Micro Simulation Study of Life Insurance Business Lauri Saraste, LocalTapiola Group, Finland Timo Salminen, Model IT, Finland Lasse Koskinen, Aalto University & Model IT, Finland
Agenda Big Data is here! I. Modelling Framework Design Computations II. The Life Business Company Overview Modelling Approach III. Results & Conclusions
I Modelling Framework
Literature and Software Finnish actuaries traditionally good at utilizing new technology - many papers Published research on individual modeling. Examples. : Antonio K. and Plat R.: Micro-level stochastic loss reserving, 2012 (Non-life real case study) Gustafsson, E.: Simulation based claim reserving in general insurance, 2012 (Includes SII considerations) Rantala, J: Estimation of IBRN claims, 1983 (Early ideas) Here computations performed by Matlab/ MathWorks and cframe (msii)/ Model IT
Background (1/2) Background Quoting England and Verrall, 2002: With the continuing increase in computer power, it has to be questioned whether it would not be better to examine individual claims rather than use aggregate data. Here we utilize micro-level insurance data for better informed decisions by performing large simulations Example: Standard PC + Graphic Cards => Tens of Billions (10 9 ) Black Scholes per Second Now it is 2013! Individual simulation is feasible, but fast computers, large memories and efficient algorithms are still crucial Fast development each year: Both algorithms and hardware. Future looks even better!
Background (2/2) When micro-level information leads to better decisions? Causal information is available Causal information is correct Information can be communicated A heavy computational application that actuaries encounter nowadays is the use of economic scenario generator (ESG). A key element of : Market consistent valuation for insurance businesses Business modeling and ORSA Available computational power can be used several purposes: ESG, Individual Simulation, Extreme Value estimation etc.
Model Overview External input Asset Valuation Full yearly distribution SII Pillar 1 MCEV SCR Economic scenarios Company decisions Liability Simulation of Cash Flows Income Statement Balance Sheet ERM and ORSA Business Decisions
Balance Sheet Liabilities Assets Based on policy-level SCR attribution and current* investment risk Equity SCR MCR Fixed income investments Priced with current* yield curve Expected value of future cash flows For policies still active in current* simulation Adjusted with evolved short rate and current* yield curve Discounted with current* yield curve Technical reserves Equity and other investments Unit Link investments Modeled as a basket of ESG indices Sum of current* unit link policy savings * Current = current simulation path and current time step
ESG (1/2) Quite simple ESG is used / not exact statistical fit. Published research articles => Parameters for the ESG model. A more sophisticated feature: ESG is its dependence on business cycle. Homogenous Markov process S(t) where the probability to stay in expansion regime S(t) = 1 is 0.97 and the probability to stay in recession regime S(t) = 1 is 0.40. We assume that the primary process is that related to GDP development. This drives the other processes inflation and equity return. Inflation in turn drives nominal interest rates. Yield-curve modeling methodology relies on a stochastic state space formulation of the Nelson-Siegel. Here model follows autoregressive processes. Interest rates incorporate regimeswitching behavior in the time-series evolution through the above inflation process.
ESG Part (2/2) Monthly inflation follows autoregressive process y(t) 0.01 = 0.63 x [ y(t-1) -0.01 ] + 0.013 x e(t), when S(t) = 1 (expansion regime) y(t) 0.08 = 0.4 x [ y(t-1) -0.08 ] + 0.038 x e(t), when S(t) = 2 (recession regime) Monthly stock returns p(t) by a two-regime mixture model. Interpretation: regime one produces bubble periods and regime two periods which bring prices back to fundamentals. p(t ) = 0.027 + 0.027 x e(t) when S(t) = 1 (expansion regime) p(t ) = - 0.3 + 0.059 x e(t) when S(t) = 2 (recession regime)
Simulation Task 1: Measuring the effect of management policies on profitability and solvency (under Solvency II) Full stochastic real world policy-by-policy simulation Customer behavior modeling Multiple policy types with embedded options Path-dependent mark-to-market balance sheet simulation Task 2: Dynamic management actions Profit sharing between dividends, customer benefits, equity Investment strategy changes based on financial position and expected liability cash flows New sales policy (run-off going concern)
Simulation cycle & High performance computing 1. Simulate or import economic scenarios 2. Go through all years T 1. Go through all customers N 1. Simulate customer s random events (death, disability ) 2. Go through all customer s contracts (usually 1) 1. Simulate contract s random events (surrender, payment, ) 2. Go through all time steps in a year (1-12) 1. Calculate contract s cash flows for M simulations 2. Terminate contracts that have encountered a termination condition for M simulations 2. Generate company balance sheet and make company decisions for D simulations The biggest loop (customers, N) can be distributed to a computing cluster Local multicore, Computing cluster, Cloud computing
ORSA features Accurate calculation of mark-to-market technical reserves without nested stochastic simulation Regression approach taken Short term simulations with mark-to-market balance sheet can be run with very high simulation count Up to 1 000 000 simulations with 1 year horizon Up to 10 000 simulations with 1-10 year horizon Accurate tail risk calculation Real World ESG has two regimes boom and recession Full simulation of the hypothetical company takes ~ 1 hour. In this presentation small samples ~ 1 minute
II The Life Business - Company Overview
Company overview (1/3) A publicly listed (hypothetical) life insurance company Solo structure with policyholders and investors as the main interest group All liabilities are euro denominated Solvency II and MCEV are closely followed Policy groups are: Pure risk policies 0% and 4% guaranteed rate savings products with all life pension payments and options to switch between guaranteed fund and unit linked funds All three policy groups have 100 000 policyholders each Estimated year premium from the insurance portfolio including new sales is around 500 million euro
Company overview (2/3) 3500 3000 Age distribution of policyholders Policyholders in 4% & UL Policyholders in 0% & UL 2500 2000 Age distribution of risk policies 2500 1500 2000 1500 1000 1000 500 500 0 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 87 90 93 0 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 82 86 90 35000 30000 25000 Number of policyholders based to amount of savings Savings 4% & UL 5000 4500 4000 3500 Number of Risk Policies by risk amount 20000 15000 10000 Savings 0% & UL 3000 2500 2000 1500 5000 1000 0 500 0 10000 50000 90000 130000 170000 210000 250000 290000 330000 370000 410000 450000 490000
Interest rates quite low, Future profits diminished EOF < BS equity SII rate declines 31.12.2012 Million Life Company Assets Investments Equities 784 Bonds 6 045 Loans 536 Cash 160 Investment backing unit-linked liabilities Company overview (3/3) 6 218 TOTAL 13 743 Equity Core equity Share capital 350 Retained earnings 936 Sub-Ordinated loans 100 Liabilities Insurance liabilities Profit sharing business 6 125 Pure risk policies 42 From policies where policyholder bears the investment risk 6 190 Solvency position Life Company Eligible Own Funds (Solvency II) 810 SCR 919 MCR 306 Solvency rate 88 % SCR (31.12.2012 Million ) 919 Adjustments -245 Operational risk 69 Basic Solvency Capital Requirement 1 096 Diversification benefits -297 Sum of risk components 1 393 Market risk 843 Interest rate 350 Equity 280 Property 0 Spread 356 Concentration 60 Currency 38 CCP - Counterparty default risk 60 Life Underwriting risks 490 Mortality 122 Longevity 244 Disability 121 Expense 119 Revision - Lapse 207 Life catastrophe 48 Health underwriting risk - TOTAL 13 743 Non-Life underwriting risk -
Overall strategy: Company policies (1/3) Main business area is insurance and capital is primarily budgeted for it Investments are seen as secondary business area getting the residual capital Now solvency position is challenging 88%. Company aims to get its solvency II position over 140% in the strategy period (5 years) It is extremely important to give benefits for the policyholders Also it is crucial to increase the share value and pay dividends. Sub strategies will be used to specify and bring the strategy to business level actions
Company policies (2/3) P & L + Balance Sheet MCEV Main Strategy Investment policy Solvency II Profit sharing policy Capital management policy Policy for overall solvency position Policy for writing new business Re-insurance policy Company decision making
Company policies (3/3) Company decision making - binding together all the policies No Solvency & business trigger Profit sharing Risk tolerance Capital management 1. (EOF/SCR > 300%) Benefits are granted amount of 30% of positive profit. Shares are payed amount that cuts the over capitalization into 300% Since the amount of (Assets - liabilities - 2/3*SCR) is invested in equitys, this increases the company Over capitalization is taken care in risk position quite much profit sharing part 2. (140% < EOF/SCR) && dmcev > 1.1 3. (140% < EOF/SCR) 4. 100% < EOF/SCR < 140% Benefits are granted 30% of the positive profit. Shares on the other hand are granted 50% of positive profit in case of substantial increase in MCEV - - Benefits and Shares are granted both 30% of the positive profit - - Benefits are halved (if positive profit) and shares are set at zero 5. 2/3 < EOF/SCR < 100% - Equity proportion is redused as EOF diminishes - Equity proportion of the portfolio is minimized now. Also worst rating class of bond holdings is lifted up into A or less risky. Sub-ordinated loan is raised with spread equal to A + alpha. This amounts up to (1.1xSCR - EOF) but makes only Tier 2 type of capital (and max is 50% of Tier 1 EOF) 6. MCR < EOF/SCR < 2/3 - No new policies will be accepted. All assets are re-allocated into A or less risky bonds. New share capital will be collected, total value an amount up to (SCR - EOF) 7. EOF/SCR < MCR Bankruptcy Bankruptcy Bankruptcy
II The Life Business - Modelling Approach
Customer behavior (1/3) 4 levels of customer behavior are modeled Level Description Example Global/Macro Company Customer Effect of global (economic) state Effect of company decisions Effect of customer properties High equity returns increase Unit Link attractiveness Good benefit history increases traditional savings product attractiveness Young age increases surrender intensity Contract Effect of contract properties Short duration increases surrender intensity
Customer behavior (2/3) Historical equity returns (UnitLink attractiveness) increase Switch Traditional UnitLink Surrenders New business Switch UnitLink Traditional Paid benefits (Traditional savings product attractiveness) increase
Customer behavior (3/3) Policy-by-policy surrenders and lapses are modelled In ESG there is a general model that estimates lapse and surrender probabilities using macro-economic data Company specific distribution is used (not real data in this exercise) The distibution of Lapse Probabilities for Savings policies 100% means the same lapse probability as in ESG for a individual Age Policy Duration 18-30 30-40 40-50 50-60 60-65 Over 65 0-2 537.60 % 470.40 % 184.80 % 151.20 % 252.00 % 84.00 % 3-5 403.20 % 352.80 % 138.60 % 113.40 % 189.00 % 63.00 % 6-10 107.52 % 94.08 % 36.96 % 30.24 % 50.40 % 16.80 % 11-15 94.08 % 82.32 % 32.34 % 26.46 % 44.10 % 14.70 % 16-20 40.32 % 35.28 % 13.86 % 11.34 % 18.90 % 6.30 % 21-30 53.76 % 47.04 % 18.48 % 15.12 % 25.20 % 8.40 % Over 30 107.52 % 94.08 % 36.96 % 30.24 % 50.40 % 16.80 %
Approximating the Solvency II position Assets and liabilities are valued using ESG Liabilities with future benefits as in next slide There is assumed to be no issues about eligibility of Tier classes (all BOF counts as EOF). SCR market risk components are calculated separately: Both interest risk and spread risk are calculated more precisely as the cash-flow structure and rating classes can t be ignored For equity risk the risk dampener is set to zero Other market risks these are assumed to develop hand-in-hand with the overall asset fluctuations. SCR life risks are estimated separately for risk-, guarantee rate savings- and unit linked policies. As life SCR risk components are calculated stressing the best estimate liabilities it is assumed that all SCR life risks follow the certain proportion (of market value liabilities) of what they are at the starting point. Total SCR is calculated based to different modules and adjustments.
Approximating market consistent technical reserves without nested stochastic simulation 1. Run 60 year market consistent simulation to calculate technical reserves at t=0 1. Store contract-level means of net cash flows and account values for each savings account and time step separately 2. Calculate time dependent regression analysis of net cash flows deviation from mean for each product segment for each decision point against each systematic risk factor in the model 2. Run 5 year real world simulation and at given time t in given simulation path k 1. For each contract that s alive, accumulate expected future cash flows and adjust with evolved account values 2. Apply regression coefficients based on current systematic risk factor values for projected net cash flows for each product segment 3. Discount the projected net cash flows with current yield curve
Approximating the decision making process No Company decision making - Algorithmic approach Solvency & business trigger Profit sharing Risk tolerance Capital management 1. (EOF/SCR > 300%) As in No 2 or No 3 depending of dmcev but after the payments an extra dividend amounting EOF - 3SCR (if positive) will be payed. This will automaticly work through reallocation processes profit sharing Overcapitalization is taken care in part 2. (140% < EOF/SCR) && dmcev > 1.1 3. (140% < EOF/SCR) 4. 100% < EOF/SCR < 140% As in No 3 but with extra 20% into dividends Normal stage - no changes Normal stage - no changes If profit before benefits & taxes is positive, 30% from this will be for benefits and 30% into dividends. Benefits increases the guaranteed funds and shares are just out of the company (but will be taken into account when measuring the company) Normal stage - no changes Normal stage - no changes As in 3. but benefits are only 15% and dividends zero Re-allocation (by Euler) takes care of reducement of market risk as there will not be any changes into the amount of Life risk No changes at this stage 5. 2/3 < EOF/SCR < 100% No benefits or shares The proportion invested into B rated bonds are allocated into less risky (A rated) bonds. The proportion in equities is reduced automaticly towards zero as EOF reaches 2/3 SCR. Sub-ordinated loans will grow and amount of (EOF - 1.1 SCR) assuming that there has not been any other new loan in current simulation (one time event in strategy period 5 years). 6. MCR < EOF/SCR < 2/3 No benefits or shares The parameter controling the maximum amount of new policies will stop all sales of new policies. Proportion invested into A rated bonds will be allocated into AAA class. 7. EOF/SCR < MCR Insolvency Insolvency Insolvency Share capital will increase by an amount of (EOF - SCR) assuming that there has not been any other new share capital in current simulation (one time event in strategy period 5 years).
III Results & Conclusions base case
III Simulation results ESG Stock returns Probabilty intervals Correlation: Solvency ratio vs. Stock return Outliers?
III Simulation results Solvency II measures Equity is 1.4 x SCR and 4 x MCR as a mean level in year four (2016) Distributions widen quite much The cumulative probability of going insolvent (EOF < MCR) is sufficiently low
III Simulation results future dividends The amount company is able to pay dividends (according to its policy) Neglible in 2013 Grows as its solvency position recovers
III Simulation results Correlations Surender vs. Inflation in 2016 OBS.: Two regimes in inflation Stock investment vs. Benefits
III Results & Conclusions case: Modifying benefits
Case Overview Paid benefits have a direct impact on company solvency (less equity) Also customer behavior and paid benefits was assumed to have a clear link between each other. Goal: Find the model s feedback for paying High benefits: 60% of company profit as benefits to policy holders Low benefits: 10% of company profit as benefits to policy holders Three measures: Achieving solvency target (measured by solvency ratio) Sustaining business (measured by policy count) Impact on shareholders
Results: Achieving solvency target The target for future benefits on the profit sharing policy has a key impact on solvency position High benefits Low benefits Mean 1.45 Mean 1.66
Number of 0% savings policies Results: Sustaining business Based on the customer behavior model there might be significant risks on setting the target for future benefits 10 3 10 3 High benefits Low benefits Mean 75 000 High benefits success to change the customer flow Mean 68 000
Results: Impact on shareholders Lower benefits has a straightforward impact on the dividend payments Finding a balance between these on the profit sharing policy will be important Cumulative dividend Cumulative dividend High benefits Low benefits Mean 2.7*10 8 Mean 4.2*10 8
III Results & Conclusions
Conclusions I Individual level simulation is feasible computational tool for life insurance modeling Flexibility to model the future cash flows realistically Better information can potentially lead to better decisions Effective communication critical! Computational power needs to be focused on essential tasks: most demanding or frequent; Full data or small sample? Specific / tailored algorithms needs to be developed for (e.g.): Contract level cash-flows Avoiding nested simulations in ERM/ORSA Parallel algorithms Fast data structures Tailored versions of known computational statistics / finance algorithms + know new methods Interactive tools needed for interactive model development / use
Conclusions II For business decisions It should be clarified which metrics to follow (here SII and MCEV levels) well-defined strategy and harmony between sub strategies appropriately taken into account in the model will help Causality structure drives results Interactive modeling a tool for scrutinizing stochastic causality Stochastic sensitivity analysis All model components need to be realistic enough so that the overall process complexity can be separated into understandable parts. The link between policyholder behavior and company was here approached via economic scenarios resulting lapses more interrelationships (like other events that cause lapses) could have also been needed For Solvency II purposes this kind of modeling would probably cover many of the ORSA requirements.
Thank You! Modeling is not always easy, but it makes actuaries real all-round experts!