Q&A Flashcards
1
Q
What factors impact the simplicity and clarity of the product from a customer point of view?
A
- underwriting and acceptance
- limits
- exclusions
- cost sharing
- pre-authorisation
- provider networks
- treatment protocols
2
Q
Consistency
A
- Assumptions in an actuarial model should relate to each other in a realistic way
- any observed correlated or interdependencies between the different variables should be incorporated into the model
3
Q
Extent of margin in pricing depends on:
A
- the extent of uncertainty associated with each parameter
- the extent of statistical risk inherent in the future experience
- the financial significance of the adverse experience
- the company’s attitude to risk
- the proprietor’s required return on capital
- the size of the company’s free assets
- the competitive nature of the market and the company’s unique selling proposition
4
Q
Pricing investigations
A
- single policy profit test
- model office pg. 1277
5
Q
Assumptions of BF
A
- for each origin year, the expected amount of claims in monetary terms, paid in each development year, is a constant proportion of the total claims, in monetary terms, from that origin year
- no explicit assumption is made for claims inflation
- the estimated loss ratio is appropriate
- the development factor to ultimate is never less than 1
- the first cohort is fully run-off, or its development to an ultimate position can be predicted with some confidence
6
Q
Assumptions of basic chain ladder
A
- assumed that undiscounted claims reserves are required
- a constant proportion of the total claim amount arising from each origin year is paid in each development year
- the pattern of inflation within the existing data can be projected into the future
7
Q
Assumptions for inflation-adjusted chain ladder
A
- undiscounted claim reserves are required
- a constant proportion of total claim amount, in real terms, arising from each origin year is paid in each development year
- claims inflation, both past and future, is x%
8
Q
Assumptions for inflation-adjusted average cost per claim
A
- undiscounted claim reserves are required
- the average amount of claim payment (in real terms) is a constant for each development year
- a constant proportion of the total number of claims from each origin year are settled in each development year
9
Q
Bootstrapping for stochastic claim reserving
A
- bootstrapping is a technique for determining the statistical properties of a quantity using the randomness that is present in the sample from the underlying population, and then applying a monte carlo approach
- it involves sampling repeatedly from an observed data set in order to create a number of pseudo-data sets that are then consistent with the original data set
- various statistics of interest can then be derived for each pseudo data set and the distribution of these statistics can be analysed further
- it is assumed that the sampled data are independent and identically distributed
- can be used to estimate the distribution of the reserve predictions
10
Q
Bootstrapping steps
A
- obtain a set of past claims data, split by origin/development year
- back-fitting a model to the past data to find the expected claims for each cell
- calculating the residual noise present in each cell
- sampling from this residual distribution to produce many pseudo-data sets
- calculating reserve projections based on each data set
- collating the reserve projections to determine the distribution, moments and percentiles
11
Q
Drawbacks of NPV
A
Comparisons based on NPV will only be valid if:
- there is a perfectly free and efficient capital market
- the risk discount rates used to discount two or more risky investments appropriately reflect their riskiness, which may be difficult to assess
- NPV is subject to law of diminishing returns
- if it were not, then company could sell infinite amount
- it says nothing about competition
- compare total NPVs for whole projects based on realistic assumptions of new business
- if the expected overhead costs are included, could give distorted view if mix isn’t as intended
- NPV doesn’t distinguish between contracts that have very different capital requirements
12
Q
Merits of passive valuation approach
A
-pg.1388
+ relatively straightforward to implement
+ tend to involve less subjectivity
+ tend to result in relatively stable accounting profit emergence
+ less exposed to systematic risk
- at risk of becoming out of date
- may provide a false sense of security when market conditions are changing significantly
- less informative in terms of understanding impact of market conditions on ability of the company to meet its liabilities
13
Q
Steps in the bootstrapping process for determining reserves using chain ladder
A
- project the triangle to ultimate using the chain ladder method
- back-fit the triangle to calculate the hypothetical past claims development, if the data has conformed precisely to the model
- calculate the triangle of raw residuals by subtracting each entry in the fitted incremental triangle from the corresponding entry in the actual incremental triangle.
- residuals based on incremental data are used in order to maintain the assumption that residuals are independent
- in practice, adjustment is applied so that they are approximately iid - obtain a pseudo-data set by calculating a new residual triangle where each cell in the triangle is picked at random from any of the points in the triangle of the adjusted residuals, and then adding this to your incremental fitted triangle
- sampling with replacement is used so that each residual can be selected more than once
- since the expected value of the residuals is zero, they can be added or subtracted
- this pseudo-dataset represents a hypothetical claims experience that could have emerged from the portfolio - use ultimate chain ladder method to estimate ultimate claims experience and get the outstanding reserves
- repeat steps 4 and 5 many times to obtain different estimates of the claim reserves for the portfolio
- the resulting distribution of reserves can be used to communicate all the uncertainty surrounding the reserves
14
Q
Problems that confront the actuary
A
CCRRISPPPPP
- policy data
- product design
- product marketability
- pricing
- return on capital
- profitability
- supervisory reserves
- investment
- capital management
- risk management
- claims
15
Q
Advantages and disadvantages of facultative
A
+ gives the insurer greater flexibility in that they can choose the reinsurance that is most suitable
+ can be used to cope with risks that otherwise fall outside the terms of a treaty
- arranged for each risk so takes time and costly
- may not be able to find appropriate reinsurance
- there can be a delay in finding cover before accepting risk
16
Q
Advantages and disadvantages of treaty
A
+ gives certainty of being able to have reinsurance cover so can do financial and risk planning with greater certainty
+ it can be simple to operate and less expensive
+ can help control solvency and growth requirements
- takes time/expense to set up in the first place
- may restrict insurer’s actions
- insurer may not get best reinsurance deals
17
Q
Insurance company risks
A
BBICCEPD+RR
- business volume + mix
- broker activity
- investment
- claims (+ anti-selection and non-disclosure)
- competition
- expenses
- persistency
- policy + other data
- regulation
- reinsurance
18
Q
How to control these risks
A
- proper management and control procedures
- control claims
- pre-authorisation
- preferred provider networks & managed care
- control anti-selection
- control expenses
- review terms
- improving renewal rates
19
Q
Compare underwriting
A
- cost
- anti-selection
- level of premiums
- speed of application
- attractiveness of cover
20
Q
Purpose of underwriting
A
SAFARI
- suitable special terms
- avoid anti-selection
- financial underwriting
- actual experience in line with expected
- risk classification to ensure that all risks are rated fairly
- identify substandard risks while aiming to accept as many on standard terms as possible
21
Q
Applications of risk adjustment
A
- budgeting
- pricing and reserving
- measuring efficiencies
- risk management
- measuring healthcare outcomes
- provider profiling
22
Q
Disadvantages of formula approach to pricing?
A
Doesn’t allow for:
- proper timing of events
- accumulation of reserves
- capital needs
- the impact of net negative cashflows
- separate inspection of claims-related and premium-related cashflows
- variation in assumptions
- changes in future experience and cannot be used to measure sensitivity of profit for such variation
- cannot allow easily for more complicated product structures, such as unit linked products and options and guarantees
23
Q
Criteria for condition to be covered under CI
A
- sufficient data to price
- clearly defined objective/unambiguous from medical point
- perceived to be serious and occur frequently
- limit anti-selection
24
Q
Advantages of OSF
A
- economies of scale in terms of non-claims costs
- opportunity to cross-sell to existing customer basis
- established company so additional reach in terms of market penetration
- will only sell the company’s products
- insurer will have better control over the sales process i.e. the risks they sell to, information conveyed
- reduces incentive for mis-selling policies
25
Pricing challenges
- product design
- availability of data
- pricing model and assumptions
- underwriting
- claims incidence and amount
- other challenges
26
Ways to test model goodness of fit
1. deviance residuals
2. standard pearson residuals
3. residual plots
27
Pricing process
- choose base period for pricing exercise such that data is...
- collect the data
- split the data into homogeneous groups
- adjust the base data
- calculate the burning cost premium for each group
- analyse data and adjust BCPs for any changes in insurer's practice or relevance of past data
- project the adjusted basis forward
- the insurer will have to allow for other loadings
28
Operational risk
1. failure of systems, processes or people
2. dominance of a single individual over the running of the business
3. reliance on third parties to carry out various functions for which the organisation is responsible
4. theft or fraud
5. non-disclosure
6. data errors
7. mistake in pricing
8. wrong amounts paid out to claimants
9. non-compliance with regulation
10. change in regulation and taxation
11. theft of equipment
12. litigation
13. reputational risks
14. control failures relating to underwriting and claims management
15. control failures relating to accounting and reporting
16. data protection breach
17. loss of key persons
18. mis-selling product
19. business interruption risk due to physical risk
29
Credit/counterparty risk
1. risk of counterparties not meeting obligations
2. downgrades of corporate bonds
3. default on corporate bonds of coupon payments or capital repayments
4. default of reinsurer
5. default of third party providing outsourced services
6. downgrade of reinsurer
7. default of intermediaries
8. default of derivatives providers
9. default of care homes in providing service
10. care homes providing unsatisfactory service
11. default on cash holdings
30
Assumptions of a classic linear model
1. error terms are independent and come from a normal distribution
2. mean is a linear combination of explanatory variables
3. error terms have a constant variance
31
Advantages of GLM
- more general than normal distribution as GLM can take on any distribution from the exponential family
- using a link function, can take account of the multiplicative nature of explanatory variables and their effects (classic assumes additive)
- the variance of the response variables is a function of its mean and can often increase with the value of the mean and as is usually the case when modelling claim amounts
32
Why monitor experience?
1. update assumptions to reflect future experience
2. identify trends
3. monitor actual vs expected experience and take corrective action
4. to provide information to management to aid business decisions
5. make more informed decisions about pricing and the adequacy of reserves
33
Claims experience investigation
- identify whether increase is due to...
- depends on there being valid data
- experience must be credible
- take account of IBNR
- take account of large or exceptional claims
- take account of seasonality of claims
- allowance needs to be made for benefit changes
- group claims into homogeneous risk cells
- risk adjustment given that case mix may have changed
- allowance must be made for demographic changes
- allowance made for changes in medical treatment/technology over the benefit years
- changes in regulation
- claims a function of frequency and severity. The analysis needs to examine each component in turn
- experience should be compared to industry or market experience to identify whether trend specific to insurer
34
General experience investigation
- change in membership - demographics and volume of members
- marketing and communication - has there been a change in the way in which benefits are communicated to members
- risk management: has there been a change in the way benefits are managed
- claims experience
35
Managed care
- term given to a process whereby an insurer intervenes in the provision of medical care
- with the dual objective of optimising the quality of the treatment for the policyholder
- and controlling the cost to maintain the affordability of healthcare
- by such means as preferred provider utilisation and claims pre-authorisation
36
List different purposes for which the government can use risk adjustment
- budgeting
- risk management
- measuring efficiency of providers
- measuring healthcare outcomes
- creating provider profiles
- reporting on costs, outcomes and future plans
37
Properties of exponential family which make them particularly useful when using a GLM
1. the distribution is completely specified in terms of its mean and variance
2. the variance of the response variable is a function of its mean
38
Two tests that can be used to analyse the significance of the additional factors
The two models are nested models
- where the scale parameter is known, the chi-squared test for the change in scaled deviance can be used
- where the scale parameter is unknown, the F statistic can be used as the ratio of the change in the deviance and the scale parameter estimate is distributed
39
Other checks to carry out before including additional factors
- consistency of the factor over time
| - consistency of the factor with other factors
40
3 ways of coping with uncertainty in financial forecasts
1. sensitivity analysis - vary key assumptions
2. scenario analysis - vary sets of assumptions
3. stochastic modelling - simulations using distributions applied to key assumptions
- appropriate method depends on purpose, audience and level of detail required
41
Why it's necessary to risk adjust efficiency factors
- hospitals treat a different mix of cases due to:
- different risk profiles of lives or
- facilities in the hospital
- e.g. some hospitals may have high proportion of cardiac cases will others have high proportion of day cases OR some hospitals may have expensive diagnostic equipment that is not available at other hospitals
- the differences in mix will impact the utlisation of resources
- as well as the cost of care
- therefore necessary to risk adjust on a like-for-like basis
42
Medical gap cover
- cover difference between cost of treatment and amount covered by conventional PMI
- can happen because of benefit limits or because of providers charging more than PMI covers
- usually focuses on providing cover for high cost events
- limited to annual amount per health event
- meets need of peace of mind
43
The factors that will influence marketability
1. attractive to the target market
2. meets needs of the target market
3. levels of service
4. distribution channel
5. competitor products
6. rating factors used by competitors
7. terms and conditions - ease of understanding
8. strictness of underwriting
44
How GLMs differ from ordinary least square regression
- similarly to ordinary least square regression, GLM can be used to model behaviour of a random variable that is believed to depend on the values of several other characteristics
- GLM is a flexible generalisation of ordinary least square regression
- generalises by allowing the linear model to be related to the response variable via a link function
- and by allowing the magnitude of the variance of each measurement to be a function of its predicted value
45
Why is multivariate modelling useful for modelling health insurance claims
- multivariate modelling is necessary when modelling multiple factors that are likely to be related or correlated to a certain extent
- examples of factors used in modelling health insurance claims that are likely to be related are:
- age and chronic condition
- age and family size
- maternity and gender
46
Advantages of Tweedie distribution for modelling PMI
- direct modelling of risk premium or incurred loss data for PMI business is problematic since a typical pure premium distribution will consist of a spike (point mass) at zero (where policies have not had claims) and then a wide range of amounts (where policies have had claims)
- the tweedie distribution is a special member of the exponential family
- it has a point mass at zero and corresponds to the compound distribution of a poisson claim number process and a gamma claim size distribution
47
PMB for CI
- cancer
- heart attack
- stroke
- coronary artery bypass graft
48
What does case mix represent
- reflects the severity of each case on a risk adjusted basis
- takes into account age, diagnosis severity
- reflects the expected relative cost of each DRG relative to the overall average cost per day in hospital taking account of risk factors
49
Factors affecting insurer's appetite for XoL reinsurance
1. volatility of claims with heterogeneous risk
2. impact of extreme claims on financials
3. availability of reinsurance
4. free assets
5. cost of reinsurance
6. predictability of experience
7. expertise in reinsurer's claims management
8. not much scope for solvency/tax arbitrage
50
Various approaches to experience rating
- can be applied prospectively or retrospectively
- prospective is based on prior period experience and applied to future rate
- retrospectively is adjustment to initial premium based on actual experience
- can be based on number of claims or amounts
- credibility approach has z factor as a rating. z is from 0 to 1 and is weighting applied to own experience vs book rates
- value of z depends on volume of claims and size of group
51
Risk adjustment
- normalizes population according to specified risk factors
| - applied to historical data
52
Risk prediction
- used to predict future costs
| - with reference to differences observed in the past
53
Case mix adjustment
- special case of risk adjustment
| - used to compare treatment costs
54
Determine whether new premium rating factor should be used
- collect claims and exposure data
- ensure sufficient volume
- apply appropriate adjustments, IBNR, inflation, high-cost claims
- build a GLM model to predict claim severity (gamma) and claim incidence (poisson)
- GLM will need to predict severity and incidence for those with new rating factor and those without
- GLM may need to adjust for other factors such as age, gender, chronicity, benefit level, to ensure no bias
- sensitivity test results and back tests on different groups of data to ensure consistent results
- if experience differs significantly, determine adjustment to premiums for those with and without the rating factor and compare this to premiums offered by competitors
- establish what data needs to be collected during underwriting process to validate new rating factor
- is this easily collected
- consider current regulation and industry practice
- use data from other sources with appropriate adjustments
55
Drawbacks of formula approach
does not allow for:
1. proper timing of events
2. accumulation of reserves
3. capital needs
4. the impact of net negative cashflows in any period
5. separate inspection of premium-related or claim-related cashflows
6. the variation of assumptions over time
56
How can long-term cashflow projection be used to aid in decision-making
- capital modelling
- renewals
- investment strategy
- long-term premium adequacy
57
Components of a full model office
- model of existing business
- new business model
- liability model
- asset model
- expense model
58
Why price regulation of healthcare industry?
• Access to healthcare is seen as a basic human right (public good
characteristics)
• The existence of health insurance may distort market price determination
mechanisms (third party payer effect)
• Consumers experience challenges in making informed decisions about
healthcare services needed. Information about the range and quality of
healthcare services relative to cost is difficult, if not impossible, for consumers to
obtain. This is compounded by:
- irregular consumption of healthcare services and lack experience in
evaluating the service
- experiences of different individuals, which are often not comparable and
therefore one cannot use the recommendations of others in making decisions
- the fact that the results are often not related to the quality of the service, but
rather to biological processes. The quality of the service therefore cannot
always be assessed by the improvement of health.
59
How to manage claims costs
1. benefit limits and exclusions
2. co-payments and levies
3. approved provider networks
4. medical savings account
5. preventative medicine and wellness programmes
6. introducing managed care systems
60
Why an employer might offer group products
1. may be a legal requirement
2. may be used as a positive message to employees at the time of recruitment
3. can promote health and ensure a speedy return to work
4. may have tax advantages over individual products
61
What does a well-functioning healthcare syste1 require?
- robust financing mechanisms
- a well-trained and adequately-paid workforce
- reliable information on which to base decisions and policies
- well-maintained health facilities and logistics to deliver quality medicines and technologies
62
Supply-side key providers
1. doctors, support medical personnel, clinical associates and nurses
2. hospitals
3. upstream service providers
63
Funders of healthcare
1. government
2. non-government organisations and donors
3. out-of-pocket expenditure by users themselves
4 trade-related employer groups
5. commercial insurance products
6. employers
64
The cost of healthcare is increasing due to
- a misalignment of incentives between the provider and the payer
- medical technology advancements combined with user expectations resulting in high cost of treatment
- cost increases resulting from a fairly static population
- fraud
65
Managed care objectives
1. reduce the cost of medical events
2. improve the quality of care provided
3. ensuring medical services are delivered in an appropriate setting
4. ensure that high-risk members are managed and receive appropriate care
5. reducing the number of unnecessary medical services
66
Risks in managed care that can be transferred between funder and provider
1. price risk
2. intensity risk
3. severity risk
4. frequency risk
5. actuarial and marketing risk
67
Principles of designing provider incentives for risk-sharing
PRESENP
1. natural response vs desired objectives
2. provider control
3. equitable risk sharing and a sense of partnership with the funder
4. simplicity
5. provider involvement
6. realistic goals
7. education and support
68
Patient concerns with managed care
- provider networks may restrict access
- providers may resent external parties imposing clinical protocols on them and influencing the way in which they practice medicine
- may compromise quality of care
- the use of formularies and other financial-based initiatives might result in additional cost being transferred from the scheme to the member
69
Measures that can be compared to benchmark to assess the quality of care
1. patient mortality rates
2. specialist referral rate
3. hospital admission rate
4. procedure complication rate
5. chronic medication adherence
6. patient questionnaire
70
Primary interests of customers
1. meeting specific needs
2. providing cash
3. providing peace of mind
4. simplicity and clarity
5. guarantees and reviewability of premiums/guarantees
71
Claims and persistency experience will depend on
- the level of financial sophistication
- income level
- level of underwriting
- who initiated the sale
72
Regulatory restrictions on insurance companies
1. a restriction on the types of contracts that an insurance company can offer
2. restrictions on premium rates or charges
3. restriction on rating factors used to calculate premiums
4. restrictions on terms and conditions
5. restrictions on the distribution channels
6. restrictions on ability to underwrite
7. an indirect constraint on the amount of business that may be written
8. restrictions on the types of assets or amount of any particular asset in which the insurance company may invest for demonstrating solvency
73
Risk adjustment is useful for:
1. risk budgeting
- funding may be distributed by DRG
2. measuring healthcare outcomes
- readmission rates
3. measuring relative efficiency
- useful for price negotiations, network selection
4. risk management
- if budget exceed, understand whether this is due to supply or demand factors
5. provider profiling
- sharing information between doctors on a risk-adjusted basis increases awareness of economic impact of clinical decisions
74
Challenges allocating limited healthcare resources
1. demographic challenge
2. technological challenge (leading to increased demand)
3. challenge of sisyphus (increasing life expectancy at older ages)
4. burden of disease
5. access to skilled medical professionals and infrastructure
6. competition or regulation in healthcare (market failures lead to the need for regulation which must be balanced with need for competition)
75
State funding methods
PAYG
- involves establishing the degree of state subsidy, estimating the coming year's tax revenue and outgo and adjust to incorporate healthcare outgo
Forward funding
- analysing level of state provision and producing model of state outgo over a future period, estimating population and workforce trends and forecasting tax revenues and calculating specific healthcare fund such that earmarked taxation will meet intended healthcare provision
Combination
76
How does state incentivise self-provision
- tax relief on premiums
- exclusions from state benefits
- reduced general taxes when insurance is taken out
- subsidies to providers to reduce the cost of healthcare premiums
77
Deterministic model features
- all parameters have fixed values
- produces results in the form of point estimates
- possible to sensitivity test the results of deterministic models by running the models with different parameter values
78
Stochastic model features
- some of the parameters are allowed to vary and have their own distribution function
- stochastic model must be run many times using random samples from the distribution functions
- model produces results in the form of a probability distribution
79
Data requirements
data should be
- relevant
- credible
- available
- reliable for the purpose intended
80
Advantages and disadvantages of independent intermediaries
Advantages
- high level of knowledge and professionalism which means that customers get good explanations
- no infrastructure cost to the insurer
- only pay for what they sell
- they have more experience in the market and will be able to market to their existing customer base
Disadvantages
- paid commission upfront which is a financial burden
- price competition
- will be competing against other companies for sales
high churn rate likely
- no control over sales approach
81
Advantages and disadvantages of tied agents
Advantages
- under insurance company's control
- non-competitive
Disadvantages
- will require some support from the insurer
- need to be trained by the insurer
82
Advantages and disadvantages of own sales force
Advantages
- even more under insurance company's control
- should provide better quality sales with lower churn
Disadvantages
- usually salaried which might be a disincentive to sell
- will require training and management by the insurer
- cost of setting up own sales force can be high
- new company so don't have existing relationships
83
Direct
Advantages
- no commission so theoretically cheaper
Disadvantages
- need some form of database
- not targeted
- lapses are generally much higher
- easier to less less complex products
84
Types of reserves
- UPR
- URR
- OCR
- IBNR
- IBNER
- equalisation reserve
- claims in transit
- mismatching reserve
- statutory solvency requirements
85
Reasons for variability in reserving estimates
- data errors
- data omissions
- data distortions such as
- external influences such as inflation
- internal influences such as changing processes
- changes in type of business sold
- random fluctuations
- methods used to calculate reserves (BF more stable)
- inconsistent number of payment runs not accounted for in the method
- change in basis
86
Chain ladder method
- uses development ratios weighted by cumulative claims values
- the basic chain ladder method applies to unadjusted paid claims using a treatment cohort
- the inflation adjusted method applies an inflation index to past claims to bring them in line with the latest year and to project claims to the expected year of payment
87
BF method
- BF method combines the estimated loss ratio with a projection method (like BCL)
- adds stability
- Bayesian approach
- based on the idea that past development for a given origin year doesnt necessarily provide more information for future claims than the loss ratio
- will be influenced by the choice of loss ratio
88
General considerations when setting assumptions
FLUNC FAR
- take care over choice of assumptions that will have the most financial significance (F)
- consider any legislation or regulatory constraint (L)
- consider the use to which the assumptions will be put (U)
- consider the needs of the client (N)
- achieve consistency between various assumptions (C)
- ensure that bases used are flexible to reflect changing circumstances (F)
- ensure parameters derived from the data are as accurate as possible (A)
- ensure that data used to derive the assumptions are relevant to the insured lives (R)
89
Margin incorporated and price charged will depend on
1. competitive nature of market
2. unique selling point
3. risk appetite
4. credibility, accuracy and relevance of data
5. free assets
90
PMI data should be adjusted for
- differences in policy conditions
- differences in negotiated hospital contracts
- future trends in provider capacity and charges
- etc
91
Expense assumptions
- based on most recent expense investigation
- assumptions split by type of expense (initial, renewal, claim or termination)
- and between per policy, per unit of premium or per unit of benefit
92
Factors that affect a product's riskiness
1. lack of historical data
2. untested market
3. policyholder options
4. high guarantees
5. complexity of design
6. high overhead costs
93
Why monitor experience
1. update assumptions for future experience
2. monitor any adverse trends in experience
3. monitor actual vs expected experience and take corrective actions as needed
5. provide management information to aid business decisions
6. make better informed decisions about pricing and the adequacy of reserves
94
Experience investigations would include
- new business and renewal rates
- mortality, sickness and other contingencies
- claim amounts
- persistency rates
- expenses
- investment return
95
Why conduct AoS
DNCER
1. show the financial effect of divergences between valuation assumptions and actual experience, exposing which assumptions are most financially significant
2. provide a check on valuation data and process, if carried out independently
3. to identify non-recurring components of surplus, thus enabling decisions to be made about the distribution of surplus
4. show financial effect of writing new business
5. comply with regulatory requirement
96
Why conduct EV analysis
1. validate calculations, assumptions and data used
2. reconcile values for successive years
3. provide management information
4. provide data for use in executive remuneration schemes
5. provide detailed information for publication in the company's accounts
97
purpose of new business analysis
check:
- new business strain
- business mix against that assumed in pricing basis
- adequacy of staffing levels
- commission paid against those assumed
98
How can the distribution process be controlled
- monitoring the sales message
- being aware of business churning
- analysing the quality of sales staff
- being aware of overgenerous commission
- monitoring premium receipt
- investing in sales training
99
Reasons for reinsurance
- limit exposure to risk
- avoid large single losses
- smooth profits
- provide expertise
- financial assistance
- increase capacity to accept risk
100
Advantages of quota share
- spreading risk
- writing larger portfolios
- encouraging reciprocal business
- financing and improving solvency
101
Non proportional mainly used in which class of business
short-term or group long-term
102
Problems confronting the actuary
CCRRISPPPPP
1. claims
2. capital management
3. risk management
4. return on capital
5. investment
6. supervisory reserves and SCR
7. policy data
8. policy design
9. pricing
10. profitability
11. product marketability
103
Risks associated with distribution channel
- commit the insurer to new conditions
- bring the insurer into disrepute
- delay premium or claim payments and become bankrupt
104
Reasons for calculating reserves
1. to determine liabilities for published accounts
2. to determine liabilities for internal management accounts
3. to determine liabilities for supervisory accounts
4. to value firm for acquisition or merger
5. to assist in premium rating
6. to influence investment strategy
7. to assist with the assessment of reinsurance arrangements
105
Reserves for long-term insurance
- reserves for in-force policies
- claims reserves
- option reserves
- for group contracts, UPR and URR
106
Reserves for short-term insurance
- UPR
- URR
- IBNR
- IBNER
- claims in transit
- OCR
- equalisation reserve
107
Principles for setting statutory or solvency reserves
- reserves must be sufficient to cover all liabilities arising from contracts
- should be calculated on a prudent basis, taking into account all liabilities
- take account of future premiums if these are contractually due to be paid
- should be calculated on a prudent basis so should include margins
- should be consistent with asset valuation
- appropriate approximations or generalisations may be allowed
- the interest rate used for calculating reserves should be chosen prudently, taking into account currencies, yields and reinvestment yields on the assets
- demographic, persistency and expense assumptions should be prudent
- the valuation calculations conducted over time should not suffer discontinuities arising from arbitrary changes to the basis
- valuation method should recognise the emergence of profits appropriately over the policies' lifetime
- valuation basis and method should be disclosed
108
Active vs passive valuation
Passive
- relatively insensitive to market changes
- basis is updated infrequently
- easier to implement
- involves less subjectivity
- relatively stable profit emergence
Active
- based more closely on market conditions
- assumptions updated on a frequent basis
- more informative in terms of understanding the impacts of market conditions
109
Disadvantages of IRR
- may not exist or may not be unique
| - cannot be related to other indicators such as sales costs or premium income
110
What factors should be considered when pricing
- profit criteria
- marketability
- competitiveness
- reserving and solvency capital requirements
- reinsurance impact
- regulatory constraints
- reviews
111
Special characteristics of group business
1. free cover
2. control of the intermediary
3. limited insured information
4. changes in the workforce
5. flexible benefits
112
Burning cost
- Estimated cost of claims in the forthcoming period
- calculated from previous years' experience
- adjusted for changes in numbers insured, the nature of cover and medical inflation
113
Steps in calculating risk premium
1. choose a base period over which to collect claims and exposure data
2. collect data, checking accuracy and appropriateness of the data
3. split data into homogeneous groups
4. calculate historical burning cost premium for each group
5. analyse the data (to identify trends etc)
6. adjust and project forward to obtain future risk premiums
114
Accelerated CI claims incidence rate
ix+(1-kx)*qx where kx is the proportion of deaths over [x;x+1] due to critical illness
115
Health and care modelling complexities
- distribution of claim frequency
- distribution of claim amount
- sales volumes
- modelling
- data limitations
116
How is modelling complicated
- the role of genetics
- trends in anti-selection
- the quality of underwriting
- the use of multi-state models
- guaranteed or reviewable product alternatives
- different approaches for different claim types
- the need to model the claim chain
- cross-subsidies between new business and renewals
117
GLM
- model the behaviour of a random variables whose value depends on the values of several other characteristics
- generalistion of normal model for multiple linear regression
- drawbacks of normal model for multiple linear regression
- it assumes response variable has normal distribution
- normal model has constant variance which may not be appropriate
- it adds together the effects of different explanatory variables
- it becomes long-winded with more than two explanatory variables
- GLM overcomes this problem by allowing response variable to take on any distribution from the exponential family and introducing a link function
118
Analysis of significance of explanatory variables for nested models
Goodness of fit of nested models:
If scale parameter known - change in scaled deviance follows a chi-squared distribution
if scale parameter is unknown - goodness of fit can be tested using F distribution
(D1-D2)/(df1-df2)(D2/df2)
119
Analysis of significance of explanatory variables for models that aren't nested (goodness of fit)
AIC = -2*loglikelihood + 2*number of parameters
| the lower the AIC, the better the model
120
Testing the appropriateness of models
pg. 679
- deviance residuals
- standardised pearson residuals
- residual plot
121
Residual plot requirements
- symmetric about x axis
- average residual of zero
- fairly consistent across width of fitted values
122
Measure of influence of data point on model result
Cook's distance
| >1, considered to merit closer examination
123
Model refinement *read through this again
- interaction terms can be included (complete or marginal)
- offsets can be used to constrain certain elements in the model in such a way that the fitted relativities for other elements adjust to compensate
- aliasing occurs within GLMs
- intrinsic aliasing occurs because of dependencies inherent within the definition of the covariates
- extrinsic aliasing occurs when two or more factors contain levels that are perfectly correlated
- smoothing parameter value (by grouping levels of factors)
124
Obtaining the predicted values from a simple GLM
1. specify the design matrix X and the vector of parameters B
2. choose the distribution for the response variable and the link function
3. identify the likelihood function
4. take the log to convert the product of many terms into a sum
5. maximize the log of the likelihood function by taking the partial derivatives with respect to each parameter, setting them equal to zero and solving the resulting system of equations
6. compute the predicted values
125
Deviance
- Deviance compares the observed value to the fitted value, with allowance for the weights, and assigning higher importance to errors where the variance should be small
- deviance residuals will be more closely normally distributed than raw residuals as the deviance corrects for the skewness in the distribution
126
Scaled deviance
Deviance adjusted by a scale parameter to give a standardised measure that can be compared to other models
127
How to measure uncertainty of a parameter estimate used in a GLM
Cramer Rao lower bound
| standard errors can be found from hessian matrix
128
Standardised pearson residual
- difference between observed response and fitted value, adjusted for the standard deviation of the predicted value and the leverage of the observed response
- possible to compare standardised pearson residuals, even where observations have different means.
- however, doesn't adjust for the shape of the distribution
129
Assumptions underlying the bootstrapping chain ladder method
- the run-off pattern is the same in each development year
- incremental claim amounts are statistically independent
- the variance of the incremental claim amounts is proportional to the mean
- incremental claims are positive for all development periods
130
Determining retention level
1. decide on some criterion for claims volatility beyond which the company cannot go
2. for differing retention levels, model the function {total claims net of reinsurance - total risk premiums net of reinsurance premiums}. This modelling can be done stochastically, varying the risk experience
3. look at the results of the modelling to determine retention that will satisfy your criterion
4. we could use this retention level - otherwise we could see if this protection could be achieved more cheaply using a risk experience fluctuation reserve
5. we could instead assume that some of the cost of the risk premium reinsurance is instead going to be spent on financing risk experience fluctuation reserve. The cost of holding reserve is M(j-i)
6. compare the protection offered under this new construction against that offered by the previous arrangement
7. try this for other arrangements and determine which combination offers the greatest protection for a given cost.
131
Revised loss ratio
= reported loss ratio + expected loss ratio*remaining development
132
GLM process
- collect claims data based on historical claims cost per benefit and option type
- ensure credibility of data per cell
- claims will need to be determined on a PLPM basis
- may need to adjust for inflation, run-off and any benefit changes
- identify risk factors that need to be accounted for
- choice of GLM and link function needs to be reasonable and appropriate
- gamma model would be a good model for claim amounts
- running the GLM will give a set of factors per risk factor, benefit amount type and benefit option
- this can be used to score StarMed
- for accuracy, sensitivity testing should be done
