Chapter 12.1 Data (very high level)

Question 1

Types of data

What type of data might an insurer work with? (2)

What kind of ‘other data’ might insurer use (5)

Answer 1

1. Policy data
   
   1. eg age at entry, policy term, duration in force, sum assured
   2. maintained by insurer
   • 2. Other data
        1. ​Internal data from other products
        2. External data
     3. Insurance industry data
     4. National stats
     5. Overseas markets

Question 2

Uses of data

We’ve covered the various data an insurer might use and what its potential sources of data might be; what would an insurer use this data for? (9)

(Remember a useful acronym from CA1/ARM)

Answer 2

Main actuarial reason for policy data:

• Regular valuation of liabilities

Uses of data: MAFIA PEPRS acronym from CA1/ARM :)

1. Marketing
2. Administration
3. Financial planning and management
4. Investments
5. Accounting
6. Provisioning/reserving
7. Experience statistics and analysis/investigations
   
   1. done to give appropriate advice
   2. inaccurate data => inaccurate advice
   3. model points:
      
      1. not always practical/cost effective to do investigation on whole book.
      2. can use model points, then scale up results
      3. the fewer the model points, less accurate
8. Premium rating or product costing
9. Risk management
10. Statutory returns

Question 3

Risk due to data: risks due to policy data

Explain how policy and other data can be a source of risk to a life insurer in terms of policyholder data (10)

Answer 3

Policyholder data: main risk is poor maintenance

• company may not maintain adequate, accurate and complete records required by actuary and so
  
  • results of supervisory valuation may be innacurate
  • other investigations may be inaccurate and so advice innapropriate
• risk when used in investigations
  
  • eg investigations modelling part/whole of business
  • risk that model doesn’t adequately reflect business
  • properly referred to as a model or modelling risk - risk that results from model office being an over-simplification of the real company in terms of the portfolio of policies currently in force and as a consequence behaving unrealistically

Question 4

Risk due to data

Discuss main issues for health and care products compared to life insurance products (3)

Answer 4

For health and care contracts specifically, the following issues are experienced (more so than for life insurance:

• Limited credibility
  
  • smaller policy volumes e.g CI and LTCI
  • lower incidence rates e.g. IP and CI
• Limited applicability of past data
  
  • Changes to products/market over time
  • Limited relevance of past data due to sensitvity to
    
    • socio economic conditions (IP)
    • medical advances (CI)
    • longevity/health at older ages (LTCI)
• Limited applicability of industry data
  
  • ​Due to heterogeneity of products and markets

Question 5

Risk due to data: risks due to other data/non-policy data

Explain how other data (ie non-policyholder data) can be a source of risk to a life insurer (5)

Comment on these risks in the context of other data which is ‘internal’ (4)

Comment on these risks in the context of external data (3)

Answer 5

Other data (internal, and or external data)

• other data may be used for actuarial assumptions eg used to determine claim incidence, surival periods, mortality
• data may
  
  • be adequate, but not suitable for the task at hand (see comment regarding external data below)
  • be purely purely inadequate or undreliable indicators of future experience
  • be of insufficient volume
  • suffer from population differences causing data to no longer be applicable to population being considered

Internal data

• even if other data is internal, it may be inadequate for many reasons:
  
  • missing data
  • inaccurate data,
  • insufficient volume,
  • inappropriate for task at hand

External data

• inadequacy of data is worsened where overseas/external data is used..
• because policy conditions may differ significantly from what’s used in company
• any external data used may be inadequate or unreliable and even when reliable it may prove to be inappropriate