A New Approach to Predict Emerging Risks Using Risk DNA Model Motoharu Dei, FIAJ, Consultant Milliman Inc. IAJ, Open Discussion Forum ERM Section December 7 th , 2012
May 29, 2015
A New Approach to Predict Emerging Risks Using Risk DNA Model
Motoharu Dei, FIAJ, Consultant Milliman Inc.
IAJ, Open Discussion Forum ERM SectionDecember 7th, 2012
Introduction
Reference Material for Today’s Presentation
Management Board of the UK Actuarial Profession invited proposals for a funded external research, and this report was awarded. “A review of the use of complex systems
applied to risk appetite and emerging risks in ERM practice”
Authors (from Milliman, the Universities of Bristol and Bath Systems Centre) N. ALLAN N. CANTLE P. GODFREY Y. YIN http://www.actuaries.org.uk/research-and-
resources/documents/review-use-complex-systems-applied-risk-appetite-and-emerging-risks
System Thinking
The methods discussed in the report are based on “System Thinking”. “Concept Mapping”, “Systems Dynamics Modelling”, “Chaos
Theory”, “Fuzzy Theory”, “Neural Networks”, “Genetic Algorithms”, “Phylogenetic Analysis”, Bayesian Network”, “Cellular Automata”, “Agent Based Modelling”, ”Network Theory”
Input
Output
Input
OutputInput
Output
System Thinking
The methods discussed in the report are based on “System Thinking”. “Concept Mapping”, “Systems Dynamics Modelling”, “Chaos
Theory”, “Fuzzy Theory”, “Neural Networks”, “Genetic Algorithms”, “Phylogenetic Analysis”, Bayesian Network”, “Cellular Automata”, “Agent Based Modelling”, ”Network Theory”
Input
Output
Input
Output
System Thinking
The methods discussed in the report are based on “System Thinking”. “Concept Mapping”, “Systems Dynamics Modelling”, “Chaos
Theory”, “Fuzzy Theory”, “Neural Networks”, “Genetic Algorithms”, “Phylogenetic Analysis”, Bayesian Network”, “Cellular Automata”, “Agent Based Modelling”, ”Network Theory”
In biology, phylogenetics (/faɪlɵdʒɪˈnɛtɪks/) is the study of evolutionary relation among groups of organisms (e.g. species, populations), which is discovered through molecular sequencing data and morphological data matrices. (Wikipedia)
Input
Output
Input
Output
Phylogenetic Approach
Emerging Risk
We should not guess in advance what we expect to see in identifying the emerging risk
Many attempts to monitor risk registers throw that away at outset Need a “model-free” approach to see emergence
Allow people to “mix” colours
Phylogenetic Approach
This approach was also presented in the SOA annual meeting in this October (Session 99 – Emerging Risk) by Neil Cantle.
The core idea is to adapt the same analysis which researchers use to analyse the “biological evolution” into the “risk evolution”. Which risk scenarios are “similar” How the risk characteristics “evolve”? Future potential risk “mutation”
Illustration for the biology phylogenetics
Lamprey
Shark
Salmon
Lizard
Characteristics Paired fin Jaws
Large dermal bones
Fin rays Lung Rasping tongue
✓
✓ ✓ ✓
✓ ✓ ✓ ✓
✓ ✓ ✓ ✓
Illustration for the biology phylogenetics
Lamprey
Shark
Salmon
Lizard
Illustration for the biology phylogenetics
By application of phylogenetic analysis (parsimony algorithm), the evolution process can be revealed using the tree-like “cladogram”: First, assume that the species evolve from nothing to existence and
one of the two branches shall be occupied by the species with the least characters, i.e., the lamprey
Second and later, repeat the selection method to find the organism that owns the least changes to the lamprey and last species.
(a) paired fins, (b) jaws, (c) large dermal bones, (d) fin rays, (e) lungs, and (f) rasping tongue
Illustration for the biology phylogenetics
Now we can dig into the evolution analysis: The shape of lamprey is stable and not likely to have changes soon. Salmon and lizard must share the same ancestor which shark does not
have relations to. ……
We can do the same thing for risk analysis!
(a) paired fins, (b) jaws, (c) large dermal bones, (d) fin rays, (e) lungs, and (f) rasping tongue
Risk Phylogenetics illustration
Data Preparation
✓
✓ ✓
✓
Animals
Biological Characteristics
Data Preparation
✓
✓ ✓
✓
Animals
Biological Characteristics
RiskScenarios
RiskCharacteristics
Characteristics - Risk CharacteristicsRisk Characteristic Code1.1 Portfolio risk selection 11.2 Portfolio Management 21.3 Claims management 31.4 Technical Reserving 41.5 Reinsurance arrangements 51.6 Longevity risk (Pension) 61.7 Pricing 72.1 Reinsurance Credit Risk 82.2 Insurance products credit risk+A23 92.3 Insurance operations credit risk 102.4 Invested assets credit risk 113.1 Asset and liability matching 123.2 Investment default 133.3 Currency risk 143.4 Basis risk 153.5 Property price depreciation 163.6 Equity risk 173.7 Interest rate risk 183.8 Commodity risk 193.9 Spread risk 204.1 Assets liquidity 214.2 Funding liquidity 224.3 Liability liquidity 234.4 FX liquidity 244.5 Intra-day liquidity 255.01 Internal fraud / Unauthorised Transactions 265.02 Internal fraud / Theft and Fraud 275.03 External Fraud / Theft and Fraud 285.04 External Fraud / System Security 29
5.05 Employment Practices / Employee Relations 305.06 Employment Practices / Safe Environment 315.07 Employment Practices / Diversity & Discrim. 325.08 Improper Business or Market Practices 335.09 Published Financial Statements 345.10 Advisory activities 355.11 Damage to Physical Assets 365.12 Bus disruption & sys failures / Systems 375.13 Transaction Capture & Maintenance 385.14 Monitoring & Reporting 395.15 Customer Intake and Documentation 405.16 Customer & Client Account Management 415.17 Trade counterparties 425.18 Vendors & Suppliers 435.19 Compliance with existing regulation 445.20 Increase in regulatory costs 455.21 Failure to implement Solvency II 465.22 Cross sector funding FSCF 475.23 Product Flaws 485.24 Expenses overruns 496.1 Regulators 506.2 Corporate responsibility 516.3 Investors / JV Partners 526.4 Media 537.1 Legal, Public Affairs & Regulatory 547.2 Macro-Economic 557.3 Changing Claims Patterns 568.1 Internal 578.2 External 588.3 General 59
Characteristics - Risk CharacteristicsRisk Characteristic Code1.1 Portfolio risk selection 11.2 Portfolio Management 21.3 Claims management 31.4 Technical Reserving 41.5 Reinsurance arrangements 51.6 Longevity risk (Pension) 61.7 Pricing 72.1 Reinsurance Credit Risk 82.2 Insurance products credit risk+A23 92.3 Insurance operations credit risk 102.4 Invested assets credit risk 113.1 Asset and liability matching 123.2 Investment default 133.3 Currency risk 143.4 Basis risk 153.5 Property price depreciation 163.6 Equity risk 173.7 Interest rate risk 183.8 Commodity risk 193.9 Spread risk 204.1 Assets liquidity 214.2 Funding liquidity 224.3 Liability liquidity 234.4 FX liquidity 244.5 Intra-day liquidity 255.01 Internal fraud / Unauthorised Transactions 265.02 Internal fraud / Theft and Fraud 275.03 External Fraud / Theft and Fraud 285.04 External Fraud / System Security 29
5.05 Employment Practices / Employee Relations 305.06 Employment Practices / Safe Environment 315.07 Employment Practices / Diversity & Discrim. 325.08 Improper Business or Market Practices 335.09 Published Financial Statements 345.10 Advisory activities 355.11 Damage to Physical Assets 365.12 Bus disruption & sys failures / Systems 375.13 Transaction Capture & Maintenance 385.14 Monitoring & Reporting 395.15 Customer Intake and Documentation 405.16 Customer & Client Account Management 415.17 Trade counterparties 425.18 Vendors & Suppliers 435.19 Compliance with existing regulation 445.20 Increase in regulatory costs 455.21 Failure to implement Solvency II 465.22 Cross sector funding FSCF 475.23 Product Flaws 485.24 Expenses overruns 496.1 Regulators 506.2 Corporate responsibility 516.3 Investors / JV Partners 526.4 Media 537.1 Legal, Public Affairs & Regulatory 547.2 Macro-Economic 557.3 Changing Claims Patterns 568.1 Internal 578.2 External 588.3 General 59
Risk Characteristic Code
1.1 Portfolio risk selection 1
1.2 Portfolio Management 2
1.3 Claims management 3
1.4 Technical Reserving 4
1.5 Reinsurance arrangements 5
1.6 Longevity risk (Pension) 6
1.7 Pricing 7
2.1 Reinsurance Credit Risk 8
2.2 Insurance products credit risk+A23 9
2.3 Insurance operations credit risk 10
Animals – Risk Scenario (Ireland)
1. Economic Downturn2. Failure to deliver the required scale and breadth of
improvement plan benefits leading to under delivery of projected 2011 UW result.
3. Business does not achieve planned growth.4. ABC integration / alignment.5. Loss of key intermediary / corporate account through
failure of intermediary or transfer of business to competitor.
…….
Animals – Risk Scenario (UK)
1. Fail to recognize and protect portfolios against the effects of larges losses and abnormal weather
2. Current review by Lord Chancellor requires reserve strengthening for Ogden lump sum awards delivery of projected 2011 UW result.
3. Adverse Bodily Injury trends continue to rise.4. Insufficient rate within Commercial Property portfolios
to achieve required risk adjusted return5. Fraud trends continue to rise
…….
Data prepared (Ireland)
Risk ID Risk1 Portfolio Risk Selection
2 Portfolio Management
3 Claims Management
4 Technical Reserving
5 Reinsurance Arraignments
IRE1 Economic Downturn. ✓
IRE2 Failure to deliver the required scale and breadth of improvement plan benefits leading to under delivery of projected 2011 UW result. ✓
IRE3 Business does not achieve planned growth.
IRE4 ABC integration / alignment.
IRE5 Loss of key intermediary / corporate account through failure of intermediary or transfer of business to competitor.
IRE6 Non-compliance with regulatory requirements, including subsidiaries.
IRE7 Inadequate Data Privacy procedures.
IRE8 Risk of adverse development of Prior Year claims on X Book.
IRE9 Repeat of catastrophic weather events. ✓ ✓ ✓
IRE10 Implementation of Periodic Payment Orders. ✓ ✓
IRE11 Failure of Software House.
IRE12 Immature capability re direct and on-line channel.
IRE13 XXX Insurance Ireland S&P downgrade.
IRE14 Outcome of test Achats by ECJ – EU gender directive decision.
“Ani
mal
s”
“Characteristics”
1 Underwriting Risk
Risk ID Risk 1
Por
tfol
io r
isk
sele
ctio
n
2 P
ortf
olio
M
anag
emen
t
3 C
laim
s m
anag
emen
t
4 T
echn
ical
R
eser
ving
5 R
eins
uran
ce
arra
ngem
ents
6 Lo
ngev
ity
risk
(P
ensi
on)
7 P
rici
ng
UK1 Fail to recognise and protect portfolios against the effects of larges losses and abnormal weather ✓ ✓ ✓ ✓
UK2 Current review by Lord Chancellor requires reserve strengthening for Ogden lump sum awards ✓
UK3 Adverse Bodily Injury trends continue to rise ✓ ✓ ✓
UK4 Insufficient rate within Commercial Property portfolios to achieve required risk adjusted return ✓
UK5 Fraud trends continue to rise ✓
UK6 Focus on top line leads to a failure to maintain underwriting, pricing and controls discipline resulting in negative bottom line impact ✓ ✓ ✓ ✓
UK7 Inadequate reserves to cover Disease (asbestos, deafness, vibration white finger) and Abuse claims ✓ ✓ ✓
UK8 The European Court of Justice rules against gender based risk pricing in insurance contracts (Achats) ✓
UK9 Periodic Payment Orders (PPOs) adversely impact current reserve levels ✓ ✓
UK10 Lack of capacity for key initiatives, deals and change programmes resulting in poor execution and / or poor integration
UK11 Systemic Credit risk event such that default levels on unsecured credit reach 1991 levels or default of a major counterparty
UK12 Poor control of Delegated Authority Schemes business results in a loss
UK13 Fail to achieve business case for key initiatives, deals, change programmes
UK14 Inflation drives adverse impact on expense base and claims cost
UK15 Fail to adapt and implement changes to the regulatory architecture, including Solvency II
Data prepared (UK)
Risk Evolution Cladgram (Ireland)1. Economic Downturn
2. Under delivery of projected UW result
9. Repeat of catastrophic weather events
10. Implementation Periodic Payments
14. Outcome - EU gender directive
8. Prior Year claims on X Book
3. Buss doesn’t achieve planned growth
4. ABC integration / alignment
7. Inadequate Data Privacy procedures
12. Immature capability re on-line channel
Lots of evolution from “7 Pricing” and “38 Trans”
1
4, 56
7
2
56
45, 54
7
52, 53
44, 46, 50
43
10, 58
37
5738
40, 41
26, 28, 29, 50, 52
38
Pricing
Portfolio Mngment
3, 8, 9, 10, 11, 13, 16, 17, 18, 19, 20, 21, 24, 55
Transaction Capture & Maintenance
Internal
Vendors & Suppliers
49
# = Code of Risk characteristics
All of these related to Pricing
All of these relate to Operation
Portfolio Risk Selection
Risk Evolution Cladgram (UK)
46, 54, 57
8, 9, 10, 11
37
49, 56
38, 40, 41
27, 28, 29
4
35, 56
54
1
3, 56
23, 26, 39
5
7
Claims Mngment
Pricing
Portfolio Risk Selection
Legal, Public Affairs & Regulatory
Technical Reserving
1. Fail to protect portfolios against larges losses and abnormal weather
6. Failure to maintain underwriting, pricing and controls
3. Adverse Bodily Injury trends continue to rise
4. Insufficient rate within Commercial Property portfolios
2. Review by Lord Chancellor requires reserves rise
8. The European Court of Justice rules against (Achats)
5. Fraud trends continue to rise
7. Inadequate reserves to cover Disease and Abuse claims 9. (PPOs) adversely impact current reserve levels
Portfolio Mngment
Reins Arrangements
Key Observations and Questions
For Ireland and UK, ‘Pricing‘(7) is the most important risk character since it defines a biggest clade.
The risk ‘Economic Downturn’(IRE1) is quite distinct due to its large number of characters: Economic downturn is a complex risk and covers many areas and this could also be argued that it is too high level and should be
split into more defined areas (like ‘housing crises‘ or ‘euro crises‘). Branches with the most characters indicate there has been
significant evolution and we are likely to see more cascading evolution, like a warm jungle is host to more forms of life than the cold tundra. If the risk ‘Inadequate Data Privacy Procedures‘(IRE7) (3 branches
and 5 characters) were to combine with ‘Immature Capability on On-line Channel‘(IRE12), what does the emerging risk look like?
Key Observations and Questions
In the UK, there are 3 risks which show no relation to any others ‘Systematic credit risk event’ (UK11), ‘Inflation derives adverse impact on expense and claim cost’ (UK14) and ‘Fail to adapt changes to the regulatory architecture such as SII’ (UK15). Risks that have not changed significantly are likely to be stable. Or this should be checked against whether the risks have not been
described in sufficient detail. We can presume “co-evolution”, which means that characters
or risks have a tendency to evolve in each other‘s presence. Like a bird that develops a long beak and a flower holding nectar that
continues to extend the long shape. For example, ‘Media‘(53) only evolves in the presence of ‘Investors/JV
Partners‘(52). So we can investigate risks that have character (52), but not yet (53) and presume some new risk events occurring.
Summary
Summary
Modern risk management is complex and ERM requires a holistic approach to make sense of the layers, interconnections and non-quantitative measures.
Using the phylogenetic analysis as the means of constructing model-free evolutionary risk trees and their interpretationenabled us to reveal the possible emerging risks.
The visualization of the risks and characters in a tree format enables a lot of fruitful observations to be quickly spotted that would be difficult and tiresome in a spreadsheet.
It also could be a good news that new and free software programs and algorithms allow easy access for actuaries to be able to construct their own risk trees.
This presentation has been prepared for illustrative purposes only. The views expressed in this presentation are those of the speaker and do not necessarily represent the views of Milliman as a firm.
No reliance or inference should be placed on the statements presented herein. Actual experience and developments may differ than those presented in this presentation.