Chapter 20 Flashcards
Setting Assumptions (13 cards)
key factors affecting the choice of assumptions (10):
- the use to which the model will be put (purpose of the model)
- Consistency between assumptions
- Legislative and regulatory requirements (assumptions may be defined in regulation or legislation)
- The needs of the clients
- purpose of valuation
- degree of accuracy (links to financial significance of assumption)
- best estimate (need to know this on order to apply margins for adverse experiences)
- implicit margins (assumptions in the background that may be critical to the credibility of results)
- explicit margins
- sensitivity and significance of assumption
2 types of assumptions:
-Demographic
* More slowly changing
* Modelled deterministically due to their slow changing nature
-Economic
* More variable and volatile
* May use stochastic model to account for the variability
Examples of demographic factors (8):
- Mortality rates (Rates of mortality before and after retirement)
- Morbidity rates (Rates of ill-health retirement, Rates of retirement in good health)
- Recovery rate
- Rates of withdrawal (lapses, surrenders)
- New entrants rates
- Proportion married (marriage rates) – affects whether a personal would get a spousal annuity or not
- Spouses’ mortality
- Salary scale (proportional increases)
Demographic can be split into 3 categories:
- Base (setting the base is where the past data is very useful)
- Trend (the way it changes in the future e.g. mortality improving, or morbidity, disability rates are getting worse may be due to modernized)
- Cycles (other factors come to play to affect the trend e.g. mortality improving can be disrupted by high unemployment rate, in economic hardship times, more people are inclined to get disabled)
Examples of economic factors (11):
- asset returns (Investment returns on the assets held)
- Discount rate (need to build in the return above the risk free rate)
- Yield curve (risk free needs to be modelled)
- Earnings inflation
- Exchange rates (if benefits with an international component to it and if the country’s inflation is influenced by the currency movements)
- Unemployment rate (more economic because it changes rapidly -
- Pension increases
- Expense inflation (based on price inflation) – influences by book growth or book shrinkage
- CPI rates
- Own expense inflation – driven by factors related by the business alone
- GDP growth
What are the sources of data - base? (7)
-internal data
-standard actuarial tables (can be used to adjust internal data)
-population data
-standard actuarial tables from similar countries
-reinsurer’s data
-population data from similar countries
-industry data
What causes low credibility of past data? (11)
- too out of date (need to adjust for the number of years it is out of date)
- target market/mix different
- underwriting strictness different
- changed claims practices
- too small a volume or conversely too heterogeneous (balance between volume and heterogeneity)
- not sufficient at certain age ranges
- credibility theory important (number of data and what you are observing) - easier to get credible data for multiple claim industries than those that only have one claim event
- raw data poor
- extraordinary events - anomalies (covid deaths in mortality rates need to be stripped out but the likelihood of it needs to be accounted for, if it is assumed to happen every 50 years, then) - either disregard the data or adjust to a more ‘normal’ value for that period
- abnormal fluctuations - changes in economic and fiscal policy (it is necessary to use the earlier data and remove the economic and fiscal conditions that differ from those that currently exist)
- random fluctuations - it is useful to convert past economic data into real terms to remove much of the fluctuations
- main source of data (cycles)
- economic impact on demographic assumptions
o much of the demographic data will also be affected by economic changes
o in a recession (more people drop below the poverty line): more stress, more suicides, higher mortality - analyse past cycles
- identify leading indicators
- main source of data (trend)
- projections from standard actuarial tables (if you have reinsurance, they would most likely tell you what these projections are)
- medical research on fundamental drivers of decrements
- comparison of standard tables over time
- medical advances and technological developments
what are the downsides of using reinsurer’s data? (3)
o the data would be very skewed towards the larger policies as those would be of interest (larger policies)
o reinsurers usually reinsure policies that are considered sub-standard lives since these are of higher risk and most times you would be pricing for standard lives (riskier policies)
o can provide the rates in which they are prepared to take on the risk (and they will price this according to their own experience and you can use that to infer the mortality experience) but the price is not only based on the risk involved but how the risk is managed
What is the main problem with using industry data?
The pooled data may not be a true reflection of the industry as not all players in the industry would contribute data
What are the 2 main problems with using population data/national statistics?
o there is usually a lag factor in national mortality experience since they are based on the census, and it is taken every 10 years
o will rarely reflect the insured population (also, census data contains lower socio-economic groups that distort the experience of lives affecting insurance contracts)
- When using past data, the actuary needs to consider how to deal with and make adjustments for (6):
- Abnormal fluctuations (one off impacts) – adjust to a more ‘normal value’ for that period
- Changes in the experience with time – adjust past data to account for these changes
- Random fluctuations – use real terms
- Changes in the way in which the data has been recorded – may lead to inappropriate assumptions if changes are not recognised
- Potential errors in the data – less likely due to better computing power but older data may have greater risk of data error to the extent in which it results in the data being useless
- Changes in the mix of homogenous groups within the past data – the information necessary to split the data reliably is unlikely to be available and splitting the data may reduce reliability.
- Changes in the mix of homogeneous groups to which the assumptions apply – may adjust for the differences in the characteristics of the individuals concerned
