Modelling Flashcards
What should be the focus for data?
Quantity and quality together
What reasons are there for poor quality data?
SRG
Systems - poor design of data system
Recording - poor control of data recording
Good processes implemented, time lag in enough data
What are the consequences of poor quantity of data?
Non-homogeneous grouping
Low credibility of grouping
What is the consequence of poor quality data?
Unreliable results
Why is the proposal form important for data?
Prime information source
Well designed/unambiguous questions
Ratings facotrs/underwriting to be included
Cross checking and validity of claims possible
Updating policy information
Historical data gathered from it
What is risk classification?
Splitting data into homogeneous groups
Why do risk classification?
Appropriate use of data Experience of group stable Inaccurate provisions if didn't do it Sensitivity testing needs to be done If group contains few model points, need to group with more at cost of homogeneous
What historical data should be used in basis in general?
Investment return, benefit growth, demographic assumptions, changes in a persons life
What parts of past investment return would you look at?
div yield
returns on asset classes
inflation data
What parts of historical beenfit growth would you use?
Salary levels for country/industry/company
nflation index
What demographic assumptions would you look at historical data for
Mortality for decrements
What changes in a persons life would you look at in using past historica data?
Leaving employment illness retiring regulation past price inflation
Why wouldn’t you use past averages in your basis?
Social and economic changes
Use what effect the changes caused in the past for your future predictions
What are unsuitable uses of unadjusted historical data?
Economic Inflation Norminal data Dividend levels Mortality data
Why shouldn’t you use past inflation in your basis?
linked to past conditions, better to use current govt bonds and projections
Why not use past nominal data for econ assumptions?
Price inflation will strip out fluctuations
What historic economic data is unsuitable for using unadjusted in your basis
Inv returns
salary
dividend growth
How would you use historical economic data in your basis
Strip out fluctuations caused by fiscal changes
How would you adjust dpast dividend levels for use in basis
Adjust for tax changes on dividend income
How would you adjust mortality data (historic) for use in basis
for mdeical advances and tends
focus on more recent data more
What are the problems with grouping past data in homogeneous groups?
Information desired unlikely to be available
Available data small so uncredible results
The balance in the homogenous group may change over time, with workers shifting
adjust it in a subjective way
What are the problems with getting data from the state or groups of companies?
Stats recorded may change
Errors are possbiel, especially in older records
Census data has ppl who can#’t afford or buy insurance
Correct mortality table should be used and adjusted
What factors determine accurace of assumptions use in basis
Purpose of assumption
Sensitivity of result to assumption
Best estimate or MAD
Individual cashfolws important (use accurate)
Capital value only important (not accurate is allowed)
Consideration of financial consequences
Expert witnesses must use same assumptions type (BE /MAD)
Awareness of implict assumptions in model like mortality not changing in open pens scheme
How do you allow for risk in assumptions?
Change risk discount rate
Stochastic discount rate
Apply MAD’s to expected values
