Illin T, Varga L (2014) An actionable definition of systemic risk in financial services validated using data from the Icelandic financial system failure.
An actionable definition of systemic risk in financial services validated using data from the Icelandic financial system failure
1. Ilin, T. and Varga, L. (2014), The uncertainty of systemic risk
Thomas Ilin, PhD student, Email: thomas.ilin@cranfield.ac.uk,
Dr Liz Varga, Principal Research Fellow, Director Complex Systems Research Centre, Email: liz.varga@cranfield.ac.uk
www.cranfield.ac.uk/som
An actionable definition of systemic risk in financial services
validated using data from the Icelandic financial system failure
$
PARTICIPANT
(regulated)
$
PARTICIPANT
(unregulated)
$
CENTRAL
AUTHORITY
FINANCIAL
SERVICE
PRODUCTS
LOAN
DEPOSIT
DEBT
SETTLEMENT
SECURITY
FUND
$$
SOVEREIGN
GDP
% Treasury Budget
LENDING (LL) / BORROWING (LB)
TAKING (DT) / PLACING (DP)
ISSUING (DI) / HOLDING (DH)
RETIRING (DR) / DISCHARGING (DD)
TENDERING (ST) / ACQUIRING (SA)
RECEIVING (FR)
Activities
Regulated
Participant Data
Profit
Assets
Liquid-Assets
Liabilities
Fines
… etc
LENDING (LL) / BORROWING (LB)
TAKING (DT) / PLACING (DP)
ISSUING (DI) / HOLDING (DH)
RETIRING (DR) / DISCHARGING (DD)
TENDERING (ST) / ACQUIRING (SA)
RECEIVING (FR)
Activities
Unregulated
Participant Data
Profit
Assets
Liquid-Assets
Liabilities
… etc
PROVIDING (FP)
Activities
LL / LB, DT / DP
Fine Policy % of Assets
Central Authority Data
Fine Rate
Ratios
Interest Rates
Liquid-Assets
… etc
FP / FR FP / FR
RATING AGENCY
Strategy:
MAX-REGULATION
MIN-REGULATION
Strategy:
GROW-PROFITS
GROW-ASSETS
REDUCE-LIABILITIES
MAINTAIN-LIQUIDITY
BALANCE
Strategy:
GROW-PROFITS
GROW-ASSETS
REDUCE-LIABILITIES
MAINTAIN-LIQUIDITY
BALANCE
Fine Policy
Strategy:
PROVIDE-LIQUIDITY
DO-NOTHING
Product Data
Function
Holdings Total
Bids Open Total
Offers Open Total
… etc
Fine
SIMULATION MODEL CONTENTS:
COUNTRY (x1)
SOVEREIGN (x1)
CENTRAL AUTHORITY (x1)
RATING AGENCY (x1)
FOREIGN PARTICIPANT (x1, sized as a multiple of native participants)
PARTICIPANT - regulated (x1 < configurable <= x100)
PARTICIPANT - unregulated (x1 < configurable <= x100)
SIFS (2 per 6 functions = 12)
% GDP
Sovereign Data
Treasury Budget
Fine Policy
Country-GDP
.. etc
$
FOREIGN
PARTICIPANT
Acquisition
DI / DH
DR / DD
ST / SA INTERNATIONAL
MARKET
DI / DH
DR / DD
ST / SA
KEY:
Reference
Value flow
Participation
Detail
$
FOREIGN
PARTICIPANT
Acquisition
Regulatory
Budget
Financial service activity-types
(SIFS Supply / Demand)
Financial service activity-types
(SIFS Supply / Demand)
Macroeconomic
events
Distress is
created by
exogenous
causes, and
operationalized in
the system.
Systemic behaviour is manifested as
external effects, sometimes causing
further distress.
SIPs – Systemically important Participants (e.g. banks, intermediaries, counterparties).
SIFS – Systemically important Financial Services (e.g. short-term funding from money-markets).
* Behaviour can also be: Contagious, Dispersive, Convergent, Expansive, Divergent, Optimal.
Global Financial System
operations
In response to increasing distress,
local operational efforts of SIPs
become focused on certain SIFS
(e.g. getting short-term funding).A
B
effects
causes
F
Distress is propagated
system-wide as problems in
the execution-level activities
of SIPs in overall supply vs
demand for certain SIFS. (e.g.
lack of short-term funding).
Failed*
D
E Which becomes emergent
operational behaviour.
C
Distress is increased by
endogenous causes within the
system’s operations.
stability
instability
effects of
instability
Purpose of research: to explain phenomena that emerge in
the operational behaviour paradigm of the global financial
system (Fig 1).
Outcomes:
• a metric of systemic risk of failure (Fig. 2)
• validation using data from the Icelandic financial crisis
2000-2009 (Fig. 3)
• multidisciplinary theory of systemic risk using a cusp-
catastrophe type (static 3d) model (Fig. 4a for the surface
and Fig. 4b with base from Fig. 4a and time in the z axis)
• a dynamical complex system model responding to
collective participation behaviour (Fig. 5a for conceptual
model and Fig. 5b for the agent based model )
Fig. 1 Operational behaviour paradigm of systemic
failure
Source: Extracted from the reported accounts of all 51 financial institutions in the Iceland database of Bankscope.
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009
tot_profit_act 1,656,700 4,312,100 8,403,100 2,276,100 6,020,600 3,549,845,139 239,599,662 206,412,248 -1,953,841,290 -607,881,900
tot_assets_act 451,393,500 924,935,100 981,676,800 565,522,000 2,939,201,500 15,999,713,336 7,811,197,543 10,435,219,325 4,996,139,748 3,887,703,500
tot_liabilities_act 468,318,500 939,790,100 995,786,700 570,690,500 3,044,863,900 16,654,898,860 8,383,680,300 11,465,867,885 5,574,049,530 4,318,728,400
tot_liquid_act 231,931,800 330,242,100 363,716,300 103,713,500 691,861,200 1,525,669,113 2,581,366,364 4,860,068,189 2,374,534,591 1,734,561,200
tot_securities_act 53,084,900 74,062,800 67,007,000 17,630,800 407,113,000 12,681,175,749 1,712,223,256 2,715,050,572 1,093,228,923 774,386,700
tot_debt_act 202,975,600 557,257,400 572,129,000 437,710,500 1,951,623,900 7,215,230,684 4,429,202,641 4,835,106,317 3,480,337,873 3,811,485,100
tot_loans_act 0 0 0 0 250,700,200 354,177,740 934,168,278 999,915,622 332,978,327 323,283,500
tot_deposits_act 92,725,900 138,312,400 136,906,900 5,807,700 136,177,400 342,577,782 511,286,489 1,081,272,784 688,948,932 362,818,900
-5,000,000,000
0
5,000,000,000
10,000,000,000
15,000,000,000
20,000,000,000
ISKthousands
National Systemic Failure Summary - Iceland Actual Financials
(total levels among all operational participants in each year)
Foreign
acquisitions.
High ratio and level
of assets in traded
international
securities.
Relief Euphoria Frustration Euphoria Relief
Prolific new
issuance of long-
term debt securities
on foreign markets.
Banking system collapsed.
State restructured remaining
banks and allowed them to
default on their external debt.
Liberalisation of the
Icelandic financial sector
and privatisation of
domestic banks
completed.
Negative reports by rating agencies limited
access to international securities markets.
Fear
Datapoints:
Sentiment:
1 2 3 4 5 6
Fig. 3 Iceland financial system failure analysis (2000-2009)
What is systemic risk?
Systemic risk is the risk of a
systemic event in a system that
produces an altered or damaged
transitional system that is
functionally impeded, which in
the extreme may no longer be
capable of functioning (authors’
summary of Zigrand, 2014).
Fig. 4b A cusp catastrophe-type model of the global financial system
Fig. 5a Conceptual model of agent-based model
So what
The method proposed here offers a way of diagnosing when the global financial system is approaching a state of operational
crisis, and understanding how that outcome could generally be avoided.
Impact
Mitigating falls in lifetime income of working age adults of $150,000 on average estimated for the 2008 global financial crisis
Potentially
Catastrophic
Systemic
Fidelity
Systemic
Failure
set
Divergent
Contagious
Optimal
Contagious Dispersive
Dispersive
DivergentOptimal
0%
When shifts in the
system’s operational
state over time are
projected from the
three-dimensional
surface of behaviour
Bt onto this two-
dimension control
surface of focus F,
and then are
described by
extensions to the
concepts of
divergence and
hysteresis from
catastrophe theory,
they provide a
categorisation of
operational behaviour.
Hysteresis
y
100%
% of systemically
important
financial services
(SIFS) that are
the focus of
concentrations in
supply intentions
x
100%
% of systemically important
financial services (SIFS) that are
the focus of concentrations
in demand intentions
( )fail
F
( )F
The two-dimensional control surface F of overall focus
on intended supply relative to demand
Fig. 5b Part of agent-based model dashboard (time on
horizontal axis) showing phenomena/system level
demand, supply and satisfaction
100%
0%
y
x
100%
z
The two-dimensional
control surface Ft of overall
operational focus on intended
supply relative to demand
The three-dimensional surface Bt
of all previous and predicted
operational states at time t. This
surface represents the operational
behaviour topology, in which each
coordinate point (x, y, z) =bt∈Bt is a
potential overall operational
effectiveness at time t. Then a current
operational state at t is defined by a
single point on this surface and a state
category derived from its placement in
a region of Bt.
Bifurcation set
(shadow of the fold
on the control surface), which is
also a catastrophic contagion set
Systemic
Failure
set
Singularity
Cusp
(can fold
either way)
Progressive
contagion
set for y
(supply)
Cross-sectional
cut in surface Bt
Catastrophic
failure of the
system
Systemic risk
mitigation effect
Progressive
contagion
set for x
(demand)
100%
% of all available systemically important
financial services that achieve a
minimum level of overall supply
satisfaction of demand