Strategic asset allocation: case study by Aspen Re using the internal model to change asset strategy

Strategic asset allocation: case study by Aspen Re using the internal model to change asset strategy Dr Marcus Foley, Aspen Re Sam Worthington, Towers Watson 27 September 2013, Brussels 2013 Towers Watson. All rights reserved.

Agenda Overview of strategic asset allocation (SAA) Aspen: a case study of SAA 2

Why are insurers revising their strategic asset allocation now? Profitability and risk management (on-going requirement) Risk management Returns Stakeholder demand Competition Diversification Achieving best practice risk management Improving risk adjusted returns more generally Improve shareholder ROE and reduce/maintain policyholder premiums React to competitors increasing their investment returns Asset risk is a good diversifier of insurance risk (eg natural cat risk) Catalysts for change Regulatory Investment markets Discounting liabilities, recognising diversification effects, use test Seek out alternative sources of return the current in low bond yield environment 4

What are the steps for implementing SAA? 1 Define SAA scope and constraints 2 Calibrate model 3 Analysis and 4 for SAA recommendations Implementation Agree to review SAA Define risk appetite* Define opportunity set, investment beliefs and general constraints Decide key risk/return metrics to use Decide SAA approach Establish in-house views on asset risk and return metrics Calibrate ESG and asset returns to your views Implement investment constraints Run model and output key risk/return stats Overlay qualitative investment analysis Scenario analysis SAA recommendations with clear rationale Benchmark results with independent views and stress tests Agree SAA changes Buy or build implementation ability Manager selection Negotiating terms Review management information dashboard Transition to new SAA Dynamic asset allocation framework? Periodic SAA review (i.e. annually) *Note defining risk appetite is a substantial project in its own right Capital team Investment team Board / executive management 5 2013 Towers Watson. All rights reserved. Proprietary and Confidential. For Towers Watson and Towers Watson client use only.

Background Background to Aspen Bermudian domiciled specialty P&C insurer and reinsurer $8mm in investible assets More than 800 people worldwide Capital modelling team Capital model 10 years old Built to inform business strategy Board commissioned SAA review in late 2012 - two approaches 1. Major investment house used their standard SAA approach 2. Aspen s capital modelling team used internal model with Towers Watson s SAA model 7

Overview of work Board objectives Long term: to increase the group s book value Short term: assess how to spend surplus capital Share buyback Increasing investment risk appetite External economic considerations Historically low interest rates Low yield on investment grade fixed income assets Recent yield rebounding of sub-investment grade assets from credit Internal considerations Capital constraints Balance sheet volatility Aggregation of asset and insurance risk 8

Asset portfolio before SAA High quality, fixed interest portfolio Interest rate duration around 2 years net of IR swap Investment holdings Government bond, agency 26% Investment grade credit 30% RMBS 15% High yield, CMBS, ABS+ Munis, derivatives 4% Equity 3% Cash 22% 9 2013 Towers Watson. All rights reserved. Proprietary and Confidential. For Towers Watson and Towers Watson client use only.

A comparison of two SAA approaches Investment house Analysis of last 3 years of returns ASSET RETURN CALIBRATION Aspen s internal team From internal, asset manager and Towers Watson forecasts Volatility RISK MEASURE Capital requirement and volatility Total return only RETURN MEASURE Yield and total return Using government bond proxy LIABILITIES MODELLED Using internal model projected liabilities Assumed to be Gaussian DEPENDENCIES Captures non-linear dependency between assets and liabilities 11

Investment house SAA summary Modelling analysis conclusions Significant diversification between fixed income instruments across the credit spectrum High expected returns for credit (extrapolate rebound since 2008) Equity unattractive due to its volatility Recommendation: a significant switch from high rated corporates to subinvestment grade credit Weaknesses Risk aggregation between assets and liabilities poorly captured Credit expected returns over-inflated 12

Aspen s modelling results: asset risk contribution to total risk 400 Breakdown of Asset Risk for Indicated Percentiles 300 200 100 Expected Profit ($m) - -100-200 -300 1 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Percentile Credit Risk Equity Risk Interest Rate Risk -400-500 -600 Interest rate risk is a material risk driver across the distribution Tail risk is most strongly affected by credit risk 13

Comparing investments: capital impact Equity had the highest long term expected return for a high level of risk Credit assets increased capital requirements most 14

Comparing portfolios: total return v RoE Most capital efficient investment options: Invest in equity (high return and medium capital impact) Switch from credit (rather than cash) Invest in agency RMBS (medium return but low capital impact as capital is insensitive to interest rate risk) 15

Qualitative analysis: stress tests Assumptions Portfolio impact ($m) Scenario Period Equity return Yield curve Spreads Base 5% Cash to equity 5% A-rated to equity 10% from treasuries and A-rated to agency RMBS and equity Base 6.7% <+50bps zero 107 134 132 137 Asian Crisis 1997 Interest rate shock 1994 Credit Crunch 2008 Dotcom Crash 2001 Yield Curve Steepening + Spread Widening Sep 1997 - Oct 1998 (13 months) Apr 1993 - Dec 1994 (20 months) May 2007 - Nov 2008 Jan 2000 - Oct 2002 (33 months) Prospective Scenario 14.8% zero 4.2% approx +350 bps -53.5% zero -45.4% zero -3.0% +100 to +250 bps significant widening Assessment of balance sheet volatility (infrequent but regular risk) Determined maximum threshold for equity investment = 5% Validation of quantitative model results = board comfort Intuitive communication of risk 103 162 168 172 zero -317-300 -263-278 extreme widening significant widening moderate widening -562-776 -706-702 -124-305 -302-297 -224-236 -203-217 16

Recommendations Portfolio reallocation Increase in total return ($m) Increase in income ($m) Increase in risk capital ($m) ROE of additional investment 5% from cash into equity 27 13 108 12% 5% from A-rated corps into equity 25 6 31 20% 10% from US treasuries and A-rated into agency RMBS and equity 30 14 30 46% Increased income due to two factors: Increased risk appetite (taking on more risk) More efficient asset allocation 18

Implementation In May investment committee recommended a 5% investment in equity Tactical consideration: easier to switch from cash than A-rated corporates Since then, market movements mean EMD yields have risen materially Used SAA analysis to identify EMD was attractive from a risk perspective. Re-ran analysis with updated EMD yields Changes implemented 2.5% switch from cash to equity 2.5% from A-rated corporate to EMD 19

Key takeaways SAA possible because capital model is widely used and understood Using the internal model liabilities can generate very different SAA results to using a liability proxy for a P&C insurer Qualitative analysis was critical in providing: validation of internal model analysis an intuitive explanation to communicate results to the board 20