Selling Risk Management: Quan2fica2on Techniques to Support Risk Decisions RIF007

Transcription

1

2 Selling Risk Management: Quan2fica2on Techniques to Support Risk Decisions RIF007 Speakers: Christopher (Kip) Bohn, Actuary, Aon Jeff Williams, Director, Risk Management & Loss Prevention, Eli Lilly and Company

3 Eli Lilly and Company Headquarters Indianapolis, IN Founded in 1876 by Colonel Eli Lilly 2014 Financial Highlights Revenue = $19.6B Research and Development = $4.7B (24% of Revenue) Elanco Animal Health = $2.3B Employees = 39,135 Highlighted Products (Legacy and Current) Prozac Cialis Cyramza TM Diabetes products

4 Learning Objectives At the end of this session, you will: Anticipate the potential pitfalls of unsupported decisions, benchmarking and complacency. Realize the benefits of utilizing analytics when making risk and insurance decisions. Prepare to sell your risk management capabilities beyond insurance decisions within your organization.

5 Pi=alls of Unsupported Decisions Selective memory Framing effects Perceptions Reasoning Nonregressive predictions Primacy and recency Selective attention Anchoring effects Inability to reason probabilistically Attribution errors Affective forecasting Confirming evidence Overconfidence Hindsight Status quo Illusion of control Escalation of Motivations Personality Decision styles Traits commitment Comfort zones Sunk cost fallacy Discounting opportunity costs Group Dynamics Habitual frames Content selectivity Wishful thinking Positive illusions Premature harmony Obedience Conformity Anonymity Attention to shared evidence Psychological safety Graphic Courtesy of John Bugalla

6 Risk Modeling A7empt to es9mate an unknown to make a be7er decision Simplifica9on of reality to help us understand characteris9cs of a system Useful when it is prohibi9ve or impossible to observe/test directly Important characteris9cs: Accuracy, Clarity, Flexibility, Efficiency Models are tools all models are wrong, some are useful Reality Model

7 Risk Modeling

8 Risk Modeling

9 Model Development 0.08 Volatility Measures Expected Value Mean, Average Most financial forecasts convey the expected value Point es9mate to represent all poten9al outcomes Does not convey idea of risk Variance Alterna9ve is Standard Devia9on Describes how far events deviate from the expected value Mean and Variance Completely describes risk for symmetric distribu9on Some9mes used for comparing op9ons Reward 99.9% 90.0% Efficient Frontier Risk

10 Model Development Advanced Volatility Measures Risks and opportuni9es open not symmetric Losses bounded below by zero Total cost of risk insured risk bounded below by transfer costs Rewards bounded below by investment amounts (theore9cally unbounded below if addi9onal dollars are con9nually pumped into a losing investment) Skewed distribu9ons Not symmetric Fat tails or Black Swam The average is no longer at the median (i.e. 50%ile) Same mean and variance can produce different results

11 Evaluating Insurance Decisions How much risk can we take and should we take? What is the value of insurance? How can we use analytics to support risk decisions for insurance portfolio retention and limits levels? Can we understand the value of risk management in terms of key financial measures such as EBITDA or Free Cash Flow? How can we develop consistency in evaluating different risks? How do we understand the role of a captive and the best potential uses of it?

12 Benefits of an Analytical Approach to Insurance Risk Decision Making Provides cost-benefit analysis and support tool for risk management and insurance decisions Provides advice to determine the impact of program structures Retained losses Insurance recoveries Losses in excess of insurance coverage Impact of non-transfer mitigation strategies

13 Insurance Risk Management Framework

14 Modeling Used by Lilly Program Cost & Structures? Loss expenses Premiums Retentions Context Exposure and/or Cost Consolidation of operations Redundancy is great but there is a cost Modeling Total Cost of Risk (TCoR) Analysis Each risk management program analyzed individually Consolidated into a portfolio view Impact Analysis of Closing Facilities 3 locations, down to 1 or 2 locations

15 Modeling Used by Lilly Context How Big Is Too Big? Can you have too much manufacturing at one site? What are the hazards? Modeling Indiana Windstorm Analysis Probability and impact of tornado striking manufacturing site Natural Hazards Windstorm, Earthquake & Flood Underwriters modeling to determine premium but are their inputs correct? Natural Hazard Modeling Windstorm Puerto Rico & Japan Earthquake Puerto Rico & Japan Flood - Indianapolis

16 Model Development Quan9fy Exposure to Risk Understand risk appe9te Construct a Monte Carlo simula9on model that considers the key risks iden9fied and the risk process map Iden9fy appropriate parameters for the model based on your and industry informa9on Data may be a challenge but not insurmountable Overlay current and proposed transfer strategies Run model to understand the distribu9on of poten9al outcome under various mi9ga9on scenarios Compare and opine on the cost vs. benefit of the various mi9ga9on strategies

17 Risk Appetite Model Commodity Prices Foreign Exchange Commodity Prices Interest Rates Revenue Expenses Income Before Taxes Risk Appe2te This flow chart depicts how changes in projected vola5lity parameters flow through the model to impact financial performance metrics

18 Actuarial Risk Model Program Expected TCOR Risk Appetite (Premium + Loss) Breach Point CAT TCOR Unlimited $22.8 million 86% $39.1 million Current $26.1 million 78% $32.2 million Alternative $23.4 million 87% $33.9 million

19 Risk Model Results Total Cost of Risk (TCoR) Analysis Most programs produced favorable results Those that did not set the stage for improvements at next renewal Impact Analysis of Closing Facilities 2 locations could be acceptable provided risk improvements were completed 1 location unacceptable without robust BCP s in place Indiana Windstorm Analysis Smaller F-value storms produced less damage than expected Larger F-value storms probability much less than expected Natural Hazard Modeling Windstorm Produced better results than what underwriter s were producing Earthquake Results appeared to be in line with underwriter s results Flood Exposure confirmed so Flood Emergency Response Plan (FERP) created

20 Selling Risk Management

21 Risk Modeling Beyond Insurance Income Statement Balance Sheet Statement of Cash Flows Financial Plan Operational Volatility Modules Financial Plan Risk Modules Infrastructure Risk Weather Risk Product Liability Volatility FX Volatility Sales Volatility Demand Volatility Key Calculation Risk Module Output Insurance Risks Operational/ Strategic Risks Risk Adjusted Financial Plan

22 Advanced Actuarial Risk Model Supplier Manufacturing Warehouse Safety Stock Earnings Per Share Supplier Supplier Supplier Raw Material Inventory Raw Material Inventory Event Manufacturing Reroute Manufacturing Reroute Warehouse Packaging Distribution and Sales Free Cash Flow Unexpected Loss Excess Capacity At Capacity Net Capacity Decrease

23 Advanced Risk Model Program Expected TCOR Risk Appetite (Premium + Loss) Breach Point CAT TCOR Unlimited $22.8 million 86% $39.1 million Current $26.1 million 78% $32.2 million Alternative $23.4 million 87% $33.9 million

24 Selling Risk Management Final thoughts Use risk transfer models as a basis to expand conversa9ons beyond insurance Data can be challenging Collec9on can be 9me consuming Consider industry data Google! Scenario workshops Will likely uncover opportuni9es for forward looking data collec9on

25 Selling Risk Management Final thoughts Break the analy9cs into bite sized pieces Many insights will be gained By connec9ng the various dots can help drive a more enterprise view of risk Drive greater risk management visibility in the organiza9on Will generate buzz and drive future projects

26 Questions?