Data, Assumptions and Modelling Flashcards

1
Q

Uses of data

A
  • -Statutory returns
  • -Investment monitoring
  • -Risk management
  • -Management info & financial control
  • -Accounting
  • -Pricing
  • -Experience stats and analysis
  • -Marketing
  • -Admin
  • -Provisioning
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Sources of data

A
  • -Tables
  • -Reinsurers
  • -Abroad
  • -Industry
  • -National statistics
  • -Existing contract experience
  • -Regulatory reports and competitor accounts
  • -Similar contract experience
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Disadvantages of industry-wide data

A
--Heterogeneity
(places, styles, processes, products)
--less Detail
--less Flexible
--less Up-to-date
--dependent on Members' data systems
--not all Contribute
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Characteristics of good quality data

A

–Complete
–Accessible
–Consistent
(format, comparable across systems - unambiguous inputs)
–Credible
–Relevant
(up-to-date, detail)
–Accurate
–Quantity of data

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Data checks

A
--Cross-check 
(against other sources)
--Consistency
(ave, min, max)
--Recon
--Spot checks
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Factors affecting assumption setting

A
  • -Financial significance of assumptions
  • -Legislative and regulatory requirements
  • -Use of the model to which the assumptions contribute
  • -Clients’ needs
  • -Consistency between assumptions
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Requirements of a good model (pt 2)

A
  • -Valid
  • -Adequately documented
  • -Rigorous
  • -Inputs to parameters need to be appropriate
  • -Arbitrage free (economic interpretation)
  • -Behave reasonably
  • -Length not too long to run or too expensive
  • -Easy to understand
  • -Communicable and transparent workings/outputs
  • -Representativeness (reflects risk profile of contracts being modelled)
  • -Independent verification of outputs
  • -Sensible joint behaviour of variables
  • -Parameters allow for all significant features
  • -Simple (parsimonious)

–Clear results (transparent)
–Adaptable (versatile in implementation)
–Refineable
–Developable
(evolution process for last 3)

...old:
--appropriate Parameters
--Communicable 
--Valid
--Verifiable
--Versatile
(applicable to different contexts)
--major Features captured
How well did you know this?
1
Not at all
2
3
4
5
Perfectly