9, Beijing

Dr. Andrew Coburn, Senior Vice President Dr. Chris Hornsby, Model Development and Longevity Model Lead Best known for our catastrophe risk models Life division since 2005 Work with most major insurance and reinsurance companies in US & Europe $2 trillion worth of insurance and capital markets transactions based on RMS Risk Models Trusted by regulators and rating agencies for over 20 yrs RMS mortality and longevity models used for rated capital market transactions Mortality Shocks Pandemic Influenza

Natural Hazards Terrorism

Industrial Accidents Emerging & other causes Infectious Diseases

Regenerative Anti-Aging Medicine Processes

Health Lifestyle Environment Medical Intervention Trend Risks A specialized team to produce solutions for the life insurance industry . Swiss Re VITA series . 2009, 2010, 2011, 2012 . Excess Mortality bonds . RMS Excess Mortality models rated by Standard & Poor’s . Swiss Re KORTIS . Successfully closed and publicized Dec 2010 . First ever successfully rated longevity bond . RMS Longevity Model rated by Standard & Poor’s . AEGON / Deutsche Bank longevity transaction . ‘Record’ size of a longevity transaction to the capital markets . Used Netherlands longevity modeling . Viewed as a “game changer” by other clients

RMS article on psycho-social metrics Longevity Risk Management: Fall 2012 for longevity risk featured on the cover RMS guest-edited edition of of this month’s special edition of Institutional Investor The Actuary on underwritten annuities RMS LifeRisks Online resource center: http://www.rms.com/liferisks

• Seminar presentations • Whitepaper series • Articles and commentary • LifeRisks Monitor • Collateral • Client Alerts and Advisories • Press releases

A Century of Change and then… 40 Years of Unrelenting Improvement Health Environment

■ Detailed analysis of the causes of the recent phase of high improvement by 4% epidemiologists and demographers Past 20 yrs ave 3.25% 3% ■ Changing lifestyles Medical - big influence from reduced smoking Intervention Lifestyle Past 50 yrs ave 2% 2.0% ■ Medical intervention a significant cause Past 100 yrs ave - preventative drugs like antihypertensives and statins 1.25% 1% - Improved treatments for heart attacks Past 160 yrs ave 0.75% ■ Health environmental causes an additional 0% minor contributor 1850 1900 1950 2000 -1% Mortality Improvements UK males aged 65 Improvements affect multiple causes of death

Deaths per thousand, Males ■ Understanding how different causes of death are changing is a 8 useful extra dimension of observation ■ The insurance industry should collect cause of death data from its policy- 7 holders 6 ■ Causes of improvement operate across several causes of death 5

4 ■ Co-morbidity and interaction between medical conditions Cancers complicates the picture 3 ■ Structural models using ‘Cause of Improvement’ have proven more 2 tractable than pure ‘Cause of Death’ models 1 Infectious Disease 0 1970 1980 1990 2000 2010 Bioscience identifies a large number of causes that contribute to mortality change

Lifestyle Regenerative Medicine • Smoking New classes of treatment for repairing • Obesity damaged systems e.g. • Other lifestyle trends • Stem cell therapy • Medical Intervention • Individualized Treatments for specific conditions: • Improvements in transplantation • Cardiovascular Disease Anti-Aging Processes • Cancer treatments • Respiratory Disease Treatments to extend life through slowing • Dementia natural processes of aging, e.g: • Other key diseases • Telomere Shortening • IGF1 Health Environment • Caloric restriction • Healthcare provision • Sanitation, housing, environmental factors Projecting Future Smoking Rates

Smoking Rates Over Time 60% Adult Males, England and Wales, Office of National Statistics

50%

40%

30%

20%

10%

0%

Future Foundation Projects 1980 1990 2000 2010 2020 2030 2040 Up in smoke: Quitting Smoking in the 21st Century Nicorette & Shire Health PR March 2005 Impact of Smoking Reduction on Mortality Improvement 20%

15% Proportion of 70 yr-old 10% men that smoke 5% Smoking ‘WipeOut’ Scenario 0% 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100

1.5% 1.0% Smoking ‘WipeOut’ Mortality Improvement 0.5% Rate for 0.0% 70 yr-old -0.5% men -1.0% -1.5% 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Medical breakthroughs and advances are sporadic

■ Observations of the process of medical advance shows that it is serendipitous and subject to significant randomness

Fj(x,t) ■ Medical progress is punctuated by reversals and setbacks – for example discovering that a treatment causes side-effects

■ It has long been observed that mortality improvement has the characteristics of a random-walk

■ Structural modelling of cause of improvement is the first model to explain WHY

0 10 20 30 ■ An important contribution of bioscience is the parameterization of this Years into future ‘meta-model of medical progress’

Historical timelines for new medical innovations to impact mortality Discovery 10 20 30 40 50 60 years Antibiotics 1910

Tuberculosis 1913

Penicillin

1929

Statins 1979

Breast cancer 1959 The RMS Longevity Risk Model Concept

Standard Lee-Carter Model Statistical projections of past experience are proven and powerful tools of actuarial expertise, but extrapolating past volatility can project medically implausible futures.

Overlaying a Vitagion Structure Enhanced statistical projection by adding future projections by causes of improvement, provides realistic medical constraints on what could happen in future.

Multidisciplinary Scientific Research Parameterization of vitagion models is informed by a detailed research program of best-of-class data, and sub-models of medical outcomes. 1.1

1.0

0.9

Mortality 0.8 Reduction Factor 0.7

0.6

0.5

0.4 1st Percentile

5th Percentile

25th Percentile

Median

75th Percentile

95th Percentile

99th Percentile Longevity Risk Characterizing the nature of Mean Curve lifestyle change and medical 99.9th Percentile progress in a full stochastic model 99.9t9h Percentile enables the uncertainty structure to be quantified 0.1st Percentile 0.01st Percentile 100%

Lifestyle

80% • Fast increase in healthier lifestyle Health • Smoking becomes de-normalized; Environment • Obesity trends slow dramatically

60% Medical Intervention

• Rapid Progress in Medical Intervention 40% Regenerative • Cancer management improves Medicine dramatically. • New monoclonal antibody drugs are Anti-Aging effective and cheap. Treatments • Continued rapid reduction of 20% premature deaths from cardiovascular disease

0% 2012 2022 2032 2042 2052 2062 What signals will indicate which path will emerge?

■ As we progress through time, the uncertainty around future years’ mortalities will reduce

■ Certain events will make particular paths more likely

■ Watch out for ‘Sentinel’ events that signal changes in the mortality regime - The end of the obesity epidemic - Smoking de-normalization - FDA approval of low-cost stem cell treatments - Radical change of the NHS - Low-cost drugs for reducing cancer mortality

The RMS LifeRisks Monitor is dedicated to keeping clients abreast of bioscience developments

$

. . ?

The Past Risk Period $

. .

The Past Risk Period

Capital

age relief survival

Risk Period time

time time 1 2 3 vs.

The Past Risk Period The Past Risk Period Commutation Period

The Past Risk Period t0 t0+20

• • t0 t0+20

realized path t0 t0+20

realized path

US survival

time

US age

Commutation survival

Risk Period Period time

GB

time survival

time

GB survival

time Loss ($units)

3.0

2.0

1.0

99.95%

tail Annuity Annuity Book scenarios

0

Standalone Annuity Risk Risk Annuity

Standalone Loss ($units) -4.0 -3.0 -2.0 -1.0 Life 0Risk 1.0 Life Book . .

vs.

Risk +Commutation Periods

ρ

100%

90%

80%

70%

60%

50%

40%

30%

20%

10%

0%

2014 2025 2010 2011 2012 2013 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2026 2027 Realized Realized Projected

survival

time

Realized Projected

survival

time

Realized Projected

survival

time

Realized Projected

survival

time

Realized Projected

$

Pay-off Function

survival

time

Realized Projected

$ Variable Realized Pay-off Function Payment Realized Realized RealizedRealized Realized VariableRealized Realized Pay-off Function Payment

σ

Variable Payment