Ch 35: Monitoring experience Flashcards
In what different ways can experience monitoring be done?
(1,5)
(2,3)
-
Directly
- want to directly analyse experience with respect to a specific item of interest, so do an analysis for this
- in this chapter we cover following direct investigations
- investigation for mortality + other contingencies
- persistency investigation
- expense investigation
- investment return investigation
-
Analysis of surplus
- investigate financial impact of any difference between actual and expected by performing analysis of surplus
- tells us
- which component is responsible for surplus arisen
- where remedial action can be taken
List reasons for monitoring experience as part of the control cycle
(6)
- Update assumptions as to future experience
- Provide management information to aid business decisions.
- Monitor trends in experience
- Monitor actual compared to expected experience and take corrective actions as needed
- Make more informed decisions about pricing and adequacy of reserves
- Develop earned asset share
List 4 reasons a life company may need assumptions regarding future experience
- Model office work
- EV work
- profitability monitoring
- financial projections
- asset-liability modelling for setting investment strategy
- determining reinsurance requirements
- Pricing
- Valuations
- Setting discontinuance terms
List ways in which experience analysis may help provide management information and take corrective actions (5)
By helping managment to identify
- profitable
- products
- sales channels
- markets
- effecient sections of business
- successful investment strategies
What are the general steps involved in an experience analysis? (broadly speaking) (4)
- Decide what type of investigation needs to be done
- direct
- analysis of surplus
- Gather required data
- Conduct analysis
- Use results
Discuss the data required for monitoring experience
Basic requirements of good data (3)
Splitting data (2)
Period (2)
Level of detail (2)
- Basic requirement is for data to be
- of sufficient volume
- consistent
- adequate to deduce trends and future experience
- Data should be split into homogenous groups
- according to relevant risk factors
- balance between homegeneity and credibility
- Period over which data is collected is very important
- sufficiently long time period for enough data volume
- …but too long time period, might not give info about recent experience
- Level of detail depends on
- volume of data available
- ideally want split at least for different contract classes
What do we mean by ‘big data’ and how have technical developments changed the insurance landscape in this regard?(2)
Give an example of big data (1)
Big data
- big data essentiallly refers to large volumes of data
- technical developments => insurers can handle/analyse large volumes of data more easily
Big data example
- banks with insurance subsidiaries selling insurance mostly to own customers (‘bancassurers’) amass large volumes of additional data on the insurance customers eg personal spending habits and travel locations
List some advantages of big data (6)
Big data advantages
- allow better understanding + analysis of risks…
- …hence better predict future behaviour
- develop more sophisticated + detailed risk classification…
- …allowing for greater ability to select preferred risks
- drive better experience through monitoring
- earlier identify changes in individual risks
- being able to intervene/influence PH behaviour
- other data sources
List some disadvantages of big data (5)
Big data disadvantages
- reputational damage
- privacy concerns
- data protection failures
- regulation changes
- regulator preventing certain data being used
- fines for misuse of data
- data issues
- collected data may be inaccurate, incomplete, or irrelevant
- modelling risk: complex models=> choice of wrong model
- expenses: collecting/analysing data vs benefits
List the types of experience investigations an actuary might conduct (4)
- Mortality and other contingencies
- Persistency
- Expenses
- Investment return
Mortality experience
List the classifications by which data (both claims and exposed to risk) would be ideally sub-divided for the purpose of analysing the mortality experience
(8)
(2)
Most useful classifications would be
- type of contract
- age
- sex
- duration from entry
- smoker/non-smoker status
- medical/non-medical status
- source of business
- location
Further classifications useful if sufficient data
- eg location broken down to postal code
- occupation (at least broad occupational group)
List classifications by which data (both claims and exposed to risk) would be sub-divided for the purpose of analysing the persistency experience (10)
Which splits are used in practice?
Subdivision and analysis of persistency experience data would usually be by:
- type of contract
- endowmnts usually > persistency than term assure
- duration in force
- usually lower at start of contract
- sales method
- more suitable product sold => better persistency
- target market
- more suitable product sold => better persistency
- frequency of premium
- monthly prem=> more chance stop paying than annual prem
- size of premium
- big annual prem may be less affordable than smaller monthly prem
- premium payment method
- debit order persistency > cash payment persistency
- original term of contract
- gender
- age
- usually worse experience for younger ages
The only first 3 are used as they are particularly important
Give 3 other factors, external to life company, that may also influence persistency rates (3)
What impact would they have on the persistency analysis?
External factors influencing persistency
- Economic situation
- Competitive situation of product
- Perceived value of product to customer
Impact on analysis
- these factors wouldn’t be used explicitly in analysis
- but may be used to understand/explain patterns in experience
Outline how full withdrawal rates can be determined for each homogenous group of lives analysed (7)
- for each homogenous group
-
divide contracts issued in company’s last financial year
- into corresponding number that survive in-force until first policy anniversary
- to give first-year persistency rate
- first year withdrawal rate = 1- first year persistency rate
- exclude deaths and maturities from calc (if material)
- repeat for subsequent years to obtain second-year, third year, etc withdrawal rates
- by looking at number surviving from number of contracts, in each group, that have their first, second, etc policy anniversary in last financial year
What 2 different ways may be used to determined persistency rates? (2)
What adjustments may need to be made to data? (2)
What other analysis may we consider regarding persistency?
Persistency rates may either be determined
- over a period: number surviving period, compared to start of period
- cumulatively: number surviving to end of period, compared to number at contract inception
Adjustments may lead to regrouping of data
- if differences for rates due to small risk cell sizes
- depending on what management wants to see
Other persistency analysis we may consider
- analysis of paid-up pols may be done as subsidiary part of overall investigation
- may analyse income/partial withdrawal rates, depending on product design