1
A Study of Incidence Experience for Taiwan Life Insurance
Jack C. Yue and Hong-Chih Huang 2010/09/10
National Chengchi UniversitySixth International Longevity Risk and
Capital Markets Solutions Conference
3
Prolonging Life Expectancy
The life expectancy of people in Taiwan has increased significantly since the end of World War II, and becomes even more noticeable after the National Health Insurance (NHI) was enforced in 1995.
Now the life expectancies in Taiwan have surpassed those in the U.S.
5
Study Objective
Two reasons contribute to the difficulty of ratemaking for Taiwan insurance companies.
Population size & Rapid improvement
We will use the experience data from Taiwan life insurance companies to explore if there are factors affecting the mortality rates.
6
Insurance Market in Taiwan (I)
Comparing to the developed countries, the history of life insurance in Taiwan is relatively short but the growth rate is fast.
On average, every Taiwan people owned 1/4 policy in 1989, 1/2 policy in 1994, and one policy in 1999.
Taiwan population is 23 millions in 2009. This indicates that the sizes of exposures were small before 1990 (less than 20 years of data).
7
201020001990198019701960
2.0
1.5
1.0
0.5
0.0
Year
Rat
io o
f ha
ving
Ins
uran
ce C
over
Time Series Plot of Ratio of having Insurance Coverage
8
Insurance Market in Taiwan (II)
On average, every Taiwan people owns at least two life insurance policies since 2008, but about 1/3 populations in Taiwan never purchase any life insurance products.
(http://www.iiroc.org.tw/report_test/1010/PDF1010_2009.pdf)
It is unknown whether the population of purchasing insurance products would have similar mortality profiles as those do not purchase.
9
Type No. of Policy %
Term Life 4,268,921 8.4%
Endowment 24,132,752 47.3%
Whole Life 22,610,097 44.3%
Life Insurance in Taiwan (1972~2008)
10
Data Description (I)
Incidence data are from Taiwan Life insurance companies, for the years 1972~2008 (27 years) collected via Taiwan Insurance Institute (TII).
Major factors include:
age, gender (2), Policy type (3),
health exam (2), principal repayment (2), benefit amount, & insurance company size (3)
11
Data Description (II)
The option of returning principal is one major attribute in many Taiwan life insurance products.
about 50% of WL and TL are with this optionWe separate 28 life insurance companies into
three groups (small, middle, & large) according to the number of policies.
small (9): # policy 100,000, < 2% share
middle (10): 10% share
large (9): # policy 400,000, 90% share
12
Empirical Mortality Rates
The mortality rates of term life, endowment, and whole life policies are very similar.
There are differences with respect to
principal repayment
insurance company size
health exam
benefit amount
14
GenderCompany
SizeTerm Life Endowment Whole Life
Male
Large Company
1,584,231 11,352,337 9,560,227
Middle Company
239,357 654,538 1,721,926
Female
Large Company
2,165,901 10,952,936 9,160,791
Middle Company
223,935 712,346 1,745,610
Table 4. Summary of Company Size
The middle companies have more on Whole life (about 65.6%) while the large companies have more on Endowment (50%).
16
Higher benefit amounts do not indicate smaller mortality rates (e.g., Term Life). Mortality rates for those with health exam are generally smaller.
17
Summary of Empirical Analysis
The mortality rates from the experience data show some interesting results. we found that Repayment of Principal, Company Size, Health Exam, and Benefit Amount are highly correlated to the mortality risk.
Taiwan insurance companies should pay more attention to pricing the life insurance products. Using a single experience life table (TSO) won’t be appropriate!
18
Mortality Improvement
We will use the Taiwan experience data on the Lee-Carter Model, to explore the mortality risk.
describes the average age-specific mortality, represents the general mortality level, and the
decline in mortality at age x is captured by .
, ,ln x t x x t x tm
x
t
x
19
The mortality improvement over time behave quite different for the population and experience data.
t
20
The differences of population and experience data in improvement rates are larger for younger age groups.
21
Evaluate the Insurance Products We will use the experience data, applying the
Lee-Carter model, to evaluate the pricing insurance products.
Compare the pure premiums of whole life, annuity, and endowment
Compare with the results from 2002 TSO, 2002 Annuity Table, and 9th Taiwan Period Life Table (9th TPL, 1999-2001 Taiwan Complete Life Table)
Interest rate 3% & ultimate age is 100 year-old
23
Issue Age 20 30 40 50
2002TSO
Male 54.6% 47.0% 39.4% 31.7%
Female 90.6% 86.7% 81.6% 75.1%
9th TPL
Male 54.1% 46.6% 38.5% 30.1%
Female 90.6% 86.8% 81.7% 75.2%
Table 5. Percentages of Over-estimate in Whole Life
24
Table 6. Percentages of Under-estimate in Annuity (Male)
Deferred period/Age 20 30 40 50
2002Annuity
0 8.2% 9.5% 10.9% 12.0%
10 26.7% 27.9% 29.3% 30.6%
20 33.3% 36.6% 41.6% 49.8%
9th TPL
0 10.9% 13.4% 16.1% 18.8%
10 35.1% 38.4% 42.1% 46.9%
20 44.4% 51.2% 61.3% 77.9%
27
Summary of Premium Analysis
The Mortality rates from the experience data are much lower than the life tables used.
Pure premiums of life insurance products are over-estimated (more obvious for the female).
Pure premiums of annuity products are under-estimated (more obvious for the female).
Insurance companies are possible to be in deficit (more than 30%) by issuing the endowment policies.
28
Conclusion and Discussions
The consumer’s behaviors in Taiwan might be different from those in other countries (need to be careful about introducing new products).
The principal repayment is an important factor, as well as benefit amount & health exam.
The experience and population data in Taiwan have different mortality improvement rates.
Using the population data to construct experience life table is feasible, but needs to be careful at the younger age groups.
29
Suggestions
The experience data in Taiwan are not fully explored and used.
Further study of mortality-related risk factors is recommended, and the premium calculation should include these factors.
Insurance companies should build their own experience tables, instead of relying on the same experience tables (e.g., 2002 TSO and 2002 Annuity Table).