Chapter 13 - Risk Flashcards
What type of data might an insurer work with?
Policy data
eg age at entry, policy term, duration in force, sum assured
maintained by insurer
Other data
+Internal data from other products
+External data
+Insurance industry data
+National stats
+Overseas markets
Define 3 risks which normally arise in investigations
e.g. when setting mortality assumptions for pricing
- Model risk:
Inappropriate/erroneous probability distribution chosen for underlying model e.g. future mortality. - Parameter risk:
Underlying model correct, but parameters inadequate (don’t reflect future experience of lives insured/to be insured) - Random fluctuation risk:
Actual future experience may not correspond with model/parameters adopted (even though these adequately reflect lives insured/ to be insured)
First 2 risks always exist (impossible to predict future with complete certainty).
Extent of 3rd depends on numbers exposed to risk: smaller number = greater risk.
The risk that actual expenses are higher than expected or allowed for e.g. due to inflation is called expense risk.
The risk that charges received are lower than expected or allowed for is more typically classified as other types of risk. Give 3 examples of what these other risks might be.
- Investment performance risk (e.g. if charges are fund-based)
- Persistency risk (e.g if charges required to recoup initial expenses are not received due to high withdrawal)
- New business mix or volume risk (e.g. if charges are linked to average size or volume of new business)
Explain how persistency may be a source of risk to a life insurer (3)
- Financial risk that surrender vaue is higher than asset share at time of withdrawal…
…made worse by mismatching initial expenses vs charges to recoup expenses
- Selective withdrawals negatively influencing mortality experience
- Reduce business volumes means risk of increasing per-policy expenses
What is the main risks from claims experience for health and care products
- mis-estimation of the transfer probabilities in the underlying multi-state model (IP and LTCI)
- rates of diagnosis (CI)