Chapter 19 - Pricing (1) - Individual business

Question 1

What is the (pure)risk premium of a short-term insurance policy

Answer 1

It is the element of the premium required to cover only the expected claim amount

Question 2

What are the factors that need to be considered when selecting a base period for a pricing exercise

Answer 2

TURD V

• Trends - The period should be long enough to indicate trends in claim frequencies and claim amounts
• Unknowns - To be the most relevant to future experience, ideally we want the most recent claims experience. However, this is also the experience with greatest uncertainty because we will need to rely heavily upon estimates of outstanding claims.
• Relevance - The further back we go into the past, the greater the danger that the experience is less relevant for the future
• Detail - For a type of cover where there is a large amount of variability in claims, a lot of data ( and hence a longer period ) will be required so that we have sufficient data
• Volume - the period should be long enough to include sufficient volume of data to be credible

Question 3

Why is data split into homogenous groups

Answer 3

• It enables greater understanding of the risk profile of each policyholder risk class and procedure/benefit
• Help ensure that dangers of cross-subsidies are avoided ie profitability will not be dependent on a particular cross-section of policyholders and so the insurer will be less exposed to changes in the business mix
• It is necessary to ensure a sufficient amount of data in each cell for credible analysis

Question 4

What is the burning cost premium (BCP) and its formula

Answer 4

The BCP is the true past risk premium of an actual portfolio of data.

BCP = Average claim amount x Claims incidence rate
or
BCP = sum (Claims) / Total exposed to risk

Question 5

In which ways might base values not be appropriate as the basis for setting new premium rates?

Answer 5

SLUIT CARR

• SEASONAL variation in claims
• LARGE or exceptional claims
• UNUSUALLY heavy/light experience
• INCOMPLETE claims
• TRENDS in claims experience
  --
• Changes in COVER
• Changes in AGREEMENTS with suppliers
• Changes in RISK
• Changes in the cost of REINSURANCE

Question 6

What do we need to allow for when projecting the past claim costs into the future

Answer 6

• Changes in policyholder profile by benefit option, considering the selective effect of membership movement
• Claims inflation
• Trends, eg changes in claims frequency or utilisation of healthcare services
• Other changes over time

Question 7

What affects the claim inflation for PMI products

Answer 7

• Professional fees and medical treatment
• The cost of pharmaceuticals
• The cost of claims

Question 8

The modelling complexities to be aware of for PMI and data limitiations

Answer 8

IN FARM

It will be necessary to estimate:

• The IMPACT of sales tax on future growth
• The existence of NCD multi-states
• Distribution of claim FREQUENCY - difficulty in defining a discrete claim
  – Claim frequency parameters are dependent on duration from policy inception, age, gender, etc
• Distribution of claim AMOUNT - indemnity
  – Distribution of claim amount is dependent on hospital capacity, medical science progress, insurance/provider details, inflation, competition, style of government
• The need to incorporate the RECLAIM of initial costs over several renewals in the pricing model
• The need to MODEL the chain of occurrence through GP referral via specialist to hospital, plus the delay to settlement
  --
  Data limitations
• Absence of insurance statistics
• Problems with family covers where the details of individuals at risk may not be known
• Problems with group arrangments, where details of individuals ( especially spouses and children) can only be estimated

Question 9

The modelling complexities to be aware of for CI insurance

Answer 9

• Possible use of separate approaches for disease-based and treatment-based
• Guaranteed and reviewable alternatives

Question 10

Influences on claims distribution to be aware of for CI products

Answer 10

• Advancements in medical science
• Earlier diagnosis
• Simpler and more readily available operations

Question 11

The modelling complexities to be aware of for LTCI

Answer 11

It will be necessary to estimate:

• Distribution of claim frequency - little experience
• Distribution of claim amount - could place a cap on the weekly/monthly benefit, but otherwise little control
• Demand - this is very dependent on political commitment

Question 12

General factors that complicate modelling

Answer 12

• The role of genetics
• Trends in anti-selection
• The quality of underwriting

Question 13

What is the process involved in deriving a risk premium

Answer 13

• Choose a base period over which to collect claims and exposure data
• Collect data, checking the accuracy and appropriateness of the data
• Split the data into homogeneous groups
• Calculate a historical burning cost premium for each group
• Analyse the data, e.g. to identify trends etc
• Adjust and project forward to obtain future risk premiums, i.e. the expected claim amount over the period that policies covered by the premium rates will be in force

Question 14

Internal data sources that are likely to be helpful in pricing PMI products

Answer 14

CRAP RUMP F

• Claims management processes, reinsurance arrangements, managed care
  arrangements and other preferred provider contracts e.g. fees to network doctors
• Recent claims experience
• Administration and other expenses (contractual arrangement with administrators and administration fee basis, commission structures and contracts)
• Profile of policyholders and their dependants
• Recent changes in cover, policyholder profile
• Underwriting process
• Marketing strategy and business projections
• Policy benefit terms and conditions
• Financial position of the insurer and the profitability of the portfolio of business (information in the financial statements and returns to regulators)

Question 15

Reasonableness checks the actuary could perform on the data used to price PMI products

Answer 15

• Reconcile the data to independent data sources such as financial statements and returns to the regulator
• Check consistency between claims data and the benefits covered
• Perform spot checks for impossible/ unreasonable values and distributions such as impossible ages or an improbable age profile

Question 16

What needs to be allowed for when projecting past claims costs

Answer 16

• Changes in policyholder profile by benefit option, considering the selective effect of membership movements (e.g. older, less healthy policyholders selecting more comprehensive benefit options)
• Claims inflation
• Trends, e.g. changes in claim frequency or utilisation of healthcare services
• Other changes in cover

Question 17

What are the time periods involved in the projection of data

Answer 17

• The period of exposure containing the base experience
• The period during which policies will be written under the new premium rates
• The full period of exposure covering all the claims that can arise from the policies written under the new rates

Question 18

Factors that should be considered when determining the appropriate level for the contingency margin

Answer 18

BP COUNCIL

• Benefit design
• Policyholder risk class distribution and changes in policyholder profile
• Credibility of claims experience
• Overall risk exposure, e.g. exposure to effects of HIV/ AIDS
• Use of reinsurance
• Number of policyholders and number of policyholders per benefit option
• Changes in managed care arrangements and other contractual arrangements e.g. policy administration
• Impact of current and future changes in legislation
• Likely variation in expenses

Question 19

Formula for accelerated CI insurance risk premium rate

Answer 19

I(x) + (1-k(x))q(x)

where Ix = the critical illness incidence rate
Kx = the proportion of deaths due to critical illness
qx = the mortality rate

Question 20

Formula for stand-alone CI insurance rate

Answer 20

i(x) x (probability of surviving the survival period)

Question 21

Reasons why accurate past claims data might not be suitable for premium rating

Answer 21

PP VMP

• PAST data is not from the class of lives insured or benefit options we want to set premiums for
• The PREMIUM structure has changed (e.g. risk premiums are to be calculated for surgery, maternity and other hospital treatments) and the past data is not credible
• The VOLUME of policy data is inadequate
• MEDICAL innovation changes the standard treatment protocol for a certain medical condition
• POLICY conditions have changed

Question 22

How the reclaim of initial costs over several renewals would be incorporated in the pricing model

Answer 22

• We need to make assumptions (from past experience) for the rate of renewal in each year following the first issue of a contract
• The excess initial expenses and commission (over the renewal levels) can then be spread over the number of years for which the policy is expected to be in force, based on these assumptions
• The same premium is then charged for new business and renewals, each covering part of the excess initial expenses
• The renewal assumptions would be conservative (low) to reduce the risk that the company suffers from unexpectedly high lapse rates and is then unable to recoup the initial expense fully

Question 23

What reasonableness checks can be performed on the data used to price products

Answer 23

• reconcile the data to independent data sources such as financial statements and returns to the regulator
• Check consistency between claims data and the benefits covered
  ]- Perform spot checks for impossible / unreasonable value