Session 105 WS, Economic Capital Management for ORSA and Beyond. Moderator/Presenter: Manchiu Chan, FSA, MAAA

Session 105 WS, Economic Capital Management for ORSA and Beyond Moderator/Presenter: Manchiu Chan, FSA, MAAA Presenters: Matthew Kramer, FSA, CERA, MAAA Christopher M. Suchar, MAAA, FCAS

Economic Capital Modeling A Stochastic Approach April, 2016 One Financial Plaza Hartford, CT 06103 www.conning.com

ERM AND ECONOMIC CAPITAL MODELING (ECM) 2

Two Parallel Tracks ERM and Economic Capital (EC) Model Growth Paths ORSA Framework Growth Paths Maturity Level Qualitative ERM Quantitative ERM Risk governance Risk appetite and tolerance limits Risk identification Risk impact assessment Often on broad scale (1 to 10) Heat maps, directional information Measuring risk impacts dollar quantification Dashboards cost/benefit relative to other financial measures Foundational ECM Robust ECM Initial models Robust enterprise models Focus on financial risks assets and underwriting Use of ESG Reflects correlation and diversification Quantifying mitigation effects All risks included Fully integrated with planning and management processes 3

Best Practices for Economic Capital Models Robust risk models on both sides of the balance sheet Economic Scenarios Calibrated to all the volatility of the 20th and 21st centuries Assets Market risk Assets Credit risk Insurance Reserving risk Insurance Underwriting risk Strategic and Operational risk Non-insurance Operations Stochastic and stress testing capability Depends on investment strategy less important for health insurers Less important for health insurers due to relatively quick benefit payouts Much more important to health insurers due to systemic changes Must be able to stochastically stress whole enterprise at once Must also be able to run deterministic stress tests Unified, integrated model of all assets and liabilities Modeling distinct business entities and at the consolidated enterprise level in the same ECM framework Modeling management actions, integrated within the model Capital Fungibility Flows of capital and funds between entities must reflect reality Liquidity risk evaluated in a consistent ECM framework The Use Test Model must be transparent and granular enough to be used by management 4

Stochastic ECMs Pros and Cons Pros Provides probability statements for capital adequacy will be required as ORSA standards advance (already required in other solvency regimes) Provides basis for allocating the cost of capital, to support better financial performance metrics critical to creating greater value for management beyond compliance Provides better framework for addressing interactions between risk factors Cons Additional work beyond what is require for a pure scenario testing approach (but the good news is all work done on a scenario testing basis can be leveraged) Additional management education required 5

BUSINESS APPLICATIONS 6

Millions Capital Adequacy Assessment To assess capital adequacy, use the ECM to project ranges of balance sheet capital The downside ends of the ranges are compared to key regulatory or rating agency thresholds need to demonstrate a small probability of capital shortfall (how small depends on audience) 5,000 Ranges of Projected Capital Blue line = average projected capital 4,500 4,000 3,500 3,000 Bottom of yellow box = 10 th percentile, i.e., 1-in-10 year downside 2,500 2,000 1,500 1,000 500 0 851 873 2011Q4 2012Q1 2012Q2 2012Q3 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 953 A+ Benchmark Capital Company Action Level Bottom of vertical line = 1,036 1st percentile = 1-in-100 year downside Compare this to minimum capital thresholds such as the blue or red line Source: ADVISE Model 7

Capital Adequacy & Risk Tolerance Key Choices Capital Adequacy Metric Policyholder Surplus Shareholders Equity Free Cash Flow Earnings Capital Adequacy Standard Regulatory or Rating Agency Threshold Debt Rating or Bond Default Threshold Time Horizon 1 Year, 3 Years, 5 Years (can produce very different answers) 8

Capital Adequacy Measure, Threshold & Time Horizon Many companies will use bond rating probability of default as a proxy/threshold for evaluating their solvency S&P Corp Bond Default Rate: Single A, 1-Year = 0.07% (i.e., 99.93% chance of not defaulting) At the 0.07% probability level, at the end of Year 1 the Capital level falls to about $600M S&P Corp Bond Default Rate: Single A, 5-Year = 0.35% (i.e., 99.65% chance of not defaulting) At the 0.35% probability level, at the end of Year 5 the Capital level falls to negative $73M Source: ADVISE model XYZ Company Capital ($ in millions) 2014 2015 2016 2017 2018 Average 651 708 739 781 815 Std Dev 27 38 67 101 146 0.07% 599 459 203 10 (389) 0.10% 600 491 224 28 (346) 0.20% 600 526 280 99 (207) 0.35% 600 539 355 148 (73) 0.50% 600 566 392 209 11 1.00% 601 595 468 320 184 2.00% 601 624 541 446 321 2.50% 601 629 561 482 381 5.00% 606 647 623 598 542 10.00% 616 663 674 690 687 25.00% 632 687 719 767 800 50.00% 650 711 751 804 850 75.00% 669 734 778 833 889 90.00% 688 753 799 857 919 95.00% 699 764 811 871 937 97.50% 708 773 821 883 953 98.00% 710 775 824 885 957 99.00% 717 783 831 894 971 99.50% 722 791 839 902 983 99.60% 724 792 842 905 986 99.80% 728 802 850 910 993 9

Determine Required Capital Calculate/find the 0.35 percentile for Capital held at Year 5 (2018) from the simulation run XYZ Company Capital ($ in millions) Take the Capital held at the beginning of the simulation (Time=0) and subtract the present value of the 0.35 percentile for Capital held at Year 5 (using 5 year treasury yield as of 12/31/2013) The result is the Required Capital, i.e. the minimum capital level as of 12/31/2013 that will satisfy the chosen risk tolerance. Source: ADVISE model 2014 2015 2016 2017 2018 Average 651 708 739 781 815 Std Dev 27 38 67 101 146 0.07% 599 459 203 10 (389) 0.10% 600 491 224 28 (346) 0.20% 600 526 280 99 (207) 0.35% 600 539 355 148 (73) 0.50% 600 566 392 209 11 1.00% 601 595 468 320 184 2.00% 601 624 541 446 321 2.50% 601 629 561 482 381 5.00% 606 647 623 598 542 10.00% 616 663 674 690 687 XYZ Company 25.00% 632 687 719 767 800 Capital ($ in millions) 50.00% 650 711 751 804 850 75.00% 669 734 778 833 889 ($ in millions) 90.00% 688 753 799 857 919 Held @ 12/31/2013 95.00% 699 764 811 871$ 937 715 97.50% 708 773 821 883 953 Less: 0.35 98.00% percentile @ 710 end of Year 775 5 (discounted) 824 885 $ 957 Required 99.00% Capital 717 783 831 894 $ 971 99.50% 722 791 839 902 983 (70) 785 99.60% 724 792 842 905 986 99.80% 728 802 850 910 993 10

Capital Allocation Approach Capital itself is not actually sub-divided and allocated to individual segments of the business. All of the capital in a business entity is, in principle, available to support each business segment. It is meaningful, however, to allocate the cost of capital to individual business segments. Each segment must bear a share of the total cost of capital for the enterprise (the cost of capital may be a certain return expected by investors, or a certain internal growth rate target). How do you fairly allocate the cost of capital in an economically rational manner? It is generally accepted that, qualitatively, the allocation should be proportional to each business segment s contribution to the enterprise s total risk. Industry practice is converging on an approach known as Co-Measures (also sometimes referred to as the RMK approach after a paper by Ruhm, Mango and Kreps) because this approach is analytically powerful, transparent and useful to a broad management audience 11

Illustrative Capital Allocation Example Capital Allocation Using Ruhm-Mango-Kreps Algorithm Through Year-End 2018 ($ in millions) (1) (2) (3) = (1) - (2) (4) (5) (6) = [(1)/(5)+1] Total Mean Tail Mean Allocation Basis Allocated ^0.2-1 Profit/(Loss) Profit/(Loss) (Total Mean - Capital Required Annualized Risk Risk Segments (Tax-Adjusted) (Tax-Adjusted) Tail Mean) Allocation Capital Adjusted ROE Profit from Investments 119 111 8 2.5% 20 48% Government 80 (147) 226 72.0% 565 3% Large Group 358 325 33 10.4% 82 40% Individual/Sm Group 93 46 47 15.1% 118 12% Totals 650 335 315 100.0% 785 13% (4) Capital Allocation Using the Allocation Basis column (3), this column calculates the proportion of each risk segment s needs to the total (5) Allocated Required Capital Total Required Capital of $785M is allocated to the risk segments based upon the Capital Allocation percentages in column (4) (6) Annualized Risk Adjusted ROE Required Capital = the minimum capital level as of the beginning of the simulation (Time=0) that will satisfy the chosen risk tolerance Measures the cost of capital for each of the risk segments Source: ADVISE model 12

Capturing Profit Measures by Risk Segment Capital Allocation Using Ruhm-Mango-Kreps Algorithm Through Year-End 2018 ($ in millions) (1) (2) (3) = (1) - (2) (4) (5) (6) = [(1)/(5)+1] Total Mean Tail Mean Allocation Basis Allocated ^0.2-1 Profit/(Loss) Profit/(Loss) (Total Mean - Capital Required Annualized Risk Risk Segments (Tax-Adjusted) (Tax-Adjusted) Tail Mean) Allocation Capital Adjusted ROE Profit from Investments 119 111 8 2.5% 20 48% Government 80 (147) 226 72.0% 565 3% Large Group 358 325 33 10.4% 82 40% Individual/Sm Group 93 46 47 15.1% 118 12% Totals 650 335 315 100.0% 785 13% (1) Total Mean Profit/(Loss) Invested Assets average cumulative profit from investments (income & gains) for ALL paths at the end of Year 5 Business Segments average cumulative underwriting profit or operating income for ALL paths at the end of Year 5 (2) Tail Mean Profit/(Loss) Invested Assets average cumulative profit from investments (income & gains) for the paths at the risk tolerance threshold at the end of Year 5 Business Segments average cumulative underwriting profit or operating income for the paths at the risk tolerance threshold at the end of Year 5 (3) Allocation Basis Total Mean Profit/(Loss) minus Tail Mean Profit/(Loss) measures each segment s shortfall at the enterprise risk tolerance level Source: ADVISE model 13

ECONOMIC CAPITAL MODELING APPROACH 14

The Economic Capital Model Is Based on P&L Forecasts The main moving parts of the ECM correspond directly to the lines of a P&L The best estimate for each line item is tied directly to the financial planning process The variability of each item is based on (1) analysis of data, (2) substantial input from business leaders and (3) economic factors The result is a model that produces realistic scenarios of possible P&L and balance sheet outcomes This will support the key metrics required for ORSA reporting and other risk-based analyses Financial Plan Best Estimate P&L For Business Segment 2012 2013 2014 Members xxx,xxx xxx,xxx xxx,xxx Avg. Prem. x,xxx x,xxx x,xxx Prem. Written xxx,xxx xxx,xxx xxx,xxx Prem. Earned xxx,xxx xxx,xxx xxx,xxx Medical Claims xxx,xxx xxx,xxx xxx,xxx Expenses xx,xxx xx,xxx xx,xxx Net UW Gain xx,xxx xx,xxx xx,xxx Variable Economic Drivers Analysis of Volatility and Dependencies of Healthcare Business Management Input from Business Units Economic Capital Model Scenarios for P&L 2012 2013 2014 Members xxx,xxx2012 xxx,xxx2013 xxx,xxx2014 Avg. Members Prem. x,xxx xxx,xxx x,xxx x,xxx 2012 xxx,xxx 2013 xxx,xxx 2014 Prem. Avg. Written xxx,xxx xxx,xxx xxx,xxx Members Prem. x,xxx xxx,xxx x,xxx xxx,xxx x,xxx xxx,xxx Prem. Prem. Earned xxx,xxx xxx,xxx xxx,xxx Avg. Written Prem. xxx,xxxx,xxx xxx,xxx xxx,xxx Medical Claims xxx,xxx xxx,xxx 2012 x,xxx Prem. Earned xxx,xxx xxx,xxx xxx,xxx 2013 x,xxx 2014 Prem. Written xxx,xxx Expenses Medical Claims xx,xxx Members xxx,xxx xxx,xxxxx,xxxxxx,xxx xxx,xxx xxx,xxxxx,xxxxxx,xxx xxx,xxx xxx,xxx Prem. Earned Net UW Expenses Gain xx,xxx Avg. Prem. xxx,xxx xxx,xxx xxx,xxx xx,xxx xx,xxx x,xxx xx,xxx xx,xxx x,xxx x,xxx Medical Claims xxx,xxx xxx,xxxxx,xxx Prem. Written xxx,xxx xxx,xxx xxx,xxx xxx,xxx Net UW Expenses Gain xx,xxx xx,xxx xx,xxx xx,xxx xx,xxx Prem. Earned xxx,xxx xxx,xxx xx,xxx xxx,xxx Net UW Gain Medical Claims xx,xxx xxx,xxx xx,xxx xxx,xxx xx,xxx xxx,xxx Expenses xx,xxx xx,xxx xx,xxx Net UW Gain xx,xxx xx,xxx xx,xxx Ranges of Possible Results 0 Potential for Loss Translates to Capital Need Gain/Loss 15

The Results of ECM Feed Back Into The Planning Process The range of potential results from the stochastic P&L is used to allocate the firm s capital based on each unit s potential to create losses for the firm The cost of that capital is then deducted from the expected profits of the unit The result is a measure of risk adjusted profit or economic profit This then feeds back into the planning process as a key input to target-setting for prices and profitability Ranges of Possible Results 0 UW Gain Potential for UW Loss Translates to Capital Need Risk-Adjusted P&L 2012 2013 2014 Members xxx,xxx xxx,xxx xxx,xxx Avg. Prem. x,xxx x,xxx x,xxx Prem. Written xxx,xxx xxx,xxx xxx,xxx Prem. Earned xxx,xxx xxx,xxx xxx,xxx Medical Claims xxx,xxx xxx,xxx xxx,xxx Expenses xx,xxx xx,xxx xx,xxx Net UW Gain xx,xxx xx,xxx xx,xxx Allocated Capital xxx,xxx xxx,xxx xxx,xxx Cost of Alloc. Cap. xx,xxx xx,xxx xx,xxx Economic Profit xx,xxx xx,xxx xx,xxx Profit Targets for Planning Process 16

Economic Capital Model Implementation Stages Inventory risk factors Prioritize by impact Identify basis for risk assumptions (actuarial data, risk assessments, etc.) Determine suitable approach for each risk Develop scenarios for each risk factor How bad can it get One year vs. multi year impacts Management/market responses Run scenarios through P&L and balance sheet Aggregate distributions of scenario results to generate capital risk metrics 17

Potential Risks & ECM Treatment Potential Risk Factors Detailed Approach Based On Actuarial/Statistical Internal Models Simplified Approach Based On Management Input/Judgment Risk Distribution Derived from an ERM Risk Assessment Medical Trend CMS Star Rating Cyber Security Risk Risk Adjustment Reinsurance Risk Corridors 18

Risk Driver Variability Government Segment GOVT: Medical Trend Rate GOVT: Simulated CMS Star Rating (Internal) 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% -2.0% -4.0% -6.0% 2013 2014 2015 2016 2017 2018 Confidence Bands 0.9-0.99 0.8-0.9 0.6-0.8 0.4-0.6 0.2-0.4 0.1-0.2 0.01-0.1 Paths Plan Sim Avg 951 662 423 533 645 5.0 4.5 4.0 3.5 3.0 2013 2014 2015 2016 2017 2018 Confidence Bands 0.9-0.99 0.8-0.9 0.6-0.8 0.4-0.6 0.2-0.4 0.1-0.2 0.01-0.1 Paths Plan Sim Avg 951 662 423 533 645 GOVT: Cyber Security Risk Dollar Impacts ($ in 000's) $70,000 Confidence Bands $60,000 0.9-0.99 0.8-0.9 $50,000 0.6-0.8 0.4-0.6 $40,000 0.2-0.4 0.1-0.2 $30,000 0.01-0.1 Paths $20,000 Plan Sim Avg $10,000 951 662 423 $0 2013 2014 2015 2016 2017 2018 533 645 Source: ADVISE model; Simulation = 1,000 paths

P&L Results Variability Government Segment GOVT Membership GOVT PremiumRevenue ($000s) 140,000 130,000 120,000 110,000 100,000 Confidence Bands 0.9-0.99 0.8-0.9 0.6-0.8 0.4-0.6 0.2-0.4 0.1-0.2 0.01-0.1 $1,900,000 $1,700,000 $1,500,000 Confidence Bands 0.9-0.99 0.8-0.9 0.6-0.8 0.4-0.6 0.2-0.4 0.1-0.2 0.01-0.1 90,000 Paths Plan $1,300,000 Paths Plan 80,000 70,000 60,000 2013 2014 2015 2016 2017 2018 Sim Avg 951 662 423 533 645 $1,100,000 $900,000 2013 2014 2015 2016 2017 2018 Sim Avg 951 662 423 533 645 GOVT MedicalClaimsExpense ($000s) GOVT OperatingMargin ($000s) $2,000,000 $1,900,000 $1,800,000 $1,700,000 $1,600,000 $1,500,000 $1,400,000 $1,300,000 $1,200,000 $1,100,000 $1,000,000 2013 2014 2015 2016 2017 2018 Confidence Bands 0.9-0.99 0.8-0.9 0.6-0.8 0.4-0.6 0.2-0.4 0.1-0.2 0.01-0.1 Paths Plan Sim Avg 951 662 423 533 645 $200,000 $150,000 $100,000 $50,000 $0 ($50,000) ($100,000) ($150,000) ($200,000) ($250,000) 2013 2014 2015 2016 2017 2018 Confidence Bands 0.9-0.99 0.8-0.9 0.6-0.8 0.4-0.6 0.2-0.4 0.1-0.2 0.01-0.1 Paths Plan Sim Avg 951 662 423 533 645 Source: ADVISE model; Simulation = 1,000 paths 20

One Adverse Path vs Plan Expectation Government Cause-and-effect modeling tells the story, leading to greater transparency & understanding... Operating Margin: 2013 2014 2015 2016 2017 2018 Expected (Plan) $ 31,708,046 $ 41,313,837 $ 46,943,544 $ 56,710,310 $ 65,987,192 $ 76,644,742 Total Revenue ( Higher / (Lower) ) $ (0) $ (0) $ - $ 0 $ (0) $ (77,690,829) Total Cost of Benefits ( (Higher) / Lower ) $ (0) $ (11,230,794) $ (29,375,587) $ (107,197,984) $ (96,990,810) $ (88,929,023) Net Admin Expense ( (Higher) / Lower ) $ - $ - $ - $ - $ - $ - At a very basic level, Operating Margin is much lower than expected due to higher Cost of Benefits and lower Revenue Actual ( Path = 951 ) $ 31,708,046 $ 30,083,042 $ 17,567,956 $ (50,487,673) $ (31,003,619) $ (89,975,109) 2013 2014 2015 2016 2017 2018 Total Cost of Benefits: Expected (Plan) $1,104,868,979 $1,204,289,981 $1,261,176,105 $1,354,775,921 $1,441,210,812 $1,539,520,398 Medical Trend Impact $ - $ 11,230,794 $ 29,375,587 $ 82,898,231 $ 132,601,170 $ 213,596,868 IT Operational Risk Impact $ - $ - $ - $ - $ - $ - Cyber Security Risk $ - $ - $ - $ 37,496,749 $ - $ 3,192,448 Membership Impact $ 0 $ 0 $ (0) $ (13,196,996) $ (35,610,359) $ (127,860,292) Higher Cost of Benefits heavily driven by unfavorable Medical Trend with some impact from Cyber Security Risk Actual ( Path = 951 ) $1,104,868,979 $1,215,520,775 $1,290,551,691 $1,461,973,905 $1,538,201,622 $1,628,449,421 2013 2014 2015 2016 2017 2018 Total Revenue: Expected (Plan) $1,210,883,595 $1,320,802,066 $1,383,693,888 $1,487,438,343 $1,583,529,875 $1,692,878,672 Membership Impact (Internal CMS Star) $ - $ - $ - $ - $ - $ - Membership Impact (Competitor CMS Star) $ - $ - $ - $ - $ - $ - Prem Rev PMPM Impact (CMS Star) $ - $ - $ - $ - $ - $ - Mgt Reactions (Mbrshp & Prem Rev PMPM) $ (0) $ (0) $ - $ 0 $ (0) $ (77,690,829) Actual ( Path = 951 ) $1,210,883,595 $1,320,802,066 $1,383,693,888 $1,487,438,343 $1,583,529,875 $1,615,187,844 Management response is to increase prices (limited) & shed membership lower membership lowers the Cost of Benefits, but also Revenue Source: ADVISE model 21

Stochastic ECMs Pros and Cons Pros Provides probability statements for capital adequacy will be required as ORSA standards advance (already required in other solvency regimes) Provides basis for allocating the cost of capital, to support better financial performance metrics critical to creating greater value for management beyond compliance Provides better framework for addressing interactions between risk factors Cons Additional work beyond what is require for a pure scenario testing approach (but the good news is all work done on a scenario testing basis can be leveraged) Additional management education required 22

Conning s Risk Solutions Team Chris Suchar, FCAS, MAAA Managing Director Mr. Suchar, FCAS, MAAA, is a Managing Director at Conning. He oversees all software-related professional services in the North American region. Mr. Suchar has extensive experience in managing and supporting deployment of economic capital models in a wide variety of insurance enterprises. Prior to joining Conning in 2010, Mr. Suchar spent ten years as a principal of DFA Capital's Insurance Modeling team and was responsible for all aspects of customer support and professional services in North America. From 1986 through 2000, Mr. Suchar was employed by Milliman as a consulting actuary. Mr. Suchar is a graduate of the University of Pennsylvania with a degree in Economics. Ph. 860-299-2156 Email: chris.suchar@conning.com Manchiu Chan, FSA, CFA, MAAA Vice President Manchiu Chan, FSA, CFA, MAAA, is a Vice President at Conning where he is responsible for providing asset-liability and integrated risk management advisory services to life and health insurance companies. Mr. Chan has over 13 years of actuarial experience in health insurance industry. Prior to joining Conning in 2013, Mr. Chan was employed by Aetna as an actuary working on Medicare and Individual Health products. Previously, he was an actuary at Anthem Blue Cross Blue Shield focusing on commercial group business, and an actuarial analyst at Swiss Re. Mr. Chan earned a MS degree in actuarial science from University of Connecticut and a MBA degree in Finance from University of Akron. Ph. 860-299-2420 Email: manchiu.chan@conning.com

Big Fish Insurance Company: A Brief ORSA Case Study Matthew J. Kramer, FSA, CERA, MAAA June 16, 2016

Everything should be as simple as possible, but not simpler. Albert Einstein 2

Agenda Introduction to ERM, ECM, and ORSA ORSA Case Study: Big Fish Insurance Company (BFIC) Intro to BFIC Section I: Risk Management Framework Section II: Insurer s Assessment of Risk Exposures Section III: Assessment of Risk Capital and Solvency Conclusion Results and feedback Pros and cons of the methodology BFIC used Ideas for future developments after BFIC s first ORSA 3

Introduction to ERM, ECM, and ORSA Enterprise Risk Management: The discipline by which an enterprise in any industry assesses, controls, exploits, finances, and monitors risks from all sources for the purpose of increasing the enterprise s short and long term value to its stakeholders. * Economic Capital Modeling Economic Capital: The amount of capital an organization requires to survive or to meet a business objective for a specified period of time and risk metric, given its risk profile. * Modeling economic capital is a key part of the ORSA. Own Risk and Solvency Assessment Required of insurance companies in most states for 2016 Insurance companies with over $500M annual premium (or groups with over $1B) Must have ERM program Must submit an annual ORSA report to the Commissioner of Insurance *American Academy of Actuaries, Insurance Enterprise Risk Management Practices 4

Case Study: Big Fish Insurance Co. (BFIC) Introduction to BFIC Privately-held regional insurance company with a focus on health insurance. 2015 ORSA compiled in Summer and Fall 2015, submitted to regulator in December 2015. Net Capital and Surplus: $200 million as of 12/31/2014 Net income of $650 million per year $600 million annual net written premium $50 million investment income Lines of business: Large group medical and drug Long term care Long term disability Self-funded ASO 5

Case Study: Big Fish Insurance Co. (BFIC) What is required in ORSA report Section I? Describe existing ERM framework How are risks identified? What is the company s risk appetite? How are risks assessed? What are the company s risk tolerances? What risk controls are in place? What are the company s risk limits? What feedback loops or monitoring processes are in place? 6

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section I BFIC s enterprise risk management (ERM) program in place since 2014: ERM committee Creates and reviews ERM policy Defines risk appetite and risk tolerance Chief Risk Officer is in charge of day-to-day ERM activities Coordinating risk assessment Communication regarding risks Creation and management of risk management strategies Develops ERM tools, practices, and policies Business unit managers are in charge of risk assessment 7

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section I Risk appetite statement emphasizes: Commitment to achieving target profit levels with no reduction in surplus. Maintenance of a 400% RBC ratio. Risk taxonomy: BFIC divides risk into six categories Credit Risk Liquidity Risk Market Risk Strategic Risk Pricing/Underwriting/Reserving Risk Operational Risk (including reputational risk, legal risk) 8

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section I Risk Identification & Risk Assessment Risk identification and assessment is primarily done by business unit managers Risk likelihood is assessed using a score from 1-10, with specific likelihood thresholds for each level Risk impact is assessed using a score from 1-10, with specific impact thresholds for each level. Impact is assessed on a residual basis. Risk score = likelihood score x impact score Risk Tolerance: risk scores over 50 trigger the need for a mitigation strategy. Risk scores over 25 trigger the consideration for a mitigation strategy. Accept risk Reduce risk Share risk Avoid risk Effectiveness of risk mitigation is reviewed periodically 9

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section II What is required in ORSA report Section II? Identify material risks Assess material risks BFIC s Section II was focused on the identification of key risks. Generally qualitative assessments and discussion in BFIC s ORSA report. Pricing/Underwriting risk identified as most important risk by BFIC (this will vary greatly by company) Recent loss ratios were above acceptable targets Identified need to balance financial performance with strategic need to grow blocks of business Identified drivers of pricing/underwriting risk, including recent drug cost trends as well as adverse performance for one jumbo group. 10

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section II Market and liquidity risk second most important risk category Identified need to protect net capital and surplus while being liquid enough to meet cash needs. Operational/Strategic risks also noted as key risks. Regulatory risk considered the top risk here ( Cadillac tax ) Strategic risk assessment focused around provider reimbursement IT risk also identified as key within this category Risk assessment acknowledged long-term nature of BFIC s risks (LTC, LTD lines of business) 11

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section III Goal of Section III: demonstrate your company has the financial resources to execute its multi-year business plan in accordance with risk appetite. Just a few of the questions BFIC had regarding Section III: What time horizon to use? How do we reflect the long-term nature of our business in our projections? How many scenarios do we need to test? What variables should the scenarios test? How do we show the interaction between variables in our scenarios? How do we reflect an assessment of operational risks? 12

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section III More questions and considerations for Section III: How can BFIC be sure it has sufficient capital to withstand an adverse environment? It is important for BFIC to understand the methodology and results: this is BFIC s own risk and solvency assessment. 13

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section III Step 1: create a baseline projection Leveraged BFIC s annual forecast Leveraged year-end reserve adequacy testing for long-term liabilities Created projection of each of the components of RBC calculation Step 2: discuss key risks with BFIC and how to reflect them in scenarios Leveraged key risk identification from Section II Step 3: collaboratively determine which variables to test in scenarios Consideration: ASOP 22 (Statements of Opinion) defines Moderately Adverse Conditions as Conditions that include one or more unfavorable, but not extreme, events that have a reasonable probability of occurring during the testing period. ASOP 22 has a narrow scope but this definition is useful in considering which scenarios to include. Many other ASOPs will apply as well. Step 4: summarize key output variables (net capital and surplus, RBC) 14

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section III 15

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section III We collaboratively developed the following deterministic stress-test scenarios for BFIC s Section III. These were presented in addition to the baseline scenario in the ORSA report. 1. Group medical and drug claims spike +10% in 2016 2. Investment income decreases 200 basis points for 2016-2018 3. LTC morbidity estimate increases 10% 4. Long term discount rates decrease by 25 basis points 5. Fully-insured jumbo group termination 6. Disaster scenario: combination of scenarios 1-5 16

Case Study: Big Fish Insurance Co. BFIC s ORSA: Section III Summary of stress test scenario results: 17

Conclusion BFIC s ORSA: Results and Feedback Key result of Section III scenario testing: BFIC is holding sufficient capital for the short term. BFIC s feedback on the economic capital modeling : ORSA process helped them achieve a better understanding of their risks. Scenario modeling made them more comfortable with the level of capital they hold. Scenario modeling was an effective method of achieving the Section III results in a way that key stakeholders at BFIC understood. Regulator feedback was minimal. This was the first year the ORSA was required in BFIC s state. 18

Conclusion BFIC s ORSA: Pros and Cons Benefits of ECM methodology used for BFIC: Gave BFIC a lot of insight into their risks. Effective way of achieving BFIC s goals. Enabled leveraging of a lot of existing work and processes. Easy to understand for stakeholders at BFIC. 19

Conclusion BFIC s ORSA: Pros and Cons Drawbacks ECM methodology used for BFIC: Leveraging year-end stress testing means using projections that are not entirely up-to-date. Interactions between variables could be reflected better. Can t assign probability to any particular scenario. Measures like value-at-risk or conditional tail expectation are therefore not applicable. Not dynamic enough to assist in day-to-day decision making. Focus on RBC as the key output means that the risk assessment has all the same pitfalls as RBC. Stochastic modeling is more meaningful for long-tail business. 20

Conclusion BFIC s ORSA: Some Ideas for Future Improvements Consider how the use test is being addressed. Risk assessment methodology is a key part of decision making. Could the economic capital also become a key part of decision making long term? Surveying risk owners to develop range of variation for key variables in the projection. More sophisticated modeling may be warranted. Could model more interaction between risks. Could incorporate a stochastic component with Monte Carlo simulation. Operational risks: could develop a severity range based on experience data. 21

Thank you Matthew J. Kramer, FSA, CERA, MAAA Matt.Kramer@Milliman.com June 16, 2016

Qualification Statement and Limitation on Use Matthew J. Kramer is a Consulting Actuary for Milliman, Inc. Matthew is a member of the American Academy of Actuaries and meets the Qualification Standards of the American Academy of Actuaries to render the actuarial opinion contained herein. The views in this presentation are those of the speaker and do not necessarily represent the views of Milliman, Inc., the American Academy of Actuaries, or the Society of Actuaries. This information is intended to give an illustrative educational overview of an Own Risk and Solvency Assessment (ORSA). This information may not be appropriate, and should not be used, for other purposes. 23