Risk

management challenges are the one important

issues facing insurance companies.Risk is a double-edged knife for

insurance companies.There are several kinds of risk in insurance companies like

model risk.This essay discusses model risk in life insurances,what are sources

of the model risk in life insurances,why does model risk arise and how can

control or avoid model risk.

2. Definitions

Definiton of model risk is risk of

loss arising from valuing financial instruments with a model that is inaccurate(Model

risk, 2011).In other words,the model that used to calculate a firm’s market

risk does not implement the tasks.Model risk is the main risk in insurance

sector. Any

deviations from expected claims and liabilities can be defined as model risk. Model risk may also

become susceptible to misuse or errors that can have significant adverse

consequences, including financial.Models are used for core financial functions

such as financial reporting, where any oversight or errors can result in

financial restatements, which can lead to the loss of investor, regulator and

policyholder confidence Inaccurate model outputs can also result in volatile,

inefficient or inadequate capital or reserve requirements required by local

insurance regulators or accounting boards(Lebel & Gagnon, 2015).

Life insurance can be defined as it

is a pecuniary benefit to the survivors of insured person upon his/her death.That

is, it

is a contract between an insurance policy holder and an insurance company,

where the insurer promises to pay a sum of money in exchange for a premium,

upon the death of an insured person or after a set period(“Life insurance,”

n.d.).In general,payment is made at time of policyholder death.The aim of life

insurance is provide a protection of financial comfort for his/her family after

policyholder dies. Life

insurance companies estimate insurance premiums through life tables.

Life tables ,which are one of the

oldest tools of demographic analysis,are tables detailing the mortality

probabilities and other statistics such as life expectancy at each age and

survival times of the population at all ages. Life tables are also known as

mortality tables.Life tables are constructed by following a cohort from birth to death.It can also be

constructed from vital registration. Widely,they are constructed for age,

sexuality, ethnic groups and occupational groups.

2.1 Sources

of Model Risk

Let’s

examine relationship between model risk and life insurance.The most important

model risk in life insurance is life tables because life insurance companies need

life tables to estimate insurance premiums.There are a lot of factors that

could be model risk for life tables.

First-one is data problems.Data that

is used for life tables may not reflect the population. Parameters are estimated

from an observed sample.Parameter uncertainty results from differences between

that sample and the population(Venter & Sahasrabuddhe, 2012). For example,Turkey

has used a life table that was developed for U.S.A because there has not been a

life table about Turkey population.As a result,Turkey has taken model risk.The

table does not reflect characteristics of Turkey population because the life

table not only has gender(female,male) but also classification between white

and black.

Secondly,changing factors over time

may lead to occurrence of model risk. Many models need the future value of some

volatility or correlation.This value is often based on historical data but

history may not providea good estimate of future value, and historical values

may themselves be unstable and vary strongly with the sampling period(Derman,

1996).For instance,Innovations in healthcare can change factors such as birth and death

rate.Changes in the effect size of factors will have a remarkable impact on the

applicability of a model.Also,sudden changing factors such as an economic crisis,earthquake and so on can

affect the applicability of a model.

Thirdly,model misspecification also leads to occurrence of model risk. Model misspecification is

the risk that the wrong model is being estimated and applied(Venter &

Sahasrabuddhe, 2012).A model might be misspecified if important variables have

been omitted and chosen a wrong functional form.For example, this is the risk that we

use an exponential model when the phenomenon follows a Pareto distribution(Venter

& Sahasrabuddhe, 2012).

Last but not least, a model could be built

correctly but it might be used for the wrong task.H?rsa(2012) mentioned that if

we assume that we have chosen a correct

model and computed a correct solution under that model,there is still the risk that the model results will be used

inappropriately.This has often been a problem in the modern history of

matematical finance where those who utilize models and their results fail to

understand their assumptions and limitations.Therefore,even though a model is a

correct model or solved correctly,it has the potential to cause problems.

2.2 Model

Risk Management

After

explaining sources of model risk that could

affect life tables, need to discuss how to avoid model risk because all kinds

of model risk in life tables lead to estimate incorrect insurance premiums.Model

risk can not be controlled or eliminated at all but at least be aware of meaning of the risk

and souces of this risk.As a result,life insurance companies should adjust

confidence level and tolerance to

estimate insurance premiums in terms of the risk.Miller(2014) mention that even

with skilled modeling and robust validation, model risk can not be eliminated,

so other tools should be used to manage model risk effectively. Among these are

establishing limits on model use, monitoring model performance, adjusting or

revising models over time, and supplementing model results with other analysis

and information.There are two ways that mitigate model risk.

Firstly,back testing is comparing

actual results for a defined period to the results forecasted by a model for

the same period in order to evaluate accuracy of the model’s

predictiveness.Back testing is an exercise that compares the actual outcome

with model forecasts during a defined period, a period of time that was not

used to develop the methodology(Lubansky, 2015).The evaluation of value at risk

is an example of back testing.In this example, actual profit and

loss is compared with a model forecast loss distribution.In general,the

comparison is performed using

statistical confidence intervals around the model forecasts.However, using back

testing could be harder for life tables.It takes a long time when using back

testing for life tables because there are 80 years old life tables.It means

that there is a massive historical data and it is not easy to examine all of

them.Also,life tables become old after a

certain year because of increasing the average life span.As a result,the data

could be unstable.It means that there could be a problem about reliability of back testing.

Secondly,reassurance can be defined as it occurs when multiple insurance companies share or transfer risk with another companies by purchasing

insurance polices from other

insurers to limit the total loss the

original insurer would experience in case of disaster .The premium paid by the

insured is typically shared by all of the insurance companies involved(Reassurance,

2008).That is to say insurance companies insure their own risk.Consequently,reassurance

encourages insurance companies to take a risk since reassurance gives the

insurer more security.