott (HP) filter and then using that gap to set a so-called benchmark buffer rate. In the BCBS/ESRB methodology, this rate serves as a guide for setting the CCyB rate.11 10 An example might be the approach adopted in the UK. The local macroprudential authority (the Financial Policy Committee, FPC) reported that under normal conditions, when systemic risks are neither depressed nor elevated, the FPC will hold a rate of 1%. The FPC intends to adjust the buffer rate gradually to minimize the potential economic costs. By doing so, the FPC aims to reduce the probability of systemic risk rise to dangerous levels on one hand and on the other to allow banks to provide loans to households as well as businesses smoothly without major setbacks. 11 Total credit comprises total loans to the private non-financial sector (households, non-financial corporations and non-profit institutions serving households) plus debt securities issued. The recommended smoothing parameter for the HP filter, X, is 400,000. The benchmark buffer rate is 0% of risk-weighted assets if the gap is less than or equal to 2 pp and is greater than zero if the gap is larger than 2 pp. The equation used to calculate 18 EU Member States are required publish a credit-to-GDP gap and a benchmark buffer rate quarterly every time they set a CCyB rate. However, they are given discretion to calculate the CCyB guide rate using a different method not necessarily based on the BCBS methodology (see the light blue boxes in Figure 2-2). Figure 2-2 The logic of the BCBS/ESRB regulatory framework for setting the CCyB rate Credit-to-GDP gap (BCBS methodology) Reference rate (Rec. B3(a), BCBS calculation) Benchmark rate (Rec. B3(b), nat. calculation) Additional credit-to-GDP gap (nat. methodology) Benchmark rate (Rec. B3(c), BCBS or nat. calculation) Guide rate CCyB rate Discretion T J Note: Dark blue boxes indicate mandatory elements and light blue boxes voluntary elements of the ESRB (2014) methodology for setting the CCyB rate. Source: BCBS, 2010, ESRB, 2014 This discretion is allowed because the original BCBS methodology would produce incorrect recommendations in many countries if applied mechanically (see, for example, Gersl and Seidler, 2011). This is true for the Czech Republic, where the use of this methodology would signal the need for non-zero CCyB rate from the start of the global financial crisis. This could be considered as dubious result as simple economic logic would suggest releasing the hypothetical buffer in case the crisis emerges (especially crisis of the extent of the one that spread out in 2008). A significantly non-zero benchmark buffer rate would also hold from 2011 Q2 and the maximum rate of 2.5% in 2013 Q2 (see the solid red line in Figure 2-3). During 2013, however, loans recorded only weak growth, property prices the rate on the basis of the gaps is: benchmark buffer rate = 0.3125*(gap) - 0.625. The benchmark buffer rate is 2.5% if the gap is greater than or equal to 10 pp. The resulting benchmark buffer rate should be calibrated in steps of 0.25 pp or multiples thereof. 19 continued to fall in year-on-year terms (as they had done since 2009 Ql) and credit standards were tightened further. These conditions can hardly be interpreted as an expansionary phase of the financial cycle. The main sources of the misleading results of applying the BCBS/ESRB methodology in the Czech economy are structural breaks in the time series related to the 1990s banking crisis, when bad loans were written off from banks' balance sheets. In addition, during early 2000s there was a structural change in the composition of credit to the private sector. Prior to period of the Czech banking crisis the main driver of the credit growth was credit to non-financial corporations whereas the main driver since then has been credit to households or related to housing. In the case of other countries, misleading results would stem from different factors. ESRB provides the estimates of the credit-to-GDP gap data for all EU countries through ESRB Risk Dashboard since 2012. For the whole period, negative credit-to-GDP gap is estimated for most economies. The data in Figure 2-3 from December 2019 suggest that positive and significant gap applies for France, Germany, Sweden and Croatia only. Nevertheless, fast credit growth and expansionary stage of financial cycle are identified by authorities in number or countries including the Czech Republic. The credit-to-GDP gap thus provides a poor indication of desirable setting of the CCyB rate. Figure 2-3 The credit-to-GDP gap and the hypothetical CCyB rate under the national and BCBS methodologies (CNB) tflfl I ■ 1 I J A I I L.....• I t • I I I I 1 I 1 1fl(] rn k sc mr cz lt ro at sk ee n pi m it uk dg lv si bc hi hu gr dk pf cs lu ic cy Source: ESRB Risk Dashboard, December 2019 Shifting from data-related issues, the use of the TIP filter to determine trends in macroeconomic variables has some drawbacks as well. To begin with, a time series trend obtained by the HP filter is dependent to a significant extent on the length of the chosen time series and the calculation is very sensitive to the smoothing parameter (lambda). A big problem as regards practical application in macroprudential policy is the end-point bias, which generates a highly unreliable 20 estimate of the trend at the end of the data period.12 Macroprudential policy, which, by contrast, requires assessment of the trend on the basis of current (i.e. end-of-period) data, would therefore be reliant on indicators subject to a high degree of uncertainty. Moreover, in case of some economies characterized by relatively short time series (e.g. post-transition countries), the credit growth is automatically incorporated into the trend by the HP filter (Cottarelli et al., 2005). This implies the method's inability to take into account economic fundamentals which affect the equilibrium stock of loans. There are other shortcomings from perspective of practical use. These are caused by unavoidable nature of the credit-to-GDP ratio because it fluctuates due to changes of both credit and GDP. In this sense, it might be quite difficult to interpret the underlying changes correctly. There are also considerable lags between changes in economic conditions and leverage adjustments. Another aspect is the filtering of credit-to-GDP may not be fully appropriate for converging economies or economies that previously experienced long-lasting credit boom. By filtering the trend in context of these economies, it is impossible to disentangle between financial deepening and credit bubble which is significant impediment. The ESRB (2014) recommendation takes such cases into consideration and allows the gap calculation to be tailored partially to the specifics of the national economy. In line with this, the CNB calculates additional gaps that may be more appropriate for macroprudential decision-making (Hájek et al., 2017). One of these is a credit-to-GDP gap based on a shorter time series excluding the structural break that occurred in the 1990s. Another is based on the ratio of bank loans to GDP and disregards other sources of credit financing (unlike the BCBS/ESRB methodology). Restricting the calculation to bank loans is logical since the CCyB is a tool targeted at the banking sector and at ensuring stable bank lending. In addition to gaps calculated using the HP filter, the authority can apply an alternative method for determining the deviation from the trend which eliminates some of the known issues with the said filtration technique. This method is based on analysis of local extremes in the time series.13 This eliminates the problem of removal of old loans from banks' balance sheets after the late-1990s crisis and (unlike the HP filter) does not lead to changes in the trend estimate as new observations come in. The corresponding gap (referred to as the expansionary credit gap) is very different from the original signal and much closer to the true course of the financial cycle (see the solid blue line in Figure 2-4). One way of dealing with end-point bias is to extend the time series into the future by means of prediction. This, however, can introduce further uncertainty into the estimate linked with the quality of the prediction. To reveal extremes indicating credit expansion, the CNB uses the difference between the present value of the ratio and the minimum value achieved in the past eight quarters. Other time periods were tested but the results remained robust. This analysis is loosely inspired by the definition of the cycle proposed in Bums and Mitchell (1946) and by the unemployment recession gap (Stock and Watson, 2010). 21 Figure 2-4 The credit-to-GDP gap and the hypothetical CCyB rate under the national and BCBS methodologies 09/06 09/08 09/10 09/12 09/14 09/16 ■ Deviation of credit-to-GDP ratio from trend (BCBS met.) ■ Deviation of credit-to-GDP ratio from trend (nat. met.) - Reference rate (Ree. B3(a), BCBS met., right axis) - Reference rate (Ree. B3(c), nat. met., right axis) (deviation in pp; right axis: rate in % of RWA) Note: The trend in the BCBS methodology is estimated using the HP filter, lambda = 400,000. The trend in the national methodology is estimated by analysis of local extremes. Source: CNB Regardless of the estimation technique, however, the credit-to-GDP gap can be only an initial guide to the position of the economy in the financial cycle. Although the credit-to-GDP ratio provides valuable information about indebtedness of the domestic private non-financial sector, there are several limitations that should be taken into account when drawing out conclusions for setting the CCyB rate. Probably the most important one is that credit-to-GDP is a lagging slow-motion variable staying above the historical norms during the initial stages of crisis. For example, a rapid decline in GDP during a recession increases the credit-to-GDP ratio and may indicate an excessive borrowing phase purely as a result of a more persistent credit cycle.14 The credit-to-GDP ratio is therefore only a very rough measure of leverage in the economy, on the basis of which it is hard to identify turning points between phases of the financial cycle in a timely manner (for more details, see Frait and Komárkova, 2012, pp. 14 and 22). The problem is partly mitigated if potential GDP, which is more stable, is used to calculate the credit indicator. However, the results for the Czech Republic are little changed in terms of identifying periods of excessive credit growth compared to the traditional calculation. For further details see Gersl and Seidler (2011). 22 2.3 Key indicators for the CCyB rate setting For the reasons given in the previous section, the recommendation of the ESRB (2014) requires national authorities to take into account other variables indicating excessive credit growth and the build-up of system-wide risk when setting the CCyB rate. To this end, the macroprudential authority should use a wider range of indicators (including composite and simple indicators of financial cycle, credit dynamics and systemic risk changes) in order to answer several layers of fundamental questions. These questions investigate (i) whether the debt in the economy is too high, (ii) where we stand in the credit cycle, (iii) why amount of credit changes and (iv) what the appropriate CCyB rate is currently and what can be expected in the future. To begin with, the question whether the debt in particular economy is too high or not, a cross-country comparison of combined debts of non-financial corporations and households could help. In such comparison, the Czech economy seems as the fourth least indebted country out of all EU member states, only Estonia, Lithuania and Romania has less indebted private sector. From a different perspective, the Czech Republic has lower level of credit-to-GDP than most advanced economies when compared at similar level of economic development measured by GDP per capita (Figure 2-5). Figure 2-5 Credit-to-GDP for similar level of economic development (GDP per capita, PPP current international USD, CZ 2015 = 32.000 USD) 200 150 100 50 & NetOm%t. (4) In the second case, where the bank is hit more seriously by a wave of shocks, its liquidity asset buffer is not sufficient to cover the net outflow in the given maturity band (equation 5) and the liquidity indicator for that maturity band is smaller than 1. In such a situation, the bank's reaction is equal to the liquidity asset buffer. The entire liquidity buffer is exhausted, i.e. the bank has a deficit liquidity position26: flgti = LRbQti, if LRbQti < NetOUT%t. (5) The result of the bank's reaction is that a stock of unencumbered assets included in the liquid asset buffer will be reduced. On the one hand, the reaction may mitigate the impact of the shock on balance-sheet liquidity, but on the other it increases each reacting bank's reputational risk as well as raising systemic risk via the simultaneous reaction of banks on financial markets. Systemic risk rises if banks exert excessive unilateral pressure on the financial market (for example, if all banks try to sell the same type of bond), leading to a fall in market liquidity. Reputational risk consists in the signalling of problems with a bank's liquidity. The growth in these two risks then feeds back in the form of a second-round shock to banks' balance sheets. The third step therefore involves calculating and applying the feedback effect in the form of an additional market shock caused by banks' reactions. This endogenous systemic shock manifests itself as an additional haircut on the asset (q) held in the liquidity buffer. We differentiate between the impact of systemic risk on non-reacting banks (qbnon) and that of systemic risk plus reputational risk on reacting banks (qbreacy In reality, the bank may first try to sell off or pledge lower quality assets even though they are subject to large market haircuts. The assumption of minimum transaction losses was chosen because the presented approach is aimed at testing the adequacy of a bank's liquidity buffer in relation to the maturity mismatch in its balance sheet. The liquidity position can be improved by accepting a short-term loan from another bank. Such "assistance" is not considered in the test given the assumption of a limit on the increase in funds. This does not apply to banks in a liquidity subgroup. 39 %i%b RQti qbnom = h*b * (£bB)\ I, where t = 1,2,3,4 (6) where q£ (h*,l) and h* reflects the market liquidity risk associated with the asset (see below), s is a market conditions indicator and B is a parameter equal to one if the bank is a reacting bank and zero if it is a non-reacting bank. A non-reacting bank is the bank that closes the gap between outflows and inflows by using cash or claim on the central bank. Those two liquid assets are not subject to haircuts and a usage of them does have no impact on markets. For parameter h*, the model uses one of three haircuts: the original haircut applied in the previous round of the test (h)27 The size of the additional haircut depends on the number of reacting banks (Zb B) banks and the size and similarity of their reaction (X& ). It is assumed that a larger number of similarly reacting banks causes greater market stress and hence a larger additional market shock. The market conditions indicator (s) in the model expresses risk aversion. This indicator is derived from the standardised distribution of risk aversion indicators using implied stock price volatility and bond spreads as proxies (Van den End, 2008). The indicator takes values in the range of (-1,1) in normal market conditions and up to 3 at times of high market stress. A higher market stress indicator magnifies the effect of the simultaneous reaction of banks. It is set by expert judgement based on knowledge of volatility and liquidity in the market concerned. Reacting banks face reputational as well as systemic risk. In their case, the additional haircut is thus larger. This type of risk (like systemic risk) is expressed using a market conditions indicator, since the signalling effect of reacting banks has a large feedback effect in the event of market stress. qbQrtiac = <4ni°mV^,where t = 1,2,3,4 (7) In a crisis, illiquid financial institutions - due to either prudential (liquidity-hoarding) or speculative (predatory)28 motives - are driven out of private credit markets or are granted liquidity at punitive rates. It is assumed in the methodology that the impacts of the shocks applied to the first maturity band and the subsequent reactions of banks will pass through to connected bands in the individual steps of the test (Q - 2, 3, 4). Here again, we consider an exogenous wave of shocks that affects the value of the assets held in the liquidity asset buffer and the size of the If h is zero, the haircut on government bond, then we use the haircut applied to the asset type in the NSFR requirement, see BCBS, 2014, p. 11. This is a speculative motive based on the assumption that high demand for cash implies low asset prices. In a crisis, when some banks are in a difficult liquidity situation, liquid banks may use their market strength and curb the provision of liquidity to illiquid banks or raise the price of that liquidity for purely strategic, healthy competitive reasons. If loan rates are too high, an illiquid bank is forced to sell off its assets, often at very attractive prices (i.e. it falls prey to predators). 40 liquidity flows via h, r and p. Additionally, however, we take into account the market stress caused by reacting banks (q).29 The liquidity indicator thus changes as follows: nt = Zt^itd-hSti-ggt-it) here = x 2 3 4 (g) yt iVetOUT^ v y It is clear that the model has limitations that prevent it from fully capturing the liquidity risk that a tested banking sector may face. For instance, it fails to take in consideration that the provision and repayment of loans are closely bound up with the creation and termination of deposits. In the test, the liquidity position of banks is improved by loan repayments (inflow) but no longer shows up as deposit termination (outflow). The model also fails to take account of direct interbank contagion, an interaction with non-bank financial intermediaries and hence the potential domino effect. The scenario considers only a simplified general interest rate shock based on the evolution of government yield curves, and only in two currencies. Specific interest rate risk is captured only endogenously through banks' reaction functions. Exchange rate risk and real estate risk are not considered at all. The model does not distinguish the type of credit and liquid lines in relation to the counterparty, i.e. it does not work with intragroup liquidity lines. And the more relevant limitation is that the model takes into account only one type of banking reaction and does not work with a banking reaction through changes in interest rates for example. The liquidity stress test should be further refinement in these areas. 3.4 Application of the model to selected Czech banks The methodology described above was applied to a representative sample of 19 banks domiciled in the Czech Republic, with various business models and bank sizes represented. The main objective was to monitor the sensitivity of the liquidity position of selected banks to a combination of shocks under the given methodology. The application was conducted on end-2016 data for the banks under review. The data was obtained from the CNB's statistics. The CNB's November 2016 macro-stress scenario and macro-stress test results (CNB, 2016b) were used to simulate the bulk of the exogenous shocks. The parameters of the exogenous shocks are summarised in Table 3-1. The parameters of the shocks, including the endogenous ones, are summarised in Table 3-2 in the Appendix. We opted for a single market indicator (s) of 1.5, implying low market liquidity (van den End, 2008). The additional haircut is applied to available-for-sale assets in the portfolio. In the case of held-to-maturity bonds, the additional haircut is only applied to the part used as collateral. 41 Table 3-1 Liquidity stress test scenario (in %) Balance-sheet item / Maturity bands <3M 3M-6M 6M-9M 9M-12M 1. Liquidity buffer Interest rate and equity shock 1,1 Q-o-q change in yield curve in pp* 1Y PRIBOR 0,3 0,0 0,0 0,0 5Y GB yield 1,0 0,6 0,5 0,4 1Y EURIBOR 0,2 0,0 0,0 0,0 5Y EUR GB yield 0,0 0,2 0,3 0,2 1,2 Haircuts from value of capital instrument 30,0 - - - 2. Inflows Size of deduction from expected inflow 2,1 Secured claims 0,9 0,9 0,9 0,9 2,2 Unsecured claims due** on NRs 2,1 2,2 2,4 2,6 on NFCs and retail SMEs 1,1 1,2 1,2 1,2 3. Outflows Expected outflow rate 3,1 Draw dow n of credit lines 5,0 5,0 5,0 5,0 3,2 Issued debt securities 100,0 100,0 100,0 100,0 3,3 Retail deposits insured 3,2 3,5 3,2 3,1 others 6,3 7,0 6,4 6,3 3,4 Liabilities to NFC insured 12,6 14,1 12,9 12,5 others 25,3 28,2 25,8 25,0 3,5 Liabilities to Fis insured 12,6 14,1 12,9 12,5 others 31,6 35,2 32,2 31,3 3,6 Growth in new loans, of which*** secured claims 0,0 1,4 1,3 1,0 due to NRs 0,0 1,0 0,6 0,4 due to NFCs and retail SMEs 2,4 0,0 0,7 0,0 Source: CNB, authors' calculations Note: The parameter values are the averages of those applied to individual banks. *The haircut is determined by multiplying the change in the yield curve by the duration of the bond portfolio. **Due claims on financial institutions were not subject to deductions in this scenario. ***The credit growth assumption is calculated using satellite models in macro stress tests of bank solvency. NFCs: non-financial corporations, FIs: financial institutions, NPs: natural persons. This table does not contain the endogenous (systemic and reputational) shocks generated in the second round of shocks. 42 The liquidity asset buffer (LR) was defined for the test as the weighted sum of cash, claims on the central bank (excluding minimum reserves), debt securities issued by domestic and foreign government, capital instruments and corporate debt securities excluding those held in credit portfolios.30 Figure 3-1 Post-stress liquidity indicators Figure 3-2 Results of the one-year horizon liquidity stress test (% of balance sheet total of bank type) 1800 1600 1400 1200 1000 800 600 400 200 0 Large banks - 1 s - 1 s -Is Befi toutfl Af Befi toutfl Af Befi toutfl Af o -z. o -z. o -z. Medium-sized Small banks Building societies banks Large banks Medium-sized Small banks Building banks societies ■ Liquidity indicators after one-year stress period * LCR (one-month stress period) Source: Authors' calculations based on Source: Authors' calculations based on CNB data CNB data Note: end-2016 data. Note: Note: end-2016 data. The column "Before" represents the pre-stress size of the liquidity buffer and the column "After" the post-stress size of the liquidity buffer. The column "Net outflow" represents the outflow of liquidity over the one-year horizon. A few banks exhausted their entire liquidity asset buffers (LR) during our one-year test, although the earliest this occurred was in the last quarter (see Figure 3-3). However, some of those banks specialise intentionally in a particular product type. They rely mostly on funding sources within their financial groups and hold hardly any liquid assets. However, the methodology also indicated that some universal banks have less stable sources in relation to their liquidity asset buffers (LR). In the case of banks that did not exhaust their liquidity asset buffers (LR), the liquidity indicator (LL) gradually decreased as the maturity bands increased in length (see Figure 3-4). However, these banks are more than sufficiently compliant with the required indicator level (LI) despite the fact that they had to use their liquidity asset buffers (LR) to cover net outflows (NetOut) from the very first round of the test. The source of resilience of most of the banks under review is their sufficient liquidity asset buffer (mostly 20% of bank's assets, see Figure 3-3), which consists mostly of zero-haircut claims on the central bank and debt securities issued by Collateral accepted was not included in the buffer due to data unavailability. 43 domestic government. For the most part, government bonds are subject not to the interest rate shocks but only to the additional haircuts in the second round of shocks (see Appendix), since a large proportion of the banks under review hold them to maturity.31 The liquidity asset buffer (LR) is fairly homogeneous across the tested banking sector, a property that may magnify the drop in its value if it is used by a large set of banks. Paradoxically, the overall endogenous shock in the form of the additional haircut (see Appendix) on domestic government bonds may thus be large by comparison with riskier assets with lower shares in the liquidity asset buffer (LR). On the one hand, a more diversified portfolio could mitigate this type of systemic risk. On the other hand, most market prices of assets are highly correlated during a crisis, so only cash or near-cash assets (such as claims on the central bank) can offer real hedging against such risk. Figure 3-3 Liquidity buffers of tested banks (% of bank's balance sheet) Figure 3-4 Liquidity indicator profiles over the tested period (%) A A * * —9 I-1-T-1-•-1-1-1-1-1-1-1-1-1-1-1 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Tested banks * Initial liquidity butler • Fbst-stress liquidity butler Source: Authors' calculations based on Source: Authors' calculations based on CNB data CNB data Note: end-2016 data. Note: end-2016 data. A more detailed breakdown reveals that claims on non-financial corporations, which banks usually provide with shorter maturities, make up the largest part of the inflows in all maturity bands. They therefore significantly exceed claims on individuals and credit institutions in maturities of one year or less (see Figure 3-5). Due to their very short maturities, inflows from claims on credit institutions are relevant only in the first maturity band of 0-3 months. By contrast, inflows from claims on households grow in importance with increasing maturity length. However, the one-year test period was too short for the simulated credit shocks to have a major impact via these claims. 31 In the case of held-to-maturity bonds, the additional haircut is only applied to the part used as collateral. 44 Figure 3-5 Liquidity inflow structure (% of bank's balance sheet; x-axis: maturity band) 7 -6 -5 - ^Secured claims I'Calims on individuals ^Claims on non-financial corporations 1 'Claims on credit institutions ■Claims on other financial institutions Figure 3-6 Liquidity outflow structure (% of bank's balance sheet; x-axis: maturity band) 14 -I Q1 Q2 Q3 Q4 a Insured retail deposits and secured liabilities h Uninsured retail deposits and unsecured liabilities s Maturing debt securities □Credit line drawdowns ■ Growth in new loans Source: Authors' calculations based on Source: Authors' calculations based on CNB data CNB data Note: end-2016 data, Q = maturity bands: Note: end-2016 data, Q = maturity bands: of 0-3 months, 3-6 months, 6-9 months of 0-3 months, 3-6 months, 6-9 months and 9-12 months. and 9-12 months. Uninsured retail deposits and unsecured liabilities to non-financial corporations and financial institutions dominate outflows at the aggregate level (see Figure 3-6). Outflows from relations with non-financial corporations far exceed those from other relations. There are two main reasons for this. The banks under review fund themselves primarily by accepting deposits from households and non-financial corporations rather than by obtaining loans from other banks in money markets. Compared to retail financing, however, corporate (wholesale) financing is considered a less stable funding source, so a relatively high outflow rate is applied to it. Banks whose sources consist mostly of corporate deposits therefore undergo severe stress in this test. Their liquidity buffers should thus be larger than those of banks with predominantly retail sources to survive the stress. 3.5 Conclusion This chapter described a liquidity stress-testing framework based on some principles of the two Basel liquidity regulatory standards the LCR and the NSFR. The model took into account the impact of both bank-specific and market-wide scenarios and includes second-round effects of shocks due to banks' feedback reactions with endogenous adverse feedback loop. We also showed how solvency and liquidity stress-testing frameworks can be interlinked, so that a complete stress-testing exercise can encompass mutually consistent shocks to liquidity, market, credit and other risks. The survival period was set on one year to monitor the liquidity position of banks over a longer period of market stress. The output of the presented stress test is a liquidity indicator which, analogously to the LCR, expresses the coverage of the net expected liquidity 45 outflow with liquid assets subject to haircuts. The liquidity indicator level is deemed adequate if it maintains a minimum value of one over a one-year period (analogously to the NSFR). The stress test methodology was applied to a representative sample of 19 banks domiciled in the Czech Republic, with various business models and bank sizes represented. The sole aim of the analysis - based on real data - was to present the methodology and monitor the sensitivity of the liquidity position of selected banks to the combination of shocks considered over a longer period. The outcomes of the model showed that the most Czech banks seemed to be resilient against presumed liquidity, market and credit shocks. However, there were four of them who exhausted their liquidity buffers, partly also due to the second-round effects. Their liquidity indicators fell below 100% minimum although not before the last quarter. This proved that there is heterogeneity among tested banks and that sufficient liquidity in a banking sector as a whole can be specious. We also compared the liquidity indicators of banking groups with their LCR requirement. The results confirm that LCR requirement as a short-term stress test is inappropriate for testing some types of business models such as building societies. It results that in a stress test shorter and longer horizon should be explored to assess whether a bank's outcomes are sensitive to this issue. 46 Appendix Table 3-2 Summary of parameter settings with use of the scenario (in %) Balance-sheet item irameterisation sourc Parameter value tor maturity band >3M-6M >6M-9M >9M-12M Inflows (p) Secured claims Claims due* on individuals on non-financial customers and retail SrVEs Liquidity butter Interest rate shock to debt securities (h) Domestic government AFS in CZK Foreign government AFS in CZK Domestic CIs' AFS in CZK Foreign CIs1 AFS in CZK Domestic corporates' AFS in CZK Foreign corporates' AFS in CZK Domestic government AFS in foreign currency Foreign government AFS in foreign currency Domestic CIs' AFS in foreign currency Foreign CIs' AFS in foreign currency Domestic corporates' AFS in foreign currency Foreign corporates' AFS in foreign currency Endogenous market liquidity shocks (r/n) macro-stress scenario macro-stress scenario i macro-stress scenario macro-macro-macro-macro-macro-macro-macro-macro-macro-macro-macro-macro- stress scenai stress scenai stress scenai stress scenai stress scenai stress scenai stress scenai stress scenai stress scenai stress scenai stress scenai stress scenai 1.35 0.56 4.31 7.05 4.15 1.45 2.10 0.68 0.84 0.81 0.69 0.37 0.79 0.88 1.44 0.70 1.54 0.69 4.43 7.19 4.18 1.40 2.07 0.57 1.19 1.08 0.97 0.40 1.18 1.22 1.59 0.69 1.06 1.71 0.99 0.33 0.49 0.14 0.63 0.57 0.51 0.21 0.62 0.65 credit credit market -market -market -market -market -market -market -market -market -market -market -market - iteresi iteresi iteresi iteresi iteresi iteresi iteresi iteresi iteresi iteresi iteresi iteresi Capital instruments (h) liquidity stress test 61.24/50 78.3/63.93 77.94/63.64 61.24/50 market - systéme anc reputat Capital instruments (q) liquidity stress test 11.24/0 28.3/13.93 41.87/29.83 -/30.59 market - systéme anc reputat Debt securities of domestic government (h) liquidity stress test 16.44/13.43 9.48 / 7.74 9.38/7.66 10.4/8.49 market - systéme anc reputat Debt securities of domestic government (q) liquidity stress test 11.44 / 8.43 12.91 /11.37 16.71 /14.1 22.25/18.24 market - systéme anc reputat Debt securities of foreign government (h) liquidity stress test 38692 7.9/6.45 8.14/6.65 8.37/6.84 market - systéme anc reputat Debt securities of foreign government (q) liquidity stress test 01.12.2000 2.9/1.45 5.07/3.18 7.21 /5.28 market - systéme anc reputat Debt securities of domestic CIs (h) liquidity stress test 62.36/50.92 47.8/39.03 47.03 / 38.4 51.53/42.07 market - systéme anc reputat Debt securities of domestic CIs (q) liquidity stress test 32.36/20.92 38.72/30.55 46.97/39.95 59.87/52.46 market - systéme anc reputat Debt securities of foreign CIs (h) liquidity stress test 63.09/51.51 46.82/38.23 46.79 / 38.2 49.42/40.35 market - systéme anc reputat Debt securities of foreign CIs (q) liquidity stress test 33.09/21.51 44.12/29.74 46.53/39.45 57.36/49.25 market - systéme anc reputat Debt securities of domestic corporates (h) liquidity stress test 63.17/51.58 46.86/38.26 46.88/38.28 49.52/40.43 market - systéme anc reputat Debt securities of domestic corporates (q) liquidity stress test 33.17/21.58 38.43/30.45 46.72/39.63 57.64/50.07 market - systéme anc reputat Debt securities of foreign corporates (h) liquidity stress test 36.74/30 46.75/38.17 46.83/38.24 49.44 / 40.36 market - systéme anc reputat Debt securities of foreign corporates (q) liquidity stress test 6.74/0 16.75/8.17 25/16.86 35.84 / 27.78 market - systéme anc reputat Outtlows (r) Credit line draw dow ns" Debt securities issued Retail deposits insured other Liabilities to NFCs secured other Liabilities to FIs secured other Growth in new loans of which secured claims of which due vis-á-vis individuals of w hich due vis-á-vis non-financial customer; expert judgement 5 non-restoration of sourc 100 LCRfloor, macro-stress III.75 LCRfloor, macro-stress 07.V LCRfloor, macro-stress 15 LCRfloor, macro-stress 30 LCRfloor, macro-stress 15 w LCRfloor, macro-stress 37.5 macro-stress scenári macro-stress scenári macro-stress scenári 3.125 VI. 25 12.V 25 12.V 31.25 III. 75 07.V liquidity liquidity liquidity liquidity liquidity liquidity liquidity liquidity credit credit credit Source: CNB, authors' calculations Note: r/n stands for reacting/non-reacting bank, h for the haircut on a liquid asset, p for the size of the haircut on the expected inflow and r for the size of the outflow. The parameter values are the average parameter values applied to individual banks. * Due claims on financial institutions were not subject to haircuts in this scenario. ** The stock of credit lines as of the test date was multiplied by the value of this parameter. 47 Chapter 4 Monetary policy after the 2007-2009 global financial crisis in the context of systemic risk By Liběna Černohorská and Petr Teplý This chapter deals with the changes in implementing monetary policy that central banks were forced to employ in response to the effects of the 2007-2009 global financial crisis (GFC). These changes were in most cases needed because conventional monetary policy instruments were no longer effective at achieving their set goals. At the same time, it was also necessary for central banks to monitor not only price stability but also financial stability, which required deployment of macroprudential policy tools. The question remains as to whether the central banks increased or decreased systemic risk of financial markets with their measures. In the first section, we explain the necessary changes that the selected central banks were forced to adopt. They began to simultaneously implement unconventional monetary policy in the form of quantitative easing, currency intervention, negative interest rates, and forward guidance. The next two sections present the basic principles of how monetary policy operated before the financial crisis, while also pointing out the areas of monetary policy that were reassessed in conjunction with the impact of the financial crisis on monetary policy. In the fourth section, we deal with the effectiveness of unconventional monetary policy in the selected countries, specifically as it relates to the Federal Reserve System (FED), the Bank of England (BOE), the Bank of Japan (BOJ), the Czech National Bank (CNB), the European Central Bank (ECB), and the Swiss National Bank (SNB). The fifth section deals with reevaluating analytic approaches to monetary policy regarding financial stability, and we also devote time to systemic risk in the context of contemporary monetary policy. The closing section summarizes the chapter and states final remarks. 48 4.1 Introduction The international financial markets are increasingly interconnected; therefore, they have great impact on individual national economies. The GFC itself began on the American mortgage market in 2007; it subsequently affected the Western European banks and then circled back to the USA. It resulted in serious disruption of the global credit markets. Losses on the world's credit markets led to the worst global economic downturn since the Great Depression. Financial crises nearly always result in steep increases in government debt. The fiscal situation in many countries was negatively influenced by massive emergency measures to stabilize financial institutions, fiscal stimulus packages, and sharp economic decline that resulted in lower tax revenues all around the world. Budget deficits became a basic component of advanced countries, and the ratio of government debt to gross domestic product (GDP) increased. Such increases in debt can even potentially result in government debt defaults. It is still unclear how monetary policy can learn from the GFC. Certain authors (e.g., Krugman, 2009, and Cochrane, 2011) state that the financial crisis revealed major deficiencies in the monetary policy enacted over the past forty years and that fundamental change was therefore necessary. Consequently, certain central banks were forced to implement unconventional monetary policy. Central banks did so when conventional (classical) monetary policy failed and interest rates were near zero. Quantitative easing, currency intervention, negative interest rates, and forward guidance are examples of unconventional monetary policy instruments (Mishkin, 2017; Mejstfik et al., 2014). Central banks implement quantitative easing by purchasing domestic financial assets - either from commercial banks or non-banking entities. For it to be considered a quantitative easing, there must also be the great increase in the central bank's balance sheet that quantitative easing entails. In practice, this primarily means the purchase of domestic government bonds. This unconventional monetary policy instrument is being or has been used by the BOE, the BOJ, the ECB, and the FED, for example. The result of quantitative easing is that governments and businesses - as the entities issuing the bonds purchased by the central bank - are able to further increase debt at low interest costs. The resulting situation influences the growth of inflation, the gross domestic product, and employment. In currency intervention, the central bank sets a given level for the currency exchange rate. In order for the central bank to achieve this currency exchange rate, it must be willing to intervene on the currency market as necessary in order to achieve its set commitment. The scope of these interventions is not shared with the public in advance. This results in weakening the domestic currency by using domestic currency to purchase foreign currency. Because the central bank alone is able to issue currency, this procedure can be conducted without limit. Currency intervention is also easier for the public to understand. The result is increasing expectations for inflation. Prices for imported goods increase due to the weakened domestic currency, and this eventually results in increasing inflation in small open economies. This monetary policy instrument was used by the CNB and the SNB. 49 In the case of negative central bank interest rates, commercial banks pay the central bank for holding their excess financial resources. The reason for implementing negative interest rates is so that banks do not allow their surplus funds to remain in their accounts at the central bank, but rather have them released into the economy, e.g., in the form of loans for other economic entities. In conjunction with the GFC, central bank interest rates dropped to zero. In 2009 Riksbank, Sweden's central bank, was the first bank to use negative interest rates in practice. Shortly thereafter, the Danish and Swiss central banks followed the Swedish lead, and the ECB also joined the banks employing negative interest rates in 2014. Forward guidance is used to signal future developments in monetary policy. With the help of forward guidance, central banks try to influence the expectations of economic entities concerning the future development of monetary policy. This instrument can be implemented in two possible ways - either the central bank can make public prognoses of monetary policy (future development) or it can define the explicit commitments it would like to achieve. For example, the ECB made it known in 2013 that its loose monetary policy would remain accommodating as long as necessary and that key rates would remain at current or lower levels over the long term. Later in this study, we will also discuss monetary policy in the context of systemic, which can be defined in several ways. For instance, ECB (2009) defined it as the risk "that financial instability becomes so widespread that it impairs the functioning of a financial system to the point where economic growth and welfare suffer materially." Alternatively, Benoit et al. (2017) offered the following definition of systemic risk: "the risk that many market participants are simultaneously affected by severe losses, which then spread through the system ". For the purpose of our chapter, we will use the definition by Benoit et al. (2017). 4.2 The basic principles of monetary policy before the GFC Until 2007, a positive general consensus prevailed among central banks when evaluating how monetary policy was being implemented. However, the question remains as to which changes in conducting financial policy were brought about by the financial crisis. Mishkin (2017) deals with the implementation of monetary policy previous to the financial crisis, when inflation was conceived as financial phenomenon. Friedman (1974) sees its causes in expansive monetary policy, i.e., in the excessive growth of money in the economy, which must be combated. Furthermore, central banks have the option of influencing inflation and should maintain it at a low, stable level, because price stability brings them great advantages. The necessary starting point for price stability is the Taylor principle, which is derived from the thesis that inflation will be stable only under the assumption that monetary policy increases the nominal interest rate by more than the growth of inflation, so that real interest rates increase along with inflation (Taylor, 1993). Numerous empirical studies (e.g., Sack, 2000; Claridaet al., 1999; 50 Levine et al., 2005; Smets and Wouters, 2003) described suitable monetary policy before the financial crisis using Taylor rules. At the same time, monetary policy should not attempt to achieve lower levels of unemployment by aiming for higher inflation, because there is no long term relationship between unemployment and inflation. Central bank independence helps central banks free themselves from political pressure and allows them to conduct expansive monetary policy. Empirical studies show (e.g., Bleaney, 1996; Alesina and Summers, 1993; and Cecchettia and Krause, 2002) that central banks achieve higher performance from a given economy when they have greater independence. Volatility on the financial markets plays an important role in the economic cycle. If a financial system is exposed to shocks that increase information asymmetry, the result is an increase in volatility on the financial markets, which leads to financial instability. Subsequently, the financial system is not able to provide financial resources for investment; therefore, the economy's performance declines. Before the global financial crisis, volatility on the financial markets was not a frequent subject of interest, and so it was not previously included in models analysing central bank policy. Mishkin (2017) claims that these cited facts are elements of what can be called "the new neoclassical synthesis." Goodfriend and King (1997), for example, deal with this issue in more detail. Before the global financial crisis, nearly all central bankers agreed with these actualities. As conceived by the new neoclassical synthesis, monetary policy strategy is termed "flexible inflation targeting" (Svensson, 1997) in academic literature. It includes a strong and reliable commitment by the central bank to stabilize inflation over the long term, often at a precise numerical rate. However, it also makes it possible for the central bank to influence economic output around a potential product over the short term. 4.3 The impact of the GFC on monetary policy Mishkin (2017) highlights that some new trends in monetary policy strategies and approaches after the GFC occurred, because of the evidence the development of the financial sector has a much greater impact on economic activity than had been previously thought. The GFC and the subsequent economic recession clearly showed the necessity of financial macroeconomic analysis, which should be included in macroeconomic models. These models should no longer be left out of central bank prognoses and analyses of monetary policy's effectiveness. Furthermore, Mishkin (2017) argues that zero interest rates constitute a serious problem, because conventional expansive monetary policy becomes ineffective when a negative shock affects the economy. Shocks from the destabilization that happens when financial systems are disrupted are much deeper than have been previously assumed. Therefore, low interest rates have become a more important monetary policy instrument for central banks than before the financial crisis. Previous to the global financial crisis, many economists believed that low interest rates were effective when combined with other unconventional monetary policy 51 instruments, because they provided a sufficient impulse for renewing economic growth (Svensson, 2018). Williams (2014) proved that unconventional monetary policy stimulates economic activity, and it is true that central banks around the world have tried to bring back full employment levels or achieve set inflation targets of 2% within individual economies. Accordingly, low or even negative interest rates are necessary for reviving economies. In this situation, central banks fall back on nontraditional (unconventional) measures for monetary policy, such as quantitative easing, which consists of purchasing assets, or currency intervention with the goal of reviving the economy, for example (as described in Černohorská, 2017; Mandel and Tomšík, 2018; Svensson, 2018; Wu and Xia, 2016; Williams, 2014; and Zamrazilová, 2014). According to Revenda (2016), quantitative easing was originally intended to support the banks' health but gradually it even managed to support banks' credit activities, which should reflect economic development and the aversion of deflation in a positive way. Stable inflation and output do not guarantee financial stability, even though before the recent financial crisis, it was the common opinion of both academics and central banks that achieving stable prices and economic output supports financial stability. This was supported by research from the authors Bernanke et al. (1999) and Bernanke and Gertler (2001), who indicated that monetary policy in support of stable inflation and production most likely also stabilizes the price of assets. This would prevent the creation of asset bubbles on the market, which would thus become less and less likely to occur. Up until 2007, the economic environment did not do enough to protect the economy from financial instability - it actually managed to make the instability even greater. The overall costs for renewing financial market stability are very high. Besides the actual expenses resulting from the global recession, there are also the additional costs that arose from the global financial crisis, which fall into two categories: financial crises are usually accompanied by slow economic growth, simultaneously resulting in higher budget deficits for governments. The cumulative losses from financial crises are very high, and it is clear that there are no exceptions - even for a global financial crisis. Mishkin (2017) has warned that monetary policy should not completely veer away from experience gained previous to the financial crisis. Currently, most of the steps for implementing monetary policy are the same as those before the financial crisis. Nonetheless, one clear lesson learned from the financial crisis is that developments on the financial markets can have a more significant effect on economic activity in individual countries than central banks previously realized. Mishkin (2017) lists the areas of monetary policy that it would be appropriate to reevaluate when implementing monetary policy itself. The steps listed are derived from the following principles of the new neoclassical synthesis: 52 1. Flexible inflation targeting 2. Monetary policy's reaction to asset price bubbles 3. The dichotomy between monetary policy and financial stability 4. International coordination of monetary policy 5. Forward guidance Before the financial crisis, central banks did not have financial market volatility included in their economic models, even though this is one of the main causes of fluctuation in the economic cycle. This resulted in the dichotomy between monetary policy and financial stability policy, which has these two types of policy being conducted separately. As Mishkin (2011) states, the recent financial crisis supports systemic regulation, with central banks becoming a suitable choice for the role. The benefit of coordinating monetary policy and macroprudential policy is another reason why central banks should take on the role of system regulator. The global financial crisis led both economists as well as central bankers to approach the implementation of monetary policy differently. It is necessary to realize that some areas must be reevaluated, and there should be a focus on inflation and monetary policy's reaction to the possible appearance of asset price bubbles. It is further necessary to concentrate on the dichotomy between monetary policy and financial stability policy. Not least, there should also be greater international cooperation on monetary policy in order to be better able to cope with volatility on the financial markets. 4.4 The effectiveness of unconventional monetary policy in the selected countries The effectiveness of monetary policy often tends to be evaluated according to how the development of monetary aggregates (including the monetary base) or central bank interest rates impact economic quantities. Monetarists start with the thesis that over the long term, monetary aggregates have been considered the deciding factor for conducting monetary policy. Friedman and Schwartz (2008) explain the economic cycle's development using changes in the money supply and inflation. As stated by Friedman (1968) as well as by Brunner and Melzer (1969), money has only short-term impact on the real economy. At the same time, they emphasize - the same as other monetarists - that monetary policy should not be used as an active tool for stabilizing the economic cycle. If the central banks had already been using standard monetary policy instruments for meeting set targets (e.g., inflation targets) and the tools used did not result in their achievement, these central banks resorted to unconventional monetary policy tools. The selected central banks took unusual steps in monetary policy (so-called unconventional monetary policy) in order to minimize the effects of the financial crisis or avert the risk of inflation. Traditional instruments primarily began to fail in 2008 in conjunction with the financial crisis, when a number of central banks nearly reached zero while setting monetary policy interest rates, and it was not possible to lower rates further in order to ease monetary conditions. The central banks needed to resort to instruments with whose consequences they had no experience. Primarily, these tools consisted of negative 53 nominal interest rates, currency intervention, quantitative easing, and forward guidance as has been previously mentioned. The FED and the BOE purchased government bonds and other private assets (i.e., they implemented quantitative monetary easing). The FED implemented quantitative easing from 2008 until 2014 at an overall amount of roughly USD 3,940 billion. The ECB focused on purchasing covered bonds between 2015 and 2018. Roughly between August and October of 2008, the central banks cited increased their volume of repos (or similar instruments) in an attempt to lessen volatility on the interbank markets and facilitate access to sources of liquidity for financial institutions. Another FED reaction to the financial crisis was to establish a number of programs that were to provide liquidity directly to borrowers and investors. Examples of these are TALF (the Term Asset-Backed Securities Loan Facility), which provided liquidity to households and small businesses and the PDCF (the Primary Dealer Credit Facility), a so-called overnight loans primarily for businessmen - in addition to many others. The risk of quantitative easing lies in the enormous increase in the central banks' balance sheet. This approach also makes it possible for governments to increase their debt and provides moral hazard for governments, which continue to increase debt because their bonds are always in demand. Furthermore, this huge growth in the amount of money increases the risk of having higher inflation, which becomes more difficult to decrease over the long term. Specifically, quantitative easing was used by the FED, the ECB, the BOJ, and the BOE. Each month, the ECB purchased bonds worth EUR 80 billion; near the end, they gradually lowered the amount to EUR 30 billion then 15 billion. It purchased a total of EUR 2.6 billion in bonds over nearly four years (from March 2015 to December 2018). Their balance sheet more than doubled to reach EUR 4.7 billion, which is nearly 35% of GDP. According to their calculations, quantitative easing was supposed to increase GDP and inflation by 0.4% annually. Within the Eurozone, each of the 19 central banks purchased bonds issued by their governments and thus also carried the risk of possible non-reimbursement, which is a real threat in some countries. Between December 2008 and October 2014, the American FED pumped USD 3.7 billion into the American economy in three phases by purchasing government and mortgage-backed securities. The FED's balance sheet increased to more than four billion dollars, which is more than eight times the pre-crisis level. Meanwhile, as early as 2009, Sweden's central bank Riksbank was the first to implement negative interest rates. Denmark's central bank followed in 2012; next was the ECB in 2014 and then the SNB, the BOJ, and finally the Central Bank of Hungary. Their rates as of September 2019 are listed in Table 4-1. Together with the SNB, the CNB began to intervene on the currency exchange market with the goal of preventing its currency's exchange rate from increasing or turning back potential deflation. Currency intervention was specific to the perspective of the individual country. The SNB conducted currency intervention since 2009, between September 2011 and January 2015 using a currency floor. The reasons why the SNB started to intervene were low inflation, the attempt to 54 support the competitiveness of its exports, and strong pressures causing the currency to appreciate. The CNB intervened on the currency exchange markets between November 2013 and April 2017 due to threatening deflation. The CNB decided to use currency intervention only after neither verbal intervention on the part of the Bank Board members nor forward guidance worked. The verbal intervention consisted of a CNB announcement that if it became necessary to further ease monetary conditions, they would do so by weakening the Czech koruna. This tool affected the market for one year by weakening the koruna by a few tens of hellers. Table 4-1 Negative interest rates of the selected central banks in September 2019 Central Bank Basic Interest Rate Deposit Interest Rate Danmarks Nationalbank 0.00% - 0.75% ECB 0.00% - 0.50% Riksbank (Sweden) - 0.25% - 1.00% Swiss National Bank - 1.25% to -0.25% - 0.75% Bank of Japan -0.10% - 0.10% Central Bank of Hungary 0.90% - 0.15% Source: Authors based on websites of selected central banks As of March 2013, the CNB indicated in advance via forward guidance how they would be proceeding with their monetary policy. In this way, they tried to influence the expectations of economic entities and thus hasten economic revival and the growth of inflation. After the effects of these unconventional nonfinancial instruments receded, the CNB decided to use currency intervention between November 2013 and April 2017, when it committed to pin the Czech koruna's exchange rate at CZK 27/EUR. Directly in the first phase, the CNB purchased euros amounting to EUR 7.5 billion, i.e., roughly CZK 202 billion, This convinced the markets, and the koruna was maintained at the given level without problem up until the summer of 2015, when the CNB gradually made another intervention that consisted of purchasing euros. The CNB intervened on the currency exchange market in the amount of USD 51.4 billion, whereas the SNB invested roughly USD 556.6 billion. The method for concluding the currency intervention differed for the two countries. In the Czech Republic, there was no significant strengthening of the koruna, whereas the Swiss franc increased by 15%. There were differences not only in the intervention programs' commencement but also in their exit phases. Representatives of the SNB left the domestic currency rate to its "fate", which also subsequently meant that it showed great volatility (Table 4-2). The reason the SNB ended its currency floor was doubts concerning quantitative easing by the ECB, which was likely to result in further pressures pushing the franc higher. The second, relatively specific reason was apprehension by SNB stockholders (the Swiss cantons) that there would be high SNB losses due to abnormal exchange rate risk considering the large amount of assets being held in foreign currencies. Afterward, the SNB used negative interest rates for easing 55 monetary conditions. Nevertheless, interventions have also been conducted in a discretionary manner. Table 4-2 Comparison of currency intervention by the CNB and the SNB CB Name CNB SNB Reason for initiation Volume Currency intervention goal Period Form of intervention Exit Communication about the exit Length of duration Regime after exit Evolution of the nominal rate after exit Ineffectiveness of conventional monetary policy instruments, risk of deflation approx. USD 90 bn Achieving an inflation target of 2% 2013-2017 A previously announced, one-sided exchange rate floor of 27 CZK/EUR Tied to meeting the inflation target From the beginning of the commitment An unspecified intervention time Managed floating exchange rate The assumed enormous strengthening of the koruna did not occur strong pressures appreciating the domestic currency as the result of an influx of capital from abroad, competitiveness of exports, the prediction of low inflation approx. USA 560 bn To weaken the domestic currency, to achieve an inflation target of 2% Since 2009 A one-sided exchange rate of 1.2 CHF/EUR in 2012-2105 Discretionary An unexpected exit An unspecified intervention time Free floating exchange rate Considerable appreciation of the franc by 15% Source: Authors based on CNB's and SNB's websites During the crisis, the CNB's Bank Board also discussed implementing negative interest rates. However, it did not consider them an appropriate instrument for the Czech Republic's monetary policy. This was a given primarily by the lack of experience with their operation on the financial markets and indirectly on the real markets. Also, there were additional concerns as to the extent to which they would need to lower the negative rate in order to eventually instigate the shift of interbank liquidity to bank loans for companies and consumers. However, there were also opinions that even though implementing negative rates would increase bank costs for maintaining their liquidity, this would not be a fundamental problem considering their profitability. In the end, the Bank Board decided not to begin experimenting with negative interest rates. 56 Quantitative easing did not make sense under the Czech circumstances, because historically there has been a so-called systemic surplus of liquidity on the Czech banking market. In other words, there is enough "interbank money" on the market and it does not make sense to create more here, when banks have not yet been able to appreciate all the available liquidity. The use of currency intervention in the form of weakening the exchange rate thus prevailed among Bank Board members as the most effective instrument for the Czech conditions. This was because the Czech economy is highly open. Czech exports represent more than 80% of GDP, with import goods having a value over 70% GDP. This means that the Czech economy is highly dependent on changes in the exchange rate - it is one of the most significant variables in our economy. On the basis of these facts, we can state that none of the central banks examined here implemented unconventional monetary policy in the same way. The common goal of these unconventional policies was to avert the outbreak of deflation, ensure price stability, support economic growth, and contribute to the stability of financial systems. 4.5 Systemic risk in the context of current monetary policy In previous sections we discussed the role of monetary policy during and after the GFC, now we will focus on systemic risk implications. Several researchers such as Acharya and Richardson (2009) or Diamond and Raj an (2009) highlighted that a loose monetary policy was one of the causes of the GFC as it resulted in risky behaviour of financial companies. Gambacorta (2009) concludes that monetary policy may influence banks' perceptions of risk in two ways: through a search for yield process and by means of the impact of interest rates on valuations, incomes and cash flows, which in turn can modify how banks measure risk. When evaluating the success of the central banks, often the factor under consideration is whether or not they achieved their goals. This evaluation is usually connected with the achievement of monetary, price, or financial stability. As Borio (2014) warned, the process of achieving financial stability is far too extensive to be ensured by monetary policy alone. For this reason, central banks implement macroprudential policy, which has the goal of achieving financial stability. Recently, economists have also been contemplating the impact of unconventional monetary policy in various countries. Borio and Disyatat (2010) have clarified the differences between various forms of unconventional monetary policy. At the same time, they provided systematic descriptions for a wide spectrum of central bank behaviour during the period of the global financial crisis, including unconventional monetary policy's effect on inflation. The recent financial crisis resulted in a fundamental reevaluation of analytic approaches to monetary policy as it relates to financial stability. The crisis demonstrated the necessity of focusing more on systemic risk and incorporating the financial sector into macroeconomic models. The shift in monetary policy towards macroprudential policy is visible in the area of regulation and oversight. There is the question of whether it is sufficient to use the achievement of price 57 stability as the criteria for evaluating the effectiveness of monetary policy. As is clear, establishing price stability alone has let to financial instability in some countries (Borio, 2011). Borio and Shim (2007) place great emphasis on central banks' macroprudential policy, which is key for financial and macroeconomic stability. In the future, central banks should aim to be more focused on macroprudential policy, which should limit the emergence of any financial instability. Macroprudential and monetary policies affect each other mutually, although both have different goals and instruments. At the same time, monetary policy should be treated entirely separately from macroprudential policy, i.e., financial stability (Svensson, 2018). Macroprudential policy instruments lower the financial system's vulnerability and increase its resilience by establishing capital and liquidity cushions, which prevent procyclicality in the financial system. Malovaná and Frait (2017) discuss how monetary and macroprudential policies may interact and potentially get into conflict. They conclude that accommodative monetary policy contributes to a build-up of financial vulnerabilities, what implies that it boosts the credit cycle. Laséen et al. (2017) conclude that a monetary policy tightening surprise does not necessarily reduce systemic risk, particularly when the state of the financial sector is fragile. More recently, Colletaz et al. (2018) find that causality from monetary policy to systemic risk in the long run in the Eurozone. As a result, they claim that central banks must be aware that a too loose monetary policy stance may be conducive to a build-up of systemic risk. Moreover, Kurowski and Smaga (2019) demonstrates on data from the UK, the Eurozone and the US that the occurrence of potential procyclical behaviour of monetary policy underlines the need for proactive macroprudential policy. Based on the previous literature survey we state that there is no unique view whether the conventional and non-conventional monetary policy measures reduce or increase systemic risk. However, one can measure systemic risk through various indicators such as Composite Indicator of Systemic Stress (CISS) developed by Hollo et al. (2012) and regularly published by the ECB. The CISS indicates a low risk (but rising) of systemic stress in recent years (ECB, 2018). 4.6 Conclusion Up until the outbreak of the GFC in 2007, central banks conducted conventional monetary policy using changes in interest rates. During the GFC all big central banks including the ECB, the FED, and the BOE were forced to begin to use unconventional monetary policy instruments, because conventional instruments failed in meeting set inflation targets. When these central banks had interest rates very close to zero, the financial systems had not yet been stabilized, and the economic situation began to rapidly degenerate. The selected central banks took unusual steps in monetary policy by using unconventional monetary policy instruments in order to minimize the impact 58 of the GFC or to turn back the risk of deflation. The FED, the ECB, and the BOE implemented unconventional monetary policy using quantitative easing. Whereas the ECB focused on purchasing covered bonds, the FED and the Bank of England purchased government bonds and other private assets. In our study we use the definition of systemic risk provided by Benoit et al. (2017) stating that systemic risk is "the risk that many market participants are simultaneously affected by severe losses, which then spread through the system. " Since there has been neither financial crisis nor financial turmoil on global scale in last years, we argue that central bank's monetary policies have not increase systemic risk until September 2019. This fact is also supported through a decline of the CISS indicator, ECB's measure of systemic risk, in the 2011-2018 period. However, the recent monetary policies might have contributed to distortions in particular markets (e.g. the Eurozone bond market or the Czech real estate market), what might result in systemic risk in the future. 59 Chapter 5 Foreign exchange market contagion in Central European countries By Luboš Komárek and Narcisa Kadlčáková This chapter examines contagion in the foreign exchange markets of three Central European countries and the Euro area. Contagion is viewed as the occurrence of extreme events taking place in different countries simultaneously and is assessed with a measure of asymptotic tail dependence among the studied distributions. Currency crisis contagion is one strand of this research. However, the main aim of the chapter is to examine the potential of "bubble" contagion. To this end the representative exchange rates are linked to their fundamentals using a cointegration approach. Next, the extreme values of the differences between actual daily exchange rates and their monthly equilibrium values determine the episodes associated with large departures from equilibrium. Using tools from Extreme Value Theory, we analyse the transmission of both standard crisis and "bubble" formation events in the examined currency markets. The results reveal a significant potential for contagion in the currency markets of Central Europe. 5.1 Introduction Recent developments in financial markets have shown that crises can have quick and often devastating effects in areas far beyond their epicentres. The speed with which the recent US sub-prime crisis reached a global dimension took economists and policy makers alike by surprise. It proved that the global nature of the current market inter-linkages makes the transmission of disequilibria across markets and regions a very likely outcome. In this chapter we look at the disequilibrium transmission within the foreign exchange markets of three Central European countries (Hungary, the Czech Republic and Poland) and the Euro area. Although no major currency crises have occurred in this region, we analyse the potential co-alignment of such crises in this 60 region. However, the main aim of the chapter is to extend traditional currency crisis analysis by looking at contagion during episodes of significant departure from exchange rate equilibrium values. This offers an insight into how likely it is that disequilibria of a "bubble" type is transmitted in a coordinated manner across the exchange rate markets in this area. Contagion during the disequilibrium formation process is examined using tools from cointegration and Extreme Value Theory (EVT). Contagion is viewed as the occurrence of extreme events taking place in different markets simultaneously, and is assessed with a measure of asymptotic tail dependence among the studied distributions. Currency crisis contagion is assessed in a standard way, by focusing on the extreme values of exchange rate return distributions. The potential of "bubble" contagion is examined by firstly linking the representative exchange rates to their fundamentals using a cointegration approach. This gives the equilibrium exchange rate values at a coarser (monthly) frequency. Next, the data is considered at a daily frequency and the extreme values of the differences between actual daily asset values and the monthly equilibrium values determine the episodes associated with large departures from equilibrium. Consequently, an EVT-based contagion approach is applied to these departures from equilibrium distributions and this forms the basis for the analysis of transmission of such "bubble" formation events among the analysed currency markets. The results reveal a significant potential for contagion among the currency markets in Central Europe, both in terms of currency crises and disequilibrium formation. We look at episodes of both depreciation (right tail) and appreciation (left tail) of the examined exchange rates. In all cases our results reveal asymptotic dependence values close to one, which proves that the contagion potential in these markets is very high. The chapter is organized as follows. Section two offers a brief description of the main approaches used and the related literature. Section three contains an overview of the main developments of the analysed exchange rates. Next section four sheds light on the methodologies employed. The main results of the empirical analysis are presented in fifth section. The final section contains main conclusions of this chapter. 5.2 Contagion and extreme value theory The empirical analysis undertaken in this chapter draws intensively from cointegration and extreme value theories. Cointegration is a standard textbook methodology that does not require further explanation. The caveat that we bear in mind, however, is that cointegration employs variables covering long time horizons and this raises the question of the existence of structural breaks in the evolution of the employed variables. The presence of structural breaks affects the decision taken with regard to the order of integration of the variables. This is an argument originally put forward by Perron (1989) and carried on in a number of subsequent chapters. The reasoning is that unit root tests have reduced power in the presence of structural breaks, meaning that such tests might be biased towards 61 the non-rejection of the unit root hypothesis even if the data were in reality stationary around a broken deterministic trend. A stumbling block in testing for unit roots with structural breaks is the fact that these two aspects are closely interrelated. Testing for unit roots requires knowledge about the existence of a structural break and vice-versa. Unless prior knowledge about the existence of an (exogenous) break is already available, deciding where to start is not obvious. A way out of this vicious circle is offered by the methodology of Perron and Yabu (2005). These two authors propose a testing procedure for the existence of a break in the trend function without prior knowledge about the stationary nature of the variables (i.e. 1(0) or 1(1)). They also indicate a method of endogenously estimating the time of the break. This is done by minimizing the sum of squared residuals from regressions run at each time spot that, besides standard regressors used in the unit root setting, also include time dummies reflecting the modelled trend changes. The methodology of Perron and Yabu is applied in this chapter to test for the existence of one endogenously determined structural break. As the results show, the existence of such structural breaks cannot be rejected for the majority of the employed variables. If the existence of a break is not rejected with the Perron and Yabu test, unit root tests allowing for a break in the trend function of the type proposed in Kim and Perron (2009) are further employed. These two authors developed a unit root testing methodology assuming the existence of one break whose time of occurrence is not a-priori known. Their break identifying method coincides with the one proposed in Perron and Yabu and thus the timing of the break is the same under both approaches. If the null hypothesis of a break is rejected, the decision about the stationary nature of the series is based on standard Augmented Dickey-Fuller (ADF) and Phillips-Perron tests. This chapter draws from the EVT part of the vast amount of economic literature related to currency and, more generally, financial crises. In the EVT approach, financial crises are viewed as rare and extreme events whose occurrence is governed by different laws than those governing the entire domain of studied asset return distributions. The focus is on the tails of the distributions. This allows the avoidance of some typical 1 misassumptions, of which the most commonly made are that (a) the analysed empirical distributions follow normal distributions and (b) the Pearson correlation is a good measure of crisis dependency. In fact, it is a common finding in the economic literature that asset returns significantly depart from the assumption of normality in the majority of markets and asset type studied. As a rule, empirical asset returns display fat tails, implying that the probability of extreme events is higher than studies based on the normal distribution usually assume. Additionally, asymptotic dependence or tail-based dependence measures are usually quite different from linear dependence measures proxied by the Pearson correlation. Embrechts and al. (2002) and de Vries (2005), for instance, proved that tail dependence may still be significant among variables with a zero Pearson correlation. It is also true that asymptotic dependence is zero 62 in the case of bivariate normal distributions with a non-zero but less than one Pearson correlation. This chapter draws inspiration from several chapters employing EVT in the crisis context. Cumperayot and Kouwenberg (2011) used EVT to search for asymptotic dependence between exchange rates and several macroeconomic variables, in an attempt to find early warning systems for currency crises. From a rather comprehensive list of macroeconomic variables, asymptotic dependence was found only between domestic real interest rates and exchange rates. Their methodology was based on the approach of Poon et al. (2004) who were the first to formalize two measures of asymptotic dependence/independence for two random variables - these will be used in this chapter too. The first measure is rather intuitive. Asymptotic dependence is examined based on the conditional probability that one variable takes extreme values given that the second variable is taking such values. If the limit of such a conditional probability goes to zero when we move more deeply into the tails of the distributions, then the two variables are said to be asymptotically independent. Otherwise, if the limit is non-zero, they are considered to be asymptotically dependent. A second measure is the measure of extreme association in the tails. It shows the speed with which conditional probability decays to zero. It has been proved (Ledford and Tawn, 1996) that this second measure equals one for all asymptotically dependent variables but is less than one for asymptotically independent ones. Consequently, the decision about asymptotic dependence is taken based on a test of equality to one of the second measure. If this hypothesis cannot be rejected, the two variables are said to be asymptotically dependent and the limiting conditional probability is computed. If the above hypothesis can be rejected, the two variables are said to be asymptotically independent and the conditional probability is zero at the limit. Poon et al.'s approach was discussed and applied in a comparative manner by Schmuki (2008) who also provided a Matlab code for its practical implementation. In this chapter, we employ Poon et al's approach and a slightly adjusted version of Schmuki's code to compute the two measures of asymptotic dependence. Contagion in other markets, using tools from EVT, has been studied by Hartmann et al. (2004). Focusing on the co-movement of extreme returns in bond and stock markets in the G5 countries, these authors found that the potential of co-crashes in stock markets and bond markets was substantial. Moreover, contagion from stock to bond markets was as frequent as flight to quality from stocks to bonds in times of crises of the former. International crisis linkages were similar to those found in the national context, a result that underscored the downside risk of financial integration. Hartmann et al. (2010) focused on contagion in exchange rate markets in relation to the statistical properties of the exchange rate fundamentals. Although interesting insights are gained from these chapters, their methodological approach is different from the one used in this chapter and will not be further commented on here. 63 5.3 Exchange rate developments and crisis episodes in CEE At the beginning of their transformation Central European economies (CEE) had limited capacity for absorbing large exchange rate fluctuations and that is why they initially preferred currency arrangements that limit the flexibility of the exchange rate (basket peg, adjustable peg, crawling peg). The main factors were poor development of markets, liberalization of prices and trade at the beginning of the transformation. Economic growth and the liberalization of the capital account attracted foreign direct investments. Subsequently, in the second half of the 90s, remarkable progress was made with respect to disinflation; economic development was accompanied by political and social pressure. Under such circumstances, many countries had to resist speculative attacks against their domestic currencies, which resulted to sharp movements of exchange rates (Chart 1). The most visible from among the Central European economies was the situation at the Czech Republic in May 1997. During this advanced transition phase, the Czech Republic, Poland and finally also Hungary switched to more flexible regimes (managed floating or free floating) with an inflation targeting framework (Table 5-1). This change in exchange rate strategy was consistent with both domestic factors (the progressive capital account liberalization in the CEEs) and external factors (the increasing risk of speculative attacks as a result mainly of Asian 1997 and Russian 1998 financial crisis). On the basis of the described changes of monetary and exchange rate regimes in CEE one can distinguish the main periods for subsequent empirical investigation. In the Czech Republic we can distinguish three key periods. The first (1993:01 -1996:02) was an exchange rate targeting period with conventional fixed parity of the Czech koruna. The second (1996:02-1997:12) period was related to transitional monetary strategy toward inflation targeting. An intermediate exchange rate regime in the form of a corridor was implemented, followed by the process of moving to a managed floating exchange rate regime- as a result of significant exchange rate turbulences in May 1997. The third (1997:12-now) period was an inflation targeting period combined with a managed floating exchange rate regime, which used FX intervention at the beginning of this period (until 2002:09) as a tool for macroeconomic stabilisation. The CNB Bank Board decided to use the exchange rate as a monetary policy instrument, and therefore to commence foreign exchange interventions, on 7 November 20 1 3.32 For the Czech Republic, as a small open economy with a long-term excess of liquidity in its banking sector, this is a more effective instrument for easing the monetary conditions than any other. The CNB Bank Board decided to end the CNB's exchange rate commitment on 6 April 2017. The decision to use the koruna exchange rate as a potential additional tool for monetary policy easing after the lower bound on interest rates was reached was made in autumn 2012 64 Figure 5-1 Development of Koruna, Forint and Zloty against USD and EUR Czech Koruna Hungarian Forint Polish Zloty 4.1.1993 27.6.1998 18.12.2003 9.6.2009 30.11.2014 -usd -eur 400 350 300 250 200 150 100 50 i A v!rj Pír 4.1.1993 27.6.1998 18.12.2003 9.6.2009 30.11.2014 -USD -EUR USD/EUR 4.1.1993 27.6.1998 18.12.2003 9.6.2009 30.11.2014 -USD/EUR Source: CNB, MNB, NBP, and Thomson DataStream. The Polish experience with exchange rate management leads us to distinguish also three main periods. In the first (1990:01-1995:05) Polish zloty plays the role of nominal anchor. The Exchange rate regime was arranged as conventional fixed parity and crawling peg with a decreasing rate of crawl. In the second (1995:05-2000:04), again as in the Czech Republic, a transitional monetary strategy toward inflation targeting was applied. The exchange rate regime was designed as a crawling corridor regime with widening fluctuation margins and a decreasing rate of crawl. The recent period (2000:04-2009:01) is characterized as a period of explicit inflation targeting and free exchange rate floating. FX intervention was not used as the tool of monetary policy. The Hungarian strategy with exchange rate regimes was slightly different compared to the Czech or Polish one. It was oriented on a long term basis to different peg arrangements, which delivered the possibility of balancing between fixed and floating exchange rates. The Hungarian case could be divided into four main stages of development. Firstly (1990:01-1995:03) they applied an adjustable peg, which also played the role of nominal anchor. The spring of 1995 saw a stabilization program, because Hungary was regarded by international financial institutions as the next candidate for financial crisis after Mexico in 1994. The second stage was (1995:03-2001:04) oriented to the application of a crawling peg, which was afterwards changed to a horizontal peg (2001:05-2008:02). From the end of 2008 Hungary applied free floating regime. 65 Table 5-1 Exchange rate regimes: Czech Republic, Hungary and Poland Czech Republic_ 03/03/1993 - 29/02/1996: Basket Peg (65% DEM, 35% USD), Band ±0.5% 01/03/1996 - 26/05/1997: Basket Peg (65% DEM, 35% USD), Band ±7.5% 27/05/1997 - present: Managed Float (07/11/2013 - 06/04/2017: Exchange Rate Commitment) Hungary_ 02/06/1993 - 15/05/1994: Adjustable Peg (50% DEM, 50% USD), Band (±0.3; ±2.25%) 16/05/1994 - 15/03/1995: Adjustable Peg (70% ECU, 30% USD), Band (±0.3; ±2.25%) 16/03/1995 - 31/12/1998: Crawling Peg (70% ECU, 30% USD), Band ±2.25% 01/01/1997 - 31/12/1998: Crawling Peg (70% DEM, 30% USD),, Band ±2.25% 01/01/1999 - 31/12/1999: Crawling Peg (70% DEM, 30% USD), Band ±2.25% 01/01/2000 - 30/04/2001: Crawling Peg, (100% EUR), Band ±2.25% 01/05/2001 - 25/02/2008: Horizontal Peg (100% EUR), Band ±15 26/02/2008 - present: Free float Poland_ 14/10/1991 - 05/03/1995: Crawling Peg (45% USD, 35% DEM, 10% GBP, 5% FRF, 5% CHF), Band ±0.6% 06/03/1995 - 15/051995: Crawling Peg (45% USD, 35% DEM, 10% GBP, 5% FRF, 5% CHF), Band ±2% 16/051995 - 24/02/1998: Crawling Peg (45% USD, 35% DEM, 10% GBP, 5% FRF, 5% CHF), Band ±7% 25/02/1998 - 31/12/1998: Crawling Peg (45% USD, 35% DEM, 10% GBP, 5% FRF, 5% CHF), Band ±10% 01/01/1999 - 25/03/1999: Crawling Peg (45% USD, 55% EUR), Band ±10% 26/03/1999 - 11/04/2000: Crawling Peg (45% USD, 55% EUR), Band ±15% 12/04/2000 - present: Free float Source: Czech National Bank, Magyar Nemzeti Bank, National Bank of Poland A summary of in-sample extreme movements of the exchange rates is displayed in Table 5-2. Although longer span data were available for the Central European countries, their extreme statistics are shown here only over the period used to assess contagion. Table 5-2 shows the lowest/highest daily changes of the exchange rates over the period January 1 st, 1999 - February 29th, 2012, together with the specific dates when these values occurred. For example, the maximum daily appreciation and depreciation values of the Czech crown were 5.737% (29th of October 2008) and 4.999%) (4th of April 2002), respectively. To get a better glimpse on how crisis events are identified in the chapter, the threshold values defining the tails are also shown. For example, in the Czech case, extreme depreciation changes are those exceeding the 1.308%) daily value which is the 95% quintile of the empirical distribution of the Czech daily exchange rate changes. 66 Table 5-2: Extreme values and tail defining thresholds of the exchange rates Left tail - Appreciation Min Date Tail Threshold Date cz -5.737% 29.10.2008 -1.256% 1.10.2010 EU -4.617% 19.3.2009 -1.054% 8.10.2004 HU -5.520% 29.10.2008 -1.452% 13.12.2004 PL -21.487% 5.1.2009 -1.339% 30.12.2010 Right tail - Depreciation Max Date Tail Threshold Date CZ 4.999% 4.4.2002 1.308% 26.10.2010 EU 3.845% 19.12.2008 1.056% 11.2.2009 HU 6.967% 10.10.2008 1.580% 26.6.2003 PL 23.061% 2.1.2009 1.523% 14.12.2007 We are aware that this "crisis" identifying method might rely considerably on in-sample information. However, perfectly objective guidelines for identifying asset crises are rarely available in empirical work. We think that our method is still superior to crisis identifying criteria of the type "plus/minus two standard deviations", which, besides the fact that they exploit the same in-sample information, may be subject to additional and often neglected limitations33. The analysis undertaken here should be viewed as an attempt to analyse coordinated extreme exchange rate movements. This could offer to policy makers in the concerned countries a first indication about the potential of synchronized exchange rate crises. 5.4 Loss absorbency in resolution as a major potential threat To test for one structural change in the trend function of a variable when information about the stationary nature of the variable is not available, we apply the methodology of Perron and Yabu (2007). Their model specification is similar to Perron (1989) and allows the implementation of three types of structural change: To mention only one, is the fact that some empirical distributions might have such fat tails that computing their second moment is not possible. In these cases, the "plus/minus two standard deviations" rule is completely flawed. 67 a change in intercept (model 1) X^a + b-t + ß-DUi+a-X^ + YAt-AX^+e, 1=1 a change in slope (model 2) k X=a + b-t + y-DT+a-X,,+Yk.-AX,.+s, i it t—i l t—l i a mixed model allowing for a change in both intercept and slope (model 3) k Xt =a + b-t + ß-DUt + r-DTt+a-Xt_1+YJArAXt_i + st i=i Here DU and DT are dummy variables controlling changes in intercept and slope, respectively: fl ift>TB \t-TB ift>TB DU, =\ and DTt=\ [0 otherwise [0 otherwise and TB is the supposed time breakpoint. The augmented form of the regression is used to correct for serial correlation in error terms. The procedure is a sequential one. It requires computing Wald statistics (Wt) for testing the null hypothesis that the coefficients of the relevant dummy variables are zero at each considered break point candidate. The Exp-functional of Andrews and Ploberger (1994) is further constructed based on Wald statistics at all considered break points: E xwp= 1 o N The ExpW functional has almost identical limit distributions under both assumptions of 1(0) and 1(1) residuals, thus providing a testing procedure with similar sizes in both cases. It also has good power properties in finite samples given the use of a bias-corrected value of the autoregressive parameter a. The critical values of the test are determined by simulations and are based on the asymptotic distributions of the ExpW test. The decision about the existence of a break is taken in a standard way, involving a comparison of the computed ExpW statistic with the critical values at the chosen significance level. In terms of EVT, a relatively standard approach is followed in this chapter. At the univariate level we assess the degree of tail fatness of the distributions using the tail index. A distribution has heavy tails if it varies slowly at infinity, in other words if a positive parameter a exists such that: 1 i rrr-tt~ - x <~^-F(t) ,x>0. This means that in the case of a distribution with fat tail, tail probabilities decrease according to a power law, which is much slower compared with the exponential decay followed in the case of the normal distribution. 68 The parameter a is called the tail index and is customarily estimated with the Hill estimator: K -i-l 1 I'* X N-I+l a = K 1=1 X N-K Here K represents the number of observations in the tail and the values in the sum are the values above the chosen tail threshold. The inverse of the parameter a (y, or the shape parameter) describes the shape of the tail. Positive values of y are characteristic for distributions with fat tails, while a y value of zero is representative for the normal distribution. For a positive y, the number of existent moments of the distribution is determined by the tail index a. Thus, the number of moments that can be reliably computed for a distribution with fat tails equals the greatest integer that is less or equal to a. Turning to multivariate EVT, a measure of asymptotic dependence can be derived starting from conditional probabilities of the type: This gives the probability that the random variable X takes an extreme value given the occurrence of an extreme event in Y. Here extremeness is defined with the q quintile, which is in general bounded by the 10% value on both ranges of the distribution. In our case, the 5% and 95% left and right ranges have been used. Asymptotic dependence in the right tail is defined with the limit of the above conditional probability when q tends to one: We follow the approach of Poon et al. (2004) who describe the asymptotic dependence structure in the bivariate case with the help of the already mentioned two measures (x>x), the first of which is a limit of the type defined above and the second is a measure of the speed of convergence of the conditional probabilities to zero. If % is non-zero, the variables are said to be asymptotic dependent and the limit % measures the degree of such dependence. If % is zero, the variables are asymptotic independent but the parameter^ measures the amount of extreme association or the speed with which extreme events converge to zero for both tails. In this chapter the approach of Poon et al. is closely followed. We first apply unit Frechet transformations to the original data in order to eliminate the impact of the marginal distributions on the bivariate distribution function but to preserve the original dependence structure. The parameters % and x are computed for the transformed series and the decision regarding asymptotic dependence/independence involves the following steps: (1) test the null hypothesis % = ^(% follows a normal distribution), (2) if this hypothesis is 69 rejected the series are asymptotic independent (%=0), (3) if X =1 cannot be rejected the variables are asymptotic dependent and compute %, the final asymptotic dependence measure. 5.5 Empirical findings The representative assets are the exchange rates of three Central European countries (Czech Republic, Hungary and Poland) and the Euro vis-a-vis the US dollar. The quest for fundamentals is based on a money-income model (see, for example Engel and West, 2003) that is summarized by the following equation: st=a0+ar (m, -mt *)+ a2 ■ (yt -yt*) + ar (pt - pt *)+ a, ■ (it -it *)+e, Here st is the logarithm of the nominal exchange rate versus the dollar, mt is a measure of money supply (Ml), yt is a proxy for output (industrial production, IP), pt is the Consumer Price Index (CPI) and it is the money market interest rate (IR). Excepting the interest rates, which enter the regression as differences from the US interest rate values, all other variables are expressed in logarithmic form and are measured relative to the corresponding US variables. Dividing the variables by the corresponding US values offers a convenient way to isolate common external shocks affecting the variables. Equation 1 can be viewed as a combination of different simple exchange rate determination models, i.e. purchasing power parity, interest parity conditions or the asset view of the exchange rates, viewing exchange rates as determined by the ratio of two monetary stocks.34 5.5.1 Unit root tests Cointegration tests can be conducted only among variables with the same order of integration. Preliminary standard Dickey-Fuller and Phillips-Perron tests including a linear deterministic trend and an intercept, suggested that all nominal exchange rates (in logarithmic form) and the majority of the macro variables considered were I (1) processes. However, we further tested for the presence of structural breaks in the deterministic functions of the variables. As already mentioned, this was done to account for the reduced power of unit root tests in the presence of structural breaks. The reduction in power of the unit root tests would imply a non-rejection of the unit root hypothesis even if the data were in reality stationary around a broken deterministic trend. The Perron and Yabu methodology was applied in considering two types of structural change: (a) a change in the growth model described by a change in the The data were collected primarily from the International Financial Statistics (IFS) database of the IMF. However, a few variables were not available there and in those cases alternative data sources were used (Datastream and the Arad database of the Czech National Bank). 70 slope of the deterministic trend (model 2) and (b) a mixed model that considers a change in both slope and intercept (model 3). Beyond accommodating one-time changes of the mentioned type, mixed models additionally offer a good approximation for trend changes, which are not one-time events but take place gradually in time. With the exception of the Euro/USD exchange rate variable, the hypothesis of a break cannot be rejected in any other exchange rate case. Thus, for the Euro/USD exchange rate, standard ADF and Phillips-Perron unit root tests were applied concerning its stationary nature. These tests suggested that this variable contained a unit root. It is worth mentioning that in many cases the existence of a break could be rejected according to model 2 but not according to model 3. This result does not necessarily deny the existence of a break in slope. It brings evidence that the change was rather a gradual adjustment and not a one-time change. The final decision about the existence of a structural break will in the end be taken based on the results obtained with model 3. In all the cases where the presence of a break was confirmed by the Perron and Yabu tests, Kim and Perron unit root tests were unable to reject the unit root hypothesis. Overall, it appears that, after controlling for the presence of a structural break, all variables are characterized by the presence of a stochastic trend. Searching for higher orders of integration, the unit root hypothesis was rejected for differenced variables in all cases excepting Polish CPI. Thus, almost all series appear to be 1(1). In the Polish CPI case, the unit root hypothesis applied to the differenced series was rejected with the Phillips-Perron test but not with ADF. It is thus arguable whether this variable is 1(2) or not. 5.5.2 Cointegration Given that the majority of variables are 1(1), we tested for the existence of a cointegration relationship of the type described in (1) using a standard Johansen methodology. In the Hungarian case both trace and rank tests supported the existence of one cointegration relationship. In the Czech, Polish and the EU cases the presence of two cointegration relationships was supported by these tests. In the Polish case, by excluding the CPI variable one cointegration relationship was supported among the remaining variables. It might be the explosive nature of the CPI that did not allow it to be cointegrated with the remaining variables. In the Czech and the EU case one cointegration relation could be found by excluding the Ml variable. In all these three cases, the variables that contained a stochastic trend not integrated with the others were eliminated from the analysis. The cointegration relationships were estimated with a Canonical Cointegration Regression (CCR) method and are displayed in Kadlčaková and Komárek (2017). They express the equilibrium relations between exchange rates and their macroeconomic fundamentals. As a final step, the equilibrium exchange rates were computed as a linear combination of the macro variables entering the cointegration relationship. 71 A graphical representation of the actual daily exchange rates and their monthly equilibrium levels is contained in Kadlčáková and Komárek (2017). The final time span differs among countries, given different time availability for different variables at the country level. When implementing the EVT approach, the time span is restricted to the longest common denominator for all variables, which is 1st of January 1999-29th of February 2012. Implementing the EVT approach requires variables that are identically and independently distributed. However, correlograms of the deviation from equilibrium series obtained so far35 (at daily frequency) showed strong evidence of first-order autocorrelation, with the potential of second-order autocorrelation in the Polish case. Additionally, the variance of these series was not constant over time, implying that the assumption of homoscedasticity was also violated. For these reasons, we filtered out the autocorrelation and heteroscedacity from the deviation series by estimating GARCH regressions in which the mean equation contained lagged terms of the specified orders and the volatility was modelled through GARCH specifications of adequate orders. In order to account for error term distributions with heavy tails, the assumed error distribution in these regressions was a Student t distribution. In the case of the exchange rate return series only the homoscedasticity assumption was not met. Thus, in this case the GARCH modelling considered only a constant in the mean equation. Table 5-3 Tail indices of the distributions Deviations from equilibrium Exchange rate returns Right Left Right Left (depreciation) (appreciation) (depreciation) (appreciation) cz 2.13 2.04 3.39 3.47 EU 2.17 2.04 4.28 4.00 HU 2.51 2.35 3.71 4.10 PL 2.17 2.03 3.45 3.87 It is clear that extreme values are present, a fact also reflected by the heavy tails of the empirical distributions. In fact, in all cases the kurtosis largely exceeds the value 3 characteristic for normal distribution (it takes values between 22 and 38) and the skewness also suggests deviations from the normal distribution36. Although the normality assumption is rejected in all these cases, the third and These residuals should not be confounded with the residuals from the cointegration tests, which should satisfy the i.i.d. condition given the inclusion of lagged terms in these tests' specifications. In fact, the tail indices of these distributions are less than 3, suggesting that the third and fourth-order moments do not even exist in these cases. 72 fourth order moments show values closer to those representative for the normal distribution (somehow less so in the Polish case). 5.5.3 Extreme value theory The mentioned EVT tools are applied to assess the degree of asymptotic dependence among different distributions. The analysis takes into account both the left and the right tails, thus separately examining depreciation and appreciation episodes, both in terms of exchange rate returns and deviations from equilibrium. The tail indices (a parameter) at the country level are given in Table 5-3 5-3. Table 5-4 Measures of bilateral asymptotic dependence a) Deviations from equilibrium series Depreciation (right tail) Appreciation (left tail) Hypothesis Hypothesis x 1 X 1 cz_ EU 0.960 Not rejected 0.924 0.924 Not rejected 0.946 cz_ HU 0.944 Not rejected 0.944 0.906 Not rejected 0.945 cz_ .PL 1.026 Not rejected 0.944 0.875 Not rejected 0.936 HU _EU 0.933 Not rejected 0.924 0.968 Not rejected 0.945 PL_ EU 1.026 Not rejected 0.924 0.922 Not rejected 0.936 HU _PL 0.925 Not rejected 0.947 0.946 Not rejected 0.936 b) Exchange rate return series Depreciation (right tail) Appreciation (left tail) Hypothesis Hypothesis X 1 X 1 CZ_ EU 0.969 Not rejected 0.929 0.962 Not rejected 0.940 CZ_ HU 0.966 Not rejected 0.947 0.950 Not rejected 0.940 CZ_ PL 0.947 Not rejected 0.872 0.950 Not rejected 0.940 HU _EU 0.962 Not rejected 0.929 0.986 Not rejected 0.950 PL_ EU 0.906 Not rejected 0.872 0.978 Not rejected 0.947 HU _PL 0.941 Not rejected 0.872 0.969 Not rejected 0.947 As can be seen from Table 5-3 both distributions display fat tails (y>0). For the exchange rate return series, the existence of third-order and in some cases fourth-order moments is assured since a is greater than 3. However, for the deviation from equilibrium series the maximum number of moments is two, meaning that only the mean and the variance can be reliably computed from their empirical distribution. Here extremeness is defined with the q quintile, which we chose to represent the below 5% and above 95% ranges of the distributions. The 73 estimated parameters % and % according to the Poon at al.'s approach are shown in Table 5-4 . Four cases are again distinguished, involving the two distributions and their left (appreciation) and right (depreciation) tails. The results suggest that significant tail dependence is present among all the pairs of exchange rate variables considered in this chapter. 5.6 Conclusion The objective of this chapter was to empirically analyse the potential for contagion in three exchange rate markets in Central Europe and the EU. Tools pertaining to Extreme Value Theory offered a suitable methodological approach and were used in conjunction with unit root tests allowing for the presence of structural breaks and cointegration. The main finding of the chapter is that the potential for contagion in the exchange rate markets of this region is particularly high. Conceived both in terms of currency crises and deviations from equilibrium, we found that all pairs of examined exchange rates exhibited high values of asymptotic dependence both on the depreciation and appreciation side. A further insight into the behaviour of exchange rates in this region was offered by the tests of structural changes implemented in conjunction with the unit root hypothesis. It is interesting to note that with only one exception, all the variables used in this chapter showed evidence for a structural break. The presence of such breaks is usually neglected in the empirical studies dealing with these markets and this might render the conclusions reached there less reliable. Another interesting result of the chapter was that support for cointegration was found among all exchange rates and the small set of macro variables that we proposed as fundamentals. This result shows that these markets function in accordance with basic theoretical models, if not on a standalone basis at least as the interplay of more factors. Based on cointegration we were also able to distinguish episodes of divergence from equilibrium. It is worth noting that these were mostly predominant during the early transition period and accentuated to some extent during the recent global financial crisis. The chapter also offered interesting insights into exchange rate developments in these countries from a long-term perspective. 74 Chapter 6 Risks associated with the transition to fixed exchange rate regimes By Mojmír Helísek 6.1 Introduction One of the source of financial instability are currency (foreign exchange, speculative) crises associated in particular with fixed exchange rate regimes. Member states of the EU replacing their national currency with the euro will (sooner or later) also have to transition to such a regime. This is one of the Maastricht convergence criteria (the criterion of exchange rate stability), which in particular means to peg the given currency to the euro in Exchange rate mechanism ERM II for a period of at least two years. A condition is to maintain the so-called "normal fluctuation margin" without "severe tension", in a manner that prevents devaluation of central parity. The objective of this chapter is to specify the circumstances under which a national currency becomes part of ERM II (including historical experiences) and find potential risks of currency crisis during the transition to a fixed exchange rate regime. After a review of the literature, there follows an explanation as to why these regimes are more susceptible to currency crises. The next section of the chapter focuses on empirical facts - what exchange rate regimes were (or have been) used in ERM II and whether they have been associated with currency crises. This empirical section is followed by a theoretical section which first characterizes three generations of currency crisis models and seeks a suitable model for ERM II conditions. The continuation of this theoretical section contains an articulation of the risks that lead to a hypothetical currency crisis in ERM II as well as the circumstances that weaken these risks. The last section is devoted to the Czech koruna become part of ERM II in terms of selecting a suitable exchange rate regime. 75 With regard to methodology, we use two approaches for assessing currency crisis risks when a national currency becomes part of ERM II. First we evaluate (through analysis and comparison) the existing empirical knowledge associated with national currencies becoming part of ERM II. We define the criteria for "currency crisis", which we then compare with indicators of its actual development: • exchange rate development, • indicators of pressure on exchange rate (i.e. "severe tension"). Has a currency crisis ever occurred in ERM II? For this we use statistical data from Convergence Reports of the European Central Bank and from national statistical databases. Second we use various theoretical models of a currency crisis (application modelling). Which of these models can be applied to the conditions of a peg in ERM II? We evaluate the selected model as follows: • in its original (general) variant for any "peg" exchange rate, • in our (specific) variant for a peg in ERM II. The result is our modified model containing specific risks and specific causes of their weakening in ERM II. The subject of our research is therefore the connection between the fixed exchange rate and the hypothetical currency crisis. We do not deal with many other consequences of fixing the exchange rate, i.e. the loss of an independent monetary policy and of the exchange rate policy, especially the danger of the so-called internal devaluation.37 6.2 Literature review Empirical studies prove that currency crises more frequently occur for currencies maintaining a fixed exchange rate regime when compared to currencies maintaining a flexible exchange rate regime. Older studies (such as Bubula and Otker-Robe, 2003) and newer studies alike (such as Zhao et al., 2014, or Melvin and Norrbin, 2017) have reached this conclusion. A range of authors38 report the risk of a currency crisis using fixed exchange rates in ERM II. We differentiate this literature into two timeframes, prior to and following accession of new EU Member States to the euro area (beginning with Slovenia in 2007). Most frequently the risk of a fixed exchange rate in ERM II is associated with concurrent free movement of capital. According to Begg et al. (2002, p. 70): 37 "Today, studies seem to recognise that without an autonomous monetary policy and a flexible exchange rate, their economies might be forced to undergo painful "internal devaluations" in cases of severe asymmetric shocks." (Gabrisch, Kampfe, 2013, p. 181). 38 For the position of Czech authors (in relation to the Czech koruna) see section 7. 76 "However, crises, particularly of the contagion type, cannot be ruled out in any scenario that combines full capital mobility with the ERM-II." "However, crises, particularly of the contagion type, cannot be ruled out in any scenario that combines full capital mobility with the ERM-II." Similar statements are made by Egert - Kierzenkowski (2003, p. 22): "In the context of fully mobile capital flows, the defensible nature of the asymmetric band,39 especially on the weaker side seems to raise some doubts in times of financial turmoil." whereas not even the 15% limit of deviations around central parity are sufficient to reverse a strong attack." Krawczyk (2004, pp. 7-9) compares the ERM mechanism (part of the European Monetary System, EMS) with ERM II. He sees risks of becoming part of ERM II in the following relevant directions: 1) In the original ERM system the parity of two currencies was always established and both countries of the exchange rate pair were involved in retaining this parity (via foreign exchange market interventions). In the ERM II system this consists of maintaining parity against the euro and the ECB (according to Krawczyk) has no obligation to maintain this parity. For this reason, even under a relatively broad fluctuation band the ERM II is just as susceptible to crises as the original ERM exchange rate mechanism. 2) Membership of new countries in the EU was conditioned on free mobility of capital. There are a range of circumstances such as loss of faith in economic policy or increased inflation expectations that lead to rapid outflow of capital, which is not compatible with maintaining a peg in ERM II. Krawczyk reaches the following conclusion: "It seems possible to argue that insisting on the ERM II participation [... ] means a disregard to the experience of the 1990s currency crises and makes the waiting period inside the ERM II likely to become a self-defeating experiment." (p. 9) For similar findings see Backe et. al. (2004, p. 6): "Acceding countries regard ERM II as an intermediate exchange rate regime, subject to risks of speculative attacks." Dyson (ed., 2006) warns against "speculative attacks" while joining ERM II: "Critics loath the ERM II as a »soft peg« prone to speculative attacks" (p. 16340). The cause of these attacks is the destabilization of investor expectations, prompted on the one hand by monetary policy, on the other hand by exchange rate fluctuations. It is explained here also why not only the Czech Republic but also Poland seeks to spend as little time as possible in ERM II. "The concern in both cases is that financial markets will use participation in the ERM II as an excuse to speculate against their national currencies." (p. 99, chapter author E. Jones) Likewise, Baldwin and Wyplosz (2006, p. 397) consider entry to The asymmetry of the fluctuation band lies in its interpretation by the European Commission: the original band of 2.25% in the direction of depreciation, and the band of 15% in the direction of appreciation, should be respected. The author of this chapter (F. Bonker) continues: "Participation in ERM II thus requires a far-reaching shift in monetary and exchange-rate policy, which can further increase the risk of a currency crisis by destabilizing the expectations of investors." 77 ERM II as "a delicate step where full capital mobility and an exchange peg may trigger speculative attacks." Concerns of currency crisis at the time of becoming part of ERM II also appear after the experiences of new member states. Michalczyk characterizes risk of speculative attacks as follows (2011, p. 128): "[...] it must be remembered that formal accession to the ERM II, although assumed to result in a higher degree of the exchange rate stability, may cause tensions in the foreign exchange market, being a consequence of speculation and the desire to "test" the authorities by market entities (vide European currency crisis in first half of the nineties)." Palankai presumes that the Hungarian forint may be considered an example of the threat to currency in ERM II. It was pegged to the euro with oscillation of ± 15% in the years 2001-2008. Free floating was then introduced "just due to the speculative threats of the financial crisis" (Palankai, 2015, p. 54). De Grauwe (2016, p. 155) considers ERM II a merely temporary regime that enables rapid acceptance of the euro. In the opposite case (when entry to the euro area is put off), participation in ERM II is undesirable. "It may then face similar problems to those the EMS experienced in 1992-1993 with speculative crises and a collapse of the arrangement." Concerns of "Spekulationsbewegungen" during participation in ERM II are also expressed by Brasche (2017, p. 227). 6.3 Fixed exchange rate and currency crisis According to the standard definition, a currency crisis is a significant depreciation of the nominal exchange rate of a given currency. This devaluation is caused by loss of investor confidence in this currency. This leads to an expectation of devaluation of this currency.41 Investors therefore transfer their assets in this currency into assets in other currencies. These represent two groups of investors. In the case of investors engaged in speculation, these trades are designated as "attacks by greedy speculators" who wish to maximize their profits. In the case of investors who diversify the assets in their portfolios, this is considered the "flight of careful investors" who want to minimize their losses. Central banks typically oppose a fixed exchange rate after a certain time, which leads to a decrease in their foreign exchange reserves and to interest rate increases. They can also implement administrative control of capital movements. However, investors ultimately force the central bank to devalue or (more often) to abandon the fixed exchange rate of the currency and its subsequent depreciation. When defining a currency crisis empirically, a distinction is typically drawn between the narrower and broader concept of a crisis (Glick and Hutchison (2011, pp. 7-8). According to the narrow concept, a currency crisis results if the exchange rate of the given currency vis-á-vis reference currency (typically USD) exceeds the average annual level of devaluation of 25% and simultaneously the increasing 41 The most common causes of devaluation of expectations (and simultaneously indicators of a currency crisis) are worsening government balance, falling central bank reserves, increasing money market rate (Babecky et al. 2012, p. 24). 78 of the rate of depreciation (year-on-year basis) is at least 10 p.p. (this criterion, which is still in use, was introduced by Frankel, Rose, 1996, pp. 352-353; other authors modify it in various ways). According to the broader concept a currency crisis is identified using an index of pressure on the exchange rate containing not only the above exchange rate but also a change to foreign exchange reserves and interest rates, always against a specific reference value (the standard concept of this index was implemented by Eichengreen et al., 1996, pp. 474-475; elaborated further by Kaminsky et al. 1998). Apart from these two concepts of a currency crisis, a currency crisis may be defined solely using "qualitative criteria" such as a forced change of parity, abandoning the fixed exchange rate, or international aid (Bordoetal., 2001). Currency crises were relatively frequent in the 1980s and 1990s, when they affected 5-10 currencies on average each year. At the beginning of the 21st century their frequency dropped significantly; it once again rose in connection with the financial crisis from the year 2007. According to Glick and Hutchison (2011, p. 19) in the years 2008-2009 a devaluation of 25% or more occurred with 23 currencies. Bush et al. (2011, p. 7) cite the frequency of currency crises as an average of 5.4 per year from 1973-1989; 2.4 per year from 1990-2009. Currency crises most often affect currencies maintaining a fixed exchange rate regime. In the case of currencies with flexible exchange rates, the defence of the exchange rate occurs to a far lesser extent (if at all) through interventions by the central banks on foreign exchange markets. Therefore, foreign exchange reserves are not exhausted, unlike from fixed exchange rates, where temporary defence of the exchange rate leads to loss of part of these reserves. A sufficient amount of foreign exchange reserves (in the case of a flexible exchange rate) then reduce investor concerns that they will be unable to convert their receivables from the given currency into a different currency. Investors therefore do not succumb to panic. "The combination of depleted reserves plus the broken promises [to maintain a fixed exchange rate - note by M. H] leaves the country very vulnerable to panic. With a floating rate system, countries can maintain their foreign reserves and thereby maintain a defence against financial panic." (Ghosh, 2001, pp. 306-307).42 A currency crisis is therefore less probable in the case of a flexible exchange rate regime. In all significant cases of currency crises such as the Mexican crisis 1994-1995, Asian crisis 1997-1998, Russian crisis 1998-2000, Brazilian crisis 1999, and the Argentinian crisis 2002, the affected economies maintained one of the variants of a fixed exchange rate (see Helisek, 2004). The same goes for other currency crises - examples are the Turkish crisis 2000-2001, the Icelandic crisis 2008, or the Russian crisis 2014-2015. According to an MMF study43 focusing on all MMF Member States in the years 1990-2001, out of a total 196 identified cases of currency crisis 73% resulted The authors of the chapter (Lesson from the Asian financial crisis, pp. 295-315) are Radelet, S., Sachs, J. Bubula, Otker-Robe (2003), p. 13. For more details see Helisek et al., 2007, pp. 57-59. 79 in crisis with currencies maintaining a fixed exchange rate, and only 27% with currencies maintaining a floating exchange rate. As part of the "fixed exchange rate" regime, the regime of a peg was most susceptible, accounting for 40% of cases of currency crisis. The following case specifies the fixed exchange rate regimes most susceptible to crisis (share of the given regime to total number of currency crises in %): - peg .................................................39.8% horizontal band....................................11.2% crawling peg.......................................10.2% crawling band......................................4.6% currency board..................................... 1.5% Another study (Zhao et al., 2014) examined the currencies of 88 countries in 1981-2010. 167 currency crises were assessed in three groups of exchange rate regimes (simplified): peg, currency board, horizontal band) narrower than ± 2%.............. 19% crawling peg, horizontal band) wider than ± 2%, managed floating .... 71% freely floating..................................................................... 11% Newer studies also reach this same conclusion (Melvin and Norrbin, 2017, p. 213): "Fixed exchange rates encouraged international capital flows into the countries ... Once pressures for devaluation began, countries defended the pegged exchange rate by central bank intervention ... and the fixed exchange rate is abandoned." Table 6-1 Exchange rate regimes in ERM II Country Regime before Regime Central parity In euro (currency ERM II in ERM to EUR area code) II from Denmark since March 1979 peg 7.46038 DKK — (DKK) vERM ± 2.25% Greece since March 1998 peg 353.109 GRD 2001 (GRD) in ERM ± 15% ** Slovenia since October 1991 peg 239.640 SIT 2007 (SIT) crawling peg ±15% Cyprus since January 2001 peg 0.585274 CYP 2008 (CYP) peg to euro ± 15% ± 15% Malta since August 2002 peg 0.4293 MTL 2008 (MTL) peg to a basket of currencies * ±0% Slovakia since October 1998 peg 38.4550 SKK 2009 (SKK) managed floating ±15% * * * Estonia since June 1992 currency 15.646 EKK 2011 (EEK) board 80 currency board to euro Latvia since February peg 0.702804 LVL 2014 (LVL) 1994 peg to SDR± 1% ± 1% Lithuania since february 2002 currency 3.4528 LTL 2015 (LTL) currency board to euru board Notes: * Shares: EUR 70%, GBP 20%, USD 10%. ** Revaluation 17. 1. 2000 by 3.5% (1 EUR = 340.750 GRD). *** Two revaluations: 19. 3. 2007 by 8.5% (1 EUR = 35.4424 SKK) and 29. 5. 2008 by 17.6% (1 EUR = 30,1260 SKK). Sources: European Central Bank, Convergence Report incl. Technical Annex (various years); Oesterreichische Nationalbank (2007), pp. 22-23; Amerini, 2003, pp. 1-8; Antal, Holub 2007, pp. 314-315; Backé et al. 2004, pp. 14-15. 6.4 Currency crisis in ERM II - empirical experience The ERM II mechanism exists since 1999 along with the creation of the euro. Joining ERM II is compatible only with certain exchange rate regimes. ECB refers to a "a number of the exchange rate strategies" that can be used as part of ERM II. However, it only explicitly references strategies that are not compatible with ERM II, specifically:44 • free floating, • managed floating without a mutually agreed central rate, • crawling peg, • pegs against anchors other than the euro. Fahrholz (2003, p. 15) adds that unilateral euroization is also not permissible for participation in ERM II. Of course, unilateral euroization is not compatible with membership in the EU either (Komárek et al., 2005, p. 25). The following limited options implicitly expressed therefore come under consideration (according to previous experiences, contained in Table 6-1): peg against the euro without a fluctuation band, peg within ERM II with standard fluctuation band ± 15%, or with narrowed band that must be defined in advance, euro-based currency board. Table 6-1 gives a list of currencies incorporated into ERM II. All exchange rate regimes listed in Table No. 1 are included among the first two regimes that according to the MMF study are most susceptible to currency European Central Bank, 2003 (p. 3); this position of the ECB is de facto assumed by the ECOFIN Report of the Council of the European Union (2000), pp. 2-3). 81 crises (Bubula, Otker-Robe, see above). The sole exception is the currency board applied twice. From subsequent evaluation we can eliminate the countries that became part of ERM II only for a short period of time: Greece - the Greek drachma transitioned from ERM to ERM II, where it remained for 24 months, Slovenia - the Slovenian tolar for a mere 30 months (at the time of evaluation it had been in ERM II for a mere 22 months, like Lithuania), Cyprus and Malta - the Cypriot pound and the Maltese lira were in ERM II for 32 months, Slovakia - the Slovak koruna was in ERM II for 37 months. On average the currencies of these countries were in ERM II for 29 months. We will also monitor only four countries that remained in ERM II for longer periods. On the one hand, we will evaluate the trend of nominal exchange rate, on the other hand the indicator of "severe tension". In total this represents the following indicators (Table 6-2): maximum exchange rate deviation around central parity to the euro (plus sign indicates depreciation in the exchange rate of the given currency, minus sign indicates appreciation of the exchange rate of the given currency); the official international reserves (a comparison of their status at the end of the monitored period with the beginning of the monitored period); the interest differential measured as the difference between the three-month interbank interest rate (CIBOR, VILIBOR, RIGIBOR, TALIBOR) and the EURIBOR, always by the end of the year). From Table 6-2 the following rating is derived: the exchange rates of the monitored changes for the entire period of remaining in ERM II were either almost stable (DKK, LVL), or entirely stable (LTL, EKK); the official international reserves reflected growth trends in all countries. Significant decreases occurred in the year 2008 (in association with the global financial crisis and the debt crisis in the euro area which led to lack of faith in the European currencies). In the annual indicator, however, this decrease was not reflected; interest rate differentials demonstrated (for the same reason) higher values solely in 2008. 82 Table 6-2 Criteria of exchange rate stability in ERM II Country (currency) ERM II involvement period Duration of stay in ERM II (months) Exchange rate deviations (%) Change of international reserves (%) Interest rate differential (p. b.) Denmark (DKK) I 1999-XII 2018 218 * +0.1 /-0.5 296.2 -0.65 -2.02 Lithuania (LTL) VII 2004 -XII 2014 126 0 172.0 0.05 -7.00 Latvia (LVL) V 2005 -XII 2013 104 +1 l-\ 189.1 -0.16-10.65 Estonia (EKK) VII 2004 -XII 2010 78 0 43.6 0.11 -4.98 vfotes: * From. January 1999 tot le end of 20 18; involvement continues. Sources: EURIBOR: https://www.emmi-benchmarks.eu/euribor-org/euribor-rates.html Denmark: Exchange rate: http://nationalbanken.statistikbank.dk/909; Reserves: http://nationalbanken.statbank.dk/nbf/125955 CIBOR: http://www.finansraadet.dk/Tal~Fakta/Pages/satser/regler-for-fastlaeggelse-af- cibor/historiske-satser.aspx Lithuania: Exchange rate: https://www.lb.lt/exchange/history.asp?Lang=E&Cid=EUR&Y=2014&M=12&D=31&id =4046&ord=l&dir=ASC; Reserves: https://www.lb.lt/en/official-reserve-assets; VILIBOR: https://www.lb.lt/en/historical-data-and-external-links Latvia: Exchange rate: https://valutaskurss-eiro.lv/kursi/LVL-lats-latvija/; Reserves: For 2005-2006 (net reserves) https://www.bank.lv/en/statistics/stat-data/net-international- reserves; next years: https://statdb.bank.lv/lb/Data.aspx?id=121 RIGIBOR: https://www.bank.lv/statistika/dati-statistika/naudas-tirgus-index/rigibid- rigibor-vesturiskie-dati Estonia: Exchange rate: http://statistika.eestipank.ee/#/en/p/VALUUTA; Reserves: http://statistika.eestipank.ee/#/en/p/l 134/r/l 122/970; TALIBOR: http://statistika. eestipank. ee/#/en/p/l 010/r/l 730 These empirical findings do not confirm concerns of the currency crisis that could result during the inclusion of the currency in ERM II. Maintaining exchange rate stability can be explained by confidence of participants in the foreign exchange market in the obligation of the central bank to retain the fixed exchange rate. According to the Danish central bank: "Officially, the krone may fluctuate by up to 2.25 per cent on either side of its central rate, but in reality the fluctuations are far smaller. This reflects the high credibility of the fixed exchange rate policy [...] The credibility of the regime means that market participants take positions which in themselves stabilise the exchange rate of the krone." (Spange, Wagner Toftdahl, 2014, p. 49)45 The intention of the central bank to keep the exchange rate The Danish central bank illustrates the behavior of participants in the foreign exchange market in the case of pressures on devaluation: "Participants in the financial markets are confident that the exchange rate of the krone will continue to fluctuate within a narrow 83 in ERM II is reinforced by an effort to retain prestige. When failing to uphold the criterion of exchange rate stability the central bank would lose trustworthiness. An exception was in the years 2007-2009, when pressure ensued on the devaluation of the central parities of these currencies in relation to the euro. An expression of "severe tension" was a decrease in international reserves (with the exception of Estonia, where reserves stagnated) and the increased interest rate differential between 2-11 p.p. Thanks to interventions on the foreign exchange markets and temporary high interest rates, fixed exchange rates could be retained in ERM II in this crisis period. The Danish central bank evaluates this period as follows: "Moreover, the instruments (i.e. the intervention on the foreign exchange markets and the interest rate adjustments performed by the central bank - M. H.) have proved to be sufficiently robust to handle extraordinary situations such as, most recently, the implications of the 2008 financial crisis and the subsequent sovereign debt crisis in several euro area member states on the exchange rate of the krone" (Spange, Wagner Toftdahl, 2014, p. 50). However, it is not certain how the European Commission and ECB would rate the fulfilment of the exchange rate convergence criteria in terms of the condition of "without severe tension". If it were taken into consideration that this was a period of global financial crisis (not internal economic problems of these countries), the criterion could be fulfilled. 6.5 Theoretical models of currency crisis and their relevance in relation to ERM II Currency crises are most often associated with abandonment of a fixed exchange rate regime. Central banks are forced into this by devaluation expectations prompted by various causes (see section 3). Currency crisis models address various combinations of these causes and the judgment of authorities on performing devaluation. These models are typically divided into three generations (e.g. Krugman, 2014; Zenker, 2014). The first generation of currency crisis models are models with fundamental causes of the crisis. If the worsening of these fundamentals occurs (often under the influence of inappropriate macroeconomic policies), it leads to an outflow of capital and a decrease in foreign exchange reserves, which forces authorities of the given country to abandon the fixed exchange rate. Crises are then considered as deserved and foreseeable. In the second generation of currency crisis models, central authorities consider whether to allow the devaluation of the exchange rate, whether to abandon their band around the central rate. [... ] In a weak krone scenario, positions are typically taken in expectation of a strengthening, which has contributed to stabilising the exchange rate of the krone close to the central rate. The stabilising positions taken by market participants have reduced the need for intervention by Danmarks Nationalbank." (Ibid, p. 53). 84 obligation to maintain a peg. For this reason these models are called "escape-clause models". Central authorities compare the benefits and costs of devaluation (see below.) Currency crises in these models are unforeseeable, the worsening of fundamental quantities need not be significant; here an important role is played by self-fulfilling expectations. The third generation of models focuses on the entrepreneurial sphere (models with business balance sheets). These models work with a range of financial indicators for businesses such as financial vulnerability, moral hazard and reinvestment, and influence of foreign debt of companies growing as a result of devaluation. The models also explain the causes of "twin crisis", i.e. concurrent currency and banking crises. We presume that the most accurate models for interpreting currency crisis risk during involvement in ERM II are the second generation models, which explain the currency crisis without significant worsening of economic fundamentals. The reason for this model selection is the necessary performance of the following: Maastricht convergence criteria (low inflation, low interest rates, "healthy" public finance) supplemental criterion, i.e. low deficit of balance of payments' current account. In these currency crisis models the authorities maintain a fixed rate regime, and yet not an irrevocable fixed rate. Under certain circumstances the fixed exchange rate can be revoked using an "escape clause", i.e. the obligation to uphold a fixed exchange rate can be cancelled. What are these "certain circumstances"? The given economy can be affected by a certain exogenic shock that leads, for example, to an increase in price level. The effort of the authorities for its stability (their restrictive policies) then leads to an increase in unemployment. A worsening of fundamental economic indicators (e.g. the aforementioned unemployment) leads investors to lose confidence in this economy and its currency. For this reason they will begin to consider devaluation from which the authorities promise that it should boost competitiveness and reduce unemployment. The pressure from investors to devalue (they are getting rid of the given currency) thereby increases until the authorities finally comply. And yet the fundamental economic quantities are not decisive in this case; rather it is the devaluation expectations that are decisive. The devaluation occurs because it is expected. Out of the various variants of the second generation models, we use for interpretation of our hypothetical currency crisis under the conditions of ERM II the approach of De Grauwe (2016, pp. 102-105). The fundamental quantity monitored by investors and according to which the trustworthiness of the given currency is assessed is that of the original DeGrauwe example, a deficit of the balance of payments' current account. A different fundamental quantity could also be applied, e.g. a deficit of public finances, decreasing foreign exchange reserves, or a growing interest rate. In Figure 6-1 this quantity is measured on a horizontal axis, denoted by the symbol s. 85 Figure 6-1 Currency crisis without fundamental causes Source: De Grauwe (2016), p. 105 Authorities are aware of the unfavourable consequences of growth of this quantity on investor decision-making, and for this reason seek to decrease this quantity using restrictive policies. And yet this leads to the increased unemployment. If the authorities perform devaluation (or abandon the fixed exchange rate), the need not enact these restrictive policies and can thereby reduce unemployment. Devaluation is therefore a benefit. This benefit is measured on the vertical axis; it is designated B (benefit). The combination of s and B is expressed by the growing curve B: the less favourable the economic trends, the higher the benefit of devaluation. The shape of curve B is explained by the principle of indifferential analysis. Curve B has two variants. The lower curve Bu does not contain devaluation expectations. The higher curve Be contains devaluation expectations, i.e. it results in a "speculative attack" (under these expectations the authorities must select a stronger restrictive policy; devaluation would offer them greater benefit). And yet devaluation is also associated with costs (designated in the figure as C, costs). We presume these costs to be fixed. In Figure 6-1 there is a horizontal line C. These costs explain DeGrauwe as a loss of the reputation of the authorities. Let us now explain the associations between s, B, and C: if the fundamental quantity under observation is strongly favourable (s < si), devaluation will not occur because C > B, if, on the other hand, it is markedly unfavourable (s > S2), devaluation will occur, because B > C, if s is on a "moderate level" around s', it will then depend on investors ("markets"), expectations. 86 In the interval "si - S2": devaluation will not occur if investors do not expect this devaluation, devaluation will occur if investors expect this devaluation. In other words, the depreciation of the exchange rate (i.e. currency crisis) is caused by expectations of this crisis. This is a self-fulfilling expectation. This model of a currency crisis therefore explains that the change in fundamental economic quantities is not important when they are at a "moderate level". The devaluation (or abandonment of a fixed exchange rate) is decided by investor expectations. 6.6 Risk of currency crisis in ERM II In our application of the model of the second generation in ERM II the currency crisis is not dependent on significant worsening of fundamental quantities. In our opinion, however, investors uncertainty and their devaluation expectations may be prompted by factors other than development of economic fundamentals. These are specific causes of loss of confidence associated with participation in the ERM II mechanism. We presume that this consists of five specific risks. 1) Potential conflict between the fulfilment of two criteria, namely the criterion of low inflation at the same time as the criterion of a stable exchange rate, where the free international movement of capital takes place concurrently. This is the trilemma of currency policy in an open economy: of three goals only two are achievable. If a strong inflow of capital occurs (see below), the retention of low inflation will not be possible without appreciation of the exchange rate. This endangers the fulfilment of the criteria of a stable exchange rate. These concerns can be weakened by the following arguments: the inflation criterion allows for exceeding the "Maastricht inflation" by 1.5 p.p., the fluctuation band in ERM II is relatively wide46, enabling up to 15% appreciation under central parity, revaluation of central parity is not in conflict with the fulfilment of the exchange rate convergence criterion. 2) Establishing central parity during entry to ERM II differently in relation to the current exchange rate. Experience with becoming part of the ERM mechanisms indicates that both undervalued parity (the case of the Greek drachma) and overvalued parity (the case of the British pound and the Italian lira) lead to subsequent exchange rate fluctuations that endanger the fulfilment of the condition of exchange rate stability. It is for this reason as well that countries that became part of ERM II selected central parity to the euro that was either exactly the same or very close to the current exchange rate. Following this tried and tested strategy That is why ERM II is sometimes referred to as the "semi-fixed exchange rate" - see for example Minenna, 2016, pp. 66, 79. 87 allows the risk of lack of investor confidence toward maintaining a fixed exchange rate to be reduced. 3) Risk of appreciation overshooting of the exchange rate. While remaining in ERM II it is possible to expect a strong inflow of capital caused by increased reliability of the economy, meeting the Maastricht criteria (low inflation, favourable fiscal indicators). Confidence can also be enhanced by the expectation of accelerating economic growth arising from the effect of the growth of trade or the effect of the decrease in risk premiums upon adopting the euro. All of these circumstances can lead to appreciation pressures, which can result in overshooting the exchange rate, i.e. significant appreciation. This leads to the opposite expectations, expectations of correction of the exchange rate for central parity, i.e. its depreciation. 4) Expected shift of euro adoption date. This may occur as a result of worsening (actual or expected) of compliance with Maastricht criteria. Investors then begin to predict cancellation of central parity and subsequent depreciation of the exchange rate. 5) Refusal of entry to the euro area. This case occurred as part of ERM II only once, with the rejection of the request by Lithuania to enter the euro area (decision of the European Union in 2006 due to failure to meet inflation criterion). As we see from the development of the LTL/EUR exchange rate, even this event did not lead to devaluation of expectations and investor "attacks" on the Lithuanian currency. LTL remained in ERM II and for the entire subsequent period met the criterion of exchange rate stability. We allow that a "speculative attack" on a currency remaining in ERM II could hypothetically ensue under the influence of the above risks. It is then possible, by our opinion, to expect measures by the central banks for keeping the exchange rate in ERM II, in particular strong intervention in foreign exchange markets. We can explain this with three arguments. First: The decision to remain in ERM II, which is dependent on strict approval procedures, will be a prestigious affair for national central authorities (governments, central banks). It is a matter of retaining their trustworthiness, both in the eyes of the domestic public (citizens, firms) and those of EU authorities. Second: If a speculative attack were successful and resulted in a change of central parity in a devaluing direction (or even to the abandonment of the fixed exchange rate), it would mean the end of meeting the exchange rate criteria. Upon subsequent entry to ERM II an additional two-year minimum requirement to remain in ERM II would begin. This would prolong the high-risk period of remaining in ERM II. 88 Figure 6-2 Modification of the 2nd generation model according to ERM II conditions Source: own elaboration Third: In the event that the exchange rate should reach the boundary of the fluctuation band, the national central bank as a matter of principle has unrestricted access to "very short-term" financing on the part of the ECB. This principle has the limitation that interventions will not be supported if they could conflict with the primary goal of the ECB, i.e. price stability. The relatively small scope of the national (e.g. Czech) financial market compared to the financial market of the euro area will nonetheless not endanger this inflation risk. Now we apply to the general 2nd-generation's model of currency crisis: the above explanation of specific risks influencing investor decisionmaking, the circumstances that weaken these risks within the ERM II regime. In this way we modify this model to the conditions of ERM II (Figure 6-2): 1) We replace the fundamental quantity on the horizontal axis (s) with "specific causes of devaluation expectations" (SC). 2) The weakening of devaluation expectations shifts curve Be downward (from Be to Besc). 3) When the currency does not meet the criterion of becoming part of ERM II for at least two years, the authority of the given country loses trustworthiness. Moreover, the two-year period must be repeated, which extends the term of persisting in ERM II. This increases the costs of abandoning the fixed exchange rate and shifts line C upward (from C to Csc). 89 From Figure 6-2 it is clear that the conditions of ERM II result in a "moderate level" shift of specific causes of devaluation expectations (SC) to the right. In other words, this means that the causes of loss of investor confidence such as an inflow of foreign capital, unsuitable setting of central parity, etc., must be very strong in order to invoke this loss of confidence. This also reduces currency risk in ERM II. 6.7 Involving the Czech koruna into ERM II The Czech Republic is one of the Member States of the European Union with a temporary derogation on to the euro's introduction. None of these countries has their currency in ERM II (Table 6-3). Concerns about possible speculative attack (i.e. currency crisis) when joining the currency to ERM II are expressed for the Czech koruna as well. E.g. Janáčková (2002, p. 777) claims that financial markets "could use a firmly set fluctuation interval for speculation against the Czech currency."47 Membership in ERM II should therefore be as short as possible. Jílek similarly states (2004, 661): "The CNB in this case [when the narrow band of 2.25% attracts speculative attacks - M. H.] would have to intervene on a massive level in order for the rate to hold. The probability that it would succeed, however, is not that great." Saroch et al. (2003, pp. 48-49) repeat the concern about the possibilities for speculative attacks and recalls the proposals of certain central banks to reduce the two-year period for remaining in ERM II. Lacina, Rozmahel, et al. also warn against currency crises (2010, p. 21): "Remaining in ERM II regime for a period of two years moreover poses a risk of speculative attacks on the currency of the candidate country".48 Likewise, Sychra (2009) warns of "potential instability" for the period in which the currency remains in ERM II; Marková (2011) does as well. A similar warning is also stated by Helisek and Mentlik (2018). Fassmann, Ungerman (2018) warn of the risk of unsustainable appreciation of the exchange rate. Table 6-3 Planned date of entry to the euro area Country Original date New date State of preparations Bulgaria "as soon as possible upon meeting the Maastricht criteria", later 2010, 2020 2022 Action plan for involving into ERM II (2018). Request for entry to ERM II 29 June 2018. Czech Republic 1 January 2010; cancelled 25 October 2006 not set National Euro Changeover Plan for the Czech Republic (2003, updated 2007) The author draws attention to "the risk of speculative attacks on ERM II entry" in her next work (2014, p. 67). Likewise Lacina et al. (2008). 90 Croatia 2023-2025 Strategy for the Adoption of the Euro in the Republic of Croatia (2018) Hungary 1 January 2010; canceled 1 December 2006 not set National Euro Changeover Plan (updated 2009) Poland 1 January 2012; not set; "as National Euro cancelled at the end soon as Changeover Plan of2009 conditions for accession are met" (2011) Romania 2015,2019, 2022 2024 National Plan for Adoption the Euro (2019) Sweden not set preparations were suspended by the rejection of the Euro by referendum on 14 September 2003 Source: European Commission, 2014, and previous reports (no newer reports were issued). Websites of NCBs banks. Haratyk (2019) Opposing votes are less frequent: "The danger of a speculative attack during the ERM II is small and negligible upon accepting the euro." (Kohout, 2004, p. 14). Concerns of a currency crisis during participation in ERM II are also expressed by the Czech central authorities creating economic policy. The Czech Republic's Euro Area Accession Strategy (2003, p. 3, emphasis M. H.) demands the shortest possible time in ERM II. The reason is as follows. "Given that participation in the ERM II... does not in itself eliminate the risk of currency turbulence, it is regarded merely as the gateway into the euro area ... staying in the ERM II for longer than the minimum required period of two years does not seem desirable". This position is developed in greater detail in the study by the Czech National Bank ERM II and the Exchange-rate Convergence Criterion (2003, pp. 4, 6, emphasis M. H.): "The ERM II... is a fixed exchange rate regime. ... In a world of massive capital flows [participation in ERM II] may be associated with potential costs as the financial markets "test" the willingness of the authorities to maintain the exchange rate within the fluctuation band". "The ERM II ... is a fixed exchange rate regime. ... In a world of massive capital flows [participation in ERM II] may be associated with potential costs as the financial markets "test" the willingness of the authorities to maintain the exchange rate within the fluctuation band". 91 Another (new) reason for remaining in ERJVIII for as short a time as possible is the requirement of the European Central Bank to enter the banking union49 along with entry to ERM II (in the case of non-member euro area countries this consists of formally establishing "close cooperation". This requirement was first articulated by the ECB in the case of the interest of Bulgaria in accession to ERM II in April 2018.50 Originally this entry to the banking union was required no sooner than at the time of entry to the euro area. The requirement was accepted by Bulgaria. The ECB expects also from other countries to meet this requirement. However, the legal binding of this requirement is questionable. The opposite opinion (compared to "the shortest possible time in ERM II") is offered by the analysis The Czech Republic and the euro area (2017). It reports that circumstances may occur that lead to an interest in rapid accession to the euro area, namely: • domestic circumstances in the form of reinforcing pressure of businesses to adopt the euro, • international circumstances in the form of separating the euro area from the rest of the EU. The rapid implementation of the euro would therefore help the Czech koruna become part of ERM II without specifying the deadline for acceptance of the euro. The above mentioned analysis marks this entry to ERM II as "technical entry". At present only Denmark is keeping its currency in ERM II. This membership in ERM II is therefore called the "Danish scenario". Making the Czech koruna part of ERM II would have other favourable impacts apart from facilitating entry to the euro area, such as: • the obligation to adopt a single European currency would be confirmed, by which the Czech Republic could enhance its trustworthiness, • its position in cooperation with the euro area would be enhanced, e.g. participation in certain euro area summits or access to certain information. 6.8 Conclusion The fixed exchange rate regime is more susceptible to currency crisis than a floating regime. All regimes allowable for ERM II are fixed exchange rate 49 Specifically this consisted (for the time being) of the first part, the Single Supervisory Mechanism - SSM, which became operational from 4 November 2014, and the second pillar of the banking union, the Single Resolution Mechanism - SRM, operational from 1 January 2016). 50 Euractiv 27 April 2018: Bulgaria's Borissov unveils secret criteria for joining the eurozone https://www. euractiv. com/section/b anking-union/news/bulgarias-borissov-unveils-secret-criteria-for-joining-the-eurozone/ 92 regimes. Worries about currency crisis as part of ERM II are confirmed by any of the following: empirical experience (nine countries in ERM II), theoretical models (we applied the second generation model). We paid special attention to long-term involvement in ERM II (we reviewed four countries - Denmark, Estonia, Latvia, Lithuania). These study results may be a jumping-off point for assessing alternatives for the shortest possible period of the Czech koruna being part of ERM II. This alternative is the "Danish scenario", i.e. joining ERM II without specifying a date for adopting the euro. 93 Chapter 7 Measuring credit risk based on CDS and bond spreads By Petr Budinsky and Michal Bezvoda Credit rating is a traditional measurement of credit risk in financial markets. This paper introduces an innovative approach based on implied ratings defined by CDS spreads. Using this approach, the credit risk can be better managed because CDS are provided on daily basis. The implied rating is compared with credit ratings provided by Moody's, S&P, and Fitch agencies. The model of implied rating deals only with sovereign ratings. 52 countries were chosen for comparison of both types of above-mentioned ratings. The model uses cumulative default probabilities (CPD) derived from CDS spreads and the main results are CPD intervals which define implied credit ratings. For those countries where the credit rating and implied credit rating are different, the chapter shows how implied rating can serve as a signal for potential upgrade or downgrade of the credit rating provided by rating agencies. The presented model is also used to verify ratings provided by Moody's, S&P, and Fitch in cases where these agencies provide different ratings for a specific country. This is especially important when some ratings are investment-grade and others are speculative-grade. 7.1 Introduction Credit rating agencies provide credit ratings for issuers of debt instruments. Issuers are governments, companies, or municipalities. The credit rating is based on the issuer's ability to repay debt and reflects its creditworthiness. The higher the credit rating, the lower the probability of default. The most respected credit rating agencies are Standard and Poor's (S&P), Moody's, and Fitch. They classify issuers into several credit rating categories. The following categories are investment-grade ratings (Table 7-1): 94 Table 7-1 Long-term credit ratings - investment grade Moody's S&P Fitch Aaa AAA AAA Aal AA+ AA+ Aa2 AA AA Aa3 AA- AA- Al A+ A+ A2 A A A3 A- A- Baal BBB+ BBB+ Baa2 BBB BBB Baa3 BBB- BBB- Source: Moody's, S&P, Fitch The highest rating is AAA and the probability of default in this case is very low. In this paper, four rating categories are used: AAA, AA, A and BBB. Moody's ratings Al, A2 and A3 are considered as being category A, and S&P/Fitch ratings BBB+, BBB and BBB- are considered as BBB. The categories in Table 7-2 express speculative-grade ratings. Table 7-2 Long-term crec lit ratings - speculative grade Moody's S&P Fitch Bal BB+ BB+ Ba2 BB BB Ba3 BB- BB- Bl B+ B+ B2 B B B3 B- B- Caal CCC+ CCC+ Caa2 CCC CCC Caa3 CCC- CCC- Ca CC CC Source: Moody's, S&P, Fitch Here the probability of default is substantially higher than for investment-grade issuers. We will use the "S" category to represent all the speculative-grade ratings in Table 2. This paper deals only with sovereign ratings to 31 January, 2017, where 52 countries were selected and categorized within the categories introduced above: AAA (8 countries), AA (10 countries), A (11 countries), BBB (13 countries), and S (9 countries), all based upon their median rating among the three above-mentioned rating agencies. A feature of credit ratings is that they do not change frequently. In one respect this brings stability, but it does not allow them to be adjusted due to actual events as rapidly as necessary. The main objective of this paper is to introduce a different type of rating based on market instruments that will allow investors to analyze the status of issuers on a daily basis. The selected market instrument is the credit default swap (CDS) which will aid in defining the implied rating. A CDS is a contract where a bond is the underlying asset and the CDS 95 functions as insurance in case the bond defaults. The riskier the underlying bond, the higher the CDS price (also called "CDS spread") will be, and so therefore the higher the probability of default. Defaulting means that some or all payments associated with the bond will not be recovered by the investor. The CDS seller is obliged to deliver missing payments to the CDS buyer. Based on the CDS spread, the cumulative probability of default (CPD) can be calculated. The CPD is the probability that the bond will default before expiration of the relevant CDS (normally 5 years). We have defined the implied rating for 52 selected countries based on their CPD. 7.2 Development of CDS CDSs offer the market an additional tool to determine the degree of credit risk. Unlike agency ratings, which are discrete and are only adjusted after time, usually in response to an important event related to underlying assets, CDS prices change in real time. The market reacts to events much faster than the time it takes for agencies to change their ratings. Georgievska et al. (2008) estimated default probabilities of emerging countries and compared them with the default rates implied by sovereign credit ratings. They detected that CRAs generally underestimated the risk of sovereign debt, and that sovereign credit ratings from rating agencies were much too optimistic. Callen et al. (2009) observed that credit ratings may have a close relationship with CDS spreads with respect to obligors sharing a common credit rating. They found that earnings of referenced firms are negatively correlated with the level of CDS prices, consistent with earnings conveying information about default risk. In accordance with Iyengar (2010, 2012), we found differences among the sovereign ratings granted by Moody's, Standard & Poor's, and Fitch. He carried out a comparison of sovereign ratings and examined their differences. Results showed that these differences are statistically significant and that they increase over time. This may lead to increased doubts about the consistency of such ratings. Budinsky, Heissler, Wawrosz (2011) dealt with the theory of equality between CDS spreads and bond spreads and they brought the evidence that this theory was valid for selected European countries before Lehman Brothers but after October 2008 it was valid for these countries only in some time periods. The above mentioned equality is presented by following equilibrium model. De Haan (2011) provided a basic background on the functioning of rating agencies. He focused on two main tasks for which rating agencies have come under criticism, namely the rating of structured instruments, and the issuing of sovereign ratings. Based on these tasks, they investigated how and whether there should be regulation. Budinsky et al. (2013) focused on the theory of equality between CDS spreads and bond spreads. This theory was valid for selected European countries before Lehman Brothers, but after October 2008, it was valid for these countries only in some time periods. Cizel (2013) argued that CDS spreads are a market-based measurement of credit risk relative to credit risk ratings. If CDS spreads represent an element of pure credit risk, and credit ratings are a relative default risk metric, then there should be a connection between the market price of credit risk and the 96 credit rating assigned to an obligor. Castellano and D'Ecclesia (2013) investigated the ability of fluctuations in CDS indexes in anticipating the occurrence of market crises. They found that CDS volatility tends to increase almost eight months before the market changes, confirming the impressive informational value of CDS changes that may reflect future expectations. Budinsky (2014) researched that implied rating based on CPD could be used to check sovereign ratings obtained by rating agencies through implied rating categories. Kiesel (2015) analysed the impact and effectiveness of regulation on the European sovereign CDS market. He focused on regulation that prohibits buying uncovered sovereign CDS contracts in the European Union. His results indicated significant change in CDS spreads prior to regulations and stable CDS spreads following the introduction of regulation. Berg (2016) was focused on monitoring 57 countries and he found that the CDS market relative to a country's debt is substantially larger for small countries, countries just above investment-grade, and countries with weaker creditor rights. Further, he came to view that the CDS market usually reacts only to negative events, and that changes in the size of CDS markets are determined by agency ratings. Budinsky et al. (2016) suggested two methods to measure credit risk. He investigated bond and CDS spreads in the equilibrium model and found that changes in economic situations may lead to the change of both bond and CDS spreads. Drago and Gallo(2016) analysed the impact of sovereign ratings announcements on the CDS market. The study concluded that agency warnings had zero to little impact on the CDS market. Based on his study, the market seems to react only to negative announcements. 7.3 Model of implied ratings Before we introduce the model for implied rating, we must place each selected country into category AAA, AA, A, BBB, or S. The median rating is introduced here based on omitting the best and worst of three different ratings (Moody's, S&P and Fitch). If at least two ratings are the same, the median rating is defined by those ratings. The median rating categories are in Table 7-3. Table 7-3 Ratings by Moody's, S&P, Fitch and Median Rating No Country Moody's S&P Fitch Median Rating 1 Australia Aaa AAA AAA AAA 2 Norway Aaa AAA AAA AAA 3 Denmark Aaa AAA AAA AAA 4 Germany Aaa AAA AAA AAA 5 Sweden Aaa AAA AAA AAA 6 Netherlands Aaa AA+ AAA AAA 7 Canada Aaa AAA AAA AAA 8 Singapore Aaa AAA AAA AAA 9 Finland Aal AA+ AA+ AA 10 United Kingdom Aal AA AA AA 11 Austria Aal AA+ AA+ AA 97 12 Belgium Aa3 AA AA AA 13 France Aa2 AA AA AA 14 South Korea Aa2 AA AA- AA 15 Abu Dhabi Aa2 AA AA- AA 16 Qatar Aa2 AA AA- AA 17 Chile Aa3 AA- A+ AA 18 China Aa3 AA- A+ AA 19 Japan Al A+ A A 20 Czech Republic Al AA- A+ A 21 Slovakia A2 A A+ A 22 Estonia Al AA- A+ A 23 Latvia A3 A- A- A 24 Ireland A3 A+ A A 25 Poland A2 BBB+ A- A 26 Israel Al A+ A A 27 Peru A3 A+ A A 28 Malaysia A3 A A A 29 Slovenia Baa3 A A- A 30 Spain Baa2 BBB+ BBB+ BBB 31 Thailand Baal BBB+ BBB+ BBB 32 Philippines Baa2 BBB BBB+ BBB 33 Romania Baa3 BBB- BBB- BBB 34 Panama Baa2 BBB+ BBB BBB 35 Mexico A3 BBB+ BBB+ BBB 36 Italy Baa2 BBB- BBB+ BBB 37 Kazakhstan Baa3 BBB+ BBB BBB 38 Colombia Baa2 BBB- BBB+ BBB 39 South Africa Baa2 BBB- BBB- BBB 40 Hungary Baa3 BB+ BBB- BBB 41 Bulgaria Baa2 BB+ BBB- BBB 42 Indonesia Baa3 BB+ BBB- BBB 43 Russia Bal BB+ BBB- S 44 Turkey Bal BB+ BBB- S 45 Vietnam Bl BB- BB s 46 Croatia Ba2 BB BB s 47 Brazil Ba2 BB BB s 48 Portugal Bal BB+ BB+ s 49 Argentina B3 B- B s 50 Egypt B3 B- B s 51 Venezuela Caa3 CCC CCC s Source: Moody's, S&P, Fitch Table 7-3 shows that all three rating agencies placed 38 countries into the same rating category and that only 13 countries (the Netherlands, Chile, China, the Czech Republic, Estonia, Poland, Slovenia, Mexico, Hungary, Bulgaria, 98 Indonesia, Russia, Turkey) have differing rating categories from two separate rating agencies. None of the countries have three different rating categories, so the ratings are very similar. The model of using CDS spreads and CPD (cumulative probabilities of default) is based on the following idea: the better the credit rating, the lower the CDS spread, and the lower the CPD. Hull, Predescu and White 2004 and Arce, Mayordomo, Pena 2011 assumed that CDS spreads should be equal to bond yield spreads. Based on this assumption, the research was conducted Budinsky (2011) where the prerequisite was set for the following statement: s = y - r, where s is defined as n-year CDS spread, y as yield on an n-year par bond issued by a reference entity and r as yield on an n-year par riskless bond. In the model may occur following states: Table 7-4 Possible states in equilibrium model State Result s > y-r The arbitrageur should decide to buy a riskless bond, to short the corporate bond and to sell CDS. The reason is that the CDS market is overvalued and the bond market underestimates the probability of a bond failure. s < y-r The arbitrageur should decide to buy a corporate bond, to buy CDS and to short the riskless bond. In this situation the CDS market is underestimated and the bond market overestimates the probability of a bond failure. s = y-r The CDS market has an equivalent predictive value to the risk of failure as a corporate bond. Source: Budinsky, Heissler, Wawrosz 2011, edited by the authors The model described in Table 7-4 was primarily designed for corporate bonds, however it can be applied also for sovereign bonds assuming that one of the sovereign bonds is defined as riskless. As a riskless sovereign bond is frequently used the German bond. The credit risk of this bond is very low whatever criterion mentioned above is used: first, rating of Germany is AAA - the highest. Second, the bond yields and CDS spreads are very low as well. 99 Table 7-5 CDS and CPD for selected countries No. Country 5 Year 5 Year CDS Median CPD (%) Spread (bps) Rating 1 Germany 1,39% 25 AAA 2 Australia 1,49% 21 AAA 3 Sweden 1,54% 25 AAA 4 Finland 1,89% 25 AA 5 Norway 1,93% 23 AAA 6 Austria 1,98% 34 AA 7 Denmark 2,03% 24 AAA 8 United Kingdom 2,03% 33 AA 9 Canada 2,08% 33 AAA 10 Netherlands 2,33% 28 AAA 11 Japan 2,33% 33 A 12 Belgium 2,48% 36 AA 13 France 3,06% 39 AA 14 Slovakia 3,16% 47 A 15 Czech Republic 3,69% 43 A 16 South Korea 4,08% 46 AA 17 Estonia 4,52% 57 A 18 Abu Dhabi 4,71% 62 AA 19 Singapore 4,85% 60 AAA 20 Latvia 5,00% 63 A 21 Ireland 5,04% 64 A 22 Spain 5,95% 78 BBB 23 Israel 6,10% 77 A 24 Qatar 6,24% 80 AA 25 Thailand 6,33% 81 BBB 26 Poland 6,43% 76 A 27 Chile 7,14% 83 AA 28 Philippines 7,37% 99 BBB 29 Slovenia 8,22% 103 A 30 Romania 8,40% 108 BBB 31 Peru 8,45% 108 A 32 China 8,64% 117 AA 33 Hungary 8,87% 123 BBB 34 Bulgaria 9,15% 143 BBB 35 Malaysia 10,11% 134 A 36 Panama 10,30% 129 BBB 37 Indonesia 11,39% 154 BBB 38 Italy 11,80% 157 BBB 39 Colombia 11,94% 161 BBB 40 Kazakhstan 11,98% 157 BBB 41 Mexico 12,25% 154 BBB 42 Russia 12,97% 176 S 100 43 Vietnam 14,17% 189 S 44 South Africa 15,23% 211 BBB 45 Croatia 15,62% 210 S 46 Brazil 17,61% 273 S 47 Turkey 18,59% 269 s 48 Portugal 18,84% 274 s 49 Egypt 27,04% 435 s 50 Argentina 29,98% 432 s 51 Venezuela 65,67% 3193 s Source: Compiled by the authors based on Deutsche Bank Research The cumulative probability of default within five years is lower than 2% for seven countries (Germany, Australia, Sweden, Finland, Norway and Austria), so their implied rating category of AAA is expected. On the other hand, three countries with CPD higher than 25% would clearly be in the S category. We now derive the exact model, which allows us to put each country into its respective implied rating category. This is done based on Table 7-4, which combines the ratings from Table 7-3 with CPD from Table 7-5. The countries in Table 7-5 are in sequence by their CPD - from the lowest to the highest. We can see that Finland, Austria and the United Kingdom (in the AA category) are distributed among the AAA category countries. Peru, China, and Malaysia, which are in the A and AA categories, are distributed among the BBB category countries. We must determine CPD ranges to maximize the number of countries with matching rating categories and implied ratings (see Table 7-6). Table 7-6 Implied rating categories 5 Year CPD (%) Implied Rating 0 - 2,09 AAA 2,10-4,19 AA 4,20 - 6,29 A 6,30-12,59 BBB > 12,60 S Source: Compiled by the authors The choice of ranges defining the intervals for CPD is not unique. Instead of 2,10, we could use 2,20 or 2,30 with the same result, or instead of 4,20, we could use 4,30 or 4,40. The solution represented in Table 7-6 is important because the ranges of intervals for the implied ratings AAA, AA, and A are 2,10, and the range of the fourth interval (BBB) is 6,30, which 3 x 2,10. 7.4 Comparison of selected countries The above introduced model will now be applied to the selected 52 countries. Using Table 7-3 for rating categories AAA, AA, A, BBB and S, we created the following tables: Table 7-7 for the AAA category, Table 7-8 for the AA category, Table 7-9 for the A category, Table 7-10 for the BBB category, and Table 7-11 101 for the S category. These tables list implied rating categories based on the CPD intervals from Table 7-6. Table 7-7 Countries with AAA median rating Country Mo- S&P Fitch Median Implied ody's Rating Rating Australia Aaa AAA AAA AAA AAA Norway Aaa AAA AAA AAA AAA Denmark Aaa AAA AAA AAA AAA Germany Aaa AAA AAA AAA AAA Sweden Aaa AAA AAA AAA AAA Netherlands Aaa AA+ AAA AAA AA Canada Aaa AAA AAA AAA AAA Singapore Aaa AAA AAA AAA AA Source: Compiled by the authors In Table 7-7 there are only two countries in the AAA median rating category (the Netherlands and Singapore) where the median rating (AAA) differs from the implied rating (AA). Implied rating is in both cases lower than the median rating. For the other seven countries, both ratings are the same (AAA). Table 7-8 Countries with AA median rating Country Moody's S&P Fitch Median Rating Implied Rating Finland Aal AA+ AA+ AA AAA United Kingdom Aal AA AA AA AAA Austria Aal AA+ AA+ AA AAA Belgium Aa3 AA AA AA AA France Aa2 AA AA AA AA South Korea Aa2 AA AA- AA AA Abu Dhabi Aa2 AA AA- AA A Qatar Aa2 AA AA- AA A Chile Aa3 AA- A+ AA BBB China Aa3 AA- A+ AA BBB Source: Compiled by the authors There are three countries (Belgium, France and South Korea) in the AA median rating category (Table 7-8) where the median rating (AA) coincides with the implied rating (AA). For the other seven countries, both ratings are different, whereas the biggest differences are found for Chile and China, with their implied ratings of BBB being significantly lower than AA. 102 Table 7-9 Countries with A median rating Country Moody's S&P Fitch Median Rating Implied Rating Japan Al A+ A A AA Czech Republic Al AA- A+ A AA Slovakia A2 A A+ A AA Estonia Al AA- A+ A A Latvia A3 A- A- A A Ireland A3 A+ A A A Poland A2 BBB+ A- A BBB Israel Al A+ A A A Peru A3 A+ A A BBB Malaysia A3 A A A BBB Slovenia Baa3 A A- A BBB Source: Compiled by t ie authors In median rating category A (Table 7-9), there are four countries (Estonia, Latvia, Ireland, and Israel) where the median rating (A) coincides with the implied rating (A). For the other seven countries, both ratings are different. Table 7-10 Countries with BBB median rating Country Mo- S&P Fitch Median Implied ody's Rating Rating Spain Baa2 BBB+ BBB+ BBB A Thailand Baal BBB+ BBB+ BBB BBB Philippines Baa2 BBB BBB+ BBB BBB Romania Baa3 BBB- BBB- BBB BBB Hungary Baa3 BB+ BBB- BBB BBB Panama Baa2 BBB+ BBB BBB BBB Bulgaria Baa2 BB+ BBB- BBB BBB Mexico A3 BBB+ BBB+ BBB BBB Indonesia Baa3 BB+ BBB- BBB BBB Italy Baa2 BBB- BBB+ BBB BBB Kazakhstan Baa3 BBB+ BBB BBB BBB Colombia Baa2 BBB- BBB+ BBB BBB South Africa Baa2 BBB- BBB- BBB S Source: Compiled by the authors In median rating category BBB (Table 7-10), there are only two countries (Spain and South Africa) where the median rating (BBB) is different than the implied rating (A for Spain and S for South Africa). For the other 11 countries, both ratings are the same (BBB), but the S&P credit ratings for Hungary, Bulgaria and Indonesia are BB+ (S category). 103 Table 7-11 Countries with S median rating Country Mo- S&P Fitch Median Implied ody's Rating Rating Russia Bal BB+ BBB- S S Vietnam Bl BB- BB S s Croatia Ba2 BB BB s s Turkey Bal BB+ BBB- s s Brazil Ba2 BB BB s s Portugal Bal BB+ BB+ s s Argentina B3 B- B s s Russia Bal BB+ BBB- s s Venezuela Caa3 CCC CCC s s Source: Compiled by the authors Nine countries are in median rating category S (Table 7-11). The median rating (S) coincides with the implied rating (S) for all of these countries, but the Fitch credit ratings for Russia and Turkey are BBB- (BBB category). Summarizing the content of the previous section, we can conclude that the median rating and implied rating are the same for 34 countries, and different for 18 countries. We can divide these 18 countries into 3 groups: • Median rating lower than implied rating - seven countries (Table 7-12). • Median rating slightly higher than implied rating - seven countries (Table 7-13). • Median rating significantly higher than implied rating - four countries (Table 7-14). Table 7-12 Countries witl l lower median rating than implied rating Country Median Rating Implied Rating Finland AA AAA United Kingdom AA AAA Austria AA AAA Czech Republic A AA Slovakia A AA Japan A AA Spain BBB A Source: Compiled by the authors All countries in Table 7-12 are investment-grade and CDSs suggest an upgrade of their credit ratings. 104 Table 7-13 Countries wit! l slightly higher median rating than implied rating Country Median Rating Implied Rating Netherlands AAA AA Abu Dhabi AA A Qatar AA A Poland A BBB Peru A BBB Malaysia A BBB Slovenia A BBB Source: Compiled by the authors Table 7-14 Countries with substantially higher median rating than implied rating Country Median Rating Implied Rating Singapore AAA A Chile AA BBB China AA BBB South Africa BBB S Source: Compiled by the authors All countries in Table 7-13 and Table 7-14 are investment-grade and CDSs suggest a downgrade of their credit rating. Substantial potential downgrades for countries in Table 7-14 mean that Singapore, Chile, and China would drop by two categories and South Africa would even obtain a speculative-grade rating, We can now investigate the accuracy of ratings delivered by rating agencies in case they differ, or if at least one of the ratings (Moody's, S&P, or Fitch) coincides with its respective implied rating. First, we notice that out of 18 countries with different median and implied ratings, there are only four countries (Table 7-15) where at least one of the ratings provided by rating agencies coincides with the implied rating. Such coincidence is marked with a plus symbol. Differences are marked with a minus symbol. S&P is the most precise rating agency in this respect, although this example of only four countries is quite small. Table 7-15 Countries with different median rating and implied ral ing Country Median Rating Implied Rating Moody's S&P Fitch Netherlands AAA AA - + - Czech Republic A AA - + - Poland A BBB - + - Slovenia A BBB + - - Source: Compiled )y the authors The same procedure will now be applied to the other 34 countries with the same median and implied ratings. 105 Table 7-16 Countries wit h equal median rating and implied rating Country Median Rating Implied Rating Moody's S&P Fitch Estonia A A + - + Mexico BBB BBB - + + Hungary BBB BBB + - + Bulgaria BBB BBB + - + Russia S S + + - Turkey S S + + - Indonesia BBB BBB + - + Source: Compilec by the authors Moody's, S&P, and Fitch deliver different ratings for only eight of these 34 countries. Moody's is the most precise rating agency for this group. Note that each of the last five countries in Table 7-16 (Hungary, Bulgaria, Russia, Indonesia, and Turkey) have at least one investment-grade rating and at least one speculative-grade rating. For these five countries, the match with the credit rating provided by Moody's and the implied rating is 100%. 7.5 Conclusion Implied rating based on CPD that is derived from CDS spreads is a powerful tool used to verify the sovereign ratings granted by rating agencies. Implied ratings are defined by CPD intervals (Table 7-6). First, implied rating could provide a signal for future upgrades or downgrades of ratings in cases where the median rating and implied rating differ. Special attention should be paid to cases where all three ratings are investment-grade, but the implied rating category is S (South Africa), or, if some ratings are investment-grade and others are speculative-grade (Hungary, Bulgaria, Russia, Indonesia, and Turkey). Second, we can find which rating agency is the most precise by using implied ratings. In cases where implied rating and median rating are the same, but the ratings from Moody's, S&P, and Fitch are different, we can use implied rating to verify the relevant rating agency when its credit rating and implied rating are the same. In case implied rating and median rating are different, one rating agency can still provide a rating that is equal to the implied rating. The deterioration of the economic situation may lead to the increase of both bond spreads and CDS spreads but they may increase differently because both markets are different - there is different counterparty risk, different liquidity and different funding costs. But in case that bonds spreads and CDS spreads are substantially different then we can use that information in the following way: 1) s > y-r indication of higher probability of default, 2) s < y-r increasing funding costs and possible depreciation of local currency. 106 Chapter 8 Insurance linked securities and their future research By Bohumil Stddnik New risks for insurance industry arise from global environment changes, undervaluation of lifetime (longevity risks) due to the significant scientific progress in medicine or, vice versa, overvaluation due to higher risk of global war conflicts with enormous impact on mortality, caused by new weapons technologies, as well as a global increase in losses due to higher volatility in climate changes. Insurance-linked securities (ILS) such as mortality-linked securities and its derivatives, longevity (survivor) bonds, mortality catastrophe bonds or natural catastrophe (CAT) bonds are defined as investment instruments which are linked to cover the insurance claims resulting mainly from life insurance events, such as longevity/mortality events, or natural catastrophes (CAT bonds) as earthquake, floods or hurricane damages; and whose values are closely connected to the probability of certain insured event. ILS have many interesting aspects of interest for investors and risk managers. They have shown low correlations with other types of investment risks, such as interest rate or currency risk or they may provide attractive yields. Its pricing is an interesting challenge for researches. 8.1 Introduction The reason of securitization of typical insurance products is the under/overestimation of the value of death/health or natural catastrophes-related claims. In the last years, the mortality improvements have become serious issue for pension funds and annuity providers to manage. The reason is that longevity has been systematically underestimated, making balance sheets riskier to unexpected increases in liabilities. On the other hand, we also newly observe issues of mortality bonds. The first mortality bond, known as Vita I, was issued by Swiss Re Group in December 2003 and was designed to reduce Swiss Re's own exposure to catastrophic mortality events, such as major terrorist attacks, avian flu pandemics, or other natural catastrophes. 107 Financial crises nearly always result in steep increases in government debt. The fiscal situation in many countries was negatively influenced by massive emergency measures to stabilize financial institutions, fiscal stimulus packages, and sharp economic decline that resulted in lower tax revenues all around the world. Budget deficits became a basic component of advanced countries, and the ratio of government debt to gross domestic product (GDP) increased. Such increases in debt can even potentially result in government debt defaults. The traditional way of transferring insurance company risk through reinsurance seems to be problematic because of the lack of the capacity and liquidity to support an estimated global exposure in excess of $20tr (Loeys et al., 2007). The solution is, as in the case of natural catastrophes, to turn to capital markets which could play a very important role, offering additional capacity and liquidity to the life insurance market. In other words: the insurance industry, using ILS, transfers life insurance companies' risk to capital markets which also allow more transparent and competitive pricing of these products. In 1992, Hurricane Andrew caused $17 billion in insured losses in Florida, and a loss figure doubles the modelling estimates at the time for the financial costs emanating from a severe hurricane. Several insurers went into bankruptcy, and reinsurance capacity was not able to satisfy the remainder. Eleven insurance companies went bankrupt, caused by more than 600 000 insurance claims filed. A new source of capacity outside traditional reinsurance was needed to fill the void. In 1996, according to Aon Securities Inc., the first catastrophe bond drawing risk-bearing capital from the capital markets to satisfy this need was developed by St. Paul Re UK. It was basically the beginning of a new story in the insurance industry which is characterized by transferring the risk from insurance companies to capital markets participants. Also the Northridge earthquake, 1994, in the San Fernando Valley region of the County of Los Angeles supported this process. Catastrophe bonds (also known as CAT bonds) are risk-linked securities that transfer from a sponsor to investors a specified set of risks as: Hurricane and tropical storm, earthquake, flood, hailstorms. CAT bond, which is also included among insurance-linked securities (ILS) with face value F is a financial instrument which is expected to provide a stream of cash payments c at the end of every period t = 1,2,...,T, where T denotes the bond's maturity, so long as a particular catastrophe does not occur. At the CAT bond's maturity, an investor receives both coupon payment and principal repayment. Provided a catastrophe occurs during the life of a CAT bond, an investor only receives a fraction of both coupon payment and principal repayment o)(F + c), where a) G [0,1] denotes the fraction received. After this payment, the bond is wound up. Edesess (2014) claims that the periodicity t of the coupon payments c is usually quarterly and the maturity ranges between 1 and 5 years with an average of 3 years. Edesess (2014) mentions 2 primary attractions of the CAT bonds. First, the risk of CAT bonds is virtually uncorrelated with other types of financial risks such as market risk, credit risk or interest rate risk. Second, the interest rates paid to the investors are rather high, consisting of the base interest on the money market funds 108 in which the principal is deposited and the premium paid for the insurance coverage. The referred paper defines four main trigger types of principal losses: Indemnity trigger, Industry loss trigger, Parametric trigger, Modelled trigger. Indemnity trigger - loss of principal triggers when there is an excess of total losses over the attachment point. Also, the exhaustion point is specified over which the principal is exhausted. This trigger favours the issuer and is not very attractive for investors. It may even cause a moral hazard, e.g., construction in flood areas. Industry loss trigger - the trigger is specified as total industry losses in excess of the pre-specified amount. Independent third party then estimates total industry losses on the insured event. The danger of moral hazard is partly mitigated. Parametric trigger - the trigger is based on the occurrence of a specific natural event, e.g., the speed of wind in excess of 100 km/h. The danger of moral hazard is completely mitigated, and thus parametric trigger favours the investors. Modelled trigger - very similar to indemnity trigger but is based on claims estimated by independent third party. The danger of moral hazard is partly mitigated. According to Edesess (2014) most CAT bonds have an indemnity or an industry loss trigger. There are several CAT bond market participants: 1. Issuers are typically insurers and reinsurers, government entities or pension funds. 2. Structuring agents assist the issuer in selecting the trigger type, and they also place the bond with investors, e.g., investment banks creating SPV (Special Purpose Vehicle). 3. Modelling agents estimate the risk of the CAT bonds. 4. Rating agencies rate the CAT bonds typically as below investment grade. 5. Secondary market in CAT bonds. 6. Investors are typically institutional investor (pension funds, endowments funds or hedge funds). CAT bonds are only privately placed. No CAT bonds can be publicly offered or traded in the USA. Only qualified institutional investors may engage in the secondary market in CAT bonds. Regarding the CAT bonds mechanics, Cizek et al. (2005) provide a thorough explanation. Sponsor creates an SPV as an issuer of bonds and as a source of reinsurance protection. The CAT bonds are then sold to investors. Raised money is immediately invested in collateral. The sponsor then makes premium payments to the SPV which together with the investment of bond proceeds make up an interest paid to investors. If there is a trigger event, the fund is immediately withdrawn from collateral and paid to the sponsor. At maturity, the remaining principal (up to 100 %) is paid back to investors. Although the CAT bond default rates have been historically very low, the spreads over US T-rates are considerably higher than those of comparably-rated junk corporate bonds. Liu et al. (2014) argue that the securitization of catastrophe risks with small probability and high loss events can bring a solution to spread the catastrophe risk. They argue that to develop effective CAT bond market, it is crucial to create accurate pricing of CAT bonds. That is the reason they unlike the vast majority of other studies on the topic employ credit risk in CAT bond valuation. Liu et al. 109 (2014) basically employ Jarrow and Turnbull method to model the credit risk and using extreme value theory thus construct a general pricing formula. Liu et al. (2014) apply their theoretical model to Property Claim Services data to finally value the CAT bonds using the Monte Carlo method. Cox & Pedersen (2010) assume that there is no correlation between the default of CAT bonds and underlying financial market variables. They further assume that the financial markets are incomplete and that the catastrophic event occurs independently of the underlying financial market variables. Cox & Pedersen (2010) propose two-step model. The first step is the estimation of the interest rate dynamics in the states of the world without the occurrence of the catastrophe. The second step contains the estimation of the probability of the catastrophe occurring. Cox & Pedersen (2010) compare the bond contract to lending money subject to credit risk, except the risk of default is, in fact, the risk of a catastrophe event happening. This comparison is of course made from the bond owner's perspective. Cox & Pedersen (2010) are then able to utilize proposed pricing methodology to assess the relative default spread on CAT bonds compared with traditional defaultable securities. Jarrow (2010) proposes a simple closed-form valuation formula for CAT bonds consistent with Libor term structure of interest rates model. The pricing in Jarrow (2010) is predominantly based on the already existing methodology for pricing credit derivatives using a reduced form model. The pricing formula established by Jarrow (2010) requires two crucial inputs - likelihood of the catastrophe occurring and percentage realized loss rate. These two inputs are easy to obtain using historical event occurrence and realized loss data. Since counterparty risk is minimized, Jarrow (2010) assumes that the issuer is default free and that he makes all bond payments in time. Cizek et al. (2005) argue that there is evidence of power-law distribution associated with catastrophe events losses which overturn the traditional assumption of lognormality of derivative pricing models. Cizek et al. (2005) model the catastrophe process as a compound doubly stochastic Poisson process. To calibrate the pricing model, Cizek et al. (2005) fit the distribution function of the incurred losses and the stochastic process governing the flow of natural events. Cizek et al. (2005) utilize 10-year catastrophe loss data provided by Property Claim Services data and argue that the claim size distributions describing property losses are often heavy-tailed. Thus, the authors employ Burr distribution for the calibration. Using the Monte Carlo simulation Cizek et al. (2005) simulate the dynamics of the CAT bond prices. Lai et al. (2014) propose a new arbitrage model which takes into account also currency exchange risk. Therefore, the authors value the CAT bonds using jump-diffusion CAT process, a stochastic process for the exchange rate and a stochastic process for both foreign and domestic interest rates. Lai et al. (2014) finally derive at semi-closed form formula for CAT bonds valuation. The authors also detected three other factors affecting the CAT bond's value - exchange rate volatility and correlations with both foreign and domestic interest rates. 110 Braun (2014) detects main determinants of the cat bond spread at issuance to be mainly expected loss, covered territory, sponsor, reinsurance cycle and the spreads on a comparably rated corporate bond. He then proposes an econometric pricing model that is applicable across territories, perils, and trigger types. Instead of CAT group we meet insurance-linked securities such as mortality-linked securities and its derivatives, longevity (survivor) bonds are defined as investment instruments which are linked to cover the insurance claims resulting mainly from life insurance events, such as longevity/mortality events; and whose values are closely connected to the probability of certain event connected to demography development process and its parameters. There is quite a literature on this topic - see for example Kabbaj and Coughlan (2007), Krutov (2006), Leppisaari (2008), Levantesi and Torri (2008), Lin and Cox (2005), Reuters (2010), Richards and Jones (2004), and Thomsen and Andersen (2007). The reason for the securitization of typical insurance products is the under/overestimation of the value of death/health-related claims. In the last years, the mortality improvements have become a serious issue for pension funds and annuity providers to manage. The reason is that longevity has been systematically underestimated, making balance sheets riskier to unexpected increases in liabilities. In the journal Nature, medical researchers at Mayo Clinic College of Medicine - led by cell biologists Darren Baker and Jan van Deursen - have made this decade's biggest breakthrough in understanding the complex world of physical aging. The researchers found that systematically removing a category of living, stagnant cells (ones which can no longer reproduce) extends the lives of otherwise normal mice by 25 percent (Baker, Bennett, Childs, Durik, Wijers, Sieben, Zhong, Saltness, Jeganathan, Verzosa, Pezeshki, Khashayarsha, Miller & Deursen 2016). See also (Gil & Withers 2016 or Childs, Dunk, Baker & Deursen 2015). On the other hand, we also newly observe issues of mortality bonds (mortality catastrophe bonds (MCB) or extreme mortality bond (EMB)) connected with higher mortality events. We observe many studies regarding number of victims in case of modern war conflict (Wydra 2015, Brighi 2015) The first mortality bond, known as Vita I, was issued by Swiss Re Group in December 2003 and was designed to reduce Swiss Re's own exposure to catastrophic mortality events, such as major terrorist attacks, pandemics, or other natural catastrophes. The volatility of mortality rates is fairly low compared with the uncertainty surrounding changes in mortality trends. Forecasting mortality trends is a challenging exercise that concerns investors willing to take on exposures to longevity risk. Biffis and Blake (2008) explicitly distinguish the role played by trends and volatility in mortality rates in determining equilibrium risk premia in longevity risk transfers. Investors currently still seem to be uncomfortable enough with longevity risk to make this a plausible situation, even for securities written on publicly available demographic indices. At the other end of the spectrum, there are holders of longevity exposures (in terms of better experience data or forecasting technologies developed by monitoring the exposures). This situation is realistic for life insurers, reinsurers and other intermediaries (e.g., pension buyout firms and investment banks) that have developed considerable expertise in managing mortality-linked cashflows. Ill ILS have many interesting aspects of interest for investors and risk managers. They have shown low correlations with other types of investment risks, such as interest rate or currency risk or they may provide attractive yields. Its pricing is an interesting challenge for researches. ILS Valuation using risk neutrality in general follows: p = 2f=1d(o,t)EQ(s(t)|in0) (i) EQ(5(t)l|fl0) is the expected value of S(i) under the risk-neutral measure Q conditional on the information lfl0 available at time 0. Another valuation approach is, for example, distortion valuation approach. Figure 8-1 3D tree 8.2 Valuation of ILS As it is written above the pricing of ILS is interesting challenge for researches. In the following text we provide certain examples in the field of valuation of ILS instruments, namely catastrophe bonds based on previous studies of Fucik & Stadnik (2017). For the valuation in the context of risk neutrality (equation (1)) one way is to use more dimensional trees. If we consider three-dimensional process, we suggest to deal with the three-dimensional trinomial tree of prices which has 9 possible ways from each node. The visualization of such tree is provided in Figure 8-1. We need to consider three-dimensional process with the following dimensions: time, interest rates development and catastrophe process development. Interest rate development. We consider the development of interest rates to be a certain random process with three possible steps from each node or the process 112 could be improved by utilizing a more sophisticated model such as Hull-White model. Catastrophe process development. We expect catastrophe process to be independent on the interest rate development. Let us suppose for example the path of the cyclone approaching the area to which is linked (Figure 10-2). This process could be also certain random process closely connected to a random walk. Figure 8-2 The path of a cyclone f^5 ^ ,L ,06:00 22/12 J3C sir „ OO:O0 13/1 00:(» 31/12jffiS^^ 00:00 1/1 W:00 14/1 3^O0:00 15/1 : ^ V ... *> o & a i ] 5 IB svlllc 00:O0 2/1 i 00:00 1 L."5»O:00 lO/l J 7/1 V| r^00:00 9/1 "i-* \ -.'Vila 00:00 8/1 10 70 80 \ 90 100 Sleeping regime 8.3.2 "Earthquake" Style Catastrophic Process Figure 8-9 describes earthquake style with a very quick run. We expect sharp peaks in comparison to cyclone style. For the valuation of both the cyclone and the earthquake style we may use the following equation (3): _ (F+c)(l-q)+