Home

Systemic Risk Monitoring (ISysMo") Toolkit2 A User Guide

1. 10 Figure 3 Unwinding of Systemic Risk Sources and Channels Real economy e lxi ay a Legend MR market risk IR Interest rate risk CR credit risk DSA debt sustainability analysis S T stress testing CCA Contingent Claims Analysis SysCCA Systemic CCA FSI Financial Soundnes indicators JPoD Joint probability of Default EVT Extreme Value Theory 11 Fundamentals based models rely on macroeconomic or balance sheet data to help assess macro financial linkages e g macro stress testing or network models By providing vulnerability measures based on actual interconnectedness and exposures these models may help build a realistic story However they often require long term data series assume that parameters and relationships are stable under stressed conditions and only produce low frequency risk estimates Market based models These models uncover information about risks from high frequency market data and are thus suitable for tracking rapidly changing conditions of a firm or sector These approaches are more dynamic but their capacity to reliably predict financial stress has yet to be firmly established Hybrid structural models These models estimate the impact of shocks on key financial and real variables e g default probabilities or credit growth by integrating balance sheet data and market prices Examples include the CCA and distance to default measures
2. Tools Binder and suggestions on how to operationalize systemic risk monitoring including through a systemic risk Dashboard In doing so the project cuts across various country specific circumstances and makes a preliminary assessment of the adequacy and limitations of the current toolkit JEL Classification Numbers G12 G29 C51 Keywords Sytemic Risk Risk Indicators Risk Monitoring Macroprudential Policy Authors E Mail Addresses NBlancher imf org SMitra imf org HMorsy imf org tiago severo gs com LValderrama imf org AOtani imf org Contents Page AIO SSE opc e ctu ttc Peck dba et da bdo cd a Dl Bia Moo dcum See dI Medial bd R20 dd US agus 3 LDIottOdBetoH co oci aa taal an AE E edet te D DU E Le utat 4 II Approaching Systemic Risk uocare e pret Gn e bois S een Ee Lore aaniars ERR RENE apie nections 6 A Whatds Systemic RISE 2v roc tria tas Dodo vati tees Ete ud Qd bus x tuetur 6 B Rey Features of tlie Doolkitzo ios pice ed o aa Pete errant dide ir Pedido 7 III Mapping Tools to the Territory A Practical Approach sene 11 A Financial institutions Is Excessive Risk Building Up in Financial Institutions 12 B Asset Prices Are Asset Prices Growing Too Fast cccesccesceseeeeceeeeeseeneeeneeees 14 C Sovereign Risk How Much is Sovereign Risk a Source of Systemic Risk 15 D Broader Economy What are the Amplification Channels among Sectors and through the Domestic Economy iii
3. bottom up and or top down approach and iv assessment of the resilience of the financial system by interpreting quantitative results FSAPs have addressed a range of risks in stress tests within the broad categories of credit risk market risk liquidity risk and contagion risk In a typical stress test in credit risk models NPLs or loan loss provisions are modeled as a function of various macroeconomic variables The analysis of market risks has used a range of different approaches Interest rate risk analysis uses pricing and maturity gaps duration and value at risk Exchange rate risk analysis focuses on net open positions Stress tests for liquidity risk have assumed shocks to deposit and wholesale funding and overseas funding Stress tests for contagion risk use data on uncollateralized interbank exposures to assess whether the failure of one bank induce failure in other banks It should be noted that stress tests have to be tailored to country specific circumstances as to the different types of risks and institutions to be subjected to stress testing the type and size of shocks applied to the stress scenario and data availability Example Since the FSAP s inception in 1999 FSAPs have been carried out at least once and for many countries more than once for over 130 countries more than two thirds of Fund membership A list of upcoming FSAPs and notes on stress test methodologies are available at the FSAP site http www imf o
4. Such an analysis offers a first assessment of sovereign risk buildup but stress scenarios used in DSA are more akin to sensitivity analysis their plausibility is not measured In addition Indicators of Fiscal Stress IFS provide a summary measure of the risk of a fiscal crisis over the medium term based on a coincident indicator of rollover pressures and on a forward looking index of fiscal stress 28 DSA and IFS can be combined with forecasting tools such as Crisis Prediction Models that aim to measure the likelihood of a fiscal crisis over a one year horizon by combining asset prices measures of external and fiscal imbalances and data on the financial household and corporate sectors In addition Schaechter and others 2012 construct a range of indicators to monitor fiscal vulnerability and identify the main underlying fiscal challenges The choice of indicators is guided by their ability to capture immediate funding pressures medium and long term funding needs and risks to the baseline debt dynamics They can be used to monitor fiscal vulnerabilities in a large set of advanced economies 29 In turn a number of tools can be used to analyze the effect of sovereign risk on financial distress Macro Stress Tests may investigate the impact of a decline in government bond prices on financial institutions both directly through their liquidity and market risk exposures and indirectly through a decline in GDP growth e g caused by fi
5. Cd cd cd cd c c oc c c c c c OQ OO OQ OON OQ QN OQ 20 Summary Credit growth has slowed down and banking stability is falling fast and below 2003 levels at end 2007 Systemic risk is starting to unwind Real Estate Prices Heat Map Equity Market Misalignments Deviation Between Market and Model Prices 200702 200703 200704 Dec 2007 Pricing Theory iscount Apr Model APTGlobal Country J cwyx NEN Summary There are mixed signals from asset markets at end 2007 Public Debt Sustainability Analysis Country X Sovereign Funding Shock Scenarios FSS percent of banking sector assets O net financing 7 5 O gross financing 9 4 10 30 sale 12 4 8 6 4 Historical Pat re aoe 0 net financing 0 gross financing 2 30 sale 0 Q G Q aD a o9 a o 200001 200003 200101 200103 200201 200203 200401 2004Q3 2005Q1 2005Q3 200601 2006Q3 200701 2007Q3 2008Q1 2008Q3 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Summary There are clear signals that fiscal risk are increasing especially from financial sector related contingent liabilities 24 GDP at Risk and Financial Stability at Risk Systemic Contingent Claims Analysis 0 00 1 500 0 50 7 1 g 1 0004 1 00 i 1 50 2 7s 2 2 00 soo E FSaR 2 50 250 m GDPaR right axis 3 00 EZEZ EELEEZEJ ZLEZE 3 50 PREPFPILAITHFEEL PT RFR ATEZE oN m oN m on Total Cont abiliti
6. denotes monthly excess return on country i s stock market index over the risk free rate and F capture various measures of risk premia The spillovers from an asset price correction on GDP are measured using a VAR specification that includes a set of macroeconomic variables a monetary policy reaction function and real house prices Finally the corporate bond valuation model estimates the impact on corporate spreads from changes in operating investing financing cash flows of bond issuers and holders driven which are driven by the business cycle market price fluctuations and financing constraints Example An econometric model has been used to estimate the price correction of residential house prices in advanced economies if the gap between 2010 house prices and their fundamental values based on changes in per capita disposable income working age population construction costs credit and equity prices and interest rates were to close over the next five years Real house prices would fall at an annual rate of between 0 5 percent and 1 5 percent on average between 2010 and 2015 and residential investment would remain depressed for several years Current cycle Previous cycles 120 House Prices GDP 120 N L T 100 M Re zB 60 s 260 40 34 0 4 8 i214 B 4 o 4 8 12140 120 Consumption Residential 120 _ Investment g io eer 100 80 5 80 60 60 40793 0 4 8 1214 8 4 O 4 8 12149 S
7. sector as well as FSIs for key nonfinancial sectors and asset prices The health of the 61 financial sector can be analyzed by looking at levels and trends in FSIs It should be noted however that interpreting developments in FSIs presents the following challenges Since FSIs are aggregated data measures of dispersion should be monitored to analyze the vulnerability of the financial system FSIs allow continuous monitoring of strengths and vulnerabilities over time and show the current financial soundness of the financial system They do not measure precisely the likelihood of or resilience against future shocks Therefore the analysis of FSIs should be strengthened by using higher frequency or more forward looking tools Example The website of FSIs http fsi imf org provides the following core set and encouraged set of FSIs of 100 countries to monitor the current financial soundness of their financial systems Financial Soundness Indicators Core and Encouraged Sets Core Set Deposit takers Capital adequacy Regulatory capital to risk weighted assets Regulatory Tier 1 capital to risk weighted assets Nonperforming loans net of provisions to capital Asset quality Nonperforming loans to total gross loans Sectoral distribution of loans to total loans Earnings and profitability Return on assets Return on equity Interest margin to gross income Noninterest expenses to gross income Liquidi
8. 2001 2002 2003 2004 2005 2006 2007 2008 2009 Financial Crisis Growth Slowdown Summary The estimated likelihood of a systemic crisis has increased but is still small 25 V KEY FINDINGS AND OPERATIONAL IMPLICATIONS 50 On balance several dimensions of systemic risk are covered well by the toolkit Tools exist to address most of the key sources of shocks and transmission channels and appear to do so relatively well along the following dimensions Impact of shocks rather than the likelihood of systemic events Long term buildup of balance sheet vulnerabilities E Spillovers across financial entities e Cross border contagion between banking systems 51 A number of operational implications emerge from the above discussion Tools should be combined to exploit their complementarities Such complementarities help to cross check and confirm the materiality of sources of systemic risk stemming from domestic macro financial imbalances e g credit boom asset price bubble unsustainable public debt and cross border linkages or individual institution exposures e g size leverage interconnectedness Therefore they help practitioners to avoid overreacting to a single signal or being lulled into a false sense of security The selection of tools should be country specific Not all tools are applicable or relevant in all country circumstances e g due to specific data requirements The use of various tools sh
9. 21u10u022042euJ jeinueui jeay pue seoud 1essy otoz enaupni pue 0 02IN aq 8002 auinquiMs psjii e ep ys aouejeq pug pue se2ud jassy z o1s IVON jeloueul4 6002 ejep uuewyaq 4 pueu pue ouog 3IUuIOUO230J9P8lA TTO iow pue 199us eoueJeg jeloueul4 nyesear Zuo pjoyasnoy pue ajzesoduod A epueuly ejep Jaays e uej eq pue moj Use 9002 ANI assay pue o IsouuaH saoud jassy jeloueul4 Z9 ezuoo eyep 1eaus e uejeq pue s id jassy eyep yaays aouejeq pue otoz alos pue pawn eSa esouldsy ainsodxa Supjueq Japsoq ssoi5 TTOZ ER e a1ezznos GERE yaays souejeq pue DUE aunsodxa Surjueq zounw reun J9pJOq sso15 ZTOZ ouanas pew jeloueul4 suljueg supjueg 8c 29 Appendix Tools Binder TOOLS FOR SYSTEMIC RISK MONITORING February 2013 Contents Page I Conditional Value At Risk COVab uie trii cate adeceexediou e ia dudo du aad 30 Ts Jorit Disttess Indie AOS outs oet ded e cred eae a a anaa E aaaea A dedu din dedu PDA id 32 TH Returns SOTO Ves dona a cance nce DE ata use dp ps natu cute o ab bcd u e a a e e Rue E RUE 34 IV DIR a Re eII INL oi ote ict s crated Gu dtl cerato dt eu cate siete Mowe tual nt ctae s ones Roue cens 36 V Market Based Probability of Detatnl ticsct ac c tcestevant aystceasariancassndcrsdiwataswucamnieeiduncunisdeateibuwets 38 VI Debt Sustainability Analysis DSA 20 svsccsssvsaeaceptacens abscess tein vx ego
10. 3 39 7 US 35 8 Full US 9 6 14 6 China 1 6 1 9 China 1 3 2 2 Taiwan 0 2 0 4 Taiwan 5 9 6 8 India 0 4 0 5 India 0 2 1 3 Indonesia 0 4 0 4 Indonesia 1 6 5 7 Malaysia 0 2 0 2 Malaysia 1 4 3 0 Philippines 0 0 0 1 Philippines 5 4 6 0 South Korea 3 1 3 2 South Korea 4 1 14 7 Thailand 0 6 0 6 Thailand 2 9 7 2 Vietnam 0 1 0 1 Vietnam 0 9 2 6 1 Assumes loss given default or lambda is 1 The figures represent the direct and indirect effects of failures 2 This results of this shock are highly sensitive to the choice of parameters The benchmark assumes lambda 1 rho 0 35 Source IMF 2011a 56 XIV SYSTEMIC LIQUIDITY RISK INDICATOR The Systemic Liquidity Risk Indicator SLRI is constructed from data on violations of arbitrage relationships in the global financial system It measures the intensity of liquidity shortages in global markets working as a high frequency indicator of tail liquidity risks It is easy to use update It should be view as a coincident indicator of systemic liquidity shortages albeit it has been shown to forecast extreme crisis events in the banking sector Tool Snapshot Attributes Description Coverage Sectors Institutions Global capital markets In general it cannot be used to evaluate liquidity risks in individual markets unless there is a high degree of segmentation Types of risk Contractions in market and funding liquidity at a global level Interpretation Main
11. SHOCK SCENARIOS The framework for sovereign Funding Shock Scenarios FSS evaluates the vulnerability of sovereigns to sudden stops situations when foreign investors stop buying or start selling off their holdings of government bonds It assesses the potential impact of foreign investor outflows on the balance sheet of the domestic banking system and how it may affect sovereign bank linkages It is easy to update and can be used along with standard debt sustainability analyses DSA Attributes Description Coverage Sectors Institutions Public sector Types of risk Sovereign risk Interpretation Main output Banking sector exposure to own government debt in percent of bank assets Other outputs Sovereign funding needs under different scenarios of foreign investor outflows Thresholds N A Time horizon Forward looking one year ahead Data requirements Sovereign gross financing needs sovereign debt investor base banking sector assets Reference Arslanalp and Tsuda 2012 Methodology The FSS aims to assess the sovereign s ability to manage a hypothetical loss of international market access a funding shock triggered by pull out of foreign investors over a year through greater reliance on domestic investors 45 The analysis relies on three parameters regarding investment decisions of foreign private investors over a one year horizon namely 1 their contribution to funding of the overall fiscal deficit a ii their rollover
12. Spillover coefficient SC and Toxicity Index TI Thresholds No specific thresholds When FISI 1 asymptotic independence among FIs as the value of BSI increases bank linkages rise Time horizon Coincident indicator of interconnectedness Data requirements CDS spreads equity prices or out of the money option prices bond spreads interbank financing cost spreads Reference Segoviano and Goodhart 2009 33 Methodology A distress dependence measure is based on estimating the Consistent Information Multivariate Density Optimizing CIMDO density of the banking system that captures time varying linear and nonlinear distress dependence among banks Denote by p x y r the CIMDO density of the financial system defined by FIs X Y and R The Joint Probability of Distress JPoD is estimated by integrating the density function over the tail of the distribution It is used as an input to construct all banking stability measures JPoD f f plx y r dxdydr The FISI reflects the expected number of FIs becoming distressed given that at least one FI has become distressed Denote by x x x the distress threshold of return for FIs x y and r P X2xj P Y2x j P R2x 1 P X lt x Y lt x mer Bank interlinkages are assessed by estimating the following conditional probabilities First the probability of distress of bank X conditional on bank Y being distressed is computed This measure j P X gt xiYzxl PY 2 x Second the PCE
13. and Lucchetta 2010 Systemic Risks and the Macroeconomy IMF Working Paper 10 29 and forthcoming NBER book Quantifying Systemic Risk Joseph G Haubrich and Andrew W Lo edited conference proceedings University of Chicago Press Dell Ariccia Giovanni Deniz Igan Luc Laeven Hui Tong Bas B Bakker J r me Vandenbussche 2012 Policies for Macrofinancial Stability How to Deal with the Credit Booms IMF Staff Discussion Note 12 06 http www imf org external pubs cat longres aspx sk 25935 Diebold Francis X and Kamil Yilmaz 2009 Measuring Financial Asset Return and Volatility Spillovers With Application to Global Equity Markets Economic Journal Vol 119 pp 15 71 2012 Better to Give than to Receive Predictive Directional Measurement of Volatility Spillovers International Journal of Forecasting Vol 28 Issue 1 January March Espinosa Vega M and J Sole 2010 Cross Border Financial Surveillance A Network Perspective IMF Working Paper 10 105 Gonz lez Hermosillo Brenda and Heiko Hesse 2009 Global Market Conditions and Systemic Risk IMF Working Paper 90 230 Gray Dale F and Andreas A Jobst 2011 Modeling Systemic Financial Sector and Sovereign Risk Sveriges Riksbank Economic Review 2011 2 Hamilton James D and Raul Susmel 1994 Autoregressive Conditional Heteroskedasticity and Changes in Regime Journal of Econometrics Vol 64 September October pp
14. controlling for other factors But at some point the 95 percentile of the distribution of countries in terms of interconnectedness increases in cross border links begin to have detrimental effects on domestic banking sector stability At a yet higher point when a country s network of interlinkages becomes almost complete the probability of a crisis goes down again This effect is stronger for funding recipient banking systems than for funding provider banking systems EE Crisis observations banking crisis dummy 1 Higher crisis Higher crisis probability area probability area nonparametric nonparametric 0 5 model model Banking crisis probability annual polynominal 02 filter order 3 0 1 ps NEN 1 A Upstream interconmectedntss ikabilties centrality Source ih k Mu oz and Scuzzarella 2011 54 XIII CROSS BORDER NETWORK CONTAGION The network analysis model measures interconnectedness among banking systems and traces a spillover path from one institution s insolvency and or funding difficulties to others It uses consolidated cross border banking statistics from BIS It uses data on actual exposures across banking systems and gives an estimate of both loss given default and the spillover direction In essence the methodology could be replicated with inter institution exposures to measure domestic interconnectedness if such data is available Tool S
15. crisis probability measure focusing on an excessive credit growth indicator It evaluates non linear effects by allowing the interaction between the latter and other risk factors including leverage noncore liabilities and asset prices It also constructs conditional crisis signals and evaluates the performance of each risk factor as an early warning indicator Tool Snapshot Attributes Description Coverage Sectors Institutions Banking sector Types of risk Credit risk market risk funding risk Interpretation Main output Time varying banking crisis probability Other outputs Marginal effect on systemic risk from individual factor indicators threshold values for risk factors type and type II forecast errors Thresholds Yes Time horizon Near to medium term predictive power Data requirements Banking crisis database from Reinhart and Rogoff 2010 or from Laeven and Valencia 2010 annual data from IFS WEO Haver and Bloomberg Reference Lund Jensen 2012 Users IMF 2011b 71 Methodology The model assumes that the binary banking crisis variable y evaluated for country i at time t is drawn from a Bernoulli distribution that depends on k systemic risk factors lagged h periods The probability of a banking crisis is specified as Pry Xs Dla xap where o is the cumulative density of a standard normal distribution probit or a standard logistic distribution logit The underlying risk factors includ
16. long to provide an adequate number of observations of extreme movements Example Trigger NY WOANDAUBWNH c VIXC MSCIWLDC Source IMF Staff estimates The table shows the distress dependence between 17 US financial institutions with 1 indicating the presence or not of contagion to others potential before the 2007 2009 crisis at the 5 percent level of significance The matrix is filled in from logit regressions of the probability of one institution being in distress conditional on another institution being in distress controlling for overall market indicators The rows are the trigger institutions followed by a constant change in the VIX and MSCI World index SC denotes spillover coefficient For example if institution 4 is the trigger then it contributes to 7 4 percent of all possible outward spillovers Overall total spillover coefficient is 81 16 17 0 30 This can be compared to another period The table can also be replicated for the marginal effects derived from the regressions in which case the intensity of spillovers can be derived 38 V MARKET BASED PROBABILITY OF DEFAULT A market based default measure provides a forward looking indicator of default risk by estimating the likelihood that an institution s future value of assets will fall below its distress point It combines market data on traded equity market cap equity return and equity volatility or traded CDS with balance shee
17. risk monitoring dashboard that can be tailored to each country s specific circumstances and key risk factors at a given point in time The illustration uses an unidentified advanced country as an example 49 The systemic risk dashboard combines complementary tools and allows to construct a comprehensive story about a country s key systemic risk at a point in time The sample dashboard for country X at end 2007 addresses the six questions successively in six chart panels and provides a summary of the key observations under each panel as follows e Panel A Credit growth has slowed down and banking stability is falling fast and below 2003 levels at end 2007 Systemic risk is starting to materialize This panel combines a low frequency indicator of credit growth change in the credit to GDP ratio with a high frequency market based indicator of systemic risk in the banking system Distance to Default Together this combination provides insights on the particular phase of systemic risk among financial institutions Consumer credit growth has fallen below 2001 levels The market price based measure shows that banking sector vulnerabilities are heightened e Panel B There are mixed signals from asset prices house prices are falling red for Country X and for countries to which Country X s banks are exposed However not all equity market models are showing misalignments for Country X and its trading partners This panel combines heat map
18. such as combinations of credit to GDP and asset valuation measures and certain high frequency market based indicators Thresholds Policymakers need clear and reliable signals indicating when to worry and when to take action and allowing them to monitor the impact of such action over time Despite recent progress further work is needed in this area System s behavior The capacity to model aggregate agent behaviors is limited in several areas such as banks approaches to internalizing the materialization or increasing likelihood of systemic risk potential reverse feedbacks and multi round effects i e perfect storms and nonlinear risk correlations during periods of financial distress More broadly the incomplete nature of the toolkit highlights the need to avoid mechanistic or narrow approaches to systemic risk monitoring The successful use of quantitative diagnostic tools depends critically on the use of sound judgment Policymakers should not be led to believe that some quantitative approaches e g stress tests or crisis prediction models are all in one tools for systemic risk assessments Indeed such assessments should bring together not only various types of tools but also qualitative information based on market intelligence or on a thorough analysis of a country s macroeconomic and financial stability frameworks TTOZ e ep 199us o2ue e sqor pueAeJ5 pain ep 1e2u leq 3 epueu pue saoud s
19. suggested by Diebold and Yilmaz 2009 DY is a time varying indicator of outward returns spillovers of institutions the contribution of one institution to systemic risk The indicator uses market data on returns CDS spreads or equity prices to estimate average not extreme contributions and is easy to use update It also has good in sample forecasting properties for systemic stress but does not identify the underlying spillover channels except those between institutions Tool Snapshot Attributes Description Coverage Sectors Institutions All financial institutions with market based high frequency data on various returns Ex equity returns CDS spread changes returns on market value of assets Types of risk Contribution of one institution to system wide spillover risk Interpretation Main output The fraction of one institution s spillover contribution to all possible spillovers of all other institutions contribution to systemic risk Other outputs The fraction of all possible spillovers received by an institution from others vulnerability to systemic risk Thresholds Yes e g 0 83 for all US institutions and 0 74 for Euro Area institutions signaling 2007 2009 crisis phase Time horizon Good for predicting near term materialization of financial system wide stress Data requirements High frequency market based financial time series flexible series of returns but limited to institutions with market data Refer
20. 0 39 28 10 25 22 66 7 2 18 31 36 02 12 35 64 31 59 21 42 405 50 05 09 18 43 59 5 3 26 34 88 01 26 04 37 25 23 09 97 14 366 12 05 04 39 63 4 4 182 42416 81 11 24 08 21 23 26 26 93 19 84 17 01 08 08 83 15 17 21 55 07 A40 39 58 29 3144 19 13 18 11 04 29 22 33 70 1 16 29 24 19 05 23 10 23 150 195 12 02 08 04 02 27 46 20 554 7 38 24 49 01 25 18 146 20 28 14 55 21 51 29 06 19 13 847 Contribution to others 240 41 63 B 46 4 102 46 95 30 60 2 60 2 18 28 35 9673 Contribution including own il Gaba 95 8 8 12 6 153 3 9 62 6 o 84 m 50 1700 Spillover Index 25 4 7 1 5 5 u 5 10 3 6 2 6 3 2 3 Go Source Based on Arsov and others 2013 The table shows the variance decomposition based on a VAR 2 lags of weekly equity returns in excess of S amp P500 returns of the top 17 United States financial institutions based on the crisis sample 2007 2011 in percent Banks in columns represent the triggers of shocks and those in rows the recipient of shocks The third row from the bottom shows the contribution of bank i in columns to spillovers into others and is the sum of all the rows under i The last row spillover index computes the same thing but as percentage of all potential spillovers into others 967 3 For instance bank 1 is the largest contributor of spillovers with 25 percent of all spillovers into others 11 percent by bank 7 From this matrix bank 1 has the most contribution and bank 7 has the second most contributi
21. 307 33 International Monetary Fund 2002 Information Note on Modifications to the Fund s Debt Sustainability Assessment Framework for Market Access Countries July 2003 Sustainability Assessments Review of Application and Methodological Refinements June 2006 FSI Compilation Guide http www imf org external pubs ft fsi guide 2006 2007 Republic of Croatia Selected Issues IMF Country Report No 07 82 2009a Detecting Systemic Risk Chapter 3 Global Financial Stability Report April http www imf org External Pubs FT GFSR 2009 01 index htm 2009b Assessing the Systemic Implications of Financial Linkages Chapter 2 Global Financial Stability Report April http www imf org External Pubs FT GFSR 2009 01 index htm 78 2010 Recovery Risk and Rebalancing Chapter 1 World Economic Outlook World Economic and Financial Surveys Washington October 201 1a Japan Spillover Report for the 2011 Article IV Consultation and Selected Issues IMF Country Report No 11 183 July 2011b Towards Operationalizing Macroprudential Policy When to Act Chapter 3 Global Financial Stability Report September 201 1c Mexico 2011 Article IV Consultation IMF Country Report No 11 250 July 2011d How to Address the Systemic Part of Liquidity Risk Chapter 2 Global Financial Stability Report April Financial Stability Board 2010 The IMF
22. 4 Systemic Risk Dashboard for a Fictitious Country X at end 2007 Summary Assessment Overall the set of tools suggests that Country X is about to face intensified financial stress although the extent of the crisis and its implications for economic growth are unclear From the dashboard presented below a policymaker could formulate a first assessment of key sources of systemic risks In this example Country X is facing financial stresses that could have a systemic impact in the financial sector Sovereign risk is heightened by contingent liabilities Contagion risks from financial sector problems in partner countries would have a large domestic impact Among asset market indicators house prices are clearly decelerating while consumer credit growth has slowed along with a sharp drop in banking stability indicators However amplification channels through the broader financial system and domestic economy while uncertain do not yet seem to play a significant role Also there is no strong signal that a full fledged financial crisis is about to materialize even though its probability is rising 23 Consumer Credit GDP change Banking System Distance to Default in y o y percentage points Risk indicator in number standard deviations 10 0 8 0 6 0 4 0 E B MDC cLou 05 ao ost oc ast g gdgg oeggoeg egdgdgB goggog oo O e amp m un r 00 QO cw TNR DAN TO WOH o0 o0 o0 000000 hn OO o Oo cc ccOcO uu c00000000000000000505 0
23. 9 March 79 Segoviano Miguel and Charles Goodhart 2009 Banking Stability Measures IMF Working Paper 09 04 Washington International Monetary Fund Severo Tiago 2012 Measuring Systemic Liquidity Risk and the Cost of Liquidity Insurance IMF Working Paper 12 194 Washington International Monetary Fund Schaechter A C Emre Alper Elif Arbatli Carlos Caceres Giovanni Callegari Marc Gerard Jiri Jonas Tidiane Kinda Anna Shabunina and Anke Weber 2012 A Toolkit to Assessing Fiscal Vulnerabilities IMF Working Paper 12 11 Sun Tao 2011 Identifying Vulnerabilities in Systemically Important Financial Institutions in a Macro financial Linkages Framework IMF Working Paper 11 111 http www imf org external pubs cat longres aspx sk 24841
24. Caja Segovia Caja Rioja 4 Caixa Banca 281 554 La Caixa Caixa de Girona 5 Catalunya Caixa 77 049 Sas Ea Caixa Catalunya Caixa Tarragona Caixa Manresa 6 Nova Caixa Galicia 76 133 27 Caixa Galicia Caixanova 7 Unicaja Unicaja 40 214 Ie Em Eo Caja de Laen 8 Unnim 28 924 a Ee Caixa Sabadell Caixa Terrassa Caixa de Manlleu 9 Kuxta 20 016 ee ta 10 CAM Caja Mdeiterraneo 74 478 L250 Le a 00 NENNEN Sources Table 2 of Ong Jeasakul and Kwoh 2012 64 XVIII THRESHOLDS MODEL The Noise to Signal ratio relies on macroeconomic and financial balance sheet data to select variables and corresponding thresholds that can signal the possibility of financial crisis materializing in the future It is easy to use update and has reasonable in sample forecasting properties for systemic stress working better in advanced countries than in emerging or developing economies Attributes Coverage Sectors Institutions Types of risk Interpretation Main output Other outputs Thresholds Time horizon Data requirements Reference Tool Snapshot Description Financial sector as a whole Can also cover alternative groups of institutions provided a group specific measure of distress is available Risk of distress for the financial system as whole Set of variables and corresponding thresholds which in combination produce early warning indicators of potential financial crisis Type l and Type ll errors indicati
25. Central Bank Central Government Banking sector Corporations Households Rest of the world Creditor Claims Liabilities Claims Liabilities Claims Liabilities Claims Liabilities Claims Liabilities Claims Liabilities Central government 0 1157 19055 6730 a E EE 40093 582 Banking sector 330 14434 6730 19055 14617 33447 49464 23298 17810 19710 Corporations 8 150 ye m 33447 14617 m 33776 11232 Households 68 0 ate 23298 49464 Rest of the world 28832 1631 582 40093 19710 17810 11232 33776 Table 2 Croatia Net Intersectoral Asset and Liability Positions in millions of Kuna December 2005 Debtor Central Bank Central Government Banking sector Corporations Households Rest of the world Creditor Claims Liabilities Claims Liabilities Claims Liabilities Claims Liabilities Claims Liabilities Claims Liabilities Central government 1 345 29191 9336 EN 2s ss 51983 465 Banking sector 4222 39566 9336 29191 34598 61175 100381 78971 67800 35969 Corporations 13 0 e m 61175 34598 73256 11134 Households 22 0 di 78971 100381 Rest of the world 54908 19 465 51983 35969 67800 11134 73256 Source IMF 2007 50 XI SYSTEMIC CCA The systemic Contingent Claims Approach CCA extends CCA to quantify the system wide financial risk and government contingent liabilities by combining individual risk adjusted balance sheets of financial institutions and the dependence between them It provides forward looking estimates Tool Snapshot Attri
26. DP Change 0 23 0 30 0 27 0 31 Source IMF staff estimate 66 XIX MACRO STRESS TESTS Macro stress tests provide quantitative analyses of system level risks and vulnerabilities FSAPSs assess a range of risks in stress tests within the broad categories of credit risk market risk liquidity risk and contagion risk Tool Snapshot Attributes Description Coverage Sectors Institutions Mainly banks but nonbanking sector including insurance sector being increasingly covered Types of risk Credit risk market risk liquidity risk and contagion risk Interpretation Main output Capital Adequacy Ratios CAR under extreme but plausible scenario Other outputs Nonperforming loans loan loss provisioning Value at Risk liquidity position and net open currency position under extreme but plausible scenario Thresholds Regulatory capital requirement Time horizon Two to five year scenario being used Data requirements Balance sheet and P L data of core financial institutions real and financial data Reference Moretti Stolz and Swinburne 2008 Methodology Typical FSAP style macro stress tests consist of four steps 1 identification of specific vulnerability or concerns ii construction of extreme but plausible stress scenario using macroeconomic model that links external shocks to macroeconomic variables iii mapping of the stress scenario into financial institutions balance sheets and income statements by 67
27. FSB Early Warning Exercise Design and Methodological Toolkit MF Occasional Paper available at http www imf org external np exr facts ewe htm Kealhofer S 2003 Quantifying Credit Risk I Default Prediction Financial Analysts Journal Jan Feb pp 30 44 Laeven and Valencia 2010 Resolution of Banking Crises The Good the Bad and the Ugly IMF Working Paper 10 146 Washington International Monetary Fund Lopez Espinosa G Moreno A Rubia A and Valderrama L 2012 Short term Wholesale Funding and Systemic Risk A Global CoVaR Approach IMF Working Paper 12 46 Washington International Monetary Fund Lund Jensen K 2012 Monitoring Systemic Risk based on Dynamic Thresholds IMF Working Paper 12 149 Washington International Monetary Fund Moretti Marina St phanie Stolz and Mark Swinburne 2008 Stress Testing at the IMF IMF Working Paper 08 206 Washington International Monetary Fund Ong Li Lian Phakawa Jeasakul and Sarah Kwoh 2013 User Guide HEAT A Bank Health Assessment Tool IMF Working Paper forthcoming Reinhart C M and K S Rogoff 2010 From Financial Crash to Debt Crisis NBER WorkingPaper No 15795 forthcoming American Economic Review Reint Gropp amp Marco Lo Duca amp Jukka Vesala 2009 Cross Border Bank Contagion in Europe International Journal of Central Banking International Journal of Central Banking vol 5 1 pages 97 13
28. WP 13 168 NIMF Working Paper Systemic Risk Monitoring SysMo Toolkit A User Guide Nicolas Blancher Srobona Mitra Hanan Morsy Akira Otani Tiago Severo and Laura Valderrama INTERNATIONAL MONETARY FUND 2013 International Monetary Fund WP 13 168 IMF Working Paper Monetary and Capital Markets Department Systemic Risk Monitoring SysMo Toolkit A User Guide Prepared by Nicolas Blancher Srobona Mitra Hanan Morsy Akira Otani Tiago Severo and Laura Valderrama Authorized for distribution by Laura Kodres and Dimitri Demekas July 2013 This Working Paper should not be reported as representing the views of the IMF The views expressed in this Working Paper are those of the author s and do not necessarily represent those of the IMF or IMF policy Working Papers describe research in progress by the author s and are published to elicit comments and to further debate Abstract There has recently been a proliferation of new quantitative tools as part of various initiatives to improve the monitoring of systemic risk The SysMo project takes stock of the current toolkit used at the IMF for this purpose It offers detailed and practical guidance on the use of current systemic risk monitoring tools on the basis of six key questions policymakers are likely to ask It provides how to guidance to select and interpret monitoring tools a continuously updated inventory of key categories of tools
29. a a A VR Ph iie is pre Du 74 30 I CONDITIONAL VALUE AT RISK COVAR The CoVaR uses market data to assess the contribution of an individual financial institution to systemic risk It is easy to use update and has good in sample forecasting properties for systemic stress but does not identify the underlying spillover channels Tool Snapshot Attributes Description Coverage Sectors Institutions All financial institutions with high frequency market data that provide various measures of return e g equity prices CDS spread or Market Value of Assets Types of risk Contribution of one institution to system wide distress Interpretation Main output The expected loss in the financial system conditional on one or many financial institutions being in distress left tail outcome Other outputs Total expected loss in the financial system conditional on one or more financial institutions being in distress the vulnerability of an institution to system wide risk Exposure CoVaR Thresholds Yes e g 7 2 percent returns on market value of assets for US institutions 0 9 1 8 for Euro Area institutions signaling 2007 2009 crisis phase Time horizon Good for predicting near term materialization of financial system wide stress Data requirements High frequency market based financial time series flexible series of returns but limited to institutions with market data Reference Main Adrian and Brunnermeier 2010 Users Arso
30. acy Ratio CCA Contingent Claims Analysis CCB Committee on Capacity Building CIMDO Consistent Information Multivariate Density Optimizing DiDe Distress Dependence DNL SRMS De Nicolo and Lucchetta Systemic Risk Monitoring System CoVaR Conditional Value at Risk DSA Debt Sustainability Analysis DSGE Dynamic Stochastic General Equilibrium DtD Distance to Default EDF Expected Default Frequency FSaR Financial System at Risk FSI Financial Soundness Indicator GDP Gross Domestic Product GDPaR GDP at Risk JDI Joint Distress Indicator JPoD Joint Probability of Default KMV Kealhofer McQuown and Vasicek LGD Loss Given Default MSCI Morgan Stanley Capital International OOS On the run Off the run Spread PCA Principal Components Analysis PCE Probability of Cascade Effects SCCA Systemic Contingent Claims Analysis SLRI Systemic Liquidity Risk Indicator VaR Value at Risk VAR Vector Autoregression VD Variance Decomposition VIX Volatility Index I INTRODUCTION l Macroprudential policymakers need to know when to act Policies to mitigate system wide risks should be based on detailed information on where and when such risks are building up and which channels may amplify their impact on the broader economy pn This paper aims to clarify the nature and use of the systemic risk monitoring tools that are currently available Building on earlier surveys it looks at all dimensions of systemic risk and assesses the tools ability to capture these dime
31. and Fiscal Stress A Proposed Set of Indicators IMF Working Paper 11 94 Washington International Monetary Fund Basel Committee of Banking Supervision BCBS 2012 Models and Tools for Macroprudential Analysis May available at https www bis org publ bcbs wp21 htm Benes Jaromir Michael Kumhof and David Vavra 2010 Monetary Policy and Financial Stability in Emerging Market Economies An Operational Framework paper presented at the 2010 Central Bank Macroeconomic Modeling Workshop Manila www bsp gov ph events 2010 cbmmw papers htm Bisias Dimitrios Mark Flood Andrew W Lo and Stavros Valavanis 2012 A Survey of Systemic Risk Analytics U S Department of the Treasury Office of Financial Research Working Paper 0001 available at http wwvw treasury gov initiatives wsr ofr Pages ofr working papers aspx Borio C and M Drehmann 2009 Assessing the Risk of Banking Crises Revisited BIS Quarterly Review March Chan Lau Jorge Srobona Mitra and Li Lian Ong 2012 Identifying Contagion Risk in the International Banking System An Extreme Value Theory Approach International Journal of Finance and Economics available at http onlinelibrary wiley com doi 10 1002 1fe 1459 abstract Cih k M Mufioz S and Scuzzarella R 2011 The Bright and the Dark Side of Cross Border Banking Linkages IMF Working Paper 11 186 Washington International Monetary Fund 77 De Nicol
32. ate loans to total loans Commercial real estate loans to total loans 62 XVII BANK HEALTH ASSESSMENT TOOL HEAT The HEAT is a tool for calculating a Bank Health Index BHI based on simple CAMELS type ratings for each bank including systemically important ones It is simple to use and update and provides a measure of relative but not absolute health of a banking system System wide health vis vis a global peer group of banks such as the G SIBs can also be assessed by taking the aggregate of each variable for all banks in the system to derive system wide BHIs or by inputting system wide ratios available from the Financial Soundness Indicators database Tool Snapshot Attributes Description Coverage Sectors Institutions Deposit takers and other financial corporations Types of risk Solvency credit and liquidity risk Interpretation Main output Bank Health Index for each bank that aggregates standardized versions of five financial ratios capital adequacy nonperforming loans ratio return on assets liquid assets ratio and leverage ratio Other outputs uH Thresholds No Time horizon Low frequency backward looking indicators Data requirements Banks financial statements from Bankscope Bloomberg or SNL Reference Ong Jeasakul and Kwoh 2012 63 Methodology For each bank five financial ratios are calculated capital adequacy total equity or Tier 1 capital to Risk weighted Assets nonperforming loans to gro
33. blic debt to GDP ratio Other outputs N A Thresholds No Time horizon Typically five years Data requirements GDP inflation public debt public revenue and expenditure interest rate on public debt and public debt composition Reference IMF 2002 and 2003 Methodology The sensitivity test on sovereign risk by DSA consists of three steps The first step sets a baseline scenario on key economic variables such as GDP growth rate and inflation rate as well as interest rate on public debt The second step projects public debt to GDP ratio using estimated flows of revenue and expenditure under the baseline scenario The final step examines the dynamics of public debt to GDP ratio under several shock scenarios including rise in real interest rate decline in GDP growth rate and a realization of 41 contingent liabilities by 10 percent of GDP The specification of shock scenario is not based on any estimates So for example the contingent liability test should be refined by several approaches such as stress test that provides an estimate of the fiscal cost of bank recapitalization in case of the materialization of various risks cross country evidence on past banking system crises that presents crude estimates of the possible contingent liabilities and other estimates of contingent liabilities by sophisticated method such as systemic CCA Example Real depreciation and contingent liabilities shocks The chart shows the dynamics of public
34. border exposures as they may indicate that on aggregate a financial system is exposed to credit risk from certain countries or is vulnerable to funding risk from cross border sources 38 A more forward looking perspective on the buildup of cross border spillover risks is provided by balance of payments and international investment position data such as data on capital inflows and outflows and on changes in banks foreign liabilities These can be combined in the T model to obtain threshold based signals of a potential financial crisis 39 Macro Stress Tests also increasingly take into account cross border linkages in identifying adverse scenarios as relevant in each country case Indeed in order to assess domestic financial institutions solvency and liquidity positions comprehensively they need to capture a range of risks e g foreign credit liquidity foreign sovereign and foreign market risks arising from cross border exposures and related risks and scenarios in other jurisdictions 40 In order to assess more deeply and dynamically the interdependences that may generate cross border spillovers among financial systems or institutions policymakers should ideally have access to the necessary data on actual interlinkages between such financial institutions and systems In this case network models can be used to gauge such spillovers due to shocks in any one or more financial institutions e g the G SIFIs or among financial sys
35. butes Description Coverage Sectors Institutions All financial institution with balance sheet data and high frequency market data equity options and CDS Types of risk Contribution of each institution to system wide distress and spillover risk in general Interpretation Main output Total expected loss in the financial system and government contingent liabilities Other outputs Unexpected loss and extreme risk in the financial system Thresholds No Time horizon Coincident indicator of interconnectedness Data requirements Daily market capitalization of each institution default barrier estimated for each institutions based on quarterly financial accounts risk free interest rate and one year CDS spreads Reference Gray and Jobst 2011 Methodology The systemic CCA can be decomposed into two estimation steps The first step uses CCA to estimate the market implied potential losses for each sample financial institution see CCA The second step uses Extreme Value Theory to model the joint market implied losses of multiple institutions as a portfolio of individual losses with time varying and nonlinear dependence among institutions and estimates system wide losses 5I In the second step firstly a nonparametric dependence function of individual potential losses is defined Then this dependence measure is combined with the marginal distributions of these individual losses which are assumed to be generalized extreme value These marginal distribut
36. cal or growth crisis events For each indicator a threshold value minimizing its noise to signal ratio is obtained and a weight assigned based on its predictive power A composite weighted indicator is thus constructed and mapped into a crisis probability defined as the percentage of crisis observations conditional on the composite indicator flagging Tool Snapshot Attributes Description Coverage Sectors Institutions Financial sector public sector real sector Types of risk Financial Crisis Sudden Fiscal Consolidation Growth Slowdown Interpretation Main output Probability of a systemic financial fiscal or growth crisis Other outputs Composite vulnerability indicator individual financial real and fiscal indicators their threshold values and their associated weights Thresholds Yes specific thresholds for each indicator Time horizon Near term predictive power Data requirements Annual data from WEO IFS OECD Bankscope Worldscope and Bloomberg Reference IMF FSB 2010 Methodology The construction of a crisis prediction measure requires two steps First a crisis event is defined A financial crisis is based on the database provided by Laeven and Valencia 2008 A fiscal crisis is defined as an abrupt fiscal consolidation within a year of at least 2 5 percent of cyclically adjusted primary balance from a negative value of at least 2 5 percentage points 73 A growth slowdown is determined by the lowest 5 percentile of the his
37. croeconomic risk arising from cyclical fluctuations The macroprudential concern stems from the presence of the aggregate risk There are many flexible parameters to mimic different types of economies extent of foreign currency lending the degree to which the central bank manages the nominal exchange rate the sensitivities of both imports and exports to the exchange rate and the ease with which the banks can raise fresh equity capital in financial markets Example The DSGE model was used in IMF 201 1b to assess the effects of countercyclical capital buffers in the presence of two types of shocks shocks related to healthy productivity gains that do not lead to crisis and shocks leading to unhealthy a house price boom that is followed by a crisis The model shows the effects of macroprudential policy on the real economy under the two shock scenarios If there is an unhealthy house price boom that has a high probability of ending in a crisis then countercyclical capital buffers CCBs can successfully cushion the crisis effects on real GDP levels left figure below However if there is a process of healthy productivity gains and policymakers mistake it for an unhealthy process like a house price boom then macroprudential policy can do permanent damage and lower the real GDP level indefinitely right figure below Effects of Macroprudential Policy on Real GDP level Countercyclical Capital Buffers on Two Types of Shocks Healthy produc
38. cs leading to a systemic financial crisis and a sovereign debt crisis D Broader Economy What are the Amplification Channels among Sectors and through the Domestic Economy 32 The interconnections and risk exposures among the financial public and other sectors can play a key role in magnifying systemic risk For instance they may give rise to concentration risks as well as compounded maturity currency and capital structure mismatches The set of Encouraged FSIs provides snapshots of household and corporate leverage and enables comparisons across countries More detailed analysis of balance sheet data in key sectors public private financial private nonfinancial household and nonresident through the Balance Sheet Approach BSA facilitates cross sectoral assessments of maturity currency and capital structure mismatches The BSA tool can be used to stress test sectoral positions by assuming shocks related to interest rates and exchange rates It also provides an indication of the likelihood that an adverse shock may get amplified into a systemic crisis 33 Credit growth episodes may also be associated with asset e g real estate price bubbles posing a greater threat to financial stability As such Asset Price models that provide indicators of such bubbles may usefully complement the above tool section B More generally combinations of credit growth leverage and asset price growth such as in 17 the Credit to GDP Based Crisi
39. d to be backward looking indicators A similar set of indicators is provided by Bank Health Assessment Tool HEAT which builds on CAMELS type financial ratios to derive individual bank indices and can be used to monitor aggregate banking soundness 15 Complementing FSIs Market Based Probability of Default measures such as Distance to Default DtD or Expected Default Frequency EDF can be used to assess with higher frequency the probability that individual financial institutions may undergo distress or fail where relevant market prices such as equity or CDS prices are available 16 Macro Stress Tests can be used to examine more closely the sources of financial institution vulnerability and to identify specific weak links in the system Macro stress tests capture a range of risks e g credit liquidity and market risks under extreme but plausible i e tail risk adverse scenarios They combine these risk factors to evaluate whether financial institutions both in aggregate and taken individually have enough capital and liquidity buffers to withstand such scenarios Key challenges in using stress test models include the calibration of appropriate and internally consistent sets of shocks across risk factors and incorporating feedback effects from financial sector problems back into the macroeconomy gt CAMELS stands for Capital adequacy Asset quality Management Earnings Liquidity and Sensitivity to market risk T
40. data on asset prices and related financial variables to assess the likelihood that the financial system as whole will enter different states regarding uncertainty and systemic risk Tool Snapshot Attributes Description Coverage Sectors Institutions Global or domestic markets on aggregate or specific market segments e g FX or interest rate markets Types of risk General degree of uncertainty risk of systemic events Interpretation Main output The probability of financial markets being in different regimes characterized by low medium of high volatility can be extended to consider more than 3 states Other outputs Estimates of the time varying volatility of the financial variables considered Thresholds No specific thresholds Rule of thumb would be to consider a systemic event when the probability of being in a high volatility state surpasses 50 Time horizon Good for predicting near term materialization of financial system wide stress Data requirements High frequency market based financial time series Reference Hamilton and Susmel 1994 Users Gonzalez Hermosillo and Hesse 2009 29 Methodology The basic methodology assumes that a certain variable Y which reflects information about general financial conditions e g the VIX the Ted spread the Euro Dollar Forex Swap etc follows a univariate ARCH Markov Switching model More specifically Y is assumed to evolve as Y a Y amp Et J 2i t is the product of a uni
41. debt to GDP ratio if one time 10 percent of pe GDP shock to contingent liabilities occurs 60 a contingent in 2010 DSA related documents of 55 gece individual countries are available on the DSA website 2 http www imf org external pubs ft dsa ind 45 ex htm 40 Baseline 35 30 2006 2008 2010 2012 2014 2016 42 VII INDICATORS OF FISCAL STRESS This methodology provides a framework to assess fiscal vulnerability and evaluate the likelihood of a full blown fiscal crisis based on the construction of a coincident indicator of rollover pressure and a forward looking index of extreme fiscal stress It can be used as an effective fiscal monitoring tool of sovereign risk based on fiscal fundamentals Tool Snapshot Attributes Description Coverage Sectors Institutions Public sector Types of risk Sovereign risk Interpretation Main output A country specific fiscal vulnerability index and a fiscal stress index Other outputs Fiscal vulnerability measures can be aggregated for advanced and emerging economies Thresholds Yes Time horizon Coincident and medium term indicators Data requirements Low frequency macroeconomic and financial data Reference Baldacci McHugh and Petrova 2011 43 Methodology This approach builds on the construction of two signaling tools First a fiscal vulnerability index measuring the deviation of a set of fiscal indicators including underlying fundamentals long term fiscal needs and rollover r
42. devpedeeneduasansVannoabecees 40 VII Indicators of Fiscal Stres Siene o ed oda etre bea Re et reete cere ge dels A eere ead 42 VII Sovereign Funding Shock Sc nani0s si scien der ctt Ete ats e e De Ee ease suu 44 IX Asset Price Models tnt cetifed ation preset ne esso enr e in obe Y ER ade ete Legi 46 AX Balance Sheet Approach isis erci scat ote Tea eho i ebrei oae da eden ca e rece Dues 48 DoT EY SUC WO C CUN conss Sen trado deu toS Popolo ada M Cea EA ee 50 XII Cross Border Interconm ctedfess io eater b event les vi ead teretes cel do a petu aeo 52 XIII Cross border Network Contagion i us etse er entes Et rege e peu a dd e Ue paret cetus 54 XIV Systemic Liquidity Risk Indicator cg e ustestaser co edes i no sea edi aM ERU d RUM TENER TUCD 56 XV Regime Switching Volatility Model essere tit eter ater desee erue adbduse e peur e esians 58 XVI Financial Soundness Indicators FSIS sss 60 XVII Bank Health Assessment Tool HEAT esesssseeseeeeeeeeee nennen 62 A VIM Thresholds Model eiie tu prios de oai ii dep i dece erea io dp aona t hraa i aaka ES piani 64 XIX Macro Stress D68 95 01 E PEE IIo itane aee aatan aba M eb eR T iaaea 66 AC GDP CRISE EE T oe eta pred ite eerie Ee Nr ERE eR ber reri E ERR E t TURAE 68 XXI Credit to GDP Based Crisis Prediction Model sse 70 AOI OC rS Pr diction MOSS aired ede erat b EE UA Queda cou dde vis wy ec Rae Te ne do den 12 DOM DSGEB IM ode
43. e a real estate vulnerability index 25 Fully fledged DSGE models are needed to quantify the systemic impact of asset price corrections by incorporating nonlinear effects and feedback loops Indeed the macroeconomic impact of asset price booms and busts depends crucially on the behavior of 15 the investor base the dynamics of household leverage and the likelihood of a credit crunch as well as feedback effects on the whole financial sector which can be aided by the construction of structural DSGE models Overall assessment 26 Overall the available toolkit provides a good set of measures for the size and impact of a potential asset price correction while its likelihood remains difficult to assess accurately especially over the near term It helps construct a variety of scenarios featuring alternative path dependent asset price dynamics that support the use of other models including stress test models see section A Yet it could be better linked to investors portfolio rebalancing decisions in order to evaluate systemic effects through asset price externalities C Sovereign Risk How Much is Sovereign Risk a Source of Systemic Risk 27 The build up of sovereign risk can be assessed through Debt Sustainability Analysis DSA which typically projects public debt GDP dynamics over 5 years under baseline and adverse scenarios e g decline in growth rate sharp rise in interest rate and sustained increase in primary deficits
44. e discusses a range of systemic risk monitoring tools They include for example tools focusing on a narrow but potentially systemically relevant sectoral perspective as well as tools to measure the risk of a systemic crisis There are four complementary ways to access and use this guide Figure 1 The authors would like to thank without implicating Jan Brockmeijer Stijn Claessens Gianni de Nicolo Dimitri Demekas Laura Kodres Jacek Osinski Ratna Sahay Amadou Sy and Jos Vi als for very helpful discussions and suggestions Serkan Arslanalp Ivailo Arsov Marcos Chamone Marco Espinosa Vega Dale Gray Deniz Igan Andy Jobst Sonia Mufioz Li Lian Ong Miguel Segoviano Juan Sole and Takahiro Tsuda for constructive comments pertaining to the tools they developed and other reviewers at the IMF The authors plan to regularly update and expand the guidance note as new tools are developed See in particular IMF Financial Stability Board 2010 IMF 20092 Basel Committee of Banking Supervision 2012 and Bisias et al 2012 An in depth discussion of six key questions on systemic risk that policymakers are likely to ask Figure 1 Is potentially excessive risk building up in financial institutions Are asset prices growing too fast How much is the sovereign risk a source of systemic risk What are the amplification channels among sectors and through the broader domestic economy What are the amplification channels through cro
45. e ei eb tos etti helabe beh Maa uaredanteasionatene 16 E Cross Border Linkages What are the Amplification Channels through Cross Border Spillovers 12 84 hibisuienietiilasutubotl ua ea eaaa atanan ei raias aeai Alpiem 18 F Crisis Risks What is the Probability of a Systemic Crisis ccccccsseeseeeseeeees 19 IV Sample Country Case Study lt isiissesisatssistacasptadasags sais ataweenedennts setsdanatsassdsaasletianyartecesspeladazads 21 V Key Findings and Operational Implications 0 cccccccccsseceseeesseeeeeceeseeesseeeseeceseseeseeensaes 25 Refefr iteS einen eie icone Mina ia tek Sole ts reed Soe ve ee acl eat eal E RA 76 Table 1 Characteristics of Different Systemic Risk Monitoring tools A Summary 27 Figures L Structure of the Guide voe at eon da dp i appe tobita cota t fd eei diis nut etu od seht aat 5 2 Buildup of Systemic Risk Sources and Channels cccccsccccssceesseeeseeessseeeeeeseseeseeesseenes 9 3 Unwinding of Systemic Risk Sources and Channels sess 10 4 Systemic Risk Dashboard for a Fictitious Country X at end 2007 sss 22 Appendix Tools Bider es oou tad speed Leo E A E o Ua DESEE 29 Glossary BSA Balance Sheet Approach CAMELS Capital adequacy Asset quality Management Earnings Liquidity and Sensitivity to market risk CAR Capital Adequ
46. e excessive credit growth measured by credit to GDP growth or credit to GDP gap equity and house price inflation banking sector leverage private credit to deposit ratio noncore liabilities foreign banking sector liabilities to M2 and fluctuations in the real effective exchange rate In the single factor analysis credit to GDP growth features as the main contributing factor to systemic risk up to three years ahead In the multivariate specification the combination of credit to GDP gap leverage and equity price inflation appear as the main determinants of systemic risk This tool also allows backing out a crisis signal threshold for alternative modeling specifications with associated critical values for the underlying risk factors It shows that combining several risk factor indicators greatly improves the accuracy of the crisis signal Example Using annual panel data for 36 countries over the period 1975 2010 featuring 26 banking crisis observations the linear combination of one period lagged credit to GDP growth and two period lagged equity price growth yield the banking crisis probability surface that is depicted below Probability of systemic banking crisis in percent Credt to GOP growth in percentage points 20 10 0 Wn x b Equity price growth in percent Source IMF 2011b 72 XXII CRISIS PREDICTION MODELS This methodology identifies a set of 23 to 25 indicators that are correlated with financial fis
47. edictors of GDP growth and the market adjusted equity return of financial firms Second joint forecasts of factors GDP growth and market adjusted equity return of financial firms are generated by Vector Auto Regressions VAR Third 8 quarters ahead VAR forecasts of predictors are used to forecast GDPaR and FSaR via Quantile Auto Regressions QARs Example De Nicol and Lucchetta 2010 examines the out of sample performance of the model specifically assessing whether the model signals a decline in GDPaR prior to 2008Q4 200901 in all G 7 countries The results show predicted changes in GDPaR and actual GDP growth go in the same direction for at least 1 quarter ahead within a three quarters horizon up to 2009Q1 in all countries The out of sample consistency of GDPaR forecasts with the future evolution of actual GDP growth for the most unpredictable event in decades suggests the potential usefulness of this model as a real time risk monitoring tool The countries covered listed by geographical areas are the following North America Canada and the United States Asia Pacific Japan Korea Australia and New Zealand Atlantic U K and Ireland Western Europe France Belgium and Netherlands Central Europe Germany Austria and Switzerland Southern Europe Italy Spain Portugal and Greece Northern Europe Denmark Finland Norway and Sweden 70 XXI CREDIT TO GDP BASED CRISIS PREDICTION MODEL This model computes a banking
48. ence Main Diebold and Yilmaz 2009 Users Arsov and others 2013 35 Methodology Vector Auto regressions VAR of the weekly returns of all institutions are used to derive DY Specifically the variance decomposition VD at a particular lag say 10 is used to derive a matrix of the portion of variance of the shocks to one institution attributable to another institution Variance decompositions allow us to assess the fraction of the 10 step ahead error variance in forecasting x that is due to shocks to xj V ji for each i The DY measure of spillover contributions of institution i is the percentage of institution 7 in the total VD of all institutions The measure is based on central moments rather than extreme tail risk movements Example FROM gt Contribution Banks TO 1 2 3 4 5 6 7 8 9 10 u 12 B 14 15 16 17 from others 1 69 24 28 03 21 18 62 27 19 14 47 06 48 24 05 25 20 391 2 16 506 34 22 23 12 32 08 61 10 43 10 48 15 10 15 17 494 3 158 20 582 02 09 20 37 36 28 13 28 13 23 08 04 08 12 418 4 53 08 11 86 04 14 25 06 11 10 02 02 09 09 04 06 1 184 5 38 12 22 06 38 09 09 13 44 15 39 17 63 20 18 11 45 66 2 6 21 o7 25 10 64 43 64 11 16 41 12 30 52 21 14 29 30 547 7 310 08 66 08 55 35 29 11 06 21 55 11 38 36 08 14 18 701 8 50 23 26 OS 12 20 60 62 37 13 29 04 05 03 03 03 14 308 9 24 79 58 25 76 08 23 27 586 09 07 08 02 03 15 41 09 414 10 85 60 04 16 37 156 64 20 62 266 35 25 67 26 44 28 06 74 1 29 22 30 08 10 34 101 26 18 55 333 1
49. es sum of ind alpha put option o aa of indi us ZEBBASRAS SSERRRRRRRRARR ZONE Summary There is limited evidence that financial sector shocks are spilling over into the real sector at this stage although spillover risk within financial institutions is slowly rising Joint Distress Indicators Crossborder Interbank Network Analysis T icis 70 4 asof 200704 2 Distress Dependence Global 60 4 X Country A Impact on Country X capital levels from pay m contagion risks stemming from credit and T X Country C 40 funding shocks in other countries in percent X Country D ao of pre shock capital signs reversed 20 Tihs o i i E a a m m A H D E FP SG H ak t M NM oO Country Jan 05 Jan 06 Jan 07 Jan 08 Jan 09 Jan 10 Summary Country X continues to be strongly connected to the rest of the world both in terms of actual balance sheet linkages of banks and potential spillover risks from market contagion Credit Based Crisis Probability Model General Crisis Probability Model as gs 085 0 35 Estimated Probability of a Systemic Bankigg 30 probabilities of Financial y 953 Crisis in Count in percent 025 Crisis and Real f o25 4 0 ij 40 4 i o20 Slowdown in CountryX N 2 402 015 4 J 015 3 5 H 3 5 j 0 10 4 j oi Change in credit to GDP ratio 0 05 4 0 05 ee el ee ee a y o 5 2 2 1 1975 1979 1983 1987 1991 1995 1999 2003 2007 2000
50. es involved in systemic risk monitoring Specifically and for practical purposes the assumption is that policymakers would start from a funnel view of the economy looking at 1 narrow sources of risk within the financial sector e g financial institutions or asset markets and then turning to 11 other sources of systemic risks or risk amplification i e in other sectors the broader economy or other countries and finally iii 12 aiming to directly measure the risk and probability of systemic events In addition to better understanding the underlying sources and severity of crisis risks such a structured approach may also help policymakers to mitigate systemic risk more effectively including through a tailored use of specific macroprudential policy tools IMF 201 1b A Financial Institutions Is Potentially Excessive Risk Building Up in Financial Institutions 13 In order to gauge risk buildup at the aggregate level one should use a combination of balance sheet data that indicate whether financial institutions are taking increasing risk with potentially systemic impact Financial Soundness Indicators FSIs provide a starting point as they focus primarily on aggregate balance sheet soundness and may help to identify sources of risk buildup e g FSIs related to sectoral credit growth and leverage 14 FSIs are collected comprehensively for many countries and cover a broad range of key risks and buffers but they ten
51. g time series with over 30 000 public companies worldwide it has identified the proportion of firms with a certain distance to default that actually defaulted within a specific forecasting window This is the expected default frequency EDF For nontraded firms with active CDS markets including sovereigns KMV offers estimates of the PD LGD and risk premium embedded in credit spreads and derives a CDS implied EDF credit measure Example The Moody s KMV methodology has been applied to estimate the individual EDF of the largest five banks in Spain and Italy The figure below shows EDF Banking Sector and Sovereign CDS the asset value weighted EDF for the Spanish and Italian banking sector The waves in bank credit distress unleashed in February 2009 and April 2011 comove positively with the spikes in government credit risk reflected in the sovereign CDS market Source Authors calculation 40 VI DEBT SUSTAINABILITY ANALYSIS DSA DSA examines the effect on the public debt to GDP dynamics of several shocks such as real interest rate shock GDP shock and a realization of contingent liabilities including financial sector bailout specified as an exogenous increase in the debt ratio of 10 percent of GDP It is easy to use and update but is not linked to any estimate of shocks Tool Snapshot Attributes Description Coverage Sectors Institutions Public sector Types of risk Sovereign risk Interpretation Main output Pu
52. his concept was developed by banking supervisors in the United States in order to assess the soundness of individual banks 13 17 From a more aggregate and forward looking perspective credit growth is often central to the buildup of macro financial risk and models such as the Thresholds Model or T model provide rules of thumb on thresholds for changes in credit to GDP and its deviation from trend that may signal a systemic financial crisis However the T model tends to produce thresholds that are fairly low in order not to miss a crisis and should thus ideally be combined with other tools that tend to yield higher thresholds e g Dell Ariccia et al 2012 so as to reduce the chance of a false signal that might lead to a costly policy mistake 18 Finally a number of tools focus on interdependences between financial institutions and assess the risk of spillovers among them In doing so these tools may also allow practitioners to identify systemically important institutions Ideally policy makers have data on actual interlinkages between financial institutions and systems In that case Network models can be used to gauge such spillovers triggered by shocks in any one or more financial institutions e g the weak links identified above Such tools can also be applied to aggregate data on cross country exposures to gauge cross border spillover risks among financial systems e g based on BIS data These models provide informat
53. ial institutions are exposed to the SLRI by running a regression of equity returns R or CDS spreads against the SLRI and other control variables Z Ri t Z t B BISLRI t e t a t o t exp zo wi SLRI t y e t 1 Note that the SLRI can affect both the mean and volatility of returns since liquidity shortage increases the riskiness of financial institutions e t is a white noise shock Example Liquidity conditions during the 2008 crisis Daily data on 36 violations of arbitrage including CIP CDS Bond basis OOS and the Bond Swap basis from 2004 until 2010 Similar data on equity returns from 53 global or regionally important banks across the globe Figure 1 shows the evolution of the SLRI over time It illustrates the sharp reduction in global liquidity around the Lehman debacle in 2008 Figure 2 shows the average by location annualized bank return volatility under normal liquidity conditions SLRI 0 light green bar and under liquidity stress SLRI 2 std below its mean dark green bar Clearly global liquidity shortfalls increase substantially the riskiness of banks Figure 1 Figure 2 ml Ta 80 0 70 1 M 60 50 2 2 40 3 30 m 20 r 4 10 5 0 Lis US UK Europe Australia India and Japan 6 and New Korea 2004 2005 2006 2002 2008 2009 2010 Pana Source Severo 2012 58 XV REGIME SWITCHING VOLATILITY MODEL The Regime Switching Volatility Model uses high frequency
54. ignal of 1 that is not followed by a crisis in the future produces a Type II error V combine the different errors to compute the NSR as NSR l VI relative to total noncrisis observations whereas 7 denotes the fraction of type I errors relative to total crisis observations The lower the NSR better the trade off between the two errors produced by the forecasting variables and their thresholds The term zt denotes the fraction of type II errors Example Annual data on credit to GDP and a crisis indicator covering 169 countries from 1970 to 2010 is used Comparison between two alternative measures of credit and their corresponding thresholds as predictors of crises The crisis indicator is based on updated data from Laeven and Valencia 2008 The table below produced for illustrative purposes shows the NSR for different lags of the forecasting variable and different thresholds defined as the number of standard deviations std above historical average for each variable calculated on a country by country basis The lowest NSR in yellow suggests that a 2 std move in the credit growth presents the best trade off between type I and type II errors 2 years before a potential crisis period Hence authorities in a given country should be alert about the possibility of a crisis materializing in the next two to three years if the credit to GDP change moves by 2 std or more Intensity IS Credit to GDP Credit to of Change s Gap G
55. ing crisis dummy from Laeven and Valencia 2008 database Reference Cih k Mufioz and Scuzzarella 2011 53 Methodology The network model examines the impact of a banking system s interconnectedness node centrality on the probability of a banking crisis using an econometric specification that controls for a set of macroeconomic and institutional variables The estimated log likelihood function is in 5 5 Pein ex Go Peali Foe x6 where P i t is the banking crisis dummy variable proposed by Laeven and Valencia 2008 and X i t the vector of explanatory variables Based on a banking system s alter based centrality notion that shows its relative importance in cross border exposures in the global banking network two measures of interconnectedness are created downstream interconnectedness asset centrality and upstream interconnectedness liability centrality The level and slope effects of interconnectedness on the likelihood of a banking crisis are estimated using a multivariate probit model approach and a nonparametric algorithm Example Using a sample of 189 banking systems over 1977 2009 an M shaped curve showing a non linear relationship between the likelihood of a banking crisis and its interconnectedness to the global banking network is obtained In a country whose banking sector has relatively few linkages to other banking sectors increased cross border linkages tend to improve that system s stability
56. ing them to overcome some of the problems with the other techniques discussed above However they rely on numerous assumptions about the structure of the economy increasing the likelihood of misspecification errors Overall Assessment 47 While combining available tools to estimate the likelihood of a crisis can be valuable to policymakers these tools taken individually are subject to important limitations The ability of asset price based models to accurately estimate crisis probability declines precipitously with time Structural models overcome these limitations but at the cost of misspecification errors which are also pervasive in reduced form statistical techniques Therefore DSGE and Crisis Prediction models can be applied to cross check whether contemporaneous increases in crisis probability emanating from financial market data are corroborated by longer term measures of risk build up Conversely authorities should use models based on high frequency asset price data to monitor the intensification of pressures if structural or econometric models have indicated in the past the increased probability of stress 21 IV SAMPLE COUNTRY CASE STUDY 48 This section aims to provide a concrete illustration of the use of the systemic risk monitoring toolkit in a fictitious country case Figure 4 The diagrammatic presentation of the key questions from the previous section provides a practical guide to policy makers in the form of a systemic
57. ing time series indicators of financial and real activity This complex model may not be overly user friendly but it captures the dynamic responses of systemic risk indicators to structural shocks and may provide useful early warnings of systemic events Moreover DSGE models provide an in depth understanding of the interactions and shock transmission across sectors and with the broader economy including by capturing inter sectoral and macroeconomic dynamics e g cyclical fluctuations However these models are particularly difficult to calibrate and interpret Overall assessment 36 Overall the available toolkit addresses several key inter sectoral linkages and related risk buildup However further efforts are needed to combine these approaches into integrated economy wide measures of systemic risk In particular there is a need to incorporate feedback and second round effects across sectors in order to fully capture sectoral risk transfers and enhance the spillover analysis One example is the gap in stress tests on links between financial sector stress and credit supply conditions the impact of these conditions on the real economy and feedback effects on financial sector stress 18 E Cross Border Linkages What are the Amplification Channels through Cross Border Spillovers 37 Encouraged FSIs related to geographical distribution of loans and foreign currency denominated liabilities are a starting point for the analysis of cross
58. int in time In a first step four main sectors are identified namely the government sector including the central bank the financial sector mainly the banking system the nonfinancial sector corporations and households and the external sector nonresidents For each sector a breakdown of claims and liabilities by currency and maturity allows to examine four general types of risks i e maturity risk currency risk credit risk and solvency risk Sectoral balance sheets provide valuable information on potential sources of risk that may be masked by the netting off of exposures under consolidated country balance sheet data Example of Tool Use The BSA framework has been applied to Croatia to examine how sectoral interlinkages have shifted with the surge in external debt to near 86 percent of GDP in 2006 A comparison of the banking sector s balance sheet position at end 2000 Table 1 and at end 2005 Table 2 reveals a significant increase in net exposures to the central bank and the government sector following a hike in reserve requirements and increased public bond issuance More significantly it reflects a sharp increase in external borrowing with a near threefold rise of foreign liabilities Overall the nonresident sector increased its exposure to Croatia by 100 billion kuna or 37 4 percent of GDP over the period Table 1 Croatia Net Intersectoral Asset and Liability Positions in millions of Kuna December 2000 Debtor
59. ion channels Market risk There is greater familiarity of financial institutions and supervisory authorities with assessing such risks including through stress testing for interest rate exchange rate or asset price shocks At the systemic level aggregate measures of market volatility can be used to assess latent vulnerabilities e g to identify periods in which markets are more likely to become more volatile 9 Underlying methodology Depending on country specific circumstances various types of tools and underlying approaches or methodologies are available Single risk soundness indicators Indicators based on balance sheet data such as financial soundness indicators FSIs are widely available and cover many risk dimensions However they tend to be backward looking and do not account for probabilities of default or correlation structures Moreover only some of these indicators can be used as early warning tools e g indicators of funding structures Market data can be used to construct complementary indicators for higher frequency risk monitoring Figure 2 Buildup of Systemic Risk Sources and Channels Sources of aggregate crisis prediction shock models Crossbord er interconne ctedness FSI other sectors BSA Note FSI stands for Financial Soundness Indicators T model Threshold Model DSA Debt Sustainability Analysis CCA Contingent Claims Analysis BSA Balance Sheet Approach
60. ion on potential spillovers through direct exposures But they do not offer information on how the system might behave during crises when both direct and indirect e g common exposures come into play 19 Complementing the above analyses or replacing them in the absence of data on direct exposures models based on market data allow for high frequency monitoring of the likelihood of spillovers between financial institutions and systemic stress within a short term horizon typically less than a year i e near or during crises They include Joint Distress Indicators JDI Financial Institutions Stability Index FISD Volatility Spillovers Diebold Yilmaz DY CoVaR Distress Spillovers DS Systemic CCA SCCA These models and indicators can be used to assess spillovers either under normal DY or extreme conditions JDI CoVaR DS Systemic CCA Moreover the Systemic Liquidity Risk Indicator SLRI provides a coincident indicator of systemic liquidity shortages during market distress These models do not trace back to the specific risk channels through which such spillovers occur but some of them help identify which institutions are more systemically important by estimating individual contributions to systemic stress Overall assessment 20 Overall when the available toolkit is applied to banks it addresses the above questions well For example the complementary tools provide rough rules of thumb on when to worry about build up of ri
61. ions are estimated via the Linear Ratio of Spacings LRS method Secondly the multivariate dependence structure of joint tail risk of potential losses is derived nonparametrically Finally after estimation of the marginal distributions and the dependence structure the following point estimates of joint potential losses at quantile q 1 a at any point time t is derived ze In 2 75 Rapt Gag j t 6 6 m 1 A w with location parameter u shape parameter and multivariate dependence structure of joint tail risk of potential losses A w Example Country Financial institutions 36 financial institutions covering bank holding companies other banks major broker dealers GSEs and insurance groups The chart shows total expected losses and government contingent liabilities Both are highest between the periods just after Lehman s collapse in September 2008 and the end of July 2009 Lehman Brothers Collapse i oo E E El ndi o2 S o Tt r OO cO OO OO cO OO oo co cO C ego c e e 9955505595988255859g55989982559595528827 2675539757255 205253 RBBSESSBSSEBEYVPR SEES om m LOr ESTR 4O02 524 223 5202355 Total expected losses sum of individual put options Total contingent liabilities sum of individual alpha value put option Total contingent liabilities GEV 5Oth percentile without government agencies Total contingent liabilities GEV 5Oth percentile with go
62. ions or sovereigns Risk transmission channels Models measuring interactions among financial entities have evolved rapidly in recent years They are designed to better capture time varying and nonlinear distress dependences e g during extreme events or the marginal contributions of individual institutions to systemic risk e The whole financial system and the economy Crisis prediction and stress test models aim to capture the risk that the entire financial system is impaired as well as macro financial linkages and feedback effects with the real economy Also general equilibrium models increasingly integrate financial sector and macroeconomic variables 8 Types of risk What are the most relevant types of risk that should be monitored and mitigated during each systemic risk phase e Credit risk This is a key source of risk in most financial systems Stress testing methodologies in particular have relied on increasingly sophisticated approaches to assess probabilities of default and potential losses if default were to occur loss given default or LGD especially in relation to various macro factors Liquidity risk Liquidity risk measurement tools have recently been developed to assess not only potential changes to financial institutions liquidity ratios but also the interactions between market liquidity e g for thinly traded illiquid assets and financial institutions funding conditions e g through collateralizat
63. is the probability that at least one FI becomes distressed given that X has become distressed This measure reflects X s systemic importance in the banking system PCE P X P R X P YNR X respectively The FISI is defined as FISI captures X s exposure to bank Y s distress P x Bx Example This analysis has been applied to estimate the stability of a set of six Swedish banks using daily CDS spreads over January 2007 October 2010 Figure 1 graphs the evolution of FISI over time Table 1 shows the PCE conditional on column i FI defaulting computed on a pre crisis date and at the event of collapse of Lehman Brothers Table 2 shows the distress dependence matrix 1 e the conditional probability of row i s FI defaulting given column j s default also computed on September 15 2008 3 Ei npn Table 1 Probability of Cascade Effects PCE from default of an FI igure 1 Sweden Financial Institutions Handels Swed Stability Index FISI SEB banken bank Nordea DnB Nor Danske 1 1 2007 0 32 0 29 0 21 0 37 0 21 0 21 9 15 2008 0 85 0 87 0 69 0 87 0 79 0 74 Table 2 Distress Dependence Matrix DiDe Handels Swed RE banken bank Nordea DnB Nor Fara o ao osef oaf oa Swed bank 0 64 0 58 0 59 0 50 0 50 Nordea 0 48 0 61 0 37 0 49 0 43 DnB Nor 0 34 0 42 0 26 0 40 0 37 1 1 2007 1 1 2008 1 1 2009 1 1 2010 041 0 44 Source IMF Staff estimates 34 III RETURNS SPILLOVERS The spillover measure
64. isk from a historical peer group average Each indicator x is standardized to zi and mapped into a cumulative normal distribution ranging from 0 to 10 Second a fiscal stress index is computed on the basis of a number of fiscal indicators exceeding endogenous thresholds that minimize noise to signal ratios of future fiscal crises weighted by their relative signaling power Example of Tool Use The table below shows the fiscal vulnerability index the fiscal stress index and a combined risk score computed as a simple average of the two indices for three clusters of fiscal variables including basic fiscal variables long term fiscal needs and asset and liability indicators The results are calculated in the Fall of 2010 for a sample of G 20 countries and Greece Ireland Portugal and Spain GIPS and are displayed separately for advanced and emerging economies Index values close to 10 indicate high levels of vulnerability while values close to 5 signal a normal degree of vulnerability Basic Fiscal Long term Fiscal Asset and Liability Variables Trends Management Overall Score Fiscal Vulnerability G20 Advanced plus GIPS 7 7 6 0 6 0 6 5 G20 Emerging Economies 6 9 5 0 6 0 6 0 Fiscal Stress G20 Advanced plus GIPS 7 2 7 3 8 3 7 4 G20 Emerging Economies 7 1 4 8 3 9 5 0 Aggregate Score G20 Advanced plus GIPS 7 5 6 6 7 2 7 0 G20 Emerging Economies 7 0 4 9 5 0 5 5 Source Baldacci and others 2011 44 VIII SOVEREIGN FUNDING
65. m of a heat map based on degrees of overvaluation or ii inputs into a model such as the T model that derives crisis signals based on a benchmark country distribution 22 More generally asset price growth features prominently as an early warning signal in Crisis Prediction Models Sustained equity price inflation or house price acceleration may reflect financial imbalances building up over time and when combined with a sharp increase in credit to GDP gap and banking sector leverage may flag a looming domestic banking crisis Credit to GDP Based Crisis Prediction Model 23 However early warning signals from asset price models are not good predictors of the timing of asset price corrections Parameters in these models are also less reliable during periods of financial stress because such parameters are derived implicitly or explicitly from fundamental based equilibrium values based on arbitrage free asset price models When such assumptions on free arbitrage do not hold as in periods of financial stress the estimated equilibrium values become less reliable 24 In addition asset price models may also help monitor the initial economic impact of a potential market correction VAR models for example can be used to estimate the response of a set of macroeconomic variables e g real GDP consumption investment or inflation to house price shocks taking into account household leverage and risk sharing provisions in mortgage contracts 1
66. napshot Attributes Description Coverage Sectors Institutions Aggregate of banking system level data from BIS Types of risk Interconnectedness and contribution of each institution to systemic spillover risk Interpretation Main output The impact on regulatory capital of one banking system from failure or funding difficulties of another system Other outputs The fraction of one banking system s spillover contribution to all possible spillovers of all other institutions contribution to systemic risk contagion path of bank failures The fraction of all possible spillovers received by an institution from others vulnerability to systemic risk Thresholds Not available Time horizon Good for assessing spillover risk and potential contribution of each institution to systemic risk Data requirements Cross border exposure data from BIS and regulatory capital from http fsi imf org Reference Espinosa Vega and Sole 2010 Users IMF 2011a 29 Methodology The data consists of a matrix of bilateral banking system exposures from the BIS Table 9B The matrix is complemented by data on regulatory capital of the countries banking sectors Using the methodology in Espinosa Vega and Sole 2010 we can then trace the network spillovers resulting from hypothetical credit and funding events to specific banking systems In particular two sets of simulations are done First is a simulation of a banking system becoming in
67. ncial institutions potential fire sales of financial assets as well as crossborder exposures and the related adverse feedback loops Figure 3 Measurement challenges During the recent global financial crisis various shock transmission channels reached an unprecedented level of complexity For example the range of potential shock transmission channels has broadened considerably reflecting the greater integration between financial institutions and markets countries and real sectors e g linkages between public and financial household or corporate and financial public and external As a result macro financial linkages and systemic risk are more difficult to measure given the potential for more complex and unpredictable scenarios greater scope for nonlinear impacts e g through illiquid markets or institutions and more unstable correlation structures and behavioral relationships B Key Features of the Toolkit 7 Focusing on risks at various levels Available tools may be used to measure systemic risk at different levels of aggregation including Individual financial institutions and markets For instance these include i market valuation tools to identify price deviations from trend or from levels implied by fundamentals focusing on assets that are relevant to financial stability e g housing equity or bond markets 11 indicators of risk taking and stress testing tools to assess the resilience of financial institut
68. ng the fraction of missed crisis relative to total crisis and the fraction of false signals relative to potential signals respectively The noise to signal ratio captures the trade off between the two types of errors Change in credit to GDP increases 2 standard deviations above its historical mean in a given country Good for predicting medium term materialization of financial system wide stress 1 to 5 years ahead Low frequency macroeconomic and financial balance sheet time series requires a measure of materialization of stress in the system Laeven and Valencia 2010 Main Borio and Drehmann 2009 Users IMF 2011b 65 Methodology The Noise to Signal Ratio NSR approach intends to select a set of macroeconomic and balance sheet variables as well as their respective thresholds to form early warning indicators EWT of potential crisis The methodology is implemented in five steps I define a crisis indicator a binary variable that assumes the value of 1 when a crisis occurs and 0 otherwise II select one or more variables that can potentially forecast crisis IIT calibrate various potential thresholds for each one of those variables IV compute a binary variable called crisis signal which assumes the value of 1 when a certain number or all of the forecasters move beyond their corresponding thresholds and zero otherwise A failure to signal a crisis that actually happens produces a Type I error whereas a false signal a s
69. nsions The paper offers suggestions on how to use the tools taking into account their nature focus and relative merits and limitations It also focuses on the systemic risk signals including their timeliness the types of risks they cover and ways of interpreting them However this paper does not analyze the direct relevance of specific systemic risk measures for the selection of appropriate macro prudential policy tools and their calibration 3 This paper offers guidance on how to select the best set of available tools under various circumstances Effective risk monitoring should be based on a clear understanding that 1 policymakers should not expect to find all in one tools because the reliability of systemic risk monitoring tools depends on the circumstances in which they are used and ii policymakers should take into account several potential sources of risk by using a range of tools at any point in time Against this background the objective of this paper is to identify those tools or combinations of tools that are most effective in measuring a specific dimension of systemic risk It provides policymakers with some general principles based on cross country analyses but it also encourages practitioners to calibrate the toolbox in view of country specific circumstances 4 The structure of this guide follows a practical approach After a brief introduction to systemic risk and the key features of the existing toolkit the guid
70. ntagion risks F Crisis Risks What is the Probability of a Systemic Crisis 43 Several tools extract information from asset prices to estimate the probability of a crisis occurring within a certain time interval Specifically the Systemic CCA and JDI can be directly applied to estimate the probability that a certain number of institutions will jointly fail in the near term thereby triggering financial instability The systemic CCA can also indicate the probability that the aggregate losses of the financial system will be above a certain specified amount Alternatively the Regime Switching Model estimates the probability that financial markets will enter into a state of high volatility or crisis Finally the SLRI model can be used to assess the probability of systemic liquidity pressures in capital markets 44 However while the above tools relying primarily on asset price data generally signal crisis events with a relatively high degree of confidence they offer only limited lead time e g a month or at most a year This may not be sufficient from a policymaker s perspective In addition they are subject to increased error risks when markets incorrectly price risks for example in the case of illiquid markets 45 In order to obtain measures of crisis probability with longer lead time policymakers should also rely on techniques that combine information on aggregate credit growth with other macroeconomic or balance sheet indicato
71. o fundamentals past valuation ratios and cross country distributions in the form of a heat map or a wheel Thresholds Overvaluation is deemed significant when it exceeds 10 percent of fundamentally implied prices one standard deviation above its historical mean or lies in the top third quartile of the cross country benchmark distribution Time horizon Contemporaneous measure of vulnerability Data requirements Quarterly monthly macroeconomic market based and balance sheet data from OECD WEO IFS Haver Analytics Knight Frank LLP Global Property Guide Consensus Forecast and Bloomberg Reference IMF FSB 2010 47 Methodology The real estate vulnerability index is constructed by combining four indicators that capture the extent of price misalignment the stress in household balance sheets the exposure to market risk from mortgage contract provisions and the impact of an asset price correction on real economic activity The corporate vulnerability index is a normalized weighted sum of leverage liquidity and profitability indicators capturing the probability of corporate distress multiplied by its potential macroeconomic impact proxied by the market capitalization of listed companies The degree of equity price misalignment is assessed using a historical time series on price to book and price to expected earnings ratios the dividend based Gordon valuation model and the following arbitrage pricing model Ri a PFs t J where R
72. of short term government debt by residual maturity B and iii their sale of ong term government debt by residual maturity y The holdings of foreign official investors are assumed to stay constant over the next year Based on this framework three scenarios are considered i foreign private investors provide no new net financing 0 0 B 100 y 0 ii foreign private investors provide no new gross financing a y 0 and iii foreign private investors provide no new gross financing and sell off 30 percent of their remaining holdings o f 0 y 30 This is the most severe scenario and is intended to replicate the average experience of Greece Ireland and Portugal during the worst part of their sovereign debt crisis A large increase in banking sector asset under the shock scenarios implies that 1 sovereign banking linkages may grow substantially with adverse effects for domestic financial stability and growth prospects due to crowding out and or ii domestic bank may face difficulty absorbing the sovereign funding needs and as a result sovereign bond yields may rise Example The results of the FSS can be analyzed through both time series and cross sectional presentations e Time series The first chart below shows the level of sovereign debt held by domestic banks before and after the shock scenarios for Belgium e Cross section The second chart shows how Belgium compares to other advanced economies in terms
73. of their vulnerability to the same shocks Belgium Sovereign Funding Shock Scenarios percent of banking sector assets Selected Euro Area Countries Bank Holdings of Own 14 Government Debt under Shock Scenarios end 2012 12 O netfinancing 7 5 0 grossfinancing 9 4 82011 30 sale 0 net financing 00 gross financing 30 sale 20 15 percent o Historical Path Hey 0 net financing Sa 0 grossfinancing 30 sale of banking sector assets 46 IX ASSET PRICE MODELS Asset price models detect signs of asset price misalignment by identifying fundamental demand and supply disequilibria using macroeconomic financial and balance sheet data They are used to assess the likelihood of market price corrections and their potential impact on the real economy Tool Snapshot Attributes Description Coverage Sectors Institutions Key asset markets and country risk impact on GDP growth from price corrections households corporates and financial institutions are covered Types of risk Market risk from shocks to residential and commercial real estate prices spikes in corporate bond spreads and equity price declines Interpretation Main output A country level index of vulnerability to downward asset price corrections Other outputs Estimated impact on GDP from a house price correction Co movement between corporate bond spreads and macrofinancial determinants Degree of overvaluation relative t
74. on to systemic stress Overall spillover index for the period is 57 or 0 57 expressed as a fraction This matrix could be repeated for windows of data to get a rolling sample in which case a time series of the DY index can be derived A more generalized spillover definition is provided in Diebold and Yilmaz 2012 36 IV DISTRESS SPILLOVERS This is an indicator of outward spillovers of institutions or markets during extreme times the potential contribution of one institution to systemic risk during crisis The indicator uses market data on returns based on either CDS spreads or equity prices to estimate extreme contributions and is easy to use update It had reasonable predictions for the interconnectedness among 25 largest banking groups in the world with pre crisis data that proved to be true during the 2007 2009 crisis It does not identify the exact spillover channels only those between institutions Tool Snapshot Attributes Description Coverage Sectors Institutions All financial institutions with market based high frequency data on various returns Ex CDS equity prices distance to default Types of risk Interconnectedness and contribution of each institution to systemic spillover risk Interpretation Main output The fraction of one institution s spillover contribution to all possible spillovers of all other institutions distress dependence among institutions Other outputs The fraction of all possible spill
75. orate and household counterparties FSIs include both aggregated individual financial institution data and indicators that are representative of the markets in which the financial institutions operate They are easy to use and update but do not measure precisely the likelihood of or resilience against future shocks Tool Snapshot Attributes Description Coverage Sectors Institutions Deposit takers other financial corporations nonfinancial corporate sector and households Types of risk Credit risk market risk and liquidity risk Interpretation Main output Capital adequacy asset quality earnings and profitability liquidity and sensitivity to market risk of the banking sector core FSIs Other outputs Soundness condition for nonbanking financial sectors nonfinancial corporate sectors and households as well as asset prices encouraged FSIs Thresholds No Time horizon Low frequency backward looking indicators Data requirements Aggregate data on balance sheet and P L of banking sector nonbanking financial corporations and nonfinancial sectors indebtedness of households data for market liquidity and asset prices Reference IMF 2006 Users Sun 2011 Methodology FSIs consist of two sets of indicators a core set of FSIs and an encouraged set of FSIs The core set of FSIs covers banking sector reflecting the central role of the banking sector in many financial systems The encouraged set of FSIs covers additional FSIs for the banking
76. ould reflect the typical phases of systemic risk o The slow buildup of risk e g through combinations of balance sheet and slow moving indicators o The identification of weak points and potential adverse shocks e g stress tests to detect weak financial institutions asset price deviation from fundamentals o The fast unfolding of crises including through amplification mechanisms e g high frequency market based spillover measures Longstanding data gaps remain an obstacle to assessing key systemic risk components including interlinkages and common exposures which is increasingly problematic in light of the growing complexity of financial crises 52 However from the perspective of guiding macroprudential policy the systemic risk monitoring toolkit is incomplete The systemic risk monitoring framework is work in progress in a number of key dimensions Tools exist to assess most sectors and levels of aggregation but they provide only partial coverage of potential risks and only tentative signals on the likelihood and impact of systemic risk events As such they may not provide sufficient comfort to policymakers Indeed a number of practical and theoretical roadblocks remain that currently limit our capacity to measure systemic risk in comprehensive and accurate ways 53 26 Early warning The forward looking properties of systemic risk measures are generally weak even though some measures appear relatively promising
77. ource IMF 2010 48 X BALANCE SHEET APPROACH This analytical framework examines the balance sheet of an economy s major sectors identifies maturity currency and capital structure mismatches and drills down intersectoral linkages across domestic sectors and to the rest of the world This is a useful tool for analyzing the resilience of the main economic sectors to specific financial shocks and the transmission of shocks across sectors Attributes Tool Snapshot Description Coverage Sectors Institutions Types of risk Interpretation Main output Other outputs Thresholds Time horizon Data requirements Reference Public sector financial sector corporate sector household sector and nonresident sector Counterparty risk exchange rate risk liquidity risk and solvency risk A balance sheet matrix displaying each sector s claims on other sectors and liabilities to other sectors Sensitivity tests to specific financial shocks N A Provides a coincident measure of interconnectedness Suitable for conducting a stress testing analysis Low frequency data from national sources Stock variables for the nonfinancial private sector are typically derived from the balance sheet positions of other sectors Allen Rosenberg Keller and others 2002 49 Methodology The BSA views the economy as a system of sectoral balance sheets displaying book valued stocks of assets and liabilities at a specific po
78. output An index variable that moves down when global liquidity dries out Other outputs Exposure of individual banks to the SLRI measured as betas on the mean and volatility of banks equity returns It can also be used to calculate a premium to be paid by banks as a compensation for the implicit liquidity support obtained from public authorities Thresholds There are no specific thresholds The index is normalized to have 0 mean and unit standard deviation Usually values above 2 indicate important liquidity shortages Time horizon Good for predicting very short term tail shocks to financial institutions No medium or long term predictive power Data requirements High frequency daily or weekly data on asset prices Reference Severo 2012 Users IMF 2011d 57 Methodology The computation of the SLRI requires time series data on various arbitrage or quasi arbitrage relationships in asset markets For example one can use data on violations of covered interest parity for many pairs of currencies the CDS Bond basis for corporate and sovereign bonds the U S Treasury s on the run off the run spread OOS etc Collect this information on a matrix X and use Principal Components Analysis PCA to extract orthogonal factors which are ranked according to their ability to explain the variation in the data The first factor is the SLRI provided it explains a significant portion of the variability in the data One can test whether banks or other financ
79. overs received by an institution from others vulnerability to systemic risk Thresholds Not available Time horizon Good for assessing spillover risk and potential contribution of each institution to systemic risk during stress Data requirements High frequency market based financial time series flexible series of returns but limited to institutions with market data Reference Main Gropp Lo Duca and Vesala 2009 Users Chan Lau and others 2012 37 Methodology It first identifies all extreme events in the data usually comprising weekly or daily returns on equities CDS spreads or market value of assets by looking at the 1 or the 5 percentile of the joint distribution of returns All returns lying in the left tail that is the ones below the thresholds are called exceedances Then distress dependence is estimated by using a logit model to account for the fatness of the tails of the distribution of exceedances In particular the probability of an exceedance is estimated conditional on exceedances in other financial institutions or centers after controlling for common shocks such as extreme conditions in the world equity markets the country s stock markets and real sector indicators The distress dependence matrices are largely static the sample periods are fairly long The analyses could also be extended to make it more time varying by repeating the exercise over a rolling window albeit one that is sufficiently
80. r Advanced Economies 2011 October 74 XXIII DSGE MODEL DSGE models can trace movements of numerous macroeconomic and financial variables in response to alternative sources of shocks A calibrated model can be used to analyze the macro financial effects of various macroprudential policy instruments like countercyclical capital buffers and loan to value ratio caps It only covers procyclicality and not interconnectedness and requires considerable experience to run Tool Snapshot Attributes Description Coverage Sectors Institutions Banks nonfinancial corporations households Types of risk Procyclicality stemming from shocks related to real estate prices lax lending standards productivity Interpretation Main output The macro financial impact of various shocks with and without macroprudential policies Other outputs Leading indicators of future financial instability Thresholds Not available Time horizon Flexible Data requirements Various depending upon calibration requirements Reference Benes and others 2010 Users IMF 2011b 75 Methodology The model embeds a banking sector along with a new Keynesian model of the real sector Key features of the banking sector include the strong role of the balance sheets of both banks and nonfinancial borrowers in the propagation of shocks and a link between the diversifiable or idiosyncratic risk faced by banks in their lending activities and the nondiversifiable aggregate ma
81. re useful for monitoring purposes in general Buildup phase Systemic risk builds up over time and this could reflect several underlying reasons The financial system may have high exposure to an overheating sector or be subject to increased risk taking e g due to competition for market share or lax supervision including through financial innovation The risk buildup could also be related to growing cross border exposures and funding sources During this phase systemic risk measures could focus on assessing the likelihood of a systemic crisis Figure 2 taking into account the evolving balance between potential financial losses and existing buffers designed to absorb these losses Shock materialization At that point the crisis is about to start Mounting imbalances or excessive risk taking make the financial system fragile and susceptible to exogenous shocks e g GDP or fiscal shocks exchange rate or housing price shock failure of a systemically important financial institution Therefore systemic risk measurement could focus primarily on assessing potential losses in both the financial system and the real sector Amplification and propagation In most crises shocks affect the broader system including financial institutions markets and other sectors and potentially other countries financial systems At that point systemic risk measurement could focus on amplification mechanisms such as interconnections between fina
82. rg external pubs ft survey so 2012 POL01 1312A htm and country FSAP document site http www imf org external NP fsap fsap aspx 68 XX GDP AT RISK The Systemic Risk Monitoring System DNL SRMS forecasts systemic real and financial risks using time series of indicators of financial and real activity It is complex to use update but has good out of sample forecasting properties for systemic stress Attributes Tool Snapshot Description Coverage Sectors Institutions Types of risk Interpretation Main output Other outputs Thresholds Time horizon Data requirements Reference Equity markets data by sector Systemic real risk is defined as the worst predicted realization of quarterly growth in real GDP at 5 percent probability Systemic financial risk is defined as the worst predicted realization at 5 percent probability of the market adjusted equity return of a large portfolio of financial firms Forecasts of indicators of systemic real risk and systemic financial risk based on the predicted density distribution of the underlying indicators Systemic risk fan charts to summarize systemic real and financial risk prospects Yes Good for predicting near term materialization of financial system wide stress A large set of quarterly time series of indicators of financial and real activity for each country including equity markets data financial monetary and banking variables related to credit condition
83. rs In particular the Crisis Prediction Model yields direct measures of the probability of a financial crisis associated with excessive credit growth or private sector leverage among other variables And the T model can signal the increased likelihood of crisis materialization without providing numeric estimates for the probability of such events However these techniques are subject to the typical limitations associated with reduced form econometric models and As noted the application of various tools to estimate the likelihood of crises requires defining ex ante what constitutes a crisis For instance bank regulators and supervisors may be interested in assessing the probability that a certain number of banks will fail at the same time or that their joint losses will be above a certain threshold Investors may be more concerned about the probability that sovereign debt or real estate prices will fall below a certain level instead 20 may in particular under estimate crisis probabilities relative to the actual occurrence of systemic events 46 DSGE models combine a broad range of variables including output consumption or asset prices and provide an in depth understanding of macro financial linkages and how these could behave under stressed conditions or in reaction to particular policy actions In addition the specification and estimation of these models may not depend on high frequency information contained in asset prices allow
84. s and price and real variables De Nicol and Lucchetta 2010 69 Methodology The DNL SRMS is a set of forecasting models estimated in real time based on developments of the methodology introduced in De Nicolo and Lucchetta 2010 The DNL SRMS is currently implemented using large sets of quarterly time series of indicators of financial and real activity with data starting in 1980Q1 for 22 advanced economies It delivers at a country level Forecasts of indicators of systemic real risk and systemic financial risk as well as forecasts of the distribution of GDP growth and an indicator of financial stress Absolute and relative risk ratings of forecasts of systemic real and financial risks Tail risk relative ratings of forecasts of indicators of key economic conditions Systemic real risk is measured by GDP at Risk GDPaR defined as the worst predicted realization of quarterly growth in real GDP at 5 percent probability Systemic financial risk is measured by an indicator of Financial System at Risk FSaR defined as the worst predicted realization at 5 percent probability of the market adjusted equity return of a large portfolio of financial firms Forecasting of GDPaR and FSaR indicators is accomplished in three steps First a large set of quarterly financial and macroeconomic variables is modeled as a multivariate dynamic factor model Estimated time series of factors summarize the joint dynamics of the series and are used as pr
85. s Prediction Model can be used to estimate relatively well the risk of systemic banking crises about two to three years in advance section F 34 A number of tools help assess more deeply the risks arising from linkages across sectors including indirectly through second round effects For instance Asset Price Models can also help measure the vulnerability of the household and corporate sectors to asset price corrections as well as the broader spillover effects on GDP section B although they do not take into account feedback loops through the impact of lower growth on asset price levels As noted above Debt Sustainability Analysis DSA also examines the impact of real economy market and financial system shocks on sovereign risk Section C and can be combined with the Systemic Contingent Claims Approach SCCA to obtain complementary and more forward looking estimates of these impacts sections A and C Macro Stress Tests assess the impact of a wide range of risks and adverse scenarios on financial institutions individually or in aggregate Importantly however feedback effects on the economy including through credit supply conditions are not appropriately covered in stress test models at this point 35 Beyond sector specific linkages some tools combine cross sectoral interdependences to assess spillovers of systemic economy wide relevance In particular the GDP at Risk model forecasts systemic real and financial sector tail risks us
86. s of house prices and equity prices to detect signs of overheating in asset markets The indicators are calculated for many countries putting Country X s situation in a cross country perspective Together with Panel A it seems that financial sector difficulties could be increasing as of end 2007 e Panel C There are clear signals that fiscal risks are increasing especially from financial sector related contingent liabilities This panel assesses sovereign bank linkages through public contingent liabilities Debt Sustainability Analysis and potential changes in banks holdings of sovereign debt under stress scenarios Debt Sustainability Analysis shows that the debt GDP could rise substantially should contingent liabilities materialize Such liabilities could be related to the financial sector The Sovereign Funding Shock Scenarios FSS show that under certain scenarios bank holdings of public debt may increase sharply leading to stronger sovereign bank linkages in the country 22 e Panel D There is limited evidence that financial sector shocks are spilling over into the real sector at this stage although spillover risk within financial institutions is slowly rising This panel focuses on risk amplification across sectors and the economy GDP at Risk and Financial Stability at Risk and adverse feedback loops between contingent public liabilities and banking sector distress Systemic Contingent Claims Analysis The Financial Stability a
87. scal consolidation and increased credit risk Complementing this approach Distress Dependence Model can also use high frequency market data to measure the probability of 16 distress of a financial institution or financial system conditional on sovereign distress The sovereign Funding Shock Scenarios FSS can be used along with DSA to do forward looking analysis to assess sovereign s vulnerability to sudden investor funding outflows and banks potential exposure to sovereign debt 30 In addition to the above some tools can help monitor the potential for negative feedback between financial sector risks and sovereign risk For example there may be concerns that the government balance sheet may not be strong enough to meet contingent liabilities reflecting the existence of explicit or implicit public guarantees leading to increased systemic risk The Systemic CCA allows gauging the impact of such negative feedback effects between sovereign risk and systemic risk Overall assessment 31 Overall the available tools allow for in depth assessments of the linkages between sovereign risk and systemic risk as they cover most risk dimensions financial institutions time horizons and country categories as well as the impact of shocks and their likelihood However they do not provide clear signals as to whether sovereign risk buildup has reached a critical level that threatens financial stability or whether it may unleash perverse dynami
88. ses highlight key sources of systemic risk the evolution of these risks over time and the underlying macro financial linkages Definition There is an evolving literature on systemic risk measurement covering a wide range of approaches In the context of this paper systemic risk is defined as risk that originates within or spreads through the financial sector e g due to insufficient solvency or liquidity buffers in financial institutions with the potential for severe adverse effects on financial intermediation and real output The objective of macroprudential policy is therefore to limit system wide financial risk IMF 201 1a by enabling policymakers to know better when to sound the alarm and implement policy responses Phases Past crisis episodes show that different sources of risk and shock transmission channels can emerge at the same time or in complex sequences including through multiple feedback effects However from an analytical perspective it may be useful to distinguish between key phases in which crisis related events unfold At the same time policymakers should be cognizant of macro financial linkages during each phase Ultimately most systemic crises involve feedback effects between the real economy and the financial sector including across countries Theoretical and empirical models dealing with interactions between the financial sector and the real economy as well as between cross border transmission channels a
89. sks in the financial sector The toolkit identifies the institutions the weak links that are vulnerable to adverse shocks in the system and market based indicators serve as good near term indicators of crisis and spillover risks between them However many of the above tools apply primarily to bank balance sheets and 14 interlinkages while as demonstrated by the current crisis a range of financial institutions including recently developed institutions such as Central Counterparties may also be systemically relevant requiring a broadened focus of the toolkit and methodologies Persistent data gaps also hinder analytical efforts to assess nonbank financial institutions Overall the combination of tools covers the impact of shocks better than their ikelihood While significant progress has been achieved more work is needed to provide firmer guidance for policymakers on risk buildup and on the design and calibration of adverse stress testing scenarios B Asset Prices Are Asset Prices Growing Too Fast 21 Asset Price Models estimate the deviation of an asset market value from its long term model based equilibrium which constitutes a measure of potential for an asset price correction the assumption being that the larger the misalignment of market prices from fundamental values the higher the probability of a price correction The Real estate market model for instance provides both i direct signals that can be presented in the for
90. solvent and being unable to repay interbank loans in others Second is a simulation of a banking system becoming insolvent not repaying loans in others and unable to rollover funding from others Example This method was applied in the context of the Spillover Report for Japan IMF 20112 Besides Japan the countries included in the analysis were Australia Austria Belgium Canada France Germany Ireland Italy Japan Netherlands Portugal Spain Sweden Switzerland United Kingdom United States China Taiwan India Indonesia Malaysia Philippines South Korea Thailand and Vietnam The table shows the impact on Japan on others if another banking system Japan fails The shocks considered are solvency shock with loss given default as 1 and a combination of a solvency and funding shock with the borrowing institution not being able to replace 0 35 fraction of its funding from a defaulting institution Capital impairment in percent of pre shock capital Impact on Japan if trigger Impact on others if country defaults Japan Credit amp Credit amp Credit shockFunding Credit Funding Trigger country 1 shock 2 Affected countries shock 1 shock 2 Australia 4 4 4 5 Australia 2 2 8 8 France 6 2 72 3 France 10 8 13 8 Germany 7 2 72 3 Germany 2 6 7 2 Ireland 3 1 72 3 Ireland 10 5 11 7 Italy 0 5 72 3 Italy 0 2 0 5 Portugal 0 0 0 0 Portugal 0 0 0 3 Spain 0 6 0 7 Spain 0 8 1 5 UK 57 6 72 3 UK 25
91. ss border spillovers What is the probability of a systemic crisis In addressing each question the emphasis is put on combinations of relevant tools in light of their relative merits and complementarities e A living inventory Tools Binder that offers a two page snapshot of each tool summarizing its key properties methodology coverage interpretation data requirements etc and providing a concrete example of its use A sample systemic risk Dashboard for a fictitious advanced country that illustrates how in a specific country context various complementary tools can be combined to monitor key sources of systemic risk Tool selection tables that summarize which tools are available for which purpose and country category thereby helping users to readily identify the most relevant tools Figure 1 Structure of the Guide Key Questions Tools Binder Policymakers should askthemselves as they assess systemicrisk Financial Finance Institutions Institutions A E in Table Asset Prices Sovereign uk Dashboard Broader Economy PN s V Cross Border ac ED Linkages 5 in Crisis Risks ESEK 5 Finally the paper concludes by highlighting how well the various dimensions of systemic risk are covered by the current toolkit and by identifying some key analytical gaps that could benefit from future research II APPROACHING SYSTEMIC RISK A What is Systemic Risk 6 Lessons from past and current cri
92. ss loans less ratio of provisions to gross loans return on average assets liquid assets to customer deposits and short term funding and tangible common equity to tangible assets Each financial ratio is normalized around the system wide or all sample banks mean and standard deviation over the three years to time t The sum of the five standardized financial ratios is the BHI relative to its peers The BHI for a country s banking system compared to a global peer group would give the relative health of a banking system Example The BHI for Spanish banks using end 2011 data was derived as a first pass analysis of their relative soundness The heat map subsequently generated using HEAT shows the differentiation in the soundness across banks within the Spanish system as well as the evolution of the financial health of the institutions over time see table The heat map shows the system wide distress in 2008 especially concentrated at the mid sized to smaller banks An overall system wide indicator can be derived by averaging the asset weighted BHIs Spain Heatmap of BHI for Selected Banks Institution Cajas de Ahorros Total Assets Overall Bank Health In millions of euro Pre Restructuring Post Restructuring Latest 2006 2007 2008 2009 2010 2011 1 Santander S A 1 292 677 oss oss GS EE M NUN 2 BBVA 622 369 EE BE o Ga M eee 3 BFA Bankia 321 188 448 E Caja Madrid Bancaja Caiza Laietana Caja Insular Caja de Avilla
93. sy 4elH Ie H 00c ye ja 19 9y ejep 199us Moi SJ0322S diequesoy 9e2uejeq e410328S UIEUJ I V ually OTOZ 8S3 4IA1 ejep M0 J Use2 peuur peur pue saoud jassy zIOZ epns esse yueq 3i qnd Unips pue djeue sav pue eseq J01s9Au PEN pue jeloueul4 A unipalA TLOZ e onad pue ysnyow jeosi4 MO jeosi4 pepeg 00c qnd a e3ep jeosly pue go MO pue zOO 3INI ids lE i pue jeuja1x3 e ep jaays eoue e 93e10d102 TEP 129u leq ugiy 1 aqui E007 paruji pue saoud jassy pue jeueul4 Jojou eax 6007 Suo pue eniw pauum saad yassy u8IH jemueutJ ne ueu 6007 Zeus penu se2ud yassy u8iH jemueutJ pue pjoqaia 6002 u eupoo5 pue ouei o8as pawn s ud yassy u8iH jemueuiJ Oroc jareuuauunig paanur 3 ueupv ejep 19aus aouejeq u8IH Jepueuij pue s d Jessy Asewiwing W sjoo Bunojiuo N xSry DIW9 SAg JUdJOJJIG Jo sonsuojoejeu BIqeL LC eui uo 3 qejreAe suoneoi qnd 484031070 ssaded aoue jlavuns e183e n nuu 19470 pue AdIjod 4W1 4 Uadeg ZU YOM M 500g Jeuunof pamalAas Jaad ui paysijqnd d suoneo i qng IPUN pa OU aSIMJAY O SSa UN Sal10Z a e9 341 104 pasn aq aA se 3ouueo JOJeIIPU BY Saljdw Juejq e sauOa e0 AY 40J pasn aq ue JOedIPU BU 724 S LA A 2ION pjoyasnoy pue ayesodiop OTOZ siay30 ejep pue sauag 31000280J28JA ejep dijqnd OTOZ 8S3 4INI peuuri 31u0u02930J28 AI pue jeioueul4 eje patton pain i jepueul uasuar punq 31Uu10u02980J28lA ejep
94. t Risk FSaR and GDP at Risk GDPaR are the worst possible realization at 5 percent probability of quarterly growth in real GDP and in the equity returns of a large portfolio of financial firms respectively At end 2007 it is unclear from the GDPaR that intensified financial sector stress could spillover to GDP growth However the Systemic Contingent Claims Analysis confirms that sovereign contingent liabilities are increasing e Panel E Country X continues to be strongly connected to the rest of the world both in terms of actual cross border balance sheet linkages of banks and potential spillover risks from market contagion This panel illustrates cross border spillover and contagion risks from two complementary perspectives Joint Distress Indicators based on market prices and Network Analysis using BIS data Network Analysis of bilateral cross border banking claims shows that the vulnerability of X from countries A and B are very high Market based Joint Distress Indicators are showing a rise in spillover risks between X and four other countries e Panel F The estimated likelihood of a systemic crisis has increased but is still small This panel directly estimates the likelihood of a systemic crisis either a banking sector crisis or a broader economic crisis based on complementary probability models The credit based banking crisis model shows an uptick in crisis probability and so does a more general crisis prediction model Figure
95. t data on outstanding debt to construct default measures at different horizons Tool Snapshot Attributes Description Coverage Sectors Institutions All financial institution with balance sheet data and equity market data Types of risk Credit risk Interpretation Main output PD risk neutral measure EDF physical measure Other outputs Distance to default Loss given default Implied Haircut and Equity based fair value CDS Thresholds No Time horizon Short term through medium term predictive power Data requirements Traded equity data equity value equity return equity volatility or CDS and balance sheet data debt face value and maturity structure Reference Kealhofer 2003 39 Methodology This methodology applies the insight by Black Scholes and Merton that views a firm s debt as an option on the asset value of the firm An option valuation approach can thus be applied to assess the default risk of a firm with traded equity or credit spreads The distance to default dtd at time t of a firm with inferred value of assets V asset volatility o face debt value D risk free rate of return r and time to maturity T t is given by TIER c A T t Under standard distributional assumptions in the stochastic process of the firm s value the risk neutral PD is characterized by PD 1 N ata dtd Moody s KMV has applied this valuation framework to compute a physical measure of default risk Using a lon
96. t variance zero mean normally distributed random variable and a time varying volatility o such that af a b The parameter g assumes different values in different states of nature It indicates whether the system is at low or high volatility regime for example The model is estimated by maximum likelihood Example VIX and probability of a high volatility regime over time This example uses daily data on the VIX between 1998 and 2008 to estimate the probability that financial markets would experience a high volatility regime when risks become systemic It shows that during the Lehman episode the volatility of VIX reached historic highs Moreover the figure suggests that markets signaled at the very beginning of the subprime crisis an elevated probability of a regime characterized by high volatility where systemic events become more likely Absolute change in VIX left scale Probability of being in high volatility state right scale 30 Beginning of rentan Russian s default WorldCom and eee oe a and LTCM Brazil s election Turkey 20 f crisis 0 8 Bear Shanghai ptearns 10 stock market 0 6 correction 0 i 0 4 Dot com T T bubble 9 14 burst 10 0 2 20 A Iah LV N j n A 0 0 1998 2000 2002 2004 2006 2008 Source Hermosillo and Hesse 2009 60 XVI FINANCIAL SOUNDNESS INDICATORS FSIs FSIs are indicators of the current soundness of the financial system in a country and of its corp
97. tems Specifically BIS data can be used to run the two network models the Cross Border Network model can be used to calculate different types of connections first round impact between financial systems and estimate the probability of a domestic financial crisis while the Cross Border Banking Contagion model can be used to run a network analysis including multiple round spillovers of solvency and funding risk from each financial system to the country These models provide information on potential spillovers through direct exposures but they do not offer information on how the system might behave during crises when both direct and indirect including common exposures come into play 41 Inthe absence of full cross exposure data or in order to complement the above analyses spillover models based on market data such as JDI Returns Spillovers or Diebold Yilmaz DY Distress Spillovers DS Systemic CCA SCCA can be used to assess potential reactions and spillovers between financial institutions across borders either under normal DY or extreme conditions JDI DS SCCA 19 Overall assessment 42 The available tools tend to capture somewhat better the impact of cross border shocks than their likelihood However data limitations with regard to cross border exposures especially among individual institutions e g G SIFIs and with other sectors in foreign countries remain a serious obstacle to in depth analyses of cross border co
98. tivity 1 Time varying capital 20 gains requirements No macroprudential policy 5 Time varying capital i requirements No macroprudential policy do Unhealthy house price boom Deviations of real GDP from baseline Deviationsof real GDP from baseline 2 S amp 5 1 ki 5 7 9 1X1 133 135 17 19 21 23 325 27 298 31 33 2 3 5 9 11 13 15 17 19 21 23 25 27 29 31 33 Quarters Quarters Source IMF 2011b Note Time varying capital requirements are designed as a rule that depends upon the growth in the credit to GDP ratio No macroprudential policy includes fixed microprudential capital requirements The baseline assumes no shock and no macroprudential policy 76 References Adrian Tobias and Markus Brunnermeier 2010 CoVaR Federal Reserve Bank of New York Staff Reports Allen M Rosenberg C Keller C Setser B and Roubini N 2002 A Balance Sheet Approach to Financial Crises IMF Working Paper 02 210 Washington International Monetary Fund Arslanalp Serkan and Takahiro Tsuda 2012 Tracking Global Demand for Advanced Economy Sovereign Debt IMF Working Paper 12 284 Washington International Monetary Fund Arsov Ivailo Elie Canetti Laura Kodres and Srobona Mitra 2013 Near Coincident Indicators of Systemic Stress IMF Working Paper 13 115 Washington International Monetary Fund Baldacci E McHugh J and Petrova I 2011 Measuring Fiscal Vulnerabilities
99. torical distribution of the gap between contemporaneous growth and a 5 year rolling average Second a set of medium term 5 year rolling average and near term variables lagged one period are identified as potential indicators Under the noise to signal nonparametric approach a threshold value minimizing the ratio of false alarms to true signals is calibrated for each indicator and a weighted composite indicator constructed where an indicator s weight correspond to its forecasting ability The value of the composite indicator is then mapped to a crisis probability defined as the percentage of crisis observations for which the composite indicator exceeds its critical threshold Example A crisis prediction model has been used in October 2011 to identify key vulnerabilities and assess systemic risk in advanced economies The graph below shows a medium risk of growth slowdown a probability above 10 percent for France Germany Italy and United Kingdom Moreover the results suggest that France and United Kingdom feature an elevated risk of fiscal crisis a probability above 20 percent On the other hand financial risk net of sovereign distress spillover or contagion effects remains contained 60 4 France 60 Germany 40 40 20 20 o o 2000 2012 2000 2012 60 T 60 United Kingdom Italy 40 40 ui dei A 0 0 2000 2012 2000 2012 nzxcrs Source IMF FSB 2010 and Vulnerability Exercise fo
100. ty Liquid assets to total assets liquid asset ratio Liquid assets to short term liabilities Sensitivity to market risk Net open position in foreign exchange to capital Encouraged set Deposit takers Capital to assets Large exposures to capital Geographical distribution of loans to total loans Gross asset position in financial derivatives to capital Gross liability position in financial derivatives to capital Trading income to total income Personnel expenses to noninterest expenses Spread between reference lending and deposit rates Spread between highest and lowest interbank rate Customer deposits to total noninterbank loans Foreign currency denominated loans to total loans Foreign currency denominated liabilities to total liabilities Net open position in equities to capital Other financial corporations Assets to total financial system assets Assets to GDP Nonfinancial corporations sector Total debt to equity Return on equity Earnings to interest and principal expenses Net foreign exchange exposure to equity Number of applications for protection from creditors Households Household debt to GDP Household debt service and principal payments to income Market liquidity Average bid ask spread in the securities market Average daily turnover ratio in the securities market Real estate markets Residential real estate prices Commercial real estate prices Residential real est
101. v and others 2013 31 Methodology Quantile regressions are used to derive time varying CoVaR Specifically the measure of contribution of an institution to systemic risk is ACoVaR the difference between the VaR of the financial system conditional on the distress of a particular financial institution i and the VaR of the financial system conditional on the median state of the institution i Quantile regressions the 5 and the 50 of the weekly returns growth in market value of assets of institution I X and the system X are estimated conditional on state variables M The Libor OIS spread and the weekly change in the yield curve defined as the spread between the 10 year Treasury bond yield and the 3 month Treasury bill yield are used in M Xi a y M E a m qn a Be x 4 gam M gem The predicted fitted values are used to derive the following at q 5 and q 50 VaR q X p M CoVaR q D od Gem p VaR q p s Finally the ACoVaR of each institution 1s simply ACoVaR 5 CoVaR 596 Co VaR 50 f Va 5 VaR 5094 Example United States Delta CoVaR 17 financial institutions 20 i Delta CoVaR Bear Stearns Country Financial Institutions 17 Delta CoVaR Whole Financial System U S financial institutions covering 94 commercial and investment banks The chart shows the time varying 12 T ACoVak of the financial system ETT and the s
102. vernment agencies Source Gray and Jobst 2011 52 XII CROSS BORDER INTERCONNECTEDNESS The network model uses annual cross border banking sector exposures to construct a measure of global interconnectedness It also estimates the impact of interconnectedness on the likelihood of a banking crisis within a one year forecast window using an econometric specification Tool Snapshot Attributes Description Coverage Sectors Institutions Banking sector Types of risk Risk of a banking crisis from contagious defaults Interpretation Main output Probability of a banking system s default conditional on its cross border interconnectedness Other outputs Combination of interconnectedness thresholds to identify high and low crisis probability areas using a nonparametric approach Thresholds An increase in upstream interconnectedness below 0 37 95 percent of the interconnectedness observations in the sample calibrated at average macroeconomic variables reduces the probability of a banking crisis When upstream interconnectedness is above 0 37 the remaining 5 percent the relationship between interconnectedness and crisis probability is more complex it is upward sloping at first only to become downward sloping again Time horizon The interconnectedness measure has forecasting ability one year ahead Data requirements Low frequency data from BIS quarterly locational banking statistics macroeconomic and bank balance sheet data and bank
103. which compare the market value of an entity s assets to its debt obligations 10 Toolkit limitations As highlighted above available tools are very heterogeneous none is universally applicable to address all aspects of systemic risk and all are subject to important underlying assumptions data issues or model risk For instance as is widely acknowledged the informational content of market prices may be undermined under certain circumstances e g both during stress and exuberant times or may not capture rising interconnectedness within the financial system More broadly and despite ongoing progress in developing and improving the toolkit efforts to integrate individual tools into a comprehensive and internally consistent quantitative framework e g across sectors types of risk or time horizons are still in their infancy IIl MAPPING TOOLS TO THE TERRITORY A PRACTICAL APPROACH 11 This section presents the existing toolkit by addressing six key questions policymakers should ask themselves as they assess systemic risk Building on the Binder presented in the Appendix which presents each tool separately the focus of this section is on the best selections and combinations of tools to address each key question taking into account the complementarities among tools and their relative strengths and weaknesses 12 The proposed sequence of key questions broadly reflects the increasing extent of macro financial linkag
104. ystemic contribution of 8 Threshold for entering crisis phase 7 2 i y Bear Stearns to overall stress A value of 7 2 represents the loss rate 4 See A DNE TEN in the system when a portfolio of nee pog od e firms moves from their median o A errem a cel a state to a distress state LOHN IVO POH HLOV DH HV D OH Hv Pg nv og m av 2003 2004 2005 2006 2007 2008 Source Arsov and others 2013 32 II JOINT DISTRESS INDICATORS The set of Joint Distress Indicators JDI includes a time varying measure of joint probability of distress JPoD between financial institutions or sovereigns with nonlinear distress dependence These indicators can be used to construct a Financial Institutions Stability Index FISI reflecting the expected number of financial institutions FIs becoming distressed given that at least one FI has become distressed It can also be used to assess banks inter linkages by computing pair wise conditional probabilities of distress The JDI provides complementary perspectives of systemic risk and FIs exposure and contribution to systemic risk Tool Snapshot Attributes Description Coverage Sectors Institutions Banks nonbank financial institutions and sovereigns Types of risk Spillover risk during distress i e expected large losses or possible default It covers credit Interpretation Main output FISI JPoD Distress Dependence Matrix DiDe and Probability of Cascade Effects PCE Other outputs

Systemic Risk Monitoring (ISysMo") Toolkit2 A User Guide

Contents

Download Pdf Manuals

Related Search

Related Contents

Systemic Risk Monitoring (ISysMo&quot;) Toolkit2 A User Guide

Contents

Download Pdf Manuals

Related Search

Related Contents

Systemic Risk Monitoring (ISysMo") Toolkit2 A User Guide