quantifying longevity risk - cirano
Post on 11-Feb-2022
8 Views
Preview:
TRANSCRIPT
1
Longevity Risk
Edouard Debonneuil - AXA Global Life
2
Longevity risk - AXA
1 Past, current and future longevity
2 Modelling and handling the longevity risk
3 Longevity risk at AXA
3
1 Past, current
and future longevity
4
Life expectancy increases by 6 hours per day
Oeppen & Vaupel. Science 2002; Christensen et al., Lancet 2009
Life Expectancy
record across countries
Life expectancy of 80 ≈expected lifespan of 100
5
Why are we living longer?
Some potential explanations
A few key discoveries Louis Pasteur ideas: microbes & boiling water
Freezers, mass consumption, antibiotics
> Improvements for young/mid-life ages since 1800
A virtuous cycle Knowledge that biomedical research & prevention saves lives more
Many old persons focus on age-related diseases
> Improvements for higher ages since 1950
6
A global longevity pandemic!L
ife
Ex
pec
tan
cy
Wealth
Evolutions from 1800 to 2007
High longevity
in developing
countries too
Some exceptions
http://gapminder.org
7
Will longevity evolve similarly in the future?
Factors for acceleration
Biomedical improvements
- Smoking recession
- Focus on old age
- Internet and globalization
- Biology of aging
Increased awareness
- Government incentives
- Associations and conferences on health and longevity
- Genetic tests
Factors for a slow down Obesity, electromagnetic waves, pollution, natural catastrophes
Resistance to antibiotics and come back of infectious diseases
> Faster? Slower? Since it could be significantly
faster, there is a real longevity risk
8
C.elegans | Humans
Pe
rce
nt
aliv
e
Chri
ste
nsen, Jo
hn
so
n, an
d V
au
pe
l: N
at R
ev G
ene
t. 2
006
Focus: an anti-aging breakthrough?
Normal worms One gene changed
The insulin-like pathway In worms, flies, yeast
Numerous life extensions in laboratories
2008: worm lifespan times nearly 10!
In mammals 2008: nearly doubled mouse lifespan (genetics+ diet restriction)
2009: a human drug (rapamycin) increases lifespan of aged mice
In Humans Natural mutations found in long-lived human cohorts, that extend
animal lives (IGF-1, FOXO or AKT): Hawaiians of Japanese descent,
Ashkenazi Jews, German centenarians, etc.
Testing on age-related diseases ongoing
Whether longevity drugs appear or remain probable,
in both cases this creates a high longevity risk
9
Imagine…
The first man to
reach age 1000 is
perhaps already alive
It could be that we change
one single gene and
double our lifespan
Science is quickly developing the
technologies needed to radically
extend the quality human lifespan.
Cynthia KenyonAubrey de Grey
There is hope that, at some time
in the future, elderly people will
be kept healthy by suppressing
the ageing process itself. http://www.LifeStarInstitute.org
Healthier elderlies as a solution.(Science 1999, PMID: 10049123; BMJ 2008, PMID: 18614506)
10
2 Modelling and handling the
longevity risk
11
What is the longevity risk?
Living long is desirable – so why “longevity risk”?
Individual risk: at retirement, will you outlive your money?
“I’ve got all the money I need for my old age…
…provided I die before 4pm today”
Groucho Marx
- Also depends on market performance & costs of health
Solution: life insurers and private and public pension funds
take your longevity risk
Cheap: they act as intermediates:
- You pay before retirement, they pay you regular income after
- Based on mutualisation of risks
- And on a priori prudent estimates of average longevity
What if people live longer than expected?
> A global risk for the whole society.
12
Demographic picture is worrying…
…For unfunded pensions Unfunded: workers pay for retired pensioners
Increased longevity + low birth rates increased longevity risk
For life insurance annuities this is less an issue Funded: money before retirement serves after retirement
Population aging might increase demand for annuities
1950 2000 2050
Percentage of 65+ :
13
High amounts at stake
Life insurance annuities
US annuity sales in 2007
- Life & term : 73 bn$
- Indexed : 25 bn$
- Variable : 183 bn$
(Morningstar, Inc. and LIMRA International)
Crucial impact of models
One more year ≈ 3-4% reserves
USA, 2010, Medicare and Social Security: by 2050,
+ 3.1 to 7.9 years of life?
+ $3.2 trillion to $8.3 trillion?
(MacArthur Foundation Research Network on an Aging Society)
14
Longevity modeling approaches
Some sound models prove wrong
Lower bounds for mortality rates- US 1928, L. Dublin: maximal life expectancy = 64.75 !
Extrapolating disease trends- They turn…
Some models proved OK: extrapolations of past mortality of general population to the future
General population- Fit past mortality (http://mortality.org)- Extrapolate (http://www.lifemetrics.com)
- Criteria to choose/reject models
Transposition to insurance- Insured population (mortality level and trend)- Extrapolation to high ages (e.g. Kannistö)
15
Which patterns should a model capture?
Distribution of age at death
(Swiss females; Jean-Marie Robine, Le futur de la longévité en Suisse) Cohort effects
1850 2000
Insured characteristics Gender, profession, smoking status,
location. 1 cigarette = -11 min of life
May not be modeled
- Genetics: 25% of variations of lifespan
Age
Time (“period”)
Improvement factors
Age and time effects
Log mortality rates
E&W males
born 1920-
1940
General population
16
Capturing patterns of past data
Various models
Interpretation: The global age-shape of mortality (β1
x)
Is positioned at lower and lower levels (κ 2t)
- With higher improvements for high ages (β2x)
Mortality is perturbated by cohort effects (γt-x)
Lee Carter (LC)
Cairns, Blake,
Down (CBD)
Lee Carter
+ Cohort (LCC)
http://arxiv.org/abs/1003.1802Main requirement: fit essential patterns
rather than patterns that won’t continue
in the future tradeoff between
complexity and robustness
Analyze fitting error
17
Extrapolating the future
Fitted past Kappat
Projections of Kappat
Catastrophic scenario
Stochastic or deterministic
Quality of past predictions
qx
t
Coherence of new
best estimate
predictions
Lee Carter (LC)
Cairns, Blake,
Down (CBD)
Lee Carter
+ Cohort (LCC)
XX
18
Handling the longevity risk
Business mix and “natural hedge” In case of high longevity, unused death benefit reserves may help
Caution and prudence
Solvency II framework Best estimate + Capital to put aside in case of high longevity
Keep an attentive eye Country by country, several times a year
Transfer excess longevity risk Longevity Swaps
Standardized financial products
19
Principle
AXA has success stories in the UK Ongoing ones in other countries
Long process.
Reinsurance may become limited in absorbing longevity risks
Customized longevity Swap
20
AXA, Deutsche Bank, J.P. Morgan, Legal & General, Morgan Stanley,
Pension Corporation, Prudential PLC, RBS, Swiss Re and UBS
Goal: create a liquid longevity market Financial markets = larger capital resources
Faster process
Standardized instruments Q-forwards (mortality) and S-forwards (survival)
Realized mortality/survival based on LLMA index
- UK as a first step
Life & Longevity Market Association (LLMA)
Fixed mortality/survival
Realized mortality/survival
Life Insurer Hedge Provider
21
3 Longevity risk at AXA
22
Longevity risk has to be handled
A risk to quantify AXA uses various models
No perfect quantification exists
Longevity risk is examined in details
AXA develops an internal Solvency II model
Attentiveness, business mix and risk transfer is key
Transferring solutions are possible AXA has done longevity risk transfers through longevity swaps
AXA participates in the LLMA to facilitate risk transfer
Accompanying progress is possible AXA Research Fund created a chair on Longevity
AXA Education and Research paper on Longevity
AXA Longevity Expert Network
23
AXA - education and research
AXA Research Fund
top related