Interbank Networks, Contagion, and Common Shocks
Transcript of Interbank Networks, Contagion, and Common Shocks
Interbank Networks, Contagion, and Common Shocks
Co-Pierre Georg
Interdisciplinary Group of Complex SystemsUniversidad Carlos III de Madrid
and
Graduate School “Foundations of Global Financial Markets”Friedrich-Schiller-Universitat Jena
CABDyN Complexity Center, Saıd Business School, Oxford UniversityOxford, 02/07/2012
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 1 / 28
Motivation
The financial system has become increasingly interconnected and complex
Supervision of individual financial institutions insufficient⇒ Network structure of interconnections matters
Systemic risk takes various forms and is highly dynamic⇒ Better understanding needed to safeguard financial stability
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 2 / 28
Motivation
The financial system has become increasingly interconnected and complex
Supervision of individual financial institutions insufficient⇒ Network structure of interconnections matters
Systemic risk takes various forms and is highly dynamic⇒ Better understanding needed to safeguard financial stability
This talk: agent-based models and network theory to address thesequestions
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 2 / 28
A Simple Bank Balance Sheet
Liabilities Assets
D
L
LC
BC
I
E
F
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 3 / 28
The Interconnectedness of the Financial System Increased
Jan 1999 Jan2001 Jan2003 Jan2005 Jan2007 Jan2009 Jan2011
34
56
7
Date
Volu
me
[Tri
llio
n E
UR
]
Figure: Deposit liabilities of euro area MFIs vs. other euro area MFIs, outstandingamounts at the end of the period, neither seasonally nor working day adjusted. Source:ECB Statistical Data Warehouse.
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 4 / 28
The Interconnectedness of the Financial System Increased
Figure: Left: Largest links on e-MID on 3 January 2007. Source: Gabrieli (2010). Right:Net pairwise correlation of stocks on 16 September 2008. Source: Diebold and Yilmaz(2011).
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 5 / 28
The Interconnectedness of the Financial System Increased
Figure: Left: Cross-border debt assets (from CPIS and BIS locational). Source: Kubelecand Sa (2010). Right: Cross-border banking (BIS locational). Source: Minoiu and Reyes(2011).
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 6 / 28
The Interconnectedness of the Financial System Increased
1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
010
020
030
040
050
060
0
Date
Volu
me
[Tri
llio
n U
SD
]
Foreign exchangeInterest rateEquity−linkedCommodityCDS
Figure: Global over-the-counter derivatives markets, notional amounts of contractsoutstanding. Source: IMF
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 7 / 28
Indirect Linkages Amplify the Risk of Fire-sales
Figure: U.S. Mortgage-Related Securities Issuance. Source: Gorton and Metrick (2010)
.
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 8 / 28
Indirect Linkages Amplify the Risk of Fire-sales
Figure: Forced Sales Discounts and Time Between Sale and Event. Source: Campbell,Giglio and Pathak (2012)
.
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 9 / 28
Do We Need Yet Another Paper?
Literature on Financial Networks
Allen and Gale (2000), Freixas et al. (2000)
Haldane and May (2011), Gai et al. (2011), Gai and Kapadia (2008)
Becher et al. (2008), Gabrieli (2011), Chang et al. (2008), Brink and Georg(2011), Markose et al. (2010)
Literature on Fire-sales
Shleifer and Vishny (1992): specialised asset holders are simultaneously indistress and sell to non-specialists
Allen and Gale (1994): endogenous market participation
Literature on Multi-Agent Models:
Iori et al. (2006), Nier et al. (2007)
However: risk-free investments, no central bank, mechanistic agentbehaviour, “fine-tuning”
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 10 / 28
What is Systemic Risk?
Definition by impact
FSB definition: “a risk of disruption to financial services that is (i) caused byan impairment of all or parts of the financial system and (ii) has the potentialto have serious negative consequences for the real economy.”
Definition by cause
B1
B2 B3 A1
A2
B1
B2
⇒
B1
B2
Figure: Left: direct connections (counterparty risk, contagion). Right: indirectconnections (common shocks, fire-sales).
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 11 / 28
Focus of this Paper
Complete dry-up of interbank markets in September 2008, central banks wereforced to unprecedented non-standard measures
⇒ Q1: Can central banks stabilize interbank markets?
Systemic risk requires macroprudential oversight in addition tomicroprudential supervision
⇒ Q2: What are robust network structures?
Different forms of systemic risk can act differently on the financial system⇒ Q3: What are the optimal policy responses?
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 12 / 28
Microfoundations of Banks Determine Model
Liabilities Assets
DHouseholds
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 13 / 28
Microfoundations of Banks Determine Model
Liabilities Assets
D
LC
Households
CentralBank
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 13 / 28
Microfoundations of Banks Determine Model
Liabilities Assets
D
LC
I
E
Households
CentralBank
Firms
CentralBank
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 13 / 28
Microfoundations of Banks Determine Model
Liabilities Assets
D
L
LC
I
E
L
Households
CommercialBanks
CentralBank
Firms
CentralBank
CommercialBanks
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 13 / 28
Microfoundations of Banks Determine Model
Liabilities Assets
D
L
LC
BC
I
E
L
Households
CommercialBanks
CentralBank
Firms
CentralBank
CommercialBanks
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 13 / 28
Microfoundations of Banks Determine Model
Liabilities Assets
D
L
LC
BC
I
E
L
Households
CommercialBanks
CentralBank
Firms
CentralBank
CommercialBanks
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 13 / 28
Agent Behaviour has to be Motivated
Banks optimize their portfolio structure and -volume according to CRRApreferences
u =1
1− θ
(
V (1 + λµ−1
2θλ2σ2)
)(1−θ)
where θ is risk-aversion parameter
Update algorithm for k = 1, . . . ,N banks and t = 1, . . . , τ update steps:
1 Obtain returns on investments, pay interest on deposits
2 Deposit in- and out-flows, required reserves
3 Settle interbank loans
4 Determine new investment level
5 Settle liquidity position
6 Pay dividends
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 14 / 28
Microfoundations of Banks Determine Model
λ
Liabilities Assets
D
L
LC
BC
I
E
L
Households
CommercialBanks
CentralBank
Firms
CentralBank
CommercialBanks
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 15 / 28
Interbank Loans Form a Network Structure
Figure: Different scale free networks
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 16 / 28
Model Parameters
Sektor Parameter
Households deposit fluctuations γFirms credit success probability pf , realized
credit return (ρ+f , ρ−
f )Commercial banks deposit interest rate rd , dividend level βk ,
expected credit success probability pb, expected creditreturn (ρ+b ,ρ
−
b ), risk aversion parameter θCentral bank main refinancing rate rb, minimum reserve
requirement r , quality of securities αk
Network parameters
number of banks N, level of interbank connections connLevel
Simulation parameters
number of update steps τ , number of simulations numSimulations
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 17 / 28
Central Bank Liquidity Stabilizes in the Short-Run...
0 250 500 750 1000
00
20
40
60
80
100
120
140
Time t
# of
Act
ive
Ban
ks
cbActivity=1.0cbActivity=0.8cbActivity=0.5cbActivity=0.45cbActivity=0.4cbActivity=0.2cbActivity=0.0
Figure: The effect of central bank activity αk on financial stability in a crisis scenario
(ρ+f = 0.09, ρ−f = −0.08)
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 18 / 28
...but the Effect is Non-Monotonic
0 250 500 750 1000
00
20
40
60
80
100
120
140
Time t
Inte
rban
k lo
an v
olum
e L
cbActivity=1.0cbActivity=0.8cbActivity=0.5cbActivity=0.45cbActivity=0.4cbActivity=0.2cbActivity=0.0
Figure: The effect of central bank activity αk on interbank liquidity in a crisis scenario
(ρ+f = 0.09, ρ−f = −0.08)
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 19 / 28
Central Bank Liquidity Stabilizes in the Short-Run
Lesson 1:
Central bank liquidity provision has non-linear effect on financial stability⇒ Close threshold value, small changes have significant impact⇒ Away from threshold value, even large changes can be ineffective
Stabilizing effect in the short-run only
Abundant central bank liquidity crowds out interbank liquidity
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 20 / 28
Some Network Structures are More Resilient Than Others
0 250 500 750 1000
00
20
40
60
80
100
120
140
Time t
# of
Act
ive
Ban
ks
connLevel=1.0connLevel=0.8connLevel=0.6connLevel=0.4connLevel=0.2connLevel=0.0
Figure: The impact of the network topology on financial stability in a normal scenario(ρ+f = 0.09, ρ−f = −0.05) in a random network.
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 21 / 28
Some Network Structures are More Resilient Than Others
0 250 500 750 1000
00
20
40
60
80
100
120
140
Time t
# of
Act
ive
Ban
ks
connLevel=1.0connLevel=0.8connLevel=0.6connLevel=0.4connLevel=0.2connLevel=0.0
Figure: The impact of the network topology on financial stability in a crisis scenario(ρ+f = 0.09, ρ−f = −0.08) in a random network
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 22 / 28
Some Network Structures are More Resilient Than Others
0 250 500 750 1000
00
20
40
60
80
100
120
140
Time t
# of
Act
ive
Ban
ks
m=1m=2m=4m=10
Figure: The impact of the network topology on financial stability in a crisis scenario(ρ+f = 0.09, ρ−f = −0.08) in a BA network
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 23 / 28
Some Network Structures are More Resilient Than Others
0 250 500 750 1000
00
20
40
60
80
100
120
140
160
Time t
Inte
rban
k lo
an v
olum
e L
connLevel=1.0connLevel=0.8connLevel=0.6connLevel=0.4connLevel=0.2connLevel=0.0
Figure: The impact of the network topology on interbank liquidity in a crisis scenario(ρ+f = 0.09, ρ−f = −0.08) in a random network
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 24 / 28
Some Network Structures are More Resilient Than Others
Lesson 2:
Network structure matters in crises
Relationship between financial stability and interconnectedness in randomnetworks is non-monotonic
Scale-free networks tend to be more stable than random networks
Interbank networks are robust-yet-fragile⇒ Size of endogenous fluctuations matter
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 25 / 28
Different Forms of Systemic Risk Require Different Answers
0 250 500 750 1000
00
20
40
60
80
100
120
140
Time t
# of
Act
ive
Ban
ks
Interbank contagionCommon shock ACommon shock B
Figure: The impact of different forms of systemic risk on financial stability in a crisisscenario (ρ+f = 0.09, ρ−f = −0.08) in a random network (connLevel=0.8)
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 26 / 28
Different Forms of Systemic Risk Require Different Answers
0 250 500 750 1000
20
40
60
80
100
120
140
160
20
40
60
80
100
120
140
160
Time t
Inte
rban
k lo
an v
olum
e L
Interbank contagionCommon shock ACommon shock B
Figure: The impact of different forms of systemic risk on financial stability in a crisisscenario (ρ+f = 0.09, ρ−f = −0.08) in a random network (connLevel=0.8)
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 27 / 28
Different Forms of Systemic Risk Require Different Answers
Lesson 3:
Common shocks can pose greater threat to financial stability
Contagion mainly reduces liquidity available in the system
Common shock mainly reduces banking capital and increases (relative) sizeof endogenous fluctuations
⇒ Different optimal responses, for different forms of systemic risk
⇒ Implications for financial Regulation
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 28 / 28
Different Forms of Systemic Risk Require Different Answers
Lesson 3:
Common shocks can pose greater threat to financial stability
Contagion mainly reduces liquidity available in the system
Common shock mainly reduces banking capital and increases (relative) sizeof endogenous fluctuations
⇒ Different optimal responses, for different forms of systemic risk
⇒ Implications for financial Regulation
Thank you!
Co-Pierre Georg (UC3M & Jena) Interbank Networks, Contagion, and Common Shocks Oxford, 02/07/2012 28 / 28
References
Francis Allen and Douglas Gale. Financial contagion. Journal of Political Economy, 108:1–33,2000.
Christopher Becher, Stephen Millard, and Kimmo Soramaki. The network topology of chapssterling. Bank of England working papers 355, Bank of England, 2008.
Nicola Brink and Co-Pierre Georg. Assessing systemic risk in the south african interbank market.Special Note in the Financial Stability Review April, South African Reserve Bank, 2011.
Eui Jung Chang, Eduardo Jose Araujo Lima, Solange M. Guerra, and Benjamin M. Tabak.Measures of interbank market structure: An application to brazil. Brazilian Review of
Econometrics, 28(2):163–189, 2008.
Silvia Gabrieli. The microstructure of the money market before and after the financial crisis: anetwork perspective. CEIS Research Paper 181, Tor Vergata University, CEIS, 2011.
Prasanna Gai and Sujit Kapadia. Contagion in financial networks. Discussion Paper 1231, Bankof England, 2008.
Prasanna Gai, Andrew Haldane, and Sujit Kapadia. Complexity, concentration and contagion.Journal of Monetary Economics, 58(4), 2011.
Andrew G. Haldane and Robert M. May. Systemic risk in banking ecosystems. Nature, 469:351–355, 2011.
G. Iori, S. Jafarey, and F.G. Padilla. Systemic risk on the interbank market. Journal of Economic
Behavior & Organization, 61(4):525–542, 2006.
Erland Nier, Jing Yang, Tanju Yorulmazer, and Amadeo Alentorn. Network models and financialstability. Journal of Economic Dynamics and Control, 31:2033–2060, 2007.