Financial Stability Measures Miguel Segoviano* and Raphael Espinoza
[email protected] RE i
[email protected] @i f
Financial Markets Group - Bank of England Conference London-January 25, 2011
*O temporall leave *On l f from the h IMF. IMF The Th views i expressed d iin this hi presentation i are those h off the h author h and dd do not necessarily represent those of the IMF, CNBV or IMF,CNBV policy. Any errors remain attributable to the authors.
Outline I. Objective. II. Main points of Modelling Framework. III. Financial Stability and Systemic Risk Assessment: Distress Dependence Amongst FIs. Financial Stability Indicators Financial Stability Indicators. Systemic Macro‐Financial ST. Second‐round effects IV. Financial Stability and Sovereign Risk. Spillover Coefficient. Index of Global Risk Aversion. Fundamentals. V. Conclusions.
2
Modeling Framework We assume that the financial system is a portfolio of FIs. Macroeconomic Risk Factors
Financial Risk Factors
Commercial Banking PoD
Pension Funds PoD
Mutual Funds POD
Develpmt Bking Insurance Cos PoD POD
Brokers PoD
Others
Financial System´s Multivariate Density EAD
LGD
0 . 2
0 . 1 5 0 . 1 0 . 0 5
0 4 2
4 2
0
Systemic Loss Simulation
0
-2
-2 -4
-4
Systemic Stress Indicators
Contagion Indicators Systemic Loss Indicators
Sovereign Risk Assessment
3
Objective Framework to estimate consistently: •
Complementary financial stability indicators.
•
Q Quantification of expected and extreme losses at the systemic level. ifi i f d d l h i l l
•
Quantification of the marginal contribution to systemic risk of Quantification of the marginal contribution to systemic risk of individual institutions.
•
Assessment of the underlying factors causing sovereign risk to spread.
•
Stress Testing Stress Testing.
4
Main Points • It is a comprehensive coverage: The methodology allows for the inclusion of banking and non‐banking financial institutions (FIs)/sectors. • It captures contagion effects: It takes into account interlinkages (direct and indirect) amongst Fis. • It captures changes across the economic cycle of distress dependence amongst FIs and sovereigns. dependence amongst FIs and sovereigns. • It integrates complementary information: It uses micro‐ f founded supervisory data and market‐based information. d d i d t d k t b di f ti • It It incorporates a wide set of factors: It accounts for a wide incorporates a wide set of factors: It accounts for a wide set of macroeconomic and financial risk factors.
5
Main Points
• It It provides robust estimations: It benefits from robust provides robust estimations: It benefits from robust estimation with restricted data (under the PIT criterion). • It can be extended to capture second round effects: It allows to take into account second‐round effects and macro‐ financial linkages. financial linkages. • It is being extended to different applications with national authorities around the world. h ii d h ld
6
Implementation Map
IMF ECB
USA , Canada / FSAP Egypt FSAP South Africa FSAP Denmark FSAP Bank of Lithuania / FSAP
Banque de France
Norges Bank
Banca d’Italia
CNBV México
Bank of Japan
Bank of Jordan
Deutshe Bundesbank Sveriges Riksbank
Central Bank of UAE Bank of Malaysia
7
Distress Dependence Segoviano and Goodhart (2009) Distress dependence between institutions is incorporated via joint Distress dependence between institutions is incorporated via joint movements of their PoDs, which in turn move in tandem due to Systemic shocks h k
Indirect Links Lending to common sectors Proprietary Trades
Direct Links InterI t -Bank Inter B kD Deposit it M Markets k t Syndicated Loans The recent crisis underlined that proper estimation of distress dependence amongst FIs in a financial system is essential for financial stability FI i fi i l i i l f fi i l bili assessment. Contagion through Idi Idiosyncratic i Shocks Sh k
G dh Goodhart, Sunirand, Tsomocos (2004). S i d T (2004) 8
The CIMDO Methodology
• • •
Problem: ‘how to estimate P(A,B) if we have P(A) and P(B)?’ ( ) ( ) ( ) We can assume a known parametric distribution (e.g. multivariate normal), and estimate/calibrate parameters using data on A and B, but it seldom fits the data… …or, we can try to “match” the data with a non‐parametric distribution ‐‐> CIMDO.
Advantages: •
Robust: Without imposing unrealistic parametric assumptions.
•
It can be estimated from partial information: From PoDs on marginals, without the need to explicitly to explicitly set correlation structures. set correlation structures
•
It characterizes the full “distributional dependence”: Rather than just linear dependence (correlations) or relations in the first few moments.
•
It embeds effects of changing macroeconomic conditions/shocks (via PoDs): It allows measurement of changes in dependence after shocks.
Source: Segoviano (2006) 9
CIMDO‐Density
Empirical I f Information ti
10
CIMDO‐Copula
Lett X and L d Y be b two t random d variables i bl with ith continuous ti di t ib ti functions distribution f ti F and d H respecitvely, it l then the Spearman Correlation of X and Y is defined and denoted by the following:
S ( X , Y ) ( F ( X ), H (Y )) 12 [C (u , v) uv]dudv 12 C (u, v) 3 2 I2 I Where I [0,1]x[0,1] and ρ(F(X),H(Y)) is the Pearson Correlation of the transformed uniform random variables F(X) and G(Y). 2
11
Complementary Indicators
SYSTEMIC INDICATORS
MICRO-BASED INDICATORS Market-based Information
Supervisory Information
Supervisory Information
Market-based Information
M to M Assets PD Credit Card PD ConsCre PD Housing Ho sing PD Commercial
Systemic Stress Indicators PoD Banks SI
PoD Banks MI
Joint Probability of Distress
Financial Stability Index
PD Governmental
C t i Indicators Contagion I di t
PD Financial
PoD Developt B k Banks
PoD Pension Funds PoD Mutual Funds
PoD Insurance C Co
PoD Brokers
Distress Dependence Matrix
Contagion Index
Spillover Coefficient
Systemic Loss Indicators Extreme Systemic Loss
Marginal Contribution to Systemic Risk
Sovereign Risk A Assessment t
Cascade Effects Probability
PoD from Supervisory Information PLD Baseline Scenario PLD Stressed Scenario St Stressed dV VaR R Expected Loss
Unexpected Loss
PLD Baseline Scenario
PLD Stressed Scenario Stressed PoD PoD
Benchmark
13
Systemic Stress Indicators: U.S. Financial Stability Index: Expected number of FIs in distress given that p g at least one became distressed (left scale).
Joint Probability of Distress (JPoD): Likelihood of common distress of all the FIs in the system (right scale).
3.5 3.0
0.025 1. 2. 3. 4.
Bear Stearns episode (3/11/08) Lehman Bankruptcy p y and AIG bailout (9/15-16/08) ( ) TARP I bill failure (9/30/08) Global central bank intervention (10/8/08)
1
2 34
0.020
2.5 2.0
Bank Stability Index (Number of banks, left scale)
0.015
1.5
0 010 0.010
1.0 JPoD (Probability of default %, right scale)
0.005
0.5 0.0 Jan 07 Jan-07
0.000 May 07 May-07
Sep 07 Sep-07
Jan 08 Jan-08
May 08 May-08
Sep 08 Sep-08
14
Systemic Stress Indicators: Mexico Financial Stability Index: Expected number of FIs in distress given that Expected number of FIs in distress given that at least one became distressed (left scale).
Joint Probability of Distress (JPoD): Likelihood of common distress of all the FIs in the system (right scale).
JPOD-FSI: México FSI BSI
4 35 3.5 3 2.5 2 1.5 1 0.5 0
1. Lehman spillover, derivatives’ market crisis and mutual funds funds’ crisis (Oct2008). 2. H1N1 crisis (March-April-2009).
JPOD JPOD
0.002
1
2
0.0018 0.0016 0.0014 0.0012 0.001 0.0008 0.0006 0.0004 0.0002 0
15
Systemic Stress Indicators They incorporate changes in distress‐dependence that are consistent with They incorporate changes in distress dependence that are consistent with the economic cycle. 30 3.0 Joint probability of distress
2.5
Average probability of distress 20 2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 J Jan-07 07
M May-07 07
S Sep-07 07
J Jan-08 08
M May-08 08
S Sep-08 08
16
Contagion Indicators: DiDe U.S. Distress Dependence Matrix: Probability that FI (row) falls in distress given that FI (column) falls in distress.
July 1 2007 September 12, 2008 July 1, 2007‐ September 12 2008 July 1, 2007
Citi
BAC
JPM
Wacho
WAMU
GS
LEH
MER
MS
AIG
Citigroup Bank of America JPMorgan Wachovia Washington Mutual Goldman Sachs Lehman Merrill Lynch y Morgan Stanley AIG Column average
1.00 0.12 0 15 0.15 0.12 0.16 0.17 0.22 0.19 0.19 0.07 0.24
0.14 1.00 0 42 0.42 0.33 0.28 0.25 0.32 0.32 0.31 0.14 0.35
0.11 0.27 1 00 1.00 0.24 0.21 0.28 0.32 0.33 0.28 0.10 0.31
0.11 0.27 0 31 0.31 1.00 0.23 0.21 0.26 0.25 0.24 0.10 0.30
0.08 0.11 0 13 0.13 0.11 1.00 0.11 0.15 0.17 0.14 0.05 0.21
0.09 0.11 0 19 0.19 0.12 0.12 1.00 0.43 0.33 0.35 0.07 0.28
0.08 0.10 0 16 0.16 0.10 0.12 0.31 1.00 0.31 0.28 0.06 0.25
0.09 0.12 0 19 0.19 0.12 0.16 0.28 0.35 1.00 0.30 0.07 0.27
0.09 0.12 0 18 0.18 0.12 0.13 0.31 0.33 0.31 1.00 0.06 0.26
0.08 0.15 0 17 0.17 0.14 0.15 0.17 0.20 0.20 0.16 1.00 0.24
September 12, 2008
Citi
BAC
JPM
Wacho
WAMU
GS
LEH
MER
MS
AIG
Citigroup Bank of America JPMorgan W h i Wachovia Washington Mutual Goldman Sachs Lehman Merrill Lynch Morgan Stanley AIG Column average
1.00 0.14 0.13 0 34 0.34 0.93 0.15 0.47 0.32 0 21 0.21 0.50 0.42
0.20 1.00 0.29 0 60 0.60 0.97 0.19 0.53 0.41 0 28 0.28 0.66 0.51
0.19 0.31 1.00 0 55 0.55 0.95 0.24 0.58 0.47 0 29 0.29 0.59 0.52
0.14 0.18 0.16 1 00 1.00 0.94 0.13 0.43 0.30 0 19 0.19 0.53 0.40
0.07 0.05 0.05 0 17 0.17 1.00 0.06 0.25 0.16 0 09 0.09 0.29 0.22
0.17 0.16 0.19 0 36 0.36 0.91 1.00 0.75 0.53 0 40 0.40 0.54 0.50
0.13 0.10 0.11 0 27 0.27 0.88 0.18 1.00 0.37 0 22 0.22 0.43 0.37
0.14 0.13 0.14 0 31 0.31 0.92 0.20 0.59 1.00 0 27 0.27 0.49 0.42
0.16 0.15 0.16 0 34 0.34 0.91 0.27 0.62 0.48 1 00 1.00 0.47 0.46
0.11 0.11 0.09 0 29 0.29 0.89 0.11 0.37 0.26 0 14 0.14 1.00 0.34
Row average 0.19 0.24 0 29 0.29 0.24 0.26 0.31 0.36 0.34 0.33 0.17 0.27
Row average 0.23 0.23 0.23 0 42 0.42 0.93 0.25 0.56 0.43 0 31 0.31 0.55 0.41
17
Contagion Indicators: DiDe Mexico Distress Dependence Matrix: Probability that FI (row) falls in distress given that FI (column) falls in distress.
July 1, 2007‐ October 22, 2008 01/01/07 Banamex BBVA Santander Banorte HSBC Inbursa Scotiabank ING Bajío
Interacciones IXE Azteca Prom Columna
22/10/08 Banamex BBVA Bancomer Santander Banorte HSBC Inbursa Scotiabank Inverlat ING Bajío
Interacciones IXE Azteca Prom Columna
Banamex
BBVA
Santander
Banorte
HSBC
Inbursa
Scotiabank
ING
Bajío
Interacciones
IXE
Azteca
Prom Renglón
1.00 0.25 0.25 0.24 0 27 0.27 0.18 0.31 0.25 0.31 0.11 0.11 0.19
0.26 1.00 0.72 0.29 0 42 0.42 0.15 0.25 0.47 0.42 0.11 0.12 0.19
0.25 0.69 1.00 0.29 0 41 0.41 0.16 0.25 0.44 0.41 0.11 0.11 0.19
0.24 0.28 0.29 1.00 0 27 0.27 0.18 0.23 0.28 0.31 0.11 0.11 0.18
0.27 0.40 0.41 0.26 1 00 1.00 0.15 0.25 0.37 0.61 0.11 0.11 0.18
0.18 0.14 0.16 0.17 0 15 0.15 1.00 0.16 0.15 0.22 0.11 0.10 0.22
0.32 0.24 0.25 0.22 0 26 0.26 0.17 1.00 0.23 0.29 0.11 0.12 0.17
0.25 0.45 0.44 0.27 0 36 0.36 0.15 0.23 1.00 0.36 0.11 0.11 0.18
0.30 0.39 0.40 0.30 0 60 0.60 0.22 0.28 0.36 1.00 0.10 0.11 0.43
0.11 0.11 0.11 0.10 0 11 0.11 0.12 0.11 0.11 0.11 1.00 0.11 0.10
0.11 0.11 0.11 0.11 0 11 0.11 0.11 0.12 0.11 0.11 0.11 1.00 0.11
0.19 0.18 0.19 0.18 0 18 0.18 0.23 0.17 0.18 0.43 0.10 0.11 1.00
0.29 0.35 0.36 0.29 0 35 0.35 0.23 0.28 0.33 0.38 0.18 0.18 0.26
0.29
0.37
0.36
0.29
0.34
0.23
0.28
0.33
0.38
0.18
0.18
0.26
0.29
Banamex
BBVA Bancomer
Santander
Banorte
HSBC
Inbursa
Scotiabank Inverlat
ING
Bajío
Interacciones
IXE
Azteca
Prom Renglón
1.00
0.49
0.48
0.46
0.49
0.39
0.54
0.47
0.53
0.29
0.29
0.40
0.49
0.46 0.47 0.47 0.50 0.38
1.00 0.83 0.52 0.63 0.35
0.79 1.00 0.52 0.62 0.35
0.48 0.50 1.00 0.49 0.37
0.58 0.60 0.49 1.00 0.34
0.34 0.36 0.39 0.36 1.00
0.46 0.48 0.46 0.49 0.37
0.62 0.62 0.50 0.58 0.35
0.59 0.61 0.53 0.75 0.42
0.27 0.29 0.28 0.29 0.28
0.28 0.29 0.30 0.29 0.27
0.37 0.39 0.40 0.39 0.42
0.52 0.54 0.49 0.53 0.41
0.50 0.48 0.52 0.29 0.25 0.41
0.45 0.67 0.61 0.30 0.26 0.40
0.45 0.64 0.60 0.30 0.26 0.40
0.42 0.50 0.51 0.28 0.25 0.39
0.45 0.57 0.72 0.29 0.25 0.39
0.35 0.36 0.43 0.29 0.24 0.44
1.00 0.46 0.51 0.29 0.26 0.39
0.42 1.00 0.56 0.29 0.25 0.39
0.48 0.58 1.00 0.29 0.26 0.61
0.27 0.29 0.28 1.00 0.24 0.26
0.28 0.29 0.29 0.28 1.00 0.29
0.36 0.39 0.59 0.26 0.25 1.00
0.45 0.52 0.55 0.35 0.31 0.45
0.48
0.54
0.53
0.47
0.52
0.41
0.48
0.50
0.55
0.34
0.35
0.44
0.47
18
DiDe: Mexico
DiDe: México- IFs Españolas MEXICO-BBVA
MEXICO-SANTANDER
Probabilida ad
1 0.8 0.6 0.4 0.2 0 Jun-2007
Oct-2007
Feb-2008
Jun-2008
Oct-2008
Feb-2009
Jun-2009
Oct-2009
Feb-2010
Jun-201
DiDe: IF´s Españolas-México
Probabilidad d
BBVA-MEXICO
SANTANDER-MEXICO
0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Jun-2007
Oct-2007
Feb-2008
Jun-2008
Oct-2008
Feb-2009
Jun-2009
Oct-2009
Feb-2010
Jun-201
Contagion Indicators: Spillover Coefficient Spillover Coefficient (SC): Vulnerability of a country/FI given distress in other countries/Fis.
P(A) P(B) P(C)
CIMDO Methodolog y
P( A / A) P( A / B ) P( A / C ) P( B / A) P( B / B ) P( B / C ) P(C / A) P(C / B) P(C / C )
P(A,B,C) JPoD P(A B); P(A,C); P(A,B); P(A C); P(B,C) P(B C) Bayes’ Bayes Law
For e.g. country (A)/FI(A):
SC(A)=P(A/B)*P(B) + P(A/C)*P(C)
20
Contagion Indicators: SC Europe Spillover Coefficient: European p p Region g 0.35 0.3 0.25 0.2
GER GRE
0.15
IRE SWE
0.1 0.05 0 03/01/05
03/01/06
03/01/07
03/01/08
03/01/09
03/01/10
21
Contagion Indicators: Contagion Index Contagion Index (CI): Toxicity of the distress of a country/FI on other countries/FIs.
P(A) P(B) P(C)
CIMDO Methodolog y
P( A / A) P( A / B ) P( A / C ) P( B / A) P( B / B ) P( B / C ) P(C / A) P(C / B) P(C / C )
P(A,B,C) JPoD P(A B); P(A,C); P(A,B); P(A C); P(B,C) P(B C) Bayes’ Bayes Law
For e.g. country (A)/FI(A):
CI(A)=P(A)+P(B/A)*P(A) + P(C/A)*P(A) 22
Contagion Indicators: Probability of Cascade Effects Probability of Cascade Effects (PCE): Probability that at least one FI becomes distressed given that a given FI becomes distressed. PCE Lehman/AIG (September 12). 100
90 Lehman AIG 80
70
60
50
40
30
20
10
9/1/2008 8
8/1/2008 8
7/1/2008 8
6/1/2008 8
5/1/2008 8
4/1/2008 8
3/1/2008 8
2/1/2008 8
1/1/2008 8
12/1/2007 7
11/1/2007 7
10/1/2007 7
9/1/2007 7
8/1/2007 7
7/1/2007 7
6/1/2007 7
5/1/2007 7
4/1/2007 7
3/1/2007 7
2/1/2007 7
1/1/2007 7
0
23
Output: Cross-region Spillovers 1.00 0.90 0.80 0.70 0.60 0.50 0 40 0.40 0.30 0.20 0.10 0.00
C it i-M exico C it i-Lat C it i-EE i EE
5/
20
09
09
7/
9/
2/
2/
20
09
09
20
20
2/
2/
20
3/
11
1/
/2
2/
/2
00
8
09
08
08
20
20 2/
9/
7/
3/
2/
08
08 20
2/
2/
20
20 2/
1/
11
5/
7
08
07
00
/2
/2
07
20 2/
20 2/
7/
5/
9/
07
20 2/
20
20
2/
2/
3/
1/
07
C it i-A sia
07
Prob ab ility
Probability of dis tre s s of C itigrou p con dition al on dis tre s s of an oth e r e n tity
Probability of dis tre s s of an e n tity con dition al on dis tre s s of C itigrou p 0.60
Prob ab ility
0.50 0 40 0.40
B A C -C it i
0.30
U B S-C it i
0.20
D B -C it i
0.10
09 20
09 3/ 9/
5/
7/
3/
3/
20
09 20
09
09
20 3/
1/
3/
3/
20
00
8
08
/2
11
9/
/3
3/
20
08 20
08
5/
7/
3/
3/
20
08 20
3/
20 3/
3/
11
1/
/3
/2
00
7
08
07
07
20 9/
7/
5/
3/
3/
20
07
07
20 3/
20
3/
1/
3/
3/
20
07
0.00
1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 0.10 0 00 0.00
C it igroup U BS DB
09
3/
20 3/
20
09 9/
09 20
3/ 5/
7/
09
09
20
3/
3/
20
8 00 /2
3/ 1/
/3 11
9/
3/
20
08
08
08
20 3/ 7/
5/
3/
20
08
3/
3/
20
08
00
1/
3/
20
7
07 11
/3
/2
07
20 3/ 9/
20 3/ 7/
20 3/ 5/
20 3/
3/
07
07
BA C
07 20 3/ 1/
Prob ab ility
C as cade Effe cts
24
Systemic Loss Indicators
Commercial Banking PoD
Pension Funds PoD
Mutual Funds POD
Develpmt Bking Insurance Cos PoD POD
Brokers PoD
Others
Financial System´s Multivariate Density EAD
LGD
Systemic L Loss Si Simulation l ti
0 . 2
0 . 1 5 0 . 1 0 . 0 5
0 4 2
4 2
0
0
-2
-2 -4
Systemic Loss Indicators
MCSR
-4
Systemic Stress Indicators Contagion Indicators Sovereign Risk Indicators
25
Systemic Loss Indicators
E Extreme Systemic Loss S i L
PLD Independent IFs PLD Stressed PoDs Independent IFs PLD Stressed PoDs Stressed Distress Dependence
Expected Loss
Unexpected Loss 25-45%
26
14.0%
10.0%
1999Q2 2000Q1 2000Q4 2001Q3 2002Q2 2003Q1 2003Q4 2004Q3 2005Q2 2006Q1 2006Q4 2007Q3 2008Q2 2009Q1 2009Q4 2010Q3 2011Q2 2012Q1 2012Q4 2013Q3 2014Q2 2015Q1 2015Q4
Forecasted PoD Under Assumed Macroeconomic Scenarios
Baseline
12.0% Min
Mean
Max
8.0%
6.0%
4.0%
2.0%
0.0%
Risk-Neutral and Subjective PoD 0.045 0.040 0.035
Comparing Risk‐Neutral Default Probability and Adjusted Real‐world Probability
0.030 0.025 0.020
RN‐Median
0.015
Real‐Median
0.010 0.005 0.000
•
For loss estimation purposes, we convert risk neutral CDS-PoDs to subjective PoDs. ((Espinoza p and Segoviano, g , 2011). )
Systemic Expected Losses
B li Baseline
• •
2007 2008 2009 2010 2011 2012 2013
$000 % Assets 27,144,218 0.22 123,251,952 0.96 58,981,703 0.45 109,916,227 0.84 77 632 877 77,632,877 0 59 0.59 33,721,172 0.26 22,758,454 0.17
T t l Total
453 406 602 453,406,602
3 50 3.50
Ad Adverse % GDP 0.19 0.85 0.42 0.77 0 53 0.53 0.22 0.15 3 14 3.14
$000 % Assets 27,144,218 0.22 123,251,952 0.96 60,233,379 0.46 127,844,040 0.98 83 131 248 83,131,248 0 64 0.64 77,552,119 0.59 59,380,397 0.45 558 537 354 558,537,354
4 31 4.31
% GDP 0.19 0.85 0.43 0.90 0 57 0.57 0.52 0.39 3 84 3.84
Potential Losses the system y could incur. Through-time pattern of losses is highly consistent with assumed macroeconomic scenarios.
Systemic Unexpected Losses
2007 2008 2009 2010 2011 2012 2013
Bas eline
Advers e
VaR 99%
VaR 99%
$000 % As s ets 162 523 198 162,523,198 1.33 1 33 356,360,895 2.78 238,209,910 1.82 334,112,073 2.56 270 050 643 270,050,643 2 07 2.07 180,355,988 1.38 148,846,513 1.14
% GDP 1 15 1.15 2.47 1.70 2.34 1 84 1.84 1.20 0.97
$000 % As s ets 162 523 198 162,523,198 1.33 1 33 356,360,895 2.78 245,110,532 1.87 368,703,692 2.82 295 548 717 295,548,717 2 26 2.26 278,878,012 2.13 242,699,313 1.86
% GDP 1 15 1.15 2.47 1.74 2.59 2 02 2.02 1.85 1.57
1,690,459,220
13.07
11.67
1,949,824,359
15.06
13.40
Memo item 2009 total equity 1,022,994,436
7.82
7.28
1,022,994,436
7.82
7.28
Total
•
Extreme Losses that the system could incur with 1 percent probability.
•
Cumulative extreme losses could be significant in both scenarios, especially compared to 2009 capital levels.
Marginal Contribution to Systemic Risk Marginal Contribution to Systemic Risk (MCSR): It takes into account of size and interconnectedness.
Commercial Banking PoD
Pension Funds PoD
Mutual Funds POD
Develpmt Bking Insurance Cos PoD POD
Brokers PoD
Others
Financial System´s Multivariate Density EAD
LGD
Systemic L Loss Si Simulation l ti
0 . 2
0 . 1 5 0 . 1 0 . 0 5
0 4 2
4 2
0
0
-2
-2 -4
Syste c Loss Systemic oss Indicators Expected Shortfall Shapley Value
-4
Systemic Stress Indicators Contagion Indicators Sovereign Risk Indicators
MCSR
31
Shapley Value •
Let F be a sub-group of members of the financial system containing the financial institution I, we define the “contribution of institution I to F” as V(F)-V(F-{I}). The Shapley Value of institution I could be viewed as the weighted average of the contributions of I over all the sub-groups of the financial system containing institution I. Let´s assume three financial institutions A, B , C Sub‐Group
Loss
Contribution A to Sub‐group
Contribution B to Sub‐group
Contribution C to Sub‐group
A B C A,B A,C B,C A,B, C Shapley Value MCSR
1 3 5 3.5 5.5 7 8.5
1 None None 0.5 0.5 None 1.5 1 0.12
None 3 None 2.5 None 2 3 2.75 0.32
None None 5 None 4.5 4 5 4.75 0.56
Permutation
Sub‐Group of all instituions before A including A Sub‐Group of all instituions before A including A
Calculation of Contribution of A Calculation of Contribution of A
Contribution of A Contribution of A
ABC ACB BAC CAB BCA CBA
A A B,A C,A B,C,A C,B,A Shapley Value of A
V(A) ‐ V(0) = 1‐0 = 1 V(A) ‐ V(0) = 1‐0 = 1 V(B,A) ‐V(B) = 3.5 ‐ 3 = 0.5 V(C,A) ‐ V( C) = 5.5 ‐5=0.5 V(C,B,A) ‐ V(C,B) = 8.5 ‐ 7 = 1.5 V(C,B,A) ‐ V(C,B) = 8.5 ‐ 7 = 1.5 Weighted Average
1 1 0.5 0.5 1.5 1.5 1
MCSR
0.12
•
Additivity. The sum of the shapley values of all the members of our financial system gives us the systemic risk of all the financial system.
•
Intuitive Indicator. The Shapley Value of a financial institution captures the systemic importance of it in only one number.
•
Flexibility. Can be applied to any measure of system-wide risk.
Marginal Contribution to Systemic Risk Marginal Contribution to Systemic Risk: g y It takes into account of size and interconnectedness. 0.4
0.016
0.35
0.014
0.3
0.012
0.25
0.01
0.2
0.008
0.15
0.006
0.1
0.004
0.05
0.002
MCSR
POD AIG
Spearman Corr AIG
Mar-09
Feb-09
Jan-09
Dec-08
Nov-08
Oct-08
Sep-08
Aug-08
Jul-08
Jun-08
May-08
Apr-08
Mar-08
Feb-08
Jan-08
0 Dec-07
0
Contagion Index AIG
Right Axis for CI
AIG Factors
Second-Round Effects Macroeconomic Factors Comm Banks PoD
Dvlpmt Banks PoD
GSEs PoD
Financial System’s Returns
Pension Funds PoD
Insurance PoD
Mutual Funds PoD
Financial System´s Tail Risk (JPoD) Macroeconomic Factor
Comm Banks PoD
Exposures
Financial Factors
Dvlpmt Banks PoD
LGDs
Systemic Loss Simulation
Marginal Contribution to Systemic Risk
GSEs PoD
Financial Factors Pension Funds PoD
Insurance PoD
Mutual Funds PoD
Financial System´s y Multivariate Density
Financial System’s Loss Distribution
Financial Stability Measures
34
Financial Stability Measures: Additional Applications
Additional Applications: 1.
Macro‐Financial stages over time.
2.
Spillovers between the banking and corporate sectors.
3.
Sovereign Risk Assessment (Caceres Guzzo Segoviano IMF WP 10/120) (Caceres, Guzzo, Segoviano, IMF WP 10/120).
35
Financial Stability Over Time •
Markov Switching VAR Definition of alternative risk zones given specific values of JPoD and d other th variables i bl 0.002
MSIAH(2)-VAR(1), 1999 (7) - 2009 (3) Jpodrev
dlhouse
0.000
-0.002 1.0
2000 2001 2002 Probabilities of Regime 1 filtered predicted
2003
2004
2005
2006
2007
2008
2009
2003
2004
2005
2006
2007
2008
2009
2003
2004
2005
2006
2007
2008
2009
smoothed
0.5
1.0
2000 2001 2002 Probabilities of Regime 2 filtered predicted di d
smoothed
0.5
2000
• •
2001
2002
Probability of being in different economic actual/assumed ((in ST)) economic shocks. Analysis of IRFs in different enconomic regimes
regimes
due
to 36
10/1/2009
7/1/2009
4/1/2009
1/1/2009
10/1/2008
7/1/2008
4/1/2008
1/1/2008
10/1/2007
7/1/2007
4/1/2007
1/1/2007
10/1/2006
7/1/2006
0.7
4/1/2006
1/1/2006
10/1/2005
7/1/2005
4/1/2005
1/1/2005
10/1//2009
7/1//2009
4/1//2009
1/1//2009
10/1//2008
7/1//2008
4/1//2008
1/1//2008
10/1//2007
7/1//2007
4/1//2007
1/1//2007
10/1//2006
7/1//2006
4/1//2006
1/1//2006
10/1//2005
7/1//2005
4/1//2005
1/1//2005
Spillovers Between Financial and Corporate Sectors
0.08
PoD: Banks & Corporates
0.07
0.06
0.05
0.04
0.03 Average Bank
0.02 Average Corp
0.01
0.00
Distress Dependence: Banks & Corporates
0.6
0.5
0.4
0.3 Bank-Bank
02 0.2 Bank-Corp
Corp-Bank Corp Bank
0.1 Corp-Corp
0
37
Sovereign Risk Assessment: Four Phases
EUR Sovereign 10Y Swap Spreads
1,000 800 600 400 200
GER
FRA
ITA
SPA
NET
BEL
AUT
GRE
IRE
POR
Sovereign Risk
Systemic Response p
Systemic Outbreak
Financial Crisis Build‐Up
0
Jul‐22010
May‐22010
Mar‐22010
Jan‐22010
Nov‐22009
Sep‐22009
Jul‐22009
May‐22009
Mar‐22009
Jan‐22009
Nov‐22008
Sep‐22008
Jul‐22008
May‐22008
Mar‐22008
Jan‐22008
Nov‐22007
Sep‐22007
Jul‐22007
‐200
38
Some Existing Literature on Sovereign Spreads • Models with Risk Aversion based on observable measures ( corporate – (e.g t Treasury spread): T d) – Fiscal situation has temporary and limited impact on sovereign spreads (Afonso and Strauch, 2004). spreads (Afonso and Strauch, 2004). – Short‐term interest rates are the main driver (Manganelli and Wolswijk, 2007) but international risk factors and liquidity premia also matter (Bernoth et al., 2004). – Sovereign spreads widen when the prospects of a domestic financial sector worsen (Mody 2009) sector worsen (Mody, 2009). Risk Aversion as a common factor: Aversion as a common factor: • Risk – common factor using a Kalman filter (Geyer et al, 2004). y g y g – Time‐varying common factor estimated with Bayesian filtering technique (Sgherri and Zoli, 2009).
39
Data and Construction of the Variables: 3 Types of ‘Measures’
• Measure of “contagion”: the Spillover Coefficient (SC). Segoviano and Goodhart, IMF WP 09/4. g • Measure of “risk aversion”: the Index of Risk Aversion (IRA). f ( ) (Espinoza and Segoviano, IMF WP, forthcoming 2011). • Country‐specific fiscal fundamentals: – Overall Budget Balance (% of GDP). Overall Budget Balance (% of GDP). – Public Debt (% of GDP). 40
The Index of Risk Aversion (IRA) Espinoza and Segoviano, IMF WP forthcoming, 2011. • Global risk aversion proxies tend to be ‘over simplistic’ (e.g. US corporate bond spreads to Treasuries). • Statistically sophisticated methods depend on the sample: global risk aversion is assigned to a common trend g or common factor f ((i.e. ‘filtering’). – common factors might be capturing ‘distress dependence’ as well as ‘risk aversion’
• O Our IRA is the ‘factor’ linking risk‐neutral probabilities (extracted IRA i th ‘f t ’ li ki ik t l b biliti ( t t d e.g. from CDS spreads) to the actual probabilities of nature . This factor is the market price of risk in situations of distress p – It does not depend on the sample used
41
The Index of Risk Aversion (IRA) • The price of an asset reflects – Market expectations of the asset´s returns – The price of risk: what investors are willing to pay for receiving income in distress states of nature. The linear (one factor) pricing and the risk‐neutral pricing formulae
1 Pt Et [mt 1 xt 1 ] t 1 ( s )mt 1 ( s ) xt 1 ( s ) t 1 ( s ) xt 1 ( s ) F 1 rt s s
t1(s)mt1(s) where is the price of a security paying $1 in state s and
t 1 ( s ) (1 rt F ) t 1 ( s )mt 1 ( s ) is the risk‐neutral probability that is given by CDS spreads
42
The Index of Risk Aversion (IRA) • Market price of risk under distress:
Et mt 1 | distress Et mt 1 | mt 1
( t ) Et mt 1 vart (mt 1 ) 1 1[ t ]
where t
t Et (mt 1 ) vart (mt 1 )
(1) (2)
• The threshold τ can be exogenous, or such that the probability that the market‐price of risk exceeds the threshold p τ is equal to q the actual probability of distress:
t Et (mt 1 ) 1[1 t ] * var(mt 1 )
(3) 43
The Index of Risk Aversion (IRA) • Market price of risk under distress:
mt
Et ( mt 1 | mt 1 ) Et (mt 1 )
t 1 (1 rt F ) t 1
1[1 t ] * var(mt 1 ) 44
The Index of Risk Aversion (IRA)
Et[mt+1 | mt+1 > τ] 3.2 3.0 2.6 2.8 2.4 2.2 2.0 1.8
45
Estimation: Methodology
• Simple GARCH(1,1) model to estimate sovereign swap (Yt) spreads as a function of: – its lag Yt‐1 – The Spillover Coefficient p – Index of Risk aversion Xt – debt/GDP and fiscal balance debt/GDP and fiscal balance
Yt Yt 1 ' X t t
t2 t21 t21 • Estimated by Maximum Likelihood 46
Estimation: Main Results GER Mean equation: Constant Lag dep.variable IRA SC Overall balance Debt ratio
-0.122*** (0 025) (0.025) 0.933*** (0.006) -0.217*** (0 034) (0.034) 0.083*** (0.026) -0.002*** (0 001) (0.001) 0.000 (0.000)
Variance V i equation: i Constant 0.000*** (0.000) ARCH term 0.381*** (0 034) (0.034) GARCH term 0.702*** (0.017) R-squared d No. of observations
0.981 0 981 1,194
FRA
ITA
SPA
NET
-0.042** (0 019) (0.019) 0.971*** (0.005) -0.202*** (0 032) (0.032) 0.176*** (0.034) -0.002*** (0 001) (0.001) 0.001** (0.000)
0.039 (0 030) (0.030) 0.981*** (0.005) -0.046 (0 038) (0.038) 0.137*** (0.035) -0.002*** (0 001) (0.001) 0.001* (0.000)
-0.140*** (0 013) (0.013) 0.943*** (0.008) -0.074*** (0 027) (0.027) 0.268*** (0.037) -0.000 (0 000) (0.000) 0.002*** (0.000)
-0.077*** (0 015) (0.015) 0.966*** (0.005) -0.118*** (0 033) (0.033) 0.154*** (0.030) -0.000 (0 000) (0.000) 0.000** (0.000)
0.000*** (0.000) 0.095*** (0 007) (0.007) 0.918*** (0.005)
0.000*** (0.000) 0.074*** (0 007) (0.007) 0.933*** (0.005)
0.000*** 0.000*** (0.000) (0.000) 0.283*** 0.089*** (0 023) (0.023) (0 009) (0.009) 0.789*** 0.926*** (0.014) (0.006)
0.984 0 984 1,194
0.993 0 993 1,194
0.992 0 992 1,194
0.986 0 986 1,194
47
Estimation: Main Results (2) BEL
Mean equation: Constant
GRE
IRE
POR
-0.028 (0 034) (0.034) 0.962*** (0.006) -0.098*** (0 032) (0.032) 0.177*** (0.025) -0.002*** (0.000) 0.000 (0.000)
-0.258*** (0 035) (0.035) 0.956*** (0.006) -0.052 (0 035) (0.035) 0.579*** (0.073) -0.001* (0.000) 0.002*** (0.000)
-0.059* (0 030) (0.030) 1.008*** (0.004) -0.053** (0 024) (0.024) 0.183*** (0.037) 0.002*** (0.000) 0.000 (0.000)
Variance equation: Constant 0.000*** 0.000*** (0.000) (0.000) ARCH term 0 113*** 0.076 0.113 0 076*** (0.011) (0.007) GARCH term 0.913*** 0.935*** (0.008) (0.005)
0.000*** (0.000) 0 066*** 0.066 (0.005) 0.945*** (0.004)
0.000*** 0.000*** (0.000) (0.000) 0 289*** 0.150 0.289 0 150*** (0.023) (0.010) 0.800*** 0.885*** (0.011) (0.005)
Lag dep. variable IRA SC Overall balance Debt ratio
-0.090*** (0 016) (0.016) 0.953*** (0.007) -0.097*** (0 032) (0.032) 0.278*** (0.024) -0.000 (0.001) 0.000* (0.000)
AUT
R-squared 0.991 No. of observations 1,194
0.993 1,194
0.996 1,194
0.998 1,194
-0.018 (0 020) (0.020) 0.950*** (0.007) -0.316*** (0 035) (0.035) 0.631*** (0.059) 0.000 (0.000) -0.002*** (0.000)
0.993 1,194
48
Global Risk Aversion, Contagion or Fundamentals? Contributions to 10-year Swap Spread 3.0 2.5 2.0 15 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 20 -2.0 -2.5
Germany
Jul 07 Sep 08 Oct 08 Mar 09 Fundamentals
Contagion
Apr 09 Sep Oct 09 Feb 10 09 Global Risk Aversion
49
Global Risk Aversion, Contagion or Fundamentals? Contributions to 10-year Swap Spread 8.0
Italy
6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0 -8.0 Jul 07 Sep 08 Oct 08 Mar Apr 09 Sep Oct 09 Feb 10 09 09 Fundamentals Contagion g Global Risk Aversion
50
Global Risk Aversion, Contagion or Fundamentals? Contributions to 10-year Swap Spread 250.0 200.0 150.0 100.0 50.0 0.0 -50.0 50.0 -100.0 -150.0 -200.0 -250.0 250 0 -300.0
Greece
Jul 07 Sep 08 Oct 08 Mar 09 Fundamentals
Contagion
Apr 09 Sep Oct 09 Feb 10 09 Global Risk Aversion
51
Global Risk Aversion, Contagion or Fundamentals? Contributions to 10-year Swap Spread 20.0
Ireland
15.0 10.0 5.0 0.0 -5.0 -10.0 Jul 07 Sep 08 Oct 08 Mar 09 Fundamentals
Contagion
Apr 09 Sep Oct 09 Feb 10 09 Global Risk Aversion
52
Estimation: Index of Risk Aversion • When Risk Aversion rises, swap spreads widen, as sovereign yields fall further below swap yields (flight‐to‐quality leads to i ld f ll f th b l i ld (fli ht t lit l d t capital flowing away from risky assets). • Not the case for some high‐debt, lower‐rated issuers (notably Greece and Italy) that do not benefit from flight‐to‐quality. • Outside the euro area, the US and the UK benefit from rising Risk Aversion. For Sweden (and to some extent Japan), risk aversion is ( ) not significant. • The impact of the IRA on US swap spreads is larger than the impact on the spreads for the other three countries (US is a “safe p p haven”).
53
Estimation: Contagion and Fundamentals • Contagion: sovereign bond yields rise when the probability of a credit event rises (because of contagion from another sovereign issuer). • High‐debt, lower‐rated sovereigns exhibit larger sensitivities as these countries are vulnerable to even remote probabilities of distress among higher‐rated issuers. • Fundamentals : significant relationship with sovereign spreads. Wh b d t d fi it i When budget deficits increase, sovereign bond yields rise i b d i ld i (versus swap yields). 54
But Contagion from Where? • Contagion can further broken down across sources of g distress. • For a given probability of distress in a specific sovereign, we find the most significant sources of contagion. • Our measure of distress dependence (contributions to the changes in each country’s SC) taken from the Contagion Matrices.
55
Contagion from Where? Systemic outbreak (1) – October 2008‐March 2009: countries weighing adversely on other sovereigns were those whose financial institutions were hit hard by the sovereigns were those whose financial institutions were hit hard by the financial crisis (Austria, Ireland, and Italy). USA JPN UK GER FRA ITA SPA NET BEL AUT GRE IRE O POR SWE AVG
USA JPN 5.2 3.7 3.7 5.3 55 5.5 48 4.8 4.1 5.1 3.5 5.3 34 3.4 51 5.1 4.1 5.0 3.6 5.5 39 3.9 58 5.8 3.4 5.9 3.8 5.9 3.2 5.6 4.1 5.1 3.5 5.0
UK 8.3 8.4 8.8 8 8 8.8 9.2 89 8.9 9.4 8.9 91 9.1 8.6 9.1 8.5 9.4 8.3
GER FRA ITA 7.2 5.6 8.1 4.5 5.0 8.9 5.1 5.4 9.7 65 6.5 88 8.8 6.2 8.7 4.9 5.1 52 5.2 57 5.7 92 9.2 5.4 6.2 9.7 5.1 5.5 8.9 46 4.6 4 8 10 4.8 10.4 4 4.4 4.7 9.7 4.5 4.5 10.2 4.9 5.4 8.6 5.2 5.7 9.7 4.6 4.9 8.7
SPA NET BEL AUT GRE IRE POR SWE 6.4 8.4 6.7 11.4 7.6 11.0 5.2 9.0 6.8 7.3 7.1 12.2 9.4 12.2 6.5 7.9 7.4 8.6 7.2 11.9 8.6 11.8 6.1 9.2 74 7.4 84 8.4 7 2 10.3 7.2 10 3 7.5 7 5 10.0 10 0 6.1 61 87 8.7 7.8 9.2 7.3 10.2 7.6 9.4 6.4 9.1 7.3 8.4 6.9 12.9 9.2 12.5 5.9 9.0 85 8.5 6 7 12.4 6.7 12 4 8.7 8 7 11.5 11 5 5.8 58 92 9.2 7.8 7.9 10.7 8.0 10.6 6.4 8.9 6.8 8.8 12.0 8.5 11.5 5.9 9.0 79 7.9 75 7.5 75 7.5 9 8 13.5 9.8 13 5 7.0 70 83 8.3 7.3 7.3 7.0 12.8 14.3 6.5 8.1 7.4 7.5 7.2 13.8 11.0 6.5 8.5 6.8 8.1 6.7 12.7 9.0 11.6 8.7 7.9 8.3 7.5 11.2 8.3 11.3 6.4 6.8 7.5 6.6 11.1 8.2 10.9 5.8 8.0 56
Conclusions • Readily implementable in terms of data needs. • Reliable in the sense of being robust under data‐restricted environments. • Interpretable, so that the approach itself and its output can be used as an input to policy development. • Incorporates distress dependence among FIs and its changes across the economic cycle. • Includes all the relevant IFs and Sectors. • Framework Framework that produces complementary measures in a consistent that produces complementary measures in a consistent manner. • Integrates Integrates complementary information (micro‐founded supervisory data complementary information (micro‐founded supervisory data and market‐based).
57
Conclusions (ctd.)
• Debt sustainability and appropriate management of sovereign balance sheets are necessary conditions for preventing sovereign risk from feeding back into broader financial stability concerns. • Rising sovereign risk requires credible medium‐term fiscal consolidation plans as well as a solid public debt management framework. • Emphasis should be given to the presence of significant contingent risk on sovereign balance sheets and the need for sovereigns to gradually disengage sovereign balance sheets and the need for sovereigns to gradually disengage from a number of measures supporting the financial sector. • Immediate steps should be taken to reduce the possibility of projecting longer‐ term sovereign credit risks into short‐term financing concerns. 58
References • •
• • • • • • •
Afonso A. and Strauch, R. (2004). “Fiscal Policy Events and Interest Rate Swap Spreads: Evidence from the EU”, ECB Working Paper, No.303. Athanosopoulou, M., Segoviano, M., and Tieman A., (2011), “Banks’ Probability of Default: Which Methodology, When, and Why?”, IMF Working Paper (forthcoming). Bernoth K., von Hagen, J. and Schuknecht, L. (2004), “Sovereign Risk Premia in the European Government Bond Market”, ECB Working Paper, No. 369. Cáceres, C., Guzzo, V., Segoviano, M., (2010), “Sovereign Spreads: Global Risk Aversion, Contagion or Fundamentals?”, IMF Working Paper WP/10/120. Cochrane, J. (2001). Asset Pricing, Princeton: NJ, Princeton University Press Codogno, L., Favero, C. and Missale, A. (2003). “Yield Codogno, L., Favero, C. and Missale, A. (2003). Yield spreads on EMU spreads on EMU governments bonds”, Economic Policy, October, 505‐32. Afonso A. and Strauch, R. (2004). “Fiscal Policy Events and Interest Rate Swap Spreads: Evidence from the EU”, ECB Working Paper, No.303. Spreads: Evidence from the EU ECB Working Paper No 303 Bernoth K., von Hagen, J. and Schuknecht, L. (2004), “Sovereign Risk Premia in the European Government Bond Market”, ECB Working Paper, No. 369. Espinoza R and Segoviano M (2011) “Probabilities of Default and the Market Espinoza, R. and Segoviano, M. (2011). “Probabilities of Default and the Market Price of Risk in a Distressed Economy”, IMF Working Paper, (forthcoming).
59
References • • • •
• •
• •
Geyer, A., Kossmeier, S., and Pichler, S. (2004). “Measuring systematic risk in EMU government yield spreads”, Review of Finance, 8, 171‐97. Goodhart, C., Hofmann, B. and Segoviano, M. (2004), “Bank Goodhart, C., Hofmann, B. and Segoviano, M. (2004), Bank Regulation and Regulation and Macroeconomic Fluctuations,” Oxford Review of Economic Policy, Vol. 20, No. 4, pp. 591–615. Goodhart, C., Hofmann B., and Segoviano M., (2006), “Default, Credit Growth, and Asset Prices”, IMF Working Paper 06/223. and Asset Prices IMF Working Paper 06/223 Manganelli S. and Wolswijk, G. (2007). “Market Discipline, Financial Integration and Fiscal Rules: What Drives Spreads in the Euro Area Government Bond M k t?” ECB W ki P Market?”, ECB Working Paper, No. 745. N 745 Mody A. (2009). “From Bear Stearns to Anglo Irish: How Eurozone Sovereign Spreads Related to Financial Sector Vulnerability”, IMF Working Paper, 09/108 Schuknecht, L., von Hagen, J., and Wolswijk, G. (2010). “Government bond risk premiums in the EU revisited. The impact of the financial crisis”, ECB Working Paper, No. 1152. Segoviano, M. (2006). “Consistent Information Multivariate Density Optimizing Methodology”. Financial Markets Group, Discussion Paper No. 557. Segoviano, M. and Goodhart, C. (2009). “Banking Stability Measures”, IMF Working Paper, 09/4. 60
References • • • •
Segoviano, M., (2006), “The Conditional Probability of Default Methodology,” Financial Markets Group, London School of Economics, Discussion Paper 558. Financial Markets Group, London School of Economics, Discussion Paper 558. Segoviano, M., (2011), “The CIMDO‐Copula. Robust Estimation of Default Dependence under Data Restrictions”, IMF Working Paper (forthcoming). Segoviano, M. and Padilla, P., (2006), “Portfolio Credit risk and Macroeconomic Sh k A li i Shocks: Applications to Stress Testing under Data Restricted Environments,” IMF S T i d D R i dE i ” IMF Working Paper 06/283. Sgherri, S. and Zoli, E. (2009). “Euro Area Sovereign Risk During the Crisis”, IMF Working Paper, 09/222. g p , /
61