The Role of Data Science in
Enterprise Risk Management
By John Liu, PhD, CFA
Question
of the Day
¡ How do you tell the difference between a
Bayesian Statistician and Data Scientist?
¡ Answer: W...
Big Data: Big Risks
¡ Healthcare
¡ Financial Services
¡ Insurance
¡ Transportation
¡ National Security
¡ Datin...
Key
Takeaways
¡ What is Enterprise Risk Management
(ERM)?
¡ What is the Role of Data Science in
ERM?
¡ What Data ...
What is
Enterprise Risk
Management?
What is
Risk Management?
¡ A structured approach to manage uncertainty
¡ Management strategies:
Risk Avoidance Risk ...
Risk Management - Defense
Insurance Approach
Reward
Do Nothing
Probability of Success
Risk Management - Offense
Opportunistic Approach
Reward
Carpe Diem
Do Nothing
Probability of Success
What is ERM?
¡ Risk-based approach to managing an enterprise
¡ Risk-aware: every major tangible and intangible
factor...
ERM
Components
Identify
Quantify
Respond
Monitor & Report
Effectiveness
Monitor
Comprehensive
Approach To
Managi...
ERM
Goals
¡ Provide holistic view across an organization
leveraging firm experience and knowledge
¡ Provide greater ...
ERM
Risk Types
¡ • Resource Capital Management
• Business Disruption, IT Operational
• Credit Exposure
• Exchange Ra...
RM vs ERM
HQ: EUR exposure Subsidiary: USD exposure
Sells EUR, Buys USD Sells USD, Buys EUR
RM: Subsidiaries/Business U...
Data Science
and ERM
ERM
Framework
Enterprise Structure, Risks Objectives & Components
Compliance
Financial
Compliance
Reporting
Hazard ...
Common
Challenges
¡ Data warehousing & sharing across entity
¡ Prioritization methodology
¡ Consolidated reporting ...
Role of
Data Science
¡ Data science methods provide:
¡ Enterprise Data Management
¡ Comprehensive warehousing
¡ D...
Typical
Corporate EDW
¡ Big data warehouse ≠ useful data (quite the opposite)
Data Management
¡ Comprehensive data warehouse
¡ Coherent data collection (maybe)
¡ Facilitate data sharing across e...
Risk
Analytics
¡ Benefits beyond Business Intelligence
Descriptive
Analytics
Predictive
Analytics
Prescriptive
An...
Rich Set of Visualization &
Reporting Tools
Aggregate Risk
Dashboards
Continuous &
Comprehensive
Risk Monitors
Sour...
Data Analytics
Applications for ERM
¡ • Scenario Analysis Operational & Stress Testing
Financial • Credit Scoring
Com...
Data Analytics
for ERM
Definition of
Risk
¡ Risk = Frequency of Loss x Severity of Loss
¡ Loss Distribution
Unexpected Loss
Traditional ERM
¡ Analytic Methods
¡ Closed-form solutions (…just like most things in life)
¡ Historical
¡ Estimat...
Modern ERM
¡ Data analytics driven
¡ Inference based methods
¡ KRI scoring
¡ Parallelization
¡ Natural applicati...
Prediction Methods
Methods
Transduction
Tail Bayesian Frequentist
Extreme-Value
Expected Deficit
Naïve Bayes
HMMs
...
Outliers, Inliers,
and Just Plain Liars
¡ Prediction problems fall in two classes:
Inliers Outliers
Inherently differ...
Main Problems with
Inlier Prediction
¡ Parametric model choice
¡ Estimation error for lower moments (mean, s.d.)
¡ ...
Main Problem with
Outlier Prediction
¡ Data Quality and Abundance
¡ To estimate low probability events, big data may ...
Value-at-Risk (VaR)
¡ Loss severity measure for a given probability and time
horizon
• Estimate potential losses (or
...
Value-at-Risk
¡ Loss severity measure for a given probability and time
horizon
1-day 95% VaR of $1m
Expect to lose no...
Tail Value-at-Risk (TVaR)
¡ Loss severity measure for a given probability and time
horizon
• Estimate potential losses...
Tail Value-at-Risk (TVaR)
¡ Loss severity measure for a given probability and time
horizon
1-day 95% TVaR of $122m
Be...
Application: Operational
Risk Management
¡ Definition: The risk of direct and indirect loss resulting
from inadequate ...
Managing OpRisk
¡ One Approach
Source: NYFed
Assess Scorecard
Internal
Loss Data
Identify
Weakness
Risk
Scenario...
Methods
¡ Scorecard
3
5
9
¡ KRI scoring models
2
3
5
¡ Useful where no severity data exists
1
2
3
Loss Dis...
Looking
Forward
ERM Trends
Source: NCSU
¡ Increasing adoption of ERM
Forensic Data Analytics
Fraud Detection Top Concern
But Low Adoption.
Source: Ernst & Young
Promise of Data Analytics
¡ EDW remains a huge issue for most corporations
¡ Legacy zombie systems
¡ IT reporting li...
Thank
you
of 42

Role of Data Science in ERM @ Nashville Analytics Summit Sep 2014

An overview of how organizations can leverage data science and predictive analytics to improve enterprise risk management. Applications for risk identification, mitigation and management will be discussed, as well as methods to facilitate strategic integration across an organization.
Published on: Mar 3, 2016
Published in: Data & Analytics      
Source: www.slideshare.net


Transcripts - Role of Data Science in ERM @ Nashville Analytics Summit Sep 2014

  • 1. The Role of Data Science in Enterprise Risk Management By John Liu, PhD, CFA
  • 2. Question of the Day ¡ How do you tell the difference between a Bayesian Statistician and Data Scientist? ¡ Answer: What’s the p-value?
  • 3. Big Data: Big Risks ¡ Healthcare ¡ Financial Services ¡ Insurance ¡ Transportation ¡ National Security ¡ Dating
  • 4. Key Takeaways ¡ What is Enterprise Risk Management (ERM)? ¡ What is the Role of Data Science in ERM? ¡ What Data Analytics are used for ERM?
  • 5. What is Enterprise Risk Management?
  • 6. What is Risk Management? ¡ A structured approach to manage uncertainty ¡ Management strategies: Risk Avoidance Risk Transfer Risk Mitigation
  • 7. Risk Management - Defense Insurance Approach Reward Do Nothing Probability of Success
  • 8. Risk Management - Offense Opportunistic Approach Reward Carpe Diem Do Nothing Probability of Success
  • 9. What is ERM? ¡ Risk-based approach to managing an enterprise ¡ Risk-aware: every major tangible and intangible factor contributing toward failure in every process at every level of the enterprise ¡ Enterprise value maximized with optimal balance between profitability/growth and related risks ¡ Management better prepared to seize opportunities for growth and value creation
  • 10. ERM Components Identify Quantify Respond Monitor & Report Effectiveness Monitor Comprehensive Approach To Managing Uncertainty Identify/Assess Internal and External Risks Risk Scoring & Modeling Respond and Control
  • 11. ERM Goals ¡ Provide holistic view across an organization leveraging firm experience and knowledge ¡ Provide greater transparency to factors that can impair value preservation and business profitability ¡ Understand & test assumptions & interpretations in business decision-making
  • 12. ERM Risk Types ¡ • Resource Capital Management • Business Disruption, IT Operational • Credit Exposure • Exchange Rate, Cash flow, Funding Financial • Privacy, Security, Safety • Regulatory and Statutory Compliance • Financial Reporting • Regulatory Reporting Reporting • Natural Catastrophe • Market Panics Hazard • Business Planning • Marketing, Reputation Strategic
  • 13. RM vs ERM HQ: EUR exposure Subsidiary: USD exposure Sells EUR, Buys USD Sells USD, Buys EUR RM: Subsidiaries/Business Units manage risks separately ERM: Manage NET exposure across entire enterprise
  • 14. Data Science and ERM
  • 15. ERM Framework Enterprise Structure, Risks Objectives & Components Compliance Financial Compliance Reporting Hazard Strategic Entity Wide Division Business Unit Comprehensive Approach Leverage Data & Analytic Resources Predictive Modeling
  • 16. Common Challenges ¡ Data warehousing & sharing across entity ¡ Prioritization methodology ¡ Consolidated reporting ¡ Timeliness ¡ Data security ¡ The risk management process itself!
  • 17. Role of Data Science ¡ Data science methods provide: ¡ Enterprise Data Management ¡ Comprehensive warehousing ¡ Data quality and abundance ¡ Risk Analytics ¡ Predictive Modeling ¡ Loss Distributions ¡ Reporting ¡ Real-time visualization, dashboards ¡ Regulatory requirements Reporting
  • 18. Typical Corporate EDW ¡ Big data warehouse ≠ useful data (quite the opposite)
  • 19. Data Management ¡ Comprehensive data warehouse ¡ Coherent data collection (maybe) ¡ Facilitate data sharing across entity ¡ No useful analytics without abundant, high quality data Data Big Data Excel BigTable PostgreSQL Cassandra, HIVE, HBase MongoDB Vertica, KDB
  • 20. Risk Analytics ¡ Benefits beyond Business Intelligence Descriptive Analytics Predictive Analytics Prescriptive Analytics What happened? What’s likely to occur? Why would it occur? Hindsight Foresight Insight Summary Statistics Data mining Heuristic Optimization web analytics, BI, credit scoring, trend operations planning, inventory reporting analysis, sentiment stochastic methods ¡ Newest: cognitive analytics = What is the best answer?
  • 21. Rich Set of Visualization & Reporting Tools Aggregate Risk Dashboards Continuous & Comprehensive Risk Monitors Source: IBM Cognos
  • 22. Data Analytics Applications for ERM ¡ • Scenario Analysis Operational & Stress Testing Financial • Credit Scoring Compliance • IT Security Anomaly Detection Reporting • Risk Dashboard Hazard • Catastrophe & Market Risk Hedging Strategic • Marketing Analytics
  • 23. Data Analytics for ERM
  • 24. Definition of Risk ¡ Risk = Frequency of Loss x Severity of Loss ¡ Loss Distribution Unexpected Loss
  • 25. Traditional ERM ¡ Analytic Methods ¡ Closed-form solutions (…just like most things in life) ¡ Historical ¡ Estimate risk using internal and external loss data ¡ Monte Carlo ¡ Estimate distribution parameters from real data ¡ Monte-Carlo sample distribution ¡ Calculate ensemble measures to estimate overall risk ¡ Simple to implement, aggregate across entity, but make complex assumptions, not robust to outliers
  • 26. Modern ERM ¡ Data analytics driven ¡ Inference based methods ¡ KRI scoring ¡ Parallelization ¡ Natural applications ¡ credit risk scoring ¡ Anti-money laundering ¡ Fraud
  • 27. Prediction Methods Methods Transduction Tail Bayesian Frequentist Extreme-Value Expected Deficit Naïve Bayes HMMs Bayes Nets Regression, Decision Trees SVM Ensemble Methods Bagging, Boosting, Voting
  • 28. Outliers, Inliers, and Just Plain Liars ¡ Prediction problems fall in two classes: Inliers Outliers Inherently different problems with different quirks
  • 29. Main Problems with Inlier Prediction ¡ Parametric model choice ¡ Estimation error for lower moments (mean, s.d.) ¡ Incorrectly conjugating priors ¡ Normal/Gaussian distributions don’t really occur in real life ¡ I.I.D.? Really?
  • 30. Main Problem with Outlier Prediction ¡ Data Quality and Abundance ¡ To estimate low probability events, big data may not be big enough Data: 150 years of daily data Predictor: 100 year flood severity Relevant Data: 1 or 2 data points
  • 31. Value-at-Risk (VaR) ¡ Loss severity measure for a given probability and time horizon • Estimate potential losses (or historical losses) • Rank losses based on severity • 95% Value-at-Risk is equal to the 95th percentile loss • Interpretation = Losses won’t exceed 65.2m 95% of time • Underestimates losses during the other 5% of time Rank Loss 1 -­‐0.1 2 -­‐0.1 3 -­‐0.3 4 -­‐0.6 5 -­‐0.7 6 -­‐0.9 7 -­‐1.1 … … 91 -­‐59.5 92 -­‐63.2 93 -­‐64.9 94 -­‐65.0 95 -­‐65.2 96 -­‐66.5 97 -­‐67.8 98 -­‐93.9 99 -­‐110.0 100 -­‐273.1 VaR
  • 32. Value-at-Risk ¡ Loss severity measure for a given probability and time horizon 1-day 95% VaR of $1m Expect to lose no more than $1m in 95 out of every 100 days Says nothing about the other 5 days out of 100. Not very reassuring, is it?
  • 33. Tail Value-at-Risk (TVaR) ¡ Loss severity measure for a given probability and time horizon • Estimate potential losses (or historical losses) • Rank losses based on severity • 95% Tail Value-at-Risk is equal to average of all losses beyond 95th percentile loss • Expect to lose on average $122m if losses exceed the 95th percentile Rank Loss 1 -­‐0.1 2 -­‐0.1 3 -­‐0.3 4 -­‐0.6 5 -­‐0.7 6 -­‐0.9 7 -­‐1.1 … … 91 -­‐59.5 92 -­‐63.2 93 -­‐64.9 94 -­‐65.0 95 -­‐65.2 96 -­‐66.5 97 -­‐67.8 98 -­‐93.9 99 -­‐110.0 100 -­‐273.1 TVaR
  • 34. Tail Value-at-Risk (TVaR) ¡ Loss severity measure for a given probability and time horizon 1-day 95% TVaR of $122m Better Measure of Risk Also known as Expected Shortfall, CVaR
  • 35. Application: Operational Risk Management ¡ Definition: The risk of direct and indirect loss resulting from inadequate or failed: ¡ Internal processes ¡ People ¡ IT systems ¡ External events Source: NYFed Operational Risk External Criminal Activity Information security failure Internal Criminal Unauthorized Activity Activity Processing Failure System Failure Control Failure Business Disruption Workplace Safety Malpractice
  • 36. Managing OpRisk ¡ One Approach Source: NYFed Assess Scorecard Internal Loss Data Identify Weakness Risk Scenarios Risk Model OpVar Risk Capital
  • 37. Methods ¡ Scorecard 3 5 9 ¡ KRI scoring models 2 3 5 ¡ Useful where no severity data exists 1 2 3 Loss Distribution Impact ¡ ¡ Estimation of severity distribution parameters ¡ MLE Not robust – data not i.i.d., biased upwards, subject to Probability data paucity & sparsity ¡ Leads to biased loss exposures and correlation assumptions ¡ Huge opportunity for inference-based analytics
  • 38. Looking Forward
  • 39. ERM Trends Source: NCSU ¡ Increasing adoption of ERM
  • 40. Forensic Data Analytics Fraud Detection Top Concern But Low Adoption. Source: Ernst & Young
  • 41. Promise of Data Analytics ¡ EDW remains a huge issue for most corporations ¡ Legacy zombie systems ¡ IT reporting lines ¡ Increased understanding by senior managers and C-suite ¡ Analytics as a Service: growing competition within consulting industry ¡ Talent Gap – same for anything Data Science
  • 42. Thank you

Related Documents