Session 83 PD, Modeling Managing and Pricing Living Benefits Risk. Moderator: Sean Michael Hayward, FSA, MAAA

Transcription

1 Session 83 PD, Modeling Managing and Pricing Living Benefits Risk Moderator: Sean Michael Hayward, FSA, MAAA Presenters: Guillaume Briere-Giroux, FSA, MAAA Sean Michael Hayward, FSA, MAAA Eric L. Henderson, FSA, CERA, CFA, FRM, MAAA

2 Modeling, Managing and Pricing Living Benefit Risks Overview of Industry Approaches 2014 Life and Annuity Symposium Atlanta May 20, 2014 Guillaume Briere-Giroux, FSA, MAAA, CFA 2014 Oliver Wyman

3 Overview of industry approaches I. What living benefits? II. What risks? III. What scenarios and what value lenses? IV. Industry modeling practices After this overview, co-speakers will expand on select modeling, pricing and risk management issues with a case study focused on a fixed indexed annuity ( FIA ) with guaranteed living withdrawal benefit ( GLWB ) 2014 Oliver Wyman 1

4 What living benefits? Lower Market Risk Higher Insurance risk type Non-elective Both elective and non-elective Mostly elective Elective Size of bubbles represents order of scale for recent new business volumes (LTC converted to single premium equivalent) Lower Insurance Risk Higher Sales data from LIMRA 2014 Oliver Wyman 2

5 What insurance risks? Product Longevity Base lapse Dynamic lapse Withdrawals or annuitization Morbidity VA GMAB Risk level VA GLWB VA GMIB High Low FIA GLWB* SPIA DIA LTC *With nursing home benefit 2014 Oliver Wyman 3

6 What market risks? Product Credit Interest rates Equity Volatility VA GMAB Fund correlation / basis risk Risk level VA GLWB High VA GMIB Low FIA GLWB* SPIA DIA LTC *With nursing home benefit 2014 Oliver Wyman 4

7 What scenarios and what value lenses? Nested stochastic Stochastic Deterministic + sensitivities Deterministic Simple Complex Economic scenarios Behavior scenarios None Static behavior scenarios Dynamic policyholder behavior Integrated dynamic behavior scenarios Size of bubbles represents order of scale for recent new business volumes (LTC converted to single premium equivalent) Real World Value lenses Risk Neutral Sales data from LIMRA 2014 Oliver Wyman 5

8 Industry modeling practices Product Stochastic equity returns (RW) Stochastic interest rates (RW) RN cost of guarantees Behavioral cohorts Dynamic behavior VA GMAB? VA GLWB VA GMIB? FIA GLWB* SPIA? DIA? LTC?? 2014 Oliver Wyman 6

9 Key points Market risks impact pricing approaches Accounting and risk management practices drive scenario layers Behavior risk drives modeling granularity and complexity Assumption modeling is becoming increasingly sophisticated 2014 Oliver Wyman 7

10 Modeling, Managing and Pricing Living Benefit Risks Modeling Considerations Indexed Annuity GLWB Case Study

11 FIA GLWBs were selected to discuss modeling considerations for living benefits Growing popularity of design Significant policyholder optionality Stochastic modeling is common Asset and liability modeling Interaction between assets and liabilities 9

12 A representative FIA GLWB model was built to support the pricing and risk discussion 10 year surrender charge design Annual point-to-point cap crediting 6% compound rollup on benefit base Representative income rates No nursing home doubler Next, we cover modeling considerations and use the case study to exemplify a sample of modeling practices 10

13 FIA modeling has come a long way Model as fixed annuity - Statutory reserves as a % of fund value Simple option budget approach - Assume perfect static hedge New product features & more computing power allowed more robust modeling - Examples in case study 11

14 But complex modeling issues remain 1. Renewal rate setting 2. GLWB modeling 3. Dynamic policyholder behavior 4. Profit bases and metrics 5. Hedging 12

15 1. Renewal Rate Setting Management has discretion in setting renewal credited rates and caps Typically captured using option budget approach - Assumes no unexpected gains/losses Defaults Hedge mismatches Requires a portfolio earned rate to set option budget - How do we model this in liability only nested projections? As modelling capabilities improve, do we need to adjust management action algorithms to reflect factors other than the asset yield? 13

16 1. Renewal Rate Setting Case Study Approach Option budget = net asset yield - spread Solve for cap using closed-form solution Consider volatility strike skew 14

17 2. GLWB Modeling GLWB elections impact the illustration of policyholder balances - Haircut does not work, need to model cohorts of election Elections are just that elective! Do we need to model them dynamically? - Implications on model memory usage and distributed processing Statutory reserve impacts - Model every possible election point in CARVM? GAAP reserve impacts - SOP 03-1, but any FAS 133 considerations? 15

18 2. GLWB Modeling Simplified Cohort Approach 16

19 3. Dynamic Policyholder Behavior Dynamic lapses need a credited rate and competitor rate - Difficult to compare caps across crediting strategies Option budget serves as a proxy Inclusion of riders changes dynamics - In-the-moneyness will impact lapses - Does inclusion of guarantees or different fee structures impact the competitor comparison? Do we have credible experience as product designs keep changing? 17

20 3. Dynamic Policyholder Behavior Case Study Approach Interest-sensitive lapses on base contract - Option budget serves as a proxy In-the-moneyness of riders dampens lapses - S curve like schedule of lapse multipliers Therefore, product reacts to both index credits and the interest rate environment - Magnifies impact of hedge or ALM mismatches 18

21 4. Profit Bases and Metrics Pricing exercises often focus on statutory based metrics - IRR, profit margin - Accurate CARVM reserves Integrated GAAP projections nice to have - GLWB riders require SOP 03-1 liabilities in addition to FAS 133 reserve Embedded value - Important for European based companies - Supplement to statutory based metric or replacement? 19

22 5. Hedging Several layers of complexity Base product credit hedging - Static call spreads - Dynamic hedging GLWB rider hedging - Futures - Delta - Call options - Gamma/Vega - Net exposure against credit hedge? 20

23 5. Hedging Case Study Approach Dynamic hedging for GLWB Additional inner loop computations for Greeks Valuation date Month 1 Month 2 Month 3 Month 4 Month 5 Γνρ Γνρ Γνρ Γνρ Γνρ Γνρ

24 Case Study Reasonableness checks using simple deterministic scenarios Case Study Pricing IRRs - Sample Deterministic Scenarios Scenarios Base only, hedge credits GLWB included, hedge credits GLWB included, hedge credits and GLWB Slow increase in yields, 2.5% index credit 12.7% 13.6% 13.8% Slow increase in yields, 3.5% index credit 13.6% 17.1% 13.4% Slow increase in yields, 1.5% index credit 11.7% 9.9% 13.6% Low rates/low index returns 9.7% 6.7% 11.9% More exposure to low rates and low equity returns when GLWB is present, magnified by dynamic behavior More stable results with GLWB dynamic hedge 22

25 Modeling, Managing and Pricing Living Benefit Risks Pricing and Risk Considerations Indexed annuity GLWB case study

26 Agenda Setting the stage Case study results and discussion Product management considerations Concluding remarks 24

27 Setting the stage Large risks vs. large risk premiums Interest rate risk is material for FIA income riders - This risk does not materially impact average pricing result Equity risk is smaller and muted by the 0% floor - Equity risk premium can have a large impact on profit metrics Start with case study to make the point 25

28 Impact of Living Benefit Hedge Reduction in range of profit margin ( PM ) outcomes per goal of hedge program Risk/reward tradeoff of hedge less compelling Remaining interest rate risk material Profit Margin 7.00% 6.00% 5.00% 4.00% 3.00% 2.00% 1.00% Interest Credit Hedge PM Economic Hedge PM Mean 2.97% 1.65% Std Dev 1.14% 0.76% 0.00% 1.00% 2.00% Interest Credit Hedge Economic Hedge 26

29 What explains this result? 1. Equity risk premium lead call options to produce excess returns 2. This can improve pricing results if the hedge notional is greater than the economic hedge Scenario Option Payoff OC 27

30 What is the economic hedge? As shown below, economic hedge ratios* vary across product types Product Type FIA no rider FIA w/fixed roll-up income rider FIA with indexed rollup income rider Economic Hedge Ratio* at issue ~80% ~55% ~70% Base FIA product: Time value of money and surrender charges Presence of guarantees: Certain benefit streams are insensitive to index credits (e.g., high rollup GLWB) *Economic hedge ratio defined here as % of account value hedged that results in stable level of accumulated surplus at end of projection 28

31 Understanding the pricing impact without hitting the grid Impact on profit margin dependent on: - Expected option return - Option budget - Liability duration If not passed through the product design, the overhedge can be considered a capital-efficient surplus investment Equivalent Yield and PM Contribution from the Equity Risk Premium % Over hedge (% of AV) 40% Option ROI 30% Current yield 5.0% Option cost (% of AV) 2.0% Over hedge pre tax yield contribution (% of AV) 0.20% Liability duration 8.0 Over hedge profit margin contribution 1.04% 29

32 Product management consideration #1: Static or dynamic hedging strategy? Building on the prior example, we show that the pricing impact of hedging the GLWB can depend on the choice of the hedging strategy Dynamic Delta hedging replication strategy gets paid less than selling the option, leading to a higher cost of hedging the GLWB Careful to not distort benefits from dynamic hedging in pricing. The result is highly dependent on economic inputs. Estimating cost of hedging GLWB Reduction in notional (% AV) to hedge guarantee Option cost saved (or replication cost received) Static Hedge 40% Delta Hedge 2% 1.8% Option payoff foregone 2.6% 2.6% Option ROI 30% 45% Current yield 5% Pre tax yield contribution foregone (% of AV) 0.20% 0.29% Liability duration 8.0 Cost of hedging (lost profit margin) 1.04% 1.48% 30

33 Product management consideration #2: Typical hedging / pricing practices explain the pull toward strong guarantees Consider a product with rollup rate = index credit + 3% (as opposed to guaranteed 6%) Traditional pricing approaches disadvantage this design: - When already hedging the account value, the modeled cost of adding the feature is the expected value of benefits (high cost since long the risk premium) - When already hedging the GLWB, the modeled cost of adding the feature is the cost of hedging the benefit (i.e., less than the expected benefit) Even with AG 33 reserving and risk benefits, the resulting design and customer value proposition is uneven (see chart) Monthly Income 13,000 12,000 11,000 10,000 9,000 8,000 7,000 Guaranteed Income Rider Participating IR (PV of Nonguaranteed income) Participating IR (Cost of funding) Guaranteed High Expected 31

34 Product management consideration #3: Calibrating the equity option ROI Higher equity option ROI increases real world pricing metrics Increase is greater when guaranteed elements are greater Calibrating ERP without monitoring the option ROI can result in different option budgets by hedge strategy and other inconsistencies - On approach is an option ROI calibration for pricing that reflects the opportunity value of making this investment, not its expected return IRR 10% ROI 30% ROI IRR Sample base policy 16.48% 17.23% 0.75% Sample GLWB 15.07% 17.14% 2.07% 32

35 Product management consideration #4: Adjusting pricing return targets Desire to maintain aggregate pricing targets Individual products equally share company over-hedge gains by adjusting product specific pricing targets - Guaranteed benefit products (with associated high over-hedge) receive a smaller share of index option gains - Base product and participating rider benefit design receive more FIA Portfolio Base Product Fixed Income Rider Over hedge (% of AV) 40% 23% 46% Option ROI 30% 30% 30% Option cost (% of AV) 2% 2% 2% Current Yield 5% 5% 5% Over hedge pre tax yield contribution (% of AV) 0.24% 0.14% 0.28% Duration Over hedge contribution to PM PM adjustment from current 1.2% 0.6% 1.5% +0.6% 0.3% 33

36 Concluding Remarks Risk management impact on pricing differs by risk Complement complex modeling with top-down analysis Be a thoughtful advisor when pricing and managing through these issues 34