Chapter 23 Reserves & solvency capital requirements Flashcards
Types of reserves:Short-term
- UPR
- URR
- IBNR
- claims in transit reserves - reserves in respect of claim reported but not assessed or not recorded.
- OCR
- INBER
- equalisation or catastrophe reserve
The role of statistical and case estimates:Long-term
- claim amount payable is known once claim is submitted, for most forms of long term insurance.
- however for annuity type contract period for which payments will carry on is not known.
- This may not be true for CI insurance, insurer will hold a reserve for claims reported but not fully settled, using amounts in policy docs and increase this by inflation where appropriate.
- reserves for benefits which provide income will be calculated using statistical methods.
- only small volume will be reserved for using case estimates. estimated using likely duration of claim.
- most of long term insurance provisions held are in respect of futre claims, acknowledging that a level premium pays cover of increasing probability of claim.
- actuaries may use deterministic or stochastic models to estimate potential claims outgo & set provisions.
The role of statistical and case estimates:Short-term
-PMI is indemnity thus the amount payable is determined by costs incurred and is not known with certainty until treatment is complete.
- statistical approach is used to estimate amount of claim.
- although certain large claims will warrant case-by-case reserving.
- This involves calculating expected total claim amounts for outstanding claims based on relevant past experience.
- each claim is unique in that many different claim causes can arise, so cost of treatment can vary considerably.
Statistical estimates
- this is appropriate for particular types of homogeneous claims where portfolio is large enough & is deemed to be stable.
- these methods estimate outstanding claims for cohorts based on historical trends and patterns, and adjusting for known or anticipated future changes.
- most statistical methods work from tabulations of claims that have recently been paid.
- portfolio might be segmented by contract type, distribution channel, location, etc.
- assumptions are made about the stability of claim development and past patterns will continue into the future.
Basic chain ladder assumptions
- BCL assumes amount of claims paid in each development year from each origin year is a constant proportion of the total claim amount from that origin year.
- BCL assumes past inflation continues into the future.
- Inflation-adjusted BCL can be used.
The Borhhuetter Ferguson method merits
- Relies on assumed run-off patten and an estimate of the ultimate claims for each cohort.
- The estimate is usually made using the loss ratio method.
- The external estimate is apportioned between the past and future (as at the date of the reserving exercise).
- it improves on the crude us of a loss ratio by taking into account information provided by latest development.
Assumptions underlying the Bornhuetter Ferguson method
- underlying method is same as BCL.
- together with that the estimated loss ratio is appropriate
- this method could be viewed as using a Bayesian approach.
Steps in calculating a reserve using BF method.
- Determine initial estimate of the total ultimate claims from each treatment month using premiums and initial expected loss ratios.
- Multipy these estimates by the proportion outstanding (1-1/f) determined from claims development table. These are estimates of the reserve for each treatment month.
- add these figures to the claims paid to date give an esitmate of the ultimate loss for treatment month.
Explain boostrapping and what it is used for?
- It can be used to estimate the variance of the IBNR reserve.
- Shows the extent to which a reserve can vary on either side of its mean.
- A reserve method is chosen and used to produce a fitted model for historical data.
- The residual values are re-sampled with replacement to generate a number of pseudo run-off triangles.
- These pseudo run-off triangles can then be used to estimate the distribution of IBNR values.
Weakness of statistical methods
- Outstanding claims might be impaired by the errors, omissions or distortions in the data, which invalidate the underlying assumptions.
- These distortions however do not mean the statistical methods should not be used.
Assumptions underlying bootstrapping the BCL
- the run-off pattern is the same for each origin period
- incremental claim amounts are statistically independent
- the variance of the incremental claim amounts is proportional to the mean
- incremental claims are positive for all development periods.
Distortions in data and results can occur due to a number of reasons
- external influences, such as inflation or changes in underlying nature of risk
- internal influences such as underwriting, claims settlement or recording procedures or reinsurance arrangements
- changes in the type of business attracted in each treatment class
- random fluctuation or large claims in a small portfolio.
Discuss the need for solvency capital
- insurance supervisors reqiure that an insurer maintain at least a specified level of solvency capital in addition to reserves or technical provisions held.
- This solvency capital can be seen as providing an additional level of protection to policyholders.
1. Against drop in asset values
2. Against adverse experience relative to that assumed in reserves - The level of SCR under regulation may be specified as a formula or it may be based on a risk measure such as VaR.
Interplay between reserves and SCR
- when considering the adequacy of reserves to be set up it is important to do this within the context of SCR and not in isolation.
- Similarly the adequacy of SCR cannot be looked at in isolation of the reserving requirements.
- in some countries reserves are set up on a relatively realistic basis ie relatively small margins from the expected values.
- there us a requirement for a substantial level of SCR determined using risk-based capital tecniques.
Value at Risk approach
- An example of a risk-based SCR approach is the use of a VaR measure.
- normally expressed at a minimum required confidence level (eg 99.5%) over defined period (eg one year).
- The supervisory balance sheet is subject to stress tests on each of the identified risk factors.
- A VaR of R10m with 99.5% means that there is only a probability of 0.5% of a loss greater than R10m.