Monitoring the experience

Reasons to monitor the experience / measure the current profitability

• This is done to review the suitability of the models and assumptions used and, if necessary, to make appropriate revisions.
  This is seen as the best source of information to help model future experiences.
• it allows for any adverse trends in the experience to be detected quickly and for corrective action to be taken. At the same time, it can be used to identify any tranches of business that are unprofitable (and plan new business strategy accordingly), e.g. stop selling unprofitable contracts, develop new tranche of business that are more profitable
• to determine sources of profit (which will assist with how to distribution the profit),
• Corrective action can be taken both in regard of new business to be sold in future and, where possible, existing in-force business.
• to determine effectiveness of risk management strategies (e.g. investment, reinsurance etc).

If an assumption is underestimated for reserving, what will happen?

• If .. is underestimated then the liabilities will be understated in the accounts and the apparent solvency will be overstated.
• Excessive profit will be disclosed, which will lead to a higher tax charge and possibly excessive dividend payments, which in turn may lead to future solvency problems.
• Investment freedom will initially appear to increase and the company may apply assets to other projects.
• Inappropriate reinsurance arrangements may be set in place.
• If too low a rate of inflation is used for pricing new business, then the premium rates charged would be too low. If the position is reassessed at a later date, adverse market reaction may make it difficult to attract good business in future and there may be problems with the regulators

Before carrying out any monitoring exercise, the actuary must have access to a reasonable volume of stable and consistent data. This will enable to actuary to

• identify any trends in the past data
• and assist with determining appropriate assumptions for future experience.

It may be necessary to adjust some of the past data to ensure consistency both within the past data itself and with respect to the future experience to be estimated (e.g. if underwriting practices or policy terms have changed recently). Compare on the like-to-like basis
Then, the data must be divided into broadly homogeneous risk groups according to the main risk factors.
What is the main problem with failing to ensure broadly homogeneous data within a particular cell?
change in mix of underlying risk will cause actual experience to differ from expected (even if all other assumptions made are borne out in practice)
However, it is also important to ensure that enough data is contained within each cell to allow a credible analysis of experience. This is a particular problem for smaller insurance companies and/or pension funds.

After sub-dividing the data into broadly homogeneous groups, the actual experience can be compared with that expected.
For statistical factors such as mortality (or withdrawal), the observed experience for each homogeneous cell is given by: \[ actual\,\,\,mortality\,\,\,rate = \frac{observed\,\,\,number\,\,\,of\,\,\, deaths}{total\,\,\, exposed-to-risk} \] This can then be compared with the expected experience for the particular cell to identify any material differences.
It is crucial that the denominator used reflects the total exposed-to-risk for the particular factor being considered (divided into the same homogeneous cell structure).

Using the results

Why / How to adjust past data

• Before using the results of an analysis of experience, it is crucial to consider whether the period under investigation was typical and likely to be representative of future experience.
• For example, the past data may have been influenced by abnormal events (although, if possible, the data should have been adjusted to reflect these before carrying out the analysis).
  • quite likely that we will have a few very large claims in the past data as a result on particularly bad experience
  • these may be left in the data unadjusted, truncated and spread over an appropriate number of years or even removed entirely (if they can be considered so exceptional as to almost certainly not occur again)
  • thus, appropriate treatment will depend on the extent to which similar claims are likely to occur in future
• Economic experience in particular may be cyclical and, where necessary, this should be reflected when determining appropriate assumptions for future experience.
• Also, there may be gradual trends evident in the past data (e.g. mortality rates have decreased steadily over time). However, consideration needs to be given to the extent to which these trends can be expected to continue in future.
  • if trends in claims numbers and/or claim amounts are detected in the base data, it is important to attach more weight to recent experience
  • trends should also be investigated to see whether or not they are likely to continue into the future
    • for hurricanes, we may expect that an upward trend in claim numbers will continue (or even worsen) as a result of global warming
    • or, we may see a downward trend in claims numbers (and amounts) as hurricane defence systems and building standards improve
  • need to have sufficient data to differentiate between a trend and a one‐off change in experience (perhaps due to random fluctuation or one‐off change in company and/or market practice)
  • need to make an allowance for claims inflation (to investigate whether claim amounts are actually increasing in real terms or simple due to inflation)
    • also, need to determine appropriate assumption for future claims experience
• Changes in the underlying risk environment
  • can be difficult to deal with: they may show up as trends and be dealt with as such. alternatively, major elements of the risk could be separated in the base data and projected separately
  • also, need to make an assumption about the future mix of risks covered
• Changes in cover provided
  • can also be difficult to allow for
  • major changes are likely to involve the perils covered (e.g. we may now exclude properties in certain very high‐risk locations) and/or limits and excesses applied to each claim
    • may also arise from changes to underwriting (e.g. requirements to have basic defences in place against flooding) and/or to claims settlement procedures
    • if a peril is no longer to be insured, then it may be possible to simply exclude these claims from the data
    • however, if a new peril is to be insured, then it may be necessary to use external data (e.g. market statistics, government statistics, data from other insurers)
• Effect of reinsurance arrangements
  • claims experience should be considered gross of reinsurance, as reinsurance arrangements may be different in future

Why do we need to monitor the experience?

• Monitoring of experience is fundamental to the effective management of the risks faced by financial institutions.
  • The environment is constantly changing and monitoring the effect of past actions can help in revising the strategy for assessing and managing the risks faced.
  • The actuary will use the results of experience analyses to make changes to the models and assumptions used for pricing, reserving and setting contribution rates.
• In practice, this is an iterative process.
  With more past experience, the actuary can develop a better understanding of the complex workings of the institution and consistently improve the models and assumptions to describe future experience.